CHAPTER 3

The Botany of Citrus and Its Wild Relatives
BY WALTER T. SWINGLE
Revised by Phillip C. Reece

      This account of the citrus fruits and their numerous wild relatives has been written in the hope of interesting both citrus growers and expert citrus investigators in this large array of plants.   Nearly every one of them has some striking character—beauty of foliage; fragrance and beauty of the flowers; bright color of the fruits; or, more practically, probable value as a rootstock for some of the commonly cultivated species of Citrus or even possible use in breeding new types of citrus fruits by hybridization.
      The vast majority of citrus fruits and their wild relatives are native to southeastern Asia, the East Indian Archipelago, New Guinea, Melanesia, New Caledonia, and Australia; another group occurs in tropical Africa.   Many of the most interesting relatives of Citrus have been collected only once or twice, and frequently flowers or fruits are still imperfectly represented even in the largest botanical museums of Europe and America.   Consequently, it has been impossible to have any consistent or uniform schedule for describing these plants in the present chapter, but each genus has been given the best treatment that the material available for study would permit.   As a result, there is presented a connected account of the whole orange subfamily and of the two tribes into which it is divided, with some remarks on the geographical distribution and probable evolutionary history of some of the more important genera of each of these tribes.   The thirty-three genera here discussed, however, have been treated in as many essays varying somewhat in method, content, and taxonomic technique with each genus.
      Curiously enough, the reader will find that the best-known material is often described with the fewest words, whereas in an imperfectly known genus, in which the species are still so inadequately known as to be difficult to separate without abundant material, it has seemed necessary to describe each species in minute detail to make sure that no possible differential character has been omitted.   It is hoped and believed that the reader will find something of interest in the discussions given under each of the thirty-three genera.
      Each tribe and every genus in the subfamily is keyed out fully so that any of them can be easily identified if material is available.   Fourteen of the thirty-three genera contain only a single species, so that the key to these genera in effect identifies these species.   The remaining nineteen genera, containing from two to many species, are supplied with keys to determine all the species except with respect to the genus Glycosmis, where only part of the species are keyed out, although all are listed.   It is much easier to get an idea of the character of the species by studying the keys than by reading the detailed descriptions, which are to be considered as material for reference.   These technical descriptions may, however, at any time become useful in the event any little-known species becomes important either for ornament or for use in practical citrus culture.
      The commonly cultivated citrus fruits belong to three genera, Citrus, Fortunella, and Poncirus, all closely related and all belonging to the subtribe Citrinae, of the tribe Citreae, of the orange subfamily Aurantioideae, of the plant family Rutaceae.   There are six other subfamilies in the Rutaceae.
      The family Rutaceae, in turn, is obviously related to two other plant families: the Simarubaceae, to which belongs the tree of heaven, Ailanthus altissima (Mill.) Swing.; and the Meliaceae, of which the Chinaberry, Melia azedarach L., is a well-known example.   These plant families are classed by taxonomic botanists along with some eighteen other families in the natural order Geraniales, of which twelve families are included in the suborder Geraniineae, which includes the Rutaceae.
      Although the classification system of Engler is followed in the arrangement of specimens in the larger herbaria in this country, it is not phylogenetic in the modem sense.   It follows a logical sequence of steps in the arrangement of its larger taxa from the simple to the complex on the premise that evolutionary lines progressed from apetaly to polypetaly and gamopetaly, apocarpy to syncarpy, hypogyny to epigyny, and actinomorphy to zygomorphy.   Engler believed that flowers that appear simple have always been simple.   This view is now rejected by most botanists.
      Hutchinson (1926) developed a presumed phylogenetic system of classification based upon many of the principles on evolutionary trends in the angiosperms adopted earlier by Bessey (1915) and subscribed to by Fuller and Tippo (1949) and most students of floral anatomy.   The major point of divergence of the Hutchinson system from earlier systems is the division of the dicotyledonous plants into two subgroups: the Herbaceae composed of families that are predominantly herbaceous, and the Lignosae whose members are predominantly arborescent.   Upon this premise the Rutaceae is placed in the Rutales, an order Hutchinson considers derived from the Celastrales.   This system of classification places the Rutaceae in a position quite remote from the Geraniales through which Engler and Diels consider them derived.
      Hutchinson's insistence that seed plants had a monophyletic origin from hypothetical proangiosperms and that families of predominantly herbaceous plants have evolved from the Ranales and families of predominantly woody plants from the Magnoliales has prevented the general acceptance of his system, but it has greatly stimulated phyletic thinking in recent decades and may ultimately lead to the development of a system based upon the works of Smith (1938) and Eames (1936) and an elaboration of the skeletal system of classification of vascular plants set forth by Tippo (1942).
      The orange subfamily includes several genera with rather numerous species that have a very primitive flower and fruit structure, much like that of other subfamilies of the Rutaceae and even analogous to that of several subfamilies in the plant families Simarubaceae and Miliaceae, which stand very close to the Rutaceae in the natural system.
      First of all, to show at a glance in clear perspective the proper taxonomic placement of the orange subfamily in relation to the six other subfamilies of the plant family Rutaceae to which it belongs, an outline of the Rutaceae is given showing all subfamilies and tribes and subtribes, followed by a concise outline of the taxonomic arrangement of the orange subfamily listing tribes, subtribes, subtribal groups, and genera, with page references to each item.
      It is worth noting that before 1943 no account of the orange subfamily describing all the genera and species of the world had been published for more than a century.   Prior to publication of the original version of this chapter (Webber and Batchelor, 1943), the last complete account of the orange subfamily by a taxonomist was published by Augustin P. de Candolle (1824, vol. 1, pp. 535-40) and contained eleven genera and forty-three species, only a third as many genera and less than a fourth as many species as are discussed here.

PRINCIPAL TAXONOMIC WORKS ON CITRUS AND RELATED GENERA PUBLISHED SINCE 1860
      In December, 1860, Daniel Oliver read before the Linnean Society of London a paper entitled "The Natural Order Aurantiaceae, with a Synopsis of the Indian Species."   It was published the following year (Jour. Linn. Soc. Bot. 5 [2]: 1-44) and covered not only all the genera and species of this order then known in India (nine genera and thirty-one species), but also fourteen additional extra-Indian species belonging to six genera, including one genus not found in India except in cultivation—a grand total of ten genera and forty-five species.   Oliver did not, however, attempt to treat the genus Citrus.  This was the first of the modem high-grade taxonomic works on the orange subfamily.
      In 1875, J. D. Hooker recognized thirteen genera in the orange subfamily with forty-three species (in Citrus he gave only four species) in the account of the family Rutaceae in his Flora of British India (vol. 1).
      In 1888, Dr. E. Bonavia brought forth a voluminous treatise, The Cultivated Oranges and Lemons, etc., of India and Ceylon, with an atlas published in 1890.   This highly imaginative work, by a genuine lover of plants, used for the most part only native names for the varieties discussed, but it aroused interest all over the world.   Bonavia's theories of the morphology and evolution of citrus fruits were highly original and ingenious, but departed widely from the critical standards of professional morphologists and taxonomists.
      In 1910, A. W. Lushington published "The Genus Citrus" (Indian Forester 36:323-53) and gave names to many of the Indian cultivated varieties studied by Bonavia.   He also named some of the citrus fruit trees figured and described by Rumphius in the seventeenth century.
      Henry N. Ridley, for many years director of the gardens and forests (including the Singapore Botanic Gardens) in the former British colony of Straits Settlements, published many articles on rare Malayan plants, and a five-volume Flora of the Malay Peninsula.   Many new species of the orange subfamily were described by Ridley in the first series of papers and coordinated with the previously known ones in his Flora.
      W. G. Craib (1926, vol. 1, pp. 215-39), who wrote the Enumeration of the Flora of Siam, discovered and described several very interesting new species of Citrus relatives.
      A. Guillaumin, taxonomist of the Musée d'Histoire Naturelle at Paris, in 1911 gave a detailed illustrated account of the Rutaceae in Lecomte's Flore genérale de l'Indo-Chine.   In the subfamily Aurantioideae he recognized eleven genera and forty-four species, of which six were in Citrus.   He also published a valuable paper on the species of Atalantia (in the broad sense) of continental Asia and on the Citrus of New Caledonia (now placed in the genus Oxanthera).
      I. H. Burkill, F.L.S., former director of the Botanic Gardens at Singapore, has taken much interest in the species of Citrus and related genera that are native in the Malayan region.   His "Enumeration of the Species of Paramignya, Atalantia, and Citrus Found in Malay," published in 1931, and his very valuable two-volume work, Dictionary of Economic Products of the Malay Peninsula (1935), were used frequently in preparing this chapter.
      E. D. Merrill, former chief of the Bureau of Science at Manila, P.I., and later administrator of botanical collections and director of the Arnold Arboretum, Harvard University, published much valuable information regarding previously published species and described many new species of Citrus relatives from the Philippines, Borneo, Indo-China, and China.
      Tyôzaburô Tanaka, who had specialized, while a student in the Imperial University, on the study of the cultivated varieties of Citrus, accompanied the senior author as guide, interpreter, and assistant through the citrus-growing regions of Japan in 1915 (and again in 1918 and in 1926).   He returned to Washington with Swingle in 1915 and was employed as his assistant at different periods from 1915 to 1930 in the former Bureau of Plant Industry of the U.S. Department of Agriculture.
      In 1928-1930, Tanaka made a trip around the world during which be photographed material of all the genera and species of the orange subfamily to be found in the principal herbaria of Europe and America.   His studies of this material led to the publication of a series of papers on the taxonomy of the subfamily Aurantioideae, entitled "Revisio aurantiacearum."   These papers constitute a very valuable contribution to the taxonomy of this group.   Tanaka's chief works are concerned with satsuma varieties (Tanaka, 1932c) and citrus species problems (Tanaka, 1954, 1959, and 1961).
      In 1896, Dr. Adolph Engler published in the first edition of Die natürlichen Pflanzenfamilien an account of the plant family Rutaceae.   His treatment of the orange subfamily included fourteen genera, and he estimated the total number of species at about seventy-one, of which six were in the genus Citrus.   Thirty-five years later Dr. Engler, who for more than fifty years had specialized on the Rutaceae, wrote, as his last botanical contribution, a revised account of this family in the second edition of Die natürlichen Pflanzenfamilien.   He included twenty-nine genera in the Aurantioideae and estimated the species at about 180, of which eleven were in the genus Citrus.
      Between 1912 and 1926, Walter T. Swingle (the senior author) published seventeen taxonomic papers on tribes, subtribes, genera, and species of the orange subfamily.   Dr. Engler adopted nine of the genera Swingle proposed and used his illustrations for several of them.   It will be evident to every reader of this chapter that Swingle has drawn freely on Engler's great store of knowledge of the orange subfamily.

POSITION OF THE ORANGE SUBFAMILY, AURANTIOIDEAE, IN THE PLANT FAMILY RUTACEAE
      The family Rutaceae belongs to the division Embryophyta Siphonogama, subdivision Angiospermae, class Dicotyledoneae, subclass Archichlamydeae (Choripetalae and Apetalae), order Geraniales, suborder 1, Geraniineae, along with eleven other plant families classed by Engler and Diels (1936, p. xl) in the following order: Oxalidaceae, Geraniaceae, Tropaeolaceae, Linaceae, Erythroxylaceae, Zygophyllaceae, Cneoraceae, Rutaceae, Simarubaceae, Burseraceae, Meliaceae, Akariaceae.   The other suborders of the Geraniales are as follows: 2, Malpighiineae (3 families); 3, Polygalineae (2 families); 4, Dichapetalineae (1 family); 5, Tricoccae (2 families); 6, Callitrichineae (1 family).   The order Geraniales is preceded by the order Rosales (including 17 families, among them Rosaceae and Leguminosae) and followed by the order Sapindales (including 23 families, among them Anacardiaceae and Sapindaceae).
      The relationship of the orange subfamily to the six other subfamilies of the Rutaceae is shown by the general key given by Engler (1931, pp. 105-11), which, translated somewhat freely from the German, reads as follows:
      Subfamily I.   RUTOIDEAE.   Carpels usually 4-5, seldom 1-3 or more, often united only by the common pistil and free below, at maturity more or less separated, opening inward by splitting the follicle (loculicidal), usually with a dehiscent endocarp, very seldom with 4-1 fleshy drupes (Pitaviinae).   Leaves and bark of twigs with schizolysigenous oil glands.   (5 tribes, 17 subtribes, 86 genera.)
      Tribe (i) Xanthoxyleae.   Trees or shrubs, mostly small, greenish or greenish-white (seldom large and clear white) flowers which are always actinomorphous (radial) and often unisexual.   Carpels only seldom with more than 2 ovules, Embryo [sic] mostly with flat cotyledons in endosperm (except Bosistoa and Pagetia).   (5 subtribes, 30 genera.)   Subtribe 1, Evodiinae (includes Xanthoxylon and Fagara) (20 g.), tropics and subtropics of Old and New Worlds; subtr. 2, Lunasiinae ( 1 g.), Monsoon region;1 subtr. 3, Decatropidinae (3 g.), Central America; subtr. 4, Choisyinae (5 g.), Central America, Pacific Islands, and Australia; subtr. 5, Pitaviinae (1 g.), Chile;
      Tribe (ii) Ruteae.   Herbs or perennial herbs, seldom shrubs, with medium-sized, always perfect flowers, which sometimes (Dictamnus) are slightly zygomorphic.   Carpels as a rule with more than 2 ovules (only in Ruta in the subgenus Haplophyllum with 2 ovules and occasionally indehiscent fruitlets).   Seeds with endosperm.   (2 subtribes, 6 genera.)   Subtr. 6, Rutinae (5 g.), subtropical and temperate regions of Old and New Worlds (includes Cneoridium, a California shrub); subtr. 7, Dictamninae (1 g.), temperate zone of Europe and Asia;
      Tribe (iii) Boronieae.   Perennial herbs or shrubs.   Carpels with only 1 or 2 ovules.   Flowers always actinomorphic, mostly perfect.   Embryo usually straight, cylindrical, usually immersed in abundant fleshy endosperm.   (5 subtribes, 19 genera.)   Subtr. 8, Boroniinae (6 g.), Australia and New Caledonia; subtr. 9, Eriostemoninae (9 g.), Australia and New Caledonia; subtr. 10, Correinae (1 g.), Australia; subtr. 11, Nematolepidinae (2 g.), Western Australia; subtr. 12, Diplolaeninae (1 g.), Western Australia;
      Tribe (iv) Diosmeae.   Mostly perennial herbs and shrubs, seldom trees (Calodendrum), always with simple leaves.   Flowers almost always actinomorphic, mostly perfect.   Seeds without endosperm.   Embryo mostly straight with fleshy cotyledons.   (3 subtribes, 12 genera.)   Subtr. 13, Calodendrinae (1 g.) East Africa (Kenya) to Cape Province (South Africa); subtr. 14, Diosminae (9 g.), Cape Province (South Africa); subtr. 15, Empleurinae (2 g.), Cape Province (South Africa);
      Tribe (v) Cusparieae.   Shrubs or tree.   Flowers actinomorphic or the corolla and the androecium zygomorphic.   Seeds with little or no endosperm.   Embryo curved, the plumule lying between the cotyledons.   (2 subtribes, 19 genera.)   Subtr. 16, Pilocarpinae (3 g.), tropical America and subtropical South America; subtr. 17, Cuspariinae (16 g.), tropical America, mostly Brazil and northern South America.
      Subfamily II.   DICTYOLOMATOIDEAE.   Leaves with many-celled but not lysigenous oil glands.   Flowers actinomorphic.   Stamens isomerous and alternate with the petals, with bractlets at the base.   Carpels united only at the base, with several ovules.   Fruits with dehiscent endocarp, 3-4 seeded.   Small trees with doubly pinnate leaves.   (1 tribe, 1 genus.)
      Tribe (vi) Dictyolomateae.   (1 genus.)   Brazil and eastern Peru.
      Subfamily III.   FLINDERSIOIDEAE.   Trees or shrubs.   Carpels 5-3, united, each with 2-8 ovules in 2 rows.   Fruit a loculicidal or septicidal capsule, with persistent endocarp.   Seed winged, without endosperm.   Leaves with lysigenous oil glands.   (1 tribe, 2 genera.)
      Tribe (vii) Flindersieae.   (2 genera.)   Eastern Australia, New Caledonia, East Indian Archipelago, Ceylon, and India.
      Subfamily IV.   SPATHELIOIDEAE.   Carpels 3, completely fused, each with 2 pendent ovules.   Fruit a winged drupe with a 3-loculed hard pit.   With oil-bearing secretory cells in the leaves, bark, and pith; lysigenous oil glands in the leaf margins.   (1 tribe, 1 genus.)
      Tribe (viii) Spathelieae.   (1 genus.)   West Indies.
      Subfamily V.   TODDALIOIDEAE.   Carpels 5-2, incompletely or completely united, or else only 1, each with 1 or 2 ovules.   Fruit formed out of 4-2 drupelets united only at the base, or which some occasionally abort, or else a drupe with a thick or thin mesocarp and a thick or thin endocarp, or a dry, winged, indehiscent fruit.   Seeds with or without endosperm.   (1 tribe, 6 subtribes, 25 genera.)
      Tribe (ix) Toddalieae.   (6 subtribes, 25 genera.)   Subtr. 18, Phellodendrinnae (2 g.), temperate and subtropical eastern Asia and tropical Africa; subtr. 19, Sohnreylinae (1 g.), Amazon Valley, Brazil; subtr. 20, Pteleinae (4 g.), tropical and temperate America; subtr. 21, Oriciinae (2 g.), tropical Africa; subtr. 22, Toddaliinae (13 g.), tropics, Old and New Worlds (include Casimiroa, a Mexican and Central American fruit tree); subtr. 23, Amyridinae (3 g.), northern South America, West Indies, Central America, Texas, Florida, tropical Africa.   [The genus Amyris, having about 30 species native to Florida, Texas, Mexico, Central America, West Indies, and northern South America, is rather closely related to the tribe Clauseneae of the next subfamily, Aurantioideae.]
      Subfamily VI.   AURANTIOIDEAE.   Fruit a berry [or hesperidium] with a leathery rind or hard shell, in tribe Citreae often with pulp formed by juicy emergenzen that arise on the carpellary walls.   Seeds without endosperm, sometimes with 2 or more nucellar [false] embryos.   Leaves and bark [of twigs and young branches] with schizolysigenous oil glands [small or sometimes large trees, rarely shrubs.]; (2 tribes, 33 genera.)   [This subfamily is given as classified by Swingle for this chapter.   Engler made a single tribe with 2 subtribes and with a total of 29 genera.]
      Tribe (x) Clauseneae.   (3 subtribes, 5 genera.)   Subtr. 24, Micromelinae (1 g.), Monsoon region and western Polynesia as far as Tonga, Fiji, and Samoan Islands; subtr. 25, Clauseninae (3 g.), Monsoon region and tropical Africa; subtr. 26, Merrilliinae (1 g.), Malay Peninsula and Sumatra;
      Tribe (xi) Citreae.   (3 subtribes, 28 genera.)   Subtr. 27, Triphasiinae (8 g.), Monsoon region; subtr. 28, Citrinae (13 g.), Monsoon region and tropical Africa; subtr. 29, Balsamocitrinae (7 g.), Monsoon region and tropical Africa.
      Subfamily VII.   RHABDODENDROIDEAE.   Flowers with a bowl-shaped concave receptacle, with obliterated calyx, 5 petals, and very numerous stamens.   Ovary free, ovoid, 1-locular, with 1 basal ovule.   Pistil attached to the side of the ovary, with a long lateral stigma.   Fruits with thin exocarp and thin endocarp.   Leaves simple.   (1 tribe, 1 genus.)
      Tribe (xii) Rhabdodendreae.   (1 genus.)   Amazon Valley.
      Total for the Rutaceae: 7 subfamilies, 12 tribes (containing 29 subtribes), with about 150 genera and 1,600 species.

THE ORANGE SUBFAMILY, AURANTIOIDEAE
      The subfamily Aurantioideae is defined briefly above in the outline of the Rutaceae, where all the subfamilies, tribes, and subtribes of this plant family are given.   All the species of the Aurantioideae are trees or shrubs with persistent (evergreen) leaves except in the three monotypic genera, Poncirus, Aegle, and Feronia, and in three species of Clausena (C. pentaphylla, C. dentata var. dulcis, and C. suffruticosa) and one of Murraya (M. alternans).   The flowers are usually white and very often fragrant.   Many of the genera bear subglobose fruits with a green, yellow, or orange peel dotted with numerous oil glands that often give an agreeable aroma when the fruit is handled.   The fruits of the genus Citrus are among the most beautiful, most fragrant, and most delicious known to man.   This subfamily contains Citrus and thirty-two other genera related more or less to Citrus, classed in two tribes and six subtribes which contain, as treated here, 203 species.   The genus Citrus and a few others closely related to it have fruits unlike any others known to botanists in being filled with curious pulp-vesicles which contain in many species a delicious juicy tissue.   The subtribe Balsamocitrinae, which belongs to the tribe Citreae, has fruits as large as oranges or grapefruits but with a hard woody shell.   These hard-shelled fruits do not contain juicy pulp-vesicles, although some of them are pleasantly aromatic and much liked by both natives and Europeans in India and Indo-China.
      Many of the remote relatives of Citrus belonging to the tribe Clauseneae have extremely small fruits very unlike those of Citrus and usually semidry and entirely inedible.   Nevertheless some of these remote relatives have been found to be graft-compatible with Citrus.
      The native habitat of the subfamily Aurantioideae is limited to the Old World.   Most of the genera are found in the Monsoon region from West Pakistan to north-central China and thence south through the East Indian Archipelago to New Guinea and Bismarck Archipelago, northeastern Australia, New Caledonia, Melanesia, and the western Polynesian islands.   Of the thirty-three genera that constitute the Aurantioideae, no fewer than twenty-nine are native to the Monsoon region and twenty-seven of them are found only there.   Five genera, belonging to two tribes and to three subtribes, are native to tropical Africa and four genera are found only there.   Only one genus, Clausena, is native both to the Monsoon region and to tropical Africa.   Many of the species of Citrus and of the genera closely related to Citrus are now found in cultivation or are grown for ornament in all the tropical and subtropical regions of the world.

Tribes, Subtribes, and Genera of the Subfamily Aurantioideae
      The thirty-three genera of the subfamily Aurantioideae are divided naturally and easily into two tribes: the Clauseneae, with five genera, including the remote relatives of Citrus; and the Citreae, with twenty-eight genera, including Citrus and all its closer relatives.   Each of these two tribes is divided into three subtribes, making six in all.   Keys are given to separate the two tribes and also the subtribes [e.g., Clauseneae; Citreae].
      As these tribes and subtribes are frequently referred to in discussing relationships, English phrases have been applied to them, in addition to their technical Latin names, that will, it is hoped, suggest rather definitely the degree of relationship of each tribe or subtribe to Citrus.
      The natural order of the tribes and subtribes of the orange subfamily is from the most simple to the most specialized, as shown in the list, except that the subtribe Balsamocitrinae, although comprising species somewhat less highly specialized than those in the subtribe Citrinae, has been placed at the end of the series.   This has been done to avoid intercalating the subtribe Balsamocitrinae between the subtribes Triphasiinae and Citrinae, an arrangement which would have obscured the close relationship clearly exhibited by some of the genera of Triphasiinae with some of the more primitive genera of Citrinae.   The list gives the number of groups under each subtribe, as well as the number of genera.   The genera having only one species (monotypic) are marked with an asterisk; the number of species and varieties in other genera is given after the genus name in parentheses.

TRIBE I.   CLAUSENEAE: VERY REMOTE AND REMOTE CITROID FRUIT TREES
      The tribe Clauseneae includes the more primitive genera of the orange subfamily.   None of the species develop spines in the axils of the leaves and the odd-pinnate leaves are at once distinguished from those of the tribe Citreae by having the leaflets attached alternately to the rachis, which does not break up into segments when the leaves fall; rachises are not winged (except in Merrillia caloxylon, Murraya alata, M. alternans, Clausena guillauminii, C. wallichii,and C. luxurians).
      Trifoliolate leaves are rarely found exclusively on any species but occur sporadically merely by reduction of odd-pinnate leaves to trifoliolate leaves, often on the same plant.   Such trifoliolate leaves do not show clearly the precise pairing of the lateral leaflets that is always shown in the tribe Citreae.   The leaves of plants of the tribe Clauseneae do not show the winged petioles often found in members of the tribe Citreae (except very narrow wings on the six species noted above).
      The ovary has two to five locules with only one or two ovules in each locule, except in Merrillia, which has five (rarely six) locules and eight to ten ovules in each locule.   The fruits are usually small semidry or juicy berries, except in Merrillia, which has ovoid fruits the size of a lemon, with a tough leathery peel.   The mature ovaries and young fruits of Micromelum have the locule walls convolute.   This genus has the petals valvate in aestivation, differing thereby from all the other genera of the subfamily.
      The bundle traces which enter the sepals and petals in plants of the tribe Clauseneae were found by Albert H. Tillson (1938, p. 9; Tillson and Bamford, 1938, p. 783) "to arise independently from the axis and to display no trace of fusion with each other."   In the tribe Citreae only three of the twenty-four genera which he studied fail to show such fusions.  
      Only a few genera (five in all) have been established in this tribe, whereas the tribe Citreae has twenty-eight.   Seventy-nine species are listed here in the tribe Clauseneae and 124 in the tribe Citreae, but the largest genus of the tribe Clauseneae, namely, Glycosmis, for which thirty-five species have been named, is still only imperfectly known.
      The tribe Clauseneae falls into three very natural subtribes, as is shown by the key.

Subtribe 1.   Micromelinae: Very Remote Citroid Fruit Trees
      The subtribe Micromelinae contains only one genus, Micromelum, with some nine or ten species, many of them very much alike and rather hard to distinguish.   All the species have odd-pinnate leaves with alternate leaflets borne on a nonarticulated rachis, except M. diversifolium, which has trifoliolate or unifoliolate leaves.   The inflorescences are corymbose-paniculate, often very large and usually flat-topped, often surrounded by leaves much larger than even the largest inflorescences.
      The leaves of Micromelum are very similar to those of Clausena and Murraya, to which genera specimens of Micromelum are sometimes assigned, even by experienced collectors and also by experienced botanists.   The flowers, ovaries, and seeds are, however, very different.   The petals of Micromelum are valvate, not imbricate, and the cotyledons are flat and folded, not thick and plano-convex as in all other genera of the subfamily Aurantioideae.   The ovaries show a curious twisting of the radial follicle walls, also unique in the orange subfamily.
      These remarkable diversities in characters are of much taxonomic importance and are proof of the ancient origin of Micromelum, a view also supported by the very wide distribution of the genus in the Monsoon region, much farther to the eastward in a truly indigenous state than any other genus of the orange subfamily.

I.   Micromelum Blume
      I.   Micromelum Blume, Bijdr. Fl. Nederl. Indie 3:137. 1825; nomen conservandum, Intern. Congr. Bot. 1930. Aulacia Lour. Fl. Cochinch. 1:273. 1790.
      Type species.—Micromelum pubescens Blume.
      Distribution.—Northeastern India; Burma to Australia and New Caledonia; Fiji, Tonga, and Samoan islands in western Polynesia.
      Small spineless trees; leaves odd-pinnate (rarely 3-foliolate or 1-foliolate in M. diversifolium); leaflets thin, alternate on wingless rachis; inflorescences usually large, terminal corymbose panicles often flat-topped; flowers small, 5-merous (except the stamens, which are twice as many as the petals); petals valvate in the bud; ovary with 2-6 (usually 3-5) locules with 2 superimposed ovules in each locule; style rather slender, narrowed at the base and articulated with the ovary, deciduous; radial walls of the locules usually (always?) curved as if twisted during development by a rotation of the outer wall of the ovary with reference to the axis; seeds with thin, folded cotyledons; fruits subglobose or ovoid, dry berries with a gland-dotted peel.
      Oliver (1861, p. 19) gave the following excellent sketch of the morphology of the genus Micromelum: "…The essential characters of the genus [Micromelum] rest especially in the broad, foliaceous, and remarkably contortuplicate cotyledons; perhaps, also, in the singular torsion of the dissepiments of the ovary, which in M. pubescens, is usually apparent immediately after the fall of the floral whorls.…The fruit is but 1- (or 2-) seeded, and the twisted dissepiments with the empty loculaments are soon closely pressed to one side by the young seed.   The style is minutely constricted and articulated at the base ovary.   In M. pubescens the ovary is very usually 5-locular; in Ceylon and Java specimens I have found an exceptional 6th cell; in the Australian plant from Port Essington, etc., it varies, 4, 5 or 3.   M. molle Turcz., I have found to be 4-locular, but probably it also varies.   The peculiar features which I have remarked of the ovary and seed are associated with other common characters.…These are especially the truly valvate or obliquely valvate aestivation of the corolla, and the terminal cymose corymbs of numerous flowers—as in Murraya seldom or never having the central axis of inflorescence elongated as it is in the paniculate Clausenae…"
      The genus Micromelum ranges from the Samoan, Fiji, and Tonga islands to Australia and New Guinea, through the East Indian Archipelago to the Philippines, Indo-China, southern China, Ceylon, northeastern India, and West Pakistan.   One species, M. minutum, ranges farther eastward into Polynesia than any other species of the orange subfamily and also occurs in New Caledonia, Australia, the Philippines, and Indo-China.   The other species have a more restricted range, and one, M. ceylanicum, is restricted to the island of Ceylon and another, M. diversifolium, is found only on Halmahera Island and its near-by companion, Batjan Island, in the Moluccas.

THE SPECIES PROBLEM IN MICROMELUM
      The number of species of Micromelum is not large—nine are here recognized—but they can be described or keyed out only with difficulty because apparently all the species are greatly variable, usually not merely in one, or in a few, but in many characters.   A cursory glance at the literature of the genus shows that taxonomists are by no means agreed on the number of species, or on the characters that define species, or on the areas they occupy.
      Fortunately, new characters are being found that promise to make possible the satisfactory definition and the easy recognition of the species when adequate material is available.   One of the most striking of these new characters is the presence of large or small oil glands that are found, usually one above each locule, at the tip of the ovary.   Very large oil glands expand greatly the ovary tip of M. falcatum; large oil glands occur in M. compressum but do not noticeably expand the ovary tip; whereas only very small oil glands are found similarly placed in M. pubescens.
      These new characters, and others already used by taxonomists, may be seen in figure 3-1, which shows microphotographs of longitudinal or cross sections of ovaries of eight of the nine species of Micromelum.   A key to the species of Micromelum is presented.

      1.   Micromelum pubescens Blume, Bijdr. Fl. Nederl. Indie 3:138. 1825. Illus. Engler, Die Nat. Pflanzenfam. 3(4):186, fig. 107,A-K. 1896; fig. 3-1,D this work.
      Type.—Java (Blume).   Rijks Herb., Leiden.
      Distribution.—Java, Sumatra, Andaman Islands, southern Burma, Malaya, Palawan Island, P.I.
      A brief description by Koorders (1912, vol. 2, p. 424), based on material from Java, the type locality, reads, translated, as follows: "Tree, up to 6 m high, or erect shrub.   Leaflets 7-11, alternate, soft-pubescent below, obliquely ovoid to broadly lanceolate, 3-17 X 2-7 cm.   Fruit 6-8 mm long, orange-yellow."   Ridley (1922, vol. 1, p. 352) describes this species as follows: "Large shrub or treelet.   Branchlets and leaf-rachis puberulous but eventually glabrescent.   Leaves 6 to 18 in. long; leaflets 9 to 15, alternate, lanceolate or ovate-acuminate, edges undulate, base slightly narrowed oblique; nerves 9 to 12 pairs, 1.5 to 3.5 in. [3.8-8.8 cm] long, 0.5 to 2 in. [1.3-5 cm] wide; petioles up to 3 in. [7.6 cm] long.   Cymes 6 to 8 in. [15-20 cm] across, peduncled, many-flowered [pubescent].   Flowers 0.25 to 0.5 in. [6-13 mm] long, greenish-white.   Calyx 5-toothed.   Ovary pubescent.   Berry ovoid to oblong, glabrescent, orange, 0.3 to 0.4 in. [7.5-10 mm] long."
      The chief diagnostic characters of this species are given in the key.   The pistil (fig. 3-1,D) is of medium size with a rather long style in proportion to the ovary.   The oil glands at the tip of the ovary, one over each locule, are unusually small and sometimes difficult to detect.   Tanaka (1931b, pp. 2, 3) contrasted this species with M. minutum, as follows: M. pubescens has (1) much larger floral organs than M. minutum; (2) stouter peduncles; (3) thicker leaflets; (4) ovary longer stalked; (5) flower buds much broader and silvery pubescent, instead of golden-yellow pubescent as in M. minutum.   From M. compressum of the Philippine Islands, Tanaka reported that it differs in four characters (see next species below).
      2.   Micromelum compressum (Blanco) Merr. Sp. Blancoanae 200. 1918. Bergera compressa Blanco, Fl. Filip. 361. 1837; Micromelum tephrocarpum Turcz. Bul. Soc. Nat. Mosc. 31:379. 1858. Illus. fig. 3-1,G.
      Type.—Philippines, Luzon Island (Blanco).   Original type lost.   Substitute type: (Merrill, Species Blancoanae, No. 884).
      Distribution.—Philippines: Luzon to Mindanao (and Palawan?) Islands.
      The best description yet published of this species is apparently Turczaninow’s original diagnosis of his M. tephrocarpum, which, translated, reads as follows: "Unarmed, very glabrous; leaves odd-pinnate, leaflets alternate, lanceolate, obtusely acuminate [at the tip], attenuate, inequilateral at the based, entire or slightly serrulate, pellucid-punctate; panicles terminal, much shorter than the leaves; fruit ovoid-oblong, acute, becoming cinereous-blue, with fewer than 3 seeds; calyx truncate, somewhat recurved, half as long as the petals; alternate filaments shorter, anthers adnate, subglobose; ovary seated upon a very short disk, style rather thick and terminated with a capitate stigma; fruit (‘berry’) with the locule walls contorted and twisted (‘contortuplicate’), 5-loculed, with only 3 seeds in the immature fruit [which is] obovate-oblong, and flattened.   From M. glabrescens Benth. it seems to differ chiefly in having fruits with attenuate tips."
      A flowering specimen collected by M. Ramos at Bohol, Luzon, P.I. (Bur. Sci., No. 42570), in the herbarium of the Arnold Arboretum, shows very small pistils only 2-3 mm long, with the ovary 1-1.2 mm long, and the style about equaling the ovary in length, faintly furrowed longitudinally, and abruptly contracted at the base where it joins the ovary, about 0.4 mm diam., stigma hemispherical, 0.8 mm diam.
      Tanaka, who has given much attention to distinguishing the species of Micromelum and has examined carefully the extant type specimens, has noted certain contrasts he observed between the species.   He stated (1931b, pp. 2, 3) that M. pubescens, common in Java, Sumatra, and north to southern Burma, differs from M. compressum: (1) in having more slender flower buds covered with very short silvery pubescence; (2) in its hairy ovary; (3) in its glabrous pedicels; (4) in it's smaller, narrower leaflets with many more lateral veins.
      2a.   Micromelum compressum var. inodorum (Blume) Tan. Trans. Nat. Hist. Soc. Formosa 22:418. 1932. Bergera inodora Blanco, Fl. Filip. 360. 1837; Micromelum molle Turcz. Bul. Soc. Nat. Mosc. 31:380. 1858.
      Type.—Philippines (Cuming, No. 1056).   Not located.   Substitute type: (Merrill, Species Blancoanae, No. 719).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines; doubtfully reported from Celebes by Tanaka (1932e, p. 419).
      Turczaninow's original diagnosis of M. molle (an undoubted synonym of Blanco's Bergera inodora) reads, translated, as follows: "Unarmed; twigs, petioles, peduncles, panicles, calyxes, and petals [covered with] short, soft pubescence; leaves odd-pinnate, 7-13-foliolate; leaflets alternate, ovate-lanceolate, acuminate [at the tip], attenuate, inequilateral at the base, densely and softly pubescent on both sides on the midrib and veins, sparingly pubescent between the veins; panicles composed of both axillary and terminal [portions] shorter than the leaves; calyx truncate, slightly 5-toothed, much shorter than the petals; fruits acute, 1-3 seeded, blackish-blue, glabrous.   From M. pubescens Bl., of which a specimen was sent by Blume himself to the [Museum] Schultesium, it differs in being more pubescent, in having larger and softer [pubescent], acutely acuminate leaflets, and in having fruits acute, not very obtuse."
      Tanaka stated (1932e, p. 418): "All botanical characters agree with the type species except the pubescence of the leaf and the shoot."   He also noted the short, thick flower buds in contrast with the small, slender flower buds of M. pubescens and M. minutum.
      3.   Micromelum scandens Rechinger, Denkschr. Akad. Wiss. Math.-Naturw., Wien 89:564. 1914. Micromelum minutum var. intermedium Tan. Med. Rijks Herb. Leiden 69:2. 1931. Illus. Tillson & Bamford, Amer. Jour. Bot. 25:782, figs. 1-11. 1938; fig. 3-1,F this work.
      Type.—Bismarck Archipelago, New Britain Island, Gazelle Peninsula, Baining Mountains (Rechinger, No. 3675).   Herb. Naturhist. Mus., Vienna.
      Distribution.—Bismarck Archipelago and Melanesia.
      The original description reads, translated, as follows: "A climbing [or clambering] shrub; leaves pinnate, leaflets large, 15-20 X 6-7 cm, with entire margins, brownish-black when dried; fruits dry berries, disposed 30-40 in cymes, with spirally twisted follicle walls, 8-12 mm long, glabrous, with many small pits, black when ripe, with a short fragment of the style at the tip, and the flat cupulate calyx, with an almost entire margin, persistent at the base."
      The type specimen of M. scandens shows one leaf 22 X 8 cm, larger than any yet measured of M. compressum, with 10 pairs of lateral veins; the leaflets of M. scandens are acuminate at the tip and have 1/5-1/4 more lateral veins than those of M. compressum, and they arise at a somewhat smaller angle with the midrib.   The persistent calyx is shallow, cupulate with indistinct lobes.   The pedicels of the fruits are 7-11 mm long, with 2 small bracts near the base.
      A specimen in fruit collected on Bougainville Island on May 25, 1930, by Kajewski (No. 1785) has a terminal fruit cluster with very stout lower branches 2.5-3.5 mm in diameter.   The notes state that the fruits are "dark red when ripe, oblong coming to a blunt point, length 1.4 cm, diameter 8 mm."   Very young fruits are 3.5-4 X 1.2-1.5 mm, cylindrical, rounded at both ends and show a few scattered, slender, white hairs.   The nearly ripe fruits are glabrous, rough with oil glands, and show at the tip a very short truncate style base 0.7-0.8 mm in diameter.
      One of the specimens collected by Kajewski in 1930 (No. 1654, from the Pupei Gold Fields, Bougainville Island) was from a tree growing up to 18 m high.   The specimen has very large, lax, more or less leafy inflorescences with long nodes below, much like those shown by the type specimen of the species.   The flower buds are oval or broadly and shortly elliptical, 5-6 X 2-3 mm, decidedly larger than those of good material of M. minutum from Tonga, the type locality, and from Fiji, New Caledonia, and Australia.   This specimen (Kajewski's No. 1654) had a flower just ready to open; its pistil is 4.2 mm long when dry, ovary almost glabrous.   The calyx is large and expanded until it is shaped like a shallow saucer, nearly pentagonal in outline.
      Swingle considered M. minutum var. intermedium Tan. from New Ireland, Bismarck Archipelago (to judge from a photograph of the types specimen sent him by Tanaka), to be a fairly typical form of M. scandens, which was based on a specimen collected by Rechinger in the adjoining island of New Britain, the eastern-most point of which is distant only some 35 kilometers from the west shore of New Ireland.   Bougainville Island lies only some 200 kilometers southeast of the southern end of New Ireland.   Tanaka had not seen the excellent material of Micromelum from the Solomon Islands in the Arnold Arboretum herbarium when he allocated just such a form to M. minutum as a variety.   Tanaka, who examined the type specimen of M. scandens at Vienna, stated (1931b, p. 3) that it is very different from M. minutum and most like M. compressum.   This species appears to be a good one, but needs further study.
      4.   Micromelum falcatum (Lour.) Tan. Bul. Mus. Hist. Nat. Paris, 2 sér. 2:157. 1930. Aulacia falcate Lour. Fl. Cochinch. 1:273. 1790; Cookia falcate DC. 1824; Micromelum octandrum Turcz. 1863. Illus. fig. 3-1,C.
      Type.—French Indo-China, Annam (?) (Loureiro).   Herb. Brit. Mus.
      Distribution.—North Vietnam, South Vietnam, Cambodia, Laos, Burma, Andaman Islands, Thailand, southern China.
      Guillaumin (1911, p. 649) described this species (under the name M. pubescens Bl.) as follows: "Shrubs or small tree, twigs woolly-pubescent, then glabrescent.   Leaves yellowish-green in color, 15-31 cm long, odd-pinnate; leaflets 7-9, lanceolate, inequilateral at base, long-acuminate-pointed, margins slightly crenulate, glabrous, except for the midrib above and the larger veins below [which are pubescent], lateral veins 6-8 on each side, prominent below, veinlets not visible; petiolules and petioles cylindric, pubescent, 4-10 mm long.   Inflorescences pubescent, shorter than the leaves; pedicels woolly, 5 mm long, having two small opposite hairy bracts toward the base; flowers 5 mm long, greenish-white; calyx woolly with 5 very short teeth; petals 5, long-elliptic, glabrous or glabrescent, 5 mm long; stamens 10, free, glabrous, the largest as long as the petals; filaments subulate; disk short, glabrous, grooved; ovary ovoid, velvety-hairy; style caducous, elongated, glabrous; stigma capitate; locules 5, with 2 ovules in each; fruit ellipsoid, fleshy, with numerous oil glands, glabrous; segments often 2-3, each with a single seed.   Flowers Dec.-Jan.   Very common."
      This species differs from all others in the genus in having very narrow, often falcate, long-acuminate leaflets, very unequal-sided at the base, usually 4 to 11 by 1.5 to 3 cm, including petiolule 3 to 5 mm long.   It also is the only Micromelum having the top of the ovary much broadened by large oil glands, found one over each locule.   This plant was the first species of the genus to be described (as Aulaciafalcate in 1790), and Loureiro's genus Aulacia antedates Blume's Micromelum by thirty-five years.   However, the genus Aulacia was not understood by botanists in spite of the fact that a specimen of it, one of the few remaining types of Loureiro's species, is preserved in the British Museum at London.   In 1919, Merrill (in a MS study on Loureiro's types) concluded that Aulacia was a synonym of Micromelum, but the Fifth International Botanical Congress, meeting at Cambridge in 1930, conserved the name Micromelum and rejected the older generic name Aulacia (Merrill, 1935, p. 221).
      5.   Micromelum ceylanicum Wight, Ill. Ind. Bot. 1:109. 1840. Illus. fig. 3-1,A.
      Type.—Ceylon (Wight, No. 1836).   Herb. Kew.
      Distribution.—Found only in Ceylon.
      Trimen (1893, vol. 1, p. 218) described this species (under the name M. pubescens) as follows: "A small tree, shoots very finely and densely pubescent; leaves imparipinnate, 8-10 in. [20-25 cm long], rachis pubescent; leaflets 9-15, alternate or subopposite, shortly stalked, 1 1/2-3 in [3.8-7.5 cm long], very oblique base, ovate-lanceolate attenuate, obtuse, irregularly finely crenate and wavy, with very numerous, rather conspicuous glands, glabrous above, nearly so or slightly pubescent beneath; flowers 1/4 - 1/2 in. [6-13 mm diam.], very shortly stalked, arranged in terminal and axillary, spreading, pubescent, dichotomous, carymbose cymes, bracts small, opposite at the bifurcations; calyx lobes broad, triangular, pubescent; petals oblong-linear, spreading, pubescent; ovary oblong, very pubescent; style rather longer, very thick, stigma capitate; fruits 3/8 in. [9 mm] long, oblong-ovoid, pointed, rough with glands, glabrous, yellow."
      This species, as can be seen in figure 3-1,A has calyx lobes extending to the middle of the ovary in a nearly full-grown flower bud.   The other species of Micromelum would not show the sepals in a section made at this level.
      In Ceylon this species is said to be rather common "in the low country, especially in the dry region."   The name of the species in Singhalese is walkarapincha.   Murraya koenigii is called karapincha in Singhalese and Clausena indica is called mogen-karapincha.   Both plants look so much like M. ceylanicum as to be easily confounded with it, according to Trimen (1893, p, 219).
      6.   Micromelum integerrimum (Buch.-Ham.) Roem. Syn. Hesper. 1:47. 1846. Bergera integerrima Buch.-Ham. ex Coleb. Trans. Linn. Soc. London 15:367. 1827; B. integerrima Roxb. ex DC. nomen semi-nudem. 1824. Illus. Beddome, Fl. Sylv. Anal. Gen. xliii-xliv, pl. 7 (flowers only). 1871; fig. 3-1,E this work
      Type.—British India, Cult. Hort. Bot., Calcutta (Roxburgh).   Herb. Brit. Mus.
      Distribution.—Northeastern India: West Bengal, Sikkim, Assam; Nepal; Burma; East Pakistan; Andaman Islands.
      Kurz (1877, vol. 1, pp. 186, 187) described this species, under the name M. pubescens, as follows: "An evergreen [shade-loving] tree, 25-30 ft. [7.5-9.14 m] tall, with a clean trunk 10-12 ft. [3-3.66 m], measuring 2-3 ft. [60-90 cm] in girth, all younger parts more or less densely puberulous; bark about a line thick, whitish, somewhat roughish; cut pale-coloured; leaves unequally pinnate, puberulous or glabrous, 1 1/2 ft. [46 cm] long, the rachis usually more or less puberulous; leaflets in 3-7 pairs with an odd one, alternating or nearly so, oblong-lanceolate to broadly lanceolate, oblique at base, shortly petioluted, 5-7 in. [12-18 cm] long, acuminate, usually somewhat waved, entire or nearly so; flowers middling-sized, whitish, on short pedicels, forming terminal divaricate or crowded pale tawny appressed-pubescent corymb-like panicles; calyx more or less obsoletely 5-teethed, pubescent; petals about 2 1/2 lin. [5 mm] long, pubescent glabrescent; berries seated on short, stalk-like torus, ovoid-oblong, also when young , glabrous, gland-dotted, yellow, turning dull orange-red, terminated by the style scar."
      H. H. Haines, the former Conservator of Forests of Bihar and Orissa (northeastern India) in his discussion of this species (1921, p. 164), under the name M. pubescens stated: "A small tree…leaves 8-18" [20-46 cm] long, with 5-11 very large leaflets and large corymbs of white flowers 0.5" [13 mm] diam., which are succeeded by foetid, ovoid, yellow or scarlet berries 0.5" [13 mm] long.…Leaflets ovate to lanceolate or elliptical-oblong, attaining 8" by 3.5" [20 X 8.8 cm], lowest sometimes only 1.5" [4 cm].…Corymbs pubescent or tomentose, often 1 ft. [30.5 cm] across…very pretty when in flower or fruit."
      The flowers and especially the pistils of this species are decidedly larger than those of all the other known species of Micromelum.   The fruits are attenuate at the base, instead of broadly rounded as in M. hirsutum and several other species.   A specimen from the South Andaman Islands, secured by King's collector in North Corbyu's Cove, now in the herbarium of the Arnold Arboretum (Tanaka's Ident. No. A 64), shows (in the dry state) pistils 7 to 8 mm long; ovary 2 mm long; style 3 mm long with many fine longitudinal furrows; stigma 1 mm high; the attenuate base of the ovary (disk?) 1.5 mm high.   Because of a long series of errors in citation, the botanical name of this species has never yet been correctly accredited either as Bergera integerrima or as Micromelum integerrimum.   Bergera integerrima was first published in 1827 by Colebrooke, who accredited it to Buchanan-Hamilton.   It was transferred to Micromelum as a species distinct from M. pubescens by Roemer in 1846, a transfer incorrectly accredited to Wight and Arnott, who merely excluded it from Bergera but did not recognize it as a valid species of Micromelum.
      7.   Micromelum hirsutum Oliv. Jour. Linn. Soc. Bot. 5(2):40. 1861. Illus. Guillaumin, in Leconte, Fl. Gén. Indo-Chine 1:646, fig. 68(8-10). 1911; fig. 3-1,H this work.
      Type.—Malay Peninsula, Penang (Wallich, No. 8516).   Herb. Kew.
      Distribution.—Burma, Andaman Islands, Thailand, North Vietnam, South Vietnam, Cambodia, Lao.
      Oliver's original diagnosis of this species reads in part, translated, as follows: "Leaflets 15-23, very asymmetrical at the base; calyx small, with 5 triangular lobes; petals hirsute without; ovary 5-locular; stigma almost as long as the style; leaves 10-16 in. [25-41 cm long]; leaflets trapezoid-lanceolate or ovate-lanceolate, often not a little acuminate, somewhat obtuse at apex, margins slightly denticulate, sparsely pilose or glabrescent below, pilose or hirsute particularly near the rib and veins above.   Inflorescences divaricate cymes, corymbose or subpaniculate, hirsute.   Pedicels very short.   Petals narrow or linear-oblong.   Fruit (young) ovoid-oblong, hispid, shortly and narrowly stipitate.   Chiefly differing from M. pubescens in the long leaves with very numerous leaflets, small calyx, and petals strongly hairy externally."
      Kurz (1877, vol. 1, p. 187) added many details to Oliver’s technical diagnosis, as follows: "A low, meagre, often simple-stemmed shrub, about 2-3 ft. [60-90 cm] high, rarely higher, all parts more or less shortly tawny hirsute or puberulous (rarely almost glabrous); leaves unpaired-pinnate, densely puberulous; leaflets in 5-10 pairs with an odd one, lanceolate or oblong-lanceolate, shortly but slenderly-petioluled, very oblique at base, acuminate, obsoletely repand-serrate, usually 3-5 in. [7.6-12.7 cm] long; flowers small, whitish, on short hirsute pedicels, forming more or less shortly tawny hirsute corymb-like panicles at the ends of the branchlets and in the axils of the upper leaves; calyx shortly tawny hirsute, deeply 5-lobed; petals hardly 2 lin. [4 mm] long; ovary densely tawny hirsute; berries sessile or nearly so, oblong or obovoid, especially while young more or less puberulous, gland-dotted, crowned by the style scar."
      "Hab.—Very frequent in the drier upper-mixed forest, and the dry and low forests, all over Burma from Ava and Pegu down to Tenasserim."
      King (1894, p. 219) added other details: "Leaves 6-12 in. [15-30 cm], rarely 15 in. [38 cm] long; leaflets 9-25, 1.5-3.5 in. [4-9 cm] long, 0.8-1.5 in. [2-4 cm] wide; main nerves 5-10 pairs.   Cymes terminal, very tomentose, often lax, 6-8 in. [15-20 cm] in diameter but sometimes condensed and only 2 in. [5 cm] in diameter."
      This species differs chiefly from the other species commonly occurring in Burma (M. integerrimum, but called M. pubescens by King and Kurz) in being a small shrub, never a tree, and in being more pubescent, especially the fruit, and in having smaller flowers.   Kurz noted that the fruits of M. hirsutum are puberulous and almost sessile, whereas those of M integerrimum (his M. pubescens) are short-stalked, glabrous berries.   Guillaumin (1911, pp. 649-50) noted that in Indo-China this species has "ellipsoid, fleshy, reddish, glabrescent or pubescent fruits."
      8.   Micromelum minutum (Forst.) Wt. & Arn. Prodr. Fl. Pen. Ind. Or. 1:94. 1834. Limonia minuta Forst. f. Prodr. 33. 1786; Micromelum glabrescens Benth. Hook. Jour. Bot. 2:212. 1843; M. pubescens var. glabrescens (Benth.) Oliv. 1861; Glycosmis subvelutina F. Muel. 1858. Illus. Banks & Solander, Bot. Capt. Cook's Voy. 1:14, pl. 36. 1900; partly reproduced in fig. 3-2 this work; also fig. 3-1,B.
      Type.—Friendly Islands (Forster).   Herb. Brit. Mus.
      Distribution.—Samoa, Tonga, Fiji, New Caledonia, northeastern Australia, Melanesia, Bismarck Archipelago, New Guinea, and north to the Philippines, Sumbawa, and Borneo; also Annam (fide Tanaka).
      This species was described, under the name M. glabrescens, in 1843 by Bentham from material collected in the Friendly Islands (the type locality) by Barclay, as follows: "Young leaves and inflorescences finely tomentose, leaflets 10-12, obliquely ovate-acuminate, minutely crenulate, adult leaves glabrous; calyx very short, 5-toothed; fruit oblong, very obtuse.   This is evidently very near M. pubescens Blume; does not quite agree with his [Blume’s] very short description.   The leaflets are quite smooth, except in a very young state.   The inflorescence is a dichotomous many-flowered terminal cyme.   The flowers appear very small, but are as yet unexpanded in the specimen before me.   The fruit is about 4 lin. [8 mm] long.   The foliaceous cotyledons are very broad, deeply emarginate and twisted, with rather a long straight radicle."   An examination of material from the type locality (Tonga Islands), from Samoa, and from the Solomon Islands shows the following characters in the dry state.   The specimen from the Tonga Islands (coll. by H. E. Parks, No. 16281, June-July, 1926, at Eua Island, now in U.S. Natl. Herb., No. 1528543) has very short pistils, only 2 mm long; the ovary is 1.3 mm long, sparingly covered with short gray hairs; the style and stigma measure 0.8 mm, the style is glabrous above, somewhat hairy below.   The specimen from the Solomon Islands (coll. by Kajewski, No. 2371, December 15, 1930, now in Herb. Arnold Arbor.) shows pistils 3.5 mm long, with the ovary 1.5 mm long, style and stigma 1.5 mm long.   The ovary is fusiform, narrowed at both ends, where it joins the disk below and the style above, and has scattered yellowish hairs that soon fall off.   The style is sparingly hairy, not evidently furrowed longitudinally, sharply constricted at the base and expanded into the flattened, discoid stigma, about 0.6 mm wide and 0.3 mm tall.   The specimens from Tutuila, Samoa (coll. by W. A. Setchell, No. 257, now in Herb. Arnold Arbor.), show pistils 3 mm long; ovary 1.25 mm long, covered with rather sparse, short, gray or yellowish hairs; style and stigma 1.2 mm long; style with similar hairs at the base concealing the constriction at the base, if present; stigma subglobose, 0.7 mm. diam.
      This species has the smallest flowers and, in particular, the shortest pistils of any species of the genus.   It ranges farther east in Polynesia than any other species belonging to the orange subfamily.   It alone reaches the Tonga and Samoan Islands.   It also ranges far to the north, from northeastern Australia through New Guinea, Celebes, the Philippines, and reaches North Borneo and Sumbawa, but not Java and Sumatra according to Tanaka (1931b, p. 2), who stated, moreover, that Annam (Vietnam) is its westernmost limit.
      As is to be expected from its very wide distribution, this species varies considerably even in such important organs as the pistil.   Only two varieties have been published as yet, but others could probably be made with equal justification.
      8a.   Micromelum minutum var. tomentosum Tan. Trans. Nat. Hist. Soc. Formosa 22:419. 1932.
      Type.—Timor (Forbes, No. 3753).   Herb. Brit. Mus.
      Distribution.—Timor; Philippines: Mindoro and Luzon islands.
      Tanaka's original description of this variety, translated, reads as follows: "Similar to Micromelum minutum in the appearance and size of its flowers and fruits but all parts completely covered with soft, golden yellow tomentum."
      8b.   Micromelum minutum var. curranii (Elm.) Tan. Trans. Nat. Hist. Soc. Formosa 22:419. 1932. Micromelum curranii Elmer, Leafl. Phil. Bot. 2:480. 1908; M. caudatum Merr. Phil. Jour. Sci. 27:26. 1925.
      Type.—Philippines, Luzon Island, Baguio (Elmer, No. 8530).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: Bontoc, Benguet, and Union provinces of Luzon Island, at 1,200-1,450 m altitude (fide Merrill, 1923, vol. 2, p. 335).
      Elmer's original description of his M. curranii (now reduced to a variety of M. minutum) reads as follows: "Shrub, with rather numerous branchlets; wood hard, covered with grayish-white mottled bark.   Leaves crowded on the young twigs, numerous, alternate, ascending, the peduncle [petiole] short yellowish pubescent when young but soon becoming glabrous; leaflets alternatingly scattered along the rachis, glabrous when mature, unequal in size, ovate to broadly lanceolate, submembranous, entire or obscurely apiculate, apex gradually acuminate, base obliquely rounded to subacute, drying brown, paler beneath, the larger blades 7 cm long by 2 cm wide; petioles [petiolules] 3 to 5 cm long, when young short yellowish pubescent but becoming glabrous; nerves 3 to 5 pairs, ascending, more prominent beneath, reticulations not visible.   Corymbose panicle terminal, 6 cm wide across the top, about as long; peduncles and pedicels strict, suberect, pubescent when in flower, becoming glabrous when in fruit, the latter about 3 to 5 mm long; calyx truncate or obscurely 5-apiculate, rim-like or saucer-shaped, pubescent when young, persistent and glabrous in fruiting state; petals 5, ligulate, rather thick, glabrous or sparsely pubescent on the outer side, 4 mm long, at least 1 mm wide, apex acute, deciduous; stamens in one series, about 9, inserted upon a prominent disk at the base of the ovary, easily detached; filaments subglabrous, fleshy and more or less flattened, pointed at the apex, subequal in length, the longer ones 4 mm in length; anthers basifixed, broadly cordate or subelliptic, nearly as wide; style thick, 1 mm long, bearing a terminal subcapitate stigma, easily detached from the ovary; ovary glabrous, oblong in outline or clavate, blunt at apex, base narrowed, immature fruits 1 cm long, thicker above the middle, prominently punctate [of] lemon color, its [seeds have the] cotyledons folded, dark green.…It is a form intermediate between M. tephrocarpum Turcz. [M. compressum] and M. pubescens Blm., but can be distinguished from either by its numerous, much smaller and glabrous mature leaves."
      9.   Micromelum diversifolium Miq. Ann. Mus. Bot. Lugd.-Bat. 1:211. 1863.
      Type.—East Indian Archipelago, Moluccas, Halmahera Island (Teijsmann).   Rijks Herb., Leiden.
      Distribution.—Known only from the type locality.
      Miquel's original diagnosis reads, translated, as follows: "Twigs, petioles, and underside of the leaves, as well as the inflorescence and the flowers, somewhat scurfy, slightly tomentose-pubescent; leaves either with one pair of leaflets and an odd one, thus trifoliolate, or unifoliolate leaflets, elliptic or oblong-elliptic, obovate, acute or somewhat obtuse [at the tip], rounded or slightly emarginate at the base, 7-8 spreading veins on both sides [of the mid-rib]; flowers in short-terminal corymbs.   Leaflets 2-4.5 in. [5-11.5 cm long]; calyx with 5 short teeth; petals 5, valvate, pubescent without; stamens 10, the alternate ones shorter, anthers ovate-cordate; style distinct, stigma convex; berry ellipsoid, somewhat stipitate-constricted at the base with 5 locules, with spirally twisted [radial] walls."
      Tanaka (1931b, p. 1), after having examined type specimens of this species in Holland, both in the Rijks Herbarium at Leiden and in the herbarium at the University of Utrecht, stated: "This extremely interesting species is not known outside of the Moluccas.   The most remarkable features are the extreme reduction of the number of the leaflets, the exceedingly villose floral organs and the completely tomentose berry."
      The origin of a species of Micromelum found almost in the middle of the area of distribution of the genus, differing radically from all the others in the great reduction of the number of leaflets, is of much interest.   The unifoliolate form of the species (and of the variety mentioned below), has reached the extreme reduction of the originally pinnate leaves to a single leaflet that has occurred in species belonging to both the tribes of the orange subfamily.   Both tribes, however, still have species with odd-pinnate leaves.
      9a.   Micromelum diversifolium var. cuneata Miq. Ann. Mus. Bot. Lugd.-Bat. 1:211. 1863.
      Type.—The East Indies, Moluccas, Batjan Island (Teijsmann).   Herb. Univ. Utrecht.
      Distribution.—Moluccas: Batjan, Obi, and Halmahera islands.
      Miquel’s original diagnosis reads, translated, as follows: "With larger glabrous leaflets [than those of the species], the terminal one acute or cuneate at the base; the lateral leaflets somewhat acute at the base with 8-10 ribs on both sides."
      Tanaka (1931b, p. 1), after examining the type specimen in Holland, wrote as follows: The "variety cuneata has a deep cupulate calyx and the reduction of fuzziness is rather noticeable; it has occasional unifoliolate leaves.   Additional material from Obi and Halmaheira [Halmahera] shows its fairly uniform nature."

Subtribe 2.   Clauseninae: Remote Citroid Fruit Trees
      The subtribe Clauseninae contains three genera, Glycosmis, Clausena,and Murraya, having very simple, more or less primitive flower and fruit structures.   The flowers of Glycosmis, Clausena,and Murraya are usually borne in dense, often large, panicled clusters at the tips of the branches.   None of the genera of this subtribe have spines.   The leaflets are alternate on the rachis, which is not articulated and does not break into segments when the leaves fall.   The flowers are 3- to 5-merous except that the stamens (always free) are twice as many as the petals.   The ovaries have two to five locules and each locule contains one to two ovules.   The fruits are small berries, either soft and juicy with few seeds immersed in mucilaginous pulp, or semidry with a clearly distinguished, gland-dotted peel.   The seeds are glabrous.
      The three genera, Glycosmis, Clausena, and Murraya, included in the subtribe range from India, Burma, and Indo-China to Borneo, the Philippines, New Guinea, and Australia.   Clausena also has a wide distribution in tropical and subtropical Africa and in the bush forests, at high altitudes, in the mountains of eastern and central Africa.   Three species and two varieties of Clausena are known from Africa.   Plotted on a globe, the distribution of the Remote Citroid Fruit Trees is very similar to that of the Near-Citrus Fruit Trees.
      Glycosmis and Clausena show very simple flower and fruit structures much like those found in some genera of other subfamilies of the family Rutaceae (such as Amyris in the subfamily Toddalioideae).   Close analogies to the flower and fruit structures shown by Glycosmis, Clausena, and also Murraya can be found outside of the Rutaceae in the Simarubaceae and Meliaceae, two plant families that stand very close to the Rutaceae in the natural taxonomic system.   The absence of oil glands in the leaves and fruits is almost the only character that separates some of the plants now classed in the Meliaceae from plants in the Rutaceae.   [A key to the genera of the subtribe Clauseninae is presented.]
      The three genera of Remote Citroid Fruit Trees are clearly though distantly related to Citrus.   Citrus has been grafted successfully on Clausena and Murraya and vice versa.   Such grafts are often short-lived, but sometimes live many years.   Grafts between Citrus and Glycosmis have not as yet been successful, but so far only a single species of Glycosmis has been tested, out of the large number known to exist.   H. J. Toxopeus (1936, p. 6) made hybrids between Citrus and Murraya, but they were weak and remained stunted.

II.   Glycosmis Corrêa
      II.   Glycosmis Corrêa, Ann. Mus. Hist. Nat. Paris 6:384. 1805. Sclerostylis Blume, Bijdr. Fl. Nederl. Indie 1:133. 1825. Chionotria Jack, Malay. Misc. 2(7):53. 1822; Dioxippe Roem. Syn. Hesper 1:33, 45. 1846; Myxospermum Roem. Syn. Hesper. 1:31, 40. 1846.
      Type species.Limonia pentaphylla Retz. (Obs. Bot. 5:24 [1789]) = Glycosmis pentaphylla (Retz.) Corr.
      Distribution.—Southeastern Asia, East Indian Archipelago, Philippines, New Guinea, northeastern Australia.
      Unarmed small trees or shrubs; new growth covered with dense, usually ferruginous pubescence; leaves 1-foliolate, 3-foliolate, or odd-pinnate; leaflets with short petiolules, alternate, long-lanceolate, more or less coriaceous; inflorescences compound, often densely racemose; flowers small, 5-merous; sepals united to the middle, with broad imbricate lobes; petals 5, white, long-elliptical or oval, imbricate; stamens 10, filaments subulate, broadened below, anthers small, often with oil glands on the back and tip; disk annular or cylindric; ovary 2-5-locular, with 1 ovule in each locule; style very short and thick, persistent, stigma simple or disk shaped; fruit a juicy berry or dry; seeds oval, thick, with a membranaceous testa; embryo with fleshy plano-convex cotyledons and a very short plumule.
      This genus includes a large number of very closely related thornless shrubs or small trees ranging from southeastern Asia and the East Indies to northern Australia.   It will probably require careful study of living plants in the garden and in the laboratory to tell whether or not the very numerous forms that have been referred to this genus constitute good species or are merely varieties.   Some species are constant, others variable.   Five species have uniformly unifoliolate leaves; two have regularly trifoliolate leaves; and several have leaves with five leaflets; two or three species occasionally have as many as 13 to 15 leaflets.   Most of the species have very small flowers, but one (G. macrantha Merr.), native to Borneo, is said by Ridley to have very large flowers.   Perhaps the most distinctive feature of Glycosmis is the fact that the young twigs and leaves are densely covered with closely-set, rusty-red pubescence, which does not seem to occur on any other genus of Citrus relatives.   The cotyledons of the seeds are thick and fleshy like those of Murraya, Clausena,and most of the other Citrus relatives, and differ widely from those of Micromelum, which are thin and folded.
      Glycosmis citrifolia is sometimes grown in greenhouses in this country, or even out of doors in the southern states.   The fruits are curious small berries having translucent, slightly pinkish pulp, surrounding greenish-brown rounded seeds, almost hemispherical when there are two seeds, but nearly spherical if there is only one.
      It is probable that, in addition to G. citrifolia, other species are to be found in culture in European and American greenhouses and gardens.   Penzig (1887, pp. 194-209, Atl. pls. 19-21) described and figured in detail the morphology and anatomy of two forms of Glycosmis cultivated in Italy, one he determined as G. pentaphylla Corrêa, the other with less certainty as G. lanceolata Kurz.   In this latter species, Penzig found the plumule of the embryo, before germination, to have abundant, rather long, slender, many-septate, brown hairs with punctate walls.   Such hairy plumules are exceedingly rare and were not known to Penzig anywhere else except in certain species belonging to the Meliaceae, a plant family closely related to the Rutaceae.
      The species of Glycosmis are still very inadequately described and need study more urgently than those of any other genus of the orange subfamily.

TENTATIVE LIST OF SPECIES OF GLYCOSMIS
      The taxonomy of Glycosmis is in such a state of confusion that it is not yet possible to key out all the species or even to tell with certainty how many should be recognized as valid.
      Engler (1931, pp. 317-18) listed twenty-one species as follows (numbers added and geographical names modernized): (1) G. pentaphylla (Retz.) Corrêa (= G. cochinchinensis [Lour.] Pierre), very widely distributed; leaves and inflorescences very variable; leaves one- to two-paired or unifoliolate, with long-lanceolate, blunt or acuminate leaflets; distributed in the whole Indian Malayan region through the Malay Peninsula and Timor to northern Australia, and north to the Philippines (among the many varieties is one with a single lanceolate leaflet, in the Khasi Hills, Assam and Burma; another [G. simplicifolia Spreng.] occurs in Java); (2) G. cyanocarpa (Blume) Sprengel, common in the Malayan region and also found in the eastern Himalayan region by Hooker fils and Thompson and sent out from Kew as G. arborea; (3) G. macrocarpa Wight, leaves with 5-1 leaflets; in southern India and Perak; (4) G. ovoidea Pierre, leaves with 5-1 leaflets; in Cambodia; (5) G. erythrocarpa Hayata; leaves with 3-1 leaflets; in Taiwan; (6) G. montana Pierre (= Tetracronia cymosa Pierre, fide Tanaka), leaves with 13 leaflets; in South Vietnam (Cochin China); (7) G. tomentella Ridley, leaves with 15 to 7 leaflets; in Malay Peninsula, Selangor; (8) G. sapindoides Lindley, with a pubescent ovary; in Andamans, Malay Peninsula, Java; (9) G. malayana Ridley, leaves always 5-foliolate; abundant in Malay Peninsula from Singapore to Perak; (10) G. monticola Ridley, leaves always 5-foliolate; rare; in Malay Peninsula, Mt. Ophir, and Gunong Mering at 1,000 m altitude; (11) G. elata Ridley, leaves always 5-foliolate; rare; in Malay Peninsula (Kelantan and Kota Bharu); (12) G. mauritiana (Lam.) Tanaka (= Limonia mauritiana Lam. [Ency., 3:51, 1789] = Limonia pentaphylla Roxb. [Pl. Corom., p. 60, 1795] = G. triphylla Wt. [1833] = G. nitida Wt. & Arn. [1834]), leaves with three or five leaflets, staminal filaments flattened; along the Coromandel coast and in other parts of India and Mauritius Island; (13) G. rupestris Ridley, leaves always trifoliolate; in Malay Peninsula; (14) G. puberula Lindley, leaves always trifoliolate; ovary pubescent; in Malay Peninsula.
      The following species (nos. 15-18) all have unifoliolate leaves: (15) G. dinhensis Pierre, South Vietnam (Cochin China); (15a) G. bonii Guillaumin [= Murraya stenocarpa], North Vietnam, Tonkin; (15b) G. pseudoracemosa [Guill.] Swingle [= G. cyanocarpa (no. 2 above)], North Vietnam, Tonkin; (16) G. crassifolia Ridley, Malay Peninsula; (17) G. lindleyana Swing. nom. nov., G. macrophylla Lindley (non Miquel) ex Ridley [Fl. Malay Penin., 1:349 (1922)], leaves unifoliolate, thin, elliptical, acuminate, 8 by 3.5 inches; ovary conic, on a large disk; type, Wallich, No. 6377, Penang local; (18) G. parkinsonii Tanaka, Burma, Tenasserim.
      Tanaka (1930a, pp. 47-49) reported fifteen species in Indo-China and Thailand, among them: (19) G. pierrei Tanaka (= Thoreldora cochinchinensis Pierre), South Vietnam (Cochin China); (20) G. craibii Tanaka (= G. singuliflora Craib, non Kurz), Thailand; (21) G. longipes (Craib) Tanaka, Thailand.   Ridley (1930, p. 79) reported one new species from Borneo (21a, G. oliveri Stapf [= G. macrantha Merr. (no. 29)]) with very large flowers.   Engler includes these species in his list.
      In addition to the foregoing list given by Engler (1931, pp. 317-18), Tanaka (1928b, 1930a, 1930b, 1931b, 1932e, 1937) has restored some old species and made a few new combinations and new species as follows: (22) G. esquirolii (Lév.) Tanaka, China, Kweichow Province; (23) G. parva Craib, Thailand; (24) G. winitii Craib, Thailand; (25) G. bilocularis Thwaites, Ceylon and peninsular India; (26) G. citrifolia (Willd.) Lindley, southern China, North Vietnam, South Vietnam, Laos, Cambodia, Thailand; (27) G. chlorosperma Blume, Java, Malay Peninsula; Borneo; Balabac Island, P.I.; (28) G. clemensii, Tanaka, North Borneo, Mt. Kinabalu; (29) G. macrantha Merrill, Borneo, Tawau; (30) G. angularis Elmer, Leyte Island, P.I.; (31) G. greenii Elmer, Philippines, Borneo, Moluccas, New Guinea, Australia; (32) G. platyphylla Merrill, Leyte and Samar islands, P.I.; (33) G. singuliflora Kurz, India, upper Assam; (34) G. macrophylla (Bl.) Miquel, Moluccas, Kei Islands; (35) G. boreana Narayanaswami, India, Assam.
      Tanaka also transferred G. bonii (no. 15a above) to the genus Murraya (as a synonym of M. stenocarpa) and considered G. pseudoracemosa (no. 15b) to be a synonym of G. cyanocarpa (no. 2) and G. oliveri (no. 21a) a synonym of G. macrantha (no. 29).   This makes the tentative list total thirty-five supposedly valid species.   There are also many varieties both old and new not listed here.
      To show how many apparently good species can be made out of a single species of the older authors, Kurz (1876, pp.-33-40) made ten species that he thought good out of what had been called G. pentaphylla (Retz.) Corrêa by Oliver (1861) and Hooker (1875).   Kurz described all ten and illustrated seven.   His key gives an excellent idea of the characters shown by the species of Glycosmis; it is reproduced with slight emendations:
      Of the foregoing ten species segregated by Kurz from G. pentaphylla (fig. 3-3), as understood by Oliver (1861) and Hooker (1875), four are listed by Kurz as having named varieties: (a) G. cyanocarpa and (d) G. trifoliata, each with four varieties; (e) G. triphylla and (h) G. arborea, each with two varieties in addition to the typical form.   Seven of these segregates, a, b, c, e, f, g, h, of the key presented are figured in two plates.
      Of these ten species, Kurz considered f, i, and k as distinct species "beyond any doubt," but was undecided whether d, e, and g should not be thrown "all into one species."   In the preceding numbered list, six of Kurz's species are included: a = no. 2; c = no. 1; e = synonym of no. 12; f = no. 33; g = no. 14; k = no. 17.   One species reduced to varietal rank by Kurz (var. beta of species a) was considered a good species by Engler (no. 8 above).
      It is very necessary that all the species and varieties of Glycosmis be subjected to a critical revision and then be fully described and provided with a workable key.   The descriptions of the older species do not suffice to separate these species from the many new ones now recognized, some of which are also inadequately described.
      One or more illustrations of the following eleven species of Glycosmis are cited in Stapf (1929-1931, vol. 3, p. 298, and vol. 6, p. 554): G. arborea var. insularis Kurz; *G. cochinchinensis Pierre (no. 1); G. cyanocarpa Spreng. (no. 2); G. cymosa Kurz; G. lanceolata Spreng.; *G. madagascariensis Corrêa; G. montana Pierre (no. 6); G. pentaphylla Corrêa (no. 1), "(sub nomen Limonia pentaphylla)"; G. puberula Lindl. (no. 14); G. singuliflora Kurz (no. 33); *G. triphylla Wight (no. 12).   (The species names preceded by an asterisk have colored illustrations and those followed by a number in parentheses are thereby assigned to the species in the numbered list above.)

      Glycosmis citrifolia (Willd.) Lindl. Trans. Hort. Soc. Lond. 6:72. 1826. Limonia citrifolia Willd. Enum. 448. 1809; L. parviflora Sims, Bot. Mag. pl. 2416. 1823; Glycosmis pentaphylla, var. beta, subvar. 2 (chinensis) Oliv. Jour. Linn Soc. Bot. 5(2):37. 1861. Illus. Sims, loc. cit. pl. 2416; Tillson & Bamford, Amer. Jour. Bot. 25:782, figs. 12, 13, 1938; Everett, Addisonia 21:pl. 687. 1940.
      Type.—(Willdenow, No. 8116), Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Southern China, North Vietnam, South Vietnam, Cambodia, Laos, Thailand; widely cultivated in the Old and New Worlds.
      Common name.—Chinese glycosmis.
      A small tree or shrub with slender, glabrous, greenish twigs with nodes 2-4 cm apart; leaves 1-, 2-, or 3-foliolate, rarely 4- or 5-foliolate; leaflets (or the single blade) elliptical or oblong-elliptical, 6-17 X 2-5 cm, usually acute at the tip and narrowly cuneate at the base, rarely somewhat rounded at base, more rarely at tip, more or less covered with rusty-brown, short pubescence when very young but soon glabrescent, lateral veins very numerous, 12-15, visible on both surfaces, variable in size, only the stronger ones running nearly straight about 3/4 of the way to the margin, at an angle of 50°-60° with the midrib; petioles 1.5-2.5 cm long, articulated with the blade in simple leaves; petiolules of lateral leaflets 4-8 mm long, with a channel on the upper side made by the decurrent leaf margins, articulated only at the base with the rachis; the terminal leaflet of pinnate leaves often articulated with a free portion of the rachis 1-2 cm long; inflorescences small, axillary, usually 2-3, sometimes 4-5 cm long, rarely terminal, sparingly branched, pedicels very short, all parts of the inflorescence covered, when young, by a dense rusty-brown pubescence; flowers small, white, 4-5-merous; sepal lobes small, subtriangular, with more or less ciliate margins; corolla campanulate, petals white, small, 3.5-4.5 mm long, bluntly cuneate at the tips, soon falling; stamens 10, filaments filiform, slightly dilated upward; ovary short, ovoid, 1.5-2 X 1.3-1.6 mm, rough or tubercular, merging into a short, thick, persistent style, 0.4-0.5 mm long, tapering upward and ending in a depressed cushion-shaped, persistent stigma, 0.5 mm wide and 0.2-0.3 mm high; fruits subglobose, translucent white or pinkish, 11-13.5 mm diam.; seeds 1-3, oblong, cotyledons green.
      This species, the Chinese glycosmis, although widely cultivated in all tropical countries and in greenhouses in the temperate zones, is still inadequately studied.   The description above was based on the type specimen (leaves only) preserved at the Berlin-Dahlem herbarium, and on material from southern China determined as this species by Tanaka.   The Chinese glycosmis is often labeled G. pentaphylla (Retz.) Corrêa in gardens or herbaria, as many authors have considered it as being merely a form of that species.   It has also been called G. cochinchinensis Pierre by some botanists.   These three species, and all the others for that matter, should be introduced into the United States and grown side by side so that they can be properly distinguished, adequately described, and tested as possible rootstocks for Citrus and as ornamental plants for greenhouse culture or for planting in subtropical gardens and parks.   Glycosmis citrifolia has become naturalized in the hammocks of Key West, according to Small (1933, p. 759) and Everett (1940, p. 29).

III.   Clausena Burm. f.
      III.   Clausena Burm. f. Fl. Ind. 87, 243. 1768. Cookia Sonner, 1782; Fagarastrum G. Don. 1832; Gallesioa Roem. 1846; Myaris Presl. 1849; Quinaria Lour. 1790; Piptostylis Dalz. 1851; Polcyema Voigt. 1845.
      Type species.Clausena excavata Burm. f.
      Distribution.—Southwestern Asia, East Indian Archipelago, Philippines, New Guinea, Australia, northeastern tropical Africa from Abyssinia south to Natal and Pondoland in Cape Province of the Republic of South Africa, central tropical Africa, and western Africa from Angola to Sierra Leone.
      Trees or shrubs without spines; leaves odd-pinnate, leaflets alternate, usually 3-7 sometimes 19-32; rachis usually not winged; inflorescences terminal or axillary, paniculate or laxly racemose, flowers often in cymose clusters, flower buds small, subglobose or short-oval or oblong, never long-cylindric; flowers small, calyx 4-5 lobed; sepals fused into a cup below; petals free, 4 or 5 imbricate in the bud, usually elliptical; stamens 10, in 2 whorls, the outer row opposite, the sepals usually longer, filaments more or less dilated or flattened below, often geniculate where the filiform apical portion joins the flattened and expanded basal portion, anthers ovate or elliptical, rarely short and subglobose; gynophore well developed, usually a perfectly glabrous hourglass-shaped column resting on an annular nectary below and supporting (sometimes clasping) the base of the ovary above; ovary with 2-5 locules, often pubescent or glandular, ovules 2 in each locule (rarely 1); style deciduous, often shorter than the ovary or equaling it in length (rarely longer), thick, sometimes merging gradually into the ovary, often sharply delimited and narrowed where it joins the ovary, stigma inconspicuous, sometimes subcapitate after the style shrivels; fruits small, subglobose or ovate, with 2-5 segments, sometimes only 1 seed maturing; all flower parts usually showing many (not all) cells strongly impregnated with tannin residues.
      The most distinctive morphological character of the genus Clausena is the gynophore, which in the typical species is a large, well-developed, hourglass-shaped structure supporting the ovary.   It is perfectly glabrous and rests upon and merges into a short annular nectary below and sometimes expands above into a thin-lipped cup-like structure that encloses the base of the ovary.   However, the gynophore varies greatly in the different groups of species and in some is so modified by expansions caused by development of oil glands that it is difficult to recognize.   Nevertheless, it is present in all species of Clausena and separates them from the species of other related genera.

RELATIONSHIPS OF THE SPECIES OF CLAUSENA WITH ONE ANOTHER
      The numerous species of this genus, still only imperfectly studied with respect to the minute flower characters, cannot be arranged now in natural sections or subgenera.   It is clear to any taxonomist who looks over carefully a large collection of the species of Clausena that they can be classified into obviously related groups, some small, some large.   Unfortunately, these groups are not clearly distinguished from one another but, on the contrary, seem rather to merge into allied groups without any sharp line of demarcation.
      The study of the flower characters and in particular of the gynoecium by means of serial microtome sections has shown that important characters are found in the pistil that may help to define the natural groups of species.
      The type species of the genus, C. excavata, has a striking hourglass-shaped gynophore which is completely glabrous.   It arises from the annular nectary below and, after contracting, again expands and may even be slightly flaring where it joins the ovary, which is strongly hirsute.   The ovary, which is 5-locular, merges into the style without any clear line of demarcation, and the style is not narrowed where it joins the tip of the ovary.
      On the other hand, in C. indica the style is abruptly contracted and countersunk into a conical depression at the tip of the ovary, which has two to five locules.   This species was made the type of the genus Piptostylis and belongs to a large group which includes many other species.
      The species C. pentaphylla at present prevents a clear separation of the two groups outlined above.   The pistil shows no sharp boundaries of gynophore, ovary, and style; the gynophore is as broad as the ovary—not hourglass-shaped—and is more or less lobed, often having short processes each ending in an oil gland.   The style may be slightly contracted where it joins the top of the ovary but is frequently more or less swollen with oil glands just at this junction.
      Clausena lansium stands apart from all the other species of the genus in having a star-shaped bud (due to its five strongly carinate petals) and a 5-angled ovary, corresponding in position to the five petal keels in a cross section of the bud.   The pistil is unusually large; the ovary is very strongly hirsute and merges into the style, which is slightly contracted at the base.   It is hourglass-shaped as in C. excavata.   Furthermore, the twigs show a central gum canal, doubtless formed by a lysigenous breakdown of the tissues.   The locules of the ovary contain scattered but well-developed hairs which arise on the dorsal and outer portion of the lateral walls and usually grow more or less toward the axis of the ovary.   One much longer hair arises at the base of each locule and grows vertically upward, nearly to the top of the locular cavity (Penzig, 1887, p. 179, pl. 16, figs. 11, 12, 16).   The fruits are apparently the largest in the genus and the locules of the ripe fruits are almost filled with a pulpy tissue which Penzig found to arise from the hypertrophy and multiplication of layers of cells nearest to the endocarp.   This species was the type of the genus Cookia, but no other species having similar characters are yet known.
      Clausena lenis also stands alone in the genus, having leathery petals and a most peculiar gynophore, no longer hourglass-shaped as in almost all the other species.   Just above the thin annular nectary there is a narrow constriction followed by a cylindrical gynophore evidently swollen by a ring of 12 to 15 oil glands at its very base; it merges above into the ovary.   The style is slender, two to three times as long as the ovary, by far the longest of any species in the genus, and contracted below, where it joins the ovary.   The leaflets are subdentate, and the largest ones are lateral leaflets attached about two-thirds of the way up the rachis.
      Clausena guillauminii has leaves showing dimorphic oil glands, the larger ones being 1/3 to 1/2 mm in diameter, and deep red by transmitted light; the flower parts show a single large oil gland at the tip of each sepal, each petal, each anther, and each locule of the ovary.   C. wallichii has almost the same distribution of the oil glands in the flower parts.   C. papuana shows great variation in the size of the oil glands in the leaflets and has the oil glands similarly distributed in the flower parts.   These species may be closely related; certainly C. guillauminii, C. wallichii and C. luxurians are.   The last-named species have the rachis slightly winged, unlike all other species of Clausena.

THE SPECIES PROBLEM IN CLAUSENA
      Of the twenty-three species of Clausena recognized here, ten were published from 1900 to 1930 and one as late as 1940.   Of the others, one was published in 1892 and eleven from 1768 to 1875.   It now appears that many of the older species, although they were described adequately enough to distinguish them from the other species known at the time, were nevertheless not described fully enough to separate them clearly from the many species recently discovered.   Then, too, the range of the older species has been extended steadily as new botanical collections have been made.   Very widely distributed species like C. excavata and C. anisata show, as might be expected, very many variations in their characters in specimens collected from widely different localities.
      For these reasons a clear-cut definition of the species of Clausena will require a great amount of work which can be done only by using all the material in the leading herbaria of the world.
      Tanaka, who assiduously studied Clausena (and also the other genera of Aurantioideae) from 1927 to 1937 in the herbaria of Europe, Asia, and America, has done much to clarify the species of this genus.   He has published notes on them in all his valuable papers included in the series "Reviso aurantiacearum."
      Swingle was permitted to borrow nearly a thousand sheets of material of Clausena from many of the leading herbaria of the Old World and the New, including all the Clausena material in the herbariums of the Royal Botanic Gardens at Kew and the Royal Botanic Garden at Calcutta.   Thanks to these loans he was able to make a close study of many of the actual type specimens of the species of Clausena.   In the study of this rich material he used a modification of Juel's method of softening herbarium material, after which it was fixed, dehydrated, imbedded, cut into serial microtome sections, double-stained, and finally mounted permanently on glass slides (Swingle, 1939b, p. 270).
      In this work, Swingle had the skillful assistance of Dr. Albert H. Tillson.   Figure 3-4, which shows longitudinal microtome sections of nine species of Clausena, makes clear how helpful this method is in studying herbarium material, especially the pistil just after the petals and stamens have fallen.   It has not, however, been possible to work out all the species satisfactorily.   Some species, like C. anisata, vary so greatly that months of work should be needed to study its numerous forms adequately, and other species are not well enough represented in herbaria for any complete study.

GEOGRAPHICAL DISTRIBUTION AND USES
      The geographical distribution of Clausena is of interest since it has the widest range of any genus included in the orange subfamily, species being found all the way from northwestern India to China and Taiwan, south through the East Indian Archipelago to Timor, northern Australia, and New Guinea.   Moreover, there is a group of three species that covers a wide range in Africa, from Ethiopia (Abyssinia) to Cape Province and through western Africa from Angola north to Sierra Leone.
      There are wide differences in the character of the growth and the height of the species; they range all the way from shrubs 20 to 40 cm high in Indo-China to trees reaching a height of 20 m (66 feet) in Africa.
      Several species of Clausena produce edible fruits, but up to now only one of them, the Chinese wampee, C. lansium, is cultivated (in southeastern China) on a commercial scale, with many varieties, the fruits varying greatly in size, flavor, and time of ripening.   One other species, the Indian wampee, C. dentata var. dulcis, is gathered on a large scale from wild trees in the state of Madras in southern India, where it is a highly appreciated fruit, said by the British experts who have studied it to be of excellent flavor.   At least two other forms are said to produce edible fruits, C. indica of India and Ceylon and C. dentata var. henryi of southwestern China.
      The Chinese wampee, C. lansium, can be used as a rootstock for Citrus, provided a few twigs of the wampee are left growing below the graft.   Other species of Clausena should be tested as stocks for Citrus and also for the edible-fruited species of Clausena.
      Clausena anisum-olens, a species common in the Philippines from Luzon Island to Mindanao Island, grows in the forests from sea level to 1,500 m altitude.   It is a small tree 3 to 6 m tall and has leaves with a strong odor of anis oil.   West and Brown (1920, pp. 211-12) stated that the suggestion had been made that the leaves could be used locally in preparing anisado, a favorite alcoholic beverage among the Filipinos.   The essential oil of the leaves was found by Brooks (1911, p. 344) to consist very largely (from 90 to 95 per cent) of methyl chavicol, which is easily converted into anethol, the true anis oil.   The leaves yield about 1.2 per cent of oil after five hours’ distillation over steam.   This essential oil is not known to occur in such a large proportion in any other plant of the family Rutaceae.
      It is highly probable that other species of this large and very widely distributed genus, when studied carefully, will be found to be of value for their fruits, for use as rootstocks, for their essential oils, or for planting as ornamentals because of their abundant white flowers and handsome foliage.
      The following four keys covering the species native to the four principal regions where Clausena is indigenous will help in giving an idea of the differences between the species.

      1.   Clausena excavata Burm. f. Fl. Ind. 87. 1768. Murraya burmanni Spreng. 1825; Amyris sumatrana Roxb. 1832; A. punctata Roxb. 1832; Gallesioa graveolens Roem. 1846; Clausena lunulata Hay. 1911; C. tetramera Hay. 1916; C. moningerae Merr. 1923. Illus. Burmann f. loc. cit. pl. 29, fig. 2; Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:664, fig. 70(8-10). 1911; Tillson & Bamford, Amer. Jour. Bot. 25:782, fig. 15. 1938; fig. 3-4,A-B, and fig. 3-5 this work.
      Type.—Java (Burmann, Dauhon Kongeere, No. 29).   Herb. Van Royen, in Rijks Herb., Leiden, sheet "908, 203-…1051" (fide Tanaka, 1932e, p. 422).
      Distribution.—India: Coromandel and northern districts to Bhutan; Burma; Thailand; southern China; North Vietnam; South Vietnam; Laos; Cambodia; Malay Peninsula; Sumatra; Java; Borneo; Philippines; New Guinea.
      Hooker (1875 vol. 1, p. 504) described this species, with slight rearrangement, as follows: "A [small] tree, branchlets as thick as a crow-quill and, as well as the inflorescence and leaflets, more or less softly tomentose; leaves 15-30 foliolate, 6-12 in. [15-30 cm] long; petiole slender, cylindric, more tomentose than the leaflets, which are very oblique at the base, petioluled, 2-3.5 in. [5-8 cm] long, ovate or lanceolate, acuminate, membranous, crenate, upper ones often falcate.   Panicle 4-12 in. [10-30 cm] high, pyramidal, branches spreading, alternate.   Flowers 4-merous, shortly pediceled, 1/6 in. [4 mm] diam., white, buds globose.   Petals oblong, glabrous.   Ovary ovoid or elliptic, sub-4-gonal stipitate, hairy or hirsute; style stout, about equaling the ovary.   Fruit 3/4 in. [18 mm] diam., broadly oblong."
      Burmann’s original description of C. excavata (l.c.) covers the stamens, in translation, as follows: "Staminal filaments 8, subulate, shorter than the corolla, the inferior part dilated and thickened, hollowed out (‘excavata')2 within."
      Guillaumin (1911, p. 661) described the stamens, in translation, as follows: "Stamens 8, one-third or a half shorter than the petals; filaments awl-shaped above, then suddenly dilated, concave and papillose at the base with a hump at the upper part; anthers oval, apiculate."    He also figured the stamens (1911, p. 664, fig. 70 [9-10]), showing the lower expanded and papillose part of the filament to be strongly curved outward and the upper slender, smooth portion to be even more strongly curved outward, almost into a semicircle.
      The gynoecium of this species, as studied from serial microtome sections, shows the following structure: pistil 3.2-4.2 mm long; gynophore very strongly developed, completely glabrous, hourglass-shaped, usually with a basal annulus 1-1.2 mm diam., then contracted to 0.7-0.8- mm diam., then expanded into a more or less cupulate top, 1-1.2 mm diam., embracing the base of the ovary, with edges often showing small oil glands; ovary cylindrical or conoid, usually strongly hirsute, 1-5 mm long, 1-1.2 mm wide, often with a few small oil glands where the style is attached; style cylindrical, not narrowed at point of attachment, but merging into the top of the ovary, glabrous, 1-1.2 mm long, 0.5-0.8 mm wide, stigma very short, flat-topped, only slightly wider than the style.   (fig. 3-4,A-B.)
      This species is the type of the genus Clausena and has the widest distribution of any species, ranging through the East Indian Archipelago to southeastern Asia.   As would be expected from its very wide range, the species has undergone many variations, until the forms from Sumatra and Taiwan do not at first sight seem to belong together in a single species.   However, the pistil of this species is distinctive, and is much the same in all these forms and very different from that of the other species of the genus Clausena.   It has a rather full, smooth, hourglass-shaped gynophore supporting the more or less strongly hirsute ovary.   The ovary tapers gradually into the thick style, which shows no constriction at its base as do the styles of many other species.
      Only one variety is here recognized, but it would be easy to designate three or four equally well characterized!
      1a.   Clausena excavata var. villosa Hook. f. Fl. Brit. Ind. 1:505. 1875.
      Type.—Burma, Rangoon (McLelland, No. 1854).   Herb. Kew.
      Distribution.—Upper Burma; also Thailand, Indo-China, Malaya, and rarely in Nepal (fide Tanaka, 1937, p. 231).
      Whole plant clothed with soft spreading hair: petals hirsute.
      This hairy form is found near the northcentral part of the range of the species.
      2.   Clausena lansium (Lour.) Skeels, U.S. Dept. Agri. Bur. Pl. Ind. Bul. 168:3. 1909. Cookia punctata Sonner. Voy. Ind. Or. 2:231. 1782; Quinaria lansium Lour. 1790; Cookia wampi Blanco, Fl. Filip. ed. 1. 358. 1837; Clausena wampi Oliv. Jour. Linn. Soc. Bot. 5(2):34. 1861. Illus. Sonnerat, loc. cit. pls. 130, 139; Jacquin, Pl. Rar. 1:pl. 101 (col.). 1797; Penzig, Studi Bot. Sugli Agrumi, Atl. pls. 15, 16. 1887; De Wildeman, Icon. Sel. Hort. Then. 6:pl. 226. 1908; Mill Tsen, Yuan I, Hort. 2:596-97. 1936, 2 text cuts with 6 figs. of fruits and 6 of leaflets; fig. 3-4,E,F,I, and fig. 3-6 this work.
      Type.—China, Canton (Loureiro).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Native in southern China and Indo-China; widely cultivated in tropical and subtropical regions.
      Common name.—Wampee.
      Oliver (1861, p. 34) gave a Latin diagnosis of this species that, translated, reads as follows: "Tree or shrub; branches at first pubescent or puberulous; leaves 5-7-9-foliolate, 20-25 cm long; leaflets ovate-elliptic, lanceolate or ovate, with petiolules 2-6 mm, more or less oblique at the base, apex obtuse or subemarginate, sometimes almost mucronulate, slightly acuminate, margin undulate-crenate or slightly serrulate, becoming glabrous or glabrescent, the midrib and veins often sparsely scabrous, glabrescent below; terminal leaflet often 6-10 cm long; flowers subsessile or shortly pedicellate, in many-flowered cymose panicles; calyx-5- (rarely 4-) merous, lobes triangular or ovate; stamens 10, alternate ones shorter, filaments dilated above the middle, flattened, subulate above, anthers oblong or elliptic, cordate-sagittate at base, with a dorsal gland; ovary shortly stipitate, glandular, very often 5-locular, ovules in pairs [in each locule], superposed, the upper one peltate on the side or toward the base, the lower one subpendulous; style very short, distinct, glabrous above, stigma 5-lobed, slightly wider than style; fruit usually with 5 locules, 5-seeded, or by abortion 1-seeded, or sterile, ovoid-globose, pubescent, 1 in. [25 mm] or less diam.; cotyledons fleshy, equal."
      This species differs widely from all the other species of the genus, as has been noted.   The wampee is a highly esteemed fruit tree in southern China, where sour, subacid, and sweet varieties are known.   The white or yellow fruits are ovoid or subglobose, about the size of a pigeon’s egg, but varying in size and shape with the variety cultivated.   The inflorescences are large panicles at the ends of the branches, so the fruits occur in rather openly-spaced clusters.   In texture, the wampee fruit is much like a loquat and is a berry without a tough peel.   An illustrated treatise in Chinese by Mill Tsen [Ts’êng Mien-chih, in Mandarin transliteration] (1936) described in detail six varieties and figured the fruit and a leaflet of each variety.   (Mill Tsen mentioned another variety [no. 6 below], which is not figured.)   Table 3-1 shows the principal characters of the fruits of these Foochow varieties (the leaves vary almost as much as the fruits but are omitted from the table).
      According to Mill Tsen, Professor Wên Wên-kuang of Sun Yat-sen University at Canton, China, listed eight varieties of the wampee grown in Kwangtung Province.   These varieties are not described in detail but are given long descriptive names such as "white-hairy-chicken-heart-sweet-wampee"; "long-chicken-heart-sour-wampee"; "yellow-hairy-chicken-heart-sour-wampee," etc.
      Although the wampee is only remotely related to Citrus, it can be grafted on a Citrus rootstock, and is thereby forced into early flowering and fruiting.   The rough lemon can be grafted on the wampee and will grow for many years if a small branch of the wampee is allowed to grow just below the graft.   Any desired Citrus fruit can be top-grafted on the rough lemon and may live for many years and may even be made to fruit.   However, the graft union of the rough lemon on the wampee is poor and badly overgrown.   Wampee on Citrus seems to make a much better and more permanent union.
      3.   Clausena pentaphylla (Roxb.) DC. Prodr. 1:538. 1824. Amyris pentaphylla Roxb. Fl. Ind. 2:247. 1832. Illus. fig. 3-4,C.
      Type.—British India, United Provinces, Cawnpore, a living plant sent to Calcutta Bot. Gard. and grown there (Roxburgh).   Herb. Brit. Mus., London.
      Distribution.—India: western Himalayan region, from Garhwal to Sikkim, northern Uttar Pradesh and Oudh.
      Brandis (1906, p. 114) gave a condensed description of the species as follows: "A strongly aromatic shrub, young shoots and inflorescences silky-tomentose, full-grown leaves pubescent.   Leaflets 5 or 7, nearly opposite, 4-6 in. [10-15 cm] long, secondary nerves prominent beneath.   Flowers yellowish, in terminal panicles.   Berry verrucose, pale orange, 1/3-1/2 in. [8-13 mm] long."   Duthie (1905, p. 138) in his Flora of the Upper Ganghetic Plain, a region where this is the only native species of Clausena, described it as follows: "A deciduous shrub; young parts silky-tomentose.   Leaves large, 3-7-foliolate.   Leaflets subopposite or alternate, 2-6 in. [5-15 cm] long, ovate or lanceolate, acuminate, entire, shortly stalked, softly tomentose beneath; main lateral nerves prominent.   Flowers 4-merous, 1/4 in. [6 mm] across, yellowish, in erect terminal downy panicles.   Sepals triangular, acute.   Petals oblong, concave.   Filaments short, broad.   Berry 1/2 in. [13 mm] diam., ovoid, papillose, pale orange."
      Good flowering material collected by J. F. Duthie in the western Himalayan region at Gonda Dun (No. 21613, Herb. Arnold Arbor.) shows the following gynoecium characters: The pistil, 2-2.2 mm long, shows no sharp boundaries of disk, ovary, style, and stigma.   The 3- or 4-locular ovary is 1-1.2 mm wide and 0.9-1 mm long, irregularly tuberculate and sparingly pilose; the style is cylindrical above, 0.35-0.45 mm wide, with no obvious stigma; below, it may be slightly contracted where it joins the top of the ovary, but very often it is swollen with rather large oil glands at or just above its junction with the tip of the ovary and in consequence may be nearly as wide as the ovary.   The disk is most irregular, usually as broad as the ovary but more or less lobed and even having short processes each ending in an oil gland.   The locules of the ovary run deeply into the disk, which in longitudinal section is not clearly separated from the base of the ovary.
      This species, remarkable for its deciduous foliage and shrubby habit, has even more remarkable pistils that depart widely in their characters from those of all other species of the genus.   Its nearest analogue in gynophore morphology is the anomalous C. lenis, which has a more or less syncopated gynoecium.
      4.   Clausena indica (Dalz.) Oliv. Jour. Linn. Soc. Bot. 5(2):36. 1861. Piptostylis indica Dalz. Hook. Jour. Bot. & Kew Gard. Misc. 3:33. 1851; Bergera nitida Thw. Enum. Pl. Zeyl. 46. 1858. Illus. Dalzell, loc. cit. pl. 2; Engler, Die Nat. Pflanzenfam. 3(4):187, fig. 108,L-O. 1896; ibid. ed. 2. 19a:321, fig. 146,L-O. 1931; Talbot, Forest Fl. Bomb. Presid. 1:195, fig. 119. 1909; fig. 3-4,K this work.
      Type.—British India, Bombay Ghats, Kanara (Dalzell).   Herb. Kew.
      Distribution.—India: western peninsula, Bombay Ghats to Anaimalai Hills, Madras State, eastern coast, Tranquebar; Ceylon.
      Hooker (1875, vol. 1, p. 505) gave a description that may be summarized as follows: "A glabrous shrub or small tree; leaves 9-11-foliolate, 4-10 in. [10-25 cm] long; petioles slender…glabrous or puberulous; leaflets polymorphous, very oblique oblong-ovate, elliptic or lanceolate, rarely almost rounded, tip rounded-obtuse and notched or acute or acuminate, margins crenulate, very dark with raised glands on both sides when dry; inflorescence a peduncled panicle, corymbosely branched above; flowers 1/6 in. [4 mm] diam., very shortly pediceled; ovary 2-5 celled, glabrous, papillose; ovules 2 in each cell; fruit 1/2 in. [13 mm] diam., globose, yellow, edible."   Talbot (1909) added a few details that can be condensed as follows: Leaflets subcoriaceous, 2-4 X 1-2 in. [5-10 X 2.5-5 cm]; petiolules 1/4 in. [6 mm] long; petals broadly ovate; stamens with filaments broad, as long or longer than the versatile anthers; ovary usually 3-loculed; fruit pulpy, edible.   Dalzell (l.c.) described and figured the style as much shorter than the ovary, thick, cylindrical, with the stigma scarcely wider than the style, which is not expanded at the base where it is situated in a depression at the tip of the ovary and wholly caducous.   He described the flowers as 4- or 5-merous but did not see any but 3-locular ovaries.
      Clausena indica is a shrub or small tree, 2 to 2.3 m high, common in the evergreen rain forests of the Ghats of western peninsular India.   In spite of its rather limited distribution, it shows great variation in the shape of the leaflets which differ greatly in outline, being "very oblique" in Ceylon, according to Trimen (1893, p. 221), but only slightly if at all oblique at the base in the type locality in the Bombay Ghats, as is shown by Dalzell's careful drawing (and even more by his type specimen in the Kew herbarium) as well as by Talbot's sketch.   The ovary varies even more strikingly than the leaflets, having regularly three locules in the type locality, according to Dalzell (l. c.) and Talbot (1909), whereas in Ceylon it always has two locules, fide Dalzell (l. c.) and Trimen (1893, p. 221).   A specimen from Ceylon sent by the Royal Botanic Garden at Peradeniya to the herbarium of the Arnold Arboretum (Tanaka's "Det. A-229") has 2-, 3-, and 4-locular ovaries in the same inflorescence.   Oliver and Hooker both found forms with 4- or 5-locular ovaries.
      As this species bears edible fruits, it should be hybridized, if possible, with the Chinese wampee and with other species that yield edible fruits.
      5.   Clausena heptaphylla (Roxb.) Wt. & Arn. Prodr. 96. 1834; and in Wall. Cat. 8508. 1848. Amyris heptaphylla Roxb. Fl. Ind. 2:248. 1832; Clausena forbesii Engl. 1896. Illus. Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:664, fig. 70(11-12). 1911; Tillson & Bamford, Amer. Jour. Bot. 25:782, fig. 14. 1938; fig. 3-4,D this work.
      Type.—British India, "about Calcutta" (Roxburgh).   Herb. Brit. Mus., London.
      Distribution.—India, Burma, Thailand, North Vietnam, South Vietnam, Laos, Cambodia and Sumatra (also Java, fide Tanaka).
      The description of this species given by Guillaumin (1911, p. 662) reads, translated, as follows: "A shrub; leaves 30-50 cm long; leaflets 7-11, unequal-sided at base (except the terminal one), membranaceous, rigid, long-oval or lanceolate, 4.5-16 X 2.5-7 cm, acute at base, acuminate at apex, entire, glabrous except the midrib on the under surface, with 6-7 pairs of lateral veins which are raised on the under surface; petioles cylindrical, with very short pubescence, glandular.   Inflorescences terminal, paniculate, pyramidal, shorter than the leaves, with alternate, slightly and shortly pubescent branches; pedicels very short with 2 basal, shortly pubescent bracts articulated at the very base.   Sepals 4, triangular, 1 mm long with ciliate margins.   Petals 4, obovate, glabrous but papillose, twice as long as the sepals.   Stamens 10, slightly shorter than the petals; filaments filiform at base then abruptly dilated below the anther; anthers attached dorsally along the middle, almost rhombohedral, without horns.   Disk very short, glabrous.   Ovary cylindric, quadrangular glabrous, as long as the stamens, 4-locular, each with 2 superimposed ovules; style cylindrical, as long as the ovary.   Fruit ovoid, 2 X 1 cm, very glandular, with 1 or 2 seeds."
      Hooker (1875, vol. 1, p. 504), who described the species from material from a wide range of localities, noted that it is a "branching bush, smelling strongly of aniseed.…Leaves 6-16 in. [15-40 cm long]…leaflets sometimes 8 in. [20 cm long]…membranous, ovate or lanceolate, acuminate, very obscurely crenulate, pale beneath, nerves and costa slender.…Flowers 1/6 in. [6 mm] diam., pedicels slender, buds glabrous, yellow-white.…Berry 1/2 in. [12 mm] long, oblong, white-reddish or pale yellow, glandular."
      Engler (1931, p. 321) mentioned in his notes on Clausena: "C. forbesii [from Sumatra] with leaves nearly one meter long, having eight paired leaflets," the longest leaves yet reported in the genus.   Probably it is merely an unusually vigorous growing form of this polymorphic species.
      This species, in spite of its abundance, needs renewed study to determine its limits of distribution in the East Indian Archipelago.   Tanaka (1937, p. 231) even reported it from the Sunda Islands and Timor.   It has not been reported from Ceylon or from Borneo or the Philippines.
      5a.   Clausena heptaphylla var. engleri (Tan.) Swing. Jour. Wash. Acad. Sci. 32:26. 1942. Clausena engleri Tan. Med. Rijks Herb. Leiden 69:6. 1931; (?) C. platyphylla Merr. Pap. Mich. Acad. Sci. 23:182. 1938.
      Type.—Sumatra (Lörzing, No. 6825).   Rijks Herb., Leiden.
      Distribution.—Known only from Sumatra.
      Tanaka's original description reads, translated, as follows: "A robust plant; rachis of leaf 36 cm long, 7-foliolate; leaflets ovate, acuminate-caudate, pointed at the base, with conspicuous veins, when dry chartaceous, very green on the upper surface, pallid below, margins serrate, veins rufescent, with very short petiolules, entirely glabrous.   Inflorescence terminal, 28 cm long, branched into many-flowered panicles.   Flowers small, 5-merous, short-pedicellate.   Calyx thin, scutellate, with triangular pilose lobes having ciliate margins.   Petals 5, caducous.   Stamens 10, filaments dilated, anthers large.   Ovary cylindrical, pilose; style linear, straight or curved, stigma somewhat capitate."
      Tanaka, in his remarks following the Latin description, stated that this plant resembles the robust form of C. heptaphylla, but the latter has 4-merous flowers, glabrous ovary, and entire leaflets."   The photograph of Tanaka's type specimen shows it to have large leaflets, 11 to 19 by 5 to 8 cm, at the apex, cuneate and only slightly unequal-sided at the base, with 10 to 12 pairs of lateral veins.   C. platyphylla, recently described as a new species from Asahan in northeastern Sumatra, seems likely to prove to be a synonym of C. heptaphylla var. engleri once a comparative study is made of the type specimens of these two plants.   Both of them have leaflets of about the same size, of similar shape, with the same number of lateral veins, and even share the peculiarity of the dried leaves being dark, shining green above and much lighter, pallid green below.   The ovary of var. engleri is pilose, whereas that of C. platyphylla shows scattered pubescence; the ovary of the species C. heptaphylla is glabrous.
      6.   Clausena dentata (Willd.) Roem. Syn. Hesper. 44. 1846. Amyris dentata Willd. Sp. Pl. 2:337. 1799, Clausena willdenowii Wt. & Arn. 1834. Illus. Wight, Icon. Pl. Ind. Or. 1:pl. 14/339. 1838; Beddome, Fl. Sylv. Anal. Gen. pl. 7. 1871.
      Type.—British India (Willdenow, No. 7293).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Ceylon; India: western peninsula, eastern Himalayan region, Sikkim to northwestern Burma; Thailand; southwestern China.
      This species was described by Hooker fils (1875, vol. 1, p. 506), with some transpositions, as follows: "A large shrub, glabrous or…pubescent; branchlets slender.   Leaves 5-13-foliolate, 6-16 in. [15-40 cm] long; petiole slender, terete; leaflets 1-2 in. [2.5-5 cm] long, oblique, oblong-ovate or ovate-lanceolate, obtuse, acute or obtusely caudate-acuminate, crenulate, membranous.   Racemes slender, axillary, 3-10 in. [7.5-25 cm] long, branches or pedicels divaricating.   Flowers fragrant, usually 4-merous, 1/3 in. [8 mm] diam., whitish; buds globose.   Petals oblong, concave, glandular, ovary stipitate, 4-angled or grooved, glabrous, ovules 2, superimposed in each cell; style short, thick.   Fruit globose, from the size of a pea to a cherry, whitish-green, pellucid."
      A specimen from southern India, Madras State, Hosur Taluk (coll. by Mrs. Kanoth Yeshoda; now in Herb. Arnold Arbor.), shows the restored flowers in serial microtome sections to have the following characters: pistil, 3.5 mm long; gynophore short, 0.5-0.6 mm long, narrowed to 0.7 mm just below the flatly rounded, nearly truncate ovary base; ovary subglobose, 1.3-1.4 mm long, 1.2-1.3 mm wide, tip abruptly rounded with a slight depression in which the nearly cylindrical style is attached; style 1.5 mm long, 0.3-0.4 mm wide, stigma short, cushion-like, 0.5 mm wide.   Cross sections show 4-locular ovary with a very short process at the tip of each locule but with no large oil gland.   The pedicels of the flowers are unusually long (4-6 mm) and slender, with small bracts at the base.
      This species, like C. suffruticosa and the three African species, C. abyssinica, C. anisata, and C. inaequalis, has axillary flower clusters, not terminal panicles as in most of the other species.   The long pedicels of the flowers are unusual in Clausena.
      6a.   Clausena dentata var. pubescens (Wt. & Arn.) Tan. Jour. Bot. Brit & For. 68:227. 1930. Clausena pubescens Wt. & Arn. Prodr. 1: 96. 1834; C. willdenowii var. pubescens (Wt. & Arn.) Hook. f. Fl. Brit. Ind. 1:506. 1875.
      Type.—Ceylon (Wight, No. 328b).   Herb. Glasgow Univ.
      Distribution.—Ceylon; India: Mysore State.
      Tanaka (1930b, p. 227) described this variety, in translation, as follows: "Leaves, inflorescences, and petioles densely villose.   Fruits globose, rather small."
      Tanaka (1937, p. 230) stated that this variety is very rare in Mysore and Ceylon.
      6b.   Clausena dentata var. longipes (Craib) Tan. Bul. Soc. Bot. France 75:709. 1928. Clausena longipes Craib, Kew. Bul. Misc. Inform. 1926:340. 1926.
      Type.—Mê Lan, Mê Hawng Sawa, Thailand, 700 m altitude, on limestone rock (Kerr, No. 5486-A).   Herb. Kew.
      Distribution.—Thailand and upper Burma.
      Differs from the species in having (1) very slender, few-flowered inflorescences ending in long, slender pedicels, 5-15 X 0.2-0.3 mm; (2) slender (1.7-2-5 X 1 mm) ovaries that merge gradually into the rather long style (2.5-3 X 0.6-0.7 mm); (3) no oil gland at the tip of each of the 4 locules of the ovary; (4) very narrow radial locule walls, not swollen where they cross to make an axial pillar; and (5) few tannin cells in the periphery of the ovary and none in the axis.
      This variety appears at first sight to be specifically distinct from C. dentata, but some of the material of the species from southern India shows very slender inflorescences and almost equal-sided leaflets very like those of var. longipes.   Craib (1926, p. 234) considered this plant to be related to C. dentata and to C. suffruticosa.
      6c.   Clausena dentata var. dulcis (Bedd.) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Cookia dulcis Bedd. Madras Jour. Lit. Sci. 2 ser. 22:73. 1861. Clausena willdenowii var. dulcis Bedd. Fl. Sylv. 44. 1871.
      Type.—British India, Madras Presidency, Anaimalai Hills (Beddome).   Herb. Kew.
      Distribution.—India: Anaimalai Hills south to Travancore.
      Common name.—Indian wampee.
      Beddome (1861, p. 73) described his new species in general terms as follows: "A tree with a delicious fruit not uncommon on the Anamallays up to 3,000 feet both in the moist woods and in the drier forests—it flowers in April and the fruit begins to ripen at the end of June—the fruit is more grateful to the taste than that of the Whampee (Cookia punctata).   The tree is well known to the hill tribes and called 'Mor Koorangee.’   I have often met Kaders carrying home on their backs basket loads of this and the fruit of Pierardia sapida, which is also abundant in these jungles."   He also noted that the "fruits are globose, size of a large cherry."   Beddome in a later publication (1871, p. 44) stated: "This is a tree 30 feet high, with a trunk several feet in girth, bearing a very delicious fruit as large as a large cherry, as succulent as a grape, and somewhat of the flavour of the black current…it is abundant about Toonacodava (elevation 2,000-3,000 ft.), both in the moist and dry forests, and is in fruit from June till August; specimens in flower were sent to Prof. Oliver, who pronounces it only a form of [Clausena] Willdenowii [= C. dentata], so I will not attempt to separate it, but I believe it quite a distinct species…in comparison with true C. Willdenowii it has leaves that are more membranaceous, highly odoriferous, more prominently dotted, and very erose toward the apex."
      This variety is not as yet clearly distinguished from the species, and most taxonomists have not recognized it as valid.   It should be noted, however, that investigations by Rao and Subramaniam (1935) have shown that the two forms of C. willdenowii [= C. dentata] growing in southern peninsular India (Madras State) differ strikingly in the composition of the volatile oil in the leaves.   Oil from the leaves of one form was shown to consist of alpha-, beta-, and di-alpha-clausenan belonging to the furfural group, of which the major constituent was alpha-clausenan.   Oil from leaves from a different locality was found to contain not alpha-clausenan but gamma-clausenan, an isomeric substance.
      Abundant good herbarium specimens from Madras State, lent to Swingle with other material of Clausena by the Director of the Royal Botanic Gardens at Kew, show a decided variation in the shape of the ovary and in the relative length of the style.   A specimen collected by J. S. Gamble (No. 16420, July, 1885, the Chingleput district of Madras State) shows abundant flowers with a slender, often conical ovary, 1 to 1.2 mm long and 0.6 to 0.7 mm wide, merging into a cylindrical style distinctly longer, 1.2 to 1.8 mm (all measured dry).   Another specimen collected by Gamble (No. 16190, Barliyár, Nilgiris, at alt. 3,000 ft. [915 m], May, 1885) shows equally abundant flowers with a more or less cubical, 4-angled ovary about 1 by 1 by 1 mm with a prominent oil gland at the tip of each of the four locules.   The cylindrical style does not merge into the tip of the ovary but is inserted between the four prominent oil glands; it is only slightly longer than the ovary, 1.1 to 1.2 mm when dry.   Another specimen, Wight's No. 354 (collected at Courtallem [Kattalan], July, 1835), has more or less cubical, 4-angled ovaries with the style distinctly longer, about 1.5 mm when dry.
      Evidently careful research is needed to distinguish the forms of this and possibly other varieties from the species.   As var. dulcis is of economic importance, it is to be hoped that its taxonomic status can be settled soon.
      Other observers agree with Beddome concerning the delicious character of the fruit of this variety growing in the western part of Madras State.   Thomas A. Bourdillon (1908, pp. 67, 68), for more than thirty years Conservator of Forests in the former native state of Travancore, stated: "I agree with Beddome that the fruit, which very much resembles a grape, is very delicious, and is well worthy of attention."   He found it "common in the evergreen forests of Travancore at an altitude of 1,000-4,000 ft. [305-1,220 m] where the rainfall is light.   It is a small tree, 30 ft. [9.14 m] high and 9 in. [23 cm] in diameter, that is leafless in March, flowers in April and fruits in May and June."
      J. S. Gamble, in his Flora of the Presidency of Madras (1915, p. 155), referring to C. willdenowi [= C. dentata], probably in large part this var. dulcis, stated: "A small tree with glabrous or pubescent leaves, black bark and white, close-grained wood.   The fruit is good and worth cultivation!"
      6d.   Clausena dentata var. dunniana (Lév.) Swing. Jour. Wash. Acad. Sci. 30:82 1940. Clausena dunniana Lév. in Fedde, Repert Sp. Nov. 11:67. 1911.
      Type.—China, Kweichow Province, Pin-fa (Cavalerie, No. 1072).   Herb. Univ. Edinburgh.
      Distribution.—China: Kweichow Province.
      Leaves 5-11-foliolate; leaflets elliptical, bluntly pointed at apex, 5-10 X 2.5-3.5 cm, very blunt (almost truncate) at base, more or less unequal-sided, inconspicuously crenulate; petiolules unusually long (3-7 mm); inflorescences axillary, slender; flower buds globose, small, 2 mm diam.; petals glabrous; pistil 2.5 mm long, glabrous; gynophore 0.8 mm long, 0.6 mm wide; ovary tuberculate, 1 X 1 cm, with 1 or 2 small or medium-sized oil glands in a tubercule that tips each locule; style cylindric, 0.8-0.9 X 0.5 mm, narrowed at base; fruits subglobose or short-ovoid, 7-8 X 6-7 mm, black.
      Little is known about this shrubby variety that grows at altitudes of from 200 to 1,200 m.   It may prove to be a good species as yet inadequately described.   The oil glands in the apical tubercles of each locule of the ovary are somewhat like those of C. dentata, although usually only one medium-sized oil gland is found in the species.   For the present we are considering this Chinese plant as a variety of C. dentata, which also has axillary inflorescences.
      6e.   Clausena dentata var. robusta Tan. Jour. Bot. Brit. & For. 68:228. 1930.
      Type.—China, Yunnan Province, Szemao; altitude 1,372 m (Henry, No. 13028).   Herb. Kew.
      Distribution.—China: Yunnan Province.
      Tanaka's original description reads, in translation, as follows: "A vigorous plant, leaflets large, broadly oval, unequal-sided rather thick.   Inflorescences suberect, robust.   Berry globose, pustulate-punctate."
      Tanaka cited two numbers as types: Henry, Nos. 11914D and 13032, both in the herbarium at Kew.   The senior author could find only No. 11914D, which consists of the tip of a fruiting branch with a single leaf and 4 axillary inflorescences, 12-18 X 3.5-4.5 cm.   The leaflets are 8.5-9 X 3-3.8 cm and are acuminate; the berries are subglobose, 7-8 mm diam., borne on a twig 5-7 mm diam., with elongate heart-shaped leaf scars, 6 X 3 mm.   Another sheet in the Kew Herbarium is Henry's No. 13028, also from Szemao, with flowering twigs with 11 foliolate leaves, 32-37 X 12-14 cm; lateral leaflets 5.5-9 X 3.5-4.5 cm, terminal 10 X 5 cm; inflorescences 12-14 X 2.5-3 cm; flower buds globose or obovoid, 2.5 mm long, 2.5-3 mm wide; pistil glabrous, 3 mm long; gynophore 0.7-0.8 X 0.5 mm; ovary tuberculate, 0.7-0.8 mm wide; style cylindric, 1.5 X 0.6 mm, stigma depressed, slightly wider than the style.   The leaflets are widely spaced on the rachis (2-4 cm) like those of C. dentata var. dunniana, from which var. robusta is not adequately separated by the brief original description.
      6f.   Clausena dentata var. henryi Swing. Jour. Wash. Acad. Sci. 30:81. 1940.
      Type.—China, Hupeh Province, Ichang (Henry, No. 4122).   Cotype: Ichang (Henry, No. 3127).   Both in Herb. Kew.
      Distribution.—China: known only from western Hupeh and Yunnan (?) provinces.
      Differs from C. dentata var. robusta Tan., from Yunnan, in having hirsute petals, ovate black fruits, and more densely hirsute young leaves.   A small tree, 2-6 m high; leaves persistent, 25-35 cm long, 13-17-foliolate; leaflets lanceolate, acuminate at tip, blunt at base, which is often strong (and always somewhat) unequal-sided, 5-12 X 2-4.5 cm; largest leaflets often laterals, attached near upper third of rachis; rachis, midribs, veins, and whole surface of young leaflets pubescent, but leaflets glabrescent above; inflorescences axillary, sometimes short, 8-10 cm, often longer, 20-25 X 4-6 cm, pubescent when young but soon glabrescent; fruits black when ripe, ovoid, 9-12 X 6-8 mm.
      The fruits of var. henryi are considered edible in central China, according to Rehder and Wilson (1914, p. 140).   It has been thought by some botanists that this plant is sometimes cultivated in Hupeh Province.   It should be introduced into culture in other regions and hybridized with species that produce edible fruits.
      7.   Clausena harmandiana (Pierre) Guill. in Lecomte, Not. Syst. 1:219. 1910. Glycosmis harmandiana Pierre, Flore Forest. Cochinch. 4(18):text to pl. 285. 1893. Illus. Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:664, fig. 70(4-5). 1911.
      Type.—French Indo-China, Cambodia (Harmand, No. 3875).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Cambodia and Laos.   According to Tanaka (1937), it was collected once at Margui, Burma.
      The original description by Pierre reads, translated, as follows: "Branches rounded, pubescent.   Leaves imparipinnate, with 6-9 opposite or alternate leaflets, which are subdeltoid, ovate-acuminate, obtuse at both ends, 5-14 cm long, 5-8 cm wide, submembranous, coriaceous, with 7-8 pairs of small lateral veins.   Inflorescences terminal, pubescent, 10-20 cm long, much branched with rather large (5-6 mm diam.), sessile flowers.   Sepals 1 mm long, very coriaceous, pubescent on the outside and ciliate.   Petals 3.5-4 mm long, very concave, obovate, veinless, red spotted.   Stamens 7-10, 2.5 mm long, with the filaments flattened below and subulate above, longer than the anther, which is oval and without an apical gland.   Disk cupuliform.   Ovary 2.5 mm long, with 5 locules, each carrying on the outside 2 large glands.   Ovules 2 in each locule."
      The description of the disk by Pierre as cupulate is difficult to credit.   A photograph of the flowering type specimen (Harman, No. 3875) in the Paris herbarium, placed at Swingle's disposal by Tanaka, led him to think that Pierre mistook a ring of small bracts at the base of the very short pedicel (only about 1 mm long) for the calyx and the calyx itself for the disk!   The length of the flower from the base of the calyx to the stigma measures about 3 mm, the calyx flattened-rotate, about 2 to 2.5 mm wide, tapering rapidly into a very short pedicel about 1 mm long (including the bluntly conical base of the calyx); at the base of the pedicel are several small bracts and sometimes one or two small flower buds which may develop later, or more often abort and fall off.   These bracts and buds are borne on a slender terminal branchlet of the inflorescence 3.5 to 8 mm long.   The ovary is subglobose or broadly oval, about 1.2 to 1.5 mm long and 1 mm wide; the style is cylindrical, 0.4 to 0.5 mm long and 0.2 to 0.3 mm wide, not contracted at the base where it joins the ovary tip.   The gynophore is shorter than in most species of Clausena, but seems to be otherwise normal.
      7a.   Clausena harmandiana var. contracta Tan. Med. Rijks Herb. Leiden 69:7. 1931.
      Type.—Java (Horsfield).   Herb. Kew.
      Distribution.—Known only from Java.
      The type specimen of this variety has 5-7 leaflets varying in size, the lower ones 3-4 X 2-2.5 cm, the larger laterals 7-9 X 3-3.5 cm and the terminal one 10 X 4 cm, all nearly equal-sided, broad-elliptical, blunt at both ends, all parts pubescent, even the upper surface of the leaflets slightly so; inflorescences terminal, very short, 8 X 3 cm; flower buds globose, 2 mm diam., yellowish to reddish-brown when dry; petals without large oil glands at tips; gynophore 1 mm long; ovary 1 mm long, 1.5 mm wide, jet black, glabrous; style 0.5 mm long, glabrous, black.
      Another specimen in the herbarium at Kew, identified by Tanaka as this variety (coll. by Zollinger, second Java voyage, No. 2878), has more glabrous leaves; ovary jet black, very prominently tuberculate; gynophore subcylindric, abruptly expanding into the ovary base; style 0.6 mm long, 0.5 mm wide, stigma depressed-capitate, 0.7 mm wide.
      This variety is apparently known only from these two specimens and is not well understood.   Tanaka (1931b) stated that it resembles somewhat C. dentata var. robusta of China, but differs in having terminal, not axillary, short inflorescences and in having a shorter style and a thicker ovary.
      8.   Clausena macrophylla Hook. f., Fl. Brit. Ind. 1:504. 1875. Clausena heptaphylla var. (?) pubescens Oliv. Jour. Linn. Soc. Bot. 5(2):30. 1861; C. hirta Ridl., pro parte (?) 1920. Illus. fig. 3-4,G.
      Type.—Burma, Trogla, banks of Saluen River (Wallich, No. 6367).   Herb. Kew.
      Distribution.—Burma; Thailand: Phuket, Tasan; Malay Peninsula.
      Hooker (1875, vol. 1, p. 504) described this species, with some transpositions as follows: "Branchlets as thick as a goose quill and, as well as the petiole and panicle, clothed with a soft, spreading, close-set velvety tomentum.   Leaves 5-foliolate, 1 ft. [30 cm] long, leaflets 5-7 in. [12.5-17.5 cm] long, equal or suboblique with very short petiolules, acuminate, obscurely toothed, small, rather distant teeth, pale on both surfaces, tomentose beneath, nerves strongly arched, midrib stout, common petiole stout, terete.   Panicle stout, erect, with the peduncle 1 ft. [30 cm] long; branches strict, erecto-patent, secondary branches short.   Flowers glabrous, crowded, 1/10 in. [2.5 mm] diam., shortly pediceled, buds globose.   Calyx lobes 4, minute, rounded-acute.   Petals broadly oblong, concave, membranous, covered with large glands.   Stamens equal; filaments much dilated below, very short; anthers large, oblong.   Ovary stipitate, glabrous, 4-lobed, tuberculed with very large glands; ovules 2, collateral in each cell; style as long as the ovary and as broad as the stigma.   Fruit immature, oblong, tip obscurely 4-angled and truncate."
      Ridley mentioned in his description of C. hirta that the leaflets are "thickly covered with translucent glands and dotted over with larger, sparser dark brown ones."   This description recalls the dimorphic oil glands of C. guillauminii and makes it very desirable to reexamine all material of C. hirta to see if any confusion has arisen with C. guillauminii, a species which also occurs in Thailand.
      9.   Clausena anisum-olens (Bl.) Merr. Phil. Gov. Lab. Bur. Bul. 17:21. 1904. Cookia anisum-olens Blanco, Fl. Filip. 359. 1837; Clausena warburgii Perk. Frag. Fl. Phil. 162. 1905; C. loheri Merr. Phil. Jour. Sci. 27:27. 1925.
      Type.—Lacking.   Substitute type: Rizal Province, Luzon, P.I. (Merrill, Species Blancoanae, No. 1012).   Herb. Bur. Sci., Manila.
      Distribution.—Philippine Islands: Luzon, Masbate, Basilan, Mindanao islands.
      Perkins (l.c.) has drawn up the only full description of this species yet published.   It reads, translated, as follows: "Shrub or tree, with glabrous…twigs 5 mm thick; leaves 15-30 cm long, 2-paired…very sparsely gray-pilose; leaflets ovate, or obovate-oblong or ovate-lanceolate, 6.5-10 cm long, 2.5-3 cm wide, apex long, narrowly acuminate, the apex itself obtuse or often subincised, base oblique, papyraceous, glabrous, veins and veinlets sparsely pilose…lateral veins 8-12 [pairs], veins and veinlets indistinct above, prominent below.   Inflorescences terminal, paniculate, 18-22 cm long, branchlets numerous, sparsely pilose, spreading, bractlets ovate, sparsely pilose; flowers 8 mm diam., with pedicels 5-8 mm long; calyx 5-merous with short lobes…ovate-acuminate, very sparsely pilose without; 1 mm long; petals with lobes…5 mm long, 2 mm wide, ovate, membranaceous, glabrous on both sides; stamens 9, as long as the petals, filaments jointed (geniculate) near the middle, the joint pubescent; disk small, smooth, ovary stipitate, glabrous, 5-locular, 1 mm diam.; style as long as the ovary, stigma sessile, 5-lobed."
      Serial microtome sections of the pistil of a specimen in the Bureau of Science Herbarium at Manila (coll. by E. D. Merrill, No. 2509, Bur. of Agric., at Lamao River) show the following characters: Pistils 5-locular, each locule ending in a short process, bearing at the end a rather large oil gland; ovary broad-oval, about 1.5 mm long, 1-1.2 mm wide; style strongly contracted at base, which is inserted into a cavity at the tip of the ovary, the rest of the style unusually broad, nearly 0.8-1 mm wide; the gynophore is long and not abruptly narrowed, about 1-1.2 mm long, 0.7-0.8 mm wide at the narrowest place, and about 1 mm at the base and the top.
      This species is confined to the main islands of the Philippines.   It has geniculate staminal filaments, bent at an angle near the middle, slender above the bend and broader below, resembling somewhat the filaments of C. excavata, but it is much less bent and much less swollen below the bend.
      It is related to C. laxiflora of Mindanao Island, Philippines, but shows important differences in the characters of the pistil and the inflorescences.
      10.   Clausena mollis Merr. Phil. Jour. Sci. Bot. 5:181. 1910.
      Type.—Philippines, Luzon Island, near Bontoc, at 915-1,220 m attitude (Curran, For. Bur., No. 16530).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: known only from Luzon Island.
      Merrill's original short diagnosis, translated and supplemented with data from the longer description (in brackets), reads as follows: "A shrub about 5 m high, all parts densely and softly pubescent; leaves 20-30 cm long; leaflets alternate or subalternate, 5-8 on each side, margins entire, base strongly inequilateral, apex shortly acuminate or…blunt, [lateral veins about 8 on each side of the midrib].   Panicles subterminal, narrowly pyramidal, flowers 5-merous, sessile or subsessile, subglomerate [on the ultimate branches]…Sepals broadly ovate, 1.2 mm long, free, densely pubescent outside.   Petals elliptic or broadly elliptic-oblong, concave, about 4 mm long, 2.5 mm wide, acute at both ends, imbricate, with few, rather large glands, the back in the upper third slightly pubescent.   Stamens 10, the filaments broad, abruptly narrowed just below the insertion of the anthers, about 1 mm long, anthers 2 mm long.   Ovary ovoid or ellipsoid, about 1.5 mm long, glabrous, prominently 5-sulcate, 5-celled, each cell with two superimposed ovules; style 1 mm long and anthers 2 mm long.   Ovary ovoid or ellipsoid, about 1.5 mm long, glabrous, prominently 5-sulcate, 5-celled, each cell with two superimposed ovules; style 1 mm long and thick [5-sulcate].   Fruits globose, white or greenish-white, 1-seeded.   [Seeds about 5 mm diam., surrounded by a gelatinous pulp, with a strong odor and taste of pine pitch.]"
      This species is doubtless related to C. anisum-olens but differs from it, not only in being densely soft-woolly all over (except the petals and the flower parts they enclose), but also in having pistils with a rather short gynophore (somewhat like that of C. laxiflora).   The filaments of the stamens are abruptly expanded from the middle down, and somewhat geniculate where the slender apical portion joins the expanded base.   Each locule of the ovary is capped by a single large oil gland.   The flowering twigs of this species are often unusually thick; a specimen collected by Ramos and Edaño (Bur. Sci., No. 7876), at Mount Masapilid, Bontoc Province, Luzon Island, in March, 1920, now in the herbarium of the Arnold Arboretum, has twigs 5 to 7 mm in diameter, with reddish-brown pith 2.5 to 3 mm in diameter.
      11.   Clausena laxiflora Quis. & Merr. Phil. Jour. Sci. 37:154. 1928. Illus. fig. 3-4,M.
      Type.—Philippines, Mindanao Island, Davao Province, Mati (Ramos and Edaño, Bur. Sci., No. 48983).   Herb. Univ. Calif., Berkeley.
      Distribution.—Known only from the type locality.
      Merrill's original diagnosis, translated with additions from his longer English description (enclosed in brackets), reads as follows: "A shrub 3-4 m high, scarcely aromatic, glabrous, or nearly glabrous.   Leaves [26-] 39 cm long, Leaflets 7-9, membranaceous, up to 16 cm long and 6.5 cm wide, ovate, elliptic, strongly inequilateral [at the base], obtusely [acuminate].   Inflorescences lax, 30-35 cm long, slightly pubescent.   Flowers 5-merous, about 5 mm diam.   [Calyx shallow, 1.5 mm in diameter, 5-lobed, the lobes broadly ovate, acute, glabrous, margins sparsely and minutely puberulent.   Petals 5, imbricate, oblong, obtuse, 3.75 to 4 mm long, 2 to 2.5 mm wide, glandular, glabrous.   Stamens 10, about 3 mm long; anthers oblong, apiculate, about 1.5 mm long; filaments somewhat enlarged below.   Ovary globose, glabrous.]"
      This species, so far found only on Mindanao Island, P.I., is obviously related to C. anisum-olens, a species distributed in the Philippines from Luzon Island to Mindanao Island.   However, C. laxiflora, besides having larger flower clusters, with longer, more slender branches and much larger leaflets than C. anisum-olens, shows important differences from the species in the structure of the pistil, as follows: (1) in having the processes at the tips of the five ovary locules without large oil glands; (2) in having a much shorter gynophore, which shows a very short basal annular portion only about 0.2 mm high and 1 to 1.1 mm wide just below the constriction; (3) in having locular cavities which descend deeply into the gynophore, reducing its length to 0.6 to 0.8 mm (instead of 1 to 1.2 mm, as in C. anisum-olens); and (4) in having a large subglobose ovary, nearly filling the flower bud, which at this stage has almost no distinct gynophore.   Tanaka (1932e, p. 422) reduced this species to a synonym of C. anisum-olens, but, in view of the striking differences in the ovary and gynophore characters noted above, it is evident that C. laxiflora should be retained as a good species.
      12.   Clausena todayensis Elmer, Leafl. Phil. Bot. 8:2805. 1915.
      Type.—Philippines, Mindanao Island, Davao, Todaya (Mt. Apo) (Elmer, No. 10530).   Herb. Bur. Sci., Manila.
      Distribution.—Known only from the type, collected on a densely wooded ridge south of the Sibulan River at 1,500 ft. [457 m] altitude.
      Elmer's original description reads, in part, as follows: "Tree, with a 10 m high and 3 dm thick stem…twigs roughened with lenticels, puberulent when young.   Leaves imparipinnate, thinly coriaceous, paler green beneath, dark green on the upper conduplicate surface, the acute to acuminate tips strongly recurved…5 to 7-foliolate…leaflets punctate beneath, slightly unsymmetric, oblongish…obscurely crenate or undulate, the base obtusely rounded, the terminal leaflet the larger, the basal pair subopposite and much reduced, the larger ones 7.5 cm long by 3 cm wide at the middle, sparsely pulverulent or puberulent on both sides; lateral pairs of veins 5 to 9, divaricat…petiolules only a few mm long, the terminal leaflet upon a 1 to 1.5 cm long rachis extension.   Inflorescence ascending, terminal, paniculate, 10 to 20 cm long; all the stalks terete, green, brown puberulent…the ultimate branchlets subtended by minute bracts; flowers solitary or usually in small terminal clusters, subtended by a bract…finely ciliate along the margins; pedicel 1 mm long, relatively thick; sepals 4, cremeus, caducous, elliptically oblong, 3.5 mm long by one-half as wide across the middle…glabrous…stamens 8, erect; filaments 2 mm long, fleshy, compressed, purplish-brown sprinkled, pointed at the apex only…anther 1.5 mm long…emarginate at the apex…ovary more or less rugose, 2 mm long, short stipitate, obscurely 4-angled and with some large glands, terminated by a circular stigma."
      Tanaka (1932e, p. 423) did not follow Merrill (1923, p. 337) in reducing this to a synonym of C. anisum-olens because, as he stated, C. todayensis has a 4-merous "small ovary, with only 8 large oil-glands somewhat like that of C. harmandiana.   The plant is, however, very gracile and not like the above [C. harmandiana]."   Pending further study, the species is here retained.
      13.   Clausena cambodiana (Pierre) Guill. in Lecomte, Not. Syst. 1:219. 1910. Glycosmis cambodiana Pierre, Flore Forest. Cochinch. 4(18):text to pl. 285. 1893.
      Type.—French Indo-China, Cambodia, Knang-Krêpeuh Mountains at 1,500 m altitude (Pierre, No. 832).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Known only from the type locality.
      The description given by Guillaumin (1911, p. 665) reads, translated, as-follows: "A small tree 5-8 m tall.   Branches glabrescent.   Leaves 15-20 cm long, with 5-11 alternate, glabrous, membranaceous leaflets, equilateral at the base, oval or oblong-lanceolate, 9-15 X 3-5 cm, cuneiform at the base, cuspidate at the apex, with 6-9 pairs of lateral veins, which like the veinlets are visible above and are raised below; petioles cylindrical, pubescent; petiolules cylindrical, about 5 mm long.   Fruiting inflorescence a terminal panicle, as long as the leaves, pyramidal, pubescent.   Fruit spherical, red-violet, about 8 mm diam., punctate because of the oil glands, very fragrant."
      In the original description Pierre mentioned that the lower leaflets are smaller and that the fruiting inflorescences are 20 cm long.
      A photograph of the type specimen (Pierre, No. 832) preserved in the Paris herbarium, made by Tanaka, shows the leaves taper to a very sharp tip 0.9 to 2 mm wide, and are very flat and entire-margined.   The fruits are borne on slender pedicels about 4 to 6 mm long, arising from nearly straight, rather stiff lateral branches of the inflorescence, 7 to 8 cm long below but much shorter above.   Most of the other species of Clausena having caudate leaflets have the margins of the leaflets minutely dentate or crenulate (as in C. dentata) or else have curved tips and very unequal-sided bases (as in C. wallichii).
      14.   Clausena brevistyla Oliv. Jour. Linn. Soc. Bot. 5(2):31. 1861.
      Type.—Australia, Queensland, Hope Islets (MacGillivray, July 18, 1848).   Herb. Kew.
      Distribution.—Australia: Queensland.
      Oliver's original description reads: "Leaves 10-15-foliolate, leaflets very oblique; ovary more or less obovate, narrowed at the base, 4-lobed above, glabrous or almost glabrous; style distinct from the ovary, very short, half shorter than the ovary, equal to the stigma [in diam.] Shrub (?).   Leaves 5-8 in. [12.5-20 cm] long.   Petiolules about 2 lin. [4 mm] long.   Leaflets oblique or rhomboid-oval or ovate, apex scarcely acuminate, margins undulate-crenate or dentate, glabrous or in young leaves pubescent near the midrib, 1-4 1/2 in. [2.5-11.5 cm] long, 10 lin. to 1 1/2 in. [20-38 mm] wide.   Calyx 4-5 parted, lobes widely ovate.   Petals 4-5, thin, broadly elliptical or rounded.   Stamens often 10, filaments thicker below and dilated, somewhat arched.   Ovary glabrous, or with few hairs, 4-5-locular.   Ovules 2 (superimposed?).   Stigma equal to the style [in diam.].   Fruit not seen.   Resembles C. heptaphylla very closely, differing in the conspicuous obliquity of the leaflets and the very short style, inserted in both species in the central apical depression of the ovary.   The flowers are often pentamerous."
      This species is the only one native to Australia and thus occupies the extreme southeastern portion of the range of the genus Clausena.
      14a.   Clausena brevistyla var. halmaheirae (Miq.) Swing. Jour. Wash. Acad. Sci. 30:82. 1940. Clausena halmaheirae Miq. Ann. Mus. Bot. Lugd.-Bat. 1:211. 1863.
      Type.—East Indian Archipelago; Moluccas, Halmahera Island (Teijsmann and De Vriese), Rijks Herb., Leiden.
      Distribution.—Known only from the original collection.
      The original description by Miquel reads, translated, as follows: "Branches, petioles, and midribs (on lower side), along with the inflorescences, covered sparingly with a thin adpressed pubescence; leaves long, with rather widely separated leaflets; leaflets (on the stronger leaves up to 11, or fewer) with short petiolules, lower ones nearly opposite, broadly obovate and shortened, the others not opposite, narrowly sublanceolate-oblong, acutely pointed toward the base, one side cut off, the other obtusely unequal-sided, chartaceous with even or slightly repand margins; sometimes very minutely subserrate, pale, black glandular-punctate below, with 10-15 spreading veins; fruiting inflorescences axillary, 1/2 longer than the leaves, with lax few-flowered branchlets; calyx 4-lobed; immature berries, 1-seeded.   Near to the species which I listed in Fl. Ind. bat., Suppl. I, p. 501, doubtfully under the name C. excavata.   Sufficiently distinct from all described species, but recalling several of them.   Leaves 1-1.5 ft. [30-45 cm] long.   Lower leaflets 3 in. [7.5 cm] long, upper ones 6-6.5 in. [15-17 cm] long."
      Tanaka (1931b, p. 6) who examined the type specimen at Leiden and other type material at Utrecht and Kew, concluded that this plant "is closely related to C. harmandiana, but is different in having narrower leaflets with distinct black dotting and a long branched panicle bearing [a] decidedly smaller berry."   With respect to the last-named character, Miquel noted that the type material (all that is known as yet) has immature fruits.
      This variety is characterized by having large fruiting inflorescences with widely spreading primary branches and sturdy terminal pedicels 2 to 4 mm long.   It has very small pistils, only 1.4 to 1.5 mm long (in dry state), with a very slender gynophore, only 0.2 to 0.3 mm in diameter, and an ovary only 0.8 mm long and 0.6 mm wide, rough with oil glands, and a cylindrical style, shorter than the ovary.   The leaflets are only slightly unequal-sided.   This variety is very distinct from C. harmandiana.
      15.   Clausena anisata (Willd.) Hook. f. Niger Fl. 1:256. 1849. Amyris anisata Willd. Sp. Pl. 2:337. 1800. Illus. De Wildeman, Ann. Mus. Congo, Bot 5 sér. 2: pls. 52, 53. 1907.
      Type.—Western Africa, "Guinea" (coll. ?).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Western Africa: Sierra Leone to Angola; central Africa: Bahr-el-Ghazal and Pemba through Usambara to northern Malawi and Tanzania on Masai plateau, reaching 3,000 m altitude.
      This Species was described by Oliver (1861, p. 34) from Nigerian material as follows: "Leaflets 10-20, panicles lax, elongate, shorter than the leaves or longer, lateral branches divaricate in a cymose manner, slender, pubescent; ovary 4-furrowed, 4-loculed; ovules paired [in each locule], obliquely superimposed or almost collateral.…Small shrub 3 ft. [92 cm] high with white flowers.   Leaves 6-12 in. [15-30 cm] long; leaflets unequal-sided, obliquely ovate or ovate-oblong, obscurely crenulate, sometimes slightly acuminate or emarginate, pubescent.   Inflorescences paniculate, 6-9 in. [15-23 cm] long, borne in the axils of the upper leaves.   Pedicels 2/3-1 lin. [1.2-2 mm] long.   Bractioles minute, ovate or lanceolate.   Calyx 4-merous, lobes ovate.   Stamens with filaments abruptly dilated in the middle, subulate above.   Ovary short-stipitate, glabrous.   Style rather long, rather thick, 4-furrowed, separating at the base."
      This species is the only one found in western Africa, but in eastern Africa it shares the central part of the range of the East African C. abyssinica.   All three African species vary greatly and have ill-defined varieties.
      Engler (1931, p. 322) reported that in western Africa C. anisata is found in the scattered bush forests of the hinterland of Cameroons (now Cameroon), where it is a tree 15 to 20 m high at an altitude of 2,100 m.   Engler found that in eastern Africa it reaches the very considerable altitude of 3,000 m (9,840 feet), where it grows in the mountain meadows as a beautiful tree of medium height with a broad crown.   This is by far the highest altitude yet reported for any member of the orange subfamily, but since Mount Kilimanjaro is situated only 30° 37' south of the equator, even at high altitudes no severe cold in winter is experienced there.
      Clausena anisata and C. inaequalis were separated by Engler (1931, p. 322) by the length of the inflorescences in relation to the length of the leaves: in C. anisata the flower clusters are much longer than half the length of the leaves; in C. inaequalis they are never more than half the length of the leaves.   De Wildeman and Durand (1901, p. 743) found the leaves of C. anisata to be 15 to 40 cm long, those of C. inaequalis to be 7.5 to 25 cm long; and the flowers of C. anisata to occur in many-flowered paniculate cymes, but those of C. inaequalis in three-flowered clusters or singly along the racemes.   These two species have similar, more or less obovate or pyriform ovaries, with a large gland at the tip of each locule.
      15a.   Clausena anisata var. multijuga Welw. ex Hiern. Cat. Welw. Afr. Pl. 1:116. 1896.
      Type.—Africa, Angola, Golungo Alto (Welwitsch, No. 1320, in part; coll., Nov. 1855).   Herb. Brit. Mus., London.
      Distribution.—Angola: reported only from Golungo Alta and Pungo Andongo, in elevated stations up to 2,200 ft. [670 m].
      Differs from the species chiefly in having leaves with more numerous leaflets (17-32).   A shrub or "small tree 7-12 ft. [2-3.6 m] high, with slender, far-reaching, almost sarmentose branches, with a very short, closely-set pubescence."   Leaves 3-10 in. [7.5-25 cm] long; leaflets 1/2-2 X 1/6-5/8 in. [1.2-5 X 0.4-1.5 cm]; inflorescences 4-9 in. [10-22.5 cm] long.
      It is doubtful whether this small-leaved multifoliolate form deserves taxonomic recognition.   No mention is made of tubercules on the ovary such as are found in a form growing a few hundred kilometers to the north in the Congo Valley; presumably the ovary in var. multijuga is smooth.
      This Angola form is said to make a small, elegant tree.   It should have value as an ornamental tree for cool subtropical regions.   The leaves are "fragrant when rubbed, emitting a very agreeable aroma like that of oranges and lemons."   The flowers are white but small.
      15b.   Clausena anisata var. mollis Engl. Die Pflanzenwelt Ost-Afrikas C:228. 1895; also in Die. Nat. Pflanzenfam. ed. 2. 19a:322. 1931.
      Type.—Africa (locality doubtful).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Africa: forest and bush steppes of Cameroon hinterland; the Lake Region; Tanzania, Mount Kilimanjaro, western Usambara, coastal region; northern Malawi.
      Leaves soft-pubescent and with blunter leaflets than the species.
      An inadequately known plant that needs study.
      16.   Clausena inaequalis (DC.) Benth. in Hook., Niger Fl. 1:257. 1849. Elaphrium inaequalis DC. Prodr. 1:724. 1824; Amyrisinaequalis Spreng. 1825; Fagarastrum inaequalis Don. 1832; Myaris inaequalis Presl. 1840. Illus. Wood & Evans, Natal Pl. 1(2):54, pl. 66. 1899; Sim, For. Fl. Cape Col. pl. 26. 1907.
      Type.—South Africa (coll. ?).   Herb. De Candolle, Geneva.
      Distribution.—Southeastern Africa: Cape Colony, Pondoland, Natal, with transition forms to C. abyssinica (fide Engler, 1931, p. 322).
      This species was described by Wood and Evans (1899, p. 54) as follows: "A shrub or small tree with dark coloured bark, which is usually thickly studded with lenticels.   Leaves unequally pinnate, 6-8 in. [15-20 cm] long; leaflets alternate or subopposite, in 4-6 pairs, petiolulate, glabrous, lateral ones very unequal-sided, the lower portion of the lamina being much smaller than the upper portion, terminal one subequal, broadly lanceolate, margin unequally crenate, attenuate to an obtuse or subacute apex; 1 1/2-2 in. [3.8-5 cm] long; 1/2-3/4 in. [13-18 mm] broad, petiolules 2-3 lin. [4-6 mm] long, pubescent, terminal one 6-9 lin. [12-18 mm] long, swollen and bent below the lamina.   Flowers paniculate, white.   Calyx small, 4-parted, sepals lanceolate, pubescent.   Petals 4, concave, free, spreading, imbricate in bud, deciduous.   Stamens 8, hypogynous, filaments free, subulate, flattened.   Anthers sagittate.   Ovary on a short cylindrical fleshy torus, obtusely 3-lobed, 3-celled, ovules 2 in each cell, collateral.   Style short, thick, deciduous.   Stigma 3-lobed.   Berry fleshy, dark purple, the epicarp thickly studded with minute glands; 1-2 seeded by abortion."
      Clausena inaequalis varies greatly, as do the other African species, C. anisata and C. abyssinica.   The first two differ in the length of the leaves, the obliquity of the leaflets, the length of the inflorescence compared to the length of the leaves, and the number of flowers arising in the ultimate division of the inflorescence; C. abyssinica is discussed below.
      17.   Clausena abyssinica Engl. Die Pflanzenwelt Afrikas 3:757. 1915. Clausena inaequalis var. abyssinica Engl. Die Pflanzenwelt Ost-Afrikas C:229. 1895. Illus. fig. 3-4,H.
      Type.—Abyssinia, plateau up to 2,300 m altitude (coll. ?).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Eastern Africa: Ethiopia; Tanzania, Mount Kilimanjaro (up to 2,800 m alt.); southern Malawi; Zambia.
      The brief original description of the variety reads, translated, as follows: "leaflets less unequal-sided, more acuminate [than in the species C. inaequalis].—3 [Abyssinian plateau] (2,300 m).—15 [Mount Kilimanjaro, Tanzania] (2,700 m).—A tree up to 10 m high in mountain forests."   The laconic description of the species in Engler's account of Clausena (1931, p. 322) characterizes C. inaequalis as having "very unequal-sided, rhombic-ovate, mostly blunt pointed" leaflets, and C. abyssinica as having "usually oblique-lanceolate leaflets and larger flowers."
      Clausena abyssinica was reported by Engler (1931, p. 322) as growing up to 2,300 m in the Abyssinian plateau and up to 2,800 m in the upper girdle-forests of Mount Kilimanjaro in Tanganyika (now Tanzania).   This altitude is exceeded only by C. anisata, which on Mount Kilimanjaro attains 3,000 m.   These two species hold the altitude record for the orange subfamily, and they must be able to grow at lower mean and maximum temperatures than any other Citrus relatives.
      A critical comparative study of C. inaequalis and C. abyssinica is urgently needed.   The original description of C. abyssinica as a species contradicts the original description of it as a variety in stating that the leaflets are more oblique than those of C. inaequalis instead of less oblique!   As a matter of fact, all three African species of Clausena (nos. 15 to 17 above) and their varieties are exceedingly variable and as yet inadequately described.
      Engler (1915, p. 757) noted the difficulty of separating the African species and varieties of Clausena in an eloquent sentence to the effect that it is as hard to separate the species as it is easy to recognize the genus, which is spread over Africa from Ethiopia to Cape Province and from upper Guinea to Angola.
      18.   Clausena papuana Lauterb. Bot. Jahrb. 55:259. 1918.
      Type.—Northeastern New Guinea, at the foot of the Bismarck Mountains (Schlechter, No. 18476).   Herb. Bot. Mus., Berlin-Dahlem; fragments in the Lanterbach Herbarium in the Herb. Bot. Inst., Univ. Breslau.
      Distribution.—Known only from the original collections made by Schlechter.
      Lauterbach's original description of this species reads, translated, as follows: "Small tree with rounded twigs, the young shoots tomentose.   Leaves alternate, imparipinnate, rachis tomentose.   Leaflets alternate…petiolulate, oblique-ovate, obtusely subacuminate, the upper half rounded at the base, the lower half acute at the base, membranaceous, punctate, showing veins on both sides in the dried material, margins inconspicuously crenulate, lateral veins (8 pairs) oblique, curving and fusing together near the margins; midrib slightly elevated below.   Panicles axillary, subterminal, as long as the leaves, main stems tomentose, branches more or less separated, ending in aggregated apical branchlets, flowers short-pedicellate, in heads (‘capitulate’).   Sepals 4, connate at the base, acutely triangular, tomentose without; petals 4, elliptic-acute, glandulose, glabrous, reflexed in anthesis; stamens 8, nearly equal in length, filaments thickened at the base, anthers elliptical, emarginate; gynophore glandulose; ovary glabrous, cylindric, somewhat quadrangular, locules 4, each with 2 ovules; style short, stigma dilated.   A small tree with twigs 4 mm diam., having gray-brown bark.   Leaves 30 cm long; rachis 3 mm diam. at lower end; petiolules 4-5 mm long.   Leaflets 5-11 cm long, 3-5 cm broad, smallest at the base of the leaf.   Inflorescence 30 cm long, stem 7 cm, branches 2-3 cm, and flower stalks 1-2 mm.   Flower buds 2.5 mm diam.; calyx lobes 0.5 mm long, petals 4 X 2 mm.   The tip of each petal bears a large oil gland which, in the bud seen from above, forms a cross with those of the other petals.   Stamens 4 mm long, of which the anther measures 1.5 mm.   Gynophore 0.5 mm high, ovary 1 mm high, the style with the stigma 1 mm long."
      Lauterbach noted at the end of the original description that "this species approaches [C.] brevistyla Oliv. from Queensland.   It differs from it in its hairiness, larger leaves with broader, blunter leaflets and much larger flower clusters arising in the axils of the uppermost leaves."
      Tanaka (1928c, p. 141) reduced this species to a variety of C. harmandiana (see p. 177), but, as he pointed out, this latter species has fewer leaflets (5 to 7 instead of 10 to 11, or more) and a less hairy rachis.   Swingle found the leaflets of C. harmandiana much less unequal-sided at the base than those of C. papuana.   The flowers of C. harmandiana have remarkably short pedicels, only about 1 mm long, whereas the pedicels of C. papuana are usually about 2 mm long.   Clausena harmandiana has a very short, inconspicuous gynophore (Pierre apparently did not see it!), probably only 1/3 to 1/2 mm long, whereas the gynophore of C. papuana is better developed, about 1/2 to 3/4 mm long, and is easily seen.   Serial microtome sections made from the type material (Schlechter, No. 18476) preserved in the Lauterbach Herbarium and kindly lent Swingle by the Botanical Institute of the University of Breslau show that each locule of the ovary is tipped by a single large oil gland (not two as in C. harmandiana); each anther, each petal, and each sepal also has a large oil gland at the tip.
      19.   Clausena guillauminii Tan. Bul. Mus. Hist. Nat. Paris, 2 sér. 2:161. 1930. Illus. fig. 3-4,L.
      Type.—Thailand, Udawn, Lôti, Wang Sapung (Kerr, No. 8782).   Herb. Kew.
      Distribution.—Thailand, Laos.
      Tanaka's original description reads, in translation, as follows: "Shrub; branchlets and leaf veins pellucid-dotted and puberulous.   Leaves 1-5-foliolate, rachis narrowly winged; leaflets alternate, thick, obovate or oblong, base acute or rather obtuse, apex cuspidate, obtuse or rather acute, lateral leaflets smaller, nearly equal-sided, terminal leaflet sometimes much wider, all sparsely black-dotted, conspicuously reticulate, slightly crenate-serrate.   Inflorescences terminal, paniculate, lateral peduncles short, about equal in length.   Flowers small, 5-merous, petals and calyx lobes pellucid-spotted at the tips.   Filaments subulate, anthers glandulose-tipped.   Ovary ciliate, with 2 series of glandulose spots, short-stipitate; style long, stigma subcapitate."
      The type and cotype material collected in Thailand by A. F. G. Kerr (his Nos. 8782 and 8388, listed by Tanaka), which was lent to Swingle by the Director of the Royal Botanic Gardens at Kew, shows the following dimensions: Compound leaves (No. 8782), 5-7-foliolate, 17-20 X 9-13 cm, rachis 9.5-11 cm long, 1-1.5 mm diam., narrowly margined (0.2-0.5 mm) between the leaflets; terminal leaflets 7.5-9.5 X 4.3-5.7 cm; lateral veins 9-12 pairs, nearly straight; lateral leaflets 3-8 X 2-4.2 cm.   Unifoliolate leaves (No. 8388), 7.5-10.5 cm long, usually long-elliptical, blunt or subacute at both ends, petioles 1-3.5 cm long, very narrowly margined; inflorescences 3-10 cm long, but only 1-2.5 cm wide.   This plant has stems only 20-30 cm high and 3.5 mm diam., springing from a root 1 cm diam., yet they were flowering freely and were bearing a few fruits when collected February 3, 1924.
      Craib (1926, p. 233) stated concerning this material: "The glands on the leaves are by far the largest of any Siamese rutaceous plant seen.   The sepals and petals all have one prominent gland much larger than the others and the ovary has five conspicuous glands at the apex and usually a few longish spreading hairs."
      Serial microtome sections of a specimen collected by Kerr (No. 8782) at Me Chan, Chien Sen, Thailand, show that there are smaller irregularly placed oil glands around the side of the ovary about at the point where the gynophore ends; there are 7, 8, or more of them, and they, along with the much larger oil glands that are borne in processes at the tip of each locule, must be what Tanaka had in mind in describing the ovary as "marked with two series of glands."
      This species is said to be a shrub growing only 20 to 40 cm tall by Thorel, who collected it in Laos.   Possibly it is killed to the ground by annual fires and springs up each year from the root.   The oil glands are dimorphic, most of them being small, but a few, scattered 3 to 4 mm apart, being much larger, l/3 to 1/2 mm in diameter, pale on young leaves but dark brown or nearly black on old leaves.   By transmitted light, such glands are bright, clear red in color, about Jasper Red of Ridgway (1912, pl. 13).   These large oil glands are often conspicuously convex and stand above the leaf surface; they are sometimes surrounded by many small veinlets that radiate out from the oil gland.
      This astonishing species merits careful study as it differs widely from almost all the others of the genus.   Two other species, C. papuana and C. wallichii, have the same distribution of oil glands on the floral organs, viz., a single, rather large oil gland on the tip of each calyx lobe, on each petal apex and another [sic] tip, as well as a short-stalked oil gland at the top of each locule of the ovary.   C. papuana has very different leaves with about twice as many leaflets and a wingless rachis, whereas C. wallichii has acuminate, crenulate or denticulate leaflets about twice as large.
      20.   Clausena wallichii Oliv. Jour. Linn. Soc. Bot. 5(2):35. 1861. Illus. Swing. Jour. Wash. Acad. Sci. 30:81, fig. 2. 1940.
      Type.—Burma, Chapedong Hill (Wallich, No. 6370).   Herb. Kew.
      Distribution.—Burma: eastern Himalayan region, Tenasserim; Thailand.
      The original description by Oliver (1861, p. 35) reads, translated, as follows: "Leaves with the rachis narrowly winged; leaflets rhomboid-lanceolate, or oblong, acuminate, glabrous, margins crenulate.   Inflorescences paniculate, many-flowered, branches close together, flowers few.   Calyx 5-merous, lobes ovate.   Petals 5, margins slightly imbricate in the bud, thin.   Stamens 10, free, filaments thickened at the middle, at first slightly arched below.   Ovary short and narrowly stalked, 5- (or 4-) locular, ovules obliquely superimposed.   Style very short, furrowed, equaling the stigma in width."   The brief diagnosis adds the following characters: "Leaves 13-17 foliolate…Style distinct from the ovary, shorter or almost as long…"
      Hooker (1875, vol. 1, p. 505) added a few characters as follows: "Leaves 8-12 in. [20-30 cm] long; leaflets 2-3 in. [5-7.5 cm] long, alternate, oblique, caudate-acuminate, with an obtuse notched tip, narrowed into a cuneate base.   Panicle 3-5 in. [7.5-12.5 cm] long, and broad, much branched…Flowers subsessile, 1/8 in. [3 mm] diam.…ovary glabrous."
      An examination of the type specimen and much other material of this species lent to Swingle from the Kew Herbarium and from the Calcutta Botanic Garden enabled him to add several characters, as follows: Leaves 20-30 X 7-12 cm; leaflets 6-10 X 2-4 cm, crenulate or denticulate, very unequilateral, turning brown on drying, with numerous evenly distributed oil glands varying greatly in size (about 0.04-0.2 mm diam.); flowers 4.5-5 mm long from base of calyx to tip of stigma, flower parts all heavily impregnated with tannin residue, glabrous, except ovary; calyx lobes, petals, anthers, and 5 (or 4) locules of the ovary each tipped with a single large oil gland; ovary tuberculate, often bearing a few rather long hairs, about 1.8-2 X 0.8-1.2 mm; style 1.8-1.9 X 0.4-0.5 mm; gynophore 0.7-0.8 mm long, 0.5-0.8 mm diam. at narrowest part; pedicel of flower 1-2 mm long with minute ciliate bracts at base; fruits subglobose or round-ovoid, 5-6 X 5 mm (immature?), borne on pedicels 2-3 mm long and 0.8-1 mm thick.
      This species has the rachis very narrowly winged (wings 0.3 to 1 mm wide), a peculiarity it shares with C. guillauminii, which usually has the rachis merely margined rather than winged, whereas in C. luxurians the rachis is plainly winged (wings 1 to 2.5 mm wide).   The floral organs, calyx lobes, petals, anthers, and ovule locules, are each tipped with a single large oil gland, in this character resembling C. guillauminii and C. papuana.   The great variation in the size of the oil glands of the leaflets resembles but does not equal that found in C. guillauminii.   There can be little doubt that C. wallichii, C. luxurians, and C. guillauminii are rather closely related species, and all of them may be distantly related to C. papuana.
      21.   Clausena luxurians (Kurz.) Swing. Jour. Wash. Acad. Sci. 30:79. 1940. Clausena wallichii var. luxurians Kurz, Jour. Asiat. Soc. Bengal 44(2):133. 1876. Illus. Swing. loc. cit., 30:80, fig. 1.
      Type.—Burma, Pegu, Choungmenah (Kurtz. No. 1995).   Herb. Bot. Gard., Calcutta.   A sheet marked with Swingle and Tillson's serial microtome sections No. 519.
      Distribution.—Southern Burma: known only from Pegu and Tenasserim.
      A shrub, with large leaves, 30-50 X 20-27 cm, 5-7- (often 6-) foliolate; leaflets small and broad at base of rachis, 6.5-7.5 X 4-5 cm, larger, lanceolate-acuminate near tip of rachis, 15-21 X 6.5-8.5 cm, base acuminate and decurrent on petiolules, entire-margined, only slightly unequal-sided, terminal leaflet symmetrical at base, largest of all, 15-21 X 6.5-9.5 cm, lateral veins varying greatly in prominence, heavier ones few (7-9), arising at angles of about 70°-75° with the midrib, smaller intermediary lateral veins usually arising at a much greater angle (often 80°-85°) with the midrib, margins entire or subcrenate, surface covered with numerous oil glands varying in size from 0.04 to 0.2 mm, rachis narrowly but clearly winged except the segment below the lowest leaflet, wing broader (1.2-2.5 mm) at top of each rachis segment where leaflet is attached, smaller (0.8-1 mm) at the base of the rachis segment; inflorescences usually terminal, short, about 8-12 X 4-6 cm; flowers white, small, about 4-5 mm diam., 2.5-3.5 mm long from base of ovary to stigma; flower buds subglobose, all flower parts glabrous, free from tannin, pale amber-colored when dry; calyx lobes, anthers, and ovary locules each tipped with a single large oil gland; pistil small, about 2.5-3 mm long; gynophore well developed, hourglass-shaped, 1 mm long, 0.6-0.8 mm wide at narrowest point; ovary tubercular at base, 1-1.1 mm long, 1-1.3 mm wide, with one large oil gland at top of each locule, not blending with the style, which is narrowest at base, 1.1-1.3 X 0.4-0.6 mm; fruits (immature) subglobose, 5-6 mm diam.
      The abundant material of this plant collected by S. Kurz at Choungmenah, Pegu (under No. 1995), was labeled by him C. wallichii var. luxurians.   A collection made by A. Meebold (No. 15090), at Wagôn, Tenasserim (in Herb. Bot. Gard., Calcutta, and also in Herb. Bot. Gard., Breslau), has somewhat smaller leaves than the Pegu material and even smaller inflorescences (6 by 3 cm).   The fact that both C. wallichii and C. luxurians have leaves with a winged rachis, an unusual character in the tribe Clausenae, has probably operated to cause most taxonomists to class them as mere forms of one and the same species.
      Serial microtome sections of the pistils show several important differences between these two species in the absolute and relative size of the flower parts; moreover, C. wallichii has abundant tannin deposits scattered through the tissues, whereas C. luxurians shows none.   The leaflets of C. luxurians are decidedly fewer and much larger than those of C. wallichii and differ also in shape and margination.   The flowers of C. luxurians, however, are decidedly smaller than those of C. wallichii and the petals are not each clearly tipped by a single large oil gland.
      22.   Clausena lenis Drake, in Morot, Jour. Bot. 6:276. 1892. Clausena kerrii Craib, Kew Bul. Misc. Inform. 1913:67. 1913. Illus. Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:664, fig. 70(6-7). 1911.
      Type.—French Indo-China, Tonkin, Tu-phap (Balansa, No. 3667).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—North Vietnam: Tonkin; Laos; Thailand: Pâyap; China: Yunnan Province.
      A rather full description of this species by Guillaumin (1911, p. 663) reads, in translation, as follows: "A shrub completely covered by soft pubescence (except the upper surface of the leaflets).   Leaves 35-40 cm long, with 11-17 alternate leaflets, very unequal-sided at the base (except the terminal one), membranaceous, oval (2.5-12 X 2.5-6 cm), acute at the base, obtuse at the tips with the margins indistinctly denticulate, glabrous above (except the midrib and lateral veins), completely soft-pubescent below, lateral veins 5-8 pairs, slightly raised on the underside; veinlets not visible, petiolules about 1 cm long, petiole cylindric, slightly glandular.   Inflorescence a pyramidal, pubescent, terminal panicle shorter than the leaves, with alternate branches, pedicels soft-pubescent, as long as the flower, without visible articulations; flowers 3-4 mm long, flower buds ovoid.   Sepals 4, small, oval-obtuse, slightly pubescent without, ciliate at the margins.   Petals 4, glabrous, oval, glandular.   Stamens 8, equal, almost as long as the petals, anthers lanceolate, with 2 linear locules each ending in a point, staminal filaments flattened and broadened, very short, glabrous.   Ovary glandular, almost completely glabrous, locules 2, with 2 superposed ovules in each, style cylindrical, as long as the ovary.   Fruit unknown."
      In his description of C. kerrii, now considered a synonym of C. lenis, Craib (l.c.) added a few details to Guillaumin's description, as follows: "Shrub up to 3 m high, twigs up to 5 mm diam., with soft short white pubescence…Leaves 50 cm long, petiole up to 6.5 cm long, lower leaflets 4.6 cm long, 3.2 cm wide, upper leaflets up to 14 cm long and 6 cm wide…Inflorescences…bracts about 1.5 mm long…Petals white (fide Kerr), oblong-ovate, 5 mm long, 2.7 mm wide.   Stamen filaments 0.75 mm long, anther 3 mm long.   Ovary 1.25 mm high, style 2.5 mm long, provided with a very few hairs, here and there."
      The Chinese form may be described as follows: Small tree, 2-3 m high; twigs stout, 5-6 mm diam., at first angled, soon becoming terete, finely puberulent, soon glabrescent; leaves large, 30-65 cm long, with 11-15 leaflets (usually 13-15), rachis minutely puberulent; leaflets very unequal-sided at base, oblong-lanceolate and acuminate or even subcaudate, cuneate at base, varying greatly in size on the same leaf, lower ones sometimes only 2-2.5 X 1.5-2 cm, those in the middle of the leaf largest, 10-13 X 3.5-5 cm, terminal leaflets 8-10 X 2-3.5 cm, lateral veins 6-8 on apical half and 5-6 on basal half on the larger leaflets, margins dentate, except near the base, midrib and lateral veins minutely puberulent, petiolules 2-4 X 1 mm, leaf margins more or less decurrent on sides; inflorescences terminal, or axillary among reduced leaves at the end of the flowering branch, 15-28 cm long and 10-15 cm wide when terminal, 5-15 X 2-4 cm when axillary, ultimate branchlets bearing pedicels 4-8 X 0.5 mm in dry state, entire inflorescence minutely puberulous; flower buds ovoid, light coffee brown in dry state; calyx very small, 4-lobed, lobes semicircular, 0.8-0.9 mm wide, 0.5-0.6 mm long, puberulous; corolla 4-merous, petals thick, coriaceous, entire, incurved at tips, 4-5 X 3-4 mm, glabrous, dotted with small oil glands; pistil 4 mm long in the flower bud, 5-6 mm long after petals and stamens fall; gynophore 1 mm high with a short annulus at the base 1.2 mm wide, 0.35-0.5 mm high, narrow constriction above annulus 0.8-0.9 mm wide, upper part of gynophore cylindrical, 1.2 mm wide, 0.7-0.8 mm high, merging with the equally wide base of ovary at top but with a complete ring of 12-15 oil glands at the base just above the constriction; ovary cylindrical below, 1.2 mm diam., slightly narrowed and rounded at top where a pit, 0.5 mm wide and 0.2-0.3 mm high, bears the style, slightly narrowed where attached, surface covered with numerous slightly prominent oil glands and sparingly hirsute, style cylindrical, 0.5-6 mm diam., and 3.5-3.8 mm long with no clearly marked stigma, ovules collateral, 2 in each locule.   Staminal filaments very short, anthers very long (4-5 mm), slender with a single elongated oil gland at back where filament is attached; fruit (immature) subglobose or slightly ovoid, strongly dotted with oil glands, 5-8 mm diam.
      In Thailand, this species has scattered hairs toward the base of the style.   The ovary is said to be entirely glabrous, but in Tonkin, North Vietnam, it is apparently occasionally slightly hairy, to judge from Drake's description.   In Yunnan Province, the ovary usually bears scattered, rather long hairs which often fall away as the young fruit develops, but the style is usually completely glabrous.   The type specimen from Tonkin shows rather blunt-pointed leaflets with undulate margins; the Siamese material (C. kerrii Craib) has acutely pointed or acuminate leaflets with faintly and distantly serrulate margins; and the Chinese plant shows acuminate or caudate leaflets with rather strongly dentate-serrulate margins.   The length of the style varies somewhat also, being shortest in the Tonkin type and longest in the Chinese material.
      This remarkable species is very different from any others of the genus, as noted above; it has evidently gone some distance on an independent line of evolution distinct from that followed by any other known member of the tribe Clauseneae.
      23.   Clausena suffruticosa (Roxb.) Wt. & Arn. Prodr. 96. 1834. Amyrissuffruticosa Roxb. Fl. Ind. 2:250. 1832.
      Type.—None found; type plant sent from Chittagong, northwestern Burma, to Calcutta Bot. Garden.
      Distribution.—India: eastern Himalayan region, Khasi Hills; Burma: Chittagong.
      An undershrub with simple stems, often only 1-2 ft. [30-60 cm] tall, with all the younger parts softly pubescent; leaves 30-40 cm long; leaflets 11-17, densely soft-pubescent when young, when older glabrescent, lower ones ovate, equal-sided, subopposite, 2-4 X 1-2 cm, with entire margins, upper ones alternate, slightly unequal-sided, ovate-oblong to rhomboid-elliptical, more or less accuminate, sometimes acute, sometimes blunt at the tip; inflorescences axillary simple or slender panicles, shorter than the leaves; flowers 4-merous, small, greenish-white, borne on rather long (3-4 mm), slender, hairy pedicels; petals about 4 mm long; pistil 3-4 mm long; ovary ovoid, 1.5-2 mm long, 1.1-1.6 wide, not borne on a slender stipe, 4-locular with small oil glands but no large one at the top of each locule; style short, cylindrical, 2 mm long, 0.4-0.8 mm wide, tip truncate, stigma not marked; fruits oblong or fusiform, nearly 2.5 cm long by 0.8-1 cm wide, drooping, glandular, orange-colored, succulent, 1 seeded.
      C. suffruticosa is a remarkable undershrub that needs further study.   It is evidently deciduous in the Chittagong Hills in northwestern Burma, as the young leaves and flowers appear together in early spring.   The twigs are often stout, 7 to 8 mm in diameter, with pith 4 to 5 mm in diameter; the leaf scars of the previous year are heart-shaped, 6 to 7 mm long and 5 to 6 mm broad.   Apparently the simple short stems arise from creeping rootstocks, perhaps because of annual fires killing the older wood to the ground.
      The pistil of this plant when studied in detail by means of serial microtome sections made by the modified Juel method (see Swingle, 1939b, p. 270) shows that it differs strikingly from most of the other species of Clausena in having no well-developed gynophore on which the ovary is borne and no cells filled with tannin residues scattered through the tissues.   In this latter character, C. suffruticosa resembles C. luxurians, which does, however, have a definite, though small, gynophore.   It also differs in several other important characters.
      Clausena dentata var. longipes has a pistil resembling that of C. suffruticosa in general morphology; i.e., it has a poorly developed gynophore and a rather long style merging gradually with the apex of the ovary.   However, cells containing tannin residues are found sparingly in the peripheral tissues of the ovary of C. dentata var. longipes and abundantly in the integuments of the ovules.
      The fruits of C. suffruticosa, which were described by Kurz (1877, vol. 1, p. 190) as "oblong, drooping, nearly an inch long, one-seeded, orange-coloured, succulent," are very different from those of any other species of Clausena.

IV.   Murraya Koenig
     
IV.   Murraya ("Murraea") Koenig ex Linn. Mant. Pl. 2:554. 1771; nomen conservandum, Regles Intern. ed. 2. 91. 1912. Chalcas L. 1767; Bergera Koen. 1771; Murraya Koen. 1774; Marsana Sonner. 1782.
      Type species.—Murraya exotica L. = M. paniculata.
      Distribution.—India; Indo-China; Thailand; southern China: Hainan; Taiwan; Malay Peninsula; Andaman Islands; Sumatra; Java; Borneo; Philippines; Celebes; New Guinea; New Caledonia; northeastern Australia.
      Common name.—Orange jessamine (or murraya).
      Unarmed trees with odd-pinnate leaves with alternate leaflets; inflorescences, rather large panicles, either axillary or terminal; flower buds cylindrical or long-ovoid; flowers rather large, 5-merous; calyx of 5 ovate or lanceolate sepals united at the base or only in the lower third; petals 5, rather large, lanceolate or linear, imbricate; stamens 10, free, elongate, filaments flattened in some species, anthers small, broadly elliptic or oval; disk annular, cushion-shaped or cylindrical, short; ovary ovoid, with 2-5 locules, each with 2 (or 1) superimposed or almost collateral ovules; style rather long and slender, finally falling off, stigma capitate; fruit a small berry, ovoid or subglobose, with mucilaginous pulp; seeds medium-sized, with a thin testa, cotyledons plano-convex.
      Murraya is a genus containing eleven species, of which all but three are rather closely related species that fall into two or three groups not as yet adequately studied.   The only relationship of Murraya that appears clear is with Clausena, a genus that is difficult to distinguish at first sight from Murraya except by its having, as a rule, much shorter styles and smaller flowers with larger anthers.   Careful study of the genus Clausena has, however, shown that all the species have an hourglass-shaped gynophore of a character not found in Murraya.   The tissues of all the flower parts in Clausena are usually highly impregnated with tannin residues deposited in some but not all of the cells.   The curious species M. alternans, which seems to be related to M. alata, has a stalked ovary unique in Murraya but of different morphology from the gynophore found in Clausena.
      The anomalous species M. stenocarpa, having unifoliolate leaves, is still too inadequately known to permit discussion of its relationships.
      It is clear that Murraya is not closely related to Citrus; nevertheless Citrus has at least once been successfully grafted on M. paniculata, a vigorous species easily propagated from cuttings.   Dr. Toxopeus, working in Java (1936, p. 6), reported having crossed Citrus and M. paniculata, but he obtained only stunted hybrids.
      Tanaka (1929a) published a monographic study of the species of Murraya (under the genus name Chalcas) which has greatly helped in understanding this genus.   He was the first taxonomist to transfer Micromelum glabrum Guill. and Atalantia stenocarpa Drake to the genus Murraya.
      A key to the eleven species of Murraya is presented.
      1.   Murraya paniculata (L.) Jack, Malay. Misc. 1:31. 1820. Chalcas paniculata L. Mant. Pl. 68. 1767; C. camuneng Burm. f. 1768; Murraea exotica L. 1771; Murraya exotica L. 1774; Marsana buxifolia Sonner. 1782; Limonia lucida Forst. 1786; Murraya sumatrana Roxb. 1832; M. odorata Blanco, 1845; M. scandens Hassk. 1866; Chalcas exotica (L.) Millsp. 1895; Murraya banati Elmer, 1915. Illus. Murray, Comment. Soc. Reg. Göttingensis 9:186-91, pl. 1. 1789; Bellenden-Ker, Bot. Reg. 5:pl. 434 (col.). 1819; Hooker, Exot. Fl. 2:pl. 134. 1825; Wight, Icon. Pl. Ind. Or. 1:pl. 96. 1840; Beddome, Fl. Sylv. Anal. Gen. pl. 7, fig. 2. 1871; Penzig, Studi Bot. Sugli Agrumi, Atl. pl. 17, fig. 15; pl. 18, figs. 12-16. 1887; Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:665, fig. 69(5). 1911; and many others (see Stapf, Index Londinensis 4:333 [1930]); fig. 3-7 this work.
      Type.—British India.   Linn. Herb. London (substitute type not labeled by Linnaeus, fide Jackson, 1912, p, 106).
      Distribution.—India, Ceylon, Burma, Indo-China, China, Taiwan, Malay Peninsula, East Indian Archipelago, Philippines, Australia, New Caledonia, Melanesian islands.
      Common name.—Orange jessamine.
      Kurz (1877, vol. 1, p. 190) described this species as follows: "An evergreen tree, 15-25 ft. [4.6-7.6 m] high, trunk 6-8 ft. [1.8-2.4 m] high, 1 1/2-2 ft. [46-61 cm] diam., the young shoots puberulous; leaves unpaired-pinnate or occasionally pinnately 3-foliolate, glossy, glabrous, or sometimes the rachis puberulous; leaflets alternate, cuneate-obovate or almost obliquely rhomboid, shortly petioluled, blunt or bluntish acuminate, 1-1 l/2 in. [2.5-3.8 cm] long, coriaceous; flowers rather large, white, in dense but small, almost sessile terminal corymbs; petals about 1/2-3/4 in. [13-18 mm] long, recurved; stamens 10, alternately shorter; ovary 2-celled, the style long with a capitate glandular stigma; berries ovoid-oblong, bluntish acuminate, nearly 1/2 in. [13 mm] long, orange-coloured, 1-2-seeded; seeds villous."
      This species is very widely distributed and doubtless extremely old.   However, it has evolved at a very slow rate, as is shown by the small morphologic differences that distinguish it from var. ovatifoliolata (noted below), which has been separated from the other forms of the species for many millions of years ever since Australia lost all land connections in the Eocene period.   The wood of this species—particularly that of the roots—is much valued by the Malays for making kris handles, Burkill (1935, vol. 2, p. 1507) noted that the trunks split unless carefully seasoned and that few pieces are larger than 8 inches (20 cm) in diameter.
      The orange jessamine is a handsome greenhouse ornamental that blooms profusely.   It has large white fragrant flowers that are succeeded by small red fruits (see fig. 3-7).   In warm subtropical climates it thrives out of doors.   It grows vigorously and can be propagated easily from cuttings.   Citrus cannot as a rule be grafted on it, although one strain of lemon (an unidentified seedling) made a good growth on it in the greenhouses of the former Bureau of Plant Industry at Washington, D.C.   There are many varieties and strains of M. paniculata in the Old World; possibly some of them would support Citrus better than the strain now grown in the United States, probably originally introduced from India.
      1a.   Murraya paniculata var. ovatifoliolata Engl. Die Nat. Pflanzenfam. 3(4):188. 1896. Murraya ovatifoliolata (Engl.) Domin, Bibl. Bot. 89:296. 1927. Illus. Bailey, Compr. Cat. Queensl. Pl. fig. 61 bis. 1909.
      Type.—Queensland (F. M. Bailey).   Herb. Bot. Gard., Brisbane (?).
      Distribution.—Australia: Queensland.
      Engler distinguished this variety from the species as having broadly oval or ovate leaflets.   Bailey (l.c.) described this form as follows: "This, our indigenous form, is of a more straggling habit with more numerous and larger oil-dots, and is often decidedly hirsute and tomentose, thus very distinct from the two Indian ones of our gardens."   The leaves are 3-9-foliolate; the twigs, calyx, petals, and ovary hirsute.
      The essential oil in the leaves of var. ovatifoliolata from Queensland was compared with that from the species growing at Dehra Dunn, India, by Penfold and Simonsen (1925, pp. 146-55).   The oil of both species and the variety was found to consist chiefly of sesquiterpenes and sesquiterpene alcohols, but "the chief constituent of the Indian oil was found to be l-cadinene…whilst bisaboline was similarly identified as one of the sesquiterpenes present in the Australian oil."   The other constituents were not identified as they did not yield any crystallizable derivatives.   Significant differences were also found in the saponification value of the two oils.   The leaves of the species grown in India yielded on steam distillation 0.01 per cent of dark-colored oil with a saponification value of 8.87; leaves of the variety from Queensland yielded 0.06 to 0.15 per cent of oil having saponification values of from 19.37 to 26.18.   There were also differences in the rotative power of the oils, that of the species being [a]D-78.2°; that of the variety from -10.2° to -12.2°.
      1b.   Murraya paniculata var. zollingeri Tan. Jour. Ind. Bot. Soc. 16:232. 1937. Chalcas paniculata var. zollingeri Tan. Bul. Soc. Bot. France 75:709. 1928.
      Type.—Bima [or Sumbawa] Island (Zollinger, No. 3351).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Central East Indian Archipelago; Bima, Sumbawa, and Timor islands.
      Leaflets 3-6, small, chartaceous, margins much reflexed; rachis thin, declinous, slightly pubescent; fruits not apiculate, often pendulous, sparsely puberulous.
      Tanaka (1929a, p. 27) is uncertain whether this is not the same as M. heptaphylla Spanoghe, in Linnaea 15:178 (1841).
      1c.   Murraya paniculata var. omphalocarpa (Hay.) Tan. Jour. Soc. Trop. Agr. 1:27. 1929. Murraya omphalocarpa Hay. Icon. Pl. Formos. 3:51. 1913.
      Type.—Formosa, Totosho (coll., Kawamakami).   Herb. Imp. Univ., Tokyo.
      Distribution.—Known only from Taiwan.
      Differs from the species in the following characters: fruits large with attenuate tips, about 21 X 12 mm (when immature, rostrate, about 15 X 5 mm); flowers larger; petals narrowed at the base; calyx lobes elongate, ovate to linear-oblong, 3 mm long; leaflets broad, 5-8 X 4-4.7 cm.
      Tanaka (1929a, p. 28) stated that "these characters may be linked by gradations" with those of the species, but as the differences are fairly large they can safely be regarded as varietal but not specific.
      2.   Murraya gleniei Thwaites ex Oliv. Jour. Linn. Soc. Bot. 5(2):29. 1861. Chalcas gleniei (Thwaites) Tan. Bul. Soc. Bot. France 75:710. 1928. Illus. Trimen, Handbook Fl. Ceylon, Atl. pl. 22. 1893.
      Type.—Ceylon (Thwaites, No. 3627).   Herb. Kew.
      Distribution.—Known only from Ceylon.
      This species was best understood by Trimen (1893, vol. 1, p. 220), who, while he was director of the Botanic Gardens of Ceylon (where alone the species is found), naturally had abundant material at his disposal.   His description reads: "A bush, much branched, bark very white, young twigs strongly pubescent; leaves imparipinnate, 1 1/2-4 in. [3.8-10 cm long]; rachis strongly pubescent, leaflets 3-7, on short pubescent stalks, upper ones much the longest, rhomboid-lanceolate, obtuse, emarginate, lower ones rotundate, all slightly crenate, glabrous; flowers 1/2-3/4 in [13-19 mm] diam., about 3-7, in short, racemose, axillary and terminal cymes, pedicels very pubescent; sepals very small, densely tomentose; petals 1/2 in. [13 mm long]; ovary on a distinct cylindrical stalk, 5-lobed, 5-celled; berry large, 1 in. [diam.], somewhat pyriform or urn-shaped, flat-topped, but mammillate in centre, 5-lobed or bluntly 5-angled, rough with large glands, shining green, 5-celled, the large inflated cells each containing 1-3 globose green seeds."
      This Ceylonese form of the orange jessamine is much like the normal form of the species except in its curious inflated star-shaped fruits.   It may perhaps be a mutation which has been able to establish itself as a satellite species.   Engler (1931, p. 320) noted that the ovary is "4- or 5-locular" and consequently the fruits must be also 4- or 5-lobed.
      3.   Murraya alternans (Kurz) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Limonia (?) alternans Wall., in Voigt, Hort. Suburb. Calcut. 139, nomen nudum. 1845; L. alternifolia ("Wall. ap. Voigt") Kurz, Jour. Asiat. Soc. Bengal 42:64. 1873; L. alternans ("Wall.") Hook. f. Fl. Brit. Ind. 1:508, 1875; L. alternans ("Wall.") Kurz, For. Fl. Brit. Burma 192. 1877.
      Type.—Burma, Pegu (Kurz, No. 2010).   Herb. Kew.
      Distribution.—Known only from type locality.   Kurz (1877, vol. 1, p. 192) stated: "Not unfrequent in the upper mixed and occasionally in the moist forests of the Pegu Yomah [Mountains, about Lat. 18°40'-19° N., Long. 96°35'-97° E.], sporadical but usually gregarious."
      A slender, simple or sparingly branched shrub, 60-120 cm [2-4 ft.] tall, with deciduous foliage; twigs slender, glabrous, with numerous shallow, longitudinal furrows and low ridges; leaves odd-pinnate, completely glabrous, 11-15 foliolate; leaflets alternate, nearly sessile, oblong-lanceolate to lanceolate, 2.5-4 cm long, somewhat unequal-sided toward the cuneately narrowed base, slightly attenuate to a blunt tip, margins slightly crenate above the middle; rachis narrowly winged; inflorescences small, short-peduncled, glabrous, axillary cymes, appearing with the new leaves; flower buds small, oblong or elliptical, 3-5 X 2-3 mm, flowers short-pedicellate; calyx nearly flat, 5-lobed, lobes triangular, glabrous, 1-1.5 mm long, 1.5-2 mm wide, with numerous oil glands; petals 5, white, upright, linear, 5-7 mm long, tapering to a point, with many oil glands in the upper half; stamens free, filaments 10, slightly swollen above, sparingly short-pilose, anthers small, oblong, 1.2-1.3 mm long; pistil glabrous, 4.5-5.5 mm long; ovary borne on a long, slender, nearly cylindrical stipe, 0.8-1 mm long and 0.4-0.5 mm wide, arising from flat top of the disk (said to elongate after flowering), the ovary itself obovate, 1.8-2 mm high, 1.1-1.2 mm wide, bluntly rounded or obcordate at top because of 2 large oil glands, 1 over each locule; locules 2, each with 2 ovules suspended from the top; style slender, subclavate, 2.2-2.6 mm long, 0.2-0.5 mm wide, stigma at tip not clearly delimited; disk annular, as broad or broader than the ovary, 1.8-2 mm wide, 0.5-0.6 mm high, fruits ovoid, glabrous, short-stalked, about 4 mm long.
      This species was assigned, somewhat doubtfully, to Limonia by Kurz and Sir Jose Hooker, but has apparently never been studied since the 1870's.   It is no longer mentioned in lists of plants covering Burma, and even the genus Limonia, to which it was assigned when discovered (and from which it has never been removed), is no longer in use (see Swingle, 1914-17, pp. 327-28).
      This extraordinary plant, frequently producing only a single slender stem, 2 to 4 feet high, is unique in the genus Murraya in having deciduous foliage.   This species together with Clausena pentaphylla, C. dentata var. dulcis, and C. suffruticosa are the only members of the tribe Clauseneae having deciduous foliage.   The leaves drop off during the hot, dry season, which in Pegu begins toward the end of February and lasts until May.   When the summer rains come on, new leaves push out on the bare stems, and with them clusters of small white flowers appear in the axils of the leaves.
      The type specimen of this species, collected in Pegu in 1872 by Kurz, was kindly lent to Swingle by Sir Arthur Hill, director of the Royal Gardens at Kew.   A study of the flower anatomy from serial microtome sections made from this Herbarium material by the modified Juel method shows clearly that it belongs in the genus Murraya and is a species near to M. alata, which occurs not far to the eastward in Indo-China and which also has leaves with a winged rachis.   It differs from all the other specie of Murraya in having the ovary borne on a slender, nearly cylindrical stipe.
      Kurz, in publishing his first description of this species in 1873, mistakenly used the specific name "alternifolia," although he professed to recognize the specific name "alternans" first printed (but not adequately published) by Wallich in Voigt in 1845.   Hooker corrected this misspelling in 1875 and Kurz accepted the correction in 1877.
      4.   Murraya alata Drake, in Morot, Jour. Bot. 6:276. 1892. Chalcas alata (Drake) Tan. Jour. Soc. Trop. Agr. 1:35. 1929.
      Type.—French Indo-China, Tonkin, Tan-keuin (Balansa, Nos. 1118, 1119).
      Distribution.—North Vietnam: Tonkin (Tan-keuin, Kien-khé, and Bac-bat), Annam.
      The brief original description reads, in translation, as follows: "Shrub, 1 m high.   Leaves up to 10 cm long, completely glabrous; petiole and rachis alate; leaflets 7, 3-4 cm long, 1-2 cm wide, oblong or rhomboid, acute or subacute at the tip, obscurely crenate; petiolule very short and inconspicuously puberulous.   Cymes axillary, 2-3 cm long, few-flowered, puberulous.   Sepals acute.   Petals oblong.   Staminate filaments linear.   Berry ovoid."
      This description was based on the type material from Tan-keuin, Tonkin, North Vietnam (Balansa, Nos. 1118, 1119).   Guillaumin (1911, p. 659) added a few characters not mentioned by Drake del Castillo, since he had additional material at his disposal from Kien-khé and Bac-bat, Tonkin, collected by Bon.   The more important characters first noted by Guillaumin are described, in translation, as follows: "…Branches glabrous, with yellow bark.   Leaflets 5-7…coriaceous, entire or undulate on the margins, lateral veins scarcely visible.   Inflorescences…2-3 cm long, branches 15 mm long, pedicels 9 mm long.   Sepals…triangular, acute, 1.5 mm long, pubescent without.   Petals glabrous, erect, obovate, 10-15 mm long, with parallel veins.   Stamens 10, the 5 longest a little shorter than the petals, and reaching to the stigma; filaments glabrous, subulate; anthers very small, attached at the base…terminated by a small grand, not easily visible.   Disk annular.   Ovary glabrous, ovoid; style subulate; stigma capitate; locules 2 each with a single ovule.   Fruits red, grouped in corymbs, fleshy, with 1 or 2 seeds which have a parchment-like testa and a single-straight embryo with fleshy plano-convex cotyledons."
      Like M. alternans, which it resembles in having leaves with a narrowly winged rachis, this species is reported as being a shrub only 1 m high.   M. alata differs from M. alternans in having persistent instead of deciduous leaves, and flowers with petals about twice as long.   The calyx lobes of M. alata are pubescent without and the staminal filaments are glabrous, both conditions being just the opposite of what is found in M. alternans.
      4a.   Murraya alata var. hainanensis Swing. Jour. Wash. Acad. Sci. 32:26. 1942.
      Type.—China, Hainan Island, strand, Haichow (McClure, No. 7611).   Herb. Natl. Arbor., Washington, D.C.
      Distribution.—China: known only from Hainan Island.
      This variety differs from the species found in Tonkin and Annam in North Vietnam in having the leaflets when young minutely puberulent on both surfaces and on the margins.   As they mature, the leaflets often become nearly glabrous, but the petiolules and the rachis remain more or less pubescent even on mature leaves, and often scattered, short cilia may be still seen along the margins near the base of mature leaflets.   The calyx lobes are pubescent and the staminal filaments are glabrous as in the species.   The leaflets seem to be rather more variable in shape (especially in width) than those of the Indo-Chinese typical form.   The fruits are ovoid, about 9-11 X 6-8 mm in herbarium specimens.
      This variety is a shrub growing from 1 to 2.5 m high along the beaches, and in the forests of Hainan from sea level to 1,300 m altitude.   The fruits are said by one collector to be "lustrous deep red" when ripe.
      5.   Murraya koenigii (L.) Spreng. Syst. Veg. 2:315. 1825. Bergera koenigii L. Mant. Pl. 563. 1771; Murraya foetidissima Teijsm. & Binn. 1864; Chalcas koenigii (L.) Kurz, Jour. Asiat. Soc. Bengal 44(2):132. 1875. Illus. Wight, Icon. Pl. Ind. Or. 1:pl. 13/334. 1838; Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:655, fig. 69(4). 1911; figs. 3-8 and 3-9 this work.
      Type.—Eastern British India (Koenig, suppl. type).   Linn. Herb., London.
      Distribution.—Indian peninsula; northeastern and northwestern India; Ceylon; Burma; North Vietnam; South Vietnam; Cambodia; Laos; China: Hainan Island, Yunnan, and Kweichow.
      Common name.—Curry leaf.
      Kurz (1877, vol. 1, p. 190) described this species as follows: "An evergreen tree, 15-20 ft. [4.6-6.1 m] high, trunk 4-10 ft. [1.2-3 m] high, 1/2-1 1/2 ft. [15-46 cm] in circum., glabrous or slightly puberulous; leaves unpaired-pinnate, the rachis usually more or less pubescent, rarely quite glabrous; leaflets in 5-10 pairs with an odd one, on a short puberulous petiolule, oblong-lanceolate or ovate, almost falcate, oblique at base, about 1-1 1/2 in. [2.5-3.8 cm] long, acuminate, more or less serrulate, membranous, glabrous, except on the midrib, which is often puberulous; flowers small, white, in terminal corymbs; petals oblong-lanceolate, acute, about 2-3 lin. [4-6 mm] long; stamens 10, alternately shorter; ovary 2-celled, the style short and thick; berries oblong, somewhat acute, the size of a small pea, l-2-seeded, bluish-black."
      This species was said by Trimen (1893, vol. 1, p. 219) to be "extremely like the small-leaved form of Micromelum pubescens [= M. ceylanicum], from which it may be distinguished by its more numerous more pubescent and small leaflets, and more compact corymbose terminal inflorescence."   The pistil is much shorter than the stamens.
      Dr. David Fairchild reported (Bur. Pl. Ind., Inventory 88, No. 68351) that "the fresh leaves form a constant ingredient of the Ceylonese curries and give them a very agreeable flavor."   He stated further (1930, p. 284) that the fresh leaves of this species are "deemed essential to all the curries of Ceylon."   They are "boiled with the curry but thrown out of it before serving."   Dr. Fairchild had a fine fruiting tree of M. koenigii growing in his tropical garden, "The Kampong," at Coconut Grove, Florida, where it was one of the most prized of his collection of citrus fruits and their wild relatives (see fig. 3-8).   He regularly used the fresh leaves in making curry.
      Trimen stated (1893, vol. 1, p. 221): "This is the familiar 'curry-leaf,' a constant ingredient in curries and mulligatawny."   He reported the species as rather rare in the wild state in Ceylon but "very much cultivated," doubtless for use in curries.
      Dr. David Fairchild obtained seeds of this species in 1926 in Ceylon.   Plants grown from this seed have thrived on the peculiar soil (a very sandy loam intermixed with porous limestone fragments of rock that often occupy half or more of the bulk of the soil) found in Miami and vicinity in the extreme southeastern part of Florida.
      6.   Murraya microphylla (Merr. & Chun) Swing. Jour. Wash. Acad. Sci. 32:26. 1942. Clausena microphylla Merr. & Chun, Sunyatsenia 2:251. 1935. Illus. Merrill & Chun, loc. cit. fig. 27. 1935.
      Type.—China, Hainan Island, Sunchuen (Chun and Tso, No. 43323).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—China: known only from Hainan Island.
      Common name.—Small-leaved murraya.
      The original description by Merrill and Chun reads, condensed and translated, as follows: "A shrub or small tree, branches glabrous, branchlets with a dispersed white pubescence;…leaves imparipinnate…5-7 cm long, petiole and rachis with scattered white pubescence of short curved hairs; leaflets, 5-10 on each side, alternate or subopposite…oblong or oblong-ovate or oblong-lanceolate, upper ones usually larger up to 2 cm long…lower ones subopposite, 3-4 mm diam.; intermediate ones 5-18 mm long, 3-6 mm wide, all of them crenulate, obtuse…base usually inequilateral; lateral veins 3-6 pairs…inflorescences terminal, sparingly pubescent…short-pedunculate, 2-2.5 cm long and wide; [flowers unknown]; fruits few, ellipsoid or ovoid-ellipsoid, up to 1 cm long, glabrous, glandular-punctate…persistent sepals 5, oblong-ovate, punctate, glabrous, obtuse or subacute, 1 mm long; pedicels scatteringly pubescent, 2-2.5 mm long."
      A specimen collected by C. Wang (No. 32740), July 4, 1933, on the Fourth Hainan Expedition of Sun Yatsen University (in Herb. Arnold Arbor.) has good flower buds, 4-5 X 2.5 mm: petals show scattered, very dark oil glands.   Serial microtome sections show the pistil to be 4 mm long, disk very short and not delimited from the ovary, which is 1.2-1.5 mm long and 0.9-1 mm wide, with 1 or 2 locules, with 1 or 2 large oil glands at the top of the locule (not in a locular process), merging into the broad style, about 3 X 0.4-0.5 mm, slightly swollen at the top into a subcapitate stigma, 0.7-0.9 mm diam., which shows a few small oil glands.   The tips of the petals also show several fairly large oil glands.   The stamens are about as long as the pistil, with linear subulate filaments.
      The finding of flowers of this interesting plant shows clearly that it is a species of the genus Murraya, of which it is a striking member, very different from any other species because of its small leaves and tiny leaflets.   The collector's notes on the Hainan specimens show that it may reach 6 m in height, and have a trunk diameter of 15 cm.   It bears white fragrant flowers.
      7.   Murraya siamensis Craib, Kew Bul. Misc. Inform. 1926(8):340. 1926. Chalcas siamensis (Craib) Tan. Bul. Soc. Bot. France 75:710. 1928.
      Type.—Thailand, Lampang, Mê Kat, alt. 250 m (Winit, No. 849).   Herb. Kew.
      Distribution.—Thailand.   Reported from four administrative districts by Craib (1926, vol. 1, p. 230).
      Craib’s original description, translated, reads as follows: "Small tree, trunk with fuscous bark having high ridges (fide Winit); young twigs with curly pubescence, year-old twigs puberulous, then fuscous, with small lenticels.   Leaves 15-24 cm long (including the petiole, 1-2.5 cm long), petiole and rachis almost terete with dense curly pubescence; leaflets 17-25, alternate or rarely subopposite, inequilateral (one half oblanceolate or lanceolate, narrowed at the base; the other half ovate, subtruncate at the base), apex obtuse, subacuminate, 2-7 cm long, 1-2.7 cm broad, chartaceous, above curly pubescent, very soon more or less glabrescent below, chiefly pubescent at the midrib, lateral veins 5-7 pairs, anastomosing near the margin above, later conspicuous, somewhat prominent below, veinlets somewhat conspicuous above, below forming a delicate rather prominent reticulation, margin subentire or more or less obscurely crenulate, petiolule about 2 mm long, covered with dense, curly pubescence.   Panicles terminal, subcorymbose, 15 cm long; pedicels 3-4 mm long with short curly pubescence.   Sepals 5, ovate, lanceolate or deltoid, obtuse, 1.25 mm long, covered with short, curly pubescence without.   Petals 5, linear-oblanceolate, incurving-apiculate, 6.5 mm long, 1.75 mm wide, glabrous, thickened at the middle, imbricate.   Disk fleshy, glabrous, slightly shorter than the calyx.   Stamens 10, alternate ones longer, filaments 4-5 mm long, narrowed toward the apex, the lower half slightly broader but scarcely complanate, anthers about 1 mm long, papillate.   Ovary slightly oblong, scarcely 1.5 mm long, glabrous, shortly stipitate; style 3-5 mm long, articulated at the base, glabrous, stigma capitate.   Fruit (immature), 1 cm diam."
      Tanaka (1929a, p. 34) gave a redescription of the material cited by Craib which adds a few points to Craib's diagnosis.   The leaves are clearly dotted on the upper surface; the terminal inflorescence is small, bearing 10-15 flowers; "flower buds small, oblong-ovoid, about 5 mm long, without glandular dots, pedicels short, about 6 mm, densely pubescent…ovary cylindric, notched, continuous with the cylindric style."   Tanaka reported that the fruit clusters are as much as 9 cm wide; the fruits are large, "black, somewhat 4-angled, indistinctly divided by shallow longitudinal grooves"; oil-dots large, more or less indistinct, pustulate; calyx lobes persistent, with long soft hairs; "pedicel about 1.5 cm long, continuous to branch of peduncle…about 2 cm long."   He found that the fruits, usually about 1.5 cm diam. may attain 2 cm, and the fresh fruit may measure still more.
      Tanaka (1929a, p. 34) stated: "This species is closely allied with C. [Murraya] koenigii, but dots on the petals are nearly indistinct, filaments are less subulate, petals much narrower, and fruit much larger and 4-grooved.   The pubescence of the plant is more pronounced than C. [Murraya] koenigii, and sometimes the whole plant is covered with white curved hairs, even…the upper surface of the leaflets."   Winit noted on his label on the type specimen that the plant is decidedly shrubby, blackish with deeply cracked bark, and has greenish-white, scented flowers.
      A specimen from Thailand, collected on December 31, 1928, by Dr. A. F. G. Kerr (No. 9740) has apparently mature, globose, or slightly depressed-globose fruits, about 1.5 to 1.7 cm in diameter in the dry state, that do not show any distinct longitudinal grooves but merely very broad, very shallow depressions between the slightly protuberant locules.
      8.   Murraya euchrestifolia Hayata, Icon. Pl. Formos. 6:11. 1916. Clausena euchrestifolia (Hay.) Kanehira, Formos. Trees 97. 1917; Chalcas euchrestifolia (Hay.) Tan. Jour. Soc. Trop. Agr. 1:32. 1929. Illus. Kanehira, Formos. Trees 104. 1917, and ibid. rev. ed. 309, fig. 263. 1936.
      Type.—Formosa, Nantôchô, (T. Itô).   Herb. Imp. Univ., Tokyo.
      Distribution.—Taiwan.
      Tanaka (1929a, p. 32) described this species as follows: "Shrub; leaflets up to 8, rachis fine, puberulent; leaflets elliptic, acuminate at both ends, shiny above, largest 9.5 X 3.4 cm, nearly glabrous, lower surface of the midrib slightly puberulent, margin entire.   Inflorescence [a] terminal, many-flowered cyme, buds small, about 4 X 1.4 mm, black dotting few on the upper half.   Calyx well lobed, about 1.5 mm diam., lobes triangular, black, glabrous down to the base, margin not pilose; pedicel slender, 4-5 mm long, pilose with soft, golden, curved pubescence.   Petals 5, oblong-oboval, about 4.5 X 1.5 mm, obtuse at the apex, not much narrowed at the base, with approximately 10 dots on outside.   Stamens 10, one shorter after the other, [longer and shorter ones alternating], longer ones about 4.5 mm, nearly as long as the petals when fully expanded, filaments linear, flat, gradually narrowed and acutely pointed; anther small, roundish, slightly cordate, about 2-3 mm diam.   Pistil slightly shorter than the stamen; ovary oval, constricted at the base toward the broad end; style cylindric, terete, often slightly curved, stigma obscure, altogether about 3.5 mm long.   Berry globose, not apiculate, largest about 1 cm across, smooth, rather sparingly large pellucid-dotted, dark colored."
      Tanaka (1929a, p. 32) stated that this plant most closely resembles M. crenulata in appearance, but the calyx lobes are acutely triangular, and the leaflets are not so oblique, lacking crenulation.   In this respect, it approaches M. koenigii more closely, although the number of leaflets is fewer and they are larger and glabrous.   It differs from both species in having a smaller flower, an obscure stigma, and a slightly constricted ovary.
      9.   Murraya crenulata (Turcz.) Oliv. Jour. Linn. Soc. Bot. 5(2):29. 1861. Glycosmis crenulata Turcz. Bull. Soc. Nat. Mosc. 31:250. 1858; Micromelum coriaceum Seem. 1865; Clausena worcesteri Merr. 1910; C. citriodora Merr. 1925.
      Type.—Philippines (Cuming, No. 335).   Herb., Leningrad.
      Distribution.—Java, Philippines, Celebes, New Caledonia.
      Tanaka (1929a, p. 30) described this species as follows: "Shrub; leaves pinnate, composed of about 6-7-paired leaflets; leaflets medium-sized [3-6 X 1-2 or 3 cm], much oblique, oval or oboval, acuminate or mucronate at the apex with blunt tip, acute at the base (roundish in the lower leaflets), very short petiolulate, veins very distinct, irregular, margin more or less crenulate, semicoriaceous, shining above, drying black, and midrib always glabrous.   Inflorescence a terminal cyme, many-flowered; flower small, but cylindric, about 5 X 2 mm, apex rounded; petal narrow, striated, with obscure black dots.   Calyx cupulate, puberulent, lobes broad and rounded, side often imbricate and ciliate on margin; peduncle slender, tomentose.   Stamens free, unequal in length, filaments long, linear, more or less dilated at the lower half, very sharp pointed at the apex, anther globose.   Pistil nearly as long as the filaments; ovary ellipsoid, apex not narrowed, rather broad and more or less ridged, base somewhat narrowed to a cylindric disk; style long, terete, cylindric, stigma globose.   Berry nearly globose and minutely apiculate, dark-colored."
      Tanaka noted that M. crenulata is allied to M. koenigii but differs from it in having fewer leaflets, which are thicker, larger, and attached to a glabrous instead of a pubescent rachis.   The fruit is subglobose and minutely apiculate, whereas in M. koenigii it is short-ellipsoid and apiculate.   M. crenulata is often confused in the field with Clausena anisum-olens and with Micromelum minutum var. curranii, but it has long flower buds and long pistils, which are never found in Clausena or Micromelum.

SPECIES OF UNCERTAIN RELATIONSHIPS
      10.   Murraya glabra (Guill.) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Micromelum glabrum Guill. in Lecomte, Not. Syst. 1:216. 1910; Chalcas glabra (Guill.) Tan. 1929. Illus. Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:646, fig. 68(11-12). 1911.
      Type.—French Indo-China, Kien-khé, Tonkin (Bon, No. 2977).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Vietnam: Tonkin and Annam.
      Guillaumin (1911, p. 650) described this species as follows: "Small shrub, 1 m high.   Branches entirely glabrous, with very odorous oil glands.   Leaves 12-46 cm long; leaflets 3-9, alternate, glabrous, obovate or oblong-lanceolate, 5-17 X 2-9 cm, inequilateral and pointed at the base, abruptly acuminate at the tip, very slightly denticulate; lateral veins, 6-9 pairs, prominent on lower surface; petiolules cylindric, glabrous, 3-4 mm long; petiole cylindric, glabrous, glandular.   Inflorescences shorter than the leaves, with very short pubescence; pedicels longer than the flowers, with 2 small bracts at the base or lower third; flower [buds] 4 mm long, white.   Calyx of 4 connate, indistinct, very shortly ciliate sepals.   Petals 4, length 6 mm, glabrous, lanceolate.   Stamens 8, of which 4 are longer; filaments dilated and pubescent above; anthers attached by the back, oval, with a few hairs.   Disk extremely short.   Ovary cylindric, warty, glabrous; style longer than the ovary, ending in a globose stigma."
      Tanaka (1929a, p. 36) in his monograph of the genus Chalcas (= Murraya) made a few additions and changes to Guillaumin's description, among others the following: "Leaflets 3-11, generally 5…texture thin, chartaceous…Rachis rather thin [slender?], about 28 cm long…Flower bud about 6 X 2.5 cm.   Ovary…distinctly constricted at about one-third from the apex…"
      This anomalous species was described as " belonging to Micromelum but was transferred to Murraya by Tanaka "on account of having a large cyme, long flower buds with prominent dots and elongated style."   The specimens from Annam have, so Tanaka found, distinctly terminal inflorescences which bear nearly 100 flowers.   The fruit is unknown.
      11.   Murraya stenocarpa (Drake) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Atalantia stenocarpa Drake, in Morot, Jour. Bot. 6:277. 1892; Glycosmis stenocarpa (Drake) Guill. 1912; G. bonii Guill. 1910; Chalcas stenocarpa (Drake) Tan. Bul. Soc. Bot. France 75:711. 1928. Illus., Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:667, fig. 71(3-6). 1911.
      Type.—French Indo-China, Tonkin (Balansa, No. 1110).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—North Vietnam: Tonkin.
      The original description of Atalantia stenocarpa by Drake del Castillo may be translated as follows: "Small tree (up to 1 m high), spineless, very glabrous; leaves subcoriaceous, shining, elliptical or oblong-lanceolate (leaf blades up to 15 cm long, 6 cm wide; petiole about 4 cm long), acute at the base, narrowed at the apex, obtuse or subemarginate, serrate.   Cymes axillary, few-flowered.   Flowers small.   Calyx lobes 5, short, ovate, acute.   Petals oblong.   Stamen filaments free, ciliate.   Ovary oblong, substipitate, with 2 ovules.   Fruit oblong, shortly attenuate at the apex."
      A redescription by Guillaumin (1911, p. 671) under the name Atalantia stenocarpa added a few details "…Leaflets 4-12.5 X 3-5 cm, cuneiform at the base, attenuate and slightly emarginate at the tip; with 9-10 pairs of lateral veins, clearly distinct from the veinlets and like them elevated above the surface below; oil glands easily visible principally on the margins and in the angles of the teeth; petiole 2-3.5 cm, with a swelling at the base of the leaf blade; ovary ovoid, borne on a low disk, with 2 locules, each with 1 ovule; style cylindric-pubescent.   Fruits fleshy, reddish, oblong, 15 mm long, attenuated at the tip."
      This anomalous species, with unifoliolate leaves, was placed in this genus by Tanaka, although originally it was placed in the genus Atalantia.   Guillaumin, because of this leaf-character, transferred the species to Glycosmis, but, according to Tanaka, "the plant lacks ferruginous pubescence characteristic to Glycosmis…The ovary…is glandular and somewhat stalked at the base and this character agrees with Clausena and not…Atalantia.   The filaments of Clausena, however, are very much attenuated, ending with a filiform tip, and the anther is extremely long.   In the present species, the filaments are dilated [at the base] and abruptly pointed [at the tip] and the anthers are not large; moreover they are pubescent, and [have] pellucid dots on the surface of the petals and calyx lobes similar to Murraya glabra."
      Tanaka (1928b, p. 711) emphasized the "exact coincidence of the characters of the stamens in the present species" with those of M. glabra.   However, M. glabra is itself a somewhat anomalous species of Murraya, and both of these species should be studied more fully.

Subtribe 3.   Merrilliinae: Large-Fruited Remote Citroid Fruit Trees
      The subtribe Merrilliinae contains only one genus with a single species, Merrillia caloxylon native to the Malay Peninsula and to the adjacent coast of Sumatra.   Merrillia differs widely from all the other genera of the tribe Clauseneae and has a superficial similarity with the genus Swinglea, one of the aberrant Hard-Shelled Citroid Fruit Trees belonging to the tribe Citreae.
      Swingle follows Engler (1931, p. 211) in placing Merrillia next to the genus Murraya, which it resembles in general aspect and in its hard handsome wood, although differing greatly in flower and fruit characters.   The flowers are the largest known in the orange subfamily but do not open fully (as do those of Citrus).   They are trumpet-shaped, becoming completely pendent during anthesis, with only the upper third of the petals recurved.   The petals composing the corolla are not only the largest known in the orange subfamily, but they are also slightly unequal in size and shape, i.e., slightly zygomorphic.   The fruits are as strange as the flowers and have a thick, radially lacunose, leathery exocarp and seeds covered with abundant, narrow, flattened somewhat deeply laciniate paleae, characters not found elsewhere in the subfamily.
      This subtribe seems to have been derived as an offshoot from the ancestors of Murraya that evolved rapidly in flower and fruit characters while retaining the general facies of a Murraya of the M. paniculata group of species.

V.   Merrillia Swing.
      V.   Merrillia Swing. Phil. Jour. Sci. Bot. 13:337. 1918.
      Type species.Murraya caloxylon Ridl. = Merrillia caloxylon (Ridl.) Swing.
      Distribution.—Malay Peninsula: southern Thailand, upper Perak; northeastern Sumatra.   Cultivated in Singapore Botanic Gardens.
      Common name.—Merrillia.
      Swingle's original description, translated, reads as follows: "Genus allied to Chaetospermum [now Swinglea]…from which it differs chiefly in the 5-6-locular ovary having an irregular lacunose pericarp; the seeds densely scaly, scales elongate, membranaceous, slightly laciniate; the leaves sessile, pinnate, the rachis narrowly winged, and the leaflets alternate.
      "Unarmed tree; leaves pinnate, sessile or subsessile; rachis narrowly winged; leaflets alternate, lower ones small, the next higher ones gradually increasing in size, the terminal ones largest; petiolule very short; inflorescence axillary, 2-1-flowered; flowers perfect, rather large, 5-merous; calyx 5-merous, small, lobes triangular-ovate; petals 5, narrow; stamens 10, free, unequal; ovary 5- (rarely 6-) locular, stipitate, merging into the elongate style, stigma capitate, ovules 8 to 10 in each locule; fruit subglobose, large pericarp very thick, coriaceous, irregularly lacunose with cartilaginous walls, segments filled with mucilaginous gum, seeds numerous in each segment (8-10), lenticular, densely scaly, scales slightly fimbriate, membranaceous, elongate, hilum carinate, subariloid.   Cotyledons hypogeous in germination, first leaves simple, opposite, broadly ovate-lanceolate."
      The very long, trumpet-shaped, slightly zygomorphic flowers of Merrillia, often 55 to 60 mm long, are unique in the orange subfamily.   The large fruit with its thick, radially lacunose epicarp and seeds covered with flat, narrow, laciniate paleae are equally strange.   These and other remarkable flower and fruit characters of Merrillia constitute one of the most striking examples of rapid divergent evolution to be found among the many that are seen in the orange subfamily, such as the highly xerophytic Eremocitrus and the anomalous Monanthocitrus with its horned fruits and spotted seeds.   The extraordinary fact is that Merrillia not only has flowers, fruits, and seeds which differ very widely from those of the other genera of the tribe Clauseneae, but fruit characters analogous to those of the genus Swinglea, an aberrant member of the Hard-Shelled Citroid Fruit Trees belonging to another tribe.
      When Swingle described Merrillia as a genus in 1918, he thought that it probably belonged with Swinglea in a special group of the subtribe Balsamocitrinae, although he directed attention to the unusual characters that make it "perhaps the most aberrant of the citrus fruits."   Swingle had the good fortune to be able for several consecutive years to study in minute detail the leaf and flower morphology of a healthy, everbloomingtree of this plant growing in the citrus greenhouse of the former Bureau of Plant Industry.   As a result of these studies Swingle considered Merrillia to be a very anomalous member of the tribe Clauseneae that has probably evolved from a Murraya-like ancestral form.
      Merrillia caloxylon (Ridl.) Swing. Phil. Jour. Sci. Bot. 13:338. 1918. Murraya caloxylon Ridl. Jour. Straits Branch Roy. Asiat. Soc. 50:113-14. 1908. Illus. Swingle, loc. cit. pls. 5, 6; Foxworthy, Malay. For. Rec. 3:169, unnumbered fig. 1927;
fig. 3-10 this work.
      Type.—Upper Perak, Kenering, at 500 ft. elevation (Robinson, No. 5548).   Herb. Brit. Mus., London.
      Distribution.—Malay Peninsula: southern Thailand, Patani Province; Federation of Malaysia: Kedah, Perak, Selangor, Pahang; northern Sumatra: Asahan.
      Common name.—Flowering merrillia (or ketengah).
      The original description of this species (Ridley, l.c.) follows: "A tree of considerable size, the branches covered with a pale flaky bark.   Leaves 8 inches [20.3 cm] or more long, with 13 leaflets; rachis flattened and winged, narrow; leaflets 3-3 1/2 inches [7.6-8.8 cm] long or less by 1 1/2 inches wide [3.8 cm], alternate, oblanceolate, obtusely acuminate with a triangular base, minutely petiolate, inequilateral, thin, bright deep green.   Flowers pale yellowish-green, several together in small panicles, in the upper axils of a branch, about an inch [2.5 cm] long.   Sepals connate, ovate-acute, 1/10 inch [2.5 mm] long.   Petals and stamens not seen.   Ovary stalked, hairy, style rather stout, hairy, stigma capitulate.   Fruit oblong, rounded at both ends, 4 inches [10.2 cm] long and three inches [7.6 cm] in diameter, the pericarp dotted and warty, greenish, eventually becoming yellow, half an inch [12.7 mm] thick, lemon yellow inside, full of long resin cells narrowed at the mouth and dilated below, cells 5, with rather thick tough walls, pulp of transparent flattened sticky fibers, olive green in colour and tasteless.   Seeds numerous, about 5 in a section, ovate-flattened, half an inch [12.7 mm] long, 1/8 inch [3 mm] thick, olive gray."
      Open flowers of this species were not seen by Ridley.   As Swingle reported in the original description of Merrillia, petals from flowers of a tree in the Singapore Botanic Gardens grown from seed from Patani Province, in extreme southern Thailand, were "35-40 X 5-10 mm, bluntly pointed at the apex and narrowed gradually to the base."   A small tree grown in the citrus greenhouse at Washington, from seeds of this tree in the Singapore Botanic Gardens, bore flowers that were even larger, having petals 4.5-5.5 cm long and 8-11 mm wide.   The flowers were pendent and never opened fully, having the imbricate petals curving outward at the tips and making a trumpet-shaped flower, greenish-white on the outside and white within.   No other species of the orange sub-family has flowers of this type.   Swingle directed attention (1918, p. 339, pl. 5) to the curious pair of nearly opposite, small, broad, and often suborbicular leaflets which are borne very near the base of the rachis, much nearer than in any other pinnate-leaved species in the orange subfamily (fig. 3-10).
      This remarkable species, called ketenggah in the Malay Peninsula, resembles Murraya paniculata in its general habit of growth and in its leaves, and even in the texture and color of its wood.   These two species are sometimes confused by the natives under the name kemuning, which properly applies to Murraya paniculata.   The flowers of Merrillia are entirely different, and its fruits are still more unlike those of Murraya, so it stands out taxonomically as a strikingly different plant obviously belonging to a distinct genus.
      This tree well deserves to be grown in greenhouses or out of doors in warm subtropical climates as an ornamental.   When a small tree, 3 to 3.5 m high, it flowers abundantly at frequent intervals, almost all the year round.   Its curious, pendent, trumpet-shaped flowers, although not conspicuously colored, are attractive because they are produced in such abundance and are pleasantly fragrant.
      In his notes following the original description of Murraya caloxylon Ridley stated that it "is famous in the Malay Peninsula for its beautiful wood," which "is of a bright yellow color ornamented with dark brown streaks and stains, fairly hard in texture and taking a good polish."   It is said to be highly prized by the Malayans for making walking sticks, kris handles, and other small objects.   A British official made beautiful furniture, boxes, etc., out of the wood.   Foxworthy (1927, p. 170) considered it a slow-growing species and thought that it would require about thirty years to reach commercial size.   It is so rare in the wild state that any considerable supply of wood would have to come from cultivated plants.
      This species, or one very closely related to it, has been found in northeastern Sumatra directly across the Malacca Strait from its habitat in the Malay Peninsula.   Mr. B. A. Krukoff, in November-December, 1932, collected good fruiting material of this tree "in primary forests on ‘red’ soil" at Hoeta Padang, near Asahan, northeastern Sumatra (Krukoff, No. 4319, in Herb. Arnold Arbor.).   Krukoff noted on the label "tree 110 ft. [33.6 m] high," i.e., nearly twice the height yet reported for M. caloxylon growing in the Malay Peninsula!   An earlier collection made at the same place in December, 1930, by Krukoff was from a small tree only 15 cm in diameter.   Mr. C. G. G. J. van Steenis collected in 1932 a leafy branch of Merrillia at 120 m altitude on Mount Maligas, Simeloengoen, eastern Sumatra (van Steenis, No. b.b. 20415, in Herb. Arnold Arbor.).   No other tree in the orange subfamily has been found to attain the height reported by Krukoff.   If such giant trees produce wood as beautiful as the smaller form growing in the Malay Peninsula, it would be worth while to test them under cultivation.

TRIBE II.   CITREAE: CITRUS AND CITROID FRUIT TREES
      The tribe Citreae includes Citrus and all of its near relatives as well as many more remote relatives that have not yet been hybridized with Citrus or grafted on it.   The Citreae may be described as follows:
      Small trees or woody, clambering lianas almost always with single or paired axillary spines at the nodes of the twigs; leaves persistent (except in three monotypic genera, Poncirus, Aegle, and Feronia), usually simple or 1-3-foliolate (usually pinnately 3-foliolate with a short petiole and the middle leaf larger, but in Luvunga palmately 3-foliolate with a very long petiole) or sometimes odd-pinnate with strictly opposite pinnae; petioles often winged; flowers in axillary clusters, or in terminal or axillary panicles or corymbs; flowers white or greenish, usually fragrant; calyx with 3-5 sepals or lobes; corolla imbricate in aestivation, with 3-5 (rarely 6) petals; stamens 2-4 or more times as many as the petals; ovary 2-18-locular, ovules 1-18 (usually 1-12) in each locule (22 or more in Burkillanthus); fruits usually orange- or lemon-like in external appearance, often small, sometimes small red or nearly black juicy berries; in the subtribe Citrinae the locules contain pulp-vesicles filled with juice but in the other two subtribes (Triphasiinae and Balsamocitrinae) the locules lack pulp-vesicles, but are usually filled with a mucilaginous gum; seeds with one or several embryos; if with several, only one (rarely two) is a true embryos, the others being false embryos that develop from nucellar buds that grow into the embryo sac; germination hypogeous or epigeous.
      Almost all the species develop single or paired spines in the axils of the leaves of vigorous shoots, especially on young trees.   The leaves are simple, unifoliolate, or trifoliolate in most of the genera, but a few genera, Hesperethusa, Feroniella, and most of the species of Citropsis have odd-pinnate leaves with the leaflets attached to the rachis in strictly opposite pairs (very different from the alternately attached leaflets of the tribe Clauseneae).   Such odd-pinnate leaves often have the petioles and the segments of the rachis between the pairs of leaflets broadly winged.   The rachis is articulated at the point of attachment of each pair of pinnae and these articulated segments fall apart when the leaf finally falls off.   The leaves of Luvunga are palmately trifoliolate with very long petioles and pulvini at the bases of the petiolules and of the petioles.   Three genera, Poncirus, Aegle, and Feronia, have deciduous leaves; all the other genera have persistent (evergreen) leaves.   The species with trifoliolate and unifoliolate leaves usually have winged petioles, whereas the species with simple leaves usually have wingless petioles, although the margins of the leaf blade are sometimes decurrent on the petiole, for part of its length.   The ovary has 2 to 18 or 20 locules, each with 1 to 18 ovules (22 or more in Burkillanthus) in each locule.   The ovaries with 2 to 5 locules usually have one or two ovules in each locule (except in Burkillanthus); those with 6 to 18 locules usually have 4 to 16 or even 18 ovules in each locule.   The fruits are often small, but they are large or very large in Citrus and in Burkillanthus and in several of the genera of the subtribe Balsamocitrinae.
      In the course of an investigation of the vascular traces of all obtainable genera of the orange subfamily, Tillson (1938, pp. 12, 25; Tillson and Bamford, 1938, pp. 785, 789, 791) found that most of the genera of the tribe Citreae studied (21 out of 24) showed fusion of the bundles which arise from the axis and extend to the sepals and petals; "the sepal midribs carry out the lateral petal bundles fused to them, and in some species the lateral sepal bundles are fused in varying degrees to the petal midribs."   The three closely related genera Aegle, Afraegle, and Aeglopsis are without fusions of either kind.   None of the many species belonging to the five genera of the tribe Clauseneae showed fusions of sepal and petal bundles.
      The tribe Citreae is distributed in southeastern Asia and the Monsoon region to the southeast as far as eastern Australia, New Caledonia, New Guinea, and the Melanesian islands; also in tropical (and subtropical) mountainous regions in eastern, central, and western Africa.
      The tribe Citreae contains, as here treated, twenty-eight genera with 124 species, which fall very naturally into three subtribes.   A key to the subtribes of Citreae is presented.

Subtribe 1.   Triphasiinae: Minor Citroid Fruit Trees
      Small trees or woody climbers, usually with single or paired spines in the axils of the leaves, but often spineless; leaves pinnately 3-foliolate (palmately 3-foliolate in Luvunga), often 1-foliolate or even simple; fruits small, often orange- or lemon-like in appearance, rarely small red or black juicy berries; fruit locules without pulp-vesicles, but usually filled with mucilaginous gum; ovary subsessile or more or less stalked, borne on a disk or on a longer cylindrical gynophore with 2-5 (rarely 1) or even 6 or 7 locules, and 1-6 or 8 ovules in each locule; fruits with gland-dotted peel, sometimes soft and berry-like, usually firm and often resembling miniature oranges or lemons but always lacking pulp-vesicles.
      Most of the eight genera of the subtribe Triphasiinae contain species with one or two axillary spines.   However, all four species of the genus Oxanthera are thornless and Wenzelia has six thornless species (also two with single spines and one with paired spines).   All the species of this subtribe have persistent evergreen leaves but they vary greatly in the different genera: Wenzelia, Monanthocitrus, Merope, and Pamburus have simple leaves with very short, wingless, non-articulated petioles; Triphasia has pinnately trifoliolate or simple leaves; Luvunga has palmately trifoliolate leaves with the leaflets attached by pulvinoid petiolules to a long petiole with a pulvinus at the base; Paramignya has unifoliolate leaves with wingless petioles usually pulvinoid at the junction with the blade (the two last-named genera have petioles articulated with the leaf blade); Oxanthera has three species with unifoliolate leaves with articulated petioles and one simple-leaved species with nonarticulated petioles.
      The eight genera of this subtribe are all native to an area partly superimposed on that in which the True Citrus Fruit Trees occur, but this area, instead of being barrel-shaped, is rather a long, obtuse-angled triangle with its acute apex cut off, having a land base in Asia extending about 4,600 kilometers from western India to southeastern China and Hainan Island and reaching some 10,000 to 11,000 kilometers to the southeast into Polynesia.   This area falls mostly in the Monsoon region.   Its middle portion is about 3,000 kilometers wide from Sumatra to the Philippines and its blunted southeastern apex, which is bounded by New Caledonia and the Fiji Islands, is about 1,800 kilometers wide.
      Of the eight genera included in this subtribe, only two, Oxanthera and Monanthocitrus, have a very restricted range, Oxanthera being found only at the very point of the triangle, in New Caledonia (and near-by Art Island), and Monanthocitrus in west-central New Guinea.   Merope (which has leaves and petioles resembling Wenzelia) occurs in brackish coastal swamps from northwestern New Guinea and Amboyna Island to the southwestern Philippines, Java, Malay Peninsula, Burma, and Bengal.   Wenzelia, Merope, and Monanthocitrus all include New Guinea in their range and may possibly have originated there.   These three genera and the related genus Oxanthera are the only ones in the subtribe with more than two (4 to 8) ovules in each locule.   The other four genera (all with one or two ovules in each locule) seem to have originated in southeastern Asia and to have spread a greater or less distance into the East Indian Archipelago: Triphasia and Luvunga range clear through to New Guinea; Paramignya ranges to Sumatra and the Philippines; whereas Pamburus has spread only as far as Sumatra, Java, and Borneo.
      The subtribe Triphasiinae contains three fairly distinct groups of genera, although most of the genera included differ strikingly from one another.   Furthermore, within a given genus, the species are far from uniform, many of them differing in characters usually considered to be of high taxonomic value.   The genus Oxanthera, for example, has only four species, but they run the whole gamut of ovarial morphology: O. aurantium has a hypomerous ovary with two locules; O. fragrans has an isomerous ovary with 5 to 6 locules; whereas O. neo-caledonica has a hypermerous ovary with seven locules.   Several other genera in the subtribe have aberrant species and doubtless others will be discovered in the future.   No other subtribe in the whole orange subfamily gives such strong evidence of containing species that have been for ages and probably still are rapidly evolving into new forms, new species, and new genera.
      The three groups of genera within the subtribe Triphasiinae are indicated in the key to the genera and are described briefly in the text.   The Wenzelia group has four genera, all with more than two ovules (4 to 8, usually 6) in each ovary locule and long, simple or unifoliolate leaves, with very short nonarticulated petioles except in Oxanthera.   The Triphasia group contains two genera, the species of which are all small trees or shrubs with short nonarticulate petioles.   Two species of Triphasia have simple leaves and one has trifoliolate leaves; the other genus in the group, Pamburus, has simple, semicoriaceous thick leaves.   The Luvunga group has two genera, both woody lianas that clamber over trees by means of recurved or hooked spines, with small or medium-sized ovate or subglobose fruits that look like miniature limes.
      The remote ancestors of our cultivated citrus trees doubtless were very like some of the plants now found in this subtribe.   Doubtless several tens of millions of years have elapsed, however, since the subtribe Citrinae arose from some ancestral species similar to those now classed in Triphasiinae.
      The forty-six species known in Triphasiinae make it the second largest of the three subtribes in the tribe Citreae.   It contains the nearer wild relatives of our Citrus Fruit Trees, all of which should be introduced into culture, studied, tested and utilized as rootstocks or in other ways by citrus growers and citrus experts the world over.   It is noteworthy that no fewer than twenty-nine species (nearly three fifths the total number!) have been described as new and also that three of the eight genera have been established since 1900.   As the rich flora of the Monsoon region is more and more actively assembled by competent collectors and studied by expert taxonomists, there is good reason to expect that many more species and doubtless some new genera will be discovered in the future.

GROUP A.   THE WENZELIA GROUP
      The Wenzelia group consists of four genera: Wenzelia, Monanthocitrus, Oxanthera, and Merope.   The first three have six ovules in each ovary locule and the last one has four ovules to each locule.   The leaves of the plants of all four genera are simple or unifoliolate.   Wenzelia, Merope, and Monanthocitrus all occur in New Guinea and are closely related genera.   Oxanthera, found in New Caledonia, in spite of its long isolation in both time and space, is still obviously related to Wenzelia.   Merope has leaves very like those of Wenzelia and Monanthocitrus, but each ovary locule, so far as known, contains four (or by abortion two?) ovules and its curious flattened seeds, the largest known in the orange subfamily, are of unusual shape, being more or less caudate.   Merope is doubtless an early offshoot from an ancestral form of the Wenzelia group.   Wenzelia and Monanthocitrus differ greatly from all the other Minor Citroid Fruit Trees in having large leaves with extremely short petioles not articulate with the leaf blade.
      In Oxanthera the leaves, the ovaries, and the stamens have varied from the primitive pattern found in Wenzelia and Monanthocitrus.   In Merope the seeds have evolved along new lines; in Monanthocitrus the seeds have continued to evolve along lines shown by certain flat-seeded species of Wenzelia until they now show characters found nowhere else in the orange subfamily.   Clymenia, a True Citrus Fruit Tree, has leaves very like those of Wenzelia but the fruit shows stalkless pulp-vesicles that pave the way to Citrus.
      All the plants of the Wenzelia group should be studied attentively to throw all the light possible on the origin of Citrus.

VI.   Wenzelia Merr.
      VI.   Wenzelia Merr. Phil. Jour. Sci. Bot. 10:272. 1915.
      Type species.Wenzelia brevipes Merr., from Leyte Island, Philippines (Wenzel, No. 998).   Herb. Bur. Sci., Manila.
      Distribution.—Southern Philippines, New Guinea, Bougainville island, Solomon Islands, Fiji Islands.
      Common name.—Wenzelia.
      Shrubs with unarmed twigs or with single or paired spines in the leaf axils; leaves large (7-28 cm long), oblong or lanceolate; petioles very short (2-5 mm long), wingless but channeled above, not clearly articulated with the leaf blade; flowers often very large; solitary in the axils of the leaves, white; calyx 5-lobed; petals 5, oblong-lanceolate or ovate, obtuse or subacute, glandular-punctate; stamens 10, free, anthers long; ovary oblong-ellipsoid, 4-5-locular with 6 (or sometimes 8) ovules in each locule, narrowed below into an annular or shallow cupulate disk; fruit obovoid or ellipsoid, 3-5 cm long, with coriaceous, glandular pericarp, segments containing mucilage without pulp-vesicles.
      The Wenzelia genus is very different from all the other Minor Citroid Fruit Trees of the subtribe Triphasiinae except Monanthocitrus in having large, or very large, simple leaves, with very short wingless petioles not articulated with the leaf blade, and in having six or eight ovules in each locule of the ovary.   The fruits are ellipsoid and the locules contain several seeds (up to six or eight) immersed in a mucilaginous material in young fruits.
      Nine species of Wenzelia are known, six from New Guinea, one from the southern islands of the Philippines, one from the Solomons, and one (the easternmost large-seeded member of the orange subfamily) from Fiji.   Seven of the species fall into two groups constituting the two subgenera described below, leaving two species still inadequately known and of uncertain relationships.
      Wenzelia is the largest and most widely distributed of the six genera limited to the southern part of the Monsoon region, Melanesia, and the Fiji Archipelago, the southeastern area of distribution of the orange subfamily.   Three of these six genera, Clymenia (Bismarck Archipelago), Monanthocitrus (New Guinea), and Eremocitrus (northeastern Australia), are monotypic; Oxanthera (New Caledonia) has four species; Microcitrus (northeastern Australia and southern New Guinea) has six species; Wenzelia (the Philippines, New Guinea, the Solomons, and Fiji) has nine species!
      A striking proof of the recent rapid advance of our knowledge of Citrus relatives is provided by the fact that all the nine species now constituting the genus Wenzelia have been described since 1900.   The first species was published in 1901, the second in 1910, the third in 1915 (when the genus Wenzelia was established), a fourth in 1918, and the other five were described early in 1940 (see Swingle, 1940a).
      It is very probable that still other species of Wenzelia will be found when New Guinea, the Spice Islands (the Moluccas), and the Melanesian islands to the east of New Guinea are explored more thoroughly.   Keys to the subgenera and species of Wenzelia are presented.

Subgenus Wenzelia
      Subgenus Euwenzelia Swing. in Webber & Batchelor, Citrus Indus. 1:216. 1943.
      This subgenus includes Wenzelia brevipes, W. tenuifolia,and W. kambarae.   These species occupy the two extremes of the area of distribution of the genus, as well as a midway location in New Guinea: W. brevipes (the type of the subgenus) is native to the southern islands of the Philippines, W. kambarae to the Lau Group of the Fijis, and W. tenuifolia to southeastern New Guinea.
      These species probably resemble the ancestral type of the subgenus Wenzelia, whereas those of the subgenus Papualimo represent a more highly specialized type of later origin.
      1.   Wenzelia brevipes Merr. Phil. Jour. Sci. Bot. 10:272. 1915.
      Type.—Leyte Island, Buena Vista near Jaro, Philippines (Wenzel, No. 998).   Herb. Bur. Sci., Manila.
      Distribution.—Southeastern Philippines: Leyte, Samar, and Bohol islands and eastern Mindanao Island.
      Common name.—Philippine wenzelia.
      Merrill’s original extended description in English, following the Latin diagnosis, reads as follows: "An erect, glabrous, unarmed shrub about 1 m high, the trunk about 3 cm in diameter.   Branchlets slender, terete, greenish-olivaceous when dry.   Leaves alternate, simple, presenting no indication of a joint between the lamina and the petiole, firmly chartaceous to subcoriaceous, oblong, pale when dry, shining, 14 to 25 cm long, 5.5 to 6.5 cm wide, distinctly glandular-punctate, entire, apex acuminate, base obtuse to rounded; lateral nerves about 10 on each side of the midrib, prominent on the lower surface, arched-anastomosing, the primary reticulations lax, prominent; petioles 2 to 4 mm long.   Flowers white, fragrant, axillary, solitary, their pedicels about 1 cm long with 2 basal, oblong-ovate, 2 mm long bracteoles.   Calyx 5 to 6 mm long, cup-shaped or somewhat campanulate, 5-lobed, glandular-punctate, the lobes very broadly ovate, rounded, about 2 mm long, wider than long.   Petals 5, imbricate, up to 3.4 cm long and at least 8 mm wide, oblong-oblanceolate, obtuse, narrowed below, prominently glandular-punctate.   Stamens 10, free; filaments linear, 1.4 cm long, equal, rather slender, cylindric; anthers oblong, basifixed, 2-celled, about 3.8 mm long.   Disk somewhat cup-shaped, 1.5 mm in diameter, forming a short gynophore.   Ovary oblong-ellipsoid, glandular, about 4 mm long and 2.5 mm in diameter, 4- or 5-celled, narrowed below into the 2 to 3 mm long stalk, and above into the cylindric style, which is about 9 mm long, continuous with the ovary, ultimately deciduous; stigma subcapitate, very obscurely lobed; ovules 6 in each cell, 2-seriate, axile.   Fruits obovoid, 3-4 cm long, base narrowed, the pericarp coriaceous-fleshy, glandular; seeds few, apparently 3 or 4 in each fruit, the cotyledons thick, fleshy, oblong-obovoid, about 1.5 cm long and 1 cm wide."
      Some of the later collections show broader leaves than those of the type specimen described by Merrill.   For example, a specimen from Bohol, collected by M. Ramos (No. 43189), has leaves 13.0 to 17.5 by 6.5 to 8.5 cm, acuminate at the tip, broadly rounded or even slightly cordate at the base, petioles very short, 4 to 5 mm long, 2 to 2.5 mm in diameter, more or less longitudinally rugose, not articulated with the leaf blades; flower buds single, axillary, 15 by 4 to 5 mm, pedicel 1 cm long, slender (0.6 to 0.8 mm diam. at base, 1 to 1.3 mm diam. at tip).
      1a.   Wenzelia brevipes var. alabatensis Swing. Jour. Arnold Arbor. 21:16. 1940.
      Type.—Philippines, Alabat Island (Ramos and Edaño, No. 48054, Bur. Sci.).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—Known only from type locality.
      Leaves narrow (22-28.5 X 5-6.6 cm), much narrower than those of the species; oil glands less conspicuous, especially on lower surface.
      This variety occupies the northernmost point of the range of the species W. brevipes.   Like the species itself and like all other species of the genus Wenzelia, it is native in an island on the western boundary of the Pacific Ocean.   There is some variation in the shape of the leaves in the species of the genus Wenzelia, but no other species shows so extreme a variation as that represented by this variety with its extremely long and narrow leaves, also peculiar in their less conspicuous oil glands.   It may prove to be an extreme form connected with the typical form of the species by intergradations.
      2.   Wenzelia tenuifolia Swing. Jour. Arnold Arbor. 21:14. 1940.
      Type.—Southeastern New Guinea, Papua, Boridi, forests, alt. about 1,220 (or 3,353) meters, "c[irca] 4,000 [or 11,000] feet" (C. E. Carr, No. 14881, 2 branches, 1 fruit, February 1, 1935).   Herb. Bot. Mus., Berlin-Dahlem; photographs and serial microtome sections, Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known only from the type locality.
      A spineless shrub, 2 m high; ultimate branches slender, 1.1-1.8 mm in diam., internodes 2-3 (rarely 5) cm long; leaves very thin, brittle when dry, more or less broadly elliptical 10-13 cm long including petiole, 5-7.2 cm wide, apex bluntly rounded and often imperfectly developed, the very tip being stunted and more or less torn, sometimes more or less irregularly emarginate, broadly cuneate at the base, margins entire, lateral veins, 10-12 pairs, arising at angles of 65°-80° with the midrib, petioles 4-6 X 1-1.1 mm; fruit ovate, 3.5 cm long, 3 cm broad, 5-loculate, radial locule walls thin, 1-1.5 mm thick but expanding into a thick rib of tissue running longitudinally in the fruit, more or less triangular in cross section, 4-7 mm in radial thickness, dorsal locule walls only 1-2 mm thick midway between the radial walls; seeds dull brown, ovate, compressed, 17-19 X 12-15 X 3.5-5 mm, more or less wedge-shaped, with a faintly, more or less concentrically marked and corrugated edge, 1-2.2 mm wide and 1-1.5 mm thick, on one side of the seed, with faint markings 0.2-0.4 mm apart, 1-4 in a locule, often occurring in closely approximated pairs; embryo monoembryonic.
      This species was known to Swingle only from the rather scanty type specimen, but it is so evidently distinct from all the other known species that he had no hesitation in making it a new species.   It is somewhat of a connecting link between the two subgenera Wenzelia and Papualimo but apparently falls in Wenzelia, as the seeds lack the paper-thin, more or less torn membrane along the free margins of the seeds found in the species of the subgenus Papualimo.
      The fruits of W. tenuifolia are circular in cross section and show five more or less triangular peripheral ribs at the distal ends of the radial locule walls, which fill the spaces between the strongly bulging dorsal locule walls.   In striking contrast to this species, W. archboldiana has fruits which are bluntly star-shaped in cross section because of the absence of any tissues filling the interlocular furrows.
      The collector’s label on the type specimen is a carbon copy (the original label was doubtless used for another specimen of the plant).   The altitude given on the label apparently reads "c. 11,000 feet" but may possibly be "c. 4,000 feet."   If the altitude is 4,000 feet (1,220 meters), it would be far above the usual upper limit for the aurantioid plants in New Guinea (about 300 meters, or 984 feet).   If the altitude is 11,000 feet (3,353 meters), it would equal the highest altitude as yet reported for any rutaceous plant from New Guinea.3
      As all the other species of Wenzelia grow at low altitudes in tropical regions, it will be of interest to learn more about the distribution of W. tenuifolia.   Perhaps its curious unusually thin leaves may be adapted for growth in very humid locations, such as occur in the cloud belts of mountains.
      3.   Wenzelia kambarae Swing. Jour. Arnold Arbor. 21:12. 1940. Atalantia sp. Albert C. Smith, Fijian Plant Studies Bul. 141:77. 1936. Illus. Swing., loc. cit. pl. 3, figs. 6-8.
      Type.—Fiji Archipelago, Lau Group, Kambara Island, limestone formation, alt. 10-100 m (A. C. Smith, No. 1265).   Herb. New York Bot. Gard.
      Distribution.—Fiji Archipelago: Lau Group, Kambara Island (A. C. Smith, Nos. 1265 and 1293), and Naiau Island (Zimmerman).   Herb. Natl. Arbor., Washington, D.C.
      Common name.—Fiji wenzelia (or moli moli).
      A spineless shrub or slender, small tree, 4-7 m high, trunk 4-8 cm diam., twig slender, 1-3 mm diam., glabrous, at first green but soon turning light brown and showing numerous irregular, more or less anastomosing, longitudinal, very narrow ridges; leaves thin, chartaceous, oblong to elongate-elliptical, varying considerably in size and shape, 6-21 X 3-8.5 cm, bluntly pointed or rounded at apex, broadly cuneate or rounded at base, margins entire or irregularly and shallowly crenate (Zimmerman's Naiau material), lateral veins very distinct on under surface, less distinct above, 8-13 on each side arising at rather widely varying angles (60°-80°) with the midrib, branching dichotomously at 3/5-4/5 of the distance to the margin, petioles 5-7 X 1.5-2 mm, tapering into the leaf blade with a very shallow channel on the upper side formed by the decurrent leaf margins, petioles on old leaves often brownish and irregularly ridged or rugose, as is the midrib on the lower surface; flower buds (coll. by Zimmerman, Naiau Island, very young) 2.5 X 2 mm, borne on slender pedicels 8-9 mm long, bearing several minute pointed bracts on the lower portion; fruits orange colored, 3-4 cm diam., 3- (or 4-?) locular with a gland-dotted but smooth peel, borne on pedicels 14-16 X 1.8 mm, with a cylindrical carpophore 1.2-2.5 mm long between the persistent calyx and the base of the fruit; seeds very large, ovoid, 18-25 X 12-15 mm, testa smooth, shiny, cream-colored, embryos 1 or 2 (sometimes more?), greenish-colored.
      This species is the easternmost large-seeded Citrus relative that occurs on any Pacific island.   Its seeds are so large that it is improbable that they would have been carried to Fiji by birds, especially since the fruits contain no pulp and little if any mucilaginous matter.   It is highly probable that the moli moli reached the Fiji Archipelago from the center of origin of the genus Wenzelia (probably New Guinea) by slowly spreading eastward over dry land, a migration that probably required millions of years to accomplish.   Its nearest known relative is W. melanesica, native to Bougainville Island in the Solomons, about 1,800 kilometers to the west-northwest.   That dry land once extended from the west as far east as Fiji in the early Tertiary epoch is considered certain by some geologists ((Brock, 1924, p. 75).
      Dr. Edward C. Zimmerman, who obtained flower buds of this plant by "shooting" the dangerous surf on the atoll surrounding Naiau Island, thus making identification possible, wrote Swingle that the natives of the Lau Group know the plant well by the name moli moli; furthermore, they told him that there were pink and red flowers, perhaps in different stages of development (all other Wenzelia flowers so far observed are white).   The trees on Naiau Island were taller (5 to 7 m) than those reported from Kambara Island (3 m), but the leaves of the Naiau plants were much smaller (6 to 10 by 2.5 to 5 cm) than those of one of the collections (A. C. Smith, No. 1265) from Kambara (11.5 to 21 by 4 to 8.5 cm).   However, another collection from Kambara Island (A. C. Smith, No. 1293) shows smaller leaves (8 to 12 by 2.5 to 5 cm).   Probably W. kambarae is a variable species showing many forms.

Subgenus Papualimo
      The subgenus Papualimo includes four species, W. platysperma (the type of the subgenus) W. paludosa, W. melanesica, and W. archboldiana, all native to New Guinea or the Solomon Islands to the eastward.
      Wenzelia paludosa (Lauterb.) Tan., which has very thin seeds (only 1 to 2 mm thick), almost certainly falls in the subgenus Papualimo, although nothing is known about the margination of the seeds, a striking character of the other species of this subgenus.
      The species constituting the subgenus Papualimo represents a striking new evolutionary tendency as shown in the development of thin seeds with very thin, laciniate membranous margins—seeds that differ widely from the ancestral type of the genus Wenzelia and from all the other genera of the orange subfamily.   This evolutionary tendency has apparently reached its highest development in the very strange monotypic genus Monanthocitrus, also native to New Guinea, which has still thinner, still more elaborately margined seeds.   The seeds are also spotted, unlike those of any other species known in the orange subfamily.
      4.   Wenzelia platysperma Swing. Jour. Arnold Arbor. 21:11. 1940. Illus. Swing. loc. cit. pl. 3, figs. 4, 5.
      Type.—New Guinea, Papua, Palmer River, 2 miles below junction of Black River, alt. 100 m, flood-plain forest undergrowth (L. J. Brass, No. 7111, branch with fruit), Herb. Arnold Arbor., Harv. Univ., Cambridge; also found in the ridge forest undergrowth at the same locality (Brass, No. 7025, branch with fruit), Herb. Arnold Arbor.
      Distribution.—Known only from the two above-cited collections, made very close together.
      A thornless, loosely branched, small tree, 3 m high; twigs slender, 1-2 mm diam., internodes 2-3 cm long, often bent slightly at each node alternately to the right and left, glabrous or very minutely and sparingly pubescent; leaves chartaceous, elongate-elliptical, 10-20 X 2.5-5.5 cm, acuminate or caudate at apex, cuneate, or broadly rounded at base, margins irregularly crenulate (and also denticulate in Brass No. 7025), lateral veins 12-14 on each side, arising at angles of 70°-80° with the midrib, branching dichotomously at 1/4-3/4 of the distance to the margin, petioles very short, 3-4 X 1.4-1.8 mm (1/40-1/60 of the length of the leaf blade), with a shallow channel on the upper side; flowers not seen, calyx persistent, subtending the fruit, deeply lobed, lobes 2 X 2 mm, broadly rounded, minutely puberulent (glabrous and apparently slightly fleshy in Brass No. 7025); fruits subglobose or ovate, 4-4.8 X 3.5 cm, 4-locular (?), reddish or pink, borne singly on a slender pedicel 10-12 mm long, arising in the axil of the uppermost leaf; seeds flattened, light brown, 12-15 X 7-9 X 2-3 mm, testa rugose on edges or else tapering into irregularly shaped extensions with somewhat torn margins, monoembryonic, cotyledons greenish.
      This remarkable species is known only from two single fruiting twigs.   Both collections show zigzag slender twigs each with a single fruit borne in the axil of the uppermost leaf.   The seeds are even flatter than those of W. archboldiana, but otherwise they are much the same.   The globose or ovate, pink or reddish, gland-dotted, aromatic fruits hang down at the ends of the fruiting branches.   There is no trace of pulp or mucilage between the crowded seeds that lie lengthwise in the carpels.
      5.   Wenzelia paludosa (Lauterb.) Tan. Jour. Arnold Arbor. 9:139 1928. Citrus paludosa Lauterb. Bot. Jahrb. 55:263. 1918.
      Type.—Northeastern New Guinea, in swampy primeval forest with sago palms and rattan at mouth of Sepik River (Ledermann, Nos. 7173 [flowers] and 7507a [fruits]).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Northeastern New Guinea: known only from the Sepik River.
      The original description of Lauterbach, translated, reads as follows: "Shrub or small tree, with smooth, slender, terete branches, and stout solitary spines; petioles short, caniculate above; leaves lanceolate, gradually narrowly acuminate, acumen obtuse, emarginate [at the tip], slightly rounded or subacute at base, chartaceous, glabrous on both sides, opaque, dull green when dry, 12-14 lateral veins, oblique, dividing near the margin, many [of them] arched-anastomosing, prominent above when dry like the midrib, margins undulate-crenate, minutely subdenticulate; flowers axillary, solitary; calyx 5-lobed, lobes ovate, rounded, ciliate; petals 5, ovate, subacute, glandular on outer side; stamens 10, filaments free, anthers linear; disk annular; ovary flask-shaped, locules many-seeded; style thick, stigma 5-lobed; fruit ovate, subacuminate, exocarp moderately thick, glandular, pulp scarcity visible; seeds up to 6 in each locule, irregularly triangular or quadrangular, much flattened, testa white, coriaceous, slightly rugose.
      "Shrub, 1-2 m high, or small tree, with light gray bark.   Branches 2-3 mm thick; petioles 2-5 mm; leaves 7-14 cm long, 2-5 cm wide; spines 10-12 mm; flower pedicel 6 mm.   There are only 2 not fully developed [flower] buds that measure 3 mm.   Fruit pedicel 10 mm; the ripe fruit, which smells like an orange, is 4 cm long by 3 cm diam.; the seeds 10-12 mm long, 1-2 mm thick."
      A later collection (Ledermann, No. 12252) was made at a hill camp on the Sepik River.   The type specimen showing fruits (Ledermann, No. 7507a) has pedicels of the young fruits 10 to 12 mm long and unusually slender for a species of Wenzelia, being only about 0.5 mm in diameter in the middle, tapering to about 1 mm at the base and also at the top where it bears the calyx without any gradual expansion into the funnel-shaped base of the calyx as in W. melanesica.   The young fruit is ellipsoid, 14 by 5 mm, narrowed into a slender stalk at the base, 3 mm long by 1.5 mm thick, and attenuate at the apex into a beak about 2 mm long and 1 mm wide.   The seeds are flattened, as are those of W. platysperma and W. archboldiana from the eastern half of New Guinea and W. melanesica from the Solomons, but unfortunately it is not known whether they are margined or not, and if so with what kind of a margin.   The petioles are very short, 3 to 4 mm long and 1.5 mm wide, with a shallow channel on the upper side that is broad at the tip and narrows regularly toward the base.   The lateral veins arise at a highly variable angle (40° to 65°) with the midrib, and branch dichotomously at from slightly less than one half to nearly three fourths of the distance from the midrib to the margins of the leaf.
      This species seems to be clearly distinct from all the others in the genus.
      6.   Wenzelia melanesica Swing. Jour. Arnold. Arbor. 21:9. 1940. Illus. Swing. loc. cit. pl. 2, figs. 6-8; pl. 3, figs. 1-3.
      Type.—Solomon Islands, Bougainville Island, Kugu-maru, Buin (Kajewski, No. 1907a).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—Known only from the type locality.
      Small thornless tree (up to 7 m); twigs glabrous, slender, 1-2.5 mm diam.; leaves thin, chartaceous, glabrous, elliptical, with very numerous minute oil glands evenly distributed over the whole surface, acuminate at the apex, which ends in a sharp point, broadly and evenly rounded at base, 10-16 X 3.5-5.5 cm, margins entire or slightly crenulate; lateral veins clearly marked below, 12-15 pairs, arising at a large angle (70°-80°) with the midrib; petioles very short, 4-5 X 1-1.3 mm, glabrous, not articulated with the blade, with a narrow, very shallow channel on the upper side, the walls of which connect with the leaf margins; flowers solitary or in small groups (2-3) in the axils of the leaves, pedicels with a few sparse, short hairs, short (4-6 mm), with very small ciliate bracts at the base, slender, expanding gradually into the funnel-shaped calyx; calyx 4-5 lobed, lobes broadly rounded or very bluntly pointed, 1.3-2 mm long, 1.5-2 mm wide, and increasingly short-pubescent toward the margins, which are abundantly short ciliate; corolla with 4-5 petals (immature in type specimen), showing (in the bud) numerous oil glands evenly distributed; stamens 8-10; pistil immature, ovary with 4-5 locules, each with 6 ovules; fruits 4-5 angled, tapering to a sharp point at the apex, 4.5 cm long, 2.3 cm wide, peel showing numerous, evenly distributed small (0.1-0.2 mm), slightly sunken oil glands; seeds thin, flat or nearly so, ovate in outline, 9-13 X 6.5-9 X1.5-2 mm, dusky gray, very thin at the edges, with a lighter colored, narrow, usually subentire, marginal membrane; embryo monoembryonic.
      This striking species of Wenzelia extends the range of the subgenus Papualimo from New Guinea to the Solomon Islands (Long. 156° E.).   It resembles most closely W. archboldiana from New Guinea but differs from it in having 4- to 5-angled fruits tapering into a sharp point.
      7.   Wenzelia archboldiana Swing. Jour. Arnold Arbor. 21:7. 1940. Illus. Swing. loc. cit. pl. 2, figs. 1-5.
      Type.—British New Guinea, lower Fly River opposite Sturt Island in rain-forest (L. J. Brass, No. 8038, fruiting branches).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—Southeastern and eastern New Guinea.   In addition to the type, two other collections from northeastern New Guinea were seen: Morobe district, Ulap, 610 m alt. (Clemens, No. 6613, flowering branches), Herb. Natl. Arbor., Washington, D.C.; and Morobe district, above Sattelberg (J. and M. S. Clemens, No. 1057, branches with young fruit), Herb. Bot. Mus., Berlin-Dahlem.
      Common name.—Archbold’s wenzelia.
      A thornless shrub or small tree, 3-5 m high; twigs glabrous, 2-4 mm diam.; leaves ovate-lanceolate or lanceolate, 16-22 X 7-9 cm, more or less abruptly acuminate at apex, cuneate at base, margins entire or slightly crenulate, lateral veins 10-14 on each side, arising at angles of 60°-70° with the midrib, usually branching dichtomously [sic] at 1/2-3/5 of the distance to the margin, petioles very short, 4-8 X 1.5-2 mm (1/25-1/40 of the length of the leaf blade), with a shallow channel on the upper side; flowers usually arising in pairs in the axil of the terminal leaf, flower buds 2.3 X 3-4 mm, pedicels 4-5 X 0.8-1 mm, expanding gradually into the funnel-shaped calyx 4-5 mm long and 3-4 mm wide at the top, calyx lobes short (1 mm long, 3 mm wide), broadly rounded, minutely pubescent on the back and margin; corolla white, petals 4 (or 5), measuring 15 X 4.5 mm in the bud, dotted with small oil glands, glabrous, margins thin and slightly scarious; pistil 12-12.5 mm long immediately after the petals fall, disk short, 0.9-1 mm high, 1.7 mm wide, abruptly constricted to 0.9-1 mm wide at junction with ovary base; ovary clavate with a sterile narrowed base, 6-7 X 1.7-3 mm, 5- (or 4-) locular, with 6 or 8 ovules in each locule, style cylindrical, 4.5-5 X 0.5-0.6 mm, not expanded at junction with ovary tip and only slightly expanded at stigma junction, stigma cushion-shaped, 1 mm high, 2 mm wide; fruits commonly paired in axil of uppermost leaf of fruiting twig, "lime green and smelling of limes," depressed-globose, 2.5-3 X 3-3.5 cm, 5-locular (in type), each locule bulged out (radial diam. 14-15 mm at junction with adjoining locule and 17-18 mm midway between the two radial locule walls) making fruit star-shaped in cross section with very blunt, rounded rays; seeds several (up to 6 or 8), crowded into each locule, flat and more or less curved, arranged more or less radially with their long diameters parallel to the axis, 15-17 X 10-11 X 1.5-3.5 mm, brownish-gray (after soaking in formaldehyde solution), monoembryonic.
      The flowers of this species were described from Clemens No. 6613, as the type specimen showed fruits only.   The most extraordinary character of the species—its numerous flattened seeds—is one which it shares with three other species, W. platysperma, W. paludosa, and W. melanesica.   This flattening of the seeds is undoubtedly not merely a result of the competition for space by the numerous seeds that develop in each locule, but is an inherited adaptation.   These flat-seeded species of Wenzelia are obviously related to Monanthocitrus cornuta, but it must be emphasized that in Monanthocitrus the seeds are not merely very much flattened, but are so compressed and crowded that they are distinctly saucer-shaped, concave on one side and convex on the other, with a thin, more or less fimbriate membrane almost entirely surrounding all but the attached side.   Furthermore, the seeds of Monanthocitrus are conspicuously spotted, a character not found, so far as known, in any other plant belonging to the orange subfamily.
      The mature fruits of the type specimen of W. archboldiana show a curious abnormality, having fusiform cracks in the outer layer of the peel arising usually near the middle of the dorsal wall of the locules just above the attachment to the pedicel.   The split in the peel reveals the more or less dried-up oil glands and other tissues, but the inner peel apparently remains intact.   Sometimes these fusiform cracks occur between the locules of the fruit, although usually not so abundantly.
      The depressed-globose, ribbed, green fruits (said to smelt like limes) usually appear in pairs in the axil of the uppermost leaf on the fruiting twig—in striking contrast to W. platysperma, where the more or less elongated subglobose or ovate pink fruits are borne singly, and hang down apparently at the tip of the twig, although in reality in the axil of the uppermost leaf.
      This species has been named in honor of Mr. Richard Archbold, who organized and led three great collecting and exploring expeditions to New Guinea.   He employed expert collectors, among them L. J. Brass, with the result that many herbarium specimens of the remarkably interesting plants of New Guinea were brought back.

SPECIES INADEQUATELY KNOWN AND OF UNCERTAIN RELATIONSHIPS
      8.   Wenzelia dolichophylla (Lauterb. & K. Schum.) Tan. Jour. Arnold Arbor. 9:139. 1928. Citrus dolichophylla Lauterb. & K. Schum. in Schum. & Lauterb. Fl. Deut. Schutzgeb. 377. 1901.
      Type.—Northeastern New Guinea, tributary river No. 9, of Ramu River (Lat. 4° 20' S., Long. 144° 38' E.), high forest, altitude 60 m (Lauterbach, No. 3108).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Known also from the upper course of the Gogol River, northeastern New Guinea, and a sterile specimen (doubtfully this species) reported by Lauterbach from Etna Bay (Lat. 4° S., Long. 134° 30’ E.) in northwestern Dutch New Guinea (now West Irian, Indonesia).
      The original description, translated, reads as follows: "Shrub with numerous slender young fruiting branches, glabrous, terete; petioles very short, furrowed above; leaves ovate-lanceolate or ovate-oblong, very acutely attenuate-acuminate, glabrous on both sides, coriaceous; immature fruits long-pedicellate, subglobose, somewhat acute, glabrous.   A shrub 2 m high.   The branches are 20-25 cm long and 2-3 mm thick at the base; they have gray to brown bark.   The petiole is 3-5 mm long, no articulation of the leaf blade with the petiole is visible; leaves 15-25 cm long and at the middle or lower third 5.5-8.5 cm broad, with 14-18 pairs of veins arising at right angles from the mid-rib, less prominent above than below; gray above, yellowish below, dull on both sides.   The pedicel is 2-2.5 cm long.   The immature fruits, still green, have a lemon-like odor."
      Lauterbach reported (1918, p. 262) on a specimen from the upper course of the Gogol River (Lauterbach, No. 1088) collected in fruit, November 26, 1890, as follows: "The nearly ripe fruits are red, lemon-shaped, 5.5 cm long and 3.5 in diameter."   These are the largest fruits yet reported in this genus and the red color is also distinctive.   Swingle examined the type specimen in the herbarium at Berlin-Dahlem many years ago and found no trace of pulp-vesicles, the locules of the dry fruits being filled with a brown, brittle, resin-like substance.
      9.   Wenzelia grandiflora (Lauterb.) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Citrus grandiflora Lauterb. Nova Guinea 8:293. 1910; Monanthocitrus grandiflora (Lauterb.) Tan. Jour. Arnold Arbor. 9:139. 1928.
      Type.—Central-western New Guinea [West Irian, Indonesia], swampy primeval forest, Bivak Island, in Noord Fluss [Lorentz River] (Lat. 5° 0' 25" S., Long. 138° 39' E.), (Versteeg, No. 1788).   Herb., Utrecht.
      Distribution.—Known only from the type locality.
      The original description of Citrus grandiflora by Lauterbach reads, translated, as follows: "Shrub 1-2 m; branches slender, spiny, 2-4 mm thick, terete, flattened when young; spines 2, about 1 cm long, borne in the axils of the leaves; leaves lanceolate, alternate, acuminate [at the tip], acumen acute, broadly pointed at base, 15-20 cm long, 5-7.5 cm wide, coriaceous, margin crenulate-subdentate, veins more prominent below, 12-13 pairs, ascending, anastomosing; petioles 5 mm long, subterete; flowers solitary (?), axillary (?), with short pedicels; calyx cupulate, 4 mm long, 5-merous, lobes triangular, subacute or somewhat truncate, 2 mm long; petals 5, white, lanceolate, broadly pointed ('late acutis') [at the tip], attenuate at base, 35 mm long, 11 mm wide; stamens 8, free, filaments 11 mm long, enlarged at the middle, anthers oblong, 3.5 mm long; disk 1.5 mm long, somewhat pentagonal; ovary substipitate, ovate, 3 mm thick, 5-locular, locules with many ovules; style 1 mm thick, 6 mm long, stigma dilated, truncate, pentagonal."
      Unfortunately, the fruits of this species are unknown, but in its flower, leaf, and petiole characters it agrees very well with Wenzelia.   It is the only species of Wenzelia that has paired spines.   However, the number of spines is variable in the genus, six species having no spines and two, W. paludosa and W. dolichophylla, having solitary spines, although the latter species is sometimes spineless.
      Wenzelia grandiflora was first described as a species of Citrus.   Many years later it was transferred to the monotypic genus Monanthocitrus because the two showed agreement in having paired spines in the axils of the leaves.   However, the flowers of Monanthocitrus are very much smaller than those of any species of Wenzelia, the petals being only one seventh as long as those of W. grandiflora!   Until fruits with mature seeds are found this latter species must stay in the genus Wenzelia.

VII.   Monanthocitrus Tan.
      VII.   Monanthocitrus Tan. Jour. Arnold Arbor. 9:138. 1928. Illus. fig. 3-11.
      Type species.Citrus cornuta Lauterb. = Monanthocitrus cornuta (Lauterb.) Tan.
      Distribution.—Southwestern Dutch New Guinea [West Irian, Indonesia]: near Alkmaar (Lat. 4° 40' S., Long. 138° 44' E.) on the Lorentz River and Signal Hill, a few miles west of Alkmaar (see van Nouhuys, 1923, pl. 4).
      Common name.—Spotseed-lime (or monanthocitrus).
      Twigs slender, slightly angled when young, soon cylindrical, internodes long (2-4 times as long as the petioles); spines paired or single (rarely 3, or wanting), very slender, about half again longer than the petioles; leaves simple, 14-22 cm long, elliptical, caudate-acuminate at the tips, broadly cuneate at the base, glabrous, margins entire or slightly wavy; lateral veins 10-15 pairs, visible on both sides; petioles very short (shorter than the spines and less than 1/20 as long as the leaf blade), wingless, flattened on the upper side, not articulated with the leaf blade; flower buds ovoid or ellipsoid, borne on slender pedicels as long or longer than the petioles; flowers small; calyx cupulate, 5-lobed, lobes acute, short; petals 5, shorter than the petioles; ovary globose, very small, with 5 locules, each locule with 6 (or sometimes 8?) ovules; style thick, twice as long as the ovary, stigma 5-lobulate; fruits globose or slightly ovoid, with thin peel showing numerous slightly depressed oil glands; locules 5, lined with smooth, cartilaginous walls, without pulp-vesicles, but containing 6 (or sometimes 8?) seeds surrounded by mucilaginous fluid; seeds packed tightly in the locule, much flattened, often concave (like a shallow saucer), testa with many small irregular dark red-brown spots; a very thin irregularly lobed membrane borders the seeds except on the attached side.
      Monanthocitrus has leaves very much like those of Wenzelia dolichophylla and several other species of that genus.   Monanthocitrus is most closely related to the species of Wenzelia belonging to the subgenus Papualimo, which have flattened seeds bearing on the unattached sides very thin, more or less torn membranous margins.   However, no species of Wenzelia has the much-spotted seeds (unique in the orange subfamily) found in Monanthocitrus, which also has flowers very much smaller than those of any species of Wenzelia—with petals only one-fifth to one-seventh as long!   Doubtless Monanthocitrus arose from an ancestral form much like that of some of the species of Wenzelia of the subgenus Papualimo, but it has traveled so far on the evolutionary highway that it apparently constitutes a good genus today.   The leaves are also much like those of Clymenia polyandra (a True Citrus Fruit Tree of the subtribe Citrinae), which has petioles of about the same length in relation to the leaf blade as those in Monanthocitrus, i.e., about one-twentieth as long.
      Thanks to the kindness of Dr. H. J. Lam, director of the Rijks Herbarium at Leiden, Holland, Swingle was able to study the type specimen on which Citrus cornuta Lauterb. was based, upon which species Tanaka based his genus Monanthocitrus.   This type specimen was collected by Versteeg (No. 1551) at Alkmaar in New Guinea in August, 1907.   It carries the fruit described by Lauterbach, showing curious slender, needle-like outgrowths from the tip of each carpel.   The fruit was still intact and when opened it showed no trace of pulp-vesicles.
      Monanthocitrus cornuta (Lauterb.) Tan. Jour. Arnold Arbor. 9:138. 1928. Citrus cornuta Lauterb. Nova Guinea 8:292. 1910. Illus. fig. 3-11.
      Type.—Southwestern Dutch New Guinea [West Irian, Indonesia], near Alkmaar (Lat. 4° 40' S., Long. 138° 44' E.) on the Lorentz River (Versteeg, No. 1551), Rijks Herb., Leiden.
      Distribution.—Known only from the type locality and from Signal Hill, a few miles west of Alkmaar.
      Lauterbach's description of Citrus cornuta (with slight changes in the order of arrangement) reads, translated, as follows: "Shrub about 1 m high; branches slender, flower-bearing ones 2-4 mm thick, terete, spiny; spines 2 or 3, about 1 cm long, in the axils of the leaves; leaves ovate-oblong or lanceolate, attenuate-acuminate, acumen up to 2 cm long, obtuse, sometimes emarginate, coriaceous, cuneate at base, entire or wavy margined, veins more prominent below, lateral veins in 10-12 pairs, ascending, thickly anastomosing by arches near the margin; petioles semiterete, scarcely channeled above, 5-6 mm long; flowers solitary, axillary, pedicellate; pedicels 5-8 mm long, subtended by 2-3 acute bracts, 1 mm long; calyx cupulate, 5-merous, lobes acute, 1 mm long, slightly ciliate under the lens; petals 5, ovate, subacute, about 5 mm long, whitish flesh-colored ('albis carnosulis'); stamens 10, free, filaments 1.5 mm long, anthers oblong, 2 mm long; disk annular; ovary globose, 1 mm diam., 5-locular, 2 ovules [? see below] in each locule; style thick, 2 mm long, stigma 5-lobed; fruit globose, rose-colored, 2.5 cm diam. when dry, fleshy, the apex with 5 solitary mammillae, each crowned by a single spine [-like appendage] up to 2 cm long.…
      "The long spines that form what may be called horns are very peculiar; as far as can be determined from dry material they arise from wart-like protuberances at the apex of the fruit and, in the dry state, fall off easily; 3 are intact; only the points of attachment of the other two are seen.   No trace of their origin can be seen on the ovaries."
      Swingle examined the type specimen of this species (Versteeg, No. 1551, collected at Alkmaar, New Guinea, August 2, 1907), lent to him by Dr. H. J. Lam, director of the Rijks Herbarium at Leiden.   It consists of a branch about 20 to 25 cm long, with six leaves (see fig. 3-11).   It bears one fruit, 25 by 32 mm, with well-marked oil glands.   When described by Lauterbach in 1910 the specimen still had three needle-like appendages arising from the top of the fruit and the bases of two more.   Doubtless there were originally five of them, one arising from each locule of the fruit.   These needle-like appendages are 20 to 22 mm long and very slender, about 0.6 to 0.75 mm in diameter at the base and 0.3 to 0.4 mm at the tip (measured when dry); they have about 20 to 22 shallow furrows, with narrow ridges between, extending lengthwise.   They arise at the apex of the fruit from wrinkled, glabrous elevations (broadly conical) of the peel about 2.5 to 3.5 mm wide and 2 to 2.5 mm high; one of these bases is found on each segment.   These remarkable needle-like outgrowths of the locules of the fruit look like slender projections of the peel and have the same color as the peel, but do not show oil glands.   The largest leaf blade measures 20.5 by 7 cm; two smaller leaves are 16 by 5.5 cm and 15 by 5 cm.   The petioles are channeled on the upper side, 7 to 8 by 1.5 to 1.6 mm, and not articulated with the leaf blade; the paired spines are straight, 3 to 12 by 0.5 to 0.8 mm.   The single fruit is oval, 30 mm long and 25 mm wide when dry; when swelled by soaking in hot water it measured 32 mm long and 28 to 30 mm wide.   The greatest length from the base to the tip of one of the five conical bases on which the needle-like outgrowths arise is 34 mm.   The dried fruit is buff to cinnamon in color (see Ridgway, 1912, pl. 39).   The pedicel is 10 mm long with two very small bracts attached 2 mm above the base.   The calyx lobes are reflexed when dry, but on soaking they extended out nearly at right angles to the pedicel, making the salverform calyx 6 to 7 mm wide and about 2.5 mm high.   The calyx lobes are about 3 mm long, thick and bluntly rounded at the tips, and become very thin at the extreme tips and sparingly ciliate.   When the fruit was cut open, the peel was seen to be thin, only 1 to 1.5 mm including the locule wall.   The outer peel is 0.6 to 0.8 mm thick; the locule walls are stiff, cartilaginous, and smooth inside.   The seeds, six (or sometimes eight) in each locule, are crowded into a bundle nearly filling the locule, which also contains mucilaginous gum, but no pulp-vesicles!   The seeds are very thin, often curved (like a shallow saucer), and have thin fimbriate margins, cream, buff, or ivory-yellow in color on all but the attached side of the seed.   Including this membrane, the seeds measure 10 to 11 by 7 to 8 by 1.5 to 2 mm.   The seed proper measures about 8 to 9.5 by 7 to 8.5 mm and is covered with an ivory-yellow testa which becomes fawn to army brown (Ridgway, 1912, pl. 40) when wet.   This testa shows many (10 to 20 or more) angular, dark red-brown spots, varying greatly in size from 0.5 to 2.5 mm, usually narrow if long.   On well-soaked seeds these spots varied from dark vinaceous to Indian red (Ridgway, 1912, pl. 27).   The inside layer of the testa in contact with the cotyledons of the embryo shows numerous very small, curved, vermicular markings of vinaceous brown (Ridgway, 1912, pl. 39).   The seeds contain a single very thin embryo, oval in outline, 8 to 9 by 7 to 8 mm, and 1 to 1.5 mm thick, vinaceous fawn to fawn color (Ridgway, 1912, pl. 40).   No oil glands can be detected in the cotyledons by superficial examination.   The very small radicle projects out 0.5 to 0.8 mm at the base of the cotyledons.   The attached edge of a well-soaked seed showed a more or less rugose ariloid tissue, 6 to 7 mm long and about 2 mm wide, which projects 1.5 to 2 mm beyond the seed.
      A paratype specimen of M. cornuta (Versteeg, No. 1551), also collected at Alkmaar, New Guinea, on August 2, 1907, preserved in the herbarium of the Botanical Museum at Berlin-Dahlem, was photographed by Tanaka, who kindly sent Swingle a print.   It is a single twig with one flower bud, 4.5 by 3 mm, on a slender pedicel about 7 mm long.   One leaf measures 2l by 7 cm; another, lacking the acuminate tip, measures 18.5 by 8.2 cm, with a petiole 8 by 1.5 mm.   The spines are all paired, straight, very slender, 8 to 14 mm long and 0.6 to 0.8 mm at the base and tapering to a very acute point.   The lateral veins arise at a large angle with the midrib, about 65° to 75°, in the middle part of the leaf blade.
      The remarkable needle-like protuberances that arise at the tips of each carpel of the fruit of the type specimen are still imperfectly known.4   In a supplementary note on this plant Lauterbach stated (1912, p. 825) that a specimen collected by von Roemer (No. 633) at Signal Hill, a few miles west of the type locality at Alkmaar, "had fruits with no trace of the peculiar spines.   Very likely it was a question, in case of the original specimen of Versteeg, No. 1551, of a mere malformation that was nevertheless of remarkable regularity."

VIII.   Oxanthera Montrouzier
      VIII.   Oxanthera Montrouzier, Mém. Acad. Sci. Lyon 10:186. 1860.
      Type species.Oxanthera fragrans Montr.
      Distribution.—New Caledonia and Art Island (80 km to the northwest of New Caledonia).
      Common name.—False orange (or oxanthera).
      Thornless small trees or shrubs with thick but small 1-foliolate or simple leaves (3-10 cm long), usually becoming glaucous; petioles either wingless or narrowly winged, articulated with the leaf blade in 3 species, not articulated in the other species; flower borne singly or in 1-3-flowered racemes in the axils of the leaves; calyx and corolla 5-merous; stamens 3-4 times as numerous as the petals; ovary with 2-7 locules, each locule containing several ovules, usually 6; fruits ellipsoid, small, about 2.5 X 1.5 cm.
      Oxanthera is a very distinct genus of Minor Citroid Fruit Trees, subtribe Triphasiinae, and differs strikingly from all the other genera of this subtribe in having glabrous, glaucous leaves bluntly rounded or retuse at the tip and cuneate at the base, borne on spineless twigs.   The leaves are thick and coriaceous in all the species except one.   To judge from the thick leaves that become glaucous as they develop, the plants of this genus are very probably somewhat xerophytic, like Microcitrus of Australia, but are not so well adapted to withstand long-continued drought and hot, dry winds as the truly xerophytic Eremocitrus, also of Australia.
      All the species of Oxanthera agree in having large, orange-like flowers, and glabrous, more or less glaucous leaves, which usually are rather blunt at the tip and cuneate at the base.   All species are thornless, and apparently all have more or less elongate fruits that are longitudinally ribbed at least when young.   Probably no other genus except Poncirus in the whole subfamily is so easily recognized at first glance.   In spite of this fact, this genus, containing four species (see key) shows unusual range of variation in other characters that in many groups of plants are extremely constant and have high taxonomic value.   Three of the four species of Oxanthera have unifoliolate leaves with clearly articulated petioles that are usually wingless, but are plainly but narrowly winged in one species.   One of the species has a hypomerous ovary with only two locules, another has an isomerous ovary with five or six locules, whereas a third species has a hypermerous ovary with seven locules.   Better material of O. aurantium is urgently needed to learn whether the ovary is always 2-locular; if so, it might well be made the type of a subgenus.   For the present it is left in the genus as an anomalous species that, nevertheless, is clearly related to the other three species.
      It is probable that Oxanthera is a highly specialized xerophytic genus that has developed from the common ancestor of the genera Wenzelia and Monanthocitrus, both of which share with Oxanthera the character of having usually six ovules in each locule of the ovary.   Oxanthera has acquired a distinctly xerophytic facies but, unlike the extremely xerophytic Eremocitrus, one of the True Citrus Fruit Trees of Australia, which has very small flowers, Oxanthera has large, white, fragrant flowers very like those of Citrus.
      1.   Oxanthera fragrans Montr. Mém. Acad. Sci. Lyon 10:186. 1860. Citrus oxanthera Beauv. Ann. Soc. Bot. Lyon 26:14. pro parte. 1901.
      Type.—L'ile Art (about 80 km northwest of the north end of New Caledonia) (Montrouzier).   Herb. Faculte de Méd. et Pharm. de Lyon.
      Distribution.—Art Island, New Caledonia.
      Common name.—Fragrant oxanthera.
      The original description by Montrouzier, translated, reads as follows: "Shrub 6-7 ft. [1.8-2.13 m] high, beautiful, twigs terete, gray.   Leaves obovate, apex acuminate, subcrenulate, thick, very much dotted [with oil glands], 2-2.5 in. [5-6.3 cm] long (including the petiole), 6-7 lin. [1.2-1.4 cm] wide.   Peduncles axillary, 1-flowered, shorter than the leaves.   Flowers white, the size of those of Citrus Aurantium, and very similar to them, very much dotted [with oil glands], exhaling a penetrating odor.   Fruit elongated.   Is this Limonia lucida Forst.?   This beautiful shrub flowers in November.   It is found in the forest.   The natives call it Dongan na diet, 'Orange Tree of the Forest.' "
      As the genus Oxanthera was originally monotypic, being founded on O. fragrans alone, Montrouzier's description of the genus applies to the species as well.   It reads, translated, as follows: "Flowers 5-merous, with imbricate aestivation, stamens 16-20, free.   Anthers erect, bilocular, incurved, mucronate at the apex.   Fruit with 5-6 locules, with many seeds in each locule.   Style elongate, thick, cylindric.   Stigma obtuse.   Leaves 1-foliolate, articulated with the tip of the petiole, petioles wingless.   A spineless shrub."
      Tanaka in discussing O. aurantium (1928b, p. 44) stated: O. fragrans has petals clearly marked with translucent points, a glabrous calyx, oblong anthers with only a small mucron at the apex, an oblong-oval ovary much furrowed ['très creusé’], and a very short style."   It differs in all these points from O. aurantium, with which Beauvisage confused it, since he included descriptions of material of both species in his Citrus oxanthera.   The type specimen photographed by Tanaka shows petioles 4.5 to 7 mm long, articulated with the leaf blade, entirely wingless but apparently with a shallow channel on the upper side.
      This plant is said to be a beautiful fragrant shrub and may perhaps prove of value as an ornamental flowering shrub for subtropical climates.   To judge from its foliage, it is probably somewhat drouth-resistant [sic].   The petals, with densely crowded oil glands, may possibly yield an essential oil for perfume.
      2.   Oxanthera aurantium Tan. Bul. Soc. Bot. France 75:712. 1928. Citrus oxanthera Beauv. Ann. Soc. Bot. Lyon 26:14. pro parte. 1901. Illus. Guillaumin, Les Citrus Cult. Sauv. 60. 1917.
      Type.—New Caledonia, Gomonen near Gatope on dry hills (Vieillard, No. 2378).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Known only from the original collections.
      Common name.—Orange flower oxanthera.
      Tanaka’s original description reads, in translation, as follows: "A shrub, twigs terete, very leafy, young twigs pubescent; leaves small, oblanceolate, not glandular [?, ‘englandulosis’], chartaceous, channeled, with many veins, lateral veins curved, arising from the midrib at more or less wide angles; apex obtuse or often emarginate, base attenuated, short-petiolate; petioles cylindric, not articulated [with the leaf blade]; inflorescences pubescent, 1-3-flowered racemes; flowers large, equaling Citrus in size, petals without oil glands, minutely papillose; calyx scutellate, pubescent without, lobes rounded, inflexed; stamens free or in part cohering, filaments much longer than the style, canaliculate, minutely papillose on the outside, anthers rather large, cordate-acuminate; ovary ovoid-attenuate, glabrous, style equaling the ovary [in length], straight, cylindric, rather thick, stigma ellipsoid, disk annular, short, stigma ellipsoid, broader than the style."
      Swingle examined two cotypes from the original type collection of this species (Vieillard, No. 2378), one in the herbarium of the Museum of Natural History, Paris, and the other, an excellent flowering specimen, in Gray Herbarium, Harvard University, Cambridge, Massachusetts.   The latter specimen shows completely thornless twigs, pubescent when young but losing the hairs and appearing glaucous with grey wax when older; leaves simple, thick, coriaceous, oblanceolate, 2-4 X 0.5-1 cm, apex rounded, usually rather deeply emarginate, margins minutely crenulate-dentate above the middle, upper surface dotted with minute oil glands, about 0.5 mm apart, which are not translucent on dry specimens, midrib slightly raised on the upper surface, strongly salient below, slightly pubescent near the base of the leaf, lateral veins visible only on the under surface, curved and more or less swollen and salient, arising sometimes at an angle of 35°-45°, sometimes at an angle of 75°-80°, with the midrib, soon curved backward toward the base of the leaf and anastomosing especially in the upper part of the leaf; petioles pubescent, not articulated with the leaf blade but merging imperceptibly into it, flattened above and showing two raised lines that continue the leaf margins, running almost to the base of the petiole, but showing no green color for 2-3 mm from the base; peduncles at flowering densely covered with gray hairs, with several leaf-like bracts, making the peduncle appear like a short leafy twig; pedicel very short (3-4 mm), not clearly demarcated from the calyx, which is densely gray-pubescent without; flowers like Citrus except for the gray-pubescent calyx and minutely papillose petals.
      Tanaka (1928b, p. 712) noted that this species differs greatly from O. fragrans in its pubescent new growth, narrower leaves with many lateral veins arising at a wide angle with the midrib, and entirely different flower parts.   Beauvisage (1901, pp. 11, 12) studied several specimens of Vieillard's No. 2378 (the type collection on which O. aurantium is based) and his report, translated, reads as follows: "I dissected several rather badly preserved flowers and found 14 to 16 stamens with mucronate anthers and connate filaments except in a very young bud, where the filaments, still short, were free.   Similarly I noted in the ovaries a few differences which could be due to the more or less advanced stage of development; the ovary was ovoid or conical; the style articulated [with the ovary] or not; the stigma channeled-ellipsoid or not; the locules of the ovary, usually with several ovules ('pluriovalées'), seem sometimes to be biovulate."
      A good specimen of the type collection (Vieillard's No. 2378) in the Gray Herbarium at Harvard University provided a flower which Swingle examined.   It was restored by prolonged soaking, then imbedded, cut into serial sections, stained and mounted permanently by a modification of Juel's method (see Swingle, 1939b, p. 270).   The ovary was skillfully cut by Swingle's assistant in this part of the work, Albert H. Tillson, so as to show half the ovary cut into many serial longitudinal sections and the remaining (longitudinal) half cut into still more numerous cross sections.   These sections showed clearly that the ovary was cylindric, about 3.5 mm long after the dehiscence of the style, and 0.9 to 1.2 mm in diameter (with two shallow furrows in the plane of junction of the carpels and two low ridges on the dorsal walls), narrowed abruptly at the base and fitting rather closely into a conical cavity in the top of the disk; it contained two long, slender locules about 2.4 by 0.3 by 0.4 mm.   The ovary was narrowed gradually at the top into the style, 0.5 to 0.6 mm in diameter at the point of dehiscence.   The serial cross sections showed altogether six ovules in the locule, arranged in two ranks, along with abundant trichomes arising from the flattened placenta that ran from bottom to top of the locule cavity along its dorsal wall.
      This species differs widely from its congeners in having a hypomerous ovary with only two locules.   Of the other species, one is isomerous with five or six locules and another is hypomerous with seven locules.   O. aurantium also differs radically from its congeners in having simple instead of unifoliolate leaves.   Its relationships are discussed above.
      This species, collected by Vieillard on arid hills in New Caledonia, shows signs of being a rather pronounced xerophyte able to grow in dry situations.   The thick leathery leaves remind one of the leaves of Eremocitrus grown on well-watered soil but are not so gray-green in color and show a different structure on the upper and lower surface, a difference that is almost lost in Eremocitrus.
      3.   Oxanthera undulata (Guill.) Swing. Jour. Wash. Acad. Sci. 30:83. 1940. Citrus undulata Guill. Bul. Soc. Bot. France 85:304. 1938.
      Type.—New Caledonia, upper Dothio Valley (Balansa, No. 3054).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Known only from the type locality in northern New Caledonia.
      Common name.—Wavy-leaf oxanthera.
      The original description of this species reads, translated, as follows: "Tree 5-6 m high, very glabrous; leaves thin, ovate-lanceolate, up to 9 X 4 cm, cuneate at the apex and base, with undulate margins, nerves very slender, almost immersed, petiole articulated [with the lamina], slender, 5-10 mm long, [wingless and] not at all margined; flowers not seen; fruits ovate, 2.5 X 2 cm, embryo with white cotyledons…"
      The notes which follow the Latin diagnosis read: "Excessively rare; a species well characterized, although imperfectly known, by its thin leaves with strongly undulate margins.   The fruit resembles that of C. neo-caledonica in form."   Unfortunately the number of locules in the fruit is not stated.
      To judge from its large thin leaves, this species may perhaps prove to be a mesophyte rather than a xerophyte, as are the other 3 species in Oxanthera.
      4.   Oxanthera neo-caledonica (Guill.) Tan. Bul. Soc. Bot. France 75:713. 1928. Citrus neo-caledonica Guill. in Lecomte, Not. Syst. 2:130 (diagnosis). 1911; and ibid. 2:390 and 406 (Latin name).5 1913. Illus. Guillaumin, Les Citrus Cult. Sauv. 61. 1917; fig. 3-12 this work.
      Type.—New Caledonia, Poume Peninsula (Balansa, No. 3374).   Herb. Mus. Hist. Nat., Paris.   Cotype: Same collection (Balansa, No. 3374).   Herb. Bur. Sci., Manila.
      Distribution.—Known only from the type locality (Lat. 20° 30' S., Long. 164° 8' E.).
      Common name.—False orange (or large-leaf oxanthera).
      Guillaumin's detailed description of this plant reads, in translation, as follows: "Spineless shrub, 1 m high, with whitish flowers; young branches with a waxy bloom that gradually disappears.   Leaves very coriaceous, glabrous even when very young, lanceolate or obovate, 6-8 cm long (including the petiole), 1-4 cm wide, narrowed toward the base, attenuate toward the apex, but not acute when lanceolate, suborbicular or orbicular when obovate, sometimes emarginate.   Midrib and lateral veins prominent on both sides but especially on the underside (like veins of the hand).   Petiole distinctly articulated [with the leaf blade], 1-2 cm long, slightly winged (less than 4 mm).   Flower [bud] (a single one) terminal on a young branch, spherical (14 mm diam.); pedicel 7 mm long, cylindric, glabrous; sepals 5, 2 mm long, nearly round; petals 6, unequal, ovate (12 X 6-12 mm), thick and fleshy, with inconspicuous, small, pellucid oil glands; stamens 24, unequal (8-11 mm), [arranged] very irregularly in fascicles, filaments rather fleshy, not flattened, anthers triangular-elongate, 2.5 mm long, the locules diverging slightly at the base; ovary subcylindric (3.5 X 3 mm), 7-ribbed, with 7 segments; style cylindric, 2 mm long, stigma cylindric.   Fruit (immature) ovoid-subcylindric, supported on pedicels covered with whitish bloom, more than 1.5 cm long and 2 mm diam.; more mature fruits 2.5 X 1.5 cm, with 7 segments, peel thin, pulp mucilaginous, seeds of unknown color.
      "Presqu'ile de Poume, in volcanic soil, May, 1871 (Balansa, No. 3374).
      "Although the color of the embryo is not known, the affinities [or this plant] seem to be especially with Citrus Aurantium L. but the form and the aspect of the leaves, of the flowers and of the ovary, clearly distinguish this species."
      Tanaka (1928b, p. 44) noted that the flowers are like those of O. fragrans and O. aurantium but that "the surface of the petals is completely uniform [not papillose or marked with evident oil glands], the ovary is shorter, the stigma not dilated and the disk much larger."   He went on to say: "The fruit is oblong, sometimes curved, pointed, with superficial ribs, with about 7 segments, containing large seeds with a single very hard embryo.   The pulp-vesicles are not very much developed and are gummy."   He was of the opinion that the lack of true pulp-vesicles excludes this species from the True Citrus Fruit Trees even though it does have more than twice as many stamens as petals.
      This species, as Tanaka remarked, shows a higher organization than most of the other Minor Citroid Fruit Trees.   It has seven locules in the ovary and, in common with the other species of Oxanthera, more than twice as many stamens as petals, instead of only twice as many, as is characteristic of the other genera of the subtribe Triphasiinae.

IX.   Merope Roem.
      IX.   Merope Roem. Syn. Hesper. 1:44. 1846.
      Type species (Monotype).—Citrus angulata Willd. = Merope angulata (Willd.) Swing.
      Distribution.—Java, Burma; also New Guinea (fide Tanaka, 1931b, p. 10).
      Small tree, twigs with stout spines paired or single; leaves simple, thick, coriaceous; petioles short, wingless; flowers solitary, axillary; calyx cupulate, 5-lobed; petals 5, imbricate, oblong-lanceolate, acute; stamens 10, free, anthers linear, as long or longer than the filaments; ovary ovoid, 3-locular, ovules 4 in each locule; fruit triangular in cross section, ovoid-acuminate, with very long, flattened seeds narrowed at the upper end to a blunt point, borne 1 in a segment and immersed in a glutinous, mucilaginous fluid.
      This remarkable genus is found widely scattered over the East Indies in brackish swamps.   It is outstanding because of its buckwheat-shaped fruit, containing very long seeds.   It is a highly specialized genus of the subtribe Triphasiinae, related to Pamburus and Paramignya, but not closely.
      Merope angulata (Willd.) Swing. Jour. Wash. Acad. Sci. 5:420. 1915. Citrus angulata Willd. Sp. Pl. ed. 4. 3(2):1426. 1801; Limonellus angulosus Rumph. Herb. Amboin. 2:110. 1741; Sclerostylis spinosa Blume, 1825; Glycosmis spinosa (Bl.) Dietr. 1840; Merope spinosa (Bl.) Roem. 1846; Limonia angulosa Wt. & Arn. ex Miq. 1859; Atalantia longispina Kurz, 1872; Gonocitrus angulatus (Willd.) Kurz, 1874; Paramignya angulata (Willd.) Kurz, 1875; Atalantia spinosa (Bl.) Koorders (non Hook. f.), 1912; A. angulata (Willd.) Engl. Die Nat. Pflanzenfam. 19a:330. 1931. Illus. Rumphius, loc. cit. pl. 32; Kurz, Jour. Asiat. Soc. Bengal 43(2):pl. 18. 1873; Valeton, Icon. Bogor. pl. 348. 1912; Swingle, loc. cit. figs. 1-2; idem, in Bailey, Stand. Cycl. Hort. 4:2038, fig. 2362. 1916; idem, in Engler, loc. cit. 329, fig. 151; fig. 3-13 this work.
      Type.—None found; based on Rumphius' description and figure cited above.
      Distribution.—Java, Burma, in tidal swamps; also New Guinea (fide Tanaka, 1931b, p. 10).
      Common name.—Merope (or kigerukkan, so-called at Noessa Kambangan Island, Java).
      Leaves coriaceous, inconspicuously veined, alternate; petioles simple.   Twigs with very strong spines, often in pairs.   Flowers borne singly or in pairs (rarely in few-flowered clusters) in the axils of the leaves; ovary stalked on a rather tall disk, 3-5-celled, with 4 pendulous ovules in each cell; stamens 10, free, anthers linear-oblong.   Fruits strongly angled, triangular in cross section; cells filled with a sticky, mucilaginous fluid (without true pulp-vesicles); seeds very large, flattened, reniform, caudate at the tip where attached.   Cotyledons in germination aerial, not increasing in size; first foliage leaves alternate, broadly ovate.   A large shrub or small tree (not a climbing shrub), growing on the seashore in tidal forests or mangrove swamps from the mouth of the Ganges to the Moluccas.   The seeds, often 2 in each of the flattened locules of the fruit, are about 35 X 15 X 5 mm, largest known in the orange subfamily.
      This remarkable plant, the sole species of the genus Merope,differs strikingly from all other species of the orange subfamily except Severinia lauterbachii in having fruits triangular in cross section, with long flat seeds tapering to a blunt point.   It is a very spiny salt-loving shrub or small tree found in tidal marshes in Java and Burma, and in the East Indian Archipelago.   As is usual with halophytes (plants able to live in more or less salty water), M. angulata has thick and leathery leaves.   Buds of the kigerukkan, inserted on grapefruit, grew vigorously in the greenhouse of the former Bureau of Plant Industry at Washington, D.C., and soon flowered (see fig. 3-13), but finally showed signs of a bad bud union, became stunted, and died.

GROUP B.   THE TRIPHASIA GROUP
      This group contains two genera with a total of four species, all of them shrubs or small trees with small ovoid or globose fruits, bright-colored (red) in Triphasia and orange-like with a thick peel dotted with oil glands in Pamburus, but always without pulp-vesicles in both genera.   The ovaries are 3- to 5-locular, and each locule contains one or two ovules.   The leaves are simple in two of the species of Triphasia, as well as in Pamburus, and trifoliolate in the other species of Triphasia; they are rather thick, more or less glaucous, and inconspicuously veined.   Little is known as yet about the relationships of these two genera with the other genera in the subtribe Triphasiinae.

X.   Triphasia Lour.
      X.   Triphasia Lour. Fl. Cochinch. 152. 1790. Echinocitrus Tan. Jour. Arnold Arbor. 9:137. 1928.
      Type species.Triphasia aurantiola Lour. = T. trifolia (Burm. f.) P. Wils.
      Distribution.—Southeastern Asia, East Indies, Philippines, New Guinea; widely cultivated in tropical and subtropical regions.
      Common name.—Limeberry (or triphasia).
      Shrubs or very small trees, 1-2 m high, with paired axillary spines; leaves usually very small (except in T. grandifolia), simple or 3-foliolate, rather thin and without evident reticulation, numerous oil glands visible on both faces; petioles very short (1/10-1/20 the length of the leaves), wingless, pubescent, flattened and slightly channeled on the upper surface, not articulated with the leaf blade; flowers single, or in groups of 2 or 3 in the axils of the leaves; pedicels slender; calyx 3-5-merous, cupulate, persistent; corolla 3-5-merous, petals linear or obovate-spathulate; stamens free, twice as many as the petals, filaments slender, sometimes broadened at the base, anthers small, linear; ovary ovoid or elliptical, tapering more or less toward the base, with 3-5 locules, ovules 1 or 2 in each locule; fruits small, with the seeds imbedded in mucilaginous pulp; seeds plump (sometimes polyembryonic in T. trifolia).
      The genus was represented from 1790 until 1925 by a single species, T. trifolia.   The discovery in the Philippines of a second species, a plant with much larger simple leaves which Merrill named Triphasia grandifolia, broadened our concept of the genus.
      It now appears that there is a third species, descried as Paramignya brassii from New Guinea by C. T. White in 1926 and made the type of a new genus, Echinocitrus, by Tanaka in 1928.   This species, although obviously closely related to the type of Triphasia, differs in a number of characters that justify placing it in a distinct subgenus, for which Tanaka’s generic name, Echinocitrus, is retained.
      Keys to the subgenera and species of Triphasia are presented.

Subgenus Triphasia
      Subgenus Eutriphasia Swing. in Webber & Batchelor, Citrus Indus. 1:237. 1943.
      1.   Triphasia trifolia (Burm. f.) P. Wils. Torreya 9:33. 1909. Limonia trifolia Burm. f. Fl. Ind. 103. 1768; L. trifoliata L. 1771; Triphasia aurantiola Lour. 1790; Limonia (?) diacantha DC. 1824; Triphasia trifoliata (L.) DC. 1824. Illus. Burmann f. loc. cit. pl. 35 (col.); Jacquin, Icon. Pl. Rar. 3:pl. 463. 1786-1793; Risso & Poiteau, Hist. Nat. Orang. pl. 108 (col.). 1818-1822; fig. 3-14 this work.
      Type.—Java (Burmann f., No. 166).   Herb. Delessert, Geneva (fide Tanaka).
      Distribution.—Southeastern Asia (?), East Indian Archipelago (?); widely cultivated in tropical and subtropical regions.
      Common name.—Trifoliate limeberry, limeberry, or triphasia.
      A glabrous shrub or small tree with terete twigs bearing paired spines in the axils of the leaves; leaves 3-foliolate, the terminal leaflet ovate with a cuneate base and a rounded emarginate tip, 2-4 X 1.5-2 cm; lateral leaflets much smaller than the terminal one (1-2-2 X 0.8-1.2 cm), broadly rounded at the tip, cuneate at the base; petiolules very short (1.5-2 cm); petioles short (3-5 mm), wingless; flowers appearing singly or 2 or 3 in the axils of the leaves; peduncles short (3-4 X 1 mm); flower buds cylindrical, 10-12 X 3-4 mm; flowers 3-merous (but with 6 stamens); sepals small (1.5-2 mm long), 3-lobed, green, persistent; petals white, 10-13 X 3.6-4.5 mm; staminal filaments slender, glabrous, 9-11 mm long, anthers oblong, 2 X 1 mm; disk annular or short-cylindric; ovary ovoid or fusiform with 3 locules, each with 1 ovule, narrowed into a slender, deciduous style with a capitate, 3-lobed stigma; fruit ovoid or subglobose, sometimes apiculate, 1.2-1.5 cm long, dull reddish-orange or crimson when fleshy ripe; peel with many small oil glands; seeds 1-3, immersed in mucilaginous pulpy flesh.
      Triphasia trifolia is very widely grown in all tropical and subtropical regions as an ornamental shrub.   It is also used for hedges.   The fragrant white flowers are soon succeeded on the branches by the small dull-red berries (see fig. 3-14).   The foliage is handsome, a shiny dark green, and the plant usually makes a round-topped shrub that does not grow too large for dooryard plantings.   Triphasia trifolia has become naturalized in certain sections of the United States, in "hammocks, fields and cultivated grounds, coastal plain, Florida to Texas," according to Small (1933, p. 760).
      la.   Triphasia trifolia forma tetraploidea.
      This form differs from the species in having larger flower parts and thicker leaves, and especially in having 36 chromosomes in all the somatic cells instead of 18, the normal number for almost all plants of the orange subfamily so far studied.
      This undoubtedly tetraploid form of the common trifoliate limeberry is like tetraploid forms of Citrus in having somewhat thicker leaves and larger flower parts.   Longley (Traub and Robinson, 1937, p. 794) first detected the tetraploid nature of this plant.   It is one of a number of seedlings (of which all the others are apparently diploid) grown from seeds of Triphasia trifolia collected by Dr. J. N. Rose at Mazatlan, Mexico, in April, 1910.   This is believed to be the first instance reported of the spontaneous origin of a tetraploid seedling of an aurantioid genus other than Citrus or Poncirus.   It is analogous to the wild-growing tetraploid form of Fortunella hindsii, native to the mountains of southern China, of which a diploid form is found in cultivation.   Frost (1938a) has found tetraploid mutations of several of the commonly grown species of Citrus arising as seedlings in California.
      2.   Triphasia grandifolia Merr. Phil. Jour. Sci. 26(4):458. 1925.
      Type.—Mindoro Island, P.I. (Ramos, No. 40822).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: known only from the type locality.
      Common name.—Unifoliate limeberry.
      Merrill's description of this species reads as follows: "An erect, nearly glabrous, spiny shrub about 2 m high, the branches and branchlets slender, terete, pale greenish when dry, the branchlets 1 to 1.5 mm in diameter, puberulent.   Leaves simple, alternate, chartaceous, oblong-ovate, 5 to 9 cm long, 2.5 to 4.5 cm wide, greenish when dry, the base broadly rounded, narrowed upward to the somewhat acuminate apex, the lower surface paler than the upper and conspicuously glandular-punctate; lateral nerves slender, spreading, about 10 on each side of the midrib; petioles about 3 mm long; stipules spiniform, rigid, stiff, 5 to 7 mm long.   Flowers white, solitary, axillary, one between each pair of spinelike stipules, their pedicels about 6 mm long, somewhat thickened upward.   Calyx 4 mm in diameter, the tube 1 to 1.5 mm long, the three broad triangular lobes obtuse to acute, punctate, about 2 mm long.   Petals 3, imbricate, glabrous, punctate, oblong-oblanceolate, about 1.5 cm long, 0.5 cm wide.   Stamens 6, their filaments narrowly lanceolate, tapering upward, 7 mm long, anthers oblong, 3 mm long.   Ovary 3 celled, short-stipitate, oblong, slight thicker than the stout stipe and style, glabrous, the style 6 mm long, stigma subcapitate.   Disk truncate, about 1 mm high.   Fruits unknown."
      Triphasia has been known since 1790 as a monotypic genus, T. trifolia, the type species, was remarkable in having three sepals, three petals, six stamens (three alternating with the sepals and three with the petals), an ovary with three locules, and trifoliolate leaves, so the genus name, Triphasia, was unusually appropriate.   In 1925, Merrill received from Mindoro Island in the Philippines a second species, clearly a Triphasia, with 3-merous flowers as in T. trifolia but with large simple leaves instead of small trifoliolate leaves.   The fruits of this interesting species are still unknown.

Subgenus Echinocitrus (Tan.) Swing., n. comb.
      3.   Triphasia brassii (White) Swing. Jour. Wash. Acad. Sci. 28:532. 1938. Paramignya brassii C. T. White, Jour. Arnold Arbor. 7:231. 1926; Echinocitrus brassii (White) Tan. Jour. Arnold Arbor. 9:138. 1928.
      Type.—New Guinea, Papua, Rigo (Brass, No. 817).   Herb. Arnold Arbor.
      Distribution.—New Guinea: known only from the original locality.
      The original description of Paramignya brassii (= Triphasia brassii) by C. T. White, translated, reads as follows: "A slender bush, attaining 2 m in height, branches drooping (fide Brass); spines numerous, single or paired, 0.8-1.5 cm long, slender, puberulous, evanescent on fast-growing large shrubs; leaves shiny, dark green (fide Brass), with short petioles; petioles puberulous, 2-3 mm long; varying greatly in the size of the blade, on flowering twigs small (1.5 X 0.8 cm), usually larger (4 X 3 cm), ovate-rhomboid, retusely acuminate at the apex, lateral veins 5-6 pairs, visible above and below; flowers sweet-scented (fide Brass), axillary, solitary, with a glabrous peduncle thickened toward the apex, about 1 cm long; calyx 5-lobed, lobes broadly deltoid, about 1 cm long, margins ciliate; petals 5, obovate, 1 X 0.3 cm, with numerous oil glands outside toward the tip; stamens 10, filaments free, 6 mm long, anthers linear, 2 mm long; ovary lageniform, short-stipitate; stipe in dried specimens slightly ribbed; style thick, ribbed, 5 mm long, stigma discoid; fruit red, of variable form, ellipsoid, pyriform or obovoid, 3 cm long, 1.5-2 cm diam., 5-locular, pulp scanty, seeds 1 or 2 in each locule, 1 X 0.6 cm."
      The simple crenate leaves of the type material of this species are usually broad at the base, often nearly trangular [sic] in outline.
      In discussing this plant, which he made the type of a new genus, Echinocitrus, Tanaka (1928c, p. 138) wrote: "…the present species is closely allied to Triphasia in its puberulent, somewhat zigzag branches armed with paired sharp spines, in the thin leaves without prominent reticulation, solitary flowers with cupulate calyx and slender petals, free filiform filaments and linear anthers, stalked ovary narrowed into a long style, and in the few-celled bright-colored berry without pulp-vesicles.   It is, however, distinct from Triphasia primarily in having 5-merous flowers, while both species of Triphasia (T. trifolia and T. grandifolia) have always 3-merous flowers.   This species is far more spiny than Triphasia, especially at the juvenile stage, and the fruit is very much larger and mostly pointed at the base, not at the apex, the segments containing distinctly biseriate seeds.   The leaves are crenate, not entire, and the pedicel of the fruit is very much elongated; calyx lobes are distinct and ovoid-deltoid, not so broad and flat-triangular as in Triphasia.   The calyx lacks pubescence but is only minutely ciliate on the margin.   The appearance of the fruit is very much like an oval kumquat (Fortunella margarita Swingle), but it lacks pulp-vesicles."
      The characters mentioned above are most of them of specific rather than generic nature.   The fact that this species is 5-merous (except that the stamens are twice as many as the petals) is not an adequate character to separate it as a distinct genus from T. trifolia and T. grandifolia, both of which have 3-merous flowers (except the stamens, which are 6-merous).   Such variation in the number occurs frequently in genera belonging to the subtribe Triphasiinae, of which Triphasia is the type genus.

XI.   Pamburus Swingle
      XI.   Pamburus Swingle, Jour. Wash. Acad. Sci. 6:336. 1916. Chilocalyx Turcz. (non Klotzsch), Bul. Soc. Nat. Mosc. 36:588. 1863. Illus. Wight, Hook. Bot. Misc. 3:291, pl. 33. 1833; fig. 3-15 this work.
      Type species.Limonia missionis Wight = Pamburus missionis (Wight) Swing.
      Distribution.—India: Madras State; Ceylon; Vietnam: Annam, Cochin China; Java; Bali.
      Common name.—Pamburus.
      Much-branched, very thorny shrubs or small trees, young twigs angled, soon becoming terete; leaves simple, very thick; petioles short, less than 1/10 the length of the leaf blade, more or less margined but not winged, not articulated with the leaf blade; inflorescences axillary racemes shorter than the leaves but longer than the petioles; flowers 4- or 5-merous; calyx cupulate, 4-5-lobed; petals 4 or 5, obovate or elliptic-oblong; stamens 8-10 (twice as many as the petals), filaments free, slender, glabrous, anthers cordate-elliptical or linear-oblong; disk cylindric, narrower than the ovary; ovary with 5 (or sometimes 4) locules, with 2 ovules in each locule; fruits globose (resembling a very small orange in appearance), with a well-developed peel showing numerous oil glands, the segments usually containing 1 or 2 seeds surrounded by a glutinous mucilage (pulp-vesicles lacking); seeds large, thick, globose or ovoid.
      Pamburus differs from Paramignya in its shrubby habit and lack of retrorse or hooked spines, and in its thick leaves with rounded tips and short non-articulated petioles without pulvinoid tips.   Its relationships to the other genera of the subtribe Triphasiinae are still obscure and should be cleared up by a detailed study of the flowers and fruits.
      Pamburus missionis (Wt.) Swingle, Jour. Wash. Acad. Sci. 6:338. 1916. Limonia missionis Wt. Hook. Bot. Misc. 3:291. 1833; Atalantia missionis (Wt.) Oliv. 1861; Chilocalyx ellipticus Turcz. 1863. Illus. Wight, loc. cit. pl. 33; fig 3-15 this work.
      Type.—British India, Madras Presidency, Talaar (Wight in Wallich, No. 6358-B).   Herb. Kew.
      Distribution.—India.
      Common name.—Pamburus.
      The original description by Wight reads as follows: "Leaves simple, elliptical-oblong, short-petioled; racemes many-flowered, in the axils of the spines, shorter than the leaves.
      "A small tree, with a round, very branching, bushy head; the branches round, smooth, bright green, the older ones armed with numerous large, strong, sharp thorns, the flower-bearing ramuli flexuose, with small, straight, axillary spines.   Leaves alternate, short-petioled, oval, sometimes emarginate and slightly crenate, oftener entire, smooth, coriaceous, dark shining green, mottled with white spots, and perforated with numerous pellucid points.   Racemes in the axils of the spines, shorter than the leaves, somewhat capitate.   Flowers numerous, fragrant, pure white.   Calyx small, 4- or 5-lobed.   Corolla 4-5-petaled, caducous.   Petals obovate.   Stamens 8 or 10, a little shorter than the petals.   Filaments subulate.   Anthers erect, large in proportion to the flowers.   Pistil: germen superior, elevated on a glandular receptacle, globular, 4- or 5-celled, with several ovules in each cell, only one of which usually arrives at maturity.   Style cylindrical.   Stigma capitate.   Pericarp a 4- or 5-celled berry; cells containing a very glutinous mucilaginous fluid, and one roundish seed,enclosed in a thick, firm, glandular, orange-like fruit.
      "This tree is found, rather frequently, in sandy soil near the sea-coast, flowering towards the end of the cool season, and ripening its fruit about August or September.   The corolla falls a few hours after expansion; hence it is difficult to procure specimens in flower.   That here figured was gathered about the beginning of March."
      Swingle's description of the species given when the genus Pamburus was established (1916a, p. 338) reads as follows: "A much-branched shrub or small tree, armed with stout straight spines, these 2-3 cm long, arising singly (or rarely in pairs?) on the side of the bud in the axils of the leaves.   Leaves oval, oblong-obovate or elliptical, 6-10 cm long, 2-4 cm broad, very thick, coriaceous, glandular-punctate, the tip rounded, sometimes slightly emarginate, the base narrowed rather abruptly into the petiole, the margin entire, becoming gray and apparently crenate in drying; lateral veins inconspicuous, tertiary ones not apparent, the two faces very similar in appearance, drying to velvety gray-green unlike those of any other member of the subfamily Citratae [Aurantioideae].   Flowers 12-20 mm in diameter, fragrant, with small pointed sepals and 5 or 4 white obovate caducous petals about 1 cm long.   Pistil about 1 cm long.   Fruit about 2.5 cm in diameter, orange-colored when ripe, with a thick peel dotted with oil glands, 5-4 celled, the cells containing 1 or 2 seeds surrounded by a sticky gum."
      This remarkable plant has leaves unlike those of any other member of the orange subfamily (see fig. 3-15).   In drying they become pale gray-green and show what seems to be a fine-textured, velvety surface.   As it grows commonly in the dry region of Ceylon and in the flat, sandy coastal lands of Madras State in India, it may prove to be the fourth genus of the orange subfamily to show somewhat xerophytic structure, the other three being Oxanthera of New Caledonia, and Microcitrus and Eremocitrus, both of the semiarid regions of northeastern Australia.
      The wood of P. missionis is used in India for making furniture.   The species should be tested as a rootstock for Citrus with Oxanthera fragrans and Eremocitrus glauca in the hope of finding rootstocks able to grow in soils with heavy content of mineral salts.   Pamburus missionis was introduced into the United States several decades ago.
      In 1941, a five-year-old tree of this species growing at the Plant Introduction Garden in Coconut Grove, Florida was about 2.5 meters high and 1.5 to 2 meters wide, with a very dense top composed of exceedingly numerous, short, divergent, very leafy twigs with stout, very sharp spines, 1.5 to 4 cm long, attached nearly at right angles to the twig every 2 to 4 cm.   The younger branches, the spines, and the more or less exposed under surfaces of the leaves were all of the same vivid green color, close to grass green (Ridgeway, 1912, pl. 17).   The upper surfaces of the leaves, also visible in large part, were a slightly darker green, near Hay’s green (Ridgeway, 1912, pl. 6).   The foliage was not only vivid green, but perfectly smooth and glossy, and showed no traces of injury from insects or diseases.   This plant seems admirably suited for making beautiful hedges of moderate height, since it stands pruning well.
      Anatomical study of freshly gathered leaves shows that the upper surfaces have two layers of palisade cells and the under surface one layer.   This doubtless explains the almost identical appearance of the upper and lower surfaces of the leaves and the almost complete masking of the veinlets.
      The plants grown at Coconut Grove thrived on rather poor, sandy soil that thinly covered a subsoil composed of porous limestone rock.   The pamburus seems to withstand nearly as much cold as the commonly cultivated species of Citrus.

GROUP C.   THE LUVUNGA GROUP
      The two closely related genera Luvunga and Paramignya, with a total of twenty-seven species, are both clambering woody lianas that hold fast to the branches of trees by means of retrorse or sometimes strongly recurved spines.   The leaves of Luvunga are large, apparently palmately trifoliolate, with very long wingless petioles; those of Paramignya are medium-sized, unifoliolate, with short petioles.   Both genera have the petioles articulated at both ends (at the junction with the leaf blade above and with the twig below).   These joints at both ends of the petioles are more or less swollen and pulvinoid, which permits each leaf to take up the best possible position to catch the light.   The ovaries are 2- to 5-locular and each locule has two ovules.   The fruits, so far as known, do not contain pulp-vesicles.
      The fruits both of Luvunga and of the typical species of Paramignya show a striking dimorphism of the oil glands, those in the peel being small, those in the deeper tissues between the peel and the locule cavities being much larger.   Possibly the deeper glands may produce not oil but mucilage that ultimately flows into the locule cavities about the seeds as the fruits ripen.   Only the study of fresh living fruits can decide the matter.
      It is highly probable that Luvunga and Paramignya are descended from a common ancestral plant with pinnate or trifoliolate leaves.   When Luvunga plants bear unifoliolate leaves, they have very much shorter petioles than do the normal trifoliolate leaves and, in consequence, look much like normal Paramignya leaves.   The flowers of both Luvunga and Paramignya look much like those of Citrus except for having fewer stamens (only twice as many as the petals instead of the four or more times as many found in Citrus).   The fruits look much like miniature lemons or limes, and, like ordinary Citrus fruits, have a peel thickly dotted with oil glands.
      Abundant material of the genera Luvunga and Paramignya studied by Tillson (1938, pp. 11, 12, 24) shows that in the development of the flower the petal midrib, as it diverges from the axis, carries, fused with it, the lateral sepal traces, an unusual type of floral vascular anatomy, characteristic of the Primitive Citrus Fruit Trees of the subtribe Citrinae, of the species of the subgenus Citrus (comprising all the commonly cultivated citrus fruit trees), and of Swinglea (see Tillson and Bamford, 1938, p. 791).
      It is possible that more thorough study of the developing fruits will show that Luvunga and Paramignya are so closely related to the Primitive Citrus Fruit Trees that they should be transferred out of the subtribe Triphasiinae to the subtribe Citrinae.   In the meantime, as the highest group (C) of the Minor Citroid Fruit Trees, the Luvunga group is placed next to the Primitive Citrus Fruit Trees (group A) of the subtribe Citrinae.

XII.   Luvunga Buch.-Ham.
      XII.   Luvunga6 (Lavanga) Buch.-Ham. Wall. Cat. 217. 1831; Wt & Arn. Prodr. 1:90. 1834. Lampetia Roem. Syn. Hesper. 1:42. 1846.
      Type species.Luvunga scandens (Roxb.) Buch.-Ham.
      Distribution.—India, Ceylon, Burma, Thailand, Indo-China, Malay Peninsula, Sumatra, Java, Borneo, Philippines, and New Guinea.
      Common name.—Trifoliate liana-limes (or luvunga).
      Woody lianas with single, recurved axillary spines; leaves 3-foliolate with long, wingless petioles; leaflets oval to lanceolate, acute or acuminate at the apex, entire, often slightly revolute on the margins; inflorescences dense axillary racemes or panicles; flowers 4- or 5-merous; calyx cup-shaped, 4-5 lobed; petals 3-5, linear or lanceolate, thick; stamens 6-10 (rarely fewer), with free, or more or less connate, filaments and linear anthers; ovary elongate-ovoid, seated on a columnar gynophore, locules 2-4, with 2 collateral or superimposed ovules in each locule; style gradually merging with the ovary, thick, sooner or later caducous, stigma capitate; fruits ellipsoid or globose, rind thick; seeds 1-3, imbedded in mucilaginous matter (pulp-vesicles not present!).
      Luvunga includes about twelve species which are very similar in general appearance.   All the species are woody vines that cling to forest trees by means of strong, recurved spines in the axils of the leaves.   They also agree in having trifoliolate leaves (occasionally unifoliolate or difoliolate), borne on very long, wingless petioles, and in bearing crowded clusters of strongly but agreeably scented, white, orange-like flowers in the leaf axils.   The fruits are small, globose or ovoid, usually rough-skinned, and yellow in color; they somewhat resemble miniature lemons or oranges in appearance but contain no juice.   Instead they are filled with a sticky mucilaginous substance in which the few medium-sized seeds are imbedded.
      The species of Luvunga are difficult to distinguish from a study of herbarium specimens and need critical study badly.   Only ten species are well enough known to be keyed out here, but two more of doubtful validity are added.   Few genera in the orange subfamily show so little diversity among the species as does Luvunga, but this genus is not closely related to any other of the Minor Citroid Fruit Trees except Paramignya, which is also a genus of woody vines that hold fast to the branches of trees with their reflexed spines.   Luvunga differs from Paramignya chiefly in having an unusual type of trifoliolate leaves on long, stiff petioles.   Occasionally unifoliolate leaves with short petioles are produced by species of Luvunga; such leaves bear a close resemblance to the characteristic leaves of Paramignya.   It is possible, even probable, that the genus Paramignya arose from an aberrant species of Luvunga which produced only unifoliolate leaves.
      The fruits of Luvunga show almost exactly the same structure as those of Paramignya.   They have small oil glands in the peel and much larger ones in the tissue between the peel and the locule cavities.   This dimorphism of the oil glands (possibly oil glands and mucilage glands?) is very evident in fruits of L. sarmentosa growing in the Botanic Garden at Buitenzorg, Java, collected by C. S. Sargent, October 16, 1903, now in the herbarium of the Arnold Arboretum (Tanaka, Det. No. A-325), and in L. crassifolia collected in Sumatra by H. S. Yates (No. 254) in 1922, now in the herbarium of the University of California, sheet 226,095 (Tanaka, Det. No. C-254).
      All the species of Luvunga are woody vines that clamber high on trees of tropical evergreen forests.   The fruits are small, 1 to 3 cm in diameter, globular or lemon-shaped, with a rough peel and a few medium-sized seeds.   These luxuriant vines, bearing thousands of densely crowded, fragrant white flowers, may prove useful as ornamentals in southern Florida, Puerto Rico, and other rainy tropical regions.
      1.   Luvunga scandens (Roxb.) Buch.-Ham. Wall. Cat. 6382. 1831. Limoniascandens Roxb. Fl. Ind. 2:380. 1832. Illus. Hooker, Bot. Mag. 86:pl. 4522 (col.). 1850; Paxton, Fl. Gard. 1:109. 1850-1851; Lemaire, Jard. Fleuriste 1:pl. 75 (col.). (copied exactly from Hooker, loc. cit.) 1851; Schnizlein, Iconogr. 3:pl. 224. 1857-1865; Pierre, Flore Forest. Cochinch. 4:pl. 288A. 1893; Guillaumin, in Lecomte, Fl. Gén. Indo-Chine 1:667, fig. 71(1-2). 1911; Basu, Ind. Med. Pl. pl. 194. 1918; fig. 3-16 this work.
      Type.—British India, hills about Silhet (Roxburgh).   Herb. Hort. Bot., Calcutta.
      Distribution.—India, Burma, Laos, North Vietnam, South Vietnam, Thailand; also occurs in Sumatra and Borneo (fide Tanaka).
      Common name.—Indian liana-lime (or luvunga).
      Twigs with long, acute, slightly recurved thorns; leaves variable; petioles 5-13 cm; leaflets slender, elliptic-oblong or oblanceolate, acute or acuminate at the tip, cuneate at the base, venation inconspicuous; inflorescences small, axillary cymes; flowers about 1.5-2 cm diam.; calyx with 4 short, truncate lobes; petals 4, white, fleshy, recurved, 12.5-18 mm long and 2-6 mm wide; stamens 8, glabrous, with the filaments connate nearly to the tips; ovary with 3 or 4 locules, ovules 2 in each locule; fruits oblong, about 25 X 20 cm ("size of a pigeon's egg"), rather smooth; seeds oval, somewhat pointed, integument thin, monoembryonic cotyledons oblong, green.
      This, the type species of the genus, is easily recognized by the smooth, entirely hairless staminal filaments that are fused together nearly to their tips into a tube and by the long, slender, pointed leaflets.   It is native from northeastern India through northern Burma, Laos, Cambodia, and North and South Vietnam.   Luvunga scandens is the best-known species of the genus, as it has been very well illustrated by Hooker and Pierre.   It can be easily grown in a greenhouse but, according to experiences in Europe, it does not flower unless given opportunity to climb high, as, for instance, in a palm house.
      2.   Luvunga crassifolia Tan. Med. Rijks Herb. Leiden 69:8. 1931.
      Type.—Malacca (Maingay, No. 285).   Herb. Kew.
      Distribution.—Malay Peninsula, Borneo, Sumatra.
      Tanaka's original description, translated, reads as follows: "Branches thick; spines curved; petioles short, somewhat thick; leaves very thick, obovate or oblong-obovate, rounded at the apex, frequently emarginate, often apiculate, obtuse at the base, margins slightly revolute, shiny above, with distinct veins, dull below, minutely verrucose.   Inflorescence many-flowered; flowers large; calyx deeply cup-shaped; filaments connate, puberulous.   Fruit large, tubercular, foveolate."
      This may be supplemented by the following notes Swingle was able to make from a specimen in the herbarium of the Arnold Arboretum, identified by Tanaka as L. crassifolia (Tan. ident., No. A-328; coll. by H. S. Yates, No. 766, east coast of Sumatra).
      This specimen has rather thick, semicoriaceous leaves, with 12-15 pairs of lateral veins visible on both surfaces, more conspicuous below, arising at an angle of 45°-55° with the midrib; terminal leaflets 21-22.5 X 7.5-8 cm, lateral leaflets 16.5-18.5 X 6-6.5 cm; petiole 11-14.5 cm long.   An inflorescence shows 6 fruits with a persistent, cup-shaped calyx, 7-9 mm wide, 2.5-3 mm high, with 5 (?) blunt, short lobes; pedicels 6-8 mm long, articulated with the short lateral branches (3-5 mm long) of the inflorescence; fruits (dried herb. spec.) fusiform, apex acute, with apparently persistent style, base abruptly rounded.
      This species differs from L. scandens in having blunter, thicker leaves, with evident venation, and in having pubescent filaments.   L. crassifolia was evidently included in L. scandens in many of the older floras of the Malayan region.
      3.   Luvunga nitida Pierre, Flore Forest. Cochinch. 4: text to pl. 288B. 1893. Illus. Pierre, loc. cit. pl. 288B (1-8).
      Type.—French Indo-China, Cambodia, Province of Tran (Pierre, No. 3876).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—South Vietnam: Pell Mountains in Chaudoc Province; Cambodia: Tran Province (the original localities cited by Pierre; not known elsewhere).
      Leaflets elliptic or oblong, abruptly acuminate, obtuse or acute at the tip, usually obtuse at the base (11-20 X 5-8 cm), having about 10 pairs of widely spaced, faint nerves; petioles 7-8 cm long; flowers 4-merous, peduncles 3-4 mm long; sepals 4, fused together, ciliate on the margins; petals 4, oblong, rounded and blunt-pointed at the tip, reflexed when fully open, stamens usually 6 or 7, filaments glabrous, slender, free except at the very base, where they are connate, anthers oblong; ovary, with 3 or 4 locules, seated on a disk 1 mm high, with 2 superimposed ovules in each locule; style rather thick, 4.5-5 mm long, summit expanded into the truncate stigma; fruit unknown.
      This species was considered a synonym of L. scandens by Tanaka (1930a, p. 52).   However, the free stamens and the broad leaflets, bluntly rounded at the base with only about ten pairs of faint veins, as well as the style gradually expanded into the truncate stigma, show that it is quite different.   It is much more closely allied to L. eleutherandra, from which it differs in having glabrous, filiform staminal filaments and larger leaflets with obtuse bases.
      4.   Luvunga eleutherandra Dalz. Hook. Kew Jour. Bot. 2:258. 1850. Illus. Engler, Die Nat. Pflanzenfam. 3(4):189, fig. 109,M-R (fls.). 1896; ibid. 19a:323, fig. 147,M-R. 1931; Talbot, Forest Fl. Bomb. Presid. 1:197, fig. 120 (lvs. and fts. [sic]). 1900.
      Type.—British India, Bombay Presidency, Syhadree Mountains (Dalzell, No. 54).   Herb. Kew.
      Distribution.—India, Ceylon (also reported, probably erroneously, from Malay Peninsula, Sumatra, and Java).
      Branches flexuose, with sharp, somewhat deflexed, nearly straight spines, 1.5-2 cm long; leaflets 7.5-13 X 3.5-6.5 cm, elliptic-oblong or obovate, shortly acuminate, base usually acute; petiole 2.5-7.5 cm long; flowers small, 4-merous; calyx cup-shaped, almost entire or obscurely 4- or 5-lobed, truncate, puberulous; petals 4, elliptic-oblong, reflexed, 6-7 mm long; stamens 8, shorter than the petals, filaments free, dilated, finely pubescent; style short, stout; fruits ellipsoid, small, 2.5-3.5 X 1.5-2 cm, peel rough and pitted.
      This small-leaved and small-flowered species is found, according to C. E. C. Fischer (1921), in the western part of peninsular India from Konkan southward to the Anaimalai Hills in the Coimbatore district in evergreen forests above 3,000 ft. altitude (between Lat. 10° 13' and 10° 32' N.).
      5.   Luvunga angustifolia (Oliv.) Tan. Bul. Soc. Bot. France 75:711. 1928. Luvunga eleutherandra Dalz. var. angustifolia Oliv. Jour. Linn. Soc. Bot. 5(2):44. 1861.
      Type.—Ceylon (Walker and Thwaites, No. 1195).
      Distribution.—Ceylon.
      Branches with spines like those of L. eleutherandra but slightly shorter (about 2.5 cm); leaves 3-foliolate with petioles 3.4-6.5 cm long; leaflets very variable, oval-lanceolate or linear-lanceolate, shortly acuminate at the apex, entire, 7.5-20 cm long; filaments free, subulate, glabrous (instead of finely pubescent as in L. eleutherandra); fruits 2 cm diam.; globular (instead of ellipsoid as in L. eleutherandra).
      This Ceylon species as described by Trimen, (1893, p. 224) and Tanaka (1928b, pp. 711-12) appears to be fairly distinct from L. eleutherandra.   A careful comparison should be made of the plant of western peninsular India on which Dalzell based his L. eleutherandra in 1850 and of the Ceylon plant on which Oliver established his variety beta (angustifolia) in 1861.   Both the species of western peninsular India and that of Ceylon were said by Talbot (1909, p. 197) to grow at altitudes of from 3,000 to 5,000 feet in evergreen forests.   L. eleutherandra occurs only about 200 miles northeast of the mountain region of south-central Ceylon where L. angustifolia is found.   L. angustifolia is said to show a tendency to dioecism.   Thwaites stated (1864, p. 48): "…this species would appear to be dioecious, as in some plants the ovaries of all the flowers are entirely without ovules."
      6.   Luvunga sarmentosa (Blume) Kurz, Jour. Asiat. Soc. Bengal 39(2):69. 1870. Triphasia sarmentosa Blume, Bijdr. Fl. Nederl. Indie 152. 1825.
      Type.—Java, Salak (Blume).   Rijks Herb., Leiden.
      Distribution.—Java; Sumatra; Borneo; Malaya; southern Burma: Tavoy in Tenasserim.
      Leaves 3-foliolate, leaflets usually small, about 12.5 cm long (but up to 30 cm long on young plants, according to Koorders, 1912, p. 426), obscure venation, elliptical or obovate (young leaflets sometimes linear, acuminate, as narrow as 30 X 3 cm), abruptly acuminate at the tip, sometimes short-caudate, but the very tip bluntly rounded, acute or broadly rounded at the base; flowers small, with free, very pubescent filaments; calyx 2-3 mm diam., with 5 triangular, minutely pubescent acutish lobes; fruits small.
      In spite of Kurz's recognition of this plant in 1870 as a species of Luvunga distinct from L. eleutherandra, it was not considered by other taxonomists to be a distinct species until Tanaka (1931b, p. 9) published very helpful notes on the species of Luvunga found in the Dutch East Indies (now West Irian, Indonesia).   He examined in the Leiden herbarium the type specimens of Blume on which L. sarmentosa was founded and was able to distinguish it from L. eleutherandra of western peninsular India and from L. angustifolia of Ceylon.
      7.   Luvunga borneensis Hochr. Cat. Hort. Bogor. 1:41. 1904; and (simultaneously?) in Pl. Bogor. Exsic. 19. 1904.
      Type.—Borneo, Sambus (Hochreutiner).   Herb. Hort. Bot., Buitenzorg, Java (now Kebun Raya, Bogor, Java).
      Distribution.—Known only from the original locality, but it is cultivated in the Botanic Garden in Bogor, Java.
      [After Hochreutiner]: "Differs from L. eleutherandra [probably L. sarmentosa is intended] having corolla and especially the calyx (5 mm high and 5 mm wide) larger; flowers 5-merous [instead of 3-4-merous]; stigma capitate and much larger; flower buds acorn-shaped; inflorescences fewer-flowered.   It differs from L. Motleyi in having petioles 7.5-15.5 cm long; petioles 8-10 mm long; leaflets 9-16.5 cm long and 5-7.5 cm wide; flowers in cymes no denser than in L. eleutherandra [L. sarmentosa?]."
      Unfortunately, no additional information has been published since 1904 and no additional collections of this plant have been made in Borneo.   Tanaka did not see the type specimen in the herbarium of the Botanic Garden at Buitenzorg (now Bogor), Java, nor did he report on Hochreutiner's authentic specimens of this species distributed in his Plantae bogorienses exsiccatae, No. 30, in 1904.   It has been possible to examine (in Washington, D.C.) an authentic flowering specimen of this species distributed in the exsiccatae by Hochreutiner as well as a specimen from another plant growing alongside in the same botanical garden.   They show the following characters: Spines on sterile shoots about 2 to 2.5 cm long, very strongly recurved and very sharp, but none on the flowering shoot; leaves trifoliolate; petioles 2.5 to 8.5 cm (usually 5 to 8 cm) long; leaflets broadly elliptical, abruptly acuminate at the apex, acumen blunt on the authentic flowering specimen and acute on the sterile plant, broadly cuneate or rounded (on small leaves) at base, with 10 to 14 pairs of parallel veins scarcely visible on the upper surface but very evident on the under surface in dried specimens; inflorescences axillary, composed of 1 to 4 short (3 to 4 cm), divaricately branched cymes; pedicels 3 to 4 mm long; calyx cup-shaped, very finely pubescent, 3.5 to 5 mm wide and 3 to 4 mm deep, with five broadly rounded lobes, finely ciliate on the margins; corolla and stamens lacking; pistils 7 to 9 mm long; ovary slender, finely pubescent.
      8.   Luvunga motleyi Oliv. Jour. Linn. Soc. Bot. 5(2):44. 1861.
      Type.—Borneo, Banjarmassing (Motley).
      Distribution.—Borneo, Sumatra.
      [After Oliver, 1861, p. 44]: Branches slender, glabrous; petioles of the size of a crow's quill, 10-12 in. [25.5-30.5 cm] long; petiolules sometimes 8-10 lin. [16-20 mm] long; leaflets obovate-lanceolate, entire or slightly undulate-crenate, 9-18 in. [23-46 cm] long, 4-6 in. [10-15 cm] wide; flowers on short pedicels, in dense, depressed, axillary cymes; calyx somewhat cupulate, with 5 small, triangular lobes or scarcely lobed; petals 5 (?); stamens 10, free; ovary with 3 or 4 locules.
      Tanaka (1931b, p. 9) noted that this species has extremely large leaflets with distinct venation, large flowers with shallow calyx, and free, very pubescent filaments.
      9.   Luvunga philippinensis Merr. Phil. Jour. Sci. Bot. 3:233. 1908.
      Type.—Philippines, Mindanao, Zamboanga (Whitford and Hutchinson, No. 9104).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: Mindanao Island; Borneo.
      [After Merrill]: A climbing vine; petioles 9-13 cm long; petiolules stout (5 mm); leaflets 15-25 X 6-10 cm, subcoriaceous, oblong to oblong-elliptical, apex acuminate, the acumen blunt or retruse, base acute, margins entire, with 8-10 pairs of irregular nerves, somewhat prominent beneath, anastomosing, "inflorescences of small, usually 3-flowered, racemose cymes, borne on the branches below the leaves or in the leaf-axils"; pedicels 4-5 mm long; flowers white, fragrant; calyx cup-shaped, about 4 mm wide, truncate or very obscurely 5-toothed, its stipe 2 mm long; petals 3 or 4, oblong, obtuse at apex, 9-10 X 3.5-4.5 mm; stamens free, filaments short, swollen, glabrous, about 6 mm long; ovary with 4 locules, oblong, quadrangular in cross section, about 3 mm long, 1.2 mm thick, scarcely narrowed into the stout style, nearly equaling the ovary in length; stigma capitate.
      Tanaka (1931b, p. 9) found that Luvunga philippinensis differs from his L. crassifolia in having indistinct venation, small flowers, free pubescent filaments, and large fruits borne on a small calyx.
      The original description was based on specimens with spineless twigs, such as are not uncommon in fruiting branches.   Undoubtedly rapidly growing young branches of this species will be found to bear the retrorse (but not coiled) spines characteristic of this genus.
      Luvunga philippinensis is well distinguished from L. papuana (the only species known to the south and east some 3,600 km distant in northeastern New Guinea), which has smaller acuminate leaflets, longer style, ovary with 2 to 4 locules, and much more strongly recurved spines, sometimes making a complete circle or even more!
      It also differs from the species which occur from 800 to 1,600 km to the southwest, in Borneo; (1) from L. borneensis, in having 3- or 4-merous flowers, with smaller calyx, longer petioles, and larger leaflets; (2) from L. motleyi, in having shorter petioles, smaller leaflets, and smaller flowers with glabrous filaments; and (3) from L. crassifolia, in having inconspicuous, irregular lateral veins, smaller flowers, glabrous filaments, and leaflets acute at the base.
      10.   Luvunga papuana Lauterb. Bot Jahrb. 55:260. 1918. Illus. Lauterbach, loc. cit. 55:261, fig. 7,A,K; Engler, Die Nat. Pflanzenfam. 19a:324. 1931.
      Type.—Northeastern New Guinea, Mai River, Sepik (Ledermann, No. 7392).   Cotype: Pfingstberg, Sepik (Ledermann, No. 7505).
      Distribution.—New Guinea.
      [After Lauterbach's descriptions and figures]: Branches terete, with axillary, very strongly reflexed, slender, sharp spines coiled in a spiral manner (sometimes more than one turn!); leaves 3-foliolate or rarely 1-foliolate; petioles short (about 10 mm on 1-foliolate leaves; 8-17 cm on 3-foliolate leaves); leaflets 10-18 X 6-9 cm, broadly lanceolate, obtusely-acuminate at the tip, acute and decurrent at the base, subcoriaceous, shiny above, dull below, margins entire, revolute, with about 9 pairs of inconspicuous lateral nerves; inflorescences fascicled, many-flowered, axillary cymes, 2-4 cm diam.; pedicels 5 mm long, calyx (in the flower bud) 1.3 mm diam.; petals 3-4, 6 X 3 mm; stamens 8, 5 mm long, with glabrous filaments 1-1.5 mm long in the flower bud; ovaries ovoid, lageniform, glandular, 2-locular, with 8-10 vertical furrows (probably between rows of oil glands); style thick, 4 mm long, merging gradually into the ovary; stigma somewhat capitate ; fruits globose, about 12 mm diam., rind thick, granular, white; seeds 1, 8 mm long, with a membranous testa.
      This species, which is the southeastern-most species of the genus and the only one found so far in New Guinea, is strikingly different from the others in having slender, sharp, very strongly recurved spines (shown by Lauterbach as arching downward to touch the stems below).   The ovary is shown as having glands apparently arranged in eight or ten vertical lines marked by slight ridges on the ovary.   The fruits are the smallest known in the genus, being only about 12 mm in diameter.   Both the description and the illustration show that some of the leaves are unifoliolate and very short-petioled (as short as 1 cm).   This species seems clearly distinct from all others known in the genus.
      A plant apparently belonging to this species was collected for the 1936 Richard Archbold Expedition to New Guinea by L. J. Brass (No. 7425, August, 1936) at Oroville camp, Fly River, in Papua, southeastern New Guinea.   A specimen in the herbarium of the Arnold Arboretum has Brass’s notes on the label which read: "fruits white, obovate, about 3-5 cm more or less long and 3 cm diam."   The strongly recurved spines and coriaceous leaves, shiny-green above, make it very probable that this plant is L. papuana in spite of its large fruits, and make it likely also that the small fruits mentioned by Lauterbach in the original description were immature.

IMPERFECTLY KNOWN SPECIES
      11.   Luvunga calophylla Kurz, Jour. Asiat. Soc. Bengal 39(2):69. 1870.
      Type.—Bangka Island (near Sumatra) at Jébús (Teijsmann, No. 3223).   Herb. Hort. Bot., Buitenzorg, Java (now Kebun Raya, Bogor, Java).
      Distribution.—Known only from the type locality.
      [After Kurz]: Leaves 3-foliolate; petioles 20-23 cm long; leaflets 25-30 X 4 cm, obovate-lanceolate, shiny on both surfaces, lateral nerves conspicuous; flowers in short, glabrous cymes; calyx large, 5-lobed, truncate, glabrous; petals, styles, etc., lacking; fruits (immature) oblong or ovate-oblong, showing the base of the style, "vesiculose-papillose."   Kurz said of it: "A very distinct species with leaves much resembling those of Zanthoxylum euneurum Miq. [= Evodia euneura Miq.]."
      The absence of flowers has prevented the accurate placing of this species.   Kurz knew Luvunga well and his opinion is of value.   L. calophylla is carried in the Index Kewensis (Hooker and Jackson, 1895-1938) as a valid species.
      12.   Luvunga tavoyana Lindl. ex Oliv. Jour. Linn. Soc. Bot. 5(2):44. 1861.
      Type.—Southern Burma, Tenasserim, Tavoy ("W. G[riffith]," Wall. Cat., p. 217, No. 6383).
      Distribution.—Known only from the original collection.
      [After Oliver, 1861, p. 44]: Leaves with petioles 7.5-15 cm long; leaflets 10-20 cm long, oblong-elliptical; flowers in short fascicles, or panicles, 1.2-5 cm long; stamens free or slightly united at the base; ovary with 3 or 2 locules.   "It occurs both unarmed and with strong recurved axillary spines."
      Oliver did not consider this to be a good species but rather a connecting link between L. scandens and L. eleutherandra that might warrant their union under one species.   Such a union seems very improbable in the light of our present knowledge of Luvunga.   L. tavoyana should be compared critically with L. calophylla and L. nitida.   It cannot be placed with any certainty at present.

XIII.   Paramignya Wight
      XIII.   Paramignya Wight, Ill. Ind. Bot. 1:108, 110, pl. 42. 1840. Atlantia Guill. (pro parte, non Corrêa), in Lecomte, Not. Syst. 1:176. 1910.
      Type species.—Paramignya monophylla Wt.; native to British India, Ceylon, and southern Burma.
      Distribution.—India; Ceylon; Burma; southern China: Hainan Island; Laos; Cambodia; North Vietnam; South Vietnam; Thailand; Malay Peninsula; Sumatra; Sumbawa; Borneo; Philippines; northeastern Australia.
      Common name.—Unifoliate liana-limes (or paramignya).
      Woody vines clambering up trees and shrubs by means of recurved spines; leaves 1-foliolate; petiole usually showing a pulvinoid, more or less swollen segment articulated with the leaf blade above, and the rigid portion of the petiole is also articulated with the twig below; flowers white, fragrant, often large, 4-5-merous; stamens twice as many as the petals, usually with elongate-linear anthers; disk an elongated columnar gynophore; ovary with 3-5 locules, 1 or 2 ovules in each locule; fruits globose or obovoid, often being narrowed toward the base, with a thick peel, seeds rather flat.
      As here used, Paramignya is a genus containing about twelve very closely related species, often difficult to distinguish from a study of herbarium material.   Two species native to Thailand, P. surasiana and P. rectispinosa, and one native to the island of Hainan in extreme southeastern China, P. confertifolia, have very small ovaries, much shorter than the styles, but otherwise they are very similar to the typical species of the genus.
      Besides the twelve typical species of Paramignya there are three aberrant species that may possibly prove upon study of good material not to belong to the genus Paramignya.   One of these species, P. lobata, a native of the Malay Peninsula, has prominently 2- to 5-lobed fruits, usually broader than long and bright red in color when ripe.   This species has very short retrorsely curved spines and is evidently a woody climber.   P. cuspidata, also native to the Malay Peninsula, similarly has very short spines but apparently they are not retrorse; it is not a high-climbing vine but rather a spiny straggling shrub that probably clambers over low trees, shrubs, and other vegetation.   It sometimes has recurved spiny lateral twigs on the branches that might aid in clambering over other plants.   The third aberrant species is P. trimera, which has a wide range from northeastern Australia and Timor Island to the islands of the southern Philippines and Indonesia.
      The fruits of the typical species of Paramignya have small or medium-sized (0.1 to 0.6 mm) oil glands in the peel, and larger, sometimes much larger, ones (0.8 to 2 mm) below the peel in the mesocarp or even sometimes in the endocarp, not far from the locule cavity.   In P. longipedunculata there are very commonly two layers of oil glands, one composed of very small (0.1 to 0.5 mm) glands in the peel and another (sometimes two layers deep) composed of much larger (0.5 to 1.5 or 2 mm) globular or ellipsoid glands in the mesocarp with the longer diameter tangential.   In this species, the two kinds of oil glands (the larger ones perhaps mucilage glands) are separated by a layer containing numerous small vascular bundles.
      The typical species of Paramignya are undoubtedly closely related to Luvunga.   Both of these genera include woody vines clambering high into trees and holding fast by means of recurved spines, and both genera have fruits with dimorphic oil glands.   Luvunga has trifoliolate leaves with very long petioles, whereas Paramignya has unifoliolate short-petioled leaves.   There are other minor differences such as the straight twigs (with few internodes) of Luvunga and the more or less zigzag twigs of Paramignya; also the presence of a well-defined pulvinus, at the base of the leaf in Paramignya, which twists so as to place the leaf in the proper position to catch incident light.
      Sometimes Luvunga produces unifoliolate leaves and then the petioles are very much shorter, not very different in length from those of some of the larger-leaved species of Paramignya, for example, P. andamanica.   Probably these two genera were derived from some common ancestor.   The aberrant species P. lobata and P. cuspidata may, however, be related to Pamburus.
      The species of Paramignya are difficult to distinguish unless cognizance is taken of all available characters.   Unfortunately, many of the published descriptions merely distinguish between the few species growing in a single region; this does not permit accurate comparison of species growing in distant regions.   For the present it is easiest to key out separately the species of the four principal regions where this genus is native.
      1.   Paramignya monophylla Wight, Ill. Ind. Bot. 1:108, pl. 42. 1840. Atalantia correae Guill. 1910. Illus. Wight, loc. cit. pl. 42; Engler, Die Nat. Pflanzenfam. 3(4):191, fig. 111,A-B. 1896; ibid. 19a:327, fig. 150,A-B. 1931; Talbot, Forest Fl. Bomb. Presid. 1:200, fig. 122. 1909; Guillaumin, in Lecomte, Not. Syst. 1:176, fig. 8(1). 1910; Swingle, in Bailey, Stand. Cycl. Hort. 5:2473, fig. 2763. 1916.
      Type.—British India, Courtallum (Wight, No. 115).   Herb. Univ. Glasgow (fide Tanaka, 1930b, p. 235).
      Distribution.—India: western peninsula from Konkan northeast to Sikkim, Bhutan, Khasiak; Burma: Tenasserim; Ceylon.
      Common name.—Indian paramignya.
      Leaves oblong or elliptic, rarely obovate, 6-12.5 X 2-4 cm, apices abruptly acute, sometimes acuminate, base rounded; petioles 12-20 mm long; flowers solitary or in few-flowered axillary clusters, large (25 mm diam.), 5-merous, fragrant; calyx cupular, with 4-5 broad, obtuse lobes; petals 4-5, lanceolate-oblong; stamens 8 or 10, free, filaments slender, compressed below, subulate above, glabrous or pubescent; pedicels slender, 6-10 mm long; ovary slightly furrowed, pubescent or glabrous, with 3-5 locules, usually with 2 ovules (sometimes 1) in each locule; style pubescent below, glabrous above; fruit ovoid or obovoid, pubescent or almost glabrous, slightly angled and sometimes compressed, 2.5-3.5 X 2.5-3.3 cm, yellow, odorous, sometimes only 1-seeded; seeds compressed.
      This species, the type of the genus, has never been adequately studied or illustrated.   The petioles are not clearly articulated between the leaf blade and the twig, but the upper portion is pulvinoid.   The fruit was said by Wight in the original description to be "pomaceous! endocarp 5-angled, between coriaceous and fleshy, surrounded by cellular pulp and rind, one- (always?) celled by the rupture and absorption of the partitions."   He also stated that the fruit is "obtusely 5-angled and furrowed between, clothed with short matted pubescence."   Oliver (1861, p. 42) described the ovary as "often slightly furrowed."
      2.   Paramignya scandens (Griff.) Craib, Fl. Siam. Enum. 1:235. 1926. Citrus scandens Griff. Not. Ad. Pl. As. 4:495. 1854; Paramignya citrifolia Oliv. Jour. Linn. Soc. Bot. 5(2):42 (excl. synonymy! [non Limonia citrifolia Roxb.]). 1861; P. griffithii Hook. f. Fl. Brit. Ind. 1:510. 1875; Atalantia griffithii Guill. in Lecomte, Not. Syst. 1:183. 1910. Illus. Griffith, loc. cit. pl. 587, fig. 1.
      Type.—Northern Burma, "Serpentine Mines," Mogung (?) Valley (Griffith, No. 519).
      Distribution.—Northeastern India: Assam State; northern Burma; Thailand.
      The original description by Griffith reads, translated, about as follows: "A spiny climbing shrub; spines smaller on twiglets or on inflorescences; twigs, petioles, and leaves pubescent; flowers white like the stamens, fragrant (with a suave odor), petals tinted green outside above the middle.   Habitat: In wooded hills near the Serpentine Mines, toward Burma.
      "A very distinct species, rare in upper Assam; common in the valley of the Mogoung, especially four leagues toward Serpentine Mines."
      Griffith (l.c., vol. 4, pl. 587) shows a flowering twig, 24 cm long and 4.5-5 mm diam., with 10 leaves, and a portion of a large twig, 10-11 mm diam., with a single leaf and a flower borne on a lateral twig, 4-5 mm diam.; also a single fruit.   No spines are shown on either twig.
      The plate shows the following characters: Leaves lanceolate, with acute or slightly acuminate (one leaf is obtuse) apices, gradually narrowed, then abruptly rounded at the base; leaf blades 9-13 X 3.5-5.5 cm, glabrous above (?) but with scattered hairs below, veins 6-7 pairs, visible on both surfaces; petioles 9-15 mm long, 2.2-3 mm diam., pubescent; flowers arising singly or 2 (or 3?) in the axils of the leaves, 18-23 mm long, pedicels 6-8 mm long, 2 mm wide, with small bractlets near the base, pubescent like the petioles; calyx 4.5-6 mm wide, 4-4.5 mm long, with acute lobes 1.5-2.5 mm long; petals strap-shaped, narrowed toward the rounded tips, 18-20 X 3-4 mm, glabrous; staminal filaments 13-15 mm long, slender, glabrous, anthers short, 3 X 1.5 mm (less than 1/4 the length of the filaments), long oval; pistil 18-21 mm long; ovary subglobose, 4.5 mm diam., glabrous; style glabrous, 14-17 X 1-1.1 mm, stigma capitate, 3-3.8 mm diam.; fruits ovoid, 25-30 mm, bluntly pointed at apex, rounded at base, finely pubescent (?), borne on a subglobose (?) gynophore (4.5 mm diam. below) which remains attached to the apparently glabrous persistent calyx, no larger than when the flower opened.   The plate gives every evidence of having been made carefully and accurately from a good flowering specimen.
      Hooker (1875, vol. 1, p. 510) described P. griffithii (evidently named in honor of Griffith) and included Griffith's Citrus scandens,which he stated has very small flowers (only 6 to 8 mm long); yet Griffith's plate shows flowers 15 to 18 mm long.   Hooker stated: "Griffith's figure is a great exaggeration, as his specimens show," but so great an exaggeration as this seems improbable, especially since the plate looks like a good cut made from the living plant.   Possibly the best material was used to make the plate and the dried specimens that were put in the herbarium do not show the flowers properly and may even belong to another species.
      Griffith described his plant as a species of Citrus, and because it was so utterly different from any known Citrus a detailed description probably seemed unnecessary.   It would not be surprising if Griffith mixed up two species of Paramignya in his material of Citrus scandens.
      It seems best for the present to let P. scandens stand as described and figured by Griffith and avoid attempting to include in it the widely different forms that have since been assigned to it.   New material from the type locality, Mogoung Valley (?) in northern Burma, is needed to settle the status of this species.
      2a.   Paramignya scandens subsp. ridleyi (Burkill) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Paramignya ridleyi Burkill, Gard. Bul. Straits Settl. 5:214. 1931; P. citrifolia var. beta (pubescens) Oliv. Jour. Linn. Soc. 5(Suppl. 2):42. 1861; P. griffithii Hook. f. 1875, in part (only as regards the Malacca specimen, fide Ridley); P. griffithii Ridley (non Hook f.), Fl. Malay Penin. 1:356. 1922.
      Type.—Singapore (Hullett).
      Distribution.—Malay Peninsula: Malacca and Singapore.
      Ridley described the Malay Peninsula plant under the name P. griffithii, as follows: "Shrubs shortly spiny, glabrous or pubescent.   Leaves coriaceous, elliptic or oblong, short, blunt-acuminate, 3 to 5 in. [7.5-12.5 cm] long, 1.5 to 2.3 in. [3.8-5.8 cm] wide, base rounded.   Flowers solitary or fascicled, axillary, slender, 0.3 in. [6 mm] long, white, fragrant.   Calyx acutely 5-lobed.   Petals oblong, 0.3 in. [7.5 mm] long, blunt.   Stamens 10, filaments hairy, longer than the linear-oblong anthers.   Ovary 5-celled; ovules 2 in each cell."
      Burkill (1931, p. 214), in making a new species for this plant, wrote: "P. Griffithii Hook. fil in loc. cit., [Hooker f., 1875, vol. 1, p. 510] only as regards the Malacca specimen: Ridley, Fl. Mal. Penin., 1, 1922, p. 356.   This species has much longer pedicels than P. scandens and is abundantly distinct.   It was collected in Malacca by Griffith, and by Hullett in Singapore.   The backs of the leaves are less hairy than those of P. scandens."
      Unfortunately, in the absence of type material, no additional details can be given for this subspecies.   It must therefore, for the present at least, be kept distinct not only from P. scandens, but from P. monophylla (= Atalantia correae Guill.), all of which were considered by Tanaka (1930a, p. 162) to be P. scandens.
      Even as matters stand, the subspecies is fairly well delimited by its broad elliptic-oblong leaves (7.5 to 13 by 3.8 to 6 cm) and its very small flowers borne singly or in few-flowered axillary fascicles, with acute calyx lobes and blunt-oblong petals 6 to 7.5 mm long, and hairy filaments longer than the anthers.
      It differs from P. scandens subsp. hispida in being glabrous or less pubescent with respect to leaves, calyx, and gynophore; from P. andamanica in having much smaller leaves, smaller flowers with larger calyx (?) and shorter pedicels; from the little-known P. citrifolia in having longer pedicels and filaments longer than the anthers; from P. armata in having much larger leaves, with longer petioles and stouter pedicels.   It differs from P. surasiana and P. rectispinosa by having styles only three times as long as the ovary.
      2b.   Paramignya scandens subsp. hispida (Pierre) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Atalantia hispida Pierre ex Guill. in Lecomte, Not. Syst. 1:182. 1910.
      Type.—French Indo-China, Cambodia, Bienhoa Province.
      Distribution.—Known only from the type locality.
      Differs from the species in having longer petioles and smaller flowers with shorter, broader petals.   It resembles more closely the subspecies ridleyi but has somewhat smaller leaves and anthers apparently shorter in proportion to the length of the filaments, and possibly longer recurved spines.
      Guillaumin (1911, p. 674) described Atalantia hispida Pierre about as follows: Branches slender, at first very pubescent, later on becoming glabrous, with spines about 10 mm long; leaves oval or oval-oblong, 6-10 X 2-4 cm, with a short acuminate tip, rounded or subcordate at the base, velvety-pubescent below, veins invisible on upper surface, parallel and close together below; petiole velvety-pubescent, up to more than 2 cm in length; inflorescences solitary, axillary; flowers 10 mm long, on velvety pubescent pedicels 5 mm long, with small bracts at the base; calyx cup-shaped, 2-3 mm long, sepals 5, oblong, acute (valvate in flower bud), pubescent without, glabrous within; petals 5, elliptical, 1 cm long, glandular, slightly pubescent without; stamens 10, 2/3 length of petals, filaments dilated, pubescent, usually free (often 2 or 3 connate), anthers elliptical, half as long as the filaments, not apiculate; gynophore cylindric, pubescent, glandular, crenulate above; ovary ovoid-globular, 5-angled, glabrous, shorter than the style, with 5 locules, each containing 2 ovules, style cylindrical, glabrous, stigma capitate, slightly 5-lobed; fruits small (according to Pierre).
      Tanaka (1928b, p. 712) considered this subspecies to be a synonym of P. scandens (Griff.) Craib, but it appears to have smaller flowers with decidedly shorter and broader petals.   It will not be possible to understand clearly these three forms here referred to P. scandens until studies can be made on collections of all three in comparable states.
      3.   Paramignya armata (Thwaites) Oliv. Jour. Linn. Soc. Bot. 5(2):43. 1861. Arthromischus armatus Thwaites, Enum. Pl. Zeyl. 47. 1858.
      Type.—Ceylon (Thwaites, No. 1187).   Herb. Kew.
      Distribution.—Ceylon (hotter parts of the island).
      Leaves oval, elliptic or elliptic-lanceolate, obtusely caudate-acuminate, rounded at the base, 5-8 X 2-3.5 cm; petioles short (8-11 mm); pulvinus 3 mm long, articulated with the petiole and with the leaf blade; flowers small (6 mm long), solitary or in few-flowered axillary clusters, 4-5-merous; pedicels very slender (0.4-0.5 mm diam.), 6-8 mm long; calyx minute, obtusely 4-5-lobed, lobes acute; petals oblong, very much imbricated; filaments linear-subulate, pubescent, slightly dilated in the middle; ovary pubescent, borne on a gynophore equaling it in height, with 3-4 locules, ovules usually 1 in each locule; style long, glabrous, cylindric, stigma globose; fruit subglobose, apiculate, 10-20 mm diam., yellow, 1-4 seeded.
      This species is said to have pulpy fruits but nothing is known about the nature of this pulp.
      Thwaites directed attention to its "jointed leaf stalk," a very striking character often overlooked by subsequent authors.   The petiole shows a stiff straight petiole proper, at the tip of which is articulated a short fleshy pulvinus.
      4.   Paramignya beddomei Tan. Jour. Bot. Brit. & For. 68:230. 1930. Paramignya armata Beddome (non Oliver), Icon. Pl. Ind. Or. 65, pl. 275 (excl. descript.). 1874. Illus. Beddome, loc. cit. pl. 275 (as P. armata).
      Type.—British India, Madras, Anamaly (Anaimalai Hills) (Beddome, No. 1052).   Herb. Brit. Mus., London.
      Distribution.—South peninsular India and Ceylon (fide Tanaka).
      Twigs very slender, conspicuously zigzag, pubescent with fine appressed hairs; leaves medium-sized, blade 70-75 X 35-40 mm, ovate, caudate, broadly rounded at base, pubescent on the midrib, especially below, petiole long (10-15 mm), sigmoid with an articulated pulvinus occupying the upper third, pubescent; flower buds large (10-12 X 3-4 mm), cylindric; flowers solitary, axillary, large (25-30 mm diam. when open) pedicels 8-12 mm long; calyx cupulate, lobes short, obtuse, auriculate; petals linear, bluntly rounded, 15 X 4-5 mm; filaments apiculate, very pubescent for their whole length, anthers linear, but contorted at maturity; ovary 4-5-lobed, cylindric, almost glabrous, but sparingly pubescent in upper portion, about as long as the disk, ovules 1 in each locule; style linear, very glandular, stigma depressed-globose, 5-lobed; fruit globose, not apiculate.
      Beddome's figure (l.c., pl. 275) shows short recurved single spines, 6 to 12 mm long, tapering to a fine point.   This species has been confused by previous students of Paramignya with P. armata, from which Tanaka separates it by many characters.
      5.   Paramignya citrifolia (Roxb.) Oliv. Jour. Linn. Soc. Bot. 5(2):42 (exel. descript.; valid for name only). 1861. Limonia citrifolia Roxb. Fl. Ind. 2:379. 1832; Paramignya citrifolia Hook. f. Fl. Brit. Ind. 1:510 (excl. synon.; valid for descript. only). 1875; Atalantia citrifolia (Roxb.) Kurz, Jour. Asiat. Soc. Bengal 39(2):69. 1870.
      Type.—Northeastern British India near Burmese boundary, Chittagong (Roxburgh, No. 340).   Herb. Brit. Mus., London.
      Distribution.—Reported from Bengal and northeastern India (now partly East Pakistan) by Tanaka (1930b, p. 61).
      Roxburgh's original description (under the genus Limonia) reads: "Shrubby; armed with recurved spines.   Leaves simple, elliptically oval, entire, obtusely acuminate.   Flowers axillary.   Berries ovate, few-seeded.
      "A very ramous, rigid, well-armed shrub, of five or six feet in height, a native of the forests of Chittagong, and with the other armed species, well adapted for fences.   Flowering time the hot season.
      "Young shoots polished.   Thorns axillary, solitary, short, somewhat recurved.   Leaves alternate, round-petioled, elliptic, with an obtuse, somewhat lengthened point, entire, smooth, but marked with numerous pellucid points, as in many Aurantiae; from four to five inches long, and from two to three broad.   Stipules none.   Flowers small, white, short-peduncled, axillary.   Bracts minute, about the insertion of the peduncles, and on them.   Calyx five-toothed, having its substance marked with pellucid points.   Petals five, oblong, smooth.   Filaments ten, distinct, short, inserted round the base of the germ.   Anthers linear, erect.   Germ [ovary] ovate-oblong, five-grooved, on the outside five-celled, each cell containing two ovula attached to the axis.   Style thick and short.   Stigma sub-peltate.   Berry ovate, of the colour and appearance of a lime, even to the little green cells in the cortex.   Seeds from one to four, separated by some few small fibers only, which are scarcely to be traced when dry, oblong, having the sides agreeing in shape with the number in the berry.   Integument single, membranaceous.   Perisperm none.   Embryo inverse.   Cotyledons conform to the seed.   Radicle superior."
      The type specimen in the British Museum (Roxburgh, No. 340; see Tanaka, 1930b) of two fruiting twigs from Korelea Hill, collected in 1811 (?), shows twigs, slender ultimate growth, 2-3 mm diam.; leaves 6-12 X 3-6 cm, elliptical, abruptly acuminate but blunt at tip, base gradually rounded, seldom very broadly cuneate, 12-14 pairs of lateral veins, inconspicuous above but visible on under side, arising at a wide angle (65°-70°) with the midrib; petioles 8-10 mm long, not clearly jointed to the pulvinus; fruit subglobose, 12 X 14 mm, pedicel 6-8 X 1.5 mm, on which the calyx, 5 mm wide but only 2-3 mm long, persists.   These notes are based on a photograph by Tanaka (No. 3789).
      Hooker, who compiled his description (without seeing any specimens) from Roxburgh's description and figure (Mus. [E.I.C.] 2243, fide Oliver, 1861), found the following characters: "…Young shoots polished.   Leaflet 4-5 by 2-3 in., quite entire.   Flowers white, about 1/3 in. diam.   Petals oblong.   Filaments quite free.   Ovary ovoid-oblong, 5-grooved; style short, thick; stigma subpeltate.   Berry ovoid, pointed, of the colour and appearance of a lime, even to the little green cells in the cortex.   Seeds 1-4, separated by a few small dry fibers only, which are scarcely to be traced when dry, oblong; testa membranous.—This plant is unknown to me; the description is taken from Roxburgh's Flora Indica and drawings.   Prof. Oliver referred to it in the preceding species [P. griffithii], which differs in the form of the stamens, and (if Roxburgh's figure of the ovary is correct) in wanting a style, but the said figures resemble what an ovary would be were the style fallen away.   I refer it to Paramignya from the long anther and pointed fruit.   Kurz has pointed out that Oliver's P. citrifolia and Griffith's Citrus scandens cannot be the same as Roxburgh's Limonia citrifolia (Jour. Asiat. Soc. Bengal, 1870, Pt. 2, 69)."
      This very imperfectly known species has been found with certainty only near Chittagong not far from the frontier between East Pakistan and northwestern Burma.   It is placed by Hooker in the group of species having "flowers about 1/2 inch [12-13 mm] long.   Calyx small with acute teeth," in which group he also places P. griffithii and P. armata.
      It would be very desirable to secure more good herbarium specimens from the type locality, Chittagong.   Until this is done, it will be very difficult, if not impossible, to determine with any certitude the affinities of P. citrifolia.
      6.   Paramignya grandiflora Oliv. Jour. Linn. Soc. Bot. 5(2):42. 1861.
      Type.—Southern Burma, Tenasserim (?) (Lobb, No. 925).   Herb. Kew.
      Distribution.—Southern Burma: Tenasserim; Andaman Islands; Malay Peninsula (?).
      Much like P. monophylla, having equally large, solitary flowers, 18-19 mm long, but broader leaves, broadly rounded at the base, pubescent below; petiole articulated with the pulvinus (?); calyx cupular, lobes well developed, often cuneate at tips and overlapping (imbricate?) at the base, ciliate; fruits unknown.
      This species is native in southern Burma and the Malay Peninsula and is much like P. monophylla, having large flowers with five petals; the styles pubescent below and the stamens pubescent and similar in shape to those of P. monophylla.   The type collection was credited to Singapore by Oliver (1861, p. 42) and is so labeled, but, according to Ridley (1922, p. 355), erroneously.   He suggested that it is from Tenasserim and stated: "it is otherwise known only from Tenasserim" (where it was collected by Wallich, No. 6361).
      7.   Paramignya andamanica (King) Tan. Bul. Soc. Bot. France 75:712. 1928. Paramignya armata var. andamanica King, Jour. Asiat. Soc. Bengal 62(2):223. 1894.
      Type.—Andaman Islands (King).   Herb. Kew.
      Distribution.—Andaman Islands, North Vietnam, South Vietnam, Sumatra, Sumbawa, Sabah (North Borneo.)
      King's original description and his type specimen in the herbarium at Kew give the following characters: Twigs slender, puberulous when young, soon becoming glabrous; spines single, short, curved, pubescent; leaves large, elliptical or elliptic-oblong, narrowed at apex into a blunt point, broadly rounded at base, 10-23 X 3.5-9 cm, with 10-12 pairs of veins making an angle of 65°-75° with the midrib; petiole 12-25 mm, pubescent on the upper surface when young, not distinctly articulated with the pulvinus; flowers 5-merous, paired or solitary, axillary; flower buds slender, 14-18 mm long; pedicels long, slender (18-20 mm), arising separately; calyx glabrous, short (3-4 mm), deeply divided into 5 broadly obtuse lobes; petals about 4-5 times as long as the calyx (15-16 mm), linear-oblong, obtuse; stamens free, almost as long as the petals, filaments rather thick and woolly for the lower three-fourths but filiform and glabrous above; ovary glabrous, ovoid, seated on a short cylindric disk not broader than the ovary; style elongate, slender, glabrous; fruits globular or turbinate, glabrous, 12-16 mm diam.
      Parkinson (1923, p. 108) reported it as "fairly frequent" in several of the Andaman Islands.   His specimens were more vigorous than those seen by King, as he described it as having curved spines about 5 cm long, leaves 10-23 X 4-9 cm, with petioles 25 mm long (all of these measurements are the largest yet reported for any species of Paramignya); flowers small (12-13 mm long), on slender pedicels 25-38 mm long; filaments pubescent; fruits 12-13 mm diam., globular or obscurely 2-lobed.   He found the leaves "almost scentless when bruised."
      Guillaumin (1911, p. 675) reported this species, under the name P. aramata, as occurring in Cochin China (South Vietnam).   He described it as having curved spines only 6-12 mm long; leaves 7.5-12 X 3.5-4.5 cm; flowers on slender pedicels 30 mm long; calyx short (3 mm), lobes 5, indistinct, rounded, ciliate; petals 3 times as long as the calyx; stamens almost as long as the petals, dilated and pubescent at the base, slender and glabrous at the tips; ovary globose, glabrous, 5-sided, with a style 3 times as long; stigma capitate, flattened above, attaining height of the stamens; fruit yellow, globose, borne on a pedicel 30-35 mm long, showing the remains of the eccentrically placed style.
      This remarkable but little-known plant reaches its best development in the Andaman Islands, although it grows also in South Vietnam, and, according to Tanaka (1931b, p. 10), in Sumatra, Borneo, and Sumbawa, which lie about 1,000 to 3,000 kilometers to the southeast.   The Andamans are situated about 1,300 kilometers nearly west from South Vietnam.
      It is curious that this species has not been reported from southern Burma, southern Thailand, and the Malay Peninsula.   Possibly some of the small-flowered forms of Paramignya reported from the Malay Peninsula belong to this species as varieties or subspecies rather than to P. scandens or P. citrifolia.
      This species, like Atalantia macrophylla (which also is native to the Andaman Islands), is remarkable for its extraordinary vigor, especially in its native habitat.
      It is the only species of Paramignya that approaches the dimensions of some of the species of Luvunga (the other genus of woody vines belonging to the orange subfamily) in the length of its leaves and spines.
      This species is related to P. armata, of Ceylon, in having very similar small flowers, but differs in having much larger leaves and spines, and, as King stated, in having fewer flowers in the leaf axils.
      8.   Paramignya longipedunculata Merr. Phil. Gov. Lab. Bur. Bul. 35:24. 1905. Atalantia longipedunculata (Merr.) Guill. in Lecomte, Not. Syst. 1:221. 1910.
      Type.—Philippines, Luzon Island, Rizal Province, Bosoboso (Ahern's collector).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: Luzon and Mindoro islands.
      Common name.—Luzon paramignya.
      Branches hairy, with spines 5-15 mm long, retrorse and slightly curved, densely pubescent when young, but the brownish, hardened tips glabrescent with age; terminal twigs dark green when dry, densely pubescent; leaves oval or elliptic-ovate, 7-10 X 5-7 cm, with the apex abruptly acute or broadly short-acuminate, base rounded, pubescent on lower surface, midrib densely pubescent below; petioles 8-10 mm long, densely pubescent, pulvinus not visibly articulated with the leaf blade or petiole; flowers solitary or in 2- or 3-flowered axillary clusters; pedicels densely pubescent, 1.5-3 cm long, subtended by slender pubescent bracts, 1.1-2 mm long, borne at the tips of short branches, 2.5-7 mm long; calyx cupular, densely pubescent, 7 mm diam., 5-lobed, lobes erect, broad, ovate, 2.5 mm long; petals 5, linear-oblong, 20 X 5 mm, densely pubescent without: stamens 10, filaments pubescent, 12 mm long, anthers 5 mm long; ovary with 5 locules; style about 15 mm long, densely pubescent; fruits broadly ellipsoid, bluntly pointed at apex, conoid at base, 2.5-3 mm X 1.5-1.8 cm, subtended by the persistent calyx.
      This is a large-flowered species like P. monophylla, but differs in being more or less densely pubescent in all parts.   It also has much longer pedicels than that species.   Merrill (1923, p. 339) reported that it grows "in forested ravines and in secondary forests at low and medium altitudes."
      9.   Paramignya mindanaensis Merr. Phil. Jour. Sci. Bot. 3:140. 1908.
      Type.—Philippines, Mindanao Island, Lake Lanao, Camp Keithley (Mary Strong Clemens).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: Mindanao Island; also Samar and Luzon islands (fide Tanaka).
      Branches glabrous except the flowering twigs; leaves oblong-elliptical, glabrous, 9-12 X 3-6 cm, apex acuminate, base rounded, veins indistinct; petioles 1 cm long; flowers, 1 or 2, axillary, about 17 mm long; pedicels slender, 10-15 mm long; calyx cupular, slightly hairy, about 5 mm diam., 5-lobed, the lobes 1.5 mm long, rounded; petals imbricate, oblong, 12 X 5 mm, glabrous; stamens 10, filaments thickened, somewhat pubescent, about 7 mm long, anthers 4 mm long; gynophore about 2 mm long and 2 mm diam., crenulate; pistil 10 mm long, somewhat pubescent; ovary 5-angled, with 5 locules; style stout; fruits (immature) 1.5-2 cm diam., usually curved, glabrous.
      This species differs from P. longipedunculata of Luzon Island in being, on the whole, less pubescent, since some parts are glabrous, and in having larger leaves and smaller flowers.   It has been referred to P. grandiflora as a synonym by Tanaka (1932e, p. 425), who stated that the type specimen of P. mindanaensis is a typical P. grandiflora.   However, one of Lobb's specimens of the latter species from Singapore, cited by Oliver in the original description, shows pedicels measuring 20 to 26 mm long, instead of 10 to 15 mm, as in Merrill's description of P. mindanaensis, and calyxes nearly twice as large as in the Mindanao species.   Inasmuch as P. mindanaensis occurs 1,500 miles east of Singapore, it would seem best to maintain it as a good species for the present.   Tanaka found that specimens from Luzon and Samar islands are much less pubescent than the type material from Mindanao Island.
      10.   Paramignya surasiana Craib, Kew Bul. Misc. Inform. 1916:261. 1916. Paramignya griffithii Hook. f. Fl. Brit. Ind. 1:510 (pro parte) (so far as concerns the specimen from Pegu collected by McClelland, fide Craib). 1875.
      Type.—Thailand, Chiangmai, Doi Sutep (Kerr, No. 2542).
      Distribution.—Thailand; northern Burma: Pegu.
      [After Craib]: Twigs puberulous; spines recurved, 7-9 mm long; leaves obovate or elliptic-obovate, apex obtusely caudate-acuminate, base broadly cuneate or rounded, 10-14 cm long (including the acumen, which may be up to 1 cm long), 5.2-7.5 cm wide, chartaceous, glabrous except for the shortly pubescent midrib, oil glands abundant, conspicuous on the lower surface, lateral veins straight, at least 10 on each side, prominent on the underside; petiole 1.5-1.7 cm long, puberulous; pedicels about 4 mm long with a few bractlets below; flower buds oblong, rounded at the apex, up to 1.4 cm long; calyx 5.5 mm long on the outside, curly pubescent without (like the pedicels), appressed pubescent within, lobes unequal in length, 1-2.5 mm long, with rounded ciliate tips; petals white, at least 1.7 cm long, glabrous; filaments 12 mm long, angular, flattened at the middle, nearly glabrous at base, elsewhere white pubescent; ovaries short, densely pubescent; style about 9 mm long, pubescent.
      This species is still inadequately known.   The fruits, said by Craib to have been collected by McClelland at Pegu in Burma, are not described by either J. D. Hooker or Craib.   The petals are very imperfectly known, and no measurements of the ovary are available.   However, since Craib described this species along with P. rectispinosa, which has an exceptionally short ovary, and furthermore described the ovary of P. surasiana as "short," it is clear that this species also has a very short ovary.   Even if the ovary of P. surasiana were twice as long as that of P. rectispinosa, it would still be less than one-fourth as long as the style.   More and better material of this species is much needed in order to place it accurately.   It was collected by Kerr in evergreen jungle in the same locality in which he discovered P. rectispinosa but at an altitude nearly 1,000 m higher, i.e., at 1,650 m instead of 660 m.
      11.   Paramignya confertifolia Swing. Jour. Arnold Arbor. 21:17. 1940. Illus. Swing. loc. cit. pl. 4, figs. 1, 2; fig. 3-17 this work.
      Type.—China, Hainan, San Tsuen Mountain (Tsang Wai Tak, No. 15523).   Herb. Univ. Calif., Berkeley.
      Distribution.—Southeastern China: Hainan Island and Kwangtung Province.
      A shrub, 3-5 m high, clambering over shrubs; young twigs slender, terete, pubescent, with internodes 2-5 cm long on vigorous shoots but much shorter, often only 8-10 mm long, on fruiting branches; spines short (3-10 mm; usually 3-6 mm), recurved, often reduced in size or wanting on the end branches, especially on fruiting branches; leaves oval or oblong to long elliptical, usually broadly rounded at the base (sometimes cuneate on longer leaves), the very tip of the acumen bluntly rounded, midrib and lateral veins visible on both surfaces but more distinct below, the 8-12 or more pairs of lateral veins arising at an angle of 60°-70° with the midrib, the space between being filled with smaller more or less parallel veinlets making elongate reticulations, margins entire or irregularly and shallowly crenulate; petioles 4-12 X 1-1.5 mm, flattened above, more or less pubescent, wrinkled in dried specimens, pulvinoid portion not articulated, the whole petiole curving more or less to place the leaf blades in best position to catch the incident light; flowers axillary, arising singly or in small clusters, sometimes in greatly reduced racemes; flower buds cylindrical, 8-10 X 2.5-3.5 mm; pedicels slender, 4-5 mm long, 0.5-0.7 mm wide, glabrous, subtended by minute sparsely hirsute bracts where the pedicels join the peduncles, which are hirsute; calyx small, 2 mm wide, 1-5 mm high, brownish buff-colored (as is the pedicel), lobes 5, triangular, with ciliate tips; petals 5, glabrous, white when fresh, yellowish-brown when dry, 7-9 or 10 X 3-4 mm; stamens 10, filaments flattened, 5-6 mm long, anthers linear, 1.8-2 mm long; disk cylindrical, not broader than the ovary base; ovary ovoid, 1.2-1.3 X 1 mm, strongly hirsute with yellowish-buff hairs, with 2 locules, each with 2 ovules, narrowed abruptly into the sparsely hirsute style, 4.5-6 X 0.3-0.4 mm wide at base and more slender above, dehiscent shortly after flowers open, stigma depressed globose, 0.5 mm high and 1-1.5 mm wide; fruits at first sub-globose, glabrous, but when full-sized becoming very rough, sometimes almost lobed with irregularly rugose folds of the peel, which shows numerous sunken oil glands having a turpentine-like odor and flavor, yellow when ripe, subglobose, 1.5-2 cm diam., apparently corky in places; seeds small, ovoid, monoembryonic.
      This species has been identified as P. scandens (Griff.) Craib, and as P. griffithii Hook. f., but it seems rather to be a new species allied to P. surasiana Craib from Thailand in having a very small ovary only a fraction of the size of that of the species found to the west of Thailand.   It differs from P. surasiana in a number of important characters and seems to be a good new species (see fig. 3-17).   The ovary of P. confertifolia is slightly less than a fourth as long as the style (see Swingle, 1940a, pl. 4, fig. 2).
      12.   Paramignya rectispinosa Craib, Kew Bul. Misc. Inform. 1916:261. 1916. Atalantia griffithii Craib (non Guill.), Aberdeen Univ. Studies 57:53. 1915 (?).
      Type.—Thailand, Chiangmai, Doi Sutep (Kerr No. 1718).   Herb. Kew.
      Distribution.—Thailand.
      [After Craib]: A climbing shrub with green, fleshy pubescent (soon puberulent twigs) [sic], up to 3.5 mm diam., with axillary, straight or nearly straight, pubescent spines nearly 2 cm long, green except for the straw-colored tips; leaves oblong or oblong-oblanceolate, rigid, chartaceous, obtusely-acuminate at the apex, gradually narrowed toward the rounded base, 10.5-14 X 2.8-4.7 cm, blades glabrous above except for the midrib, which is finely pubescent (very rarely subglabrous) especially at base, soft-pubescent below, with many pellucid dots [oil glands], lateral veins about 15 on each side, anastomosing inconspicuously, faint above but prominent below; petioles 10-11 mm long, slightly channeled above, pubescent like the twigs; flowers white (fide Kerr), solitary; pedicels about 3 mm long, pubescent, with small branchlets near the base; calyx 2 mm long, pubescent without, glabrous within, 5-lobed, lobes deltoid or narrow-deltoid, rather obtuse, slightly shorter than the tube; petals oblong, 12 X 3 mm, glabrous, conspicuously glandular dotted; filaments 8 mm long, pubescent, glandular like the petals, anthers 2 mm long; disk slightly shorter than the calyx; ovary about 1 mm long, densely pubescent; style 7 mm long, with divergent hairs; stigma capitate.
      Craib compared this species with P. griffithii Hook. f. [= P. scandens], "from which it differs in having spines nearly 2 cm long, straight or nearly straight."   This species is known only from Craib's original description, which shows it to differ markedly from the other species of the genus, not only in the long, straight (or nearly straight) spines, but also in its short calyx, very short ovary, and short anthers.   The type specimen was collected in evergreen jungle at 660 m above sea level.

SPECIES OF DOUBTFUL RELATIONSHIPS
      13.   Paramignya cuspidata (Ridl.) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Atalantia cuspidata Ridl. Jour. Straits Branch Roy. Asiat. Soc. 82:174. 1920.
      Types.—Malacca, Sungei Hudang (Ridley, No. 8391), and Dindings, Lumut (Ridley, No. 7944).   Herb. Hort. Bot., Singapore.
      Distribution.—Known only from the type localities in the Malay Peninsula.
      The original description reads as follows: "A spiny bush; branchlets pubescent; spines short, blunt, straight.   Leaves thin-textured, not coriaceous, ovate-elliptic, cuspidate, blunt, base round, pubescent on the back, midrib sunk above, elevated beneath, the nerves about 6 pairs, very fine, forked at tip, 2-3 in. [5-7.6 cm] long, 0.5 in. [12 mm] wide; petiole 0.2 in. [5 mm] long, pubescent, jointed in the middle.   Flowers solitary; pedicels 0.25 in, [6 mm] long.   Sepals round, pubescent.   Petals 5, linear, oblong, 0.4 in. [1 cm] long, white.   Stamens linear, oblong.   Ovary cylindric, hairy.   Style stout, glabrous.   Stigma orbicular.   Fruit ovoid, blunt, 0.8 in. [2 cm] long; rind thin, pulp scanty; seeds 2, large flattened."
      Thanks to the kindness of I. Henry Burkill, former director of the Singapore Botanic Gardens, Swingle secured a photograph from Kew Gardens of one of the type specimens collected by H. N. Ridley (No. 7944) at Lumut, Dindings, March, 1890.   A close study of this photograph shows that the petiole is twisted just as those of other species of Paramignya, because of a pulvinus just below the leaf blade, noted by Ridley who stated: "petioles…jointed in the middle."   This refers to the junction of the pulvinus with the lower rigid portion of the petiole.   As the plant does not have recurved spines like true species of Paramignya, it would seem at first sight not to belong in this genus.   However, an inspection shows that P. cuspidata has what seem to be leafy side branches that arise at right angles and sometimes curve backward.   The pulvinate pubescent petioles, the recurved side branches, and the general character of the leaves seem to indicate either an ancestral form from which the true Paramignya have evolved or else a degenerate form of Paramignya that has largely lost its climbing habit and no longer has recurved spines.   This species is connected with the group of normal long-spined species of Paramignya through P. lobata, which is a woody climber with short reverted spines.
      14.   Paramignya lobata Burkill, Gard. Bul. Straits Settl. 5:214. 1931. Atalantia hispida Ridl. (non Pierre), Fl. Malay Penin. 1:357. 1922; A. monophylla Ridl. (non DC.), Jour. Straits Branch Roy. Asiat. Soc. 26:26. 1894.
      Type.—Johore, Tumpat near Kota Pahat (Lake and Kelsall. Nov. 11, 1892).   Herb. Hort. Bot., Singapore.   Cotype: Kelantan, Tumpat near Kota Bakru (Ridley, Feb., 1917).   Herb. Hort. Bot., Singapore.
      Distribution.—Malay Peninsula.
      A climbing vine, with short (4-8 mm), retrorsely curved, stout, sharp spines with strong bases oval in cross section; twigs not angled but with slight longitudinal ridges, glabrous, apparently also slightly glaucous, internodes 1-3 cm long; leaves thin but more or less coriaceous, oblong-oval or long-elliptical, gradually acuminate, the extreme apex of the acumen bluntly rounded or subacute, bases broadly rounded or even slightly cordate, 6-12 X 2.2-5 cm, margins subentire, slightly undulate, surface densely and evenly dotted with oil glands, midrib depressed above, strongly salient below, lateral veins faint above, and not strongly marked below, 12-15 pairs arising from the midrib at an angle of 55°-60°, connected by coarsely reticulate veinlets; petioles 5-8 X 1-2 mm, wingless, but with a narrow furrow on the upper surface, glabrous, not clearly articulated with the leaf blade and not jointed in the middle, the entire petiole more or less pulvinoid and able to bend and twist to put the leaf blade in proper position to catch incident light; flowers solitary, axillary; sepals 5, oblong-acute, becoming triangular, ciliate or velvety-pubescent without; petals 5, elliptic, slightly pubescent without, 7-8 mm long; stamens 10, filaments free, velvety-pubescent; disk a cylindric gynophore, velvety-pubescent; fruits red, thin-walled, 1-1.5 cm long, 1-2 cm diam., deeply 2-5-lobed, depending on the number of seeds formed; pedicels 6-8 X 0.7-1.5 mm, more or less longitudinally ridged in the dried specimens; seeds about 9 X 6 X 4 mm, apparently monoembryonic.
      Thanks to the courtesy of R. E. Holttum, then director of the Botanic Gardens, Singapore, who sent two type specimens on which this species was based, Swingle had the best possible material to study except for the flowers, which are known only from Ridley's description of this plant cited above.   The strikingly lobed fruits of this species are quite unlike those of any other species of Paramignya, and are for that matter unique in the tribe Citreae, although they are somewhat similar to the lobed fruits of Murraya gleniei of the tribe Clauseneae.
      Paramignya lobata resembles the shrubby P. cuspidata in having short, stout spines.   Both of these species need further detailed study to determine more definitely their true affinities.
      15.   Paramignya trimera (Oliv.) Burkill, Gard. Bul. Straits Settl. 5:213. 1931. Atalantia trimera Oliv. Jour. Linn. Soc. Bot. 5(2):24. 1861; Triphasia monophylla DC. 1824; Atalantia (?) recurva Benth. 1863.
      Type.—East Indian Archipelago, Timor Island (L'Eschenault, Tanaka's Det. No. Q-1719).   Herb. Kew.
      Distribution.—Java; southern Philippines: Mindoro and Mindanao islands; Timor; northeastern Australia.
      A very full description of this species based on material from Timor (the type locality) published by Decaisne (1834, v. 439) under the name Triphasia monophylla D. C., reads in translation as follows: "Twigs erect, rounded, bark glabrous, verrucose, yellow-gray when young, becoming green.   Leaves simple, 1-2 in. [2.5-5 cm] long, 3/4 in. [18 mm] wide, oblong-ovate, subentire, sometimes emarginate at the apex, midrib prominent below, coriaceous, shiny, punctate with numerous pellucid glands, petiolate; petiole 4 -2 lin. [8-4 mm] long, somewhat glabrous.   Spinescent stipules [spines] acute, 4-6 lin. [8-12 mm] long, horizontal, with a greenish cortex, becoming gray.   Inflorescence racemose, axillary, half as long as the leaves, 8-2 flowers; pedicels very short, somewhat glabrous, with 3 bracts.   Calyx persistent, 3-toothed, glandular-punctate, with subrounded teeth, concave, margin ciliolate.   Petals 3, rarely 4, alternating with the calyx, 2 lin. [4 mm] long, obovate, concave, subcoriaceous, without veins, glandular, punctate in the middle, smooth.   Stamens 6, shorter than the petals, filaments thick, flattened at the base, free, without glands, equaling the ovary in length.   Anthers oblong-sagitate, obtuse at the apex, subglandular, punctate, longitudinally dehiscent.   Pistil almost equal to the stamens in length.   Ovary borne on a subsinuate disk, more or less cone-shaped, 2-locular, 1 pendent ovule in each locule.   Style thick, glandular-punctate like the ovary.   Stigma flattish, obscurely 3-lobed.   Fruit (immature) fleshy, obovoid, crowned at the tip with the persistent stigma, filled with gum, 2-locular, 1 seed in each locule, seeds attached at the internal angle of the segments."
      A specimen collected on Mataja Island, Mindanao, Philippines, by Kienholz (Bur. Sci., Manila, No. 15525, in Herb. Univ. Calif.), was assigned to this species by Tanaka (1932e, p. 427).   It has leaves 8-10 X 5-6 cm, oval in outline, broadly rounded at the apex, and rather abruptly emarginate; the wingless petioles, 5-7 mm long, are not articulated with the blade, wrinkled when dry, doubtless pulvinoid in the fresh state.   At each node is found a single short sharp recurved spine, 5-7 mm long.   The young ovaries are stipitate at the base, ovoid in outline, tapering into a very short style, less than 1/4 the length of the ovary, which is 2 mm long and 1 mm wide.   The leaves are much larger than those reported by Decaisne from Timor.
      This species of Paramignya is an anomalous one and may prove not to belong to this genus.   It has, however, recurved spines and a cylindrical gynophore like Paramignya.   The locule walls seems [sic] to have minute hair-like secreting emergenzen unlike any other member of this genus yet studied.

Subtribe 2.   Citrinae: Citrus Fruit Tress
      The thirteen genera of the subtribe Citrinae (the Citrus Fruit Trees), which include sixty-five species, differ strikingly from the other members of the orange subfamily in having pulp-vesicles, very peculiar structures that arise from the locule walls (especially the dorsal wall) and grow into the locular cavity, developing into sacks filled with numerous large, very thin-walled cells full of a watery juice.   No such structures are found in any other plants of the family Rutaceae or in related families, and no close homologies are known among any of the higher plants.   Other genera of the subfamily Aurantioideae have secretory glands on the locule walls, from which arises the mucilaginous gum that fills the locular cavity of the ripe fruit in most of the genera.
      The taxonomic work that has been done since the beginning of this century on the genera most closely related to Citrus has shown that the pulp-vesicles not only are of primary importance in the classification of the very close relatives of Citrus (the True Citrus Fruit Trees)7 but are even more important in the taxonomic study of the genera included in the Near-Citrus Fruit Trees and in the Primitive Citrus Fruit Trees.   The presence or absence of pulp-vesicles is indeed of paramount importance in the correct classification of the small-fruited Citrus relatives and has permitted making a sharp distinction between the two groups of small-fruited members of the subtribe Citrinae and the deceptively similar Minor Citroid Fruit Trees belonging to the subtribe Triphasiinae.
      The thirteen genera of the subtribe Citrinae fall naturally into three groups, A, B, and C.   Group A, which we may call the Primitive Citrus Fruit Trees, comprises five genera, Severinia, Pleiospermium, Burkillanthus, Limnocitrus, and Hesperethusa.   Pleiospermium has pulp-vesicles that are small, ovoid or cylindrical, blunt-tipped, and usually distinctly corticated.   Limnocitrus has pulp-vesicles that are slender, fusiform, with very acuminate tips, somewhat contracted at the base where they are attached to the dorsal wall of the locule.   Severinia and Hesperethusa have peripheral pulp-vesicles that are irregular in shape and size and little differentiated in structure.   Finally, Burkillanthus, which is a very strange genus, has pulp-vesicles of a type not found as yet in any other plant in the orange subfamily.   The walls of the pulp-vesicles are rather thick and semitranslucent.   As the fruit ripens, the pulp-vesicles apparently collapse entirely and may perhaps set free a mucilaginous fluid.   Burkillanthus has leaf, spine, and flower characters greatly resembling those found in the genus Pleiospermium, from which it differs chiefly in having much larger fruits with a tough rind and in having very numerous ovules (22 to 26) in each of the five locules, instead of two ovules as in Pleiospermium.   Probably Burkillanthus arose from a Pleiospermium-like ancestral form.
      Group B, the Near-Citrus Fruit Trees, comprises two genera, Citropsis and Atalantia, both showing well-organized pulp-vesicles with broad sessile bases and slender conical sides tapering to a rather acute apex pointed toward the center of the fruit.
      Group C, the True Citrus Fruit Trees, includes six genera with highly organized pulp-vesicles that show a slender stalk at the base (except Clymenia) and a more or less expanded fusiform body with the apex more or less acutely pointed.   The genus Clymenia has short, often subglobose pulp-vesicles with short, usually very broad stalks or almost sessile.   Nevertheless, in all its other characters, Clymenia is a True Citrus Fruit Tree, greatly resembling Citrus in external character, with simple leaves like those of Citrus medica but with venation somewhat like that in leaves of Wenzelia and Monanthocitrus of the subtribe Triphasiinae.
      The genus Hesperethusa in group A resembles Citropsis in group B in its leaf and seed characters but has much smaller fruits with much less definitely organized pulp-vesicles.   The genus Atalantia of group B resembles the genera Fortunella and Eremocitrus of group C in having small fruits with few locules, each containing only one or two ovules.   All the other genera in group C have larger fruits, with four or more (usually six or eight) ovules in each locule.
      The twelve genera that, with Citrus, constitute the subtribe Citrinae are of extraordinary interest since all of them are undoubtedly rather closely related to Citrus.   Some of them throw light on the origin and evolutionary development of the True Citrus Fruit Trees, as they represent more or less accurately the remote ancestral forms from which Citrus arose.   Every effort has been made to discover all possible characters that serve to demonstrate the evolutionary progress and present relationships of these plants.   For this reason tens of thousands of serial microtome sections were made by the modified Juel technique from herbarium specimens.   The new characters thus revealed have made possible the taxonomic placement of many genera and species up to now only imperfectly known, and sometimes known only from a single scanty herbarium specimen.   For example, in all the herbariums of the world, the new genus Clymenia, closely related to Citrus, and the new species Citropsis tanakae were each represented by a single specimen, each having only a single flower bud.   Nevertheless, by application of the modified Juel technique, it was possible to classify both of these plants with certainly.
      Thanks to this minute and detailed study of all known plants of the subtribe Citrinae, a study that has extended over many years, it became possible to discuss, with all pertinent evidence assembled, the phylogeny of the True Citrus Fruit Trees, a problem of great scientific interest because of the highly specialized pulp-vesicles filled with fragrant, delicious juice.   In the orange subfamily these organs are found only in the subtribe Citrinae and in full perfection only in the genus Citrus; nor are they known to occur in any other of the half-million or so species of higher plants now recognized by taxonomic botanists.   Keys to the groups and genera of the subtribe Citrinae are presented.

GROUP A.   THE PRIMITIVE CITRUS FRUIT TREES
      This group includes the five genera of the subtribe Citrinae, Severinia, Pleiospermium, Burkillanthus, Limnocitrus, and Hesperethusa, which show the most primitive pulp-vesicles, lacking the stalks of the pulp-vesicles of group C (the True Citrus Fruit Trees) and the definite conical shape with broad sunken bases of those of group B (the Near-Citrus Fruit Trees).
      Until three decades ago, taxonomists had not realized that the plants belonging to these five genera contained in their fruits various primitive forms of pulp-vesicles, those very remarkable structures heretofore known only in the True Citrus Fruit Trees, where they attain their most perfect development, and in the obviously related Near-Citrus Fruit Trees, where they are somewhat simpler but still impossible to overlook.   As a result of this discovery, made during the original preparation of this chapter, these five genera were transferred to the subtribe Citrinae and took their place among the near relatives of Citrus.
      Severinia has six species, all with simple leaves and with a cup-shaped disk in which the base of the ovary is partially immersed.   The very primitive pulp-vesicles are peripheral and much like those of Hesperethusa described below.
      Pleiospermium has five species: one with trifoliolate leaves, rarely with two leaflets or even one; another species with usually unifoliolate or difoliolate leaves, rarely with three leaflets; the third, fourth and fifth species always with unifoliolate leaves.   All these species have winged petioles that are rather long (one-eighth to one-sixth the length of the leaf blade) and articulated with the blade.   The locules of the fruit contain numerous small, ovoid or cylindrical pulp-vesicles, with bluntly rounded or acute tips, usually having an outer cortical shell of coarser tissue surrounding a central core of more delicate tissue that seems to disintegrate, as the fruit ripens, into a more or less spongy, oily, or resinous mass.
      Burkillanthus looks like a Pleiospermium with giant, hard-shelled fruits.   The uni-, di-, or trifoliolate leaves resemble those of Pleiospermium dubium but are larger; the petioles are narrowly winged and articulated with the leaflets, as in all the species of Pleiospermium.   The flower is larger than in Pleiospermium and the ovary is very different in morphology, having 22 to 26 ovules in each of the five-locules, the largest number known in any plant of the orange subfamily.   The fruits of Burkillanthus are much larger than those of Pleiospermium and also differ in having a firm leathery or woody rind.   They contain very numerous, very large seeds and peculiar thick-walled pulp-vesicles that seem to secrete a mucilaginous fluid.   Burkillanthus is probably descended from a Pleiospermium-like ancestor but has become so different as to constitute a very distinct genus.
      Limnocitrus also resembles Pleiospermium slightly but differs in having thick, veiny, simple leaves with very short petioles (one-tenth as long as the leaf blade) that are pulvinoid for their whole length.   The ovary differs strikingly from that of Pleiospermium.   The fruits of Limnocitrus differ from those of Pleiospermium in having very numerous, very slender, fusiform pulp-vesicles contracted at the base into a short, broad stipe; they also show a central core of whitish droplets of oil (or wax?).   These pulp-vesicles do not show the clearly corticated structure of those of Pleiospermium, or the irregular outlines of those of Hesperethusa, which are also much larger and fewer.
      Hesperethusa presents striking analogies in its leaf characters with Citropsis, the African cherry-oranges.   The pulp-vesicles in Citropsis are, however, much more highly organized than those of Hesperethusa, and show a high specialization of both structure and arrangement in the locule.   The pulp-vesicles of Hesperethusa are irregularly rounded or even polygonal from natural pressure, without specialization of form or regularity of arrangement as far as can be learned from the material now available for study.   Hesperethusa, like Citropsis, can be grafted on Citrus, and vice versa.
      This group of Primitive Citrus Fruit Trees offers promise of throwing light on the origin and evolution of the pulp-vesicles, unique organs that are found only in the subtribe Citrinae and that reach their highest development in the genus Citrus and the other closely related genera comprising the True Citrus Fruit Trees.
      In the course of a comparative study of the vascular traces of the flowers of all obtainable genera of the orange subfamily, Tillson (1938, pp. 12 and 25; see also Tillson and Bamford, 1938, pp. 787 and 791) found that all the genera comprising the Primitive Citrus Fruit Trees show fusion of the lateral sepal bundles with the petal midrib.   This unusual type of floral vascular anatomy is found elsewhere in Luvunga and Paramignya, in the species of Citrus belonging to the subgenus Citrus (comprising all the commonly cultivated citrus fruit trees), and in Swinglea.

XIV.   Severinia Tenore
      XIV.   Severinia Tenore, Ind. Sem. Hort. Neap. 3. 1840; also in Atti Terza Riunione Degli Sci. Ital. 502. 1841. Helie Roem. Syn. Hesper. 1:42. 1846.
      Type species.—"Citrus buxifolia Hort." = C. buxifolia Poir., in Lamarck, Encycl. Meth. 4(4):580 (1797) = Severinia buxifolia (Poir.) Tenore, loc. cit.
      Distribution.—Southern China, Malay Peninsula, East Indian Archipelago to the Philippines, Kei Islands, and New Guinea.
      Common name.—Box-orange or severinias.
      Leaves simple, conspicuously parallel-veined, with very short wingless petioles 1/15-1/10 the length of the leaf blade, not articulated with it; flowers small; in simple clusters in the leaf axils or in corymbs or panicles; sepals 3-5-lobed; petals 3-5; stamens 6-10, free; disk cup-shaped, enclosing the base of the ovary; ovaries with 1-5 locules, with 1 ovule pendent in each locule; fruits small, berry-like and juicy or semidry with a well-defined peel dotted with oil glands; locules with rudimentary pulp-vesicles, polyhedral through mutual pressure, often crushed by the developing seeds; seeds smooth, plump, single in each locule.
      The types [sic] species and all the other species here referred to as Severinia have in recent decades been considered to be species of Atalantia.   Atalantia, however, differs decidedly from Severinia in having well-formed, conical pulp-vesicles and much larger flowers, often with the filaments of the stamens more or less connate.   The typical species of Severinia have the ovary more or less sunken in the disk, which is, in consequence, usually broader than the ovary.   As noted above, Severinia has very primitive, stalkless, peripheral pulp-vesicles very like those of Hesperethusa; in neither genera is there indication of the contents breaking down into a more or less resinous mass, such as is seen in Pleiospermium.   Severinia differs decidedly from Hesperethusa in having simple leaves and wingless petioles.   The leaves of Severinia are strongly veined, the finer veins forming more or less uniform reticulations, not clearly marked in Hesperethusa.   A still more striking difference is in the disk, which is cup-shaped in Severinia and annular in Hesperethusa.   Severinia resembles Atalantia in the shape, size, and venation of its leaves and in having wingless petioles, but differs decidedly in having nonarticulated petioles and more primitive pulp-vesicles without definite form, those of Atalantia being conical and pointing toward the center of the fruit with a rather broad base, more or less sunken in the tissues of the dorsal locular walls.   Severinia also resembles somewhat several of the genera in the subtribe Triphasiinae, especially the genera Paramignya and Pamburus, from which it differs in possessing rudimentary pulp-vesicles.
      Severinia lauterbachii is very different from the first five species of the genus and is as yet only imperfectly known.
      Severinia buxifolia and S. disticha are known to have served as rootstocks for Citrus, and it is probable that other species can be so employed.
      The six species of Severinia are keyed.
      1.   Severinia buxifolia (Poir.) Tenore, Ind. Sem. Hort. Neap. 3 (?). 1840; also in Atti Terza Riunione Degli Sci. Ital. 501-503. 1841. Citrus buxifolia Poir. in Lamarck, Encycl. Meth. 4:580. 1797; Limonia monophylla Lour. (non L.) Fl. Cochinch. 1:271. 1790; Citrus emarginata Desf. 1829; Limonia bilocularis Roxb. 1832; Sclerostylis atalantioides Wt. & Arn. Prodr. 93. 1834; Atalantia loureiriana Roem. Syn. Hesper. 1:44. 1846;8 Helie atalantioides (Wt. & Arn.) Roem. Syn. Hesper. 1:42. 1846; Sclerostylis buxifolia Benth. Hook. Jour. Bot. 3:326. 1851; Atalantia buxifolia Oliv. Jour. Linn. Soc. 5 (Suppl. 2):26. 1861; Severinia monophylla ("L.") Tan. Bul. Mus. Hist. Nat. Paris, 2 ser. 2:163. 1930.9 Illus. Seemann, Bot. Voy. Herald, pl. 81. 1852-1857; Penzig, Studi Bot. Sugli Agrumi, Atl. pl. 11, figs. 6-17; pl. 12, figs. 1-21. 1887; Swingle, Jour. Wash. Acad. Sci. 5:656, 657, figs. 1, 2. 1916; Tillson & Bamford, Amer. Jour. Bot. 25:784, figs. 21-25. 1938; fig. 3-18 this work.
      Type locality.—In the vicinity of Canton, China (Sonnerat).   Herb. Mus. Nat., Paris.
      Distribution.—Southern China: Kwangtung Province, Hainan Island; North Vietnam; South Vietnam; Taiwan; Hong Kong; Cambodia; Laos.
      Common name.—Chinese box-orange.
      Bentham's description (1851, p. 326) may be translated as follows: "Low shrub, or in our gardens dwarf tree, branches spreading, young twigs obscurely angular or somewhat compressed, always puberulous, soon glabrous and terete.   Spines axillary, strong, usually much shorter than the leaves, only rarely longer.   Leaves 1-1 1/2 in. [2.5-3.8 cm] long, 6-9 lin. [12-18 mm] wide, obtuse, narrowed at the base into a short, narrow, rounded petiole, coriaceous, rigid, densely dotted with oil glands, glabrous, with numerous parallel veins arising from the midrib, much more conspicuous than in many species [of Sclerostylis = Atalantia].   Flowers between small bracts in the axils of the leaves, sessile, or rarely borne on a very short pedicel, never solitary, usually 2-3-fasciculate.   Calyx widely campanulate, short, smooth, with 5 rounded, very obtuse lobes.   Petals 5, scarcely 2 lin. [4 mm] long, oblong, erect, glabrous.   Stamens 10, shorter than the corolla, filaments dilated, attenuate at the apex, anthers ovate, parallel locules.   Ovary sessile on a cupulate disk, subglobose, fleshy, with small locules in the centers, ovules attached to the central axis slightly below the apex.   Style very short, stigma thick, ovoid, pulvinate.   Berry depressed globose, becoming black at maturity, seeds large, ovoid."
      The original description of Poiret (1797, pp. 580, 581) reads, in translation, as follows: "Citrus buxifolia, leaves subsessile, ovate-retuse, flowers racemose, very small…I do not know its height, but the branches I have examined in the herbarium of Citizen Lamarck, would indicate that it is not tall.   With its short, stiff branches, its spines and the shape and firmness of its leaves, it has the aspect of Rhamus pyracantha.   The wood is hard, white, with a smooth green bark.   It has many spreading branches with stiff straight spines, yellowish at the tips.   The leaves are few, alternate, oval, almost sessile, much like those of box, but twice as large, obtuse at the apex, emarginate, narrowed at the base, remarkable by the very close veins, which are prominent and parallel, coriaceous, with entire margins.   The petioles are simple and very short.   The flowers occur in small bunches toward the ends of the branches.   The corolla is white and very small.   This plant is native to China and was observed there by Sonnerat, who sent the specimens to Citizen Lamarck."
      This, the type species of the genus Severinia, is not an uncommon plant in wooded areas in southern China as far north as the Tropic of Cancer.   It is also found in Hong Kong and Hainan, and in Indo-China from the northern border of Laos and North Vietnam to latitude 10° S.   It is a small, spiny tree with crooked, thorny branches and dense, evergreen foliage (see fig. 3-18).   It is esteemed by the Chinese for its leaves, which are used in making yeast cakes.   For this reason it is called 'tsau 'ping lak10 in Cantonese, meaning "wine cake thorn."
      A careful search for variations in S. buxifolia, made by G. W. Groff and F. A. McClure, formerly of the Horticultural Department of Lingnan University at Canton, led to the discovery of many forms of the species differing in height, character of growth, size and shape of the leaves, number and length of spines, et cetera, and also in physiological characters, such as congeniality with Citrus in grafting experiments.   So far, none of these forms have been given taxonomic names, but some appear so distinct that when they are studied more carefully it may prove desirable to recognize them as subspecies, varieties, or forms.
      Certain broad-leaved forms growing near the seashore at Da Nang (Tourane) in South Vietnam (Lat. about 16° N.) have shown not only great vigor under culture but also high resistance to boron and injurious saline solutes often found in the ground water where Citrus is grown in semiarid situations.
      Inasmuch as Citrus can be grafted readily on S. buxifolia, and since some plants so grafted have lived over thirty years without showing any sign of bad graft union, it is clear that more study should be given to the use of this species as a rootstock.   Besides its important physiological distinction from Citrus in respect to boron absorption and sensitivity toward saline solutes in the ground water, the very fact that S. buxifolia is so remote a relative of Citrus as to be quite immune to many Citrus diseases recommends it for trial as a rootstock.
      With regard to the relative boron tolerance of the lemon and the box-orange, Eaton and Blair (1935, p. 413) stated: "The lemon (Citrus limonia Osbeck) is highly sensitive to boron; in fact few plants are thought to be more sensitive, and it accumulates much boron in its leaves.   The Chinese box orange, Severinia buxifolia, on the other hand, in confirmation of the observations of Swingle, Robinson, and May, is relatively tolerant to boron.   Severinia has been found to accumulate much less boron in its foliage than does the lemon."
      Eaton and Blair conducted experiments on boron tolerance of the lemon and of the Chinese box-orange, as well as of reciprocally grafted plants (i.e., lemon on box-orange roots and box-orange on lemon roots) which Swingle supplied them with the help of Eugene May, Jr.   Striking proof that the box-orange has much higher resistance to boron-poisoning than the lemon is given in their Table II reproduced below (table 3-2).
      Eaton and Blair commented as follows on the data presented in the table: "Lemons [grafted] on box orange roots had about one-third as much boron in their leaves as when grown on their own roots.   The data indicate that the boron concentrations in the box orange leaves are increased threefold when the plant is grown on lemon roots.   There were no box orange plants on their own roots in the 2 and 4 ppm cans, but the plant in 6 ppm on its own roots had 390 ppm, whereas the box orange on lemon roots in 4 ppm had 877 ppm of boron in its leaves.   Lemon leaves on their own roots were more severely injured than when on the box orange roots."
      This record of experiments, made after years of experience in growing Citrus in nutrient solutions containing known amounts of boron, is of great value in showing the striking difference in the toxicity of boron for Citrus roots and for Severinia roots, although both genera of plants grow wild in situations that are similar with respect to both climate and soils.   (See also discussion of boron tolerance of Eremocitrus.)
      The possibility of growing Citrus on a plant as remote, taxonomically, as the box-orange has served to direct attention to the need of testing all the wild relatives of Citrus in the hope of finding others able to thrive as the box-orange does, in physical and biological environments that Citrus roots cannot endure, and at the same time able to serve as rootstocks for the commonly cultivated species of Citrus.
      1a.   Severinia buxifolia brachytic form.
      Common name.—Dense-leaved box-orange.
      Twigs spineless or nearly so, internodes very short, often less than 1 cm long; leaves densely crowded because of the short internodes; flowers and fruits as in the species but produced in less abundance.
      This form, analogous to the myrtle-leaved variety of the sour or Seville orange, Citrus aurantium var. myrtifolia, is even more like the common box (Buxus sempervirens) in general appearance than the normal form of the Chinese box-orange.   This dense-leaved variety is, like the parent species, able to endure high temperatures, hot winds, and other unfavorable climatic conditions that would injure or kill the true box tree.   It cannot, of course, endure severe cold, like the common box, and is injured when the temperature falls as low as 14° F (-10° C).   This beautiful, vivid green, nearly thornless shrub deserves careful trial as a substitute for the common box for hedges in warm-temperature and subtropical climates.
      This form seems to arise by mutation from the mother species.   As might be expected, intermediate forms occur as, for example, the one shown in figure 3-18, which has some nearly thornless branches and others with long spines borne on the same plant.   Even the typical forms of this mutation are usually not entirely thornless but have short, slender spines that are, however, so completely hidden under the very abundant leaves that the spines are scarcely felt when a branch is touched.   Doubtless by careful selection superior, dense-leaved, thornless forms of this variety can be developed that will be of high ornamental value.
      2.   Severinia disticha (Blanco) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Limonia disticha Blanco, Fl. Filip. 356. 1837; L. corymbosa Bl. Fl. Filip. ed. 2. 251. 1845; Sclerostylis nitida Turcz. Bul. Soc. Nat. Mosc. 31:249. 1858; Atalantia nitida Oliv. Jour. Linn. Soc. 5 (Suppl. 2):25. 1861; A. disticha (Blanco) Merr. Phil. Gov. Lab. Bur. Bul. 27:28. 1905.
      Type.—Wanting.   Substitute type: Luzon (Merrill, Species Blancoanae, No. 594).   Herb. Bur. Science, Manila.
      Distribution.—Philippines: Luzon to Mindanao islands; Sabah (North Borneo): Banguey Island.
      Common name.—Philippine box-orange.
      Young twigs angular, sometimes with scanty gray pubescence soon glabrous and terete, sometimes apparently glabrous from the beginning; twigs spineless but occasionally with paired, slender, sometimes pubescent, spine-like paraphylls rarely equaling the petioles in length; leaves 3-8 (rarely 10-11) cm long by 2.5-4 (rarely 1.5-2) cm wide, oval, ovate-lanceolate or sometimes fusiform, acuminate or acute at the apex, acumen often blunt and frequently emarginate, narrowed gradually into a cuneate base which merges into the petiole, margin subentire sometimes slightly crenulate and frequently wavy in dried specimens, lateral veins visible above but slightly more prominent below, extremely numerous, parallel, straight, occasionally showing 15-20 pairs, arising at an angle of 55°-60° with the midrib; petioles 3-9 (usually 6-8) mm long, flattened above with a furrow made by the decurrent margins of the leaf, sometimes pubescent, sometimes almost glabrous, especially on the under side; inflorescences many-flowered, axillary or rarely terminal racemes, 3-6 cm long (rarely 8-9 cm long, and much branched); pedicels slender, 3-5 mm long, with minute bractlets at their juncture with the peduncle; both peduncles and pedicels pubescent; calyx persistent, sepals 5, imbricate, broadly rounded, margins ciliate; petals 5, 3.5-4.5 mm long, 2-3 mm wide, oblong or obovate; stamens 10, nearly as long as the petals, filament flattened, broad, free, anthers short, small, ovate; ovary subglobose, small (1.5-2 mm long), 2-locular, with 2 large glands at apex, 1 over each locule; disk cupulate, partly enclosing the base of the ovary; style as long or longer than the ovary, stigma ovoid, slightly broader than the style; fruits subglobose, 6-8 mm diam. when mature, peel yellowish-green, with many large, translucent oil glands, 0.5-0.75 mm diam.; seeds thick, monoembryonic, testa exceedingly thin, embryo olive-green when dry and covered with numerous minute oil glands, 0.1 mm diam.
      This species, very abundant in the northern Philippine Islands, is so like the box-orange, S. buxifolia, that when it was found the latter could be used successfully as a rootstock for Citrus tests were also made with S. disticha.   It was found to serve as a rootstock almost as well as the box-orange and, like it, to be resistant to amounts of boron in the soil moisture that would be injurious to Citrus grafted on the commonly used rootstocks belonging to the genus Citrus.   Swingle made these tests over thirty years ago in a greenhouse at Washington, D.C. by watering potted seedling plants of Citrus, Severinia, and other genera of the orange subfamily with water containing 2 ppm of boron.   Because of evaporation from the soil and transpiration from the plant, the boron accumulated in the soil rather rapidly and soon the Citrus seedlings showed the characteristic yellowish-green mottling of the leaves, followed by severe injury and even death.   The seedlings of Severinia buxifolia and S. disticha remained healthy for many weeks after those of the Citrus species showed severe injury.   Severinia disticha has fruited freely in Miami, Florida.
      3.   Severinia linearis (Blanco) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Limonia linearis Blanco, Fl. Filip. 357. 1837; Atalantia linearis Merr. Phil. Jour. Sci. 1 (Suppl. 3):200. 1906.
      Type.—Wanting.   Substitute type: Luzon, Rizal Prov. (Merrill, Species Blancoanae, No. 746).   Herb. Bur. Sci., Manilla [sic].
      Distribution.—Philippines: Luzon Island.
      Common name.—Narrow-leaf box-orange.
      Merrill's description of Atalantia linearis reads as follows: "A shrub 1 to 3 m high.   Branches light gray, glabrous, the young branchlets greenish, pubescent or puberulent, terete.   Leaves alternate, linear to narrowly linear-oblong, or linear-lanceolate, glabrous, shining, coriaceous, 2 to 7 cm long, 3 to 10 mm wide, the base acute, the apex blunt, retuse, entire, the margins often somewhat recurved, the midrib prominent, the lateral nerves numerous, scarcely more distinct than the dense reticulations; petioles glabrous or pubescent, 3 mm long or less.   Inflorescence of terminal and axillary panicles, 3 to 5 cm long, pubescent, the branches short, few-flowered.   Flowers white, short-pediceled, and 6.5 mm long.   Calyx short, regularly 5-lobed, the lobes imbricate, broadly suborbicular, rounded, about 2 mm long, 2.5 mm wide, pubescent, the margins ciliate.   Petals 5, free, oblong, the apex rounded, 6 to 6.5 mm long, and 3 mm wide, somewhat narrowed at the base.   Stamens 5 [10]; filaments broad, 4 mm long; anthers about 1-4 mm long.   Ovary glabrous, 3-celled, sessile, the disk shallow.   Fruit glabrous, globose, white, glandular-punctate, about 8 mm in diameter.…
      "On cliffs and boulders at an altitude of about 40 m along the river, frequently in situations submerged at high water associated especially with Eugenia mimica Merr., and sometimes with Homonoya riparia Lour.…A species at once recognized by its very narrow leaves, which are retuse at the apex.   Blanco's original description of this species is very short, his material being from the Island of Maricaban, Province of Batangas, Luzon, flowering in the month of July.   In this description he states that the leaves are minutely serrate, probably from the fact that in specimens with recurved leaf margins, the slightly raised veins on the upper surface appear like minute teeth.   The note following the description of the species in the first edition of the Flora di Filipinas is excluded in the second, this note referring to a similar form observed by Blanco in the Province of Bulacan, Luzon.   I am of the opinion that the form here described is identical with Blanco's Limonia linearis, and accordingly his specific name is adopted and the species is redescribed under Atalantia."
      This curious species is apparently a satellite species derived from S. disticha that has become adapted to growing in situations exposed to flooding.   Perhaps the narrow leaves resist the tearing action of flood waters rushing down narrow valleys.   The lateral veins of the leaves run less regularly parallel than those of the leaves of S. disticha.   Perhaps this, too, is an adaptation to prevent splitting of the leaves by rushing waters.
      Extreme specimens of S. linearis are easily distinguished by the narrow leaves, but broad-leaved forms are hard to distinguish from narrow-leaved forms of S. disticha.   A broad-leaved specimen of S. linearis (coll. by M. P. Sulit, April, 1926; Tanaka’s ident. No. C-305, Forestry Bur., No. 30320, Herb. Univ. Calif.) has leaf blades measuring 9 to 11 cm in length, and 1.5 to 2 cm in width!   However, the venation of the two species differs slightly, lateral veins of S. linearis running more irregularly and distinctly less parallel than those of S. disticha.   It is possible that natural hybrids may be in play here.
      4.   Severinia paniculata (Warb.) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Atalantia paniculata Warb. Bot. Jahrb. 13:340. 1891; A. maritima Merr. Phil. Jour. Sci. Bot. 9:293. 1914; A. disticha var. paniculata Tan. Jour. Arnold Arbor. 9:141. 1928.
      Type.—Ceram Laut (Warburg, No. 20132).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Moluccas: Kei Islands, Ceram Laut; southern Philippines: Mindanao, Negros, and Panay islands; North Borneo (Sabah): Biak; Sumbawa and Java (fide Tanaka, 1928c, p. 141).
      Common name.—Bouquet box-orange.
      Warburg's original description of this species reads in translation as follows: "A shrub with gray-pubescent [terminal] branches, soon glabrate, with yellow wood; leaves oblong-ovate, obtuse or emarginate at the apex, margin obscurely crenulate; petioles short, pubescent, nearly terete; inflorescences terminal, paniculate, pubescent, pedicels short; calyx 5-lobed, lobes pubescent and ciliate, subtruncate at the tips; petals nearly glabrous, lanceolate, obtuse; filaments long, glabrous, anthers broadly cordate; ovary with 4 locules; style deciduous; fruits globose, with the calyx persistent, 2-4-seeded.   The youngest twigs are reddish-yellow, the somewhat older twigs brown after the gray pubescence has fallen, often with a whitish bloom; the petioles are 8 mm long, the leaves 9-12 cm long and 3-5 cm wide, with a large number of parallel veins of which some are more prominent, uniting near the margins in arches.   The whole inflorescence is 4-6 cm long, branched from the base, the flower pedicels are from 2-4 mm long, the calyx is about 2 mm long, the sepal tips 1 mm long, the petals and stamens are 4 mm long; the largest fruit of my specimen [Warburg, No. 20132] is 12 mm in diameter."
      This species is easy to distinguish from the related species because of its pubescence, its lack of thorns, and its terminal paniculate inflorescences.
      Severinia paniculata was discovered on Ceram Laut and Kei Islands in the Moluccas west of New Guinea growing on dry areas in the brush.   Tanaka considered it to be identical with Atalantia maritima, growing in the southern Philippines, but Swingle could not follow him in making the species a variety of S. disticha, although it is closely related to that species.   It differs in having larger leaves and smaller, paniculate rather than racemose inflorescences, and in having ovaries with four instead of two locules.
      5.   Severinia retusa (Merr.) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Atalantia retusa Merr. Phil. Jour. Sci. 1(Suppl. 4):200. 1906.
      Type.—Philippines, Palawan Island, Puerta Princessa (Curran, No. 3609).   Herb. Bur. Sci., Manila.
      Distribution.—Known only from the type locality.
      Merrill's original description of this species reads as follows: "A shrub about 3 m high, with oblong-elliptical-ovate, subcoriaceous, glabrous, strongly retuse leaves, racemose inflorescences and 5- to 7-merous, the stamens 10 to 15, free, the ovary 1-celled.   Branches brownish-gray, glabrous, the branchlets green, glabrous.   Leaves 5 to 9 cm long, 2.5 to 4 cm wide, shining, broad at both ends, scarcely narrowed above, the base rounded; nerves numerous, anastomosing; petioles 8 mm long or less, the spines short or wanting.   Racemes in the upper axils, 1.5 cm long in anthesis, densely flowered, puberulent.   Calyx lobed, the teeth 5 to 7, short, broad, regular, their margins ciliate, obtuse.   Petals 5 to 7, glabrous, oblong, obtuse, 5 to 5.5 mm long, 2.5 mm wide.   Stamens 10 to 15, unequal, the filaments free, 2.5 to 3.5 mm long, anthers broadly ovate, slightly exceeding 1 mm in length.   Ovary oblong, glabrous, 1-celled, 1-ovuled; style much shorter than the ovary, including the stigma about 1 mm long.   Disk thickened, ring formed.   Fruit (immature) ovoid, glabrous.
      "A species apparently related to Atalantia disticha (Blanco) Merr., differing from the latter in its leaves not being narrowed above, and in its flower characters.   In some cases a very rudimentary second cell was observed in the ovary, showing that the species is perhaps sometimes with 2-celled ovaries.   The very short style is another character, as well as the variable number of calyx teeth, petals and stamens.   In other species of the genus the ovary is from 2- to 5-celled, the petals and calyx teeth 3 to 5, and the style equalling [sic] or longer than the ovary."
      This species is described as having an ovary with only one locule, but a specimen in the National Herbarium at Washington, sheet No. 439318, from Palawan (J. Bermejos, Bur. Sci., No. 195), shows ovaries with two locules.   Penzig (1887, p. 158) reported finding, very rarely, ovaries of S. buxifolia that were "monocarpellate without the slightest trace of a second ovarial follicle."   A specimen of Atalantia maritima (considered by Tanaka [1932e, p. 426] to be a synonym of S. paniculata) in the National Herbarium at Washington, sheet No. 1294389, from Zamboanga, Mindanao Island, P.I. (A. Lomente, Forestry Bur., No. 25288), shows only a single carpel in the ovary and a single stylar canal.   Three of the typical species of Severinia occasionally (or regularly) show only a single locule in the ovary and probably the other closely related species, S. paniculata, may occasionally show a hypomerous or even a monocarpellate gynoecium.11

SPECIES OF UNCERTAIN RELATIONSHIPS
      6.   Severinia lauterbachii Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Atalantia litoralis Lauterb. (non Guill.) Nova Guinea 14:146. 1924.
      Type.—Northwestern New Guinea [West Irian, Indonesia], Schouten Islands, Biak, near Warsa (Lat. 2° S., Long. 137° E.), growing on coral limestone strand (Feuilleteau de Bruyn, No. 320).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Known only from the type locality.
      Common name.—Triangle-fruit severinia.
      The original description, translated, reads as follows: "Small tree, about 4 m high, with slender, flattened, glabrous twigs, 4 mm thick.   Leaves elliptic or broadly lanceolate, subacutely rounded at the apex, sometimes emarginate or subacute, base subacute, margins obscurely crenulate, 12-16 cm long, 6-8 cm wide, coriaceous, glabrous above and below, lateral veins 16, obliquely subparallel, curved and anastomosing near the margin, veinlets subparallel, reticulate, when dry more prominent below like the midrib; petiole flattened on upper surface, 6-7 mm long.   Inflorescences axillary, paniculate, many-flowered, 3-6 cm long, small branchlets minutely pubescent.   Flowers pedicellate, pedicels 3-4 mm long; buds obovoid; calyx 5-lobed, lobes rounded, ciliate, 1 mm long; petals 5, oblong, rounded, white, glandular, glabrous, margins subpellucid, 5 mm long, 1.5 mm wide; stamens 10, 5 of them shorter, 3 or 5 mm long, both [lengths] connate at the base, filaments dilate, glabrous anthers cordate, about 1 mm long; disk short, 10-crenulate; ovary triangular, each angle crowned with a large gland, 3-locular, 1.3 mm long; style clavate, 2 mm long, stigma decurrent, somewhat 3-lobed."
      This species was said by Lauterbach to differ from Atalantia paniculata Warb. (here S. paniculata) in having much larger leaves and larger flowers with triangular ovaries, each having three locules.   The term "10-crenulate" applied to the disk of this species probably indicates that it is cupulate (as in other species of Severinia), but in drying it has become separated from the base of the ovary and acquired a wavy margin.   It is possible that this curious species does not belong to the genus Severinia, but it is impossible to place it with certainty until the flowers and fruits are better known.
      Merope angulata, widely distributed in tidal swamps, also has triangular (sometimes 4- or 5-angled) fruits, but it has very large seeds, stout paired spines in the leaf axils, an ovary with two to four pendent ovules in each locule, and a cylindric disk definitely not cupulate or 10-crenulate.   Merope angulata is probably very different from S. lauterbachii, although the latter species is also a strand plant with triangular fruits.
      This species was named Atalantia litoralis by Lauterbach in 1924, but this name was preoccupied by Atalantia littoralis (Miq.) Guill., a new combination made eleven years previously by Guillaumin (1913, p. 441) when he transferred Paramignya littoralis Miquel to the genus Atalantia.   The variation in spelling, litoralis or littoralis, does not make these names distinct, as these are merely variant spellings of the same Latin adjective.   Lauterbach's specific name, therefore, has no standing and must be supplanted by the new name which is used here.

XV.   Pleiospermium (Engl.) Swing.
      XV.   Pleiospermium (Engl.) Swing. Jour. Wash. Acad. Sci. 6:426. 1916. Limonia Sec. Pleiospermium Engler, Die Nat. Pflanzenfam. 3(4):189. 1896.
      Type species.Limonia alata Wt. & Arn. = Pleiospermium alatum (Wt. & Am.) Swing.
      Distribution.—Southern India, Ceylon, Sumatra, Java, Borneo.
      Common name.—Orangeasters.
      Small trees; twigs with 1 or 2 spines in the axils of the leaves or unarmed; leaves 3-foliolate, 2-foliolate, or 1-foliolate, margins entire; petioles narrowly winged or nearly wingless (broadly winged in P. latialatum); inflorescences axillary or terminal panicles of few-flowered groups of flowers arising in the axils of the leaves; flowers 12-24 mm diam.; calyx 4-5-lobed, sepals deltoid or lanceolate-acuminate (in P. longisepalum, long, strap-shaped, persistent, reflexed); petals 4 or 5, oblong or linear-oblong, rounded at the tips, 6-12 mm long; stamens 8 or 10 (twice as many as the petals), filaments free, glabrous; anthers oblong or linear-oblong; pistil subsessile or short-stalked; disk cupulate, enclosing base of ovary for 0.5-0.8 mm; ovary cylindric or ovate, with 4-5 locules, ovules 2 in each locule; style slender or thick, not sharply delimited from the ovary, ending in a more or less capitate stigma with 4 or 5 stylar canals, with 2 small oil glands between each pair; fruit globose or oblong, 20-30 mm diam., with slender rudimentary pulp-vesicles, 2-10 mm long, that are more or less corticated and show more or less disintegration of the contents into an oily or resinous mass as the fruits ripen; seeds about 1 cm long, flattened.
      This genus includes some of the most puzzling species of the subtribe Citrinae.   These species have been referred to several different genera, usually with a question mark!   None have been examined thoroughly in the living condition and it is very desirable that a study be made of the structure of the fruit and in particular of the peculiar pulp-vesicles, which at some stages of their development have very much the appearance of stalked oil glands.
      The two always unifoliolate species from northeastern Borneo, P. longisepalum and P. latialatum, and the one from Sumatra, P. sumatranum, are somewhat like Atalantia in appearance, but differ in having well-developed winged petioles plainly articulated with the leaf blade, as in the two other species of Pleiospermium which have tri-, di-, or unifoliolate leaves.   The very curious new genus Burkillanthus has a marked superficial resemblance to Pleiospermium in its leaf characters but differs widely in its fruits.   It is probably a large-fruited form derived from a Pleiospermium-like ancestor.   is remotely possible that Citrus itself may prove to be another large-fruited genus derived from a remote ancestor much like some of the existing species of Pleiospermium.   The anomalous trifoliolate leaves of Poncirus, with winged articulated petioles, unique among the True Citrus Fruit Trees, may perhaps be explained as a reversion to the trifoliolate leaves found in the two more primitive species of Pleiospermium, occasionally in P. dubium and regularly in P. alatum.
      The pulp-vesicles in the genus Pleiospermium, although very primitive in comparison with those of Citrus, show a very interesting series of stages in the evolution of these unique and taxonomically highly important organs.   In
P. dubium the pulp-vesicles are long-ovoid or cylindrical, each with a very distinct tissue surrounding a definitely limited central body that looks very much like a rutaceous oil gland, and, like such an oil gland, degenerates at length into a granular, apparently structureless mass of oily or resinous (?) substance.   In P. sumatranum the pulp-vesicles are short, only about 1 to 2 mm long, with blunt tips, and also are plainly corticated.   The central mass of cells breaks down into an unorganized mass that looks much like that of an oil gland in the immature peel of a citrus fruit.   The pulp-vesicles of this species seem to secrete a liquid that fills the locules of the fruit and hardens on drying to a dark reddish-brown, brittle, resinous mass.   In the two Bornean species, P. longisepalum and P. latialatum, the pulp-vesicles are longer than in P. dubium and much longer than in the similar-looking P. sumatranum, and are of a slender conical shape often ending in a sharp point at the apex.   Their cortical tissue is much thinner than in P. dubium and less clearly cellular.   The constituents do not break down so completely.   The pulp-vesicles of these two species are more like those of Citropsis and Atalantia but have smaller bases not so deeply sunken into the dorsal wall of the locule.   The series of pulp-vesicles in the different species of Pleiospermium connect with the still more highly developed pulp-vesicles of Limnocitrus, which are no longer conical but are definitely fusiform and do have a very short stalk.
      Perhaps no other genus of the orange subfamily is of greater interest than Pleiospermium in throwing light on the origin and development of the unique and highly specialized pulp-vesicle organs which are found only in the subtribe Citrinae, and which reach their highest specialization in the genus Citrus, giving the delicious flavor characteristic of its cultivated varieties.   It would be desirable to study the development of the pulp-vesicles in Pleiospermium and the related genera Limnocitrus and Burkillanthus from living material.
      The discovery of three species of Pleiospermium having unifoliolate leaves greatly resembling those of Citrus in shape, size, color, and texture brings them and their trifoliolate congeners into prominent notice as possible persisting representatives of ancient ancestral forms in the descent of Citrus.   Like the living species of Citropsis in tropical Africa, the species of Pleiospermium show all stages in the origin of unifoliolate leaves from trifoliolate leaves (or pinnate leaves in Citropsis), something that without doubt occurred, ages ago, in the development of Citrus.   A key to the species of Pleiospermium is presented below.
      1.   Pleiospermium alatum (Wt. & Arn.) Swing. Jour. Wash. Acad. Sci. 6:428. 1916. Limonia alata Wt. & Arn. Prodr. 1:92. 1834. Illus. Wight, Ill. Ind. Bot. 1:pl. 41. 1840; Beddome, Fl. Sylv. pl. 7, fig. 3(1-9). 1871.
      Type.—Southwestern British India, "foot of the Neelgherries," Madras State (Wight, No. 324).
      Distribution.—Western peninsular India: Madras State; Ceylon.
      Common name.—Ceylon orangeaster.
      Small trees with glabrous branchlets, sometimes with a single stout, straight spine up to 25 mm long at the side of the bud in the axil of the leaf, sometimes spineless, especially the fruiting branches; leaves 3-foliate [sic] (rarely 2- or 1-foliolate); lateral leaflets much smaller, all leaflets rather thick, glossy above, finely netted-veined beneath, lateral veins slender, not very conspicuous, making a very wide angle with the midrib, margins entire, the apex obtuse or subacute, rarely acuminate, bluntly rounded at the very tip, often emarginate, the leaflet blade abruptly narrowed to a cuneate base; petiole variable, 25-38 mm long, narrowly winged, articulated with the blade; inflorescences in clusters in the axils of the leaves or in terminal much-branched hoary panicles; flowers small, about 12-15 mm diam., 4-5-merous, pedicels rather short, finely pubescent; flower buds cylindric, rounded at the tip, more or less pubescent; calyx small, 4-5-lobed, sepals deltoid, finely pubescent; petals oblong or ovate, obtuse, entire, sparingly covered with fine pubescence without; stamens free, 8 or 10 (twice as many as the petals), filaments free, anthers large, erect, linear-oblong; pistil subsessile, seated on a low cupulate disk which clasps the base of the ovary for 0.5-0.8 mm; ovary obovoid, glabrous, 3.5-4.2 X 1.6-1.8 mm, 4-5 locular, with 2 ovules in each locule; style slender, gradually merging with the tip of the ovary and capped by a slightly thicker, subglobose stigma; fruits globose, 2.3-2.5 cm diam., like small oranges in appearance, the segments each containing 1 or 2 seeds surrounded with an aromatic, mucilaginous fluid, peel rough, dotted with numerous oil glands, seeds ovate, somewhat flattened, with a hard, smooth testa.
      This species occurs rather commonly in the hot dry parts of southern India and Ceylon.   The fruit is said to be very bitter; the wood is hard and heavy, yellowish, even-grained.   The leaves have a strong lemon odor when crushed.
      2.   Pleiospermium dubium (Bl.) Swing. Jour. Wash. Acad. Sci. 6:429. 1916. Limonia ? dubia Blume, Bijdr. Fl. Nederl. Indie 1:133. 1825; Paramignya blumei Hassk. Tijdschr. Nat. Gesch. 10:137. 1843.
      Type.—Java, Kuripan (Blume), Rijks Herb., Leiden (sheet No. 908203-1318).
      Distribution.—Java.
      Common name.—Java orangeaster.
      Twigs angular when young, with straight, sharp spines, 2-2.5 mm long, 1 or 2 in the axils of each leaf, older twigs often spineless; leaves 1-, 2-, or 3-foliolate; leaflets thin, elliptical, cuneate at the base, acute or acuminate at the apex, the very apex often blunt, or emarginate, the middle leaflet (or single leaflet) 7-9.5 X 2-3.6 cm, with 8-12 faint veins visible on both surfaces on each side; petioles 7-22 mm long, narrowly winged above but nearly wingless at base, plainly articulated at both ends; flowers in short, 5-15-flowered, axillary racemes; ovary obovoid, hirsute, 2-2.8 X 1.2-1.7 mm, borne on a cylindrical gynophore somewhat like that of Paramignya or Merope; fruits with rudimentary pulp-vesicles arising from the ovary wall, elongate-ovoid or irregularly conical, 2.5-3.5 mm long, 0.4-0.6 mm broad at the base, completely covered with a cellular cortex which connects at the base with the dorsal wall of the locule but is not usually imbedded in it and never deeply imbedded; mature fruits subglobose, 2-2.5 cm diam.
      This species, known only from Java, is of great interest from the fact that, because of its leaves, which have from one to three leaflets, it serves to connect the usually trifoliolate Indian species, P. alatum, with the unifoliolate species, P. samatranum, P. longisepalum, and P. latialatum, which all have leaves remarkably like those of Citrus.   This is true also of the unifoliolate leaves of P. dubium.
      The pulp-vesicles of this species are each sharply divided into a cortical layer, composed of rather small cells, and a central portion made up of larger cells which finally disintegrate into a spongy mass, oily or resinous (?) in nature, which looks much like the contents of the oil glands in the peel of the young fruit (see above).
      Blume's type material was collected on "limestone hills near Kuripan, Buitenzorg Prov." in Java, but neither flowers nor fruits were mentioned by him.   The type specimen in the Rijks Herbarium at Leiden, No. 908203-1318, and a cotype specimen in the Herbarium of the Museum of Natural History at Paris both show trifoliolate leaves; the Leiden branch has paired axillary spines as described by Blume, the Paris specimen is spineless.
      Plants of P. dubium cultivated in the Botanic Garden at Buitenzorg (now Bogor), Java, have flowered and fruited for many years.   Dried herbarium specimens from these trees were distributed by B. P. G. Hochreutiner, as No. 111 in his Plantae bogorienses exsiccatae, about 1904 (this number is found in the Kew Herbarium and also in the Paris herbarium).   On his printed label Hochreutiner stated: "This plant is without any doubt the original of Hasskarl['s Paramignya blumei]" (Swingle, 1916b, p. 429).   Herbarium specimens from these Buitenzorg plants are preserved in the herbarium of the Bureau of Science at Manila and in the National Herbarium (sheet No. 448333) at Washington.   From them were taken the flower and fruit characters given in the description above.
      Pleiospermium dubium is a small tree or shrub, attaining a height of about 4 meters.   The inflorescences were said by Koorders (1912, p. 426) to be 5 to 15-flowered, short, axillary racemes.   This plant has probably the most polymorphic foliage of any species of the orange subfamily.   It often has branches bearing only trifoliolate leaves and others with exclusively unifoliolate leaves.   However, sometimes trifoliolate and unifoliolate leaves are disposed hit-and-miss on a branch.   Some trifoliolate leaves develop only one of the side leaflets.
      This species and Citropsis gabunensis of tropical West Africa, which has almost equally polymorphic leaves, offer excellent material for a study of the critical evolutionary stages passed through ages ago by the remote ancestors of Citrus.
      3.   Pleiospermium sumatranum Swing. Jour. Arbor. 20:260. 1939. Illus. Swingle, loc. cit. pl. 3, figs. 3, 4; fig. 3-19 this work.
      Type.—Sumatra, Paanus (?) (Korthals, No. 960).   Rijks Herb., Leiden.
      Distribution.—Known only from Korthals' collections, probably from a single locality.
      Common name.—Sumatra orangeaster.
      Shrub or small tree; twigs cylindrical, slender (1-2 mm diam.), brownish-gray (in dried herb. spec.), fruiting twigs spineless, internodes 2-3 cm long; leaf blades broadly lanceolate, 8-17 cm long, 5-8 cm wide, acuminate at apex, acumen about 9-13 mm long, often curved slightly to right or left, 3-4.5 mm wide at base, but at the very tip 1.5-2 mm wide, blunt or slightly emarginate, base cuneate (angle 75°-85°), blade 10-16 X 5-6.5 cm, petiole 18-20 mm long, winged above, 3.5-6.5 mm wide, but wingless for 4-5 mm up from attachment at base, lateral veins 10-12 but of unequal prominence, some strongly elevated above lower surface, others faint and not prominent, making angles of 60°-75° with the midrib, margins entire, very faintly undulate, oil glands very numerous, evenly spaced all over the blade, 0.2-0.5 mm apart, plainly visible below, almost invisible above, petioles 10-20 mm long, 3-7 mm wide, winged at top, wingless at base; inflorescences axillary, simple or rarely bifurcate, apparently sometimes terminal on short lateral twigs; pedicels of nearly ripe fruits 1-1.5 cm long, 2-3 mm wide; flowers unknown; calyx with 5 acute lobes, 1-2 mm wide and 2-2.2 mm long, minutely pubescent below, margins sparingly short-ciliate; fruits subglobose, 18-20 mm diam., with 3 locules lined with a dense layer of short pulp-vesicles, 1-2 mm long, 0.4-0.7 mm wide, peel with numerous oil glands 1.5-1.8 mm thick.
      This species of Pleiospermium has remarkably small pulp-vesicles lining the dorsal walls of the three locules of the fruit.   The pulp-vesicles are very numerous, ovoid, bluntly rounded at the tip, 1 to 2 mm long, 0.4 to 0.7 mm wide, slightly narrowed at the base but not imbedded in the locule wall, cortex very thin, composed of indistinct cells, contents sharply delimited from the cortical layer and soon breaking down into a spongy mass of oily or resinous (?) matter looking almost exactly like the contents of the oil glands in the peel of the fruit (see fig. 3-19).   Apparently these very small pulp-vesicles secrete a translucent liquid that completely fills the locules of the fruit.   In dried herbarium specimens this liquid hardens to a dark reddish-brown, resin-like mass that splinters when struck a sharp blow.
      This Sumatran species of Pleiospermium and its two close relatives from Banguey Island and Borneo, P. longisepalum and P. latialatum, all have leaves very similar to those of Citrus except for the more or less acuminate leaf tips.
      A fungus, found on the fruits of the type specimen, was identified by Bitancourt and Jenkins (1938, p. 388) as a species of Elsinoë much like the scab fungi, E. fawcettii and E. australis, that attack Citrus.
      4.   Pleiospermium longisepalum Swing. Jour. Arnold Arbor. 20:259. 1939. Illus. Swingle, loc. cit. pl. 3, fig. 5.
      Type.—Banguey Island (off northeastern point of Borneo) (Castro and Melegrito, No. 1348).   Herb. Natl. Arbor., Washington, D.C. (sheet No. 45800).
      Distribution.—Known only from the type collection and from Castro and Melegrito, No. 1713, from the same locality.
      Common name.—Banguey Island orangeaster.
      A small tree or shrub, fruiting twigs angular, 2-3 mm diam., straight, glabrous, grayish-green, spineless, vegetative twigs at first angular then cylindric, 2-3 mm diam., both vegetative and fruiting twigs with faint striae decurrent from the nodes; leaves 1-foliolate, leaf blades 7.5-11.5 X 3-4.5 cm, elliptical with acuminate apices turned slightly to one side, but the actual tip (about 2 mm wide) abruptly rounded or even slightly retuse, margins entire, lateral veins weak, 8-10 irregular pairs, with irregular reticulations between; petioles 10-14 mm long, clearly articulated with the leaf blades and with the twigs, winged, 3-5 mm wide at top, tapering into a wingless base constituting about 1/4-1/3 the total length, wingless basal portion often twisted and apparently pulvinoid; flowers borne singly in the axils of the leaves but the tip of flowering branch merging into a racemose inflorescence with the leaves reduced or wanting; pedicels 20-25 X 1.5-2 mm with several minute bracts borne singly, faintly striate longitudinally and grayish-green like the twigs; pedicels of sterile flowers persistent, very slender, 15-20 X 0.75-1 mm; sepals persistent, gray-green, strap-shaped, 9-12 mm long, 2 mm wide at the base, 0.5-0.8 mm wide at the apex, which is bluntly rounded or slightly emarginate, straight and strongly reflexed below the ripening fruit, showing one median nerve on the inside, margins entire; fruits ovoid, tipped by the umbonate, persistent, very short style-base, glabrous, gray-green, 20 X 15 mm, with 3 locules, having 1 or 2 ovules in each locule; pulp-vesicles very rudimentary, 4-6 mm long (arising from the dorsal ovary walls and tapering gradually to a blunt point); peel thin (1.5-2 mm), gray-green, with numerous, slightly sunken, minute oil glands; seeds 1 or 2, about 11 X 6 X 4 mm, monoembryonic.
      This very curious plant was briefly noted by Merrill (1926, p. 377), who published Swingle’s observations made on a specimen of it and his own notes made shortly after its discovery.   At first sight it seemed to be allied to Atalantia because of its unifoliolate leaves.   However, it does not have the well-developed, sessile pulp-vesicles characteristic of Atalantia but instead has rather rudimentary, corticate ones somewhat like those found in P. dubium.   Then, too, its winged petioles are unlike any found in Atalantia, Severinia, or related genera, and are almost exactly like those of other species of Pleiospermium.   The long, persistent, reflexed ligulate sepals are unique in the orange subfamily, although the evidently related species, P. latialatum, has similar sepals, persistent, cuneate, and partially reflexed, but shorter.   The other three species of Pleiospermium have short sepals much like those of many other members of the subtribe Citrinae.
      5.   Pleiospermium latialatum Swing. Jour. Arnold Arbor. 20:261. 1939. Illus. Swingle, loc. cit. pl. 3, fig. 6.
      Type.—North Borneo (Sabah), Tawao (Elmer, No. 21542).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—Known only from the original collection.
      Common name.—North Borneo orangeaster.
      A small tree or shrub, ultimate twigs slender, 1-2 mm diam., at first slightly angled, soon terete, very finely puberulent, more or less longitudinally wrinkled, nodes 1-1.8 cm long; spines lacking (at least on fruiting branch); leaves 1-foliolate, glabrous, shiny above, dull below, oblong-elliptical or lanceolate, acuminate-caudate at apex, end of acumen linear for 4-6 mm and 1.5-2 mm wide, ending abruptly with an emarginate tip, broadly rounded or very bluntly cuneate at the base; leaf blade 9-15 X 3.7-7 cm, margins entire, lateral veins 10-13 pairs, arising at an angle of 70°-75° with the midrib, curving slightly toward leaf apex, oil glands small, very numerous; petioles (all detached) short, broadly winged, obcordate, 1.6-2.4 X 1-2 cm. narrowed into a pulvinoid and minutely wrinkled terete base, 1-2 mm long, 1.25 mm wide with a narrow channel above formed by the decurrent wings; flowers, not seen, arising singly (?) at the tips of short axillary branches which bear leaves or bracts below; calyx puberulent without, sometimes persisting under young fruits, lobes narrowly cuneate or somewhat rounded, apparently irregular in size, acute or rounded, 2-4 mm long, 2-5 mm wide, diverging merely at right angles to the pedicel under the young fruit; young fruits subglobose, 10-14 mm diam., glabrous, peel green, with numerous oil glands, slightly raised above the surface with a small central depressed (crateriform); locules 3; pulp-vesicles conical, very slender, 3-4.5 mm long, acutely pointed, attached to the dorsal wall of the locule into which their bases are somewhat sunken, cortication very thin with indistinct cells, contents not degenerating to an oily or resinous mass as in P. dubium but showing axile droplets of oily substance; seeds 2 or 3, large and plump.
      This remarkable species is known from a single collection from Tawau, in North Borneo (about 175 km southwest of Shibutu Island in Sulu Archipelago, Philippine Islands).   It is obviously related to P. longisepalum but differs strikingly in its very broadly winged petioles, which are also relatively short, giving them a very pronounced obcordate outline.   The sepals are sometimes (perhaps always) persistent under the developing fruit, but are only about half as long as those of P. longisepalum and are not completely reflexed, but stand out approximately at right angles to the general direction of the pedicel.   Only young fruits were available for study but they show clearly that the pulp-vesicles of this species, much like those of P. longisepalum, are very numerous, slender-conical in shape, and slightly immersed in the dorsal wall of the locule.   The cortical layer of the pulp-vesicle is present but not strongly developed and apparently sometimes this cortex is so weak that the upper portion of the pulp-vesicle splits open as the fruit develops.   This species has leaves that strikingly resemble those of Citrus in general appearance.

XVI.   Burkillanthus Swing.
      XVI.   Burkillanthus Swing. Jour. Arnold Arbor. 20:255. 1939.
      Type species.Citrus malaccensis Ridl. = Burkillanthus malaccensis (Ridl.) Swing.; from Malacca, Malay Peninsula.
      Distribution.—Malay Peninsula: Malacca; northern Sumatra: Asahan.
      Common name.—Burkillanthus or Malay ghost-lime.
      Leaves 3-, 2,- or 1-foliolate; leaflets thin, with numerous strongly marked, subparallel, lateral veins; petioles variable in length, articulated, narrowly winged or sometimes merely margined, with pulvinoid bases; petiolules very short, pulvinoid; inflorescences, few-flowered clusters in the leaf axils; flowers large, 5-merous with 10 stamens; pedicels short, about the length of the styles; calyx deeply 5-lobed; petals 5, spathulate; ovary obclavate, minutely and sparsely pubescent, merging abruptly into the much narrower style, which is nearly as long as the ovary, with 5 locules, each with 22-26 ovules arranged in 2 rows; fruits very large, ovoid, more than twice as long as the petioles, with a thin peel dotted with numerous oil glands covering a thin, hard, woody endocarp; pulp-vesicles sessile (not deeply counter-sunk) on the dorsal locule walls, very long, swollen near the base, then cylindrical, then tapering to an acute apex, corticate, with translucent inner core not breaking down into an oily or resinous (?) mass; seeds very large, with a thin, slightly wrinkled testa, immersed in mucilaginous gum and mingled with the collapsed rudiments of the pulp-vesicles; embryo single with light buff (not green) cotyledons.
      This strange plant seems to be an evolutionary freak of nature perhaps descended from some ancestral form more or less like a trifoliolate Pleiospermium such as P. alatum or a uni-, di-, or trifoliolate species like P. dubium.
      The pulp-vesicles of Burkillanthus are stalkless and somewhat resemble those of Pleiospermium; however, they are very much larger and less obviously corticate, with a central core which does not disintegrate into an oily or resinous (?) mass resembling that of an oil gland.   These pulp-vesicles of Burkillanthus seem to represent a very interesting developmental stage about midway between the pulp-vesicles of Pleiospermium and those of Clymenia, one of the True Citrus Fruit Trees.
      It is remotely possible that the curious, cavernous lining of the radial locule walls in Swinglea may, upon study of fresh material, prove to be composed of still more primitive structures analogous to the pulp-vesicles of Burkillanthus.   the [sic] pistil of B. malaccensis (see fig. 3-20) is extraordinarily like that of Swinglea glutinosa in size, shape, number, and arrangement of the ovules, et cetera, except that in Swinglea the ovary is hypermerous, having 8 to 10 locules (not five as in Burkillanthus) and oil glands occur in all parts of the pistil, including the stigma (see fig. 3-20).   However, the fruits of Swinglea have a very different gross structure from those of Burkillanthus and the seeds of Swinglea are not smooth but are covered with flattened laciniate paleae.
      Cockrell (1935) found that the wood of B. malaccensis, as it grows near Asahan, Sumatra, is almost exactly like that of Micromelum pubescens (one of the Very Remote Citroid Fruit Trees), differing from it only slightly in texture and not showing any "'reasonably constant basis for separation."   However, the wood of Burkillanthus differs decidedly from the wood of all the other genera of rutaceous plants, 16 in all, that he studied, including eight genera of the orange subfamily, viz.: Glycosmis, Clausena, Murraya, Merrillia, Feronia, Aegle, Atalantia, and Citrus.   He did not, however, have wood of Pleiospermium or of Swinglea to compare with that of Burkillanthus.
      Burkillanthus malaccensis (Ridl.) Swing. Jour. Arnold Arbor. 20:257. 1939. Citrus malaccensis Ridl. Fl. Malay Penin. 1:359. 1922. Illus. Swingle, loc. cit. pls. 2 (1/2 nat. size, not nat. size) and 3 (figs. 1, 2); figs. 3-20 and 3-21 this work.
      Type.—Malay Peninsula, Malacca Territory, Nyalas (Goodenough, No. 1273).   Cotype: Malacca State, Bukit Sedanan (Derry, No. 1106).   Both in Herb. Bot. Gard., Singapore.
      Distribution.—Malay Peninsula: Malacca; Sumatra: Asahan.
      Common name.—Burkillanthus or Malay ghost-lime.
      A tree up to 14 m high, some branches with stout, straight, usually paired spines, others spineless, with leaves mostly 1- or 2-foliolate, rarely 3-foliolate; leaflets subcoriaceous but thin, broadly lanceolate or elliptic-lanceolate, acute or more or less acuminate at the tips, which are blunt at the very apex, terminal leaflets and 1-foliolate leaves variable in size, often large, 12.5-27 X 4.5-11.5 cm, lateral leaflets much smaller, 6.5-8 X 3-4 cm, with short, pulvinoid petiolules 2-2.5 mm long, lateral nerves (of larger leaflets) about 12-18 on each side arising at a very wide angle (70°-85°) with the midrib, running subparallel and only slightly curved for 2/3 or 3/4 the distance from the midrib to the margin, then branching dichotomously, base broadly cuneate, articulated with the petiole by means of a pulvinoid petiolule, margin subentire, very slightly coarse-crenate; petioles of the 2- or 3-foliolate leaves 4-5 cm long, often narrowly winged (3-6 mm in total width) at apex, narrowing toward the base, often merely margined even at the apex; petioles of 1-foliolate leaves usually shorter, 1-3, rarely 4, cm long, sometimes almost wingless when short; inflorescences few- (usually 1-4-) flowered, axillary clusters, flowers large, 5-6 cm diam., borne on pedicels 5-7 mm long; calyx of 5 lanceolate-acuminate, acute sepals with large oil glands, free nearly to the base; petals 5, subspathulate, with oil glands in the broadened upper part, 20-25 mm long, 3.5-4.5 mm wide, tapering into a long slender base 1-1.5 mm wide; stamens 10, filaments free, long, slender, anthers rather small, oblong; ovary obclavate, 6.5-7 mm long, 2.5 mm diam. at the broadest part (about 2/3 the distance from the top of the disk to the base of the style, at a point shortly below the tips of the locules) and 1.8-2 mm diam. at the base, with 5 locales, each with 22-26 ovules in 2 rows in the upper 2/3 of the locular cavity (see fig. 3-20), the entire ovary covered with short sparse pubescence and with a few small oil glands dotted over the upper half only; style slender, 5.5-6 X 0.5-0.7 mm, with small oil glands, glabrous, merging abruptly into the tip of the ovary without any definite demarcation, swelling gradually at the tip and surmounted by a depressed globose stigma, 1 mm high, 1.6 mm wide, without oil glands; disk short, cylindrical, 0.6-0.7 mm high, 1.4-1.5 mm wide; fruits large, ovoid, 10-11 X 8-9 cm, outer peel thin (2-3 mm), roughened with numerous slightly depressed oil glands, fitting closely on woody endocarp, 2-3 mm thick, very hard in dried fruits, that shows on its outer surface numerous slightly depressed areas 1.5-3 mm diam.; pulp-vesicles sessile, 15-33 mm long, with bases 2.5-4 mm diam., tapering into long cylinders, 1.5-2 mm diam., which often end in very acute tips; corticate with small tangentially compressed peripheral cells and larger, thinner-walled central cells somewhat elongated in their radial diameter; locules of the ripe fruit almost filled with large seeds, apparently imbedded in mucilaginous gum, perhaps secreted by the pulp-vesicles; seeds broadly obovate, tapering to an acute point at the base where the seed is attached, 22-27 X 16-18 X 7-10 mm, smooth, cream-colored, with a cinnamon brown chalazal cap covering the broad end of the seed for 6-8 mm, monoembryonic with pale buff (not green) cotyledons, testa thin, papery, irregularly wrinkled on dried seeds.
      Burkill published (1931, p. 218) very interesting notes on this species, as follows: "The characters of this extremely interesting plant have not been brought into adequate prominence.   It was collected in the year 1893 twice in the interior of the Territory of Malacca, and is represented in the herbarium of the Botanic Gardens, Singapore, by a sheet of each gathering.   Derry obtained it with flowers in January, and Goodenough with fruit in July.   Derry's locality is given as Bukit Sedanan; Goodenough's as Nyalas; these places are 6-8 miles apart.   Upon the two sheets there are ten leaves; and upon each sheet one is compounded by having a single lateral leaflet.   Other leaves have a scar at the apex of the narrowly winged petiole and therefore count as compound; others have no scar, and obviously are simple.   Trifoliate leaves do not exist; but it would be curious if they are never produced.   There is a conspicuous joint between the petiole and the leaf-blade or blades.   The flower suggests that of Merrillia caloxylon Swingle.   The sepals are in shape as those of Atalantia and have large glands.   The petals are nearly spathulate, and with glands in the broadened upper part.   The stamens apparently are 10, free, with long filaments and rather short anthers.   The ovary is hairy and fluted with five grooves; it is hard to distinguish the exact place where the gynophore passes into it.   The style is as long as the ovary and ends in a large stigma.   There are numerous ovules.   For other points the original description may be consulted."
      This remarkable plant was first collected in flower by R. Derry (at that time in charge of the Malacca forests) at Bukit Sedanan, near Malacca, in January, 1893.   On July 9 of the same year, J. Goodenough collected at Nyalas, only some eight miles distant from Bukit Sedanan, a specimen with ripe or nearly ripe fruits (see fig. 3-21) (fide Burkill, in letter dated January 13, 1937).   Thanks to the kindness of R. E. Holttum, then director of the Botanic Gardens at Singapore, Swingle was privileged to have the loan of the two type specimens noted above.
      E. D. Merrill directed attention in June, 1938, to a tree which he had assigned to the subfamily Aurantioideae, collected by B. A. Krukoff (No. 4233) in November-December, 1932, in the Nassihi Forest Reserve near Asahan in northern Sumatra (about 335 km west-northwest of Malacca).   Herbarium specimens kindly lent by the New York Botanical Garden and the Arnold Arboretum showed that this plant was almost identical with the type specimens of Citrus malaccensis Ridl.   The twigs of Krukoff's specimens were in part spineless but in part armed with stiff, stout, straight, abruptly pointed, usually paired thorns 2 to 3 cm long, and 2 to 3 mm in diameter at the base.   The leaves were uni-, di-, or trifoliolate.   A giant unifoliolate leaf in the New York Botanical Garden herbarium had a petiole 2 cm long and a broadly lanceolate blade, 27 by 11.5 cm, short-acuminate at the apex, cuneate at the base.   It had 17 or 18 principal lateral veins, and about 9 or 10 shorter laterals, arising at angles of 70° to 80° with the midrib (see Swingle, 1939a, p. 262, pl. 2).   The ovoid fruit measured about 11 cm long and 9 cm wide.   The inner side of the dorsal walls of the locules were thickly covered with cylindrical or subclavate (?) saclike pulp-vesicles, as described above.   No similar pulp-vesicles are known in any other genus of citrus fruit yet studied.
      In a letter to Swingle, dated April 19, 1940, R. E. Holttum reported that a collector whom he had sent to Bukit Sedanan, the type locality of B. malaccensis, found this species growing there in quantity "in primitive forest, but only on the ridges, where the shade is less than on the valley slopes."   The trees were about 30 feet (9.15 m) high and bore half-grown fruits.
      In June, 1941, a large fruit, slightly over 10 by 16 cm, was sent to Swingle by air express from the Singapore Botanic Gardens.   It reached Washington in good condition, which made it possible to study fully developed pulp-vesicles in the living state.   This fruit contained more than 120 very large, mature seeds.   (See fig. 3-21.)
      This remarkable tree should be studied in the living state and possibly tested as a rootstock for Citrus.   It is called by the Malays limnan hantu ("ghost lime"), a name also applied to wild forms of Citrus according to Burkill (1931, p. 221), perhaps because its large fruits are not unlike those of some species of Citrus in size and shape.

XVII.   Limnocitrus Swing.
      XVII.   Limnocitrus Swing. Jour. Arnold Arbor. 21:2. 1940.
      Type species.—Paramignya ? littoralis Miq. = Limnocitrus littoralis (Miq.) Swing.
      Distribution.—Java: Bali; South Vietnam.
      Common name.—Limnocitrus or swamp-orange.
      A shrub with glabrous twigs having stout, axillary, single spines at the nodes; leaves simple, coriaceous, glabrous, broad-elliptical or obovate, bluntly pointed at the tip and broadly cuneate at the base; petioles short (less than 1/10 as long as the leaf blade), pulvinoid, not articulated with the leaf blade; inflorescences terminal, short, corymbose; flowers 4- or 5-merous, borne on very short, glabrous pedicels; flower buds oblong, greenish; calyx 4-5-lobed, lobes triangular, obtuse; petals linear-lanceolate, obtuse, white; stamens 10, filaments free, glabrous, anthers linear; disk annular; ovary oblong-ovate, with 15-20 narrow hirsute ridges, with a single large oil gland over each of the 4 or 5 locules, each having 2 ovules; style slender, sparingly hirsute, stigma slightly capitate, distended with large oil glands between the stylar canals; fruit globose or subglobose, rather large (3.5-4 cm diam.), with a gland-dotted peel, becoming orange-colored when ripe, filled with very numerous, slender, fusiform pulp-vesicles, slender and acutely pointed at tip and abruptly truncate at the broad base about 1/3-2/3 as wide as the broadest part, which is near the base, with a slender axile [sic] column of whitish granules (oil or resin drops?) in the center; seeds large, flattened-oval, with a hard, creamy-white testa, monoembryonic, embroyo green, cotyledons dotted with oil glands.
      This remarkable plant has been placed in several different genera by the few botanists who have discussed it.   It has certain resemblances to the species of Pleiospermium but differs from them in several important characters: (1) in having simple leaves with very short pulvinoid petioles; (2) in having slender, fusiform pulp-vesicles with very slender, acute tips and slightly narrowed at the base where attached; (3) in having thick, sparingly veiny leaves; (4) in having straight, very stout, single spines in the leaf axils.
      Limnocitrus is distinguished from Atalantia by its very slender-tipped pulp-vesicles, which are not conical and do not have broad bases deeply immersed in the locule wall, as in Atalantia, but are fusiform and narrowed into a definite, though very short, basal stipe.   The leaves are thicker and differently veined and the whole plant has a different facies from that of Atalantia.
      In many ways the pulp-vesicles of Limnocitrus are intermediate between those of the True Citrus Fruit Trees (Citrus and the five most closely related genera) and those of the Near-Citrus Fruit Trees (Citropsis and Atalantia) and some of the Primitive Citrus Fruit Trees, especially Pleiospermium.   In most of its other characters, Limnocitrus stands very much alone.   The leaves are different in texture, venation, and general appearance from those of all the other twelve genera in the subtribe Citrinae.   However, similar, although not identical, leaf and twig characters are found in Merope and Pamburus, two monotypic genera of the Minor Citroid Fruit Trees, subtribe Triphasiinae, which also grow in brackish marshes.
      Limnocitrus littoralis (Miq.) Swing. Jour. Arnold Arbor. 21:3. 1940. Paramignya ? littoralis Miq. Ann. Mus. Bot. Lugd.-Bat., 1:211. 1863, and Teijsmann & Binnendijk, Natuur. Tijdschr. Nederl.-Indie 27:41. 1864; Limonia littoralis (Miq.) Backer, 1911; Atalantia littoralis (Miq.) Guill. 1913; Pleiospermium littoralis (Miq.) Tan. Bul. Mus. Hist. Nat. Paris, 2 ser. 2:162. 1930. Illus. Valeton, Icon. Bogor. 4:163, pl. 349. 1912; Swingle, loc. cit. pl. 1; fig. 3-22 this work.
      Type.—Java, Rembang, on seashore (Teijsmann).   Rijks. Herb., Leiden.
      Distribution.—Java: Bali; South Vietnam: Annam, Cochin China.
      Common name.—Limnocitrus or swamp-orange.
      This species was based on a specimen collected by Teijsmann on the seashore at Rembang on the north coast of Java.   Miquel's original description reads, in translation, as follows: "Spines solitary, straight, arising nearly at right angles to the twig; leaves simple, with very short petioles, acute or acutish at the base, broad- or obovate-elliptical, apex obtuse or somewhat protracted-obtuse and emarginate, coriaceous, pellucid-punctate, reticulate, venose below, [margins] very slightly crenulate; leaves 2-3 in. [5-7.6 cm] long; flowers in terminal tufts, forming much contracted cymes, with very short, thick pedicels; calyx unequally 5-toothed; petals 5, elliptic-oblong, imbricate; stamens 8, filaments very short, anthers linear, marked with an oblong gland on the back, joined together in the bud; ovary inserted on a short disk, 5-angled, sparsely pilose, 5 locular, ovules 2 in each locule, subcollateral; style thick, stigma slightly capitate, obscurely 5-lobed; fruits globose, 2- or 3-seeded, about the size of apricots, with distinct, empty locules."
      Another description published almost simultaneously by Teijsmann and Binnendijk (doubtless based in part on independent observations of the species by Teijsmann, who discovered it at Rembang) supplements Miquel's original description.   Omitting obvious duplications, this reads, in translation, as follows: "Branches cinereous, twigs very glabrous; spines 1/2 in. [12.5 mm] long, stout; leaves emarginate at the apex, rarely obtuse; racemes [sic] short, sessile…calyx 4-5-lobed; filaments 10, free; ovary pubescent, 4-5-locular, ovules paired; fruits baccate, very fetid, 2.3 cm diam., locules by abortion 1-seeded."
      In the notes following the technical description they add, regarding the fruit, that it is "depressed-globose, about the size of a lime, becoming orange-colored…4-5-locular…pulpy; rind, leathery-glandular.   Seeds large, testa membranaceous, white.   Cotyledons fleshy, plano-convex, green."
      One of the most important characters of the species, overlooked by Miquel but brought to light by Teijsmann and Binnendijk, is that the locules of the fruit contain pulp.   This was mentioned a few years later by Kurz, who lived for some years in Java, where he doubtless saw Teijsmann's specimens collected at Rembang.   Kurz wrote (1875, p. 136, footnote): "The berries of P[aramignya] littoralis Miq., a species closely allied to P. angulata [= Merope angulata], have pulp, but the dried ones appear pulpless."
      Fortunately Swingle was able to supplement the above-quoted descriptions and a later one by Valeton (l.c., p. 163), all based on a single collection made at Rembang, Java, from excellent material from Indo-China.   Flowering specimens studied were from plants growing in a hedgerow near houses at Tourane, Annam (coll. by J. and M. S. Clemens, No. 3263), in Herb. Arnold Arboretum; fruiting material was from Nhatrang, Annam (Aug. Chevalier, No. 30526, February 6, 1916), in Herb. National Arboretum.   This material showed the following characters:
      A shrub or small tree 2 m high with stout, straight, single spines; leaves simple, glabrous, broadly oval, bluntly pointed at both ends, margins entire or faintly crenulate; leaf blades 5-7.5 X 3-4 cm, with 8-11 pairs of lateral veins, arising at angles of 40°-60° with the midrib, usually straight but sometimes bending to the right or left to unite with the neighboring lateral vein, tertiary or quaternary veinlets forming coarse, irregular, inconspicuous reticulations, oil glands small, very numerous; petioles cylindrical, 3.5-7 X 1-1.5 mm, pulvinoid, minutely puberulous on the upper side inside the shallow channel bounded by the decurrent leaf margins; inflorescences terminal (sometimes also small ones in axils of last few leaves), corymbose, 3-4 X 2.5-5 cm, pubescent, peduncles abundantly and divaricately branched, divided into nodes (5-10 X 1.5-2 mm) by thick, blunt-pointed, pubescent, caducous bracts, 1-5 X 0.8 mm, pedicels 2.5-3.5 mm long, expanding gradually into the base of the calyx without any articulation; calyx glabrous without, 4-5-lobed, lobes bluntly pointed, finely pubescent within, with short-ciliate margins; petals white within, green without in the bud and greenish-white when expanded, linear-oblong, blunt-tipped, 10-11 X 3 mm, glabrous except for scattered puberulence near the tip; stamens with glabrous filaments, anthers long and slender, 4.5 X 1 mm, with a large dorsal oil gland; disk cup-shaped, 0.5-0.6 mm high, 1.5-1.6 mm wide, enclosing loosely the base of the ovary for 0.2-0.25 mm; pistil 10-11 mm long; ovary obovoid, 4.5-5 mm high, 1.7-2 mm wide, flattened at top, obscurely ribbed in middle portion with 15-20 slight ridges from which arise scattered long hairs, pentagonal at the top because of large sunken oil glands (1 over each locule), locule [sic] 4-5, each with 2 pendant collateral ovules; style 6 mm long, 1.8-2 mm thick, terete, sparingly pilose above, increasingly so toward the base, stigma depressed capitate, 1.8-2 mm wide, 0.7-0.8 mm high, with 4-5 radially arranged, linear stylar canals, with 2 (rarely 1) large oil glands between each adjoining pair of stylar canals; fruits subglobose, 37-40 mm diam., peel thin, dotted with oil glands, containing numerous, very slender, fusiform pulp-vesicles, 9-11 mm long, gradually tapering to 0.5-0.8 mm diam. at the base where attached to dorsal wall of locule, the broadest portion (1.6-1.8 mm) of the pulp-vesicle is situated about 1/10-1/6 of the distance from the base to the tip, the slender apical portions often matted and tangled in dried specimens; seeds long obovate, 17-18 X 9-11 X 3.5 mm, with a smooth, firm, yellowish-gray testa, monoembryonic; cotyledons green, gland-dotted.
      This little-known species apparently has a fairly wide distribution, as it has been found in Annam and Cochin China in South Vietnam, as well as in Java and Bali.   However, it has been collected only a few times since it was discovered by Teijsmann some seventy-five years ago at Rembang in Java, where apparently it has not been seen since.
      It has, however, been collected several times near Nhatrang in Annam (see fig. 3-22), and once at Tourane.   It grows in tidal swamps subject to overflow with brackish water at spring tides.   Guillaumin (1913, p. 442), in describing the plant as found in the vicinity of Nhatrang, wrote: "The fruits, as a matter of fact, resemble small oranges about 4 cm in diameter, with five locules almost completely occupied by two very large seeds surrounded by pulp formed by succulent hairs, as in the genus Citrus and in Atalantia citroides."   He went on to say that it is called kim-do-um by the natives at Nhatrang and that, since it grows on sea beaches, it is probably salt-resistant.
      Limnocitrus, from its habitat, is almost certain to possess a high tolerance to salinity in the soil moisture.   Citrus can be grafted on Merope, but has made only unsatisfactory growth.   It is probable that Citrus would grow better when grafted on Limnocitrus, to which it is much more closely related than to Merope (a member of another subtribe).   Such a rootstock might be expected to be useful wherever Citrus is grown in soils showing a high salinity in the soil moisture, as has been found to be true with the xerophytic Eremocitrus, native to more or less saline flats in the semiarid plains of Australia.

XVIII.   Hesperethusa Roem.
      XVIII.   Hesperethusa Roem. Syn. Hesper. 1:31. 1846. Limonia Sec. III, Hesperethusa Engl. Die Nat. Pflanzenfam. 3(4):190. 1896.
      Type species (monotype.—Limonia crenulata Roxb. = Hesperethusa crenulata (Roxb.) Roem.; from British India.
      Distribution.—Northern India, Burma, southwestern China, Thailand, South Vietnam, Laos, Cambodia.
      Common name.—Hesperethusa.
      Slender, thorny trees or shrubs; leaves persistent, odd-pinnate; petioles and rachis broadly winged (petioles sometimes more narrowly winged); inflorescences axillary, few-flowered; flowers small, 4-merous (except stamens, twice as many as petals); ovary with 4 locules, 1 ovule in each locule; fruits small, 1 or more of the 4 segments with a single seed, surrounded by scanty, bitterish pulp containing large rudimentary sessile pulp-vesicles of irregular size and shape; seeds hard, smooth, germinating with epigeous cotyledons.
      Hesperethusa has apparently no near relatives, differing widely from all the other genera of the Citrus Fruit Trees except Citropsis in having odd-pinnate leaves with three to four pairs of small, opposite, crenate leaflets and broadly winged petioles and rachis segments.   Pinnate leaves were undoubtedly borne by the remote ancestors of all the Citrus and Citroid Fruit Trees (the tribe Citreae).   The characters persist today only in two out of the thirteen genera (Hesperethusa and Citropsis)of the second subtribe, Citrinae, and in two out of the seven genera (Feronia and Feroniella) of the third subtribe, Balsamocitrinae.   None of the eight genera in the first subtribe, Triphasiinae, have pinnate leaves.   In other words, only two of the three subtribes of the tribe Citreae have genera with pinnate leaves, and out of a total of twenty-eight genera in the tribe only four have pinnate leaves.
      Hesperethusa has some other points of similarity with the African genus Citropsis aside from pinnate leaves.   The seeds of both genera have a hard testa, cream-yellow in color, composed of radial prismatic cells, some with very thick walls.   The little known Citropsis daweana from Portuguese East Africa has the smallest leaflets of any of the species of Citropsis, with widely winged rachis segments but nearly wingless petioles, making the leaves look somewhat like those of certain forms of Hesperethusa crenulata.
      The very small fruits of this monotypic genus show rudimentary pulp-vesicles attached to the dorsal wall and apparently also to the base of the locule.   There are relatively few of these pulp-vesicles in each locule of the developing fruit; they are not clearly differentiated into basal and apical portions but are irregularly rounded and seem to be thin-walled; they vary greatly in size, and are often polygonal from mutual pressure.   These rudimentary pulp-vesicles can be seen in the sections12 of the fruits of paratypic material in the Kew Herbarium which was collected by Roxburgh in the last quarter of the eighteenth century and from which he drew up the description of Limonia crenulata, the type species of the genus.
      The pulp-vesicles of Hesperethusa that arise from the dorsal wall of the locule appear like small vesicles adhering to the wall by a broad base and are broader than high.   The few pulp-vesicles that arise from the base of the locule are more or less slender and elongated but do not have any clearly marked stalk.   The shapes taken by the pulp-vesicles seem to be governed by the free space available into which they can expand as they develop.   The pulp-vesicles attached to the dorsal wall are somewhat like those of Pleiospermium in shape but they do not show any definite cortical layer and furthermore do not show any degenerations of the central portion into an oily or resinous (?) mass.
      Hesperethusa can be grafted on Citrus and such grafts live and grow for many years.
      Dr. Anna E. Jenkins (1936, pp. 71, 73) discovered on the fruits of Hesperethusa from India a fungus, Elsinoë sp., very like the scab fungi, E. fawcettii and E. australis, that attack species of Citrus.
      Hesperethusa crenulata (Roxb.) Roem. Syn. Hesper. 1:38. 1846. Limonia crenulata Roxb. Pl. Corom. 1795; L. acidissima Auct. (non L.). Illus. Rheede & Draakestein, Hort. Malabar. 4:31, pl. 14. 1679; Roxburgh, loc. cit. 1:pl. 86; Talbot, Forest Fl. Bomb. Presid. 198, fig. 121. 1909; Engler, Die Nat. Pflanzenfam. 3(4):190, figs. J, K, L. 1896; Swingle, in Bailey, Stand. Cycl. Hort. 3:1478, fig. 1825. 1915; fig. 3-23 this work.
      Type.—Peninsular British India (Roxburgh).   Herb. Brit. Mus., London (fide Tanaka).
      Distribution.—West Pakistan, India, Burma, Thailand, southwestern China, South Vietnam, Laos, Cambodia.
      Common name.—Hesperethusa.
      Talbot (1909, pp. 198, 199) described this species as follows: "A spinous, glabrous, small tree: spines straight 0.5"-1" [12-25 mm] long.   Leaves imparipinnate; leaflets 2-3 pairs, 1"-2" [2.5-5.1 cm] by 0.5"-1" [1.2-2.5 cm], conspicuously gland-dotted, sessile, ovate, emarginate, crenulate; terminal leaflet usually largest; rachis and petiole broadly winged, jointed; joints of rachis obovate.   Racemes short, axillary, pubescent, subumbellate, few-flowered.   Flowers 4-merous.   Calyx small, glandular, 4-lobed; lobes ovate.   Petals 4, fragrant, white, 0.25" [6 mm] long, elliptic.   Disk annular or stipitate.   Ovary 4-celled with 1 pendulous ovule in each cell; style short.   Fruit globose, 1-4 seeded, 0.25" [6 mm] in diameter, black when ripe; pedicels 0.75" [19 mm] long."
      Hesperethusa has a wide range from West Pakistan to Burma, southwestern China on to southern Indo-China (Cambodia).   In West Pakistan and India, it occurs commonly in the sub-Himalayan tract from Ravi River (Long. 75° 30' E.) eastward, ascending to 4,000 feet [1,220 m] altitude.
      This species is a small tree reaching a height of 25 to 30 feet (7.6 to 9.14 m); it grows commonly on dry hills or in dry jungles.   It was said by Haines (1921, vol. 2, p. 163) to be "subdeciduous at the time of flowering."   The wood is largely used in Bihar and Orissa, India, for cart axles.   Several authors have stated that the spines are sometimes paired and that the leaves often have four pairs of leaflets.   The inflorescences often bear from one to several foliage leaves at the base, and leaf spurs (Kurztrieben), with very short internodes, are often borne in the axils of the leaves of older twigs.   The petioles of the leaves on such Kurztrieben are often more narrowly winged than the rachis segments of the same leaves and are sometimes almost wingless.   The fruits are said by many authors to be very acid, but Haines (1921, vol. 2, p. 163) reported them to be "intensely bitter (not acid)."13   Guillaumin (1911, p. 660) spoke of the seeds being "immersed in a mucilage," but no one seems to have realized up to now that the locules of the fruit contain rudimentary pulp-vesicles.   Hesperethusa crenulata produces a hard, close-grained, light yellow wood in India, and its leaves, fruits, and roots are used in making medicine in that country.   According to Watt (1890, vol. 4, p. 642) the fruit is exported from India "to the Arabian coasts, where it is used as a condiment with fish, meat, etc., being powdered along with the spices commonly used in cooking."
      Hesperethusa crenulata is promising as an ornamental because of its beautiful, feathery, green foliage (see fig. 3-23).   Grafts of it on Citrus were still growing in the greenhouses of the Agricultural Research Service at Washington after 20 years, although the union of the stock and scion was somewhat swollen.   It can also be grafted on Swinglea glutinosa (the Philippine tabog), which belongs to the subtribe Balsamocitrinae.

Group B.   THE NEAR-CITRUS FRUIT TREES
      This group includes two genera, Citropsis and Atalantia, that show well-developed pulp-vesicles having broad sessile bases and conical sides that taper regularly to the acute apices.14   These pulp-vesicles are arranged more or less radially with the bases at the periphery of the fruit locules attached to the dorsal walls of the locules and imbedded in the inner layer of the rind.   The conical pulp-vesicles usually point toward the center of the fruit unless deflected by the seeds, which (if present) are attached at the inner angle of each locule.   The genus Citropsis, native to tropical and subtropical Africa, has eleven species, all but one with pinnate or trifoliolate leaves resembling in several important characters those of the genus Hesperethusa of group A; Atalantia, native to southeastern Asia (including Ceylon), likewise has eleven species, but all have unifoliolate or simple leaves much resembling those of Citrus of group C in size and shape but differing in their more leathery texture and finer venation.
      Both Citropsis and Atalantia are obviously related to the True Citrus Fruit Trees of group C, and some three species of Citropsis and two or three special of Atalantia (all that have been tested) can be grafted on Citrus, and vice versa.   These two genera are intermediate in character between the Primitive Citrus Fruit Trees of group A and the True Citrus Fruit Trees of group C; like the genera in group A they contain many obvious connecting links between Citrus and the Minor Citroid Fruit Trees of the subtribe Triphasiinae.

XIX.   Citropsis (Engl.) Swing. & M. Kell.
      XIX.   Citropsis (Engl.) Swing. & M. Kell. Jour. Agr. Res. 1:421. 1914. Limonia Sec. citropsis Engler, Die Nat. Pflanzenfam. 3(4):189. 1896.
      Type species.—Limonia preussi Engl. (from Elephant Lake, Cameroons) = Citropsis articulata (Willd.) Swing. & M. Kell. (ranging from Togo to Cameroons and to Belgian Congo).
      Distribution.—Not uncommon from Uganda to Nigeria and Sierra Leone, and south to Republic of the Congo, Angola, and Mozambique.
      Common name.—African cherry-oranges.
      A genus of Near-Citrus Fruit Trees found in tropical Africa.   Shrubs or small trees with 1 or 2 well-developed spines (C. angolensis is said to be spineless) in the axils of the leaves; leaves pinnate, sometimes 3-foliate [sic], rarely 1-foliolate (simple in C. tanakae); rachis and petiole usually broadly winged (wingless in C. le-testui), flowers 4- or 5-merous, usually in dense axillary clusters or short racemes; stamens twice as many as the petals, filaments free, flattened laterally and in some species cohering laterally at the base in groups; ovary with 4 or 5 locules, each with a single ovule; fruit spherical, small (2-3 cm diam.), with pulp-vesicles broad at the base where they are attached and tapering to an acute tip; seeds large (10 X 6 X 4 mm); cotyledons hypogeous in germination.
      This genus is certainly closely related to Citrus, in spite of the fact that all but one species of Citropsis have pinnate or trifoliolate leaves.   It belongs, along with Atalantia, to group B, the Near-Citrus Fruit Trees, of the subtribe Citrinae.   Both of these genera have small, subglobose fruits, 1.5 to 3 cm in diameter, with 3 to 5 locules, in which occur broad-based, sessile, conical, radially arranged pulp-vesicles.   All the species of Citropsis have leaves, or leaflets, very similar to those of Citrus in shape, venation, and texture, whereas the leaves of the species of Atalantia are thick, more leathery, and much more veiny than those of Citrus.
      The seeds of Citropsis are plump and broadly ovoid with a very hard, smooth testa.   The hilum is an oval or ovate aperture bordered by the incurved testa, which is here very hard and thick.   No other genus of the subtribe Citrinae has seeds as plump and as hard-shelled.   The flowers of Citropsis differ greatly in the various species with respect to size and proportion of parts, and especially with respect to the morphology of the pistil.   The leaves differ even more than the flowers in the different species, but fortunately for the taxonomist the leaf and flower characters are not closely correlated; that is, a pair of species having very similar leaves, like C. schweinfurthii and C. angolensis, may show easily recognized differences in the flowers.   Citropsis schweinfurthii and C. gilletiana have similar flowers, with only slight differences in the pistil, but very different leaves.
      The species of Citropsis look like missing links, i.e., surviving forms of the remote ancestors of Citrus and may, in fact, be such forms that migrated tens of millions of years ago from the southeastern Asiatic homeland of the tribe Citreae and found safe and congenial refuge in the tropical forests of Africa, where they have persisted with little change ever since.   Once safely in Africa, the ancestors of Citropsis were freed from competition with the numerous species of related genera of the subtribe Triphasiinae as well as from the danger of hybridization with vigorous new genera that evidently arose in rapid succession in southeastern Asia and the East Indian Archipelago.   Thanks to this age-long isolation, even today the leaflets of Citropsis (and especially the unifoliolate leaves of certain forms of C. gabunensis) resemble very closely the leaves of Citrus (and doubtless those of the common ancestor of Citrus and Citropsis) in shape, texture, venation, and color.   That this resemblance of the leaves is no mere analogy is proved by the fact that the species of Citropsis, so far as they have been tested, graft readily on Citrus, and vice versa.   Citropsis and Pleiospermium are the two genera that most nearly represent today the remote ancestors of Citrus.
      The species of Citropsis are all native to tropical and subtropical Africa, from Sierra Leone to Angola on the west coast and from Uganda south to Mozambique in East Africa, as well as to many, if not all, of the regions of central Africa lying between.   As this immense territory includes vast areas still only inadequately explored, it is highly probable that several, perhaps many, more species of Citropsis will be discovered in the future.
      A striking proof of the inadequacy of our botanical exploration of tropical Africa is shown by the fact that, of the eleven species of Citropsis here recognized, eight were discovered after 1900 and five of these after 1920.   The last three species to be found—C. gilletiana, C. tanakae,and C. daweana—were described as recently as 1940!   The two last-named are strikingly different from all the other known species and from each other.   They are not mere small species made by splitting large species, but are new and aberrant members of the genus, which when better known may need to have new subgenera made to include them.   One subgenus, Afrocitrus, has already been named to include three allied species, all growing in western tropical Africa.
      In the Congo, it was found that Gillet’s cherry-orange, C. gilletiana, the most vigorous species in the genus, growing to be a tree 10 meters high, could be used to advantage as a rootstock for cultivated varieties of Citrus because of its resistance to foot rot (very injurious to cultivated varieties of Citrus there).   This species proved to be an even better rootstock than the sour orange for the orange, the mandarin, the pummelo, and the lemon (see below).   It is one of the more important rootstocks for Citrus found in this century.
      In view of the beauty of their foliage, the refreshing fragrance of their flowers, and the attractive aroma of their tiny fruits, coupled with their rapid growth and free flowering while still small trees, an effort should be made to test all the species of Citropsis as greenhouse ornamentals in the cooler climates and as out-of-door ornamentals in subtropical regions.   Keys to the subgenera and species of Citropsis are presented.

Subgenus Citropsis
      Subgenus Eucitropsis Swing. & M. Kell. Jour. Arnold Arbor. 21:125. 1940.
      Type.C. Schweinfurthii.
      Leaflets usually large, petioles and rachis segments broadly winged; ovary ovoid without a single large oil gland at the top of each locule; stigma medium-sized, depressed-globose, without large oil glands; staminal filaments glabrous.
      Most of the species of this subgenus have large leaves, and three of them, C. articulate, C. gilletiana, and C. latialata, have leaves and leaflets larger than those of any other members of the tribe Citreae except possibly the palmately trifoliolate leaves of certain species of Luvunga.   The species of this subgenus are rather sharply distinguished from those of the subgenus Afrocitrus by the ovary and stigma characters.
      1.   Citropsis schweinfurthii (Engl.) Swing. & M. Kell. Jour. Agr. Res. 1:426. 1914. Limonia schweinfurthii Engl. Notitzbl. Bot. Gart. Berlin 1:28. 1895. Illus. Swingle & M. Kellerman, loc. cit. figs. 1-2, 4(C)-7, pl. 49; Swingle, in Bailey, Stand. Cycl. Hort. 2:780, fig. 972. 1914; Engler, Die Nat. Pflanzenfam. 19a:348, fig. 158,K (copy of fig. 4,C above). 1931; fig. 3-24 this work.
      Type.—Uando, northwestern Uganda (Schweinfurthii, No. 3656, 1870).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Northeastern tropical Africa, northwestern Uganda, and northeastern Republic of the Congo, at altitudes of 600-1,600 m.
      Common name.—Uganda cherry-orange.
      Small tree, ultimate branches at first angled, soon terete, often with single or paired spines 1-2.5 cm long; leaves medium-sized, 3-5-foliolate, 12-30 X 12-20 cm, leaflets oblong or lanceolate, tapering rather abruptly to a short, blunt acumen, cuneate at base, margins denticulate to subentire, terminal leaflets usually narrowly cuneate at base, 10-15.5 X 3-5.5 cm, lateral leaflets more broadly cuneate at base, 8-14 X 3.5-5.5 cm, petioles broadly winged but usually slowly narrowed and rounded at both ends, 5.5-10 X 2-3 cm, rachis segment similar in shape to petioles but shorter and narrower, 4-9 X 1.5-2.8 cm; inflorescences axillary, short (1-2 cm), many flowered, pedicels 2 mm long, 1 mm wide at base, gradually thicker upward and about 2 mm wide at base of calyx, flowers white, fragrant, usually 4-merous, about 2.2-2.5 cm when fully expanded, calyx with 4 thick, short, blunt lobes, petals 4, 12-15 X 4-5 mm, broadly rounded at tips, stamens 8, 10-11 mm long, flattened and broad 3/4 of way to top, 0.5 mm thick, 1.4-1.6 mm broad; disk 1 mm high, 3 mm broad, surrounding narrowed base of ovary with a shallow (0.3-0.4 mm), thick-walled cup with small scattered oil glands on inside near base, pistil 10-11 mm long (including disk), ovary ovate, 4-locular, 2 mm high, 2.5 mm wide, rounded at tip with style often (always ?) attached to one side of center of tip!, style short, stout, 6-7 X 1.1-1.4 mm, expanded slightly at tip, stigma cushion-shaped, 0.9-1.1 mm high, 2-2.5 mm wide, 4-lobed at tip, with very small oil glands between stylar canals; fruits subglobose, lemon-yellow when ripe, fragrant, 2-2.2 cm diam., locules filled with numerous, sessile pulp-vesicles that occupy all free space, seeds slightly roughened, 9-12 X 7-8 mm, monoembryonic.
      Citropsis schweinfurthii was grown for many years in the United States and thrived in the Washington greenhouses of the U.S. Department of Agriculture, where it flowered profusely and set fruit freely while still a small tree.   The leaflets are narrower than those of most of the other species except the related C. angolensis of Angola and the very different C. daweana of Mozambique.   The leaflets of the Uganda cherry-orange show 6 to 8 lateral veins, slightly elevated below and slightly depressed above.   The surface of the leaflets is nearly plane.   The fruits of C. schweinfurthii are often borne in clusters of two to five on short axillary inflorescences; they are almost spherical or very slightly flattened at the top and 22 to 25 mm in diameter.   The peel is very thin, 1 to 1.5 mm thick.   The ripe fruit ranges in color from Empire yellow to Apricot yellow (Ridgway, 1912, pl. 4).   The pulp, when seen in cross section, is straw yellow, amber yellow, or even mustard yellow (Ridgway, 1912, pl. 16) near the peel.   The pulp-vesicles fill all the space in the cells of the ovary not occupied by the plump seeds, of which 1 to 4 may develop, never more than one in each locule.   The pulp-vesicles contain in their center a long column of oil droplets, irregular in diameter and not sharply delimited.   The oil droplets are extremely variable in size, though most of them are small.   The mawkish taste of the pulp is doubtless due to these droplets of oil.
      The Uganda cherry-orange can be grafted readily on Citrus (Swingle and M. Kellerman, 1914, p. 435, pl. 49).   It can also be made to grow, but less vigorously, when grafted on the wood apple, Feronia limonia, or on the Philippine tabog, Swinglea glutinosa.   It is well worth growing as an ornamental in the greenhouse, as it not only has handsome foliage, but also bears, for months at a time, a great profusion of white bloom looking much like orange flowers, with a strong but agreeable odor about midway between that of citrus flowers and that of the North American wild plums (Prunus americana and other species).
      2.   Citropsis angolensis Exell, Jour. Bot., Brit. & For. 65(Suppl. 1):53.
      Type.—Angola, Amboine (Grossweiler, No. 4454).   Herb. Brit. Mus.   Fragments and serial microtome sections, S. and T. Nos. 273-A, 273-B, and 273-C, 12 slides, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known with certainty only from the type locality.
      Common name.—Angola cherry-orange.
      The original description by Exell reads as follows: "A tree with brown branchlets; leaves trifoliolate, leaflets elliptical or obovate, cuneate at base, margins remotely and irregularly crenate, glabrous on both surfaces, petioles broadly winged, elliptical; flowers pedicellate, pedicels glabrous, in clusters in the axils of the leaves; sepals 4, triangular, small; petals 4, oblong, imbricate, rounded at the tips; stamens 10, filaments free, flattened, glabrous; ovary 4-locular, style scarcely thickened at the base.
      "A tree, total height 15 feet [4.6 m]; shortly branched from the base.   Leaflets 8-12 X 4-6 cm; petiole 7-8 X 2.5-3 cm; pedicels 5-7 mm long; petals 12-14 X 5-6 mm; filaments of stamens 8-9 mm long; anthers 4 mm long affixed near the base; ovary 4 mm long and 2.5 mm in diam., style 6 mm long; young fruits 8 X 5 mm.
      "The species is nearest to C. Schweinfurthii (Engl.) Swing. and M. Kell., but has rather longer pedicels and a slightly longer and slenderer style—it is, moreover, apparently devoid of spines."
      The two flowers from the type specimen were kindly donated to Swingle by the Curator of the Herbarium of the British Museum.   A nearly mature flower bud was cut into several hundred serial microtome sections (both cross and longitudinal) and a fully mature flower from which the petals and stamens had just fallen was restored to natural state and photographed.   The following characters were observed:
      Pistil slender, 1.2 cm long, with a very slender style, 6 mm long, borne on a slender pedicel, 0.7 mm diam. at base and 1 mm diam. at top; calyx very small (2.3 mm diam.), lobes triangular, acute, with several (2 or 3) small oil glands near the tip; stamens free, filaments flattened, anthers 3-3.5 mm long, apparently without an oil gland near the tip; disk small, 1 mm high, 1.3-1.5 mm wide, and nearly cylindrical on mature, dry flowers, but bulging to 1.8 mm at mid-height in nearly mature flower buds (and on a restored mature pistil), not cupulate and not enclosing the ovary base; ovary truncate-fusiform (ellipsoid), 3.5(-4 ?) mm high, 1.3 mm wide at base, 1.8 mm wide in middle in dry flowers (2-2.5 mm wide on a fully restored mature pistil), 4-locular, narrowed at tip and merging gradually into style base; style slender, 6 mm long, 0.8-1.3 mm wide, expanded rather abruptly into the subglobose stigma which is 1.1 mm high, 1.6 mm wide when dry (1.2 mm high, 2 mm wide in restored mature pistil), with 1 or 2 small oil glands between the stylar canals.
      The flower characters of this species show clearly that, in spite of the general resemblance of its leaves to those of C. schweinfurthii, it is in fact a very distinct species, differing decidedly in its spineless twigs, in its longer ovary and its longer, more slender style arising from the center of the top of the ovary (not from one side as in C. schweinfurthii), and, even more strikingly, in its very different, noncupulate disk, which does not enclose the base of the ovary even to the slightest extent.
      This species of Citropsis can, as a matter of fact, be distinguished only with difficulty from C. schweinfurthii by any of the easily visible, grosser characters of the leaves, flowers, or fruits and has in fact been considered by experienced taxonomists to be a mere synonym of it, not even worthy of being distinguished as a variety.   At first glance the only distinguishing character is the absence of spines in C. angolensis and their abundance in C. schweinfurthii.   However, when the flowers are studied critically, striking differences appear which not only separate clearly the Angola cherry-orange from the Uganda species but also distinguish it from all the other known species of the genus Citropsis.
      3.   Citropsis articulata (Willd.) Swing. & M. Kell. Jour. Agr. Res. 1:433. 1914. Citrus articulata Willd. in Spreng., Syst. Veg. 3:334. 1826; Limonia preussii Engl. Notitzbl. Bot. Gart. Berlin 1:28. 1895; Citropsis preussii (Engl.) Swing. & M. Kell. Jour. Agr. Res. 1:423. 1914. Illus. Swingle & M. Kellerman, loc. cit. p. 424, fig. 3; p. 425, fig. 4,A; Engler, Die Nat. Pflanzenfam. 19a:348, fig. 158,F,G,H. 1931.
      Type.—In the mountains 50-75 km north of Accra (Isert. 1786).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Ghana, Togo, Cameroon, Republic of Congo (?).   This species widely spread in tropical West Africa, is everywhere a shrub, according to Engler (1931, p. 347).
      Common name.—West African cherry-orange.
      A shrub or small tree with solitary axillary spines, 0.5-2 cm long, even on fruiting branches; leaves large or very large, 15-33 X 8.5-25 cm, leaflets broadly oval or oblong-lanceolate, 13-19 X 7.5-10.5 cm, rounded at apex and abruptly narrowed into a blunt acumen often very short (3-5 mm) but sometimes longer (8-12 mm) and acute at the very tip, broadly or narrowly cuneate at base, lateral leaflets slightly smaller than the terminal one, 10.5-15 X 5-7.5 cm, apex like that of terminal leaflets but base more broadly cuneate or rounded, margins serrulate or shallowly crenulate, petioles 5.5-8.5 X 1.7-3.5 cm, narrowly elliptical, usually narrowed at both ends but sometimes broader at the tip or even obovate (as in the type specimen); inflorescences about 3 cm long (1/3-1/2 as long as the petiole); flowers white, 4-merous, calyx 4 mm wide, lobes triangular, acute, petals 4, linear-oblong, 18 X 4 mm, stamens 8, filaments 8 X 1 mm, anthers 3.5 mm long, affixed near the base, disk 1 mm high, ovary 4-locular, ovoid-acuminate, 3-4 mm high, 2.5 mm wide, style long (12-15 mm), slender, expanding gradually and merging with top of ovary, stigma subglobose, 2.5 mm diam., 4-lobed; fruits subglobose with a truncated, conical style base, 1.5-2 mm long, persisting at apex.
      This was the first species of Citropsis to be described.   It was discovered by P. E. Isert, who called it a new species of lemon with jointed leaves: "une nouvelle espèce de citroniers avec des feuilles articulées."   One of Isert's specimens was sent to the herbarium at Berlin, and Willdenow named it Citrus articulata in 1826, but did not describe the flowers, which had perhaps fallen off the "many-flowered peduncles" he mentioned.   In 1895, Engler proposed as a new species Limonia preussii, found growing on the shores of Elephant Lake in what is now Cameroon.   The description was the first full and accurate one published of a Citropsis.
      The very defective type specimen of Citropsis articulata, consisting of a short twig and two petioles in the Berlin-Dahlem herbarium, is all that is known to be extant of Isert’s material.   This type specimen, when studies by Willdenow, must have been more complete than it is now, for his original description reads, in translation: "C.[itrus] with large, articulated, obovate, leaf-like petioles, leaves oblong, peduncles many-flowered."   No inflorescences are now to be seen and no leaves (or leaflets).   Possibly, however, Willdenow mistook detached oblong petioles for leaves.   However, Engler (1915, p. 758), who had access to the type specimen of Citrus articulata and abundant material of his own species, Limonia preussii, said: "…indeed I cannot avoid combining my Limonia preussii [with Citropsis articulata] as the width of the petiole wing shows no specific differences."
      Engler gave C. articulata a very wide range, from the type locality north of Accra in modern Ghana through near-by Togo (Koli near Kame) east to what is now Cameroon (Elephant Lake, type locality of Limonia preussii) and south to the present-day Republic of Congo (Brazzaville) and the Republic of the Congo (Leopoldville) (including Limonia demeusii De Wild. and L. poggei Engl., of which the flowers are not yet know).   Future study of flowering material is needed, however, to determine whether C. articulata has so wide a range or not.
      For the present, Engler's view concerning the identity of C. articulata is accepted, but flowering specimens from the type locality might even show it to be identical with C. mirabilis, native to the ivory Coast, adjoining Ghana to the west, in which event the name C. preussii would be applied to what is here called C. articulata.
      Citropsis preussii, here considered as synonymous with C. articulata, is a shrub with very large leaves, nearly as large as those of C. gilletiana, which grows to be a tree sometimes 10 meters high.
      4.   Citropsis mirabilis (A. Chev.) Swing. & M. Kell. Jour. Agr. Res. 1:432. 1914. Limonia mirabilis A. Chev. Bul. Soc. Bot. France 58(mem. 8d):144. 1912. Illus. Swingle & M. Kellerman, loc. cit. p. 425, fig. 4,B (pistil only); Engler, Die Nat. Pflanzenfam. 19a:348, fig. 158,J (copy of above). 1931.
      Type.—West Africa, Ivory Coast, between Sanroa and Quode on Koué River (an affluent of the Sassandra River) (Chevalier, 1909, No. 21609).   Herb. Chev., Paris.
      Distribution.—West Africa: Ivory Coast.   Up to now this species was known only from the type locality.
      Common name.—Ivory Coast cherry-orange.
      A much-branched and very spiny shrub, 2-5 m high, completely glabrous, spines 2-4 cm long; leaves 3-5-foliolate, leaflets oblong or broad-elliptical, 9-16 X 5-9 cm, the terminal leaflet larger than the lateral ones, apex narrowed into a short, blunt acumen, base cuneate or badly rounded (when the terminal leaflet is borne singly on a short rachis segment, as in some 3-foliolate leaves); inflorescences small (5-10-flowered), axillary racemes, 2-3.5 cm long, pedicels slender, 3-5 mm long, 0.5 mm wide at base and gradually expanding to 1 mm diam. at top; flowers white, 4- (rarely 5-) merous, sepals greenish, 1 mm long, petals linear, 13-15 X 4-5 mm, soon falling, stamens free, filaments flattened, disk very short (0.5 mm), forming a loose, very shallow cup (0.2-0.3 mm deep) at base of ovary, pistil in fully developed flowers 11-13 mm long (including disk), ovary slender, 2-2.5 X 1-1.2 mm, style 6-9 mm long, slender at base (0.2-0.3 mm diam.), gradually expanding upward to 0.6-0.7 mm at top, stigma 0.5-0.8 mm high, 1-1.2 mm wide, 4-lobed, with 2 small oil glands in each space between stylar canals; fruits (only immature seen) borne on pedicels.
      This species much resembles C. articulata, differing chiefly in the somewhat shorter style not expanding at the base and not merging into the top of the ovary.   Possibly C. mirabilis, when better understood, will be ranked as a variety of C. articulata occupying the westernmost part of its wide range.
      5.   Citropsis gilletiana Swing. & M. Kell. Jour. Arnold Arbor. 21:116. 1940. Illus. "W. G. S[mith (?)]," Gard. Chron. 3 ser. 53:378, fig. 159. 1913; Goossens, Bul. Agr. Congo Belge 15:161, fig. 45. 1924; Swingle & M. Kellerman, loc. cit. pl. 1, figs. 1-7; pl. 2, figs. 1-5; fig. 3-25 this work.
      Type.—Washington, D.C., Bureau of Plant Industry greenhouse, tree grown from seeds sent by Père J. Gillet from type locality, Kisantu, Belgian Congo (Swingle, C.P.B. No. 7800 G = P.E.I. No. 109622), flowering branch; also several paratypes of leafy shoots, flowering and fruiting branches, fruit, etc., cut from type tree; photographs and serial microtome sections S. and T. Nos. 501-A, 501-B, 501-C, and 501-D, 40 slides with 1,339 sections from type tree, in Herb. Natl. Arbor., Washington, D.C.
      Topotypes.—Belgian Congo, Kisantu, Jardin Botanique (J. Gillet); 2 branches with very young fruits, March 5, 1913, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Reported (Goossens, 1924, p. 162) from Ubangi and Equateur regions of northwestern Belgian Congo (now Republic of the Congo) and cultivated at the Jardin Botanique d'Eala.
      Common name.—Gillet's cherry-orange.
      Tree 8-10 m tall, young branches glabrous, 2.5-4 mm diam., internodes 2-6.5 (usually 3.5-4.5) cm long, spines slender, acute, 2-5 cm long, solitary or in pairs; leaves very large, 3-5-foliolate, 16-37.5 X 10-36 cm, terminal leaflets 9-21 X 5-14 cm, lateral ones (always smaller than the terminal) 7-17 X 4-9.5 cm, petioles obovate, 6-10 X 3.5-7.2 cm, rachis segments (1 or 2) elliptical or obovate, 3.5-9.5 X 1-5-7.5 cm (usually 3.5-6 X 1.5-4 cm); inflorescences racemose, axillary, short, 5-12 mm long, flower pedicels 3-5 mm long; flowers white, numerous, about 2.5 cm diam., but often opening incompletely because densely crowded in short inflorescences 4-parted, calyx lobes triangular, acute, 2 X 2 mm, petals 16-18 X 4.5-5.2 mm, stamens 8, filaments glabrous, flattened, free, disk 0.8-0.9 mm high, 2.2-2.4 mm wide, cupulate (0.3 mm deep), pistil (including disk) 10-13 mm long, ovary ovoid or barrel-shaped, 3.5-4 X 2.5-3 mm, 4-locular, locules with 1 ovule, style 4.5-6 X 1.2-1.3 mm, stigma depressed globose, 1.2-1.3 mm high, 2.5 mm wide; fruits subglobose, 24-25 mm diam., with pedicels 7-9 mm long, lemon-colored when ripe, peel thin (1-1.5 mm), locules with very numerous pulp-vesicles filled with granules of yellowish wax; seeds white, smooth, usually 2-3 (often none), ovoid, 9.5-10 X 5-6 mm, testa very hard, hilum ovate, 1-2 mm diam. with smooth edges, monoembryonic.
      The type tree of C. gilletiana was grown in the citrus greenhouse of the former Bureau of Plant Industry from seeds received from Père Gillet in March, 1913.   Unlike C. schweinfurthii, growing alongside in the greenhouse, the young tree did not flower for many years and then only sparingly; later it flowered and fruited profusely.   The type tree produced enormous leaves, both trifoliolate and 5-foliolate.   A trifoliolate leaf at the base of a vigorous water sprout was 28.5 cm long and 31 cm wide.   The terminal leaflet measured 20.5 by 14 cm, the two lateral leaflets 16.5 by 9.5 cm.   The largest leaflets have 7 to 10 principal lateral veins, raised on the under surface and sunken on the upper surface.   The surface of the leaflet is usually more or less bullate because of the upward curving of areas of the leaf surface limited by the principal lateral veins and their cross veinlets.   The winged petiole, cordate in outline, measured 8.5 by 7.2 cm.   A large 5-foliolate leaf higher up on the same shoot measured 37.5 cm long and 36 cm wide; the terminal leaflet (somewhat deformed) was 18 by 10 cm.   The two higher lateral leaflets were 18.5 to 19 by 10.5 to 11 cm.   One of the lower leaflets measured 16 by 10.25 cm (the other was imperfect).   The elongate-elliptical winged rachis segments measured 9.5 by 4.5 cm, the elliptical winged petiole about 10 cm long and about 7.5 cm wide (one side defective).   These are the largest leaves yet reported on either a Near-Citrus or a True Citrus Fruit Tree, though they are greatly exceeded in size by the leaves of some species of Clausena.
      In trifoliolate leaves, the terminal leaflet is usually decidedly longer than the adjoining laterals and frequently much wider, from one-fourth to one-third or even more; when the terminal leaflet, as often happens, is borne on a separate winged rachis segment, it is sometimes even two-fifths wider.
      The stigma in cross section shows eight small oil glands (two between each pair of stylar canals).   The fruits are often crowded, three to six in a dense cluster in the axils of the leaves of the fruiting branches; they are usually subglobose, 2 to 2.5 cm in diameter, but are sometimes depressed-globose, 2 cm long and 2.3 to 2.4 cm wide; they become yellow-orange and are fragrant when ripe (see fig. 3-25).
      Gillet’s cherry-orange, the largest and most vigorous of all the known forms of the genus Citropsis, has been reported by Belgian expert horticulturists and pathologists (Goossens, 1924, pp. 156-62, figs. 42-45: Staner, 1929, pp. 364-66, fig. 170; and Pynaert, 1935, pp. 305-14), on the basis of experiments carried out at the Eala Botanic Garden near Equatorville in Belgian Congo (now the Republic of the Congo), to be immune to a very destructive form or foot rot, supposed to be caused by the brown-rot fungus (Phytophthora citrophthora), that attacks the lower trunks and upper roots of the species of Citrus grown in the Congo, including the sour orange, Citrus aurantium, usually found to be very resistant to foot rot (see Fawcett, 1936, p. 177).   It appears that in the Congo foot rot is greatly aggravated in severity by the larvae of a longicorn beetle, Monohammus sp. (see Goossens, 1924, p. 159, and Staner, 1929, pp. 365-66), which attacks first the cambium layer and later the wood of the base of the trunk.   Citropsis gilletiana, besides being immune to the attacks of the foot-rot fungus, is not attacked by this beetle.   It has been found that Gillet's cherry-orange makes an excellent rootstock for the cultivated varieties of sweet orange, mandarin, grapefruit, and lemon which were all found to grow more vigorously when grafted on Gillet's cherry-orange than on sour orange rootstocks.   Furthermore, citrus trees grafted on Gillet's cherry-orange rootstocks were completely immune to the Monohammus beetle and also to foot rot, provided the Gillet's cherry-orange stocks were grafted high enough above the ground to prevent infection from the soil.
      Pynaert (1935, p. 313) reported that the native Citropsis growing about the Eala Botanic Garden in Belgian Congo had been introduced into Belgium, and that a Jaffa orange grafted on a plant of Citropsis growing in a greenhouse of the Colonial Garden at Laeken, Belgium, had made a particularly vigorous growth and in 1935, when still only 2 meters high, had produced a number of fruits.   An excellent half-tone figure of the wild Citropsis found near the Botanic Garden in Belgian Congo, published by Goossens (1924, p. 161, fig. 45), shows clearly that this cherry-orange is the same as the one sent to Swingle by Père Gillet from Kisantu, about 700 kilometers down the Congo River from Eala.   It has the terminal leaflet distinctly larger than the laterals, agreeing with C. gilletiana (but not with C. latialata).   Goossens reported that near Eala this Citropsis is a tree growing to a height of 8 to 10 meters (26 to 33 feet) that bears tiny, orange-like fruits, 2 to 2.5 cm in diameter, with a resinous, fragrant pulp in which three or four seeds are imbedded.   The fruits produced by the type tree of Gillet's cherry-orange growing in the citrus greenhouse at Washington agreed very closely with those described by Goossens, and the foliage, spines, and fruit clusters of this type tree resembled exactly what is shown in Goossen’s photograph of the Citropsis growing wild about Eala.
      6.   Citropsis latialata (De Wild.) Swing. & M. Kell. Jour. Wash. Acad. Sci. 28:533. 1938. Limonia poggei var. latialata De Wild. Ann. Mus. Congo Bot. 5 sér.1:160. 1904. Illus. De Wildeman, loc. cit. pl. 43; Swingle & M. Kellerman, Jour. Arnold Arbor. 21:pl. 2, figs. 6-10. 1940; fig. 3-26 this work.
      Type.—Ikongu village, Sankuru River basin, Lualaba-Kasai district, south-central Belgian Congo (L. Gentil. No. 1).   Herb. Jard. Bot. l'Etat Brussels; photographs and serial microtome sections, S. and T. No. 397-A, slides 1-6, 429 cross sections, and No. 397-B, slides 1-2, 26 longitudinal sections of a mature flower from the type specimen, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known with certainty only from the type collection.
      Common name.—Ikongu cherry-orange.
      De Wildeman's original French description of Limonia poggei var. latialata reads, translated, as follows: "Shrub 3-4 m tall, with slender, cylindric branches, with imparipinnate leaves, having 2 pairs of lateral leaflets.   Leaves distant 1-5 cm from each other, with the petiole and rachis winged; petioles 3-8.5 cm long and 1.5-5 cm broad, upper [pair of] leaflets distant 2-8.5 cm from the lower [pair of] leaflets; rachis reaching 4.5 cm in width.   Leaflets oblong, narrowed into a very short petiolule at the base, cuneate at the apex and at the base, the midrib dividing the leaflet in 2 more or less unequal parts.   Leaf-blades 6-15 cm long and 2-7 cm wide, irregularly denticulate on the margins; the terminal leaflet regular, cuneiform at the base, of the same size as the lateral leaflets; axillary stipules changed into solitary or paired thorns, more or less developed, reaching 2.5 cm in length.   Flowers white, with the odor of orange [flowers], in axillary or terminal inflorescences; ovaries ovoid, terminated by a style twice as long, surmounted by a trilobed stigma."
      A note states that the type material (L. Gentil, No. 1) was in flower when collected but "the petals fell off during the preparation."   Evidently the flowers were mature.
      A single flower from the type specimen (see Swingle, 1940c, p. 120) was restored by the modified Juel method, imbedded in paraffin, and cut into serial microtome sections.   First 26 longitudinal sections were cut until a section was obtained exactly through the center; then the paraffin block was turned at right angles and 429 cross sections 20 mu thick were cut, including the entire length of the pistil, disk, calyx, and pedicel.   These sections give the following characters:
      Calyx is 4-merous, the lobes about 1.2 mm wide and 1.5 mm long, thick in the middle but with thin edges; disk shallow, cup-shaped, 0.3-0.4 mm tall and 1.1-1.2 mm wide, pistil 6.4 mm tall, including the short (1 mm), very shallowly cupulate disk, ovary ovoid, 1.8 mm tall and 1.6 mm wide, merging rather abruptly into the style, which is 0.8 mm diam. at the junction with the ovary and nearly 1 mm diam. at the stigmatic junction; stigma cushion-shaped, 1.5 mm wide and 0.5-0.6 mm high, more or less 4-lobed, with 4 medium-sized oil glands between the 4 stylar canals.
      Thanks to the generous loan by Dr. W. Robyns, director of the National Botanic Garden at Brussels, of one of the very few flowers of the type specimen, it was possible to clear up the long-standing confusion of this species with C. gilletiana, from which it can now be clearly distinguished by many characters of high taxonomic value in this genus.   Citropsis latialata is a shrub only 3 or 4 meters tall instead of a tree 8 to 10 meters high, it has somewhat narrower leaflets and the terminal leaflet is of the same size or only very slightly larger than the adjacent lateral leaflets (in C. gilletiana, on the contrary, the terminal leaflet is always larger than the lateral ones).   The pistil of C. latialata is only a little more than half as long as that of C. gilletiana and has a decidedly shorter and much less deeply cupped nectary, as well as a more flattened stigma with fewer but larger oil glands.
      The long-standing confusion of these two species was due, without doubt, to the very scanty type material of C. latialata and the lack of any other material of this species from the type locality.   De Wildeman's excellent lithographic plate figuring the type specimen of C. latialata, natural size, shows five pistils in a close cluster at the end of a leafy twig.   This is fully confirmed by a photograph of the type filed in the herbarium of the National Arboretum.   These pistils must have been full grown, as the petals were in place when the material was collected, but fell off during the drying of the type specimen.   As noted above, a single flower from the type specimen sufficed to distinguish clearly the two species C. latialata and C. gilletiana.

Subgenus Afrocitrus
      Subgenus Afrocitrus Swing. & M. Kell. Jour. Arnold Arbor. 21:126. 1940.
      Leaflets small, acuminate or caudate at apex, petioles and rachis segments narrowly winged or wingless; ovary clavate or obovoid, with a single very large oil gland at the top of each locule; stigma subglobose, distended by large oil glands; staminal filaments sparsely pubescent on inner side.
      The three species that constitute this subgenus, Citropsis gabunensis (type of Afrocitrus), C. zenkeri, and C. le-testui, are easily distinguished from all the other species of Citropsis by their long, acuminate leaves, narrowly winged or wingless petioles and rachis segments, and the striking ovary and stigma characters.
      Citropsis tanakae also has very small flowers borne on slender pedicels and very slender styles each ending in a subglobose stigma swollen by several large oil glands, all of which are characters very like those of the three species named above.   However, C. tanakae differs from them in lacking the large oil glands at the tip of the ovary, one over each locule, and also in having simple leaves with very short, wingless petioles, not articulated with the leaf blade.
      7.   Citropsis gabunensis (Engl.) Swing. & M. Kell. Jour. Agr. Res. 1:430. 1914. Limonia gabunensis Engl. Notitzbl. Bot. Gart. Berlin 1:28. 1895. Illus. Swingle & M. Kellerman, loc. cit. 425, fig. 4,D-E; De Wildeman, Ann. Mus. Congo Bot. 5 sér. 1:pl. 50. 1904; Engler, Die Pflanzenwelt Afrikas 3:760, fig. 354,L,N. 1915; Engler, Die Nat. Pflanzenfam. 19a:348, fig. 158,L,N (last two, copies of Swingle & Kellerman). 1931; fig. 3-27 this work.
      Type.—Sibange Farm, Menda region near Libreville, Gabon, French Congo (H. Soyaux, No. 105).
      Distribution.—Gabon Republic, Republic of the Congo, and Spanish Guinea.
      Common name.—Gabon cherry-orange.
      A shrub or small tree, 1.5-6 m high, ultimate branches slender, 1.5-3 mm diam., with internodes 2-5 cm long, with single or paired slender spines, 2-2.5 cm long even on fruiting branches; leaves glabrous, 1-5 foliolate, variable in size, 5-foliolate leaves 15.5-19 X 12-15 cm, 3-foliolate leaves 10-13 X 7-10 cm, 1-foliolate leaves 12-17 X 4-7 cm (including the petiole 2-4 cm long), leaflets oblong or elliptical, long-acuminate at the apex or caudate, very variable in size, 1-foliolate leaflets very large, blade 7-14.5 X 3-7 cm, petiole 2-3.5 X 0.4-1.5 cm, with a broadly rounded base, 5-foliolate leaflets usually 8-10 X 4-5 cm, but sometimes smaller, 3.5-8 X 1.5-3.5 cm, terminal leaflet often borne on a short rachis segment and then abruptly narrowed at base, but if borne at apex of first rachis segment with 2 lateral leaflets then narrowly cuneate or acuminate at base and longer than lateral leaflets, petioles variable in size and shape, 1.8-5 X 0.3-2 cm, linear or narrowly elliptical if narrow, obovate if broad, rachis segments often similar in shape to petiole, 3.5-5.5 X 0.4-2 cm; inflorescences short, axillary flower clusters, 1.5-2.5 cm long, peduncles apparently unbranched, 8-9 mm long, pedicels long and very slender, 7-11 X 0.4-0.6 mm (in dry state), short-pubescent, flowers very small, usually 4-merous, buds 5-6 X 2 mm, open flowers about 10-13 mm diam., calyx lobes 4, triangular, acute, with a single medium-sized oil gland near the apex, petals 4, about 6 X 2.5 mm, stamens free, filaments flattened, sparingly pubescent on inner side; disk very small, about 0.5 mm high, pistil very short (4-6 mm, including the disk), ovary obovoid, about 1.5 X 0.8-0.9 mm, 3-4-locular with a large, more or less protuberant oil gland at the top of each locule, style very slender, not expanded at base, about 2 mm long, 0.3-0.4 mm diam., stigma subglobose, 0.9-1 mm diam., slightly 4-lobed, with about 4 large oil glands; fruits subglobose, about 1.8-2.2 cm; borne on pedicels 8-12 X 1-2 mm, nearly filled with 3 or 4 ovoid, smooth, hard seeds, 10-12 X 8-10 X 6-8 mm, with an even-margined hilum 2 X 1 mm, pulp-vesicles almost suppressed the seeds in most fruits.
      This species is the type of the subgenus Afrocitrus, which includes also C. zenkeri and C. le-testui.   These three species are evidently related to each other and clearly delimited from the six typical species in the subgenus Citropsis.
      Citropsis gabunensis has very small flowers, the smallest of any species of the genus.   No other species of Citropsis shows as much variation in the number and size of the leaflets.   The type material from Sibange Farm, Gabon (and also abundant material collected by Père Klaine near Libreville, Gabon, now in the Paris herbarium), shows whole branches with very large unifoliolate leaves, although more often some of the leaves have two or three leaflets.   The typical 5-foliolate form is also common at both places; it often shows a short winged rachis segment bearing the terminal leaflet, something found only in trifoliolate leaves of other species of Citropsis.   The large unifoliolate leaves look astonishingly like those of Citrus in every character—size, shape, venation, texture, and color; they even have winged petioles very like those of some forms of Citrus aurantium and C. grandis.
      The unifoliolate species of Pleiospermium, P. sumatranum, P. longisepalum, and P. latialatum, also have leaves very like those of Citrus and closely resembling those found on this form of Citropsis gabunensis.   Doubtless both Citropsis and Pleiospermium contain species that resemble closely remote ancestral forms from which Citrus has evolved.
      7a.   Citropsis gabunensis var. lacourtiana (De Wild.) Swing. & M. Kell. Jour. Arnold Arbor. 21:124. 1940. Limonia lacourtiana De Wild. Ann. Mus. Congo Bot. 5 sér. 1:159. 1904. Illus. De Wildeman, loc. cit. pl. 50.
      Type.—Bombay, Sankuru (Lualaba-Kasai), south-central Belgian Congo (L. Gentil, No. 1), Herb. Jard. Bot. l'Etat, Brussels; photographs and serial microtome sections of type specimens, S. and T. Nos. 532-A, 533-A, 533-B, 533-C, and 533-D, 18 slides; also leafy twig and fruit of type collection, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Republic of the Congo: known only from the type locality in the Sankuru River valley.
      Common name.—Sankuru cherry-orange.
      Fruits subglobose, 1.8-2 cm diam., yellow-orange when ripe, peel 2-2.5 mm thick, rather soft, pulpy tender, juicy, and of very agreeable flavor, seeds few or none, plump, subglobose, bluntly conical at one end, 11 X 9-10 mm.
      The type specimen of Limonia lacourtiana was kindly lent to Swingle by Dr. W. Robyns, director of the Jardin Botanique de l'Etat at Brussels.   It consists of a single fruiting branch collected in May, 1902, with a number of half-grown and three nearly mature fruits, most of them seedless, but one (or two?) with a single yellowish-brown (Buckthorn brown, Ridgway, 1912, pl. 15), rather large, plump, short-ovoid seed, 11 by 8.5 to 9 mm, with a hard testa having a smooth-edged, ovate hilum 2 by 1 mm.   Gentil's original field label attached to the type sheet calls the plant "un mandarinier sauvage" and notes its "fruit délicieux."   The three to six fruits occur together in dense axillary clusters and are borne on pedicels 8 to 13 mm long, which in turn are densely crowded on a very short peduncle only 4 to 10 mm long.
      Three topotype specimens in the herbarium of the National Botanic Garden at Brussels collected nearly a year later by Em. and M. Laurent consist of two sterile branches with paired axillary spines, 2 to 2.5 cm long, and a fruiting branch from a small tree, 6 to 7 m high, looking extremely similar to the type specimen, with eight immature fruits, all but one of them seedless.
      Flowers of the variety are unknown, but a specimen in the Brussels herbarium from the same general Sankuru region, collected by A. Sapin in September, 1906, consists of two flowering twigs with axillary clusters containing three to six very small flower buds, about 5 by 3 mm, with slender pedicels, 0.4 to 0.6 mm diam.   These have pistils (immature?) which when restored by the modified Juel technique measured 4.4 mm long (including the disk); disk 0.6 to 0.7 mm high, 1 to 1.25 mm wide, cupulate, enclosing narrowed base of ovary for 0.2 to 0.25 mm; walls of the cup unusually thick (0.4 to 0.6 mm), containing six to ten medium-sized oil glands arranged approximately in a ring at a level even with or slightly below the bottom of the cup; ovary 3-locular, 1.3 mm long, 1.1 mm wide at top, 0.5 mm wide at bottom, with a medium-sized oil gland at the top of each locule; style 2 mm long, 0.5 to 0.6 mm wide; stigma 1 mm long, 1.2 mm wide, with large oil glands between the stylar canals.   The anthers have a single large oil gland near the tip of the connective.   These measurements and the disposition of the oil glands in the ovary, stigma, and anthers agree very closely with those found in the flower buds of a specimen of typical C. gabunensis preserved in the herbarium of the Museum of Natural History at Paris, collected by Père Klaine at Libreville, Gabon.   However, Sapin’s material from Sankuru has glabrous staminal filaments, whereas those of the species have scattered, slender hairs on the side facing the ovary, as do those of C. le-testui.   If this absence of pubescence on the filaments of the variety is found to be a constant character, it will be useful in separating it from the species at seasons when no mature fruits are to be found.
      The two species of Citropsis that have and fruited in the United States, C. schweinfurthii and C. gilletiana, both produce an abundance of small orange-like fruits, but they are not edible, as the pulp-vesicles contain numerous granules of a waxy substance of disagreeable flavor.   Doubtless C. gabunensis, belonging as it does to a very different group (the type of the subgenus Afrocitrus discussed above), does not produce this ill-flavored wax.   The species itself has fruits so full of plump seeds as to leave very little space for pulp.   The variety lacourtiana, on the contrary, has seedless fruits that are filled with high-flavored pulp.
      8.   Citropsis zenkeri Engler, Die Pflanzenwelt Afrikas 3(Heft 1):760. 1915. Illus. Engler, loc. cit. p. 759, fig. 354,A-E; Engler, Die Nat. Pflanzenfam. 19a:348, fig. 158;A-F. 1931.
      Type.—Bipindi, southern Cameroons (Lat. 3° 20' N., Long. 11° 45' E.).   Herb. Bot. Mus., Dahlem-Berlin.
      Distribution.—Known only from the type locality.
      Common name.—Zenker's cherry-orange.
      Fruiting twigs with slender, acute, solitary, axillary spines, 1.5-2.5 cm long and 2-2.5 mm diam. at base; leaves 5-foliolate, 25 X 23 cm, leaflets 11-13 X 5-5.8 cm, ovate-acuminate or short-caudate, acumen 15-20 X 4-1 mm, acute at apex, base broadly cuneate, with very short (1-2.5 mm) petiolules, petiole 5.5-6 cm long, narrowly winged, 3-4 mm wide on either side of the midrib, rachis segments about 5 cm long, narrowly winged, total width at broadest place, 8-9 mm; inflorescences short, axillary clusters of numerous flowers with pedicels 3-4 X 0.5-0.7 mm, borne in dense clusters (5-8?) on short, unbranched, glabrous peduncles, 3-4 mm long; flowers medium-sized, 15-18 mm broad when open, 4-merous, calyx lobes deltoid, bluntly pointed, roughened on the back by numerous oil glands, petals tapering regularly to an acute apex, 8-10 mm long, with numerous oil glands, stamens 5-6 mm long, anthers 3-3.5 mm long, pistil 7.5-8 mm long (including short, narrow disk), ovary obovoid, 4-locular, 2-3 mm long, 0.6-0.8 mm wide at base but expanded at apex to 1.5-2 mm in width by 4 large protuberant oil glands (1 at the top of each locate), style cylindrical, not expanded at junction with the tip of the ovule, 3.5-4 mm long (including the stigma), 0.4-0.5 mm wide, stigma subglobose, 1.5 mm diam., obscurely 4-lobed, swollen with 4 large oil glands.
      This description is based partly on the short original description but mostly on the excellent illustrations that accompany it (Engler, 1915), which are made more valuable by comparable figures of five other species of Citropsis.
      The Zenker's cherry-orange is evidently closely related to C. gabunensis and to C. le-testui, but has even larger, more protuberant oil glands at the top of the ovary above each locule.   The fruits are unknown, but should be collected in the hope of finding an edible form with few seeds like C. gabunensis var. lacourtiana.
      9.   Citropsis le-testui Pellegrin, Bul. Mus. Hist. Nat. Paris 27:446. 1921. Illus. Pellegrin, Mém. Soc. Linn. Normandie 26:42, fig. 3. 1924.
      Type.—Ndougou in Ngouyé Valley, French Congo (Lat. 2°-3° S., Long. 10° 30' E.) (Le Testu, No. 2286).   Herb. Mus. Hist. Nat., Paris; photographs of type, also 2 leaflets and 3 flower buds of type material with serial microtome sections of 1 bud, S. and T. Nos. 708-A (9 slides, cross sections), 708-B (3 slides, longitudinal sections), in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known only from the type locality.
      Common name.—Le Testu's cherry-orange.
      The original description, translated (and slightly amended from the equally detailed French redescription published (3 years later, in 1924) reads: "A small tree, 3 meters high, branches gray-green, glabrous, slender, more or less spiny.   Spines 1 or 2, erect, axillary, 2-3 cm long.   Leaves usually 3-foliolate, sometimes 5-foliolate ([the two pairs] separated 3-6 cm).   Petiole nearly terete, glabrous, 6-8 cm long (or in 5-foliolate leaves, 12-15 cm long), articulated with the rachis segment.   Leaflets ovate, 6-12 cm long, 3.5-7.5 cm wide, apex attenuate-acuminate, acumen obtuse, 1-2 cm long, abruptly contracted at base into a petiolule 3-4 mm long, glabrous, rigid, membranaceous, glandulose-punctate, midrib prominent below, depressed above, 6 primary lateral veins on each side, ascending and anastomosing far from the margin then curving to meet it, the smaller veinlets reticulate, slender, prominent below in dried specimens.   Inflorescences short, axillary panicles, very shortly tomentose, peduncles 0.8 mm long, pedicels 5 mm long.   Bracts ovate, acute, 1 mm long.   Calyx thick, glandulose, shortly-villose without, 4-lobed, lobes deltoid, acute, 2 mm long with scarious margins.   Petals 4, glandulose, glabrous, oblong, concave, apex subangulate, 7 mm long, 3.5 mm wide.   Stamens 8, filaments linear-lanceolate, flattened, shortly-villose, 4 mm long, anthers introrse, glabrescent, 4 mm long, oblong-saggitate [sic], apex acute and long-acuminate.   Disk prismatic, apex somewhat 4-lobed, 1 mm high, glandulose, glabrous.   Ovary glandulose, glabrous, obpyramidal, somewhat 4-lobed at the apex, 2 mm high, 4-locular, locules 1-ovulate; style terete, glabrous, 2-3 mm high, stigma capitate, somewhat 4-lobed.   Fruit subglobose, slightly flattened at the apex, 1.5 cm high, 2 cm broad."
      Pellegrin in his comments following the description stated that the flowers are white, the leaves usually 3-foliolate.   Citropsis le-testui is a small tree, 3 m high, relatively abundant on the slopes of the hills, and never seen growing in wet situations; the fruit is yellow when ripe.
      This remarkable species differs from all others of the genus Citropsis in having completely wingless petioles and rachis segments, which gives the leaves a curious aspect.   As noted by Pellegrin, the leaves of this species much resemble those of Balsamocitrus dawei, a very different plant of the subtribe Balsamocitrinae, not at all closely related to Citropsis.
      The figure of C. le-testui cited above shows the stamen to be 8 mm long, the filament flattened, about 4 mm long, 0.75 mm wide, 0.5 mm thick at the widest part, and short-pubescent on the inner side.   The anther is shown to be about 3 to 3.2 mm long, 1 mm wide, and 0.75 mm thick, tapering to a sharp point at the apex.
      Fortunately, a flower bud ready to open was available for study from the type specimen.   After being restored by the modified Juel technique and cut into serial microtome sections, the following characters of the flower parts were observed:
      Nearly mature flower buds 9-10 X 4-5 mm; calyx lobes 3 (4?), 1 mm wide, 1.5 mm long, subacute, much thickened by numerous oil glands which are larger and more closely juxtaposed in the free lobes; corolla 3- (4-?) merous, petals linear, 9-10 X 5 mm, with sparsely scattered, medium-sized oil glands; stamens 8, about 7 mm long, filaments free, 4-4.5 mm long, subclavate, about 1 mm wide, thickest near the apex (0.4-0.5 mm), with many slender hyaline, scattered hairs on the ventral surface (facing the ovary), anthers 3-3.5 mm long, 0.6-0.8 mm wide, with a large, elliptical, acute-pointed oil gland at the apex, 1 mm long, 0.3-0.4 mm wide; pistil (in nearly mature bud) 7 mm long (including the disk), disk 1-1.2 mm long, cupulate and enclosing base of the ovary for about 0.4 mm, without noticeable oil glands; ovary 3- (4-?) locular, obconical, 2.6-3 mm long, 0.5-0.6 mm wide below, at junction with disk (at bottom of cup), tapering regularly upward, 2 mm wide at flattened top, which has 1 large, ovoid, somewhat protuberant oil gland (0.6-0.8 mm diam.) over each locule and numerous smaller oil glands between them; style arising abruptly, narrowed and slightly countersunk in the center of the flattish upper face of the ovary, 3-3.2 mm long, 0.5 wide at base, gradually expanding to 0.8 mm wide at junction with the stigma, sparsely covered with small, scattered, superficial oil glands; stigma subglobose, 1.2 mm high, 1.5-1.6 mm wide, with several (4-5?) large oil glands, 0.9-1 mm vertical diam. and 0.3-0.6 mm transverse diam., sometimes smaller (through pairing).
      Thanks to this authentic type material, it is possible to assert definitely that the pistil of C. le-testui is very like those of C. gabunensis and C. zenkeri, and in general is intermediate between them in its characters.   The flat-sided ovary and disk described and figured by Pellegrin (l.c.) were doubtless distorted by shrinking suffered during desiccation.

SPECIES INADEQUATELY KNOWN AND OF UNCERTAIN RELATIONSHIPS
      10.   Citropsis tanakae Swing. & M. Kell. Jour. Arnold Arbor. 21:121. 1940. Illus. Tanaka, Kankitsu No Kenkyû (Citrus Studies) 78, fig. 71, left (as Citropsis citrifolia, a nomen nudum). 1933; Swingle & M. Kellerman, loc. cit. pl. 3, figs. 1-5.
      Type.—Africa, Sierra Leone (Afzelius).   Herb. Univ. Uppsala, Sweden; photographs and serial microtome sections, S. and T. No. 226-A, 3 slides, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known only from the type locality.
      Common name.—Sierra Leone cherry-orange.
      Ultimate branchlets slender, 1.5-3 mm diam., with solitary, axillary, slender, short spines, 2-10 mm long; internodes 2-3 (or 4?) cm long; leaves simple! broadly lanceolate, tapering gradually, or slightly acuminate, into a short, thick, blunt acumen, 8-10 X 6-3 (or 4) mm, base broadly cuneate, with the margins slightly decurrent into upper part of petiole, lateral veins numerous, 10-12 on each side, arising at angles of about 75°-80° with the midrib, margins regularly but shallowly crenulate on upper half, subentire below; petioles very short (3.5-4 mm), wingless, slender (1-1.5 mm wide), glabrous, not articulated with the leaf blade; inflorescences very short, few-flowered, axillary; flower buds small (immature?), about 8 X 3 mm, borne on short, slender pedicels, 2-3 mm (or more?) long, 0.5-0.75 mm diam.; calyx roughened with numerous oil glands, glabrous, calyx lobes 4, short (0-3-0.4 mm), with thin scarious edges, with a single rather large oil gland near the subacute tip, petals 4, about 7 X 3 mm, glabrous, with many medium-sized oil glands, more abundant at the pointed tips, stamens 8, about 5.5-6 mm long, staminal filaments glabrous, somewhat flattened radially, broad at the base where they cohere in groups of 2 or 3 for some distance (about 2.8 mm), anthers about 3 mm long with a single small oil gland near the top of the connective; disk cylindrical, glabrous, short, shallow cup-shaped, fitting rather closely over the base of the ovary for about 0.1 mm, with a few small oil glands in walls of cup, pistil 5.6 mm long (including disk), ovary 4-locular, ovoid, 0.93-1 mm long, 0.8-1 mm wide, rounded at top, without a large oil gland over each locule, style long and slender (4.6 mm long, 0.3-0.4 mm wide), contracted where it joins the ovary in which it is very slightly countersunk, slightly expanded at the apex where it merges into depressed globose stigma, 0.7 mm high and 0.9 mm wide, containing several (5-7) large oil glands.
      This species is known only from a single twig, about 20 cm long, with seven leaves and a single flower bud, collected in Sierra Leone between 1794 and 1796 by the Swedish botanist Afzelius, who identified it as Citrus medica, doubtless because the petioles were not articulated with the leaf.
      This unique flower bud and a single leaf were kindly sent to Washington by the custodian of the herbarium of the Botanical Museum of the University of Uppsala, Sweden.   The flower bud was cut into 454 serial cross sections, each 20 mu thick, by a modification of the method published by Juel (1918, p. 14) while he was professor of botany at the University of Uppsala.   This very old material could not be restored as well as usual, but nevertheless it was possible to work out the flower structure in minute detail (see Swingle, 1940c, pl. 3, figs. 1-5) and to draw up the preceding description, which proves beyond doubt that this plant is not a species of Citrus but a new species of Citropsis.
      This remarkable species is unique in the genus because of its simple leaves with very short wingless petioles (only one-twentieth as long as the leaf blade and not articulated with it).   All other species of Citropsis have odd-pinnate leaves, trifoliolate to 7-foliolate or occasionally (in C. gabunensis) unifoliolate leaves, with petioles that are, however, very evidently winged and never less than one-eighth as long as the leaf blade.   In C. le-testui the petioles (and rachis segments) are wingless, but are very many times longer.
      The leaves of C. tanakae are very like the leaflets of the other species of Citropsis, which always have very short, wingless, nonarticulated petiolules.   Citropsis tanakae has flowers somewhat resembling those of the species in the subgenus Afrocitrus in having a subglobose stigma distended with large oil glands borne on a very slender style, but the ovary lacks the large oil glands, one at the top of each locule, such as characterize the species of this subgenus.
      Besides being one of the most distinct species of the genus, C. tanakae is also the northernmost species in Africa, occurring some 500 to 600 kilometers northwest of C. mirabilis, a very different species, native to the Ivory Coast in West Africa.
      This beautiful Sierra Leone cherry-orange has been named in honor of Professor Tyôzaburô Tanaka, who, as soon as he saw the type specimen in the Uppsala herbarium, recognized that it belongs, not to the genus Citrus, as Afzelius supposed, but to Citropsis.
      11.   Citropsis daweana Swing. & M. Kell. Jour. Arnold Arbor. 21:123. 1940. Illus. Tanaka, Kankitsu No Kenkyû (Citrus Studies) 77, fig. 70, right (sub nomen nudum, Hesperethusa villosa). 1933; Swingle & M. Kellerman, loc. cit. pl. 3, fig. 6; fig. 3-28 this work.
      Type.—Madanda Forest, Portuguese East Africa (Dawe, No. 443).   Herb. Brit. Mus., London; photograph and fragments in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known only from the type locality (now Mozambique).
      Common name.—Mozambique cherry-orange.
      A shrub or small tree, 3 -5 m high, branchlets at first angular, soon cylindrical, 2 -4 mm diam; internodes 2-3 cm long; spines axillary, solitary, short, straight, 1-2.5 cm long, 2-4 mm diam. at base, shorter ones blunt-pointed, the longer ones (1.5-2.5 cm) sharp-pointed; leaves 5-7-foliolate; leaflets elliptical to rhomboid or ovate, bluntly rounded at the apex, terminal one narrowly cuneate at base, 5-5.8 X 2-2.4 cm, lateral ones broadly cuneate or broadly rounded at base, 2-4.5 X 1.5-2.4, with 4-6 rather inconspicuous veins, sparsely and evenly dotted with small translucent oil glands, sparingly pubescent on both sides and on the margins with very slender, colorless ciliate hairs, margins finely and shallowly crenate or subentire, petiolules very short (1-1.8 mm); petioles nearly or quite wingless.   1.2-2.2 cm long, 1-1.5 mm wide; first rachis segments 1.5-2 cm long, with spathulate wings, 3-6 mm wide and rounded at the distal end, usually narrowed to a subacuminate base at the proximal end, second rachis segments elliptical, broader (about 2.5 X 0.8 cm), not spathulate; flowers and fruits unknown.
      This remarkable species is known only from the type material collected by M. T. Dawe (when director of agriculture of the Companhia de Moçambique) in the Madanda Forest, a rubber-producing region, which lies 100 to 200 kilometers to the southwest of Beira, between the Lucite River (Lat. 20° S.) and the Save River (Lat. 21° 30' S.), to the west of the Sofala lands (about Long. 34° E.) and to the east of the Mossurize district (about Long. 33° E.).
      Unfortunately the flowers and fruits of this species are unknown; however, the leaf characters are so distinctive that it cannot be mistaken for any other Citropsis.   The leaves, including the petioles and rachis segments, are sparingly pubescent on both surfaces, whereas in all the other species they are glabrous; the leaflets are the smallest known in the genus.   The petiole-rachis wings become progressively wider toward the tip of the leaf, that is, the petiole is wingless or nearly so, the first rachis segment narrowly winged (wider at the tip), and the second segment still wider.   This is just the reverse of what is found in the other winged species of Citropsis.   The scanty type material seems to show Kurztrieben, like the leaf and flower spurs of Poncirus trifoliata and some of the Hard-Shelled Citroid Fruit Trees (subtribe Balsamocitrinae).
      The leaf, petiole, and rachis characters of this species show a great similarity to those of certain forms of Hesperethusa crenulata from peninsular India.   However, as noted above, certain forms of Citropsis gabunensis have leaves and petioles greatly resembling those of Citrus but have flowers and fruits unmistakably belonging to Citropsis.   Only the study of flowers and fruits can reveal the exact relationships of C. daweana.
      The Mozambique cherry-orange has the appearance of being a somewhat xerophytic plant; the other species are evidently mesophytes and usually grow in tropical rain forests.
      This is without question the least known species of the genus Citropsis and at the same time the most distinct one!   When flowers and fruits are available for study it will very probably be found to constitute a new subgenus, if not a new genus.   It is of great phylogenetic interest as a measure of the nature and rate of evolution that has occurred since Citropsis was cut off from any connection with its relatives in southeastern Asia, eastern India, and the Monsoon region.

XX.   Atalantia Corrêa
      XX.   Atalantia Corrêa, Ann. Mus. Hist. Nat. Paris 6:383. 1805. Rissoa Arn. Nov. Act Acad. Caes. Leop.-Carol 18:324. 1836; Malnarega Raf. Sylv. Tell. Mant. 143. 1838.
      Type species.Limonia monophylla Roxb. non (L.) = Atalantia monophylla DC.
      Distribution.—Southwestern Asia: India, Ceylon, Burma, Cambodia, Laos, North Vietnam, South Vietnam, southern China; also Sumatra, Banka (?), Java (?).
      Common name.—Atalantia.
      Small trees or shrubs, twigs more or less angled when young, soon cylindrical, glabrous in some species, more or less hairy when young, with single, stout, sharp spines in the axils of the leaves, or spineless, especially fruiting branches of old trees (some species [A. rotundifolia] nearly spineless); leaves 1-foliolate (or rarely simple), more or less coriaceous, strongly veined, and usually with conspicuous reticulations between the lateral veins; petioles wingless, often pubescent when young, articulated with the leaf blade, short, often less than 1/10 as long as the leaf blade; flowers small, axillary or rarely terminal, fascicled, or in racemose corymbs or panicles; calyx 3-5-lobed, or forming a continuous envelope over the corolla, stamens, and pistil, and splitting more or less irregularly into 2 or 3 parts as the flower expands; petals imbricate; stamens 6-10, free, or more or less connate into a staminal tube which is sometimes a closed cylinder with the anthers attached to the rim; anthers short, broadly ovoid; disk annular, short; ovary ovate, or subglobose, with 3-5 locules, with 1 or 2 ovules in each locule; fruits small, subglobose, with numerous broad-based, sessile, slender, tapering pulp-vesicles filling all the space in the segments left by the seeds; peel thin, greenish-yellow on ripe fruit, dotted with oil glands; seeds oblong, green within, sometime polyembryonic.
      The genus Atalantia contains small trees somewhat resembling Citrus in general aspect, bearing fragrant white flowers and globose fruits with the appearance of diminutive greenish-yellow oranges.   The pulp-vesicles are, however, different from those of Citrus in being sessile instead of stalked.   The leaves of Atalantia, although almost always unifoliolate like those of Citrus, are very different in having much more prominent, more numerous lateral veins, and veinlets forming reticulations between the lateral veins.
      Besides the typical species of Atalantia, which have numerous sessile, conical pulp-vesicles, there are at least two species which belong to the subgenus Rissoa, based on the type species A. ceylanica.   This species has very large seeds which almost completely fill the locules of the fruit, leaving very little space for pulp-vesicles.   Atalantia rotundifolia is closely related to A. ceylanica and is perhaps a spineless brachytic mutation of it.   A third species, A. guillaumini, somewhat resembles A. ceylanica and like it has large seeds almost completely filling the locules of the fruit.   However, so far as could be determined from the scanty fruits examined, no pulp-vesicles are present.   It may be found that there has been an evolutionary tendency of the species of the subgenus Rissoa to lose the pulp-vesicles, resulting in the development of completely pulpless species obviously closely related to Atalantia in spite of lacking one of the most important characters of the genus and even of the subtribe Citrinae.
      It has in the past been supposed that the genus Atalantia was founded by Corrêa da Serra to include Limonia monophylla L., now believed to be a synonym of A. ceylanica.   Corrêa cited no species under his new genus Atalantia but almost certainly had in mind the "Limonia monophylla L." described and figured by Roxburgh (1795, vol. 1, pp.59-60, pl. 83), since he spoke in detail of "singular monadelphy" of the stamens with the filaments fused into a complete cylinder, just as had been figured by Roxburgh ten years previously.   That he was familiar with Roxburgh's great work on Plants of the Coast of Coromandel is shown by the fact that immediately after discussing the genus Atalantia he mentioned Limonia arborea and L. pentaphylla of Roxburgh (two species on which Corrêa based his new genus Glycosmis)and cited Roxburgh's plates of them which follow immediately after the plate of Limonia monophylla.
      In view of these facts, it appears that the genus Atalantia is, in fact, based on Roxburgh's description and illustration of a plant which he thought was Limonia monophylla L., but which is now known to be another species, very probably Atalantia monophylla DC.   Keys to the subgenera and species of Atalantia are presented.

Subgenus Atalantia
      Subgenus Euatalantia Swing. in Bailey, Stand. Cycl. Hort. 1:426. 1914.
      Type.Atalantia monophylla DC.
      Stamens connate, forming a cylindric tube, or connate in groups, rarely free; fruits with numerous well-developed, juicy pulp-vesicles that fill all the space in the fruit not occupied by the few medium-sized seeds; stipuloid paraphylls absent.
      Seven of the species of Atalantia fall in this subgenus.   In general they much resemble the species of the subgenus Rissoa.
      1.   Atalantia monophylla DC. Prodr. 1:535. 1824. Limonia monophylla Roxb. (non L.) Pl. Corom. 1:59. 1795; (?) Turraea virens Hellen. (non L.) Act. Holm. 9:308. 1788; (?) T. spinosa Willd., in L., Sp. Pl. ed. 4/5. 2:554. 1899 (= T. virens Hellen.); Atalantia spinosa (Willd.) Tan. Jour. Bot. Brit. & For. 68:232. 1930 (name preoccupied!); (?) Mal-naregam, Rheede, Hort. Malab. 4:27. 1679; (?) Malnarega malabarica Raf. Sylv. Tell. Mant. 143. (1838); (?) Atalantia malabarica (Raf.) Tan. Jour. Indian Bot. Soc. 16:233. 1937; Atalantia floribunda Wt. Icon. Pl. Ind. Or. 4:16. 1850. Illus. (?) Rheede, loc. cit. pl. 12; (?) Hellenius, loc. cit. pl. 1, fig. 1; Wight, loc. cit. pl. 1611. 1874; Beddome, Fl. Sylv. Anal. Gen. pl. 7, fig. 5. 1871; Engler, Die Nat. Pflanzenfam. 3(pt. 4):fig. 111,C-D. 1896; ibid. 19a:fig. 150,C-D. 1931; fig. 3-29 this work.
      Type.—British India, Coromandel (Roxburgh).   Brit. Mus.?
      Distribution.—Common in India (except in Himalayan region and in Bombay State), Ceylon, Burma, Thailand, Cambodia, Laos, North Vietnam, South Vietnam.
      Common name.—Indian atalantia.
      A small, much-branched tree with rounded twigs usually slightly hairy at first, becoming glabrous, with single, stout, sharp spines, 10-15 or 20 mm long, or unarmed; leaves bright green above, paler and reticulate-veined below, ovate-lanceolate or elliptical, variable in size, 3-15 X 2-4 cm (usually 4-8 or 10 X 2.5-3.5 cm), margins entire or slightly undulate, apices obtusely rounded, often emarginate, bases broadly cuneate, articulated with the wingless petioles that are 5-10 mm long and 1.5-2 mm wide, flattened above, veins strongly marked on the under surface, 10-14 on each side, branched, forming reticulate veinlets and also forked near the margins and anastomosing; flowers long-pediceled on short racemes (or clustered) in the axils of the leaves, pedicels 6-15 mm long, finely pubescent or glabrous, merging into the calyx; calyx more or less irregularly and deeply 2-cleft, and glabrous or finely pubescent; corolla 4-5-merous, petals white, glabrous, bluntly rounded at apex, 8-10 mm long; stamens 8-10, united into a tube 6-8 mm long with the anthers borne on the free tips of the filaments; pistil slender, 6-7 X 1 mm, ending in an abruptly expanded, 3-4-lobed stigma, 1.2-1.5 mm diam.; style caducous; fruits globose, 1.5-2 cm diam., yellowish-green when ripe, locules filled with numerous cuneate, sessile pulp-vesicles which are broad-based and tapering to a point at their free end, toward the center of fruit; seeds usually only 1 to a locule; often some of the locules do not develop seeds.
      The most remarkable character of this species (found also in the closely related species A. macrophylla) is its very curious calyx.   This is a hollow, balloon-like envelope, not articulated with the pedicel and without any sign of sepals.   Soon this undifferentiated envelope splits more or less irregularly into 2 (rarely 3) lobes, and exposes the corolla, stamens, and pistil to view.
      The nomenclature of this species is in a confused state.   Limonia monophylla of Linnaeus has been assumed by almost all taxonomists to be this species.   However, Trimen (1893, p. 227) listed L. monophylla L. as a synonym of Atalantia ceylanica, and stated that "the figure of Burmann [1737, pl. 65, fig. l]…appears to represent this rather than A. monophylla…Hermann's specimen…is certainly A. zeylanica [ceylanica]."   Airy-Shaw (1939, p. 291) again examined Hermann’s specimen (which is in fact the type specimen of Limonia monophylla L.) in the British Museum and saw no reason to doubt Trimen's findings.   Airy-Shaw then showed that A. monophylla was based by De Candolle expressly on "Limonia monophylla Linn." of Roxburgh (1795, vol. 1, pp. 59-60, pl. 83) and that L. monophylla L. was not cited.   However, the plant described and figured by Roxburgh was certainly not A. ceylanica and hence was not Limonia monophylla L.   Airy-Shaw assumed that Roxburgh's plant from the Coromandel Coast was, in fact, A. monophylla as understood by Hooker (1875, p. 511), and accepted the name A. monophylla DC. for it.   However, Roxburgh described and figured a plant having a "four-five parted" calyx, a character not found in A. monophylla as understood by Hooker.   That species has, instead of a normal calyx, a bladder-like continuous membrane covering the flower bud which, as the bud expands, splits irregularly.   Possibly both Roxburgh and the artist who made the beautiful colored plate of the Coromandel plant failed to see the irregularly ruptured calyx and, without careful study, assumed that it was 4- or 5-parted as in most of the other species of Atalantia and other related genera.   Roxburgh described the leaves as "end-nicked" and "from two to three inches long, and one or one and a half broad"; nevertheless he published a plate showing a flowering branch with 23 leaves, none of them emarginate at the apex, measuring 1 7/8 to 2 9/16 inches long, and 5/8 to 1 1/16 inches wide, only one leaf slightly exceeding the minimum width given in the description and more than three-fourths of them narrower than 1 inch!   In view of these facts, no great importance should be attached to the fact that neither Roxburgh nor the artist noticed at that time the abnormal splitting of the calyx, something no botanist had noticed in any species of Atalantia or in any other plant belonging to the orange subfamily.
      For the present, it seems best to assume that Roxburgh was describing and figuring the true A. monophylla of Hooker and that A. monophylla DC. is the oldest valid name for it.
      Atalantia monophylla has very hard wood that has been recommended as a substitute for boxwood (Gamble, 1915).   The fruits are said to yield a warm oil used in treating chronic rheumatism by native doctors.   One Indian author (Nadkani) stated that the fruits "make a nice pickle" (Lewis, 1934, p. 80).
      When watered frequently, this species makes a luxuriant growth in the greenhouse (see fig. 3-29).   It should be tested as a hedge plant in warm subtropical climates and in dooryard gardens as an ornamental, since it has dense deep-green evergreen foliage and bears a great profusion of fragrant flowers and many small globose yellow fruits.   Atalantia monophylla can be grafted on Citrus, and vice versa.
      2.   Atalantia macrophylla (Oliv.) Kurz. Jour. Asiat. Soc. Bengal 44(2):136. 1875; idem, For. Fl. Burma 1:195. 1877. Atalantia monophylla var. macrophylla Oliv. Jour. Linn. Soc. Bot. 5(Suppl. 2):24. 1861.
      Type.—Tenasserim (Helfer).   Herb. Bot. Gard., Calcutta (?).
      Distribution.—Southern Burma: Tenasserim; Andaman Islands; also Bangka Island and Java (fide Engler).
      Common name.—Giant Andaman atalantia.
      This species was at first considered to be a subspecies of A. monophylla, since it has, like that species, the calyx irregularly split, usually to the base on one side; it differs, however, in having 2 ovules in each locule of the ovary (A. monophylla usually has only 1), and in having larger and broader leaves (5-15 X 3.5-6.5 cm).   Kurz (1877, vol. 1, p. 195), who raised A. macrophylla to specific rank, found that it had fruits much larger than those of A. monophylla; he stated that they are "the size of a wood apple" [which measures about 3.5-5 cm diam.].   The pedicels of the short racemes are long (6-15 mm), as in A. monophylla, but are very stout, stiff, and compressed.   Atalantia macrophylla is a small tree, 8-10 m in height, occurring in the Andaman Islands and Tenasserim.   Kurz described it as glabrous in all parts and almost spineless.
      Unfortunately, this very interesting Atalantia has been very inadequately studied and has never been figured.   Kurz showed clearly that the two species agree in possessing this remarkable character by putting both A. monophylla and A. macrophylla into the first division of his key to the species of Atalantia, which reads: "Calyx irregularly lobed, split to the base on one side."   All other Aurantioideae have flower buds with a more or less regularly (usually 3 to 5-) lobed calyx.   This character, shared in common by the two species, is an almost certain indication of their very close relationship.
      Parkinson reported (1923, p. 108) that the trunk of this species measures 2 to 3 feet (61 to 91 cm) in girth at breast height (19.3 to 29.1 cm diam.), an unusual girth for an Atalantia.   This Andaman Island form (which Parkinson called A. monophylla) has globose fruits 2.5 cm in diameter; it is common everywhere in deciduous forests.
      Engler (1931, p. 328) recognized A. macrophylla as a good species distinct from A. monophylla.   Hochreutiner (1904, pp. 50-51) considered the form from Bangka Island as a variety of A. monophylla differing "in having somewhat larger leaves, less hairy twigs and especially in having a larger calyx with convex acute lobes; in the typical A. monophylla the calyx lobes are smaller and truncate-scarious."   A specimen examined in the Botanic Garden at Buitenzorg (now Bogor), Java, was a round-topped tree, 12 m high, with a deeply furrowed trunk 1.8 m in circumference (59 cm diam.), forking at 1 m from the ground and branching profusely at 2 to 3 m from the ground.   No information is given by Hochreutiner concerning the size of the fruits or the number of ovules in each locule of the ovary.
      Atalantia macrophylla is the only species of the subgenus Atalantia that is found south of the Asiatic mainland; it seems to have spread westward to the Andamans and southward to Bangka from extreme southern Burma, retaining its large size and exceptional girth, but showing diminished (although still large) fruit size.   The typical form of the species, growing in Tenasserim, has the largest fruits of any of the Near-Citrus Fruit Trees (included in the genera Citropsis and Atalantia).
      3.   Atalantia racemosa Wight,15 Hook. Jour. Bot. 1:64, pl. 122. Jan., 1834; A. racemosa Wt. & Arn. Prodr. 1:91. Apr., 1834, nomen semi-nudum. Sclerostylis racemosa Wt. 1838?; S. parvifolia Wt. 1838?; Lampetia racemosa (Wt.) Roem. Syn. Hesper. 1:42. 1846. Illus. Wight, loc. cit. pl. 122; idem, Ill. Ind. Bot. 1:pl. 71. 1838?; idem, Icon. Pl. Ind. Or. 1:pl. 71. 1840; Beddome, Fl. Sylv. Anal. Gen. xlvii, pl. 7. 1871; Talbot, For. Fl. Bomb. Sind. 1:p. 202, fig. 123. 1909; Swingle, Jour. Arnold Arbor. 21:133, pl. 4, fig. 5. 1940.
      Type.—Southwestern British India, "Madera…Alpine country" (Wight).   Herb. Univ. Glasgow.
      Distribution.—Southwestern India: Konkan Coast to Travancore; Ceylon.   It is much more common in the Bombay Presidency than A. monophylla, which is there rare (fide Cooke, 1903, p. 187).
      Common name.—Bombay atalantia.
      This species was described by Hooker (1875, vol. 1, p. 512) as follows: "Leaflet oblong-elliptic or ovate-oblong, tip obtuse or 2-lobed, flowers racemed, rarely fascicled, racemes pubescent or glabrous, calyx 4-lobed, lobes acute or obtuse, filaments more or less connate, ovary 2-3 celled, ovules 2 collateral.…A small tree, armed or unarmed, very similar to A. monophylla, but the flowers, instead of being long-pediceled in short racemes or fascicles, are shortly-pediceled on racemes sometimes 3 in. long, but at others almost reduced to corymbs; the leaves are usually larger, and often broader, though quite small in Wight's figure; the flowers are of the same size, but often very crowded and the buds very globose, they are usually 4- but sometimes 3-5-merous, and the filaments are free or variously united; the ovary is sessile on a small disk.—The great distinction between these species is in the more or less regularly 4- (3-5-) lobed calyx of this, the lobes of which are acute or obtuse.   Fruit globose, 3/4 in. [19 mm] diam.   One of Wight's specimens has ovate leaves, cordate at the base.   Wallich’s 6358-E, from Penang (G. Porter), in fruit only, with leaves cordate at the base, may be this or the following [A. ceylanica], or something different from both."
      This species, like A. monophylla, has the stamens fused into a tube.   In A. racemosa, however, the stamens are often fused all the way to the tips into a cup-like cylinder surrounding the ovary and the anthers are attached directly to the edge of the cup, whereas in A. monophylla the tips of the stamens are free and only the bases (about two-thirds to three-fourths of their length) are fused together into a tube which is somewhat irregular because the free tips vary in length.   The flower buds of A. racemosa are globose and the calyx differentiates into four or five (rarely three) lobes, whereas in A. monophylla the flower buds are oblong or ovate, pointed at the tips, and the calyx shows no division into sepals but splits more or less irregularly into two or three lips often to the base on one side as the bud opens.   The fruits are much like those of A. monophylla.   They are globose, about 18 to 20 mm in diameter, and show the three to five lobes of the persistent calyx.   The flesh is composed of pulp-vesicles.
      3a.   Atalantia racemosa var. henryi Swing. Jour. Arnold Arbor. 21:127. 1940. Illus. Swingle, loc. cit. pl. 4, figs. 1-4.
      Type.—China, Yunnan Province, Szemao, altitude 1,220 m (Henry, No. 12930).   in [sic] Herb. Arnold Arbor.; photographs and serial microtome slides, S. and T. Nos. 660-A, 660-B, 660-C, and 660-D (14 slide [sic] with 464 cross sections and 156 longitudinal sections).   in [sic] Herb. Natl. Arbor.   Paratype: Same collection, in Natl. Herb., sheet No. 459320.
      Distribution.—China: Szemao and Chi-li district in southern Yunnan Province near the Indo-Chinese border at altitude 1,100-1,800 m.
      Common name.—Henry's atalantia.
      A small tree, 4-7 m high, ultimate branches slender, 2-3 mm diam., soon terete, spineless; leaves glabrous, lanceolate, short-acuminate at apex, acumen 5-8 X 4-3 mm, tip blunt or even slightly emarginate, cuneate at base, lateral veins very numerous, 15-30 on each side, not all clearly marked, arising at angles of 50°-70° with the midrib, margins subentire, petioles 5-10 X 1.2-2 mm, glabrous, coriaceous, more or less wrinkled, with a deep narrow channel, 0.5-0.8 mm wide on upper side, articulated with the leaf blade; inflorescences single axillary racemes, 1-2 cm long, with 5-15 flowers borne on slender pedicels 3-8 mm long, peduncles, pedicels, and calyx pubescent, flower buds subglobose, 3-4 mm diam., calyx lobes rounded, 1.8-2 mm long, 2-3 mm wide, margins thin and ciliate, petals oblong, rounded at apex, stamens 8, filaments glabrous or very sparingly short ciliate, cohering irregularly in groups, sometimes almost to the tips, in other flowers free almost to base, anthers about 2 mm long, attached near the middle to the narrowed filament, connective bearing 1 medium-sized oil gland near tip; disk cup-shaped, 0.5 mm high, 2 mm wide, pistil 3.5-4 mm long, ovary ovate, 1.5 X 1.5 mm, 4-locular with 2 ovules in each locule, with 1 medium-sized oil gland at top of each locule, style short and thick, 2 mm long (including stigma), 0.7-0.9 mm wide, stigma not clearly distinguished from the style, about 0.9-1 mm wide, with 4 stylar canals with 2 medium-sized oil glands in each space between two stylar canals; fruits subglobose, 1.5-2 mm diam., borne sparingly (1-2?) on each raceme on pubescent pedicels 3-5 X 1.5-2 mm, with numerous sessile pulp-vesicles (scanty in ripe fruit from seed pressure?), seeds 1-3, ovoid-oblong, 12 X 7 mm, monoembryonic.
      This variety resembles the typical A. racemosa of southwestern peninsular India and Ceylon in its racemose flower clusters, but it has larger leaves, flowers with longer pedicels and usually with a 4-locular ovary (not 2- or 3-locular) and with the staminal filaments rarely completely connate into a long tube.
      It is remarkable that so far as is known neither this variety nor the species itself has been found in the vast region, some 2,700 kilometers across, that separates the Mekong River valley of southern Yunnan from the Western Ghats in Bombay State, India.   There are many instances of discontinuous distribution of species or of varieties of a single species belonging to the tribe Citreae but few are as striking as this one.
      4.   Atalantia citroides Pierre ex Guill. in Lecomte, Not. Syst. 1:178. 1910. Illus. Guillaumin, loc. cit. p. 176, fig. 8(2); p. 177, fig. 9(3, 4); idem, in Lecomte, Fl. Gén. Indo-Chine 1:pl. 24, figs. C and 5, 6. 1911; fig. 3-30 this work.
      Type.—French Indo-China, Cochin China, Mount Dinh (Pierre, No. 4011).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—South Vietnam: Cochin China; Cambodia.
      Common name.—Cochin China atalantia.
      Guillaumin's original description of this species reads, in translation, as follows: "Trees 5-10 m high, glabrous, spines straight, few, 5 mm long, in the axils of the leaves; branches grayish.   Petioles glabrous, 1 cm long.   Leaves coriaceous, ovate (5-9 X 2-4 cm), entire, attenuate at the base, attenuate and emarginate at the apex, primary and secondary veins prominent above and below, oil glands visible only by being pellucid [to transmitted light].   Flowers very glabrous, in the axils of the leaves, pedicels glabrous, 1 cm long, with bracts at the base; calyx cupulate, with 3 or 4 lobes, obtuse at the apex; petals 4, reflexed, ovate, attenuate at the apex; stamens 10, almost as long as the petals, more or less equal, connate for 2/3 of their length, anthers ovate, apiculate; disk annular at the base of the ovary; ovary ovoid-elongate, terminating in an articulated cylindric style, caducous at maturity, stigma capitate, locules 3-5, ovules 1-2 [in each locule], collateral.   Fruit very similar to a small orange, 2 cm diam., pulp made up of succulent vesicles, seeds ellipsoid, about 1 cm long."
      The specimens from Cambodia were said by Guillaumin to differ from those from Cochin China in having the young twigs, as well as the peduncles and pedicels of the flower clusters, slightly pubescent, as shown in his figure C, cited above.
      This species has the filaments of the stamens united into a cylindrical tube like A. monophylla and A. racemosa.   The tips of the filaments (about one-third of their length) are free and the anthers are oval and apiculate, not triangular, as in A. monophylla.   The petals are narrowed at the tip, not broadly rounded.   The inflorescences resemble those of A. monophylla, having rather long-pediceled flowers clustered on short inflorescences in the axils of the leaves.   The calyx is much like that of A. racemosa, with three to four sepals.   The fruits resemble small oranges about 20 mm in diameter, with a rough peel and the flesh formed of sessile, conical pulp-vesicles.   The seeds are ellipsoid, about 10 mm long.
      This species grows to be a medium-sized tree, 5 to 10 m high, and is probably larger and more vigorous than any other species of the subgenus Atalantia except A. macrophylla.
      Although this species looks much like A. monophylla and has flowers with similar staminal tubes, it differs greatly in having a calyx with distinct sepals, unlike the curious balloon-like calyx of A. monophylla and A. macrophylla that splits irregularly into two or three lobes as the flower bud expands.   Tanaka (1930a, p. 163) made this species a synonym of A. monophylla, overlooking this important difference.   Guillaumin, who studied A. monophylla, also widely distributed in Indo-China, considered it to be clearly distinct from A. citroides.
      5.   Atalantia wightii Tan. Bul. Soc. Bot. France 75:714. 1928.
      Type.—Nilgiri Hills, Madras, southern British India (Bentham, No. 1531).   Herb. Kew.
      Distribution.—India; northern Canara, Bombay; Mysore and Madras (fide Tanaka, 1937, p. 234).
      Common name.—Nilgiri atalantia.
      Tanaka's original description reads, translated, as follows: "Leaves oblong-ovate, coriaceous, acute at the tips, rounded at the base, never attenuate; [stipuloid] paraphylls linear, very short; racemes few-flowered; calyx lobes very narrow and acute, glandular; stamens free (at length sometimes crowded); style evenly cylindric; stigma obtuse."   A typical specimen collected in Nilgiri Hills, India (Bentham, No. 1531), preserved in the Rijks Herbarium at Leiden, Holland, of which Tanaka provided Swingle a photograph, shows leaves 6-10.5 X 2-4.5 cm (exclusive of petioles), acute at the tip and rounded or subcuneate at the base, lateral veins numerous, 15-20 or more pairs arising at a wide angle (60°-70°) with the midrib, margins entire; petioles very short, 4-6 X 1-2 mm, wingless.   Another specimen of Bentham's type collection (No. 1531), preserved in the herbarium of the British Museum, London, shows the flowers in axillary racemes or corymbs, with 8-10 or more very small flowers borne on slender pedicels 4-5 mm long.
      A specimen with good flowers and young fruits in the herbarium of the Arnold Arboretum at Jamaica Plain, collected by Father Anglade in southern India in the Kodaikanal region, Pulney [or Pulni] Hills about 75 kilometers northwest of Madura, identified by Tanaka ("Det. A-382") as Atalantia wightii, has smaller leaves than the type specimens, blunt at the apex and emarginate, with the leaf blades 4-7 (or 8?) X 1.5-3.8 cm, with fewer lateral veins (11-18); petioles 3-6 mm long; flowers small, densely crowded in axillary racemes, pedicels very slender, 3-4 mm long; calyx lobes subacute at apex; petals linear, tapering to an acute apex, 5.5-6.5 X 2.5-2.8 mm.
      In properly restored material from this specimen cut into serial sections the floral organs appear as follows: Stamens 4.5-5 mm long, filaments 2.8-3.2 mm long, 0.1 mm wide at base, tapering upward, 0.25-0.3 mm thick, cohering in groups near the base, anthers without an oil gland in the connective; pistil (including disk) 4.5-5 mm long; disk short, shallow-cupulate, 0.5-0.6 mm high, 1.5 mm wide; ovary ovoid, 2 mm long, 1.6 mm wide, merging into the thick style, which is about 2 mm long, 1 mm wide, and merges into a stigma of nearly the same width, slightly 2-lobed for about 0.4-0.5 mm at the top; young fruits fusiform, 8-9 X 5-5.5 mm (when dry), in restored microtome cross section 6-7 mm diam., 2-locular with 2 collateral ovules in each locule, and a few pulp-vesicles developing in the locule walls.
      It seems probable that A. wightii is somewhat related to A. simplicifolia and A. roxburghiana.   Like these two species, it needs more study of the flower and fruit characters.
      6.   Atalantia simplicifolia (Roxb.) Engl. Die Nat. Pflanzenfam. 3(4):192. 1896. Amyris simplicifolia Roxb. Fl. Ind. 2:244. 1832; Atalantia caudata Hook. f. Fl. Brit. Ind. 1:513. 1875.
      Type.—"Pulo Penang," Malaya? (coll.?).   Herb. Brit. Mus., London.
      Distribution.—Northeastern India: Khasi Hills; also, fide Tanaka (1930b, p. 232), in [northern] Burma but not yet found in Malaya or even in southern Burma (Tenasserim).
      Common name.—Khasi Hills atalantia.
      Hooker's original description of A. caudata, collected by Hooker and Thompson at Churra, altitude 1,220 m [4,000 ft.], Khasi Mountains, northeastern India, now considered a synonym of A. simplicifolia, reads as follows: "Leaflet elliptic-lanceolate caudate-acuminate with a broad 2-lobed tip; flowers small in very short glabrous racemes, calyx very small 4-5 lobed, ovary 2-celled, ovules 2 collateral…A glabrous unarmed shrub, with slender branches.   Leaflets 3-4 in. [7.5-10 cm], much narrower and more narrowed at each end than in any congener, margin quite entire, tip sometimes dilated.   Racemes 1/2 in. [13 mm], peduncles and pedicels much more slender than its congeners.   Flowers 1/4-1/3 in. [6-8 mm] diam., white, fragrant, usually 4-merous, buds rather longer than broad.   Calyx very small.   Petals obovate-oblong.   Stamens quite free; anthers ovoid.   Ovary sessile on a disk narrower than itself; stigma subcapitate.—A very distinct species, distributed as A. Roxburghiana? by T. Thompson and myself."
      The type specimen of Roxburgh's Amyris simplicifolia, preserved in the herbarium of the British Museum, shows the following characters: leaves elliptical-lanceolate, 6.5-9.5 cm long (excluding petioles), 2-3.6 cm wide, acuminate at the apex, with an acumen 6-8 X 4-1 mm, the tip blunt and minutely emarginate, cuneate at the base and narrowed toward the petiole; petioles 5-6 X 11.5 mm with a narrow (nearly closed?) canal on the upper side, articulated with the blade, primary veins 10-16 with about as many fainter ones between, arising at angles of 50°-60° with the midrib; inflorescences axillary, peduncles 1-3 cm long, 0.4-0.6 mm wide, pedicles almost all fallen (about 3.5 mm long?).
      The type specimen of A. caudata preserved in the Kew Herbarium looks very much like-the type of Amyris simplicifolia except that the leaves are slightly narrower, 2-2.8 cm wide, and have a longer acumen, 8-11 X 4-2 mm, with a blunter, more deeply emarginate apex.
      This species is still incompletely known and in particular the flowers and fruits are not adequately described.   It is without doubt related to A. roxburghiana but seems to be specifically distinct.   It has been found so far only in the Khasi Hills in Assam and in near-by parts of northwestern Burma, more than 2,500 kilometers distant from the type locality of A. roxburghiana at Malacca; it has not yet been found in southern Burma (Tenasserim) about 1,800 kilometers distant.
      The nomenclature of this species is confused.   The type specimen in the British Museum is credited to Pulo Penang, but the species has never been found anywhere near Penang.   The original description of Amyris simplicifolia Roxb. agrees fairly well with the type specimen, but a drawing of the species, made by Roxburgh and published later by Wight (1840, pl. 72) under the name Sclerostylis roxburghii, does not agree either with the description or with the type specimen, as it shows short petioles not articulated with the blade, the margins of which are decurrent well down the petiole.   It also shows only a few (five to seven) main lateral veins and about as many more fainter and shorter ones, whereas the type specimen shows many more veins, both strong and faint.
      If the figure was made from some other plant, not this species, and if the type specimen came, not from Penang, but from Assam in northeastern India, or from northwestern Burma, then undoubtedly the name A. simplicifolia will hold.
      7.   Atalantia roxburghiana Hook. f. Fl. Brit. Ind. 1:515. 1875.
      Type.—Malay Peninsula, Malacca (Griffith).   Herb. Bot. Gard., Calcutta?
      Distribution.—Malay Peninsula, also Indo-China; fide Tanaka, varieties of this species also occur in Thailand and southeastern China; also, fide Burkill (1931, p. 216), in northern Sumatra.
      Common name.—Malayan atalantia.
      A sprawling thornless shrub, ultimate branchlets slender, 1.5-2.5 mm diam., soon terete, glabrous; leaves large, lanceolate or long-elliptical, tapering toward both ends, but often acute or shortly subacuminate at apex and sometimes broadly cuneate or rounded at base, 9-21 cm long (including petiole), 2-4 cm wide, glabrous, subcoriaceous, with abundant veins visible on both surfaces, primary veins very abundant, 10-12 stronger ones and as many fainter ones arising at an angle of 70°-80° with the midrib; petioles short, 6-10 mm long, 1.3-1.8 mm wide, with a very narrow, almost closed canal, 0.3-0.7 mm wide, on the upper side, glabrous, articulated with the leaf blade; flowers not seen; inflorescences (bearing fruits) axillary, 1-6 cm long, peduncle often simple, bearing pedicels 5-6 mm long, 0.6-1 mm wide; calyx lobes 4, persistent, glabrous, rounded, margins thin, sparingly ciliate; fruits subglobose, 1.5-2 cm diam., young fruits 2-locular, with abundant sessile, conical pulp-vesicles, older fruits often nearly full of oblong seeds, 10-12 X 4-6 mm; oil glands in peel rather sparse and often large, 0.3-1 mm diam.
      This seems to be a very well-defined species, easily recognized by its being spineless and having unusually long leaves with short petioles, less than one-tenth as long as the blade, which have a nearly closed canal on the upper side.   It is closely related to A. simplicifolia, which has smaller acuminate or caudate leaves.   Two varieties of A. roxburghiana are described below.
      The nomenclature of this species and of A. simplicifolia is badly confused, so badly confused that it cannot be settled finally without further study of the type material of both A. roxburghiana and A. simplicifolia.   It is possible that the name here used can be considered as legal since it was based, in part, on fruiting material that doubtless belongs to the species as here described.   Unfortunately, the original description was also based on a misleading illustration of Amyris simplicifolia Roxb. which does not agree with the original description of Amyris simplicifolia or with the type specimen of it preserved in the British Museum.
      7a.   Atalantia roxburghiana var. kwangtungensis (Merr.) Swing. Jour. Arnold Arbor. 21:129. 1940. Atalantia kwangtungensis Merr. Phil. Jour. Sci. 21:496. 1922.
      Type.—China, Kwangtung Province, Tung Sing (Ts'oong, No. 1936).   Herb. Bur. Sci., Manila.
      Distribution.—Known only from the type locality.
      Common name.—Kwangtung atalantia.
      The original description, in English, reads as follows: "A glabrous shrub, the branches terete, the branchlets somewhat compressed, unarmed.   Leaves subcoriaceous, olivaceous, oblong-elliptic, 10 to 14 cm long, 4 to 6 cm wide, subequally narrowed to the acute base and to the acute or somewhat obtusely acuminate apex, the upper surfaces slightly shining, the lower somewhat paler; primary lateral nerves about 14 on each side of the midrib, prominent on the lower surface, the ultimate reticulations rather close and distinct on both surfaces, the glands invisible except by transmitted light; petioles about 5 mm long.   Inflorescences axillary, very short, of solitary or fascicle-like racemes, the rachis 3 mm long or less.   Flowers white, about 6 mm long, their pedicels stout, 1 mm in length; bracteoles ovate, obtuse, about 1 mm long, the margins ciliate.   Calyx lobes 5, orbicular, rounded, 1.5 to 2 mm wide, the margins slightly ciliate.   Petals 5, 4 to 5 mm long, oblong-elliptic, rounded, glabrous.   Stamens usually 8, the filaments about 3 mm long, flattened, glabrous, and united for the lower one-half; anthers 1.2 mm long.   Ovary glabrous; style 1 mm long."
      This variety differs from the species in having usually 5-merous flowers (instead of 4-merous) and staminal filaments cohering from the base for about half their length instead of being free.   Tanaka (1930b, p. 232) referred A. kwangtungensis to A. roxburghiana and Merrill (1931, p. 311) accepted this reduction.   In view of the presence of characters in A. kwangtungensis not yet found in typical A. roxburghiana, it appears possible, even probable, that the southern Chinese form deserves to be recognized as a variety, as done here, pending further study of the A. roxburghiana complex of forms.
      7b.   Atalantia roxburghiana var. kerrii Swing. Jour. Arnold Arbor. 21:129. 1940.
      Type.—Thailand, Sam Roi Jawt (Kerr, No. 10943).   Herb. Univ. Aberdeen; photographs, fragments, and 12 serial microtome cross sections of a fruit, S. and T. No. 684-A, slides 1-4, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Known only from the type locality.
      Common name.—Kerr's atalantia.
      A small tree up to 5 m tall, ultimate branches at first green and slightly angular, soon brownish, terete and faintly striate longitudinally; internodes 1.2 cm long; leaves thick and leathery, broadly oval or elliptical, smaller ones lanceolate, 9-15 cm long (including the petiole), 3-8 cm wide (usually 6-8 cm), tapering to a blunt apex, broadly cuneate at base with numerous small oil glands scattered over the whole surface of the leaf blade but scanty near the margins which are entire or faintly crenulate and slightly thickened at the very edge, lateral veins 7-12 pairs, nearly straight or slightly curved, forked at 5/6-6/7 of the distance to the margin, making angles of 45°-60° (rarely 65°-70°) with the midrib, with numerous small lateral veinlets that anastomose; petioles stiff, glabrous, wrinkled, 8-13 mm long, with a shallow channel on the upper side, 1-2 mm wide, 0.4-0.5 mm deep; inflorescences axillary, paniculate (?), 2-10 cm long, peduncle 1-2 mm diam., pedicels 5-8 X 1.25-1.8 mm; fruits subglobose, 1.5-1.8 cm diam., peel covered with large, slightly sunken oil glands, 0.5-0.9 mm diam., 3-locular, pulp-vesicles numerous, sessile, 3-5 mm long, filling all the space not occupied by the seeds; seeds ovoid 11.5-12 X 10-11 X 6-8 mm, with a very thin, papery testa (which swells and separates from the embryo in hot water), monoembryonic.
      This remarkable Atalantia has large, broad, coriaceous leaves; unfortunately it is known only in the fruiting stage.   It differs strikingly in its leaf characters, not only from A. roxburghiana, but also from the other species of Atalantia known from the Indo-Chinese region.   As the species of this genus are, most of them, variable in many of their characters, it seems best to consider this striking form as a variety of A. roxburghiana Hook. f., not uncommon in the Malay Peninsula and reported from Indo-China, at least until flowers can be secured for study.   This variety is named in honor of Dr. A. F. G. Kerr, who did so much to make known the rich flora of Thailand.

Subgenus Rissoa
      Subgenus Rissoa (Arn.) Swing. in Bailey, Stand. Cycl. Hort. 1:426. 1914. Rissoa Arn. Nov. Act. Acad. Caes. Leop.-Carol. 18:324. 1836.
      Type.—Rissoa ceylanica Arn. = Atalantia ceylanica (Arn.) Oliv.
      Stamens free; fruits with the locules nearly filled with large seeds, with only a few sessile pulp-vesicles; leaves often accompanied at the base with a pair of small stipuloid paraphylls.
      The species of this subgenus are very like those of the subgenus Atalantia in general appearance but differ from them in having free stamens, and nearly dry fruits with large seeds, almost filling the locules.   Only two species of Atalantia (A. ceylanica and A. rotundifolia) are now known to belong to this subgenus, but probably several others will be placed here when their fruit structure is better known.
      8.   Atalantia ceylanica (Arn.) Oliv. Jour. Linn. Soc. Bot. 5(Suppl. 2):25. 1861. Rissoa ceylanica Arn. Nov. Act. Acad. Caes. Leop.-Carol. 18:324. 1836; Limonia monophylla L. Mant. Pl. Alt. 237. 1771; Sclerostylis ceylanica Wt. 1840; S. arnottiana Wt. 1840. Illus. Swingle, in Bailey, Stand. Cycl. Hort. 1:426, fig. 434. 1914 ; fig. 3-31 this work.
      Type.—Ceylon (Wight No. ?), Glasgow Univ. Herb. (fide Tanaka, 1930b, p. 232).
      Distribution.—Ceylon, low country; southern India.
      Common name.—Ceylon atalantia.
      This species was described as it occurs in Ceylon by Trimen (1893, p. 227) as follows: "A much-branched bush, usually armed with very sharp, stout spines, 1/4-1 in. [6-25 mm] long, bark smooth, brown, young parts glabrous, l[eaves] very variable, 1 1/4-3 in. [3-7.6 cm long], lanceolate or oval-oblong or somewhat ovate, rounded at base, emarginate or 2-lobed at apex, entire, coriaceous, veiny, petiole short, thick, glabrous; fl[owers] 1/2 in. [13 mm diam.] or rather more, on somewhat slender, glabrous ped[icels], usually crowded in short, very finely pubescent, corymbose, or racemose cymes much shorter than the l[eaves], bracts minute; cal[yx] lobes broad, shallow, often obscure, finely ciliate; pet[als] 4 (rarely 3 or 5), oval, with a broad claw, obtuse, thick, slightly concave, glandular; stam[ens] 8, alternate ones longer, quite distinct, fil[aments] flat, rather wider than anth[ers], anth[ers] cordate-ovate, apiculate; ov[ary] sessile, smooth, 2-celled, with 2 collateral ovules in each cell; style short, stout, stigma clavate; berry 1/2-3/4 in. [13-19 mm diam.], globular-ovoid, 2-4-seeded."
      Hooker (1875, p. 511) described it as follows: "leaflet obtuse, notched or 2-lobed, rarely obtusely acuminate, flowers very shortly racemed, calyx 4-lobed, filaments free or 2 or 3 of them more or less combined, ovary 2-celled, ovules 2-collateral."   He mentioned in the notes that the ovary is "sunk in the annular disk."
      Atalantia ceylanica is remarkable for its nearly dry fruits, with the very few pulp-vesicles almost crowded out by the large, plump seeds that almost completely fill the locules.   Small stipuloid paraphylls, 4 to 6 mm long, often occur singly or in pairs at the base of the petiole.   This species seems to be related to A. guillaumini, from Indo-China, which apparently has lost its pulp-vesicles and has still larger seeds that completely fill the locules of the fruit.
      9.   Atalantia rotundifolia (Thwaites) Tan. Bul. Soc. Bot. France 75:714. 1928. Sclerostylis rotundifolia Thwaites, Enum. Pl. Zeyl. 46. 1858; Atalantia ceylanica var. rotundifolia (Thwaites) Oliv. 1861.
      Type.—Ceylon, Muratte (Thwaites, No. 3295).   Herb. Kew; photographs, fragments, and serial microtome sections of type material, S. and T. Nos. 652-A, 652-B, 652-C, and 682-D, 17 slides, in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Ceylon: "Montane zone at about 4,000 ft. [1,220 m] and upwards"; also in Madras, India, and eastward in Indo-China (Vietnam), Annam and Cochin China (fide Tanaka, 1928b, p. 714).
      Common name.—Dwarf Ceylon atalantia.
      Thwaites's original description reads (translated with the addition of the English account) as follows: "Spineless; leaves rounded or ovate-rounded, retuse, coriaceous; inflorescences racemose, grouped at the ends of the branches.   A small tree, very much branched, glabrous.   Leaves 0.75-1.75 in. [19-45 mm] long, 0.5-1.4 in. [12-39 mm] wide; petiole 1-2 lin. [2-4 mm] long.   Racemes axillary and terminal, solitary, or in clusters of 2 or 3, 5-10-flowered, as long as the leaves.   Flowers 4-merous.   Calyx profoundly 4-lobed; lobes rounded.   Ovary oblong, short-stalked, attenuate into a cylindric style of equal length ending in a compressed-dilate top with stigmatic margins; locules 2, 2-ovulate [Latin to here].
      "Closely allied to the preceding species [= A. ceylanica], but distinguished at once by the absence (apparently) of spines, and the different shape of the leaves.   The calyx is deeply four-lobed, whereas in S. [Atalantia] ceylanica it is scarcely more than undulated; the style is also more slender in the present species, and the ovary more decidedly stalked.   In both species the latter organ is surrounded at its base by a free, glandular annulus."
      A photograph of Thwaite's [sic] type, No. C.P. 3295, in the Rijks Herbarium at Leiden, sent Swingle by Tanaka, shows that this species has the leaves densely crowded on the twigs, because of the shortness of the internodes.   No spines can be seen.   Many of the leaves are elliptical in shape, abruptly rounded at the narrow base but usually emarginate (often rather deeply) at the apex, giving the appearance of an obcordate leaf that had been greatly compressed laterally.   The leaves appear to be tufted at the ends of the short twigs, where the internodes must be very short, probably only 4 to 6 mm long in some instances.
      The whole appearance of this plant is reminiscent of the spineless or nearly spineless, short-internoded mutations of Severinia buxifolia—for example, the brachytic form—and the myrtle-leaved mutation of the sour or Seville orange, Citrus aurantium var. myrtifolia.   Possibly this species arose from A. ceylanica as such a brachytic, spineless mutation which in time bred true and persisted at higher altitudes.   Atalantia rotundifolia therefore is probably to be considered as a satellite species of A. ceylanica.   It is possible that A. rotundifolia will some day be used for hedges or other ornamental purposes in moist subtropical regions.

SPECIES OF UNCERTAIN RELATIONSHIPS
      10.   Atalantia guillaumini Swing. in Lecomte, Not. Syst. 2:159. 1911. Atalantia disticha Guill. (non Merr.) in Lecomte, Fl. Gén. Indo-Chine 1:673. 1911. Illus. Swingle, loc. cit. 2:162. fig. 1.
      Type.—French Indo-China, southern Tonkin, Lang-hé, on Mount Dên (Bon, No. 4047).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—North Vietnam: known only from southern Tonkin, from Mount Dên, Mount Nam-cong, Mount Luong-xa, and Mount Thin-chan (all from collections made by H. Bon.)
      Common name.—Tonkin atalantia.
      Swingle's original description, translated, reads as follows: "Leaves ovate-lanceolate, apex obtuse, base rounded or deltoid with many lateral veins running parallel [margins entire]; petioles articulated at both ends, not winged, not [deeply] channeled; calyx persistent, sepals 4-5, subtriangular, with sparsely ciliate margins; fruits solitary, 2.4-2.8 cm diam., globose, with 3 or 4 segments, segments 2- (or 1- ?) seeded, destitute of pulp-vesicles; seeds large, with membranous testa."   [Latin diagnosis to here; continuing, in French, the translation reads:] "A small tree, 3-4 m high, with spiny, pubescent branches.   Twigs more or less angular when very young, soon becoming rounded and glabrous.   [Fruiting twigs] spineless.   Internodes 1-2 cm long.   Leaves large, oval or oval-lanceolate, rounded at the tip and base; the small leaves lanceolate, more or less pointed at the tip and base; leaf blades 55-85 mm long, 25-60 mm wide, always clearly distinct from the petiole (which is, moreover, articulated at the point of junction with the leaf blade), glabrous, without oil glands except at the margins; lateral veins numerous, 20-30 [sometimes fewer], parallel and almost straight for 3/4-5/6 of their length, confluent by anastomosis near the margin; petioles 5-9 mm long, flattened above with the edges slightly elevated, brown in dry specimens, with many or few very slender hairs; stipules [stipuloid paraphylls] short, narrow, 2 mm long, 0.5 mm wide, veinless.   Inflorescences few-flowered, spicate, 3-4 cm long; axillary or else bearing only isolated flowers or a single fruit in the axils of leaves near the tips of the young twigs; flowers not seen; fruit pedicels 6-8 mm long, 1.5-2 mm diam.   Calyx persistent, composed of 4-5 distinct sepals that are more or less triangular, sparsely ciliate on the margins.   Fruits solitary, spherical or slightly pointed, 2.4-2.8 cm diam. with 1-3-fertile segments; seeds 2 in each segment, very large, 12-15 mm long, 8-9 mm wide, rounded at both ends, more or less angular from mutual pressure, with a very thin membranous testa; pulp-vesicles absent, skin glandular and rough, but thin."
      Guillaumin (1911, p. 673) described this plant (under the name Atalantia disticha [Blanco] Merrill) as a "tree 4 m high, spineless," but went on to say, "branches…with small erect spines," and later on, "petioles…furnished with 2 erect stipules [stipuloid paraphylls], in the form of spines reaching 5 mm [in length] or in the form of minute linear leaf blades [paraphylls] showing a central vein and a few veinlets."
      This species much resembles A. ceylanica (1) in having large fruits filled with very large seeds; (2) in the general character of the leaves, which have few or no oil glands except along the margins; and (3) in having frequently one or more paraphylls more or less like stipules at the base of the petiole, usually more spinelike than in A. ceylanica.   The ripe fruits of A. guillaumini do not show pulp-vesicles, or at least Swingle did not find any in the type specimen.   The ripe fruits of A. ceylanica usually show traces of a few pulp-vesicles, although almost all of them are crowded out by the very large seeds that almost completely fill the locules.   The still larger seeds of A. guillaumini may perhaps obliterate all the pulp-vesicles in many of the fruits, or even in all of them.   The study of young fruits of this species should show whether a few pulp-vesicles are present or whether they have been completely suppressed.   If the latter should prove to be true, a very interesting and important taxonomic problem would be raised, since the possession of pulp-vesicles of a special type is an important character of the genus Atalantia.
      Severinia disticha, to which this species was assigned by Guillaumin (1911, p. 673), is a very different plant, having small fruits with rudimentary, round, stalkless pulp-vesicles crowded into the periphery of each locule.   A single seed of Atalantia guillaumini is larger than an entire fruit of Severinia disticha.   Severinia disticha has, moreover, the cup-shaped disk enclosing the lower part of the ovary, as in other typical species of Severinia.
      11.   Atalantia hainanensis Merr. & Chun, ex Swingle, Jour. Arnold Arbor. 21:20. 1940. Illus. Swingle, loc. cit. pl. 4, figs. 3-7.
      Type.—China, Hainan Island, Po-Ting (How, No. 72807).   Herb. Arnold Arbor, [sic] Harv. Univ., Cambridge; photographs, fragments, and serial microtome sections, S. and T. Nos. 261-A, 261-B, 261-C, 12 slides, in Herb. Natl. Arbor., Washington, D.C.   Cotype: Same locality (How, No. 73976), Herb. Arnold Arbor.; serial microtome sections S. and T. Nos. 263-A and 263-B, in Herb. Natl. Arbor.
      Distribution.—Known also from Yaichow, Hainan Island.   Not uncommon, usually in forests near streams at altitudes of 100-640 m.
      Common name.—Hainan atalantia.
      A small spineless shrub, 1-2 m high; young twigs slightly angled but soon becoming cylindrical; leaves simple, persistent, thick and coriaceous, elliptical or broadly elliptical, slightly acuminate or bluntly pointed at the tip, often emarginate, cuneate at the base, narrowing gradually into the petiole, very variable in size, usually 6-15 X 2.5-6 cm, but sometimes as large as 20 X 8 cm including the petiole, margins entire or faintly undulate; petiole not articulated with the leaf blade, 5-10 mm long, more or less pulvinoid, with a channel on the upper side formed by the decurrent leaf margins; inflorescences axillary, few-branched panicles, short (1-1.5 cm); pedicels slender, 2-3 X 1 mm; flower buds small, 3-3.5 mm long, 2-2.5 mm wide; calyx with 5 triangular lobes, thickened, roughened with oil glands except at the margins, which are thin and ciliate; petals 5, white; stamens 10, filaments flattened, connate at the base, free above, anther with 1 large oil gland in the connective; ovary small, 1.5 X 1 mm, ovoid, with 2 locules, each with 2 collateral ovules, top of ovary with 1 large oil gland above each locule; style slender, 1.25 X 0.3-0.4 mm, stigma nearly isodiametric with the style, with 2 pairs of large oil glands; disk cupulate, about 0.35 mm deep, glabrous, completely surrounding the basal 2/5 of the ovary; fruits (young) ellipsoid, 8-10 X 4-5 mm, peel green, with numerous oil glands, surmounted by the small persistent style, 1 X 0.6-0.7 mm; seeds 1 (or more?) to a fruit, large, ellipsoid, about 10 X 7 X 5 mm, monoembryonic (?), cotyledons with numerous oil glands.
      This curious plant, clearly a very distinct new species, is hard to place in the absence of mature fruits.   It has leaves varying greatly in size (2.5 to 19.5 by 0.8 to 7.5 cm) which have much the same general aspect as those of Severinia buxifolia, a species it further resembles in having a cupulate disk and a 2-locular ovary with a large oil gland over each locule.   One specimen not considered in the leaf measurements given above (How, No. 73818) from Po-Ting, Hainan, in the herbarium of the Arnold Arboretum, shows relatively very small leaves, only 2.5 to 5.5 by 0.7 to 1.5 cm, narrowly elliptical, very blunt-pointed and retuse at the tip.   Young fruits with persistent calyces on this specimen were, however, almost identical with those of the large-leaved specimens from the same locality in Hainan.   However, serial microtome sections made of the immature fruits of the cotype (How, No. 73976) show the locule walls lined not with a clear-cut inner layer of stalkless subglobose pulp-vesicles but rather with large and variable-sized oil glands (?), not all in a definite layer but with a few of them scattered between the much smaller oil glands of the peel and the inner locule wall.   Study of the fruits of this species at all stages of development may show that these structures have a merely superficial resemblance to and no true homology with the primitive pulp-vesicles of Severinia buxifolia and other typical species of Severinia.   The typical species of Atalantia have sessile, broad-based, conical pulp-vesicles growing out from the dorsal locule walls which with the seeds fill the locules completely.   However, A. ceylanica, the type species of the subgenus Rissoa, shows in the mature fruit very few pulp-vesicles, perhaps because the very large seeds almost completely fill the locules.   Atalantia hainanensis, like the anomalous A. guillaumini, of which only the very large, subglobose, fully mature fruits are known, does not seem to have any pulp-vesicles.   These two anomalous species of Atalantia need critical study at all stages of the developing fruits in order to be certain of their taxonomic placement.

GROUP C.   THE TRUE CITRUS FRUIT TREES
      This group includes six genera, Fortunella, Eremocitrus, Poncirus, Clymenia, Microcitrus, and Citrus, all marked by having orange or lemon-like fruits with highly specialized, slender-stalked, usually more or less fusiform pulp-vesicles filling all the space in the segments of the fruit not occupied by the seeds, and also by having at least four times as many stamens as petals.   All the genera have persistent unifoliolate or simple leaves except the monotypic genus Poncirus, which has trifoliolate, deciduous leaves.   Clymenia differs in having simple leaves borne on short, wingless petioles, with numerous, nearly straight veins prominent on the lower surface, somewhat resembling the venation of the leaves of Wenzelia and Monanthocitrus of the subtribe Triphasiinae.
      All the genera in this group except Clymenia have fusiform pulp-vesicles borne on very slender stalks, sometimes short, sometimes long.   In Clymenia the pulp-vesicles are subglobose or pyriform in shape, usually having subglobose apices and being narrowed into somewhat contracted bases; the bases, however, are never reduced to slender stalks of variable length, as they are in the pulp-vesicles of the other genera of this group.   All the genera except Clymenia have all, or at least the majority, of the pulp-vesicles attached to the dorsal walls of the locules.   A few species of Citrus (subgenus Papeda) have a few or sometimes many pulp-vesicles attached to the radial walls, but only on the distal portion of these lateral walls nearest the periphery of the fruit.   Clymenia differs from all the species of the subgenus Citrus of the genus Citrus in that many of the pulp-vesicles, probably a large majority of them, are attached to the radial walls of the segments and occur on these walls nearly to the center of the fruit (three-fourths to four-fifths of the way).   Although Clymenia has much simpler pulp-vesicles, without specialized apices, the other fruit characters are very like those of Citrus, to which it is obviously related.
      The structure of the pulp-vesicles is of great taxonomic importance in the classification of the genera of the subtribe Citrinae.   Found only in this subtribe, these unique structures reach their highest development in group C, the True Citrus Fruit Trees.   In contrast with group B, the Near-Citrus Fruit Trees, in which the pulp-vesicles are conical in shape, tapering into more or less acute apices and having broad, sessile bases, the pulp-vesicles of the True Citrus Fruit Trees are never conical in outline, but are fusiform, being narrowed at the base (except in Clymenia) into slender stalks (varying considerably in length in different pulp-vesicles of the same locule) and into a point at the apex.   In Clymenia the pulp-vesicles are different from any others occurring in the True Citrus Fruit Trees in being pyriform and in not being slender-stalked at the base nor having acute apices.
      The pulp-vesicles of all the genera included in this group, with the possible exception of Clymenia (not yet available for study in the living condition), contain droplets of oil.   These oil droplets are very abundant in the pulp-vesicles of Poncirus, Microcitrus, and the species of Citrus belonging to the subgenus Papeda, but are fewer in the pulp-vesicles of Fortunella, Eremocitrus, and the species of Citrus belonging to the subgenus Citrus.   Similar oil droplets (or even wax particles) occur also in the pulp-vesicles of the Near-Citrus Fruit Trees, Citropsis and Atalantia.
      Three genera in group C, Clymenia, Eremocitrus, and Poncirus, are monotypic.   Two other genera, Fortunella and Microcitrus, have four and six species respectively.   All five of these genera have a limited area of distribution much more restricted than that of Citrus, which is the dominant genus of the subtribe with the largest number of species (sixteen are recognized in this chapter).
      The region occupied by the six genera comprising the True Citrus Fruit Trees is a long, barrel-shaped area about 9,000 kilometers (5,590 miles) long and 3,200 kilometers (1,990 miles) wide in the center, tapering to about 2,600 kilometers (1,615 miles) at the ends.   This barrel-shaped area has its long axis slanting from the northwest (northeastern India to north-central China) to the southeast (east-central Australia to New Caledonia).   At its broadest part the area extends from Java to the eastern shores of the Philippine Islands.   If, as some experts believe, Citrus is native to southern Japan, there would be an extension of the northern corner of this area to include Citrus tachibana.   Citrus is native to the whole of the area except in the southernmost corner (northeastern Australia), where it is replaced by Microcitrus and Eremocitrus, and in the extreme northernmost corner (northern China), where it is replaced by Poncirus.   Fortunella occurs in southeastern China, where Citrus is probably also native, and Clymenia occurs in the Bismarck Archipelago (the extreme easternmost part of the area), where one species of Citrus is probably also native.
      The six genera of group C are all obviously closely related and all (except the newer, little studied genus Clymenia) have been grafted on one another and hybridized with one another.
      Fortunella resembles Atalantia (a Near-Citrus Fruit Tree of group B) in having 3 to 5- (rarely 6- or 7-) locular ovaries with only two ovules in each locule, also small fruits and thick, stiff leaves, but differs in having highly organized, slender-stalked pulp-vesicles.   Eremocitrus has ovary and fruit characters much like Fortunella, but it has undergone striking xerophytic adaptations that have culminated in the acquisition of leaves very diverse in structure from those of the other genera of the True Citrus Fruit Trees.   Poncirus stands alone in group C in having trifoliolate, deciduous leaves and winter buds well protected by bud scales.   It has pleiomerous ovaries with six to eight locules, with many ovules in each locule.   In these ovarial characters it agrees with Microcitrus and Citrus.   Poncirus undoubtedly represents an ancient offshoot from the True Citrus Fruit Trees that pushed north into China and in so doing acquired deciduous leaves and bud scales and developed great resistance to winter cold.   Clymenia is probably the most primitive of all the genera comprising the True Citrus Fruit Trees.   Although it has fruits (said to be sweet and edible) that appear much like those of Citrus, it differs widely from Citrus in having more primitive pulp-vesicles and in having leaves that resemble those of the genera Monanthocitrus and Wenzelia of the subtribe Triphasiinae, the Minor Citroid Fruit Trees.   Microcitrus is a primitive genus related to Citrus; perhaps some of its species are very like the ancestral species from which Citrus developed.
      Of the six genera comprising the True Citrus Fruit Trees, Fortunella, especially the subgenus Protocitrus, is the most simple and primitive of the main branch that ends in Citrus.   Microcitrus represents a higher stage of evolution and is still closer to Citrus.   Eremocitrus and Poncirus are genera profoundly modified, the one to withstand semiarid climates and the other winter cold.   Clymenia is the Cinderella of the group; it shows some affinity with the Papeda subgenus of Citrus and even some relationship to certain of the Primitive Citrus Fruit Trees, namely, Pleiospermium, Limnocitrus, and Burkillanthus, on the basis of which a new theory on the origin of the True Citrus Fruit Trees is possible (see under Clymenia; also under Pleiospermium).
      Four of the six genera in group C are clearly adapted to fit exceptional climatic or soil conditions, conditions under which Citrus grows poorly or fails completely.   Eremocitrus is a pronounced xerophyte, able to grow in semiarid regions in northeastern Australia that are too dry to support Citrus.   Microcitrus is semi-xerophytic and can endure long drouths.   Poncirus has become deciduous and is able to endure very severe cold in winter, a condition fatal to Citrus and even to Fortunella.   It also has developed well-protected flower buds that form during the early summer and push into bloom from old twigs the following spring.   Fortunella can withstand long spells of warm weather in winter or early spring without showing new growth, thereby escaping all danger of injury from spring frosts, often very injurious to Citrus.   These four genera can grow where no species of Citrus can thrive; perhaps all of them are in fact survivals that escaped destructive competition with Citrus by occupying territory where the latter genus could not live.   The other relative of Citrus in this group, Clymenia, is so little known as yet that it is not possible to state whether it has any physiological or structural adaptations that permit it to survive where Citrus would fail.   Perhaps, like the four other genera of limited geographic range, it will on further study show adaptive structures and physiological peculiarities that will give it an advantage over Citrus in the regions where it occurs in a wild state.
      There is every indication that Eremocitrus, Microcitrus, Poncirus, and Fortunella, because of their ancient and deeply inbred adaptations to special climatic and soil conditions, will prove important in breeding new types of citrus fruits and new rootstocks able to resist disease and able to endure unfavorable climatic and soil conditions that no Citrus species can withstand.   It is for this reason that every effort has been made to present as full an account as possible of these close relatives of Citrus, all of which can very probably be hybridized with Citrus and also with one another.

XXI.   Fortunella Swingle
      XXI.   Fortunella Swingle, Jour. Wash. Acad. Sci. 5:167. 1915.
      Type species.—Citrus margarita Lour. = Fortunella margarita (Lour.) Swing.
      Distribution.—Southern China, probably now only in cultivation.
      Common name.—Kumquat.
      Shrubs or small trees; young branches angular, the older ones rounded; spines borne singly at one side of the bud in the axils of the leaves, or wanting; leaves 1-foliolate, rather thick, blunt-pointed or even retuse, acute or rounded at the base, veins evident above, scarcely showing beneath, lower surface pale green, densely glandular-dotted; petioles narrowly winged or merely margined, sometimes not articulated with the leaf blade; flowers borne singly or in few-flowered clusters in the axils of the leaves, hermaphrodite, 5-merous (rarely 3-, 4-, or 6-merous); flower buds, 8-10 mm long, more or less angular in cross-section; petals 5 (rarely 4 or 6), white, acute, 8-12 mm long; stamens 16 or 20, polyadelphous, cohering irregularly in bundles; filaments broad, but tapering at the tip; pistil seated on a well-marked cylindrical disk; ovary subglobose, with 3-7 locules, with 2 collateral ovules in each locule; ovary merging gradually or abruptly into the short style, this usually shorter than the ovary, sometimes shorter than the stigma; stigma capitate, symmetrical, cavernous within because of the large, deep-seated oil glands (1/4-1/5 the diam. of the stigma); fruits small, usually 2-4 times as long as the petiole, ovoid or globose; peel rather thick, fleshy, aromatic, and sweet flavored, containing large immersed oil glands; segments 3-7; pulp-vesicles small, fusiform or subglobose, stalked, containing an acid juice; seeds ovate in outline, smooth; embryo pistache green, germination with hypogeous cotyledons; first foliage leaves broadly ovate, subsessile, opposite as in Citrus.
      The genus Fortunella resembles Citrus in the general appearance of the stems, twigs, spines, leaves, flowers, and fruits and in having the polyadelphous stamens cohering in bundles and normally four times as numerous as the petals.   It differs from Citrus: (1) in having an isomerous or hypomerous ovary normally with three to five, rarely six or seven, locules (not polymerous with 8 to 15 or more locules); (2) in having two collateral ovules in each locule (not 4 to 12); (3) in having a cavernous stigma containing a few large, deeply immersed, lysigenous oil glands, usually in pairs, oval in cross section, with the radial diameter longer, about one-fourth to one-fifth the diameter of the stigma (in Citrus the homologous oil glands are so much smaller [one-tenth to one-fifteenth the diameter of the stigma in C. sinensis] that they do not give a cavernous character to the stigma); (4) in having the under surface of the leaves pale green, nearly veinless, and with very numerous, small, deep-green glandular dots; (5) in having very small fruits with acid pulp and a sweet, edible, more or less pulpy skin; (6) in having small, more or less angular flower buds.
      The kumquat bears abundant orange- or flame-to-orange-colored fruits of small size, often less than an inch in diameter.   Fortunella margarita and F. japonica, widely cultivated in China and Japan and in all subtropical regions, have fruits with a relatively thick, fleshy, sweet, edible peel, with four to seven segments filled with mildly acid pulp.   A third species, with long, slender leaves and long petioles, F. polyandra, commonly cultivated in the Malay Peninsula, has larger globose fruits with a thin peel.   A fourth species, F. hindsii, which grows wild in the mountains of southern China and on the island of Hong Kong, has very small globose fruits with three to four segments.
      The kumquat orange, though described by early Chinese writers on agriculture, remained virtually unknown to Europeans until recent times.   The kumquat is mentioned in many early Chinese works and described in some detail by Han Yen-chih (1923) in his treatise on the oranges written in 1178.   Later works of both Chinese and Japanese authors treat of it fully, often with fairly good illustrations.
      The first vague description of the kumquat orange in European literature was published by Ferrari in 1646 in his Hesperides and was based on reports made to him by Alvaro Semedo, a Portuguese Jesuit who lived for twenty-two years in China.   Ferrari's successors, Sterbeeck, Volckamer, Risso and Poiteau, and other authors of monumental illustrated works on citrus fruits, have added nothing to our knowledge of the kumquat.
      Full descriptions of the round and oval kumquats were published by Hume (1903, 1909), but not until 1912 was there a good account of these plants published in Europe, when Trabut (1912) described them and distinguished them from the so-called "chinosis" or "chinotto" (Citrus aurantium var. myrtifolia), with which they had been confused by Volckamer (1708) and many subsequent European writers.   Two years later, Trabut (1914) published in Algeria a fuller illustrated account of these plants and gave an explanation of the failure of the kumquat to become known and be propagated in continental Europe, namely, that seedlings do not thrive and, furthermore, that viable seed is hard to secure; moreover, he stated that the kumquat, when grafted on sour or Seville oranges (the stock generally used), did not succeed at all.
      One of the kumquats, F. polyandra, is native to tropical regions; the other three grow in cool subtropical or even warm temperate regions.   The type species of Fortunella, F. margarita, and the closely related F. japonica not only are resistant to cold when in a dormant condition but also exhibit the highest degree of winter dormancy of any of the True Citrus Fruit Trees.   These two species of kumquats can endure fairly warm weather in winter or early spring lasting for many days or even for some weeks without starting new growth (Swingle, 1910, 1913e; Swingle and Robinson 1923, p. 229.)
      Fortunella approaches Atalantia, and differs strikingly from Citrus, in having only two collateral ovules near the top of each locule (Citrus has 4 to 12 ovules).   However, Fortunella differs from Atalantia in having four times as many stamens as petals (instead of twice as many) and agrees with Citrus in having similar twig, leaf, spine, flower, and fruit characters.   The leaves, however, show very many more oil glands on the underside than are present in any species of Citrus, often ten times as many.   In many superficial fruit characters Fortunella agrees with the Australian desert lime, Eremocitrus glauca.   The seeds are very different, however, and the stem, twig, leaf, and flower characters are so strikingly different that it is not possible to regard these genera as being very closely related.

THE SPECIES AND HYBRID PROBLEM IN FORTUNELLA
      The species of Fortunella, in spite of their small number, have been very inadequately studied, largely because, until recently, only one species was cultivated in either Europe or the United States and the others were represented only by scanty herbarium material in a few of the largest herbaria.
      The discovery that a small-fruited plant, obviously a kumquat, was masquerading under the name Atalantia hindsii led Swingle to a study of the group, with the result that in 1915 he proposed a new genus, Fortunella, to include all the species of kumquats that had hitherto been placed in Citrus and Atalantia.   Important characters were brought to light that separated Fortunella sharply from both Citrus and Atalantia.   Further studies have disclosed the fact that a number of varieties of kumquats found in cultivation in China and Japan are merely hybrids due to chance cross-pollinations by insects in village or dooryard groves.   The Meiwa variety (called Chintan in China), at first considered by Swingle to be a good species (Fortunella crassifolia), and the Changshou kumquat, named as a species (F. obovata) by Tanaka, are both doubtless mere garden hybrids not entitled to rank as species.   This decision helps greatly in clarifying the taxonomic status of the genus.
      There are also intergeneric hybrids resulting from cross-pollinations between Fortunella and Citrus found in cultivation in China and neighboring countries.   One such hybrid, the Calamondin, has been erroneously named as a species of Citrus (it was called C. microcarpa by Bunge and C. mitis by Blanco).   However, the whole complex intergeneric hybrid problem of Fortunella has been materially clarified by the making of many accurately safeguarded cross-pollinations of Fortunella with Citrus, Poncirus, Microcitrus, and Eremocitrus.   This work, conducted under Swingle's direction in the Agricultural Research Service, led to the creation of a bewildering group of strange hybrids that would have been very disturbing to the taxonomist if their origin were not definitely known.   They are discussed below under Fortunella hybrids.
      Several species of Fortunella have developed a degree of resistance to winter cold and at the same time an even more important physiological peculiarity, namely, a very pronounced winter dormancy that permits them to pass through weeks of warm weather without starting growth or flowering.   This quality, possessed in much higher degree by the kumquats than by any other citrus fruit tree (not excluding the winter-hardy Poncirus), makes them of prime importance in the breeding of new types of hardy citrus fruit trees (especially for producing acid citrus fruits) able to grow in much colder regions than the lemon or lime (both notoriously deficient in winter dormancy and hence easily pushed into growth by a few days of untimely warm weather in winter or early spring).
      Fortunella hybrids will in all probability be produced in large numbers in the future.   Their study is likely to prove of great importance in bringing about a just appreciation of the hybrid problem in the taxonomy as well as in the breeding of citrus fruit trees.   Since Fortunella, with its few species and its already well-understood intergeneric hybrids, constitutes a veritable microcosm of the citrus world, it will doubtless prove of great help in understanding the much more complex and still only imperfectly studied species and hybrid problems in the genus Citrus itself.   A key to the subgenera and species of Fortunella is presented.

Subgenus Fortunella
      Subgenus Eufortunella Swing. in Webber & Batchelor, Citrus Indus. 1:346. 1943.
      Ovary with 4-7 locules; fruits more than 1.5 cm diam., with many pulp-vesicles.
      1.   Fortunella margarita (Lour.) Swing. Jour. Wash. Acad. Sci. 5:170. 1915. Citrus margarita Lour. Fl. Cochinch. 2:467. 1790; C. aurantium [var.] olivaeformis Risso ex Loisel.-Deslong. Nouv. Duhamel 7:95. 1816; C. aurantium var. japonica Hook. f. Curt. Bot. Mag. 3 ser. 30:pl. 6128. 1874; C. aurantium subsp. japonica var. globifera subvar. margarita Engl. Die Nat. Pflanzenfam. 3(4):199. 1896. Illus. Hooker f. loc. cit. pl. 6128 (col.); Siebold & Zuccarini, Fl. Japon. 1:35, pl. 15, fig. 3. 1835; W. G. Smith, Gard. Chron. 2 ser. 2:337, fig. 72. 1874; same figure, ibid. 3 ser. 7:393, fig. 58. 1890; also same figure, redrawn in Gartenflora 31:pl. 1097 (col.). 1882; Swingle, loc. cit. 5:171, fig. 2; idem, in Bailey, Stand. Cycl. Hort. 2:1269, fig. 1563(1); 1270, fig. 1564(1). 1915; Hume, Cult. Citrus Fruits 118, fig. 76. 1926.
      Type.—Canton, China, cultivated (coll. by Loureiro, but lost).
      Distribution.—Known only in cultivation; doubtless native in southeastern China.
      Common name.—Oval kumquat.
      Loureiro's original description (translated from the Latin) reads as follows: "Differs from the species of Citrus in its ascending, spiny branches, its linear petioles, its 5-locular, oblong fruits.
      "Habitat and notes: A tree 4 ft. [1.22 m] high, branches spiny, ascendant; leaves lanceolate, entire, shining, few; petioles linear; flowers white with 5 petals, fragrant; peduncles sparse, few-flowered; fruit oblong-ovate, yellowish-red, glabrous, 8 lines [16 mm] long; 5-locular; peel thin, pulp vesicular, sweet, edible.
      "Habitat: Canton, China, not rare.   Not seen by me in Cochin China."
      Hume (1926) gave the following description: "Tree dwarf, eight to twelve feet, bushy; young branches somewhat angled, light green; leaves 1 1/4-3 1/2 X 3/8-1 1/4 inches [3-8.8 X 1-3 cm], lanceolate, apex obtuse; base acute or obtuse; margins crenate down about half-way from the apex; veins inconspicuous, surface dark green, glossy; lower lighter; borne on rather stout, usually very slightly margined, petioles, 1/4 to 5/8 inch [6-16 mm] in length.   Fruit small, obovate or oblong, 1 1/4-1 3/4 X 3/4-1 3/16 inches [3-4.4 X 1.8-2.5 cm], golden yellow; stem short; calyx small; rind smooth, aromatic, spicy; oil glands large; juice acid, sparse; sections usually five; seeds 2 to 5, oval 1/2 inch [13 mm] long, greenish; cotyledons two, green; season October-January.
      The oval kumquat differs from the round kumquat chiefly in the following respects: (1) the leaves are larger, more acute at the base, less pallid, and more veiny below; (2) the ovary has usually four or five (not four to seven) locules; (3) the fruit is oval, not globose; (4) the style is persistent, not caducous; (5) the seeds are larger and especially longer, with a rougher testa.   It differs also in being distinctly more vigorous and attaining a greater height (3 to 4 m); in the somewhat brighter orange color of its fruits; and especially in the harsher, more biting flavor of the peel, which evidently contains an ethereal oil more like that of the common orange than is that of the round kumquat.
      The type specimen of Citrus margarita seems to have been lost, but Loureiro's description is very good and can apply only to an oval kumquat very similar to the one commonly cultivated in all the warmer parts of the world.
      Robert Fortune, the celebrated explorer of the tea regions of China, brought back plants of the oval kumquat from China and delivered them to the Royal Horticultural Society's Garden in London on May 6, 1846.   He stated (1848) that he had found numerous groves of kumquats thriving to perfection, along with tea, on the lower hills of Chusan Island (off the coast of Chekiang Province, near latitude 30°) where the climate is far too cold to permit the culture of ordinary sweet oranges or mandarins.   He added: "this shows, therefore, that the Kum-quat is of a much hardier nature than any of the plants belonging to the orange tribe with which we are acquainted in gardens."
      In a later Publication Fortune (1870, p. 50) asserted that the kumquat needs a hot summer, 80° to 100° F and "will bear without injury 10° to 15° of frost, and perhaps even a lower temperature."
      The oval kumquat reached the United States by 1850 (see Hume, 1926, 114), but the round kumquat was not known in either Europe or the United States until near the end of the nineteenth century.
      2.   Fortunella japonica (Thunb.) Swing. Jour. Wash. Acad. Sci. 5:171. 1915. Citrus japonica Thunb. Nov. Act. Soc. Sci. Upsal. 3:199. 1780; also Fl. Japon. 292. 1784; C. madurensis Lour. 1790; C. inermis Roxb. 1832; C. aurantium subsp. japonica var. globifera Engl. Die Nat. Pflanzenfam. 3(4):199. 1896. Illus. Thunberg, Icon. Pl. Japon. 2:pl. 5. 1800; Siebold & Zuccarini, Fl. Japon. 1:pl. 15. 1835; Fortune, Jour. Roy. Hort. Soc. London 3:240. 1848; Swingle, loc. cit. 5:168, fig. 1; p. 171, fig. 3; idem, in Bailey, Stand. Cycl. Hort. 2:1269, fig. 1563(2); 1270, fig. 1564(2). 1915; Hume, Cult. Citrus Fruits 116, fig. 74. 1926.
      Type.—Japan (in culture) (Thunberg).   Herb. Thunberg, Bot. Mus. Uppsala.
      Distribution.—Known only in culture, doubtless native in southern China.
      Common name.—Round kumquat.
      The first fairly complete description of this species, under the name Citrus japonica, was published in 1784 by Thunberg in his Flora japonica; it reads in translation, as follows: "petiole winged, leaves acute, shrubby stem.   Japanese: Kin Kan, vulgo Fime tats banna, Kaempf., Am. ex., Fase. V. p. 801.   Growing here and there, often cultivated for its little fruits.   Stem shrubby, compressed-subangulate, erect, glabrous, scarcely a foot high.   Branches and branchlets alternate, compressed-angular, spinose, glabrous, erect, green.   Spines solitary, in the axils of the leaves, erect divergent.   Leaves few, with winged petioles, ovate, somewhat acute, entire, somewhat concave, glabrous, dark green above, paler below, erect, spreading, 1 in. [2.5 cm] long, with very minute glands ('poris').   Petioles winged.   1 line [2 mm] long.   Flowers axillary, often solitary, rarely paired, pediculate, nodding.   Pedicels glabrous, hanging down, 1 line [2 mm] long.   Parianth, 1-phyllous, green, glabrous, minute, 5-toothed.   Petals 5, white, oblong, somewhat concave, spreading, somewhat claw-like.   Filaments 19, subulate, compressed, erect, in 5 more or less coherent bundles, connate, forming a cylinder, shorter than the corolla, nearly equal in length, white.   Anthers oblong, small, yellow.   Ovary superior, subglobose, glabrous.   Style solitary, cylindrical, slightly shorter than the stamens, greenish white.   Stigma simple, globose, yellow, striate, many-locular within.   Fruit with fleshy peel, vesicular pulp, 9-locular [sic], the size of a cherry.   Differs from the other species of Citrus especially in being a very small shrub with minute fruits; thus it can scarcely be considered as a variety of orange.   It resembles Citrus medica in the axillary flowers, but differs in the winged petioles; it differs from the orange in the axillary flowers, which are solitary or paired, never in panicles.   Fruit ripens in December and January, is very sweet, agreeable and edible."
      This description was evidently drawn from a very small plant, perhaps an artificially dwarfed one, such as are commonly grown in Japan.   The very small leaves with axillary spines, and the fruits "the size of a cherry" strongly indicate that Thunberg was describing the round kumquat and not the oval kumquat (F. margarita).   His plate of this plant published in 1800 shows a flowering twig with small leaves and well-developed axillary spines.
      Hume (1903 and 1926) described this species as follows:
      "Tree similar to Nagami [F. margarita], except that it is slightly thorny, and has the leaves somewhat smaller and rounder at the apex.   Leaves oval; apex obtuse; margin crenate halfway down the length; veins slightly more conspicuous than in Nagami; borne on short rigid, inconspicuously winged petioles, 1/4 to 1/2 in. [6-13 mm] in length.   Fruit spherical or somewhat oblate, 1 to 1 1/4 inches [2.5-3 cm] in diameter; golden yellow, short stalked; calyx small; rind smooth, thin, spicy to the taste and aromatic when bruised; oil cells large; pulp sparse; juice acid; sections four to seven; seeds one to three, small, oval, greenish; cotyledons two, greenish.   Season same as Nagami."

SPECIES OF DOUBTFUL VALIDITY
      3.   Fortunella polyandra (Ridl.) Tan. Studia Citrol. 6:33. 1933. Atalantia polyandra Ridl. Fl. Malay Penin. 1:359. 1922; Fortunella swinglei Tan. Bul. Soc. Bot. France 75:714. 1928; Citrus polyandra (Ridl.) Burkill, Gard. Bul. Straits Settl. 5:219. 1931; C. swinglei Burkill ex Harms, in Engler, Die Nat. Pflanzenfam. 19a:459. 1931.
      Type.—Malay Peninsula (Ridley).   Herb. Bot. Gard., Singapore.
      Distribution.—Malay Peninsula; Southern China: Hainan Island.
      Common name.—Malayan kumquat.
      Ridley described this species as follows, under the name Atalantia polyandra: "Unarmed glabrous bush, branchlets smooth with a ridge decurrent from the petiole; leaves thinly coriaceous, unifoliolate; leaflets lanceolate, bluntly acuminate, base narrowed; main nerves slender, about 10 pairs; intermediate nerves nearly as thick, numerous, irregular, all inarching; [leaflet] 4-6" [10-15 cm] long 1.25-2.75" [3.2-7 cm] wide; petioles 0.25 to 0.75" [6-19 mm] long, winged above; flowers few, usually 2 in an axil, pedicels 0.25" [6 mm] long; calyx lobes 5, ovate-acute, short; petals 5, linear-oblong, blunt, 0.5" [11 mm] long; stamens 24, connate, in a tube, some shorter than the others, tube nearly 0.5" [11 mm] long, cylindric; anthers small-ovoid; disc fleshy, cushion-shaped, forming a gynophore; ovary on the disc, oblong, glandular, 5-celled; style short, stout; stigma oblong, clubbed, 5-ribbed; fruit globose with numerous large glands 1.75" [4.5 cm] through (not ripe)."
      In his description of Fortunella swinglei, based on a specimen from the Malay Peninsula, Tanaka stated: "fruits small, globose, 1.5 cm diam., with 5 locules; peel thin, smooth, minutely punctate; seeds ovoid, 10 X 15 mm, flattened, apex rounded, base very acute."
      A kumquat collected in May, 1922, by F. A. McClure (No. C.C.C. 9449) along a woodcutters' path in dense forest on the south slope of Five Finger Mountain, now in the National Arboretum Herbarium at Washington (C.P. Or. Herb. No. 7714), was identified by Tanaka (slip SW-316) as Fortunella swinglei.   The leaves are slender, lanceolate, acuminate, with the very tip emarginate; the petioles are slender, 12-14 mm long, and very narrowly winged above.   Fruits are not known.   It looks like the wild form from which the Malayan kumquat has been developed by culture or by hybridization.
      This species is as yet inadequately known.   The cultivated form common in the Malay Peninsula, where it is called limau pagar, or "hedge lime," has much larger fruits than any other species of Fortunella, and likewise a thinner peel.   It may possibly prove to be a limequat, i.e., a hybrid of a Fortunella and some variety of Citrus aurantifolia, but, if so, it is abnormal in having only five locules in the ovary and only two ovules in each locule.   It also has an abnormally high number of stamens, which may indicate hybridity.

Subgenus Protocitrus
      Subgenus Protocitrus Swing. Jour. Wash. Acad. Sci. 5:174. 1915.
      This subgenus differs from the subgenus Fortunella: (1) in having the ovary hypomerous (3- or 4-loculed, not 4- to 7-loculed); (2) in having in mature fruits on the inside of the ovary wall between the stalks of the pulp-vesicles a number of minute, wart-like, pale yellow, cellular masses; (3) in having the dissepiments of the fruit dry, and the peel thin and not very fleshy; (4) in having shorter, broader, more brachytic flowers; (5) in having leaves with the veins more prominent on both faces, and less pallid below.
      The two most important characters distinguishing the subgenus Protocitrus from the subgenus Fortunella are the few-loculed ovary and the dimorphic emergenzen arising from the ovary wall of the fruit, viz., ordinary pulp-vesicles and verruciform tufts of loosely aggregated, more or less colored cells.   These curious organs are analogous to the short-stalked, clavate slime glands found in Citrus, but are much more easily seen as they are not intermingled with the developing pulp-vesicles; they also persist longer, being visible in half-grown fruits.
      4.   Fortunella hindsii (Champ.) Swing. Jour. Wash. Acad. Sci. 5:175. 1915. Sclerostylis hindsii Champ. ex Benth. Hook. Jour. Bot. 3:328. 1851; Atalantia hindsii (Champ.) Oliv. ex Benth. 1861. Illus. fig. 3-32.
      Type.—Hong Kong (Hinds, no number, coll. 1841).   Herb. Kew.
      Distribution.—Hong Kong; China: Kwangtung and Chekiang provinces.
      Common name.—Hongkong wild kumquat.
      A spiny shrub or small tree; twigs slender, angled when young; leaves oval-elliptical, tapering sharply at both ends, dark green above and faintly venose, paler and not venose below; petioles winged, often emerging into the lamina of the leaf without a separative joint; flowers short, broad, not opening very widely; pistil very short; style shorter than the ovary; stigma large, cavernous; ovary with 3 or 4 locules; ovules 2 in a locule; fruits small, 1-1.5 cm diam., subglobose, bright orange or scarlet-orange when ripe (the color of a tangerine orange); pulp-vesicles very few, small, fusiform; seeds thick, oval or ovate in outline, plump, 9-11 X 7-8 X 5-6 mm, pistache green in section.
      The Hongkong wild kumquat, Fortunella hindsii, differs from the round kumquat (F. japonica) and the oval kumquat (F. margarita) in a number of morphological characters, some of them being of decided taxonomic significance in this group.   Longley (1925, p. 347) found the pollen mother cells of this species to contain eighteen instead of nine chromosomes as in all other wild-growing species of the orange subfamily.   This plant is in the tetraploid state, very probably in an autotetraploid state.
      Frost (1926, pp. 380-81) found that tetraploid forms of the commonly cultivated species of Citrus (sweet orange, tangerine and mandarin, grapefruit and lemon) show "unfavorable tree and fruit characters," such as thickened leaves, lower vigor, slowness to bloom, coarser rind, and lowered acid content in the fruits.
      Although F. hindsii has long been known to the Chinese and is not uncommonly cultivated in both China and Japan, it was not discovered by Europeans until about a century ago.   At first it was not recognized by taxonomic botanists as being closely related to the other species of kumquats (then placed in Citrus) but was considered to be a species of Atalantia (probably because of the small size of the fruit), in spite of the fact that the stamens are four times as numerous as the petals, whereas in Atalantia the stamens are only twice as numerous as the petals.
      The earliest known Chinese reference to Fortunella hindsii is found in the first monograph on citrus fruits ever published in any country, the famous Chü lu (Orange Monograph) of Han Yen-chih (1923), first published in 1178.   In the translation made by Hagerty, the paragraph about wild kumquats reads as follows: "Chin chü [golden chü orange].   The chin chü grows along the sides of mountain paths; compared with the chin kan [golden mandarin, the ordinary round kumquat], this fruit is very much smaller, but is similar in color and form…We sometimes find the tree bearing enough to fill several pint measures.   The pulp segments of the chin chü cannot be divided.   It has but one seed but the taste of its pulp is such that it cannot be eaten.   The chin chü trees are only suitable to be placed in pots on balcony railings.   Many gardeners cultivate and sell them for this purpose."
      Han Yen-chih then went on to say that this little tree is also called shan chin kan (mountain golden mandarin), meaning wild kumquat, and quoted a song about it written by Chou Mei-ch'êng, a poet of the Sung dynasty, who must have written before A.D. 1178, the year Han Yen-chih’s work was published.   A few centuries later, in the Ming dynasty (1368-1644 A.D.), we find Chinese authors using the name chin tou (golden bean) for this species.   This name is found in Min shu (Records of Fukien), compiled by Ho Chiao-yüan, who lived toward the end of the fifteenth century.   In describing the kinds of kumquats, he wrote: "there is also the chin tou…growing in the mountain forests; when preserved in honey they are very good."
      T’u Pên-tsun (1600?), an author of the Ming dynasty, in an essay, P'ing shih yüeh chih (Calendar of Garden Flowers), named clearly three kinds of kumquats: "the three little friends, the chin tou [or golden bean, Fortunella hindsii], the chin kan [or golden mandarin, F. japonica] and the chin chü [or golden orange, probably F. margarita]."   Swingle (1922 and 1929a) pointed out that this was apparently the first definite recognition of the fact that these are different kinds of kumquats, now considered to be three distinct species.
      4a.   Fortunella hindsii var. chintou Swing. Jour. Arnold Arbor. 21:130. 1940. Illus. fig. 3-33.
      Type.—Plant introduced from Japan and grown in the U.S. Plant Introduction Greenhouse near Glen Dale, Maryland (C.P.B. 907, F.P.I. No. 71241).   Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Cultivated in the warmer regions of China and Japan as an ornamental plant.   Doubtless native, or originated under culture in southeastern China.
      Common name.—Golden-bean kumquat.
      Differs from the parent species, F. hindsii, in having larger, thinner, somewhat narrower leaves, 3.5-8 X 1.5-2.5 cm, shorter, more slender spines, and larger, slightly depressed, globose fruits, 12-15 mm in diam., and in having the normal diploid number of chromosomes (9 in the gametes and 18 in the somatic cells) instead of twice as many (18 and 36), as the parent species has.   It has also distinctly smaller flowers with petals 5-6 X 2.5-4 mm instead of 6-7 X 4-5 mm, as in the tetraploid species growing alongside; blunter, much shorter calyx lobes (0.5-0.6 mm long instead of 0.8-1.2 mm, as in the species) and a somewhat narrower and evidently shorter disk than in the parent species.
      This striking dwarf kumquat is probably a cultivated variety derived from the wild species.   It is unique among the Citrus Fruit Trees in being the diploid state of a normally tetraploid wild species.   As is remarked under Citrus, autotetraploid cytonomic states of the commonly cultivated species of Citrus have been found by Frost (1925, 1926, 1938a, 1938b) to occur not uncommonly in California among seedlings of the sweet orange, tangerine, mandarin, grapefruit, and lemon.   Such autotetraploid seedlings usually show many or all of the following characteristics, as compared with the parent species: a tendency to slower growth, later bearing, and being less floriferous and less fruitful; also thicker and broader leaves and winged petioles, more thorny twigs, shorter fruits having thicker rind with larger oil glands, less juice, and larger, especially longer seeds.
      Swingle was able to observe minutely this cultivated diploid variety during a trip to Japan and China in 1926 and to bring back living plants with him to this country.   This cultivated form, called by the Chinese Chin-tou (pronounced kindzu in Japan), meaning "golden bean," has just such characters as would be expected in a diploid reversion from an autotetraploid species.   It may, however, prove to be a cultivated form of a species which still persists as a diploid plant.

THE THREE CLASSES OF FORTUNELLA HYBRIDS
      The hybrids of Fortunella are many and are both intrageneric (inside the genus) and intergeneric (between Fortunella and other genera).   The hybrid question comes up at every turn in classifying cultivated citrus fruit trees and is the chief hindrance to phylogenetic taxonomy of the species of Citrus and Fortunella.   It can be given a simpler and at the same time a more comprehensive treatment in a discussion of Fortunella than of Citrus, for the latter has many more species and many more hybrids that are often less easily recognized.
      There are at least three distinct classes of Fortunella hybrids.   (1) intrageneric hybrids between species of Fortunella; (2) bigeneric hybrids between species of Fortunella and species of Citrus, Poncirus, Microcitrus, or Eremocitrus; and (3) trigeneric hybrids.

INTRAGENERIC HYBRIDS
      Two cultivated varieties of kumquats, the Meiwa (or Chintan) kumquat of China and Japan, widely distributed, and the Changshou (longevity) kumquat growing in Chekiang and Fukien provinces of China, are probably to be classed as hybrids between two species of Fortunella.
      Meiwa kumquat.
      According to Tanaka (1922, p. 248), the Meiwa kumquat was so named in Japan because it was introduced from China during the reign of the Japanese emperor Meiwa (1764-1771).   It is usually called Chintan (golden bullet) in China and is widely grown in Chekiang Province (Hu, 1934, pp. 22-23).
      The Meiwa kumquat is figured and described in Swingle (1915c, pp. 172-74, fig. 4) and in Hume (1926, p. 117, fig. 117).   The fruits are broadly oval or sometimes subglobose, 25-35 X 25-28 mm, ordinarily with 7 segments, and usually some of them seedless.   The peel is very thick, nearly twice as thick as that of either the round or the oval kumquat, and is very sweet and good flavored.   The leaves are also thicker than those of the common round or oval kumquats and have 4-5 or even more layers of palisade cells on the upper surface instead of 2 or 3 layers.
      The nearly thornless twigs, the often nearly seedless fruits and the increased thickness of the peel, the irregularly thickened segment walls, and the thickened leaves with more layers of palisade cells are in a way an intensification of kumquat characters that may have resulted from a chance hybridization of the round and the oval kumquat, or a back-cross with a Fortunella of a Citrus-Fortunella hybrid.   The broadly oval shape of the fruits of the Meiwa kumquat is about midway between the oblong fruits of Fortunella margarita and the globose fruits of F. japonica.
      Swingle named this variety as a new species, Fortunella crassifolia (1915c, p. 172), but suggested that it might be of hybrid origin.   He later considered it to be a garden hybrid not entitled to rank as a species.
      Changshou kumquat.
      The Changshou (longevity) kumquat is a dwarf variety commonly grown as a potted plant in China, in the provinces of Wenchow, Fuchow, Chekiang, and Fukien.   A condensation of Hu's description (1934, p. 24) follows.
      Fruits obovate with the tip concave, about 3 cm long and 3 cm wide, with 8 segments.   The peel is very thin (1.5 mm) and has the same flavor and odor as that of the round kumquat, F. japonica.   The seeds are few (rarely more than 2 or 3) and polyembryonic.   This variety was fully described by Tanaka (1932c, pp. 151-52, in Japanese; and 1933a, pp. 38-40, in English).   He reported the plant as thornless, the fruits as "about 3 cm broad, and slightly over 3 cm in height, obovoid, sometimes nearly globose, but broadest at the apical part; apex abruptly and rather deeply concave…"   The peel is said to be 2.5-3 mm thick, soft, sweet and edible; segments 5-6; the pulp, "juicy, subacid, refreshing."   Seeds sometimes absent but often rather abundant if present, polyembryonic with more embryos than those of the Meiwa kumquat.
      The dwarf habit, absence of spines, the frequent seedlessness of the fruits, the concavity at the apex, and the large number of embryos in the seeds, all point to this as being a chance hybrid between two species of Fortunella.   It was named as a species, F. obovata, by Tanaka (1933a, p. 38), but specific names for forms suspected of being garden hybrids or mutations are of doubtful taxonomic validity and should not be accepted if proof is lacking that the plants are not mere garden hybrids, propagated only by man's help.
      It is probable that other kumquat hybrids exist in China, since differences in the articulation of the petioles with the leaf blades are observable in herbarium specimens of Fortunella collected in China.

BIGENERIC HYBRIDS OF FORTUNELLA WITH CITRUS AND PONCIRUS
      The citrus experts of the Agricultural Research Service, using adequate precautions to exclude all stray pollen, have made a number of bigeneric hybrids of Fortunella.   Apparently, some such hybrids have arisen without human help in the dooryard or village groves of mixed genera, species, and varieties of citrus fruit trees.   It is probable that some of these chance seedlings found in the Far East are second-generation kumquat-citrus hybrids or else backcrosses of an F1 Fortunella X Citrus hybrid on one of the parental species.
      Limequats [Citrus aurantifolia X Fortunella sp.].
      These hybrids have been described by Swingle (1914-1917, vol. 4, p. 1882, fig. 2176); by Swingle and Robinson (1923, pp. 235-38, pls. 4, 5); by Hume (1926, pp. 134-35, fig. 88); and by Hodgson (see chap. 4, this work).
      Most of the named varieties of limequats, Eustis, Lakeland, etc., are hybrids of Fortunella japonica with Citrus aurantifolia 'Mexican,' the round kumquat being the pollen parent.   Another named variety of limequat, the Tavares, has the oval kumquat as the pollen parent (Citrus aurantifolia 'Mexican' X Fortunella margarita).
      Orangequats [Citrus reticulata c. satsuma X (Fortunella japonica X F. margarita 'Meiwa’)].
      The orangequat is a complex, artificial, safeguarded hybrid between the satsuma orange and the Meiwa kumquat (the latter, the pollen parent, is here considered a hybrid of the round and the oval kumquat).   A named variety of this hybrid, the Nippon orangequat, was described in Robinson and Savage (1934, p. 11).   The fruit is broadly oval or obovate in shape, 3.8 to 5.0 cm in diameter by 5 to 6.3 cm in length, with thick, spongy peel of mild flavor and with acid pulp.
      Citrumquats [Poncirus trifoliata X Fortunella japonica].
      The citrumquat proved an exceedingly difficult hybrid to make.   For several years no viable seeds were obtained.   When at last a few were secured, the young seedlings all died promptly.   Finally, in 1909, Swingle found a few belated "June blooms" on a Poncirus trifoliata tree, nearly all of which set fruit when pollinated with pollen from Fortunella japonica.   From these cross-pollinated trifoliate oranges nearly a hundred seeds germinated.   Many of the seedlings died while still very young but a few lived and a very few made healthy, vigorous plants that were obviously hybrid plants and not nucellar seedlings of the female parent.   (See fig. 3-34.)
      This hybrid looks much like a citrange in twigs and foliage.
      The citrumquat is a combination of two citrus fruit trees possessing to the highest degree two very different kinds of hardiness.   Poncirus trifoliata is a deciduous tree able to endure winter temperatures as low as -23° C or -25° C (-10° F to -13° F), whereas the kumquat, with its highly developed winter dormancy, is able to withstand long spells of warm weather in winter or early spring without putting out tender new growth easily killed by temperatures of even only a few degrees below the freezing point.   This winter dormancy of the kumquat was noted by Swingle and Robinson (1923, p. 230) in their discussion of citrangequats.
      The extreme difficulty experienced in getting healthy hybrids between Fortunella and Poncirus is further proof of the remote degree of relationship of these genera.   However, failure to get viable hybrids on the first trial should not be taken as conclusive proof that such hybrids cannot be obtained if the parent plants (particularly the seed parent) are in proper condition when cross-pollinated.
      Calamondin (orangequat?) [Citrus reticulata var. austera ? X Fortunella sp. ?].
      This Chinese citrus fruit tree, illustrated in figure 3-35, widely cultivated in the Philippines and also grown in Hawaii and the United States, is very probably an orangequat that arose in China by insect cross-pollination of a sour, loose-skinned mandarin orange and a kumquat, perhaps Fortunella margarita.   In the Philippines, it is known under the name Calamonding.   This hybrid was described under the name Citrus mitis Blanco by Swingle (1914-1917, vol. 2, p. 784) and by Hume (1926, pp. 133-34, fig. 87) and under the name Citrus microcarpa Bunge by Tanaka (1933b, p. 184).   It was figured in Ochse (1931, pp. 131-32, col. pl. 50).   It is reasonably certain that the Calamondin is a Citrus X Fortunella hybrid and should not be considered as a valid species (see Swingle, 1942, p. 26).
      This hybrid has depressed-globose fruits with a very thin peel, that becomes loose as the fruit ripens, and with intensely acid pulp.   The segments number only seven to ten.   It may possibly turn out to be a back-cross of an F1 Citrus X Fortunella hybrid on Citrus reticulata.   This could probably be determined by experimental hybridization.   This hybrid enters into the parentage of the interesting trigeneric hybrid, the Altamaha or Glen citrangedin.
      Tetraploid Calamondin.
      This interesting form of the Calamondin was discovered in Formosa by N. Nakamura, one of Tanaka's graduate students.   He published (1934) an account of his examination of a large number of cultivated forms of Citrus, including at the end of the list several forms of the Calamondin and two other hybrids.   In all, he examined seventy-six species or cultivated varieties of Citrus and four citrus hybrids.   Only one of these varieties (the one now in question) showed the tetraploid number of chromosomes.   He called this variety "Shikinari-mikan" and considered it to be a tetraploid form of the Calamondin.   He noted that the pollen mother cells show 18 bivalent chromosomes in the first division of the pollen mother cell, and 18 monovalent ones in the second division (Nakamura, 1934, p. 169, figs. 9, 10).   He stated that the characters of this variety correspond in general to those of the diploid form of Calamondin, commonly cultivated in Japan; however, the tetraploid variety has "fruits…larger in size and sweeter, and…leaves…much thicker than those of the diploid individuals.…Its pollen grains are also larger than those of the diploid," but more of them (about 30 per cent) are sterile.
      Unlike many tetraploid forms of Citrus, this tetraploid Calamondin grows as vigorously as the diploid form, or even more vigorously.

COMPLEX BIGENERIC HYBRIDS
      Procimequat [(Citrus aurantifolia 'Mexican’ X Fortunella japonica) X Fortunella hindsii].
      This very interesting complex hybrid, which has a triploid chromosome number, is the result of a carefully safeguarded cross-pollination of the Eustis limequat with Fortunella hindsii, a tetraploid species made by Eugene May and the writer expressly to obtain a triploid hybrid.   Longley (1926, pp. 543-45, fig. 1) found it to be triploid, with 27 chromosomes in the somatic cells (18 supplied by the male parent, the Hongkong wild kumquat, and nine by the limequat).
      The limequat fruits have from six to nine segments, as might be expected from a hybrid of the round kumquat (with four to seven segments) with the Mexican lime (with 10 to 12 segments).   The Hongkong wild kumquat fruits have only three or four segments.   The ovaries of the procimequat hybrid under consideration usually show from four to five segments.
      The leaves of these hybrids are small but some of them show fairly vigorous growth (see fig. 3-36).   The fruits set abundantly even on small young plants and are small and subglobose, much like those of Fortunella hindsii but a little larger and a much paler orange in color when ripe.   These fruits are not seedless, as was expected, but produce some nucellar bud embryos, as do many citranges after the development of the ovules has been stimulated by pollination.   Triploid limes are usually seedless.
      This hybrid is interesting because it throws light on bigeneric Fortunella X Citrus back-crosses such as are possibly represented by the Malayan hedge lime discussed.   The procimequat is in reality intermediate between a true bigeneric back-cross and a trigeneric hybrid, because Fortunella hindsii belongs to a subgenus, Protocitrus, with many important taxonomic characters separating it from the true Fortunella species placed in the subgenus Fortunella.
      The name "procimequat" (given here for convenience) is derived from Pro[to]c[itrus X L]imequat.

TRIGENERIC HYBRIDS OF FORTUNELLA
      Several remarkable trigeneric hybrids having Fortunella as one parent have been made by the citrus experts of the former Bureau of Plant Industry.   These hybrids are of great interest because of the light they throw on the origin of strange complex hybrids that arise without human aid in the dooryard citrus groves of China, Japan, and India.
      Citrangequats [Fortunella sp. X (Citrus sinensis X Poncirus trifoliata)].
      These were the first trigeneric hybrids to be produced artificially from definitely known parents.   Swingle succeeded in making this cross in 1909 by using pollen from Willits and Rusk citranges (Poncirus trifoliata X Citrus sinensis) on properly safeguarded flowers of the oval kumquat (Fortunella margarita) and the round kumquat (F. japonica).   Several of these citrangequats were described and illustrated by Swingle and Robinson (1923, pp. 230-33, pls. 1, 2, 3).   The Thomasville, Sinton, and Telfair citrangequats, described and figured by Hume (1926, pp. 40-42, fig. 45), are the best known of these hybrids.
      Citrangequats vary greatly in size, color, flavor, etc.   The Thomasville citrangequat has obovoid fruits borne on long pedicels, 1.5 to 3 cm long and 2 to 2.5 mm in diameter at the base, but swollen and pulvinoid at the top, 8 to 10 mm, and permanently curving as the fruit matures until it hangs down, making an angle of 45° to 90° with the base of the pedicel.   This character is unknown in any of the three parent species and shows how taxonomically new characters can arise in complex hybrids (see further discussion under Poncirus hybrids).
      Citrangedins [(Fortunella sp. ? X Citrus reticulata var. austera ?) X (Citrus sinensis X Poncirus trifoliata)].
      This highly complex trigeneric and 4-specific hybrid is the result of a cross of the Calamondin, itself a hybrid of Fortunella and Citrus, with the Willits citrange, a hybrid of Poncirus and Citrus.   It bears small, subglobose fruits of a brilliant orange-yellow color.   The fruit was described and figured by Swingle, Robinson, and Savage (1931, pp. 14-15, pl. 9) and the plant’s extreme resistance to cold was noted in Swingle, Robinson, and Savage (1932, p. 7).   The 4-specific hybrid often shows one- or two-lobed leaves on young vigorous shoots, and sometimes difoliolate or trifoliolate leaves.   The leaves on fruiting twigs are almost always unifoliolate and look very like those of a Fortunella-Citrus hybrid.   This hybrid shows clear traces of all three genera that enter into its parentage but is strikingly different from any species of Fortunella, Poncirus, or Citrus.
      If its parentage were not known, the Altamaha or Glen citrangedin might even be considered a new genus, as well as a new species, by those who name "chance seedlings" of obvious hybrid ancestry as good species!
      Faustrimedins [Microcitrus australasica X (Fortunella sp. X Citrus reticulata ‘Calamondin’)].
      The faustrimedin is a hybrid of the Australian finger-lime with the Calamondin, itself a hybrid between the kumquat and some variety of orange of the mandarin group (Citrus reticulata).   It is thus a trigeneric hybrid of Citrus, Fortunella and Microcitrus.   The faustrimedin has small leaves much like those of the Australian finger-lime and is hardy like the Calamondin.
      This hybrid, first made by Oliver (1911, p. 40, pl. 1, fig. 2), may be of use in obtaining still more complex hybrids containing the blood of Eremocitrus or Poncirus.   It is a vigorous grower and shows some promise as a rootstock.   It is easily propagated from cuttings.   The fruits are like finger-limes but are shorter, nearly seedless, and pleasantly acid.

XXII.   Eremocitrus Swingle
      XXII.   Eremocitrus Swingle, Jour. Agr. Res. 2:86. 1914.
      Type species.—Triphasia glauca Lindl. = Eremocitrus glauca (Lindl.) Swing.
      Distribution.—Australia: central and southern Queensland; northern New South Wales.   From near the coast at Broad Sound, Queensland, as far west in scattered colonies as Long. 142° E. in New South Wales, or possibly Long. 141° E. in Queensland.
      Common name.—Australian desert lime.
      Leaves gray-green, thick and leathery with appressed grayish hairs on both surfaces; spines solitary in the axils of the leaves, sometimes lacking on older branches; flowers single or few in the axils of the leaves, 3-5-merous, petals narrowed at the base, stamens free, 4 times the number of the petals, ovary with 3-5 locules, 2 ovules in each locule, style short; fruit oval or pyriform with a fleshy peel (as in Citrus), having oil glands, pulp acid, with stalked subglobose pulp-vesicles, seeds small with hard wrinkled testa; cotyledons hypogeous, first leaves cataphylls.
      This genus is in many ways the best characterized and most distinct of any of the near relatives of Citrus.   It is the only pronounced xerophytic plant in the whole orange subfamily.   Both leaves and twigs are gray-green, like sagebrush; the leaves are small, thick, bifacial with palisade parenchyma on both faces, and covered with a thick cuticle.   The stomata are deeply sunken in pit-like depressions in the surfaces of both leaves and young twigs.   There is a somewhat persistent sparse coating of thick-walled appressed hairs on the leaves and twigs not present in any other citrus fruit tree.   In the event of severe drouth, the leaves fall and the gray-green twigs carry on the very restricted photosynthesis possible under such conditions.   The young plants have greatly reduced linear leaves that are, in fact, cataphylls.   The young seedlings grow very slowly until an extensive root system is formed; then, if adequate moisture is found, the stem grows upward as rapidly as that of any citrus tree and little by little larger leaves are formed until some may measure 50 to 65 by 10 to 20 mm and nearly 2 mm thick!   The young plants have single sharp, stiff spines in the axils of the leaves.   These are usually 2 to 3 cm long.   On certain particularly vigorous wild trees growing in central New South Wales, spines of enormous length have been produced.   They are very stout, somewhat up-curved, and reach a length of 6 to 9 cm with a vertical diameter at the base of 5 to 7 mm and a transverse diameter of 3 to 4 mm.   The flowers are much like those of Microcitrus but smaller, usually 3- to 5-merous with six to ten entirely free stamens with slender filaments.   The ovary has three to five locules with two ovules in each locale, a character not found in any other True Citrus Fruit Tree except Fortunella!   The fruit is small, subglobose, ovoid or obovoid, 15 to 25 by 12 to 15 mm.   The subglobose pulp-vesicles have short, slender stalks and are loosely aggregated, not cohering as tightly to each other as in the common citrus fruits.   The seeds are small and contain only a single embryo.
      This remarkable xerophytic citrus fruit tree grows wild in central and northern New South Wales and in southeastern Queensland.   In New South Wales, it is sometimes found in semiarid regions even as far west as the Mootwingee Range (Lat. 31° 45' S., Long. 142° 10' E.), about 80 miles northeast of Broken Hill in extreme western New South Wales.
      Eremocitrus glauca (Lindl.) Swing. Jour. Agr. Res. 2:88. 1914. Triphasia glauca Lindl. in Mitchell, Jour. Exped. Trop. Austr. 353. 1848; Atalantia glauca (Lindl.) Benth. Fl. Austr. 1:370. 1870. Illus. Fairchild, U.S. Dept. Agr. Yearbook 1911:pl. 45, fig. 1. 1912; Swingle, loc. cit. text figs. 1-7; idem, in Engler, Die Nat. Pflanzenfam. 19a:332, fig. 153,a-o. 1931; Maiden, Forest Fl. N. S. Wales 7:245, pl. 248, figs. A-H. 1921; Flanders, Calif. Citrog. 17:278, 6 unnumbered half-tone text pls. 1932; fig. 3-37 this work.
      Type.—Australia, Queensland, Dublin County, near junction of Maranoa and Merivale rivers (Mitchell, 1846).   Kew, Brit. Mus., and Gray Herb. (Swingle, 1914b, p. 90).
      Distribution.—Australia: central and southern Queensland; northern New South Wales.   From the Burdekin River (Lat. 20° S.) in northeastern Queensland to the coast at Broad Sound (Lat. 23° S.), thence south along Long. 152° E. to the New England district in northeastern New South Wales, thence southwest to Dubbo in central New South Wales (Lat. 32° 20' S., Long. 149° 35' E.), thence westward (rather common) to Long. 147° E., or even 146° E., and also occurring in scattered groups much farther west in the hills in the semiarid region as far as Long. 142° E. in New South Wales and possibly Long. 142° E. or even 141° E. in Queensland.
      Common name.—Australian desert lime.
      The typical form of this species as it occurs in southeastern Queensland, Australia, is a small tree or a large shrub, sometimes growing as a thorny bush only a few feet in height.   F. M. Bailey described it as "a rigid glaucous shrub of 2 or 3 ft., often armed with straight or incurved axillary spurs of 1/2 in. or under…"   However, Flanders (1932, p. 278) figured a giant tree at Chinchilla, Layton County, in Queensland, some 130 miles west-northwest of Brisbane, that had a single trunk, about 25 feet in height and measuring 20 inches in diameter at 2 feet above the ground.   The round crown was about 28 feet across.   Such a tree undoubtedly grew where its roots had access to ground water.
      The leaves are gray-green, very thick and leathery, usually oblong, linear- or elongate-obcuneate, obtuse at the apex, often emarginate, 25-40 X 4-10 mm.   Both surfaces show appressed, grayish hairs and have a thick cuticle with sunken stomata.   The leaves are more or less paraheliotropic (on the edge to the light) and have no clearly defined upper and lower surfaces (Swingle, 1914b, figs. 5, 6, 7).   Flowers very small, 3-5 mm long and 5-8 or 10 mm across, borne singly or in groups of 2 or 3 in the axils of the leaves on slender pedicels, 4-6 mm long, flowers 3-5-merous, with 4 (sometimes 3) times as many stamens as petals, stamens short, filaments slender, 4-5 mm long (sometimes slightly united at the base according to Bailey); pistil short, 3-4 mm long, ovary ovate with 3-5 locules; fruits subglobose or obovoid, 7-12 X 8-10 mm, pulp-vesicles small, subglobose, on short, slender stalks and nearly free from one another; seeds small, oval, 5-6 X 3-4 X 2.5-3 mm, testa rough, monoembryonic.   Young plants growing in the greenhouse at Washington, D.C., had very slender, straight or slightly upcurved thorns, 25-35 mm long but only 1-1.5 mm diam.; young trees growing out of doors at Riverside and Indio, California, show much stouter spines, 2-3 mm diam., but seldom as long as 35 mm.   The twigs on young plants are usually very slender, 1-2 mm diam., and for several years bear very narrow, slender leaves or cataphylls, 18-30 X 1-2 mm, and make very slow growth; as soon as an extensive root system has developed, the young tree grows upward rapidly, producing larger leaves on slightly stouter twigs, 2-3 mm diam., that often show only very short spines or none.
      The trees growing at Indio under irrigation began to flower and set fruit when they were 12-15 feet high and about 8-10 years old.   At that age the branches were comparatively short and the trees compact with but little spread (at 20 years of age one tree was 20-22 feet high and the top spread 12-16 feet).   The flowers appear very abundantly in March, and by late May or early June the fruits begin to drop while still green or slightly yellowish-green.   The fruits are juicy, thin-skinned, and rather pleasantly acid.   Two forms occur: one strain has subglobose fruits, the other obovoid or short-ovoid fruits.
      The Australian desert lime, Eremocitrus glauca, is a pronounced xerophyte, the only one found thus far in the orange subfamily (fig. 3-38).   It is able to withstand severe drouth and hot dry winds.   Under such conditions the leaves fall off and the leafless gray-green twigs (resembling those of the smoke tree, Dalea spinosa, and the paloverdes, common in the arid regions of the southwestern United States and northern Mexico) carry on photosynthesis on a reduced scale.   The seedling develops an enormous root system before making vigorous aerial growth and developing full-sized leaves (young seedlings bear only very slender cataphylls).   The roots are able to endure rather high concentrations of salts in the soil moisture and are much less susceptible to boron poisoning than are those of Citrus.   When in a dormant condition this species is able to withstand, without injury, temperatures of ten or more degrees below zero Fahrenheit (-24° C or lower).   It probably possesses a higher zero point of growth and, in consequence, a greater dormancy in late winter and early spring than Citrus.   It apparently ranks next to, but does not equal, the kumquats in possessing both winter dormancy and resistance to cold.
      There is good reason to believe that Eremocitrus has spread slowly westward in northeastern Australia over a long period of time, probably some 20 million years or more, during which time this plant has become adapted to withstand a semiarid climate and to grow in soils having an appreciable saline content in the soil moisture.   Trees of Eremocitrus glauca have been found growing wild in northeastern Australia up to east longitude 142° 10’, a distance inland of 800 to 960 kilometers (500 to 600 miles).   The western half of this range falls in a region classed as "Dry Steppe" by the Australian experts Andrews and Maze (1933), whose maps show that Eremocitrus grows wild as far west as the climatic zone which according to their designation has an average of from six to eight arid months.   Their No. 8 line (indicating eight arid months in the year), which separates the steppe climate from the desert climate in Australia, runs through the westernmost limit of the range of Eremocitrus.   Convincing proof of the drought resistance of Eremocitrus is found in certain thrifty groves of the desert lime growing in regions in Australia where salt bushes thrive and where trees of some other species die during prolonged drought before the Eremocitrus trees show even slight injury.16
      In spite of years of effort on the part of several generous Australian cooperators, the attempt to obtain seed from the Eremocitrus trees growing wild in the drier parts of Queensland and New South Wales proved unsuccessful up to 1941.   A few mature fruits were found west of longitude 148°, but they contained no viable seeds.   Fortunately, in December, 1941, C. T. White collected ripe fruits of Eremocitrus glauca containing viable seeds in the vicinity of Charleville, Queensland, in the Warrego River Valley (Long. 147° 40' to 145° 30' E.).   The trees from which the seeds were obtained were growing under semiarid conditions.   These seeds germinated and the seedlings grew well in the greenhouses of the U.S. Department of Agriculture near Washington, D.C.   The complete absence of seed in fruit of this species growing in its natural habitat is apparently the rule, according to information supplied in a letter dated December 2, 1964 to Dr. J. R. Furr, U.S. Date and Citrus Station, Indio, California, from E. C. Levitt, New South Wales Department of Agriculture, Sidney, Australia.   As recently as 1964, the E. glauca trees growing at the U.S. Date and Citrus Station have been the source of seed supplied to Australian citrus specialists for their studies on the suitability of E. glauca as a rootstock and its tolerance to salinity.
      Eremocitrus glauca can serve as a rootstock for Citrus aurantium and other species of Citrus; when used for the sour orange excellent growth was obtained in the Washington greenhouses.   Eremocitrus also grafts readily on Citrus.

BIGENERIC AND TRIGENERIC HYBRIDS OF EREMOCITRUS
      Eremocitrus trees growing near a collection of Citrus species and hybrids in the U.S. Date and Citrus Station at Indio, California, have fruited heavily (fig. 3-39).   A small percentage of the seedlings from them have proved to be hybrids.   In all, several score of such hybrids have been found and many of them propagated from cuttings (fig. 3-40).   Efforts to pollinate castrated and bagged Eremocitrus flowers from Citrus or other genera have not so far succeeded, probably because of the great profusion of blossoms produced by Eremocitrus, of which doubtless only one in a hundred or so set fruit.
      Swingle believed that two kinds of Eremocitrus hybrids are easily distinguished: bigeneric hybrids, in which a Citrus species is the pollen parent, and trigeneric hybrids, which have a citrange for the pollen parent.
      In the revision of this chapter, the junior author has maintained Swingle's classification of bigeneric and trigeneric hybrids.   However, it should be noted that Dr. J. R. Furr, U.S. Date and Citrus Station, Indio, California, has reservations concerning such classification.   In a letter to Dr. Walter Reuther, the editor of this volume, dated June 7, 1965, he stated:
      I have some reservations regarding Swingle's supposed eremolemon, eremorange, eremoradias, etc.   As well as I can remember the seedlings of the old E. glauca tree that used to grow at the U.S. Date and Citrus Station looked about the same after the Meyer lemon tree growing very close to it was removed as they did before its removal; but there were grapefruit trees nearby all of the time, so it is possible that we were getting E. glauca X grapefruit hybrids.
      I have been puzzled by the E. glauca seedlings I have grown.   Usually there is a mixture of types in the open-pollinated seedlings grown from our E. glauca trees ranging from vigorous, upright plants with lanceolate leaves to very small, weak plants with thorny branches and only vestigial leaves.   The weak ones usually die in the flats before they are large enough to plant in the field.   I supposed that the weak thorny plants were possibly selfed E. glauca seedlings and that the vigorous ones were probably hybrids with citrus.   We have never caged a tree of E. glauca to see if it will produce any seed without cross-pollination.   One year I did pollinate flowers of E. glauca with several different kinds of citrus.   On different branches of E. glauca, flowers were pollinated by hand with pollen from sweet orange, pummelo, rough lemon, and Shekwasha.   No seed [sic] were obtained from flowers pollinated by pummelo.   Seedlings that appear to be hybrids were obtained from the other crosses; that is, the seedlings were highly variable and there seemed to be family differences between the different lots of seedlings of supposedly different pollen parents.   In these supposed crosses, the E. glauca flowers were not emasculated; the pollen was just daubed on the stigmas of many open flowers, which may have been self-pollinated at the same time.
      I think Swingle was probably right in supposing that many of the E. glauca seedlings grown from the original old E. glauca specimen at the U.S. Date and Citrus Station were hybrids.   I doubt that he had adequate evidence to justify his assumption that he could tell what the hybrids were.   Bees do work E. glauca vigorously, and if trees of E. glauca and the citrus species were quite close together, I should think there would be some cross-pollination.
      At the same time, Dr. Furr notes in closing his discussion that: "Swingle probably grew or had grown by his subordinates more seedlings of E. glauca than anyone ever grew before or since his day."

BIGENERIC HYBRIDS
      Eremolemons [Eremocitrus glauca X Citrus limon 'Meyer lemon'].
      All the bigeneric hybrids between Eremocitrus and Citrus have unifoliolate leaves, intermediate in size between the two parental species but thin and bifacial like those of Citrus.   Many years ago, all citrus trees except a Meyer lemon were removed from the vicinity of a fruitful Australian desert lime in the Coachella Valley at the U.S. Date and Citrus Station.   Desert conditions produced about one-fifth hybrids, probably with Meyer lemons as the pollen parent.   They grew vigorously, and when 8 to 10 feet high flowered and fruited rather abundantly, producing subglobose, slightly oblate, yellowish fruits, 18 to 25 mm in diameter, which have five or six segments and usually one to three seeds.   Upon germination, these seeds grew rapidly and produced plants which looked exactly like the original hybrid.   Doubtless these eremolemon seeds contained only nucellar-bud embryos which reproduced the mother F1 hybrid exactly.   This Coachella eremolemon, as it is called, has a high tolerance for boron and doubtless a greater tolerance for salinity in the soil moisture than any species of Citrus.   Its seedlings are remarkably uniform, grow rapidly when young, take Citrus buds readily, and have taproots like the Citrus rootstock used by nurserymen.
      Eremoranges [Eremocitrus glauca X Citrus sinensis].
      Doubtless some of the many hybrids grown from seed from the Eremocitrus trees at the U.S. Date and Citrus Station were of this parentage.
      Eremoradias [Eremocitrus glauca X Citrus aurantium].
      Perhaps a few hybrids developed at the U.S. Date and Citrus Station were of this parentage, as a sour orange tree grew only a dozen yards away from one of the most fruitful trees of Eremocitrus glauca.

TRIGENERIC HYBRIIDS
      Citrangeremos [Eremocitrus glauca X (Poncirus trifoliata X Citrus sinensis)].
      These hybrids showed trifoliolate or difoliolate leaves and usually also some unifoliolate leaves.   They were undoubtedly the result of the pollination of Eremocitrus flowers from a row of citrange trees (Savage, Morton, and Rustic) growing nearby in the U.S. Date and Citrus Station at Indio.   Many of these hybrids had very narrow leaflets, which perhaps was the reason they were weak and of slow growth.
      Both the bigeneric and the trigeneric hybrids of Eremocitrus are very unlike any known citrus fruit trees, but evidently they cannot properly be named as new genera and new species.   If it proves possible to cross-pollinate successfully a citrangeremo [= Eremocitrus X (Poncirus X Citrus)] with another trigeneric hybrid already existing, the faustrimedin [= Microcitrus X (Fortunella X Citrus)], no fewer than five genera of True Citrus Fruit Trees will have been combined in the resulting hybrid, viz., Citrus, Fortunella, Microcitrus, Eremocitrus, and Poncirus!   Naturally such ultra-complex hybrids, if obtained, would not be recognized by taxonomists as constituting true genera or true species.

XXIII.   Poncirus Raf.
      XXIII.   Poncirus Raf. Sylva Tellur. Mant. 143. 1838. Pseudaegle Miq. Ann. Mus. Bot. Lugd.-Bat. 2:83. 1865-1866.
      Type species (monotype).—Citrus trifoliata L. = Poncirus trifoliata (L.) Raf.
      Distribution.—Central and northern China; widely cultivated.
      Common name.—Trifoliate orange.
      Small trees with single, stout, axillary spines and palmately 3-foliolate, deciduous leaves, twigs dimorphic; (a) normal twigs with internodes as long or longer than the petioles, (b) foliage spurs, which develop from dormant buds on twigs of previous year, with extremely short internodes (less than 0.5 mm long) bearing 1-5 normal foliage leaves but no spines; flowers developing singly at the nodes in spring from scale-covered flower buds formed early in the previous summer on last year's twigs that lose their leaves during the winter; sepals 5, petals 5, long, slender, spathulate or clawed; stamens 4 or more times the number of petals, with free, glabrous, slender filaments; ovary subglobose, pubescent, seated on a short, shallow, cup-shaped disk broader than the ovary; ovary with 6-8 (sometimes fewer or more), usually 7 locules; ovules 4-8 in 2 rows in each locule; style short, thick, merging into the slightly clavate stigma; fruits globose or short-pyriform, subsessile, 3-5 cm diam., finely pubescent, with lemon-yellow, soft, orange-like peel with abundant oil glands; pulp-vesicles elongate-conical, slender-stalked, bearing scattered, secretory, hair-like organs, and filled with acid pulp containing numerous droplets of acrid oil (or oily wax) in the center; seeds oval, plump, often with several supernumerary nucellar embryos; germination hypogeous; young seedlings developing at first bract-like cataphylls, then intermediate forms that soon merge into normal 3-foliolate leaves.
      This remarkable genus, although evidently closely related to Citrus, Fortunella, Microcitrus, and Eremocitrus, has many strikingly aberrant characters.   In the first place, it differs from all the other True Citrus Fruit Trees, which are found in only tropical or subtropical regions, in having penetrated far into the temperate zone in northeastern Asia; in so doing it has become a deciduous tree with small leaf buds and larger scale-covered flower buds (formed in early summer) that pass the winter on the leafless terminal twigs and open before (and sometimes with) the leaves early in the following spring.   The trifoliolate leaves are doubtless a survival of the foliage of some remote ancestral plant.   All the other True Citrus Fruit Trees have unifoliolate leaves (except Clymenia, which has simple leaves) that persist on the tree for two or more years and cannot endure severe cold.
      The pith of the stem shows transverse plates composed of thick-walled cells, not as yet found in any other genus related to Citrus (see Swingle, 1909).   The pulp-vesicles carry scattered hair-like organs that bear at their expanded tips a number of rounded, thick-walled cells with numerous small, oblique fissures.   These organs secrete a viscous fluid that allows the pulp-vesicles to slip past one another.   These organs are unknown in any other genus of the orange subfamily.   The immature fruits contain ponciridin, a glucoside analogous to hesperidin, but not found in Citrus.
      Poncirus hybridizes freely with Citrus.   Such hybrids, called citranges, citrandarins, etc., are usually ovule-sterile but occasionally produce some fertile pollen grains.   Poncirus has also been hybridized (after overcoming many difficulties) with Fortunella and citranges have been hybridized with Eremocitrus.
      The pulp-vesicles of Poncirus contain numerous droplets of oil.   Penzig (1887, pp. 161-62) discovered these inclusions and found that some of these oil globules were semisolid and more or less brittle.   Oil droplets are also found in the pulp-vesicles of Citrus, being very abundant in the species of Citrus which belong to the subgenus Papeda.
      Poncirus is not easy to place in a phylogenetic series along with Citrus and other related genera arranged so as to show the course of evolutionary progress.   Doubtless many "missing links" between it and the ancestors of Citrus have perished, leaving Poncirus as perhaps the most isolated and aberrant genus of all the True Citrus Fruit Trees.
      Poncirus trifoliata (L.) Raf. Sylva Tellur. Mant. 143. 1838. Citrus trifoliata L. Sp. Pl. ed. 2. 1101. 1763; C. trifolia Thunb. Fl. Japon. 294. 1784; C. triptera André (non Desf.), Rev. Hort. 57:518. 1885; Aegle ? sepiaria DC. 1824; Pseudaegle sepiaria (DC.) Miq. Ann. Mus. Bot. Lugd.-Bat. 2:83. 1865-1866; Limonia ? trichocarpa Hance, 1882. Illus. Kaempfer, Amoen. Ext. 5:801. 1712; Hooker, Bot. Mag. pl. 6513. 1880; Nicholson, Gard. Chron. 3 ser. 27:269. 1900; Penzig, Studi Bot. Sugli Agrumi, Atl., pls. 13, 14. 1887; André, Rev. Hort., 57:pl. facing p. 516 (col.). 1885; Georgeson, Amer. Gard. 12:525. 1891; Swingle, U.S. Dept. Agr. Bur. Pl. Ind. Circ. 46:figs. 1-6. 1909; idem, in Bailey, Stand. Cycl. Hort. 5:2751, 2752, figs. 3123-25. 1916; Tillson & Bamford, Amer. Jour. Bot. 25:784, figs. 29-33. 1938; fig. 3-41 this work.
      Type.—Based on Kaempfer's description and plate.
      Distribution.—Central and northern China; widely cultivated in Japan, Europe, and North America.
      Common name.—Trifoliate orange.
      Small, much-branched tree, twigs of rapid growth, strongly angled, becoming cylindrical with age; spines single, stout, straight, sharp, oval in cross section (vertical axis longer); twigs dimorphic: (a) normal twigs with internodes 1-5 or even 8 cm long with a single leaf at each node, in the axil of which is a bud and often a stout spine; (b) foliage spurs, arising from the nodes of the twigs of the previous year's growth (but above the spine and 5-8 mm above the scar of the fallen leaf in the axil of which it was laid down as a bud), with 1-5 (usually 2-4) excessively short nodes (less than 0.5 mm long), each bearing a normal foliage leaf but no spine, tipped with a bud 1-2 mm wide, covered with green, overlapping bud scales with more or less ciliate margins (such foliage spurs may persist for two [or more?] years developing 2-4 or more leaves each year); leaves palmately 3-foliolate; terminal leaflet 3-6 X 1.5-2.5 cm, shallowly crenulate, serrate above the middle, cuneate at the base, midrib slightly raised on both surfaces and gradually expanding below near the base for 2 or 3 mm into a pulvinoid swelling entirely included in the leaf blade and not free except at the extreme base (i.e., no petiolule), lateral leaflets 2.5-5 X 1-2 cm, sessile, lower half wider, petioles 1-2.5 cm, wing 2-3 mm at broadest part, tapering to the base; leaflets on leaf spurs very similar but with petioles slightly longer in proportion to the length of the middle leaflet; spines stout, straight, acute, 1-6 cm long, 3-10 mm diam. at base, 2-4 mm diam. in the middle, oval in cross section (vertical diam. nearly twice the transverse in the lower third of the spine); flowers produced in spring on twigs of previous year's growth from subglobose flower buds, with bud scales usually ciliate on the margins, flowers single, nearly sessile, 5- (rarely 4- or 6-) merous; calyx composed of 5 small, nearly free, ovate, cucullate, persistent sepals, 8-10 X 3-4 mm; corolla composed of 5 white petals, 18-30 X 8-15 mm, flat when fully open, subspathulate, often narrowed into a claw at the base, soon deciduous; stamens 20 or more (up to 60); filaments free, unequal in length, slender, tapering gradually toward the tips; ovary subglobose, 2 mm diam., pubescent, with 6-8 (usually 7) locules; ovules numerous, in 2 rows in each locule; style short, 1-1.5 mm long, merging into a rather large, subclavate stigma; fruits almost sessile, globose, ovoid or slightly pyriform (sometimes with a blunt nipple at the tip set in a depression surrounded by a ring-shaped furrow), 3-5 cm diam., dull lemon-colored and fragrant when ripe, finely and densely pubescent, peel 5-10 mm thick, with numerous oil glands, rather rough; seeds ovoid, plump, very numerous, leaving very little space for the pulp, which is composed of slender-stalked, narrow, elongate-fusiform pulp-vesicles filled with very acid juice and having a slender column of numerous small droplets of acrid oil in the center.   The pulp-vesicles bearing hair-like appendages, expanded at the tips; these usually show scattered thick-walled cells having numerous small slanting fissures, which secrete a viscous fluid.
      The trifoliolate orange shows surprisingly few variations considering that it has been grown in China for thousands of years and in Japan since at least the eighth century.   Palmately 4- or 5-foliolate leaves are seen very rarely on old trees.   In the parks of Washington, D.C., a small-flowered form occurs that has shorter petals than the normal form and these are occasionally seen bearing the rudiments of anthers on their margins.   In other words, such flowers show partial staminody of the petals.
      The curious dwarfed variety with very narrow, almost threadlike leaflets, called hiryô in Japan, where it is cultivated as a dwarf potted tree, is described below.   The Comte de Castillon (1877, p. 73, figs. 11, 12) reproduced Japanese figures and gave Latin varietal names to two other sports of this species grown in Japan—the variety microcarpa, which has minute fruits, and the variety punctata, which has leaves dotted with golden-yellow spots, called in Japan sunago or "gold-dust" trifoliate orange.   These interesting garden sports probably do not merit recognition in taxonomic botany as varieties.
      In spite of the variations mentioned above, it is probable that Poncirus trifoliata, of all the True Citrus Fruit Trees, is the most stable of the species that have been cultivated for long periods by man.   This may perhaps be due to the fact that this aberrant genus is not very closely related to the other genera in the group; if it crosses with them, the resulting hybrids (citranges, etc.) are almost always sterile.   Thanks to this circumstance P. trifoliata has preserved its nature with little change in spite of long culture.
      The trifoliate orange has been used very extensively in Japan and somewhat less so in the United States as a rootstock for the satsuma and other cultivated citrus fruit trees.   It is also used occasionally for hedges (Swingle, 1911) and is commonly grown as an ornamental in Asia, Europe, and North America, especially in regions that are too cold to permit the culture of ordinary citrus fruit trees out of doors.
      Poncirus trifoliata has been crossed with Citrus making hybrids called citranges that have proved valuable as vigorous, hardy, disease-resistant rootstocks for ordinary citrus fruit trees.
      The Chinese have for centuries used the dried fruits of the trifoliate orange in their medical practice.   Apparently both the immature fruit, called chih-shih,and the fully ripe, sliced, and dried fruits, called chih-k'o, are utilized, but for different purposes.
      Poncirus trifoliata var. monstrosa (T. Ito) Swing. in Bailey, Stand. Cycl. Hort. 5:2952. 1916. Citrus trifoliata var. monstrosa T. Ito, Encycl. Jap. (Nippon Hyakka Daijiten) 2:1056. 1909.
      Type.—Cultivated in Japan.
      Common name.—Flying-dragon trifoliate orange.
      Tokutaro Ito's original diagnosis in English reads as follows: "Citrus trifoliata var. monstrosa T. Ito, nov. var.   Branches and spines crooked and curved, leaves very slender."   A very dwarf variety, with tortuous slender branches with curved spines; the leaves are very small with the leaflets often reduced to linear filaments on which the oil glands appear as node-like thickenings.
      In Japan this curious monstrosity, called hiryô, or flying dragon, is esteemed for culture as a dwarf potted plant.

PONCIRUS HYBRIDS
      Citranges [Poncirus trifoliata X Citrus sinensis].
      As the genus Poncirus is monotypic, all its hybrids are intergeneric, i.e., either bigeneric or trigeneric.   The first adequately safeguarded hybrids of Poncirus and Citrus were made by Swingle in 1897.   These hybrids, called citranges, showed such striking differences, even between sister hybrids grown from a single fruit, that a number of them (Rusk, Willits, Morton, Coleman, Savage, Rustic, and Saunders) were named, described, and illustrated.   (See Swingle and Webber, 1898, p. 400, text fig. 13; Webber and Swingle, 1905, pp. 223-35, pls. 10-16, text figs. 12 and 13; Webber, 1907, pp. 329-36, pls. 17-20; Swingle, 1910, pp. 36-41, pls. 1-2; idem, 1913e, pp. 83-87, pls. on pp. 84, 86, 88, 90, 92; idem, 1913f, pp. 382-86, figs. 1-4; idem, 1927b, pp. 19-21.)   Several named varieties of citranges are described by Hodgson in this work.
      Citrumelos [Poncirus trifoliata X Citrus paradisi].
      Citrumelos are very similar to citranges.   At least one of them, the Sacaton, produces large numbers of seeds containing only nucellar embryos and consequently yields identical F1 seedlings in great numbers, like the Troyer citrange (see Hodgson, this work).   The Sacaton is a promising rootstock.
      Citrandarins [Poncirus trifoliata X Citrus reticulata].
      Citrandarins are much like citranges except that the mandarin orange parent greatly reduces the amount of bitter oil found in droplets inside the pulp-vesicles.   In citranges this oil usually makes the fruit too bitter to eat.
      Citremons [Poncirus trifoliata X Citrus limon].
      Many of these hybrids show abnormally small leaves in germinating and soon die (Swingle, 1913f); others are broad-leaved.   One, the Highgrove, is a very fast grower with fruits like a lemon but rougher and more seedy.
      Citradias [Poncirus trifoliata X Citrus aurantium].
      These hybrids greatly resemble citranges but seem to be even more vigorous and cold-resistant.   The Brownell citradia (fig. 3-42) grows vigorously and fruits freely in northern Alabama at Tuscaloosa (Lat. 33° 12’ N.).   It makes a good stock for the commonly cultivated citrus fruits.
      Citrumquats [Poncirus trifoliata X Fortunella japonica, or F. margarita].
      These hybrids are very difficult to make, germinate badly, and nearly all die while still very small.   Nevertheless, a very few of them have been obtained that grow vigorously (see under Fortunella).

SECOND-GENERATION CITRANGES
      Segentranges [(Poncirus trifoliata X Citrus sinensis) F2].
      Almost all Poncirus X Citrus hybrids are completely ovule-sterile, although almost all of them produce a little fertile pollen.   Only two of the many citranges that have been studied (the Phelps and the Sanford) regularly produce true second-generation seedlings, all of which show extraordinary variability.   Some of the seedlings are almost like the one or the other parent species, and all degrees of intermediates occur.   In the experimental plantings these hybrids have been called segentranges, for convenience, but this name covers such a range of forms as to be of little practical value (see Swingle, 1910, p. 37, pls. 3, 4 [showing Sanford citrange F2 seedlings]; Swingle, 1927b, p. 20; and Hodgson, this work).

BACK-CROSSES OF PONCIRUS HYBRIDS ON THE PARENT SPECIES
      Citrangors [Citrange X Citrus sinensis].
      These hybrids are often very like the sweet orange in leaf characters; the fruits have not as yet been studied.   The orange-like seedlings resemble also the orange-like second-generation seedlings of the fertile citrange varieties.   Some of the citrangors show trifoliolate leaves and look much like citranges (see Swingle, 1910, p. 36, pls. 7, 8).
      Cicitranges [Citrange X Poncirus trifoliata].
      These hybrids often look much like Poncirus trifoliata in both leaf and fruit characters.   The fruits are often more juicy and contain fewer seeds than those of P. trifoliata but are always more or less bitter from the droplets of acrid oil in the pulp-vesicles.   All the cicitranges are trifoliolate, but some are citrange-like with broad lateral leaflets.

TRIGENERIC PONCIRUS HYBRIDS
      Citrangequats [Citrange X Fortunella sp.].
      These vigorous and variable hybrids are discussed in this chapter under Fortunella.
      Citrangedins [Citrange X (Fortunella X Citrus) 'Calamondin’].
      A remarkable hybrid of this parentage is discussed in this chapter under Fortunella.
      Citrangeremos [Citrange X Eremocitrus glauca].
      These hybrids are discussed in this chapter under Eremocitrus.

CITRANGES AND OTHER PONCIRUS HYBRIDS IN DOORYARD AND VILLAGE GROVES
      All Poncirus hybrids are marked by their trifoliolate or difoliolate leaves and hence are easily recognized even when the seedlings are too young to have flowered or fruited.   There is every reason to expect that similar spontaneous hybrids due to cross-pollination by insects between Citrus and Poncirus should be found; as a matter of fact, such hybrids do occur in China, Indo-China, and Malaya.   Also it is to be expected that intergeneric hybrids between Poncirus and widely different species of Citrus should exist; in fact, these are found in the homeland of citrus fruits, in southeastern Asia, Japan, and throughout the Monsoon region.
      Swingle saw in the dooryard gardens in China trifoliolate citrange trees and also trifoliolate, difoliolate, and unifoliolate forms that were probably citrangors; other forms, obviously Poncirus hybrids, have been found in China and in Indo-China.   It is therefore to be expected that back-crosses and second-generation seedlings, as well as complex 3-parental hybrids, should occur not uncommonly as "chance seedlings" in dooryard citrus gardens in the Far East.
      Fortunately the vast amount of carefully safeguarded hybridization work that has been done in the United States on the True Citrus Fruit Trees has shown what types of hybrids may be expected to occur, and as a matter of fact do occur, when mixed plantings of species of Citrus, Fortunella, and Poncirus in dooryard or village groves are cross-pollinated by insects (which visit the flowers with great assiduity because of their abundant supply of nectar).   Chance seedlings are almost certain to arise from seeds dropped in such dooryard or village groves and their chance seedlings can and do persist, flower and fruit.   If the fruit is valued, such hybrids are propagated by grafting and a new cultivated variety is born.
      Knowing, as we do, the exact origin of the remarkable hybrids of Poncirus listed above, as well as of the many other, even more complicated, hybrids listed under Fortunella and under Eremocitrus, we are able to appreciate how unscientific it would be if "new species" (and even the "new genera" that would be needed in some instances) were created to include these hybrids (whether they arose artificially by human design or spontaneously through insect pollination), thus giving them the same taxonomic status as genuine species that arise in nature by slow evolutionary processes.

XXIV.   Clymenia Swingle
      XXIV.   Clymenia Swingle, Jour. Arnold Arbor. 20:251. 1939.
      Type species (monotype).—Citrus polyandra Tan. = Clymenia polyandra (Tan.) Swing.
      Distribution.—Known only from the type locality near Namatanai in northern New Ireland, Bismarck Archipelago, northeast of New Guinea.
      Common name.—Clymenia.
      This genus is allied to Citrus; it differs in having (1) subsessile pulp-vesicles (somewhat narrowed at the base but not borne on slender stalks) attached in great numbers to the lateral segment walls for 3/4 the distance from the peel to the axis; (2) very numerous stamens (50-100) with free, slender filaments; (3) ovary with a very short, stout style (less than 1/2 the length of the ovary); and (4) leaves with very short petioles, from 1/10 to 1/20 the length of the leaf blade, with which they are not articulated.   The flowers arise singly in the leaf axils and are borne on straight, rather stout pedicels, slightly longer than the petioles of the subtending leaves; calyx persistent, 5-lobed, lobes reflexed at apex in dried material, suberect at sinuses, thick and coriaceous, wrinkled without, glabrous within; corolla 5-merous, petals imbricate, dotted with numerous small oil glands; stamens very numerous (10-20 times as many as the petals), filaments free, slender, anthers oblong; ovary ovoid with 14-16 locules, ovules several in each locule; style very short (about half the length of the ovary); stigma capitate; disk broader than the ovary, about half as long; fruit ovoid, size of a small orange, skin thin, orange-like, dotted with oil glands; pulp-vesicles attached in part to the dorsal, but principally to the side walls of the segments; seeds very numerous, small, subglobose or oval, glabrous, embryo green.   Small trees, branches spineless, twigs subangular when young, then cylindrical; leaves thin, smooth, with numerous subparallel lateral veins, acuminate-caudate at apex, cuneate at base, tapering into the very short petiole, furrowed above, and not articulated with the leaf blade.
      This remarkable new genus (of which only one species is known) is in its general aspect so much like Citrus that it was at first mistaken for a sweet lime.   It bears sweetish edible fruits about the size and shape of a sweet lime (5 to 6 by 7 to 8 cm), even to a short nipple at the apex.   The ripe fruits are yellow in color, again like sweet limes.   However, Clymenia differs very widely from Citrus in many important taxonomic characters and is undoubtedly a distinct new genus.   It is especially noteworthy in having a type of pulp-vesicle not known in any other citrus fruit.
      The type of this genus, Citrus polyandra, was described in Latin rather fully by Tanaka in all its characters except the detailed structure of the pulp-vesicles and the manner of their attachment in great numbers to the side walls of the fruit segments.   The pulp-vesicles are short, plump, blunt, oval or subglobose, sessile or very short-stalked.   What is also important from a taxonomic standpoint is that these pulp-vesicles are attached in very large numbers all along the side walls of the segments, nearly to the inner angle touching the central axis of the fruit.   The leaves of Clymenia are very unlike those of any other True Citrus Fruit Tree (Citrus, Microcitrus, Eremocitrus, Fortunella,or Poncirus) and strongly resemble those of Monanthocitrus, a genus of the Minor Citroid Fruit Trees described from New Guinea by Tanaka.   Monanthocitrus has, however, a much simpler floral organization with only twice as many stamens as petals, instead of 10 to 20 times as many, as in Clymenia.   The fruits of Clymenia are about the size of a small orange.   The greatly enlarged disk bearing very numerous stamens is unique in the orange subfamily.
      Clymenia polyandra (Tan.) Swing. Jour. Arnold Arbor [sic] 20:253. 1939. Citrus polyandra Tan. Studia Citrol. 2:158, 163. 1928; Citrus medica subsp. limonum var. limetta Lauterb. (non "[Risso] Engl.") Bot. Jahrb. 55:264. 1918. Illus. Tillson & Bamford, Amer. Jour. Bot. 25:786, figs. 34-36. 1938; Swingle, loc. cit. pl. 1, figs. 1-5; fig. 3-43 this work.
      Type.—Bismarck Archipelago, New Ireland, Namatanai, Nukonoku (Peekel, No. 408).   Herb. Bot. Mus., Berlin-Dahlem.   Fragments of type: Herb. Natl. Arbor., Washington, D.C., sheet No. 46011.   Topotype: New Ireland, Buratamtabai, Namatanai (C. S. Gee, 5/6/37, with the help of Father Peekel!), Herb. Natl. Arbor, Washington, D.C., sheet No. 46010.
      Distribution.—Bismarck Archipelago: known only from the vicinity of Namatanai, New Ireland.17
      Common name.—Clymenia.
      Tanaka's original description (in Latin) may be translated about as follows: "Branches triangular, glabrous, without spines; leaves oblong, caudate-acuminate, abruptly narrowed at the base, irregularly serrate; petioles very short, not articulated [with the leaf blade], and not winged, but furrowed [on the upper side]; flowers single in the axils of the leaves, with rather long, somewhat thick, pedicels; calyx saucer-shaped, obscurely toothed; stamens up to 100, free, arising in many rows outside the disk; filaments linear; anthers oblong; ovary ellipsoid; style thick, very short; stigma capitate, somewhat pentagonal; fruit oblong, somewhat lemon-shaped, with an inconspicuous mammilla at the tip; peel thin, segments numerous; pulp vesicular, yellow, rather sweet; pulp-vesicles short, blunt, very numerous; seeds obovoid, superposed."
      The twigs are spineless and the leaves are very peculiar, being elliptical, acuminate at both ends, with the base merging into a very short, slender, wingless petiole which is not articulated or in any way marked off from the blade.   The flowers are produced singly in the axils of the leaves, petals short and broad (7 X 7 mm); stamens very numerous (50-100) in many whorls, with short filaments (5 mm), pistils shorter than the stamens, ovary ovoid, with a broadly rounded tip not merging into the style, ovary with 14-16 locules, style short (only half as long as the ovary), with capitate stigma; fruit lemon-shaped, 4.5 X 6-7 cm, with a short nipple at the apex; pulp sweet; seeds numerous, superposed, polyembryonic.
      Tanaka deserves the credit for having first noticed the great divergence of this plant in many of its essential taxonomic characters from any other of the known True Citrus Fruit Trees.   However, he considered it to be a hybrid of Citrus macroptera and C. medica, which is said by Lauterbach to be "commonly planted at the [European ?] stations."
      After studying Citrus macroptera and other members of the subgenus Papeda and observing the hybrids between them and species of the subgenus Citrus that are common in the Philippines, Swingle became convinced that a hybrid such as Tanaka thought his Citrus polyandra to be (Citrus macroptera X C. medica) could not possibly show a wingless petiole, even though one supposed parent, the citron, has an almost completely wingless petiole, because the other supposed parent, C. macroptera, has a very long, very broadly winged petiole (certainly the largest petiole to be found in Citrus).
      The extraordinarily close resemblance of the leaves and petioles of this species to those of Monanthocitrus convinced Swingle that this plant was an entirely new type of citrus fruit tree, possibly having descended from a remote ancestral species common also to Monanthocitrus.   Swingle thereupon published Clymenia as a new genus to include this plant (see fig. 3-43).
      This species, called a-mulis by the natives in the village of Namatanai in New Ireland, is cultivated for its sweet fruits.

XXV.   Microcitrus Swingle
      XXV.   Microcitrus Swingle, Jour. Wash. Acad. Sci. 5:570. 1915.
      Type.Citrus australasica F. Muell. = Microcitrus australasica (F. Muell.) Swing.
      Distribution.—Of the 6 known species, 5 are found in eastern Australia (Queensland and northern New South Wales) and the other (M. warburgiana) in southeastern New Guinea.
      Common name.—Australian wild lime.
      Twigs minutely puberulous; leaves dimorphic, those on young seedlings greatly reduced cataphylls gradually merging into small, linear or elongate-elliptical leaves and finally into the normal foliage of the species; mature leaves more or less coriaceous, strongly veined, glabrous except in midrib near base; petioles short, wingless, finely pubescent, especially on the upper flattened side; spines slender, acute, single (or in pairs in M. inodora and M. maideniana) in the axils of the leaves; flowers small, 4-5-merous, stamens entirely free, ovary with 4-8 locules; ovules several, 4-8 or more in each locule; fruits round-ovoid or cylindric; pulp-vesicles noncoherent, slender-stalked, subglobose or ovoid (except in M. australis, in which they are elongate-fusiform, tapering above), filled with acid pulp having minute droplets of acrid oil aggregated along the axis; seeds small, ovate in outline; cotyledons hypogeous in germination.
      This genus is obviously related to Citrus, from which it differs, however, in having dimorphic foliage, free stamens, an ovary with four to six (or eight) locules, and more or less coriaceous, strongly veined leaves.   Moreover, it is related to the highly specialized genus Eremocitrus, which also has dimorphic foliage and free stamens.   Eremocitrus, however, differs from Microcitrus in always having thick, coriaceous leaves with thick cuticle palisade tissue and stomata on both faces and an ovary with only three to five locules having only two ovules in each locule.
      The six known species of Microcitrus are the result of millions of years of slow evolution from a primitive ancestral type.   From this ancestral type arose the genus Eremocitrus, with marked xerophytic adaptations and two ovules in each locule of the ovary, and the genus Microcitrus, with more numerous ovules in each locule.
      This ancestral type probably resembled somewhat M. warburgiana, the New Guinea species, which has small leaves and small, nearly spherical fruits.   From such an ancestral form, one line of evolution produced the so-called native orange, round lime, or Dooja (M. australis), that grows to a large tree and has subglobose fruits much larger than those of M. warburgiana, with long, slender, pointed, more or less twisted pulp-vesicles; another line of evolution culminated in M. inodora and M. maideniana, highly specialized forms showing adaptations to tropical rain forests, with large leaves and paired spines; a third line of evolution led to the small-leaved, somewhat xerophytic species M. australasica and M. garrowayi, both with long-ovoid or very elongate-cylindric fruits.
      These remarkable citrus fruits are extremely interesting, in that they show how evolution has proceeded in regions isolated as Australia and New Guinea have been during the last twenty or thirty million years since they were cut off from all other land masses.18   The evolution of other citrus fruits is not so easily followed, since Citrus, Fortunella, and Poncirus did not originate in regions that were geographically isolated in definitely dated geologic eras.   A key to the species of Microcitrus is presented.
      1.   Microcitrus australasica (F. Muell.) Swing. Jour. Wash. Acad. Sci. 5:572. 1915. Citrus australasica F. Muell. Fragm. Phytogr. Austr. 1:26. 1858. Illus. G. W. Oliver, U.S. Dept. Agr. Bur. Pl. Ind. Bul. 202:pl. 15, figs. 5, 6; pl. 2, fig. 1. 1911; F. M. Bailey, Comprehen. Cat. Queensl. Pl. 84, fig. 63. 1913; Swingle, loc. cit., figs. 1-4; fig. 3-44, E this work.
      Type.—Moreton Bay, Queensland (F. von Mueller).   Present location (?).
      Distribution.—Australia: coastal regions, southern Queensland; northern New South Wales.
      Common name.—Australian finger-lime.
      First leaves of young seedlings small, 3-5 X 2-4 mm, ovate, with very short wingless petioles borne on numerous slender, horizontal twigs with internodes often only 2-4 mm long, with sharp, slender spines, 7-12 mm long, at the nodes; the upright twigs of young plants have similar leaves and spines but the internodes are sometimes 5-8 mm long; the upright, strongly angled, vigorous twigs of mature trees, which have internodes 8-12 mm long with single sharp spines, 5-10 mm long, have lozenge-shaped leaves, 22-25 X 14-15 mm, obscurely toothed in the upper half, gradually narrowed to a blunt, truncate base with which the petiole articulates; lateral veins faintly marked above, more clearly below, 6-10 pairs, arising at an angle of 35°-45° with the midrib, lower ones longer, upper ones soon branching, all with several anastomosing curved cross-veins; leaves tapering into a truncate or emarginate tip, those on lateral twigs 8-20 X 2-10 mm, the smaller ones abruptly truncate, often irregularly emarginate; spines slender, short, 3-8 mm long; flowers arising singly in the axils of the leaves (or occasionally in pairs); flower buds subglobose or obovate, 4 X 5 mm, borne on very short pedicels, 1-2 mm long; flowers usually 5-merous but sometimes 3-4-merous; sepals free, small, concave, minutely ciliate; petals oblong, 7-8 mm long; stamens 20-25, free, pistils short, stout, ovary with 5-7 locules; ovules numerous, 8-16 (or even 20) in each locule; fruits cylindric-fusiform, 6.5-10 X 1.5-2.5 cm, often slightly curved, narrowed at both tip and base, often showing a blunt protuberance at one or both ends; peel rough with numerous oil glands, greenish-yellow at maturity; pulp-vesicles nearly free or loosely cohering, long-stalked, ovoid, ending in a very blunt or rounded tip, about 1.7-3 X 1.2-1.5 mm, borne on a slender stalk 1-2 mm long; seeds numerous, small, 6-7 mm long, ovoid, monoembryonic, usually flattened on one side and often showing small, shallow depressions on the other faces (probably caused by mutual pressure of the pulp-vesicles exerted during development of the fruit); testa smooth (not wrinkled as in Eremocitrus glauca); cotyledons hypogeous in germination.
      This, the type species of the genus Microcitrus, has several characters separating it from all of the species of Citrus.   It is also very clearly separated from the other species of Microcitrus by its very long, slender fruits that are unique in the orange subfamily (see fig. 3-44,E).   They have numerous seeds (or rudiments) in each segment (more numerous than in any other true citrus fruit).   The pulp is composed of loosely grouped, long-stalked, subglobose or pyriform pulp-vesicles, tapering bluntly to the tip.   These pulp-vesicles are very different from those of M. australis but are much like those of M. inodora; they also resemble somewhat those of Citrus (Papeda) hystrix, as noted by Penzig (1887, pp. 214-17), who was the first to describe the anatomy of the fruits, seeds, and pulp-vesicles of M. australasica.   The very acrid pulp has a harsh aftertaste, probably due to droplets of acrid oil in the pulp-vesicles, such as have been found in the Australian round lime.   Francis (1929, p. 174) stated that the fruits, both of this species and of M. australis, "have a lemon-like flavor accompanied by a taste of a turpentine nature."
      The finger-lime of the coastal regions of northeastern Australia is a tall shrub or small tree and makes a handsome ornamental.   The seedlings are very spiny.   The first leaves are minute linear cataphylls; these gradually merge into juvenile foliage which, in turn, merges into the mature foliage, the leaves of which are smaller than those of any other True Citrus Fruit Tree with the exception of Eremocitrus glauca when the latter occurs in very dry situations.   The branches on young plants often stand out nearly at right angles to the stem, giving the young plant the appearance of a very dwarfed fir tree.   This species is precocious; Oliver (1911, p. 15) figured a small branch of a two-year-old seedling bearing two fruits.
      Uphof (1932, pp. 138-39), in an examination of five trees of M. australasica, found that most of the flowers were male and that only a very few were perfect.   He examined the flowers on eight twigs and found, out of 161 flowers, that 151 were male and that only ten were perfect.
      Microcitrus australasica is of interest in hybridization work because of the great vigor and the extraordinary production of slender leafy twigs that have been shown by the Sydney hybrid.
      1a.   Microcitrus australasica var. sanguinea (F. M. Bail.) Swing. Jour. Wash. Acad. Sci. 5:574. 1915. Citrus australasica var. sanguinea F. M. Bail. Dept. Agr. Queensl. Bot. Bul. 18:8. 1892. Illus. Penzig, Studi Bot. Sugli Agrumi, Atl. pl. 21, fig. 13. 1887.
      Type.—Queensland (Bailey ?).   Herb. Bot. Gard. Brisbane.
      Distribution.—Australia: southeastern Queensland.
      Common name.—Red-pulp finger-lime.
      This variety differs from the species in that the pulp-vesicles at maturity vary from pink to red in color.
      There are similar "pink" or red varieties of oranges (the so-called "blood oranges") and grapefruit, some of which are known to have arisen as budsports.   However, the red-pulped variety of the Australian finger-lime is found growing wild and can be propagated from seed; it seems to have originated without the aid of man.
      HYBRIDS
      Faustrimedins [Microcitrus australasica X (Fortunella sp. X Citrus sp.) 'Calamondin'].
      This trigeneric hybrid is discussed under Fortunella.

      2.   Microcitrus australis (Planch.) Swing. Jour. Wash. Acad. Sci. 5:575. 1915. Limonia australis. A. Cunn. nomen subnudum. 1839; Citrus australis Planch. Hort. Donatensis 18. 1854-1858; C. planchoni F. Muell. 1872. Illus. Penzig, Studi Bot. Sugli Agrumi, Atl. pl. 21, fig. 8-13. 1887; Campbell, Agr. Gaz. N. S. Wales 10:1168. 1899; Guilfoyle, Austr. Pl. pl. facing p. 123. 1911; Bailey, Comprehen. Cat. Queensl. Pl. 84, fig. 62. 1913; fig. 3-44,A-B this work.
      Type.—Moreton Bay, Queensland (Leichhardt).   Herb. Mus. Hist. Nat., Paris.
      Distribution.—Northeastern Australia.
      Common name.—Australian round lime.
      Leading terminal twigs straight, very angular, with internodes 1-1.5 cm long, with single slender acute thorns 5-10 mm long; lateral twigs short, slightly angled, with internodes 5-10 mm long, and spines 2-6 mm long; leaves entire, glabrous, or sparingly ciliate on the margin near the base; leaves articulated with the petioles; larger leaves on leading twigs 3-4 X 2-3 cm, obovate to obcordate (sometimes lozenge-shaped on less vigorous leading twigs); leaves of lateral twigs diamond-shaped, 2-3 X 1.2-1.8 cm; veins visible on both surfaces, but more conspicuous below, 12-15 pair arising at an angle of 25°-30° with the midrib and frequently branching, with veinlets arising at a very acute angle and running parallel to the veins; leaves rounded, emarginate or bluntly pointed at tip, cuneate at the base; petioles articulated with the leaf blades, finely pubescent and flattened an the upper side, 2-3 mm long; flowers arising singly in the axils of the leaves, small, 10-15 mm diam., 4- or 5-merous, with 16-20 stamens with free filaments; fruit globose or subglobose, 2.5-3.5 or 5 cm diam., with 6 segments (Schaler's green to grass green; Ridgway, 1912, pl. 6); pulp-vesicles proper (exclusive of stalk) 6-10 X 2.5-3.5 mm, lightly coherent in mature fruits, very pale greenish (water-green, Ridgway, pl. 41) near the peel but much brighter green (light grass-green to grass-green, Ridgway, pl. 3) in the center, irregularly slender-pyramidal or fusiform, more or less angular from pressure, tapering gradually into long, more or less contorted blunt tips, 1-1.5 mm wide, containing numerous large droplets of strong-flavored oil, stalks often very short, sometimes 2-3 mm long, texture of pulp-vesicles firm enough to cut into slices with a sharp knife (much firmer than in any species of Citrus); seeds flattened, containing a single embryo; no clearly defined chalazal cap in spite of the acid nature of the pulp (in Citrus the species with acid pulp have strongly colored chalazal caps).
      This species, called "Dooja" by the aborigines and round lime or native orange by the Australians, is native to northeastern Australia.   It is readily distinguished from all the other species of Microcitrus (and from Eremocitrus as well) by its globose or slightly pear-shaped, rough-skinned fruits, 2.5 to 5 cm in diameter (about the size of a large walnut), and by its short-stalked pulp-vesicles that taper gradually into blunt tips often more or less deformed and twisted by mutual pressure (see fig. 3-44,B).
      The early descriptions of this species and of M. australasica are so meager as to leave doubt concerning which species was being discussed.   Fortunately the original descriptions of Citrus australis, published by Planchon in a rare work (Hortus donatensis, 1854-58, p. 18), was based on a specimen in the herbarium of the Muséum d'Histoire Naturelle at Paris, collected by Leichhardt in the vicinity of Moreton Bay in southeastern Queensland.   A study of this type specimen makes clear the proper application of the name M. australis.
      Penzig (1887, pp. 210-17, pl. 21, figs. 8-13) described in detail the anatomy of this species.   He found that the pulp-vesicles were similar to those of the commonly cultivated citrus fruits except that they contain, along their central axis, large masses of yellowish oil.   These pulp-vesicles are very different in shape from those of the finger-lime, M. australasica, or those of the other species of Microcitrus.
      The Australian round lime, or Dooja, grows to be a tree 30 to 60 feet high and is the largest of any of the native citrus fruit trees.   The fruit is said to be used for preserves in Queensland.   This species, which has been introduced into the United States, grows well in California, where it is now fruiting.   It is exceeded in vigor, however, by the Sydney hybrid, the result of a natural cross of this species with the finger-lime, Microcitrus australasica.
      HYBRIDS
      Sydney hybrid19 [Microcitrus australis X M. australasica]. Illus. fig. 3-44,C and D.
      Twigs very slender and very numerous, usually only lightly angled, very minutely and rather sparsely puberulous; bud scales brownish, inconspicuously ciliate; nodes 5-13 mm long, with a single slender, sharp, axillary spine, 4-12 mm long; leaves on vigorous leading twigs 20-30 X 8-13 mm, glabrous, narrowly elliptical or lozenge-shaped, with undulate or irregular, shallowly crenate-dentate margins, narrowed at the apex, sometimes with an emarginate tip, cuneate at the base, with numerous lateral veins arising at a small angle (25°-35°) with the midrib, with some cross and some subparallel veinlets, narrowed into very short petioles, 2-3 mm long which are sparingly and finely pubescent above; leaves on lateral twigs 10-20 X 3-8 mm, linear-elliptic or narrowly lozenge-shaped, the smaller leaves entire, the larger with undulate or irregular, shallowly dentate margins, petioles 1.2-2 mm long; fruits elongate-obovoid or ellipsoid, 35-50 X 20-28 mm, with the tip abruptly rounded, numerous slightly protuberant oil glands, circular in outline and about 0.5 mm diam., peel 2 mm thick, yellowish-green when ripe (near lime green, Ridgway, pl. 31); inner portion 0.4-0.5 mm thick, smooth, of firmer texture than the outer, segments 5-8 central core solid, very narrow, ovule traces numerous; pulp-vesicles ovoid, tapering to an acute point above, broadly rounded at base, 4.5-7 X 2-4 mm, borne on slender stalks, 2-7 mm long, attached to inner wall of the ovary but not to segment walls, not adhering to one another, containing an acid juice; seeds none.   The ripe fruits have a spicy, not disagreeable odor.   The flower buds are spinel red, and the new growth is very dark dull dusky purple (Ridgway, 1912).
      This remarkable hybrid was grown from seeds of the Australian round lime, Microcitrus australis, or of the finger-lime, M. australasica, sent to the U.S. Department of Agriculture's Agricultural Research Service by the late J. E. Maiden, director of the Botanical Garden at Sydney, Australia.   More than 200 meters of twigs, both large and small, were borne on a single branch, 3 cm in diameter, of a Sydney hybrid growing near Riverside, California, which survived for several decades on land no longer irrigated where citrus trees made little or no growth.
      The leaves are intermediate in shape between the two parent species, but, owing to the enormous number of twigs and consequent great profusion of leaves, they are not intermediate in size between those of the parent species, but nearer that of the small-leaved parent, the finger-lime.   The leaf margins are more crenate than those of either parent.   The small twigs of both parents are minutely ciliate, but the larger angled twigs, especially of the finger-lime, are almost glabrous.   In the hybrid, however, the twigs are more pubescent than those of either parent and are more slender and more upright.
      This hybrid has been found to support the orange and other commonly grown citrus fruits very well in greenhouse culture, and it can be propagated readily from soft-wood cuttings of summer growth.   As it excels all known citrus fruit trees in the length of twigs it produces, it should be tested as a stock for citrus fruits grown on foothill soils low in organic nitrogen.   On account of its extremely abundant branchlets covered with miniature leaves, the Sydney hybrid deserves trial as an ornamental, especially in the orange-growing section of the southeastern states.

      3.   Microcitrus garrowayi (F. M. Bail.) Swing. Jour. Wash. Acad. Sci. 5:574. 1915. Citrus garrowayi F. M. Bail. Queensl. Agr. Jour. 15:491. 1904. Illus. Bailey, Comprehen. Cat. Queensl. Pl. 84, fig. 65 (leaves and fruit). 1913.
      Type.—Queensland, Mount White, Cape York Peninsula, near Coen (Lat. 13° 55' S., Long. 145° E.), altitude 1,300 feet (R. C. Garroway).   Herb. Bot. Gard. Brisbane.
      Distribution.—Known only from the type locality.
      Common name.—Garroway's Australian wild lime.
      Bark with minute, irregularly interrupted longitudinal ridges; young twigs very slender, 0.8-1.5 mm diam., cylindric or rarely slightly angled when very young, with very small more or less upward curving, hyaline hairs; internodes 4-15 mm long (usually about 6-10 mm), with very slender spines, 2-5 mm long, or spineless; older twigs below last growth show gray, corky, longitudinal, more or less confluent streaks, 0.5-1 mm wide, that show fissures (these corky streaks become more and more abundant and finally run together, so that two-year-old twigs show little or no green bark); leaves of cotype specimen (from Mount White, Queensland) narrow, thick and leathery, lozenge-shaped to broadly lanceolate, 30-45 X 15-20 mm (small leaves 20-25 X 10-12 mm), bluntly pointed, rounded or even slightly emarginate at apex, cuneate at base, with 5-7 pairs of veins, visible on both surfaces of the leaves, arising at the midrib and occasionally branching, margins nearly entire, slightly undulate; leaves on leading twigs of greenhouse specimen 20-25 X 12-18 mm, lozenge-shaped or broadly ovate, bluntly rounded or slightly emarginate at apex, cuneate at base, margins slightly dentate or crenulate especially toward the apex, with a few hyaline hairs on the margin at the base of the leaf and still fewer on the lower portion of the midrib; petioles 2-4 X 0.8-1 mm, wingless, flattened above, articulated with the blade, finely pubescent on the upper flattened side, especially at the edges; fruit of cotype specimen ellipsoid or cylindric-ellipsoid, 4 X 2 cm, with numerous sunken oil glands.
      Bailey described this species as follows, under the name of Citrus garrowayi: "Fruit upon the slender branchlets oblong, 2 1/2 in. long, 14 lines diameter.   Oil glands large, giving a tuberculose appearance to the fruits; ultimately these glands sink, and the fruit appears then to be lacunose; rind very thin; cells 4 or 5; pulp of a sharp agreeable acid; seeds 3-angular, white, free, with more or less very short hairs, about 3 lines long and 2 lines thick in the centre.   The rough rind of the fruit somewhat resembles that of C. australis, but the form of fruit is nearer to that of C. australasica; the fruit of this latter, however, is nearly smooth and the glands small."
      Microcitrus garrowayi seems to be a good species, distinct from M. australasica and M. australis, occurring farther to the south in eastern Australia.   Bailey called this the Mount White lime.   It has been introduced into the United States.
      4.   Microcitrus inodora (F. M. Bail.) Swing. Jour. Wash. Acad. Sci. 5:577. 1915. Citrus inodorus F. M. Bail. in Meston, Rept. Govt . Sci. Exped. Bellenden-Ker Range 34. 1889; C. inodora F. M. Bail. Syn. Queensl. Fl. 3d Suppl. 12. 1890. Illus. Bailey, Queensl. Fl. 1:pl. 10. 1899; idem, Comprehen. Cat. Queensl. Pl. fig. 64. 1913; C. T. White, Jour. Hered. 13:120, figs. A-E. 1922.
      Type.—Queensland, Harvey's Creek, Russell River (F. M. Bailey).   Herb. Bot. Gard. Brisbane; part of type material in Herb. Natl. Arbor., Washington, D.C.
      Distribution.—Australia: northeastern Queensland.
      Common name.—Large-leaf Australian wild lime.
      Leaves broadly oval or lanceolate, or with cuneate bases and more or less acute or even caudate at the apex, 8-18 X 4-10 cm, with very numerous, parallel, lateral veins arising at an angle of 60°-80° with the midrib; petioles very short, 4-8 mm long, wingless, not articulated with the blade; flowers small (as in other species of Microcitrus) but reported by Bailey to be odorless; stamens free but much more numerous (over 30!) than in the other species of Microcitrus; fruit oblong or elliptical (somewhat lemon-shaped), 5-6.5 X 3-3.5 cm, with a conical base, segments 8, on drying sometimes showing ribs corresponding to the walls separating the segments; pulp-vesicles stalked like those of other species of Microcitrus, rather loose; seeds small, somewhat pear-shaped, 6-8 X 4-5 mm.
      This species is a shrub or small tree 2 to 4 meters high with the trunk 4 to 5 cm in diameter.   The twigs are angular like those of Citrus sinensis (the sweet orange) and have one or two slender, very sharp spines, 6 to 12 mm long.   It has been found growing wild only along Harvey's Creek, Russell River, where, according to White (l.c., p. 119), it is common in the lowland rain forests at the foot of the Bellenden-Ker Range in northern Queensland near Cairns (Lat. 17° S.).   This is a very rainy region, the average annual precipitation being 170 inches.   In February, 1922 (?), the rainfall was 62 inches!
      5.   Microcitrus maideniana (Domin) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Citrus maideniana Domin, Bibl. Bot. 89:297. 1927. [Beiträge z. Fl. Austr. 851].
      Type.—Northeastern Queensland, rain forest along Harvey's Creek (Domin, December 1909, to January 1910).   Herb. Univ. Praha, Czechoslovakia.
      Distribution.—Known only from the type locality.
      Common name.—Maiden's Australian wild lime.
      The original diagnosis reads, in approximate translation: "A low erect shrub, 1.5-2.5 m high; young twigs angular-compressed; spines numerous on the twigs, two in the axil of each leaf, of which one spine is occasionally shorter; leaves lanceolate or ovate-lanceolate, pale green, midrib rather prominent; fruit borne on a short thick pedicel, obovoid-ellipsoid, when ripe about 4.5 cm long and 2.5 cm wide or sometimes less, yellow, smooth, with 8 obtuse ribs, not clearly visible in ripe fruits, apex deeply sunken; young fruits deep green with closely set dusky green spots.
      "A new species named in honor of the late Professor J. H. Maiden of Sydney; it appears from the description to be related to Citrus inodora F. M. Bail. from the same locality, but, besides the smaller fruits, the deeply sunken apex shows it to be distinct."
      It is highly probable that this species is very closely related to Microcitrus inodora, native to the same locality, with which it agrees in having paired spines and more or less clearly eight-ribbed fruits (doubtless having eight segments).   The deeply depressed apex of the fruit is the only clearly distinctive character known as yet.
      6.   Microcitrus warburgiana (F. M. Bail.) Tan. Bul. Soc. Bot. France 75:714. 1928. Citrus warburgiana F. M. Bail. Contr. Fl. Brit. N. Guin. 1. 1903. Illus. Bailey, loc. cit. first unnumbered plate, lower half (leafy twigs and fruit); fig. 3-45 this work.
      Type.—Southeastern British New Guinea, Milne Bay (W. E. Armit).   Paratype: U.S. Natl. Herb., Washington, D.C. (from F. M. Bailey!).
      Distribution.—Known only from the type locality.
      Common name.—New Guinea wild lime.
      The original description reads as follows: "Branchlets angular, soon becoming terete, apparently quite glabrous; axillary spines erecto-patent, about 2 lines long on the specimen seen.   Leaves rhomboid-lanceolate, 1 1/2 to 2 1/2 in. long, 5 to 11 lines broad, deeply emarginate and the margins crenulate for 2/3 the way down; primary nerves rather numerous and very oblique, the intermediate ones and reticulate veins rather prominent.   Petioles light colored, 2 to 3 in. long, very narrowly winged.   (No flowers collected.)   Fruit axillary on a stout peduncle, 1 1/2 lines long, globose, about 3/4 in. diam., oil glands concave, crowded, rind thin, cells 6.   This species seems to be nearly allied to C. Medica, var. aruensis, Warbg., Engl., in Bot. Jahrb., 1891, page 340."
      A twig of the type collection sent to Swingle by Bailey shows characters as follows: Twigs 30 cm long, 2 mm diam. at base, with very slender terminal twigs only 0.5-1 mm diam., internodes 1-2.5 cm long (rarely 5 mm on the slenderest lateral twigs), slightly angled at first but soon cylindrical, minutely puberulous at first but apparently soon glabrescent; leaves elongate-elliptical, leaf blades 4-6 X 1-1.8 cm, coarsely but shallowly crenulate-serrate, bluntly rounded at apex and often slightly emarginate, bases cuneate, subentire, lateral veins 10-12, connected by numerous smaller subparallel reticulating veinlets; petioles subulate, very short, 2-3 mm long, 0.5 mm thick, 0.8-1 mm wide, articulated with the leaf blade, with slightly depressed channel on the upper side bordered along the upper half by a slightly crenulate membrane (in drying, sometimes curved upward and making a nearly closed channel), sparingly and minutely puberulous on the upper side and on the margins.
      The small globular fruits, the elongate-elliptical, crenulate leaves, and the narrowly subcrenulate-margined, very short petioles separate this New Guinea species very clearly from the other species of Microcitrus.   This is the only species of the genus outside of Australia, and it occurs nearly a thousand miles away from M. garrowayi, the Australian species of Microcitrus growing nearest to New Guinea.
      The fruits are small, globose, with six segments (fig. 3-45).   Bailey's description, however, says nothing concerning the character of the pulp-vesicles, if any!   The petioles differ from those of the other species of Microcitrus in having a very narrowly crenulate margin or wing.   The veins are numerous, making only a small angle (35° to 50°) with the midrib, and soon branch, forming a reticulation of veinlets not unlike those of M. garrowayi, although the latter species has smaller leaves with fewer veins.   The ultimate twigs, especially the short lateral branchlets, are remarkably slender, the smallest lateral twigs being 1 to 1.5 mm or even only 0.5 to 0.8 mm in diameter.

XXVI.   Citrus L.
      XXVI.   Citrus L. Sp. Pl. 1:401. 1753. Citrophorum Neck. Elem. 1:401. 1790; Sarcodactylis Gaert. f. Fruct. Sem. 3:39, pl. 185. 1805; Papeda Hassk. Flora, 25, Beibl. 2:42. 1842.
      Type species.—Citrus medica L.
      Distribution.—Southeastern Asia, East Indian Archipelago, Philippines, New Guinea, New Caledonia, Bismarck Archipelago, Solomon Islands, Fuji (?), Samoa (?).
      Small trees; young twigs angled, soon cylindrical, with single spines in the axils of the leaves but older branches often spineless; leaves 1-foliolate, usually thin, not coriaceous, lateral veins few and without conspicuous reticular veinlets between; petioles usually more or less winged and articulated with the leaf blade (except in C. medica, where they are wingless or merely margined and not articulated with the leaf blade); flowers single in the axils of the leaves or in short, axillary, corymbose racemes, perfect or staminate by more or less complete abortion of the pistil; calyx cup-shaped, with 4-5 lobes; petals 4-8 (usually 5), thick, linear, strongly gland-dotted, imbricate in the bud; stamens usually about 4 times as many as the petals but in some species usually 6-10 times as many; disk annular, short; ovary subglobose and sharply distinct from the much narrower style, or else truncated, fusiform, or subcylindrical and merging gradually into a style nearly as thick as the upper portion of the ovary, locules 8-18 (usually 10-14), with 4-8 or more ovules in each locule in 2 collateral rows; style cylindrical, expanding abruptly into the subglobose or oblate-spheroid stigma; fruit a hesperidium with the segments containing seeds near the inner angle and the rest of the space filled with stalked fusiform pulp-vesicles, filled with a very watery, large-celled tissue; around the segments is a white endocarp, outside of which is the peel dotted with very numerous oil glands and turning yellow or orange at full maturity; seeds obovoid or flattened obovoid, more or less angular, containing 1 or many embryos, either white or green.
      In young developing fruits of Citrus short-stalked, club-shaped slime-secreting organs develop rapidly along with the pulp-vesicles.   These, however, soon mature and liberate, from collapsing superficial cells of the head, slimy matter which probably enables the developing pulp-vesicles to slide freely over one another (see Tschirch and Oesterle, 1898, p. 303, pl. 70, figs. 36, 36a, 36b, 36c, and 38).   Similar organs probably occur in the other genera of the True Citrus Fruit Trees, but in the subgenus Protocitrus of the genus Fortunella these slime-secreting organs are sessile and are not intermixed with the developing pulp-vesicles.
      The genus Citrus, thanks to the progress made in the taxonomic study of the orange subfamily in the last half century, can now be seen in proper perspective.   It is the culmination of a very long period of progressive evolution that certainly began before Australia was cut off from land connection with New Guinea and Asia, probably more than 20 million years ago.   The genus Microcitrus, closely related to Citrus, occurs both in southern New Guinea (one species) and in northeastern Australia.   A related genus, Eremocitrus, originated in Australia and became a xerophyte adapted to grow in semiarid regions.   Citrus occurs in New Caledonia, but only one species is known there or in neighboring islands to the east and southeast.   The genus may have originated in the New Guinea-Melanesia region, but its evolution into many different species took place chiefly on the mainland of southeastern Asia.   In fact, it is only there that the most highly developed species of Citrus can be considered as indigenous.   Only the aberrant species belonging to the subgenus Papeda have developed in the East Indian Archipelago, and in the Philippines, New Guinea, and Melanesia.   A few of these species reached the Asiatic mainland, and from these sour and bitter-flavored, almost inedible species have doubtless developed the fragrant, delicious-flavored species of the subgenus Citrus, in many ways the finest fruits known to man.
      The superiority of these finer species of Citrus has led to their extinction as wild plants because their fruits made them so attractive that the trees were transplanted to village gardens.   This led to cross-pollination by bees and other insects, with the resultant obscuring of species limits to the extent that in many parts of southeastern Asia and in the East Indian Archipelago it is difficult to find any clear-cut species of Citrus.   There appears instead, an endless array of complex hybrids, often combining not two, but three or even four, different species in varying proportions.   This situation has led to much confusion in attempting to define species of Citrus.   As will be shown later, it is necessary to abstain from naming the principal cultivated varieties of Citrus as independent species.   Unfortunately, names have sometimes been given even to almost sterile varieties, doubtless of hybrid origin, that depend wholly on man's care for their perpetuation.

THE SPECIES CONCEPT IN THE GENUS CITRUS
      About a century and a half ago, several treatises were published on the citrus fruit trees.   The first of these was the Traité du Citrus, by Georges Gallesio (1811), followed by the work of J. A. Risso (1813) and J. L. A. Loiseleur-Deslongchamps and Etienne Michel (1818), and finally by the Histoire naturelle des orangers of Risso and A. Poiteau (1818-1822).   In these voluminous works, every cultivated variety was given a name consisting of a Latin phrase of from three to ten, or even more, words.   As these polynomials did not fit into the Linnaean system of plant nomenclature then in universal use by botanists, they were never incorporated into the taxonomic literature, especially since botanists considered that hybrids and monstrosities originating in orange houses or gardens were not deserving of technical botanical Latin names.
      The view of professional botanists on this plethora of names is perhaps best indicated by quoting a paragraph relating to the genus Citrus written by Daniel Oliver.   In his monograph on all the other genera of the orange subfamily occurring in India, Oliver (1861, p. 1) gave his reasons for not attempting to treat the species of Citrus.   He wrote: "Feeling it utterly hopeless usefully to define them [the species of Citrus], I have thought it best to leave the Oranges, Lemons, Limes and their allies as I found them.   It has been difficult to form, for purposes of comparison, any satisfactory approximate estimate of the number of species in this very variable and widely cultivated genus.   From the data afforded by Risso, Loureiro, Wight and Arnott, Miquel, and others, and from what I have myself seen of dried or living specimens, I have assumed five as about the probable number."
      Fourteen years later, in 1875, J. D. Hooker in his Flora of British India included Citrus among the thirteen genera that he recognized as comprising the orange subfamily, but recognized only four species in Citrus, in India proper, Ceylon, and Burma!
      As late as 1896, Adolph Engler, who for many years had specialized on the Rutaceae, recognized only six species of Citrus for the entire world, and in these he included three species now placed in other genera.   Finally in 1931, as his last and greatest taxonomic contribution, Engler revised his account of the Rutaceae for the second edition of Die naturlichen Pflanzenfamilien.   In this he included eleven species in the genus Citrus and some six or seven species (which he had placed in Citrus in 1896 but now considered to be distinct) in the two genera Poncirus and Microcitrus.
      In the meantime, early in the present century, evidence of the most convincing sort, based on many hundreds of accurately safeguarded cross-pollination experiments made by the citrus experts of the former Bureau of Plant Industry of the U. S. Department of Agriculture, began to be published.   This evidence showed that all the species of Citrus tested, and also species of two other related genera, could be hybridized without difficulty and that these hybrids often manifested astonishing variability in the first generation.   What was even more surprising, many of these hybrids, especially those between rather remotely related species, bore apparently normal seeds that reproduced the parent hybrid exactly.   In reality it was found that such hybrids often produced no viable egg cells, but, if pollinated, tiny buds from the nucellar tissue of the mother plant would grow into the embryo sac and there develop into apomictic embryos that, of course, reproduced the mother plant exactly (Traub and Robinson, 1937).
      It is therefore highly probable that in China, Japan, India, and other Oriental countries accidental hybridization has occurred in dooryard or village plantings of assorted citrus fruit trees through the action of bees and other insects which seek the abundant nectar of Citrus flowers.   It is also reasonable to suppose that if hybrids did arise and if chance seedlings came into fruiting they would be propagated by the villagers if the trees bore fruit of superior quality.
      Many year of intimate contact with the citrus hybrid populations gave Swingle and his colleagues a clear understanding of the characters and nature of some of the more valuable of these hybrids between different species of the commonly cultivated citrus fruit trees.   Armed with this experience, Swingle made three trips between 1915 and 1926 to the citrus-growing regions of Japan, China, and the Philippine Islands, where he paid special attention to hybrids and mutations of pummelos, oranges, and other citrus fruits.   He recognized in Oriental villages many of the types of hybrids that had been made in the United States by carefully safeguarded cross-pollinations!   In the Orient, these had originated as accidental hybrids, doubtless due to cross-pollination by insects.   For example, citranges, hybrids of the trifoliate orange and the sweet orange, were found growing in dooryards in the Yangtze Valley region of China, and tangelo-like fruits, undoubtedly hybrids of pummelos and loose-skinned citrus fruits, were not uncommon in Japan!   After the mass of evidence regarding the origin and nature of citrus hybrids was brought to light and published in detail in well-illustrated papers, evidence that proved the possibility of producing large numbers of striking new forms of citrus fruit trees by cross-pollination of the commonly cultivated species of Citrus, it is indeed surprising that A. W. Lushington in India, in 1910, and T. Tanaka in Japan, from 1924 to date, should have given species rank to scores of citrus forms that undoubtedly are hybrids.
      Tanaka (1935b, p. 576) realized that many of his "secondary or miscellaneous" species are chance hybrids.   In discussing the fifteen so-called "primary elements or species" of Citrus, Poncirus, and Fortunella,20 and the "secondary or miscellaneous species derived from them," he stated that the lack of secondary species among the pummelos (Citrus grandis) of southern China, "is to be explained by the lack of cooperation of compatible elements (species) acting as the mating pair to originate new species."21   Actually, one of the conditions that hinders the hybridization of C. grandis with other species is that the anthers lie against the stigmatic surface and release their pollen before the bud opens.   Later Tanaka stated: "Compatible combinations of primary species22 encourage the creation of such secondary elements…of great significance to the evolution of the Citrus flora of the world."
      In 1932 Tanaka (1932d, p. 263) made the following statement concerning the Calamondin orange: "The admittance [acceptance] of Citrus microcarpa [= C. mitis, a distinct bigeneric hybrid between Citrus and Fortunella] is another of the endless examples, which present no contradiction to the writer's system of nomenclature…"   Tanaka also stated explicitly (1927a, p. 54) that new forms of Citrus to which he had given taxonomic rank as genuine new species might have arisen from "chance seedlings."   He considered that the Japanese Natsudaidai, which he held to be a good species (C. natsudaidai Hayata), was "unquestionably a chance seedling originated at Hagi in Yamaguchi Prefecture."   He also considered the satsuma orange to be "botanically a good species, Citrus unshiu Marc.," that had originated some centuries earlier in western Japan as a chance seedling of one of the famous oranges of Hwangyen, Chekiang Province, China.   In discussing the cultivated Ichang lemon (probably a hybrid of the Ichang papeda, C. ichangensis Swing., with some other species of Citrus grown in central China), which he named C. wilsonii (1932b, p. 37).   Tanaka stated: "The present species unquestionably has Citrus junos in one parent and seems to be a cultigen originated through the chance seedling."   There is no proof, however, that any one of the chance seedlings discovered and propagated by man as cultigens could have persisted and become established as a wild species.
      The taxonomic delimitation of species, subspecies, and forms in Citrus and closely related genera is complicated by three peculiarities in the reproduction of citrus fruit trees: (1) polyembryony through formation of nucellar-bud embryos, (2) rejuvenation by neophyosis of such nucellar-bud embryos of more or less senescent varieties long propagated asexually, and (3) the spontaneous production of autotetraploid forms.
      Hybrids of Citrus, both intergeneric and intrageneric, are frequently egg-sterile; yet, when pollinated, they often set seeds.   These seeds are devoid of sexually derived zygotic embryos ("gametic" embryos of some authors), but contain one or more embryos within the embryo sac apomictically produced by ingrowing buds arising from the adjoining undifferentiated nucellar tissue of the female parent.   This phenomenon was first described by Strasburger (1878) and later was discussed by Frost (1626) and Traub (1936).   Such nucellar-bud seedlings reproduce the mother plant exactly, so that many violent F1 Citrus hybrids, such as citranges (C. sinensis X Poncirus trifoliata) and tangelos (C. paradisi X C. reticulata), reproduce perfectly from seeds and sometimes are grown on a large scale.   Taxonomists have found such egg-sterile hybrids to be confusing, since they do not show the usual wide variation in the second generation characteristic of normal F2 seedlings but behave like very constant true species.
      In many species of Citrus, such nucellar embryos may occur together with the zygotic embryo.   In such cases, the zygotic embryo may produce a plant that varies somewhat in its characters from the mother plant, whereas the nucellar embryos reproduce the mother variety almost exactly.   Nucellar-bud seedlings as well as sexual seedlings apparently restore the variety to its primitive or juvenile condition, because such plants not only develop spines abundantly but show greater energy of growth and greater hardiness and sometimes produce somewhat larger or otherwise superior fruit.   Two nucellar strains of a well-known mandarin orange (Dancy tangerine) were named the Weshart and Trimble tangerines by Webber and Swingle (1905, pp. 238-40 and pl. 20-22).   The Silverhill satsuma is another striking example of a nucellar seedling of the Owari satsuma (see Swingle, 1931).
      Similar rejuvenation of nucellar seedlings was observed in experimental work with the Washington navel orange where there could be no confusion between the zygotic and the nucellar embryos arising from the same seed, since the zygotic embryos were citranges obtained through pollinations with Poncirus trifoliata.   This rejuvenation of the old cultivated varieties of Citrus has been called neophyosis (see Swingle, 1932, 1933).   Nucellar strains of old varieties rejuvenated through neophyosis might easily be mistaken for taxonomically distinct varieties or even subspecies, because of their vigorous growth, their large leaves and fruit, and their luxuriant development of spines which are often lacking on varieties long propagated vegetatively.
      Finally, thick-leaved seedlings of several species of Citrus were found by Frost (1925, 1926) to be autotetraploid forms of the parent species; the fruits have a thicker rind with larger oil glands, are usually more depressed-globose, and have less juice, and more and larger seeds than the diploid parental species from which they originated.
      As a matter of fact, from the standpoint of phylogenetic taxonomy, which records the evolutionary descent of natural wild species, only Tanaka's "primary elements or species" deserve consideration as true species; chance seedlings, mutations, and hybrids, propagated usually by grafting, should be given a status similar to cultigens of apple or rose varieties that no taxonomist of standing has proposed to name as good species, although they often exhibit important differences in flowers or fruits or other organs as well.
      Once this nomenclatorial confusion is cleared up, Tanaka's studies of the myriad cultivated races of Citrus and Fortunella in the Far East will be appreciated at its full value and he will receive recognition as having greatly advanced the study and classification of citrus fruits.   This task can then be carried on without upsetting the scientific taxonomy of Citrus and Fortunella, which is based on the native wild species originally propagated spontaneously from seed without human help.
      A. E. Kozhin, the well-known Russian citrologist, published a notable paper on the citrus fruit (1931) in which he gave particular attention to the very extensive literature on the subject and to the taxonomic history of Citrus and the genera most closely related to it.   He strongly objected to the practice of considering "garden species" as species in the taxonomic sense.   He contrasted such false species, some of them seedless and unable to live at all without man's help and most of them propagated by clones, with true species.   He said in part: "…a Linnean species is a whole system of elementary forms which, continually crossing with one another, keep within the limits of definite specific characters.   In drawing up such a system of forms, the geographic moment, i.e., the conditions of habitat, play an important role…From this point of view the attempt to introduce into the system as equivalent members (as has been done by Tanaka), the wide-compassed Linnean species and such narrow forms as garden species, frequently seedless ones, is a failure."
      More recently Stebbins (1950) discussed the species concept in agamic complexes, and stated that "…the species concept…maintained by the majority of those who desire a truly biological concept of species, centers about the possibility for exchange of genes between members of the same species and separation of different species by barriers to the exchange of genes.…Such a concept cannot be applied to agamic complexes.   Free exchange of genes between apomicts is prevented by the very nature of their type of reproduction, while the origin of many apomictic clones is from genotypes which have combined the genes of previously isolated sexual species, and which without apomixis would not be able to persist because of their sexual sterility.   It is not strange, therefore, that systematists have not been able to agree on the boundaries of species in…[apomictic genera]."   Stebbins continued: "In attempting to set up species like those in sexual groups, they are looking for entities which in the biological sense are not there.   Nor are those 'splitters' who make a separate species out of every apomictic clone…likely to provide any better concepts of the variation patterns in the genera concerned.…One criticism is that the number of apomicts in any well-developed complex is so large that recognizing them as separate species makes the comprehension of the group as a whole difficult or impossible; one cannot see the forest for the trees!   But more important is the fact that in many complexes the great majority of the apomicts, and in nearly all complexes at least some of them, are only partly or facultatively apomictic.   From time to time they reproduce sexually, and on such occasions a whole series of new clones, or 'species', may arise in the offspring of a single individual."   Once "splitting" is begun on the basis of fruit characters or small detail, it would be possible to justify thousands of species in the genus Citrus.   Furthermore, this number would be constantly augmented by additional "species" as the citrus breeder produces new apomictic hybrids each year.   As Lawrence (1951) has pointed out, the presence of apomixis makes possible the survival of hybrids that would otherwise be eliminated under natural conditions by their sterility.   Lawrence stated, "The perpetuation of these apomictic hybrids has resulted in some descriptive taxonomists treating each biotype as a morphologically distinct and seed producing species."
      Two major systems of citrus taxonomy and nomenclature have developed.   Swingle's system is presented here.   The other is the system proposed by Tyôzaburô Tanaka (1927c, 1935a, 1954, and 1961), and supported by Yuichirô Tanaka (1948).   Each taxonomic treatment is based upon years of research and study by very competent men of widely divergent viewpoints.
      Swingle began the first known attempts at artificial hybridization of citrus in 1893.   His experience as a citrus breeder and his observations of large seedling populations produced by controlled cross-pollination lead to the development of his system of classification.   He assembled an extensive collection of Citrus material for study upon which he based this system.
      Reece, the junior author, who was given the task of revising this chapter for this edition of The Citrus Industry, makes no pretense of competence in the field of taxonomy.   On the other hand, he has devoted many years to Citrus breeding.   He considers that Swingle's system comes closest to the truth of Citrus taxonomy, and he cannot accept the Tanaka system with its excessive splitting.
      In following up his interest in Citrus taxonomy, Tanaka has traveled widely and observed a wide range of citrus.   He is familiar with both the original literature and herbarium material.   Swingle and Tanaka became associated and collaborated together at various times between 1915 and 1930 in this country and Japan.   Nevertheless, their divergent viewpoints are apparent in their systems and conclusions.   Their disagreement is confined principally to the species problem in the genus Citrus.
      Although Tanaka in 1935 recognized only fifteen primary species in Citrus, Poncirus and Fortunella, he also stated, in an English summary of a Japanese paper (Tanaka, 1935a): "The author [Tanaka] enumerated 74 species as such well-established species of Citrus, Fortunella and Poncirus in the present paper, but the mobilization of all species from various Citrus districts is now on the way of progress."   In a later paper (1937), he listed no fewer than four additional "species" of Lushington (published in 1910) and described six new species himself, as well as four new varieties, all from India.   In several papers since 1935, he mentioned en passant many other new species or varieties without describing them.   This brought the total to about 100, but the end was by no means in sight.
      Tanaka (1954) established the foundation for his classification system with the publication of Species Problem in Citrus.   In that work, he classified the forms of Citrus into two subgenera, eight sections, thirteen subsections, eight groups, two subgroups, two microgroups, and 145 species.   Seven years later (Tanaka, 1961), his system was expanded to include two new subsections, Eulimonellus and Megacarpa, one new group, Paranobilis, and twelve new species.   Further refinements have continued, however, and a classification chart sent by Tanaka to some of his colleagues in 1966 contains five new subsections and includes 159 species.   In contrast to the Tanaka system, the Swingle classification recognizes two subgenera: the subgenus Papeda containing six species, and the subgenus Citrus (formerly Eucitrus) with ten species.   A comparison of the major classification systems of the genus Citrus is presented in table 3-3.   Systematics are given for the Tanaka classification of 1961, which contained 157 species, and the Swingle classification of 1943 followed in this chapter.   In addition, the table includes columns for species recognized by the pre-Englerian, Englerian, and early Swingle and Tanaka systems.
      Tanaka recognizes thirty-five species of mandarins.   In some cases the only distinguishing character in his key is leaf size or fruit size.   Anyone familiar with Citrus production knows that size is easily influenced by rootstock, soil, amount of moisture, fertilization, and weather conditions prevailing during development of a particular flush of growth.
      Tanaka's arguments for granting species rank to some of the mandarins are interesting and often quite unique.   In regard to making the satsuma a separate species from Citrus reticulata, he says: "Pulp qualities are…the real object of human interest to consume the fruit.   The public knows more about such specific distinction than a few technicians because they consume more.   To the Japanese, who eat 500,000 tons of satsuma fruit a year, they would never believe it to be a mere variety or clone of either the Ponkan (C. reticulata) or Kunenbo (C. nobilis), as is true to Americans not admitting the grapefruit as a variety of the pummelo (Shaddock).   It is a human belief and science should endorse it."   (See Tanaka, 1954, pp. 20-21.)
      In support of his argument that the Ponkan, the Dancy tangerine, and the C. crenatifolia of Lushington are separate species and not varieties of C. reticulata, he comments that "if these three should belong to a single species, 350,000,000 Indian people would not believe in science."   However, the objective of science is to seek the truth rather than the support of human beliefs.   Nevertheless, few biologists doubt that new species can result from hybridization, but the new species should be able to perpetuate itself by sexual methods.   It should not be dependent upon man for some type of vegetative propagation.   For this reason a number of Tanaka's "species" are not considered valid by the authors.   Furr and Reece (1946) found "Citrus temple and C. clementina of Tanaka" unable to reproduce their type even by nucellar embryony because nearly all embryos are zygotic.   Reece and Childs (1962) in a study of a seedling population of Persian lime (C. latifolia Tanaka) have shown that this variety is apparently a hybrid between a seedy lime (C. aurantifolia Swing.) and citron as the other parent.   If, indeed, a seedy lime and a citron were the parents of the Persian lime, the appearance of the many lemon types of seedlings in the population supports Swingle's hypothesis that the lemon may be of "hybrid origin, perhaps having the citron and the lime for parent species."
      Santiago (1962), in a guide to Malayan nomenclature of citrus species, followed Swingle's system of classification with a few exceptions.   Bhattacharya and Dutta (1956), in their comprehensive monograph on the citrus fruits of Assam, followed Swingle's system in a general way, but accept C. nobilis Lour., C. karna Raf., C. limetta (Risso) Lush., C. jambhiri Lush., and C. megaloxycarpa Lush., five species of the Tanaka system that were not accepted by Swingle.   They find the nomenclature of one of these Tanaka species invalid (C. pennivesiculata Tanaka = C. megaloxycarpa Lush.), and they name one new species found in Assam (C. assamensis Dutta and Bhattacharya, sp. nov.).   This is known as ada-jamir or the ginger citrus of India because people consider its aroma similar to that of ginger.   It is taxonomically close to C. hystrix DC., but differs from it to an extent considered sufficient by these authors to deserve a separate rank.
      Several other systems of citrus classification have been put forward by various authors.   Perhaps the one that has received the least attention by contemporary students of Citrus taxonomy is the system proposed by Marcovitch (1926), which bases classification on leaf and flower characteristics only.
      Wolfe (1959) has criticized both Swingle's conservatism and Tanaka's system with its excessive splitting.   He takes an intermediate position and recognizes eleven species not considered valid by Swingle and Reece.   These species are indicated in table 3-3 (see footnote).   The authors' reasons for their rejection of these species are stated elsewhere in the text under the discussion of the several forms whose taxonomic rank is in dispute.
      The late R. W. Hodgson made two trips to the Orient and studied the principal Citrus collections in India for nearly a year.   He also studied a large number of Citrus introductions from many parts of the world.   He worked in close association with Tanaka for nearly a year when Tanaka was in the United States as a Fullbright research scholar.   Hodgson (1961; Hodgson, Singh, and Singh, 1963a, 1963b, and 1963c) published his nomenclature of Citrus which recognized twenty-three Citrus species of Tanaka's system.   Swingle's treatment was criticized by Hodgson on the basis that it denies species standing to ancient, well-known, distinctive forms of economic importance.   However, such criteria are not sound reasons for species standing if they are apomictic and show evidence of their hybrid nature.   It is true that Swingle's treatment may involve speculation concerning the probable parentage and even error in proper placing of the form in the attempt to arrive at a natural system of classification.   However, if they are made in an attempt to understand the natural relationships of biological organisms, do such errors merit greater censorship than a more artificial system that names as species forms that are known or probable hybrids, obvious apomicts, or horticultural forms that can only be perpetuated by vegetative propagation?
      Distinctive or easily recognized form is not an adequate reason for naming species.   The situation in Citrus is somewhat analogous to the situation in Canis domesticus so well expressed by Dobzhansky (1955): "The longer a species has been domesticated the more numerous and more diversified are its domestic varieties.   Thus the dog is the most ancient domestic animal, and has produced the largest number of breeds.   Furthermore, some dog breeds differ so much in appearance, as well as in temperament and behavior, that it is hard to believe that they all belong to the same species.   Compare the little Chihuahua with a Great Dane; or an ebullient Fox Terrier with a stolid Saint Bernard.   Nevertheless, all these breeds do belong to one species, since they are capable of exchanging genes, and often do so, either directly by occasional hybridization, or via the intermediate breeds where crossing is impossible because of extreme difference in size.   The dog species has a common gene pool, subdivided, to be sure, into the gene pool of the various breeds."   Zoological taxonomists consider that those distinctive forms are "cultivated varieties" not different species.   Certainly the related forms or "breeds" of citrus (the species of Tanaka) have a common gene pool since they can and do exchange genes through hybridization.
      The study of D. Singh, and C. A. Schroeder (1962) on the taxonomic and physiological relationships of the so-called mandarin-lime group of Citrus caused them to conclude: "…the so-called mandarin-lime of Citrus is a misnomer in that lime parentage is not indicated.   The possibility of mandarin-rough lemon parentage is suggested."   They contend that the binomial Citrus limonia Osbeck should be employed as a species designation for members of this group.   Determination of the actual parentage of this group is not easy to establish, but hybrid origin is not only possible, but highly probable.   Apomixis strengthens this probability.   The percent of polyembryony in germinated seeds does not, however, indicate the degree of apomixis in these varieties.   It only indicates the number of seeds in which two or more embryos were of sufficient size and strength to emerge on germination and single embryos are also of apomictic origin in many instances.   On the basis of probable hybridity and apomixis, the binomial of Citrus limonia as a species designation for this group is denied.
      The choice as to which of these taxonomic systems to follow will depend upon the personal viewpoint of the student of citrus.   The biological facts in the case will not be altered by arguments for or against either system.   However, serious difficulties arise when taxonomists rank apomictically perpetuated biotypes with sexual species as Heslop-Harrison (1956) has pointed out.   Some taxonomists following such methods have described as species over 1,100 forms in the hawthorn genus, Crataegus, over 4,000 in the blackberry genus, Rubus, and over 10,000 in the hawkweed genus, Hieracium.   It would be a simple matter to embark upon a similar path and describe an astounding number of Citrus "species," but the multitude of binomials, unnecessarily complicated keys, and cumbersome system of nomenclature would make the comprehension of the group difficult indeed.
      In the two decades since World War II, truly remarkable progress has been made in describing the chemical constituents in citrus tissue.   Because of their great economic importance and adaptability to industrialization, the common citrus fruits have been the subject of more intensive biochemical investigation than perhaps any other horticultural crop.   As a result, hundreds of compounds, particularly those found in fruits and seeds, have been isolated, identified, and in many cases assayed (Braverman, 1949; Bartholomew and Sinclair, 1941; Sinclair, 1961; U.S. Dept. of Agriculture, 1962).   More recently such research has been extended to citrus varieties and species of little or no economic importance and also to some of the near relatives of citrus.
      The material summarized in table 3-4 provides an example of this type of research and indicates the great variety of flavonoids found in the genus Citrus and some relatives.   Similar research is in progress on essential oils (U.S. Dept. of Agriculture, 1962; Stanley, 1963a; Scora, 1966; Scora, et al., 1966; Scora and Newman, 1967; Scora and Torrisi, 1966), coumarins (Stanley, 1963b), carotenoids (Yokoyama and White, 1965), limonoids (Dreyer, 1965), and other compounds found in leaves, fruits, or seeds of the Rutaceae family.
      In time the pattern distribution of some of these compounds may prove to be distinctive within certain genera, species, or other groupings.   Such chemotaxonomic considerations, when added to the classic morphologic and cytogenetic bases of classification, should contribute toward a more penetrating insight into the species problem of citrus.
      A key to the subgenera and species of Citrus is presented.

Subgenus Citrus
      Subgenus Eucitrus Swing. in Webber & Batchelor, Citrus Indus. 1:395. 1943.
      Pulp-vesicles with few and minute, or sometimes no, oil droplets (never containing acrid oil); petioles wingless or narrowly winged, or, if broadly winged, then never 3/4 as broad as the leaf blades; flowers large and fragrant; stamens cohering in bundles.
      The genus Citrus is divided into two very distinct subgenera, Citrus and Papeda, easily distinguished by leaf, flower, and fruit characters.   The subgenus Citrus includes all the commonly cultivated species of Citrus, all of which have pulp-vesicles filled with pleasantly acid, subacid, or sweet juice, free, or almost free, from droplets of oil.   On the contrary, none of the species of Citrus belonging to the subgenus Papeda have edible fruits, as the pulp-vesicles have dense aggregations of droplets of acrid oil that form an axile column and usually give the juice a very disagreeable, acrid, bitterish flavor.
      In the course of a comparative study of the vascular traces of the flowers of all obtainable genera of the orange subfamily, Tillson (1938, pp. 21 and 30; also Tillson and Bamford, 1938, pp. 788 and 790) brought to light a striking difference between the species of Citrus belonging to the subgenus Citrus and those classed here in the subgenus Papeda.   The commonly cultivated species of Citrus, the sweet orange, the mandarin, the grapefruit, the lime, and the lemon, all belonging to the subgenus Citrus, show "not only fusion of the sepal midrib with the lateral petal bundles [as is usual in the tribe Citreae], but also fusion of the lateral sepal bundles with the petal midrib.   When the petal midrib reaches its position beneath the petal base, one or more branches from it continue into the base of the calyx, forming the lateral sepal bundles."   On the contrary, Tillson found that the species of Citrus belonging to the subgenus Papeda showed no fusion of the lateral sepal bundles with the petal midrib (this was true even of C. ichangensis, which differs from the other species of the subgenus Papeda in having large flowers, much like those of the species of the subgenus Citrus).   In this character of its floral anatomy the subgenus Citrus differs not only from the subgenus Papeda, but also from all the other genera, Fortunella, Eremocitrus, Poncirus, Clymenia, and Microcitrus, all closely related to Citrus, that comprise the True Citrus Fruit Trees.
      1.   Citrus medica L. Sp. Pl. 2:782. 1753. Citrus tuberose Mill. Gard. Dict. ed. 8. 1768; C. odorata Roussel, Fl. Calvados 194. 1796; C. fragrans Salisb. 1796; C. cedra Link, 1831; C. cedratus Raf. 1838; C. crassa Hassk. 1844. Illus. Risso & Poiteau, Hist. Nat. Orang. pls. 96, 97, 99, 100 (all col.). 1818-1822; Bentley & Trimen, Med. Pl. 1:pl. 53 (col.). 1880; Swingle, in Bailey, Stand. Cycl. Hort. 2:779, fig. 971. 1914; and many others.
      Type.—Wanting, but plant seen and studied by Linnaeus.
      Distribution.—China and India southward; cultivated in many subtropical regions.
      Common name.—Citron.
      A shrub or small tree of irregular habit of growth; twigs angled and purplish when young, soon cylindrical, glabrous, with stout, short, single spines in the axils of the leaves; leaves glabrous, elliptic-ovate or ovate-lanceolate, bluntly pointed or rounded at the tips, cuneate or rounded at the base, margins serrate; petioles short, wingless or narrowly margined, not clearly articulated with the leaf blade; inflorescences short, few-flowered racemes; flower buds large, purplish; flowers perfect or male by more or less complete abortion of the pistil; petals 5, pinkish on the outside; stamens very numerous, 30-40 or even 60 as found by Webber (1923, pp. 112-20); ovary large, bulged, cylindrical, with 10-13 locules, tapering into the thick style, which is sometimes persistent; fruit large, oblong or oval, surface smooth or more often rough and bumpy, fragrant, yellow when ripe, rind very thick, segments small, filled with pale greenish pulp-vesicles with acid or sweetish pulp; seeds numerous, small, 9-10 X 4-5 X 3-4 mm, pointed at the base, smooth; embryo white.   (See Hodgson's description of the cultivated varieties of citron in this work.)
      The citron was the first citrus fruit to reach the Mediterranean region.   Apparently it was introduced to the eastern Mediterranean area following the invasion of Persia by Alexander the Great about 325 B.C.   Theophrastus (writing about 310 B.C.) called the citron the Median or Persian apple.   He said that it was inedible but very fragrant and a remedy for rheumatism and sore mouth, as well as a repellent to drive away moths.   Engler stated (1931, p. 338): "Since the fruits had the same uses as the wood of the Sandarak tree, Callitris quadrivalvis, the [ancient] name of this wood 'Citrus' was transferred to the fruit as Mala citrea."   This renaming of the Median apple as the Citrus apple led to the transfer of the name "Citrus" first to the citron and later to other citrus fruits.
      The native home of the citron has not been determined with certainty.   The citron is commonly supposed to be indigenous to India but J. D. Hooker, who said (1875, p. 514) that he had no doubt about the citron's being truly wild when he found it growing "mainly on dry sunny slopes [in Sikkim] totally unsuited for any kind of cultivation…," later came to doubt its being indigenous.   Bonavia (1888, p. 70) stated: "I am still in doubt whether it [the citron] is indigenous in India.   It does not appear to have any ancient Sanskrit name and the number of varieties, if they are variations, on the western seacoast is suggestive.   It is curious that they should be found in the area which came most in contact with foreigners."
      The citron has been grown since ancient times in China, but Chi Han, minister of state under the Emperor Hui Ti, in a work written about 300 A.D. (Nan fang ts'ao mu chuang) mentioned the arrival, in 284 A.D., as tribute to the Chinese Emperor, of 40 Chinese bushels of citrons from Ta-ch'in (a name usually meaning the Roman Empire).   He stated: "…the Barbarians value the citron very highly.   It is aromatic and its flesh is very thick and white…"   This early Chinese record of the citron would indicate that it was not indigenous to China but had been introduced from the West.
      The early advent of the citron in Media and Persia, and its subsequent slow penetration into India and China, could be explained easily if the citron should prove to be a native of southern Arabia.   The bael fruit of India, Aegle marmelos, has no close relatives in Asia, but three closely allied genera, Aeglopsis, Afraegle, and Balsamocitrus, are found in Africa.   Citropsis, an African genus of the Near-Citrus Fruit Trees closely related to the Asiatic genus Atalantia, has eleven species.   It would not be surprising to find midway between India and Africa, in some mountain oasis within the tropical zone in Arabia, the citron growing in a wild state.   Over a century ago Wellsted (1838, vol. 1, pp. 126-52) found gardens in the Jebel Akhbar Mountains (150 km southwest of Mascat) where grapes grew abundantly; also "pomegranates, citrons, almonds, nutmegs and walnuts with coffee bushes."   Over thirty years ago Bartram Thomas (1932, map, p. 101) explored thoroughly the Qara mountain range, about 1,100 km farther to the southwest, and found it to be "an Arcadia of luxuriant forests that clothe steep mountains with perennial streams."   In these mountains, situated in a summer rain belt along the coconut-fringed shores of the Arabian Sea, Thomas found giant, large-fruited wild fig trees and "wild-growing, bitter, limes" fruiting abundantly, as well as an extensive growth of frankincense trees at elevations of from 2,000 to 2,500 feet above sea level.   Search should be made in this region, between eastern Hadhramaut and Oman, for the native home of the citron.
      The gradual increase in use of the citron can be traced in early literature.   Theophrastus, writing at Babylon about 310 B.C., said the citron "was not eaten."   Plutarch, writing between 81 A.D. and 96 A.D., stated that "many substances which in the past people would neither taste nor eat, are considered today as very agreeable.…Shall we mention the cucumber, the melon, the Median apple, and pepper?" (Tolkowsky, 1938, p. 91).   By the second century the epicurean Apicius Caelius could recommend the following dishes as being exceptionally delicate: (1) the white inner part of the peel of citron made up into a salad, and (2) small pieces of citron peel served with fish mixed with herbs, vinegar, oil, and spices (Tolkowsky, 1938, p. 59).   Before long citrons had become a prized article of food in Rome, and in 301 A.D. the records show that their sales prices were officially fixed by Diocletian at values ranging from twelve to sixteen times the price of melons.
      A method of candying citron peel was finally discovered in the Mediterranean region.   This involved the softening and clearing of the peel, before in was candied, by fermentation in sea water through the addition of a mixed culture of a yeast and a bacillus.   Candied citron peel ultimately supplanted almost completely the use of fresh peel.   However, as the fresh citron peel, like that of lemon and orange, contains hesperidin (Penzig, 1887, p. 286), it is probable that it will prove to be a good source of vitamin P because of its thick mesocarp, in which the hesperidin is found.   The peel should be tested for the making of "citrin" (see also table 3-4 and below).
      1a.   Citrus medica var. sarcodactylis (Noot.) Swing. Pl. Wilson. 2:141. 1914. Citrus sarcodactylis Noot. Fleurs, Fruits, Feuill. Java 1:pl. 3. 1863. Illus. Nooten, loc. cit., pl. 3; Swingle, in Bailey, Stand. Cycl. Hort. 1:781. 1914.
      Type.—Lacking in Linnean Herbarium.
      Distribution.—Widely cultivated in China, Japan, Indo-China, and India.
      Common name.—Fingered citron.
      Like the species except in the fruit, which is split into a number of finger-like sections.   Usually pulp is lacking, or if present is very scanty.
      The fingered citron is well known and highly esteemed for its fragrance and beauty in China and Japan, where it is called "Buddha's Hand Citron" (Fo Shou kan in Chinese, Bushu-kan in Japanese).   It is used by Chinese and Japanese for perfuming rooms and clothing.   It is also grown as a dwarf plant, of which good fruiting specimens are highly prized for ornamental purposes.
      1b.   Citrus medica var. ethrog Engl. Die Nat. Pflanzenfam. 19a:338. 1931.
      Type.—Europe (Engler?).   Herb. Bot. Mus., Berlin-Dahlem.
      Distribution.—Western Asia and Mediterranean countries; known only in cultivation.
      Common name.—Etrog citron.
      Fruits small, ellipsoid to fusiform, rough, with a nipple ending in the persistent style with the persistent stigma.
      The "Etrog" of the Jews, used in the Feast of Tabernacles, is not mentioned in the Bible.   It probably did not reach Palestine until after the time of Alexander the Great and was not used by the Jews in fulfilling the prescriptions as given in Leviticus 23:40.   Immanuel Löw (1924, p 286) stated that its use had been recorded from the time of Alexander Jannaeus (104-78 B.C.).
      On the basic of the detailed prescriptions governing the character of the "Etrog" suitable for use in the Feast of Tabernacles, it is highly probable that the present-day Etrog is a survival of an ancient form of the citron, perhaps much like the original form of the species from which the ordinary citron has been developed by selection.
      2.   Citrus limon (L.) Burm. f. Fl. Ind. 173. 1768. Citrus medica var. limon L. Sp. Pl. 2:782. 1753; Limon vulgaris Mill. 1768; Citrus limonum Risso, 1813; C. medica var. limonum Hook. f. Fl. Brit. Ind. 1:515. 1872. Illus. Loiseleur-Deslongchamps, Nouv. Duhamel 7:pl. 28. 1818; Risso & Poiteau, Hist. Nat. Orange. pl. 70 (col.). 1818-1822; Hayne, Arzn. Gew. 11:pl. 27. 1830; Bentley & Trimen, Med. Pl. 1:pl. 54. 1880; Ford, Bot. Gaz. 104:288-305. 1942; and many others.
      Type.—Europe? (L.?).   Herb. Linn. in Linn. Soc. London.
      Distribution.—Southeastern Asia (?); widely cultivated in all subtropical countries.
      Common name.—Lemon.
      Small thorny trees, young leaves and flower buds reddish; leaves pale green, long-ovate, pointed at the tip, margins serrate or subserrate; petioles narrowly winged or margined, plainly articulated with the leaf blade; flowers reddish-tinted in the bud; petals white above, purplish below; stamens numerous, 20-40; ovary subcylindric or barrel-shaped, tapering into the thick deciduous style; fruit oval with a broad, low apical papilla, with 8-10 segments; peel yellow when ripe, rather thick, prominently glandular-dotted; seeds small, ovoid, pointed, smooth, white within.
      The origin of the lemon is a mystery.   Tanaka (1929b, p. 342) suggests that it was not introduced to China until the Sung Dynasty (760-1297 A.D.), and it is still rare in India (according to Biraghi, 1935).   It was actively spread in the Mediterranean region by the Arabs about 1000 to 1200 A.D.   Like the sour orange, it became widely and favorably known as a medicinal agent.   According to Glidden (1937, p. 382), the lemon was first described in detail by Ibn-Jami, a physician at the court of Saladin (1171-1193 A.D.), in a treatise on the medical uses of the lemon (now lost; parts, however, have been preserved in quotations).   The lemon was considered by Linnaeus to be a variety of the citron (C. medica) and obviously closely related to it.   Probably the lemon should be considered as a satellite species of the citron; possibly it may prove to be of hybrid origin, perhaps having the citron and the lime for parent species.   As is true with the grapefruit, it is difficult to explain the origin of the lemon as a hybrid, as it crosses readily with other species of Citrus and yet, when self-pollinated, reproduces itself from seed with only small variations.   For this reason it is left here as a distinct, but probably satellite, species of Citrus.23
      The glucoside hesperidin was found in fruits of the lemon tree by Pfeffer (1874, p. 529).   A comprehensive account of the biochemistry of the glucosides, organic acids, sugars, vitamins, essential oils, and other constituents found in the lemon is given in The Lemon Fruit by Bartholomew and Sinclair (1951) and in U.S. Department of Agriculture Agricultural Handbook 98 (1962).
      The true lemon is of recent introduction and is still very rare in southern China, as was shown by Kwok Wa-shau, who in 1921 and 1922 made, under Swingle's direction, for the Bureau of Plant Industry (now encompassed by the Agricultural Research Service) a detailed survey of all the citrus fruits cultivated or found growing wild in the vicinity of Canton.   The so-called white lemon, pak ning-mong in Cantonese, and the red lemon, hung ning-mong, both of which grow commonly in southern China, are not true lemons, but are hybrids, perhaps between the lemon and the mandarin orange (see below).
      The origin of the lemon is still doubtful unless it proves to be a hybrid or sport of the so-called lemon of India.   The rough lemon, commonly used as a rootstock in the United States, is apparently widely naturalized or possibly indigenous in India and Pakistan, where it is called jambhiri.   According to Bonavia (1888-1890, p. 61), it was mentioned by the Emperor Baber in his memoirs, composed in 1519 A.D. (Bonavia figured the fruit in pls. 131, 132).   Hodgson (1937, p. 513) reported finding in India a form of jambhiri "indistinguishable from our Florida rough lemon."   It grows in a seminaturalized state along the Mazoe River in Southern Rhodesia, South Africa, where it is called the Mazoe lemon.   The rough lemon has been found to show a very high percentage of nucellar embryos, which would indicate a hybrid origin.   True lemons, on the contrary, show only a small proportion (10 to 15 per cent) of nucellar embryony.

HYBRIDS OF THE LEMON
      Lemonange ? [Citrus limon X C. sinensis ?].
      The only hybrid of this supposed parentage that has come into culture in the United States is the Meyer lemon, grown as a potted plant in Peking, China, and introduced into this country in 1908 by the late Frank N. Meyer, then agricultural explorer of the Foreign Plant Introduction Service of the former Bureau of Plant Industry.   It was figured and described by McKee (1927, pp. 218-21, text figs. 48, 49); see also Traub and Robinson (1937, p. 751).   It is probably a hybrid between the lemon and some other species of Citrus.
      Lemonimes [Citrus limon X C. aurantifolia].
      These hybrids, made by safeguarded, cross-pollinations, are intermediate between the two parent species.   The best known of these hybrids, the Perrine lemon (C. aurantifolia 'Mexican' X C. limon 'Genoa'), is noteworthy in having high resistance to lemon scab (caused by Elsinoë fawcettii) and also to lime withertip (due to Gloeosporium limetticolum), the worst diseases afflicting the two parent species.   The Perrine lemon is described and figured in Swingle, Robinson, and Savage (1931, p. 15, pl. 10), in McLennan (1937, 1 text fig. and col. front cover), and in Traub and Robinson (1937, p. 781, fig. 15).
      Lemandarins [Citrus limon X C. reticulata ?].
      Probably the Otaheite orange, the hung ning-mong, or "red lemon" of the Chinese of Canton, and the lemon called pak ning-mong or "white lemon" are both hybrids of this character, possibly complex hybrids.   Tanaka has shown (1925, pp. 107-26, and 1933b, col. pl. 8) that the name Citrus limonia Osbeck belongs or was assigned to one of these hybrids (not to the hung ning-mong, as he supposes, but to the pak ning-mong), but such hybrid fruits propagated artificially by man cannot be considered as true taxonomic species.   They arose doubtless as garden hybrids in southern China through crosses and pollinations effected by insects between the lemon, possibly introduced by the Arabs, and the native mandarin orange.
      Rumphius (1741, vol. 2, p. 101, pl. 26, fig. 2) figured a lemon called Lemon Martin, grown in the East Indian Archipelago by Europeans, that is almost certainly a hybrid between a true lemon, probably brought in by the Portuguese, and some other species of Citrus (possibly C. grandis or C. hystrix); nevertheless it is called a lemon by the Europeans who live in the East Indies and by other Europeans.   The same thing seems to have happened in southern China to the true lemon, probably transported there by the Arabs and then degenerated by hybridization.

      3.   Citrus aurantifolia (Christm.) Swing. Jour. Wash. Acad. Sci. 3:465. 1913. Limonia aurantifolia Christm. in L., Pflanzensyst.…nach d. Houttuyn. 1:618. 1777; Limonellus sive Limon Nipis Rumph. Herb. Amboin. 2:107. 1741; Limonia acidissimia Houttuyn. (non L.) Natuurl. Hist. 2(2):444. 1774; Citrus lima Lun. 1814; C. acida Roxb. 1832; C. spinosissima Meyer 1818; C. notissima Blanco, 1837; C. limonellus, Hassk. 1848; C.medica var. acida Hook. f., Curtis' Bot. Mag. 3 ser. 40:pl. 6745 (col.). 1884. Illus. Rumphius, loc. cit. pl. 29; Merian, Metam. Insect., Surinam. 17, pl. 17 (col.). 1705; Hooker f. loc. cit. pl. 6745 (col.); Bonavia, Cult. Orang. Lem. India Cey., Atl. pls. 228, 229. 1890; Hume, Cult. Citrus Fruits 19, 20, figs. 18, 20. 1926; Ochse, Fruits Fruitcult. Dutch East Ind. pl. 44 (col.). 1931; European iconographies of Citrus usually have misleading illustrations of this species!
      Type.—East Indian Archipelago, Amboyna, no type preserved; based on Rumphius' description and plate and on Merian's plate.
      Distribution.—East Indian Archipelago; widely grown in all subtropical regions.
      Common name.—Lime.
      A small tree with rather irregular branches; twigs with short, stiff, very sharp spines; leaves small, 5-7.5 cm long, elliptic-ovate or oblong-ovate, obtusely pointed at the tip and rounded at the base, margins crenulate, pale green, petioles narrowly winged, spathulate; inflorescences axillary, short, lax racemes of 2-7 flowers (rarely single); flowers small, white in the bud, calyx cupulate, 4-5 lobed; petals 4-5, 8-12 X 2.4-4 mm; stamens 20-25; ovary depressed, globose, with 9-12 segments, not merging into the style but clearly set off from it; style soon deciduous; stigma depressed, globose; fruits small, oval or subglobose, often with a small apical papilla, greenish-yellow when ripe; peel very thin, prominently glandular-dotted; seeds small, oval, white inside.   (See Hodgson's descriptions of limes in this work [excluding the Rangpur and Kusaie varieties, which belong not to this species but probably to C. reticulata, or else are hybrids of it with the lime].)
      The lime is apparently indigenous in the East Indian Archipelago.   From there it has spread by human help to the Asiatic mainland and to many other tropical or subtropical regions of the world.   It is a distinct species, not closely related to any other species of Citrus.   However, limes hybridize freely with other species of Citrus and many such hybrids are found in the East Indies.   The Tahiti lime (see chap. 4, this work), which is widely grown in Florida and California, was found by Bacchi (1940) to be in a triploid cytonomic state; Uphof (1931) had already discovered that it produces no pollen grains or viable ovules.

HYBRIDS OF THE LIME
      Lemonimes [Citrus aurantifolia X C. limon].
      These hybrids, looking much like lemons, are discussed above.
      Limequats [Citrus aurantifolia X Fortunella japonica, or F. margarita].
      These lime-like hybrids are discussed under Fortunella.

      4.   Citrus aurantium L. Sp. Pl. 2:782. 1753. Aurantium acre Mill.Gard. Dict. ed. 8. 1768; Citrus florida Salisb. 1796; C. vulgaris Risso, 1813; C. bigarradia Loiseleur-Deslongchamps, 1818; C. bigaradia Risso & Poiteau, 1818-1822; C. amara Link. 1831; C. karna Raf. 1838; C. communis Le Maout. & Decaisn. 1868; C. aurantium var. bigaradia Hook. f. 1875. Illus. Risso & Poiteau, Hist. Nat. Orang. pl. 30 (col.). 1818-1822; Berg & Schmidt, Darst. Beschr. Off. Gew. 3:pl. 31. 1861; Bentley & Trimen, Med. Pl. 1:pl. 50. 1880; Tschirch & Oesterle, Anat. Atl. Pharm. pls. 69, 70. 1898.
      Type.—Europe ? (Linnaeus ?).   Herb. Linn. in Linn. Soc. London.
      Distribution.—Southeastern Asia; widely cultivated in all subtropical regions.
      Common name.—Sour or Seville orange; bigarade.
      A medium-sized tree up to 10 m high, with a rounded top; twigs angled when young, with single, slender spines, often short, or stout spines up to 5-8 cm long on rapidly growing shoots; leaves medium-sized, ovate, bluntly pointed at tip, broadly rounded to cuneate at base; petioles 2-3 cm long, rather broadly winged, often 1.2-1.8 cm wide at top, but sometimes narrower, 1 cm or less, narrowing rapidly to the wingless base; flowers large, very fragrant with oil of neroli; 5-12 per cent male (staminate only); fruits subglobose, usually slightly depressed at both base and top, peel thick, with a rather rough surface, becoming brilliant orange with a reddish tint at maturity; locules 10-12, filled with sharply acid pulp and numerous seeds; fruit becoming hollow at center as it matures, and then able to float in water.   (See Hodgson's account of the cultivated varieties of the sour orange in chap. 4 of this work).
      This species was introduced into the Mediterranean region from the East and for many centuries was the only orange known to Europeans.   During this long period of culture it became very well known and much appreciated as a medicinal agent; the fruits were used for flavoring and for marmalade, and the flowers for perfumery.   It was the orange of late medieval Europe.   Good high-flavored varieties of the sweet orange, C. sinensis, did not reach Europe from southeastern Asia until the fifteenth century.   From that time on there was more or less confusion over the name of the sour orange.   The pharmacologists persisted in calling it C. aurantium and the citrologists, then the botanists, called it C. bigaradia.   Many botanists considered both the sour and the sweet orange as merely varieties of a single species.   As a matter of fact, the sour orange (C. aurantium) and the sweet orange (C. sinensis) are very distinct botanical species, not merely cultivated varieties of one species.
      The chief morphological differences are as follows: In the sour orange the petioles are much more broadly winged than in the sweet, and the leaf blades are narrower and more acutely pointed at the apex, and less rounded and more cuneate at the base.   Ruggieri has shown (1935) that the petioles of the sour orange are much the longer, averaging 25.89 mm, whereas those of the sweet orange average only 15.91 mm; in other words, the sour orange petioles average 63 per cent longer than those of the sweet orange.   The fruits of the sour orange are of a brighter orange color and have a rougher peel; moreover, in the sour orange the oil glands are situated beneath minute sunken areas in the peel, whereas in the sweet orange the tissue covering the oil glands is often convex.
      Uphof reported (1932, pp. 133-35, fig. 4) that he found no male flowers on several cultivated varieties of the sweet orange (C. sinensis), but that he did find from 5 to 12 per cent of male flowers on the sour seedling oranges of Florida (C. aurantium).   The sour orange, according to Uphof, "probably is very close to the sweet orange but from the standpoint of the production of male flowers constitutes a transition, so to speak, toward the lemons, limes and citrons."   Uphof found no male flowers on cultivated varieties of the grapefruit (C. paradisi).   Many trees of the Dancy tangerine (C. reticulata) showed no male flowers (one small tree overloaded with a very heavy bloom had one single male flower among the thousands examined).   Tangelos (hybrids of C. reticulata with C. paradisi) also showed no male flowers.
      In view of these facts, it is clear that the occurrence in appreciable numbers of male flowers in C. aurantium constitutes an important differential character separating this species from C. sinensis.
      The ethereal oil in the leaves, flowers, and fruits of the sour orange is of very different odor than that in the sweet (more agreeable and aromatic in the sour orange) and has a different composition.   Also the oil recovered from the petals of the sour orange, neroli oil, finds a different use in perfumery and has a higher value than that of sweet orange flowers.   This oil, considered by perfumery experts to be "indispensable to finer perfumery," is said to owe its high value to small amounts (only 0.4 to 1.0 per cent) of "a nitrogenous compound of exceeding fragrance," methyl anthranilate, NH2 · C6H4 · COOCH3 (see Gildemeister and Hoffmann, 1922, pp. 93 and 96; and Finnemore, 1926, pp. 436-41).   This remarkable substance is not found in the oil extracted from the petals of the sweet orange (Theulier, 1902).   The pulp of the sour orange is intensely sour, with a bitterish aftertaste, in contrast to the sweet, agreeable flavor of the sweet orange.   Furthermore, the peel of the sour orange, which is official in the British Pharmacopœia (the peel of both the sour and the sweet orange is official in the U.S. Pharmacopœia), contains three glucosides, according to Tanret (1886, p. 518): (1) 4 to 30 parts per 1,000 of isohesperidin (= naringin ?), having the same percentage composition and the same rotary polarization as hesperidin but a bitter taste (hesperidin of the sweet orange is tasteless) and a very different solubility; (2) 15 to 25 parts per 1,000 of aurantamarin (to which, so Tanret asserts, most of the bitter taste of the sour orange peel is due), differing slightly in composition and in solubility from both naringin (?) and hesperidin; and (3) from a mere trace to six parts per 1,000 of hesperidin.   In the sweet orange only this last glucoside, hesperidin, is found, but it is present in much larger quantities than in the sour orange.24
      The sour orange also shows physiological differences from the sweet orange.   It stands winter cold better and has almost complete immunity to the foot rot, or mal di gomma, so destructive to the sweet orange in some localities.   However, the sour orange is severely attacked by the scab fungus, Elsinoë fawcetti, which does not attack the sweet orange.
      Besides the morphological, chemical, and physiological differences mentioned above, there are further anatomical difference discovered by Ruggieri (1935) between the sour and the sweet orange in the separative layer of articulation that lies between the petiole and the leaf blade, which may be summed up as follows: (1) The pith in the lower articulation joint where the petiole joins the twig is much more flattened from the top to the bottom in the sweet orange than in the sour.   In the sour orange the ratio of the pith to the woody cylinder averages 1:1.4, measured in a horizontal direction, and 1:2.2, in a vertical direction, whereas in the sweet orange the ratio of pith to woody cylinder is 1:1.5 horizontally (about the same as in the sour orange) and 1:4.5 in the vertical (dorsiventral) direction, or only half as thick as the pith of the sour orange petiole.   (2) The cells of the interior layers of the cortical parenchyma of the upper articulation joint where the petiole joins the leaf blade are isodiametric and are 12 to 20.5 mu in diameter in the sour orange but are 20 to 27 mu in diameter in the sweet orange, or nearly one-half larger than in the sour orange.   (3) The pericycle fibers that form a more or less interrupted sheath around the woody cylinder are strongly thickened in the sour orange but are little thickened, if any, in the sweet orange.
      B. Miyazaba, S. Matsubara, and T. Kawaida (1928, p. 189, figs. 4, 5) have also found other anatomical characters that separate the sour orange from the sweet.   Comparing the common sour orange (kaisei-to) of Japan with the Washington navel orange (tento) in leaf structure, they found that the sour orange leaf is thin (197 to 274 mu, the average of the ten measurements being 243.38 mu), whereas the sweet orange leaf is about 11.1 per cent thicker (247 to 334 mu, the average of the ten measurements being 270.33 mu).   The two layers of palisade tissue are about the same thickness in the two species, but the spongy tissue shows a wide variation, being only 148 to 189 mu thick in the sour orange leaf (the ten measurements averaging 167.28 mu) but ranging from 165 to 231 mu in the sweet orange leaf (the ten measurements averaging 197.51 mu), or nearly 17 per cent thicker.   They also found that the number of stomata is somewhat greater in the sweet orange epidermis than in that of the sour orange, averaging 23.44 in the microscope field of vision for the sweet orange and 20 for the sour orange, or 17 per cent more for the sweet orange.
      In view of this array of anatomical, physiological and chemical differences between the sour and the sweet orange it is obvious that they are distinct species even if the gross morphological differences between them are small.

VARIETIES AND MUTATIONS OF THE SOUR ORANGE
      Hodgson has given in this work an account of the forms of the sour orange known in the United States, including the Bittersweet, with sweet fruits; the Paraguay, with subacid fruits; and the Bergamot, grown in Italy for the manufacture of bergamot oil from the fruits.   These three varieties are of unknown origin and opinions with respect to whether they are mutations of the sour orange, like variety myrtifolia, or hybrids.   Experimental hybridization would probably settle the question promptly.
      The orangeries of Europe, which came into vogue early in the fifteenth century and for two centuries were among the most prized possessions of everyone who could afford one, were filled with every form of citrus that could be found, among them many forms of the sour orange.   Tolkowsky (1938, p. 186) observed: "However, it was the sixteenth and part of the seventeenth century…that gave citrus trees, both in theory and practice, their final status as an element of the first importance in the Italian garden."   He also stated that the first orange house or "orangery" in northern Europe, where citrus could not be grown out of doors, was built by order of Charles VIII, king of France, in his own château at Amboise after his famous expedition to Italy in 1494 and 1495.   "The example set at Amboise was soon followed in other royal castles and in those of many members of the nobility."   The most famous orangeries, according to Tolkowsky (pp. 200, 209), were those built for Louis XIV at Fontainebleau and Versailles and finally, in 1674, the one for Madame de Montespan at Clagny, where by 1675, so Madame de Sévigné reported to her daughter, "there is a whole forest of orange-trees in huge boxes."
      In 1818-1822, Risso and Poiteau in their famous folio work, Histoire naturelle des orangers, published descriptions and colored plates in natural size of 23 varieties of sour oranges called bigarades.   Of these, eleven had already been named, described, and figured by Ferrari in his great work the Hesperides, published in 1646.   A glance at these plates suffices to show what a wealth of forms, both normal and teratological, were grown in Europe.   Many of these forms are still to be found in the citrus collections of the Old World, and a few new forms have appeared in recent works on the European varieties of citrus fruit trees (Riccobono, 1899, pp. 141-83; lnzenga, 1915, pp. 19-26; Galli, 1928, pp. 10-13; and others).   A striking new strain of the sour orange, named the Oklawaha, was originated within the United States.   It has large fruits, 7.5 to 10 cm in diameter with a rind 6 to 10 mm thick, that are rich in pectin, making it an excellent marmalade variety (see Swingle, Robinson, and Savage, 1932, pp. 9-10; and Traub and Robinson, 1937, pp. 782-83).
      The Trabut variety of sour orange (named in honor of L. Trabut, who brought together at Maison Carrée, Algeria, early in the twentieth century, a very fine collection of cultivated varieties of citrus) is a European strain bearing normal bitterish-sour fruits but differing from all other published descriptions or figures of sour oranges in having very large, obdeltoid petioles, much broader (2 to 3 cm) than those of typical sour orange leaves.   It was introduced into the United States by David Fairchild in 1925, and distributed as F.P.I. No. 63550.   It, like the Oklawaha variety just noted, is a strain of sour orange very readily recognized by an easily seen morphological character.
      The Daidai is a cultivated variety of the sour orange common in Japan that differs little from the ordinary sour orange except in being dwarfish in habit, and having blunt-tipped leaves and a persistent calyx that continues to grow as the fruit develops, finally turning yellowish or orange-colored as the fruit ripens.   This variety shows almost complete immunity to citrus canker in the Philippine Islands, where ordinary sour orange is very susceptible, as Swingle observed in 1915.   The Daidai is also known in China, where it is grown in Kiangsu and Chekiang provinces for its flower buds, which are used to mix with tea leaves.   It is likewise grown in Canton, especially in the famous Fa t’i flower gardens, as a potted plant.   Hu (1934, p. 47) described it as follows:
      "After a growth of 5 or 6 years the Tai-tai [pronounced 'dai-dai'] tree reaches a height of 1 m; its branches grow sparsely and are of spreading habit; the leaves are elliptical or ovate, apex obtuse, base rounded; leaf blades are very thick, petioles winged; leaves are broad and thick.   Fruit is compressed-globose, measures 5.4 X 6.4 cm; peel is of orange red color.   Fruit pulp is in 10 segments, membranes are thick and white; fruit flesh is light yellow.   Seeds ellipsoid, apex cuneate, monoembryonic; cotyledons white in color; chalaza purple.   Fruit juice is very sour, so the fruits are not fit to eat in their raw state.   They [the Chinese orchardists] gather only the flower buds of this species which they fire-dry to make a scenting material for tea leaves."
      Hu failed to mention that the Chinese Tai-tai has enlarged, persistent calyx lobes, but a description drawn up by Kwok Wa-shau, Swingle's former assistant working at Canton, China, from plants introduced into Kwangtung Province (about 1915) from Soochow, Kiangsu Province, reads: "The calyx is remarkably large, being strikingly different from that of the other species of Citrus, bluish-green in color, the 5 sepals forming a somewhat protruding cup…The rind is extremely thick and the pulp very sour."
      The Chinese do not use the Tai-tai as a rootstock; instead they use the following sorts that grow to a larger size:
      The Vermillion globe sour orange (chu-luan), which grows to a height of 5 meters and has long, narrow, acutely pointed leaves.   Hu (1934, p. 46) described it as having compressed-globose fruits, 8 cm long and 9.5 cm wide, with many (30 to 40) seeds.   It is used as a rootstock for sweet oranges.
      The Leather-head sour orange (p'i-t'ou ch'êng), which has elliptical leaves, blunt at the tip, small depressed-globose fruits, 4.4 cm long and 6 cm wide, with rough peel and with numerous (about 20) seeds.   It is used as a rootstock for sweet oranges at Huangyen, Chekiang Province.
      Tanaka, who studied the citrus fruit trees grown in the coastal regions of China opposite Taiwan, reported (1932a, p. 29) that the sour orange rootstocks used in Fuchow and Huangyen "are very different from our [Japanese] Dai-dai."   Probably the sour orange used as a rootstock at Huangyen which Tanaka mentioned was the form called Leather-head sour orange by Hu.   Tanaka, who is familiar with the sour orange used as a rootstock in the United States, stated further that "our Daidai…is the most inferior grafting stock.   On the other hand, the sour orange in the United States and Italy is an excellent grafting stock, as well as that of Chekiang…"
      Benemerito (1938), like Kwok (1922), found no typical sour orange growing in Kwangtung Province, only the mutation Tai-tai, which has a large and persistent calyx.   He listed four other varieties that he assigned to Citrus aurantium, but three of them, which have very narrowly winged petioles, are probably hybrids, whereas the fourth, which has sometimes broadly winged and sometimes narrowly winged petioles, was considered by Benemerito himself as probably a hybrid between the sour orange and a mandarin.   Some of these hybrids are worth testing as rootstocks.

POSSIBLE ORIENTAL HYBRIDS OF THE SOUR ORANGE
      Besides the four varieties from China and Japan, discussed above, that seem to be mutations of C. aurantium, there are other forms, probably accidental garden hybrids of unknown parentage.   The following are known from Taiwan and Japan:
      Hyonkan.
      This form has large leaves (13 by 6.5 cm), acute or acuminate at the tip, and small, subglobose (6 cm high, 7 cm wide), thick-skinned fruits with a solid core.   The calyx in the flowering stage has slightly enlarged sepals; these, however, do not elongate as the fruit grows.   Hayata (1919, pp. 27, 28, text fig. 16) named it Citrus daidai after a nomen nudum of Siebold printed in 1830, but he stated: "It may be [a] hybrid between C. aurantium and C. sinensis."   It is reported as being cultivated in Taiwan and as being much like a Japanese variety called Kwai-seikan ("very green [sour] orange").   Certainly no evidence is adduced to show that this variety merits a species name.
      Nanshô daidai.
      This is a form found, very rarely, growing semiwild in the forests of Taiwan in three localities, Nanchwang (formerly Nanshô), Gaogan and Taitung districts.   Tanaka (1932a, pp. 20-21) stated that the "Nanshô daidai resembles the Daidai [the Japanese form of] Citrus Aurantium in the shape of the fruits, but the leaves are somewhat similar to the Naruto-mikan."   He went on to say that the fruits are round in shape with a slight nipple at the apex, have an acid taste like the Yama-mikan, and have a thick rind like a citron.
      The Nanshô daidai differs from the ordinary Japanese Daidai, according to Tanaka (1926d, pp. 54-58), "in having much longer leaves [blades about 12 by 5 cm], lanceolate, sharply and gently pointed at the tips, thin, inodorous; petioles broadly winged; calyx glabrous, fruits flattened, globose, peel with smaller oil glands, pulp delicate, very sour and more deeply yellow."   In this article the Nanshô daidai was published in Japanese as Citrus taiwanica Tanaka and Shimada.   Tanaka (1926a, p. 345), a month later, published the Nanshô daidai in English as a new species, Citrus taiwanica Tan., and stated, "at present it seems to have become extremely scarce."   It is by no means clear that this is a good species; it is probably a hybrid of C. aurantium with some other species of Citrus having long leaves.   Tanaka thought that the "very uniform" seedlings of this species might "serve as stocks for commercially semitropic citrus fruits."   This very uniformity of the seedlings may be due to the hybrid nature of the Nanshô daidai, as many Citrus hybrids yield perfectly uniform seedlings from nucellar buds that replace the true embryo in the seeds.   In a later publication Tanaka (1933b, p. 172, fig. 138) included a halftone illustration of the type specimen showing flowering twigs; another text figure (1933b, p. 136, fig. 112) illustrates the depressed-globose, thick-skinned fruit, which shows 10 segments.
      Yama-mikan ("Mountain mandarin").
      Concerning this form, T. and Y. Tanaka (1932, p. 3) stated: "Yamamikan…is used as a rootstock in Hyûga Province [in Japan], and it corresponds to the American sour stock in its vigor and habit."

ACCURATELY SAFEGUARDED HYBRIDS OF CITRUS AURANTIUM
      Apparently the only sour orange hybrid known that was made by properly safeguarded cross-pollination is the following bigeneric one:
      Citradia [Citrus aurantium X Poncirus trifoliata].
      This beautifully intermediate trifoliolate hybrid has great vigor and hardiness (see also under Poncirus).

      4a.   Citrus aurantium var. myrtifolia Ker-Gawl. [John Bellenden] Bot. Reg. 4:text to pl. 346. 1818. Citrus myrtifolia Raf. Sylva Tellur. Mant. 141. 1838; C. sinensis Pers. (non Osb.) Syn. Pl. 2:74. 1807. Illus. Ker-Gawler, loc. cit. pl. 346 (col.); Risso & Poiteau, Hist. Nat. Orang. pl. 50 (col.). 1818-1822.
      Type.—England (Ker-Gawl.).   Herb. ?
      Distribution.—Known only in cultivation: supposed to have been brought from China into the Mediterranean regions in the seventeenth century.
      Common name.—Myrtle-leaf orange.
      Twigs with very short internodes, usually 1-2 cm long, often without spines; leaves small, about 1/2-1/3 the size of the normal foliage of the species; fruit small, often about 1/4-1/2 the size of that of the species.
      This plant is so strikingly different from the common orange that Charles Darwin (1868, p. 100) considered it to be one of the most distinct species of Citrus.   He stated: "…I can adduce another case: the myrtle-leaved orange is ranked by all authors as a variety, but is very distinct in general aspect: in my father's greenhouse, during many years, it rarely yielded any seed, but at last produced one; and a tree thus raised was identical with the parent form."
      The myrtle-leaved variety of the sour orange probably arises as a mutation of the common form of the sour orange.   Swingle once found at Leesburg, Florida, sprouts of this myrtle-leaved variety growing from the roots near the base of the trunk of a very old sour orange rootstock on which a large sweet orange had grown from a graft.   Doubtless this old sour orange was originally, like the many others found in the same grove, dug up from under the live oak trees in a hardwood grove on rich soil (hammock) where sour orange trees grew in a semiwild condition.
      One important horticultural race of sour orange, the chinotto of Italy, belongs to this variety.   It is grown for its small fruits, which are candied.

THE TAXONOMIC PROBLEM OF CITRUS AURANTIUM AND ITS FORMS
      Citrus aurantium is unusual and can almost be said to show mimicry of the sweet orange, C. sinensis.   It has required careful search to find morphological differences to separate the two species; yet they show great differences in their physiological requirements and limitations, in their graft compatibility to other citrus fruit trees, in their resistance to disease, and especially in the chemical composition of their essential oils, glucosides, etc.
      If we admit that the two species are distinct in spite of their mimicry, we then face the even more difficult problem of distinguishing the similar-looking but physiologically diverse forms, whether subspecies, varieties, or strains, found within the species.   Two forms of sour orange so similar as to look almost identical may have profoundly different values as rootstocks for the lemon and perhaps for other cultivated citrus fruit trees.   This problem, already bafflingly complex, is further complicated by the hybrids of C. aurantium, which make an already bad taxonomic situation still worse.

      5.   Citrus sinensis (L.) Osbeck, Reise Ostind. China 250. 1765. Citrus aurantium [var.] sinensis L. Sp. Pl. 2:782. 1753; Aurantium sinensis Mill. Gard. Dict. ed. 8. 1768; Citrus aurantium Lour. (non L.) Fl. Cochinch. 2:466. 1790; C. aurantium Risso (non L.), Ann. Mus. Hist. Nat. Paris 20:181. 1813; C. aurantium [var.] vulgare Risso & Poit. Hist. Nat. Orang. 33. 1818-1822; C. aurantium [var.] dulce Hayne, Arzn. Gew. 11:pl. 28. 1830. Illus. Risso & Poiteau, loc. cit. pls. 3, 4 (col.); Hayne, loc. cit. pl. 28 (col.); Bentley & Trimen, Med. Pl. 1:pl. 51 (col.). 1880; and many others.
      Type.—Europe (Linnaeus), lost ? (no authentic specimen in Linnean Herbarium, fide B. Daydon Jackson [1912, p. 58]).
      Distribution.—China, Indo-China, possibly other southeastern Asiatic regions.
      Common name.—Sweet orange.
      A medium-sized tree with a rounded top and regular branches; twigs angled when young, usually with slender, somewhat flexible, rather blunt spines in the axils of the leaves; leaves medium-sized, pointed at the apex, rounded at the base; petioles narrowly winged, articulated both with the twig at the base and with the leaf blade at the tip; flowers in small racemes or singly in the axils of the leaves, medium-sized; calyx with 5 lobes; petals 5; stamens 20-25; ovary subglobose with 10-13 locules; style slender, clearly delimited, soon falling; fruits subglobose, oval or flattened globose; peel thin, tight, not bitter, central axis (pith) solid; seeds cuneate-ovoid with rough-margined plane surfaces, white inside; embryos usually numerous, varying greatly in size.   (See Hodgson's descriptions of the cultivated varieties of the sweet orange in this work.)
      In 1914, in discussing Chinese species of Citrus, Swingle stated (1914a, pp. 150-51): "The common sweet orange, often confounded with the sour or Seville orange, is in fact quite distinct from the latter.   The fruits of the sweet orange have a solid core, never becoming hollow like that of the sour orange; the petioles are narrowly winged in the sweet and broadly winged in the sour orange; the leaves and flowers of the two species have a very distinct odor.   The two species show decided differences in their soil requirements and in their susceptibility to the attacks of fungous diseases.   Many other minute but constant divergences are shown between these two oranges in all their organs.   These two plants, then, superficially so similar, are in reality very unlike and should by no means be united as varieties of one species."   In the discussion above under C. aurantium is has been shown that many anatomical differences between the sweet and the sour orange have recently come to light, differences that go far to show that these are in reality very distinct species that happen to possess certain superficial resemblances concealing many profound differences.
      Citrus sinensis is probably native to southeastern Asia, in northeastern India or, more likely, in southern China and Indo-China.   It has been so widely cultivated for so long that it is no longer known in a truly wild condition.
      Tolkowsky (1938) theorized that the sweet orange first appeared in Roman gardens as early as the first century A.D., eventually disappearing in the barbaric invasions.   The sweet orange mentioned in literature of the fifteenth century appears to have reached Europe over the Genoese trade route, followed by superior varieties brought around the Cape of Good Hope by Portuguese navigators in the sixteenth century.
      Before the advent of the sweet orange into the Mediterranean regions, citrus fruits were esteemed there for medicinal use, but the introduction of this delicious and beautiful fruit aroused an intense interest in Citrus and led many people of means to build orangeries wherein they grew all sorts of citrus fruits.   Everyone in Europe heard of these wonderful fruits.   By the middle of the seventeenth century (1646) Ferrari had written his Hesperides, a beautifully illustrated account of the citrus fruits then known.   Around this was woven the story of Hercules, taken from Greek mythology, who was supposed to have found oranges when he gathered the golden apples of the Hesperides.
      The fruits of the sweet orange contain a glucoside, hesperidin, which was discovered by Lebreton in 1828 but was first recognized as a glucoside by A. Hilger (1876) and E. Hoffmann (1876).   It is allied to, but different from, the naringin of the pummelo and the grapefruit, the aurantamarin of the sour orange, and the ponciridin of trifoliate orange (see table 3-4).   Much interest was aroused by the discovery of "citrin," or vitamin P, related to hesperidin, in the juice and peel of the lemon (and sweet orange) by Szent-Györgyi and his colleagues (see Rusznyák and Szent-Györgyi, 1936; Armentano et al., and Szent-Györgyi, 1936; and Rosenberg, 1942).   Further information on the composition and biochemistry of the sweet orange may be found in Sinclair (1961) and the U.S. Department of Agriculture's Handbook No. 98 (1962).

HYBRIDS OF CITRUS SINENSIS
      Tangors [Citrus sinensis X C. reticulata].
      These hybrids, which look much like the sweet orange, bear orange-like fruits often of high quality.   They are discussed under C. reticulata.
      Citranges [Citrus sinensis X Poncirus trifoliata].
      See under Poncirus.
      Citrangors [Citrus sinensis X (C. sinensis X Poncirus trifoliata)].
      See under Poncirus.
      Citrangequats [(Citrus sinensis X Poncirus trifoliata) X Fortunella japonica, or F. margarita].
      See under Fortunella.
      Citrangedins [(Citrus sinensis X Poncirus trifoliata) X (Citrus reticulata var. austera ? X Fortunella sp.)].
      See under Fortunella.
      Citrangeremos [(Citrus sinensis X Poncirus trifoliata) X Eremocitrus glauca].
      See under Eremocitrus.

      6.   Citrus reticulata Blanco, Fl. Filip. 610. 1837. Citrus nobilis Andrews (non Lour.), Bot. Repos. 9:pl. 608. 1809; C. nobilis var. major Kerr, Bot. Reg. 3:pl. 211. 1817; C. deliciosa Ten. Ind. Sem. Hort. Neap. [9]. 1840; C. nobilis var. genuina Tan. Bot. Mag. Tokyo 26:204. 1912. Illus. Andrews, loc. cit. pl. 608; Kerr, loc. cit. pl. 211; Du. Breuil, in Risso & Poiteau, Hist. Nat. Orang. ed. 2. 49, pl. 29 bis. 1871-1872.
      Type.—Wanting.   Substitute type: Philippines, Luzon (Merrill, Species Blancoanae, No. 402).
      Distribution.—Philippines, southeastern Asia; widely cultivated in all subtropical regions.
      Common name.—Mandarin orange.
      A small spiny tree with slender twigs; leaves broadly or narrowly lanceolate; flowers arising singly or in small clusters in the axils of the leaves; fruits depressed globose or subglobose, with thin, loose peel easily separating from the segments, bright orange or scarlet-orange when fully ripe; seeds small, pointed at one end, embryo green.   (See Hodgson's description of the cultivated varieties of the mandarin orange in this work.)
      In discussing the Chinese species of Citrus in Plantae Wilsonianae (Swingle, 1914a, pp. 141-49), Swingle showed that the true mandarin orange with its thin, loose peel is very different from Citrus nobilis, which Loureiro described as having a "thick, succulent, sweet, edible and irregularly tuberculate rind."   As Swingle wrote at that time, "this description is impossible to reconcile with the ordinary mandarin oranges."   It does describe the Cochin-Chinese orange that is called King (or King of Siam) in the United States.   This latter shows many other important differences from the typical mandarin orange—differences so great as to force the authors to the conclusion that it is a hybrid between a mandarin orange and some other citrus fruit, possibly a sweet orange or a pummelo, or even a pummelo X sweet orange hybrid.   H. Brenier (1917, p. 214), in commenting on the confusion in the names of Indo-Chinese citrus fruits, stated (in translation): "It may well be that hybridization has occurred.   It is indeed very probable that the hybrid King of Siam (C. Aurantium X nobilis [i.e., C. sinensis X C. reticulata]), an orange with high-colored pulp mentioned in the Flore générale de l'Indo-Chine, is not the only hybrid that has originated in our peninsula."
      In 1931, Matlack reported that the chromoplasts in the peel of the King orange are very different from those of the mandarin orange and similar to those of the sweet orange.25   Another discovery, made by Nelson (1934), is of a fully methylated flavonol in the peel of the tangerine orange.   Nelson stated: "It seems to be the first instance of a fully methylated flavonol occurring in nature."   It is doubtless analogous to hesperitin (derived from hesperidin of the sweet orange), naringenin (derived from naringin of the grapefruit and the pummelo), and ponciritin (found in flowers of Poncirus trifoliata by Hattori, 1936).
      It thus became necessary to find another botanical name for the mandarin orange.   The oldest valid published name for the mandarin type of orange is Citrus reticulata, so named for the irregular network of soft, white fibers found between the loose peel and the juice-filled segments.   This name was published by Blanco in his Flora de Filipinas in 1837.   Because his type specimens have disappeared, it was not entirely clear what plant Blanco had in mind until Merrill (1918) published his study of Blanco's plants and issued sets of herbarium specimens including substitute types for Blanco's new species.   Merrill's studies and substitute types show clearly that Blanco's C. reticulata is, in fact, a mandarin orange.   The Blanco species name antedates by three years C. deliciosa Tenore, which was based on the so-called Willowleaf mandarin.   The Philippine C. reticulata is fortunately not an extreme and highly specialized form of this species, and hence is excellently fitted to be the type of this widespread and variable species of Citrus.   Probably a dozen or more so-called species have been published for forms of mandarin oranges and their hybrids and numerous varieties.
      6a.   Citrus reticulata var. austera Swing. Jour. Wash. Acad. Sci. 32:25. 1942.
      Type.—China, Kwangtung Province, Ch'ao-chou (Groff, No. 233, 1918).   Herb. Lingnan Univ., Canton, China.
      Distribution.—China: Swatow region, Kwangtung Province.
      Common name.—Sour mandarin.
      Differs from the sweet mandarin orange in having smaller fruits with intensely acid pulp.   The type of this variety is the sour mandarin called sün kat in Cantonese, propagated from seed in the Swatow region of Kwangtung, where it is commonly used as a rootstock for grafting.   G. W. Groff in 1918, in a manuscript report covering his work on citrus done in China under Swingle's supervision for the former Bureau of Plant Industry of the U.S. Department of Agriculture, described the sün kat as follows:
      "Fruits slightly depressed-globose, 2.9-3.3 cm long, 3.3-3.6 cm diam., with smooth, loose peel about 4 mm thick, capucine yellow (Ridgway, pl. 3) when ripe; oil glands small, round, far apart, fragrant; segments 9, easily separated; segment walls thin, tender, white; core 6-8 mm diam., soft; pulp deep chrome yellow (Ridgway, pl. 3), composed of small, short, pulp-vesicles, clinging together but irregularly arranged and easily broken; juice reddish yellow, very sour; seeds about 9, rounded at one end, pointed at the other, showing white parallel lines from base to tip; leaves lanceolate-elliptical, blades 6.8 X 2.5 cm, rather acutely cuneate at the base and narrowed to a blunt apex, with about 10 pairs of lateral veins; petioles nearly wingless."
      This variety is widely grown about Swatow, China, where it is used as a rootstock upon which to graft the Ponkan or mi-tong-kan ("honey pot orange") and other famous varieties widely exported from Swatow.
      Probably some of the other sour mandarins called kat by the Cantonese are forms of this variety.   Some of the so-called kat varieties with large fruits, which as they ripen may become sweet enough to eat, are probably hybrids between variety austera and the sweet mandarin, C. reticulata, or the sweet orange, C. sinensis.
      The so-called Rangpur lime (see chap. 4 this work) may upon further study prove to be a form of C. reticulata var. austera or a dilute back-crossed hybrid of it.   It is probable that the sour mandarin has hybridized with several other species of Citrus in China; hybridized with C. ichangensis, it doubtless gave rise to the widely cultivated Yuzu, a sort of cold-resistant lemon-substitute widely cultivated in Japan and northern China; hybridized with a kumquat (some species of Fortunella), it gave rise to the well-known Calamondin, widely cultivated in China and the Philippines and well known in citrus variety collections in the United States.

HYBRIDS OF CITRUS RETICULATA
      The following types of hybrids between mandarin oranges and other citrus fruit trees are known:
      Tangelos [Citrus reticulata X C. paradisi].
      These hybrids, which look very much like oranges and are often beautifully colored and of delicious flavor, were first made by Swingle, using safeguarded cross-pollination technique.   They were described and illustrated by Webber and Swingle (1905, pp. 235-37, pls. 17-19); by Webber (1907, pp. 336-37, pl. 21); by Swingle 1913e, pp. 87-89, pls. on pp. 94, 95); and by Hume (1926, pp. 106-08).   Additional tangelos that had proved worthy of being named as horticultural varieties were described and illustrated by Swingle, Robinson, and Savage (1931, pp. 1-13, pls. 1-7); by Traub and Robinson (1937, pp. 766-67, figs. 6, 7); and by Hodgson in this work (chap. 4 text, and fig. 4-64).
      Similar hybrids have originated in China, Indo-China, Japan, and other Far Eastern countries by insect cross-pollination of mandarin oranges and pummelos growing in dooryard groves of mixed varieties of citrus fruit trees.   The Natsumikan or Natsudaidai of Japan is almost certainly such an accidental hybrid; it cannot be named as a new species of Citrus unless similar new species names are coined for the many strikingly diverse tangelo hybrids that have been made in the United States and with accurate knowledge of their parentage.   The name, Citrus natsudaidai Hayata, given by Hayata (1911-21, vol. 8, p. 29, fig. 19), cannot be accepted as a proper one for a chance hybrid known only in cultivation.
      Tangors [Citrus reticulata X C. sinensis].
      These hybrids have been made by safeguarded cross-pollination; among them is a promising tangelo-like fruit named the Umatilla (see Swingle, Robinson, and Savage, 1931, pp. 11-12, pl. 8).   The King orange of Indo-China, long grown in the United States, is doubtless a tangor (see Brenier, 1917, p. 214; quoted above).   The Temple orange of Florida is probably another tangor (see Traub and Robinson, 1937, p. 775).   There are without doubt many tangors besides the King that are masquerading as oranges in China, Japan, and other eastern Asiatic countries, where they have originated as chance hybrids.   As there are many more varieties of mandarin oranges commonly grown in the Far East than in the United States, it is to be expected that even more variation will be found in Oriental tangors.
      The King tangor was made the type of a new species which Loureiro (1790, vol. 1, p. 266) published as Citrus nobilis.   As Loureiro's herbarium specimens have almost all been lost, this species was misunderstood by taxonomists, who considered it to be a true mandarin orange, i.e., what is now called C. reticulata.   It is therefore necessary to abandon the name C. nobilis (see discussion above).
      Citrandarins [Citrus reticulata X Poncirus trifoliata].
      These trifoliolate hybrids are discussed under Poncirus.
      Calamondin [Citrus reticulata var. austera ? X Fortunella sp.].
      See under Fortunella.
      Yuzu [Citrus reticulata var. austera ? X C. ichangensis].
      See under Citrus ichangensis.
      Faustrimedin [(Citrus reticulata 'Calamondin' X Fortunella sp.) X Microcitrus australasica].
      See under Fortunella.

      7.   Citrus grandis (L.) Osbeck, Dagbok Ofwer Ostin. Resa 98. 1757. Citrusaurantium [var.] grandis L. Sp. Pl. 2:783. 1753; C. aurantium [var.] decumana L. Sp. Pl. ed. 2. 1101. 1763; C. decumana L. 1767; Aurantium decumana Mill. 1768; Citrus pamplemos Risso, 1826; C. maxima (Burm.) Merr. 1917. Illus. Loisleur-Deslongchamps, Nouv. Duhamel 7:38, 42. 1818; Risso & Poiteau, Hist. Nat. Orang. pls. 62, 63. 1818-1822; Poiteau, Pomol. Franc. 2:pl. 344. 1846; Nooten, Fleurs, Fruits, Feuill. Java 1:pl. 3. 1863; Tanaka, Kankitsu No Kenkyû (Citrus Studies), pl. 3 (col.). 1933; Tillson & Bamford, Amer. Jour. Bot. 25:786, figs. 37-38. 1938; and many other plates (see Stapf, Index Londinensis 2:223 [1930]).
      Type.—China, Canton (Osbeck), in Herb. Mus. Stockholm (fide Tanaka).
      Distribution.—Southeastern Asia, East Indian Archipelago; widely cultivated in all subtropical regions.
      Common name.—Pummelo.
      A large, spiny, round-topped tree with angular twigs, often pubescent; leaves large or very large, oval or elliptic-oval, with a blunt point at the tip and a broadly rounded base, often subcordate and even slightly overlapping the winged petiole; midrib and large veins often pubescent; petioles broadly winged, and more or less cordate, usually pubescent; flowers very large, borne singly or in axillary clusters or in subterminal inflorescences; sepals and petals 5; stamens 20-25, with large linear anthers; ovary globose, sharply delimited from the deciduous style, with many segments; fruit large or very large, subglobose, oblate-spheroid or subpyriform; seeds large, thick, wrinkled.   (See Hodgson's descriptions on horticultural varieties of the pummelo in this work.)
      Citrus grandis is one of the most distinct and most easily recognized species of Citrus.   It is separated from the other species of the genus by a number of easily seen characters.
      Its thick, often pubescent, angular young twigs, with huge leaves borne on broadly winged, more or less heart-shaped petioles, its very large flowers and giant pale yellow fruits often the size of a child's head, make the species impossible to mistake.   The fruit usually has a thick peel and the pulp-vesicles are much larger than those of other species of Citrus.   Instead of cohering with one another, they easily fall apart.   The membranes that enclose the segments, although thin, are so strong that they can be peeled off the enclosed mass of pulp-vesicles easily.   If this is done carefully, the segment remains intact in spite of having lost its covering membrane.
      Besides the morphological differences distinguishing it from the other species of Citrus, the pummelo was discovered to have an important chemical difference (later found to be shared by the grapefruit): it contains naringin, a bitter glucoside related to the nearly tasteless hesperidin of the sweet orange and to the bitter aurantamarin of the sour orange, but differing from both.   (Full information on naringin is given by Will, 1885 and 1887; see also Zoller, 1918, Poore, 1934; and table 3-4.)   Certain Oriental pummelos have a slightly higher content of vitamin C than oranges or grapefruit.   Pummelos are highly esteemed in both China and Thailand.
      8.   Citrus paradisi Macf. Hook. Bot. Misc. 1:304. 1830; also Fl. Jamaica 1:131. 1837. Citrus decumana var. racemosa Roem. Syn. Hesper. 1:67. 1846; C. decumana var. patoniana Riccob. Boll. Ort. Bot. Palermo 7:211. 1908; C. maxima var. uvacarpa, Merr. & Lee, Amer. Jour. Bot. 11:383. 1924. Illus. Swingle, in Bailey, Stand. Cycl. Hort. 3:1392, pl. 50 and fig. 1744. 1915; Tanaka, Kankitsu No Kenkyû (Citrus Studies), pl. 4 (col.). 1933; and many others.
      Type.—Wanting (fide Tanaka, 1932e, p. 432).
      Distribution.—West Indies; now cultivated in all subtropical regions.
      Common name.—Grapefruit.
      A large, round-topped tree with dense foliage; twigs angular when young, glabrous or nearly so; leaves larger than those of the sweet orange, smaller than those of the pummelo, ovate, bluntly tipped and broadly rounded at the base, glabrous or nearly so; petioles rather broadly winged but not so broad as those of the pummelo, oblanceolate to obovate in shape, the somewhat broadly rounded tip touching the very broadly rounded base of the leaf blade; flowers large, borne singly or in small clusters in the axils of the leaves; calyx 5-lobed; petals smaller than those of the pummelo, often larger than those of the sweet orange; fruits larger than those of the sweet orange but smaller than those of most pummelos; seeds smaller than those of the pummelo, white, not yellowish, not ridged as are those of the pummelo.   (See Hodgson's description of grapefruit varieties in this work.)
      The grapefruit apparently originated in the West Indies.   In spite of careful search, it has not been found native in the Old World, where the parent species is largely grown and where many horticultural varieties are known.
      Another alternative is that the grapefruit may be a hybrid of the pummelo with the sweet orange.   The morphological characters of the twigs, leaves, flowers, fruits, and seeds would suggest this.   The most important argument against such an origin of the grapefruit is the absence of any breakup of self-pollinated grapefruit seedlings into orange-like and pummelo-like forms.   The fairly high fertility of the grapefruit when pollinated by Citrus reticulata 'Dancy' makes the absence of reversions in the self-pollinated seedlings of the grapefruit hard to explain.   On the other hand, the hybrids of C. reticulata 'Dancy' with grapefruit show great variability; some of them are grapefruit-like and others somewhat orange-like.   In view of this fact it seems best to retain the grapefruit, for the present at least, as an independent but satellite species immediately following C. grandis, with which it is so closely allied that many Citrus taxonomists consider it a variety of that species.   Patrick Browne (1756, 1789) was perhaps the first to discuss the grapefruit, under the name "Forbidden Fruit" or "Smaller Shaddock."   Lunan (1814, pp. 171-73) was the first to use the term grapefruit.   He wrote: "There is a variety known by the name of Grapefruit, on account of its resemblance in flavor to the grape."   Tanaka (1926b) gave the taxonomic history of the grapefruit (in Japanese) and in the English summary a good technical description of its botanical characters.
      The fruit of the grapefruit contains a bitter glucoside, naringin,26 first discovered in the pummelo.   According to Zoller (1918, p. 371), naringin is found in large quantities in the peel of the grapefruit (from 0.66 to 0.80 grams per 100 grams of fruit).   It occurs without any hesperidin but seems able to give rise to a form of vitamin P, according to Szent-Györgyi (1938).   It is different from the glycosides of hesperidin, found in the sweet orange, and aurantamarin, in the sour orange.
      It must be admitted that the true nature of the grapefruit is still unknown.   It is to be hoped that the mystery of its origin can be settled by some of the newer methods now used in taxonomic research.

HYBRIDS OF CITRUS PARADISI
      Tangelos [Citrus paradisi X C. reticulata].
      These orange-like fruits, some of them beautiful in color and of delicious flavor, are discussed under Citrus reticulata.

      9.   Citrus indica Tan. Studia Citrol. 2:164. 1928. Illus. Tanaka, Kankitsu No Kenkyû (Citrus Studies), 94, fig. 80. 1933.
      Type.—British India, Khasia, Churra, altitude 2,000-4,000 feet (Hooker f. and Thomson, 15/8/50).   Herb. Kew.
      Distribution.—Eastern Himalayan region: "a really wild Citrus found in Nowgong District, Khasi Hills and Manipur in Assam" (Tanaka, 1937, p. 235).
      Common name.—Indian wild orange.
      The original description, translated into English, reads as follows: "Branches terete, spiny, glabrous; leaves oblong or lanceolate, thick, sub-entire, attenuate at the apex, acute at the base, veins curved, minute, indistinctly reticulate; petioles articulated, linear, sulcate; flowers not seen; fruits small, broadly obovoid or sub-pyriform, solitary on terminal twigs, pedicels very short; calyx 'hypocrateriform,' lobes triangular, acute; cortex thin, segments few (?), pulp vesicular, vesicles fusiform; seeds large, suborbicular, smooth, mono-embryonic."   In his discussion Tanaka stated that this species "has leaves resembling those of C. sinensis, but [the] small, fig-shaped fruit containing extremely large seeds is entirely different from any [other] Citrus."
      In the Japanese account of C. indica, Tanaka gave measurements of various organs and the details not treated in his Latin diagnosis.   His detailed description in Japanese (1928a, pp. 160-62), as translated by S. Katsura, reads: "Twigs round in cross-section, rather stout, somewhat zigzag ('wild-goose-procession form'), surface light green, glabrous, not wrinkled; thorns strong, arising at a broad angle, sharp-pointed; leaves oblong or lanceolate, pointed at both ends, apex somewhat caudate, the extreme point sharp or slightly emarginate, the upper side green with indistinct veins, the underside light-colored with fine curved veins, without convex oil glands, margin subentire; petioles short, with narrow wings articulated [with the leaf blade]; leaves near the top of the branches measure 12.2 X 4.1 cm, petioles 0.8 X 0.2 cm; flowers indistinct [only calyx seen]; fruits terminal, borne singly on the ends of branchlets, pedicels extremely short and strong, articulated with the calyx; calyx funnel-shaped, sepals triangular, acute, thick, glabrous; young fruits depressed at apex, slightly attenuate at the base, short, pyriform, about 2 cm diam., peel very thin [red, fide Tanaka, 1937], segments few; pulp-vesicles spindle-shaped, stalked, very delicate; pulp very slimy [fide Tanaka, 1937]; seeds very few, very large, almost round [in outline], flattened, monoembryonic, 11 X 9 X 3.5 mm."
      In discussing specimens collected at Khasia by Hooker and Thomson in August, 1850, and preserved in Kew Herb. (Tan. Ident. No. 1851), Herb. Bot. Mus., Berlin-Dahlem (Tan. Ident. No. D-1282), and Rijks Herb., Leiden (Tan. Ident. No. R-1016), Tanaka stated that they "have branches with numerous spines and luxuriant growth of leaves.   The calyx persists after the flower has fallen.   Flowers borne single in the axils of the leaves, with extremely short pedicels.   Calyx tubular (fistulose ?) articulated with the pedicels.   Ovaries vase-shaped and scarred where the style has fallen off, resembling ovaries of Lavanga [=Luvunga].   Disk noticeable, below the ovary."   In his remarks in Japanese, Tanaka stated: "The most outstanding point [of this new species] is the morphological aspect of the flower parts.   The Leiden specimen has sepals pointing [out] like claws with oil cells on the upper side resembling those of Chalcas [=Murraya], the flowers are evidently borne singly.
      "It resembles Metacitrus [Citrus junos, C. ichangensis, and the loose-skinned oranges, C. reticulata and its cultivated varieties and hybrids] in its short pedicels, the thin skin of the fruits and the large seeds like C. yuzu and C. ichangensis but in its depressed shape it resembles Pleiospermium [one of the Primitive Citrus Fruit Trees].
      "In the shape of the articulation of the calyx and pedicels, it resembles Citron and also in the round branches in cross section, but its leaves are very different because they have no dentations at all.   The leaves show analogies to those of the sweet orange ? ('amodaidai') group in shape, but are longer and have far more indistinct fine, parallel and curved veins.   The form of the large leaves borne scattered on strong shoots resembles that of leaves of the genus Paramignya, in short it is still undetermined to which group this species belongs."
      In a later publication, however, Tanaka (1937) stated: "It is really a wild Citrus found in Nowgong District, Khasi Hills and Manipur in Assam.   It belongs correctly to Metacitrus, as was predicted by the author, having small sized fruit with thin reddish rind and very slimy pulp, much similar to that of Citrus depressa Hayata.   Perhaps of value as a rootstock."
      This remarkable species has several unusual characters which indicate that it may be a truly indigenous wild species.   There is always the chance, however, that C. indica is in fact a hybrid of the remarkable, truly wild Citrus species of northeastern India, C. latipes, which probably has very large seeds like its close relative, C. ichangensis, of central and southwestern China.   If so, hybrids between Indian C. latipes with some species of the subgenus Citrus could be expected to have very large seeds like the Ichang lemon, which very probably is a hybrid of C. ichangensis and C. grandis.
      10.   Citrus tachibana (Mak.) Tan. Bul. Sci. Fak. Terk. Kjushu Univ. 2:52. 1926. Citrus aurantium var. tachibana Mak. Jour. Jap. Hort. Soc. No. 75:2, pl. 1896; C. nobilis var. spontanea Tokutaro Ito, Jour. Coll. Sci. Univ. Tokyo 12:361. 1900; C. aurantium subsp. nobilis var. tachibana Mak. Bot. Mag. Tokyo 15:167. 1901. Illus. Makino, loc. cit. 1896; fig. 3-46 this work.
      Type.—Lost.   Paratype: Japan, Tosa (T. Makino).   Herb. Hokkaido Imp. Univ. (Tanaka ldent. label Sa-2).
      Distribution.—Southern Japan: Yamaguchi and Kochi prefectures, south to Kagoshima Prefecture; Ryukyu Islands; Taiwan.
      Common name.—Tachibana orange.
      A description of this species by T. Makino, as translated by S. Katsura, reads: "Tree stands over 10 feet.   Branches and leaves grow thickly.   Strongly resistant to frost or snow.   Fruit somewhat flattened, 2-3 cm lateral diam.   Skin smooth, oil glands scattered beneath the skin.   Segment cases 6-7; juice bitter and almost inedible.   Seeds 1-2 in a segment, and rather large in size.   Fruits at first green but turning yellow in the late autumn.   The flavor of the skin resembles that of Yuzu.   Flowers the same as other Citrus plants in time of blooming, shape and color.   In the rainy season of summer the petals begin to turn white.   Needle-like thorns are found on the branches.   Petioles not winged."
      A description of the paratype in the herbarium of Hokkaido Imperial University made by Tanaka reads as follows: "Branches small and slender; thorns about 3 mm long, acute.   Leaves long, ovate-elliptical, subcoriaceous, broadly acuminate, obtuse and incised at the tip, somewhat broad and convex at the base, indistinctly dentate at the margin, midrib slender, straight and distinct beneath, veins almost indistinct, oil glands indistinct; petiole short, small, with linear wings which seem to be on the verge of degeneration.   Flowers axillary, solitary, small.   Pedicels 2 mm long, slender, glabrous; scales at the base triangular, ciliate at the margin.   Calyx 3 mm in diameter; sepals somewhat recurved outward, densely ciliate at the margin.   Disk large, ring-form, subcarnose, depressed at the apex.   Ovary almost globular, attenuate at the base, about 2 X 2 mm in size."
      Tanaka (1922, pp. 246-48) published in English a very interesting account of the tachibana orange, about which cluster many of the oldest Japanese traditions, showing that the Japanese people have known this fruit since the very beginning of their history.   As this species is not edible, it would doubtless not have attracted so much reverent attention had it not been the only citrus fruit known to the early Japanese.
      The widespread occurrence of the tachibana, from southern Taiwan to the southwestern province of the main island of Japan, makes it very probable that it is in fact a wild species that has persisted since prehistoric times.   Tanaka (1931a) published in Japanese a detailed account of the occurrence of C. tachibana in a wild condition in southern Japan, on the Ryukyu Islands, and also of its discovery in southern Taiwan where it grows at much higher altitudes than in Japan but where the temperatures are much the same as at sea level along the northern limits of its range.
      This species is very similar in many of its characters to the mandarin orange, C. reticulata.   The great antiquity of the tachibana in Japan seems to preclude its being a hybrid or "chance seedling" of recent origin.   It probably should be considered as a satellite species of C. reticulata and somewhat closely related to it.

POSSIBLE HYBRIDS OF CITRUS TACHIBANA
      Shekwasha (also spelled Sheequasha, Shiikuwasha, Sekwasa, or even Seequassia).
      This plant was described by Hayata (1919, p. 16) as follows (in translation): "Twigs very green, applanate-triangular, obtusely bent at the node, sometimes with spines 1-1.5 cm long in the axils of the leaves.   Leaves ovate-oblong, 8-9 cm long, 3.5-4 cm wide, slightly acuminate or broadly obtuse and retuse at the apex, broadly and obtusely triangular at the base, margins subentire, slightly crenulate; petioles 8 mm long, very narrowly winged.   Fruits terminal, with very short peduncles, depressed-globose, 23 mm long, 4.5 cm wide, rounded in outline or sometimes slightly lobed, deeply sunken at the apex and slightly sunken at the base, with 7-9 locules, pericarp very thin, 1 mm thick, luteo-flavescent, rather smooth, flesh acid.   Seeds 1 cm long, 6 mm wide, apex acute, rostrate tip obtuse; embryo pale greenish."
      Tanaka (1927c, p. 31) added a few points in his short description: "Bushy tree about 4 m high.   Leaves small, broad-elliptic, obtuse, petiole short, almost wingless.   Fruit small, oblate, ends concave, very smooth, oil cells fine, reddish orange.   Rind thin, turgid.   Segments many.   Center hollow.   Pulp deep-colored, sweetish and rich in pectin, vesicles finely netted.   Seeds very plump, finely striated, teguman [sic] at chalaza rosy, polyembryo green."   He figured the Shekwasha (1933b and 1926c, pp. 190-93, figs. 1, 2) in comparison with Citrus tachibana, from which it differs strikingly in having larger oil glands distinctly sunken in the peel, also fewer locules that separate from each other and from the peel in the ripe fruit.   Also Tanaka (1927b, p. 199) contrasted the Shekwasha with the Koji orange of Japan (which he called "Citrus leiocarpa Hort. nov."), from which it differs in never occurring north of Sanbok (in the Ryukyu Islands), and "by having pointed leaves, reddish rind, finely round-netted pectiniferous pulp-vesicles, large seeds of brownish coat and dark brown chalaza part of tegumen."
      The Shekwasha is found, according to Tanaka (1927c, p. 31), in a wild or semiwild state in the Ryukyu (Luchu) Archipelago south of Sanbok and in Taiwan.   Swingle brought back to the United States in 1927 a plant of the Shekwasha given him by the Tanaka Citrus Experiment Station at Tanushimaru, Kyushu, Japan, to which was assigned the Foreign Seed Introduction number 71180 and C.P.B. number 936.   In this country the fruits of this variety when fully ripe are bright orange-colored and the less mature fruits are orange-buff or capucine yellow; the pulp is near primuline yellow (Ridgway, 1912, pl. 16).   The seeds are pointed-obovoid and somewhat compressed, 11 to 13 by 7 to 8 by 5.5 to 6 mm, with a very finely striate but nevertheless remarkably smooth and even testa for a Citrus; the chalazal cap is reddish-brown (Ridgeway, pl. 27); the embryo is pale greenish.   Tanaka reported (1926c, p. 192) that the pulp of the Shekwasha contains much pectin.
      The Shekwasha grows so vigorously in the peculiar sandy-loam and porous-limestone soils of extreme southeastern Florida that it should be tested there as a rootstock.   The fruits are poor-flavored.
      Hayata (1919, p. 16) named the Shekwasha as a valid species of Citrus (C. depressa); in the same paper, however, he named as "new species" of Citrus several other forms that, according to his own statement, are probably hybrids.   The Shekwasha has just the characters that would result from the hybridizing of the native C. tachibana of Taiwan and the Ryukyu Islands with some form of the mandarin orange (C. reticulata) introduced from the Chinese mainland.

Subgenus Papeda
      Common name.—Papedas.
      Pulp-vesicles with very numerous droplets of acrid oil; petioles always large and broadly winged; flowers small, usually less than 2 cm diam.; stamens free, or, if cohering in bundles, then flowers larger (1.5-3 cm diam.) and petioles very long, 1 3/4-3 times longer than broad.
      This subgenus comprises a number of truly wild species of Citrus, many of them still to be found growing in primeval forests of the Monsoon region.   There are two sections in the subgenus Papeda, the section Papeda, comprising the typical species, and the section Papedocitrus, which is intermediate in character between the subgenus Citrus and the section Papeda, having flowers much like those of the subgenus Citrus and leaves like those of the section Papeda.   The fruits are almost never eaten, but the acrid juice of the fruits of the species belonging to the section Papeda is widely used by primitive peoples, especially Malays, Melanesians, and Polynesians, as a hair wash.
      As noted above in the discussion of the subgenus Citrus, the species of Citrus belonging to the subgenus Papeda have been shown by Tillson (1938, pp. 21, 30; also Tillson and Bamford, 1938, pp. 788, 790) to possess a decidedly simpler type of vascular anatomy of the flowers than do the species of the subgenus Citrus.   In this important character the subgenus Papeda agrees with the other five genera (other than Citrus) included in the True Citrus Fruit Tree group.   This fact will doubtless prove important in studying the course of evolution of the commonly cultivated species of Citrus, all of which belong to the subgenus Citrus.   This makes it very desirable to learn much more than we now know regarding the species of papedas in their relationship to the subgenus Citrus, on the one hand, and to the genera Clymenia, Microcitrus, Eremocitrus, Poncirus, and Fortunella, on the other.   Tillson's discovery of a primitive character in the papedas serves to emphasize how very different they are from the true oranges of the subgenus Citrus.
      The species which belong to the section Papeda have small flowers, 1.2 to 1.7 (rarely 2) cm in diameter, with free stamens not cohering in groups.   The pulp-vesicles of all the species contain numerous globules of very acrid oil and are sometimes attached to the radial locule walls for one-half to three-fourths the distance from the dorsal wall to the center of the fruit.
      To facilitate an understanding of the subgenus Papeda, the detailed key was prepared to give a perspective view, as it were, of the taxonomic characters of the species that could not easily be obtained from independent descriptions, no matter how detailed they might be.

Subgenus Papeda Section Papedocitrus
      Flowers large (1.5-3 cm broad when open); stamens at first coherent nearly to the tips, but separating later into a few bundles, connate only at the base; winged petioles long-elliptical with more or less parallel sides; seeds crowded, often very short, very thick, and angular.
      The section Papedocitrus comprises two species growing in southwestern China, northeastern India, and northern Burma, far to the north of the range of the species that constitute the section Papeda.   Section Papedocitrus is intermediate in its flower characters between the section Papeda and the subgenus Citrus, but the pulp-vesicles contain numerous droplets of acrid oil and the winged petioles often equal the area of the leaf blade, just as in the section Papeda.
      The fruits of the species of this section do not seem to be used by primitive peoples for washing the hair, but it is believed by some that in China the fruits of C. ichangensis were anciently, and perhaps still are, used in medicines.   Striking hybrids have arisen in China between this species and species of the subgenus Citrus.   One of these hybrids, the Ichang lemon, is cultivated on a small scale by the Chinese.   The Yuzu, which shows many points of similarity with C. ichangensis, is doubtless another of these hybrids; it has been known in China since ancient times and is widely grown both in northern China and in Japan.
      11.   Citrus ichangensis Swing. Jour. Agr. Res. 1:4 (excluding fruit). 1913. Illus. Swingle, loc. cit. pl. 1, figs. 1, 4-7 (excluding figs. 2, 3 of fruit); fig. 3-47 this work.
      Type.—China, Hupeh Province, Hsing shan district (E. H. Wilson).   Herb. Arnold Arbor., Harv. Univ., Cambridge.
      Distribution.—West-central and southwestern China: Hupeh, Szechwan, Kweichow, and Yunnan provinces.
      Common name.—Ichang papeda.
      A spiny shrub or small tree usually 5-15 ft. [1.5-4.6 m] high; twigs angular when young, with stout, sharp spines, 1.5-2.5 cm long, often reduced or lacking on flowering twigs; leaves narrow, 4-6 times longer than wide, mostly 8-11.5 X 1.8-3 cm; petioles very large, broadly winged, obovate or oblong-spatulate, evenly rounded at the tip and narrowed abruptly at the base, usually 3.5-6 X 2-3 cm; leaf blade ovate-acuminate, often more or less caudate, emarginate at the tip and evenly rounded or bluntly pointed at the base, usually 3-6 X 1.8-3 cm, often not equaling the winged petiole in area; flowers 2.5-3 cm diam., 5-merous; sepals thick, 3 mm long and 3 mm wide, with minutely ciliate margins; petals oblong, 1.5-2 X 0.5-0.8 cm, white; stamens 20, at first all connate to the tips, finally breaking up into several bundles, about 10 mm long; pistil about 10 mm long; style very short, caducous; stigma nearly as large as the ovary; ovary with 7-9 locules, ovules numerous in each locule; fruits small, glabrous, 3-4 cm diam. in dried specimens (probably 3.5-5 cm when fresh), peel rough, 2-4 mm thick in dried specimens; seeds large, very thick, 12-18 X 12-18 X 7-9 mm in dried material, very blunt at both ends, with a chalazal cap about 12 X 5 mm, apparently monoembryonic.
      This species differs from its congeners in having large, very thick seeds and slender leaves 4-6 times longer than broad, with very large, winged petioles often as large or larger than the blade.   It differs from C. hystrix in having oblong, rather than triangular, winged petioles and much larger flowers with connate stamens.   (See fig. 3-47.)
      This remarkable plant,27 which grows in a truly wild state in central and southwestern China, is doubtless the most cold-resistant of all the evergreen species in the orange subfamily.   It differs greatly from the other species in the subgenus Papeda in having large flowers, partly connate stamens, and large, very thick seeds, which are nevertheless monoembryonic.   The leaves also differ from those of all other species of Papeda in having extremely long but rather narrow winged petioles, often exceeding in area the slender, pointed, lanceolate leaf blades.
      In 1926, Swingle brought to this country a healthy young seedling of this plant from Hupeh Province, grown for him at Nanking, China, by C. C. Hu.   As C. ichangensis is the hardiest known evergreen species of Citrus, it is of great interest for use in breeding cold-resistant hybrids, such as the Ichang lemon and Yuzu.

HYBRIDS OF CITRUS ICHANGENSIS
      Ichandarins [Citrus ichangensis X C. reticulata cv. satsuma group].
      An ichandarin was produced by the citrus experts of the former Bureau of Plant Industry (now encompassed by the Agricultural Research Service) by applying the pollen of the Ichang papeda to properly prepared flowers of the satsuma orange.   Naturally all chances of foreign pollen reaching the crossed flowers were excluded, and of course the stamens of the satsuma flowers had been cut away before any pollen had matured.
      The leaf blades of this hybrid are broadly elliptical, 6 to 8.5 by 2 to 4 cm, slightly acuminate at the tip and cuneate at the base (angle 70° to 85°), margins minutely serrulate, dentate above the middle.   The leaf blades differ only slightly from those of the Yuzu, a hybrid in which the leaf blades are more acuminate at the tip and more broadly rounded at the base.   The winged petioles are obovate, oblanceolate, or elliptical, broadly rounded at the tip and gradually narrowed into the base when the petiole is merely margined.   The petioles are so much like those of the Yuzu that it is often impossible to distinguish between them with certainty.   This hybrid is of interest chiefly in that it throws light on the origin of the Yuzu, a well-known Chinese and Japanese citrus fruit tree.
      It is very probable that if some strain of C. reticulata other than the satsuma, for example the tangerine or the Willowleaf mandarin, had been used as one of the parents the leaf blade would have been more acuminate at the tip and more broadly rounded at the base, i.e., like the leaf blade of the Yuzu, which is doubtless such a hybrid.
      Yuzu [Citrus ichangensis X C. reticulata var. austere ?].
      Illus. Takahashi, Kankitsu Saibai 112, fig. 31. 1912; Makino, Shokubutsu Dzukan 1064, fig. 2022. 1917; Tanaka, Jour. Hered. 13:pl. facing p. 243. 1922; Hu, Nogyo Oyobi Engei 5:1637-1638, figs. 34-37. 1930; idem, Calif. Citrog. 16:544, pl. 2, fig. 15. 1931; idem, Jour. Agr. Ass. China Nos. 126-127:58, fig. 15. 1934; Murakoshi, Nai-gai Shokubutsu Genshoku Dai-zukan 5:118, pl. 24, fig. 251 (col.). 1935.
      The Yuzu is a medium-sized spiny tree; leaves lanceolate-acuminate, with rounded bases but with pointed, usually acuminate tips, slightly crenulate-margined toward the tips, leaf blades 5-7 X 2.5 X 3.5 cm; winged petioles obovate, 18-30 X 6-15 mm, with entire or very faintly crenulate margins; fruits depressed-globose, usually with 10 locules, 5-7 cm diam., 4.5-5.5 cm high, with a rough, bumpy peel, greenish in color when ripe; pulp very acid, and somewhat bitterish; seeds plump, about 12-14 X 7-8 X 6-7 mm.
      This well-known citrus fruit tree of China and Japan is doubtless another ichandarin resulting from an accidental cross-pollination of some cultivated variety of the mandarin orange by the Ichang papeda, probably accomplished by some insect.   The Yuzu is unlike either parent species in many important taxonomic characters; however, there has been little opportunity to learn its characters or experiment with it since one parent, C. ichangensis, is a wild species (apparently never cultivated in China) that was not discovered until 1913.   The hybrid nature of the Yuzu was not known even in the Orient, where C. ichangensis is a botanical curiosity.   As has been described above, hybrids that show astonishing similarity to the Yuzu have now been produced in this country between the Ichang papeda and the satsuma orange (a form of C. reticulata).
      The Yuzu is sparingly cultivated in north-central China, in Kiangsu, Chekiang, Hupeh, and Kansu provinces, and in the plateau regions of southwestern China and as far south as Yunnan Province.   It is more commonly grown in Japan, both for its acid fruits, which are used as a substitute for lemons or limes, and as a rootstock for the satsuma and other cultivated varieties of citrus fruits.
      Meyer, in October, 1914, found this hardy citrus fruit tree in northwestern China, in the latitude of Atlanta, Georgia, at Hsi-Chi village, near Siku (Lat. 33° 44' N., Long. 104° 30' E.), in the southern part of Kansu Province.   It was growing at an altitude of 610 to 1,372 meters (2,000 to 4,500 ft.) along with walnuts, persimmons, pomegranates, and the Trachycarpus palm.   Meyer (1918) described the fruit as follows: "The fruits were loose-skinned, round flattened, the size of mandarin oranges, color of rind light yellow; rind full of oil glands, smelling like a fine lemon; segments separating easily; fairly juicy and of an agreeable sharp sour taste; contains plenty of large seeds."   Meyer's photograph, taken at His-Chi, Kansu Province, was published by Tanaka (1922, pl. facing p. 243).
      Two strains of this hybrid grown in north-central China were described by Hu (1934, pp. 47-48).   They are: (1) hsiang ch'êng (aromatic ch'êng), a name current in Chekiang and Kiangsu provinces; and (2) Lo han ch'êng (Buddhist disciple Ch'êng), a name used in T'ang-ch'i, Chekiang Province.   Hu (1930) has shown the first in his figures 34 and 35 and the second in figures 36 and 37.
      Tanaka has directed attention (1933b) to certain Chinese records which seem to prove that this plant was known and cultivated in ancient China under the name yu.   This name is still used for it in Japan but not in China, where the name is now applied to the pummelo (C. grandis), a very different species having much larger fruits with agreeably flavored sweet pulp.   The Yuzu is now called ch'êng tzu in China and was so called as early as 1108 A.D. by Tang Shên-wei in his great illustrated herbal, the Chêng lei pên ts'ao, published in that year.
      In the "Spring and Summer Annals" of Lü Pu-wei, who died in 237 B.C., the yu of that epoch was described as follows (as translated by Michael J. Hagerty): "Some are sweet and some are sour.   The sour are called hu kan or Barbarian sweet.   At present the common people sometimes speak of the ch'êng as yu but this is wrong."   This last sentence proves, as noted by Hagerty, that there were already in the third century B.C. two different fruits called yu.   It seems probable that the Japanese name yuzu, which corresponds to yu tzu in the Chinese spoken language, has been kept with its ancient meaning in Japan, but that the meaning has for many centuries been lost in China.
      The Yuzu was named Citrus junos (as a good species) by Tanaka, but, as has been shown above, it is very probably a hybrid of the Ichang papeda and some Chinese cultivated variety of the mandarin orange and therefore cannot be recognized as a good botanical species.
      The Yuzu is worthy of trial as an acid fruit for the home garden in subtropical or warm-temperate climates that are too cold to permit the growth of other acid citrus fruits.   The Yuzu was formerly used widely in Japan as a rootstock for the satsuma orange, but now the trifoliate orange is almost exclusively used in nurseries which propagate the satsuma on a large scale, doubtless partly because seeds of the Yuzu are only to be obtained in moderate quantity from fruits that have a market value, whereas trifoliate orange seeds can be had in large quantity from fruits that are of no value.
      Nagai and Takahashi (1928, pp. 2018-22, and 1925 [in Japanese]) found that of the five following rootstocks—Yuzu, sweet orange, sour orange, trifoliate orange, and Japanese summer orange (Natsudaidai)—the Yuzu gave the best results when approach-grafted in order to rejuvenate Thomson navel trees on trifoliate orange rootstocks.   The sweet orange did nearly as well, but it is subject to foot rot, a disease that does not attack the Yuzu.
      Shangyuan, or Ichang lemon [Citrus ichangensis X C. grandis var. ?].
      Illus. Swingle, Jour. Agr. Res. 1:figs. 2-3. 1913; Hu, Calif. Citrog. 16:544, fig. 18. 1931; idem, Jour. Agr. Ass. China Nos. 126-127:61, fig. 29. 1934.
      Distribution.—China: Lower valley of Yangtze River in Hupeh, Anhwei, Kiangsu, and Chekiang provinces; sparingly cultivated.
      An erect, much branched tree with lower branches short, giving the tree a columnar outline; twigs glabrous, spiny; leaves ovate, ovate-lanceolate, or ovate-acuminate, apices sometimes minutely emarginate, sometimes bluntly pointed, margins obscurely crenulate, leaf blades about 3 times as long as the petioles, 6-9 X 2.5-3.5 cm, glabrous, tips rounded, margins subentire; petioles more or less broadly winged, 20-25 X 6-12 mm; fruits large, 8-11 X 7-10 cm, broadly obovate, usually with a low mammillate tip surrounded by a shallow furrow; peel thick, 7-10 mm, lemon yellow when ripe, very fragrant; oil glands concave; segments 10 or 11 with numerous large seeds, 4-6 per segment, usually 16-20 (sometime longer) X 9-12 X 7-10 mm, tapering into a flattened truncate base, sometimes polyembryonic; pulp very acid, but with a strong, aromatic aftertaste; pulp-vesicles fusiform, 8-10 X 2-4 mm, on stalks 2-8 mm long, with a central streak of minute granules of acrid oil.
      The Shangyuan, or "Ichang lemon," which was introduced into the United States about 1914, has proved to be very resistant to winter cold.   In southern Alabama it has endured without harm temperatures that injured or killed most of the satsuma orange trees of the same age growing near by, although the satsuma is a very cold-resistant variety.   At first the plant was supposed to be C. ichangensis but was soon found to be very different, as it has very much larger fruits and very much smaller winged petioles; it is, in fact, a hybrid.   There is some variation among different trees of the Ichang lemon growing in China, such as is found among other violent hybrids of Citrus.
      Hu in 1930 named this hybrid Citrus grandis var. Shanyuan; the next year he corrected the spelling of the varietal name to read Shangyuan (1931, p. 544).   He noted the similarity of this tree to the pummelo but stated that the fruits are smaller and have a bitter taste.   He considered it to be a hybrid of the pummelo.   He also reported (1930) that it is grown in the lower valley of the Yangtze River, in Anhwei, Kiangsu, and Chekiang provinces; later (1931) he stated that the fruit, though ornamental, is not eaten and that the tree is hardy enough to be grown in Nanking.
      Hu's varietal name for this hybrid is a phonetic rendering of the Chinese name for it, namely, Hsiang yüan, or "Fragrant ball," an allusion to its pleasing aroma.   Tanaka (1930a, p. 32) changed Hu's name to a botanical one, Citrus hsiangyuan Hort.   Later he named the hybrid Citrus wilsonii (1932b, p. 37), but stated: "The present species unquestionably has Citrus junos in one parent and seems to be a cultigen originated through the chance seedling."
      There is no valid reason for calling this plant a botanical species, as it is undoubtedly a chance hybrid that arose under cultivation.   Its resemblances to the Yuzu, remarked by Tanaka, are doubtless due to the fact that both the Yuzu and the Ichang lemon, or Shangyuan, are hybrids of C. ichangensis with two different species of the subgenus Citrus.   Fortunately, the International Code of Botanical Nomenclature (Lanjouw, 1961, p. 31) now gives a very simple method of citing such sports and cultigens under the name of the species from which they arose by giving the variety name in some living language (not in Latin) following the species name and surrounded by single quotes.   Under this rule any cultivated form of a species of Citrus can be cited after the species name under its common name; for example, C. sinensis 'Hamlin' or C. reticulata 'Dancy.'   Many of the complex hybrids of Citrus can best be handled in this way until their origin is worked out experimentally.
      In China, the ripe fruits of the Ichang lemon are used to perfume rooms or cabinets.   In the United States, the fruit has been occasionally used to make lemon pie, the heat of baking evidently driving off the acrid oil that gives a disagreeably strong odor to "lemonade" made from the juice; some who have made pies from Shangyuan juice prefer them to regular lemon pies.
      The Ichang lemon is a very hardy plant and deserves trial as a rootstock in this country, where it grows vigorously.   It produces very numerous large seeds that should make it easy to grow in the nursery.   It also produces pollen freely and hybridizes spontaneously with the satsuma orange if growing near by; doubtless it could be crossbred with other forms of citrus fruits as well.   It should be used in breeding new hybrids for trial as rootstocks or as acid fruits for home use in the cotton belt.

      12.   Citrus latipes (Swing.) Tan. Studia Citrol. 2:155. 1928. Citrus ichangensis subsp. latipes Swing. Jour. Agr. Res. 1:12. 1913. Illus. Tanaka, Kankitsu No Kenkyû (Citrus Studies), 79, fig. 72, also 167, fig. 136(1). 1933; fig. 3-48 this work.
      Type.—Northeastern British India, Khasi Hills, Living Bridge (Hooker f. and Thomson).   Herb. Kew.
      Distribution.—Northeastern India: Khasi Hills; northern Burma; grows in the mountains at considerable elevation, 500-1,830 m [1,640-6,000 ft.].
      Common name.—Khasi papeda.
      A thorny tree similar to C. ichangensis but having leaf blades more variable in size and shape and with the tips subacute or even bluntly rounded, not apiculate or subcaudate with blunt points as in C. ichangensis.   The flowers, instead of being borne singly in the axils of the leaves as in C. ichangensis, are sometimes, at least, borne in small axillary racemes with 5-7 flowers.   These latter flowers are much smaller, are 4-merous (instead of 5-merous as is the large single flower of C. ichangensis), and all the parts are smaller.   However, the fruits are borne singly and resemble those of C. ichangensis except for having a thicker peel, of which the inner layer is chalky white just below the outer green layer.   The seeds are also smaller and more numerous than those of C. ichangensis and are arranged 5-7 in each segment.
      The following is a detailed description of flowers and fruits (after Tanaka, 1928a, p. 155): Inflorescences racemose, having 5-7 flowers; peduncles short, 5-8 mm, pubescent; pedicels 7-8 mm long, moderately thick, becoming thicker at the upper end where they join the calyx, striated, glabrous; flower buds medium-sized, ovoid, about 7 X 9 mm, rounded at the apex, showing oil glands indistinctly at the surface; calyx discoid, 4 mm diam., lobes 4, broad, semicircular, wrinkled, slightly convex at the tip and curving outward, rather thick, slightly pubescent at margins, glabrous above; flowers, when open, 1.5 cm diam.; petals 4, spreading horizontally, thick, oval; stamens 18-20, 7-8 mm long and about 1 mm wide, filaments almost free at the tip but arranged in 6-7 groups each composed of 3-4 united together at the bases, thick, flat, wrinkled, pointed at the tip; anthers oval or elliptic-oval, about 1.5 mm long, pointed, with small gland at the apex; pistils shorter than stamens, about 7 mm long; ovaries depressed-globular, 2 mm diam.; styles rather stout, 2.5 mm long, 1 mm diam., distinctly articulated with the ovary; stigmas large, depressed-globose, 1.5-2.5 mm diam., with 4 sometimes indistinct furrows; fruits medium-sized, almost globular, slightly compressed at both ends; oil cells moderately small, more or less equal in size, rather dense, convex; peel somewhat thick, composed of rather dense tissue; central column (axis) of even width, moderately large; segments 9, fairly large, outer margin very much rounded, inner margin also rounded; pulp-vesicles few, fairly large, spindle-shaped, not short, relatively well developed with a thick membrane, stalks very short; seeds numerous, arranged 5-7 in each row, rounded, moderately large, parallel and horizontal.
      This species was founded on plants collected in the Khasi Hills of northeastern India, but very similar forms are found in the mountains of northern Burma.   The type specimen in the Kew Herbarium on which Swingle based C. latipes (1913c, p. 12) has a fruit in cross section measuring 5 cm in diameter, peel 5 to 6 mm thick, with nine segments.
      In view of the extreme cold-resistance of the typical C. ichangensis (and its hybrid, the so-called Ichang lemon) it would be well worth while to introduce C. latipes into the United States and see if it also is hardy and perhaps of value for use in breeding new hardy acid citrus fruits and hardy rootstocks.

Subgenus Papeda Section Papeda
      Subgenus Papeda Section Eupapeda Swing. in Webber & Batchelor, Citrus Indus. 1:432. 1943.
      Flowers small (less than 2 cm, often less than 1.5 cm diam.); stamens free from the beginning; winged petioles broad, and more or less triangular; seeds small or medium-sized, not angular, much longer than broad.
      The species of the section Papeda of the subgenus Papeda are remarkable in having very large winged petioles and small flowers (often very small) with free staminal filaments.   The pulp-vesicles contain numerous droplets of acrid, bitter oil that gives the pulp a very disagreeable taste.   The pulp-vesicles are often attached not only to the dorsal walls of the locules, but also to the radial walls for one-half to three-fourths the distance from the dorsal wall to the central core of the fruit.   In this character these species approach Clymenia.
      As a result of the widespread use of the fruits by natives, these species have come into culture on a small scale about villages throughout the East Indies and in Melanesia and Polynesia.   In this way these species have been exposed to cross-pollination not only with other species of the Papeda group but also with edible-fruited species of the subgenus Citrus grown as dooryard trees.
      Rumphius, near the close of the seventeenth century, described and figured many forms of the Papeda group found growing in Amboina and in other islands of the East Indian Archipelago; his observations were published by Jan Burmann in 1741-1755 (Herbarium Amboinense, vols. 2 and 5).   Wester (1913; 1915, 1917a, 1917b, and 1924) published good descriptions and photographic illustrations of many forms of the section Papeda group found growing wild or cultivated on a small scale in the Philippine Islands, especially in the southern islands, Bohol, Cebu, Negros, and Mindanao.   Several forms are likewise found in the southern half of Luzon, the northernmost island.
      It is clear from the works of Rumphius and Wester that the species of the section Papeda have been very extensively hybridized, almost as much so as those of the subgenus Citrus.
    However, many of the species of the section Papeda group still occur in a truly wild state in the primeval forests of southeastern Asia, the East Indies, New Guinea, and neighboring islands.   In these species and varieties it is thus possible to see the unobscured results of many million years of evolution, whereas in the edible-fruited species, varieties, and forms belonging to the subgenus Citrus the results must be surmised rather than directly observed.   The papedas, because of their acrid, disagreeable flavor, have not been considered by citrus experts of sufficient value to merit study.   However, hybrids between the species of the subgenus Papeda and species of the subgenus Citrus have been known to yield rootstocks very well adapted to support the commonly cultivated citrus fruit trees.   Several of the wild species of papedas are very vigorous and deserve trial as rootstocks for the edible species of citrus.   Nothing is known about the vitamin C content of the papedas or about which glucosides, if any, they contain.
    13.   Citrus micrantha Wester, Phil. Agr. Rev. 8:20, pls. 5c, 6b. 1915. Citrus macroptera var. micrantha (Wester) Tan. Trans. Nat. Hist. Soc. Formosa 22:430. 1932. Illus. Wester, loc. cit. pls. 5c, 6b, and Phil. Bur. Agr. Bul. 27:pl. 12. 1913.
    Type.—Philippines, Bohol (Wester, No. 4829).   Herb. Bur. Sci., Manila.
    Distribution.—South Philippines.
    Common name.—Small-flowered papeda.   Native name: Biasong.
    The original description of this species by Wester reads as follows: "A tree attaining a height of 7.5 to 9 meters, with comparatively small but sharp spines; leaves 9 to 12 centimeters long, 27 to 40 millimeters broad, broadly elliptical to ovate, crenate, rather thin; base rounded or broadly acute; apex acutely blunt pointed; petioles 35 to 60 millimeters long, broadly winged, up to 40 millimeters wide; wing area sometimes exceeding leaf area; flowers small, 12 to 13 millimeters in diameter, white, with a trace of purple on the outside, 2 to 5, in axillary or terminal cymes; petals 4; stamens free, equal, 15 to 17; ovary obovoid, locules 6 to 8; style slender; distinct; fruit 5 to 7 centimeters long, 3 to 4 centimeters in transverse diameter, averaging 26 grams in weight, obovate to oblong-obovate, somewhat compressed towards base; apex blunt pointed; surface fairly smooth or with transverse corrugations, lemon yellow; skin comparatively thick; pulp rather juicy, grayish, acid; aroma similar to that of samuyao; juice cells short and blunt to long, slender and pointed, sometimes containing a minute greenish nucleus; seeds many, flat, pointed, more or less reticulate."
      This species stands out from all others of the subgenus Papeda because of its very small 4-merous flowers, only 1.2-1.3 cm diam. when open.   The fruits are oval or obovate, blunt-pointed at both ends, 5-7 cm long but only 3-4 cm wide, with very few locules (only 6-8).   The leaves have broadly ovate blades, blunt-pointed at both ends, 3.5-6 cm long, 2.7-4 cm wide.   The broadly winged petioles, measuring 3.5-6 X 3.5-4 cm, have a nearly straight or slightly curved top and gradually curved sides, tapering to a blunt base.   Seeds many, flat (2.5-3 mm thick).
      This species is sparingly cultivated in the southern Philippine Islands (Cebu, Bohol, Negros, and Mindanao), where it is called biasong.   Wester stated that the fruit of this species is used as a hair wash but is not eaten and is of no economic importance.
      13a.   Citrus micrantha var. microcarpa Wester, Phil. Agr. Rev. 8:21, pl. 7b. 1915. Illus. Wester, loc. cit. pl. 7b, and Phil. Bur. Agr. Bul. 27:pl. 14. 1913.
      Distribution.—Southern Philippines: Cebu and Bohol.
      Common name.—Small-fruited papeda.   Native name: Samuyao.
      The original description of this variety by Wester reads as follows: "A shrubby tree, 4.5 meters tall, with slender branches and small, weak spines; leaves 55 to 80 millimeters long, 20 to 25 millimeters broad, ovate to ovate-oblong or elliptical, crenulate, thin, of distinct fragrance, base rounded to broadly acute; apex obtuse, sometimes notched, petioles 20 to 30 millimeters long, broadly winged, about 14 millimeters wide, wing area somewhat less than one-half of the leaf blade; flowers in compact axillary or terminal cymes, 2 to 7, small, 5 to 9 millimeters in diameter, white, with trace of purple on the outside; calyx small, not cupped, petals 3 to 5; stamens 15 to 18, free, equal; ovary very small, globose to obovate; locules 7 to 9, style distinct; stigma small, knob like; fruit 15 to 20 millimeters in diameter, roundish in outline; base sometimes nippled; apex an irregular, wrinkly cavity; surface corrugate, greenish lemon yellow; oil cells usually sunken; skin very thin; pulp fairly juicy, acid, bitter with distinct aroma; juice cells very minute, blunt, containing a small greenish nucleus; seeds small, flattened, sometimes beaked."
      The samuyao was found by West and Brown (1920, pp. 208-09) to yield from the rind "a clear, almost colorless oil which is very fragrant.   It should be useful as a perfumery oil."   The peel yielded 1.29 per cent of oil.   They stated that the crushed fruits are used by the women of Cebu for cleansing the hair and are added to coconut oil to give it fragrance when applied to the hair.   The tree bears within five years from planting and produces fruits during the entire year, but more abundantly during the rainy season.   West and Brown (1920, fig. 70) show the fruits as distinctly pyriform, broadly rounded at the tip and tapering into a narrow, conical base.
      This variety has the smallest fruit and the smallest flowers of any True Citrus Fruit Tree known but curiously enough has more segments in the fruit (seven to nine) than C. micrantha (six to eight), although the fruits of the latter are much larger, 5 to 7 by 3 to 4 cm, instead of 1.5 to 2 cm, as in the variety microcarpa.
      Wester stated (1915, p. 22) that in Bohol Island, P.I., "a somewhat more vigorous variety of the samuyao was found that is called samuyao-sa-amoo" which has "fruits a little larger, and smoother, and longer than broad, otherwise similar to the samuyao."   This description calls to mind Rumphius' Limonellus aurarius, the goldsmith's lime, which occurs sparingly in Amboina, Moluccas, and Ternate, a few hundred miles directly south of Bohol.   According to Rumphius (1741, vol. 2, p. 109, pl. 30), the fruits of this plant are the smallest of all the limes (his figure shows ovoid fruits 2.5 to 3 by 2 to 2.5 cm) and the leaves are also small, only 3 inches long in all, about one-third of which represents the cordate, winged petiole.   The fruits are not eaten but are used by goldsmiths to clean gold objects.   The fruits also enter into native medicines.
      Both C. micrantha and its variety microcarpa are of much scientific interest, as they are perhaps the most primitive members of the subgenus Papeda, itself the most primitive group of species of the genus Citrus.
      14.   Citrus celebica Koord. Fl. N. O. Celebes, Mededeel. uit 's Lands Plantent. 19:639, also 370. 1898. Illus. Koorders, Suppl. Fl. N. O. Celebes 2:pl. 19. July, 1922; loc. cit. 3:pls. 10-11. December, 1922.
      Type.—Northeastern Celebes (Koorders).   Hort. Bot. Buitenzorg, Java (now Kebun Raya, Bogar, Java).
      Distribution.—Northeastern Celebes, southern Philippines.
      Common name.—Celebes papeda.
      Leaves with small, lanceolate, subcrenulate blades, 4.5-5.5 X 2.5-3.5 cm; winged petioles larger than the blades, 5-6.5 cm long, 3-3.4 cm wide, oblanceolate, with a subcordate or slightly curved upper end and gradually rounded sides tapering into a deltoid base, having from 5-10 mm of the petiole almost wingless; margins faintly crenulate; flowers occurring in 3- to many-flowered spikes or small cymoid clusters in the axils of the leaves; flowers medium-sized, 15-17 mm when open, 4-5-merous; ovary reversed-pyriform, tapering sharply below, broadly rounded at the tip with many (17-20) locules; style short, 5-6 mm; stamens short, free; fruits large (10 cm diam.), globose, with a very thick peel (about 3 cm); seeds small, probably monoembryonic.
      A small tree 5 m high, growing in a wild state in primeval forests at Karowa, in extreme northeastern Celebes (Minahasa) (Lat. 1° 15' N., Long. 124° 45' E.); fruits inedible.
      The original description of this species published in 1898 was not adequate to show its nature clearly, as the description did not give the number of locules in the ovary, an important character in the subgenus Papeda.   Twenty-four years later C. celebica was fully figured; then it became clear that this species was distinct from C. macroptera.
      14a.   Citrus celebica var. southwickii (Wester) Swing. Jour. Wash. Acad. Sci. 28:533. 1938. Citrus southwickii Wester, Phil. Agr. Rev. 8:16-17, pls. 3c, 4c. 1915. Illus. Wester, loc. cit. pls. 3c, 4c, and Phil. Bur. Agr. Bul. 27:pl. 16b. 1913. (fide Wester, loc. cit. 1915 = C. southwickii).
      Type.—Philippines, Bohol (Southwick, Wester).   Herb. Bur. Sci., Manila.
      Distribution.—Philippines: Bohol and Mindanao.
      This plant appears to be a variety of Citrus celebica, with which it agrees in having very numerous locules (15-19) in the ovary.   It also agrees well in flower characters, especially in the short style and the ovary tapered toward the base, but differs in having smaller fruits with thin peel, and in having the blade of the leaves decidedly larger than the winged petioles and often much blunter-pointed.   The winged petioles are similar to those of C. celebica but decidedly smaller.   Both the species and the variety have smooth fruits.   The difference in the diameters of the fruits is misleading, since the fruits of variety southwickii, because of their thin skin, have more pulp than the decidedly larger but very thick-skinned fruits of the species.
      The variety southwickii was found by E. F. Southwick in Bohol, where it is not uncommon and where it is cultivated.   It was also found by Wester in Baganga (Lat. 7° 35' N., Long. 126° 35' E.), in eastern Mindanao, about 450 miles north-northeast of Karoa in northern Celebes.   It has been introduced into the United States.
      It is possible that this plant may prove, upon more detailed study, to be a complex hybrid of C. celebica with one or two other species.   The thin, smooth skin and the abundant, though acrid, pulp and the short, broad-winged petioles arouse suspicion that it may be a complex hybrid of C. celebica with a pummelo (C. grandis), or with some other species of the subgenus Citrus.
      This variety is promising for breeding purposes because of its fair-sized, juicy, smooth, thin-skinned fruits.   The bitter taste of the pulp, due to the acrid oil in the pulp-vesicles, can doubtless be removed by hybridization with the species of the subgenus Citrus that have little or even no oil in the pulp-vesicles, such as C. reticulata.   It is highly probable, even if variety southwickii proves to be a dilute hybrid of the pummelo, that wild forms of C. celebica will be found native in the southern islands of the Philippines, since southern Mindanao is only about 300 miles north of northeastern Celebes; Bohol, where variety southwickii occurs, is less than 400 miles distant.

OTHER POSSIBLE HYBRIDS OF CITRUS CELEBICA
      Alemow.
      This hybrid, named C. macrophylla by Wester (1915, p. 16, pls. 3b, 6c), is sometimes cultivated in Cebu, P.I.   It has large leaves, with the blades 12 to 14 cm long and 6 to 8 cm wide, with much smaller, subtriangular, short-winged petioles, measuring up to 3.5 cm wide near the top.   The fruits are very large, 8.5 to 10 cm in diameter, subglobose to oblong, more or less narrowed at the base, with a rough, transversely-corrugated, but rather thin skin.   The fruit has 13 to 16 segments and rather dry, sour pulp, considered inedible even by the natives.
      This seems to be a hybrid of C. celebica, or some other species of the subgenus Papeda, with a species of the subgenus Citrus, probably a pummelo (C. grandis).   It has shown some promise as a rootstock for lemons in California.
      Kabuyao.
      This is another possible hybrid of C. celebica.   It is called kabuyao in Luzon Island, P.I., and kopahan in Bohol Island.   It was described by Wester (1915, pp. 17-19, pl. 5b, fig. 1; and 1924, p. 93, pl. 31c) as typical C. hystrix.   It differs from that species, however, in having more numerous stamens (30 to 36) and more locules (13 to 18) in the ovary.   The fruit is subglobose or short-pyriform, 7 to 9 cm long and 7 cm in diameter, with a smooth skin; the pulp-vesicles contain a "nucleus," doubtless composed of droplets of acrid oil such as are found in all species of the Papeda group and in most of their hybrids.
      Amongpong.
      This plant is found in Bohol Island, P.I., and is evidently similar to the kabuyao, but according to Wester (1915, p. 18) it has only 26 to 30 stamens and a larger fruit, 10 cm in diameter.
      These two hybrids, the kabuyao and the amongpong, and other similar fruits may perhaps be hybrids of C. celebica (which has many locules in the ovary) with C. macroptera or with C. hystrix, both members of the subgenus Papeda.   The increased number of stamens and the larger fruits might result from such hybridization.

      15.   Citrus macroptera Montr. Mém. Acad. Sci. Lyon 10:187. 1860. Citrus papuana F. M. Bail. Contr. Fl. Brit. N. Guin. 1903; C. aurantium subsp. saponacea Safford, Contr. U.S. Natl. Herb. 9:226. 1905. Illus. Bailey, loc. cit. (text fig.); Tanaka, Kankitsu No Kenkyû (Citrus Studies), 131, fig. 106. 1933 (?); fig. 3-49 this work.
      Type.—New Caledonia, Art Island (Montrouzier).   Herb. Fac. Med. Pharm., Univ. Lyon, France.
      Distribution.—Thailan