JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY EDITORIAL BOARD C. E. KOBUSKI, Editor I. M. JOHNSTON I. W. BAILEY KARL SAX VOLUME XXXIV JAMAICA PLAIN, MASS. 1953 Reprinted with the permission of the Arnold Arboretum of Harvard University KRAUS REPRINT CORPORATION New York 1968 eee DEC 121968 Pa! Cle) 022 DATES OF ISSUE No. 1 (pp. 1-96) issued January 15, 1953. No. 2 (pp. 97-190) issued April 15, 1953. No. 3 (pp. 191-300) issued July 15, 1953. No. 4 (pp. 301-425) issued October 15, 1953. Printed in U.S.A. TABLE OF CONTENTS STUDIES IN THE BORAGINACEAE, XXIV. A. THREE GENERA SEGRE- GATED FROM LITHOSPERMUM. B. SUPPLEMENTARY NOTES ON LITH- OSPERMUM. By Ivan M. Johnston A RECLASSIFICATION OF LIBOCEDRUS AND CUPRESSACEAE. With two plated. By Hui-Lin Li . Stupies oF PaciFIc ISLAND PLANTS, XIV. NOTES ON THE FIJIAN SPECIES OF CyRTANDRA. By A. C. Smith A NEw FIJIAN SPECIES OF a With two plates. By I. W. Bailey and A. C, Sm A Taxonomic REVISION oF Popocarpus. VII. THE AFRICAN SPECIES OF PopocaRPUS: SECTION AFROCARPUS. By Netta FE. Gray ............ Tur MorPHOLOGY AND RELATIONSHIPS OF IDENBURGIA AND NOUu- HuysIA. With two plates. By /. W. Bailey and B. G. L. Swamy THE VEGETATION ON CoASTAL DocTtooTH LIMESTONE IN SOUTHERN Cusa. With one plate. By Richard A. Howard and Winslow R. Briggs Stupies oF Paciric ISLAND PLANTS, XVI. NoTES ON FIJIAN RUuBI- ACEAE. By A.C. Smith STUDIES IN THE THEACEAE, XXVII. MiscELLANEOUS NEW SPECIES IN THEACEAE. By Clarence E. Kobuski ILEX IN TAIWAN AND THE LiukIu IsLaNps. By Shiu-ying Hu ............ A Taxonomic REvISsION oF Popocarpus. VIII. THE AFRICAN SPE- CIES OF SECTION EUPODOCARPUS, SUBSECTIONS A AND E. By Netta E. Gray ... Tur IDENTITY OF VALENTINIA ILICIFOLIA Swartz. By Richard A. Howard and Claude E. Smith, Jr. ... New SPECIES AND DISTRIBUTION RECORDS FOR LAS VILLAS PROVINCE, Cua. By Richard A. Howard and Winslow R. Briggs ..........0....... Tur STRUCTURE AND DIAGNOSTIC SIGNIFICANCE OF CRATERIFORM BoRDERED PITS IN THE VESSELS OF CERCIDIUM. By iis (COSZO. ea tee a PLANTAE PAPUANAE ARCHBOLDIANAE, X XI. THE PAPUASIAN SPECIES ’ or Macaranca. By Lily M. Perry . egpenstastaussiuns STUDIES IN THE BORAGINACEAE, XXV. A REVALUATION OF SOME GENERA OF THE LITHOSPERMEAE. By Ivan M. Johnston 163 176 182 187 258 RESULTS OF THE ARCHBOLD EXPEDITIONS PAPUAN NOTHOFAGUS. With twenty-two figures and frontispiece. By C. Steenis G. G. J. van THE MorPHOLOGY AND RELATIONSHIPS OF THE CHLORANTHACEAE. With three plates and sixty-four text-figures. By B. G. L. Swamy THE ARNOLD ARBORETUM DURING THE FISCAL YEAR ENDED JUNE 30, 1953 BIBLIOGRAPHY OF THE PUBLISHED WRITINGS OF THE STAFF AND STUDENTS JULY 1, 1952—JUNE 30, 1953 STAFF OF THE ARNOLD ARBORETUM 1952-1953 INDEX TITLE-PAGE AND TABLE OF CONTENTS .... 375 JOURNAL OF THE ARNOLD ARBORETUM Vor. XXXIV JANUARY 1953 NUMBER | STUDIES IN THE BORAGINACEAE, XXIV A. THREE GENERA SEGREGATED FROM LITHOSPERMUM Ivan M. JOHNSTON CONCERNED HERE are three species of the Mediterranean and adjacent areas which have been accepted as members of the genus Lithospermum, viz., L. apulum (L.) Vahl, L. microspermum Boiss., and L. callosum Vahl. These three plants share a distinctive type of nutlet, and despite their many differences appear to be more closely related to one another than to any of the other species referred to Lithospermum. A consideration of their characters gives reasons for treating them as three monotypic genera. The nutlets in our three species, as compared with those of true Litho- spermum, are rather small and have a thin pericarp. Above the small, obliquely basal attachment they have a ventral keel formed by the over- lapped margins of a closed ventral suture. In contrast the thick-walled. usually much larger and more broadly attached nutlets of Lithospermum have the ventral suture more or less obliterated. Its margins are tightly joined, commonly more or less confluent, and not well defined nor merely overlapping. Although our plants are not closely related to Megastoma., Sericostoma, and Echiochilon, the behavior of the ventral nutlet-suture in that group of interrelated genera is suggestive in interpreting that in our three species. In the genera mentioned the suture may be open nearly to the nutlet apex or be closed in varying degree by the overlapping of its margins. This variation is associated with the observable changes from a narrow elongate to a broad low gynobase and the accompanying shift from a distinctly lateral to a more and more basal nutlet-attachment. The suture in our three species, closed simply by the overlapping well- defined margins, gives every evidence of being the end product of a comparable series of changes in nutlet-attachment. In other words our three species probably had immediate ancestry with distinctly lateral rather than basal nutlet-attachment. In contrast, bony basifixed nutlets having the ventral suture obscured or obliterated by fusing and confluence of the seam-margins are present not only in true Lithospermum but also in a wide circle of obviously related genera, including such diverse plants as Onosma, Cerinthe, Echium, Alkanna, etc. This type of nutlet is a stabilized one and 2 JOURNAL OF THE ARNOLD ARBORETUM | [vot. xxxiv characteristic of a large and obviously old group of genera. It is not one from which the nutlets of our three species could have been derived. The indications are, therefore, that our plants are not directly related to Lithospermum or its immediate relatives. Although having nutlets of a similar basic type, the three species differ from one another in many significant details, and each of them has one or more unusual developments that separate it from all species of Lithosper- mum as well. There seem the best of reasons, hence, for giving each of them generic recognition. Their salient characters are revealed in the following synopsis. Moltkiopsis. Calyx circumscissile at the base. Corolla blue, evidently longer than the calyx, hairy outside, glabrous inside except for hairs on the swollen annular nectary low in the tube. Stamens affixed high in the corolla, two of them having shorter filaments and a slightly lower attachment than the other three, fila- ments much longer than the anthers. Pollen ellipsoid, 25-30 & 20-25 u, in lateral profile with sides convex or nearly straight or with one of the sides obscurely angled, broadest at the middle; pores three, about the equator. Style elongate, more or less exserted; stigma obscurely bilobed. Nutlets smooth and shiny, somewhat asymmetric, straight. Plant a suffruticose perennial with pallid, pungently bristly foliage and white exfoliating bark on the older stems; cymes small. Mairetis. Calyx tubular, elongate. Corolla blue, very small, scarcely longer than the calyx, with a villulose band inside at the base of the tube but otherwise completely glabrous; nectary not developed. Stamens affixed at equal heights high in the corolla, filaments twice as long as the anthers. Pollen spherical, 30-33 m in diameter, with usually eight equatorial pores. Style reaching up to the corolla-throat; stigma simple. Nutlets sparsely warted, nearly straight. Plant an annual herb with inconspicuously bracted cymes. Neatostema. Calyx divided. Corolla yellow, very small, the lobes minutely puberulent, throat with a continuous or broken band of hairs, inside of the tube villulose and bearing a ten-lobed, more or less hairy basal nectary. Stamens borne just above the nectary very low in the corolla; filaments much shorter than the anthers. Pollen ellipsoidal, 25-30 X 23-26 u, in lateral profile with sides slightly angled, broadest at the middle; pores eight, equatorial. Style very short, less than half the length of the corolla; stigma bilobed. Nutlets very coarsely warted, angulate, bent. Plant an annual herb with con- spicuously bracted cymes. Moltkiopsis, gen. nov. BORAGINACEAE — LITHOSPERMEAE. Calyx 5-partitus subsessilis maturitate sub basim loborum circumscissus, lobis lanceolatis conniventibus. Corolla caerulea elongata anguste tubuloso- infundibularis extus villulosa intus (nectario excepto) glaberrima, lobis aequalibus adscendentibus oblongis vel ovatis apice rotundatis, fauce am- pliata nuda nullo modo appendiculata vel glandulifera, tubo elongato basim versus nectario annulato tumido non rariter sparse villuloso praedito. Stamina e fauce corollae breviter exserta fere ad apicem tubi corollae in- aequaliter inserta, filamentis inaequalibus, duobus antheris aequilongis altitudine sinus corollae attingentibus, tribus antheris sesquilongioribus alti- 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 3 tudine supra medium lobis corollae attingentibus, antheris oblongis medio- affixis aequalibus apice rotunda. Pollinia ellipsoidea medio poris 3 donata. Stylus filiformis tarde exsertus stamina evidenter vel vix superans. Stigma depresso-capitatum simplex vel obscure bilobatum. Nuculae saepe 2 vel 3 laeves nitidae lanceo-ovoideae subasymmetricae apice acutae basi angus- tatae truncatae cicatrice plana parva transverse elliptica donatae dorso et latere rotundatae ventre angulatae. Carina ventralis nuculae margines sulci longitudinali clausi imbricata sed haud confluentes composita. Gynobasis depressa subplana. — Planta humilis perennis suffruticosa ramosa hirsuta. Caules plures erecti cortice albo vetustiore exfoliato obtecti. Folia pallida alterna. Cymae terminales abbreviatae bracteatae. Flores caerulei. — Nomen derivatur a Moltkia et dys, habitus, propter similitudinem cum genere Moltkia. Moltkiopsis ciliata (Forsk.) comb. nov. Lithospermum ciliatum Forsk. Fl. Aegypt.-Arab. 39 (1775); C. Christensen, ansk Bot. Ark. 4°: 14 (1922). Moltkia ciliata (Forsk.) Maire, Cat. Pl. Moroc. 4: 1102 (1941). Lithospermum angustifolium Forsk. Fl. Aegypt.-Arab. 39 (1775). Lithospermum callosum Vahl, Symb. Bot. 1: 14 (1790).—A renaming of L. angustifolium Forsk. Moltkia callosa (Vahl) Wettst. Oesterr. Bot. Zeit. 67: 368 (1918). Lithospermum callosum var. asperrimum Bornm. Mitt. Thiringisch, Bot. Verein 6: 58 (1894). This desert plant, which ranges across northern Africa and east to Iran, has by some been treated as a member of Moltkia but by most botanists accepted as a species of Lithospermum. Its nutlets, as well as its triporate pollen and circumscissile calyx, distinguish it quickly from all members of both genera. Because of its fruticulose habit and elongate blue corollas its gross habit is most suggestive of Moltkia. Its corollas, however, differ from those of Moltkia in a number of important details. They are hairy rather than glabrous outside. They have a well-developed nectary in the tube. They bear stamens of two sizes, not at a single but at two levels in the throat. The stamen-attachment is of special interest. As in Alkanna, two of the five stamens have filaments shorter and attached slightly lower in the corolla-throat than the other three. Unlike Alkanna, however, the two shorter and lower stamens are not juxtaposed nor are they abaxial. They are separated by a long stamen and appear to represent the two adaxial lateral members of the androecium. They are not abaxial laterals as in Alkanna. The behavior of the calyx in Moltkiopsis is also unusual. As the fruit approaches complete maturity the calyx becomes circumscissile just above the base, its connivent lobes drop away, and the ripe nutlets are completely exposed. After the nutlets are shed all that remains of the flower on the old inflorescence is the knob-like base of the subsessile calyx and the flattened gynobasic surfaces on its truncate summit. This condi- tion is unique among the herbaceous Boraginaceae. Over most of its wide range the plant has blue subtubular corollas 4 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv 10-15 mm. long. No evidence of cleistogamy has been detected. The only striking variation noted in the flower is that present among some, not all, plants collected in the region about the Persian Gulf. In these the corollas are very small (3—7 mm, long), the anthers are imperfectly devel- oped, and the style is very short (2.5-6 mm.). The cymes on such plants tend to be very slender, the flowers very crowded, and the bracts unusually small. Although they appear to have functional stigmas, no fruit has been found on plants producing the small crowded flowers described. The con- dition described may be the result of virus infection. Possibly also it might be an expression of gynodimorphism comparable to that known in Echium vulgare and Anchusa officinalis. In those species occasional indi- viduals produce not the normal hermaphrodite flowers, but only small functional female ones. Representatives of the small-flowered form of Moltkiopsis have been seen from Bushir, southern Persia, 1868, Hauss- knecht (Brit. Mus.), Bahrain Island, Fernandez 489 (Kew). and Kuwait. Dickson 224 (Kew) Mairetis, gen. nov. BORAGINACEAE — LITHOSPERMEAE. Calyx elongatus secus costas sparse hirsutus alibi sparse strigosus fere ad medium lobatus, sub anthesin subsessilis corolla paululo brevior tertiam partem inferioribus tubulosus, lobis erectis gracilibus donatus, in statu fructifero 4—5-plo accrescens breviter pedicellatus tubo quam lobis erectis cuneatis 1—2-plo longiore donatus. Corolla minuta symmetrica extus glaber- rima, lobis caeruleis erectis vel adscendentibus longitudine corollae 6—7-plo brevioribus, fauce brevi glabra subcampanulata nullo modo appendiculata vel glandulifera, tubo elongato flavo intus basim versus villuloso alibi glabro nectarium haud gerenti. Stamina in fauce corollae aequaliter affixa apice altitudine basim sinus corollae attingentibus; filamentis aequalibus lineari- bus antheris duplo longioribus; antheris ellipticis vel oblongis apice mu- cronulatis infra medium affixis, thecis infra medium liberis et laeviter divergentibus. Pollinia globosa secus aequatorem poris saepe 8 donata. Stylus filiformis altitudinem staminum attingentibus in statu fructifero nuculis longioribus, stigmate capitato haud lobato abundanter papillato. Nuculae saepe 4 cinereae rosaceaeve sparse sed distincte verrucosae ovato- lanceolatae dorse latereque rotundae ventre obtusae carinatae (carina margines sulci longitudinali clausi imbricatas sed haud confluentes com- posita) basi angustatae cicatrice obliqua parva transverse elongata donatae. Gynobasis plana vel laeviter concava.— Planta annua herbacea erecta strigosa. Folia parva alterna. Cymae multibracteatae terminalis scor- poideae, maturitate elongatae secundae inconspicue bracteatae. Corolla minuta fauce et limbo caerulea, tubo flavo.— Nomen, derivatum a Maire et 7s, datum est in honorem cl. René Maire (1878-1939), optime de flora Africae septentrionalis praesertim Moroccae meritae. Mairetis microsperma (Boiss.), comb. nov. Lithospermum microspermum Boiss. Diag. ser. 2, 3: 135 (1856) and FI. Orient. 4: 218 (1875). 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 5 Lithospermum Webbii Coss. & Dur. Bull. Soc. Bot. France 22: 64 (1875). nomen; Ball, Jour. Linn. Soc. Bot. 16: 574 (1878), nomen in syn. A monotypic genus known only from Morocco and the Canary Islands, The plant is especially notable for its elongate tubular fruiting calyces. At anthesis the calyx is tubular for about a third of its length. At maturity it is greatly accrescent, several times longer than broad and tubular for a half to two-thirds its length. The nutlets, hence, are matured at the bot- tom of a deep investing calyx-tube. Among some anchusoid borages calyx- tubes are well developed. It is, however, a condition rare among other herbaceous borages and is entirely unknown in Lithospermum or any of the closely related genera. The stigma is also unusual. It is capitate and terminal and unlike the stigmas usually developed in the Lithospermeae is not only simple but shows no suggestion of lobing. Although very distinct, Mairetis seems to have its closest relationship with Neatostema and Moltkiopsis. As to annual habit and roughened nut- lets it is most like Neatostema. Its corolla, because of its color, elongate form, long filaments, and elevated stamen-attachment, is suggestive of that of Moltkiopsis, although very different in size. The tiny corolla, scarcely longer than the calyx, is completely glabrous except for hairs low inside the tube on the site of the undeveloped nectary. The stamens are affixed at equal height low in the small corolla-throat and have filaments of equal length. The position of the stamen-attachments is usually marked by small depressions on the outer surface of the corolla. Neatostema, gen. nov. BORAGINACEAE — LITHOSPERMEAE. Lithospermum § Oxyspermum Visiani, Fl. Dalmatica 2: 246 (1847). Calyx 5-fidus, maturitate accrescens, lobis calycis fructiferi conniventi- bus conspicue costatis. Corolla anguste infundibuliformis symmetrica extus villulosa, lobis adscendentibus vel patentibus ovato-oblongis apice rotundis utrinque minutissime abundanterque glanduloso-puberulentis, fauce nullo modo appendiculato vel glandulifero pilis in annulo circumferentialiter angusteque dispositis vel in locis 5 congestis praedito, tubo intus villuloso supra basim nectarium 10-lobatum villulosum gerenti. Stamina 5 basim versus tubi corollae paulo supra nectarium aequaliter affixa; filamentis perbrevibus antheris 4-plo brevioribus, antheris parvis oblongis utroque obtusis rotundisve vel rare basi retusis infra medium affixis. Pollinia ellip- soidea medio poris 8 donata. Stylus brevissimus antheras haud superans maturitate nuculis duplo brevior, stigmate terminali bilobato lobis sub- globosis. Nuculae saepe 4 brunneae angulatae evidenter verrucosae paulum curvatae utroque angustatae basi cicatrice parva obliqua donatae dorse convexae saepe verruculis longitudinaliter 4-serratis ornatae ventre ob- tusae. Carina ventralis nuculae margines sulci longitudinali clausi im- bricatas sed haud confluentes composita. Gynobasis fere plana perdepresse pyramidali.— Herba annua hispida erecta humilis. Folia parva alterna. Flores parvi flavi in cymas terminales densas foliaceo-bracteatas mox 6 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv elongatas scorpioideas dispositi.— Nomen derivatur a véaros, infimus, et o7jpa, Stamen, quod stamina basim versus corollae affixa sunt. Neatostema apulum (L.), comb. nov. Myosotis apula L. Sp. Pl. 1: 131 (1753 Lithospermum apulum (L.) Vahl, anes Bot. 2: 33 (1791); Stroh, Beih. Bot. Centralb. 58 °: 204 (1938). Rhytispermum apulum (L.) Reichenb. Icon. Fl, Germ. 18: 67, t. 1313, f. 8-14 Lithospermum inte f. cleistogamum Murbeck [Contr. Fl. Moroc. 2:] Lunds . Arsskrif. n. f. Avd. 2, 19(1): 23 (1923). A well-known plant widely distributed about the Mediterranean and eastward to Mesopotamia. It also occurs on the Canary Islands. Its corollas are distinctive. The stamens, on very abbreviated filaments, are borne just above the basal nectary in the tube and hence extremely low in the corolla. The corolla-throat has no intruding appendages but does have five small densely villous spots, or the latter more or less united to form a narrow villose band. The corolla-lobes on both surfaces are micro- scopically glanduliferous and velvety-puberulent. The tube is hairy on the inner as well as outer surface. The style is extremely short and at maturity is greatly overtopped by the nutlets. The nectary is very well developed and consists of ten projecting somewhat villulose closely juxta- posed quadrate lobes. The nutlets are coarsely roughened and angulate. One of their distinc- tive features is the development of two adjacent paralleling lines of warts down the middle of the dorsum. The body of the nutlet is somewhat bent and the attachment appears to be oblique and suprabasal. Dissection shows, however, that inside the nutlet the tip of the cotyledons is directly above the attachment- surface. Whatever the appearance, the attachment is morphologically basal. As in Mairetis, the gynobase of Neatostema is flat or slightly higher in the middle and heres very depressed pyramidal. At extreme maturity, as a result of shrinkage of tissue, the gyncbase may actually become lowest at the center. Another rather unusual feature of Neatostema is the frequent develop- ment of cleistogamic flowers. Such flowers have tiny obconic corollas which eventually fall off unopened. Their calyx and fruit, as well as the cymes in which they are borne, are normal in appearance. Cleistogamy is known among a few American species of Lithospermum but has not been reported in the Old World species of the genus nor in related genera. The expanded pollen grains of this genus are not unlike those of various species of Lithospermum and related genera, but the behavior of the dry grains as they expand in lactic acid is very different. In lactic acid the grains first become cylindric and because of their low-convex polar ends are almost oblong in lateral profile. Their most distinctive feature at this stage, however, is a localized strong deep narrow constriction halfway between the poles. In this well-marked deep equatorial groove the pores appear as eight sac-like depressions. These features all disappear as the 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 7 grains swell to full size. The expanded grain is ellipsoidal, broadest at the equator, and in lateral profile with the sides slightly angulate. B. SUPPLEMENTARY NOTES ON LITHOSPERMUM Durrnc the past summer I had the opportunity of visiting Kew, London, and Paris, and examining the representation of Lithospermum in the large herbaria there. Through the courtesy of the curators, critical specimens of the genus which I selected were subsequently loaned me for close study at the Arnold Arboretum in conjunction with other critical specimens kindly sent at my request from Vienna. While preparing my recent paper on Lithospermum, Journal Arnold Arboretum 33: 299-366 (1952), I had no material available of two Afghan species, Arnebia inconspicua Hemsl. & Lace and A. speciosa Aitch. & Hemsl., nor was I aware that Dr. Rechinger had just published several closely related species also from Afghanistan. Material from the sources previously mentioned now makes it possible to describe and discuss these plants. Having seen more speci- mens, it is also possible to modify and extend my previous descriptions of a few other Asiatic species. Unhappily, a change of name for one of the most widely known species in the genus must also be recorded. Lithospermum cyrousianum (Parsa), comb. nov. Arnebia cyrousiana aes Kew Bull. a A (1948). Arnebia echioides (L.) DC. Prodr. 10: 46). Lithospermum Tournefortii Johnston, Pag rea Arb. 33: 336 (1952). The type of Arnebia cyrousiana, in the Kew Herbarium, has been studied. Its corollas have been dissected and its pollen examined. The specimen has all the characters of the plant well known to gardeners as Arnebia echioides, and at most is only a minor phase of the same species. The single plant mounted on the type-sheet has long-styled flowers, but both short- and long-styled corollas are represented in the pocket asso- ciated with it. Of all the synonyms of Arnebia echioides only Parsa’s binomial has a specific epithet that can be used when the species is trans- ferred to the genus Lithospermum. The binomial Lithospermum cyrousia- num must replace Lithospermum Tournefortii, the name recently proposed by me when I was unaware of the identity of Parsa’s species. Lithospermum fimbriopetalum (Stocks) Johnston, Jour. Arnold Arb. 832/320 Ci9a7). Arnebia a Stocks in Hooker, Jour. Bot. & Kew Miscl. 3: 180, t. 6 (1851). eo Bungei Boiss. Fl. Orient 4: 215 (1875). Lithospermum Bungei (Boiss.) Johnston, Jour. Arnold Arb, 33: 326 (1952). Annual herb 5-15 (—40) cm. tall, sparingly branched. Leaves thickish, sparsely hairy or glabrous on the lower surface. Flowers heterostylic. Corolla yellow, 25-30 mm. long; limb 10-20 mm. broad, the lobes broad, 8 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv rounded, margins distinctly lobulate or lacerate; throat bearing no glands nor appendages; tube slender, 20-24 mm. long, surpassing the calyx 5-8 mm., above the base inside bearing a membranous more or less hairy collar-like nectary 0.5-1 mm. high. Style slender, forked at apex, lobes 0.5 mm. long. Stigmas 2, compressed, as broad as long, obscurely bilobed. Long-styled flowers with corolla-tube cylindric, abruptly enlarging in diam- eter 6-8 mm. below the summit; anthers 2—2.3 mm. long, borne low in the upper third of the corolla-tube; pollen moderately constricted at the middle, 42-50 28-33 y; style almost as long as the corolla-tube. Short- styled flowers with the tube gradually ampliate; anthers 2—2.3 mm. long, borne at summit of tube with their upper third exserted into the throat: pollen with sides straight and parallel or slightly concave, 46-56 33-41 yw; style two thirds as long as the corolla-tube. Calyx 15-17 mm. long at anthesis (with linear lobes), becoming very strongly accrescent (with lanceolate or ligulate lobes) in fruit, at maturity 25-30 mm. long with the major lobes 3-5 mm. broad near the base, glabrescent with the midrib and a few veins becoming evident, at times with papillate excres- cences on the midrib of the lobes near the base. Nutlets olivaceous or rubiginous, tuberculate, 3-3.5 mm. long, 3 mm. broad at the base, beaked above the middle, lower half of dorsum swollen on either side of a broad medial depression, upper half with a low rounded medial keel, venter angulate, attachment-surface broadly triangular, basal. Gynobase very depressed. BALUCHISTAN: Upper erne ee Stocks 977 (Kew, Type of A. fimbriopetala) ; Gival, 5-6000 ft.. A.V. Monro ex Duthie (Kew). PERSIA: 20 mi. from Isphan on Shiraz road, 6000 ft.. open hills, Apr. 1944, A. C. Trott 906 (Kew); pr. Sser-tschah, March 12, 1859, A. Bunge (G, tsoTYPE of A. Bunget). A careful study and comparison of the above cited collections has shown conclusively that Arnebia fimbriopetala and A. Bungei are conspecific. The species is most closely related to Lithospermum detonsum of northern Persia and Transcaucasia, agreeing with that more northern and western plant in habit of growth, in large, much accrescent calyces, and in size, form, and markings of nutlets. It differs sharply from L. detonsum in sev- eral respects. Unlike that species its flowers are decidedly heterostylic. Furthermore, its corollas are not merely larger but also have a more elon- gate tube, conspicuously longer than the calyx, and lobes that are not entire but evidently lobulate or lacerate-fimbriate on the margin. pate grea Szechenyi (Kanitz) Johnston, Jour. Arnold Arb, 33: 330 (1952). My previous description of this species was based on a collection with short-styled flowers from Hsuin Hwa, Kansu (R. C. Ching 731), supple- mented by the original description (of a long-styled plant) given by Kanitz. Additional information concerning this species of western China 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV y is now available through the study of three collections made by Father Licent. These were obtained at Ta-la-chi, Kansu (ca. 150 km. n. e. of Lanchow), no. 4017, and at Sain-Nor, at the base of the Scharanarin-ula {Khara-narin ula] northwest of the Ordos in southern Inner Mongolia, almost 600 km. north-northeast of Lanchow, nos. 13612 and 13621. The first collection, at the British Museum, contains both long- and short- styled flowers. Of the two other collections, both at Paris, Licent 13612 is a short-styled plant with fruit; Licent 13621 is the long-styled plant. These collections of Licent are less robust plants than those previously described, having noticeably more slender stems and narrower leaves. Their stems, 2.5 mm. thick at the base, are 2-4 dm. long and may become de- cumbent. The largest middle cauline leaves are 2—2.5 cm. long and only 5 mm. broad. The bracts in the cymes are elongate, foliaceous, and broader than the calyx-lobes. The corollas are 16-19 mm. long and retain evi- dences of having been yellow and provided with evanescent dark spots. The corolla-tube is 11-14 mm. long. There are no glands on the inner surface of the throat as reported previously. The style is not forked at the apex. Its two stigmas are terminal and juxtaposed. In the short-styled flowers the corolla-tube is gradually ampliate, the anthers (3 mm. long) are borne at the summit of the tube and partially exserted into the throat. the pollen (39-43 X 25-26) is barrel-shaped, having rounded ends and sides that are parallel or only very slightly concave; the style reaches up to the middle of the corolla-tube. In long-styled flowers the cylindric corolla-tube has a slight but distinct increase in diameter just above the middle. the anthers (1.3-2 mm. long) are borne at the middle of the corolla-tube, the pollen (30-33 X 16-20u) is evidently constricted at the middle, and the style reaches up into the corolla-throat. The mature nut- lets, 2.5 mm. long and 1.5 mm. broad, are weakly constricted just above the base and are abundantly tuberculate. Some of the wart-like roughen- ings are rounded but others bear a minute point or short stout hair and can almost be described as muricate. The attachment surface of the nutlet is broad, flat, and obliquely basal. The venter of the nutlets is angulate. The dorsum is rounded and obscurely keeled only towards the apex. Lithospermum Hancockianum Oliver in Hooker, Icones 25: t. 2457 (1895); Johnston, Jour. Arnold Arb. 33: 354 (1952). Among some duplicates recently received from Paris are six hitherto unstudied collections of this remarkable species from eastern Yunnan. They provide more numerous and better corollas for dissection than were previously available to me. I am, accordingly, able to correct and emend my previous description of the plant. The species appears to have weakly dimorphic flowers. The ten collections at hand sort into two types. One has the style evidently exserted from the corolla-tube, usually 3-5 mm., and the other has the style equaling the tube in length or 1-2 mm. shorter. In both types the anthers (1.5-1.7 mm. long) are borne at the summit of the corolla-tube. Those in the long-styled flowers tend to have their tips reaching only to the summit of the corolla-tube, while those in short-styled 10 JOURNAL OF THE ARNOLD ARBORETUM _ [vol. xxxiv flowers tend to have the upper third of their length projecting above it. All the flowers on a given plant appear to belong to one or the other of these two types. There is no accompanying difference in pollen. Minute stipitate glands are abundant on the corolla-tube behind the anthers, in a band extending 1 mm. below to 1 mm. above the filament-attachments. They occur sparingly in the tube as far down as 5 mm. below its summit. The filaments are broadly affixed, attenuate, and laterally compressed and accordingly unguiculate. The style is perceptibly forked at the extreme apex, and its two oblong stigmas may become divergent. The glabrous nectary in the tube is not continuous but made up of ten projecting, closely juxtaposed quadrate lobes. The nutlets are white, smooth, sparingly pitted, stout pointed ovoid, 2.5—3 mm. long and 2 mm. thick. They have a lineate constriction just above the broad obliquely basal attachment-surface. The dorsum is convex and is keeled only near the apex. The venter is angulate and evidently keeled. The gynobase is depressed pyramidal. Lithospermum Lindbergianum (Rech. f.), comb. nov. Macrotomia Lindbergiana Rech. f. Ann, Naturhist. Mus. Wien 58: 58 (1951). Plant perennial, with a stout taproot crowned by a small caudex: branches of caudex short, stout, clothed with persistent leaf-bases. Indu- mentum cinereous, composed of numerous very slender, usually spreading bristles 2-5 mm. long and an abundance of very minute, frequently re- trorse hairs 0.2—0.5 mm. long. Flowering stems 4—6 cm. long, erect, slender, simple, 1-2 mm. thick at the base, arising from the center of a cluster of functional leaves, terminated by the inflorescence. Leaves with an evident midrib, veinless; basal ones linear-oblanceolate, becoming 4 cm. long, 2-3 mm. broad below the acute apex; cauline leaves few, attenuate, gradually reduced up the stem, uppermost 2-3 cm. long, 1.5—-2 mm. broad near the middle. Inflorescence a dense capitate terminal cluster ca, 4 cm. in diam- eter, comose from an abundance of slender protruding sepals; bracts shorter than the calyx. Calyx at anthesis 23-24 mm. long, hispid; lobes linear- attenuate and more or less flexuous, 21-22 mm. long, 0.8 mm. broad at the base, surpassing the corolla 2-4 mm. Calyx in fruiting state 30 mm. long, inside below the middle densely appressed white villose. Flower heterostyled with only the short-styled form known. Corolla 19-21 mm. long, slender and elongate, abundantly pubescent outside, inside glabrous except on the nec- tary; tube 16-17 mm. long, ca. 1.5 mm. thick for most of its length, upper 3 mm. of length swollen (and staminiferous) and 2—2.5 mm. thick, slightly constricted at the very summit, inside just above the base bearing a distinct membranous collar-like nectary 0.3—0.4 mm. high which is sparsely vil- lulose on the inner face; throat funnelform, short, 2-3 mm. long, spar- ingly glanduliferous inside; limb ca. 4 mm. broad, composed of ascending triangular-ovate lobes ca. 1.7 mm. wide and 2 mm. long. Anthers 2.5 mm. long, borne just below the summit of the corolla-tube. Pollen 33-35 25-26 yp, in lateral profile with the sides parallel or practically so. Style 11-12 mm. long, reaching up to slightly above the middle of the corolla- 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 11 tube, forked at the summit, the lobes 0.8—1 mm. long; stigmas 2, unequal, spathulate. Nutlets (only sub-mature ones seen) erect, lanceolate, 3.5 mm. long, 2.5 mm. broad below the middle, surface dull, densely and very minutely papillate, venter with prominent keel, dorsum with an obscure partial keel and probably irregularly tuberculate when completely mature, attachment broad and obliquely basal. Gynobase depressed pyramidal. AFGHANISTAN: Bamian, Ajdaha, May 24, 1947, K. Lindberg s.n. (TYPE, ienna). This very distinct species is known only from a single collection ob- tained in the mountains about 100 km. northwest of Kabul, Afghanistan. Judging from the totality of its characters the plant is evidently a mem- ber of the group of species containing L. Benthami, L. superbum, L. nobile and L. euchromon. It comes from the same floristic area and has very similar elongate corollas with purpurescent limb. It differs from all these relatives in the small size of the plant and in having a well-developed collar-like nectary low in the corolla-tube. The nectary is especially note- worthy. In the related species mentioned the corolla-tube bears no nectary nor any rudiment of one such as tufts of hairs nor discolored or slightly swollen tissue. Those authors who have assigned such species to the segregate genus Macrotomia have usually laid great stress on the absence of a nectary. Though the present species in all other technical characters, as well as in general appearance, gives every evidence of close affinity with the species mentioned, it would be divorced from them in any classifica- tion in which the presence or absence of a nectary is made arbitrarily into a crucial character. We have here, in fact, further evidence that Macro- tomia is neither a natural nor useful segregate; cf. Jour. Arnold Arb. 33: 313: €1952). The present species agrees with its obvious relatives not only in the form of the corolla but also in the type of coloration. The corollas found on herbarium specimens of these species show no evidence of decided orange or yellow coloration so common in other groups of the genus. In drying they tend to become brownish, pinkish, or whitish, but in most inflores- cences some are to be found having purple or purplish lobes and throat. The indications are that the corollas of this group of species have striking color-changes associated with the state of maturity. A collector of L. nobile reports the corollas as at first dark brown but later yellow. Aitchison, Jour. Linn. Soc. 18: 19 (1880), states that L. euchromon is ‘“‘remarkable for its flowers varying from greenish yellow to deep purple-black.” Al- though collectors of L. inconspicuum report the corollas as creamy or pale yellow, some of the corollas on their dried specimens have decidedly purple or purplish throat and limb. The five species most closely related to L. Lindbergianum are distin- guished from it in the following key: Inflorescence globose or broader than long; leaves with midrib only. Corolla with a well-developed collar-like nectary in the tube; stems less than 12 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxtv 1 dm. tall. arising from the center of a leaf-cluster; leaves narrow, 2-3 mm. broad; calyx linear-attenuate as much as 0.8 mm. wide at base ‘ite des Lithospermum Lindbergianum. Corolla without a nectary. in the tube. Plant gray or silvery, the herbage not glanduliferous; calyx-lobes subulate- linear, 1 mm. wide at base; corolla with stipitate glands in the throat: stems less than 1.5 dm. tall. arising from the center of a basal leaf- cluster; leaves less than 5 mm. broad ... Lithospermum inconspicuum. Plant green or somewhat tawny, the herbage usually glanduliferous; calyx-lobes lanceolate, 1-3 mm. broad; corolla without glands in throat; stems to 4 dm. tall, arising lateral to the basal ae of leaves; leaves -10 mm. broad. Lithospermum suche tenon. Inflorescence A fires a cylindrical thyrse e evidently ee longer than broad; lower leaves with 3—5 evident longitudinal ribs. Thyrse continuous, with numerous exserted bracts; pred lobes very slender exuous. thospermum Bentham. ties feud only the lowermost bracts protrudent ; calyx-lobes stiff. Basal leaves hispid; corolla hairy outside. not glanduliferous ....... NouG bE Petes Ea.9 da heated as Lithospermum speciosum. Basal leaves strigose: corolla glanduliferous but usually otherwise glabrous Lithospermum nobile. Lithospermum inconspicuum (Hemsl. & Lace), comb. nov. Arnebia inconspicua Hemsl. & Lace, Jour. Linn. Soc. 28: 326 (1891). Arnebia argyrea Rech. f. Ann. Naturhist. Mus. Wien 58: 59 (1951). Plant with herbage and stems pale and somewhat silvery from a dense mostly appressed indument composed of abundant pallid hairs of two sizes, coarse hairs 1-2 mm. long and minute very slender hairs 0.2-0.5 mm long; perennial with a rather dense caudex formed of crowded branches ensheathed by old leaf-bases. Leaves firm, oblanceolate, 1-5 cm. long, 2-5 mm. broad, with evident midrib but no veins. Stems arising from the center of a leaf-rosette, usually 3-15 cm. long, 1-3 mm. thick at the base, simple, terminated by the capitate inflorescence, rarely much reduced in length and bearing the inflorescence in the center of the leaf-rosette. Middle and upper stem-leaves not numerous, not conspicuously smaller than the basal leaves. Cymes congested into a dense terminal capitate cluster 1.5-3.5 cm. in diameter at anthesis. Inflorescence lacking con- spicuous bracts but frequently with some of the uppermost stem-leaves crowded at its base, rarely developing a small secondary cluster of cymes in leaf-axils below the major terminal cluster. Calyx 12-17 mm. long; lobes subulate-linear, 10-16 mm. long, ca. 1 mm. broad at the base, reaching up to the throat of the corolla. Flowers heterostylic. Corolla narrow and elongate, 12-19 mm. long, surpassing the calyx 2-3 mm., whitish or yellow, frequently somewhat purpurescent in drying, outside evidently appressed hairy, inside glabrous or with scattered hairs on the corolla-lobes; limb ascending, 2.5-4.5 mm. broad; tube-nectary not de- veloped; throat short, ca. 2 mm. long, funnelform; lobes ascending, acute, ovate-triangular, 1-2 mm. long; throat somewhat glanduliferous, at least 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 13 in the short-styled flowers; tube 9-14 mm. long, in long-styled flowers cylindric 12—-14.5 mm. long and 2 mm. thick, in short-styled flowers 1— 1.5 mm. thick below the middle and above very gradually broadening into the throat. Anthers 2 mm. long, borne on very short filaments either above the middle of the corolla-tube or in the corolla-throat. Pollen of short-styled flowers 40-43 X 24-28 y, in lateral profile with sides straight and parallel or at most only slightly concave. Style evidently forked (branches slightly unequal), reaching to slightly above middle of corolla- tube or into the corolla-throat, short styles ca. 7 mm. long with lobes 1 mm. long, long styles ca. 15 mm. long with lobes 2-3 mm. long. Stigmas 2, distinct, compressed, nearly as broad as long. Fruiting state of plant and its fruit unknown. BALUCHISTAN: Zahru, Stocks 866 in pt. (Kew); without locality, 1891-94. C. F. Elliott (Kew). AFGHANISTAN: Kabul, Gul Tara, fl. creamy white to yellow, Neubauer 364 (Vienna); Cabul, H. Collett 18 (Kew); Nozi. fl. Rare yellow, 10000 ft., W. Koelz 12015 (Vienna, TYPE of A. argyrea); Obeh, 16 M. Koie 4403 (Vienna): Tscharikar, Top Tara, Neubauer 610 (Vienna) ; ree fl. pale yellow, 7000 ft.. Koelz 11485 (Vienna). A very distinct and readily recognizable species notable for its rather silvery indument and dense capitate inflorescence. Evidently a member of the same group of species as L. Benthami and L. euchromon, although not obviously related to any particular member of it. In technical characters it departs from all its allies in having stiputate glands on the inner surfaces of the corolla-throat. Lithospermum speciosum (Aitch. & Hemsl.), comb. nov. Arnebia speciosa Aitch. & Hemsl. Proc. Linn. Soc. 18: 81 (1880) and 19: 179. t. 24 (1882). Plant perennial, coarse, very bristly. Stems simple, erect, becoming 4 dm. (6 dm. fide Aitch. & Hemsl.) tall and 8 mm. thick at the base, arising from the center of the basal leaf-cluster, hispid, its bristles 2-4 mm. long and towards the base of the stem retrorsely deflexed but elsewhere widely spreading. Leaves firm, those below the middle of the stem with three strong longitudinal ribs, hispid, the hairs 1-2 (—3) mm. long, slender, stiff, spreading to ascending and sometimes retrorse, those on the upper leaf-surface arising from minute discoid bases. Basal leaves to 10 cm. (15-25 cm. fide Aitch. & Hemsl.) long, narrowed above the broad sheathing dye-stained base, broadest (7-10 mm.) above the middle and then gradually narrowed to the acute apex. Cauline leaves gradually reduced up the stem, the middle ones lanceolate, 5—7 cm. long, 5-8 mm. broad near the base. In- florescence an elongate interrupted cylindrical thyrse, 15-25 cm. long, 5—7 cm. thick, composed of a small terminal cyme and 10-15 slightly larger short- neduncillate cymes borne in the leaf-axils along the upper half of the stem. Cymes densely flowered, at first glomerate and 3—5 cm. broad. but 14 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv in age tending to elongate and become evidently scorpioid (and eventually perhaps 4—7 cm. long); bracts shorter than the adjacent calyces; subtend- ing leaves surpassing only the lower cymes. Calyx hispid, 17-18 mm. long at anthesis, 21 mm. long in fruit, if bearing stipitate glands these very few and inconspicuous; lobes ca. 15 mm. long, 1.5—2 mm. broad near the base. reaching up to the tip of the corolla-lobes or surpassing them 1-2 mm. Flowers heterostylic, but only the short-styled form seen. Corolla 16-17 mm. long (20-24 mm. fide Aitch. & Hemsl.), elongate, outside evidently hairy, especially on the lobes, bearing few if any stipitate glands, inside glabrous and devoid of basal nectary; tube 13 mm. long, 1.2 mm. thick at the middle, above gradually ampliate and becoming 3 mm. thick at the summit; throat short and broad; limb about 7 mm. broad, lobes ascending, 2.5 mm. broad and 2 mm. long, apex rounded. Anthers 2 mm. long, borne at the summit of the tube and partly exserted into the throat. Pollen (poor condition) 39-42 x 23-28 y, sides evidently concave, pores 6 at each end. Style 5 mm. long, reaching up to the middle of the corolla-tube, obscurely forked at the apex, lobes ca. 0.3 mm. long. Stigmas 2, compressed, broader than long. Nutlets gray or brown, erect, 4-5 mm. long, attached obliquely by the broad base to a low-pyramidal gynobase, with a narrow very prom- inent ventral keel, apex truncate and sometimes toothed, back 2—3 mm. broad below the middle, then narrowed to the laterally compressed apex, usually with evident longitudinal ridges paralleling the dorsal keel, surface minutely tuberculate and irregularly verrucose. PAKISTAN: hill behind Kaiwas, Kurrum Valley, 9-12000 ft., open grassy spots, July 3, 1879, Aitchison 720 (TYPE, Kew). I know this species only from the type specimen which is described above. The plant is related to L. Benthami of the Himalayas of north- western India and Kashmir and to L. nobile of the high transverse ranges of eastern Afghanistan. It seems to be more like L. Benthami, with which it agrees in nutlets and pollen. It differs, however, in being clothed in stiffer, more spreading hairs and in having broader, stiffer, non-flexuous calyx- lobes and a not continuous but distinctly interrupted thyrse with not numerous but few if any salient bracts. From L. nobdile it differs in the form of nutlets, smaller pollen, lack of evident stipitate glands, hairy outer corolla-surfaces, hispid rather than neatly strigose basal leaves, and per- haps more rounded tips of the corolla-lobes. The color of the corolla is not reported by the collector. As with specimens of L. nobile, however, some of the non-faded younger corollas on the type-specimen have distinctly purple lobes and throat. Lithospermum nobile (Rech. f.), comb. nov. Arnebia nobilis Rech. f. Ann. Naturhist. Mus. Wien 58: 58 (1951). Plant coarse, perennial. Stems erect, 2-5 dm. tall, as much as 8 mm. thick at the base, arising from the center of a basal cluster of leaves, below the middle usually antrorsely strigose, above the middle usually with some spreading hairs and commonly glanduliferous. Basal leaves firm, 8-23 cm. 1953] JOHNSTON, STUDIES IN THE BORAGINACEAE, XXIV 15 long, with 3-5 longitudinal ribs, narrowed just above the broad sheathing dye-stained base, broadest (3-22 mm.) above the middle, apex slenderly acute, indument of short strongly appressed pallid hairs, hence neatly strigose. Cauline leaves gradually smaller up the stem, middle ones broadest at the base, 2-8 mm. wide, attenuate, usually clothed with ap- pressed hairs and frequently ciliate on the margins. Inflorescence an elongate interrupted thyrse, cylindric, 5-12 cm. broad, at least several times as long as broad, composed of a small terminal cyme and numerous short-pedunculate lateral cymes arising from the leaf-axils along the upper half of the main stem and occasionally even down to near its base. Cymes densely flowered, at first glomerate and 2-3 cm. in diameter, even- tually elongating, becoming scorpioid and 5—15 cm. long in fruit, hispid and glanduliferous; bracts shorter than the adjacent calyces; upper stem-leaves soon surpassed by the subtended cymes, only the lower ones protruding from the thyrse. Calyx hispid, frequently somewhat tawny, bearing scattered minute stipitate glands, 15-18 (-20) mm. long at anthesis, be- coming as much as 25 mm. long in fruit; lobes 1-2 mm. broad, acute, in long-styled flowers 1-6 mm. shorter than the corolla, in short-styled flowers equaling the corolla or surpassing it 1-2 mm. Flowers heterostylic. Corolla elongate, 16-20 (—27) mm. long, outer surface usually bearing scattered stipitate glands but otherwise usually glabrous, inside glabrous, without basal nectary in the tube; corolla-tube 12-15 (—20) mm. long, 1-1.5 (-2) mm. thick below the middle, in short-styled flowers gradually ampliate above the middle and becoming 2.5-3 mm. thick at the summit, in long-styled flowers with upper 5 mm. cylindric and 2.5-3 mm. thick; throat short, abruptly enlarged and rounded, inside without gland or appendages; limb 5-7 (-10) mm. broad, lobes usually more or less triangular and commonly 2.5 mm. broad and 2—2.5 mm. long, ascending, the margins and tip of the lobes somewhat recurved. Anthers 2.3-2.5 (—3) mm. long; in short-styled flowers borne high in the tube (with their bases ca. 1 mm. below its summit) and protruding up into the corolla- throat; in long-styled flowers borne low in the upper cylindric section of the corolla-tube, their bases 5 mm. below the summit of the tube. Pollen with 6 pores at each end of the grain, in short-styled flowers subcylindric with rounded ends, in lateral profile with sides parallel or only very slightly concave, 49 & 33-35 ; in long-styled flowers evidently constricted at the middle, 33-38 X 18-23 p. Style reaching up to slightly above the middle of the corolla-tube or to the summit of the tube or even into the base of the corolla-throat, forked at the apex, lobes unequal, 0.5—1 (—1.5) mm. long; stigmas 2, compressed, broader than long. Nutlets gray, erect, usually only one developing, ca. 7 mm. long, broadest (2.5—3 mm.) at or just above the base, gradually narrowed upwards into a somewhat rostrate incurving apex, hence narrowly subconic in form, ventral and dorsal keel obscure except on the upper 2-3 mm. (i. e., on the beaked apex), surface very minutely tuberculate, longitudinally striate and rugose, the rugae irregular and sparsely verrucose; attachment-surface broad, obliquely basal. Gyno- base low pyramidal. 16 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxrv AFGHANISTAN: Sanjadabad, May 1880, H. Collett 22 (Kew); Sanjadabad, 7000 ft., April 1880, H. Collett 77 (Kew); Dscheratu, May 7, 1949, H. F. Neubauer 836 (Vienna, Type of A. nobilis); Paghman, rocky slope, 7000 ft., fl. at first dark brown, later yellow, leaves silvery, July 14, 1935, W. R. Hay 350 (Kew); Sar-i-chasma, 2700 m., June 3, 1948, L. Edelberg 1832 (Vienna): Hauz-i-Mahik, 2500 m., July 12, 1948, M. Koie 3166 (Vienna); Farakulum. 3000 m., July 19, 1948, M. Kdie 3167 (Vienna). A species of the high mountains (Hindu Kush) of eastern Afghanistan and most closely related to the more southerly ranging L. speciosum of the Pakistan-Afghanistan border. Distinctive of the species are its tidy and decidedly strigose basal leaves, the numerous stipitate glands on the inflorescence, calyx, and outer corolla-surface, the large pollen, and the narrowly conic longitudinally striate nutlets with incurved beak. Among the collections cited only one (Hay 350) has hairs on the outer surface of the corolla. Only one fruiting collection has been seen (Edelberg 1832). The elongate erect nutlets, broadest at or near the base and gradually narrowed upwardly into a slightly incurved beaked apex, have a form unique in Lithospermum. The roughening of their surface, numerous narrow longitudinal usually verrucose ridges which are separated by lineate grooves, is also unique. Another unusual feature is the suppression of the ventral keel except on the beak of the nutlet. ARNOLD ARBORETUM, HARVARD UNIVERSITY. 1953] LI, LIBOCEDRUS AND CUPRESSACEAE 17 A RECLASSIFICATION OF LIBOCEDRUS AND CUPRESSACEAE Hutr-Lin Lr! With two plates THE GENus Libocedrus, of the Cupressaceae, is well known as having a remarkably disjunct range. As currently interpreted, the species are widely scattered in lands bordering the Pacific Ocean: one in southern Chile, two in New Zealand, two in New Caledonia, three in New Guinea, one in southern China, ene in Formosa, and one in Pacific North America. Such a generic range is indeed unique among the conifers, as it covers more or less equal areas in the Northern and Southern Hemispheres. All other genera are confined either to the northern or southern lands or have but occasional outlying species extending beyond the equator. The northern and southern species were combined together because of the elongate, basically attached scales and the bi-winged seeds with the wings very unequally developed. These characters, as will be noted below, actually differ to a certain extent between the northern and southern species. It is possible that the resemblance is superficial and due to parallel variation. Two authors, Koch (19) and Kurz (20), noted some differences between the northern and southern species and established independently in the same year 1873 two genera, Heyderia and Calocedrus respectively, for two of the northern species, but their work was soon disregarded by most subsequent authors. In 1926, Pilger (28), noting the discrepancy between the northern and southern species, divided the genus into two subgenera — Heyderia, containing the northern species, and Eulibocedrus, the southern. However, he did not emphasize the significance of their fundamental differences, which, in my opinion, are of more value than subgeneric differentiation would indicate. To clarify the taxonomy of the group of species currently included in Libocedrus, it will be helpful to trace briefly the bibliographical history of the genus. The first species of this assemblage were discovered in Chile and named in 1824 as Thuja chilensis D. Don and Juniperus uvifera D. Don (in Lambert, 21). Another was collected in New Zealand and first named Dacrydium (?) plumosum by D. Don in 1828 (In Lambert, 21, ed. 2). In 1842, W. J. Hooker (12) transferred Dacrydium (?) plumosum to the genus Thuja, renaming it Thuja doniana. In 1843, Hooker (13) described ‘This study was initiated several years ago at the National Taiwan University, Formosa, China, and was completed in the Department of Botany, Smithsonian Institution, to the authorities of which the author is indebted for their generosity in placing facilities at his disposal. Thanks are due to the curator of the herbarium of acknowledgment is 5 made for their kindness in reading the manuscript and offering valuable suggestions. 18 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv independently a new species from Chile which he called Thuja chilensis: this is regarded as the same species as that described previously as chilensis D. Don, which is based on a different type; the use of the same specific epithet is a coincidence, Hooker being unaware of Don’s species. In 1844, Hooker (14) described a Thuja tetragona from Chile, which has turned out to be the same species as Don’s Juniperus uvifera; in this in- stance, Hooker did know of Don’s species, but was not sure of its identity with his own, inasmuch as Don had described the fruit as that of a true Juniperus. These three species of the Southern Hemisphere, attributed by Hooker to Thuja, which is otherwise a genus exclusively of the Northern Hemisphere, were made the basis of the genus Libocedrus by Endlicher in 1847 (5). I am following Pilger (28) and others in choosing L. doniana (Hook.) Endlicher, now correctly known as L. plumosa (D. Don) Sargent, as the type species. The other species of Chile, Z. tetragona (Hook.) Endlicher, later known as L. uvifera (D. Don) Pilger, which differs from the other species in the 4-ranked, more or less equal leaves, and tetragonous and fastigiate branches, remained in the genus Libocedrus until Florin (7) segregated it to form the monotypic genus Pilgerodendron in 1930. In 1853, Torrey (35) described a species from California as L. decurrens, the first species from the Northern Hemisphere attributed to the genus Libocedrus. This species differed from the southern species in having three instead of two pairs of cone-scales: a smaller outer pair, a much longer fertile middle pair, and an inner sterile pair fused together into a single plate, a structure not found in the southern species. Noting these differ- ences, Koch (19) in 1873 made the Californian species the type of a genus Heyderia. In the same year that Koch proposed the genus Heyderia, Kurz (20) described a plant from Yunnan, China, as Calocedrus macrolepis, a new genus considered by him to be related to Libocedrus and Thujopsis, differ- ing from the former in the seed structure. Both Calocedrus and Heyderia were combined with Libocedrus by Bentham and Hooker (1) in 1880, although they noted that these species differ from the southern species of the genus in having the innermost pair of scales sterile and connate. Since that time these northern species have remained in the genus Libocedrus. In 1867, a second species from New Zealand was named by J. D. Hooker as L. bidwillit (15). Another species, L. austro-caledonica Brongn. & Gris (2), was discovered in New Caledonia, considerably extending the generic range. The range was further extended by discovery of species in New Guinea and Formosa. Three species have been reported to occur in New Guinea, L. papuana F. Muell. (25), L. torricellensis Schlechter (ex Lauter- bach, 23), and L. arfakensis Gibbs (11). The Formosan plant identified as L. macrolepis since 1902, has been described as a distinct species, L. formosana Florin (6). Recently two more species were discovered in New Caledonia, L. chevalieri Buchholz (3) and L. yateensis Guillaumin (38). The discovery of these different species from widely separated areas in both the Southern and Northern Hemispheres would make this, in its geographical range, a unique genus in the conifers. 1953] LI, LIBOCEDRUS AND CUPRESSACEAE 19 A careful study of herbarium material and literature shows that these northern and southern species are different in their cone-structures and other characters. In all the southern species, including those from New Guinea, there are four scales in two pairs of very different sizes. The outer pair is small and sterile, while the inner pair is about two or three times as large and is fertile, each scale bearing one or usually two seeds at the base. In the three northern species, L. decurrens, L. macrolepis, and L. formosana, there are three pairs of cone-scales. The first pair is small, sterile, and recurved at tip when mature. The second pair is very long, over three to six times the length of the outer pair and fertile, each scale bearing two seeds at the base. The third pair is as long as the second pair or slightly longer and is sterile, the scales being connate throughout their entire length into a flat thin plate. Of more significance is the fact that the disposition of these cone-scales ‘n the northern and southern species is different. The southern species have their four cone-scales meeting at the edges, or in other words, these are valvately disposed. In this respect they differ fundamentally from Thuja and other related genera of the north. In the three northern species of Libocedrus, the three pairs of scales are imbricately disposed, the outer pair overlapping the inner. This disposition is exactly the same as in Thuja, which also has the innermost pair of cone-scales connate into one piece. In Thuja and the northern species of Libocedrus, the innermost pair is about the same size as the middle fertile pair. But as it is partly covered by the latter, the exposed parts are narrow and it thus appears to be smaller. Actually it is often slightly longer. This difference in the disposition of the cone-scales indicates in my opinion that the affinities of the southern species of Libocedrus are, among the existing genera of the conifers, not with Thuja, but with Diselma, Widdringtonia, Fitzroya, and other related genera of the Southern Hemi- sphere. In all these genera, the cone-scales meet at their edges and are not imbricate. On the other hand, the three northern species are undoubtedly very close to Thuja. The cone of these species approaches closely that of Thuja, which, as noted above, has a similarly fused pair of scales in the center but with numerous outer sterile scales. Reduction in the number of these outer cone-scales would result in nearly the exact condition that we have in the cones of the northern species of Libocedrus. With a broader generic concept, these species could be included in Thuja, and this was the view actually expressed by Voss in 1908 (36). However, the leaves of these species are in apparent whorls of four instead of being strictly decussate as in Thuja. This difference in the organization of leaves on the stem as well as the fewer number of scales and unequally developed seed- wings keep these species out of Thuja. The northern species of Libocedrus also differ from the southern species in having the scales mucronulate on the back of the tip, in having the seeds less unequally 2-winged, both wings being well-developed but with one about twice as long as the other, and in the generally larger number of stamens. In the southern species, the scales are provided with a short or 20 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv long, more or less sharp spine or a bract at the back toward the tip or base or center, and the seed has a long, well-developed wing on one side and merely a narrow membranaceous margin on the other. In vegetative structures, the northern and southern species of Libocedrus are also strikingly different. In the southern species, the leaves are ver unequal and strictly decussate, alternately long and very short, the dorso- ventral being very small, and the lateral much larger, keeled and contiguous toward base. The lateral leaves only very slightly overlap the next above, and scarcely overlap laterally the dorso-ventral leaves. In the northern species, the leaves are more strongly decurrent, in apparent whorls of four, and the pairs are of about equal length, the dorso-ventral being narrow, and the laterals keeled, overlapping the next above at their bases and also the dorso-ventral ones laterally, while they themselves are not con- tiguous below. The leaves are more clearly imbricate in arrangement, reflecting the arrangement of the pistillate cone-scales. In view of the important differences of the three northern species from the southern species of Libocedrus in both vegetative and reproductive characters, it seems desirable to consider the two groups as generically dis- tinct. There are two generic names, Heyderia Koch and Calocedrus Kurz, published in the same year, available for the northern species. Calocedrus appeared in July, 1873; the exact date of publication of Heyderia is not known. As Heyderia was chosen by Pilger for the subgenus he proposed for the group, it is here adopted in preference to Calocedrus, at least until an earlier date for the latter can be definitely established. The three species from New Guinea differ from the other southern species in certain important characters in both the staminate and ovulate cones, There are four scales in the ovulate cone and these are valvate, as in the other southern species. However, these scales are only slightly woody when mature and bear a large, triangular or ovate appendage on the back near the base or below the center. In the other species usually a short or long spine, or sometimes a triangular bract, is borne either toward the tip or above the center. In the New Guinean species, these appendages are thickened and they assume a slightly shield-like appear- ance. According to Gibbs (11), in her description of L. arfakensis, as the cone increases in size, a swelling appears between the two fertile scales. This swelling gradually develops into two ovate-oblong projections which displace the apices of the scales. The outer scales are modified in the same manner but to a lesser extent. However, as to the outer scales, it appears that they are formed by two bracts coalescing together, the outer becom- ing the smaller ovate appendage and the inner the scale proper. In L. torricellensis, Schlechter (ex Lauterbach, 23) actually described these ap- pendages as bracts attached to the scales: they are found nearly at the base of the outer scales. The seed-scale complex in the Cupressaceae is shown to develop by the intimate fusion of an axillary seed-scale complex, the sterile part of the flower, which faces the cone-axis and bears ovules in its basal regions, and a bract (10). Together they form the “ovuliferous scale.” In the 1953] LI, LIBOCEDRUS AND CUPRESSACEAE 21 mature cone, the anterior side of the former is generally suppressed and fused to the bract, and the apophysis of the latter becomes the spine or bract-like appendage on the back of the ovuliferous scale. In the New Guinean species, this fusion is apparently less complete and the inner sterile part of the flower more developed than the bract. As a result the bract appears to be very distinct and of relatively larger size as compared with other Cupressaceae. A case somewhat similar to these species is found in Libocedrus bidwillii Hook. f. of Tasmania, which has an ovulate cone more or less approaching these species than other species of Libocedrus. In the staminate flowers, the New Guinean species are very distinct in having numerous scales, spirally arranged instead of decussate. The cells are also more variable in number, varying from three to six. The spiral arrangement of the stamens is a unique character. It transcends the Cupressaceae and suggests some relationship with the Taxodiaceae. In vegetative appearance, these species, with their small decussate leaves of very dissimilar alternate pairs, approach most nearly Chamaecyparis. On the basis of these distinctive characters in both the staminate and ovulate cones, the New Guinean species are here segregated as a genus distinct from ‘Libocedrus, sensu Stricto. In this connection, it is worthy to note that Peirce (27), in his study of the wood anatomy of the Cupressaceae, noted particularly the hetero- geneous nature of the genus Libocedrus, sensu lato. He studied wood specimens of the following species: L. bidwillii (New Zealand), L. chilensis (Chile), L. decurrens (North America), L. macrolepis (China), L. papuana (New Guinea), and L. uvifera (Chile), the last being the type of Pilgero- dendron, proposed by Florin (7) as a separate genus. With regard to the latter, Peirce noted that “the woody anatomy has failed to disclose any features that would warrant giving generic status to that species and not to some of the others, for Libocedrus is comparatively heterogenous.” The wood anatomy of the other cupressaceous genera studied by him is all uniform and homogeneous within the genus. Perhaps an attempt to in- vestigate further the wood anatomy of Libocedrus sensu lato and Pilgero- dendron along the line of the present scheme of classification will bring out different results. To summarize, the resemblances of the elongated cones in the northern and southern species of Libocedrus are superficial only, probably due to parallel variation. Two groups of species differ from each other in cone- structure as well as vegetative characters. Different affinities are indicated and it is therefore desirable to treat these two groups of species as gener- ically distinct. Among the southern species, those from New Guinea differ from the rest in certain important characters of the ovulate as well as staminate cones and are treated as representing a separate genus. In addition to these three genera, there is Pilgerodendron, an earlier segregate from Libocedrus made by Florin. The classification of the four genera in question is as follows. 22 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxIv Heyderia K. Koch, Dendrol. 2(2): 179. 1873. Calocedrus Kurz in Jour. Bot. 11: 196. t. 133. July 1873. Libocedrus sensu Benth. & Hook. Gen. Pl. 3: 426. 1880, p. p., non Endlicher. Libocedrus subgen. Heyderia Pilger in Engler & Prantl, Nat. Pflanzenfam. 13: 389. 1926. Type species: Heyderia decurrens (Torr.) Koch. Trees; branchlets distichous, strongly compressed. Leaves scale-like, closely and distinctly imbricate, decussate, strongly compressed and de- current, free only at the obtuse and minutely pointed tip, the pairs of about equal length, the dorso-ventral narrow, the lateral keeled, over- lapping the dorso-ventral laterally, not joining together themselves. Flowers monoecious, solitary, terminal on different branchlets. Staminate flowers oblong, consisting of 6-16 decussate scales; anthers sessile, of 4 cells pendulous from the subpeltate, broadly ovate, pointed scale. Ovulate flowers oblong, formed of 6 erect woody imbricate persistent scales, the upper and lower pairs sterile, the middle pair only fertile; ovules 2 at the base of each fertile scale, erect. Mature cone oblong, more or less truncate, maturing the first year, the scales 6, in 3 pairs, woody, imbricate, mucronulate at the back near the tip, the lower pair small, ovate, sterile, recurved at tip, the middle pair much larger, 3-6 times as long as the outer or more, oblong, fertile, erect, the upper pair linear, about as long as or slightly longer than the middle pair, sterile, connate together inte a flat woody erect plate. Seeds 2 to each fertile scale, erect, compressed, with 2 unequal, lateral, erect, oblong wings, the larger to nearly as long as the scales, the other about half as long; cotyledons 2. Three species, one in Pacific North America, one in Formosa, and one in Yunnan, Hainan, and northern Burma. 1. Heyderia decurrens (Torr.) K. Koch, Dendrol. 2(2): 179. 1873. Libocedrus decurrens Torrey in Smithson. Contr. Knowl. 6(1): 7. f. 3. 1853. Thuja craigiana Murray in Rep. Oreg. Exped. 2: ¢. 2. 1854. Thuja gigantea sensu Carriere in Rev. Hort. 1854: 224. 1854, non Nuttall. Thuja decurrens Voss in Mitt. Deutsch. Dendr. Ges. 1907(16): 88. 1908. North America, Oregon and California to Lower California, scattered among other coniferous trees at 1,000—2,500 meters. The following forms are known in cultivation: Heyderia decurrens f. compacta (Beissner) comb. nov. Libocedrus decurrens compacta Hort. ex Beissner, Handb. Nadelh. 30. 1891. Heyderia decurrens f. glauca (Beissner) comb. Libocedrus decurrens glauca Beissner in Jager & eee os ed. 2, 472. 1884. Heyderia decurrens f. aureo-variegata (Schwerin) comb. n Libocedrus decurrens aureo-vartegata Schwerin in Mitt. Deutsch. Dende Ges. 1907(16): 256. 1908. 1953 | LI, LIBOCEDRUS AND CUPRESSACEAE 23 2. Heyderia formosana (Florin) comb. nov. Libocedrus formosana Florin in Svensk Bot. Tidskr. 24: 126. 1.2: t..2..1930. Libocedrus macrolepis sensu Forbes & Hemsl. in Jour. Linn. Soc. Bot. 26: 540. 1902, p. p., non Benth. & Hook Libocedrus macrolepis var. formosana Kudo in Jour. Soc. Trop. Agr. (Formosa) 3: 16. 1931. Formosa. scattered in broad-leaved forests, rarely forming pure forests, in ravines and on mountain slopes at 150—1,900 meters, in the northern and central part of the island. 3. Heyderia macrolepis (Kurz) comb. nov. Calocedrus macrolepis Kurz in Jour. Bot. 11: 196. ¢. 133. 1873. Libocedrus macrolepis Benth, & Hook. Gen, Pl. 3: 426. 1880. Thuja macrolepis Voss in Mitt. Deutsch. Dendr. Ges. 1907(16): 88. 1908. A rare tree, at about 1,400—1,600 meters, southwestern Yunnan and along the Burmese border; also in Hainan. Pilgerodendron Florin in Svensk Bot. Tidskr. 24: 132. 1930. Trees or shrubs, evergreens, the branchlets tetragonous. Leaves small, scale-like, ovate, quadrifarious, decussate, of about equal size, imbricate, adnate below, free and more or less spreading above, dorso-ventrally com- pressed, keeled on the back. Flowers dioecious, solitary, terminal on branchlets. Staminate flowers relatively large, subcylindric, with elongate scariose-margined basal leaves; scales large, decussate, imbricate; anthers of 4-8, usually 6 cells, pendulous from the subpeltate, short-stalked, erect, scales. Ovulate flowers with 4 elongate, decussate basal leaves, ovoid to ellipsoid, formed of 4 decussate persistent scales, slightly fleshy when young and cohering at margins, the outer smaller, sterile; ovules 2 at the base of each fertile scale, erect. Mature cone ovoid, the scales 4, in 2 pairs, separate, valvate, woody, with a long erect-incurved spine on the back near the tip, the lower scales small, oblong-lanceolate, sterile, the upper scales obovate-oblong, fertile, about 3 times as long as the lower, the axis often projected in the center of the cone into a very short ovoid or rarely subcylindric column. Seeds solitary or 2 at the base of each fertile scale, obtusely triangular, very unequally winged laterally, the longer wing erect, elongate, much longer than the seed, the other very short. One species in southern Chile. 1. Pilgerodendron uviferum (D. Don) Florin, loc. cit. 133. 1930. Juniperus uvifera D. Don in Lamb. Pin. 2: 17. 1824. Thuja tetragona Hook. in Lond. Jour. Bot. 3: 148. t. 4. 1844. Libocedrus tetragona Endlicher, Syn. Conif. 44. 1847. Libocedrus cupressoides Sargent, Silva N. Amer. 10: 134. 896. Libocedrus uvifera Pilger in Engler & Prantl, Nat. Pflanzenfam. ed. 2. 13: 389. 1926. Southern Chile, western slopes of the Andes from Valdivia southward to Terra del Fuego. 24 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv Libocedrus Endlicher, Syn. Conif. 42. 1947. Libocedrus subg. Eulibocedrus Pilger in Engler & Prantl, Nat. Pflanzenfam. ed. 2. 13: 389. 1926. Type species: Libocedrus doniana (Hook.) Endlicher = L. plumosa (D. Don) Sargent. Trees or shrubs; branchlets distichous and compressed, sometimes tetragonous when young. Leaves scale-like or short needle-like, imbricate only at the tip, decurrent, decussate or 3-ranked, compressed, very unequal, the dorso-ventral very small, the lateral much sae keeled and contiguous below. Flowers monoecious, solitary, terminal on different branchlets. Staminate flowers oblong, of 6-10, decussately arranged scales; anthers sessile, of 4 cells pendulous from the peltate or subpeltate scale. Ovulate flowers ovoid, formed of 4 decussate, erect, woody, persistent scales, the outer smaller, sterile; ovules 2 at the base of each fertile scale, erect. Mature cone oven. obtuse, maturing the first year, the scales 4, in 2 pairs, valvate, woody, with a short or long spine or small triangular bract-like appendage on the back above the center or near the tip, the lower scales small, ovate, sterile, the upper scales ovate-oblong, fertile, about 2 or 3 times as long as the lower. Seeds solitary or 2 at the base of each fertile scale, compressed, very unequally winged laterally, the lower wing erect, oblong, to nearly as long as the scale, the other narrow, reduced; cotyledons 2. Five species, widely scattered in regions bordering the Pacific in the Southern Hemisphere, one in southern Chile, two in New Zealand, and two in New Caledonia. 1. Libocedrus plumosa (D. Don) Sargent, Silva N. Amer. 10: 134. 1896; Druce in Rep. Bot. Exch. Cl. Brit. Isles 1916: 633. 1917. Dacrydium (?) plumosum D. Don in Lamb. Pin. ed. 2. App. 143. 1828. Thuja doniana Hook. in London Jour. Bot. 1: 571. t, 18, 1842. Libocedrus doniana Endlicher, Syn. Conif. 43. 1847. New Zealand, northern and southern islands, in forests, rare. 2. Libocedrus bidwillii Hook. f. Handb. N. Zeal. Fl. 257. 1867. New Zealand, northern and southern islands, on mountain slopes to 2,000 meters. 3. Libocedrus chilensis (D. Don) Endlicher, Syn. Conif. 44. 1847, Thuja chilensis D. Don in Lamb, Pin. 2: 19. 1824. Thuja chilensis Hook. London Jour. Bot. 2: 199. 1843. Thuja adina Poepp. & Endl. Nov. Gen. et Sp. 3: 17. t. 220. 1845. Southern Chile, on slopes of mountain valleys, at about 950-1,500 meters. 4. Libocedrus austro-caledonica Brongn. & Gris. in Bull. Soc. Bot. France 18, 140. 1871. New Caledonia, eastern slopes of Mt. Humboldt at about 1,000 meters. 1953 | LI, LIBOCEDRUS AND CUPRESSACEAE 25 5. Libocedrus chevalieri Buchholz in Bull. Mus. Hist. Nat. Paris 21(2) : 283. 1949. New Caledonia, western slopes of Mt. Humboldt, at 1,450—1,550 meters. 6. Libocedrus yateensis Guillaumin in Bull. Mus. Hist. Nat. Paris 21: 45, 1949. New Caledonia, right bank of Blue River, at about 200 meters. Papuacedrus Li, gen. nov. Arbor alta; ramulis oppositis distichis compressis; foliis decussatim oppositis, adpresse quadrifariatim subimbricatis, difformibus, lateralibus longioribus complicato-carinatis subfalcatis acutis vel subacutis, pro parte maxima adnatis, apice solum liberis, facialibus minutis, planis adpressis squamiformibus triangularibus vel basim ramulorum versus oblanceolatis acutis vel acuminatis; strobilis in diversis ramis monoicis; strobilis staminibus in ramulis terminalibus solitariis cylindricis, antheris 16—co- seriatis spiraliter dispositis, squamiformibus late ovatis subpeltatis breviter stipitatis chartaceis, loculis 2-6 globosis deorsum 2-valvatis; strobilis ovulatis in ramulis brevibus erectis, elongato-ovatis, squamis 4, decussatis, valvatis, 2 exterioribus sterilibus; strobilis ovulatis maturis ovoideis, squamis demum sublignosis, 2 exterioribus sterilibus ovatis vel oblongis ad basim bractea ovata acuta adnata munitis, 2 interioribus ad basim 2-ovulatis, longioribus lanceolatis subacutis vel rotundatis, exterioribus 2- vel 3-plo longioribus, infra medium bractea triangulari patula praeditis; seminibus 4 ellipsoidalibus plus minusve compressis lateraliter alatis, ala altera elongata, altera subobsoleta. Type spPecIES: Libocedrus papuana F. Muell. A genus of three species in New Guinea and Molucca. 1. Papuacedrus papuana (F. Muell.) comb. nov. Libocedrus papuana F. Muell. in Trans. Roy. Soc. Vict. N. S. 1(1): 32. 1889. Thuja papuana Voss in Mitt. Deutsch. Dendr. Ges. 1907(16): 88. 1908. New Guinea, in northern and southern parts at 1,700—3,000 meters, and Molucca, at about 2,000 meters. 2. Papuacedrus torricellensis (Schlechter) comb. nov. Libocedrus torricellensis Schlechter ex Lauterbach in Bot. Jahrb. 50: 53. f. 2, H-N. 1913. New Guinea, Torricelli Mountains, at about 900—1,000 meters. 3. Papuacedrus arfakensis (Gibbs) comb. nov. Libocedrus arfakensis Gibbs, Phytogeogr. & Fl. Arfak Mts. 84. f. 6, a—b. 1917. New Guinea, Arfak Mountains, on ridges and in forests, at 2,300—2,600 meters. 26 JOURNAL OF THE ARNOLD ARBORETUM _ |{vot. xxxiv CLASSIFICATION OF CUPRESSACEAE With the above redefinition of the genera Libocedrus, Papuacedrus, Heyderia, and Pilgerodendron, a problem which follows is their proper classification within the family Cupressaceae. These genera are segregated mainly on the basis of the ovulate cones, and the structure of the ovulate cone has long been considered as important in classification and of great significance in interpreting relationships in the conifers. A general review of the structure and evolution of the ovulate cone in the Cupressaceae is given by Florin (8,10). He considers the cones in some species of Juniperus and in Microbiota, in which only one fertile axillary complex and one single ovule is developed, as the most strongly reduced. It is interesting to note that although the northern and southern genera assigned to the Cupressaceae are grouped in separate taxa in most systems of classification of the conifers, Libocedrus has had various dispositions. The first important system was proposed by Endlicher in 1847 (5), in the work where Libocedrus was first established. In his order Cupressineae, the Actinostrobeae include these southern genera: Widdringtonia, Frenela, Actinostrobus, Callitris, and Libocedrus; while the Thujopsideae include the northern genera Biota, Thuja, and Thujopsis. At that time, it should be noted, only three species of Libocedrus were known, all from the Southern Hemisphere, and therefore the scope of the genus was clear and definite and its relationship was correctly indicated by the author. In later years, when the northern species of the genus Libocedrus were discovered, practically all authors of later systems, such as Eichler (4), Neger (26), Vierhapper (37), and Saxton (32), included Libocedrus with the northern genera, apparently interpreting the genus on the basis of the northern species only. A radical change was made by Pilger (28), who combined the northern and southern groups of genera, long referred to two different groups by all authors, into one subfamily Thujoideae under the Cupressaceae. His system of classification of the whole family is as follows: Subfamily I. Thujoideae: Actinostrobus, Callitris, Tetraclinis, Callitropsis, Widdringtonia, Fitzroya,‘Diselma, Thujopsis, Thuja, Libocedrus, Fokienia. Subfamily II. Cupressoideae: Cupressus, Chamaecyparis. Subfamily III. Juniperoideae: Arceuthos, Juniperus. The latest system. by Janchen in 1950 (17), classifies the Cupressaceae as follows: Subfamily I. Juniperoideae Tribe Junipereae: Arceuthos, Juniperus. Subfamily II. Cupressoidea Tribe 1. Cupresseae: Cane. Chamaecyparis, Fokienia. Tribe 2. Thujopsideae: Pilgerodendron, Libocedrus, Microbiota, Biota, Thuja, Thujopsis. Tribe 3. Actinostrobeae: Diselma, Fitzroya, Widdringtonia, Neocallitropsis, Callitris, Tetraclinis, Actinostrobus. 1953 | LI, LIBOCEDRUS AND CUPRESSACEAE af Pilger’s system, appearing in a standard reference work, is widely known, but, compared with other systems, it is unsatisfactory in that it combines the usually separated and widely different Actinostrobeae and Thujopsideae into one. This arrangement may serve to obviate the problem of placing the questionable Libocedrus, but it obscures the different and divergent trends of development of these genera. Also unsatisfactory is Pilger’s placing of Fokienia in his Thujoideae instead of Cupressoideae, as the cone of Fokienia is essentially the same as that of Chamaecyparis; the two should undoubtedly be closely associated, as in Janchen’s system, where Cupressus, ene aka and Fokienia constitute the tribe Cupresseae. Saxton (30, 3h, 32), working with embryogeny, considered that the Cupressaceae Zee be divided into at least two subfamilies: the Calli- troideae, containing the genera Actinostrobus, Callitris and Widdringtonia, and the Cupressoideae, containing the remaining genera of the family. Tetraclinis and Fitzroya have affinities with both groups, but they should probably be considered as belonging to the Cupressoideae. Moseley (24) based his evaluation of characteristics on the reproductive morphology and embryogeny, considered by him as of phylogenetic im- portance in 12 genera, and proposed an entirely different system for the Cupressaceae. The characters are listed in a chart, mostly presented in pairs, considered by him as either primitive or advanced. His modification of Pilger’s system is as follows: (genera starred are considered by him as of doubtful status). Subfamily Cupressoideae: Cupressus, Chamaecyparts. Subfamily Juniperoideae: Juniperus, Arceuthos*, Microbiota*. Subfamily Thujoideae: i la Biota, Tetraclinis, Fitzroya, Thujopsis. Thuja, Diselma*, Fokien Subfamily Callitroideae: jee om Callitris, Widdringtonia, Callitropsis*. Considering the number of primitive characteristics, Moseley regards the Cupressoideae as lowest in the family and the Callitroideae as the most highly evolved. In the Thujoideae, Libocedrus and Biota possess the greatest number of primitive characteristics, while Thuja and Fitzroya are the most advanced. Callitroideae was originally established by Saxton and upheld by Moseley on the basis of these morphological characters: archegonia lateral in position, the absence of a prosuspensor in the embryo, a proembryo that completely fills the archegonium, the absence or obscurity of an archegonial jacket, and a proembryo which is not in definite tiers. Moseley’s phylogenetic scheme is very different from, and in some cases diametrically opposed to the various systems proposed by taxonomists on the basis of external morphology, especialy that of the cone. Juniperus, with fleshy fused cone-scales and wingless seeds, is considered by all others as more advanced than those genera in the Thujoideae, which have dry distinct cone-scales and winged seeds (10). In Moseley’s system, the order is reversed. He considers that “the Callitroideae possess outstanding characters [as mentioned above] that distinguish them from the other 28 JOURNAL OF THE ARNOLD ARBORETUM - [vot. xxxiv groups of the Cupressaceae.”” However, when he compares Fitzroya and Tetraclinis, two genera in his Thujoideae, with the Callitroideae, he finds in Fitzroya, four, and in Tetraclinis three important characteristics in common with the latter group. It remains to be seen whether characters in embryogeny will be sufficient to explain the phylogenetic trends in the Cupressaceae. At the present judgment on the phylogenetic significance of these characters is still con- troversial. Thomson (34, Radforth, 29) disagrees with Buchholz’s idea, which Moseley follows, in the interpretation of certain fundamental phenomena in embryogeny pertinent to the phylogeny of conifers. In some cases, such as polyembryogeny, Thomson’s view of the phylogenetic significance is just the reverse of that of Buchholz. Using mainly the characters of the cone-scales, a revised system of classification for the Cupressaceae is presented below. Among the genera of the family, there are two main types of cones. In one group, the cone- scales are present in pairs or in whorls of three or four. Generally there are two pairs, and only more rarely two ternate whorls or two quadrate whorls. The scales are always thick and usually woody, and are valvate in arrangement, as the scales come into contact at their edges and do not overlap. All genera with cones of this type occur in the Southern Hemi- sphere, with the exception of the isolated Tetraclinis. These genera may be considered as constituting one subfamily, the Callitroideae (PLATE 1). In this subfamily, there are three genera with ternate whorls: Fitzroya, Actinostrobus, and Callitris. In Fitzroya, there are three whorls of scales, the innermost being very rudimentary and minute, the middle largest and each scale bearing about two or three 3-winged seeds, and the outermost smaller and sterile. In Actinostrobus, there are two whorls surrounding a slightly protruding axis, the outer and inner scales being of about equal size and each bearing two 3-winged seeds. In Callitris, there are also two whorls, the inner being the larger, each scale bearing many winged seeds, and the outer ones slightly smaller and bearing fewer seeds. Both Actinostrobus and Callitris may sometimes possess a residuum of very rudimentary scales in the center of the cone as in Fitzroya. All three genera have 3-ranked leaves. These genera, of comparatively more primitive character in having ternate scales, few to many, winged seeds, and sometimes a whorl of rudimentary scales in the center, are clearly of close relationship and are here classified as representing one tribe, the Actino- strobeae. It should be noted that Fitzroya was described by the publishing author, Hooker, as having imbricate cone-scales. Pilger (28) described the scales as somewhat imbricate. So far as I can make out from herbarium specimens, the thick, coriaceous scales are valvate as in the other southern genera. For a definite determination it will be necessary to have fresh The other southern genera have 2-, 4-, or 8-ranked leaves, and all have cones consisting of two pairs of scales, except Octoclinis and Neocallitropsis (Callitropsis). The cone of these genera has eight scales in two whorls, with a short axis protruding in the center. Each of the inner scales bears 1953] LI, LIBOCEDRUS AND CUPRESSACEAE 29 two, winged seeds. The leaves in these genera are in whorls of four. In Widdringtonia there are four scales of equal size, each bearing many, winged seeds. In Diselma, Papuacedrus, Pilgerodendron, and Libocedrus, there are two pairs of scales. In Diselma, the two pairs are of about equal size, one sterile and one fertile, the latter bearing at the base of each scale two or three winged seeds. In Pilgerodendron, Papuacedrus, and Libocedrus the two pairs of cone scales are of unequal size, the outer much smaller and sterile, and the inner larger and fertile, bearing one or two seeds at the base of each scale. These southern genera represent another tribe, the Libocedreae. Papuacedrus, as noted above, differs from other genera of the Cupres- saceae in the spiral arrangement of the scales in the staminate cone. This character may be of phylogenetic significance, as it indicates relationship with the Taxodiaceae. In this connection mention may be made of two genera of the Taxodiaceae with outstanding characters. Metasequota has decussate scales in the ovulate cones, a character transcending the Taxo- diaceae and suggesting relationship with the Cupressaceae (16, 33). Athrotaxis, of Tasmania, the only genus of the Taxodiaceae of the Southern Hemisphere, has either spirally or decussately arranged leaves and sub- spirally arranged staminate scales, characters somewhat intermediate be- tween the Taxodiaceae and Cupressaceae. These genera, together with Papuacedrus, offer great possibilities in elucidating the relationships be- tween these two families by further investigation. A third tribe, the Tetraclineae, contains the more or less isolated genus Tetraclinis of northern Africa. It is the only genus of the Northern Hemi- sphere with valvate cone-scales. There are two pairs of cone scales, of equal size but of slightly different shape. The young scales are somewhat fleshy. The vegetative characters mostly closely approach Heyderia and Thujopsis. The cotyledons are three to five, instead of usually two as in other genera, and this character suggests Juniperus. Tetraclinis thus shows characters intermediate between the northern and southern genera, but its basically valvate cone-scales indicate closer relationships with those of the south. In the northern genera, excepting Tetraclinis, the cone-scales show more varied development, but basically the arrangement is imbricate. These genera constitute another subfamily, the Cupressoideae. The scales occur in pairs with the exception of Juniperus and Arceuthos, where the scales are generally present in threes but sometimes also in twos. In this subfamily three tribes are discernible (PLATE 2). In the first tribe, Cupresseae, including Cupressus, Chamaecyparis, and Fokienia, the cones, which are essentially globose, bear three to eight pairs of shield-like scales. Most of these scales except usually the outermost and innermost ones, are fertile, each bearing two to many, winged seeds. In the number of scales and seeds, this group is undoubtedly the most primitive. Cupressus has six to twelve scales, the fertile ones bearing many seeds each. Chamaecyparis also has six to twelve scales, but the fertile scales bear only three seeds each. Fokienia has a larger number of scales, varying from twelve to sixteen, but the fertile ones bear only two seeds 30 JOURNAL OF THE ARNOLD ARBORETUM - [voL. xxxiv each. Because of the thickness of the cone-scales they do not appear distinctly overlapping at the edges. Pilger (28) considers these scales as valvate. However, the outer scales cover the inner ones at almost the entire length and their disposition, much in the same manner as in Heyderia, is clearly imbricate. In Fokienia, the seeds are more unequal-winged than in the other two genera. This genus is probably more advanced than the other two and serves as a link with the following tribe, which it also resembles very closely in vegetative characters. The second tribe, Thujopsideae, is characterized by fewer scales, of flat or concave, generally elongate shape. 7'hujopsis has six to eight thick scales, the innermost and outermost pairs being sterile. The fertile scales bear two winged seeds each. In TAuja, there are eight to twelve scales, with the innermost pair sterile and often fused into a plate. The middle pairs bear two winged seeds each at the base of each scale. In Biota,? sometimes included in TAuja, the scales are six in number, thick in texture, with the inner pair fused and sterile and the outer bearing one or two wingless seeds each. This genus probably connects with the following tribe, which has fleshy scales and wingless seeds. From Thuja, further reduction in the number of scales and seeds resulted in Heyderia, with only three pairs of scales, the inner fused and sterile, the middle fertile, and the outer much smaller and also sterile. The seeds are unequally winged. The last tribe, Junipereae, consisting of Arceuthos and Juniperus, some- times combined into one genus, has fleshy cone-scales, separate at first but fused together at maturity. The scales usually appear in two whorls of three each, but occasionally also in pairs. The seeds are wingless. This is undoubtedly the most highly evolved group of the whole family. Although the fleshy connate scales are distinctive, this tribe is clearly linked with the last tribe, inasmuch as Biota, with fleshy scales and wingless seeds, is somewhat intermediate. It is thus better treated as an advanced, specialized tribe of the northern subfamily, with imbricate scales, than as a subfamily by itself. All genera of the subfamily Callitroideae, with the exception of Tetraclinis, are of the Southern Hemisphere; while all genera of the sub- family Cupressoideae are of the Northern Hemisphere. Tetraclinis occurs in northern Africa, within the range of the hypothetical Gondwana land as with all the rest of the southern genera. The geographical range shows that the two subfamilies have developed independently for a very long time. This pattern of distribution is in accord with that of other group of conifers, where the genera or higher categories are either of the north or of the south. The extraordinary geographical range that has been accredited to Libocedrus, sensu lato, was based upon a misconception of significant generic characters. Fossil records have shown that the coniferous floras of the Northern and Southern Hemispheres have been distinct from each other since very * Microbiota decussata Komarov is an uncertain genus and species. Rehder (Man. Cult. Trees Shrubs, ed. 2. 55. 1940) suggests it as probably only a variation of Biota orientalis (L.) Endl. retaining the juvenile foliage up to the fruiting stage. 1953 | LI, LIBOCEDRUS AND CUPRESSACEAE 31 ancient times. In North America, for instance, Mesozoic and Cenozoic fossils all pertain to genera of the present northern type, such as Cupressus. Juniperus, Thuja, Taxodium, Sequoia, Abies, Larix, Picea, Pinus, Psuedotsuga, Tsuga, Cephalotaxus, and form genera related to these modern ones (18, 22).* Florin (9), in a detailed analysis of the Tertiary fossil conifers from the southern lands, shows the genera to be the same as those now existing in the Southern Hemisphere, and that none of the genera typical of the Northern Hemisphere mentioned above were present. Florin has convincingly proved that the separation of the northern and southern types has existed since the late Palaeozoic. He also indicates. however, that certain genera of the southern group might have forced their way northward into the region primarily occupied by the northern group and vice versa, which is also reflected by the recent distribution of some genera. A system of classification for the family Cupressaceae is tabulated below. The synonymy of suprageneric groupings in the Cupressaceae, as well as in other conifers, is very complicated. Practically all the tribes given below have been treated at one time or another as families or subfamilies. In a very ancient group like the conifers, the existing genera, mostly of relic nature, naturally do not show intimate relationships between them. as do many more modern groups. The rather burdensome synonymy reflects the varied opinions expressed by many authors. It is suggested that in the conifers, a broader outlook must be taken in presenting systems of classification in order to show the relationships between the existing genera and to render the systems useful for practical purposes. As the synonyms have been given very fully by Janchen (17) in a recent publica- tion, they are not repeated here. Family CUPRESSACEAE Neger I. Subfamily CALiitRomEAE Saxton in New Phytol. 12: 253. 1913. A. Tribe Actinostrobeae Endlicher, Syn. Conif. 3. 1847, p. p. 1. Actinostrobus Miquel (2 species in western Australia). 2. Callitris Ventenat (About 20 species in Australia, Tasmania, and New Caledonia). 3. Fitzroya W. J. Hooker (1 species, in southern Chile). B. Tribe Libocedreae Li (Actinostrobeae Endlicher, op. cit., p. p.). Arbor vel frutex; foliis decussatim oppositis; squamis ovulatis 4 vel 8, 2-seriatis, similibus, aequalibus vel inaequalibus. 1. Neocallitropsis Florin (Callitropsis Compton) (1 species, New Caledonia ). 2. Octoclinis F. Mueller (1 species, southwestern Australia). ’ A few doubtful records of fossil material pertaining to Callitris and Podocarpus are known from the Tertiary of the Northern Hemisphere. Florin noted (9, p. 73) that r suspicion,” and (9, p. 83) that “All the Tertiary fossil coniferous remains from Europe and North America supposed to belong to Callitris appear to be referable to Aa | the northern genus Jetraclinis. 32 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv ios) . Widdringtonia Endlicher (5 species, South Africa and southeastern Tropical Africa). Diselma J. D. Hooker (1 species, Tasmania). Papuacedrus Li (3 species, New Guinea, Moluccas). Pilgerodendron Florin (1 species, southern Chile). . Libocedrus Endlicher (5 species, southern Chile, New Zealand, New Caledonia). C. Tribe Tetraclineae Li (Tetraclinaceae Hayata in Bot. Mag. Tokyo 46: 27. 1932). 1. Tetraclinis Masters (1 species, Morocco, Algeria, Tunisia). II. Subfamily CupressowearE K. Koch A. Tribe Cupresseae Neger 1. Cupressus Linnaeus (About 12 species, North America, Asia to eastern Mediterranean). 2. Chamaecyparis Spach (About 6 species, North America, Japan, ormosa). 3. Fokienia A. Henry and H. H. Thomas (1 species, southeastern China to Tonkin), B. Tribe Thujopsideae Endlicher 1. Thujopsis Siebold & Zuccarini (1 species, Japan). 2. Thuja Linnaeus (5 species, eastern Asia and North America). 3. Biota D. Don ( 1 species, northeastern Asia). 4. Heyderia K. Koch (3 species, Pacific North America, Formosa, Hainan, southwestern China to northern Burma). C. Tribe Junipereae Neger 1, Arceuthos Antione & Kotschy (1 species, Europe to western Asia). 2. Juniperus Linnaeus (About 60 species, widely distributed in the Northern Hemisphere). Inn +S SUMMARY The genus Libocedrus is found to consist of two diverse groups of species with basic differences in the cone structure. The genus should be limited to those species, all of the Southern Hemisphere, with four valvate cone- scales. Three species from New Guinea have ovulate scales bracteate below and spirally-arranged staminate scales, indefinite in number, and are segregated from the other species as a distinct genus Papuacedrus. The three northern species, with six, imbricate cone-scales, constitute another genus, Heyderia. The arrangement of cone-scales seems to be an important character in the classification of the Cupressaceae. As a result of the reclassification of Libocedrus, the family Cupressaceae can be reorganized as consisting of two subfamilies. The subfamily Callitroideae is composed of genera with valvate scales and can be divided into three tribes: Actinostrobeae, with ternate scales, Libocedreae, with paired or quadrate scales, and Tetraclineae, with paired dissimilar scales. The sub- family Cupressoideae is composed of genera with imbricate scales and can be divided into three tribes: Cupresseae, with thick, shield-like scales, Thujopsideae, with flat, more or less concave scales, and Junipereae, with fleshy scales coalescing at maturity. All genera of the Callitroideae, with the exception of the isolated Tetraclinis, are of the Southern Hemisphere, 1953] LI, LIBOCEDRUS AND CUPRESSACEAE 33 while all genera of the Cupressoideae are of the Northern Hemisphere. This distribution pattern, together with their basic difference in the cone- structure, indicates that the two groups are probably of remote relation- ship, having been long isolated and having developed independently, like many other groups of conifers. _ ho S LITERATURE CITED BENTHAM, G. and J. D. Hooker, Genera Plantarum. 3 vol. London, 62-83. BRONGNIART, A. and A. Gris. Supplément aux coniféres de la Nouvelle- Calédonie. Bull Soc. Bot. France 18: 130-141. 1871. BucHHOLz, J. T. Additions to the coniferous flora of New Caledonia. Bull. Mus. Nat. Hist. Paris 21: 279-286. 1949 E1cHLer, A. W. Coniferae. In Engler and Prantl, Nat. Pflanzenfam. II. 1: 28-116. 1887. ENDLICHER, S. L. Synopsis Coniferum, Sangalli. Frorin, R. Die Koniferengattung Libocedrus ae in Ostaisien. Svensk Bot. Tidskr, 24: 117-131. 30. . Pilgerodendron, eine neue Koniferen Gattung aus Siid-chile. Svensk Bot. Tidskr. 24: 132-135. 1930. Die Koniferen “ Oberkarbons und des unteren Perms. Paleon- tographica 85. Heft 1-7. 1938-45. ———. The once fossil conifers of south Chile and their phytogeograph- ical significance. Kungl. Svensk Vatenskapsakad. Handl. HI. 19(2): 1-107. 1940. . Evolution in cordaites and conifers. Act. Hort. Berg. 15: 285-388. 1951. Grpss, L. S. Dutch N. W. New Guinea. A Contribution to the Phytogeog- raphy and Flora of the Arfak Mountains, etc. Taylor & Francis, London. 1917 fe Hooker. W. J. On a new species of Thuja. and on Podocarpus totara of New ia London Jour. Bot. 1: 570-575. 1842. ————. Figure and description of a new species of Thuja, from Chile. London Tear Bot. 2: 199-200. 1843. — —. Description, with a figure, of a new species of Thuja, the Alerse of Chile. London Jour. Bot. 3: 144-149. 1844. 5. Hooker, J. D. Handbook of the New Zealand Flora. London. 1867. 6 Hu, H. H. and W. C. Cuenc. On the new family Metasequoiaceae and on Metasequoia glyptostroboides, a living species of the genus Metasequoia found in Szechuan and Hupeh. Bull. Fan Mem. Inst. Biol. N. S. 1: 151-163. 1948 JANCHEN, E. Das System der Coniferen. Sitzungsber. Oest. Akad. Wiss. Math.-Nat. Kl. Abt. I. 1949(3): 155-162. 1950. KNow.ton, F. H. A catalogue of the our and Cenozoic plants of North America. U. S. Geol. Surv. Bull. 696. 1919. Kocu, Kk. Dendrologie, 3 vol. Erlangen. en Kurz, S. On a few new plants from Yunan [sic!|. Jour. Bot. Brit. & For. 11: 193-196. 1873. LAMBERT, A. B. A Description of the Genus Pinus. 1803-24. ed. 2. 3 vol. 1828-39. London. 34 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv 22. LaMorte, R. S. Supplement to catalogue of Mesozoic and Cenozoic plants of North America 1919-37. U. S. Geol. Surv. Bull. 924. 1944. 23. LAUTERBACH, C. Neue Pinaceae Papuaseins. Bot. Jahrb. 50: 46-53. 1913. 24. Mosetey, Jr., M. F. Contribution to the life history, morphology, and phylogeny of Widdringtonia cupressoides. Lloydia 6: 109-132. 1943. 25. MUELLER, F. von. Records of ce alae of Sir William Macregor’s high- land plants from New Guinea. Trans. Roy. Soc. Vict. N. S. 1: 1-45. 1889. 26. NeEGER, F .W. Die Nadelholzer (Koniferen und ubrigen Gymnospermen. Sammlung Goschen 355: 1-185. 27. Pertrce, A. S. Systematic anatomy of i: woods of the Cupressaceae. Trop. Wo ods 49: 5-21. 1937. 28. Pitcer, R. Coniferae. In Engler and Prantl, Nat. Pflanzenfam. ed. 2. 13: 121-407. 1926. 29. RaAprortH, N. W. Robert Boyd Thomson’s morphological and ne studies bearing on Gymnosperm taxonomy. Bot. Gaz. 110: 1948 30. Saxton, W. T. Contributions to the life history of nee rea Bot. 24: 557-569. 1910. 31 . Contributions to the life history of Tetraclinis articulata, Masters, with. some notes on the phylogeny of the Callitroideae. Ann. Bot. 27: 577-606. 1913 3 The classification of conifers. New Phytol. 12: 242-262. 1913. 33. STERLING, C. Some features in the morphology of Metasequoia. Am. Jour. Bot. 36: 461-470. 1949. 34. THomson, R. B. “Polyembryogeny”: sexual and asexual embryo initiation and food supply. Trans. Roy. Soc. Canad. Ser. 3, Sect. V. 34: 143-169. 1945, 35. Torrey, J. Plantae Fremontianae. Smithsonian Contr. Knowl. 6: 1-24. 1853. 36. Voss, A. Coniferen Nomenklatur Tabelle. Mitteil. Deutsch. Dendrol. Ges. 1907(16 ): 88-95. 1908. 37. VIERHAPPER, F. Entwurf eines neuen Systems der Coniferen. Abh. Zool. Bot. Ges. Wien V. 4: 1-56. 1910 38. GuUILLAUMIN, A. Contribution a la flore de la Nouvelle-Calédonie, XCII. Plantes récoltées par M. J. Bernier (complément). Bull. Mus. Hist. Nat. Paris 21: 453-461. 1949. Morris ARBORETUM, PHILADELPHIA, PA. EXPLANATION OF PLATES PLATE 1. Diagram showing the hypothetical types of cones of the genera of the subfamily Callitroideae and their probable relationships and directions of evolution. PLATE 2. Diagram showing the hypothetical types of cones of the genera of the subfamily Cupressoideae and their probable relationships and directions of evolution. Jour. Arnotp Ars. VoL. XXXIV PiaTeE I & LIBOCEDREAE Ve LIBOCEDRUS PILGERODENDRON PAPUACEDRUS ise TETRACLINEAE 7 OISELMA re ox 4) UD | ~~ ame TETRACLINIS Oa ma NEOCALLITROPSIS > a OCTOCLINIS ACTINOSTROBEAE SAD =< CAI ~~ Oxf FITZROYA UF [\) = ACTINOSTROBUS 86 CALLITRIS Li, LrsocEDRUS AND CUPRESSACEAE Jour. ARNOLD Ars. VoL. XXXIV PriateE II JUNIPEREAE ee JUNIPERUS ieee GOD THUJOPSIDEAE an () = a. 0) eae THUJA ae S77 THUJOPSIS | | C_) CUPRE SSEAE QD SS (esa) > O(fEo)0 ’ kamr ’ Pad ge ; ( 0} oe ESL — FOKIENIA a Ee CHAMAECYPARIS Cc) CUPRESSUS Li, LriBoCEDRUS AND CUPRESSACEAE 1953] SMITH. STUDIES OF PACIFIC ISLAND PLANTS. XIV 37 STUDIES OF PACIFIC ISLAND PLANTS, XIV NOTES ON THE FIJIAN SPECIES OF CYRTANDRA A. C. SMITH THE ONLY REVISION which treats the vast genus Cyrtandra (Gesneri- aceae) in its entirety is that of C. B. Clarke (in DC. Monogr. Padn..3: 1-303. 1883), and this of course is entirely out of date. In it the 167 recognized species of Cyrtandra are arranged in 13 sections, which have seemed to most subsequent students highly artificial. The total number of described species of the genus is now at least double the number known to Clarke. It may prove quite impossible to revise Cyrtandra for its entire range, unless some competent student is willing and able to spend many years at this task alone and to supplement his work by intensive field study. Regional solutions, however, are quite feasible, and these may follow the precedent set by J. F. Rock for the Hawaiian species (in Am. Jour. Bot. 4: 604-623. 1917, 5: 259-277. 1918, 6: 47-68, 203-216. 1919), in which Clarke’s sections are ignored. The purpose of the present paper is in no sense revisional ; it is intended only to place on record the ten species collected by me in 1947 1 that appear to be undescribed, and to note pertinent data concerning a few of the older species. With these ten new species, Cyrtandra is now represented in Fiji by 45 species, as contrasted with the 20 species of the region known to Clarke. For convenience, the sequence here followed is that of Clarke, but the writer hopes in the near future to revise the Fijian species, and at that time more suitable groupings can be devised. It is not to be supposed that all the Fijian C yrtandrae are now known; on the contrary, further intensive collecting in the wet montane forest is certain to disclose undescribed entities. The genus is second in size only to Psychotria, in Fiji, and it definitely includes some of the most beautiful species of undershrubs and low trees to be found in the Melanesian forests. Specimens cited in the present paper are deposited in the herbaria of the Arnold Arboretum (A) and the U. S. National Museum (US). Cyrtandra victoriae Gillespie in Bishop Mus. Bull. 74: 25. fig. 34. 1930. Vitt Levu: Mba: Mt. Ndelaiyod, on the escarpment west of Nandarivatu, Smith 5081 (A, US); upper western slope of Mt. Tomanivi, Smith 5915, 5916, 5917 (all A, US): Ra: Ridge from Mt. Namama (east of Nandarivatu) toward Mt. Tomanivi. Smith 5697, 5722 (A, US). The cited specimens are from shrubs or low trees 2-5 m. high, growing in dense wet forest; the flowers are large and conspicuous, the white corolla ‘These collections were made under the auspices of the Arnold Arboretum of Harvard University and the John Simon Guggenheim Memorial Foundation, with the aid of grants from the Penrose Fund of the American Philosophical Society and the Bache Fund of the National Academy of Sciences. 38 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv having a pale yellowish pubescence. This very striking species, typified by a Gillespie collection from Mt. Tomanivi, is one of the most beautiful Fijian Cyrtandrae and is fairly frequent at elevations of 900-1300 m. in north-central Viti Levu. The leaves vary tremendously in size during their development, the chief vegetative characteristic of the plant being the dense indument of multicellular eglandular hairs (3-8 mm. long) that covers all the young parts and is often long-persistent, A few supplemen- tary notes may be added to Gillespie’s description, as follows: Calyx copiously sericeous within toward base with hairs 3-4 mm. long (not glabrous as described), the lobes 4 or 5, lanceolate, up to 2 cm. long, the 2 or 3 upper ones joined by the tips or laterally connate; corolla up to 4 cm. long, 3-5 mm. in diameter at base, broadened at throat to about 15 mm. in diameter, copiously pilose without like the calyx, less densely short- pilose within, the hairs several-celled and usually gland-tipped, inter- spersed with very short glandular hairs, the lobes 5-6 mm. broad; disk glabrous, about 1.5 mm. high; ovary glabrous, the style stout. copiously glandular-pilose. Cyrtandra tomentosa A. C. Sm. in Sargentia 1: 116. 1942. Vitr Levu: Mba: Hills east of Nandala Creck, about 3 miles south of Nandarivatu, alt. 850-970 m., Smith 6237 (A, US) (shrub 2 m. high, in dense forest; fruit green). The cited specimen, the second known of the species, agrees excellently with the type, Degener 14889, in all respects. Its leaves are slightly larger (petioles up to 6 cm. long; blades up to 35 & 9 cm.). The type locality, Nandrau, falls in the present province of Nandronga & Navosa and is slightly south of the region cited above. Cyrtandra amicta sp. nov. decurrentibus, apice breviter acuminatis, margine undulatis ac etiam irregulariter et minute calloso-serrulatis, supra glabris vel interdum pilis multiseptatis ad 2 mm. longis subpersistenter pilosis, subtus nervis et interdum facie pilis aureis debilibus 0.1—0.5 (-1) mm. longis molliter pilosis demum subglabratis, costa valida supra elevata subtus prominente, nervis secundariis utrinsecus 15-20 arcuato-adscendentibus copiose anas- tomosantibus supra inconspicue subtus valde elevatis, nervis tertiariis et rete venularum laxis supra immersis subtus prominulis; inflorescentiis axillaribus multifloris compacte cymosis multibracteatis laxe capitatis ad 10 cm. diametro, bracteis omnino liberis submembranaceis oblongo- 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 39 lanceolatis, 25-30 mm. longis, 7-10 mm. latis, subacutis, intus basim versus pilis aureis debilibus pluriseptatis 2-3 mm. longis villosis cetera elabris; pedicellis crassis teretibus glabris sub anthesi 6-10 mm. longis; calyce amplo membranaceo campanulato-infundibulari sub anthesi 35—40 mm. longo et basim versus 6-10 mm. diametro, superne latiore, extus glabro, intus basim versus ut bracteis debiliter piloso, inaequaliter 5-lobato, lobis deltoideis vel deltoideo-lanceolatis 5-12 * 3-10 mm.; corolla mem- branacea inconspicue nervata infundibulari sub anthesi 40-45 mm. longa et basim versus 5—7 mm. diametro, superne ampliata, extus glabra (inter- dum juvenili tubo pilis ut bracteis floccoso-villosa mox glabrata), intus glabra, lobis 5 leviter inaequalibus semiorbiculari-oblongis in alabastro late imbricatis sub anthesi patentibus 7-10 mm. longis latisque; staminibus glabris tubo corollae supra medium affixis, filamentis crassis leviter com- planatis 3-5 mm. longis, antheris oblongis 4—5 mm. longis utroque rotunda- tis; disco crasse carnoso glabro annulari-cupuliformi 1.5-2 mm. alto apice undulato; ovario graciliter ovoideo glabro vel superne parce piloso, stylo crasso tereti sub anthesi circiter 15 mm. longo pilis patentibus 0.3-0.7 mm. longis pluriseptatis capitato-glandulosis copiose piloso, stigmate subclavato demum bilobato,. lobis complanatis; fructibus juvenilibus (calyce caduco) elongato-ovoideis in sicco rugulosis. Vitt Levu: Mba: Eastern slopes of Mt. Koroyanitu, Mt. Evans Range, Smith 4243, 4244 (A, US); upper western slopes of Mt. Tomanivi [ Mt. Victoria], alt. 1100-1150 m., Sept. 6, 1947, Smith 5914 (A type, US); hills east of Nandala Creek, about 3 miles south of Nandarivatu, Smith 6223 (A, US); Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Rewasau, Smith 5654 (A, US). The cited specimens are from often simple-stemmed shrubs 1.5—4 m. high, occurring at elevations of 850-1150 m. in dense forest. The bracts, calyx, and corolla are white, the fruit becoming white at maturity and up to 2.5 cm. in length. The leaves are characteristically congested toward the apex of the plant, concealing the densely clustered axillary inflor- escences. Cyrtandra amicta is strongly characterized by its glabrous flowers and its remarkably long calyx, which approximates the corolla in length and conceals it except for the tip. Its relationship is with C. vitiensis Seem. (§ Decurrentes), which it resembles closely in having its large leaf-blades gradually narrowed and long-decurrent at base. However, C. vitiensis has the calyx only about half as long as that of the new species, while its corolla is sericeous without. I believe that C. vitiensis is represented in my 1947 collections by nos. 5103, 5315, and 5848, from Mt. Tomanivi and vicinity; these numbers differ from the new species in the stated floral characters, in having the calyx uniformly long-pilose within, and in their more obvious and persistent foliage-indument. Cyrtandra occulta sp. nov. Frutex ad 2 m. altus, caule simplici valido apice ad 2 cm. diametro saepe fistuloso primo debiliter piloso mox glabrato; foliis oppositis magnis. 40 JOURNAL OF THE ARNOLD ARBORETUM [VOL. XXXIV petiolis crassis canaliculatis 2-8 cm. longis ad basim alatis mox glabratis. laminis in vivo carnosis in sicco subcoriaceis vel chartaceis supra fusco- viridibus subtus pallidioribus, obovatis vel oblanceolatis, 30-40 cm. longis, 10-16 cm. latis, basim versus gradatim angustatis et in petiolum longe decurrentibus, apice ut videtur acutis, margine undulato-crenulatis etiam irregulariter calloso-denticulatis, utrinque praecipue nervis pilis aureis debilibus 0.5—2 mm. longis molliter pilosis, supra mox glabratis et minute pustulosis, subtus demum subglabratis, costa valida supra elevata subtus prominente, nervis secundariis utrinsecus 18-22 curvatis obscure anastomo- santibus supra subplanis subtus prominentibus, nervis tertiariis et venulis laxis subimmersis; inflorescentiis axillaribus multifloris compacte cymosis subcapitatis 3.5-5 cm. diametro multibracteatis, bracteis extimis in involucrum latum cupuliforme connatis submembranaceis utrinque copiose tomentosis, pilis pallide aureis debilibus pluriseptatis extus 2—3 mm. longis demum caducis intus 5-8 mm. longis densissimis persistentibus, bracteis interioribus numerosis membranaceis oblongis ad 3 & 1.5 cm similiter pilosis; pedicellis crassis (2-3 mm. diametro) teretibus sub anthesi 5-10 mm. longis persistenter debiliter pilosis (pilis 2-3 mm. longis plurisep- tatis) ; calyce membranaceo campanulato sub anthesi 20-22 mm. longo et basim versus circiter 10 mm. diametro superne latiore, utrinque pilis eis bractearum similibus densissime sericeo, apice in lobos deltoideo-lanceolatos inaequaliter fisso; corolla membranacea infundibulari sub anthesi 30-35 mm. longa et basim versus 5—7 mm. diametro, superne ampliata, extus pilis 5-7 mm. longis septatis copiose et persistenter sericeo-tomentella, intus praeter basim versus ubique copiosissime glanduloso-pilosa (pilis minutis patentibus, stipite ad 0.2 mm. longo vel subnullo plerumque 2—4- septato apice capitato-glanduloso), lobis 5 subaequalibus oblongis rotundatis 5-7 mm. longis latisque; staminibus glabris tubo corollae medium versus affixis, filamentis crassis leviter complanatis circiter 5 mm. longis, antheris oblongis circiter 4 mm. longis utroque rotundatis; disco crasse carnoso glabro annulari-cupuliformi 1.5-2 mm. alto margine integro; ovario graciliter ovoideo glabro, stylo crasso tereti sub anthesi 10-15 mm. longo superne pilis pluriseptatis 0.2-0.5 mm. longis capitato-glandulosis copiose piloso, stigmate subpeltato bilobato: fructibus juvenilibus (calyce caduco) elongato-ovoideis apice acutis. Vitt Levu: Mba: Upper western slopes of Mt. Tomanivi [ Mt. Victoria |, alt. 1100-1150 m., Sept. 6, 1947, Smith 5913 (A TYPE, US) (simple-stemmed shrub 2 m. high, in dense forest; bracts greenish white: corolla and young fruit white): Naitasiri: Northern portion of Rairaimatuku Plateau. between Mt. Tomanivi and Nasonggo, alt. 870-970 m., Smith 5792 (A. US) (simple-stemmed shrub 1-2 m. high, in dense forest; bracts white). The species here described closely resembles in foliage C. vitiensis Seem. and C. amicta, above described, and is probably also referable to § Decur- rentes. It is readily characterized by its large subcapitate inflorescence surrounded by coalescent and copiously pilose outer bracts, by having its large calyx densely long-sericeous on both sides. and by its copiously 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS. XIV 41 sericeous corolla, which is densely glandular-pilose within. An examined isotype of C. vitiensis (Seemann 277, from the vicinity of Namosi) at the Gray Herbarium has lost the outer bracts of the inflorescence, and so a comparison on this score cannot be made. However, C. occulta differs ob- viously from C. vitiensis in its densely pilose calyx. From C. amicta the new species is readily separated by its large coalescent bracts, its more copious inflorescence-indument, its shorter calyx, and the glandular in- dument of the inner surface of its corolla, as well as by obvious differences in leaf-texture, surface, and venation. Cyrtandra leucantha sp. nov. Frutex ad 3 m. altus, ramulis gracilibus apices versus subquadrangu- laribus 3-5 mm. diametro pilis brunneis pluriseptatis 0.5—-0.8 mm. longis patentibus copiose indutis demum subglabratis; foliis oppositis apices ramulorum versus subcongestis, petiolis gracilibus leviter canaliculatis vel semiteretibus 3—6 cm. longis ut ramulis juvenilibus pilosis, laminis in sicco papyraceis supra fusco-viridibus subtus pallidioribus, oblanceolatis vel obovatis, 20-37 cm. longis, 7.5—14 cm. latis, inferne gradatim angustatis et in petiolum longe decurrentibus, apice breviter acuminatis vel cuspidatis. margine copiose sed irregulariter calloso-serratis, supra pilis multiseptatis 1-1.5 mm. longis copiose patenti-pilosis, margine pariter ciliatis, subtus etiam copiose pilosis (pilis nervorum similibus, eis facie pallidioribus inter- dum unicellularibus 0.3-0.5 mm. longis), costa et nervis secundariis utrin- secus 9-13 arcuato-adscendentibus supra subplanis subtus prominentibus vel valde elevatis, nervis tertiariis et rete venularum intricato supra im- mersis subtus prominulis; inflorescentiis axillaribus cymosis multifloris congestis, bracteis liberis papyraceis ellipticis vel suborbicularibus, 6-8 mm. diametro, apice rotundatis, margine distali undulatis, utrinque glabris vel basim versus obscure pilosis, pedunculo brevi, pedicellis gracilibus sub anthesi 6-9 mm. longis glabris; calyce submembranaceo cylindrico-cam- panulato, sub anthesi 16-17 mm. longo, basim versus circiter 5 mm. diametro superne ad 8-9 mm. diametro gradatim ampliato, extus praeter basim versus pilis numerosis patentibus 0.5-1 mm. longis (stipitibus 3—6-septatis gracilibus apice capitato-glandulosis raro eglandulosis) prae- dito, intus glabro basi sericeo excepto, lobis 5 suberectis subaequalibus deltoideis circiter 3 X 4 mm. apice obscure calloso-mucronulatis; corolla membranacea cylindrica sub anthesi circiter 20 mm. longa et basim versus 4 mm. diametro, faucibus paullo ampliata, extus superne ut calyce glandu- loso-pilosa, intus glabra, lobis 5 subaequalibus patentibus suborbicularibus circiter 4 mm. diametro; staminibus tubo corollae supra medium affixis, filamentis gracilibus 2-3 mm. longis, antheris ellipsoideis circiter 3 mm. longis; disco glabro breviter tubuloso circiter 1.5 mm. alto; ovario graciliter ovoideo glabro, stylo crasso circiter 8 mm. longo parce glanduloso-piloso, stigmate complanato bifido; fructibus ellipsoideis maturis circiter 15 mm. longis in sicco rugulosis, apice styli basi apiculatis, calycis limbo et disco mox caducis sed pilis calycis basis subpersistentibus. 42 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxrv Vitt Levu: Mba: Western slopes of Mt. Nanggaranambuluta [ Lomalangi], east of Nandarivatu, alt. 850-1000 m., Smith 4767 (A. US) (mbeta kai: simple shrub 1-1.5 m. high, in dense forest; fruit white), 6312 (A type, US) (Oct. 2, 1947; shrub 3 m. high, in dense forest; calyx and corolla white); vicinity of Nandarivatu, alt. 800-900 m., Degener & Ordonez 13523 (A, US) (shrub to 2 m. high, in dark wet forest; calyx and corolla white): Nandala, near Nandarivatu, Degener 14836 (A, US) (mbeta; sparingly branched shrub 2 m. high, in dense forest; bracts white). The last two specimens cited were originally distributed as C. glandulosa Gillespie, a species similar in the indument of its calyx and corolla. Further examination indicates that these plants do not represent C. glandulosa, which, however, appears to occur in my present collection in no. 4129 (A, US), from the Mt. Evans Range of northwestern Viti Levu. The new species here described differs from C. glandulosa in the much shorter indu- ment of its branchlets and leaves. in having its leaf-blades long-attenuate at base and the petiole comparatively short and slender, in its much smaller inflorescence-bracts, its comparatively small flowers, its white rather than yellow corolla, and its caducous calyx. The fact that the calyx of C. leucan- tha is not saccate and persistent would, in Clarke’s system, remove it from the alliance of C. glandulosa. A closer relative of the new species is per- haps C. desvoeuxii Horne ex Clarke (represented in my present collection by no. 5101 [A, US], from Mt. Tomanivi), which Clarke placed in his S Aureae. Cyrtandra desvoeuxii is a very robust species, with flowers ap- proaching those of C. glandulosa in size. It is possible that C. glandulosa also belongs in § Aureae rather than in § Campanulaceae, where it was placed by Gillespie. Cyrtandra chlorantha sp. nov. Frutex ad 4 m. altus, ramulis gracilibus obscure quadrangularibus apices versus 3-4 mm. diametro pilis patentibus 2-4 mm. longis multiseptatis brunneis eglandulosis (vel interdum capitato-glandulosis) copiose indutis demum subglabratis; foliis oppositis, petiolis gracilibus leviter canaliculatis 3—4 cm. longis ut ramulis juvenilibus pilosis, laminis in sicco papyraceis supra fusco-viridibus subtus pallidioribus, lanceolatis vel anguste ellipticis, 14-19 cm. longis, 5—8 cm. latis, basi attenuatis et in petiolum longe decur- rentibus, apice cuspidatis, margine dentibus irregularibus 1 vel 2 per cen- timetrum calloso-serratis, supra dispersim pilosis (pilis multiseptatis 1.5— 3 mm. longis), subtus similiter pilosis vel pilis faciei brevioribus, margine pariter ciliatis, costa supra leviter elevata subtus prominente, nervis secun- dariis utrinsecus 7-9 adscendentibus supra planis subtus elevatis, rete venularum supra subimmerso subtus prominulo; inflorescentiis axillaribus cymosis paucifloris, pedunculo ad 1 cm. longo subglabrato, bracteis liberis papyraceis ellipticis vel ovatis, 12-16 mm. longis, subacutis, margine undulatis, utrinque pilis multiseptatis 1-2 mm. longis plerumque capitato- glandulosis parce patenti-pilosis, pedicellis gracilibus sub anthesi 5-8 mm. longis superpe ut calyce pilosis; calyce submembranaceo campanulato sub anthesi 15-18 mm. longo et basim versus 5 mm. diametro superne ad 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 43 10 mm. diametro ampliato, extus pilis 2-3 mm. longis pluriseptatis capitato- glandulosis (glandulis interdum caducis) copiose et uniformiter patenti- piloso, intus similiter sed parcius pilosis, lobis 5 inaequalibus deltoideis 2—5 mm. longis obtusis; corolla membranacea cylindrica viridi sub anthesi circiter 22 mm. longa et basim versus 5 mm. diametro, extus ut calyce copiose et subtiliter glanduloso-pilosa, intus glabra, apice in lobos 5 sub- aequales late imbricatos suborbiculares circiter 5 mm. diametro abrupte incrassata; staminibus tubo corollae supra medium affixis, filamentis cir- citer 5 mm. longis, antheris ellipsoideis circiter 3.5 mm. longis; disco cupuliformi glabro circiter 1 mm. alto; ovario ovoideo glabro, stylo circiter 0 mm. longo pilis ad 0.5 mm. longis glanduloso-piloso, stigmate bifido. e Levu: Naitasiri: Northern portion of Rairaimatuku Plateau, between Tomanivi [Mt. Victoria] and Nasonggo, alt. 870-970 m., Aug. 21, 1947, Smith 5789 (A TYPE, US) (shrub 3-4 m. high, in dense forest; bracts and calyx pale green; mature corolla bright green; anthers white; young fruit green). The new species is of the general relationship of C. leucantha, described above, and for the present appears best referred to § Aureae. Cyrtandra chlorantha is readily distinguished from C. leucantha by its longer vegeta- tive indument, smaller leaves, larger and pilose inflorescence-bracts, the markedly longer hairs of its calyx and corolla, and the striking bright green color of the latter. With C. desvoeuxii Horne ex Clarke and possibly C. glandulosa Gillespie, the two new species here described make up a group readily characterized, among the Fijian species, by their glandular- pilose flowers. Cyrtandra bracteolosa sp. nov. Arbor gracilis ad 5 m. alta, habitu partibus juvenilibus minute et evanescenter furfuraceis ee glabra, ramulis gracilibus apices versus subquadrangularibus 2-3 mm. diametro inferne subteretibus cinerascen- tibus, internodiis 1.5—5 cm. longis; foliis oppositis, petiolis gracilibus leviter canaliculatis 1-2.5 cm. longis, laminis chartaceis in sicco viridibus, lanceo- latis, (11—) 14-19 cm. longis, (3.5—) 4-6 cm. latis, basi attenuatis et in petiolum decurrentibus, apice gradatim longe acuminatis, margine dentibus 1—2 cm. remotis calloso-apiculatis manifeste undulato-crenatis, costa supra paullo elevata subtus prominente, nervis secundariis utrinsecus 5—8 arcuato- adscendentibus supra subplanis subtus elevatis marginem versus inter- connexis, rete venularum laxo supra immerso subtus prominulo; inflores- centiis e nodis defoliatis orientibus cymosis, aliquot dichotome ramulosis, ad 6 cm. longis, paucifloris (floribus plerumque mox caducis), pedunculo ramulisque teretibus gracillimis (0.4-0.7 mm. diametro), pedunculo 10— 25 mm. longo, internodiis 7-15 mm. longis; bracteis bracteolisque ad nodos Sanne abi pene vel submembranaceis oblongo-linearibus, 4— 7 mm. longis, 0.8—1.2 mm. latis, obtusis 1-nerviis, distalibus dorso saepe obscure furfuraceis; nee é nodis apicalibus solitariis, pedicello tereti 3-5 mm. longo; calyce membranaceo cupuliformi 4-5 mm. longo demum subrotato caduco, manifeste nervato, ad medium 5-lobato, lobis subaequali- 44 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv bus late deltoideis subacutis; corolla non visa; fructibus juvenilibus ellip- soideis levibus, stylo crasso circiter 5 mm. longo, stigmate incrassato bilobato. im1 Levu: Mba: Hills east of Bare Creek, about 3 miles south o Nandarivatu, alt. 850-970 m., Sept. 25, 1947, Smith 6232 (A TypE, US) ee tree 5 m. high, in dense forest : young ae green). The cited specimen, although lacking corollas, seems definitely to rep- resent a new species related to the species of § Polynesieae with freely branching inflorescences; this group in Fiji includes C. denhamii Seem.., C. ciliata Seem., C. taviunensis Gillespie, and perhaps C. montana Gillespie. The new species differs from all of these in its narrow, linear bracteoles. These bracteoles bear a superficial resemblance to the calyx-lobes of C. ciliata, but actually the calyces of the two species are very dissimilar, and foliage differences are also pronounced. In comparison with C. brac- teolosa, C. taviunensis has much larger bracteoles and calyx, while C. mon- tana has the leaves and bracteoles considerably smaller and the calyx larger. Cyrtandra denhamii, apparently the closest ally of the new species, is very imperfectly known, but it is described as having an elongate peduncle, oblong bracteoles about 1 cm. long, and a subspathaceous calyx. Cyrtandra trichophylla sp. nov. Arbor gracilis ad 6 m. alta, ramulis apices versus subteretibus 3-6 mm. diametro pilis pallidis ad 1.5 mm. longis obscure pluriseptatis copiose indutis, demum incrassatis ad 15 mm. diametro glabratis cinereis rugulosis; foliis oppositis, petiolis crassis (3-5 mm. diametro) leviter canaliculatis 1—2.5 cm. longis ut ramulis juvenilibus pilosis glabratisque, laminis in vivo carnosis in sicco subcoriaceis fuscescentibus, obovatis vel obovato-ellipticis, (12—) 14-28 cm. longis, (4-) 5.5—11 cm. latis, basi acutis vel attenuatis et in petiolum decurrentibus, apice breviter acuminatis vel cuspidatis, mar- gine dentibus 1-3 per centimetrum inconspicue calloso-serratis, supra glabris (juventute dispersim sericeis), subtus pilis subtilibus pallidis 0.8— 1.5 mm. longis obscure pluriseptatis densissime molli-sericeis, costa supra plana vel leviter elevata subtus prominente, nervis secundariis utrinsecus 7-11 adscendentibus vel erecto-patentibus supra planis subtus valde eleva- tis, rete venularum utrinque immerso; inflorescentiis trunco vel ramulis infra folia enatis vel interdum axillaribus, simpliciter cymosis, plerumque 3(interdum 2- vel 4-)-floris, pedunculo pedicellisque gracilibus longitu- dine subaequalibus (vel pedunculo interdum subnullo) sub anthesi 10-- 15 mm. longis copiose pilosis (pilis subpatentibus pluriseptatis 1-2 mm. longis) demum subglabratis; bracteis apice pedunculi in involucrum papy- raceum monophyllum coalitis ad 10 mm. longis latisque, apice obtusis, utrinque ut pedicellis copiose pilosis (indumento intus_persistentiore) demum deciduis; calyce subcarnoso campanulato sub anthesi 15-17 mm. longo et basim versus 6-7 mm. diametro superne ad 10-12 mm. ampliato utrinque (intus saepe densius) pilis eis pedicelli similibus copiose sericeo, lobis 5 subaequalibus deltoideo-lanceolatis acutis 5—7 mm. longis; corolla 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 45 infundibulari sub anthesi circiter 25 mm. longa, tubo subcarnoso basim versus 3—5 mm. diametro faucibus ampliato extus superne ut calyce piloso intus glabro, lobis 5 subaequalibus sub anthesi patentibus submembranaceis oblongo-suborbicularibus 7-8 mm. diametro intus inconspicue pilosis (pilis debilibus 0.3-0.5 mm. longis stipitatis capitato-glandulosis); staminibus apicem tubi corollae versus affixis, filamentis 2-3 mm. longis, antheris ellipsoideis circiter 2 mm. longis: disco carnoso breviter cylindrico glabro 1—2 mm. alto; ovario anguste ellipsoideo glabro, stylo crasso circiter 10 mm. longo superne parce glanduloso-piloso, stigmate complanato obscure bilo- bato; fructibus ellipsoideis in vivo ad 4 2.5 cm. in sicco rugulosis. calyce caduco. Vitt Levu: Ra: Ridge from Mt. Namama (east of Nandarivatu) toward Mt. Tomanivi [Mt. Victoria], alt. 1050-1120 m.. Aug. 18, 1947, Smith 5698 (A TYPE, US) (tree 5 m. high, in dense forest; corolla and stamens white; truit when fresh white, ellipsoid, about 4 x 2 cm.. greatly shrinking in drying); Mba: Hills east of Nandala Creek, about 3 miles south of Nandarivatu, alt. 850-970 m.. Smith 6246 (A, US) (slender tree 6 m. high, in dense forest; calyx green, with brown pubescence; corolla and anthers white: fruit white. when fresh about 4 2.5 cm.. greatly shrinking in drying). A species of § Polynesieae, the new species forms, with C. involucrata Seem. and C., gillespieana A. C. Sm. |C. monticola Gillespie |. a group char- acterized by simple inflorescences with the two bracts joined to form a cup-like involucre. Cyrtandra trichophylla differs from its relatives in the prevailingly obovate, larger, and thick-textured leaves, of which the lower surface is very densely sericeous with long, pale, obscurely septate hairs. and in its comparatively large calyx and corolla, the former being copiously sericeous on both sides. Cyrtandra esothrix sp. nov. Frutex vel arbor gracilis ad 6 m. alta, partibus novellis copiose sed minute brunneo-puberulis, ramulis gracilibus apices versus subquadrangu- laribus demum glabratis, internodiis 1-2.5 (—4) cm. longis; foliis oppositis. petiolis gracilibus leviter canaliculatis vel semiteretibus 1-3 (—3.5) cm. longis ut ramulis puberulis glabratisque, laminis papyraceis in sicco viridi- bus vel fusco-viridibus, lanceolatis vel obovato-ellipticis, 10-18 cm. longis. 3-6 cm. latis, basi attenuatis et in petiolum decurrentibus, apice acuminatis et calloso-apiculatis, margine dentibus 1-3 per centimetrum calloso-mucro- nulatis conspicue serratis, supra glabris vel nervis parce puberulis, subtus nervis (et plerumque facie) primo pilis 0.1-0.3 mm. longis copiose puberu- lis demum subglabratis, costa supra subplana subtus prominente, nervis secundariis utrinsecus 7 vel 8 adscendentibus supra planis subtus elevatis. rete venularum laxo subtus prominulo; inflorescentiis cymosis axillaribus etiam infra folia enatis breviter pedunculatis plerumque unifloris raro bifloris, bracteis bracteolisque oppositis inconspicuis deltoideis vel oblongis 1.5-2.5 mm. longis obtusis mox caducis, pedicellis gracilibus sub anthesi 5-15 mm. longis cum pedunculo bracteisque ut ramulis juvenilibus puberu- 46 JOURNAL OF THE ARNOLD ARBORETUM - [voL. xxxiv lis demum glabratis; calyce subcarnoso in alabastro ovoideo et areal mucronulato, sub anthesi 8-12 mm. longo in lobos 2 vel 3 ovatos 3-4 m longos cuspidatos irregulariter rumpente, extus glabro vel basim versus parce puberulo, intus pilis crassis adscendentibus fulvis subrigidis 3—5- septatis 0.8-1.5 mm. longis copiose sericeo, mox caduco; corolla mem- branacea cylindrica inconspicue curvata ubique glabra sub anthesi ad 22 mm. longa, basim versus 5—6 mm. diametro superne ampliata, lobis 3 inferioribus rotundato-oblongis 5-6 mm. longis, 2 superioribus paullo minoribus; staminibus glabris tubo supra medium insertis, filamentis circiter 3 mm. longis, antheris oblongis circiter 2 mm. longis utroque rotun- datis; disco carnoso annulari glabro circiter 1.5 mm. alto; gynoecio glabro, ovario subcylindrico-ovoideo, stylo crasso sub anthesi circiter 4 mm. longo demum elongato, stigmate subcapitato; fructibus anguste ovoideis ad 2 cm. longis in sicco rugulosis. Vitt Levu: Naitasiri: Northern portion of Rairaimatuku Plateau, between Mt. Tomanivi [| Mt. Victoria] and Nasonggo, alt. 870-970 m., Sept. 18, 1947, Smith 6134 (A type, US) (slender tree 4-6 m. high, forming dense thickets in forest; calyx white in bud, at length pale green; corolla and young fruit white); a: Western and southern slopes of Mt. Tomanivi, alt. 850-1150 m., Smith 5098, 5267 (A, US) (shrubs 3-4 m. high, in dense forest; calyx pale green; corolla and fruit white). A member of § Polynesieae, C. esothrix appears to be most closely re- lated to C. coleoides Seem., a species with the same type of irregularly splitting calyx and a similar, but sparser, vegetative indument. The new species differs from C. coleoides most obviously in having the calyx copi- ously sericeous within (rather than glabrous), the several-celled hairs con- spicuously protruding as soon as the calyx-bud breaks, and in having the flowers usually one per inflorescence (rather than several). Cyrtandra coleoides is a not infrequent species in upland Viti Levu, being represented in my present collection by nos. 5259, 5628, and 5813: its leaves are slightly larger than those of the new species and have the margins merely undulate rather than sharply serrate, while the indument of the nerves on the lower surface is evanescent. Cyrtandra greenwoodiana sp. nov. Arbor vel frutex ad 6 m. altus, ramulis gracilibus sub anthesi subquad- rangularibus 1-3 mm. diametro et pilis fusco-brunneis pluriseptatis 0.1— 0.3 mm. longis copiose et arcte indutis, mox glabratis subteretibus, in- ternodiis 1-2 cm. longis; foliis oppositis, petiolis gracilibus semiteretibus (0.7—) 1-3 cm. longis ut ramulis pilosis glabratisque, laminis in sicco chartaceis saepe viridibus subtus pallidioribus, oblongo-ellipticis, (4—) 6-15 cm. longis, (2—) 4—6 cm. latis, basi obtusis, apice abrupte cuspidatis vel breviter acuminatis, margine dentibus irregularibus 1 vel 2 per cen- timetrum apiculatis manifeste crigr metal serratis, supra glabris, subtus costa nervisque secundariis ut petiolis (saepe sparsim) pilosis alioqui glabris, costa et nervis secundariis ares 5-8 curvatis patentibus supra sub- 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 47 planis subtus valde elevatis, rete venularum subimmerso vel subtus laxe prominulo; inflorescentiis axillaribus unifloris, pedunculo gracili 5-12 mm. longo primo ut ramulis juvenilibus arcte piloso sub fructu subglabrato, bracteis apice pedunculi binis minutis lanceolatis 1-2 mm. longis arcte strigillosis caducis, pedicello pedunculum subaequante vel sub fructu ad 15 mm. longo similiter strigilloso; calyce submembranaceo inaequilater- aliter campanulato, in alabastro clauso et conspicue caudato-apiculato, sub anthesi irregulariter fisso saepe subspathaceo, 15-16 mm. longo, 5—7 mm. diametro, utrinque pilis eglandulosis pluriseptatis (extus 0.2-0.3 mm. longis caducis, intus ad 0.7 mm. longis persistentibus) induto, lobis 5 lanceolatis 5-7 mm. longis, saepe omnino ad apicem connatis, interdum 1 vel 2 liberis aliis connatis, calyce e basi mox caduco; corolla membranacea infundibulari sub anthesi lobis inclusis 20-25 mm. longa et basim versus 3-5 mm. diametro superne ampliata, tubo utrinque glabro, lobis 5 sub- aequalibus oblongo-suborbicularibus circiter 5 mm. longis intus subtiliter et minute glanduloso-pilosis; staminibus in faucibus corollae insertis, fila- mentis 2—3 mm. longis, antheris ellipsoideis longitudine aequalibus; disco annulari-pulvinato haud 0.5 mm. alto; ovario anguste ellipsoideo glabro, stylo crasso circiter 7 mm. longo obscure glanduloso-piloso, stigmate in- crassato bilobato; fructibus ellipsoideis in vivo circiter 2.5 cm. longis in sicco rugulosis, basi styli subpersistente Vit1 LEvu: Mba: Mt. Nairosa, eastern flank of Mt. Evans Range, alt. about 1050 m., May 14, 1947, Smith 4412 (A Type, US) (tree or shrub 4-6 m. high, in crest thickets at base of ultimate pinnacle; corolla and fruit pure white, the mature fruit about 2.5 < 1.5 cm., shrinking in drying); northern portion of Mt. Evans Range, between Mt. Vatuyanitu and Mt. Natondra, alt. 700-900 m., Smith 4304 (A, US) (slender tree 5 m. high, in dense forest; calyx pale green; corolla and fruit white); Mt. Evans Range, alt. about 970 m., Greenwood 1260 (US) (shrub 3-4 m. high, in thick forest; flowers white). Cyrtandra greenwoodiana is characterized by the close (and often sparse) indument of its vegetative parts and inflorescence, its one-flowered inflores- cence with minute bracts subtending the solitary pedicel, and its large, unevenly cleft, often one-sided and subspathaceous calyx, which is densely short-pilose within. A member of § Polynesieae, its closest ally seems to be C. pritchardii Seem., from which it differs in its more obvious indument, its slightly smaller eaves its consistently one-flowered inflorescence, and its much larger calyx and corolla, The calyx of C. pritchardii is only about 7 mm. long and is subequally 5-lobed, and its corolla is about 16 mm. long. It is probable that some of the specimens from central Viti Levu identified as C. pritchardii will be better referred to the new species, but a final circumscription of Seemann’s species (type from Ovalau) may be delayed until the genus is revised for Fiji. Another species of this rela- tionship, C. spathacea A. C. Sm. (type from Kandavu), is essentially glabrous throughout and has much larger, many-nerved leaves. It is a pleasure to name the new species for Mr. William Greenwood, a long-time resident of Fiji, who has collected many unusual species on the Mt. Evans Range of northwestern Viti Levu. 48 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxx1v Cyrtandra harveyi Seem. Fl. Vit. 182. 1866: C. B. Clarke in DC. Monogr. Phan. 5: 281. 1883. Vanua Levu: Mathuata: Southern base of Mathuata Range, north of Natua, Smith 6761 (A, US); slopes of Mt. Numbuiloa, east of Lambasa, Smith 6343 (A, US), 6529 (A, US). Cyrtandra harveyi (§ Polynesieae) is one of the most frequent species of the genus in Fiji at low elevations, occurring from near sea-level up to 500 m. in the forest undergrowth, especially on Vanua Levu. In addition to the recently collected material above cited it is represented by: Vanua Levu: Harvey (type at Kew, isotype at Gray Herbarium), Smith 367. 1847, Degener & Ordonez 13899, 13969, 14088; Viti Levu: B. E. Parham 17, Tabualewa 15596. The specimens are from shrubs or slender trees 1-4 m. high; the corolla is cream white to pale vellow, and the mature fruit is white and about 2 x 1 cm., with a quickly caducous calyx. Although the available specimens show some variability in leaf-size and marginal serration, there seems no doubt that they represent the same well circumscribed species. The branchlets, petioles, pedicels, and calyx are copiously and closely sericeous with dark golden subascending (less commonly subspreading) several-celled hairs 0.2-1 mm. long, both leaf- surfaces being similarly pilose but eventually subglabrate; the inflorescence is axillary or lateral below the leaves, openly cymose, pedunculate, incon- spicuously bracteate, 2—4-flowered; the calyx is closed and apiculate in bud, eventually 10-12 mm. long and unequally (sometimes deeply) 5- lobed; the corolla is 20-25 mm. long, subequally 5-lobed, copiously to sparsely spreading-pilose with several-celled gland-tipped hairs 0.1—0.5 mm. long; the stamens are highly placed on the corolla: and the style often equals the corolla in length, having scattered glandular hairs distally. Cyrtandra xanthantha sp. nov. Frutex ad 4 m. altus, ramulis subteretibus vel obscure quadrangularibus apices versus 1.5-3 mm. diametro et pilis patentibus fusco-brunneis pluri- septatis 0.8-1.5 mm. longis copiose indutis, demum glabratis, internodiis brevibus 5-8 mm. longis, nodis subincrassatis: foliis oppositis, petiolis gracilibus basim versus canaliculatis 5-15 mm. longis ut ramulis juvenil- ibus patenti-pilosis, laminis chartaceis in sicco fusco-viridibus lanceolatis. (8—) 11-18 cm. longis, (3—) 4—5.5 cm. latis, basi acutis sed haud decur- rentibus, apice calloso-acuminatis, margine dentibus circiter 2 per cen- timetrum calloso-apiculatis inconspicue denticulatis. utrinque pilis plurisep- tatis 0.5-1 mm. longis dispersim patenti-pilosis, costa nervisque secundariis utrinsecus 7—9 subadscendentibus supra subplanis subtus valde elevatis. rete venularum supra immerso subtus prominulo; inflorescentiis axillaribus simpliciter cymosis 3—4-floris, pedunculo pedicellisque longitudine sub- aequalibus 5-10 mm. longis gracilibus ut ramulis juvenilibus copiose pilosis. indumento pedicelli pallido, bracteis mox caducis non visis: calyce mem- branaceo campanulato sub anthesi 5—6 mm. longo, extus ut pedicellis piloso (pilis pallidis 0.5-1.3 mm. longis multiseptatis eglandulosis), intus glabro. 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 49 fere ad basim 5-lobato, lobis deltoideis subacutis, 3 interdum altiuscule connatis; corolla membranacea infundibulari sub anthesi lobis inclusis 22- 25 mm. longa et basi 4-5 mm. diametro superne leviter ampliata, extus pilis subtilibus pluriseptatis 0.2-0.4 mm. longis capitato-glandulosis copiose sed inconspicue induta, intus glabra, lobis 5 subaequalibus oblongo-subor- bicularibus 4-6 mm. longis; staminibus apicem corollae tubi versus insertis, filamentis 2-3 mm. longis, antheris ellipsoideis circiter 2 mm. longis; disco carnoso cupuliformi glabro circiter 1.5 mm. alto; ovario anguste ellip- soideo glabro, stylo 7-10 mm. longo subtiliter glanduloso-piloso, stigmate incrassato bifido; fructibus immaturis ellipsoideis ad 1.5 cm. longis rugu- losis styli basi persistente coronatis. Vitt Levu: Mba: Slopes of the escarpment north of Nandarivatu, alt. 550— 800 m., Sept. 29, 1947, Smith 6277 (A Type, US) (shrub 3-4 m. high, in woods along stream; calyx white; corolla pale yellow, the lobes at first greenish, then yellow). Cyrtandra xanthantha is a species of § Polynesieae, characterized by its simple, few-flowered inflorescence with as bracts, its comparatively small and deeply 5-lobed calyx being eglandular-pilose without and gla- brous within, and its corolla being 22-25 mm. long and finely glandular- pilose without. Its closest relatives seem to be C. harveyi Seem. and C pritchardii Seem. From C. harveyi it differs in the more obvious and more definitely spreading indument of its vegetative parts and inflorescence, its smaller and comparatively short-petiolate leaves, and its smaller, more deeply lobed calyx. Cyrtandra pritchardii is a more distant ally, differing obviously from the new species in its closer and sparser indument (essen- tially lacking on the flowers) and its longer-petiolate, more conspicuously toothed leaves. Cyrtandra jugalis sp. nov. Frutex vel arbor parva ad 5 m. alta, ramulis gracilibus subteretibus vel obscure quadrangularibus apices versus copiose pilosis (pilis patentibus fuscis 0.4-0.7 mm. longis 3—7-septatis) inferne glabratis; foliis oppositis, petiolis leviter canaliculatis (5—) 10-25 (-30) mm. longis ut ramulis juvenilibus pilosis superne angulatis haud alatis, laminis chartaceis in sicco fusco-viridibus subtus saepe pallidioribus, lanceolatis vel lanceolato-ellip- ticis, (6—) 10-19 cm. longis, (1.5—) 2.5-5 (—6) cm. latis, basi acutis vel subattenuatis, apice gradatim acuminatis et calloso-apiculatis, margine den- tibus 2 vel 3 per centimetrum calloso-mucronulatis conspicue et irregu- lariter serratis, supra glabris, subtus molliter pilosis (pilis fusco-aureis 0.3-0.5 mm. longis inconspicue pluriseptatis saltem nervis longe persisten- tibus), costa supra subplana subtus prominente, nervis secundariis utrin- secus 5-8 longe adscendentibus supra planis subtus inconspicue elevatis, rete Suleman laxo subimmerso; inflorescentiis axillaribus vel ramulis defoliatis enatis simpliciter cymosis plerumque 3-floris pedunculatis, pedun- culo gracili 2-6 cm. longo raro subnullo cum pedicellis bracteisque ut ramulis juvenilibus piloso (pilis patentibus vel subadscendentibus 0.2— 50 JOURNAL OF THE ARNOLD ARBORETUM - [vot. xxxiv 0.7 mm. longis) ; bracteis apice pedunculi binis ovato-lanceolatis, 5-13 mm. longis, 1.5-4 mm. latis, subacutis, basi non vel haud connatis, demum caducis; pedicellis gracilibus sub anthesi et fructu (8—) 15-25 mm. longis; calyce subcarnoso in alabastro ovoideo longe apiculato, sub anthesi cam- panulato 10-14 mm. longo apice ad 12 mm. diametro utrinque ut pedicello piloso, ad medium 5-lobato, lobis subaequalibus deltoideo-lanceolatis acu- minatis 5-8 mm. longis; corolla membranacea infundibulari sub anthesi lobis inclusis 18—24 mm. longa basi 2.5—4 mm. diametro superne ampliata. extus pilis pallidis pluriseptatis 0.4-1 mm. longis molliter patenti-pilosa, tubo intus glabro, lobis 5 subaequalibus suborbicularibus sub anthesi patentibus 5-8 mm. diametro intus copiose glanduloso-pilosis (pilorum stipitibus pluriseptatis 0.1-0.3 mm. longis); staminibus apicem corollae tubi versus insertis, filamentis crassis 1-2 mm. longis, antheris oblongis 2—3 mm. longis utroque rotundatis; disco carnoso glabro pulvinato-cupuli- formi 1—1.5 mm. alto apice undulato; ovario elongato-ovoideo glabro, stylo crasso tereti sub anthesi 4-8 mm. longo ut corollae lobis glanduloso- piloso, stigmate subpeltato; fructibus calyce mox caduco in vivo ellipsoideo ad 1.5 & 1 cm. in sicco ruguloso. Vit1 Levu: Mba: Summit of Mt. Nanggaranambuluta | Lomalangi], east of Nandarivatu, Smith 4878 (A, US); ridge between Mt. Nanggaranambuluta and Mt. Namama, alt. 1050-1120 m., June 30, 1947, Smith 4994 (A typE, US); western slopes of Mt. Tomanivi [Mt. Victoria], Smith 5319 (A, US); Ra: Ridge from Mt. Namama toward Mt. Tomanivi, Smith 5695, 5696, 5709, 5715 (all A, US); Nandronga & Navosa: Vicinity of Nandrau, Degener 14904 A); northern portion of pera taara i Plateau, between Nandrau and Nanga, Smith 5569 (A. US). The cited specimens are from shrubs or slender. freely branching trees 2—5 m. in height, occurring in dense forest at elevations of 725-1120 m. The corolla, stamens, and style are pure white to cream-white, and the mature fruit is white. Recorded local names are mindra (Smith 4994) and mbetambeta (Degener 14904). The described species, of § Polynesieae, is quite frequent on the long ridge that curves from Nanggaranambuluta to Tomanivi and dominates north-central Viti Levu. It seems strange that Gillespie did not obtain material of the species, or at least that he did not describe it. I have examined type material of most of his species, and all are well described and figured; the only one that suggests C. jugalis is C. prattii, a species with larger and more copiously nerved leaves, an elongate inflorescence with solitary lateral flowers, pedicels only about 4 mm. long, and a some- what larger, unequally lobed calyx. Closer relatives of the new species are C. anthropophagorum Seem. and C. hornei Clarke, both of which have comparatively small flowers. This group of species is characterized by its indument of several-celled eglan- dular hairs and its few-flowered inflorescence with free, opposite, rather small bracts. In characters of indument, C. jugalis is intermediate be- 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XIV 51 tween C. anthropophagorum and C. hornet, the former having more copious and softer pubescence or, both vegetative and inflorescence parts and the leaves less obviously serrate, the latter being soon essentially glabrate on vegetative parts, calyx, and corolla. DEPARTMENT OF BOTANY, U. S. NATIONAL Museo. SMITHSONIAN INSTITUTION. 52 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv A NEW FIJIAN SPECIES OF CALYPTOSEPALUM I. W. BarLey AND A. C. SMITH With two plates AMONG THE PHANEROGAMS collected in Fiji by the junior author in 19471, one specimen from upland central Viti Levu could not, upon pre- liminary study, be referred to any family known to occur in the region. Of the plant in question, Smith 5339, several duplicate herbarium speci- mens and a wood sample from the trunk are available. The herbarium material bears foliage and essentially mature fruits, but one sheet shows very young fruits, with the carpel scarcely developed beyond its flowering condition and with the perianth and disk intact, and also young flower buds. The plant being apparently dioecious, no staminate flowers are avail- able, and there is no trace of stamens or staminodes in the pistillate flower. The specimen clearly represents a genus not previously recorded from Fiji or the adjacent archipelagos. The desirability of applying a name to it has led us to consider all plausible systematic positions for this Fijian plant, and we have reached the conclusion that it is best placed as a congener of Calyptosepalum sumatranum S. Moore. Reasons for this con- clusion and a discussion of the position of the genus follow a formal de- scription of our new species and an analysis of its salient characters. Calyptosepalum pacificum sp. nov. Arbor dioica ad 12 m. alta, ramulis cinereis subteretibus praeter partes novellissimas minute et fugaciter sericeas glabris valde corticeo-lenticellatis; stipulis binis basi petiolorum lateralibus deltoideis acutis circiter 1 mm. longis latisque extus minute fulvo-sericeis mox caducis, cicatricibus minutis; foliis simplicibus alternatis, petiolis nigrescentibus leviter canaliculatis in sicco rugulosis crassis (1.5—-2 mm. diametro) 1—1.5 cm. longis; foliorum laminis coriaceis siccitate brunneis vel fusco-olivaceis, ovato-oblongis vel late ellipticis, (6—-) 7-12 cm. longis, (3—) 3.5—7 cm. latis, basi late obtusis vel subtruncatis et saepe paullo inaequilateralibus, in petiolum subito breviter decurrentibus, apice obtusis vel subacutis, margine integris et leviter incrassatis, utrinque glabris vel interdum pilos paucos dispersos subpersistentes (eis carpelli similes) subtus nervis gerentibus, pinnatinerviis, costa valida utrinque prominente et rotundata, nervis secundariis utrinsecus 5-8 irregulariter arcuato-adscendentibus et copiose anastomosantibus utrinque valde elevatis, basalibus confertis, rete venularum copioso utrinque “Under the auspices of the Arnold Arboretum of Harvard University and the John Simon Guggenheim Memorial Foundation, with the aid of grants from the Penrose Fund of the American Philosophical Society and the Bache Fund of the National Academy of Sciences. 1953] BAILEY & SMITH, CALYPTOSEPALUM 53 prominulo; floribus (¢ solis visis) pluribus (plerumque 2-5, interdum cum aliis abortivis), e glomerulis parvis irregularibus natis, glomerulis axillaribus vel supra cicatrices in ramulis defoliatis dispositis; bracteolis sub floribus paucis (videtur 2 basi pedicelli oppositis) late reniformibus, circiter 0.5 mm. longis, 1.5—2 mm. latis, extus obscure sericeis, margine ro- tundato minute ciliolatis; pedicellis rectis teretibus in sicco rugulosis paullo post anthesin 4-6 mm. ones: parce strigilloso-puberulis demum glabratis, apice in receptaculum complanatum 2—3 mm. diametro subito incrassatis, sub fructu ad 2 mm. diametro et 8 mm. longis; perianthio tepalis 4 decus- satis composito, tepalis reniformi-ovatis apice rotundatis, 2.5—-3 mm. longis, 3-3.5 mm. latis, e basi plurinervatis, glabris, 2 exterioribus papyraceis, 2 juteroabus membranaceis, omnino evanescentibus, cicatricibus linearibus vel transverse ellipticis inconspicuis; disco annulari carnoso glabro, primo subpatente haud 0.5 mm. alto apice truncato, demum rotato et margine irregulariter sinuato; staminibus et staminodiis nullis; ovario supero ellipsoideo vel obovoideo uniloculari, extus minute papilloso et copiose piloso (pilis pallide fuscis simplicibus pluriseptatis ad 0.5 mm. longis subadscendentibus vel subappressis, apicem ovarii versus reflexis) ; stigmate subsessili terminali carnoso peltato leviter infundibulari 2-3 mm. diametro, margine ventraliter inciso, mox caduco (raro subpersistente), cicatrice parva rotundata; ovarii loculo solitario, placenta ventrali superne valde incrassata, ovulis 2 collateralibus anatropis, e parte incrassata placentae pendulis; fructibus ellipsoideis vel obovoideis maturitate ad 3 cm. longis et 1.5 cm. latis, basi et apice rotundatis, raro stigmate coronatis, apice plerumque inconspicue cicatricosis, pericarpio coriaceo persistenter piloso, semine solitario (ovulo altero abortivo) in sicco valde contracto, in vivo videtur magno carnoso elongato-obovoideo apicem versus connecto, testa tenui, endospermo copioso, embryone magno. FIJI: Virt tevu: Mba: Valley of Nggaliwana Creek, north of the sawmill at Navai, alt. 725-850 m., July 21, 1947, Smith 5339 (type at Arnold Arboretum, 2 sheets, duplicates at U. S; National Museum, etc.) (donggau; tree 12 m. high, on edge of dense forest; tepals dull yellow; mature fruit orange). The new species differs from C. sumatranum S. Moore, the type and only previously known species of Calyptosepalum, in several obvious characters, although the known flowers, being staminate in one case and pistillate in the other, cannot be too critically compared. The Fijian plant has com- paratively long petioles and leaf-blades that are thicker in texture, pro- portionately broader, obtuse to subtruncate at base, and obtuse or subacute at apex. In contrast, C. swmatranum has its leaf-blades narrowed at base and short-acuminate at apex, with nerves and veinlets less obvious than those of C. pacificum. As regards floral characters which presumably are specific in nature, it may be noted that the disk in C. sumatranum is pilose and the ovary-rudiment (in ¢ flowers) glabrous, whereas in the new species the disk is glabrous and the ovary copiously pubescent. 54 JOURNAL OF THE ARNOLD ARBORETUM © [voL. XxxIv THE FLOWER The several duplicate specimens of the single collection of this Fijian tree bear mature or nearly mature fruits. The type specimen has, in addition, small immature fruits and a few flower buds in early stages of enlargement. A detailed study of the exomorphic and endomorphic char- acters of the flower buds and of the youngest fruits enables one to visualize the general form of the flower at anthesis. It consists of two pairs of fleshy decussate tepals, a glandular-appearing, annular disk, and a single carpel which contains two pendent anatropous ovules and terminates in a broad, here shaped stigma, TEXT-FIcs. /—/. ere are no rudiments of stamens r staminodes and no structures that might be interpreted as vestiges of ae It should be noted in this connection, however, that subsequent to anthesis and during the enlargement of the torus and of the voung fruit, TEXT-FIGURES 1-4. Fic. 1. Young fruit. showing retention of the decussate tepals at an unusually late stage of development. There is a corky scar at the apex of the carpel left by the abscission of the stigma, * 12. Fic. 2. Young fruit, showing retention of the stigma, also disk and corky scars left ig the abscission of the tepals, & 8. Fic. 3. Cleared stigma, showing complex vasculature, < 1 Fic. 4. Longitudinal section of carpel, showing two pendent anatropous ovules, x 10. 1953] BAILEY & SMITH, CALYPTOSEPALUM 55 the disk becomes distorted and broken and gives at times a false impression of being the bases of additional appendage’. Abscission of the tepals and stigma appears to occur shortly subsequent to anthesis, leaving four corky scars at the base of the enlarging carpel and a circular embossed one at its apex. In exceptional cases only are the tepals and stigma retained for a time during subsequent stages of development, TEXxT-Fics. 7 an Each of the four decussate tepals has a branching reticulate venation derived in most cases from three principal, independent vascular strands or traces. The disk is unvascularized, but the carpel in its somewhat enlarged form in the youngest available fruits has a strikingly complex and highly specialized vasculature. In transverse sections cut at the base of such a carpel, there is an outer ring of numerous, small, precociously branching vascular strands, a small, compact, more or less centrally located eustele, and a single, large, detached bundle. The outer bundles extend upward in the wall of the carpel, branching repeatedly and forming a_ basket-like venation of much complexity. The central eustele is in continuity at lower levels with the eustele of the pedicel. It extends upward in the ventral side of the carpel to the level of attachment of the ovules, where it dissociates into a number of principal strands. Two of these curve abruptly downward and ramify in the integuments of the ovules, two extend upward and ramify in the stigma, and four to six curve upward and then downward in the wall of the carpel. The large detached bundle extends upward as a discrete strand in the dorsal side of the carpel. It branches at higher levels and has an extension into the stigma. Evidence from a young flower bud indicates that the differentiation of the ventral eustele and of the discrete dorsal strand is initiated during early stages of the ontogenetic development of the flower, whereas that of the peripheral bundles occurs during subsequent stages of the enlargement of the torus and carpel. There are no vestiges of a suture in the ventral side of the older carpels, but the stigma has a deep cleft in its ventral side, TExT-FIGs. 2 and 3. THE FRUIT One ovule aborts, at least in a majority of cases, and the mature fruit contains a single elongated seed, TExT-FIcs. 5 an The older fruits have no clearly defined outer epidermal layer, but in- stead are clothed externally by a dense mat of short, multicellular papillae. This papillose surface appears to be derived during development of the carpel and the young fruit by repeated periclinal and anticlinal divisions of all of the cells of an originally unicellular epidermal layer. Among the papillose projections are elongated hairs which become more widely spaced as the fruit matures. Both the papillae and the hairs vary considerably in size and form. Among the former are conspicuously capitate and glandular-appearing ones. The hairs are multicellular and frequently are internally septate. They may be thick-walled or thin-walled, straight or markedly undulate, or recurved parallel to the surface which bears them. As seen in sectional view, PLATE-FIG. 7, the pericarp consists of five more 56 JOURNAL OF THE ARNOLD ARBORETUM _ [VOoL. xxxiIv or less clearly defined layers. Passing from without toward the locule these are: (1) the papillose layer referred to in the preceding paragraph, (2) a zune of small, thin-walled parenchymatous cells, (3) a zone composed largely of stone cells, (4) a broad central zone of large, thin-walled parenchymatous cells and (5) a compact zone of thin-walled parenchyma whose cells tend to become elongated or oriented parallel to the inner surface of the pericarp. The vascular bundles are distributed within the TEXT-FIGURES 5—7. Fic. 5. Seed, showing oe x 3.3. Fic. 6. Seed, showing extensive hilum, x 3.3. Fic. 7. Embryo broad central layer and have numerous branches which extend outward toward the sclerenchymatous zone. The proportions and the detailed con- figurations of these layers change during successive stages of the enlarge- men. of the fruit. In particular, the layer of stone cells becomes broader and dunser. Furthermore, the vasculature of the pericarp becomes increas- ingly complex and massive. THE SEED The seed, even in apparently mature fruits, contracts so extensively in drying and becomes so deformed that it appears to have aborted. However, it re-expands upon soaking and resumes its original plump, elongated form, 1953] BAILEY & SMITH, CALYPTOSEPALUM 57 TExT-Fic. 5. It contains a well-formed, normal-appearing embryo with pseudo-palmately veined, basally auriculate cotyledons, TExtT-F1c. 7. The embryo is embedded in a massive endosperm whose thin-walled cells con- tain no visible accumulations of starch, fat or oil. It is the excessive con- traction and re-expansion of these large, delicate cells which produces the pronounced changes in volume of the seed during drying and wetting. The testa is thin and soft and contains an externally conspicuous vasculature which resembles that of the ovules in greatly expanded form, compare TeExtT-Fics. 4 and 5. THE LEAF The simple, entire, pinnately Yeined leaf is coriaceous and glabrous appearing when fully expanded. However, a few hairs tend to persist, from the early juvenile stages, upon the lower surface of the leaf and to be scattered along the principal veins. These hairs resemble the thicker-walled, recurved ones that occur on the carpel and fruit. The base of the young leaf is flanked by a pair of small, triangular, caducous stipules which leave corky scars following abscission. The lower epidermis is composed of angular cells of relatively uniform size. The stomata have no special subsidiary cells, and the guard cells are not extensively subtended by the surrounding epidermal cells. There is a conspicuously differentiated layer of large hypodermal cells which becomes discontinuous at times. The palisade layer consists of several rows of relatively short elements, and the thin-walled spongy mesophyll contains large intercellular spaces. The venation is massive and conspicuous, even the smaller veinlets being jacketed by sclerenchyma and in turn by thin- walled crystal-bearing cells, PLATE-FIG Three traces, related to a trilacunar node of the stem, enter the base of the petiole, where they quickly become associated in a vascular cylinder or eustele, PLATE-FIG. 2, which extends throughout the petiole and the mid-rib of the lamina. Within the pith of this vascular cylinder, in its course through the petiole, there are one to three, commonly two, am- phivasal bundles. The cells of the parenchyma of the petiole, particularly many of the cortical cells and the rays of the secondary phloem, contain crystals of calcium oxalate. Small druses predominate in the phloem and large, single, rhombohedral crystals in the cells of the cortex. The vascular cylinder is jacketed by large, very thick-walled fibers. THE STEM The pith and the eustele in the internodal parts of the stem have a symmetrically four-lobed form as seen in transverse sections, the con- figuration of the tissues being correlated with an approximately one-half phyllotaxy. The median traces of the leaves are situated in two opposite lobes and the lateral traces in the two intervening lobes. The median trace of each leaf consists of two widely separated halves in the subtending 58 JOURNAL OF THE ARNOLD ARBORETUM _ |{voL. xxxiv internodal part of the stem, but commonly divides forming three or four independent strands at the nodal level. Further divisions of these strands, as also of the lateral traces, occur precociously in the base of the leaf, where orientation into a foliar eustele takes place. The pith is homogeneous and composed of relatively short cells, many of which contain single rhombohedral crystals of calcium oxalate. The secondary xylem is formed by a primitive type of cambium having elongated fusiform initials with extensively overlapping ends. The wood parenchyma is apotracheal in distribution, PLATE-FIc. 3, varying from diffuse to loosely aggregated tangentially. The relatively thin-walled. angular vessels occur singly and in small radial clusters. The long, slender vessel members have numerous scalariform perforations in their extensively overlapping ends. The intervascular pitting in the tangential facets of adjoining vessels is minute, closely crowded and multiseriate, and exhibits numerous transitions between opposite and alternating seriations. Vestiges of scalariform pitting are of not infrequent occurrence, particularly in the smaller vessels. The pitting between vessels and wood parenchyma or rays is of essentially similar size and form, with, however, numerous evidences of unconformity. The imperforate tracheary elements or “fibers’’ have such excessively thickened walls that the lumina are occluded and the pits are few in number with minute vestigial borders. The first-formed secondary xylem has vertically extensive uniseriate rays composed of tall erect cells. In addition, there are more or less numerous biseriate and triseriate rays composed of more nearly cubical cells or slightly procumbent ones. The wood from the outer part of old stems resembles the secondarv xylem of twigs taken from herbarium specimens, but exhibits the usual increase in cell size that occurs during enlargement of stems. The rays. PLATE-FIG. 5, are lower, the multiseriate (2- to 4-seriate) ones being pro- portionally more numerous and composed of radially more extensive pro- cumbent cells. Many of the parenchymatous cells of both the earlier and the later formed wood contain crystals of calcium oxalate, the rhombohedral form predominating. Prior to the initiation of cambial activity, each bundle of the eustele — as seen in transverse sections of the stem — is capped externally by thick- walled fibers. Subsequently, the entire vascular cylinder becomes enclosed in a composite ring of sclerenchyma, formed in part by the original fibers and in part by stone cells derived from tangentially intervening, thin- walled parenchyma. The secondary phloem, when first formed, is soft and is composed of parenchyma, sieve tubes and rays, no fibers being formed among the derivatives of the cambium. As the zone of phloem widens in enlarging stems more or less extensive sclerification of parenchyma occurs in its outer part. The form of these secondary changes in the older phloem varies considerably in different parts of the tree. Successive rings of sclerenchyma are formed in the bark of the main stem of the tree. These rings are comparatively widely spaced and are composed of sclereids. The soft intervening tissue contains more or less numerous patches of such sclerenchyma derived in part from phloem parenchyma and in part by the 1953 | BAILEY & SMITH, CALYPTOSEPALUM 59 sclerification of ray cells. At least in certain of the upper branches of the tree, all of the vertically oriented parenchyma tends to develop lignified secondary walls, whereas most of the cells of the rays do not. The crystals of calcium oxalate differ markedly in form in different parts of the phloem. Druses occur abundantly and characteristically in the un- modified parts of the rays. On the contrary, the zones and patches of stone cells have numerous cells which contain single crystals of rhombohedral form. DISCUSSION The summation of reproductive and vegetative characters, in the case of this Fijian tree, is peculiar and unusual, and the problem of determining the true relationships of the plant is a difficult one. When all of its salient exomorphic and endomorphic characters are taken into consideration, it does not fit readily into any existing family of the dicotyledons. However. it closely resembles a Sumatran tree collected by H. O. Forbes in 1880-82. subsequently described and placed in the monotypic genus. Calvptosepalum, by S. Moore (3) in 1925. The species, Calyptosepalum sumatranum S. Moore, is based upon two collections, Forbes 2847 and 2862, Both sets of herbarium specimens bear male flowers only, but a few detached fruits were included under Forbes 2862. Unfortunately, these fruits are of two entirely different kinds. Therefore. it is essential to determine which of them actually belongs to Calyptosepalum. Both of the Forbes collections, 2847 (isotype) and 2862, are represented in the Gray Herbarium, and our observations are based upon these speci- mens, supplemented by examination of a fruit from Forbes 2862 at the British Museum. Sections of the stems of Forbes 2847 and 2862. made by Swamy (5) in connection with his anatomical investigation of the Santala- ceae, were available for detailed study. Significant similarities and differences between the vegetative parts of the Fijian and the Sumatran plants are the following: The buds of both trees aré naked with a 4 phyllotaxy and precociously developing triangular stipules. Those of the Fijian plant are less con- spicuously hairy except at their apex, and the stipules are more massive and leave more evident scars following their earlier abscission. The stems of both plants have a characteristically 4-lobed pith and eustele as seen in transverse sections, a morphological feature that is correlated with a 4 phyllotaxy and a trilacunar nodel anatomy. The cellular structure of the pith, xylem, cambium, phloem, and cortex is fundamentally similar in both cases, the only conspicuous histological difference being the occurrence of porous, as well as of scalariform, perforation plates in the vessels of the Sumatran plant. The major patterns and the minor details of the vas- cularization of the petiole and lamina of the simple entire leaves are strikingly similar. The petioles of both plants have cylindrical eusteles. surrounding a pith which contains from one to three vascular strands. the 60 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv medullary bundles of the Fijian plant being more extensively amphivasal than those of the Sumatran species. The venation of the lamina, as revealed in cleared specimens, is massive and conspicuous, even the smaller veins being jacketed by sclerenchyma and in turn by crystal-bearing cells. The stomata of the Sumatran plant are more extensively subtended by sur- rounding epidermal cells, and hairs and bases of hairs are more numerous on the under surface of fully matured leaves. In addition, there are more numerous small druses in the palisade layer. Both plants are characterized by having hairs of a similar structural type, viz. multicellular and unbranched, but commonly recurved parallel to the surface which bears them. Furthermore, both are characterized by forming two types of crystals, viz. druses and single rhombohedral crystals. A summation of exomorphic and endomorphic evidence from the vegetative organs indicates that the Fijian tree and the Sumatran one are closely related. Such differences as occur in the leaves and stems are of no greater magnitude than may be anticipated in related, but geographically widely separated, species. In fact, without convincing evidence from the reproductive organs, there are no valid arguments for placing the two trees even in separate genera. It is unfortunate that the female flower of the Sumatran plant and the male flower of the Fijian tree are not available, and that one is forced at present to depend upon comparisons between flowers of different sexes. The size, external form, and axillary distribution of the flowers are similar in both cases. Both kinds of flowers have a perianth consisting of two pairs of fleshy, deeply concave, decussate tepals. Both have a conspicuous disk whose form is determined by internal pressures and spacial relationships within the developing flower bud. At anthesis, the expanded disk of the male flower bears the imprints of the four stamens that are crowded inward and downward upon it in the closed bud. That of the female flower, being confined to the space between the base of the carpel and the tepals, expands outwardly rather than inwardly and has an undulating contour in con- formity with the fleshy bases of the tepals. The disk of the male flower is pilose, whereas that of the female flower is glabrous and glandular appear- ing. The carpel of the female flower is conspicuously pilose during the earlier stages of its ontogeny, whereas its sterile homologue in the male flower is nearly glabrous. However, the hairs of both flowers are of a fundamentally similar structural type. As noted earlier, the detached fruits included under Forbes 2862 are of two different morphological kinds. In the case of the specimen at the Gray Herbarium, the fruit obviously does not belong to Calyptosepalum as indicated by the character of the persistent gamosepalous calyx which subtends it. The angularities in the outer contour of the calyx demonstrate that it is composed of five concrescent members. Furthermore, the hairs on its external surface are unicellular and aggregated in clusters, in marked contrast to the multicellular (i.e. internally septate) and diffusely dis- tributed ones of Calyptosepalum. The fruit is broadly ovoid, with a mas- sive pericarp and a persistent, fleshy, 4-lobed and 4-ridged stigma. It 1953] BAILEY & SMITH, CALYPTOSEPALUM 61 has four locules, each of which contains a seed with conspicuously sclerenchymatous testa. The fruit from Forbes 2862 at the British Museum is detached from its pedicel and has a corky scar at its apex, formed by the abscission of its stigma. It is of more ellipsoidal form and evidently is the kind of fruit that was figured and described by Moore (3). It is derived from a single enlarged carpel and contains a single seed which is attached at a relatively low level of its ventral side. As in the case of the Fijian fruits, the seed contracts and expands extraordinarily during drying and soaking. It has a thin soft testa with coarsely conspicuous vasculature, and a massive endosperm which is lobed internally. No embryo is visible. Although the fruit and seed of Forbes 2862 from the British Museum closely resemble those of the Fijian plant in their salient exomorphic char- acters and in the cellular structure of the testa and endosperm, there are obvious anatomical differences in the pericarp. The outer surface of the Sumatran fruit is glabrous, with a thick cuticle which projects inwards between the epidermal cells. Thus in surface view, the epidermis has a reticulate pattern superficially resembling that which occurs on the under side of the leaves of certain Sapotaceae. The sclerenchymatous layer is internal, instead of external, to the vasculature. The stone cells, many of which are of irregular size and form, occur in closely adjacent patches and give to the internal surface of the pericarp an appearance of being com- posed of a miniature mosaic. Forbes’ field notes indicate that fruits of Calyptosepalum sumatranum were collected. It seems likely, therefore, that the fruit from Forbes 2862 at the British Museum is one of these, and that the fruit on the corre- sponding specimen at the Gray Herbarium was added to that sheet by mistake. Evidence from the flowers, fruits, and seeds supports anatomical data from the vegetative parts and indicates that the Fijian tree is closely related to Calyptosepalum. The question whether the two plants should be in- cluded in the same genus or should be placed in two separate, but closely allied, genera is one that may not be finally answered until more extensive collections of the reproductive parts are made. However, on the basis of material now available, we find no valid reason to separate the two plants generically. Differences between them are of a sort to be anticipated in any angiosperm genus of reasonable circumscription, and so we refer the new Fijian species to Calyptosepalum with a fair degree of confidence. Moore (3) concluded that the affinity of Calyptosepalum is with Henslowia, and therefore placed the Sumatran tree in the Osyrideae of the Santalaceae. It is listed under that family by Pilger (4) who con- sidered it to be a genus of uncertain affinities. As shown by Swamy (5), Calyptosepalum cannot be included in the Santalaceae, even as a primitive and structurally less specialized representative of that family. Further- more, the genus has a number of salient characteristics which exclude it from the Olacaceae. Indeed, a summation of evidence from both the vegetative and reproductive parts suggests that Calyptosepalum and its 62 JOURNAL OF THE ARNOLD ARBORETUM _ |[vot. xxxiv Fijian relative do not belong in any previously described family. Should they be placed in a new family: This raises a question of major taxonomic significance and one upon which general agreement should be attained. If Casuarina, Leitneria, Myzo- dendron, Grubbia, Octoknema, Lactoris, Cercidiphyllum, Eupomatia, Gomortega, and other isolated end-products of morphological specializa- tion are to be segregated in independent small families, then it obviously is consistent to deal with Amborella, Trimenia, Euptelea, Trochodendron, and many other genera in a similar manner. Although such a procedure leads to a multiplication of small families—to which many systematic botanists object — it is preferable to loading truly homogeneous families with discordant elements which prove upon thorough study to be out of place. Classification should be determined not by an inherently and rigidly “conservative” or “radical”? concept of families, but by the char- acter of the plants under consideration. In this connection, we may well remember that modern concepts of family delimitation were evolved by botanists familiar, for the most part, with the floras of the Northern Hemisphere. In the regions best known to them, the students who outlined our angiosperm families found many large and obviously closely related groups of genera. Because of the ap- parently natural composition of such groups as the Compositae, Rosaceae. Orchidaceae, Gramineae, etc., it is probable that the taxonomists of a century ago became inured to a concept of large and comparatively few families, and that they subconsciously acquired an aversion to “splitting” at the family level. This.aversion, as is well known, persists among the curatorial staffs of modern herbaria. Every change in family delimitation and every proposal of a new family means, to the curator of a herbarium, a change in his system, accompanied by a shift of specimens, the preparation of new covers, case-labels, index-cards — in short, such a change means more work for a small and often already harassed staff. It is not sur- prising, therefore, that herbarium botanists with great pertinacity cling to the notion that there are already “enough” families. However, it is not a function of systematists to decide a priori upon the number of families. If the plants of the southeastern Asia-Malaysian area had been as well known to the phylogenists of a century ago as were the plants of Europe and North America, can anyone believe that our tradi- tional families would have their present circumscriptions? Workers in this area and also in tropical Africa are finding that, if they apply to their floras the criteria of family-delimitation that have served for the North Tem- perate floras, two courses are open to them. First, they may expand cur- rent family concepts beyond all degree of usefulness, to the point where relic genera are quite lost in a maze of vague and often contradictory char- acterizations, or second, they may propose distinct families for such genera. That the second course seems to be increasing in popularity may cause herbarium workers considerable anguish, but one must conclude that it offers the more legitimate solution of a difficult problem. The multiplica- tion of families caused by the application of uniform standards of family- 1953 | BAILEY & SMITH, CALYPTOSEPALUM 63 criteria is “unfortunate” only in that it causes inconvenience to those in charge of our large herbaria. If such a development leads to a better comprehension of the interrelationships of genera and families, then it should be welcomed as fortunate, in spite of transient physical and psvy- chological inconveniences. Another alternative to the proposal of small families, in the case of phanerogams of uncertain affinities, may be the temporary assignment of aberrant genera to a special category, comparable to the Fungi Imperfecti of mycologists. In such a category might be placed plants of uncertain or obscure affinities, the inclusion of which in any existing family would cause an undue expansion of the basic family-characters. This category could also accommodate those plants as yet incompletely known, lacking. for instance, material for adequate cytological, embryological and anatom- ical investigations. To place such plants in a group of “‘Phanerogamae Im- perfectae” would at least call them to the attention of regional students and would save them from the obscurity of being misplaced in some large and unsuitable family. As previously stated, available exomorphic and endomorphic evidence makes it difficult to include Calyptosepalum in any existing family of the dicotyledons. As demonstrated by Swamy (5), the salient features of Calyptosepalum do not fall within the range of anatomical or morpholog- ical characters of the Santalaceae, and it may be excluded likewise from the Olacaceae. Furthermore, Dr. R. A. Howard concurs in our conclusion that a summation of evidence from different organs of the plants excludes Calyptosepalum from the Icacinaceae. Although the genus exhibits certain similarities, on the one hand, to the Olacales and, on the other hand, to the Celastrales of Hutchinson, there is no convincing summation of evi- dence to justify placing it with certainty in either of these orders. We believe that Calyptosepalum is another relic genus which will ulti- mately have to be placed in an independent family. This should not be done, however, until more adequate and extensive material of its two species are available for detailed exomorphic and endomorphic investiga- tions. For the present, it had best be placed in a special category of plants of uncertain affinities, as has been suggested (1,2) in dealing with other relic genera from Austromalayan and Indomalayan regions. ACKNOWLEDGMENTS We are much indebted to Dr. George Taylor, of the British Museum (Natural History), for the opportunity of examining a fruit from its sheet of Forbes 2862, and to the Gray Herbarium for the privilege of studying its specimens of the Forbes collections. 64 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL. xxxiv LITERATURE CITED BatLey, I. W. The anatomical approach to the study of genera. Chronica Botanica. Ined. _ N . L. Swamy. The morphology and ca of Iden- burgia and Nouhuysia, Jour. Arnold Arb. 34: 77-87. 1953. 3. Moore, S. Dr. H. O. Forbes’s Malayan Plants. Jour. Ny 63: Suppl. 91-92. 1925, Pitcer, R. Santalaceae. E. & P. Nat. Pflanzenfam. ed. 2. 16b: 91. 1935. Swamy, B. G. L. The comparative morphology of the Santalaceae: node, secondary xylem, and pollen. Amer. Jour. Bot. 36: 661-673. 1949. a BIOLOGICAL LABORATORIES, HARVARD UNIVERSITY. DEPARTMENT OF epies U.S. NationaL Mu SMITHSONIAN Fearon. EXPLANATION OF PLATES PLATE I Fic. 1. Transverse section of carpel at the level of the attachment of the two ovules, X 24. Fic. 2. Transverse section of the petiole, showing cylinder of vascular tissues and two included amphivasal bundles, x 50. PLATE II Gc. 3. Transverse section of the secondary xylem, * 50. Fic. 4. Cleared leaf, showing characteristic pattern of venation, X 50. Fic. 5. Tangential longitudinal section of the secondary xylem, showing extensively overlapping vessel members and heterogeneous rays. X 107 PLATE I Jour. ARNOLD Ars. VoL. XXXIV ge 9 *ee8 oa eo ae" MG M MITH, CALYPTOSEPALU Y¥&S BAILE aT Ss BAILEY & SMITH, CALYPTOSEP ALUM ee" *.s ee Pe ry 3 a s S ape 2s a a, hey ae = a) 4, 4 é “a Bi AIXXX ‘TOA ‘aay GTONUYy “‘ANOf II Lv1q 1953] GRAY, REVISION OF PODOCARPUS 67 A TAXONOMIC REVISION OF PODOCARPUS VII. THE AFRICAN SPECIES OF PODOCARPUS: SECTION AFROCARPUS NETTA E. Gray* Tue SECTION Afrocarpus was set-up aS a group distinct from section Stachycarpus in the first paper of this series (4). The six species included are Podocarpus falcatus (Thunb.) R. Br., P. gracilior Pilger, P. Mannii Hook., P. usambarensis Pilger, P. gracillimus Stapf and P. Dawet Stapf. They are all located in areas of Africa near and south of the Equator. Extensive work on the geographic distribution of the species and on the taxonomic details has been done by Stapf (16), Chalk et al. (5) and Robyns (13) representing British and Belgian explorations in their respective territories. Pilger (12) placed this group of species in the subgenus Stachycarpus, part B, and authors since have followed this treatment. However, dissatis- faction with this arrangement has been often voiced, particularly by Florin (7) and Orr (11). These species are distinctly different in their leaf anatomy and female cones, being clearly separated in any key, and are, in themselves, ‘a compact and morphologically homogenous group” (11). They are mostly large trees providing important lumber in areas of abundance. The leaves in all species are amphistomatic with about the same number of stomata on the upper and lower surfaces. Abundant hypodermal fibers are found at the margin and midrib and gathered in groups between the stomatal rows on both upper and lower sides of the leaf. The palisade parenchyma may be equally distributed on both sides or may be more abundant on the surface exposed to the sun, whether abaxial or adaxial. Often it is poorly developed. The single vascular bundle has one resin canal below it, and is flanked on either side by wings of transfusion tissue which extend far into the blade of the leaf, sometimes more than half-way to the margin. This latter feature has been noted by Orr (11) and Robyns (13) particularly. The only other group in which a similar arrangement has been found is in the section Polypodiopsis, P. minor and P. Rospigliosst. In the parenchyma, around the vascular bundle and transfusion tissue, are numerous, often very large, conspicuous fibers. There is no accessory trans- * The author of this paper greatly regrets the loss of her associate in this study, Professor John T. Buchholz, deceased July 1951. She wishes to express her apprecia- tion for his assistance in the preparation of this paper, especially the cleared and mounted whole leaves. Also she gratefully acknowledges the permission and assistance which was given to him to examine and photograph specimens of the section Afrocar pus preserved in the herbaria of the Royal Botanic Gardens, Kew, the British Museum and Cambridge University. 68 JOURNAL OF THE ARNOLD ARBORETUM - [VvoL. xxxiv fusion tissue, but often large pitted sclereids of similar origin develop in the mesophyll between the end of the transfusion tissue and the margin of the leaf (4, PLatE I, FIc. The abundance, size and position of these sclereids were noted to vary among the samples. A detailed study of these patterns was made in whole mounts of leaves which had been cleared in NaOH and stained with saf- franin. It has been found that the sclereids are shortest at the base and tip of a leaf, and largest in the middle. They are usually most abundant in the base of the leaf, less so in the remainder, often decreasing toward the tip or even again increasing in number at the very tip. In addition, the number of sclereids increases with the age of the leaf. Thus, the data of TABLE | are of doubtful value in distinguishing species because of the abundance of over-lapping sizes in leaves having sclereids under the maxi- TABLE 1. SCLEREID LENGTH IN SECTION A frocarpus Species Min. Max. P, gracillimus 50u 95m P 50u 230u P. gracilior 70p 2635p P. Dawei 60u 310u P. usambarensis 65 322u P. Mannu SOu 3638u mum and the data are not consistent either as to species or geographical region. In a similar examination of the sclereids in the leaves of the Podocarpus species, in section Stachycarpus (4) from New Caledonia, New Zealand and Australia, patterns of size and abundance sufficiently different to separate the species lines were found. No such clear criteria seem to be apparent in the case of the African group of species. The following key, adapted from Chalk et al. (5), adding where useful, the characters of leaf anatomy, includes all the species in this group. It is preferable to base a key upon one prepared by individuals familiar with the living trees in the field, where keys are so necessary for identification. Key To SECTION A frocarpus Leaves small, 1-5 cm. long, 1-4 mm. broad on adult trees; inner woody layer of seed 3 mm. thick or less Leaves 2—5 cm. long, 2-4 mm. broad on adult trees (up to 7 or 10 cm. by m. on young plants and long shoots). Seed glaucous green, globose, 0.6—1.2 cm. diam. on stalk 0.6-1.2 cm. long: outer layer of seed-shell coriaceous, inner layer very hard, woody, sa 1 mm. thick. S. Africa: Cape Province, Natal. Transvaal, S. Mozambique..... 2. ee ae igi = nici more than 1. 5m: lone, inner woody layer. more than 1 n ae «lipo globose, purplish-brown and slightly bluish pruinose, 1.6- long, 1.4-1.6 cm. diam.; outer layer of seed-shell coriaceous: 1953] GRAY, REVISION OF PODOCARPUS 69 inner layer very hard woody, up to 2 mm. thick; branchlets angular from the decurrent leaf-bases. Abyssinia, Kenya, Uganda.......... 2. P. gracilior. woody, light colored, 3 mm. thick. Uganda........ : F Leaves 1.8-2.5 cm. long, 1-1.5 mm. broad, straight or slightly falcate, acute, densely crowded at the ends of the branchlets; seed up to 2 cm. long, woody inner layer of seed-shell 1-2 mm. thick. Transvaal........... 4. P. gracillimus. long. Tanganyika, Belgian Congo.............--- . P. usambarensis. Leaves sessile, 7.5-15.0 cm. long, 5-12 mm. wide; seed subpyriform, subsessile, about 3.8 cm. long. West Africa...................... 6 P. Mannit. Podocarpus falcatus (Thunb.) R. Br. ex Mirb. in Mém. Mus. Paris 13: 75. 1825 (nomen); Endlicher, Syn. Conif. 219. 1847, Parlatore in D C. Prodr. 16 (2): 511. 1868; Pilger in Pflanzenreich IV. 5 (Heft 18): 72. 1903, in Nat. Pflanzenfam. ed. 2, 13: 245. 1926; Bernard in Beih. Bot. Centralblatt 17: 275. 1904; Wilson in Jour. Arnold Arb. 9: 143. 1928; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 262. 1931: Chalk et al., For. trees and timbers of Brit. Emp. 1: 23. 1932, 3: 86. 1935; Dallimore & Jackson, Handb. Conif. 44. 1923, 1931, 67. 1948: Stapf in Fl. Cap. 5 (Sect. 2, suppl.): 10. 1933. Podocarpus Meyeriana Endlicher, Syn. Conif, 218. 1847; Parlatore in DC Prodr. 16 (2): 512. 1868. Podocarpus elongata Carriére, Traité Conif. ed. 2: 671. 1867, in part; Sim, Forests & For. Fl. 335. 1907; Marloth, Fl. S. Afr. 1: 101. 1931-32. Taxus falcata Thunb., Prodr. Pl. Cap. 117. 1800, et Fl. Cap. (ed. Shultes) 547. 1823. This species is a tree, often very large, with minute terminal buds, leaves erect, 2-5 cm. X 3-5 mm., apex shortly acute to almost obtuse (juvenile somewhat larger); male flowers short, solitary or 2—3 together; seed 6-12 mm. diam., globose, with peduncle 6-12 mm. A complete description, in- cluding figures, may be found in Chalk et al. (6). Podocarpus falcatus, from the Cape of Good Hope vicinity, was first described as Taxus falcata from a specimen collected by Thunberg in 1773-74, at the same time as P. elongatus later to be used as the type for the genus. Both these specimens are in the Botanical Museum at Uppsala. We have seen photographs and examined leaf samples and can designate them as authentic. Toward the latter part of the nineteenth century these two species, with leaves of similar size and shape, became rather hopelessly confused both in the literature and among foresters, and they are still occasionally interchanged in published work. As late as 1943, Smith (15) found it necessary to include these in publishing a verification of the 70 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv nomenclature of certain African species of Podocarpus. In 1867, Carriére published a description under the name of P. elongata Carr. which was only partly P. falcatus. Parlatore, in 1868, described P. falcatus under that name but the description of the female flowers was of another species. In 1874, Bertrand (2) published his detailed anatomy of conifer leaves in which he described P. falcatus under the name of P. elongata L’Herit. and placed the former name in synonymy. Mahlert, in 1885, using much of Bertrand’s work on conifers, described P. falcatus in detail under the name of P. elongatus L’Herit. Stiles (17) reconsidered this same data and fol- lowed the same misconception. In 1891, Van Tieghem (18), in a paper setting up Stachycarpus as a genus, did not include any African species, but he placed P. falcatus in Eupodocarpus and listed P. elongatus as a species having palisade parenchyma on both sides of the leaf. Pilger (12), in 1903, showed in his taxonomic key and description a clear concept of P. falcatus R. Br. as different from P. elongatus L’Herit. However, in his preliminary discussion of leaf anatomy preceding the taxonomic part, he repeated the errors of Mahlert and others in describing for P. elongatus the leaf anatomy of P. falcatus. Later authors continued the confusion. Sim (14) and Marloth (10) redescribed this species in its correct position to affirm its place to African investigators, as did Stapf (16) and Chalk et al. (6) in their listings of strictly South African trees. Orr (11), in making a study of the leaf anatomy of the genus, mentioned the confusion in the literature and found that Bernard labelled a text figure of P. elongatus as P. falcatus which is the reverse of the usual error. Orr (11) indicated that this species can be differentiated by the leaf anatomy: “hypodermal fibers are much larger than those of any other species and the groups into which they are segregated project more deeply into the underlying mesophyll.” We are not able to confirm this observa- tion as statistical treatment of measurements shows no consistent varia- tion from those of the other species. In a study of the size and abundance of sclereids it was found that those of P. falcatus were larger than those of P. gracillimus and mostly smaller than those of P. gracilior, but in this latter case, the largest sclereids of P. falcatus overlapped the smaller in P. gracilior, making this data of doubtful use in differentiating these species. Two varieties have been described by Pilger on the basis of leaf size. Podocarpus falcatus var. latifolius has leaves 2.5~3 cm. long, 5-6 mm. wide, with the apex short-rotundate. It is represented by a single specimen from Kaimansgat, Drege 6182. We have examined a specimen, Drege 6184, which does not seem to differ from the species. Podocarpus falcatus var. pondoensis is one with small leaves, up to 2 cm. long and 3 mm. broad. represented by the single specimen Bachmann 69. We have seen no speci- men comparable to this and it may be referred to P. gracillimus which will be discussed later. These varieties have been ignored by Stapf (16) and Chalk (6) in studies of these African trees. DISTRIBUTION: Southern part of the African continent in the following districts: Cape Province, Natal, Portuguese East Africa. Transvaal and British Bechuanaland. 1953] GRAY, REVISION OF PODOCARPUS 71 SPECIMENS EXAMINED: SOUTH AFRICA: Transvaal: Burtt-Davy 20248 (A*). Natal: Northern Zululand, Boocock in 1923 (A—3 sheets); The Doyle, Wilson in Jan. 1922 (A), Wilson in Feb. 1922 (A—2 sheets). Transkeian Territory: Mananza Forest, Mt. Ayliff, Cochrane in 1920 (K). Griqualand East: Buswayo Forest, Manina. Merwe Pret. For. Dept. Herb. 2268 (K), Merwe Pret. For. Dept. Herb. 2269 (K); Instubani Forest, Fraser Pret. For. Dept. Herb. 62A (K), Fraser Pret. For. Dept. Herb. 2226 (K); Emkazeni For. Res., Household Pret. For. Dept. Herb. 1956; Waterfall Farm, Wilmot Pret. For. Dept. Herb. 2205 (K). Cape Province: East London, Kuntze in 1894 (K, NY); Queenstown Dist., Junction Darm., Galapin 8179 (K); King Williamstown Div., Kaffraria, Cooper 1297 (+K); Alexandria Forest, Strauch 4137 (A); Somerset East, Boschberg, Burchell 3174 (K), Burchell 3189 (K), MacOwan 1561 (+F, K, Ph); Albany Dist., Prior in 1848 (K), Dyer 3325 (K); Alicedale, Pulgram Pret. For, Dept. Herb. 2609 (K); Uitenhage, Elon River, Drege 6184 (+Mo), Drege 1839 (K), Kamaehs, Long 1280 (K), eastern valley Blauberg, Smuts 852 (K), Swartkop River, Prior in 1847 (K), Eklon & Zeyher sn. (A, +F, tMo, UC); Humansdorp, Hankey, Fourcade 3319 (K), Klein River, Long 1350 (K); Ribeckscastle, Vleermuysdrift, Thunberg s.n. ($UPS); Outeniqua Mts., Montagu Pass, Rehmann 213 (+BM); Knysna, Kaymansgate, Drege in 1839 (K), Plattenberg For., Bowie s.n. (K), Wilson in 1922 (+A—S sheets, UC), Vandu Wats, Burchell 5293 (K), Setchell in 1927 (UC), Mund & Maire s.n. (K), in 1840(?) (K) ; George Division, Gat River, Drege s.n. (K), Burchell 5761, 6068, 8909 (K). In Cape CoLoNy BUT WITH NO LOCALITY: Drege s.n. (+Mo, NY), Prager 785 (+CAS), Anon. s.n. ex Scheid- weiler Herb. (+BR), Burchell 5245 (GH), Burchell 4632 (GH), Basil in 1865 (NY), Anon. leg. M. & M. (ex HB. Rg. Br.) (GH, NY), Anon. s.n. (GH), Stewart sn. (K). No xocatity: Anon, ex Herb. A.N.S. (+Ph), Anon. 10.9 (+Mo), Anon. ex Herb. Jan Gen 840% ex Bernhardi Herb. (+Mo—4 elements on sheet), A. Brown in 1862 (+Mo), Anon. s.n. (BR), Molliana in 1833 (+BR), H.S. in 1834 (+BR), Anon. ex Gord. Herb. (K). CULTIVATED: Cape of Good Hope, “Hort. Daudin” 1851 (+DS in part); Natal, Kirstenbosch, Bolus in 1941 (t+Ill); France, Riviera, Schneider in 1903 (A), “Hort. Monac.” (BR in part); Italy, Rome, for Engelmann in 1870 (+Mo), Naples for Engelmann in 1869 (+Mo); Brazil, Rio de Janeiro, Whitford 27 (+F, GH, +US); Australia, Bowman in 1921 (A); China, Hongkong, Chun 5235 (¢UC), Sargent in 1903 (+A). Podocarpus gracilior Pilger in Pflanzenreich IV. 5 (Heft 18): 71. 1903, in Nat. Pflanzenfam. ed. 2, 13: 245. 1926; Battiscombe, Descr. * The following symbols indicate the herbaria having the specimens cited: Academy of Natural Sciences of Philadelphia (Ph), Arnold Arboretum (A), British Museum (BM), Brussels Botanical Garden (BR), University of California at Museum (Field Museum) (F), Gray Herbarium (GH), University of Illinois Herbarium (Ill), Kew Herbarium (K), Missouri Botanical Garden (Mo), New York Botanical Garden (NY), Rutgers University Herbarium (NJU), Stanford University Dudley Herbarium (DS), United States National Herbarium (US). Uppsala Botanical Museum (UPS), Yale University Herbarium (YU). + This symbol preceding the abbreviated name of an herbarium following the specimens examined signifies that the details of the leaves of this specimen have been examined in cross-section. 72 JOURNAL OF THE ARNOLD ARBORETUM _ |[vot. xxxiv Cat. of common trees & woody plants of Kenya Colony 1. 1926; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10 (1): 262. 1931; Chalk, Burtt- Davy & Desch, Forest trees and timbers of Brit. Emp. 1: 23. 1932: Giordano in L’Alpe 24: 417. 1937; ae in Madrono 6: 119. 1941; Stapf in Fl. Cap. 5 (Sect. 2, suppl.): 1933 This species differs from P. falcatus chiefly in its more restricted range in the forests of tropical Africa. Pilger differentiated it primarily by the larger male cones, 1.5~3 cm. long, its scales with acute apices (instead of obtuse), the more patent, fairly loosely arranged leaves, and the larger seeds, 12-20 mm. long. Stapf (16) finds the leaves mostly long tapering to a sharp point and narrower. He notes the similarity to P. falcatus and suggests that it may be only a geographical variety. Podocar pus gracilior is cultivated as an evergreen exotic in California and has been distributed widely in tropical-plant houses. The tree has been called P. elongatus, an error which was corrected by Buchholz in 1941 (3). At the same time several additional differences from P. falcatus were noted. Podocarpus gracilior has green twigs, the leaves are longer and wider, and the pollen cones are twice as long. In the description which Battiscombe (1) included with his listing of this species from Mt. Elgon in Kenya Colony, the small seed must be an error and 1% in. long was meant. DIsTRIBUTION: Eastern Africa, on mountain slopes in tropical regions. SPECIMENS EXAMINED: ETHIOPIA: Addis Ababa, Armbruster in 1915 (K): Mt. Chillalo, Scott s.n. (K); Galla Pass, Gillett 5132 (K); West Sidamo, Drake-Brockman 205, Brock- man 244 (K); Djem-Djem Forest, Cooper sn. (K); Kena Valley, Alafa, rig 7445 (K); Chere, Quartin-Dillon & Petit 105 ee Schimper 1160 PE) (K); ee 1539 (+BR); Mearns 92 (+BR); Roth sn. in 1841 is Stordy sn. (K). KENYA: ella ie Oljoro-o-Nyon and Narok Rivers, Mearns 373 (+US), a 382 (CU, +F, GH, NY—2 sheets), Mearns 1382 (GH, NY); NW Kenya, Hutchins 400, ean (K); ars Plains, 20 m. from Kenya, Hutchins 399 (K); Coles Farm, Fries 801 (+BR, K); Myeri Dist., Decie in 1926 (+BM); Walenso, A. Donaldson Smith 217 (+BM); Mau escarp- ment, Wilson in Dec. 1921 (A—S sheets); beyond Londoni, Wilson in Dec. 1921 (A—4 sheets); SW atl ee ea sm. in 1906 (K); Nairobi Dist., Gardner 1110 (K), Wilson in Nov. 1 (A—4 sheets, 1 pence Wilson in Jan. 1922 (A—2 sheets); Fuller 683 aS ae in 1928 (DS); Curtis 901, 1071 (A). UGANDA: Kaburon, Mt. Elgon, Eggeling 2474 a Eeteling 2479 (+BR); Mt. Elgon, Lugard 64, 64a (K), Jackson 352a (K); di & Ravine Dists., Johnston in 1900 (K); Eldoma Ravine, Johnston in or (K), in 1900 (K), Whyte in 1898 (K); Mt. Debasien, Karamoja, Thomas 2953 (K); Masai County, Ty Kipia, Thomson in 1884 (K); Ruwenzori Exped., Scott-Elliott 6959 (K). TANGANYIKA TERRITORY: Mt. Kilimanjaro, Gereragua, Zimmer- mann 1783 (K); Mt. Meru, Greenway 4434 (K):; Kondoa Dist., Salanga Hill on Bereku Ridge, Burtt 10694 (K—2 sheets), Burtt 1070 (K—2 sheets), Kinyassi Mtn. Burtt 910 (K); Mt. Hanang, Nangwa, Greenway 7602 (K); Bukoba, Minsiro Forest, Wigg 303 (K). LocaLiry UNKNowN: Burtt 1174 (+K 1953] GRAY, REVISION OF PODOCARPUS 73 in part); ex Scheidweiler Herb. (BR); Anon, s.n. (+BR); Parker s.n. (A); Anon. s.n. (tMo). CULTIVATED: Kenya: Nairobi Arboretum, Eggeling 3570 (K), Kirsten- bosch, Graham in 1941 (+Ill). Tanganyika; Cons. Forests in 1932 (A, tBR, K, tMo, NY). Cameroons: Maclaut 344 (K). Cape of Good Hope: ex Herb. John Miers (| BM), Forest Nursery, Wilson in 1921 (A-seedlings). India: Dehra Dun. Raizada 5028 (K). Switzerland: Schneider in 1903 (A). England: Kew, Temp. House, Cook in 1937 (+Ill). U.S.: California, Rosecroft Gardens, Buchholz in 1941 (+Ill—2 sheets); Santa Barbara Bot. Gard. Moran 2529 (Dearing Herb., +Ill); San Diego, Wangenheim in 1941 (Ill) ; Pasadena, Coolidge Nursery, Buchholz in Jan. 1941 (A, Ill), in Feb. 1941 (A, Ill); Berkeley Bot. Gard. Nelson in 1941 (Ill—2 sheets); Santa Barbara, Van Rensselaer 1727 (Ill—2 sheets), 1801 (Ph). Podocarpus Dawei Stapf in Fl. Trop. Afr. ed. Prain, 6(2)5 342, 1917; Florin in Svenska Vet.-Akad. Handl. Ser. 3, 10: 262. 1931; Chalk, Burtt-Davy & Desch, For. trees & timbers of Brit. Emp. 1: 23. 1932. A typical member of section Afrocarpus, it differs from P. falcatus, P. gracilior and P. gracillimus by the subglobose, chestnut-brown seed with inner woody layer 3 mm. thick. Chalk (5) also differentiates it from P. usambarensis, found in the same area, by the smaller leaves, but this be- comes a very doubtful criterion after noting the ecological variations in the latter species seen in the specimens and described in detail by Robyns C13); DistRIBUTION: Tropical Africa, Uganda. SPECIMENS EXAMINED: Uganda: south Budda, near Kagera River, Dawe 961 (+K, Type); Podo, Mt. Debasien at Mareyo, Eggeling 2704 (+BR); Masaka, Katua, Eggeling 5566 (K—3 sheets and seeds in alcohol); Kigezi Dist., Kayonso forest, Fyffe in 1928 (K—2 sheets), Eggeling 4180 (K-seedling). Belgian Congo: near Matemba, between Lubango & Luofu, w. of Lake Edward, Chapin 264 (+NY). CULTIVATED: Uganda: Entebbe, Eggeling 5712 (with seeds in alcohol) (K), Eggeling 5711 (K). Natal: near Durban, Wood 3005 (+K). Podocarpus gracillimus Stapf in Fl. Trop. Afr., ed. Prain, 6 (273 343. 1917: Chalk, Burtt-Davy & Desch, For. trees and timbers of Brit. Emp. 1: 23. 1932; Stapf in Fl. Cap. 5 (Sect. 2, suppl.): 14. 1933. This species is represented primarily by a single specimen, Nelson 423 from Transvaal, with small leaves (1.8-2.5 cm. long by 1—1.5 mm. broad) and a seed 20 mm. in length. The leaf anatomy was inadvertently described by Laurent (8) in connection with his description of P. rostratus. We have examined a photograph of this specimen and a leaf from it, which shows, as he described, palisade parenchyma on both sides, sclerified fibers in the parenchyma and hypoderm in discontinuous groups, and indicates that it is amphistomatic. He also described three resin canals under the vascular bundle, but we are unable to verify this observation as we find only one in that position. If three resin canals do occur, it is the only member of Afrocarpus having them. Orr (11) says that it has the 74 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxrv largest vascular fibers, but these vary so much in size among all the species and even in a single section from a leaf that it is a doubtful diagnostic feature. In the statistical study of the abundance and form of the sclereids of the accessory transfusion tissue, it was found that these were by far the smallest and shortest, not exceeding 95, in length. In Stapf’s (16) later esimate of the status of P. gracillimus, it is a “very doubtful and incompletely known species. It may represent merely a state of P. gracilior, in which the reduction of the leaves and breadth character- istic of the fruiting stage has been carried to excess.” Sim (14) had pre- viously suggested this explanation. A similar state was collected by A. Whyte in the Eldama Ravine, Kenya, “along with typical P. gracilior.” Stapf adds Burtt-Davy 5083 as a possible example and we can include several specimens from American herbaria which are listed below. DIsTRIBUTION: South Africa, in Transvaal and Natal. SPECIMENS EXAMINED: TRANSVAAL: Houtschberg, Nelson 423 (+K): NATAL: Zululand, Bal- landen 2925 (A—2 sheets); Wilson in 1922 (A). CULTIVATED: Transvaal, in Helpmakaar Arboretum, Burtt-Davy 20227 (+A). Podocarpus usambarensis Pilger in Pflanzenreich VI. 5(Heft 18): 70. 1903, in Nat. Pflanzenfam. ed. 2, 13: 245. 1926; Stapf in Fl. Trop. Afr. ed Prain, 6(2): 341. 1917; Dallimore & Jackson, Handb. Conif. 57. 1931, 84. 1948; Florin in Svenska Vet.-Akad. ser. 3,10: 262. 1931; Chalk, Burtt-Davy & Desch, For. trees & timbers of Brit. Emp. 1: 23. 1932; Robyns in Inst. Roy, Colon. Belge, Bull. 6(1): 236. 1935, in Flore du Congo Belge et du Ruanda-Urundi 1: 5-6, 1948. Podocarpus falcatus Engl. in Pflanzenwelt Ostafr. C: 92, 1895. This species is distinguished by Pilger (12) and Chalk (5) from P. falcatus and P. gracilior by the thicker inner woody layer of the seed and from P. Mannii by the shorter leaves and smaller, more globose seed. How- ever, it closely resembles the two former species and Robyns (13) has given a very pertinent discussion (translated) of the status of this species: “Tt is true that P. wsambarensis has an area of distribution more tropical than P. gracilior, that it is only described from Abyssinia, Uganda, or Kenya, but it is possible that these groups represent in reality two geo- graphic varieties of one and the same species. We note, moreover, that Stapf has shown that the diagnostic characters of P. gracilior and P. falcatus are confused and that it is reasonable, also, to consider P. gracilior as a geographical variety or the tropical form of P. falcatus, A comparative study of the three species in question will lead then, very probably, to their fusion into a single species, which should be called P. falcatus, according to the laws of nomenclature.” We would concur with Robyns in this matter and perhaps add P. gracillimus and P. Dawei to the series. However, we believe that P. 1953] GRAY, REVISION OF PODOCARPUS 75 usambarensis can be distinguished by the large seeds even though this may not be a character of specific rank. Also the leaves usually seem to be stiffer and more abruptly pointed, not as falcate as P. gracilior. We have tried to confirm Orr’s (11) statements that the accessory trans- fusion sclereids are “more numerous and conspicuous here,” and that the hypoderm ‘“‘has the appearance of being interrupted at greater intervals”’ and is in an “unbroken layer above and below vascular bundle.” We do not find these to be any more true in this than in the other species. DistriBuTion: Tropical Africa, in Belgian Congo and Tanganyika. SPECIMENS EXAMINED: BELGIAN CONGO: Lac Mokoto, Claessens 52 (+BR). RUANDA- URUNDI: Rubongera, Lestrade 2 (+BR). TANGANYIKA: Kilimanjaro, north side, Schlieben 5129 (+ex Berlin Herb. Florin, +BR, Ill, +YU); Mt. Usambara, Holst 2467 (+K); Kondoa Dist., Burtt 1069 (K—2 sheets), Burtt 1174 (K). CULTIVATED: Tanganyika: Runjwa Dist., Davies, Dept. Agr. D211 (+K):; Jamaica: Cinchona, Nichols 167 (+F, GH, +Mo, NY, US, YU); Schwaby in 1943 (Ill): Buchholz in 1946 (+Ill); Chrysler 1460 (NJU); Schetdweiler s.n. Podocarpus Mannii Hook f. in Jour. Linn. Soc. 7: 218. 1864; Pilger in Pflanzenreich IV. 5 (Heft 18): 70. 1903, in Nat. Pflanzenfam. ed. 2. 13: 245. 1926: Henriques in Biol. Soc. Brot. 5: 216. 1887; A. Nobre in Boletim da Soc. Brot. 7: 115. 1889, in Pflanzenwelt Ostafr. C: 92. 1895 (not Engl.); Stapf in Fl. Trop. Afr., ed. Prain, 6 (2): 341. 1917: Seybold in Bot. Abh. 6: 49. 1925; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 262. 1931; Dallimore & Jackson, Handb. Conif. 50. 1931, 74. 1948; Chalk, Burtt-Davy & Desch, For. trees & timbers of Brit. Emp. 1: 23. 1932; Chevalier in Rev. Bot. Appl. 19: 411. 1939. This species has the largest leaves and seeds of any species in the group. The leaves are 12-16 cm. long by 7-11 mm. wide (shorter on flowering shoots), and the seed, narrowed at the base, solitary, is found up to 3.8 cm. long on a peduncle 2-3 cm. long. The auxiliary sclereids also reach the greatest length, 368». Podocarpus Mannii is a rare species found native only on St. Thomas Island, cultivated in the forests of Cameroons and in California in Griffith Park. Herbarium specimens were collected in 1944 from 2 yr. old seedlings raised in this latter place from seeds obtained from Buenos Aires, pre- sumably from cultivated trees. DISTRIBUTION: Western tropical Africa. north of the Equator. SPECIMENS EXAMINED: WEST AFRICA: St. Thomas Island: Mann 1065 (tyPE) (+GH—2 sheets, +K—2 sheets), Spangler in 1897 (+BR), Exell 339 (+BR). CULTIVATED: Cameroons: Buea, Dartel 557 (A), Maitland 343 (K), Fairchild SPD74656 (K), Anon, 345 (K); California: Griffith Park, Buchholz in 1944 (+III—2 sheets). w N —_ - nr add SF - ed o — — 17. 18. JOURNAL OF THE ARNOLD ARBORETUM [VOL. XXXIV LITERATURE CITED Battiscombe, E. A descriptive catalogue of some of the common trees and woody plants of Kenya Colony. London. 1926. BERTRAND, C. E. Anatomie comparée des tiges des feuilles chez les Gnétacées et les Coniféres. Ann. Sci. Nat. Paris, ser. V. 20: 5-153. 1874. Bucuuotz, J. T. Podocarpus gracilior in cultivation. Madrofio 6: 119-122. 1941. BucHHoLz, JoHN T. ann Netra E. Gray. A taxonomic revision of Podocarpus. I. The sections of the genus and their subdivisions with special reference to leaf anatomy. Jour. Arnold Arb. 29: 49-63. 1948; V. The South Pacific species of Podocarpus: section Stachycar pus. ibid. 32: 82-92. 1951. Cuatk, L., J. Burtt-Davy & H. E. Descu. Forest trees and timbers of the British Empire I. Some East African Coniferae & Leguminosae. Clarendon Press, N.Y. 66 pp. 1932. Cyak, L., M. M. Cuarraway, J. Burtt -Davy, F. S. LaucHton & M. H. Scort. III. Fifteen South Ana high forest timber trees. Forests & se of Brit. Emp. Clarendon Press, N.Y. 1935. . Fiortn, R. Untersuchungen zur Stammesgeschichte der Coniferales u. Cordaitales. Erster teil. Morphologie und Epidermizstrukture der Assimilationsorgane bei den rezenten Koniferen. Svenska Vet. -Akad. Hand. ser. 3, 10: 1931 LAURENT, M. L. Les Podocarpus de Madagascar. Ann. Fac. Sci. Marseille 23% 52-64. 1915. MAHLERT, ApOLpH. Beitrige zur Kenntnis der Anatomie des Laubblatter der Coniferer mit besonderer Beriicksichtigung des Stalt6ffnungs- Apparates. Bot. Centralbl. 24: 278-282. 5 MarotH, R. The Flora of South Africa, with synoptic tables of the genera of the higher plants. Vols. 1-4. Cape Town & London. 1913-1932. - Orr, M. Y. The leaf anatomy of Podocarpus. Trans. Proc. Bot. Soc. Edinburgh 34: 1-54. 1944. Privcer, R. Taxaceae. Engler Pflanzenreich IV. 5 (Heft 18): 1-124. 1903. - Rosyns, W. Sur les espéces de Podocarpus du Congo Belge et du Ruanda- as. Urundi. Inst. Roy. Col. Belge, Bull. 6: 226-242. Stu, T. R. The forests and forest flora of the Colony of the Cape of Good Hope. Aberdeen. 1907. SmiTH, C. A. Nomenclature of Podocarpus species. So. Afr. For. Assoc. Pretoria Jour. 10: 23-28. 1943 . STAPF, O. Podocarpaceae. FI. Cap. 5 (Sect. 2, suppl.): 3-14. Ed. Sir Arthur Hill. 1933. STILEs, W. The Podocarpeae. Ann. Bot. 26: 442-514, 1912. VAN TIEGHEM, Ph. Structure et affinités des Stachycarpus, genre nouveau de la famille des Coniféres. Bull. Soc. Bot. Fr. 38: 162-176. 1891. Emory UNIVERSITY, GEORGIA. 1953] BAILEY & SWAMY, IDENBURGIA AND NOUHUYSIA (is THE MORPHOLOGY AND RELATIONSHIPS OF IDENBURGIA AND NOUHUYSIA I. W. BAILEY AND B. G. L. Swamy With two plates INTRODUCTION IN A PREVIOUS PAPER (5), we excluded /denburgia and Scyphostegia from the Monimiaceae. Since doing so, we have devoted considerable at- tention to additional investigation of the morphological characteristics of these peculiar genera, and to the problem of determining their relationships. It soon became evident that /denburgia is closely related to, if not actually congeneric with, Nouhkuysia. In writing to Professor van Steenis for addi- tional material, we learned that he had a paper in press (7), in which he concludes that the two genera are identical and places the combination in a new tribe, the Nouhuysieae of the Guttiferae. He recognizes three dis- tinct species of Nouhuysia, N. papuana Lauterbach, N. arfakensis (Gibbs) van Steenis and NV. pauciflora (A. C. Sm.) van Steenis; Idenburgia novo- guineensis Gibbs, I. pachyphylla Gilg & Schlechter, and J. elaeocarpoides xilg & Schlechter being reduced to synonymy with NV. papuana Lauterbach.! Upon the basis of our detailed study of the exomorphic and endomorphic characters of the vegetative and reproductive parts of Jdenburgia and Nouhuysia, we are in agreement with van Steenis regarding the identity of the two genera, but we question whether Nouhuysia actually belongs in the Guttiferae. Furthermore, there are questions regarding speciation within the genus that appear to need extensive investigation. It seems desirable accordingly to summarize salient features of the morphological and ana- tomical data that we have accumulated during the last few years. MATERIAL Our observations are based upon the examination of material from the following 25 collections of Nouhuvsia and Idenburgia: Versteegh 1668 (type of NV. papuana Lauterbach); Clemens 3828 (type of J. pauciflora A. C. Sm.), Kostermans 2217 [N. arfakensis (Gibbs) v. Steenis], 2198; Clemens 2240, 2422, 3122, 3978, 3978A, 4194A, 5122, 54996, 7023, 8754, 8835, 11134, 12353; Brass 11078, 12661; Brass & Versteegh 10472; Gibbs 5654; Neth. Ind. For. Serv. bb 24085, bb 25014; Pulle 647; Rutten 2240. "Since our manuscript went to press, we have learned through Dr. A. C. Smith of a paper by S. Hatusima (Bot. Mag. Tokyo 65: 107-111, 1952.) in which he likewise reduces Idenburgia to Nouhuysia, This paper is not available as vet in the libraries of Greater Boston. 78 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL. xxxiv STYLOIDS A significant criterion in all of the collections that we have examined is the occurrence of visible calcium oxalate dominantly in the form of styloids. Although the styloids vary markedly in size, attaining at times a length of 350 microns and a breadth of 30 microns, their characteristic crystallo- graphic form is that illustrated in Fig. 6. They are square or rectangular in sectional view, and are formed more or less abundantly in the lamina of the leaves and tepals and in the cortex and secondary phloem of the stem and petiole. Furthermore, they may occur, at least in certain cases, in the pith of the stem, in the connective of the stamens and in the pericarp of the fruit. In leaves and tepals, aberrations of form due to twining or the development of more than one crystal in a single cell are statistically of relatively infrequent occurrence. Transitions to raphides, such as may be observed for example in certain of the Parietales and Rubiales, do not occur. Nor have we been able to detect the presence of mucilage in the crystal-bearing cells. Transitions to aggregates of smaller crystals or to druses do occur, however, at times in the phloem and pith of the stem, e.g. Brass 11078. In the lamina of the leaves and tepals — where the styloids commonly are large and have their major axis oriented parallel to the surfaces of the lamina — they are contained within cells that are markedly different from the surrounding cells and therefore appear to be idioblasts. Their outlines are visible in surface view of herbarium specimens when boiled leaves and tepals are examined under an ordinary binocular microscope with adequately controlled surface illumination. It is significant in this connection that, in clearing dried leaves for detailed study of their internal structure in transmitted light, the styloids tend to disintegrate and may disappear under prolonged treatment in 3% sodium hydroxide at 52°C. Therefore, it is essential to verify observations based on cleared material by the examination of thin sections of tissues that have not been subjected to the action of either acids or alkali. LEAVES Although the simple, extipulate leaves with glandular-serrate margins vary considerably in size, in the form of their apex and in hairiness, they are characterized in all collections by having a fundamentally similar pat- tern of vasculature. Three independent and relatively widely separated vascular strands or traces at the trilacunar nodes of the stem, F7g. 10, enter the base of the petiole, where they aggregate in the characteristic vascular configuration illustrated in Fig. 12. Thus, throughout most of the petiole and the midrib of the lamina the vascular tissues are distributed in the form of a trough with acutely inturned margins. The venation of the lamina is fundamentally pinnate and arcuate-reticulate in all cases. The veins and veinlets are more or less heavily jacketed with thick-walled fibers, and therefore tend to be embossed and conspicuous in dried specimens. The 1953] BAILEY & SWAMY, IDENBURGIA AND NOUHUYSIA 79 stomata are of relatively constant size, form and structure, and are sur- rounded by ordinary epidermal cells, Figs. 8 and 9, no special subsidiary cells being formed in relation to them. The cell walls of the lower epi- dermis, as seen in surface view of the lamina, commonly tend to be con- spicuously undulate, Fig. 9, but this character is less constant than the preceding ones, Fig. 8. A much wider range of variability occurs, how- ever, in the cells of the mesophyll. The upper epidermis is subtended by palisade tissue, no hypodermal layer being formed in any of the leaves that we have examined. The palisade tissue is subtended by a small-celled, relatively compact form of spongy parenchyma, the rest of the mesophyll being composed of larger cells with more extensive intercellular lacunae. The cells of the palisade commonly are distributed in a single layer, but more than one such layer is encountered in certain cases, viz. Brass 11078 and Brass & Versteegh 10472. The cells may be slender and typically elongated (Kostermans 2198, Brass 12661) or they may be short and poorly differentiated from the subtending compact form of spongy meso- phyll (Clemens 3828, 3978). The cells of the spongy mesophyll in com- parable parts of the lamina vary in size and form from one collection to another. They may be thin-walled throughout the lamina, thin-walled with occasional sclerotic modification, Fig. 2, a mixture of thin-walled and thick- walled, or prevailing thick-walled, Figs. 1 and 7. Similarly, the cells that jacket the veins and veinlets may be either thin-walled or thick-walled. The intercellular lacunae also vary markedly in size, compare Figs. 1 an STEMS The fundamental structures of the stem, viz. pith, xylem, phloem and cortex, are remarkably similar in all collections that we have examined, such minor differences as occur being of a quantitative, rather than of a qualitative, nature. The most significant diagnostic criteria, other than the occurrence of styloids particularly in the phloem, are the primitive structure of the xylem and the character and distribution of the scleren- chymatous cells in the cortex. The secondary xylem is formed by a primitive type of cambium having long fusiform initials with extensively overlapping ends. The vessels are slender, thin-walled and angular, Fig. 13, and are distributed either singly or in occasional radial clusters. The vessel members are long with exten- sively overlapping ends and with unusually numerous scalariform perfora- tions, Fig. 14. The pitting between vessels in lateral contact is scalariform and transitional to opposite, as is that between vessels and parenchymatous cells. The more slender vessel members closely resemble scalariformly pitted tracheids and exhibit transitions between imperforate and perforate scalariform bordered pits. The fiber tracheids have numerous pits with conspicuous borders in both their radial and their tangential walls. The orifice of these pits may be included, slightly extended, or much extended by cracking of the secondary wall. The distribution of the wood paren- chyma is apotracheal-diffuse. The rays are conspicuously heterogeneous, 80 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv the uniseriate ones being composed of very tall erect cells, and the biseriate and triseriate ones of lower cells which tend to become procumbent in the outer wood of older stems, Fig. 13. There is a more or less compact zone of thick-walled fibers in the so- called pericyclic layer of the stem. This is jacketed by a broad zone of cortical sclerenchyma, composed of large cells of somewhat irregular form and orientation. These cells, although of a sclerenchymatous nature, are characterized by having relatively thin, lignified walls and unusually large mina. FLOWERS AND FRUITS The floral morphology of the various species of /denburgia as well as of Nouhuysia papuana conforms to a similar plan with little variation. The exomorphic characters of the tepals, stamens, pistil and fruit have been described and adequately illustrated by Gibbs (2), Gilg and Schlechter (3), and van Steenis (7). The chief variable factor appears to be in the number of stamens present in a flower, /. arfakensis with six standing at one extreme and /. elaeocarpoides with 10-13 representing the other ex- treme. /denburgia pauciflora has flowers decidedly of smaller size in rela- tion to other species of the genus and is distinctive in the possession of profuse dense tomentum on both its vegetative and reproductive parts. The vasculature of the flowers also presents a rather stabilized pattern in the two genera. The median vascular strands, Text-fig. 3, md, of the two pairs of decussately arranged tepals are derived directly from the axial cylinder; the marginal veins, mr, of the adjacent tepals, however, are formed by four commissural strands, com. Stamens, sf, receive a single strand each of which broadens slightly while traversing the connective. The remainder of the vascular tissue of the axis, consisting generally of eight to 12 strands, supplies the gynoecium, g. Two of these strands fuse together to form the “ventral” vein which runs vertically in the tissue separating the two locules, v in Text-figs. 2 and 3, and finally bifurcates, each branch supplying one ovule. The other strands of the gynoecium traverse the ovary wall and in the region of the stigma undergo consider- able proliferation and anastomosis, Text-fig. 2. The gynoecium in all the species of 7denburgia and in Nouhuysia exam- ined by us is bilocular, lodging a single anatropous ovule in each locule, Text-fig. 1. The funicle is rather massive and characteristically papillose externally. Generally both ovules develop to produce two-seeded fruits. In certain collections, however, single-seeded fruits have been described (3,6). An examination of such specimens has revealed that this condi- tion is due to a failure of one of the ovules to develop to maturity. The seed structure of the two genera under consideration are remarkably alike. the most noteworthy feature being the pronounced centripetal ruminations of the stony layer of the seed, Fig. 17. 1953] BAILEY & SWAMY, IDENBURGIA AND NOUHUYSIA 81 acorns ocean, i ee Pas cece cetememes tame rm mene ~ = te en anne ota ae cane Hoes wo sve ene nememt teses re one? ee cn ee emt tee —_——" / eee ene, ee, Fic. 1. Gynoecium of /denburgia cut longitudinally to expose the ovules. Fic. 2. Longitudinally cut half of the gynoecium of Jdenburgia showing the pat- tern of vasculature. v = ventral vein. Fic. 3. A generalized vascular diagram of the flower of Jdenburgia. com: = commissural strand; g = gynoecium; if = inner Ae md = median strand; mr = marginal strand; ot = outer tepal; st = stamen; v = a vein 82 JOURNAL OF THE ARNOLD ARBORETUM {| vot. xxxtv POLLEN Erdtman is quoted by van Steenis as stating that the pollen of Koster- mans 2198 is 3-porate, oblate (14.5 & 20.5), with an exine 1.5, in thick- ness, pattern + obscure. We have examined the pollen from eight dif- ferent collections. Gibbs 5654 and Kostermans 2198 are characterized by dominantly oblate, 3-porate pollen, Fig. 3, with a minor admixture of 4-porate grains. The inner exine is greenish and highly refractive in lactic acid and the outer exine exhibits a barely detectable sculpture in surface view. The pollen of Rutten 2240 is similar but differs in being dominantly 4-porate with a minor admixture of 5-porate grains. Clemens 3828 (type of J. pauciflora A. C. Sm.) is dominantly 3-porate but differs in its con- spicuously smaller size. Kostermans 2217 (1. arfakensis Gibbs) and Clemens 2422 have a conspicuously granular- ae outer exine in sur- face view, the former being prevailingly 3-porate, and the latter 4-porate and conspicuously smaller, Fig. 5. In marked contrast to the pollen of these collections, is that from Brass 12661 and Clemens 5499b which is inaperturate, spherical and much larger, Fig. 4. It is significant, however, that all of the various forms of pollen are characterized by having a rela- tively thick, conspicuously hyaline inner exine. DISCUSSION The combination of exomorphic and endomorphic characters of both the reproductive and vegetative parts is closely similar in all collections, and there can be no doubt that /denburgia and Nouhuysia are congeneric. Such morphological variations as occur tend to intergrade more or less extensively, V. pauciflora (A. C. Sm.) van Steenis being at one extreme of variability, for example in hairiness, and NV. arfakensis (Gibbs) van Steenis in leaf form. When these specimens are excluded as easily recognizable species, the remaining collections of Nouhkuysia present a problem of con- siderable taxonomic difficulty. Should all of them be assigned to one variable species, V. papuana Lauterbach, as concluded by van Steenis, or should additional species be segregated as attempted by Gibbs and by Gilg and Schlechter? Unfortunately, such conspicuous differences in the internal structure of leaves as those illustrated in Figs. 7, 2 and 7 are at present of questionable utility. The structural differences between sun-leaves and shade-leaves trasted with arid, environments are known to be considerable. Therefore extensive additional collections must be made in order to determine whether such foliar structural differences are due to genetic factors or merely to the effects of varying environmental factors. It is significant in this con- nection, however, that the differences in pollen morphology, illustrated in Figs. 3 and 4, exceed the limits of variability that may be anticipated in a single species. Unfortunately, owing to the limited number of avail- able collections with mature pollen. it is not possible at present to 1953| BAILEY & SWAMY, IDENBURGIA AND NOUHUYSIA 83 establish valid correlations between differences in pollen morphology and other diagnostic criteria. Although the available material is inadequate for a satisfactory solution of the problem of speciation within Nouhuysia, it provides a summation of evidence that is significant in any discussion of the relationships of the genus. That the genus does not belong in, or in close relationship to, the Trimeniaceae or Monimiaceae is indicated by fundamentally significant morphological and anatomical differences in all of its organs. “Ethereal oil cells,’ which occur so characteristically throughout the Monimiaceae and the Trimeniaceae, are absent, as are large mucilage cells of the type which are a significant diagnostic criterion in Trimenia and Piptocalyx. Crystallization of calcium oxalate in the form of styloids, which is so typical of Nouhuysia, is alien to the Monimiaceae and allied families. Further- more, the genus has a dominantly trilacunar nodal anatomy in contrast to the unilacunar structure which occurs throughout the Monimiaceae, Lau- raceae, Gomortegaceae, Trimeniaceae, Amboreltaceae and Austrobailey- aceae. In addition, the pollen morphology of NouAuysia is fundamen- tally unlike that of these families. Although Trimenia, Piptocalyx and many genera of the Monimiaceae have retained a relatively primitive form of secondary xylem, they exhibit a tendency for the elimination of wood parenchyma and for the formation of septate fibers that is absent in Nouhuysia. The similarities in the flowers are superficial. The form and particularly the vasculature and other internal structures of the tepals, stamens and pistil of Nouhuysia prove upon detailed examination to be fundamentally different from those of Trimenia and Piptocalyx. Similarly a summation of available evidence appears to exclude Nou- huysia from close relationship to the Guttiferae. The genus lacks the schizogenous secretory receptacles and the subsidiary cells oriented parallel to the guard cells of the stomata that are such characteristic diagnostic features of that family. The xylem is at a conspicuously more primitive level of structural specialization. Furthermore, the trilacunar nodal struc- ture is alien to the Guttiferae which are dominantly unilacunar. In addi- tion, styloids are not known to occur in the Guttiferae whose form of crystallization is of a different type. Nor have outstanding similarities in the reproductive parts been described which might be interpreted as neutralizing such significant differences. s in the case of Calyptosepalum, discussed in another paper (1), Nouhuysia exhibits a combination of reproductive and vegetative charac- ters that negates its inclusion at present in any existing family. It appears to be another of the numerous relics of an ancient woody dicotyledonous flora that have survived in New Guinea, New Caledonia, Fiji and adjacent regions. What to do with such genera as they successively come to light is a problem of major taxonomic importance. When thoroughly investi- gated, they promise to reveal much new evidence regarding the morpholog- ical characteristics of primitive angiosperms, and, therefore, should not be relegated to a position where they become lost from view. To refer them at once to existing families not only tends to conceal them, but also is 84 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxIv extremely misleading in many cases. Lauterbach’s (4) brief description and premature reference of NouAuysia to the Guttiferae is undoubtedly responsible in part for the genus being overlooked and for the description of subsequent collections as species of a new genus, viz. /denburgia Past experience with relic genera, e.g. Casuarina, Degeneria, Gina dra, Amborella, Trochodendron, Tetracentron, Eupteles. Tllictum, Cercidi- phyllum, etc., indicates that many, if not most, of them will have to be placed eventually in separate families, regardless of the results of such a procedure in multiplying the number of small families. This should not be attempted, however, until adequate collections of material are available for detailed and comprehensive investigations of all parts of the plant or until there is a summation of evidence which justifies the exclusion of a genus from existing families and the establishment of a new one. In the meantime, it appears wiser to place such relic genera in a special category of plants of uncertain or undetermined affinities comparable to the Fungi Imperfecti of mycologists rather than to continue to toss them about from one exist- ing family to another as is commonly done at present. Certain relic genera, when thoroughly studied, prove to be related to existing families, e.g. Degeneria and Himantandra to the Magnoliaceae or Amborella, Trimenia, Gomortega and Austrobaileya to the Monimiaceae and Lauraceae, and therefore may justifiably be placed in independent families in close proximity to them. Others, e.g. Casuarina, Cercidiphyllum, Euptelea, Trochodendron and Tetracentron, appear to be relics of more remotely related plants having no close surviving relatives and had best be retained in a general category of uncertain affinities, at least until their true relationships are revealed by paleobotanical evidence he genus Vouhuysia belongs for the time being at least in such a cate- gory of plants of uncertain or undetermined affinities. ACKNOWLEDGMENTS We are indebted to the Arnold Arboretum for the loan of all of its collections of Jdenburgia and Nouhuysia, and to Professor van Steenis for his kindness in sending us pollen-bearing flowers of Gibbs 5654, Rutten 2240 and Kostermans 2217. The junior author wishes to express his keen appreciation of helpful privileges extended to him by the Director and the staff of the herbarium at Kew. LITERATURE CITED — Bartey, I. W. anp A. C. SmitH. A new Fijian species of Calyptosepalum Jour. Arnold Arb. 34: 52-66. 1953. 2. Gisps, L. S. A contribution to the phytogeography and flora of the Arfak Mountains, etc. 135-140. London, Taylor and Frances. 7 3. Gitc, E. AnD R. SCHLECHTER. Die Monimiaceen-Gattung Idenburgia. Eng- ler’s Bot. Jahrb. 58: 244-248. 1923. 4. LAUTERBACH, C. Guttiferae. In Lorentz’s Nova Guinea. 8: 843. 1912. 1953] BAILEY & SWAMY, IDENBURGIA AND NOUHUYSIA 85 5. Money, L. L., I. W. Bartey anp B. G. L. Swamy. The morphology and relationships of the Monimiaceae. Jour. Arnold Arb. 31: 372-404. 1950. 6. SmituH, A. C. Studies of Papuasian plants. II. Jour. Arnold Arb. 22: 231- 252. 1941. 7. VAN STEENIS, C. G. G. J. Reduction : two endemic monotypic Papuan genera. Acks Bot. Neerl. 1: 93-98. 1952. EXPLANATION OF PLATES PLATE I Fic. 1. Clemens 54996. Cleared leaf, showing large intercellular spaces in the mesophyll and heavy thickening of the walls of its constituent cells, & 150. Fic. 2. Clemens 12353. Cleared leaf, showing thin-walled mesophyll and scat- tered sclereids, X 150. Fic. 3. Kostermans 2198. Oblate, triporate pollen, x 440. Fic. 4. Brass 12661. Spheroidal, inaperturate pollen, x 440. Clemens 2422. Oblate, tetraporate pollen, & 440. Fic. 6. N.J.F.S. bb. 25014. Cleared leaf, showing styloid, * 150. Fic. 7. Kostermans 2198. Cleared leaf, showing thick- walled mesophyll and relatively small intercellular spaces, * 150. PLATE II Kostermans 2198. Cleared leaf, showing stomata and form of the sur- ee epidermal cells, X 310. Fic. 9. Brass 12661. Cleared leaf, showing stomata and form of the surrounding epidermal cells, & 310. Fic. 10. Clemens 3978. Transverse section of the stem at the level of a trilacunar node, X 11. Fic. 11. Clemens 2240. Transverse section of a fruit with two seeds, Fic. 12. Clemens 3978. Transverse section of the mid-vein of the leaf, showing form of vascular tissues, X 19. Fic. 13. Clemens 2422. Transverse section of the secondary xylem, & 40. Fic. 14. Clemens 2422. a sei aac section of the xylem, showing the scalariform perforations, * 1 BIOLOGICAL LABORATORIES, HARVARD UNIVERSITY. Jour. ARNoLp Ars. VoL, XXXIV PLATE I BAILEY & SwWAMy, IDENBURGIA AND NOUHUYSIA PLaTE II Jour. ARNOLD ArB. VoL. XXXIV * ™ eae SS “<. Call ae ARE BE EN aye 2% Cie -, IDENBURGIA AND NOUHUYSIA BAILEY & SWAMY, 88 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv THE VEGETATION ON COASTAL DOGTOOTH LIMESTONE IN SOUTHERN CUBA RIcHARD A. HowarD AND WINSLOW R. Briccs With One Plate ONE OF THE GEOLOGIC FEATURES of the southern coast of Las Villas province in Central Cuba is an uplifted bench of weathered limestone known in Cuba as diente de perro or dogtooth limestone. The name describes well the lacerated, sharp-edged, hard, metallic-ringing limestone rock (Pv. I, Fic. 1). Similar formations exist in a few other spots in Cuba, on the islands off Haiti and the Dominican Republic, and near Ponce in Puerto Rico; but they appear to be found principally on the southern coasts. They are missing in the Lesser Antilles and Jamaica, and are not comparable to the limestone formations of the Bahamas. The formation is covered with a tree and shrub vegetation which has never been described adequately. Gleason and Cook (Scientific Survey of Puerto Rico and the Virgin Islands 7: 158. 1927) describe briefly a “xerophytic forest” on the Ponce limestone. Ekman (Ark. for Botanik 22A(16) : 4. 1929) lists the components of this vegetation as found on Navassa Island. The senior author has considerable unpublished data for the vegetation of Beata Island, off Hispaniola, which is also growing on dogtooth limestone. These four areas seem comparable in aspect of the vegetation and ie dei but the floristic composition of the vegetation is different in each c The present paper records the elements comprising the ii woody vegetation on the dogtooth limestone in Cuba. Coastal exposures of dogtooth limestone are found in Las Villas province on either side of the entrance to Cienfuegos Bay. The exposures, paralleling the coast, extend east from Punta Caballos to the mouth of the San Juan river, and west from Castillo de Jagua to the edge of the Cienaga de Zapata. The flora in this area has been collected by several botanists in the past few decades. John George Jack of the Arnold Arboretum collected intensively during the late 1920’s in the region discussed in this paper, but he failed to leave any significant notes on the composition of the woodland formation. Combs (Trans. Acad. Sci. St. Louis, 7: 393-491. 1897) studied the area west of Cienfuegos Bay, primarily around Castillo de Jagua, but he too failed to describe the vegetation. Ekman apparently never visited the area during his work in Cuba, and Leon and his associates have not de- scribed it. The area selected for study is a portion of the Gavilan tract owned by the Publication No. 11, Journal Series from the Atkins Garden and Research Laboratory of Harvard University Soledad, Cienfuegos, Cuba 1953 | HOWARD & BRIGGS. DOGTOOTH LIMESTONE 89 Central Soledad. It is west of the Rio Gavilan and east of Punta Lobos and the mouth of Cienfuegos Bay. The specific area is known locally as the Potrero Seboruco and is adjacent to the small Playa de Gavilan. The south coast of Las Villas province in this area has an undulating shoreline. In a very few of the embayments are sandy beaches. The major portion of the shoreline consists of limestone benches slightly below, at, or slightly (1-3 feet) above sea level. The benches represent a relatively small vertical land adjustment of .recent origin. When exposed they are covered with a low shrub growth, littoral in nature, consisting of Rachicallis americana, Strumpfia maritima, and Conocarpus erectus. Directly behind this coastal formation is a sharply uplifted block of limestone of Miocene age. It is the vegetation of this block that is considered in the present paper. The average uplift appears to be between 15 and 25 feet. The southern edge of the block is wave cut, with a series of erosion caverns indicating that the area, within relatively recent time, was at the sea margin. The uplifted limestone block dips slightly to the north, and ends in an open lateritic plain of reddish soil with a savannah type of vegetation. The limestone block parallels the coast, and the region studied is approximately one-half mile wide. The formation has been cracked by subsequent land movements. Some of the larger sections are flat-topped, perfectly smooth and show little sign of weathering except at the fractures. At the coastal margins, however, sections of the block are severely cracked and show both large and small chasms, narrow and wide, extending down to sea level. One chasm (Pt. I, Fic. 2) is 150 feet long and 25 feet deep and averages 25 feet in width. In this chasm a shallow lake of brackish water has developed. The bottom of the shallow lake is of soft debris and muck, and it was impossible to walk in it. The smaller sections of the main block, especially those at the southern edge, are severely weathered with each section rounded to the middle (Px. I, Fic. 1). The rock is extremely hard and the results of weathering have been peculiar. A matrix of extremely sharp and hard ridges and spines has been developed (PL. I, Fic. 1). According to Lewis (Amer. Assn. Petr. Geol. Bull. 16: 533-555. 1932) this weathering is not due to the erosion of a softer rock from a harder matrix, as the rock is essentially homogeneous, but rather due to extremes of temperature, humidity,. and rainfall, external physical conditions. Loose fragments of the rock will ring with a characteristic metallic sound when struck. Footing is treach- erous on these naked sharp ridges, and the name “dogtooth limestone” for such an area is well deserved. Similar lacerated surfaces are found only on the edges of the larger fracture blocks. The entire surface in this area can be described as naked rock. The mass is so cracked and weathered into a porous structure that only rarely have very small, cup sized pockets of soil accumulated. The leaves and branches present on the surface are dry and no humus has been formed. However inhospitable the area seems, the limestone block is covered with a shrub and tree vegetation. In most areas this vegetation is open. The trees average 35 feet in height and are unbranched for the first 15 feet. The 90 JOURNAL OF THE ARNOLD ARBORETUM - |[voL. xxxiv boles are thick and massive for the heights of the trees, and are generally crooked. The crowns are symmetrical and the effects of the wind have not caused the lopsided growth expected in coastal vegetation. The shrub vegetation is approximately 12 feet in height, usually densely branched and predominantly spiny. Herbaceous vegetation is almost entirely lacking and no ferns were seen. Shade loving plants are conspicuously absent under the thin and open canopy. Epiphytes are relatively abundant but only members of the Bromeliaceae were found. Vines are few but the individual plants are usually extensive. Root development of all plants found on the dogtooth limestone is con- spicuous, perhaps because one is not accustomed to seeing the roots of tropical trees. Unless the plants are growing in or near a rock chasm the roots usually crawled along the surface to a distance greater than the radius of the crown before penetrating into cracks in the rocky substratum. The trees seem to be successful in such an area because the roots are able to penetrate to a depth where moisture is available. In order to determine accurately the composition of this vegetation we ran a transect through the vegetation on the limestone outcrop. The senior author ' had surveyed the region several years before in order to determine the most undisturbed locality and to learn the nature of the flora. Several survey trips were made before and after the date of the transect to ascertain the extent to which the transect was typical. It was determined that the area studied was characteristic of the vegetation occurring on the lime- stone, and that relatively few more species were to be expected than were seen on the transect. The transect made was 880 yards long and 2 yards wide. Every plant within this strip, the area of which is 14 acre, was counted. Stem diameters of all woody species were determined at breast height. The vegetation of the transect area consists of 74 species and 521 indi- vidual plants. excluding only the epiphytes and parasites. The detailed composition is shown in the following table: Trees 22 species 29.7% of total — 231 individuals 44.3% of total Shrubs 35 species 47.2% of total — 224 individuals 42.9% of total Herbs: 9 species 12.1% of total — 30 individuals 5.7% of total Vines 8 species 10.8% of total — 36 individuals 6.9% of total Total 74 species 99.8% 521 individuals 99.8% While shrubs dominate the area from the standpoint of the number of species involved, the trees are slightly more significant from the standpoint of number of plants. The herbaceous species, vines, and epiphytes make up a relatively unimportant aspect of the vegetation on the dogtooth lime- stone. ‘The authors are grateful for the assistance of the other members of the course in Tropical Botany of Harvard University who helped on this survey. Special apprecia- tion is acknowledged for the suggestions and help of Mr. Earl E. Smith in the recording method used. They also wish to thank the management of Central Soledad for permission to study the area. SPECIES Gymnanthes lucida Cordia oe Leucaena glauc Hebestigma ere oe Wrightii Adelia ricinella Tabebuia peropil Torrubia disco Exostema oa Capparis cynaphallophori Colubrina is Vitex divar Bursera sara Luhea spec Hypelate a Securinega Acidoton Rondeletia pedicellaris THE SIGNIFICANT COMPONENTS OF THE'V LESS THAN nee 4” Se id 6” 7 8” yg? 6a «| ag 135” 16% Torar 31 18 9 2 60 3 ; 1 40 26 1 3 1 1 32 10 1 4 l 4 1 2 1 2 26 6 3 l 10 5 2 | 1 9 7 1 8 5 1 ! 7 1 2 ! 1 1 | 7 3 1 1 5 4 l 5 5 l ! 5 l 1 1 3 1 1 2 1 1 1 1 = J te ad 1 1] ss 1 é : 1 ION 4 DOGTOOTH LIMESTONE IN CUBA LISTED BY DIAMETERS OF THE BOLES AND EQUENCY OF OCCURRENCE. lesol ANOLSAWIT HLOOLDOC ‘SODIAd ® GAVMOH 92 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxtv The trees were tabulated for their frequency and size. All trees with a trunk diameter of less than three inches were considered together. Any plant with an unbranched bole of ten feet was considered to be a tree possessing a potentially useful log. Casearia sylvestris, represented by three specimens, and Forestiera laevigatus by one specimen, have the necessary length of trunk but none of the specimens exceed three inches in diameter at breast height. One large specimen of Ficus laevigatus defied classification in possessing multiple trunks, several of which were eight inches in diameter. A lone specimen of Ceiba pentandra towers over the rest of the trees on this limestone and has a trunk diameter of 36 inches; however, the wood of the Cezba is considered worthless. The remaining 18 arboreal species are listed in the following table, in order of their frequency of occurrence. All size figures represent diameters at breast height It can be seen from the table that Hebestigma cubensis and Torrubia discolor are the only potential timber trees in the transect. Gymnanthes lucida, Cordia Gerascanthus, and Leucaena glauca appear to be the only trees seeding themselves, as indicated in the number of saplings en- countered. None of the species in the above list is currently considered valuable as a timber tree. The shrubs of the transect have trunks less than six feet long to the first branch, or are characteristically bushy in appearance. They comprise the following species. The number of individuals counted in the transect follows each species. Erythroxylum areolatum 72 Lantana reticulata 4 Argythamnia candicans 28 Leucocroton microphylla 3 Pithecellobium Hysterix 15 Capparis flexuosa 3 Cordia globosa 14 Schaefferia frutescens 3 Croton lucidus 14 Malpighia cubensis 3 Hyperbaena racemosa 13 Guettarda elliptica 3 Acacia hie tana 10 Erythroxylum rotundifolium 3 Duranta repens 7 Plumiera obtusa 3 Capparis Grisebachii 5 Randia spinifex 2 Boerreria succulenta 4 ia specimens of the following species were found: Citharexylum spinosum, Amyris elemifera, Boerreria virgata, Caesalpinia glaucophylla. Phyllanthus neopeltandrus, Guettarda Combsti, Bumelia glomerata, Guaia- cum sanctum, Savia sessiliflora, Comocladia deniata, Guettarda calyptrata, Anthacanthus tetrastichus, Pluchea odorata, Rhacoma Crossopetalum, Celtis iguanaea, and Morinda Royoc. Erythroxylum areolatum dominates the shrubs in the transect area. The plants of this species average eight feet in height and fruit heavily. They usually occur in groups. It is probable that birds are responsible for the dispersal of the red-colored fruits. Argythamnia candicans, generally a 3—4’ woody shrub, is represented by hundreds of seedlings. The fruits of this member of the Euphorbiaceae are elastically dehiscent. Argythamnia 1953 | HOWARD & BRIGGS, DOGTOOTH LIMESTONE 93 is the only plant that appears to be spreading rapidly. Comocladia dentata is more abundant outside of the transect area, particularly in the inland lateritic soil zone. Celtis iguanaea is represented by one shrub, rooted far down in one of the larger chasms and climbing on the limestone rocks before scrambling into the tree tops. Some of the branches of this plant are 2” in diameter and reach a length of 25 feet. The herbaceous plants and cacti are represented by nine species. Hibiscus pilosus is most abundant with individuals counted. Setaria distantiflorum, represented by three individuals in the transect area, is growing on very small pockets of soil in the dogtooth limestone. Aventis pulsilla, Acalypha chamaedrifolia and Portulaca pilosa are all on the limestone, but only single individuals were seen. Sterile plants of Harrisia, Selenicereus, and Cephalocereus were found. Opuntia Dillenii is represented by one flowering and fruiting plant which forms an extensive sprawling mass. Of the eight species of vines in the transect area Stigmaphyllon Sagraea- num is the most abundant with fifteen individuals counted. Serjania subdentata was found nine times, and Triopteris rigida and Passiflora suberosa four times for each. Two extensive plants of Acacia tenuifolia, a viciously spiny species were found, and single specimens of Urechites lutea, Gouania polygama and Tournefortia peruviana. One parasite, Phthirusa purpurea was found growing on Tabebuia pterophila. Three species of Tillandsia were the only epiphytes seen. These plants are numerous and were not considered in the transect count. Tillandsia fasciculata is the most abundant, while 7. flexuosa and T. tenuifolia are infrequent. In an area such as the dogtooth limestone, where soil cover is lacking, moisture penetration is high, the canopy thin, and the temperature high, evaporation and transpiration are assumed to be critical factors in the environment. Various morphological specializations are expected in plants capable of living in such an environment. These specializations are gen- erally expressed in reduced total leaf surface, microphyllous leaves which are usually associated with the presence of spines, thick cuticles, and ilicioid leaves. If the leaves are large, they usually have soft laminal tissues and are apt to wilt and droop easily. Of the species found in this area, five of the trees, five of the shrubs, and one of the vines possess microphylls. Such leaves are generally of hard thick tissues. Spines, either modified branches or of stipular origin, are found in one of the trees, seven of the shrubs and one of the vines. Ilicioid leaves occur in one species of the trees and one of the shrubs. The wilting habit was especially obvious. On the hot sunny day when the transect was made nearly 50% of the broad leafed species appeared to be in a wilting condition. Exact rainfall data for this specific area are not available. The site is to the west of the normal path of rain storms from the Trinidad-San Juan mountains. The rainfall is probably in the range of 30-35 inches, which is commonly regarded as insufficient, in the Caribbean area, to support other than a thorn-shrub type of woodland. The lack of surface accumula- O4 JOURNAL OF THE ARNOLD ARBORETUM _ [VOL. XxxIVv tion of soil and moisture may account for the lack of herbaceous plants and the deep penetration of the roots of the shrubs and trees. It is probable that the woody vegetation found on the dogtooth lime- stone is a fairly stable type for this habitat under existing environmental conditions. While the sea side cliff development indicates former proximity to the Caribbean Sea there is, at present, no indication of the littoral or halophyte shrub vegetation characteristic of the low limestone coastal formations in Cuba The current vegetation of the area meets the description of a thorn woodland proposed by Beard (Ecology 25: 140. 1944) but does not fit into any of the fascies he suggested. In general aspect it agrees with the descriptions published by Ekman for the dogtooth limestone areas of Navassa Island, and by Gleason and Cook for the Ponce limestone. In floristic composition and species dominance, however, it is quite different. Ekman described the forests of the dogtooth limestone of Navassa Island as “low, the trees are stunted, though often with thick trunks. Characteris- tic of Navassa is the small number of species involved. Only four species of trees could be labelled as common, to wit, Ficus populnea, Sideroxylon foetidissimum, Coccoloba laurifolia, and Metopium Browne. Scatter occur: Pseudophoenix navassana, Bumelia navassana, while Fagara martinicensis and Colubrina feruginosa are rare. Shrubs are few, e.g. Pisonia discolor var., Schoepfia obovata, and Duranta erecta, while vines and epiphytes are nearly absent.” (Ark. for Botanik, 22A (16): 4. 1929). Gleason and Cook (Sci. Surv. Puerto Rico and Virgin Is. 7: 158. 1927) report the composition of the “‘xerophytic forest” on the Ponce limestone as follows: ‘Trees, Bursera simaruba, the commonest trees, Bucida buceras, Ficus laevigatus, Amyris elemifera, Capparis cynophallophora, and Pisonia albida. Rare or scattered trees of Guaiacum officinale and Tabebuia heterophylla. The shrub layer comprises a large number of species of which Lantana involucrata, Coccolobis laurifolia, Helicteres jamaicensis seem to be the most abundant. Other common species are Moluchia tomentosa, Croton rigidus, Croton lucidus, Exostema caribaeum, Varronia angustifolia, Savia_ sessiliflora, Pithecellobium Unguis-cati, Ricinella Ricinella, Eugenia ligustrina, Eugenia buxifolia, Schaefferta frutescens, Hypelate trifoliata, Reynosia uncinata, Samyda dodecandra, Citharexylum fruticosum, Plumeria alba, Adelia Bernardia, Krugiodendron ferreum, Elaeodendrum xylocarpum, Comocladia Dodonea, Canella Winteriana, Jacquinia Berterii and Tournefortia microphylla.” A comparison of the list of plants found on the Las Villas dogtooth limestone to those listed above by Ekman and by Gleason and Cook emphasizes the similarity of aspect among the dogtooth limestone floras as well as the differences in species and dominants comprising them. THE BIOLOGICAL LABORATORIES, HARVARD UNIVERSITY. 1953] HOWARD & BRIGGS, DOGTOOTH LIMESTONE 95 EXPLANATION OF PLATE Fic. 1. The face of one of the smaller chasms in the Gavilan dogtooth lime- stone showing the many cracks and the eroded surface with sharp projections. Fic. 2. A larger chasm in the limestone block showing a shallow brackish lake. Jour. ARNOLD Ars. VoL. XXXIV ‘ Ye ye A PLATE I ETATION ON DoGTooTH LIMESTONE G Howarp & Briccs, VE JOURNAL OF THE ARNOLD ARBORETUM VoL. XXXIV APRIL 1953 NUMBER 2 STUDIES OF PACIFIC ISLAND PLANTS, XVI NOTES ON FIJIAN RUBIACEAE ! A. C. SMITH THE PRESENT PAPER is intended primarily to place on record the appar- ently undescribed species of Fijian Rubiaceae collected by the writer in 1947.2, Twenty species are here described as new, in the genera Gardenia, Ixora, and Psychotria. No effort is here made toward a revision of the family on a local scale, but field observation inclines me to believe that such a revision will not prove difficult. When all the Fijian types are brought together for comparison with the abundant material now available, it should be possible to delimit and correlate the local species with accuracy. In spite of the fact that it is the largest family of phanerogams in Fiji, the Rubiaceae does not present insurmountable taxonomic problems. Differ- ences of opinion as to the rank of categories may remain for some time, but the taxa appear in general to be well delimited. The place of deposit of specimens is indicated by: A (Arnold Arbore- tum), US (U.S. National Museum). HEDYOTIS L. Hedyotis tenuifolia Sm. in Rees, Cycl. 17: no. 19. 1811. Oldenlandia tenuifolia sensu Forst. f. Fl. Ins. Austr. Prodr. 11. 1786; non Burm. FI. Ind. 37. pl. 14, fig. 1. 1768. Vanua Levu: Mathuata: Seanggangga Plateau, in drainage of Korovuli River, vicinity of Natua, alt. 100-200 m., Smith 6886 (A, US) (on edge of pond in open rolling country; corolla white). I am indebted to Dr. F. R. Fosberg for pointing out the identity of this interesting little plant, which, in my observation, is rare in Fiji; at least 1No. XV of this series was published as vol. 30, part 5 (pages 523-573), of Contribu- tions from the U. S. National Herbarium, 1953. * Under the auspices of the Arnold Arboretum of Harvard University and the John Simon Guggenheim Memorial Foundation, with the aid of grants from the Penrose Fund of the American Philosophical Society and the Bache Fund of the National Academy of Sciences. 98 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxIv I have seen it only in the cited locality. However, Seemann noted the species as “Common on roadsides throughout Viti.” Smith’s binomial of 1811 is based upon Forster’s interpretation of Oldenlandia tenuifolia Burm. and his concept is based entirely upon For- ster’s specimen from Tanna, in the New Hebrides. Oldenlandia tenuifolia Burm. is said by Smith to be a very different plant, which he refers to Hedyotis herbacea. Of course, by modern nomenclatural practice, Smith could not have based his binomial upon Forster’s interpretation of an earlier name to the exclusion of the actual type of that concept, and therefore Forster cannot be cited as the parenthetical author. However, there is no obstacle to the acceptance of Smith’s Hedyotis tenuifolia as a new species, dating from 1811 and based entirely upon Forster’s Tanna specimen. Pre- vious mention of the occurrence of the species (as Oldenlandia tenuifolia Forst. non Burm.) in the New Hebrides and Fiji is to be found also in the following works: DC. Prodr. 4: 425. 1830; Seem. FI. Vit. 126. 1866; Drake, Ill. Fl. Ins. Pac. 186. 1886; Guillaumin in Bull. Soc. Bot. Fr. 74: 698. 1927. GARDENIA Ellis Gardenia candida sp. nov. Arbor ad 7 m. alta, ramulis gracilibus, inferne subteretibus cinereis glabris, internodiis distalibus leviter angulatis 2-3 mm. diametro pilis palli- dis 0.1-0.3 mm. longis puberulis; stipulis longe persistentibus in tubum 9-11 mm. longum pilis ad 0.5 mm. longis puberulum vel setulosum apice undulatum connatis; petiolis liberis semiteretibus superne leviter canali- culatis 2.5—3.5 cm. longis ut ramulis juvenilibus puberulis; foliorum laminis chartaceis in sicco viridibus, ellipticis, (12—) 15-20 cm. longis, (5—) 6-9 cm. latis, basi acutis et in petiolum decurrentibus, in acuminem obtusum ad 1 cm. longum apice cuspidatis, margine integris, supra costa nervisque secundariis plus minusve puberulis exceptis glabris, subtus pilis patentibus pallidis 0.2-0.4 mm. longis uniformiter et persistenter puberulis, costa valida supra elevata subtus prominente, nervis secundariis utrinsecus 15—17 paten- tibus marginem versus curvatis et interconnexis utrinque elevatis, nervis tertiariis numerosis inter secundarios transversis et rete venularum utrinque prominulis; floribus magnis in axillis distalibus solitariis, pedicello crasso subtereti sub anthesi 1—-1.5 cm. longo cum calyce pilis pallidis 0.3—0.7 mm longis copiose patenti-piloso; calyce infundibulari sub anthesi lobis inclusis 3.5-4 mm. longo et apicem versus circiter 2.5 mm. diametro, tubo parvo obconico, limbo chartaceo vel subcoriaceo elongato dorso 4-alato, alis semiobovatis circiter 2.5 cm. longis et 1 cm. latis conspicue nervatis apice retundatis apicem limbi ipsi scariosum et undulatum 5—10 mm. excedenti- bus; corolla in vivo subcarnosa hypocrateriformi, tubo 4.5—5 cm. longo et basi circiter 4 mm. diametro faucibus ad 1 cm. ampliato, extus glabro vel obscure puberulo, intus ad lineas staminibus alternatas pilis debilibus adscendentibus 1-2 mm. longis copiose piloso, limbo patenti sub anthesi rotato 9-11 cm. diametro ad basim 8-lobato, lobis late imbricatis oblongis 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 99 ad 5 cm. longis et 2—2.5 cm. latis apice rotundatis; staminibus 8, antheris sessilibus anguste oblongis circiter 17 mm. longis et 1.5 mm. latis, dorso complanatis, a tubo corollae leviter exsertis; stylo crasso clavato quam tubo corollae paullo breviore demum in stigmata 4 circiter 1 cm. longa dorso minute strigillosos fisso. Vanua Levu: Mathuata: Seanggangga Plateau, in drainage of Korovuli River, vicinity of Natua, alt. 100-200 m., Nov. 25, 1947, Smith 6640 (A TYPE, US) (ndrenga; tree 7 m. high, in patches of forest in open rolling country; corolla-lobes pure white, the tube greenish). Species of Gardenia are among the more obvious components of the flora of the Mathuata coast of Vanua Levu. Their abundance in the patches of forest that occur here and there in this comparatively dry region, together with the pervasive fragrance of their flowers, imparts a characteristic charm to the landscape. Species known to occur in Mathuata are G. vitiensis Seem., G. gordonii Baker, G. grievei Horne ex Baker, and G. Aillii Horne ex Baker. The species here described cannot be confused with any of these nor with the several others known from Viti Levu. Superficially G. candida is most like G. gordonii, but it differs in the indument of its leaves, pedicels, and calyx, and in its even larger flowers. MASTIXIODENDRON Melchior Mastixiodendron pilosum A. C. Sm. in Jour. Arnold Arb. 26: 109. 1945. Vanua Levu: Mathuata: Southern base of Mathuata Range, north of Natua, alt. 100-250 m., Smith 6781 (A, US) (slender tree 8 m. high, in dense forest; mature fruit dull yellow). The third known collection of the species agrees very well with the two earlier ones, which are similarly in fruiting condition. Previously the species has been known with certainty only from the province of Mbua, Vanua Levu. IXORA L. Ixora pedionoma sp. nov, Arbor ad 2 m. alta, ramulis gracilibus apices versus complanatis 1—1.5 mm. diametro pilis 0.1-0.3 mm. longis patenti-puberulis, vetustioribus subteretibus cinereis glabratis, internodiis 5-10 mm. longis; stipulis liberis e basi late ovato aristatis 4-5 mm. longis ut ramulis puberulis; foliis glabris subsessilibus, petiolis canaliculatis 1-3 mm. longis, laminis chartaceis siccitate olivaceis lanceolatis, 4-9 cm. longis, 1-2 cm. latis, basi attenuatis et in petiolum decurrentibus, apice in mucronem callosum 0.5—1 mm. longum angustatis, margine integris leviter incrassatis, utrinque levibus, costa supra valde elevata subtus prominente, nervis secundariis utrinsecus 8-12 subadscendentibus utrinque prominulis vel supra planis, rete venu- larum immerso vel subtus prominulo; inflorescentiis in ramulis brevibus terminalibus 3-floris, basi bracteis binis foliaceis subtentis, bracteis papy- raceis ovatis, (15—) 18-25 mm. longis, (10—) 12-15 mm. latis, basi sub- 100 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv cordatis, apice cuspidatis, reticulato-nervatis; pedicellis teretibus 1—1.5 mm. longis ut calyce copiose pallido-puberulis; calyce sub anthesi circiter 3 mm. longo, limbo erecto, lobis 4 deltoideis obtusis haud 0.5 mm. longis; corolla hypocrateriformi glabra, tubo gracillimo (haud 1 mm. diametro) circiter 25 mm. longo, lobis 4 patentibus oblongis subacutis 8—8.5 mm. longis circiter 3 mm. latis; staminibus exsertis glabris, filamentis ligulatis 1—1.5 mm. longis, antheris anguste oblongis 4-5 mm. longis; stylo gracil- limo exserto parte inclusa puberulo, stigmatibus complanatis circiter 2 mm. longis; fructibus ellipsoideis ad 5 mm. diametro persistenter puberulis calycis limbo coronatis. VaNuUA Levu: Mathuata: Seanggangga rg in drainage of Korovuli River, vicinity of Natua, alt. 100-200 m., Nov. 28, 1947, Smith 6703 (A TYPE, US) (shrub 1—2 m. high, along river in patches : forest in open rolling country; corolla and style white). he new species (of § PAylleilema) is a very close relative of /. nandari- vatensis Gillespie, in comparison with which it has its branchlets distally and its shorter stipules puberulent, its leaves comparatively short-petioled, proportionately narrower, and smooth rather than bullate in texture, its inflorescence-bracts larger, and its corolla with a very slender, longer tube and larger lobes. Ixora decora sp. nov. Arbor ad 4 m. alta, ramulis apices versus gracilibus 1.5—3 mm. diametro complanatis parce puberulis mox glabratis, vetustioribus teretibus cinereis, internodiis plerumque 2—5 cm. longis; stipulis glabris vel minute puberulis in tubum brevem subcoriaceum circiter 2 mm. longum biaristatum (aristis circiter 2 mm. longis) caducum connatis, basi intus pilis circiter 0.7 mm. longis ornatis; foliis glabris, petiolis gracilibus canaliculatis 8-16 mm. longis, laminis rigide chartaceis in sicco fusco-olivaceis, oblongo-lanceolatis, (6—) 9-13 cm. longis, (2—) 2.5—4 cm. latis, basi acutis vel attenuatis et in petiolum decurrentibus, apice rotundatis vel obtusis vel obtuse cuspidatis apice ipso obscure calloso-mucronulato, margine integris, costa supra im- pressa vel in sulcula elevata subtus prominente, nervis secundariis utrinse- cus 12—14 patentibus marginem versus anastomosantibus supra subplanis subtus prominulis, rete venularum plerumque immerso; inflorescentiis ter- minalibus trichotome corymbosis multifloris ad 9 cm. longis et 13 cm. diametro, radiorum pedunculis gracilibus glabris ad 5 cm. longis, ramulis infimis subtrichotomis haud articulatis minute et sparse puberulis, brac- teolis paucis inconspicuis subulatis ad 1 mm. longis; pedicellis sub fructu 0.5-3 mm. longis, bracteolis basi ovarii insertis minutis 0.2—0.3 mm. longis; calyce ovario incluso cupuliformi minute puberulo paullo post anthesin 1—-1.5 mm. longo, limbo minuto, lobis 4 rotundatis haud 0.2 mm. longis; corolla in alabastro 4-lobata lobis extus minute puberulis; fructibus glo- bosis in sicco rugulosis glabris ad 7 mm. diametro, calycis limbo minuto coronatis. 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 101 Vanua Levu: Mathuata: Southern slopes of Mt. Numbuiloa, east of Lambasa, alt. 100-350 m., Oct. 27, 1947, Smith 6376 (A TypE, US) (tree 4 m. high, in open forest; fruit red). The new species (§ Pavettopsis) is of the general relationship of /. elegans Gillespie, having a similar type of inflorescence, stipule, etc. How- ever, it differs from J. elegans in its shorter stipules, its oblong-lanceolate leaf-blades that are proportionately narrower and rounded or obtuse at apex (the apices in /. elegans being gradually narrowed and predominantly acute), and the very obscure indument of its inflorescences. Another spe- cies of this relationship, /. twbiflora A. C. Sm., differs from J. decora in its more conspicuously aristate stipules, as well as in foliage characters and in its glabrous calyx with comparatively conspicuous lobes. Ixora myrsinoides sp. nov. Arbor ad 8 m. alta, ramulis teretibus cinereis glabris apicem versus circiter 2 mm. diametro interdum in internodiis distalibus inconspicue complanatis; stipulis subcoriaceis glabris circiter 5 mm. longis e basi late ovata aristatis, basi lateraliter haud connatis, arista dorsali 2-3 mm. longa; foliis glabris, petiolis gracilibus canaliculatis 1.3-2.3 cm. longis, laminis chartaceis in sicco olivaceis oblongo-ellipticis, (10—) 12-16 cm. longis, (5—) 6-8 cm. latis, basi obtusis sed subito angustatis et in petiolum manifeste decurrentibus, apice acutis et calloso-mucronulatis, margine integris et leviter recurvatis, costa supra paullo impressa subtus prominente, nervis secundariis utrinsecus 8-12 patentibus leviter curvatis marginem versus anastomosantibus supra subplanis subtus elevatis, rete venularum utrinque manifeste prominulo; inflorescentiis terminalibus subsessilibus trichotome corymbosis sub anthesi floribus inclusis ad 5.5 cm. longis et 9 cm. diametro, pedunculo 2-3 mm. longo, bracteis primariis oblongo-lanceolatis vel obo- vatis 7-15 mm. longis, radiorum pedunculis ad 1.5 cm. longis et ramulis infimis subtrichotomis glabris, bracteolis ad 2 mm. longis infimis subulatis minoribus; pedicellis sub anthesi 2-4 mm. longis vel subnullis, bracteolis basi ovarii insertis 0.3-1 mm. longis; calyce ovario incluso anguste cupuli- formi sub anthesi 3.5-4 mm. longo obscurissime puberulo, limbo erecto lobis 4 oblongo-deltoideis obtusis 0.7—0.9 mm. longis inclusis 1.5 mm. longo intus minute strigilloso; corolla subcarnosa glabra hypocrateriformi, tubo circiter 15 mm. longo et basi 1.3 mm. diametro superne paullo ampliato, lobis 4 patentibus oblongis obtusis circiter 9 mm. longis et 4 mm. latis; staminibus exsertis glabris, filamentis gracilibus 2-3 mm. longis, antheris anguste oblongis 7-8 mm. longis; stylo gracillimo exserto medium versus obscure puberulo superne incrassato, stigmatibus circiter 2 mm. longis; fructibus subglobosis in sicco rugulosis glabrescentibus ad 9 mm. diametro calycis limbo coronatis. Vitt LEvu: Mba: Northern slopes of Mt. Namendre, east of Mt. Koromba [Pickering Peak], alt. 750-900 m., May 27, 1947, Smith 4529 (A type, US) (tree 8 m. high, in dense forest; corolla white, the lobes faintly pink-tinged; fruit at length purple). 102 JOURNAL OF THE ARNOLD ARBORETUM | [voL. xxxiv The closest relative of 7. myrsinoides (§ Pavettopsis) is apparently J. tubiflora A. Sm., from which it differs in the shorter aristae of its stipules, its much more compact inflorescence, its puberulent calyx with slightly shorter lobes, and its corolla with a shorter tube but larger lobes. Other related species, /. elegans Gillespie and J. harveyi (A. Gray) A. C. Sm., are readily distinguished from /. myrsinoides in details of leaf-shape, calyx-lobes, and in their comparatively small corollas. Ixora greenwoodiana sp. nov. Arbor gracilis ad 10 m. alta vel frutex, ramulis gracilibus cinereis sub- teretibus apices versus 1.5-2 mm. diametro inconspicue puberulis mox glabrescentibus, seh plerumque 1—3 cm. longis; stipulis subcoriaceis mox glabris 3.5-5 mm. longis e basi late ovata aristatis, basi lateraliter breviter connatis, a 1.5—3 mm. longa; petiolis gracilibus canaliculatis 7-12 mm. longis minute puberulis vel in sulcula glabris; foliorum laminis chartaceis in sicco olivaceis, elliptico- vel obovato-lanceolatis, 9-13 cm. longis, 3.5—5 cm. latis, basi attenuatis et in petiolum decurrentibus, apice cuspidatis et calloso-mucronulatis, margine integris et anguste recurvatis, supra glabris, subtus pilis pallidis haud 0.05 mm. longis persistentibus minute sed dense et uniformiter puberulis, costa supra leviter impressa subtus prominente, nervis secundariis utrinsecus 8-12 erecto-patentibus leviter curvatis inconspicue anastomosantibus supra prominulis subtus ele- vatis, rete venularum supra prominulo subtus plerumque immerso; inflores- centiis e ramulis brevibus terminalibus trichotome corymbosis sub fructu ad 6 cm. longis et 10 cm. diametro, pedunculo 3—5 mm. longo arcte puberulo, bracteis primariis obovatis foliaceis ad 2 cm. longis, radiorum pedunculis ad 2 cm. longis et ramulis infimis subtrichotomis minute puberulis; pedi- cellis paullo post anthesin et sub fructu ad 2 mm. longis vel subnullis, brac- teolis basi ovarii insertis subulatis 0.3—-1 mm. longis puberulis; calyce ovario incluso oblongo-cupuliformi paullo post anthesin 2.5-3 mm. longo dense sed minute puberulo, limbo suberecto lobis 4 deltoideis obtusis 0.5—0.7 mm. longis inclusis circiter 1 mm. longo intus. obscure strigilloso; corolla non visa; fructibus subglobosis ad 8 mm. diametro in sicco rugulosis ob- scure puberulis calycis limbo coronatis. Vitr Levu: Mba: Eastern slopes of Mt. Koroyanitu, Mt. Evans Range, alt. 950-1050 m., May 2, 1947, Smith 4249 (A type, US) (slender tree 10 m. high, in dense low forest; fruit becoming pink); Mt. Evans Range, alt. about 1150 m., Greenwood 939 (A) (shrub, to 3 m. high; young leaves golden-brown, the flower- buds deep red). The new species (of § Pavettopsis) suggests J. harveyi (A. Gray) A. C. Sm. in the size and shape of its leaves and their narrowed bases, but it differs in the close but uniform indument of the lower surfaces of leaves (the leaves of J. harveyi being strictly glabrous), in its more ample and puberulent inflorescence, and in its pilose calyx with comparatively short lobes. Other Fijian species of § Pavettopsis with leaves pilose beneath are I. maxima Seem. and /. pubifolia A. C. Sm., but both of these have more 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 103 pronounced foliar and inflorescence indument and substantially larger leaves. Jxora greenwoodiana differs from J. myrsinoides, described above, not only in the foliage indument, but also in the short petioles, the con- siderably smaller leaf-blades more gradually narrowed at base, and the smaller calyx-limb. Ixora bullata Turrill in Jour. Linn. Soc. Bot. 43: 25. 1915. Vanua Levu: Mathuata: Summit ridge of Mt. Numbuiloa, east of Lam- basa, alt. 500-590 m., Smith 6414 (A, US) (shrub or slender tree 2-4 m. high, in dense forest; inflorescence compact, deep red). The extraordinary species described by Turrill bears little resemblance to other Fijian /xorae; our specimens have some of the leaves even more extreme in size (up to 40 & 3 cm.) than those originally described. Appar- ently the species is otherwise represented only by the type, collected in southeastern Viti Levu (im Thurn 359, at Kew). PSYCHOTRIA L. Psychotria, the largest genus of phanerogams in Fiji, has had referred to it more than 60 binomials based on material from the archipelago. Some of these have been reduced outright and others are now referred to varietal status, but there remain in Fiji at least 55 described species of Psychotria which seem readily distinguishable. Particular efforts were made during my 1947 trip to obtain material of as many Psychotriae as possible, with a view to clarifying the status of various inadequately known species. Many, if not most, of the older species were re-collected, and it was gratifying to observe that certain species, hitherto considered rare, are in fact locally quite abundant. Of other species a second or third known collection was obtained, but there still remain several Fijian species known only from the type collection. It is hoped that future collectors will pay particular atten- tion to this complex genus, which, in my opinion, can be locally understood by the examination of all types in connection with abundant new material. Most of the species appear to be sharply delimited, and one may suspect that much of the confusion in herbaria is due to misinterpretation of types or attempts to expand species concepts beyond reasonable limits. In the present treatment I describe 15 species as new; thus the genus now in- cludes approximately 70 Fijian species, a number which will almost cer- tainly be substantially increased by future collecting. Careful consideration of all the Fijian species of Psychotria with various types of enlarged calyces inclines me at present to agree with Fosberg’s reduction of my genera Calycodendron and Eumorphanthus (cf. Sargentia 1: 125. 1942). The calyx appears to be a highly variable feature in Psychotria, although it is reasonably constant within species. The Fijian species that have the calyx-limb much exceeding the tube show several trends of development, but these are not necessarily stages along one line of evolution, and consequently the single character of a showy calyx-limb probably does not justify generic segregation. An extreme development 104 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv of the calyx-limb is found in P. eumorphanthus and approached in P. confertiloba, these species also being characterized by very large, 4-angled fruits. To this group two new species are added in the present paper. Species with a large calyx-limb abruptly flattened, or essentially rotate nearly from the base, are very striking on the basis of this character; this group (“Calycodendron’’) includes P. glabra, P. magnifica, P. fragrans, and P. rufocalyx. A somewhat similar calyx-limb, but campanulate or crateriform rather than subrotate, occurs in P. pubiflora, P. milnei, and P. gibbsiae. Species which resemble the latter group in this character but have the calyx smaller are P. neurocalyx, P. calycosa, and P. gillespieana. Psychotria vitiensis must also be associated with this group, although its calyx-limb is erect and fusiform, abruptly flaring only at the apex. The species here mentioned, together with several novelties of this relationship, would have to be associated in Calycodendron if that genus were main- tained, but I think that Fosberg is probably correct in implying that no sharp line can be drawn between this group and Psychotria proper. In the following treatment the species of the “Calycodendron-Eumor phanthus” alliance are discussed toward the end. Psychotria broweri Seem. F]. Vit. 135 (as P. browerit). 1866. Viti Levu: Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Nanga, alt. 725-825 m., Smith 5483 (A, US) tree 4-5 m. high, in dense forest; fruit red). The cited specimen is the only one of my 1947 collection that represents P. broweri, a species too broadly interpreted in herbaria. Seemann originally cited as representing this species his numbers 244 and 254, both from the island of Moturiki and both represented in the Gray Herbarium. The description agrees with no. 254, which Gillespie has indicated as an isotype. Number 244 represents a different species and has been so in- ein by Gillespie; it has a single pedunculate inflorescence and small leav eee broweri is characterized by its glabrous foliage and _ in- fructescence; its leaves are narrowly elliptic, up to 25 & 9 cm., acute to attenuate at both ends. Its cymes are strictly sessile, with several (up to 6) short radiating branches, and its fruits are turbinate, up to 12 mm. long and sharply several-carinate when dried. My no. 172, from Kandavu, is also typical of this species, while my no. 1023 (from Koro) and Gillespie 3476 and Degener 15335 (both from Viti Levu) probably represent it. In herbaria, identifications of this species and of the related P. bracken- ridgei A. Gray, P. turbinata A. Gray, and P. pritchardii Seem. are often questionable. However, study of the types of these and of certain more recently described species will permit adequate circumscriptions. Psychotria brackenridgei A. Gray in Proc. Am. Acad. 4: 44 (as P. brackenridgit). 1860. Vitrt Levu: Mba: Northern slopes of Mt. Namendre, east of Mt. Koromba [Pickering Peak], Smith 4548; western and southern slopes of Mt. Tomanivi, 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 105 Smith 5137, 5266, 5305, 5748; Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Rewasau, Smith 5616. The cited specimens (all A, US) are from shrubs or slender trees 2—5 m. high, occurring in dense forest at 725-1150 m. altitude, and all bear fruits that are at first pale green and eventually orange to red. These specimens may confidently be referred to Gray’s concept, as represented by the type at the Gray Herbarium (no Exploring Expedition material of this species is in the U. S. National Herbarium). In foliage the species is very similar to P. broweri Seem., like which it has strictly sessile cymes. In comparison with Seemann’s species, P. brackenridgei has a more ample inflorescence with persistently pilose branchlets and calyces. This species has been mis- identified (by the writer and others) as P. storckii Seem. and P pritchardii Seem.; older collections from Vanua Levu and Taveuni, as well as Viti Levu, fall into a reasonable concept of P. brackenridget. Further study is needed to analyze the differences between this species and the very closely related P. turbinata A. Gray. Psychotria pritchardii Seem. FI. Vit. 135. 1866. Vanua Levu: Mathuata: Seanggangga Plateau, in drainage of Korovuli River, vicinity of Natua, alt. 100-200 m., Smith 6744 (A, US) (slender tree 4 m. high, in patches of forest in open rolling country; corolla-bud white) The cited collection is the only one of the species obtained by me in 1947. Psychotria pritchardti, as represented by Seemann 259 (isotype, Gray Herb.), from Taveuni, is readily characterized by its essentially glabrous habit, its large, fleshy, auriculate, subpersistent stipules, its lanceolate-elliptic leaves with blades up to 17 6 cm., its elongate in- florescence with a reflexed peduncle up to 18 cm. long, its slender pedicels 6-10 mm. long, and its shallowly undulate calyx-limb. Also si easel the species are Gillespie 4826 and Smith 743, from Taveuni, and Degener & Ordonez, 13820, 14015, and 14097 and Smith 1606 and 1617, all pee Vanua Levu. Certain specimens from Viti Levu so identified are in need of further study before they can be definitely referred here. Psychotria furcans Fosberg in Sargentia 1: 133. 1942. Vitt LEvu: Mba: Mt. Evans Range, Smith 4091, 4277, 4354, 4372; upper slopes of Mt. Koromba [Pickering Peak], Smith 4660, 4661; northern slopes of Mt. Namendre, east of Mt. Koromba, Smith 4550; slopes of the escarpment north of Nandarivatu, Smith 6038, 6270; Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, Smith 5477, 5608. The cited specimens (all A, US), collected in dense forest or hillside thickets at altitudes of 550-1075 m., were from shrubs or slender trees 2-10 m. high; the corolla is white or distally pale pink, and the fruit is red to bright orange. The species is typified by Degener 14450, from the Nandarivatu region, and Fosberg has questionably referred to it another specimen from lowland Vanua Levu. It now seems that P. furcans is one of the more abundant Psychotriae in upland Viti Levu, and I would also 106 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv refer here Degener 14369, 14625, and 14637, which were otherwise iden- tified by Fosberg. The type is among the smaller-leaved specimens, but no differences of consequence are noted, except that the indument of the inflorescence is persistent to a varying degree. The petioles vary in length up to 7 cm., the leaf-blades may rarely be as large as 23 « and the peduncle, usually sharply reflexed, is up to 6 cm. in length. I doubt if this species is as closely related to my P. edentata and P. pachyantha as implied by Fosberg, and I should rather relate it to P. pritchardii Seem., from which stipular characters and the shorter inflores- cence distinguish it. The forked stipules indicate a relationship of P. furcans to P. broweri, P. brackenridgei, and their more immediate relatives, from which the pedunculate inflorescence readily separates it. Psychotria chrysophylla Fosberg in Bull. Torr. Bot. Cl. 67: 423. 1940. Vitt Levu: Mba: Slopes and summit of Mt. Ndelaiyod, on the ae west of Nandarivatu, Smith 5078; southern slopes of Mt. Ndelainathovu, on sam escarpment, Smith 4931; western and southern slopes of Mt. Tomanivi, Smith 5296; Ra: Ridge from Mt. Namama (east of Nandarivatu) toward Mt. Toma- nivi, Smith 5702; Naitasiri: Northern portion of Rairaimatuku Plateau, mes tween Mt. Tomanivi and Nasonggo, Smith 6130; Nandronga & Navosa: Pegs ‘lets of Rairaimatuku Plateau, between Nandrau and Nanga, Smith 5545, 5 The cited specimens (all A, US) agree very closely with the original description, based only on the type, St. John 18344, collected in the upper Wainimala drainage near the center of Viti Levu. The type locality falls into the present Province of Naitasiri and lies about eight miles south of the last locality listed above. My specimens were from trees or shrubs 4—5 m. high, occurring in dense forest at elevations of 725-1150 m.; the corolla is pure white and in one case a fragrance of the flowers was noted. The only recorded local name (no. 5545) was kau simbala. Slightly to amplify the original description, it may be noted that the petioles vary in length from 0.5 to 3 cm., the leaf-blades are sometimes as much as 15 cm. long and 5.5 cm. broad, and the peduncle may be as long as 6 cm. The variation is no more than normal for a species of Psychotria, and in fact this is a very well-marked species, distinguished as originally noted by the uniform pubescence of the inflorescence and the rather regularly oblong leaf-blades which dry a characteristic yellow-green, at least on the upper surface. Psychotria st.-johnii Fosberg in Bull. Torr. Bot. Cl. 67: 423. 1940. itl Levu: Mba: Slopes of Mt. Nairosa, eastern flank of Mt. Evans Range, Smith 4025, 4049; hills east of Nandala Creek, about 3 miles south of Nandari- vatu, Smith 5927; hills between Nggaliwana and Nandala Creeks, Smith 5839; hills between Nggaliwana and Tumbeindreketi Creeks, Smith 5890, 5991; Nan- dronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Rewasau, Smith 5648. 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 107 The cited specimens (all A, US) are from shrubs or slender trees 1-8 m. high, occurring in dense forest at elevations of 725-1050 m.; the calyx and corolla at anthesis are pure white and the fruit is red to bright orange. This material agrees very closely with the two specimens previously known of the species, both collected by St. John near the center of Viti Levu, about 10 miles south of the last locality listed above and in the present Province of Naitasiri. The species is very sharply marked by the short reddish indument of its long, calyptrate stipules and lower leai- surfaces, by its acuminate, pale-nerved leaves, and by its white flowers with the calyx very inconspicuously lobed. Although the flowers are usually glabrous without, the reddish indument of the inflorescence- branches sometimes (as in my no. 5890) extends to the calyx-tube. Psychotria degeneri Fosberg in Sargentia 1: 134. 1942. Vitr Levu: Mba: Western slopes of Mt. Nanggaranambuluta [Lomalangi ], east of Nandarivatu, Smith 4771, 4773; western and southern slopes of Mt. Tomanivi [Mt. Victoria], Smith 5246, 5301; Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Nanga, Smith 5540. The cited specimens (all A, US) agree very well with the type and the three other collections originally cited; they were taken from shrubs or low trees 2—5 m. high, growing in dense forest at 725-1150 m. elevation. The inflorescences (branches, pedicels, calyx, and corolla) are white, or the pedicel and calyx may be faintly pink-tinged, and the fruit is red to orange. This recently described species, like P. st.-johnii (discussed above), proves to be fairly common in the undergrowth of the rain-forest of upland Viti Levu. It resembles P. st.-johnii in its strongly veined leaves and its delicate, sessile inflorescences, but it is most readily distinguished by its longer indument, which extends to the calyx and corolla, the individual hairs being reddish and several-celled. Psychotria diffusiflora sp. nov. Arbor vel frutex ad 4 m. altus, praeter corollam et interdum costam foliorum ubique glaber, ramulis gracilibus teretibus apices versus ut petiolis luteo-glanduloso-lineolatis, internodiis distalibus 5-15 mm. longis; stipulis papyraceis maturitate 17-25 mm. longis in vaginam connatis superne in tubum brevem haud 1 mm. longum contractis, apice minute liberis recurvatis inconspicue bifidis, caducis, vagina ellipsoidea inflores- centiam juvenilem obtengente ad 12 mm. lata uno latere erumpente; petiolis leviter canaliculatis vel semiteretibus in sicco rugulosis 1-2 (—2.5) cm. longis, laminis foliorum papyraceis vel subcoriaceis copiose immerso- glanduloso-punctatis vel -lineolatis siccitate fusco-olivaceis, ellipticis vel elliptico-oblanceolatis, (6—-) 7-12.5 cm. longis, (2.5—) 3-5 cm. latis, basi acutis et in petiolum manifeste decurrentibus, apice cuspidatis vel breviter acuminatis (apice ipso gracili 5-15 mm. longo calloso obtuso), margine integris vel undulatis leviter recurvatis, costa supra paullo elevata subtus prominente et interdum pilis rubellis multiseptatis circiter 1 mm. longis 108 JOURNAL OF THE ARNOLD ARBORETUM | [vot. xxxiv patenti-pilosa, nervis secundariis utrinsecus 10-13 patentibus curvatis pallidis marginem versus inconspicue anastomosantibus supra prominulis vel subplanis subtus valde elevatis, rete venularum intricato utrinque subimmerso vel plano; inflorescentia terminali cymosa copiose ramulosa multiflora ad 10 cm. longa et lata epedunculata vel pedunculo ad 8 mm. longo, ramulis plerumque 5 divaricatis, pedunculis secundariis gracilibus ad 7 cm. longis, bracteis bracteolisque lanceolatis 0.5—-2 mm. longis caducis, pedicellis gracilibus sub anthesi 7-9 mm. longis; calyce obconico-cupuli- formi sub anthesi 1.5—-2 mm. longo et apice diametro, tubo brevi, limbo suberecto membranaceo 0.5—0.8 mm. longo inconspicue nervato truncato vel obscure 5-denticulato; corolla membranacea infundibulari sub anthesi lobis inclusis 6-7 mm. longa faucibus pilis circiter 0.5 mm. longis copiose barbellatis exceptis glabra, tubo brevi haud 2 mm. longo, lobis 5 patentibus vel reflexis oblongis obtusis circiter 4 mm. longis et 1.5 mm. latis; stamini- bus 5 faucibus affixis exsertis, filamentis gracilibus circiter 3 mm. longis, antheris oblongis circiter 1.5 mm. longis; disco conspicuo annulari-pulvinato circiter 0.7 mm. alto; stylo gracili quam corolla breviore apice bifido; fructibus ovoideis circiter 7 mm. longis et 5 mm. latis, in sicco 4-angulatis et leviter complanatis, calycis limbo persistente coronatis, pyrenis hastatis circiter 6.5 & 4.5 & 2.5 mm., basi obtusis, medium versus abrupte con- tractis, superne angustatis, ventre levibus, dorso basi levi excepto unicarinatis. Viti Levu: Mba: Southern slopes of Mt. Tomanivi [ Mt. Victoria], alt. 850- 1150 m., Smith 5247 (A, US) (tree 4 m. high, in dense forest; pedicel, calyx, and corolla pure white), 6087 (A type, US) (Sept. 18, 1947; shrub 3 m. high, in dense forest; stipules greenish white; calyx and corolla white). A representative of Gray’s subgenus Piptilema, the new species is characterized by its glabrous habit, its large calyptrate stipules with very minute free portions, and its ample and many-flowered, spreading in- florescence. Its only close ally seems to be P. filipes A. Gray, from which it differs in its shorter petioles, in having its leaf-blades acute and decurrent at base (rather than narrowly but obviously cordate) and with more numer- ous secondary nerves, and in its long-pedicellate flowers, which are con- siderably larger, especially as to the corolla-lobes. My understanding of P. filipes is based entirely upon the type (US 62336), which was redescribed and discussed by Fosberg in Sargentia 1: 129. 1942. Psychotria tetragona Seem. FI. Vit. 137. 1866. Vitt Levu: Mba: Slopes and summit of Mt. Ndelaiyoéd, on the escarpment west of Nandarivatu, alt. 900-1053 m., Smith 5057, 5058 (A, US). The cited specimens agree very well with an isotype in the Gray Herbarium, Seemann 252, from Ovalau. The species is characterized by its glabrous habit, lanceolate leaves, forking stipuies which are basally connate around the inflorescence-bud, very compact inflorescence, small calyx, conspicuous and pulvinate disk, and clavate corolla about 15 mm. 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 109 long. Also to be placed here are Degener 14611, 15057, and 15167 (all A, US), from Viti Levu. Similar to P. tetragona, but differing in its broader leaf-blades (3-6 cm. broad) is Smith 4352 (A, US) (Viti Levu: Mba: Northern portion of Mt. Evans Range, between Mt. Vatuyanitu and Mt. Natondra, alt. 700-900 m.). My no. 4551 (A, US) (Viti Levu: Mba: Northern slopes of Mt. Namendre, east of Mt. Koromba, alt. 750-900 m.) is similar to no. 4352 except that the costa of the lower leaf-surface is copiously hispidulous with hairs 0.5—0.8 mm. long. In inflorescence characters these two speci- mens are hardly distinguishable from typical P. tetragona, and at present I am inclined to consider them extreme representatives of this species; it may be noted that both occur on isolated forested hills rising from the grassland of western Viti Levu. Psychotria tetragonoides Fosberg, a species of this alliance somewhat suggested by the last two numbers discussed above, is clearly separable from P. tetragona by the indument of its foliage and flowers; it is not represented in my collection. Psychotria ampullacea sp. nov. Arbor gracilis ad 5 m. alta, ramulis subteretibus gracilibus praeter cicatrices stipularum hispidulas glabris, internodiis distalibus 6-10 mm longis; stipulis glabris papyraceis 2—2.5 cm. longis in vaginam ampullaceam connatis, vagina basi ventricosa inflorescentiam juvenilem obtegente superne in tubum circiter 1 cm. longum valde contracta, stipularum partibus liberis circiter 3 mm. longis bifidis; petiolis gracilibus glabris 1-3 cm. longis subteretibus rugulosis; foliorum laminis chartaceis in sicco fusco- olivaceis, lanceolatis vel oblanceolatis, (9Q—) 12-17 cm. longis, (2—) 3.5-5 cm. latis, basi longe attenuatis et in petiolum decurrentibus, apice gradatim acuminatis, margine anguste recurvatis, supra glabris, subtus pilis 0.7—1 mm. longis rubellis multiseptatis copiose (demum sparsiore) patenti-pilosis, costa supra elevata subtus prominente, nervis secundariis utrinsecus 11—13 curvatis patentibus inconspicue anastomosantibus supra leviter subtus valde elevatis, rete venularum subimmerso; inflorescentia terminali sub- capitata 9—12-flora, floribus pedicellatis, pedicellis teretibus 2—3 mm. longis copiose pilosis (pilis pallidis patentibus 0.5-0.7 mm. longis multiseptatis) ; calyce campanulato 8—9 mm. longo et apice circiter 4 mm. diametro, tubo brevi haud 1.5 mm. longo ut pedicello piloso, limbo erecto submembranaceo glabro conspicue nervato 5-lobato, lobis deltcideo-lanceolatis suberectis acutis 1-2 mm. longis; corolla submembranacea infundibulari sub anthesi circiter 20 mm. longa, extus lobis inclusis pilis ad 1 mm. longis ut pedicello copiose patenti-pilosa, intus glabra, lobis 5 erectis oblongis 4-5 mm. longis obtusis; staminibus glabris medium corollae versus insertis, filamentis gracilibus 4-5 mm. longis, antheris oblongis 2.5-2.8 mm. longis haud exsertis; disco conspicue pulvinato glabro circiter 1 mm. alto; stylo gracili glabro quam corolla breviore apice bifido. 110 JOURNAL OF THE ARNOLD ARBORETUM | [vot. xxxiv Viti Levu: Nandronga & Navosa: Northern portion of Rairaimatuku Plateau, between Nandrau and Rewasau, alt. 725-825 m., Aug. 11, 1947, Smith 5659 (A type, US) (slender tree 5 m. high, in dense forest; calyx pale green; corolla white). The species here described, together with the following, seems super- ficially to be related to P. tetragona Seem. and P. tetragonoides Fosberg, because of its congested, subcapitate inflorescence. However, the calyces of the two older species are very small, scarcely 1.5 mm. in length, and in- conspicuously dentate, whereas the calyces of the two species here de- scribed as new are comparatively elongate and obviously lobed. In texture, the calyx-limb of P. ampullacea suggests that of some of the species that have been referred to Calycodendron, but in those species the inflorescence is ample and freely branched. Probably these two new species belong in the general alliance of P. tetragona, but they seem to have no immediate relatives. Further distinguishing characteristics of P. ampullacea are the indument of the leaves, pedicels, calyx-tube, and corolla (the hairs being divided into many short cells by transverse septae), and the form of the stipular sheath, which is flask-shaped, swollen at base to enclose the inflorescence- bud and narrowed to a very slender elongate throat, the stipules being free and bifid only at the extreme apices. Psychotria tomaniviensis sp. nov. Frutex ad 4 m. altus, ramulis gracilibus glabris subteretibus vel in sicco angulatis, internodiis distalibus 5-10 mm. longis; stipulis papyraceis 5-8 mm. longis basim versus connatis, superne 3—5 mm. liberis et profunde bifidis, glabris vel apicem versus puberulis; foliis glabris, petiolis canalicu- latis 5-10 mm. longis vel subnullis, laminis chartaceis in sicco fusco-viridi- bus oblanceolatis, (7—) 10-16 cm. longis, (2.5—) 3.5-4.5 cm. latis, basi attenuatis et in petiolum longe decurrentibus, apice obtuse cuspidatis, margine integris et anguste recurvatis, costa supra subplana vel leviter elevata subtus prominente, nervis secundariis utrinsecus 12-15 curvatis patentibus inconspicue anastomosantibus supra subplanis subtus elevatis, rete venularum subimmerso; inflorescentia terminali subcapitata multiflora, bracteolis sub floribus lanceolatis circiter 2 mm. longis ciliolatis caducis, pedicellis teretibus circiter 1 mm. longis glabris; calyce campanulato sub anthesi circiter 10 mm. longo et apice 4 mm. diametro lobis interdum obscure puberulis exceptis glabro, tubo brevi haud 2 mm. longo, limbo erecto submembranaceo manifeste nervato 5-lobato, lobis oblongis 2—3 mm. longis subacutis; corolla submembranacea infundibulari sub anthesi circiter 22 mm. longa, extus glabra vel juventute pilis conicis haud 0.1 mm. longis minute puberula, intus medium versus pilis albidis mollibus copiose tomen- tella, lobis 5 erectis oblongis circiter 5 mm. longis obtusis; staminibus glabris, antheris subsessilibus apicem tubi corollae versus oblongis circiter mm. longis; disco conspicue pulvinato glabro circiter 0.7 mm. alto; stylo gracili glabro leviter exserto apice bifido; fructibus subglobosis circiter 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 111 15 mm. diametro, pericarpio crasse carnoso, pyrenis obovoideo-oblongis circiter 8 & 6 X : mm., ventre levibus, dorso superne conspicue 3-carinatis, apice profunde cavatis, basi rotundatis. Vit1 Levu: Mba: Western slopes of Mt. Tomanivi [Mt. Victoria], alt. 850- 1150 m., July 17, 1947, Smith 5294 (A type, US) (shrub 3-4 m. high, in dense forest; calyx green; corolla pure white; fruit red). The new species differs from the above described P. ampullacea, which seems to be its closest ally, in its much smaller stipules that do not form the flask-shaped sheath so noticeable in P. ampullacea, in its glabrous leaves and much shorter petioles, in its glabrous pedicels and calyx-tube, in having the external indument of the corolla composed of very minute conical hairs and evanescent, while within the corolla is tomentellous rather than glabrous, and in its subsessile anthers and elongate style. Such characters as length of filament and style, and perhaps internal pubescence of the corolla, may sometimes be correlated with actual or incipient dioecism in Psvchotria. Psychotria valleculata sp. nov. Arbor ad 4 m. alta sub fructu cicatricibus-stipularum obscure ferrugineo- setulosis exceptis ubique glabra, ramulis subteretibus crassis, internodiis longitudine diversis; stipulis subcoriaceis deltoideo-lanceolatis 6-7 mm. longis, basi connatis, superne liberis, apice integro minute mucronulatis; petiolis crassis rugulosis semiteretibus 8-18 mm. longis, laminis subcoriaceis in sicco pallide olivaceis, oblongo-obovatis, (6—) 8-10.5 cm. longis, (2.5) 3.5—5 cm. latis, basi attenuatis et in petiolum longe decurrentibus, apice rotundatis, margine integris et anguste recurvatis, costa utrinque valde elevata, nervis secundariis utrinsecus 8-10 patentibus anastomosantibus utrinque elevatis, rete venularum subimmerso; inflorescentia non visa: infructescentia terminali capitata, receptaculo coriaceo rotundato vel pedicellis haud 1 mm. longis vel subnullis; fructibus obovoideis in vivo ad 11 mm. longis et 8 mm. latis, in sicco indistincte 4-angulatis, calycis limbo mox caduco non viso, disco pulvinato interdum conspicuo circiter 1 mm. alto, pyrenis subhastato-ovoideis circiter 7 < 4 < 2 mm., ventre complanatis vel basim versus inconspicue costatis, dorso conspicue uni- carinatis, basi acutis, paullo supra medium valde contractis, apice rotun- datis, parte basali obcordiformi carina excepta complanata, parte apicali conspicue angustiore. Vitr Levu: Nandronga & Navosa: Southern slopes of Nausori High- lands, in drainage of Namosi Creek above Tumbenasolo, alt. 300-450 m., June 4, 1947, Smith 4721 (A type, US) (tree 4 m. high, in dense forest of valley; fruit red). In spite of the lack of flowers, this collection seems patently to represent a new species, characterized by its oblong-obovate, rounded leaves and its strictly capitate infructescences. In the latter character and in its general 112 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxIv aspect the new species suggests P. tetragona Seem., but its stipules are of another type and its leaves are quite different; it is further characterized by its obovoid fruits, of which the pyrenes are unicarinate, sharply con- tracted near the middle, and broadly obcordiform in the basal part. It is perhaps referable to Gray’s subgenus Piptilema, but it is only remotely related to the species of this group discussed by Fosberg (in Sargentia 1: 126. 1942). Psychotria evansensis sp. nov. Arbor ad 5 m. alta vel frutex, ramulis crassis subteretibus novellis minute puberulis exceptis glabris, internodiis distalibus 1-3 cm. longis; stipulis papyraceis 3-5 mm. longis deltoideis obtusis, basi in vaginam brevem connatis, superne liberis, mox caducis; foliis magnis, petiolis semiteretibus 2-5.5 cm. longis primo minute puberulis demum_ glabratis, laminis papyraceis in sicco fusco-viridibus anguste oblongis, 16-26 cm. longis, 5-8.5 cm. latis, basi attenuatis et in petiolum longe decurrentibus, in acuminem 1-2 cm. longum gradatim angustatis, margine integris vel undula- tis anguste recurvatis, immerso-luteo-glandulosis, supra glabris, subtus minute puberulis glabratisque, costa supra subplana vel leviter elevata subtus prominente, nervis secundariis utrinsecus 12—18 curvatis patentibus anastomosantibus supra subplanis subtus elevatis, rete venularum intricato supra immerso subtus prominulo; inflorescentia non visa; infructescentia terminali cymosa fructibus inclusis 5-9 cm. diametro pedunculata, pedunculo gracili 2—4 cm. longo et ramulis primariis 4 vel 5 divaricatis 1— 2.5 cm. longis pedicellisque minute puberulis, fructibus apices ramulorum versus pluribus, pedicellis gracilibus 1-3 mm. longis; fructibus in vivo obovoideis 11-12 mm. longis 9-10 mm. latis, in sicco valde 6—10-costatis, basi obtusis, apice rotundatis vel paullo depressis, calycis limbo persistente inconspicuo circiter 1 mm. longo submembranaceo fere ad basim 5-lobato lobis subacutis, pericarpio carnoso copiose immerso-luteo-glanduloso, pyrenis semi-obovoideis circiter 10 & 7.5 & 5 mm., basi rotundatis, apice truncatis, ventre complanatis rugulosis basim versus costatis, dorso con- spicue 3-carinatis, carinis crista rotundatis rugulosis, sulcis profundis, marginibus incrassatis interdum carinis similibus. Vitt Levu: Mba: Slopes of Mt. Nairosa, eastern flank of Mt. Evans Range, alt. 700-1050 m., in dense forest, Smith 4024 (A, US) (shrub 3 m. high), 4072 (A type, US) (Apr. 28, 1947; slender tree 5 m. high; fruit red). Although it lacks flowers, the species here described is evidently a relative of P. forsteriana A. Gray, which it resembles in general appearance, leaf-shape and -texture, and the obovoid fruits with a very short calyx- limb, immersed-glandular pericarp, and 3-carinate pyrenes. The new species differs from P. forsteriana in its larger leaves, obviously pedunculate infructescences, and much larger fruits. The mature fruits of P. forsteriana are 6-7 mm. long and 5—6 mm. broad, while the pyrenes do not exceed 6 < 5 < 2 mm. apice rotundatis demum reflexis; staminibus 5 in faucibus affixis, filamentis glabris brevibus, antheris oblongis circiter 2.5 mm. longis apice obtusis, loculis basi saepe divergentibus; disco conspicuo hemisphaerico-pulvinato circiter 1.5 mm. alto luteo-glanduloso- lineolato; stylo filiformi, stigmatis lobis complanatis. Vitt Levu: Mba: Immediate vicinity of Nandarivatu, alt. 800-900 m., Smith 5031 (A type, US) (July 2, 1947; tree 4 m. high, in dense forest along stream; corolla white), Degener & Ordonez 13598 (A, US, etc.) (tree, in rain- forest); western slopes of Mt. Nanggaranambuluta [Lomalangi], east of Nan- darivatu, alt. 850-1000 m., Smith 6303 (A, US) (compact tree 4 m. high in dense forest). The new species differs from its closest ally, P. gibbsiae S. Moore (dis- cussed above), in having its stipules connate into a tubular or flask-shaped sheath, with the free tips comparatively minute, in having its inflorescence- branches, pedicels, and calyx spreading-pilose with comparatively long, reddish, multiseptate hairs (or glabrate, but not closely puberulent), in having its bracteoles glabrous except for the ciliolate margins, and in having its calyx campanulate from the base, whereas in P. gibbsiae it is slightly contracted at the apex of the tube and then flaring slightly into the campanulate limb. It should be noticed that no. 6303, identical with no. 5031 in every other respect, has its calyx glabrous, indicating that the degree of indument is not as dependable as its type. Psychotria vaughanii sp. nov. Frutex, ramulis crassis teretibus apicem versus pilis haud 0.1 mm. longis copiose sed minute cinereo-puberulis, nodis distalibus 1-3 cm. longis; stipulis papyraceis 3-4 mm. longis (immaturis?) dorso parce puberulis 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 119 inferne in vaginam brevem connatis, partibus liberis circiter 2 mm. longis bifidis; petiolis crassis semiteretibus 0.7—1.4 cm. longis ut ramulis juvenili- bus dense puberulis, laminis coriaceis in sicco luteo-viridibus, oblanceolatis vel anguste ellipticis, 12-19 cm. longis, 4-7 cm. latis, basi acutis et in petiolum decurrentibus, apice obtuse cuspidatis, margine undulatis et anguste recurvatis, supra glabris, subtus minute sed copiose puberulis, costa valida supra leviter elevata et obscure sulcata subtus prominente, nervis secundariis utrinsecus 12-14 patentibus curvatis marginem versus obscure anastomosantibus utrinque inconspicue elevatis, rete venularum immerso; inflorescentia terminali epedunculata e basi 3-ramulosa multiflora plerumque 4-plo divisa, ramulis pedicellisque copiose puberulis (pilis fulvis circiter 0.1 mm. longis), bracteolis papyraceis lanceolatis ad 1 mm. longis margine ciliolato et irregulariter fisso excepto glabris caducis; floribus e nodis ultimis plerumque 3, pedicellis gracilibus sub anthesi 3-5 mm. longis; calyce sub anthesi circiter 5 mm. longo et apice diametro extus copiose puberulo vel superne demum glabrato, tubo minuto cupuliformi, limbo campanulato inferne erecto breviter tubuloso circiter 3 mm. diametro apice subito patente, lobis 5 deltoideis rotundatis circiter 1.5 & 2 mm.; corolla submembranacea infundibulari sub anthesi 11-12 mm. longa et basim versus circiter 1.5 mm. diametro superne ampliata, extus inconspicue sed copiose puberula, intus inferne glabra medium versus pilis pallidis circiter 0.5 mm. longis copiose tomentella superne puberula, lobis 5 erectis oblongis circiter 1.5 mm. longis apice rotundatis et cucullatis; staminibus 5 apicem tubi corollae versus insertis glabris, filamentis gracili- bus circiter 1 mm. longis, antheris oblongis circiter 2 mm. longis; disco pulvinato circiter 0.6 mm. alto; stylo gracili quam corolla breviore, stigmatis lobis complanatis. Vitr LEvu: Naitasiri: Prince’s Road, mile 7 [approximate bay aes of Tholo-i-suva], alt. about 150 m., Nov. 10, 1946, J. H. Vaughan 3323 (TYPE in herbarium of British Museum) (common shrub, with terminal clusters of tubular white flowers). The species here described suggests P. fragrans (Gillespie) Fosberg in the size and spreading apex of its calyx-limb. However, the calyx-limb of P. fragrans is subrotate from its base, rather than campanulate, and is conspicuously glandular-lineate, while the plant is essentially glabrous in habit, has comparatively small leaves, a pedunculate inflorescence, and a longer corolla. Actually P. vaughanii in calycine characters more nearly approximates P. gibbsiae S. Moore, a species with much smaller leaves and a more compact and fewer-flowered inflorescence. The new species is fur- ther distinguished by the uniform puberulence of its vegetative parts and inflorescence and by having its narrowly campanulate calyx-limb flaring only at the apex. Psychotria mundula sp. nov. Frutex ad 2 m. altus ubique praeter corollam glaber vel partibus novellis minute puberulis, ramulis gracilibus teretibus in sicco rugulosis, internodiis 120 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv distalibus 2—6 mm. longis; stipulis papyraceis ovato-oblongis 3.5—6 mm. longis in vaginam inferne connatis, superne liberis saepe scarioso-margi- natis, dorso brevi-calcaratis vel in apicem subacutum vel acuminatum interdum puberulum productis; foliis apices ramulorum versus confertis parvis, petiolis gracilibus semiteretibus 2—7 mm. longis, laminis subcoriaceis obscure immerso-glandulosis in sicco viridi-olivaceis, lanceolatis vel ellip- ticis, (2—) 3—5.5 cm. longis, (0.5—) 1-2 cm. latis, basi attenuatis et in petiolum decurrentibus, apice acutis vel breviter acuminatis et calloso- apiculatis, margine integris anguste recurvatis, costa supra subplana sub- tus elevata, nervis secundariis utrinsecus 7-10 inconspicuis patentibus vel subadscendentibus curvatis anastomosantibus supra planis subtus promi- nulis, rete venularum immerso; inflorescentia terminali cymosa compacta pauciflora epedunculata vel pedunculo ad 2 mm. longo, sub anthesi floribus inclusis 2—3 mm. longa, ramulis plerumque 3 patentibus ad 1.5 cm. longis saepe 3-floris, floribus sessilibus, lateralibus apice ramulorum secundariorum ortis, bracteolis sub floribus 1-2 mm. longis deltoideis acutis vel interdum apice trifidis; calyce infundibulari sub anthesi limbo incluso 5—7 mm. longo et apice 3-4 mm. diametro, tubo brevi, limbo papyraceo erecto lobis in- clusis 3-4 mm. longo, lobis 5 vel 6 deltoideo-lanceolatis 1-1.5 mm. longis incrassatis acutis interdum apice patentibus; corolla tenuiter carnosa gracili infundibulari sub anthesi lobis inclusis 13-15 mm. longa, extus glabra, intus faucibus et lobis pallide puberula, lobis 5 vel 6 oblongis obtusis 3-4 mm. longis sub anthesi patentibus; staminibus 5 vel 6, filamentis gracilibus ad 1 mm. longis vel subnullis, antheris oblongis 2—2.3. mm. longis utroque obtusis; disco annulari-pulvinato 0.6—0.8 mm. alto; stylo filiformi interdum corollam subaequante, apice bifido. Vanua Levu: Mathuata: Northwestern slopes of Mt. Numbuiloa, east of Lambasa, alt. 500-590 m. [also noted on summit], Nov. 6, 1947, Smith 6533 (A TYPE, (shrub 2 m. high, in wind-swept thickets; corolla white). VuitT1 : Mountains near Lautoka, alt. 480-550 m., Greenwood 1226, 1227, 1270, 1271 (all US) (shrubs about 50 cm. high, on dry open hillsides; flowers white). The very distinct species described above is characterized by its com- pact habit, the crowded leaves nearly concealing the small, few-flowered inflorescences. Each branch of the 3-parted inflorescence usually bears three sessile flowers, but the lateral flowers, borne on short secondary branchlets, are sometimes flanked by additional and mostly undeveloped flowers. The erect calyx-limb, with longitudinal thickenings terminating in the lobes, and the slender corolla with the throat puberulent also charac- terize the species. In its basic floral characters, the new species seems most closely related to P. gillespieana A. C. Sm. [Calycosia laxiflora Gillespie | (represented in my present collection by nos. 5780 and 5184, from Mt. Tomanivi, the type-locality), a species with much larger stipules which have laminar forked apices, larger and longer-petiolate leaves, a much more ample and freely divided inflorescence, and larger flowers. Green- wood’s specimens were taken from two plants, his nos. 1227 and 1270 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 121 representing one and nos. 1226 and 1271 the other, the second of these having slightly the narrower leaves. Psychotria leptocalyx sp. nov. Arbor ad 5 m. alta, ramulis gracilibus subteretibus et petiolis minute sed dense puberulis (pilis griseis ad 0.05 mm. longis), internodiis distalibus 8-15 mm. longis; stipulis papyraceis 15-17 mm. longis extus copiose puberulis, basim versus in vaginam connatis, partibus liberis laminaribus 10-12 mm. longis conspicue bifidis; petiolis gracilibus semiteretibus 1-2 cm. longis, laminis subcoriaceis in sicco fuscis, oblanceolatis vel elliptico- obovatis, 9-16 cm. longis, 3—6.5 cm. latis, basi attenuatis et in petiolum longe decurrentibus, apice obtusis vel obtuse cuspidatis, margine integris et anguste recurvatis, supra glabris, subtus minute sed uniformiter griseo- puberulis, costa supra leviter elevata et sulcata subtus prominente, nervis secundariis utrinsecus 11—13 arcuato-adscendentibus marginem versus anas- tomosantibus supra paullo subtus valde elevatis, rete venularum immerso; inflorescentia terminali sub anthesi ad 12 cm. longa multiflora plerumque 4-plo divisa manifeste pedunculata, pedunculo gracili ad 4 cm. longo et ramulis inflorescentiae pedicellisque copiose puberulis, pedunculis secun- dariis 3 radiatis 2—3.5 cm. longis; bracteolis papyraceis oblongis 1—-1.5 mm. longis in lobos 3 plerumque fissis, dorso puberulis, margine ciliolatis, brac- teis similibus 3—4 mm. longis mox glabratis et caducis; pedicellis sub anthesi gracilibus ad 6 mm. longis; calyce sub anthesi 6-7 mm. longo et apice 4—5 mm. diametro extus copiose puberulo, tubo pyriformi circiter 1.5 mm. longo et 2.5 mm. diametro, apice constricto, limbo papyraceo obscure nerv- ato, parte basali fusiformi erecta 3-4 mm. longa et circiter 2.5 mm. diametro superne leviter contracta deinde subito patente, lobis 5 oblongis rotundatis circiter 1.5 2 mm.; corolla submembranacea infundibulari circiter 20 mm. longa (vel ultra?) et basim versus 2 mm. diametro superne ampliata, extus basi excepta cinereo-puberula, intus medium versus pilis ad 0.6 mm. longis dense tomentella, lobis 5 erectis oblongis 3—4 mm. longis apice obtusis cucullatis; staminibus 5 faucibus affixis glabris, filamentis gracilibus 1-2 mm. longis, antheris oblongis circiter 3 mm. longis; disco pulvinato cir- citer 0.5 mm. alto; stylo gracili quam corolla breviore, stigmatis lobis complanatis. Vitrt Levu: Mba: Hills east of Nandala Creek, about 3 miles south of Nandarivatu. alt. 850-970 m., Sept. 25, 1947, Smith 6215 (A typr, US) (tree 5 m. high, in dense forest; pedicel and calyx-tube greenish white, the lobes white; corolla white). In its comparatively narrow, fusiform calyx-limb, P. leptocalyx suggests P. vitiensis Fosberg [Calycosia monticola Gillespie], which, as Fosberg has pointed out (in Sargentia 1: 127. 1942), is distinct from P. calycosa A. Gray. From P. vitiensis, however, the new species is readily distinguished by the dense, although minute, puberulence of its vegetative and inflores- cence parts, and by its larger leaves and more freely branching inflorescence. Also suggestive of the new species and resembling it in indument is P. 122 JOURNAL OF THE ARNOLD ARBORETUM _[VvoL. xxxiIv vaughanii, described above, which has the inflorescence epedunculate, the calyx proportionately broader, with the basal part of the limb campanulate rather than fusiform, and the corolla apparently shorter. Psychotria eumorphanthus Fosberg in Sargentia 1: 127. 1942. Viti Levu: Mba: Western slopes of Mt. Tomanivi [ Mt. Victoria], alt. 850— 1000 m., Smith 5091 (A, US) (tree 6 m. high, in dense forest; calyx-limb green). This species (Eumorphanthus fragrans A. C. Sm. in Bishop Mus. Bull. 141: 158. fig. 81. 1936) has previously been known only from the type collection, from Taveuni. The cited Viti Levu specimen, in fruit, agrees in all respects with the type and similarly has a subpersistent calyx-limb about 2 cm. long; the only apparent difference is that no. 5091 has the secondary nerves of the leaves slightly closer, 12-15 per side rather than 9-13. Unless the discovery of flowers should prove otherwise, I think that the range of the species may safely be extended to include Viti Levu. Psychotria leptantha sp. nov. Arbor ad 12 m. alta, ramulis gracilibus subteretibus glabris (vel novellis interdum parce puberulis) apices versus 1.5-2 mm. diametro, internodiis distalibus 5-10 mm. longis; stipulis lanceolatis 6-8 mm. longis obscure puberulis vel glabris, basim versus lateraliter cohaerentibus, apice sub- acutis, mox caducis; foliis patentibus glabris, petiolis semiteretibus gracili- bus 1-3 cm. longis, laminis papyraceis in sicco viridibus, interdum ut ramulis stipulisque ehiria circiter 0.2 mm. longis ornatis, ellipticis vel oblanceolatis, 7-12 cm. longis, 3-5 cm. latis, basi acutis vel attenuatis et in petiolum eis apice cuspidatis (apice ipso ad 5 mm. longo obtuso vel subacuto), margine integris, costa supra leviter elevata et basinr versus anguste canaliculata subtus prominente, nervis secundariis utrinse- cus 6—9 arcuato-patentibus marginem versus anastomosantibus supra sub- planis subtus elevatis, rete venularum laxo utrinque prominulo vel supra immerso; inflorescentia terminali pauciflora compacte cymosa sub anthesi corolla inclusa ad 9 cm. longa et lata plerumque e basi 3-partita bracteis pluribus deltoideis obtusis 2-3 mm. longis caducis subtenta, ramulis gra- cilibus laxe pilosis (pilis pallidis 0.1-0.3 mm. longis obscure pluriseptatis) , lateralibus 1-1.5 cm. longis apice 2- vel 3-floris, ramulo centrali semel vel bis partito plurifloro; bracteis sub floribus minutis circiter 1 mm. longis caducis, pedicellis teretibus sub anthesi 5-9 mm. longis ut ramulis pilosis in calycem gradatim crassatis; calyce graciliter infundibulari sub anthesi 1.5—2 cm. longo basim versus piloso vel puberulo alioqui glabro, tubo minuto, limbo magno erecto submembranaceo conspicue longitudinaliter nervato apice circiter 5 mm. diametro inconspicue 5-lobato, lobis inaequali- bus deltoideis acutis haud 1 mm. longis; corolla hypocrateriformi sub- membranacea, tubo graciliter cylindrico sub anthesi 3.5—4 cm. longo et cir- citer 1.5 mm. diametro faucibus paullo ampliato superne extus puberulo et intus subtiliter arachnoideo-tomentello cetera glabro, lobis 5 patentibus anguste oblongis vel obovatis 13-15 mm. longis circiter 4 mm. latis mani- 1953] SMITH, STUDIES OF PACIFIC ISLAND PLANTS, XVI 123 feste nervatis apice obtusis extus puberulis intus glabris; staminibus 5 tubi apicem versus insertis et leviter exsertis, filamentis membranaceis ligulatis ad 5 mm. longis vel omnino tubum adnatis, antheris anguste oblongis circiter 3.5 mm. longis obtusis; disco annulari-pulvinato glabro circiter 0.8 mm. alto et 1.5 mm. diametro; stylo gracili glabro circiter 2 cm. longo in stigmata circiter 2 mm. longa fisso; floribus post anthesin saepe caducis, fructibus 1 vel 2 per inflorescentiam, pedicellis sub fructu 1—1.5 cm. longis persistenter patenti-pilosis; fructibus in vivo carnosis et ellipsoideis ad mm. (maturis?) in sicco 4-angularibus, basim versus et basi per- sistente calycis limbi puberulis, parte majore calycis limbi mox caduca, pyrenis (maturis?) triquetro-ovoideis circiter 1 cm. longis latisque, basi rotundatis, apice subacutis, ventre complanatis levibus, dorso unicarinatis. Vitt Levu: Mba: Slopes of Mt. Nairosa, eastern flank of Mt. Evans Range, alt. 700-1050 m., in dense forest, Smith 4048 (A type, US) (April 26, 1947; tree 8 m. high; calyx and corolla pure white), 4423 (A, US) (tree 12 m. high, in fruit). Psychotria leptantha is a sharply marked and very attractive species, characterized by its long, narrow calyx-limb and its very slender, elongate corolla-tube. It is closely allied only to P. eumorphanthus Fosberg, from which it obviously differs in being more delicate and slender throughout, in its much smaller leaves, slightly smaller and more slender calyx-limb, corolla with a slightly shorter tube and much smaller lobes, and shorter style. Fosberg (in Sargentia 1: 125, 127. 1942), in reducing my genus Eumorphanthus to Psychotria, allies its single species to P. confertiloba A.C. Sm. [P. umbraticola Gillespie]. This is perhaps the correct disposi- tion, since the two species concerned are quite similar in basic corolla and fruit characters. However, the calyx-limb of P. confertiloba does not ex- ceed 7 mm. in length, whereas in both P. eumorphanthus and P. leptantha it approaches or exceeds 2 cm. Psychotria confertiloba proves to be a very common species in upland Viti Levu, being represented in my present col- lection by nos. 4542, 4800, 5011, 5129, 5252, 5313, 5560, and 5790, and also by Degener 13567, 14280, 14610, 14754, and 14830. To the Vanua Levu specimens cited by me as P. confertiloba (in Bishop Mus. Bull. 141: 152. 1936) may be added Smith 6470, 6650, and Degener & Ordonez 14010. Psychotria gracilior sp. nov. Arbor ad 10 m. alta sub fructu ubique glabra, partibus novellis, stipulis, petiolis inflorescentiae ramulisque cystolithis 0.1-0.3 mm. longis saepe copiose ornatis; ramulis gracilibus subteretibus apices versus 1.5-2 mm. diametro, internodiis distalibus 5-7 mm. longis; stipulis lanceolatis 1—1.5 cm. longis infra medium lateraliter connatis apice liberis et calloso-acumi- natis mox caducis; foliis patentibus, petiolis gracilibus semiteretibus 1—2 cm. longis, laminis papyraceis in sicco viridibus oblanceolatis, 8-11 cm. longis, 2.5—4 cm. latis, basi attenuatis et in petiolum longe decurrentibus, apice in acuminem gracilem 5-10 mm. longum obtusum angustatis, margine 124 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv integris et leviter recurvatis, supra cystolithis dispersim ornatis, costa supra leviter elevata subtus prominente, nervis secundariis utrinsecus 7—9 mar- ginem versus obscure anastomosantibus supra subplanis subtus elevatis, rete venularum laxo supra plano subtus prominulo; infructescentia terminali compacte cymosa fructibus inclusis ad 4 cm. longa et 6 cm. lata e basi 3-partita bracteis caducis subtenta, ramulis lateralibus (interdum caducis cicatricibus indicatis) 8-12 mm. longis fructus 2 vel 3 gerentibus, ramulo centrali plerumque semel partito; fructibus pluribus per infructescentiam, pedicellis gracilibus 6-12 mm. longis, fructibus in vivo carnosis ellipsoideis maturitate ad 2 & 1.5 cm. in sicco 4-angularibus, utroque obtusis vel subacutis, calycis limbo subpersistente cylindrico papyraceo erecto 4—5 mm. longo longitudinaliter nervato coronatis, limbo apice irregulariter erosulo vel inconspicue 5-lobato, lobis deltoideis obtusis circiter 0.5 mm. longis; pyrenis triquetro-oblongo-ellipsoideis, 11-12 mm. longis, 6.5-8 mm. latis, 3.5-4 mm. crassis, basi rotundatis, apice cuspidatis, ventre complanatis levibus vel obscure sulcatis, dorso unicarinatis. Viti Levu: Mba: Hills between Nandala and Nukunuku Creeks, along trail from Nandarivatu toward Lewa, alt. 750-850 m., Sept. 22, 1947, Smith 6186 (A type, US) (tree 5-10 m. high, in dense forest; fruit at length deep red). Although the specimen cited above lacks flowers, it clearly represents a new species of the general alliance of P. leptantha, above described, from which it differs most obviously in its comparatively short calyx-limb. Psy- chotria gracilior further differs from P. leptantha in its completely glabrous habit, its comparatively narrow leaves, and its apparently smaller fruits, of which the pyrenes are differently proportioned. In spite of the general similarity of these two species in foliage, perhaps a closer alliance of P. gracilior is with P. confertiloba A. C. Sm., as indicated by the similarity in size of the calyx-limb. However, P. confertiloba is comparatively coarse in habit, with broader leaves generally obtuse at apex and with spreading secondaries, and with larger fruits. DEPARTMENT OF BOTANY, U.S. NATIONAL MusEUM, SMITHSONIAN INSTITUTION. 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 125 STUDIES IN THE THEACEAE, XXVII MISCELLANEOUS NEW SPECIES IN THEACEAE CLARENCE E. KosuskI STUDY AND WORK on a Single genus is never truly completed. The pub- lication of such a work often incites an interest for even further study, as the author begins receiving unexpected material for identification — ma- terial which often changes his opinion of his own work supposedly com- pleted. Such is the case in the genus Adinandra. Shortly after the publication of a review of the genus more than seven hundred additional specimens were received, mostly from Indonesia. These were sent primarily for my studies to be included in the Flora Malesiana. Several new species have been discovered among these specimens, and these are being in- cluded here, since considerable time will elapse before their descriptions will appear in the Flora. Also included are a few new entities in Eurya and Freziera. All are oriental except a single species of Freziera from Colombia. Adinandra anisobasis, sp. nov. Arbor magna, 20-30 m. al juvenilibus rubris, teretibus, adpresso-pubescentibus; ramulis maturis glabris, teretibus, brunneo-rubris; gemmis terminalibus conicis, dense aureo-adpresso-pubescentibus. Folia coriacea, late ovata vel elliptica, 8— 14 cm. longa, 5—7 cm. lata, supra nitida, glabra, subtus pallidora, pubes- centia, punctata, apice obtusa, late acuminata vel rotundata, basi inae- qualia, subrotundata, rare obtusa, margine integra, venis 10—12 paribus, undique elevatis, marginem intra anastomosantibus arcuantibusque, peti- olis brevibus, ca. 2 mm. longis vel subsessilibus. Flores axillares, solitarii; pedicellis teretibus, ca. 1.5 cm. longis, brevi-adpresso-pubescentibus; brac- teolis 2, persistentibus, oppositis vel suboppositis, late ovatis vel deltoideis, subaequalibus, ca. 2.5 cm. longis latisque, adpresso-pubescentibus; sepalis 5, imbricatis, late ovatis vel subrotundatis, duobus exterioribus ca. 5 mm. longis et 4 mm. latis, margine integris, dorso ad marginem dense adpresso- pubescentibus, tribus interioribus ca. 7 mm. longis et 5 mm, latis, margine scariosis, integris, dorso medio adpresso-pubescentibus; petalis 5, cremeo- albis basi connatis, obovatis, ca. 15 mm. longis, 5.5—7 mm. latis, integris, apice rotundatis, exteriore medio adpresso-pubescentibus; staminibus ca. 35, 3-vel 4-seriatis, filamentis ad basim corollae adnatis, 5-7 mm. longis, supra pubescentibus, ad basim glabris, antheris ca. 5 (—6) mm. longis, dense brunneo-pubescentibus; ovario globoso, glabro, ca. 4 mm. diametro, 5-loculari, mutiovulato, apice abrupte in stylum attenuato, stylo integro, glabro, ca. 10 mm. longo, stigmate punctiformi. Fructus maturus non visus. 126 JOURNAL OF THE ARNOLD ARBORETUM _ [VOL. XxxIv BRITISH NORTH BORNEO: Mt. Kinabalu: Penataran river basin, Penataran river, in jungle near Lobang, alt. 3500 ft., J. & M. S. Clemens 34154 (Bogor, type; AA, fragment), July 25, 1933 (tree 70 ft.; diam. 1-2 ft.; flowers cream, the anthers brown). — Penataran river basin, Wun River, alt. 3500 ft., J.& M.S. Clemens 34035 (Bogor), July 22, 1933 (tree 80 ft. high., diam. 2 ft.; flowers cream-white, the anthers brown). The outstanding characteristic of this species is the unequal base of the shining subsessile leaves. In this character it resembles A. cordata, also found on Mt. Kinabalu. However, the latter species can be separated by (1) the truly cordate leaf-base; (2) larger (20 & 10 mm.) glabrous corolla-lobes; and (3) larger calyx-lobes (12 18 mm.), broader than long. Cited here may be a third specimen, unnumbered, collected by J. & M. S. Clemens at Penibukan, 4000 ft. altitude, on Mt. Kinabalu. This speci- men is sterile but agrees with the species in foliage and branch characters. Adinandra endertii, sp. nov. Arbor; ramis rubro-griseis, teretibus glabris (innovationibus leviter pubescentibus), gemmis terminalibus longo-conicis pubescentibus. Folia membranacea vel chartacea, elliptica, glabra, subtus punctata, 4-5 cm. longa, 1.3-1.7 (-—2) cm. lata, apice acuta, basi cuneata decurrentiaque, margine integra vel subintegra, venis 7—9 paribus, petiolis circiter 3 mm. longis, subtus pubescentibus. Flores axillares, solitarii; pedicellis teretibus, tenuibus, circiter 2 cm. longis, glabris (basi saepe leviter adpresso-pubes- centibus); bracteolis 2, oppositis, late ovatis vel subrotundatis, inaequal- ibus, 1.5-2.5 mm. longis, 2.5-3.5 mm. latis, persistentibus, in medio crassioribus, ut videtur glabris (sub magnificatione leviter adpresso-pubes- centibus); sepalis 5, imbricatis, glabris (sub magnificatione leviter ad- presso-pubescentibus), late ovatis, apice rotundatis, subaequalibus, ca. 7 mm. longis, 5—8 mm. latis; petalis 5, ovatis, ca. 6 mm. longis (in gemmis), dorso medio dense pubescentibus: staminibus numerosis, 3-seriatis ?, fila- mentis glabris, ad basim corollae adnatis, antheris dense hirsutis; ovario subplano, glabro, 5-angulato, apice in stylum abrupte attenuato, 5-loculari, multiovulato. Fructus non visus. EAST BORNEO: W. Koetai: near Kemoel, forest in very mountainous country, alt. 1600 m., F. H. Endert 4135 (Arnold Arboretum, TyPE; Bogor, ISOTYPE), Oct. 14, 1925 (tree with greyish green buds). In general characters this species resembles the Philippine species more closely than the Indonesian species. The smaller chartaceous elliptic leaves and the smaller, less robust flowers and floral parts remind one of A. apoensis Elmer of Mindanao. The Philippine species, however, can be separated by the coriaceous leaves, broadly rounded at the apex, with the margin serrate along the upper half. The filaments in A. apoensis are very hirsute — as hirsute as the anthers. 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 127 Adinandra laronensis, sp. nov. Arbor ?; ramulis maturis teretibus, griseis, glabris; ramulis juvenilibus teretibus, rubris, adpresso-pubescentibus; gemmis terminalibus conicis, dense aureo-adpresso-pubescentibus. Folia coriacea, elliptica vel sub- elliptica, 7-13 cm. longa, 2.5-4 cm. lata, supra glabra, subtus glabres- centia, apice late acuta vel obtusa, basi cuneata, margine integra vel minute glanduloso-denticulata, venis ca. 12 paribus, marginem intra anastomo- santibus arcuantibusque, petiolis brevibus, 2-3 mm. longis, glabris. Flores ignoti. Fructus immaturi axillares, solitarii; pedicellis teretibus, 1.5—2 cm. longis, glabris; bracteolis 2, oppositis, persistentibus, glabris, rotun- datis vel subrotundatis, ca. 3 mm. longis latisque; sepalis 5, imbricatis, glabrescentibus (juvenilibus adpresso-pubescentibus) , rotundatis, 5—7 mm. longis latisque; petalis non visis; fructibus immaturis globosis vel sub- globosis, glabris, 5-locularibus, multiseminatis; seminibus nigris, nitidis, 2 mm. longis, reticulatis. CELEBES: Gouvt. Celebes en Onderh.: Malili, near Larona, “Mantri Boschwezen” bb.2399 (Bogor, TYPE), Oct. 13, 1922.— Same locality, “Mantri Boschwezen” 1875 (Bogor). This species owes its proposed status to its very distinct leaves. Ordi- narily one would not suggest a new species on such sparse material, since only a single attached immature fruit is to be found. However, the rather thick coriaceous elliptic leaves make it stand out from the other species on the island. A third specimen (506.3225) collected at the same locality and by the same collector, “Mantri Boschwezen,” obviously an unknown assistant, might be cited here. The leaves in this third specimen are wider. Adinandra loerzingiana, sp. nov. Arbor 25 m. alta; ramis griseis, teretibus, glabris, cicatricibus magnis, ca. 4 < 3 mm., ramulis novellis rubris, leviter adpresso-pubescentibus; gemmis terminalibus longis, adpresso-fulvo-pubescentibus. Folia sessilia, crasso-coriacea, glabra, elliptica vel subelliptica, 10-15 cm. longa, 5—7 cm lata, apice obtusa, subtiliter late acuminata, basi late cuneata, margine integra, venis prominentibus, primariis 20 (vel plus) paribus, marginem intra anastomosantibus, venis secundariis inter primarias na ea Flores axillares, solitari!: pedicellis glabris, curvatis, teretibus, 2—2.5 c longis; bracteolis Z, supoppositis, persistentibus, glabris, deltoideis, inae- qualibus, ca. 2 mm. longis latisque; sepalis 5, imbricatis, glabris, subro- tundatis, inaequalibus, exterioribus duobus minoribus, ca. 5 mm. longis et 7 mm, latis, interioribus tribus ca. 5 mm. longis et 12-13 mm. latis, margine late scariosis; petalis 5, basi connatis, glabris, cremeis, obovatis, inaequal- ibus, exterioribus ca. 15 mm. longis et 9 mm. latis, interioribus ca. 15 mm. longis et 12-13 mm. ee staminibus ca. 40, ut videtur 4-seriatis, inaequalibus, 10, 11, 12 et 13 mm. longis, filamentis 4-5 mm. longis basi connatis, glabris vel apice leviter pubescentibus, antheris inaequalibus, 6, 7, 8, et 9 mm. longis, dense setiferis; ovario late conico, ca. 3 mm. longo, 128 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL. xxxIv 5 mm. diam., ad apicem adpresso-pubescente, 2-loculari, pauci-ovulato, stylo glabro, integro, ca. 1.5 cm. longo. Fructus globosus vel subglobosus, glabrescens vel apice leviter adpresso-pubescens, 1.5-2 cm. diam., 2-locu- laris, pauci-seminatus, seminibus 2—10 (rare uno semine in quoque loculo), subplanis, hippocrepiformibus, 7-8 mm. longis, nitidis. SUMATRA: Gouvt. Oostkust Sumatra: Afd. Sibolangit, near Bandar baroe, rare in primary forest, alt. 850 m., J. A. Loérzing 6862 (Bogor, TYPE), Sept. 7, 1919 (tree 15 m. with hard wood; flowers fragrant, cream-colored) ; same general locality, J. A. Lérzing 7040 (Bogor), Jan. 7, 1920 (tree 25 m. high). This species is characterized by the sessile, thickly coriaceous, many- veined leaves, the persistent bracteoles, the two-celled ovary and fruit, and the few ovules and seeds. In some dissections a single seed was found in each locule of the fruit. The leaf-scars are unusually large, measuring as much as 5 mm. in diameter. This large scar is due to the unusually thick- ened base of the sessile leaves. The only other two species known to have two-celled ovary and fruit are A. nunkokensis Kob. from Mt. Kinabalu, British North Borneo and A. subsessilis Airy-Shaw of Sarawak. In both A. nunkokensis and A. sub- sessilis the flowers are briefly pedicellate, longer (5 mm.) in the former species and subsessile in the latter. In both species the leaves are long- acuminate and the veins are fewer in number (6-8 pairs in A. nunkokensis and 12-14 pairs in A. sudsessilis). Adinandra kjellbergii, sp. nov. Arbor (immatura) 6 m. alta; ramulis multis, teretibus, glabris, inno- vationibus leve adpresso-pubescentibus, gemmis terminalibus parvis, coni- cis, dense adpresso-pubescentibus. Folia coriacea, obovata, 5-9 cm. longa, 2—4 cm. lata, apice obtusa vel rotundata, late acuminata, basi cuneata, supra nitida glabra, subtus pallidiora, sparse adpresso-pubescentia, margine integra vel glanduloso-denticulata, venis undique obscuris, ca. 8 paribus, petiolis 7-10 mm. longis. Flores axillares, solitarii; pedicellis tenuibus, 1.5—2 cm. longis, sparse adpresso-pubescentibus vel glabrescentibus: brac- teolis 2, oppositis, adpresso-pubescentibus, inaequalibus, una late rotun- data, 2.5 mm. longa et 4 mm. lata et altera late ovata, 2.5 mm longa et 3 mm. lata, apiculata; sepalis 5, imbricatis, rotundatis, 7-9 mm. longis, ca. 8 mm. latis, dense adpresso-pubescentibus; petalis 5, albis, basi con- natis, obovatis, ca. 12 mm. longis, 6-8 mm. latis, exteriore medio aureo- adpresso-pubescentibus; staminibus ca. 40, bi-seriatis ut videtur, ca. 7 mm. longis, filamentis ad basim corollae adnatis, connatis, ca. 3 mm. longis, dense pubescentibus, antheris ca. 4 mm. longis, dense pubescentibus; ovario globoso, glabro, ca. 3 mm. diametro, apice subtruncato, 5-loculari, multiseminato. Fructus non visi. CELEBES: Gouvt. Celebes en Onderh.: Makale, Rantepao, near Doa, alt. 1000 m., J. van Ziyl de Jong 29/vz (Bogor bb.20268), Feb. 9, 1936 (young tree 6 m. high with white flowers). — Todjamboe, rare in rain forest, alt. 800 m., 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 129 G. Kjellberg 2281 (Mus. Bot. Stockholm, Type; Bogor, isotype), Aug. 12, 1929 (tree ca. 15 m In this species all the floral parts with the exception of the style and ovary are densely appressed-pubescent. In the stamens both the anthers and the filaments (except the portion touching the corolla) are densely pubescent. The ovary is unusual in that the apex is quite truncate rather than tapering, so that the style rises abruptly rather than gradually as is usually the case. As the flowering material of the type was very sparse, only a single floral dissection was possible. However, the abundant pubescence on the floral parts is distinctive for any Adinandra in the region. Adinandra masambensis, sp. nov. Arbor nondum adulta 25 m. alta; ramulis teretibus, robustis, rigidis, rubro-brunneis, partibus juvenilibus pubescentibus. Folia crasso-coriacea, elliptica vel ovato-elliptica, 7-12 cm. longa et 4—6.5 cm. lata, apice obtusa vel rotundata, subemarginata, basi rotundata vel subcordata, supra glabra subtus juventute pubescentia, glabrescentia, costa supra canaliculata, sub- tus elevata, margine integra, venis undique obscuris, petiolis crassis, teretibus, ca. 5 mm. longis. Flores non visi. Fructus (juvenes) axillares, solitarii; pedicellis ca. 2.5 cm. longis, apice 3 mm. crassis; bracteolis 2, oppositis, persistentibus, adpresso- pubescentibus, latissimis, sublunatis, x 7 mm. latis, adpresso-pubescentibus, margine subciliatis; sepalis 5, imbricatis, crassis, subrotundatis, adpresso-pubescentibus subaequalibus, 9-10 mm. longis et 11-12 mm. latis; ovario globoso, ca. 5 mm. diametro, glabro, 5-loculari, multiovulato; stylo fracto. CELEBES: Gouvt. Celebes en Onderh.: Masamba, near Rato, rare, alt. ca. 1000 m., F. Grot 6 (Bogor, bb. 26298, TYPE), Sept. 11, 1938 (young tree 25 m., diameter 30 cm.; flowers white). Outstanding characteristics of this new species are (1) the thick coriace- ous pubescent leaves, almost elliptic in shape, subretuse at the apex, rounded or subcordate at the base, mot decurrent into the petiole, with obscure vein- ing; (2) the persistent bracteoles, much broader (9 mm.) than long (5 mm). Its closest relative appears to be A. cordata from British North Borneo. This latter species can be separated by its glabrous leaves, sharply oblique- cordate at the base, obtusely acuminate at the apex, with ten to twelve conspicuous pairs of veins, and its sericeous ovary. Neither flowers nor mature fruit were available for this study. However, the glabrous, five-celled fruit, although immature, and the unusual leaves show the species to be distinct from its nearest relatives and so worthy of description. 130 JOURNAL OF THE ARNOLD ARBORETUM _[VvoL. xXxxIv Adinandra rantepaoensis, sp. nov. Arbor ca. 15 m. alta; ramulis paucis, teretibus, glabris; gemmis termi- nalibus glabris. Folia crasso-coriacea, elliptica vel ovato-elliptica, 13-19 cm. longa, 5.5-8 cm. lata, apice late ovata vel obtusa, subacuminata, basi cuneata in petiolum decurrentia, margine integra, venis 8—10 paribus, undique conspicuis, By ee ea ad marginem anastomosantibus, peti- olis ca. 1.5 cm. longis, glabris. Flores solitarii, axillares; pedicellis glabris, teretibus, 2.5-3.5 cm. longis, 2.5-3 mm. crassis, apice curvatis; bracteolis 2, oppositis, crassis, glabris, subaequalibus, latioribus quam longis, semi- rotundatis, 4—5 mm. longis, 7-9 mm. latis, integris; sepalis 5, imbricatis, glabris, crassis, subrotundatis, exterioribus duobus 12 mm. longis et 14 mm. latis, interioribus tribus 9 mm. longis et 11 mm. latis; petalis (im- maturis) 5, obovatis, ca. 15 mm. longis et 12 mm. latis, apice sub- acutis, dorso dense pubescentibus (margine excepto); staminibus ca 60, ut videtur 3-seriatis, ca. 10 mm. longis, dense pubescentibus, filamentis connatis, ca. 7 mm. longis, basi ad corollam adnatis, antheris ca. 3 mm longis; ovario globoso vel subgloboso, 6-7 mm. diametro, dense luteo- pubescente, 5-loculari, multiovulato, pericarpio crasso; stylo 7 mm. longo, glabris basi excepto; stigmate punctato. Fructus non visi. CELEBES: Gouvt. Celebes en Onderh.: Rantepao, Bala, Balambang, alt. 1600 m., J. K. Thenu 19 (bb. 11857) (type, Bogor), Sept. 6, 1927 (tree ca. 15 m.). — Same locality, J. K. Thenu 8 (bb. 11846) (Bogor). This distinctive species is characterized by large heavy-coriaceous gla- brous leaves with eight to ten widely spaced primary veins. The leaf-base is cuneate and decurs into the petiole, nearly to its base. The floral parts are all large. A single stamen adhering to the lobe of the calyx shows that at maturity the longer stamens may measure up to ten mm. The corolla, which was described above from a bud, may be considerably extended at anthesis. The cell cavity of the ovary is rather small and appears to be confined to a small area near the base of the ovary. It measures only ca. 2mm. in diameter. The pericarp is very thick, measuring about 2 mm. thick at the base. The upper two thirds of the ovary is dense and of the same texture as the pericarp, with no cell structure obvious. Adinandra rubiginosa, sp. nov. Arbor 25 m. alta, trunco angulatissimo, 40 cm. diametro; ramulis tere- tibus, rubro-brunneis, pubescentibus; gemmis terminalibus dense hirsutis. Folia coriacea, oblongo-ovata, 10-15 cm. longa, 3.5-4 cm lata, supra glabra, rubiginosa, subtus dense rubro-brunnea hirsuta, apice acuta, basi cuneata, margine integra, venis obscuris, petiolis crassis, ca. 5 mm, longis et ca. 3 mm. diametro. Flores axillares, solitarii; pedicellis crassis, re- curvatis, teretibus, 5 mm. longis, 3 mm. diametro; bracteolis 2, oppositis, persistentibus, crassis, semi- rotundatis, inaequalibus ca. 3 mm ongis 4 4 mm. latis et 2 mm longis & 4 mm. latis, dorso pubescentibus, margine ciliatis; sepalis 5, imbricatis, inaequalibus, duobus exterioribus rotundato- 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 131 ovatis, 5 mm. longis, basi 5 mm. latis, pubescentibus, tribus interioribus subrotundatis, glabris, ca. 7 mm. ona 5 mm. latis, margine scariosis; petalis (immaturis) 5, dorso glabro; staminibus (immaturis) ca 30, . seriatis ?, filamentis basi connatis, ad basim corollae adnatis, subtus, glabris, super hirsutis, antheris elongatis, ca. 3 mm. longis, dense hirsutis, apice apiculatis; ovario semi-globoso, 5-loculari, multiovulato, apice abrupte in stylum attenuato, stylo glabro, integro, stigmate punctiformi. Fructus (immaturus) globosus, glaber, 5-locularis, multiseminatus, stylo 13 mm. longo. BORNEO: Subdiv. W. Koetai: near Kemoel, in forest on steep ravine wall, alt. 1100 m., F. H. Endert 3869 (AA, TYPE; Bogor, IsoTYPE), Oct, 10 1925 (tree 25 m. high, 25 cm. diam.; trunk very angular; flowers white). This species is characterized by dark red, coriaceous, oblong-ovate leaves glabrous on the upper surface (even in the terminal bud) and densely hirsute on the lower surface, becoming glabrescent. The pubes- cence on the younger leaves is striking in that it appears to be densely distributed in distinct longitudinal striae. Although not an unusual oc- currence in other pubescent species of the family, this character is unusual in Adinandra. Also characteristic is the short thick pedicel. Although both flowers and fruit were available for this study, it is un- fortunate that neither was mature. The petals and stamens were described from a bud, making accurate measurements quite impossible. However, the diagnostic characters were discernible, showing this species to be distinct. Adinandra steenisii, sp. nov. Arbor parva, ad 5 m. alta; ramulis multis, congestis, glabris, juventute adpresso-pubescentibus; gemmis terminalibus luteo-adpresso-pubescen- tibus. Folia coriacea, rotundata, parva, 1-2 cm. longa, 0.8-1.5 cm. lata, apice rotundata, subretusa, basi cuneata, supra glabra, subtus cito glabres- i F axillares, solitarii; pedunculis teretibus, ca. 1 cm. longis, glabris; ‘bracteolis 2, oppositis, subaequalibus, latioribus quam longis, semi-rotundatis, 1.7— 1.8 mm. longis, ca. 3 mm. latis, glabris; sepalis 5, imbricatis, glabris, sub- aequalibus, rotundatis, 6-8 mm. longis, ca. 8 mm. latis. Baca globosus, glaber, 8-10 mm. diametro, 5-locularis, multiseminatus, stylo glabro, ca. 10 mm. longo, stigmate punctato. CELEBES: Gouvt. Celebes en Onderh: Route Rante Lemo-Angin Angin via Poka Pindjang, alt. 2500 m., C. G. G. J. van Steenis 10303 (TYPE, Bogor), June 19, 1937 (small tree + 5 m. high). This species is very distinctive and can be easily recognized by its small habit (tree to 5 m. high), the very small rotund leaves (1-2 cm. 132 JOURNAL OF THE ARNOLD ARBORETUM _[VoL. xxxIv long), with the petiole measuring only 1-2 mm., and the small fruit. In general, all the parts seem to be in miniature. The flowers were not seen, and the fruit as described above probably was immature. However, the species should be easily recognized in the future This species is named after C. G. G. J. van Steenis, the collector. Adinandra subauriculata, sp. nov. Arbor ad 22 m.; ramis griseis, teretibus, crassis, glabris; gemmis ter- minalibus ignotis. Folia crasso-coriacea, ovata, glabra, 10-17 cm. longa, 5—6 cm. lata, apice obtusa, basi rotundata, subauriculata, margine minute glanduloso-denticulata, costa subtus conspicua, venis lateralibus ca. 20 primariis paribus, gracilibus ad marginem anastomosantibus, venis secun- dariis inter primarias frequentibus, petiolo brevi, crasso, ca. 5 mm. longo. Flores non visi. Fructus axillares, ut videtur solitarii; pedicellis glabris, teretibus, brevibus, 0.5—1.5 cm. longis, crassis, ca. 2-3 mm. diam.; bracteolis caducis; sepalis 5, imbricatis, persistentibus, ovatis vel late ovatis, ca. 1 cm. longis et 7 mm. latis, leviter adpresso-pubescentibus. Fructus globosi vel subglobosi, leve pubescentes vel glabrescentes, ca. 2 cm. diam., 3- loculares, multiseminati, seminibus subplanis hippocrepiformibus, nigris, nitidis, ca. 5 mm. longis, 3-4 mm. latis. SUMATRA: Res. Sumatra’s Westkust: Soeliki, in old forest, alt. 1300 m., Neth. Ind. For. Serv. bb. 6538 agg Bogor; 1soTyPEe, Arnold Asbovetum), Feb. 16, 1924 (tree 22 m. high, 8 m. to first branch, 40 cm. diam. breast high; fruit green, malodorous, bitter). The outstanding characters of this species are (1) the thick-coriaceous leaf with a rounded subauriculate base and a short thick petiole; (2) a pubescent three-celled fruit with many (up to eighty) flattened hippocrepi- form shiny seeds. The latter character is unusual in the genus in that most species with three-celled fruits are known to possess very few (not more than twenty) large seeds, usually thicker than those found in this species. Several fruits were carefully dissected and all showed at least sixty well- developed seeds. The closest relative appears to be A. cordifolia Ridley, found in Sarawak and British North Borneo. In this species the fruit is five-celled, the leaves are distinctly cordate (unequally so) at the base, subsessile or with a very brief petiole. Adinandra subunguiculata, sp. nov. Arbor ad 15 m. alta; ramulis maturis innovationibusque glabris, tere- tibus; gemmis terminalibus conicis, aureo-adpresso-pubescentibus. Folia coriacea, glabra, obovata vel subelliptica, 9-12 cm. longa, 4-6 cm. lata, apice late acuminata vel obtusa, basi cuneata, margine integra vel minute denticulata, venis 12 paribus vel plus, obscuris, petiolis 4-5 mm. longis. Flores axillares, solitarii; pedicellis glabris, 3.5—4.5 cm. longis, ad apicem 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 133 accrescentibus; bracteolis 2, alternatis vel suboppositis, inaequalibus, bracteola superiore rotundata, ca. 2 mm. longa et 3 mm. lata, glabra, bracteola inferiore deltoidea, ca. 1.5 mm. longa et basi 1.5 mm. lata, glabra; sepalis 5, imbricatis, subrotundatis, subaequalibus, 5-6 mm longis, 6-8 mm. latis, glabris, interioribus rare exceptis, dorso medio minutissime adpresso- puberulentis, cito glabrescentibus; petalis 5, glabris, inaequalibus, exterioribus obovatis, 14 mm. longis, 11 mm. latis, interioribus sub- unguiculatis, 11 mm. longis, 8 mm. latis; staminibus 40-45, 4- seriatis ut videtur, 4-7 mm. longis, filamentis connatis, glabris, ad basim corollae adnatis, 2-3.5 mm. longis, antheris setosis, 2-3.5 mm. longis; ovario conico vel subgloboso, glabro, 5-loculari, multiovulato, stylo glabro, ca. 1 cm. longo. Fructus globosus, glaber, ca. 1 cm. diametro, 5-locularis, multi- seminatus; seminibus nigris, nitidis, reticulatis. CELEBES: Gouvt. Celebes en Onderh.: MALiLr: near Tabaramo, alt. 600 m., A. Hoornstra 32 (Bogor bb.9710), Apr. 9, 1926 (tree 17 m., with white flowers). KeNpartI: N. Kendari, rain-forest, alt. 150-300 m., G. K. Kjellberg 641 (Bogor, TYPE), March 3, 1929.— Poehara, rain-forest, alt. 100 m., G. K. Kjellberg 698 (Bogor), March 6, 1929.— Peninsula, southeast of Lepo-Lepo, O. Beccari 1190, 1190A, 1190B (Firenze), July 1874. — “BoETON EIL MOENA”: Laboenti, alt. 15m., C. P. Burghaut 85 (Bogor bb.5860), July 28, 1923 (tree 14 m. with white flowers). Morena: Labouti, A. G. Waturandang 250 (Bogor, bb.21788), January 22, 1937 (tree 15 m. with white flowers). This species is one which cannot be placed in close association with any other species of the genus, yet has no truly outstanding characteristics of its own to separate it from all others! The bracteoles are usually sub- alternate. All the floral parts except the setose anthers are glabrous, the terminal bud is tawny appressed-pubescent, yet the early leaves and the young branchlets are glabrous or quickly become so. The ovary is five- celled with very many characteristic small ovules. From the label of Bogor 5860 I cannot be certain of the place of col- lection — whether from Moena or the near-by island of Boeton. However, I am inclined to feel that it was collected on Moena, considering the almost duplicate label of Bogor 21788. Adinandra dubia, sp. nov. Frutex; ramulis juvenilibus teretibus, sericeis; ramulis maturis glabris vel glabrescentibus, griseis, teretibus; gemmis terminalibus conicis, dense aureo-sericeis. Folia coriacea, ovata vel elliptico-ovata, 6-11 cm. longa, 2-3.5 cm. lata, supra nitida, glabra, subtus pallidiora, puhescentia: apice acuminata, basi cuneata, margine denticulata (juventute glanduloso-denti- culata), venis ca. 10 paribus, undique leve elevatis, marginem intra anasto- mosantibus arcuantibusque, petiolis 7-10 mm. longis, subtus pubescentibus. Flores axillares, solitarii; pedicellis teretibus, 5-8 mm. longis, recurvatis, sericeis ; bracteolis as suboppositis, longo-triangularibus, 6—6.5 mm. longis, basi 3— 4 mm. latis, adpresso- -sericeis, margine glanduloso-denticulatis, cito caducis; sepalis 5, imbricatis, inaequalibus, dense sericeis, late ovatis, ex- 134 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL.: xxxIv terioribus 6-7 mm. longis, 5-6 mm. latis, interioribus 4-5 mm. longis, 2.5—4 mm. latis; petalis 5, dorso medio adpresso-sericeis, oblongo-ovatis, 3-5 mm. longis, ca. 2 mm. latis; staminibus ut videtur i-seriatis, paucis (ca. 20) ad 2.25 mm. longis, Rlamentia 0.5 mm. longis, connatis, ad basim corollae adnatis, glabris, antheris 0.5—1.5 mm. longis, apice retusis, interiore dense hirsutis; ovario minuto, 1.5 mm. diametro, 5-loculato, multiovulato, glabro vel minute pubescente, stylo ca. 2.5—3 mm. longo, glabro vel minute adpresso-pubescente. Fructus ignotus. SUMATRA: Res. Sumatra’s Westkust: Ophir District, northwest slope of Talamau, alt. 2100 m., H. A. B. Biinnemetjer 910 (Bogor, TYPE), May 28, 1917 (shrub). Adinandra dubia, as the specific name intimates, is a provocative species. Every so often a worker is confronted with a specimen which seems to con- form to a certain genus in all respects, yet instinctively a doubt persists as to its real position. Such is the case with A. dubia. The flowers are very small, with the corolla measuring only three to five millimeters long and two millimeters wide. The bracteoles are nearly petaloid and are actually larger (ca. 6 mm. long) than the petals, a most unusual feature in this genus. Minute are the stamens, the longest ones measuring only 2.5 mm. in length. They are also very few in number yet so closely joined together in the specimens dissected that an accurate count could not be obtained. The copious pubescence usually found on the anthers in most species of the genus is sometimes so sparse and scattered in A. dubia as to be almost negligible. Closely allied and perhaps belonging here is a specimen collected by Mareman | Forest Res. Inst. 6216] at the Hoogvlakte area near Lake Toba in the Tapanoeli Res. Collected at a lower altitude [900 m.], the habit is that of a tree 21 meters high, which in itself is not unusual. However, the immature fruit is quite conical in shape and densely sericeous. So also is the style. This very noticeable difference in pubescence, especially since the Mareman specimen was collected in fruit, cannot be overlooked. A reversed situation would be quite understandable. Eurya perserrata, sp. nov. Frutex 3 m. (fide collectori) altus, ramulis teretibus, innovationibus rufo-pilosis. Folia subcoriacea, oblongo-elliptica vel oblongo-ovata, 14—18 cm. longa et 4.0—-6.5 cm. lata, apice acuminata, basi late cuneata, subtus pilosa, venis utrinque conspicuis, margine acre glanduloso-serrata, serra- tionibus inaequalibus ad 2 mm. longis, sessilia vel subsessilia. Flores é 2-3 in axillis foliorum, albi; pedicellis 1.5-2 mm. longis; bracteolis minutis, sepaloideis, ca. 1 mm. longis; sepala 5, imbricata, inaequalia, sub- rotundata, concava, 3-3.5 mm. longa et 2—2.5 mm. lata, extus intusque strigosa, margine scariosa, ciliata; petala 5, imbricata, inaequalia, oblongo- obovata, 5-6 mm. longa et 2.5—3 mm. lata, obtusa; stamina ca. 15, fila- mentis ca. 2.5 mm. longis, antheris ca. 1 mm. longis. Flores ¢ et fructus ignoti. 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 135 CHINA: Yunnan: Muchielung, Salwin-Kiukiang Divide, in dense mixed forest, alt. 2500 m., JT. T. Yu 21035 (Arnold Arboretum, TYPE), Nov. 21, 1938, shrub 3 m. with white flowers. Unfortunately, neither pistillate flowers nor fruit are available for study in this species. However, the serration on the leaves in the staminate plants is so outstanding that when the pistillate specimens are found they will be easily recognized. The serrations are approximately 2 mm. long (or more), surely the most conspicuous feature on the specimen. No other species in the genus approaches E. perserrata in this character. The rufous pilose terminal buds, the sessile or near-sessile leaves, and the large flowers are other outstanding characters. Eurya pseudocerasifera, sp. nov. Arbor magna (30 m., fide Tsai), ramulis teretibus, ee pubes- centibus. Folia oblongo- elliptica, coriacea, decurrentia, 9-13 cm. longa et 3—5 cm. lata, apice acuminata, basi cuneata, supra cde subtus opaca, margine integerrima, glabra, basi costae excepta, venis elevatis subtus conspicuis, petiolis ca. 5 mm. longis, pubescentibus. Flores @ axillares, 3-fasciculati, albi; pedicellis ca. 3 mm. longis, pubescentibus; bracteolis minutis, sepaloideis, pubescentibus; sepalis 5, imbricatis, inaequalibus, concavis, subrotundatis, ca. 2 mm. longis et 2 mm. latis, pubescentibus, marginibus scariosis ciliatisque: petalis 5, imbricatis, innesualibus, ca, -3 mm. longis et 2 mm. latis, obtusis, apice rotundatis vel subemarginatis; ovario globoso, piloso, ca. 2 mm. diametro, 3-loculari, multiovulato, stylo 3- vel 4-partito, ca. 5 mm. longo, glabro, 1/2 libero: flores ¢ ignoti Fructus (7. 7. Yi 17251) baccatus, globosus, 4—5 mm. diametro, glabres- cens, multiseminatus; semina minuta, nigro-nitida, reticulata, ca. 1 mm. diametro. CHINA: Yunnan: Lung-ling Hsien, in woods, alt. 2300 m., H. T. Tsai 55002 (Arnold Arboretum), Jan. 7, 1934 (tree 20 ft. high with white flowers). — Mong-ka, in forest, alt. 1750 m., H. T. Tsai 56882 (Arnold Arboretum, TYPE), Feb. 9, 1934 (large tree 100 ft. high with yellow fruit). — Chenkang, Snow Range, Tapingchang, common in ravine among forest trees, alt. 2350 a Teter od (Arnold Arboretum), Aug. 6, 1938 (tree 20-25 ft. high). — Taron-Taru Divide, Tangtehwang, in mixed forest, alt. 1800 m., T. T. Yu ue (Arnold Arbore- tum), Aug. 27, 1938 (tree 30 ft. high). This species is characterized by the entire leaves, the pubescent ovary, and the long style (five millimeters), three- or four-parted, the branches free for one half the length. Also most unusual is the size, recorded by Tsai (56882) as a big tree attaining a height of a hundred feet. None of the three other specimens examined and cited above is recorded as taller than thirty feet. This species resembles EF. cerasifolia (D. Don) Kob. superficially in the entire shining leaves and the conspicuous veining. However, in E. cerasi- folia the leaves are not truly entire. Serration may be found on the upper half of the leaf, and then often confined to a single side. Also in E. cerast- 136 JOURNAL OF THE ARNOLD ARBORETUM - [voL. xxxIv folia, the ovary is glabrous, the style is shorter (two millimeters) and joined nearly its whole length Another species in close relationship is E. pittosporifolia Hu. The origi- nal author recorded this species as a tree six meters high with glabrous branchlets, oblanceolate leaves, long-acuminate at the apex and narrow- cuneate at the base, entire or crenulate on the upper two thirds. The sepals were reported as glabrous and the fruit densely ferrugineous-sericeous. The styles were listed as five and distinct. In examining the isotype of E. pitto- sporifolia (C. W. Wang 78362), one finds also that the terminal buds are distinctly pilose and the sepals strigose. Freziera cuatrecasasii, sp. nov. Arbor magna, ramis teretibus, undulatis, crassis, pubescentibus, lenti- cellatis, lenticellis numerosis, prominentibus, elevatis, subsuberosis, ramulis hornotinis dense fulvo-pilosulis. Folia coriacea, ovata, 18-23 (—33) cm. longa, 7-10 (— 14.5) cm. lata, apice abrupte acuminata, basi subrotundata, conspicue inaequalia, margine integra, costa supra canaliculata, subtus basi ad 5 mm. elevata, supra glabra (juventate fulvo-pubescens, cito glabres- cens), subtus pubescens, venis lateralibus 30-37 paribus undique promi- nentibus, reticulatis, petiolis ca. 1 cm. longis, pubescentibus, alatis. Flores axillares et cauliflori in axillis defoliatis pedicellati in fasciculum peduncu- latum aggregati; pedicellis 3 mm. longis vel minus, curvatis, crassis, pubes- centibus; bracteolis 2, oppositis, subrotundatis, ca. 3 mm. longis latisque, dense fulvo-pubescentibus; sepalis 5, imbricatis, adpresso-pubescentibus, rotundatis, ca. 4 mm. longis et 5 mm. latis, margine anguste scariosis, in- tegris; petalis 5, albis, inaequalibus, 6-7 mm. longis, ca. 4.5 mm. latis, exterioribus duobus ovatis, non apiculatis, interioribus tribus apice acutis; staminibus ( @ fl.) ca. 15, uniseriatis, ca. 3 mm. longis, filamentis ca. 1 mm. longis, paucis claviformibus, antheris ca. 2 mm. longis; ovario glabro, conico, ca. 2 mm. diametro basi, apice attenuatis per stylum ca. 3 mm. in stigmate, 5-loculari, multiovulato. Fructus ca. 7 mm. diametro, glaber, 5-locularis, ailtisenduntus. COLOMBIA: Dept. del Valle, Cordillera Occidental, Hoya del rio Digua, Rio San Juan, 1300-1500 m. alt., abaj o de Queremal a la derecha del rio entre km. 52-53, J. Cuatrecasas 23085 (aeacle Arboretum, TYPE; Chicago Nat. Hist. Mus., IsoTYPE), Mar. 27, 1947 (gran arbol; hoja coridcea rigida, verde oscura brillante haz, verde palida enves; sépalos verdes; pétalos blancos). This species is most closely allied to F. guatemalensis in its pedunculate, fasciculate flowers, and to F. Smithiana in its large multiveined leaves and dense fulvous pubescence. It may be separated from the latter species by the pedunculate flowers, the shorter pedicels and petioles, the entire margin of the leaves, the presence in quantity of the conspicuous lenticels, and the somewhat undulate branches. The outstanding features for this species are the huge coriaceous leaves measuring as much as 33 cm. long and 14.5 cm. wide, conspicuously veined, the pedunculate flowers (with the scales of former flowers below) both in 1953] KOBUSKI, STUDIES IN THE THEACEAE, XXVII 137 the axils of the leaves and on the stem in the defoliated axils. The large leaf-scars, along with the so-called cauline flowers, give to the older branches a somewhat undulate appearance which is missing in the younger branches and branchlets. ARNOLD ARBORETUM, HARVARD UNIVERSITY. 138 JOURNAL OF THE ARNOLD ARBORETUM _[vot. xxxiv ILEX IN TAIWAN AND THE LIUKIU ISLANDS SHIU-YING Hu INTRODUCTION IN PREPARING A FLORA OF OKINAWA, Dr. E. H. Walker, Associate Curator, Department of Botany, Smithsonian Institution, asked me to comment on the portion of his manuscript devoted to /lex. Although in my studies of the Chinese species of that genus I included some Taiwan forms, chiefly those occurring on both the island and the mainland of China, my study of the insular material was not exhaustive. As I possessed even less knowledge of the Liukiu Islands species I scarcely felt competent to accept his invitation. Yet my knowledge of the Chinese species of Jlex enabled me to assure him that the two Chinese species included in his manuscript do not occur in the Liukiu Islands. As to the other species I was uncertain and requested him to send me as much Okinawan material as possible. In the identification of these specimens I was naturally forced to consider all the known species of Taiwan and the Liukiu Islands. The results of this study are included in this paper. This paper is supplementary in nature to my publications on THE GENUS ILeEx IN CuinA.! Descriptions, synonyms, and the citations of literature and specimens relating to the species treated therein are not repeated here; however, references to that former work have been included. The speci- mens marked US are deposited in the United States National Herbarium, and those marked G are in the Gray Herbarium. All other cited specimens, whether designated by A or not, are in the Herbarium of the Arnold Arbore- tum. Through Professor E. D. Merrill I have been able to obtain photographs and fragments of types from Dr. Siro Kitamura, Director of the Botanical Institute, Kyoto University, Kyoto, Japan, and Dr. H. Hara of the Botani- cal Institute, Tokyo University, Tokyo, Japan. To both of these men and to the Curators of the above-mentioned institutions I should like to express my deep appreciation for their help in clarifying certain nomenclatural problems in my study of the Jlex of Taiwan and the Liukiu Islands. PHYTOGEOGRAPHICAL SIGNIFICANCES The geographic distribution of the Taiwan and Liukiu Islands species of Tlex reflects the floristic relationships of the areas covered, and to a certain extent it also illustrates the affinity of the floras of these islands to those of the Chinese mainland and Japan, as well as that of the Philippine Islands. * Journal of the Arnold Arboretum 30: 233-344, 348-387, 1949; 31: 39-80, 214-— 240, and 241-263, 1950. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 139 In Taiwan twenty-two valid species occur. Twelve of these occur also on the Chinese mainland. Six of the twelve are confined to China and Taiwan, while four, /. goshiensis Hayata, J. micrococca Maxim., 1. pedun- culosa Miq., and J. rotunda Thunb., are widely spread in China, occurring also in Japan and, in some cases, in Korea or Indo-China. Two other species, I. asprella (Hook. & Arn.) Champ. and J. formosana Maxim., occur also in the Philippines. Among the species confined to Taiwan and China (and closely adjacent regions), two, /. bioritsensis Hayata and J. yunnanensis Franch., are restricted to the higher altitudes of western China and central Paigati, and the others, /. cochinchinensis Lour., I. ficoidea Hemsl., /. pubescens Hook & Arn., and J. triflora Blume var. ‘kanehirai (Yamamoto) S. Y. Hu, are subtropical species which occur in the coastal provinces on the mainland and in the southern part of Taiwan. Some of these also occur in Hainan and in northern Indo-China. It is worthy of note that, although there are four widely spread species that Taiwan shares with Japan and China, there is only a single taxon, /lex sugeroki Maxim. var. brevi- pedunculata (Maxim.) S. Y. Hu, that is confined to Taiwan and Japan. There is no species that is confined to Taiwan and the Philippines. It is also interesting to note that six species, 7. buergeri Miq., I. chinensis Sims, I. crenata Thunb., J. latifolia Thunb., J. macropoda Migq., and J. serrata Thunb., which are common in Japan and eastern China, have not been found in either Taiwan or the Liukiu Islands. Thus for the /lex species of Taiwan, approximately 54% are Chinese elements, 23% are endemic, 18% are shared with the flora of the Liukius, and a very minor percentage are Japanese elements. The genus is strongly marked and easily recognizable and comprises an extraordinary number of distinct species, both deciduous and evergreen. They are in general very specific in their association with different types of vegetation. For exam- ple, J. bioritsensis Hayata is associated with plants constituting the broad- leaved forests of the secondary order, while 7. yunnanensis Franch. var. parvifolia (Hayata) S. Y. Hu is found only along the edge of the coniferous forests. Due to this fact certain species can be used to a considerable ex- tent as indicators of the type of vegetation in their native habitats. There- fore it may perhaps be a legitimate assumption that the proportion of Chinese, Japanese, and endemic elements existing in the //ex flora of Taiwan is likely to prevail in other genera. As illustrated by the species of Ilex, the Taiwan flora is strongly affiliated with that of the China main- land, while the affinity between the flora of Taiwan and that of Japan is weak. The chief Philippine Islands elements are not found in Taiwan proper at all, but are limited to Botel Tobago. There are twelve valid species of J/ex in the Liukiu Islands, four of which are endemic. Four species are limited to Taiwan and the southern Liukiu Islands, and one occurs only in Japan and the northern Liukiu Is- lands. Ilex goshiensis Hayata has a range extending from the Liukius and Taiwan north to Japan and south to Hainan Island. There is also in the Liukiu group one widely spread Chinese subtropical element, J. ficoidea Hemsl., and another widely spread Sino-Japanese species, /. rotunda Thunb. 140 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xXxxIv If the genus //ex may be taken as a criterion, the flora of the Liukiu Islands is characterized by a rather high degree of endemism. Binomials like /lex mertensii Maxim. and Jlex matanoana Makino, which are species endemic to the Bonin Islands, appear repeatedly in litera- ture concerning the floras of Taiwan and the Liukiu Islands. As a matter of fact, however, the Bonin Islands species are morphologically very dif- ferent, and their affinities with the plants of the group of small islands extending southward toward the Micronesias and the Caroline Islands seem to be closer than with the plants of Taiwan and the Liukiu Islands. At least it appears to me that none of the Taiwan and Liukiu material should carry the names applied to the species of the Bonin Islands. KEY TO SUBGENERA, SECTIONS, AND SPECIES A. Leaves deciduous, branchlets with abbreviated shoots, each bearing a fascicle of leaves and flowers; lenticels conspicuous on the current year’s growth... abd do OFS RREESGA Cae een S B. Inflorescence a_trichotomously ae ee cyme bearing 15 or more flowers; fruit small, with smooth pyrenes, each longitudinally canaliculate on the back. om ~ Siz na a. =) i S in) in g BB. Inflorescence a solitary flower or a simple 3- flowered cyme, often fasciculated with the leaves at the end of an abbreviated shoot; fruit medium-sized, with striate and sulcate pyrenes; ne woody or stony. . ca pekaa ae dee oeees t. Prinoides. C. Leaves ovate, entirely glabrous, obtuse. or " shortly parr broadly sc saadiaes at the apex; ovary with no evident style; pyrenes woody, PIRI. kc pee 92 8 eee 80 Ee Fea 2. usanot. CC. Leaves ovate-elliptic, hirsute on the | nerves above, acuminate, the acumen up to 12 mm. long; style evident; pyrenes stony, striate deeply sulcate...................... 20-005. 3. IJ. asprella. AA. Leaves ait branchlets without abbreviated leafy shoots; lenticels usually absent on the current year’s growth. .......... ubgen. EUILEX. B. Pistillate and staminate inflorescences both solitary in the axils of leaves on the current year’s growth; pyrenes smooth, with coriaceous or see aia endocarps, slightly concave, unicanaliculate or 3-striate and 2-sulcate on the back. ........................ Sect. Lioprinus. °C. sip eatin cymose; pyrenes concave or unicanaliculate at the k. D. Fruit ovoid-globose, 5-7 mm. in diameter; pyrenes 4-5 m long, dorsally concave; pedicels slender; leaves a ee on both surfaces (except var. hakkuensis). ....4. J. lonicerifolta. DD. Fruit ellipsoid, 8-10 mm. in diameter; pyrenes 8 mm. long, deeply unicanaliculate on the back; pedicels stout; leaves So Pah ¢6/ i eo a ce oe are ear matudai, CC. Inflorescence oe pyrenes 3-striate and 2-sulcate at the back; leaves entire. ...................0..000-. 6. J. rotunda. CCC. Inflorescence ree ate; the pyrenes smooth, ware ena the back; the fruit ellipsoid- scaled 5-7 mm. in dia Se uae he AOE Ee eS s Oba ends bee ie nae mene 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 141 BB. Pistillate flowers solitary, axillary in the scales or leaves on the current year’s growth, very rarely fasciculate; staminate flowers fasciculate on the second year’s growth, rarely solitary and axillary in the scales or leaves at the base of current year’s growth; pyrenes smooth or slightly roughened; the endocarp coriaceous. ................. Sect. Paltoria. C. Leaves not punctate beneath. D. Petioles 8-17 mm. long, up to one third the length of the lamina; leaves usually entire, rarely guese the midribs plane or slightly ae ADOVE: «<5 youn ee ee I. pedunculosa. DD. Petioles 2- . long, up to one peer ‘the length of the lamina ; ee: usually serrate, the midrib elevated and pubescent a E. Pyrenes 4; branchlets thickly ferruginous-pubescent ; leaves aristately serrate, extending almost to the base. ........ nt ae aes 9. J. yunnanensis var. parvifolia. EE. Pyrenes 4, 5, or 6; branchlets puberulous; leaves serrate or crenate only near the apex, the basal half entire Ts «5s. ..10. J. sugeroki var. bresuibedencilate: CC. Leaves punctate beneat h. D. Pistillate flowers and fruit solitary; leaves elliptic or oblanceo- late, apex acute or shortly acuminate. E. rae elliptic, the apex acute; petioles 7-10 mm. long; a. one third the length of the lamina. ey ee ee ee ll. J. maximowicziana. EE. Leaves oblanceolate; Beenie acuminate; petioles 3-6 mm. long, ca. one eighth the length of the lamina... . I e's 3 2c ins an 2g a2, df; mutchagora. DD. Pistillate flowers and fruit Gea tes leaves obovate or oblong, the apex rounded. ...13. /lex triflora var. kanehirai. BBB. Pistillate and staminate in poreceeniens both fasciculate, axillary on cond year’s growth; pyrenes rugose, pitted or striate with elevated striae; endocarp woody or stony. C. Pyrenes 4 or 2; endocarp woody or stony; eget branches of the inflorescence uniflorous. .................. Sect. Aquifolium. D. Leaves entire or spiny margined. E. Fruit large, 9-12 mm. in diameter; pyrenes stony, irregu- larly yap A ee pitted; leaves entire, acuminate, (4—) 6-7 (-8) cm. long. F. Fruiting ae 4-9 mm. long. .....14. J. integra. FF. Fruiting pedicels 1-3 mm. long. .15. J. brachypoda. EE. Fruit small, 4-8 mm. in diameter; pyrenes striate and sulcate; leaves rigid, spiny, when ey rounded at the apex or rarely short acuminate, 1-4 c ong. F. Leaves ovate or quadrangular, ne on the fruiting branchlets with 2 or 3 strong spines on each side; fruiting pedicels 2 mm. long; pyrenes 2. ......... Crt er hee ee rears 16. J. bioritsensis. FF. Leaves obovate or elliptic; entire or re fruiting pedicels 4-5 mm. long; pyr G. Leaves dimorphous, those on ae fees branch- lets entire and those on the sprouts sinuate and 142 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxIv spiny, the entire ones obovate, rounded at the Ct i an a 17. J, dimorphopla. GG. Leaves entire, elliptic, obtuse or shortly acu- minate at the apex. .......... 18. J. susubii DD. Leaves serrate or crenate. E. Fruiting pedicels 2-3 mm. lon F. Petioles 4-9 mm. long, 12- 22 times shorter than the length of the lamina............. . formosana. FF. oo 8-16 mm. long, 5-12 times shorter than the WON ns ce esa bas Sena pete ts 0. J. ficoidea. EE. Fruiting mere (4-) 6-8 (-14) mm. eaves obovate, rarely elliptic, obtuse, rt recuse, rarely broadly and shortly acuminate at the apex fruiting pedicels 9-14 mm. long. ..21. J. paeeieee FF. Leaves oblong-elliptic, velliptic or obovate-elliptic, apex acuminate; aba pedicels (4—) 6-8 mm. long. G. Fruit 10-11 m n diameter; leaves shortly and abruptly aa a the lateral nerves prominent on both surfaces 22. TI. uraiensis. GG. Fruit 5-6 mm. n diameter: leaves caudate, the lateral nerves sgh ran on both surfaces. ee ee .23. TI. warburgii. CC. Pyrenes 4, 5, or 6, the endocarp coriaceous or sublignescent; in- dividual branches of the fascicles of the staminate plant usually cymose or umbelliform. .......... _ Sect. Pseudoaquifolium. D. Branchlets ridged, in cross- -section ‘appearing quadrangular; pyrenes 6, 3-striate and 2-sulcate, the endocarp sublignescent; leaves chartaceous or membranaceous, serrate or subentire, pubescent on both surfaces. ..... 24. JI. pubescens. DD. Branchlets subterete; pyrenes 4 or o7 smooth the endocarp bro E. Leaves 9-16 cm. long, punctate Aeneas fruit with a columnar-mammiform stigma; pyrenes smooth. ........ 25. 0, cochinchinensis. EE. ‘Leaves 2 a: cm. - long, epunctate, fruit with a capitate or discoid stigma; pyrenes 3-stria F. Leaves suborbicular or oo. elliptic, the apex shortly produced, usually retuse; individual Daas of the inflorescence 1- or 3-flowere I. goshiensis. FF. Leaves ovate-elliptic or elliptic, aceite individ- ual branches of the inflorescence uniflorous. DANS Ae ab ehe os bade cate ds hayateiana. 1. Ilex micrococca Maxim. in Mém. Acad. Sci. St. Pétersb. VII. 29 (3): 39, pl. 1, fig. 6. 1881. —S. Y. Hu in Jour. Arnold Arb. 30: 261. 1949. Ilex micrococca var. Shien Hayata, Ic. Pl. Form. 3: 55, pl. 9. 1913, et . Form. 6 (Suppl.): 13. (Gen. Ind. Fl. Form. 13). 1917. — Ito, Taiwan Shokub. Zus. fig. a Bee es haa Cat. Gov. Herb. 317. 1930. — Suzuki in Masamune, Short Fl. Form. 125. 1936 Hayata differentiated his variety by the “much narrower leaves with acute or obtuse base which is never rounded as is the case with the type.” 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 143 The base of the leaf in this species varies according to the position of the leaf on the branchlet. The leaves on the lower part of the branchlet have a broader base, which appears rounded, and those on the apical portions have a narrower base which is obtuse or acute. Such variations occur on specimens collected in Japan and China as well as on those from Taiwan. Moreover, the Taiwan specimens that I have examined appear to be iden- tical with Hayata’s type (photograph) of this variety, N. Konishi, Aug. 1906, from Uraisha. All these specimens agree with the Chinese and Japanese material in the size, texture, and margin: of the leaves, in the inflorescence, and in the characters relating to the fruit and pyrenes. I see no reason for maintaining this Taiwan variety. In Taiwan, //ex micrococca occurs only in the northern and west-central parts of the island. It has been recorded from Sozan,.Toyen, Uraisha, the mountains Daiton, Taihei and Gosi, and Lake Jitugeti%an. It is a common tree in the woods, growing to a considerable height (up to 23 meters). In the whole family there is no species that equals /. micrococca in the length of its fruiting peduncles and in the large number of fruits in each infructescence. For this reason, this species is well worthy of introduction into southern American gardens for the fruiting branches. The chartaceous leaves, the conspicuous lenticels on the current year’s growth, and the occasional occurrence of abbreviated shoots in this species suggests its relationship with /. macrocarpa Oliv., a deciduous species. Of all the specimens that I have examined, I have not seen any with leaves on the second year’s growth. For this reason I have placed the species in the subgenus Prinos. Field data for the winter habit of the species are needed. When such data are available, it may prove necessary to change the subgeneric status of this species. bo Ilex kusanoi Hayata in Jour. Coll. Sci. Tokyo 30: 55 (Mat. Fl. Form. 55). 1911; Ic. Pl. Form. 1: 132. 1911, et 6 (Suppl.) : 13 (Gen. Ind. FI. Form, 13). 1917.— Kanehira, Form, Trees 123. 1917, rev. ed. 375, fig. 331. 1936. — Sasaki in Trans. Nat. Hist. Soc. Form. 18: 330, 1928, et Cat. Gov. Herb. 317. 1930. — Suzuki in Masamune, Short Fl. Form. 125. 1936. — Sonohara et al., Fl. Okinawa 92. 1952 Ilex taiwaniana Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 58 (Mat. Fl. Form. 1911; Ic. Pl. Form. 1: 135. 1911, et 6 (Suppl.): 13 (Gen. Ind. Fl. Form. 13). 1917. — Kanehira, Form. Trees 127. 1917. Ilex poneantha Koidzumi, Pl. Nov. Amami-Ohsim. 13. 1928, syn. nov. Deciduous tree (?), entirely glabrous, with both elongated and ab- breviated shoots; second year’s growth of the elongated shoot 3—5 mm. in diameter, when dry castaneous, shiny, with conspicuous lenticels, the cur- rent year’s growth subterete, up to 20 cm. long, 2 mm. in diameter; ab- breviated shoots 1 cm. long, 2.5 mm. in diameter, rugose with leaf-scars and scars of the pedicels, bearing 1—4 leaves, 3-13 flowers. Leaves 6-15 mm. apart on elongated shoots, fasciculate at the end of abbreviated shoots: stipules minute, broadly deltoid; petioles 5-8, rarely up to 10 mm. long, 144 JOURNAL OF THE ARNOLD ARBORETUM [voL. xxx1v one tenth up to one sixth the length of the lamina, the distal portion winged with decurrent leaf-base, canaliculate above; lamina chartaceous, oliva- ceous, ovate, 4—-6.5 cm. long, 2.5—4 cm. wide, acute at the base, obtuse or shortly broad-acuminate at the apex, remotely crenulate-serrate, the midrib slightly impressed above, prominent beneath, the lateral nerves 5—7 pairs, obscure above, evident beneath, with reticulation obvious beneath. In- florescence fasciculate, in the axils of scales or small leaves at the base of the elongated shoots or at the apex of the abbreviated ones. Staminate flowers solitary or in a simple 3-flowered cyme, peduncles 3 mm. long, pedicels 3-4 mm. long, prophyllus basal, ciliate; calyx patelliform, 4- up to 6-lobed, the lobes ovate, erose, ciliate; corolla 10 mm. across, the petals oblong, 4 mm. long, 2.5 mm. wide, connate at the base; stamens shorter than the petals, the anthers oblong; rudimentary ovary pulvinate, de- pressed at the center. Pistillate flowers 3 in each fascicle, the pedicels 12— 17 mm. iong; calyx patelliform, 2.5 mm. across, with 5 or 6 lobes, the lobes rounded and ciliate; corolla rotate, 5 mm. across, connate at the base, the lobes ovate, 2 mm. long, 1.5 mm. wide; staminodes two thirds the length of the corolla, the sterile anthers sagittate; ovary globose, 1 mm. in diam- eter, the stigma mammiform. Fruit not seen. TAIWAN: Taito, S. Kusano 8, July 1908 (photograph of the TYPE); Kashioto [Kwasyoto], G. Nakahara 1025 (photograph of type of /lex taiwaniana Hayata). BoTEL Tosaco: Kotosyo [Kotosho], 5. Sasaki, Feb. 7, 1920. LIUKIU ISLANDS: Amami-Oshima, J. Ohba 171 (LECTOTYPE of Jlex poneantha Koidz., photograph and fragments). Ilex kusanoi Hayata is closely related to Ilex macropoda Miq., but the latter species has relatively longer petioles, puberulent leaves, solitary pistillate flowers with the pedicels 6-7 mm. long, and uniflorous individual staminate flowers in small fascicles. Hayata distinguished this species from his /. tatwaniana by its “much thinner” leaves, but in his descriptions for both species he used the same term, “chartaceo-mebranacea.” As the types of these two binomials were collected at different seasons, there may be a reason for the difference in the texture of the leaves. The fragment of the type of /lex poneantha Koidz., kindly supplied by Dr. S. Kitamura of Kyoto, Japan, represents a staminate plant of Ilex kusanoi Hayata. 3. Ilex asprella (Hook & Arn.) Champ. ex Benth. in Hook Jour. Bot. Kew Gard. Misc. 4: 329. 1852. Henry in Trans. As. Soc. Jap. 24 (Suppl.): 26 (List Pl. Form. 26). 1896. — Mat. & Hayata, Enum. PI. Form. 81. 1906. — Ito, Taiwan Shokub. Zus. fig. 486. 1928, — Sasaki, Cat. Gov. Herb. 316. 1930. — Suzuki in Masamune, Short Fl. Form. 124. 1936.—S. Y. Hu in Jour. Arnold Arb. 30: 269, 1949, TAIWAN: Taipei, H. Keng 1024. This species is widely distributed in the warm temperate, subtropical, and tropical regions on the mainland of China and on Taiwan, and extends southward to northern Luzon in the Philippines. In Taiwan it has been 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 145 reported from the mountains Daiton, Gosizan and Sozan, and from Toyen of Taihoku prefecture in the north, Nanto of Taichu prefecture in the west, Karenko [Kwarenko] and Taito of the east coast, South Cape at the south- ern extremity, and Mt. Niitaka in the center of the island. It is a common shrub about three meters high with a trunk up to 6 cm. in diameter, oc- curring in thickets and along roadsides from sea level up to 1000 meters altitude. Its white flowers appear in early March or April. The fasciculate staminate flowers, the solitary long-pedicellate pistillate flower, the globose fruit with evident style and capitate stigma, and the striate pyrenes sug- gest close relationship with //ex longipes Chapm. of southeastern North America, which ranges from Florida to Alabama and Tennessee. These species of //ex provide another proof of the affinities of the floras of eastern Asia and eastern North America. 4. Ilex lonicerifolia Hayata, Icon. Pl. Form. 3: 54, pl. 8. 1913, et 6 (Suppl.): 13 (Gen. Ind. Fl. Form. 13). 1917.— Sasaki, Cat. Gov. Herb. 317. 1930. — Suzuki in Masamune, Short Fl. Form. 125. 1936. —§. Y. Hu in Jour. Arnold Arb. 30: 290. 1949. The range of this species is limited to a narrow band across the center of Taiwan, Lat. 23° 50’-24° 20’ N. The type material was collected from Pokupokusha, Kwarenko, on the east coast. Additional specimens have been collected from Mt. Daisetu and from Nanto and Lake Candidius (Jitugetutan) of Taichu prefecture, at an altitude of 750 meters. The plant has been reported as a tree up to 17 meters high. It flowers in May. The mature fruits are red. It can be readily recognized by its pubescent entire leaves, cymose inflorescences, medium-sized ovoid-globose fruits, and smooth pyrenes which are shallowly concave on the back. 4a. Ilex lonicerifolia var. hakkuensis (Yamamoto) S. Y. Hu in Jour. Arnold Arb. 30: 291. 1949 Ilex hakkuensis Yamamoto Suppl. Ic. Pl. Form. 1: 32, fig. 14. 1925. — Sasaki, Cat. Gov. Herb. 316. 1930. — Suzuki in Masamune, Short Fl. Form. 125. 1936 This glabrous variety was first collected at Hakku by B. Hayata in April 1916. Material of this variety has also been collected from Rengeti and Lake Jitugetutan of Taichu prefecture. 5. Ilex matsudai Yamamoto Suppl. Ic. Pl. Form. 1: 37, fig. 17. 1925. — Sasaki, Cat. Gov. Herb. 317. 1930. — Kanehira, Form. Trees 377, fig. 333. 1936. Ilex lonicerifolia Hayata var. matsudai Yamamoto in Jour. Trop. Agr. 5: 55. An evergreen tree, entirely glabrous; the third and second years’ growth 3-4 mm. in diameter, more or less rugose with numerous conspicuous lenticels, the current year’s growth 2 mm. in diameter, longitudinally 146 JOURNAL OF THE ARNOLD ARBORETUM | [vot. xxxIv striate, the terminal buds ovoid, with glabrous and ciliate scales. Leaves occurring even on the third year’s growth, 1—2 cm. apart, the stipules ob- scure; petioles 8-15 mm. long, rugose, canaliculate above; lamina sub- coriaceous, olivaceous and brown, slightly shiny above, opaque beneath, oblong-elliptic, the lower ones often broad-elliptic or suborbicular, 4—9.5 cm. long, 2.5-4 cm. wide, acute or rarely obtuse at the base, very shortly broad-acuminate, rarely obtuse at the apex, the acumen 3—5 mm. long, almost as wide; margin entire, more or less recurved when dry; midrib plane above, elevated beneath, the principal lateral nerves 9-11 pairs, obscure above: elevated beneath, reticulate near the margin. Flowers not known. Infructescence cymose, solitary, in the axils of the leaves on the current year’s growth, with 3 fruits, the peduncles 3-11 mm. long, dorso- ventrally compressed, glabrous; pedicels 5 mm. long, very minutely puberu- lent at the base. Fruit ellipsoid, 8-10 mm. long, 6-8 mm. in diameter, the persistent calyx explanate, 4 mm. in diameter, with 5 or 6 rounded, ciliate lobes, the stigma discoid. Pyrenes 4 or 5, smooth, 8 mm. long, 2— mm. wide on the back, dorsally deeply and widely unicanaliculate, in cross-section U-shaped, the endocarp sublignescent. TAIWAN: Koshun [Kosyun], Mt. Hiiran, 2 Matsuda in 1919 (photo and fragment of TYPE); same locality, K. Yamada The evergreen entire leaves, the cymose eee and the deeply unicanaliculate pyrenes of this species suggest a close relationship with Ilex maclurei Merr. of Kwangtung and northern Indo-China. Both of them are tropical species, but the latter can be distinguished by its large and thicker leaves and the compound cymose inflorescence. 6. Ilex rotunda Thunb., Fl. Jap. 77. 1784. — Henry in Trans. As. Soc. Jap. 24 (Suppl.): 27 (List Pl. Form 27). 1896.— Mat. & Hayata, Enum. Pl. Form. 82. 1906. — Sasaki, Cat. Gov. Herb. 318. 1930. — Ito, Taiwan Shokub. Zus. fig. 487. 1928. — Suzuki in Masamune, Short FI. Form. 126. 1936. — S. Y. Hu in Jour. Arnold Arb. 30: 308. 1949. Ilex koshunensis Yamamoto, Suppl. Ic. Pl. Form. 1: 36, fig. 16. 1925.— Suzuki in Masamune, Short Fl. Form. 125. 1936. Tlex sages Yamamoto, op. cit. 1: 39, fig. 19. 1925. — Sasaki, Cat. Gov. Herb. 18. 1930. — Suzuki, op. cit. 126. 1936. TAIWAN: Toyen, Kayahara, Hayata & Sasaki (photograph of type of Jlex sasakii Yamamoto); Koshun, Botansha, Nakahara 941 (photograph of TyPE of Ilex koshunensis Yamamoto). LIUKIU ISLANDS: Amami-Oshima, Y. Hosoyamada, July 29, 1927; without precise locality, C. Wright 184 (G) This is the most widely distributed species of //ex in eastern Asia, rang- ing from Korea and Japan southward to the Liukiu Islands, Taiwan, and the mainland of China and Indo-China. In Taiwan it occurs on the moun- tains of Horan, Sitisei, and Taihei, and at Toyen, Sirin, and Sitiku of the 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 147 Taihoku prefecture, at Suwo of the Giran prefecture, Nanto, Mt. Noko, and Lake Candidius of the Taichu prefecture, Daibu of the Taito pre- fecture, and Koshun (Kosyun) of Tainan prefecture. It has been reported as a bush three meters high, as well as a tree of ten meters. The scarlet fruit matures in November. The differences in the climatic and edaphic conditions prevailing in the large area covered by the wide range of the species naturally induce vari- ations in the size, shape, and texture of the leaves. As to the Taiwan material, several binomials have been given to the various collections. As the types of these species are compared with a very large number of speci- mens from the extensive range of the species, many intergrades are noted. Thus with specimens from widely separated regions, the characters em- ployed to separate Thunberg’s from the more recently published species become insignificant and cease to be of value even for distinguishing vari- eties. 6a. Ilex rotunda Thunb. var. microcarpa (Lindl. ex Paxt.) S. Y. Hu in Jour. Arnold Arb. 30: 310. 1949. Ilex microcarpa Lindl. ex Paxt. in Fl. Gard. 1: 43. 1850. In Taiwan this variety occurs in the northern and west-central parts of the island. It differs from typical JJex rotunda Thunb. in the puberulous inflorescence. The fruit of this variety is red. Masamune in 1935 described a yellow-fruited variety, /lex rotunda Thunb. var. sinensis (Trans. Nat. Hist. Soc. Form 25: 13) from Hongkong and Nanhoi district in Kwang- tung province. As it also has puberulous inflorescences, its relationship is probably closer to this variety than to the typical Jlex rotunda Thunb. There is a possibility that Masamune may have observed immature fruits. 7. Ilex tugitakayamensis Sasaki in Trans. Nat. Hist. Soc. Form. 21: 153, fig. 3. 1931. — Suzuki in Masamune, Short Fl. Form. 126. 1936. —§. Y. Hu in Jour. Arnold Arb. 30: 288. 1949. This unique species with its subcapitate umbelliform infructescence is known only from the type collection. It grows on Mt. Tugitaka. It has a peculiar position in the classification of the units of the section to which it belongs. As the fruiting pedicels all originate from the enlarged end of ihe peduncle, the inflorescence is umbelliform, and the species should belong to the Series Umbelliformes (Loes.) S. Y. Hu. But species of that series all possess three striate and two sulcate pyrenes, while the pyrenes of this species are smooth, coriaceous, and flattened or slightly concave on the back. In this section such pyrenes are found only in representatives of the Series Chinenses S. Y. Hu. More material is needed for the clarification of the taxonomic position of this species. Its coriaceous entire leaves and the much reduced fruiting pedicels suggest a relationship with Jlex lancilimba Merr. of Hainan Island. Future students may find it desirable to segregate these species in a distinct series of the Section Lioprinus (Loes.) S. ¥. Hu. 148 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv 8. Ilex pedunculosa Miq. in Vers]. Med. Kon. Akad. Wet. II, 2: 83. 1868 [1866] (Repr. 19. 1866), et in Ann. Mus. Bot. — -Bat. 3: 106. 1867. —S. Y. Hu in Jour. Arnold Arb. 30: 334. 1949 Ilex morit Yamamoto, Suppl. Ic. ie iis 1: 38, fig. 18. 1925. — Suzuki in Masamune, Short Fl. Form. 125. 1936. ae ih ei Yamamoto, l.c. 34, . 15, 1925. — Sasaki, Cat. Gov. Herb. 930. — Suzuki, l.c Slip sige Beir Central Mountain Range, U. Mori, Dec. 1908 (photograph of of J. morti Yamamoto); Taichu, Nanto, Saramao, E. Matsuda, Aug. 11, ino Eieiak of TYPE of J. impressivena Yamamoto », Yamamoto published two species from central Taiwan. He distinguished I. mori from I. pedunculosa Miq. by its smaller leaves and / pressivena by the impressed midribs of the leaves. These variations also occur in specimens collected in China and Japan. After a comparative study of the Taiwan, Chinese, and Japanese specimens, I can only conclude that the two Formosan forms are conspecific with Miquel’s species. In Taiwan, the species occurs in the west-central part of the island. The white flowers appear in August. 8a. Ilex pedunculosa Miq. var. taiwanensis S. Y. Hu in Jour. Arnold Arb. 30: 336. 1949. This is a small-leaved form endemic to northern Taiwan. The small white flowers appear in July. The small size and the brunneous-nigrescent color of the dried leaves of this variety remind one of lex sugeroki Maxim. var. brevipedunculata (Maxim.) S. Y. Hu, but the latter has much shorter petioles. 9. Ilex yunnanensis Franch. var. parvifolia (Hayata) S. Y. Hu in Jour. Arnold Arb. 30: 341. 1949. Ilex parvifolia Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 57 (Mat. Fl. Form. 57). 1911; Ic. Pl. Form. 1: 134, fig. 19. 1911, et 6 (Suppl.): 13 (Gen. Ind. Fl, Form. 13). 1917. — Sasaki, mn Pais Herb. 318. 1930. — Suzuki in Masamune, Short Fl. Form. 125. Ilex transarisanensis Hayata ex a. tae Trees 127. 1917. This is a high mountain form. In Taiwan it occurs at altitudes of 2500— 3300 meters in the central and southwestern parts of the island. It is com- mon along the edges of the forests between Mt. Arisan and Mt. Niitaka. A compact shrub, 1.5—5 meters high, with shiny dark green foliage, it bears white flowers in June and bright red berries from November to February. It should be an excellent species for foundation plantings, low screens, or hedges. 10. Ilex sugeroki Maxim. var. brevipedunculata (Maxim.) S. Y. Hu in Jour. Arnold Arb. 30: 343. 1949. Ilex sugeroki Maxim. forma is tects Maxim. in Mém. Acad. Sci. St. Pétersb. VII, 29 (3): 36, pl. 1, fig. d. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 149 i taisanensis Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 57 (Mat. Fl. Form. 57). 1911; in Ic. Pl. Form. 1: 134. 1911, et 6 (Suppl.): 13 (Gen. Ind. FI. Form. 13). 1917.— Sasaki, Cat. Gov. Herb. 318. 1930.— Suzuki in Masamune, Short Fl. Form. 126. 1936 This variety occurs on the northwestern slopes of the Central Mountain Range, at Hori and Bioritu [Byoritu]. The few specimens that I have examined do not have the characteristic olivaceous or brown shiny leaves of the dried Japanese material. This may be due to the technique used in their preparation. 11. Ilex maximowicziana Loes. in Nov. Act. Acad. Caes. Leop.-Carol. Nat. Cur. 78: 339 ee Aquif. 1: 339). 1901. — Sonohara et al., Fl. Okinawa 92. 1952 Ilex crenata sensu Ito & Matsum. in Jour. Coll. Sci. Tokyo 13: 367 (Tent. FI. Lutch. 367). 1900, non Thunb. 1784. Ilex crenata Thunb. var. scoriatum Yamamoto, Suppl. Ic. Form. 1: 31. 1925, non var. scoriarum W. W. Smith. 1917. Ilex scoriatulum Koidzumi in Bot. Mag. Tokyo 43: 389. 1929. — Sasaki, Cat. Gov. Herb. 318. 1930. — Kanehira, Form. Trees, rev. ed. 381, fig. 338. 1936 (7. scoriatum).— Suzuki in Masamune, Short Fl. Form. 126. 1936. An evergreen tree 3 m. high with a trunk 4 cm. in diameter; branchlets longitudinally ridged, the third year’s growth 3 mm. in diameter, the lenti- cels obscure, the current year’s growth 1.5 mm. in diameter, subquadrangu- lar, very minutely puberulous, the terminal buds usually well developed, subconic, puberulent. Leaves occurring also on the second year’s growth, 3— 10 mm. apart, the stipules aciculate, 0.7 mm. long, persistent; petioles 7— 10 mm. long, one third or one fourth the length of the lamina, puberulous, deeply grooved above; lamina coriaceous, olivaceous, shiny above, opaque and punctate beneath, elliptic or broad-elliptic, 2-4.5 cm. long, 1.3-2.3 cm. wide (up to 8 cm. long and 3.4 cm. wide, according to Loesener), acute at both ends, rarely the lower ones obtuse or rounded at the apex; evenly crenulate-serrate at the margin; midrib impressed above, elevated beneath, the lateral nerves obscure on both surfaces. Staminate inflorescences fasciculate, the individual branches 1- or 3-flowered, the flowers white, 4-merous; the calyx patelliform, the corolla rotate, the stamens shorter than the petals, with oblong anthers, the rudimentary ovary subglobose, with a rostellate center. Pistillate inflorescence solitary in the axils of small leaves or scales on the lower portion of the current year’s growth; pedicels 7 mm. long, the flowers not known. Fruit globose, 8 mm. in diameter, the persistent calyx 3.5 mm. in diameter, 4-lobed, the lobes rounded, ciliate; the stigma discoid. Pyrenes 4 in number, 5 mm. long, 4 mm. wide, the endocarp coriaceous, smooth, 5-striate, the striae impressed. TAIWAN: Taito, Taririku-sya, S$, Sasaki in May 1924 (photograph). LIUKIU ISLANDS: Ishigaki, J. L. Gressitt 608 (TopotyPE). Irumuti, S. Kawagoe on July 27, 1919. 150 JOURNAL OF THE ARNOLD ARBORETUM _ [VoL. XxxIv Ilex maximowicziana was based on the collections of Warburg and Tashiro from Ishigaki. The species has been overlooked by botanists interested in the flora of Taiwan and the Liukiu Islands. In 1925 Yamamoto interpreted the collections of Soma, Matsuda, and Sasaki from Taiwan, and Tashiro’s collection (type material for /lex maximowicziana Loes.) from Liukiu as Ilex crenata Thunb. var. scoriarum (scoriatum) W. W. Smith. In 1929 Koidzumi, presumably on the basis of the same collections cited by Yama- moto, raised the variety to specific rank, and not realizing that there was an earlier valid name for the Liukiu plant, called it Jlex scoriatulum. This species is very closely related to //ex viridis Champ. ex Benth. The differences between them rest on very minute and technical characters. In I. maximowicziana the petioles are relatively longer, equalling one third or one fourth the length of the lamina, and the pyrenes are smooth with im- pressed striae, while in //ex viridis the petioles are relatively shorter, being one thirteenth to one sixth the length of the lamina, and the pyrenes are rugose with slightly elevated striae. The insular taxon is not well known. More material from Taiwan and the Liukiu Islands may prove it to be con- specific with the mainland form. In Taiwan, this species occurs at the south- ern end of the island. 12. Ilex mutchagara Makino, in Bot. Mag. Tokyo 27: 75, 1913.— Sasaki, Cat. Gov. Herb. 318. 1930, pro parte. — Sonohara et al., FI. Okinawa 92. 1952. An evergreen shrub up to 3 m. high, glabrous, the branchlets subquad- rangular, deeply grooved, the third year’s growth 1.5—2 mm. in diameter, the lenticels lacking, the leaf-scars semicircular, very slightly elevated; current year’s growth 1 mm. in diameter, the terminal buds weakly de- veloped. Leaves occurring also on the second year’s growth, 2-5 mm. apart; the stipules callose, deltoid, 0.5 mm. long; petioles 3-6 mm. long, up to one eighth the length of the lamina, deeply narrow-grooved above. Lamina subcoriaceous, olivaceous, shiny above, opaque and punctate be- neath, oblanceolate or obovate-elliptic, (2—) 4-5 cm. long, I1-1.5 cm (rarely up to 1.8 cm.) wide, cuneate at the base, shortly and broadly acuminate or rarely rounded at the apex, the acumen obtuse at the tip, remotely crenulate-serrate on the anterior half; midrib impressed above, elevated beneath, the lateral nerves obscure on both surfaces. Staminate inflorescence fasciculate, the individual branches 3- or 1-flowered, the peduncles 3—7 mm. long, minutely puberulent, the pedicels 2—3.5 mm. long, subpuberulent, the flowers 4-merous; calyx patelliform, the lobes sub- orbicular, ciliate; corolla rotate, the petals broad-elliptic, the stamens with ellipsoid anthers; the rudimentary ovary minute. Pistillate flowers not known. Fruit solitary, in the axils of the bracts at the base of the current year’s growth, the pedicels with 2 prophylla at the middle, the berries globose, 7-8 mm. in diameter, the persistent calyx explanate, 3 mm. across, the lobes rounded, sparsely ciliate, the stigma thickly discoid. Pyrenes 4, smooth, 5 mm. long, the endocarp coriaceous. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 151 LIUKIU ISLANDS: Okinawa, Nago, E. H. Wilson 8071. Makino based his description chiefly on material collected from Okinawa, principally from Nago. Wilson’s collection appears to have come from the type locality. Makino also cited material from Kume and Iriomote. Speci- mens from these islands belonging to this species have not been available for my examination. As Makino observed, this species is closely allied to Jlex crenata Thunb. The oblanceolate, loosely arranged leaves and glabrous branchlets are very distinct. It is endemic to the Liukiu Islands, where it is rare even at the type locality. Its white flowers appear in February. 13. Ilex triflora Blume var. kanehirai (Yamamoto) S. Y. Hu in Jour. Arnold Arb. 30: 332. 1949. Ilex crenata Thunb, var. kanehirai Yamamoto, Suppl. Ic. Pl. Form. 1: 31, fig. 11, 1925 Ilex kanehirai (Yamamoto) Koidz. in Bot. Mag. Tokyo 43: 389. 1929. — Sasaki, Cat. Gov. Herb. 317. 1930. — Kanehira, Form. Trees, rev. ed. 375. 1936. — Suzuki in Masamune, Short Fl. Form. 125. 1936. Tlex ae Makino var. kanehirai (Yamamoto) Masamune in Trans. t. Soc. Form. 25: 253. 1935. — Kanehira & Hatusima in Trans. Nat. Hist. Form. 29: 156. 1939. — Sonohara et al., Fl. Okinawa 92. 1952. Regarding the status of this taxon, there is considerable controversy among botanists. Yamamoto first placed it as a variety of Jlex crenata Thunb. Koidzumi in 1929 raised it to specific rank, and both Kanehira and Sasaki agreed with him. Masamune maintained it as a variety of /lex mutchagara Makino. Kanehira also adopted this concept. /lex mutchagara Makino is characterized by oblanceolate leaves with a shortly acuminate apex and solitary fruits. The present species has fasciculate fruits and obovate leaves with rounded apex. Its relationship should be closer to Ilex triflora Blume, a species with fasciculate fruits, which is widely spread in China This variety was first described from Giran in northeastern Taiwan. Ad- ditional material has since been collected from Kosyun at the southern extremity of the island. It also occurs on the island of Hainan and in Fukien Province of the Chinese mainland. 14. Ilex integra Thunb., Fl. Jap. 77. 1784. — Sieb. & Zucc. in Abh. Bay. Ak. Wiss. Math. Phys. 4 (2): 148 (FI. Jap. 1: 40). 1845. — Maxim. in Mém. Acad. Sci. St. Pétersb. VII, 29 (3): 28, 41, pl. 1, fig. 3. 1881. — Forbes & Hemsl. in Jour. Linn. Soc. Bot. 23: 116. 1886.— Ito & Matsum. in Jour. Coll. Sci. Univ. Tokyo 12: 368. 1900.— Loes. in Nov. Act. Acad. Caes. Leop.-Carol. Nat. Cur. 78: 270 (Monog. oy 1: 270). 1901. — Matsum. & Hayata, Enum. Pl. Form. 82. 1906. Sasaki, Cat. Gov. Herb. Form. 317. 1930.— Suzuki in Masamune, Short Fl. Form. 125. 1936. — Rehder, Man. Cult. Trees and Shrubs, 152 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv ed. 2, 548. 1940; Bibliogr. er Trees and Shrubs 400. 1949. — Sono- hara et al., FI. Oki inawa 92. 1952. Othera japonica Thunb. (praes., resp. Lodin), Nov. Gen. Pl. 56. 1783. Ilex othera Sprengel, Syst. Veg. 1: 496. 1826. Ilex asiatica Sprengel, 1.c. Prinos integra Hook. & Arn. in Bot. Beechey Voy. 261. 1853. Small evergreen trees 5—8 m. high, the trunk 12-20 cm. in diameter, glabrous; branchlets subterete, striate, the third year’s growth 3-4 mm. in diameter, the lenticels lacking, the leaf-scars suborbicular, conspicuous, the current year’s growth 2-3 mm. in diameter, the terminal buds well de- veloped, conic, glabrous. Leaves occurring also on second year’s growth, 3-15 mm. apart, the stipules obscure; petioles 1-1.5 cm. long, up to one third the length of the lamina, narrowly canaliculate above; lamina cori- aceous, olivaceous and brunneous, obovate to obovate-elliptic, rarely oblan- ceolate, 4—7 cm. long, 1.5—2.5 cm. wide, cuneate, rarely obtuse at the base, abruptly short-acuminate, the acumen 5 mm. long, entire, the midrib plane above, evident beneath, the lateral nerves 6—9 pairs, obscure, rarely evident on both surfaces. Inflorescence fasciculate, the flowers 4-merous, the in- dividual branches of the fascicles uniflorous. Staminate inflorescence: pedicels 4—5 mm. long, glabrous, bracts ovate, glabrous, ciliate, the prophylla basal, ciliate; calyx patelliform, 4 mm. across, the lobes ovate, 1.75 mm. au obtuse and ciliate; corolla rotate, 10-11 mm. across, the petals oblong, 5 mm. long, 2 mm. wide, connate at the base; stamens shorter than the petals, the anthers oblong: rudimentary ovary globose, slightly depressed, often 4-lobed at the apex. Pistillate inflorescence: pedicels 6-8 mm. long, bracts ovate, the prophylla 2, basal, ciliate; calyx patelliform, 3 mm. across, the lobes rounded; corolla erect, the petals obovate, 5 mm. long, 5 mm. wide, the apex rounded, ciliate; staminodes shorter than the petals, the sterile anthers minute, cordate; ovary ovoid, 4 mm. long, 3.5 mm. wide, the apex flat with discoid stigma. Fruit large, globose or oblong-ellipsoid, 9-19 mm. long, 10 mm. in diameter, the exocarp thick, the stigma discoid. Pyrenes rugose, striate and sulcate, the back oblong in outline, 7-10 mm. long, 4 mm. wide, slightly depressed along the middle, the endocarp stony. LIUKIU ISLANDS: Okinawa, Kunigami, E. H. Walker, S. Sonohara, S. Tawada & T. Amano 7005; same locality, S. Sonohara, S. Tawada & T. Amano 6309. Nakano-shima, Linshoten Isl. Group, S. Kawagoe, Aug. 20, 1912 JAPAN: Hondo: Tokyo, E. Elliott 7; M. Mizushima 1088. Sagami, M. & U. Mizushima 911; K. Miyabe, April 1882; K. Sakurai, April 12, 1906. Mino, K. Shiota 70, 6542. Yokohama, Maximowics in 1862 (G); E. H. Wilson 6415. Kamakura, E. H. Wilson 6608, 6609. Nagasaki, Oldham. Without precise locality, Herb. Thunberg, fragment of type material; Siebold ex Herb. Lugduno- Batavo 101570 (A, G); Zollinger. Kyushu: Tanega-shima, E. H. Wilson 6134. Kago-shima, E. H. Wilson 6167. KOREA: Quelpaert, U. Faurie 1638, 1639, 1647; T. Taquet 622, 623, 2718, 2719, 4146; E. H. Wilson 9513. Oo-ryong-too, E. H. Wilson 8540. Port Hamil- ton, C. Wilford in 1859 (G). 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 153 I have also examined material cultivated in Japan, the Royal Botanic Gardens at Kew, the Golden Gate Park, San Francisco, and in gardens in Augusta, Ga., and Federal Point, Fla. In Japan the tree is widely culti- vated for ornamental purposes. The young leaves are eaten as greens. The bark is ground to obtain a sticky substance which is used for bird-lime. Loesener recorded this species from China, as represented by Fortune 57. Among all the Chinese material of //ex that I have examined, there is no element that is close to this species, and it appears that any occurrence of it in China is not spontaneous. There exists a great variation in leaf-shape and pedicel-length. The pedicels of all the Liukiu material that I have examined are on the short end of the normal curve. They are 6-7 mm. long. Among the Japanese and Korean elements there is a long-pedicellate form. The fruiting pedicels are as long as 10-15 mm. In such cases the prophylla are inserted on the pedicels, about one third of the way from the base. 15. Ilex brachypoda sp. nov. Frutex sempervirens, usque 3 m. altus, ramulis glaberrimis, bienniis 4 mm. diametro, teretis, rugosis, lenticellis obsoletis; foliis coriaceis, integer- rimis, obovatis, vel oblongo-obovatis, 3-8 cm. longis, 1.5—4.5 cm. latis, basi acutis vel cuneatis, apice obtusis vel breviter acuminatis, costa supra plana, subtus prominula, nervis lateralibus utrinque 5 vel 6, obscuris; in- florescentiis paucifasciculatis, floribus 4-meris; ¢ ignotis; @ unifloris, pedicellis 2-3 mm. longis, prophyllis basalibus, ciliatis; calycibus patelli- formibus, 4 mm. diametro, lobis ciliatis; corolla erecto-patenti, petalis liberis, ovatis, 4 mm. longis, 3 mm. latis; staminodiis petalis brevioribus, antheris parvis, cordiformibus, ovario magno, ovoideo, 4-5 mm. longo, 3- 4 mm. diametro, stigmate discoideo. Fructus globosus, 10 mm. diametro, stigmate umbilicato; pyrenis 4, rugosis et lapidosis, dorso 8 mm. longis et 4 mm. latis, leviter depressis. LIUKIU ISLANDS: Okinawa, Genka, Mt. Kunchon, £. H. Wilson 8118. Oosima, C. Wright, US 15798 (US, TYPE). According to Wilson this is a common shrub in Okinawa. Its white flowers appear in early March. It is closely related to Jlex integra Thunb.. from which it can be distinguished at once by its subsessile fruits. 16. Ilex bioritsensis Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 53. 1911. — Sasaki, Cat. Gov. Herb. 316. 1930. — Suzuki in Masamune, Short Fl. Form. 124. 1936. —S. Y. Hu in Jour. Arnold Arb. 30: 366. 1949. This species is characterized by its small spinose leaves and paired fruits, each with two pyrenes. It was first recorded from Bioritsu [Byoritu], on Mt. Tokuzyo-taizan. Additional material has been collected from Tonbara of Nanto in Taichu prefecture and Mt. Noko of Taito prefecture. It is 154 JOURNAL OF THE ARNOLD ARBORETUM _[VoL. XxxIv now also known from central and western China in mixed forests at alti- tudes of 1700-3960 meters. In Taiwan it grows to be a small tree up to seven meters high. As a result of repeated cutting for firewood, it is often seen as a shrub about two meters high. Its compact shiny evergreen foliage and its brilliant red fruits make it a good garden specimen. 17. Ilex dimorphophylla Koidzumi, Pl. Nov. Amami-Ohsim. 12. 1928. An evergreen shrub, the old twigs light gray, slightly rugose, the current year’s growth pilose. Leaves occurring also on the third year’s growth, 4— 12 mm. apart, the petioles 0.5-3 mm. long, canaliculate and puberulous above; lamina coriaceous, olivaceous, glabrous, entire or on young shoots 3- to 6-sinuate-dentate on each side, spiny, obovate, obovate-elliptic or broad-elliptic, 1-3.5 cm. long, 7-17 mm. wide, obtuse or rounded at the base, mucronulate-acute to rounded at the apex, the midrib plane above, slightly elevated beneath, the lateral nerves 4 or 5 pairs, obscure on both surfaces. Flowers not known. Infructescence fasciculate, axillary on the second year’s growth; pedicels 5 mm. long, puberulous, the prophylla 2, basal, ovate, ciliate; young fruit globose, 3 mm. in diameter, 4-loculate, the stigma 4-lobed discoid, the persistent calyx 2.5 mm. across, 4-lobed, the lobes ovate, rounded at the apex, ciliate. Pyrenes not known. LIUKIU ISLANDS: Amami-Oshima: Mt. Yuwandake, S. Tashiro in March 1924 (Herb. Univ. Kyoto, Lectotype, photograph and fragment in A); same locality, G. Koidzumi in April 1923 (Herb. Univ. Kyoto, paratype; photograph and fragment in A). When Koidzumi published this species, he cited three specimens (Z. Tashiro in 1924, H. Ohba in 1925, and G. Koidzumi in 1923), not desig- nating the type. Tashiro’s collection represents a four-year-old shoot. The leaves are elliptic, all spiny, and with extremely short petioles. Its ap- pearance reminds one of the Taiwan species /lex bioritsensis Hayata. Koidzumi’s collection is a fruiting specimen. Its leaves are all entire and the petioles are 3 mm. long. It resembles a Hongkong species, //ex champi- onii Loes. As leaf-dimorphism is a common feature in many species of /lex (as in the case of J/ex cornuta Lindl., where both spiny and entire leaves occur on the same plant), I think that Koidzumi was probably justified in maintaining these heterogeneous elements with entirely different aspects as conspecific. Regarding the typification of this species Dr. S. Kitamura of the Botanical Institute of Kyoto wrote, “I selected the specimens collected by S. Tashiro as the lectotypus. This specimen is sterile. There is Koid- zumi’s handwriting . . . greater part of his original description coincides well with this specimen. I selected the specimen collected by G. Koidzumi as paratypus.” This species appears to be close to /lex goshiensis Hayata, which has abruptly short-acuminate, retuse, entire leaves 2.8-4.8 cm. long and 1.5— 2.5 cm. wide. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 155 18. Ilex suzukii S. Y. Hu in Jour. Arnold Arb. 30: 376. 1949. TAIWAN: Mt. Taiheizan, S. Suzuki, Aug. 7, 1928 (US). This species is endemic to Mt. Taiheizan, Taihoku, Taiwan. It has a superficial resemblance to J/ex goshiensis Hayata because both species have small entire leaves. But the nerves on the leaves of Hayata’s species are obscure beneath, the fruits have thick discoid stigmata, and the endocarps of the pyrenes are coriaceous, while the leaves of this species are prominently reticulate beneath, the stigmata of the fruits are navel-like, and the endo- carps of the pyrenes are woody. According to the characters presented in the fruit and pyrenes, this species should be placed in Series Denticulatae S. Y. Hu of the Section Aquifolium Gray. But all the other species in this series have denticulate leaves, while this one has entire ones. More ma- terial is awaited for the clarification of its taxonomic position. 19. Ilex formosana Maxim in Mém. Acad. Sci. St. Pétersb. VII 29 (3): 46. 1881.— Henry in Trans. As. Soc. Jap. 24 (Suppl.): 27 (List PI. Form. 27). 1896.— Matsum. & Hayata, Enum. Pl. Form. 81. 1906. — Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 54 (Mat. Fl. Form. 54). 1911, et in Ic. Pl. Form. 1: 131. 1911. — Sasaki, Cat. Gov. Herb. 316. 1930. — Suzuki in Masamune, Short Fl. Form. 125. 1936.—S. Y. Hu in Jour. Arnold Arb. 31: 68. 1950. This species was based on material collected by Oldham in Taiwan. It has been considered endemic to that island by some botanists, among them Kanehira. The type material has elliptic-lanceolate leaves with obscure reticulation on the lower surface and an acuminate apex, and small fruits on very short (2-3 mm. long) pedicels. As our knowledge of the flora of mainland China increases, we know that a species so characterized is wide- spread in the warmer regions of China. In Taiwan it occurs in forests at altitudes of 600-700 meters, where it grows to a tree up to twelve meters high. The red fruits remain on the tree throughout the winter, from Decem- ber to March. 20. Ilex ficoidea Hemsl. in Jour. Linn. Soc. Bot. 23: 116. 1886. —S. Y. Hu in Jour. Arnold Arb. 31: 72. 1950 Ilex buergeri Miq. var. glabra Loes. in Nov. Act. Acad. Caes. Leop.-Carol. t. Cur. 89: 286 (Monog. Aquif. 2: 286). 1908. Ilex glomeratiflora Hayata, Ic. Pl. Form. 3: 53. 1911; 5: 15, fig. 6. 1915, et 6 (Suppl.): 13 (Gen. Ind. Fl. Form. 13). 1917.— Yamamoto, Suppl. Ic. Pl. Form. 1: 32. 1925, pro parte. — Sasaki, Cat. Gov. Herb. 316. 1930. — Suzuki in Masamune Short Fl. Form, 125. 1936. Ilex arisanensis Yamamoto Suppl. Ic. Pl. Form. 1: 30, fig. 10. 1925. — Cat. Gov. Herb. 316. 1930. — Kanehira, Form. Trees, rev. ed. 370. Suzuki in Masamune, Short Fl. Form. 124. 1936. Ilex warburgit sensu Yamamoto, Suppl. Ic. Pl. Form. 1: 40. 1925, non Loes., 1901. 156 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL. xxxIv Ilex buergeri Migq. var. glabra Loes. was based on Faurie 39 from Taiwan. In the herbarium of the Arnold Arboretum there are two sheets of this number. Their glabrous stems, caudate leaves, and very short fruiting pedicels are characteristic of J/ex ficoidea Hemsl. In making this variety a synonym of /lex warburgii, Yamamoto misinterpreted the latter species. The type of Jlex glomeratiflora Hayata represents a staminate plant of Ilex ficoidea Hemsl., which has very short pedicels, 1.5 mm. long. Yamamoto wrongly identified the long pedicellate fruiting material of /lex uraiensis Yamamoto as the female plant of /. glomeratiflora, and at the same time determined U. Faurie 186 as Ilex arisanensis. This last taxon represents a plant with short fruiting pedicels. Both Sasaki’s and Faurie’s collections were from Arisan. They represent the staminate and pistillate plants of the same species, /lex ficoidea Hemsl. 21. Ilex liukiuensis Loes. in Nov. Act. Acad. Caes. Leop.-Carol. Nat. Cur. 78: 336 (Monog. Aquif. 1: 336). 1901.— Sasaki, Cat. Gov. Herb. 317. 1930. — Masamune & Suzuki in Ann. Rep. Taihoku Bot. Gard. 3: 61. 1933. — Sonohara et al., Fl. Okinawa 92. 1952. Ilex mertensii sensu Ito & Matsum. in Jour. Coll. Sci. Univ. Tokyo 12: 369 (Tent. Flor. Lutch. 369). 1900.— sensu Sonohara et al., Fl. Okinawa 92. 1952, non Maxim., 1888. Evergreen trees up to 8 m. high, entirely glabrous; branchlets of the third year’s growth 3 mm. in diameter, rugose with large, slightly elevated leaf- scars and scars of the inflorescences, the lenticels lacking, the current year’s growth 1.7 mm. in diameter, striate. Leaves occurring also on the third year’s growth, 3-10 mm. apart, the stipules minute, often obsolete; petioles 10— 16 mm. long, up to one third the length of the lamina, narrowly canalicu- late above; lamina coriaceous, brunneous, rarely olivaceo-brunneous, obo- vate, or oblong-elliptic, those near the apex sometimes elliptic, 3—7.5 cm. long, 1.8—3.5 cm. wide, acute at the base, obtuse, rounded, retuse or broadly short-acuminate at the apex, the margin recurved and obviously thickened, subentire or remotely crenulate, the midrib plane above, elevated beneath, the lateral nerves 5 or 6, obscure above, prominent and noticeably reticulate beneath. Staminate inflorescence unknown. Pistillate inflorescence fascicu- late, 2—4-flowered, the pedicels 9-14 mm. long, each with two prophylla inserted 1-2 mm. above the base; calyx patelliform, 2 mm. across, 4- or 5-lobed, the lobes rounded, ciliate; corolla rotate, 8 mm. across, the petals more or less free, oblong, 3.5 mm. long, ciliate at the apex; staminodes half the length of the petals, the anthers sieges ovary ovoid, 1.5 mm long, the stigma discoid. Fruit globose, 6 mm. in diameter, the persistent calyx explanate, the stigma navel-like, 4-lobed. ue 4, the back oblong in outline, 4.5 mm. long, 3 mm. wide, the middle longitudinally depressed, rugose and obliquely striate, the endocarp woody. LIUKIU ISLANDS: Iriomotto [Iriomote], Warburg in 1887 (fragment of the Type, A); same island, between Shira-hama and Sonai, E. H. Walker & S. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 157 Tawada 6528. Ishigaki Island, E. H. Walker & S. Tawada 727. Okinawa, Nago- Dake, T. Kanashiro 1697 (US), 1812. I have also seen some sterile material from Yokohama Nursery. The woody, rugose, dorsally impressed pyrenes, the relatively long- petiolate, obovate or oblong-elliptic leaves, the fasciculate fruit, and the absence of lenticels on the third year’s growth of lex liukiuensis all suggest a close relation with J. graciliflora Champ. of Hongkong. Besides the geo- graphical separation, the latter species differs in having sparsely puberulous branchlets, pedicels 4—6 mm. long, and thickly coriaceous leaves In Liukiu the plant occurs on forest ridges. The fruit begins to turn red in middle August. 22. Ilex uraiensis Yamamoto in Jour. Soc. Trop. Agr. Taiwan 4: 486. 1932. — Suzuki in Masamune, Short Fl. Form. 126. 1936.—S. Y. Hu in Jour. Arnold Arb. 30: 382. 1949. Ilex mutchagara sensu Sasaki, Cat. Gov. Herb. 318. 1930.— sensu Kanehira, Form. Trees, rev. ed. 378, fig. 335. 1936, non Makino, 1913. Ilex uraiana Hayata in sched. ex Kanehira, I.c., in syn. Ilex kelungensis sensu Kanehira & Hatusima in Trans. Nat. Hist. Soc. Form. 29: 156. 1939, non Loes. 1901. TAIWAN: without precise locality, W. R. Price 262 (fragment). Ilex uraiensis Yamamoto was based on material collected from Uraisha and Sozan of northern Taiwan, with the type from Sozan. This material was interpreted as /lex mutchagara by Sasaki, as indicated by his citations, in 1930, and by Kanehira in 1936. As Kanehira was the Director of the Department of Forestry, Government Research Institute, and Sasaki the Curator of the Herbarium of the same department, it was natural for them to base their misinterpretation on the same material of /. mutchagara Makino. Sasaki cited no synonyms. Kanehira overlooked Yamamoto’s publication and credited the binomial to Hayata as /lex uraiana Hayata in sched. Both Kanehira and Sasaki were mistaken in maintaining that the Uraisha and the Sozan specimens were conspecific with the Liukiu /. mutchagara Makino. Kanehira’s illustration, based on northern Taiwan material, was certainly not Makino’s Liukiu species. In the Arnold Arbore- tum we have several collections with both flowering and fruiting specimens from Sozan, the type locality of /lex uraiensis Yamamoto. Wilson 10288, a staminate plant in full bloom, matches Kanehira’s figure in every respect. The leaves of this specimen appear similar in size and shape to those of J. mutchagara Makino. But the lower surface of Makino’s species is “‘brown- ish-puncticulate” and that of the Sozan material is not. Moreover, the endocarp of the pyrenes of the Liukiu species is ‘‘smooth, thin, coriaceous,’ while that of the Sozan species is “irregularly striate, wrinkl ed and stony.” The punctate leaves, black fruits, and s mooth coriaceous ha carps of Ilex mutchagara Makino place it in the section Paltoria, while the crenulate-serrate leaves, the fasciculate red fruits and the irregularly striate, 158 JOURNAL OF THE ARNOLD ARBORETUM _[VoL. xxxIv wrinkled and stony endocarps of Jlex uraiensis Yamamoto place it in the Denticulatae series of the section Aquifolium. The resemblance in size and shape of the leaves in these two species is very superficial. Again, in 1939, Kanehira and Hatusima interpreted J/ex uraiensis Yama- moto as conspecific with Jlex kelungensis Loes. The latter species was based on Warburg’s collection made in November 1895 at Kelung, a port in the northeastern part of Taihoku. Its elliptic-lanceolate leaves with obscure reticulate veinlets on the lower leaf-surface, and the small fruits on very short pedicels (2—3 mm. long), are characteristic of /lex formosana Maxim., to which it belongs. Jlex uraiensis Yamamoto, on the other hand, is characterized by its prominently reticulated leaves and larger fruits on pedicels 6-8 mm. long. The size and shape of the smaller leaves of this species appear to resemble those of the isotype of Jlex kelungensis Loes., which I interpret to be conspecific with J. formosana Maxim. It was prob- ably on account of this superficial resemblance that Kanehira and Hatu- sima made J. uraiensis Yamamoto a synonym of J. kelungensis Loes. 22a. Ilex uraiensis Yamamoto var. formosae (Loes.) S. Y. Hu in Jour. Arnold Arb. 30: 383. 1949. This variety differs from the typical //ex uraiensis Yamamoto in its smaller obovate leaves with rounded or obtuse apex. 22b. Ilex uraiensis Yamamoto var. macrophylla var. nov. Frutex, ramulis glabris; foliis coriaceis, ellipticis, raro obovato-ellipticis, 7-10.5 cm. longis, 3.5 cm. latis, basi acutis, apice brevi-acuminatis, acumine 5-10 mm. longis; fructibus fasciculatis, pediceltis 6-8 mm. longis. TAIWAN: Kiirun, 7. Tanaka & Y. Shimada 17797 (TYPE). This large-leaved variety differs from the typical //ex uraiensis Yama- moto in which the elliptic or obovate-elliptic leaves measure only 3-7 cm. long and 1.5—2.5 cm. wide. 23. Ww Ilex warburgii Loes. in Nov. Act. Acad. Caes. Leop.-Carol. Nat. Cur. 78: 326 (Monog. Aquifol. 1: 326). 1901.— Yamamoto, Suppl. Ic. Pl. Form. 1: 40. 1925.— Suzuki in Masamune, Short Fl. Form. 126. 1936. A small evergreen tree up to 4 m. high; branchlets glabrous, the third year’s growth 34.5 mm. in diameter, the lenticels obscure, the leaf-scars suborbicular, the current year’s growth 1.5—2 mm. in diameter, subterete, the terminal buds conic, pubescent. Leaves 1—2 cm. apart, occurring also on the second year’s growth, the stipules obscure, the petioles 8-15 mm. long, up to one fifth the length of the lamina, deeply canaliculate, minutely pu- berulent or glabrescent above; lamina coriaceous, olivaceous, oblong-elliptic, rarely broad elliptic, 4-8.5 cm. long, 2.4 cm. wide, obtuse or rounded at the base, abruptly caudate at the apex, the acumen 8-15 mm. long, 2-3 mm wide, the width rather uniform, remotely crenulate-serrate, the midrib 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 159 impressed above, glabrous or very minutely puberulent, elevated beneath, the lateral nerves 7 to 9 pairs, obscure above, evident beneath, reticulate near the margin. Flowers not known. Infructescence fasciculate or subrace- mose, the central axis up to 6 mm. long, the pedicels of individual fruits 4—8 mm. long, the prophylla basal or medium, the fruit globose, 5-6 mm. in diameter, the persistent calyx explanate, 2 mm. across, the lobes ciliate, the stigma navel-like. Pyrenes 4, short and robust, rugose, striate and sul- cate, the back ovate in outline, 4 mm. long, 3 mm. wide, the endocarp woody. TAIWAN: Taihoku, Shirin, Taihoku Univ. Herb. no. 11132; Shirin to Sozan, E. H. Wilson 10298 LIUKIU ISLANDS: Ischigaki ieee pales \ ae of TYPE). Yaeyama Gunto, Iriomote, E. H. Walker & S. Tawada This species is closely related to Jlex ficoidea mie which is charac- terized by its very short (2-3 mm. long) fruiting pedicels. 24. a Ilex pubescens Hook. & Arn., Bot. Beechey Voy. 167, pl. 35. 1833. — Yamamoto, Suppl. Ic. Pl. Form. 1: 39. 1925, — Sasaki, Cat. Gov. Herb. 318. 1930.—S. Suzuki in Ann. Rep. Taihoku Bot. Gard. 1: 154. 1931, et in Masamune, Short FI. Form. 125. 1936.—S. Y. Hu in Jour. Arnold Arb. 31: 220. 1950. TAIWAN: between Goshyo & Suizya, S. Suzuki 5799; Taihokuzyu, Bunzan- gun, Suzuki-Tokio 8822; same locality, S. Suzuki on April 30, 1933. Baran, Kudo-Sasaki 15126. This species was first published on the basis of material collected from Hongkong. It has a wide range of distribution on the mainland of China, from Chekiang and Anhwei southward to Kwangtung and Kwangsi. In Taiwan it has been reported from Sozan, Heirinbi, Taihei, and Urai in the northern part of the island, from Nanto and Lake Candidius in the central part of the island, and from Kosyun and South Cape at the southern ex- tremity of the island. It is a common shrub in thickets along the roadside at altitudes 600-750 meters. Its pinkish flowers appear in late March and its scarlet fruit lasts on the bush all through the winter. 25. Ilex cochinchinensis (Lour.) Loes. in Nov. Act. Acad. Caes. Leop.- Carol. Nat. Cur. 78: 230 (Monog. Aquif. 1: 230). 1901.—S. Y. Hu in Jour. Arnold Arb. 31: 239. 1950. Ilex ee Loes. op. cit. 359. 1901. — Sasaki, Cat. Gov. Herb. 316. 1930. in Masamune, Short Fl. Form. 124. 1936. In Taiwan this species occurs only at the southern extremity of the is- land. It is found also in Hainan and northern Indo-China. Its large entire punctate leaves and its fasciculate fruits are very characteristic. In com- menting on Jlex ardisioides Loes., Hayata wrote in 1911, “a species im- perfectly known to me.” Two years later he published /lex cleyeroides, which is a synonym of Jlex cochinchinensis (Lour.) Loes. 160 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiIv 26. Ilex goshiensis Hayata in Jour. Coll. Sci. er Ri 30: 54 her Fl. Form. 54). 1911, et Ic. Pl. Form. 1: 1.—S. Y. in Jour. Arnold Arb. 31: 248. 1950. Ilex hanceana sensu Ito & Matsum. in Jour. Coll. Sci. Univ. Tokyo 12: 367 (Tent. Fl. Lutch. 367). 1900.— sensu Hayata, Ic. Pl. Form. 6 (Suppl.): 13 (Gen. Ind. Fl. Form. 13). 1917.—sensu Sasaki, Cat. Gov. Herb. 317. 1930.— sensu Sonohara et al., Fl. Okinawa 92. 1952, non Maxim., 1881. LIUKIU ISLANDS: Okinawa, T. Miyagi (fragment); Kunigami, Tanyu-dak, S. Sonohara, S. Tawada & T. Amamo 6308; Genka Mt., E. H. Wilson 8091 (A, US) JAPAN: Kyushu: Hingo, Koonose, K. Mayebara 320 (fragment); Mt. Kir- shima, Z. Tashiro, May 8, 1913 (sterile). This species has been misinterpreted as //lex hanceana Maxim. by many botanists. Through Dr. H. Hara of the Botanical Institute, University of Tokyo, I have obtained fragments of the type material for comparison, accompanied by an excellent photograph. The elongated pedicels, the fasciculate umbelliform inflorescences, with the individual branches 3- to 7-flowered, and the globose fruit with slightly elevated discoid stigmata, are distinct features of this species. It is very different from //ex hanceana Maxim., which has paired fruits on very short (1.5 mm. long) pedicels. In Taiwan and Liukiu this species appears to be a small tree up to 6 meters high. It is common at altitudes of 100-600 meters. Its fruit turns red in August (fide Gressitt) and remains dull red until March (fide Wilson). 27. Ilex hayataiana Loes. in Fedde Rep. Spec. Nov. 55: 333. 1941. — S. Y. Hu in Jour. Arnold Arb. 31: 249. 1950. LIUKIU ISLANDS: Okinawa: Yona, J. W. Conover 1139, 1840 (US); Kunigami, Nago-dake, FE. H. Walker, S. Sonohara, S. T. Tawada & T. Amano 6192. Irumuti (Iriomote), 3, Murayamei in 1927. This species differs from /lex goshiensis Hayata in having narrower leaves with acuminate or caudate apex and fasciculate uniflorous individual branches in the inflorescences. It has been reported to be abundant at altitudes of 230-330 meters, where it grows as a small tree along the banks of streams in the shade of taller trees or steep walls of the valleys. DOUBTFUL AND EXCLUDED SPECIES ILEX CINEREA sensu Ito & Matsumura in Jour. Coll. Sci. Univ. Tokyo 12: 369 (Tent. Fl. Lutch. 369). 1900. —Sonohara et al., Fl. Okin. 92. 1952, non Champ., 1852. Ilex cinerea Champ. ex Benth. is endemic to Hongkong. Its large oblanceolate leaves with very short (2-4 mm. long) petioles and its very short (2-3 mm. long) pedicellate fruits are very characteristic. Many East Asian specimens belonging to //ex ficoidea Hemsl. or J. formosana Maxim. 1953] HU, ILEX IN TAIWAN AND THE LIUKIU ISLANDS 161 have been erroneously named /lex cinerea Champ. It is quite likely that authors dealing with the flora of the Liukius may have made similar errors, since I have no evidence, from the material which I have examined, to lead me to believe that the species has ever been collected outside Hongkong. ILEX MATANOANA sensu Sasaki, Cat. Gov. Herb. 317. 1930. — sensu Masa- mune & Suzuki in Ann. Rep. Taih. Bot. Gard. 3: 61. 1933, non Makino, 1913. Ilex matanoana Makino was first published from the Bonin Islands. Its small coriaceous obovate leaves with retuse apex and prominently reticulate nerves are very characteristic. Judging from the material I have studied, the species of //ex in the Bonin Islands are very different from those of Taiwan and the Liukiu Islands. In fact, there is no Bonin Island element of //ex found in the flora of Taiwan and Liukiu. Sasaki ascribed specimens collected from Sozan, Giran, and Kosyun to this species. Masamune & Suzuki recorded it as occurring on Kizan, a small volcanic island northeast of Taiwan. They cited no specimens. As these collections are not available for my examination, I can only question their interpretation. Ilex seed Hayata in Jour. Coll. Sci. Univ. Tokyo 30: 56 (Mat. FI. Form. 56). 1911, et in Ic. Pl. Form. 1: 133. 1911. — Kanehira, Form. Trees, Ist ed. 125. 1917 = Symplocos nokoensis (Hayata) Kane- hira, Anat. Charact. Ident. Form. Woods 151. 1921, et Form. Trees, rev. ed. 596, fig. 553. 1936. ILEX EURYAEFOLIA Mori & Yamamoto in Jour. Soc. Trop. Agr. Form. 4: 485. 1932. This species was based on T. Suzuki 7332, collected from Mt. Taihei in the northeastern part of Taiwan. According to the description of the au- thors, the plant has ‘‘Folia coriacea . . . longe eliptica vel oblongo-oblan- ceolata vel subspathulata ... apice [obtusa], ... margine ad totam crenato-serrata sed raro inferne integra . . . petiolis brevibus 3-5 mm longis . . . Inflorescentia fructifera axillaris, umbellaeforme fasciculata . Fructus globosi, 4 mm. in diametro, apice stigmate rostrati .. . evens 4.” ILEX RARASANENSIS Sasaki in Trans. Nat. Hist. Soc. Form. 21: 154. 1931. This species was based on Sasaki’s own collection from Mr. Rarasan of Taihoku prefecture. According to his description the plant has “Leaves sempervirens, alternate and fascicules near the top of the branches. . . elliptic or ovate, entire or loose serrate . . . acuminate or caudate . . Inflorescens [sic] umbellate panicles, terminal axilis [sic] of the branches, peduncles umbell [sic] ... Drupe ...6 mm. in diameter, shining chestnut colour when ripe. Seed [sic] 6 . .. 3-4 mm. long, 2 mm. in diameter.” 162 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxIv Kanehira in 1936, without giving any reason, excluded /. rarasanensis Sasaki and J. euryaefolia Mori & Yamamoto from the Aquifoliaceae in his Formosan Trees. In order to ascertain the status of these taxa, I wrote to Professor W. F. Chu, Head of the Department of Forestry, National Taiwan University, asking him for the photographs and fragments of the type material of these species. He very kindly searched in the Herbarium of the Department of Botany of the same University and also in the Government Herbarium of the Department of Forestry, Government Re- search Institute, of which Sasaki was curator. Unfortunately no traces of these species could be found. For the present they have to be classified among the doubtful species. ARNOLD ARBORETUM, HARVARD UNIVERSITY. 1953] GRAY, REVISION OF PODOCARPUS 163 A TAXONOMIC REVISION OF PODOCARPUS VIII. THE AFRICAN SPECIES OF SECTION EUPODOCARPUS, SUBSECTIONS A AND E Netta E. Gray * Tue sEcTION Eupodocarpus of the genus Podocarpus is well represented on African mountain slopes from the southernmost coast to well north of the equator in both the eastern and western regions, wherever relatively cool and wet climates are found. In many regions the species are found in pure stands, often of magnificent timber which has been very important economically. In other parts the members are less abundant, interspersed among other trees, but still constituting an item in the lumber industry. The five species on the continent (all in subsection A) are Podocarpus elongatus (Ait.) L’Hérit., P. Henkelii Stapf, P. latifolius (Thunb.) R. Br. and P. milanjianus Rendle, which have been compared by Stapf (10) and Chalk et al. (3), and P. ulugurensis Pilger, which was subsequently de- scribed (7). Podocarpus madagascariensis Baker, endemic on the island of Madagascar, is also in subsection A and has been treated separately by Laurent (4). Podocarpus rostratus Laurent, also confined to the island of Madagascar, is in the new subsection E. Our studies show that P. rostratus has close affinities with section Eupodocarpus, but we recommend placing it in a separate monospecific subsection. Details leading to this decision are to be found under the discussion of the species. The leaf anatomy of the group consisting of P. elongatus, P. latifolius, P. Henkelii, P. milanjianus and P. madagascariensis is quite similar, dis- playing only very minor variations. It always shows two marginal resin canals, hypoderm between the stomatal rows on the lower side, the Florin ring in the stomatal subsidiary cells, and toothed or pitted walls in the epidermal cells. In addition, we find that P. elongatus can usually be distinguished from the others of the group by the thicker palisade parenchyma of more than one layer and no auxiliary sclereids in the mesophyll. In most specimens there is only one vascular resin canal. In most leaves the upper hypoderm is occasionally interrupted and it has been found that these interruptions * The author wishes to express her greatest appreciation to the late Professor John T. Buchholz for assistance in the preparation of this paper. She wishes to acknowledge the assistance which was given to him, with permission to examine and photograph specimens of section Eupodocarpus preserved in the herbaria of th ew Botanic Gardens, the British Museum and Ca ge University. Also she especially thanks Prof. J. Leandri of the Paris Museum Herbarium and Prof. Carl G. Alm of the Uppsala Botanical Museum for locating certain critical specimens. A special grant from the Emory University Research Committee aided the author in the completion of this part of the problem. 164 JOURNAL OF THE ARNOLD ARBORETUM _[VvoL. XxxIVv are accompanied by one to six short rows of stomata in the upper epidermis. Thus, P. elongatus is differentially amphistomatic, a unique character in the section Eupodocarpus. Vascular sclereids or fibers in P. elongatus are rare, and the stomatal rows seem closer together than in P. latifolius and P. Henkelii. Hypodermal fibers average 12-18 » in diameter, somewhat smaller than those in the other species. In some specimens of P. Jatifolius and P. Henkelii the leaf characters seem to overlap those of P. elongatus, but there are generally three vascular resin canals, continuous upper hypoderm, larger hypodermal fibers (15— 23 » average diameter), stomatal rows somewhat farther apart, palisade parenchyma only one cell-layer thick, auxiliary sclereids in the mesophyll, and no stomata on the upper side. Podocarpus milanjianus, usually having the margins of the leaves revo- lute, often has the upper hypoderm interrupted at the point of the turn. There is often much increased sclerification, this appearing in auxiliary sclereids in the mesophyll, heavier walls in the accessory transfusion tissue and more abundant upper and lower vascular fibers. Podocarpus mada- gascariensis, however, shows the most extreme sclerification, affecting most of the mesophyll. The specimen of P. ulugurensis which we examined shows the striking difference of five instead of three vascular resin canals. The two extra ones appear in or very close to the transfusion tissue. Externally, this is the only species with a definitely sulcate upper midvein. Key to SEcT. Eupopocarpus (AFRICAN SPECIES), SUBSECTIONS A AND E Leaves very small; many marginal resin canals, accessory niger tissue absen Subsection E. P. rostratus. Leaves larger; 2 marginal resin canals, accessory transfusion tissue hie es with 5 vascular resin canals; upper midrib sulcate. .. P. ulugurensis. Leaves with 1-3 vascular resin canals; upper midrib never sulcate Leaves narrowly lanceolate, 3.5—6.5 cm. long, 3.5-4 mm. wide: acute to obtuse at tip. . inks owalbeh@edeeenntaaw as See: Leaves elongate and broader. dult leaves very large, 7-18 cm. long. Male cone very short, 13 mm. long. ves 8-18 cm. long, 8-18 mm. wide; seed 20 mm. OO o2a5 yas teases ees madagascariensis, maa ee 5-7 cm. long, 6 mm. wide; seed 2 . madagascariensis var. sca me Male cone adece 20-45 mm. ..... . Henkelit. Adult leaves smaller, juvenile often as large as ‘above. Leaves gradually attenuate........... _.... P. milanjianus. Leaves short rotundate, angustate at tip. Leaves 5-11 mm. wide. ................ P. latifolius. Leaves 12-17 mm. wide. ...... P. latifolius var. latior. Podocarpus elongatus (Ait.) L’Hérit. in Pers. Syn, 2: 580. 1807; L. C. . Richard, Comm. Bot. Conif. 13. 1826; Endlicher, Syn. Conif. 218. 1847; Carriére, Traité Conif. ed. 2, 671. 1867; in part; Parlatore 1953] GRAY, REVISION OF PODOCARPUS 165 in DC. Prodr. 16 (2): 511. 1868; Van Tieghem in Bull. Soc. Bot. France 38: 169. 1891; Pilger in Pflanzenreich IV. 5 (Heft 18): 89. 1903, in Nat. Pflanzenfam. ed. 2, 13: 247. 1926; Dallimore & Jackson, Handb. Conif. 44. 1923, 1931, 66. 1948; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 279. 1931; Chalk, Burtt-Davy & Desch, For. Trees & Timbers Brit. Emp. 1: 24. 1932; Stapf in Fl. Cap. 5 (Sect. 2, suppl.) : 8. 1933. Podocarpus Thunbergii var. angustifolia Sim fide Dallimore & Jackson, Handb. Conif. 44. 1923, 1931 Taxus elongata Aiton, Hort. Kew. 3: 415. 1789; Thunberg, Prodr. Fl. Cap. 117. 1800. Taxus capensis Lamarck in Encycl. 3: 229. 1789. Podocarpus elongatus is found on the Cape of Good Hope on Table Mountain above the city of Capetown, and hence was one of the plants early brought from that region. Pilger (6) distinguished it from P. lati- folius and P. milanjianus by its narrow lanceolate-linear leaves. Stapf (10) separated it from P. Henkelii by its short straight leaves and fleshy receptacle and from P. latifolius by leaves only up to 4 mm. wide. Chalk (3), on the other hand, distinguished it from P. H enkelii and P. latifolius by its red receptacle (which seems to be in error) and from P. milanjianus by leaves shorter and much narrower (2.5-5 cm. & 2-3 mm.). Of the species studied by Laurent (4), P. elongatus has the shortest, narrowest leaves, shorter male cones than those of P. Jatifolius and P. milanjianus, the shortest peduncle on the female cone, and the smallest seed. According to the descriptions, the twigs of P. elongatus are drooping while those of P. latifolius are erect with leaves stiff. We would distinguish it from P. latifolius by the consistent absence of auxiliary sclereids, the one vascular resin canal (rarely 3), the interrupted upper hypoderm, and the 1—6 upper rows of stomata. The thick leaves show no tendency to be revolute. The differences from P. latifolius seem to be chiefly relative. Many of the sterile specimens referred to P. elongatus which show wider leaves and no stomata in the upper epidermis belong probably to P. latifolius, which has considerable range in leaf size and also a much wider geographi- cal distribution. Both these species may be found in the same area, often being collected by an explorer on the same day. Their close similarity and the existence of a number of specimens intermediate between the two definitely suggest hybridization. Podocarpus elongatus seems to be chiefly a plant of western Cape Province. MacOwan 1958, in both the Gray and Kew Herbaria, has two elements on it, the portion with larger leaves being P. latifolius and that with small leaves being P. elongatus. This latter has both a single vascular resin canal and a few stomata on the upper surface. In all cases it must be borne in mind that for more than fifty years P. elongatus (Ait.) L’Heérit. and P. falcatus R. Br. (in section A frocarpus) were almost hopelessly confused in the literature, and the identity of the 166 JOURNAL OF THE ARNOLD ARBORETUM _[VoL. xxxIv actual material discussed must be checked with greatest care. Podocarpus elongatus is the only species in the section Eupodocarpus with scattered stomata occurring on the upper side of the leaf. This may have been noted by some earlier workers, giving credence to the incorrect determina- tion P. falcatus, a species with leaves always equally amphistomatic. Since Podocarpus elongatus (Ait.) L’Hérit. was used by Persoon as the type species for the genus, a most diligent search was made for the type specimen. An original Thunberg specimen of this species is in the Botani- cal Museum at Uppsala with other plants he collected in and near Cape- town in 1773-1774. It isa male specimen with almost mature pollen cones. A female specimen, with immature ovules, was found in the Paris Museum, collected by Sonnerat (Herb. Jussieu 17135) in 1774 in the company of Thunberg on the occasion of his visit to Capetown and ascent of Table Mountain, the site of the former collection. This specimen was probably collected at a later season than the Thunberg specimen. Confusion arose when it was found that a specimen in the British Museum, indicated as the probable type for Aiton’s Taxus elongatus in his Hortus Kewensis, was really P. macrophyllus var. Maki from Japan. This mistaken identity is reasonable as the male specimens of the two entities are very similar to the naked eye although the leaves of P. elongatus are usually narrower. In this case, the leaf anatomy carries the proof of identity. No doubt there are other good specimens in the British Museum that we have not seen, as Stapf (10) mentions one by Masson who accompanied Thunberg in his original travels. Nor have we seen the specimen used by L. Richard for his drawing of P. elongatus with seeds, in the publication of 1826 (9). We feel that the designation of the Thunberg (male) specimen at Uppsala Botanical Museum and the Sonnerat (female) specimen at the Paris Mu- seum as lectotypes will make this species better understood by later in- vestigators. In genera composed of dioecious species it is desirable to designate both male and female specimens. It is to be hoped that the mis- identified specimen will be removed from a type-folder. DisTRIBUTION: South Africa, in forested areas in southwestern Cape Province primarily. SPECIMENS EXAMINED: SOUTH AFRICA: Cape Province: Clanwilliam Division: Grarsangen Mt., Pillans 8676 (K); * Oliphant’s River, Kradouw Krantz, Pillans * The following symbols indicate the herbaria having the specimens cited: Academy of Natural Sciences of Philadelphia (Ph), Arnold Arboretum (A), British Museum and at Los Angeles (LA), California Academy of Science (CAS), Chicago Natural History Museum (Field Museum) (F), Cornell University (CU), Gray Herbarium (GH), University of Illinois Herbarium (Ill), Royal Botanic Gardens, Kew (K), Missouri Botanical Garden (Mo), New York Botanical Garden (NY), Rutgers Uni- versity Herbarium (NJU), Stanford University Dudley Herbarium (DS), United States National Herbarium (US), Uppsala Botanical Museum (UPS), Yale University Herbarium (YU). 1953] GRAY, REVISION OF PODOCARPUS 167 5297 (K), Pearson 5328 (tK), Wilson in 1922 (A—4 sheets) ; Krakadow Pass, Cedarburg Mts., Wilson in 1922 (+A); Ceres, Munchen 11764 (+NY), Pearson 15551 (LA). Ma Imesbury: Paarl Berg, Prior in 1946 (K), Drege 1839 (K); Stellenbosch, Miller s.n. (K), Garside 1044 (+K); Robertson, Banks of Breede River, For. Dept. Pretoria Herb. 1247 (K), Wilson in 1923 (+A—5 sheets). Swellendam Division: Breede River, Burchell 7480 (K), Schlechter 5652 (K); Table Mtn. sheeted: s.n. ($UPS), Sonnerat ex Herb. Jussieu 17135 (TE). Albany Division: w. of Grahamstown, Zeyher 3448 (K), MacOwan 1958, in part (+GH, +K); Olijantshoek, Zeyher s.n. (+A); Steudel, Eptingee, Ecklon i in 1834 ({NY); Bain’s Kloof, Hutchinson 1008 (+K); Kaffraria, Dohne, Sim 19620 (+A). LOCALITY NOT INDICATED: ex Hook. Herb., Pappe s.n. (K), Anon. in 1906 (A), “Hort. Daudin” in 1851 (+DS), Prager sw (+CAS), Ecklon & Zeyher s.n. (+Mo), Marloth 11488 (+A), “EM” 81 ({M CULTIVATED: Algeria For. Sta., Wilson in 1922 (A), Piquetberg, Bosch Kloof, Compton 9498 (fIll). US., Mass., Arnold Arboretum 482.29, Judd in 1930 (A). Podocarpus latifolius (Thunb.) R. Br. ex Mirb. in Mém. Mus. Nat. Hist. Paris 13: 75. 1825; Berthold, Vergleichenden Anatomie der Coniferen-Blattern 12, 16. 1875; Zimmerman in Flora 63: 2. 1880; Pilger in Paaeengeich VS (Heft 18): 90. 1903; in Nat. Pflanzen- fam., ed. 2, 13: 247. 1926; Florin in Svenska Vet.-Akad. Hand. ser. 3, 10: 279. 1931; Stapf in Fl. Cap. 5 (Sect. 2, suppl.) : 5. 1933; Dalli- more & Jackson, Handb. Conif. 48. 1923, 1931, 72. 1948; Chalk, Burtt-Davy & Desch, For. Trees & Timbers Brit. Emp. 1: 23. 1932; Chalk et al., For. Trees & Timbers Brit. Emp. 3: 81. 1935; Adamson in Fl. Cap. Penin. 31. 1950. Podocarpus Thunbergit Hook. in London Jour. Bot. 1: 657. 1842; Endlicher, S onif. 217. 1847; Carriére, Traité Conif. ed. 2, 710. 1867; Mahlert in Bot. Centralbl, 24: 281. 1885; Parlatore in DC. Prodr. 16 (2): 511. 1868. Podocarpus Thunbergii var. latifolia Sim fide Dallimore & Jackson, Handb. Conif. 48. 1923, 1931; Van Tieghem in Bull. Soc. Bot. France 38: 169. 1891. Podocarpus Sweetii C. Pres] in Bot. Bemerk. 110. 1844. Nageia latifolia O. Kuntze in Rev. Gen. 2: 800. 1891 (not Gordon). Taxus latifolia Thunb. Prodr. Fl. Cap. 117. 1794-1800 et Fl. Cap. (ed. Schultes) 547. 1823. This species is very large, often found very abundantly in the south- eastern part of Cape Province, and important economically as a lumber tree. Its foliage differs from P. Henkelii in the shorter and smaller rigid leaves on erect twigs and fleshy receptacle, and from P. elongatus by the wider a with three vascular resin canals and no stomata on the upper s Sera latifolius shows wide variation in its foliage as would be expected in a species of such wide distribution. It seems to be midway between P. elongatus and P. milanjianus, with many sterile specimens diffi- + This symbol preceding the abbreviated name of an herbarium following the speci- mens examined signifies that the details of the leaves of this specimen have been ex- amined in cross-section. 168 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxIv cult to classify because of extreme characters. The difference between juvenile and adult foliage, which has already been noted in a number of species, adds to the range of characters involved. A detailed description may be found in Chalk (3) and Stapf (10). Of the varieties of this species which have been described, Stapf (10) and Chalk (3) disposed of most. However, the variety P. latifolius var. latifolia still standing deserves comment. With its long narrow leaves it seems that this should be a variety of P. elongatus, while the erect leaves around rigid twigs show a relation to P. latifolius. However, the leaves “1-2 in. long X 2-3 lin. wide” are long narrow leaves and certainly belie Chalk’s and Stapf’s suggestion that it is the common form in the Cape Colony with its “short, wide leaves.” Robyns (8) says that the green color of the receptacle, which has been given by certain authors as a distinctive character for P. latifolius, should not receive consideration. Stapf (10) indicates that, in a mature state, this receptacle is vividly colored red as in P. milanjianus and he describes it as resembling a “small dark red cherry in color and shape.” It is hard to understand why Chalk (3), after using and quoting Stapf, used a green receptacle for a key character only a year DIsTRIBUTION: Limited at present to South Africa, most abundant in the southeastern part, probably ranging farther north but no collections seen. SPECIMENS EXAMINED: SOUTH AFRICA: Transvaal: Zautpansberg, Hutchinson & Gillett 4283 (K—2 sheets), Houseman 5249 (K), Obermeyer 1228 (+F):; Pietersburg, Hout- boschberg, Burtt-Davy 1194 (K); The Downs, Sabie, Rogers 21910 (K); Help- makaar Arboretum, Burtt-Davy 20252 (+A, +BR), Leeman 105 (+K); Blauberg, Smuts 906 (K); Waterberg, J. B. Pole Evans in 1933 (K); Nylstroom, Burtt- Davy 4549 (+K), Galpin 11663 (+K). Swaziland: Forbes Reef Bush, Burtt- Davy 2748; Kaaphe Kloof, Rogers 21089 (K), Burtt-Davy 2455 (K). Orange Free State: Drakensburg, Cooper 1111 ({BM, K, NY). Natal: Zululand, Ngoya Forest, For. Dept. 4266, Chilvers (+A), Wilson in 1922 (A—2 sheets) ; Nat'l Park, Drakensburg, Godman 263 (+BM), Hutchinson 4488 (K); Cham- paign Castle, Meebold 13140 (NY); Durban, near garden, Wilson in 1922 (A), Hout in 1859 (K); Donneybrook, Wilson in 1922 (A); The Boyle, Bews in 1922 (A), Wilson in 1922 (A—2 sheets) ; Van Reenery Prop., Kuntze 1800 (K), Gerard 127 (K), Sanderson sn. ({K). Cape Province: East Griqualand: Umzimkulu, Mhlonga Forest, For. Dept. Herb. Pret. 2168, 2170, 2189 (2 & 4) Kaufmann (K); Maclear Dist., Pot River Berg, Galpin 6831 (K); Emkazene Forest, Ingwangwane, For. Dept. Herb. Pret. 1957, Houshold (K); Instubani Forest, For. Dept. Herb. Pret. 2227, Frazer (K); Buswayo Forest, Manina, For. Dept. Herb. Pret. 2266, Merwe (K). Albany Division: near Grahams- town, MacOwan 1408 (F, K, YU), MacOwan 1958 in part (+GH, +K), Wilson in 1922 (A—2 sheets); w. of Grahamstown, Zeyher 3885 (K); Atherston, Anon. 89 (tK). Uitenhage Division: Zeyher 3880 (K). Knysna Dist.: Katzees Kraal, Burchell 5223, 5254 (+GH, K); Deepwalls Crown Forest, Wilson in 1922 (tA—2 sheets); Harberville Forest, Keet 524 (+K), Bowie s.n. (+K), Burchell 5223 (+K), Wilson in 1922 (A—2 sheets), Munde & Maire 1840 (K). George & Knysna: Bowie sm.(K). George Dist.: Drege 1639 (K), 1953] GRAY, REVISION OF PODOCARPUS 169 Burchell 5843 (+GH, K), Burchell 3505, 5843 (K). Cape Dist.: Table Mtn., near Brown’s Krall, Prior in 1847 (K), Wilson in 1922 (+A—6 sheets), Geehout 12 (A), 223 (A), Zeyher 3533 (A); Stellenbosch, Harvey s.n. (+BM); Murchi- son, Wood 3028 (K); Newland woods, Wolley Dod 2729 (K); Groot Vader’s Bosch, Anon. in 1906 (A), Thunberg s.n. ({UPS); Orange Kloof, Gamble 22002 (K); Oliphant’s River, Pillans 5297 (+K) ; Zeyher 3882 (+K), 3883 (+K), Munde & Maire sn. (+K). Cape Colony but no locality: Anon, in 1879 (CAS), Drege sn. (+Mo), Ecklon & Zeyher s.n. (+A), Burtt-Davy 4549 (+K), Corney Westliche, Prager 90 (+CAS), Anon. sn. (K). No locality indicated: H.R.P. ex Herb. A.C.Limingani Comitis A (DS), Anon, s.n. (K). CULTIVATED: South Africa: Tokai, near Capetown, G.A.W. in 1912 (K), Eames in 1939 (CU); Cleremont Garden, The Hill, Wilson in 1922 (A—S5 sheets); Mission de Bunia, Gilbert 505 bis (+BR); Newlands, Kirstenbosch, Compton in 1941 (tIll). England: Kew, Cook in 1937 (+111), Buchholz in 1950 (+tIll). Podocarpus latifolius var. latior Pilger in Pflanzenreich IV. 5 (Heft 18): 90. 1903; Stapf in Fl. Cap. 5 (Sect. 2, suppl.) : 7. 1933; Chalk et al., For. Trees & Timbers Brit. Emp. 3: 82. 1935. A variety with very wide leaves, which is easily recognizable. DISTRIBUTION: Cape Province on mountain slopes, apparently quite rare. SPECIMENS EXAMINED: SOUTH AFRICA: Cape Province: Vogelgat, Schlechter 9542 (K, +Ph); Table Mt., Wilson in 1922 (+A), Drege, in part, ex Bernhardi Herb. (+Mo). Podocarpus Henkelii Stapf ex Dallimore & Jackson, Handb. Conif. 47. 1923, 1931, 71. 1948; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 279. 1931; Chalk et al., For. Trees & Timbers Brit. Emp. 1: 23. 1932, 3: 84. 1935; Stapf in Fl. Cap. 5 (Sect. 2, suppl.) : 9. 1933. This tree is closely associated with P. latifolius, but can readily be dis- tinguished by the larger, drooping, attenuate and often falcate leaves. Its description, distribution and regeneration is treated at length by Chalk et al (3). In certain areas it is found abundantly (Natal, East Griqualand) where it forms 90% of the trees in pure open forests. In leaf anatomy it shows a range of variation similar to that of P. latifolius, and juvenile specimens of P. latifolius are often difficult to distinguish from it. DIsTRIBUTION: South Africa, in Natal, Swaziland, northern Transvaal and East Griqualand. SPECIMENS EXAMINED: SOUTH AFRICA: Transvaal: Barbarton 2467 (+K). Natal: Swaziland, Forbes Reef, Burtt-Davy 2748a (+K), Pretoria For. Dept. 1880, Houshold (K); Swartzkop, Sim in 1921 (A—3 sheets); Donnybrook, Wilson on Feb. 16, 1922 (A—7 sheets); Kirstenbosch, Sim 1252/14 (tll); Pietermaritzburg, Lindeberg in 1936 (tA), Wilson on Feb. 12, 1922 (A), Wilson on Feb. 14, 1922 (A—3 sheets ¢, 3 sheets 2). Griqualand East: Mt. Ayliff, Ft. Donald, 170 JOURNAL OF THE ARNOLD ARBORETUM _[vor. xxxiv Balembu Forest, Cochrane in 1920 ({K—2 sheets); Nguhi Forest, Cochrane in 1920 (K); Gsewaleni Forest, Pret. For. Herb. 2172, Cochrane (K); Insikini riverside, Dawson 1441 (+K). Cape Province: Knysna, W. A. & C. B. Setchell in 1927 ({UC). Locality unknown: Ad No. 1282, Ser. I, ex Herb. L. van den Bossche (+BR). CULTIVATED: Africa: Tokai near Capetown, Wilmot in 1912 (K). Eng- and: Kew, Buchholz in 1950 (+Ill). United States: New York Bot, Garden in 1926 (+NJU). Podocarpus milanjianus Rendle in Trans. Linn, Soc. ser. 2, 4: 61. 1844; Pilger in Pflanzenreich IV. 5 (Heft 18): 92. 1903, in Nat. Pflanzen- fam. ed. 2, 13: 247. 1926; Dallimore & Jackson, Handb. Conif. 50. 1923, 1931, 75, 1948; Battiscombe, Desc. Cat. of Common Trees and Woody Plants of Kenya Colony 1. 1926; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 279. 1931; Chalk, Burtt-Davy & Desch, For. Trees & Timbers in Brit. Emp. 1: 24. 1932; Robyns in Inst. Roy. Col. Belge, Bull., 6: 226. 1935, in Fl. du Congo Belge du Ruanda-Urundi 1: 6. 1948; Chevalier in Rev. Bot. Appl. 19: 411. 1939. Podocarpus milanjianus has a very wide tropical distribution from 15° s. lat to 10° n. lat. and across the entire width of the continent. It has wider leaves than P. elongatus, but no anatomical differences in the leaves from those of P. latifolius and P. Henkelii. However, the natural areas of these species are so far removed from each other that there is little chance of confusion. Podocarpus milanjianus shows considerable variation in the foliage due to its appearance in all horizons of mountain forest, even into subalpine regions where the reduction in size of both the foliage and the tree is considerable. In general appearance, the leaves seem to be more spreading than those of P. latifolius, but the form of the young specimens is very similar. The leaf anatomy shows no consistent difference of specific importance other than that the rows of stomata seem closer together. In comparisons made by Laurent (4), P. milanjianus exceeds P. lati- folius and P. elongatus in the greatest dimensions of leaves, male cones, peduncles, and length of seed. Chalk et al. (2) include it in their key to African Podocarpus, distinguishing it from P. elongatus by the larger leaves, but from P. latifolius only by the red receptacle which has been mentioned before as an error. In a recent study of specimens accumulated at the Brussels Herbarium, Robyns (8) found such extreme variation that at first he thought he had more than one species, but his detailed discussion gives adequate evidence that this is within the range of normal variation of the species. Moreover, there is extreme difference between adult and juvenile foliage. Hence, Robyns abandons Pilger’s forma arborescens and forma typica, and sub- stitutes the more descriptive forme adulte, forme éricoide, and forme juvénile. He illustrates these by photographs of both foliage and mature tree habit in the case of forme éricoide. ISTRIBUTION: Africa, from Cameroons and southern Sudan, south through Belgian Congo, Uganda, Kenya, Tanganyika, into N yasaland and 1953] GRAY, REVISION OF PODOCARPUS 171 southern Rhodesia, on mountain slopes and alpine zones from 1900-3500 m. altitude. SPECIMENS EXAMINED: SUDAN: Immatong Mts., Ras Logoforsk, Chipp 104 (+K); Mt. Kivetti, Chipp 85 (K); Lomwaga Mt., E. Acholi, Greenway & Hummel 7281 (K), T. Smith Sabei Bugishu, Thomas 2633 (K). BELGIAN CONGO: Ruwenzori: Nyam- bitaba, Humphreys 1402 (+BM); s. of Butahu River, Bambune, Chapin 83 (+NY); westside, Chapin 113 (BR, +NY), Bequaert 3703 (BR), Humbert 8903 (BR); Lukubuku valley, Eggeling 1259 (K); Fishlock & Hancock 160 (+K), Dawe 547 (K), Scott-Elliott sn. (K), Doggett s.n. (K), Lebrun 4570 (BR, K), Purseglove 347 (K), Esmans 15 (BR); Tschiaberimu massif, between Kasindi & Lubango, Lebrun 4762 (+BR, K); Mt. Tsiaberimu, Prince Leopold 49, 50, 51 (+BR); Kahuzi Massif, Humbert 7716, 7716bis, 7716ter (+BR), Scaetta 1408 (+BR, K); Mt. Kahuzi, Scaetta in 1928 (+BR), Lebrun 5528 (K); Virunga Mts., Mushubangabo Volc. Burtt 3158 (K); s. slope, Chambra- gongo Saddle, Burtt 3169 (BR, K); Nyirangongo Volc., Humbert 7951 (BR); Nyamlagira Volc., Humbert 8148 (BR), Heutmann in 1938 (BR). Ruanda: Nyamusha-Kogunge, Scaetta 1722 (+BR); Rutengeria, Lestrade 1 (+BR). KENYA: Mt. Aberdare, R. E. & C. E. Fries 798 (BR, K); Mt. Kenya, west slopes, Mearns 1297 (+F, GH, NY—2 sheets, US); Lari, Wilson in 1921 (A, +UC-seedling), (+A), (A—4 sheets); Nairobi, Kisuru, Dummer 1565 (K); Kikuga Hills, Comm. at Morubasa (K); Muzherengira, Elliott 251, 255 (K); Whyte in 1898 (K—2 sheets); Hutchins in 1907 (K—3 sheets); Hutchins s.n. (K); Cooper 843 (BR); Imp. For. Inst. 15a (A). TANGANYIKA: Mt. Kiliman- jaro, Bismarck Hill, Greenway 3879 (BR, K); s. slope between Umbwe & Weru- weru rivers, Greenway 3180 (K); Schlieben 4862 (BR); Pare Mts. Mushange, Kipare, Greenway 6544 (K); East Usambara Mts., Greenway 4673 (K), 4899 (K); Uluguru Mts., Lupanga Peak, Morogoro, Burtt 4714 (A, K), Schlieben 3153 (BR, K); Kirunga Volc., Kassner 3202 (K); Kondoa, Iringa Dist., Burtt 1068 (K); Kinguassi Mt., Burtt 920 (K), 971 (K), 1174 (K), 1345 (K—2 sheets), Imp. For. Inst. 70, Wigg (A), Mbalamu, Adamson 71, 72 (K). NYASA- LAND: north of Lake Nyasa, Kyimbila Dist., Stolz 387 (K); Rungnee Sta., Stolz 363 (GH, K); Mt. Malosa, Whyte in 1896 (+K); Mwanemba Point, McClownie 171 (K), Johnston in 1896 (+K), Topham 914 (K); below and w. of Lake Chiuta, Cunningham 10 (K); Mt. M’lanji, Burtt-Davy 22023 (+A, +F), 22007 (BR), 22145 (BR), Stolz 373 (GH), Whyte 34, 39 type ({BM), Whyte in 1891 (+K), Buchanan 949 (K—2 sheets), 969 (+K), Clements in 1924 (A), Greenway 3001 (K), Greenway 4673 (+K), Greenway 6300 (K). S. RHODESIA: Gazaland, Chimanimani Mts., Swynnerton 1962 (K); Umkali, Uumba, Galpin 9242 (K), Depts Agr. 1167 (K); Mukungwa Valley, Greenway & Trapnell 5550 (K). ANGOLA: Nkanda Hills (Congo), Dawe 110 (+K); Dist. Bie, Cuito Rapids, N. Lisboa, Gossweiler 10975 (K); Sierra da Chella, Humpata, Humbert 16628 (+BM). CULTIVATED: Tanganyika, Comm. For. 465/33 (A, BM, BR, #F, Ill, K, NY, 172 JOURNAL OF THE ARNOLD ARBORETUM _jvot. xxxiv +tY); Horti Thenensis, Luja 77, Ser. 11] (+BR); Nursery Kinohop, Aberdeen Mt., Wilson in 1921 (A-seedlings) ; N. Rhodesia, Bot. Garden, Burtt-Davy 21883 (BR). Podocarpus ulugurensis Pilger in Notizbl. Bot. Gart. Mus. Berlin 12: . 1934. This tree, recently described from a single collection, resembles P. milanjianus except for several rather striking differences. The leaves of P. ulugurensis, contrary to the foregoing species, have a distinctly grooved upper midrib. According to the description, the male cones have thickly coriaceous scales 4 mm. long and the cone on the type specimen measures 3 cm. long and 4 mm. wide. Study of the leaf anatomy shows extreme thickening of the cell walls in all regions. Most surprising of all, two ad- ditional resin canals are seen in the transfusion tissue, a feature found elsewhere only in certain specimens of Eupodocarpus from the South Pacific, P. longifoliolatus in New Caledonia and P. decipiens in the Fiji Islands. DIsTRIBUTION: In the Uluguru Mts. of Tanganyika, 2200 m. elevation, in fog forests. SPECIMEN EXAMINED: TANGANYIKA: Uluguru Mts., Schlieben 4224 type (+BR). Podocarpus madagascariensis Baker in Jour. Linn. Soc. 21: 447. 1885; Pilger in Pflanzenreich IV. 5 (Heft 18): 92. 1903, in Nat. Pflanzenfam. ed 2, 13: 247. 1926; Laurent in Ann. Fac. Sci. Marseille 23: 53. 1915; Dallimore & Jackson, Handb. Conif. 50, 1923, 1932; 74. 1948; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 279. 1931. Podocarpus madagascariensis, as indicated by Baker, is closely related to the foregoing species. It is a small tree indigenous to the island of Mada- gascar. Numerous specimens were collected by Perrier de la Bathie in 1912. An excellent study of the species and comparison with P. elongatus, P. lati- folius, and P. milanjianus were made by Laurent (4). The size of foliage leaves and peduncle exceeds that of all the other species, and the seed (20 mm. long) is twice as large as that found in the other three species. The length of the male cone, 13 mm., is the least of any species in this group. Pilger’s (6) suggestion that this species belongs to section Stachy- carpus is unfounded. The leaf anatomy shows the greatest degree of thickening of cell walls, to the extent that most of the mesophyll, including much of the palisade parenchyma, is thickened, with pitted walls, and much of it is devoid of cell contents. The large seed is blackish and oval in shape, 20 mm. long and 8 mm. in diameter. DISTRIBUTION: On mountain slopes on the island of Madagascar. 1953 | GRAY, REVISION OF PODOCARPUS 173 SPECIMENS EXAMINED: MADAGASCAR: Massif de Beampingaratra (Sub-Est), Mt. Papanga, Hum- bert 6398 (+A), Baron 2794, 3129, 3441 (+K); Chouvenot 107 (BM, +K); Kiknife Hills, Warburg 549 (K). Podocarpus madagascariensis var. rotundus Laurent in Ann. Fac. Sci. Marseille 23: 59. 1915. This variety grows at an altitude higher than that of the species. It possesses somewhat smaller leaves but larger seeds (28 mm. long). DIsTRIBUTION: Madagascar, on basalt at 1500-1800 m. altitude. SPECIMEN EXAMINED: MADAGASCAR: Parker s.n. (+Kew). Podocarpus rostratus Laurent in Ann. Fac. Sci. Marseille 23: 60. 1915; Florin in Svenska Vet.-Akad. Handl. ser. 3, 10: 262. 1931; Orr in Trans. Bot. Soc. Edinburgh 34: 11. 1944 This extremely rare, and perhaps now extinct, species from the island of Madagascar, is different from all other podocarp species. It is a small tree, 8-10 m. high, with somewhat shiny leaves, 2.3-3.6 cm. long and 1.25-2 mm. wide, sessile, straight, very narrowly lanceolate, gradually attentuate toward apex. Male cones solitary and sessile or in groups of 2 or 3 on a slender, 5—6.5 mm. peduncle, 10-20 mm. long and 3 mm. wide, with scales triangular and apiculate. Female cones and seeds are unknown. The leaf anatomy shows a single midvein with two very small lateral groups of transfusion tissue, all of which is surrounded by a thick layer of parenchyma definitely distinguishable from the rest of the mesophyll of the leaf by the dense cell contents. Below the vascular bundle are 3 resin canals, one central and two lateral. Laurent (4) found a thin sheet of “irrigation tissue” on each side of the vein separating the palisade from the spongy parenchyma. We do not find this to be lignified and pitted as in true accessory transfusion tissue. There is a continuous hypoderm on the upper side of the leaf, often more than one layer thick, and it is present also between the rows of stomata below. Differing from all other podo- carps, it has a number of resin canals, 6-10, around the edge of the leaf just inside the hypoderm (Trext-Fic. 1). These were described rightly by Laurent (4) as varying in number from one leaf to another and even within the same leaf according to the position near the base, middle, or apex. Also the arrangement of the stomata into two bands on each side of the midrib, on the lower side of the leaf, is a distinctive character. The stomata show a definite Florin ring. The lack of the female structures prevents exact placing of this species taxonomically. Laurent (4) tried to place this species among its relatives. He compared it with P. gracillimus, Nelson 423 (which has been clearly shown to belong in section Afrocarpus), its foliage being the same shape, but P. rostratus differs from it, as expected. Because P. gracillimus was originally in the 174 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv section Stachycarpus, as expressed by Pilger (6), Orr (5) attempted to place it as an aberrant member of that section. The differences listed by him, and which we have confirmed, indicate clearly the section Eupodo- TEXT-FIGURE 1. Podocarpus rostratus Laurent. Cross-section of leaf showing relative thickness of (C) cuticle, extent of (H) hypodermal fibers, location of marginal and vascular (RC) resin canals, and the special layer of (M) mesophyll surrounding the vascular bundle and transfusion tissue. * 50 carpus where continuous upper hypoderm and the three vascular resin canals are compatible. In comparison with other species, we find P. rostratus most closely resembles P. acutifolius (New Zealand) in sub- section D, which is found also in Chile and Australia. In all except two characters of the leaf anatomy P. rostratus matches P. acutifolius, even to the presence of the Florin ring in the stomatal subsidiary cells. We are placing this species in a new subsection E of Eupodocarpus in recognition of the following anatomical features: (1) the presence of a variable number of marginal resin canals, different from the condition in all other podocarps; (2)-the double stomatal bands on the lower side of the leaf, likewise not found in any other podocarps; (3) the three vascular resin canals found in African and Asiatic Eupodocarpus but not in sections Afrocarpus nor Stachycarpus. DIsTRIBUTION: Madagascar, on high summits of Mt. Tsaratana, in forests almost destroyed by fire. SPECIMEN EXAMINED: MADAGASCAR: Mt. Tsaratana, H. Perrier de la Bathie 10443 (tex Florin, Riksmuseet, Stockholm). 1953] GRAY, REVISION OF PODOCARPUS Be — . onn mn +> Ww i) so 10. LITERATURE CITED BUCHHOLZ, JOHN T. & NetTa E. Gray. A Taxonomic Revision of Podo- carpus. I. The sections of the genus and their subdivisions with special reference to leaf anatomy. Jour. Arnold Arb. 29: 49-63. 1948. . CHALK, L., J. Burtt-Davy & H. E. Descu. For. Trees & Timbers Brit. p 1932. . CHALK, L., M. M. Cuattaway, J. Burtt-Davy, F. S. LAucHton, & M. H. Scott. For. Trees & Timbers Brit. Emp. 3: 69-90. 1935. LaurENT, M. L. Les Podocarpus de Madagascar. Ann. Fac. Sci. Marseille 23: 52-64. 1915. . Orr, M. Y. ee leaf anatomy of Podocarpus. Trans. Proc. Bot. Edinburgh 34: 1-54. . Pricer, R. foe Pflanzenreich IV. 5 (Heft 18): 1- a 1903. PILceR, R. Taxaceae. Notizbl. Bot. Gart. Mus. Berlin 12: 1934. Ropyns, W. Sur les especies de Podocarpus du Congo oe a du Ruanda- Urundi. Inst. | Col. Belge, Bull. 6: 226- — 1935 RICHARD, L. C. A Comm. Bot. Conif. 18 STAPF, O. Seen elanes Fl. Cap., ed. Hill, 5 ae 2, suppl.): 3-14. 1933. AGNES ScoTT COLLEGE, D ECATUR, GEORGIA. 176 JOURNAL OF THE ARNOLD ARBORETUM [voL. xxxiv THE IDENTITY OF VALENTINIA ILICIFOLIA SWARTZ RIcHARD A. HowarD AND CLAUDE E. Situ, JR. THE ILICcIomD CASEARIAS of Cuba have been studied recently by Fr. Marie-Victorin (Contrib. Inst. Bot. Univ. Montreal 49: 19-56. 1944) who discussed the confusion existing between Casearia ilicifolia Vent. and Valentinia ilicifolia Sw. At that time Marie-Victorin did not have at his disposal information on the type specimens of these two species, nor the recent collections from Hispaniola which allow a more satisfactory descrip- tion and disposition of these species. The genus Valentinia was described by Swartz (Prod. 63. 1788 and Flor. Ind. Occ. 2: 689-90, t. 14. 1800) based on descriptions and illustrations by Plumier (Cat. in add. 46. 1703, and ed. Burm. 160, ¢. 167. 1755) and Plukenet (Phytographie 196, fig. 3. 1691) as well as new material which Swartz illustrated. The genus contained one species, Valentinia ilicifolia. The genus and species were transferred to Casearia and placed in the sec- tion HEXANTHERAE by Bentham and Hooker (Gen. Pl. 1: 796. 1867). The section HEXANTHERAE of the genus Casearia had been established earlier by Endlicher (Gen. Pl. 917, 1840) for a single species, Casearia ilicifolia Ventenat. Endlicher had considered the Swartz genus Valentinia as of doubtful position. Bentham and Hooker, in referring Valentinia to Casearia, implied the identity of Valentinia ilicifolia Sw. and Casearia ilicifolia Vent. but did not specifically transfer the Swartz species. Eichler in 1871 [Flor. Bras. 13 (1): 462. 1871] cites Valentinia ilicifolia Sw. in synonymy with Casearia ilicifolia Vent. It appears, however, that Urban (Symbol. Ant. 8: 446. 1920) was the first to publish the actual combination Casearia ilicifolia (Sw.) Vent. implied by the earlier workers and generally accepted in the literature today. Urban was in error in making or accepting such a combination, for Casearia ilicifolia Vent. and Valentinia ilicifolia Sw. are actually two distinct species although both belong to the genus Casearia. It should be noted here that Urban cited in synonymy with Casearia ilicifolia (Sw.) Vent., Valentinia ilicifolia Sw. as the basonym and also Casearia comocladifolia Vent. Urban apparently recognized discordant elements in the original pub- lication of Valentinia ilicifolia which Swartz had based on a Plumier refer- ence and figure as well as material of his own. The Plumier figure was con- sidered by Urban in several subsequent discussions. In 1919 (Fedde Rep. Spec. Nov. 15: 403-4. 1919) he concluded Plumier’s illustration could not be a Malpighia as Plumier had suggested and might possibly be an unde- scribed species of a different or possibly new genus but that it was not the same as Valentinia ilicifolia Sw. In his consideration of Plumier’s life and writings in 1920 (Fedde Rep. Beih. 5: 75. 1920) Urban continued to call the Plumier figure Valentinia ilicifolia Sw. but reported it as a tree of dubious affinities. Finally in 1922 (Fedde Rep. Spec. Nov. 18: 365-6. 1953] HOWARD & SMITH, VALENTINIA ILICIFOLIA 177 1922) Urban decided that the Plumier illustration was comparable to re- cently acquired herbarium specimens and these he described as Sloanea ilicifolia spec. nov. He cited in synonymy “V. ilicifolia Sw. quoad syn. Plum.” Unfortunately Urban added to the confusion by using the same specific name and called this species Sloanea ilicifolia. We conclude that Urban did not intend to imply a new combination here as he referred the species Valentinia ilicifolia Sw., represented by the Swartz description and material, to Casearia in the same year. Marie-Victorin, without access to the type specimens concluded it was difficult to determine what the Swartz species actually was and accepted Urban’s disposal of the name Valentinia ilicifolia as a new combination in Casearia. We now have photographs of the type collections of Valentinia ilicifolia Sw. and Casearia ilicifolia Vent. In addition we have been aided by Dr. Ivan Johnston who examined the Swartz material of Valentinia ilicifolia in the British Museum. The Swartz material is definitely a member of the genus Casearia. How- ever, to transfer Valentinia ilicifolia Sw. to the genus Casearia as past workers have done would create a later homonym for Casearta ilicifolia Vent., a completely distinct species, hence a new name must be found. Gomez and Molinet recognized this transfer difficulty and created the name Casearia Valentinia for Valentinia ilicifolia Sw. (Gomez de la Maza, Flor de Cuba 33. 1887), however an earlier name is available. In 1803 Ventenat described a pubescent leafed Casearia which he called C. ilicifolia and a glabrous leafed species which he called Casearia comocladifolia (Choix de pl. 44. 1803). The latter was correctly recognized by Urban as identical with Valentinia ilicifolia Sw. and cited in synonymy when Urban referred the Swartz species to Casearia as Casearta ilicifolia (Sw.) Vent. (Symbol. Ant. 8: 446. 1920). Marie-Victorin also accepts C. como- cladifolia Vent. for the common glabrous, usually membranaceous ilicioid leafed Casearia of Hispaniola and Cuba; however, he did not recognize this was the same as Valentinia ilicifolia Sw. The correct nomenclature should be: Casearia comocladifolia Vent. Choix de pl. 44. 1803. Valentinia ilicifolia Sw. Prod. 63. 1788; Flor. Ind. Occ. 2: 689-90, t. 14. 1800. Casearia Valentinia Gomez and Molinet, Gomez de la Maza, Flor. de Cuba The pubescent leafed Casearia ilicifolia Vent. is well described and illustrated in the original publication (Choix de pl. 44. 1803), and is clearly distinct. This species was discovered by Turpin near Monte Cristi in Santo Domingo around 1800. Turpin himself did the illustration for Ventenat. It is clear from the original description that Ventenat was describing a new species and not transferring the earlier Swartz name to Casearia as Urban, Moscoso (Cat. Flor. Doming. 386. 1943), Gilg. (Nat. Pflanzenfam. ed. 2, 21: 453. 1925) and others imply. The correct nomen- clature for this species is: 178 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxIv Casearia ilicifolia Vent., Choix de pl. 44. 1803. Samyda ilicifolia (Vent.) Poiret, Lam. Encycl. Suppl. 5: 31. 1817. The lack of any collections of this species since 1800 have handicapped most recent considerations of this problem. In 1946, the senior author relocated the species near Monte Cristi in the Dominican Republic. Addi- tional material was collected in 1950 and it was determined that the species is abundant in restricted locations and has only been overlooked by past collectors in the dry thorn shrub vegetation around Monte Cristi. The following description of Casearia ilicifolia Vent. has been compiled from a field knowledge of this species and the additional collections cited below. Casearia ilicifolia Vent. Spreading bushes 10 feet tall or trees 15-18 feet tall with trunk diameter averaging 4 inches at breast height. Branches occasionally geniculate and the youngest twigs densely pilose. Petioles 2-3.5 mm. long, densely pilose. Leaves oblong in outline, 5.5 & 3.0 to 7.0 * 3.0 cm. long and broad, coria- ceous; apex truncate-emarginate tipped by two spines to slightly rounded; base truncate to truncate-cordate; margin strongly sinuate, spinose, the 10-13 lateral spines and ultimate leaf margins distinctly cartilaginous; primary veins 6—9 pairs, arcuate anastomosing near the margin with strong branch veins running into the spines; the blade lightly pubescent becoming glabrate and shining above, densely persistently pilose below. Flowering pedicels 3-4 mm. long; hypanthium 0.5 mm. long. Sepals 6, united at the base, pink, 6.5 & 2.3 to 7.5 & 3.5 mm. long and broad, densely short pilose outside, sparsely pubescent inside; corolla wanting; filaments and staminodes 8, thinly united at the base, the union 1.5 mm. long, the free portions of the filaments 2 mm. long, the free portions of the staminodes 1.5 mm. long, staminodes attenuate, filaments and staminodes sparsely short pilose, anthers affixed near the base, 1 mm. long; ovary globose, 3.0-3.5 mm. in diameter, short pilose above the middle, glabrous below, style 2.5—3.0 mm. long, sparsely short pilose, stigma capitate, 3-sided, apex depressed; fruiting pedicels 5-8 mm. long, the mature fruit a yellow- orange berry, depressed globose, 1 cm. in diameter, splitting along 3 lines, the pulp red, the seeds 4-8, yellow. SPECIMENS SEEN: DOMINICAN REPUBLIC: prov. Monte Cristi: Villa Isabella, Howard 12520 (GH); Banks of the Rio Yaqui del Norte, south of Monte Cristi, Howard 9570 (GH). Collected in flower and fruit in July and August. Previous workers have been troubled by the earlier cited ranges of these species. Swartz, in the original publication of Casearia comocladifolia (Valentinia ilicifolia Sw.) gave the location as “habitat in sterilissimis petrosis Hispaniola, versus Oceanum. In Cuba circa Havanam.” The species has never been relocated near Havana although collections are known from the Oriente Province of Cuba. In Hispaniola the current loca- 1953 | HOWARD & SMITH, VALENTINIA ILICIFOLIA 179 tions for this species are Gonaive Island and near Port au Prince in Haiti and near Barahona and Monte Cristi in the Dominican Republic. The Barahona location is a new record and is based on the collection Howard 12599. This was made from a 15 foot tree which had pink flowers which appeared while the plant was in full leaf. It was collected in flower in August and was growing in the thorn shrub area north of the town of Barahona. Marie-Victorin refers both Casearia ilicifolia and Casearia comocladifolia to the section HEXANTHERAE DC which he emtended. The primary charac- ter listed by Marie-Victorin is the production of flowers before the leaves appear. The two species considered here, which the senior author has seen in the field, may be leafless when flowering; may produce flowers on ter- minal leafless branches while the rest of the plant possesses leaves (Howard 9570), or may be in full leaf when flowering with the flowers hidden in the dense foliage (Howard 12559). The character of hysteranthous flowers as used by Marie-Victorin is not reliable. The number of stamens in the species of the section HEXANTHERAE is also variable. Marie-Victorin distinguishes Casearia comocladifolia by having six stamens. However, Marie-Victorin’s own dissections of a Gray Herbarium specimen of Casearia comocladifolia (Wright 12) with the open flowers spread on cards show flowers with six, seven and eight stamens and so indicated in Marie-Victorin’s handwriting. The original plate given by Swartz in describing Valentinia ilicifolia shows eight stamens and the recent collections of this species from the Dominican Republic have most of the flowers with eight stamens, a few with seven and a very few with six stamens. Recent collections of Casearia ilicifolia also have flowers with eight stamens. The name of the section HEXANTHERAE is deceptive. We can return now to a consideration of Sloanea ilicifolia Urban. In the original publication of this species Urban cited in synonymy “Valen- tinia ilicifolia Sw. quoad syn. Plum.” In his citation of specimens Urban refers first to the Plumier figure and then to collections. No type was specifically selected. The original description is to a large extent based on a flowering specimen collected by Pére Straessle at Morne Bellefon- taine. All other specimens cited are indicated as sterile. The Straessle collection is presumably destroyed having been at Berlin. Only one of the other collections, Leonard 3797a, is available in this country. In 1929 Urban and Ekman [Arkiv Bot. 22A (17): 26. 1929] described a second species of Sloanea with ilicioid leaves which they called Sloanea castor. The distinctions between Sloanea ilicifolia and Sloanea castor do not seem reliable; the flowers and fruits are similar and the leaves of the two species show all intergradations in the material now available to us. Furthermore the ranges of the two entities coincide. It appears to us that Sloanea castor Urb. and Ekman must be reduced to synonymy with Sloanea ilicifolia Urb. An emended description of this species, based on the available collections, follows. 180 JOURNAL OF THE ARNOLD ARBORETUM _ [VvoL. xxxIv Sloanea ilicifolia Urban, Fedde Rep. Spec. Nov. 18: 365. 1922. Sloanea castor Urban and Ekman, Arkiv Bot. 22A (17): 26. 1929. Tree 15—20 m. tall. Branchlets with brown or grey-brown bark, scarcely striate to rough. Twigs more or less sulcate, scarcely to densely light brown puberulent. Leaves alternate to opposite, often clustered on many twigs towards the ends of the branchlets; stipules caducous, 3.5—-4.5 mm. long, linear-lanceolate, light brown pubescent; petioles 3.0-12.0 mm. long, terete, scarcely incrassate at both ends, light brown puberulent; blade very vari- able in size and shape, 3.5 & 2.2 to 12.5 & 9.5 cm. long and wide, elliptic- lanceolate to broadly oval, stiff coriaceous, midrib and secondary veins impressed, scantily puberulent to glabrous above, prominent, puberulent to glabrous beneath, secondary veins 4 to 8, arcuate-ascending, generally ter- minating in a spine at the leaf margin, tertiary venation irregular, occa- sionally terminating in a marginal spine, base rounded to cordate, apex rounded to acute to short acuminate, the midrib prolonged into a spine, margin irregularly and deeply spinose dentate. Inflorescences axillary to leaf scars, one to several flowered; flowers 3.0-4.0 mm. long, about 4.0 mm. in diameter, sepals generally 6, to 3.0 mm. long, 0.5—1.25 mm. wide at the base, lanceolate-deltoid, obtuse, densely puberulent within and without; stamens 2.0-3.0 mm. long, filaments 1.5-2.0 mm. long, puberulent to pubescent, anthers 0.75-1.0 mm. long, deltoid, puberulent, dehiscing linearly most of their length, connective prolonged into a short knob above the anther sacs; pistil to 3.0 mm. long, ovary 1.0-2.0 mm. long, densely pubescent, indistinctly 4-angled, 4-loculed, gradually giving rise to the style, style often divided to the ovary to form two parts each of which may be divided again at the apex, pubescent at the base, glabrous above. Capsule brown, 1.5-1.75 cm. long, 1.25-1.5 cm. in diameter, sub- globose to ellipsoidal, 1-loculed, 1—-2-seeded, 3—4-valved; valves with velu- tinous outer surface densely covered with flexible spines; spines to 1.5 cm. long, densely antrorsely puberulent, straight, abruptly acute. Seeds en- closed almost completely in a deeply 3-lobed red aril which is firmly at- tached to the chalazal quarter of the seed. SPECIMENS SEEN: DOMINICAN REPUBLIC: Prov. of Barahona, Monteada Nueva, S.E. of Polo, alt. 3500 ft. R. A. Howard 12355, 50 ft. tree with small buttresses, d.b.h. 3 ft., fruit brown, Aug. 3, 1950 (GH); Prov. of Azua, Cordillera Central, Las Lagunas, 2250 ft., E. L. Ekman H6366, quite common tree, flowers past anthesis, July 13, 1926 (S). HAITI: Vicinity of Mission, Fonds Varettes, alt. about 3000 ft., E. C. Leonard 3797a (co-TyPE), occasional tree, steep wooded mountain slope, sterile, April 17—-May 4, 1920 (NY, S); Massif de la Hotte, western group, Les Roseaux at Nan-Patates, 3000 ft. alt., H. L. Ekman 10690 (type of S. castor), common tree, in flower and fruit Sept. 17, 1928 (S, US): Massif de la Hotte, western group, Jeremie, Source-Cahouane, alt. 600 ft., E. L. 1953] HOWARD & SMITH, VALENTINIA ILICIFOLIA 181 Ekman H10234, in fruit July 4, 1928 (S); Massif de la Hotte, western group, Torbee, La-Marie-Praux, alt. 2100 ft., E. L. Ekman H5399, sterile on Dec. 8, 1925 (S); Massif de la Selle, Morne Tranchaut, Fourcy, alt. 4620 ft., E. L. Ekman H12835, sterile on Aug. 4, 1924 (S); Massif du Nord, Port-de-Paix, high ridge of Haut Piton, alt. 3000 ft., E. L. Ekman H3706, sterile on April 6, 1925 (S); Petit Source, Morne de Commissaires, alt. 4560 ft., L. R. Holdridge 1930, tree 15 m. tall, d.b.h. 3 dm., in flower Sept. 1, 1944 (GH, US); Riviere Glace, alt. 2250 ft., L. R. Holdridge 2219, tree 20 m. tall, d.b.h. 3 dm., in flower and fruit Aug. 7, 1945 (US). The figures given by Plumier and Plukenet agree with the specimens cited above. UNIVERSITY OF CONNECTICUT, Storrs, CONNECTICUT AND BIOLOGICAL LARORATORIES, HARVARD UNIVERSITY. 182 JOURNAL OF THE ARNOLD ARBORETUM _[VoL. XXXIV NEW SPECIES AND DISTRIBUTION RECORDS FOR LAS VILLAS PROVINCE, CUBA Ricuarp A. HowARD AND WINSLOW R. Briccs SEVERAL NEW SPECIES and significant extensions of known ranges of distribution were encountered in the course of general collecting and field work in southern Las Villas province of Cuba in the summer of 1951. These are reported in this paper. The authors, instructor and member of the class in Tropical Botany of Harvard University, wish to express their appreciation to the other mem- bers of the class for their assistance during this work. Duplicate specimens are to be distributed and will cite as collectors, R. A. Howard, W. Briggs, I. Lane, P. Kamb, and R. Ritland. The many services supplied by Dr. I. D. Clement, economic botanist in charge of the Atkins Garden and Research Laboratory which was our headquarters for this work, are gratefully acknowledged by the authors speaking for the entire group. Dorstenia Lanei sp. nov. Herba acaulis; rhizomatibus erectis, cylindricis, crassis, 13-18 mm. longis, 3-4 mm. crassis; stipulis 0.4-0.6 mm. longis, ad 0.2 mm. latis, an- guste acuminatis, minute puberulentis. Folia peltata, purpureo-grisea, orbiculari-ovata, 18 X 15, 15 < 12 vel 9 X 8 mm., apice rotundata, basi saepe truncata, ad medium inserta, margine undulata vel crenulata, supra glabra sicco paullo scabrida, subtus puberulenta, petiolis glabris 4-7 cm. longis basi etiolatis glabrisque, apice purpureis puberulentisque. Inflores- centia cyathiformia; pedunculis 8-20 mm. longis, ad apicem puberulentis; receptaculo 4-5 mm. diametro, peltato, puberulente, purpureo-rubiginoso, margine 6-10 dentibus triangularibus horizontalibus inaequalibus, 0.40.8 mm. longis; floribus ¢ et @ intermixtis, floribus ¢ verdis, floribus 9 albis. CUBA: Las Villas Province: 10kms. south of Santa Clara on the road to Manicaragua, Howard, Briggs, et al. 69 (typE, GH). Collected in flower July 5, 1951. This is a unique species of Dorstenia, distinctive in its habit in the field and the first species known to us to occur on serpentine soil. The erect rhizome of Dorstenia Lanei is entirely buried so that the blade of the pel- tate leaf appears to be flat on the surface of the ground and the receptacle opens flush with the surface of the soil. The peduncle and petioles are all etiolated at the base and show pigmentation only at the immediate Publication No. 13, Journal Series from the Atkins Garden and Research Laboratory of Harvard University Soledad, Cienfuegos, Cuba 1953] HOWARD & BRIGGS, NEW SPECIES FOR CUBA 183 apices. The venation of the leaf blade is palmate from the point of at- tachment of the petiole. All the veins bifurcate once before reaching the margin. The midrib is not particularly distinct, appearing of the same size as the other veins and is not pinnately branched. Dorstenia Lanei, named for Irwin Lane who discovered the first speci- men in the field, is most nearly allied to D. erythrandra Wr. ex Griseb. but differs in habit, size and leaf venation. Erythrina Elenae sp. nov. Arbor ad 10 m. alta; truncus 30 cm. diametro, spinis suberis, magnis, persistentibus; ramulis aculeatis, spinis 5-6 mm. longis. Folia pinnata, pinnis 3; petiolis inermibus, 6.5-11.5 cm. longis, glabris; petiolulis 2-3 mm. longis, 0.4—0.7 mm. crassis, fuliginosis, puberulentis vel glabris, rugosis; stipellis 0.5 mm. longis, fuliginosis, glandulosis; foliolis 3, inermibus, gla- bris; foliolis terminalibus lanceolatis vel lanceolato-ovatis, 3-13 cm. longis, 0.4—2.8 cm. latis, late acuminatis, basi rotundatis, infra medium latissimis, nervis lateralibus 8-20, nervis secondariis prominente reticulatis, margine integro; foliolis lateralibus similibus aliquando brevioribus. Flores non visi. Inflorescentia subterminalis, racemosa, lignosa, ad basin tumida, ad apicem attenuata, ad 13 cm. longa; pedicellis lignosis, 4-7 mm. longis, 1.52.5 mm. crassis. Legumina usque 2.5 cm. stipitata, 5-8 cm. longa, 7-8 mm. lata, moniliformia, fuliginosa, sublignosa, glabra, apice 1.8-2 cm. arcuato-rostrata. Semina 2-5, ovalia, 6-8 mm. longa, scarlatina. CUBA: Las Villas Province: limestone hillside 4 mile west of the spot where the Camino de la Sur crosses the Rio San Juan along the south slope of the Trinidad Mountains. Howard, Briggs, et al. 377 (type, GH). Collected in fruit July 17, 1951. The lanceolate to lanceolate-ovate glabrous leaflets of Erythrina Elenae set this species so distinctly apart from any other species reported from the Antilles that a satisfactory comparison is not possible with any of them. The specimens were collected in fruiting condition and until the flowers are known the species can not be assigned to any section in the genus. Erythrina Elenae is respectfuly named for Mrs, William Claflin, nee Helen Atkins, in appreciation of her active interest in the flora and the study of botany in Cuba. Tetrazygia aurea sp. nov. Frutex vel arbor parva, ad 5 m. alta. Folia anguste oblonga, lamina 4— 5 cm. longa, 1—-1.5 cm. lata, apice attenuata demum incurvata, in mucrone 0.4-0.5 mm. longo terminata, basi rotundata, supra nitida, glabra, subtus ferrugineo-stellata margine incurvata, nervis primariis 3, subtus promi- nentibus, supra prominenter impressis, nervis secundariis rectis angulis divergentibus. Inflorescentia terminalis, paniculata, floribus 5, raro 7; calyx undulatus; petala 5—6, 8-10 mm. longa, 5—6 mm. lata, rosea, obovata vel orbicularia, unguiculata; stamina 10-12, antheris 5 mm. longis, fila- 184 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv mentis 7 mm. longis; ovarium 5- vel 6-loculare, ovulis plurimis; stylis gracilibus, 11 mm. longis, ad apicem attenuatis, stigma punctiforme. eng depresso-globosus, 5- vel 6-locularis, 7-8 mm. diametro, pedicello 2 longo (fructus pedicellusque persistente ferrugineo- stellatus) semina ae hlubin cuneiformia, 1 mm. longa, pars una adusta, pars altera castanea. CUBA: Las Villas Province: Gavinas, Trinidad Mountains, Howard 6447 (type, GH); Las Vegas de Matagua, Trinidad Mountains, Jack 5953; Buenos Aires, Trinidad Mountains, Leon, Jack & Rowe 13936, Smith, Hodgdon & Gonzales 3383; El Purial on Rio Banao, Lomas de Banao, Ekman 16236. The cited collections have all been named Tetrazygia elaeagnoides, a species which differs from 7. aurea in having a cinereous pubescence, gla- brate fruits, dull and acuminate leaves, 4-parted flowers and a strongly 4- lobed fruit. The specific epithet is derived from the golden color of the leaves. Pectis Ritlandii sp. nov. Herba prostrata. Folia oblongo-linearis, 5-8 mm. longa, 1—2 mm. lata, apice acuta et spinulosa, margine 3—4 paribus setarum, praecipue basin versum, non ad apicem, glandulosa, glandulis submarginalibus, uniseriatis, 4—9 per marginem. Capitula solitaria, terminalis; pedunculis 5-11 mm longis, glabris; involucro campanulato, 4-5 mm. longo; phyllaribus 5, lineari-obovatis, acutis vel obtusis, basi gibbosis, carinatis, dorso glandu- losis, glandulis oblongis irregulariter dispositis, margine scariosis hyalinis; floribus ligulatis 5, ligulis luteis, ad 5 mm. longis; floribus discoideis 4-8, corollis ad 3.5 mm. longis; pappi setis numerosis, inaequalibus, maturitate ad 2 mm. longis, scabris, pilulis minute excunventibiss, CUBA: Las Villas Province: on coastal rocks between Punta Lobas and Pasa Caballos, Howard, Briggs, et al. 357 (type, GH); Castillo de Jagua, Howard 4219. Pectis Ritlandii is named in honor of Richard Ritland who collected the first specimens. The species is most closely related to Pectis Leonis Rydb. which is known only from savannahs north of the Sancti Spiritus moun- tains. Pectis Leonis differs from the present species in having the stem and branches rough pubescent; the glands in the leaves about 16 in num- ber, and the margins of the phyllaries purplish. Xylosma Shaferi (Wils.) comb. nov. Myroxylon Shaferi P. Wilson, Torreya 30: 73. 1930. Tillandsia argentea Griseb. CUBA: Las Villas Province: forest on hillslope west of Rio San Juan crossing on the southern slopes of the Trinidad mountains, Howard, Briggs, et al. 367. 1953] HOWARD & BRIGGS, NEW SPECIES FOR CUBA 185 This small silvery Tillandsia was described by Grisebach on a Wright specimen collected at Monteverde in Oriente province. It has since been reported from Jamaica in the Greater Antilles. The current collection is the first record of its occurrence in Las Villas province. Cattleyopsis Lindenii (Lindl.) Cogn. CUBA: Las Villas Province: thorn shrub on the Camino de la Costa, west of Juragua, Howard, Briggs, et al. 245. This species has a wide distribution in Cuba, the adjacent Bahama Is- lands, and Jamaica but has not been reported previously from Las Villas. Tetramicra erosa Carabia CUBA: Las Villas Province: dry hillside 10 kms. south of Santa Clara on serpentine soil, Howard, Briggs, et al. 293. Previously known from the Oriente province of Cuba this is the first record of the species from Las Villas. Croton prostratus Urban CUBA: Las Villas Province: dry hillside 10 kms. south of Santa Clara on serpentine soil, Howard, Briggs et al. 78. The type collection and a single additional collection of this species were made by Ekman on Sierra de Nipe in Oriente province. The current collection was made from a few plants growing in association with Croton nummularifolius in an open savannah. Leucocroton revolutus Wright CUBA: Las Villas Province: dry hillside 10 kms. south of Santa Clara on serpentine soil, Howard, Briggs et al. 107. The type collection of this species was made in Pinar del Rio province. It is also known from the Oriente province. This collection represents the first material from central Cuba. Linociera bumelioides Griseb. CUBA: Las Villas Province: thorn shrub along the Camino de la Costa south of Juragua, Howard, Briggs et al. 238. A widely distributed species not previously recorded from Las Villas province. Heliotropium hypogaea Urb. and Ekman CUBA: Las Villas Province: Camino de la Costa west of Juragua, Howard, Briggs et al. 226. Heliotropium hypogaea was described by Urban and Ekman based on material collected on Gonave Island off Hispaniola. The collection cited 186 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv above is the first record of this interesting plant from Cuba. A small colony of this species was found growing on sandy soil in a thorn shrub thicket. The plants were in flower on July 11, 1951, and the corollas were white with yellow throats. Mature fruits had elongated peduncles and were pushed into the ground as described by Urban and Ekman. The species is distinctive in the genus Heliotropium in having these hypogeous fruits. Tecoma microphylla (Lam.) Urban CUBA: Las Villas Province: thorn shrub along the Camino de la Costa, south of Juragua, Howard, Briggs et al. 256. Like the preceding species Tecoma microphylla has been found only on Gonave Island off Hispaniola. This is the first record for this handsome shrub in Cuba. Guettarda rigida A. Rich. CUBA: Las Villas Province: dry hillside 10 kms. south of Santa Clara on serpentine soil, Howard, Briggs et al. 73. This species has been collected previously in Matanzas province and the current collection represents the first record from Las Villas. Machaonia microphylla Griseb. CUBA: Las Villas Province: forested hillslope west of the Rio San Juan crossing on the southern slopes of the Trinidad mountains. Howard, Briggs et al. 397; San Blas — Buenos Aires area, Howard 6537. This handsome 15 foot shrub has attractive white flowers with a strong and pleasant odor. The plants were extremely attractive to bees and might well be cultivated as ornamentals or honey plants. The collection made by the Tropical Botany class will be distributed as an Exsiccata of the Gray Herbarium. The two collections cited above are the first records of this species from Las Villas province. UNIVERSITY OF CONNECTICUT, SToRRS, CONNECTICUT A ND BIOLOGICAL LABORATORIES, HARVARD UNIVERSITY. 1953] COZZO, STRUCTURE OF PITS IN CERCIDIUM 187 THE STRUCTURE AND DIAGNOSTIC SIGNIFICANCE OF CRATERIFORM BORDERED PITS IN THE VESSELS OF CERCIDIUM Dominco Cozzo THE PECULIAR PITS in the vessels of Cercidium australe Johnston were first described and figured by Tortorelli & O’Donell (1937), who considered them to be ‘“‘vestured pits” such as occur throughout most of the Legumi- nosae with the exception of the Bauhineae, Bailey (1933). In connection with my investigations (1950, 1951) of Argentine Leguminosae, I noted certain unusual characteristics of these pits which led me to believe that they merited detailed reinvestigation. The vessels of C. australe are studded internally with projections which resemble miniature volcanic cones. Each of these projections contains a craterlike cavity that extends from its apex through the thick secondary wall of the vessel into the chamber of a bordered pit. According to the terminology adopted by the International Association of Wood Anatomists (1933), an extended opening through a thick secondary wall — which pro- vides a means of communication between the lumen of a cell and the chamber of a bordered pit — is called a “pit canal.” A pit canal has an ‘inner aperture” that opens into the lumen of the cell, and an “outer aper- ture” that leads into a “pit chamber.’’ Thus, the projections in the vessels of C. australe are not a form of vesturing, but are excessive inward exten- sions of localized parts of the secondary wall which surround the pit canals. True vesturing ! of the bordered pits in C. australe is confined largely to the rim of the outer aperture of the pit canal. The detailed structure of this aberrant type of bordered pit in C. australe, Fig. 1, A and B, differs from the usual type in the following respects. In the case of tracheids and vessels with thin secondary walls, the area of the wall which jackets the pit chamber is embossed inwardly beyond the general contour of the wall which surrounds the lumen of the cell. With increasing thickness of the secondary wall and reduction in size of the pit chamber, this embossing effect is submerged and concealed. In very thick-walled vessels and fiber tracheids, having circular bordered pits, the outer aper- ture of the pit canal tends to be circular, but of conspicuously smaller diameter than the circular outer contour of the pit chamber. The pit canal flares toward the lumen of the cell by an enlargement of one of its diameters, and the inner aperture usually is more or less narrowly elliptical or slit- like. In the vessels of C. australe, on the contrary, the inner aperture of 1In my opinion, the term “ornate” is preferable to “vestured.” In any case, the Spanish term “orladas” as applied to this type of structure should be changed to “ornadas,” the correct translation of both “vestured” and “ornate.” 188 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv ‘ A ¢, 9%, °, Se te 4 4 - iz) e. id e. ‘@ : © é TEXT-FIGURE 1. Crateriform bordered pits in surface and sectional views. (A) Sectioned parallel to the long axis of the outer aperture of the pit canal. (B) Sectioned at right angle to (A). (a-a) Contour of pit chamber, (b-b) con- tour of outer aperture, (c-c) contour of inner aperture, (d-d) coincident diam- eters of inner and outer apertures, (e) vestured rim of outer aperture. the extended pit canal is small and circular. Furthermore, the pit canal flares outwardly, being broadly elliptical at the level of its outer aperture. TAXONOMIC CONSIDERATIONS Crateriform bordered pits occur in the vessels of the first-formed, as well as the later-formed, secondary xylem of the stem. Therefore, it is possible to study their occurrence in small twigs from herbarium speci- mens. Their presence or absence in material obtained from the Arnold 1953] COZZO, STRUCTURE OF PITS IN CERCIDIUM 189 Arboretum (AA), Gray Herbarium (GH), Museo Argentino de Ciencias Naturales (BA), Yale Forestry School (YF) and the Wood Collection of Harvard University (HU) is as follows: CRATERIFORM Pits PRESENT Cercidium australe Johnston: Argentina, Mendoza, Mexia 4377 (GH); Argentina, La Rioja, Cozzo (BA, 52235); Argentina, Salta, Cozzo (BA, 52722), Venturi 9507 (AA). Cercidium praecox (R. & P.) Harms: Argentina, Jujuy, Ledesma, Venturi 5343 (GH); Peru, Lambayeque, J. West 3576 (GH); Equador, Loja, Hitchcock 21331 (GH); Venezuela, Pittier 12945 (AA), Pittier 1928 (YF, 12458); Venezuela, Llavo, Curran and Haman 1251 (GH); Mexico, Sonora, La tinajo, Hartman 241 (GH); Mexico, Sonora, Abrams 13287 (GH). CRATERIFORM Pits ABSENT Cercidium andicola Gris.: Argentina, Jujuy, DeCarles (BA, 27/1102); Argentina, Jujuy, Maimara, Lorentz & Hieronymus 746 (GOET, TYPE); Argentina, Jujuy, Humahuaca, Schreiter 11085 (GH); Bolivia, Toldos bei Bermejo, Fiebrig 2493 (GH). Cercidium floridum Benth.: Mexico, Nuevo Leon, 7. C. and E. M. Frye 2391 (GH); Mexico, Sonora, Wiggins and Rollins 272 (HU, 25775); Mexico, Sinaloa, Gentry 7016b (GH); U.S., Arizona, Pringle 1881 (AA). Cercidium macrum Johnston: Mexico; Victoria, Tamaulipas, Palmer 125 (GH); U.S., Texas, Palmer 12303 (AA). Cercidium microphyllum (Torr.) Rose & Johnston: U.S., Arizona; Brass 14360 (GH); US., California, Epling, Haines and Stewart 1933 (AA); Pringle 1882 (AA); HU 9678. Cercidium molle Johnston: U.S., Gulf of California, Johnston 3877 (AA Cercidium peninsulare Rose: U.S., Gulf of California, Carmen Island, Johnston 3802 (GH). Cercidium sonora Rose & Johnston: Mexico, Sonora, Abrams 13280 (GH). Cercidium texanum Gray: U.S., Texas, Buckley 1881 (AA). The constant occurrence of crateriform bordered pits in two species of Cercidium, and their absence in eight other species, provides a diagnostic character of considerable significance; one which may be utilized by taxono- mists in any future revision of the genus and its species. The character is so peculiar and unusual that it is indicative of close relationship between C. australe and C. praecox. It serves to differentiate these species sharply from C. andicola, as well as from North American representatives of the genus. 190 JOURNAL OF THE ARNOLD ARBORETUM _[voL. xxxiv ACKNOWLEDGMENTS I am indebted to Professor I. W. Bailey — in whose laboratory I have been working as a Fellow of the John Simon Guggenheim Foundation from Argentina — for aid in translating this paper from Spanish into English. I am also indebted to the Directors of the Arnold Arboretum, the Gray Herbarium, the Museo Argentino de Ciencias Naturales, the Systematisch-Geobotanisches Institut Goettingen, and the Yale Forestry School for the privilege of studying material in their collections. LITERATURE CITED Bartey, I. W. (1933). The cambium and its derivative tissues VIII. Structure, distribution and diagnostic significance of vestured pits in dicotyledons. Jour. Arnold Arb. 14: 259-273. Cozzo, D. (1950). Anatomia del lefio secundario de las Leguminosas Papilio- noideas eiaereoere Rev. Mus. Arg. Cien. Nat. 1: 223-361. . (1951). Anatomia del lefio secundario de las Leguminosas arias y Cesalpinioideas Argentinas. Rev. Mus. Arg. Cien. Nat. 2: 63-146. ie Association of Wood Anatomists (1933). Glossary of terms used in describing woods. Tropical Woods 36: 1-12 ToRTORELLI, L. A. AND C. A. O’DONELL (1937). or punctuaciones orladas de “Cercidium praecox.”” Rev. Arg. Agron. 4: 197-201. JOURNAL OF THE ARNOLD ARBORETUM VoL. XXXIV JuLty 1953 NUMBER 3 PLANTAE PAPUANAE ARCHBOLDIANAE, XXI * THE PAPUASIAN SPECIES OF MACARANGA Lity M. PERRY In 1876 THE GENUS Macaranga Thouars was first reported from New Guinea by F. Mueller. He published a single species, M. aleuritoides, in his Descriptive Notes on Papuan Plants. As exploration continued on the island other species were collected. Seven species and two varieties were recorded by Warburg as additions to the Papuan flora in Botanische Jahr- bicher, 13: 349-352. 1891. The first comprehensive flora of the north- east region of New Guinea was published by K. Schumann & Lauterbach, Fl. Deutsch. Schutzgeb. Siidsee, 1900. This contained fourteen species of Macaranga. In 1910 and 1912 J. J. Smith, Nova Guinea (Botanique), added several species from Netherlands New Guinea. The first compre- hensive work on the genus after that of Miller Argoviensis in DC. Pro- dromus 15(2): 987-1016, is that of Pax & K. Hoffmann in Das Pflanzen- reich, 63 (IV. 147. VII). Here, there are twenty-four species recorded from New Guinea. Five years later, 1919, in Pflanzenreich, 68 (IV. 147. XIV), Pax & K. Hoffmann added twenty more to the previously known species from that region. These were based chiefly on the large collections of Ledermann from the Sepik (Kaiserin Augustafluss-Expedition) River region. Since then only occasional new species have been reported. Mean- while the collections of the Archbold Expeditions have been accumulating. These, with a considerable loan from the Bogor Herbarium, the Carr ma- terial from the British Museum, and the Papuan Macaranga from the Brisbane Herbarium, and that of Lae, form the basis of the work here presented. It brings together in one place the species of the Papuasian region. The study is admittedly incomplete. In determining specimens of a variable and dioecious genus, such as Macaranga, it is important to have material from both the staminate and the pistillate plants. Unfortunately, all too often this fact has not been recognized, and, as a result, species have been described from collections representing a single sex. It is diffi- cult indeed to match a pistillate unknown with the description of a stami- * Botanical Results of the Richard Archbold Expeditions. See Jour. Arnold Arb. 32: 369-389. 1951. 192 JOURNAL OF THE ARNOLD ARBORETUM _ [VoL. xxxiv nate specimen, and even more so vice versa. This same difficulty presents itself when a key is to be drawn for the species. Another difficulty is the lack of sufficient material to estimate the variability within a species with any degree of accuracy. A question one often ponders is — what constitute good specific characters in this genus? Some of the findings which I con- sider pertinent are recorded below: Stipules: The stipules seem to be fairly constant in size, outline and pubescence. Since usually they are shed rather quickly, good specimens should have at least two terminal buds. In several species the stipules are connate forming a single scale covering the terminal bud in the axil of the uppermost leaf. Such stipules leave a ring-like scar. In all species reported in this article as having connate stipules, a stipule has been removed from a bud for verification. There are other species which apparently bear large leaves clustered at or near the apex of the branchlets; here the stipules are large and it is not easy to determine whether they are connate or separate. However, if one is so fortunate as to have a part of the branchlet showing stipular scars, it may be noted that when the stipules are in pairs broad enough to surround the young branchlet, one is slightly higher than the other at the point of meeting, so the scar is slightly uneven at that point. Leaves: In the group of species with palmately veined leaves, the char- acter of lobed versus entire leaves is often unsatisfactory and one to be used much more cautiously than it has been in the past. Great variation is also encountered in the size and width of leaves even on the same branchlet. Staminate inflorescence: Only a few staminate inflorescences are simple; most are panicles. The bracts subtending the branchlets should possibly be regarded as a negligible character, as these in most instances appear to have fallen from the herbarium specimens. However, the bracts subtending the glomerules are persistent and, although showing variation, are reasonably reliable. The presence or absence of patelliform glands appears to be quite constant, although the glands may vary in size, shape and position. Within limits the number of stamens tends to be a good character, but several dissections are necessary to establish the number accurately and it is preferable to have more than one gathering to work upon. Most anthers are four-locular, but in a few species both three-locular anthers and four-locular anthers occur in the same flower. Again in some species three-locular anthers are constant, while in one Macaranga two- locular anthers are found. In at least seven species the calyx is lobed only to the middle, although in most species the sepals are free or the calyx is tripartite. Pistillate inflorescence: The bracts appear to fall very quickly from the pistillate inflorescence; however, if several of them can be seen, they offer a certain parallel with those found in the staminate inflorescence. The calyx is distinctive, at least for closely related groups, but this too is displaced by the growing flower and seldom observed. The unilocular char- acter of the ovary appears:‘to be dominant in certain species, but in M. ovatifolia Merr. from the Philippines, and here reported from the Solomon Islands, the bilocular ovary is almost as frequent as the unilocular ovary on 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 193 the same specimen. Occasionally specimens have both smooth capsules and capsules irregularly marked with small tubercles. The majority of the species have capsules with processes. Any outgrowth which appears longer than broad is here regarded as a process. In M. induta the processes are broad in proportion to their length and obtuse. Some species have rather fine and relatively long, almost filiform, processes. On the other hand, M. caudata Pax & K. Hoffmann is marked by very short processes, almost short enough to be termed tubercles. However, they have minute hairs at the apex similar to those found along the long processes in other species. In M. clemensiae and M. fragrans the base of the process is broad. An- other feature of the processes to be noted here is that in some species they tend to slough off as the capsule matures, so that the capsule, unless care- fully observed, sometimes appears smooth. Pubescence: The type of pubescence rather than the quantity is here regarded as a criterion. The terminal bud, the lower surface of the leaves, and the inflorescence usually have the same type. Often pubescence rubs off, but that found on the terminal bud and the inflorescence is fairly con- stant. In the key given below I have not attempted to follow the phylogenetic order of Pax & K. Hoffmann, but rather to use those superficial characters which might aid in the identification of future collections. I am deeply grateful to the directors and curators of the various herbaria who have given me the privilege of studying their material. I am indebted to Dr. E. D. Merrill for the photographs of two types and carbon rubbings of leaves of several other species not represented here. I also express my appreciation for the kindness of Dr. Taylor in having Mr. Stearn check a type-specimen in the British Museum for me. The institutions from which material for this study was used are indicated as follows: A — Arnold Ar- boretum; BM — British Museum; BO — Bogor Herbarium; BR — Bris- bane Botanic Museum and Herbarium; F — Herbarium Universitatis Flor- entinae; G— Gray Herbarium; LAE — Department of Forests, Lae, New Guinea; MICH — University of Michigan; NY— New York Botanical Garden KEY TO THE SPECIES A. ees connate (the terminal bud with a single scale). acts oe the ¢ glomerules not patelliform-glandular on the upper surfac M. belensis. BB. Bracts eubentie the 3 glomerules patelliform- glandular on the upper surface .. Leaves conspicuously bullate. ... . 2. M. eymae. CC. ae not bullate, or if bullate only observed with the aid of a lens. D. Capsule smooth: leaves lanceolate or ovate-lanceolate. ye & ate hee aly Se es Gee ores 9-4 daca by Re ON reiteriana. DD. Capsule processed (not known in M. pleytei); leaves ovate, elliptic or lobed. E. Anthers 2-locular. ......... 4. M. cucullata. EE. Anthers mostly 3- locular, 4- locular ones aes een oc- curring. 194 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv F, nha peltate; pubescence on the branchlets ca. 4 mm. ere Pere ee ee ee ere re. 5. M. pleytei FF. cae not peltate; pubescence on the branchlets not G. Leaves biauriculate at the apex of the petiole. H. Bracts subtending the 3 glomerules linear, dilated at the apex and bearing a single orbi- cular patelliform gland; bracts subtending the 2 flowers lanceolate or linear, narrowed at the base; calyx of the 2 flower 6-8 mm. long; styles 0.8-2 cm. long. ........... ee Tee eee errr 6. M. aleuritoides. HH. Bracts subtending the ¢ glomerules 3-lobed, the two basal lobes rounded, the middle lobe suborbicular and patelliform-glandular; bracts subtending the 2 flowers ovate; calyx of the 2 flower hardly 4 mm. long; styles ca. 0.5 Oi WO 2 6c a ceva nig 7. M. papuana. GG. Leaves not auriculate at the apex of the petiole. H. Leaves elliptic-ovate, somewhat narrowed towards the apex and base, 3-nerved at the base; stipules 1.5-3 cm. long; sepals of the é flower about 0.5 mm. long; stamens with 3-locular anthers; calyx of the @ flower about 3.5 mm. long. ....8. M.. tessellata. HH. Leaves ovate or subrhombic-ovate or lobed, palmately 5-nerved; stipules 5-11 cm. long; sepals of the ¢ flower 1 mm. long; stamens with 3-locular (some with 4-locular) anthers; calyx of the 2 flower 1 cm. or more long. Ree eGo 54:8 ekS ee BeOS 9. M. platyclada. AA. Stipules separate (the terminal bud with two scales). sie ood Lagrapacien glandular on the upper surface (2 inflorescence un- n M. astrolabica, M. bullata, and M. stenophylla, & inflorescence ber in M. advena). C. Capsules with processes. D. Leaves broadly (0.5-6.5 cm.) peltate. E. Leaves bullate. ...................05, 17. M. bullata. EE. Leaves not bullate. F. Stamens with 3-locular — stipules 2— : cm. long; leaves 1.5—6.5 cm. peltat ......10. M. fallacina. FF. Stamens with 4-locular eee stipules less than 2cm G. Stamens 15 or 16. ........ 12. M. stenophylla. GG. Stamens not more than 12. H. Leaves 1.3-2.5 cm. peltate; capsule 2-locular with 1 long (S-7 mm.) processes. Or ee ere ll. M. salomonensis. HH. Leaves 0.4—0.8 cm. peltate; capsule 1-locular with shorter (up to 2.5 mm.) processes. I. Stipules lanceolate or lance-ovate, 1-1.7 cm. long; inflorescence velutinous; bracts 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 195 subtending @ flowers subfoliaceous; style 6-9 mm. long. ........ 27. M. similis. II. Stipules linear- lanceolate or linear-ob- long, 0.4-0.7 cm. long; inflorescence fer- rugineous- ae bracts subtending 2 oo not foliaceous: style 4 mm. 29. M lon punctata DD. rigoe apeltate or only slightly (0. 4 0. a cm. n.) peltate. E. Leaves 3—5-nerved or at least 3-nerved at the base. F. Ovary 2-locular (in M. strigosa, var. carrii sometimes 3-locular). G. Stipules 3—5.5 cm. long; capsules with very short (0.2-0.4 mm.) processes, ...... 15. M. caudata. GG. Stipules not more than 2 cm. long; capsules with processes 1.5—4.5 mm. long (not known in M. warburgiana). H. Calyx of @ flowers urceolate or subtubular and denticulate, not splitting into almost even lobes; leaves not bullate except in M. strigosa. I. Stamens 11-18. ....13. M. rufibarbis. II. Stamens 2-8. J. Stamens 2-4 with 3- or 4-locular anthers, mostly 3-locular. K. Leaves cordate at the base; petioles fulvous-setose with re- trorse hairs; : Paes Non pea a ae M. strigosa. KK. Leaves cere at mA base; petioles glabrous; ¢ panicle iC) Oe ae or erTe 18. M. warburgiana. JJ. Stamens 4-8 with 4-locular anthers. K. Leaves deeply bifoveate at the base; ¢ bracts thickish and cucullate or ere Saale 8 Me ce fears, fa ok 19. M. bifoveata. KK. Leaves with 4 flat glands at the base; é bracts ovate, acutish or obtusish, not usually thick- ened. ...... 20. M. latifolia. HH. Calyx of @ flowers conic, apparently split- ting into 2 or 3 regular lobes; leaves more or less bullate. ..... 21. M. glandulifera. FF. Ovary 1-locular (processes often caducous as fruit matures, G. Bracts of the 2 inflorescence foliaceous; pubes- cence velutinous or Ses -pilose, not crisply tomentose nor setu H. Bracts sbtending the ¢ oer dentate; stamens 3 or 4. ........ M. densiflora. Z3: Py dalechampivides. JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv HH. Bracts subtending the ¢ glomerules entire; 2 I. Pistillate bracts lanceolate or linear- lanceolate, entire. ...24. M. urophylla. II. Pistillate bracts ovate or triangular-ovate denticulate or dentate J. Stipules narrowly linear, 4-10 mm. long; leaves rounded, or truncate and retuse, or very slightly cordate at the base. K. Style 7-10 mm. long; capsule with processes 1.5—2.5 long. ....25. M. involucrata. KK. Style 4-6 mm. long; capsule with processes hardly 1 mm. long. ..26. M. schleinitziana. JJ. Stipules lanceolate or lance-ovate or slightly obovate-oblong; leaves cor- date at the base. 27. M. similis. GG. Bracts of the @ inflorescence not foliaceous; pubescence crisply tomentose or setulose H. Capsule with very short (0.4-0.6 om pineoseee pubescence setulose; stamens 2 0 28. M. eerie. HH. ae with processes about 2 mm. long; pubescence crisply tomentose; stamens 3—10 29. M. punctata. EE. Leaves pinnately nerved. apsule 2-locular; stamens 2-4; @ bracts linear- lanceolate, narrowed towards the base. Spa aed 3-45 Hd haces 14. ; decipiens. FF. Capsule 1-locular; stamens 1 15 (not known in M. na G. eaves ovate- ghey or narrowly linear apex ong; pules ovate; @ bracts rhombic-ovate and oe at the base, patelli- form-glandular within the margin. raga M. advena. GG. Leaves ianceniate; acuminate; stipules linear- subulate; 9 bracts, at least the lower ones (some- times the upper small and lacking glands), ovate with entire or dentate apex, reflexed, and bearing near the base 2 large oval ae age bans pee PRE Oo Poa e be Rees Oe ustifolia, CC. Capsules smooth or marked irregularly with sees ee (not known in M. astrolabica D. New growth and sndorescendé glabrous or inflorescence slightly puberulous; leaves lanceolate, subcoriaceous; capsule smooth, glabrous. E, Stamens 2. ...................005; 32. M. astrolabica. EE. Stamens 10-16. ............ .. 54. M. gracilis. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 197 DD. New growth and inflorescence brownish sublanate- haere . eaves ovate, rather stiff; capsule tomentulose, smooth metimes with a few tubercles. 33. M. sterephyla, BB. Bracts ae paenane glandular on the upper “surface (2 unknown in M. seheeieeue ha M. hoffmannii, M. kostermansi, M. longicaudata, tad womersleyi; & unknown in M. inermis, M. ee and M. magnifolia) : anthers usually 4-locular. C. Leaves obviously bullate; capsules with processes. D. Leaves pea bullate, iy elevations on the upper surface about 5 mm. long, 2-3 mm. broad. ...... 34. M. clemensiae. DD. ie more finely ie the elevations on the upper surface not more than 3 mm. long, 1-2 mm. broad. E. Leaves caudate-acuminate, the acumen almost linear and 2-3 cm. long; venation clos F. Staminate inflorescence Gndluding the flowers) densely cana tomentose; lower surface of the leaves densely , the veins villous. ... 35. M. womersleyt. FF. Secvusie. inflorescence hirtellous; flowers glabrous; lower surface of the leaves not tomentose, all the veins hirtellous. — . 36. M. longicaudata. EE. Leaves gradually or subabruptly acuminate, but not caudate; venation not so clos F. Leaves 0.5-1 cm. ee ovary 1-locular Be ete Chek: e- 9c « _ leonardii. FF. Leaves rounded or slightly “cordate at the base, scarcely peltate; ovary 2-locular. 38. M. carri. CC. Leaves not or only very slightly bullate; capsules smooth or with processe Dp, Cae peltate. E. Leaves broadly (1.5-20 cm.) peltate; stipules medium to large (1-13 cm. long). F. Stamens with 3- napa anthers; a bilocular, with processes 1.5-3 mm. long. ....10. M. fallacina. FF. Stamens with 4-locular ne capsules 2—5-locular. G. Stipules large, 4-13 cm. long; leaves large, 34— 76 cm. long; capsules with short (0.5-2 mm.) tomentulose processes, or tuberculate, 2- o 3-locular; ee (where known) tomentulose on both s H. Capsules densely ae with short (0.5- 2 mm.), tomentulose process or ee oe 39. fragrans. HH. Capsules densely granulo- gene ree and somewhat irregularly tuberculate; tubercles 0.5-0.7 mm. high, 1-1.5 mm. broad, pube- scent. 40. M. ipsa GG. Stipules smaller, 1-3 cm. long; leaves 7-33 c long; capsules with long (3-15 mm.) ae processes, 2—5-locular; subtending bracts glabrous or somewhat velutinous H. Bracts subtending the ¢ glomerules ovate, deeply concave and entire or sinuate; stamens 198 JOURNAL OF THE ARNOLD ARBORETUM [ VOL. XXXIV 2 or 3; ovary (3-) 4-5-locular; leaves usually bearing 1-4 maculate elands on the basal nerves between the insertion of the petiole and the lower margin 41. M. quadrislondulosa, HH. Bracts subtending the é glomerul bicular-ovate, pectinate-lacinulate or dentate; stamens 4-14; ovary 2-3-locular; leaves usually without maculate glands . bi bth Sunde Out pd de oe ow oH : rius EE. Leaves narrowly (not more than 1 cm.) peltate; stipules usually less than 1 cm. lon es . Bracts subtending the ¢ glomerules deeply concave; stamens 2 or 3; capsules with slender processes. . 41A. M. guadriglandulosa, var. variabilis. _ Bracts sending the & glomerules only slightly concave; stamens 3-15; capsules smooth or slightly and se mc be ie processed only in M. induta. G. Branchlets with new growth glabrous; 4 inflo- rescence glabrous (only pubescent part of plant mentioned is the sparsely pilose midrib on the lower surface of the leaves); primary veins 2-4 pairs above the basal ones. . 43. M. hoffmannii. Branchlets with new growth and_ inflorescences tomentose; primary veins 5-9 pairs above the basal ones. H. Ovary 2- or 3-locular, mostly 2-locular. I. Leaves at maturity almost glabrous on the lower surface; capsules mostly 1-locu- lar, rarely 2-locular. 44. M. versteeghii. II. Leaves at maturity tomentose on the c G @ J. Stamens 3-5, anthers with a few brownish wrinkled hairs attached to them; pubescence of the lower surface of the leaves uneven under a lens. 45. M. trichanthera. JJ. Stamens 10-15, anthers glabrous; felt) under a lens; capsule eoueaeg or smooth. .46. M. induta HH. ps 1-locular, rarely 2- loc ular iz elliptic-ovate, almo st glabrous ms maturity; stipules oblong-lanceolate. yea ahs woe AA, II. Leaves subrhombic- or suborbicular- ovate, somewhat tomentose on the lower surface; stipules linear-lanceolate. 47. M. albescens. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 199 DD. Leaves apeltate. E. Stamens 20 or less F. Capsules with processes; stamens 2 or 3. G. Bracts subtending the ¢ glomerules entire H. Bracts ovate, 3-5 mm. long, deeply concave, abrous. 3 .o eae eek oe eee M. Re er var. variabilis. HH. Bracts densely hairy outside, about 1.3 mm. long, acuminate, somewhat clasping the rachis 48. M. nova-guineensis. GG. Bracts cohtanuding the ¢ glomerules dentate, uberulous 49. M. subpeltata. FF. Capsules sMootl: ‘staméneee 17 G. Bracts subtending the ¢ glomerules dentate or lacinulate. H. Bracts dentate with short wages mm.) teeth; ovary 1- or 2-locular. 50. M. ovatifolia. HH. Bracts lacinulate with: laciniae 1—2.5 mm. I. Ovary 1-locular. ...52. M. Rciessin II. Ovary 2-locular. ... .51. en iell GG. Bracts subtending the 46 glomerules e H. Lower surface of the leaves pane "wath rusty brown tomentum. ..... as Lah ME ae e603 34.42 M._ trichanthera. HH. Lower surface of the leaves not tomentose. eaves sparsely pilose or setulose on the upper surface, short-villous beneath; 53 — - Leaves glabrous on ‘the upper surface and mostly glabrous beneath; stamens 5-17. J. Stamens 10-17. K. Leaves caudate-acuminate with the acumen 1.5-3 cm. long, 3-nerved at the base with 4-6 additional pairs of primary veins above the basal ones, not maculo- cae on the upper K a Ee oO 2 a oO n ° o oO & es ce io) c wn < Q je) oe ? the upper surface at the base ener oe 55. M. haplostachya. JJ. Stamens 5-9. K. Leaves at the base suddenly contracted to about 2—2.5 mm broad and extended down- 200 JOURNAL OF THE ARNOLD ARBORETUM _ [vot. xxxiv d 2.5-3 mm. to join the petiole. 56. . lanceolata, KK. Leaves obtusely cuneate or L. Leaves very small, 2.5- 4.5 cm. long, 0.9-1.3 cm. broad; 4 inflorescence simple or sparsely ~~ . M. kostermansi LL. ae larger, 6-1 long, 2-11 cm. oan 4 inflorescence branching from near the base (known ee in the varieties). 58. M. inermis. EE. Stamens 32-59. ............... 59. M. pleiostemona. 1. Macaranga belensis sp. nov. Arbor 10-12 m. alta; ramulis novellis, foliis novellis petiolis, stipulis, et inflorescentiis molliter pilosis vel subvillosulis, pilis 1.3-2 mm. longis; laminis firme chartaceis, 5-13 cm. longis, 2—5.5 cm. latis, ovatis, sensim longe acuminatis, acumine 1—2.5 cm. longo, versus basin angustatis, basi retusis vel emarginatis et supra maculari-2—6-glandulosis, maturis supra glabris olivaceis, subtus pallidioribus costa venisque paulum pilosis, utrin- que flavo-granulari-glandulosis, inconspicue trinerviis vel penninerviis, venis primariis utrinsecus costa 6—9 oblique adscendentibus subtus prominulis et in axillis crispe pilosulis, venis secundariis et rete manifestis; petiolo 1-3.5 cm. longo; stipulis connatis cito caducis 1.1—2.3 cm. longis; inflorescentiis é simplicibus 5—7.5 cm. longis (incl. basi 1-2.5 cm. nudis) dissite glom- erulo-florigeris, bracteis extus dense pilosis, intus non patellari-glandulosis, ovatis inferioribus longe acuminatis ca. 7—10-floris; floribus ¢ pilosis et granulari-glandulosis; calyce tripartita, lobis circiter 2 mm. longis; stamini- bus 11-24, antheris 4-locularibus; floribus ¢? ignotis. NETHERLANDS NEW GUINEA: common in secondary forest, Bele River, m. NE. of Lake Habbema, alt. 2200 m., Brass 11212 (type, 6, A), Nov 1938 (tree 10-12 m. high). The basal glands of the leaves are below the two basal lateral veins, which extend only about one-third of the length of the leaf. These are not more prominent than the other lateral veins, but in the larger leaves the basal laterals support secondary veins more remote than the upper laterals, hence suggesting three-nerved leaves. 2. Macaranga eymae sp. nov. Habitus ignotus, probabiliter frutex vel arbor; ramulis teretibus apicem versus subadpresse villoso-hirsutis, pilis leoninis; laminis firme chartaceis, 13-30 cm. longis, 3.8-11.7 cm. latis, lanceolatis vel anguste ovatis sensim 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 201 acuminatis, acumine 1.5—3 cm. longo, basi breviter cordatis, supra prope apicem petioli glandulis maculiformibus 4-6 ovalibus instructis, margine integris vel undulatis, novellis dense villoso-hirsutis, maturis glabrescenti- bus, supra in costa tantum, subtus costa et venis et venules villoso-hirtellis, supra minute bullatis, pinnatinerviis vel basi breviter trinerviis, venis primariis utrinsecus costa 18-25 oblique adscendentibus prope marginem arcuatim conjunctis utrinque perspicuis; petiolo 1.5-5.5 cm. longo sub- adpresse hirtello; stipulis connatis, 3-6.5 cm. longis, cito caducis, adpresse villoso-hirsutis; inflorescentiis hirtellis; ¢ 7—9 cm. longis, 2.5-3 cm. pedun- culatis, ramis inferioribus ad 5 cm. longis adscendenti-patentibus, bracteis sub ramos lineari-lanceolatis caducis, bracteis florigeris persistentibus e basi dilatata angustatis deinde in laminulam fere semiorbicularem intus patelliformi-pluriglandulosam protractis, supra basin ca. 2 mm. longis, 5-7-floris; calyce ¢ sub anthesin fere 1.5 mm. longo, sepalis 3 ovatis, staminibus 3—7 (saepe 4), antheris 4-locularibus; 9 simplicibus, pedunculo 3-5 cm. longo, apice flores 1 vel 2 proferentibus, bracteis oblongis basin versus petioliformi-angustatis, circiter 1 cm. longis, parte superiore intus patelliformi-pluriglandulosis; floribus @ ca. 1 cm. pedicellatis, calyce tubu- loso, 6.5 mm. longo, parte inferiore subglobosa ovarium arcte cingente, ca. 5 mm. diam., parte superiore brevi truncata, deinde rumpente et deciduo; ovario globoso dense adpresse molliter echinato, hirtello, 2-loculare, stig- matibus 2 linearibus plumosis circiter 6 mm. longis; capsula immatura sine echinis 9 mm. diam., 6 mm. alta, echinis numerosis subulatis leviter curva- tis pilosis, fere 5 mm. longis. NETHERLANDS NEW GUINEA: Wissel Lake region, SW. corner of Lake Paniai, Camp Moeie, alt. 1750 m.. Eyma 4966 (3 & 2, A. Type; BO), July 28, 1949. 3. Macaranga reiteriana Pax & K. Hoffmann in Pflanzenr. 85 (IV. 147. XVIII) : 185. 1924. Tree 6-20 m. tall; young parts (branchlets, petioles, young leaves, stipules, and in a lesser degree inflorescences) brownish golden-hirtellous with hairs 0.7—1 mm. long and antrorsely subappressed; the slender branch- lets glabrate; stipules connate and caducous, 2.5—5 cm. long; petiole 1—2.5 cm. long; leaf-blades firmly chartaceous or thinly coriaceous, 4.5-10 cm. long, 1.5—4 cm. wide, lanceolate or ovate-lanceolate, entire and acuminate, obtuse at the base, above near the attachment of the petiole dotted with 2-6 small flat dark glands, glabrous above except for the sparsely hairy midrib, beneath more densely hairy on the midrib and veins and densely punctate with brownish depressed minute glands, 8-10 primary veins on either side of the midrib, with secondary veins and reticulations easily seen on the upper surface but inconspicuous below; the simple ¢ inflorescence 4—6 cm. long with scattered glomerules of flowers, the bracts subtending the glomerules about 2 mm. long with shallow concave base about 1 mm. deep, elongated on the dorsal side to form a shortly spatulate lobe with the apex patelliform-glandular (many minute glands) within and about equal- 202 JOURNAL OF THE ARNOLD ARBORETUM _ [voL. xxxiv ing the narrowed portion in length; ¢é flower with 3 or 4 sepals 0.5-0.8 mm. Jong and granulo-glandular; stamens 3 or 4, the anthers 3- or 4-locular; ? flower (not seen) solitary and bibracteate at the apex of the peduncle; calyx fusiform enclosing the densely glandular ovary, the style plumose. NETHERLANDS NEW GUINEA: occasional on slopes in primary forest, 15 km. SW. of Bernhard Camp, Idenburg River, alt. 1510 m., Brass & Versteegh 11977 (3, A), Jan. 1939 (tree 19 m. tall, diam. 35 cm., crown not wide-spread- ing; flowers grey; bark 5 mm. thick, black, fairly smooth; sapwood rose, heartwood red-brown). This staminate collection agrees well with the original description; the only differences I find are that none of the leaves in the collection cited is broader than 2.7 cm., whereas the maximum width given in the original description is 4 cm., and that the anthers are as often 3-loculed as 4-loculed, not mostly 4-loculed. These features are only indicative of variation, and not in any way of specific value. The terminal bud of this species is long and very narrow (3 or 4mm.), much like that of M. longestipulata (Kurz) Muell. Arg. from Ternate in the Moluccas. Unfortunately only the pistil- late form of the latter has been collected, and apparently from the original description of M. reiteriana only the very young pistillate flowers of this are known. As between the two species there are no processes on the ovary, whereas they are characteristic of the fruit of M. longestipulata. 4. Macaranga cucullata J. J. Smith in Nova Guin. 8: 237, t. 60. 1910; lic. 789. 1912. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 369. 1914, Tree up to 8 m. high; young branchlets, petioles, peduncles, and branches of the inflorescence yellowish hirsute with (sometimes stiffish) hairs 1-6 mm. long, quickly glabrescent or glabrate; stipules connate, caducous, up to 13 cm. long, appressed hirsute (sometimes sparsely so) on the outer surface; petioles 8-43 cm. long; leaf-blades large and subcoriaceous, 18-63 cm. long, 14-44 cm. broad, 3—5-lobed or entire and elliptic-ovate with the apex of entire leaves or lobes abruptly and narrowly 1—4 cm. acuminate, distantly callose-dentate on the margin, broadly rounded on either side of the deeply cordate base and bearing at the sinus 2 narrow and rounded- undulate auricles with a few flat oval glands on the upper surface, strongly 3-nerved at base with 12-16 pairs of prominent primary veins above the basal ones, the secondary veins almost parallel; young leaves densely lanate-hirsute on the upper surface, later glabrate, densely and minutely dark-glandular beneath with the venation appressed hirtellous; ¢ inflores- cence paniculate, 13-17 cm. long, the peduncle 3—5 cm. long; primary bracts stipule-like, caducous; bracts subtending the glomerules attached to the rachis forming a cup-like base 1-1.5 mm. wide and deep, the free part an oblong or spatulate-obovate lobe 1-1.5 mm. long, pilose outside, bearing a single orbicular patelliform gland within; ¢ flowers short- pedicellate; sepals 3 or 4, ovate, concave, 0.5—0.7 mm. long; stamens 3-7 with 2-locular anthers; ? inflorescence short, 3—3.5 cm. long, few-flowered, 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 203 hirsute; capsule depressed saree globose, densely glandular, with pubescent subulate processes 2—2.5 mm. long. NETHERLANDS NEW GUINEA: pave forest, North River, Versteeg 1064 (4, BO, type), May 1907; primary forest, North R iver, Von Roemer 431 (2, BO), Oct. 1909; common in seral rain forest on edge of river floodplain, Bernhard Camp, Idenburg River, alt. 50 m., Brass 13826 (46, cl on 1938 (tree 6-8 m. high with very large cordate leaves about 55 & 44 ¢ PAPUA: occasional in subseral rain forest on river bank, Palmer River, 2 miles below junction Black River, alt. 100 m., Brass 7256 (6, A), July 1936 (small tree 3 or 4 m. high with erect branches; leaves 63 X 39 cm., red beneath; panicles red, anthers white). The only @ collection is Von Roemer 431. The 6 inflorescences of the other specimens are so much alike that I feel sure that they belong to a single species. It should be noted, however, that the leaves of the Brass collections are entire and larger, whereas those of Versteeg 1064 and Von Roemer 431 are 3—5-lobed mostly. In view of the leaf variation shown in the excellent series of specimens belonging to M. aleuritoides F. Muell., cited above, the presence or absence of lobes cannot be regarded as a specific character. Brass 7256 is more nearly glabrous than the other specimens; on the other hand Brass 13826 is more hairy (with irritant hairs) than any of the others; here, even on the upper surface of the leaves, are scattered longish hairs, and the upper branchlet is almost hispid. 5. Macaranga pleytei sp. nov. Arbor parva 8 m. alta; ramulis teretibus novellis setulosis, setulis paten- tissimis flavescentibus ca. 4 mm. longis; laminis novellis, petiolis. stipulis et pedunculis dense hirtello-setulosis, setulis saepissime patentissimis 2.5—4 mm. longis et pilis 0.2—0.5 mm. longis intermixtis; laminis chartaceis 6.5— 14.5 cm. longis, 4.5-12 cm. ee orbiculari- ovatis abrupte acuminatis, acumine 0.5—1 cm. longo, basi 0.6—1.9 cm. peltatis, supra in nervis inter insertionem petioli et marginem basalem glandulis maculiformibus 4—6 in- structis, in sicco margine revolutis, glabrescentibus, maturis supra sparsim schulasis et interdum leviter rugnlost: subtus minute cerino- -glandulosis etiam costa et venis et venulis hirtellis, palmatinerviis, venis primariis supra basalibus utrinsecus costa 7—9 oblique adscendentibus prominulis, secun- dariis et rete perspicuis; petiolo 4.5-11 cm. longo; stipulis connatis cito caducis, 3—5 cm. longis, setulis subadpresse adscendentibus; inflorescentiis g paniculatis, 11-18 cm. longis, pedunculo 8.5-11 cm. longo, rhachi et ramis hirtellis, ramis inferioribus 2.5-4 cm. longis; bracteis sub ramos breviter triangularibus acuminatis, bracteis florigeris ca. 13-floris, glom- erulos non superantibus rhachidem semiamplectentibus semiorbiculari-ovatis apiculatis, parte superiore intus patellari-pluriglandulosis; floribus ¢ sae- pissime alabastris, extus pauciglandulosis; calyce tripartita; staminibus 4, antheris 3-locularibus; floribus @ ignotis. 204 JOURNAL OF THE ARNOLD ARBORETUM __ [VoL. xxxiv NETHERLANDS NEW GUINEA: Misool Island: in thinned forest near Fakal, alt. 40 m., Pleyte 1117 (é, A, Type; BO), Oct. 7, 1948 (tree 8 m. tall, 6 cm. diam., flowers yellow). Although the pistillate plant of this species is not known, the connate stipules, the 3-locular anthers, the leaf-venation, and the somewhat coarse pubescence suggest a relationship with the section Dimorphanthera. It is readily distinguished from the other species of this section by the peltate leaves 6. Macaranga aleuritoides F. Mueller, Descr. Notes Pap. PI. 1876. — Pax & Hoffmann in Pflanzenr. 63 (IV. 147. VII): 366. ie Macaranga riparia Engler in Bot. Jahrb. 7: 463. 1886; Forschungsr. Gaz. 35, . 11, 1888. — K. Schumann & Lauterbach, Fl. Deutsch. Schutzgeb. Stidsee 397. 1900. — J. J. Smith in Nova Guin. Bot. 8: 235. 1910; lc. 789. 1912. Tree up to 20 m. high; young branchlets, stipules, petioles, peduncles, and branches of the inflorescence puberulous and appressed-pilose with hairs 0.6-1.7 mm. long, quickly glabrate; stipules connate, caducous, 3.5—9 cm. long; petioles 1-38 (usually 5-15) cm. long; leaf-blades often large and subcoriaceous, 13-45 cm. long, 7-42 cm. wide, sometimes entire and ovate but usually with 3 erect-spreading and narrowly acuminate lobes, the middle one triangular-ovate and much larger than the oblique, mostly entire (occasionally bilobed) lateral ones, inconspicuously callose-dentate on the margin or entire, truncate to cordate or in a few instances rounded at the base and bearing at the apex of the petiole 2 rounded-undulate auricles with a series of 1—4 flat oval glands above often extending to the blade, strongly 3-nerved at the base with 10-18 less prominent pairs of primary veins obliquely spreading-ascending and shortly arcuate just within the margin and joined throughout by distinct and almost parallel second- aries, the latter connected in turn by finer reticulations; the upper surface of the young leaves cinnamon-colored tomentose becoming glabrate, the lower remaining permanently puberulous and rather closely dark glandular- punctate; ¢ inflorescence paniculate, 7.5-24 cm. long including the com- pressed 1-6 cm. peduncle, the primary bracts stipule-like and caducous, the bracts subtending the glomerules 2-5 mm. long, linear, dilated at the apex and here bearing a single orbicular gland within; ¢ flowers almost sessile, the sepals 3, ovate, the stamens 5-15, the filaments exserted, the anthers 3-locular; 92 inflorescence mostly branching (sometimes simple), 5.5-29 cm. long with the compressed 3—8 cm. peduncle, the primary bracts stipule-like, the secondary ones variable, lanceolate or linear and narrowed at the base, within variously dotted along the margin or towards the apex with orbicular glands, outside pilose; calyx of @ flower pilose, 6-8 mm long, subglobose at the base and cylindrical in the upper part with a trun- cate apex, quickly deciduous, the plumose-papillose 2 or 3 styles 0.8—2 cm. long and recurved at the apex, the subglobose capsule 2- or 3-locular, ca. 1.5 cm. diam., densely glandular and bearing minutely hairy subulate processes 3 or 4 mm. long. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 205 MOLUCCAS: Key Islands, West Elal, Jensen 342 (6, A), May 1922. NETHERLANDS NEW GUINEA: Pionier bivouac, alt. about 50 m., Docters van Leeuwen 9373 (4, A, BO), June 1926; same camp, Van Eechoud 104 (sterile, BO), Nov. 1939; Albatros bivouac, alt. 50 m., Docters van Leeuwen 9629 (2, BO), July 1926; near Prauwen bivouac, alt. 90 m., Lam 1037 (6 & 2, BO); Hollandia, alt. about 20 m., Gjellerup 310 (6, BO), Aug. 1910; swampy forest, Ransiki, Vogelkop, alt. 10 m., Kostermans 2895 (6, ae Aug. 1948; on edge of rain forest Nabire, Kanehira & Hatusima 11421 (9 BO), Feb. 1940; common in rain-forest a growths on ground aa pes Bernhard Camp, Idenburg River, alt. 50 m., Brass 13823 (6 & 9, A), April 1939 (tree 6-8 m. high; material from a 8 and @ trees); ene Camp, alt. 50 m., Meyer-Drees 428 (2, BO), Mee 1938. Misool: in thinned forest at Fakal, alt. 50 m., Pleyte 960 (4, A, BO), Sent. 20, 1948 (tree 8 m. high, 10 cm. diam.: flowers dirty yellow) ; west of Fakal, Pts 30 m., Pleyte 1124 (6, A, BO), "Oct. 1948 (tree 10 m. high, 10 cm. diam.). Aroe Islands: primary forest and thickets, Poelau Wokam, Tadiinnetae Buwalda 5044 (é, A, BO), 5046 (2, A, BO), May 1938; same locality, Neth. Ind. For. Serv. bb.25314 (9, A), May 1938. PAPUA: Palmer River, 2 miles below junction Black River, alt. 100 m Brass 7284 (2, A), July 1936 (dioecious tree 10 m. high; this and Homalanthus species the dominant species in rain-forest second growths); plentiful in seral shrubberies on muddy river banks, Oroville Camp, Fly River (30 miles above D’Albertis Junction), Brass 7435 (6, A), Aug. 1936; plentiful on river banks flooded by tides, Lower Fly River, east bank opposite Sturt Island, Brass 8186 (3, A), 8187 (@, A), Oct. 1936; on old farm lands, Buna, Lane-Poole 164 (6, BR), July 1922; forest, ase i about 450 m., Carr 12819 (6, A, BM), July 1935; rain- forest second growth, Kubuna, alt. 100 m., Brass 5636 (@, A, NY), Dec. 1933; Thu, Vailala i. ' Brass 904 (9, A), Feb. 1926; Bomgwina River, Brass 1622 (9, A), June 1926. NORTHEAST NEW GUINEA: mountain bush, Boana, alt. about 1050 m., Clemens 8306 (&, A), June 1938; Boana, alt. 750 to 1350 m., Clemens 41465 (2, A), May—Nov. 1940; mountain bush above Markham River, vicinity of Kajabit Mission, alt. about 600 m., Clemens 10893 (8, A), Aug.—Dec. 1939; Bulu, Schlechter 16078 (4, A), May 1907; Kenejia, alt. 150 m., Schlechter 18435 (8, A), Oct. 1908; forest bank of Kerame River, 25 miles above Sepik River, Herre 331 (?, BO, NY); on rain-forest margin, Yalu, White, Dadswell & Smith NGF 1648 (@, A, BR, LAE); in brown sandy loam on alluvial flat on bank of Munim Water, Yalu, alt. about 30 m., 2nd Austral. For. Surv. Co. NGF 275 (92, A, BR, LAE), July 1944; between Lae and Yalu, White, Dadswell & Smith NGF 1519 (@, BR), July 1944; near Aitape, Smith NGF 1216 (@, A, BR, LAE), Jan. 1945. BISMARCK ARCHIPELAGO: New Britain: in Nodup area, Gazelle Peninsula, Waterhouse 237 (9, NY), May, June 1934. Duke of York Is., without further locality, Bradtke 339 (sterile, BR), 1918. SOLOMON ISLANDS: Bougainville: Torokina, Robinson & Volk ae 818 (°, A, BR, LAE), Jan. 1945; Kieta, alt. 80 m., Kajewskt 1576 2, A), Mar. 1930; Kugu-maru, Buin, alt. 150 m., Kajewski 1804 (9, A, BO), June 1930; Karngu, Buin, Kajewski 2227 (6, A), Oct. 1930. San Gristo- bal: Waimamura, Brass 2579 (6 & 9, A), Aug. 1932. 206 JOURNAL OF THE ARNOLD ARBORETUM _ [VOL. xxxIv This common species of second-growth forest shows great variability in the size of the leaves and the length of the petioles; a few leaves are entire, and some of the very large ones have the lateral lobes somewhat bilobed, although they are only 3-nerved at the base (cf. Gjellerup 310, Docters van Leeuwen 9373, White, Dadswell & Smith NGF 1648, Clemens 10893, Schlechter 16078, Herre 331, and Brass 13823). The species is readily recognized by the three-lobed leaves with small rounded auricles close to the base of the petiole, the long apical bud covered by the connate stipules, the linear bracts (with an orbicular gland at the apex) subtending the glomerules of 4 flowers, and the three-locular anthers. The single specimen cited from the Moluccas is the only one that I have seen from outside the Papuasian area. 7. Macaranga papuana (J. J. Smith) Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII) : 368. 1914. Macaranga hispida var. papuana J. J. Smith in Nova Guin. Bot. 8: 234. 1910. Macaranga ovalifolia Ridley in Trans. Linn. Soc. II. Bot. 9: 148. 1916.— _ Hoffmann in Pflanzenr. 85 (IV. 147. XVII): 186. 1924. Tree 7-17 m. high, 30-32 cm. diameter; branchlets tomentulose- puberulous, tardily glabrescent; stipules connate, caducous, 2—5.5 cm. long, sericeous outside; leaf-blades chartaceous, 7.5-18 cm. long, 5.5—-15 cm. wide, ovate with entire margin, and acuminate at the apex, rounded or truncate or slightly cordate at the base and bearing at the apex of the petiole 2 rounded undulate auricles with 1-3 oval glands above, pinnately veined (3-nerved at the base) with 8-12 primary veins obliquely ascending from either side of the midrib, the secondary venation distinct on the lower surface, glabrous above except for the puberulous midrib, minutely dark- glandular and crisply puberulous beneath on all veins; young leaves densely short-pilose with hairs scarcely 1 mm. long; petiole 2—10 cm. long, tomentu- lose-puberulous like the branchlets; 4 inflorescence 10-24 cm. long (in- cluding peduncle 1.5-3 cm.), tomentulose-puberulous or pilose, sparsely 2-6-branched with 2 or 3 long slender branches bearing dense many- flowered glomerules about 5 mm. apart; bracts subtending the glomerules 3-lobed, pilose on the outside, with the two lateral lobes rounded and the dorsal lobe suborbicular and patelliform-glandular within; ¢ flowers pu- berulous and sometimes granulo-glandular on the exposed part, the pedicels hairy, 1 mm. long; calyx 0.5-0.7 mm. long, 3-parted; stamens 4-8 with 3-locular anthers; @ inflorescence simple, immature, few-flowered; axis 2.5-5 cm. long, the bracts subtending the flowers ovate and patelliform- glandular within; calyx sparsely pilose, hardly 4 mm. long, truncate at the apex, deciduous, the ovary 2-locular, the 2 styles about 5 mm. long, pubes- cent on the dorsal surface and densely papillose above; capsule about 1 cm. diameter (probably immature), granulo-glandular and rather closely covered with pubescent processes 2—4 mm. long. NETHERLANDS NEW GUINEA: Merauke River, Jaheri s. n. (6 & 2, BO, type), Mar. 25, 1901; Camps I to III, Wollaston Exped., Kloss s, n. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 207 (type of M. ovalifolia Ridl., 6, BM; photo A); Babo, Matatula 19 (Neth. Ind. For. Serv. 6b.21815) (9, BO); Hollandia, Bernhard Camp, about 50 m. alt., Meyer-Drees 471 (Neth. Ind. For. Serv. bb.25718) (6, A, BO). PAPUA: Central Division, Mageri near Sogeri, G. Angell NGF 4197 (6, A). This species is undoubtedly closely related to M. hispida (Blume) Muell. Arg., but it is readily distinguished by the auricles and the lack of a deep sinus so characteristic of the Moluccan species. Also it is less hispid. 8. Macaranga tessellata Gage in Nova Guin. Bot. 12: 481, t. 186. 1917. — Pax & K. Hoffmann in Pflanzenr. 85 (IV. 147. XVII): 184 (sphalm tesselata). 1924. Tree about 16 m. high; apices of the branchlets, young leaves, stipules, and petioles densely hirtellous with hairs 0.4-1 mm. long, soon glabrate; stipules connate, caducous, 1.5—3 cm. long; petioles 1.7—3.5 cm. long; leaf- blades chartaceous, 13—18.5 cm. long, 6.5—-10 cm. broad, elliptic-ovate with undulate or slightly dentate margin, somewhat narrowed at both ends, rather abruptly 1-1.5 cm. acuminate at the apex and shallowly cordate or rounded at the base, on the upper surface near the attachment of the petiole bearing 2 small oval glands or one or none, glabrous above except for the pubescent midrib and primary veins, hirtellous beneath on the midrib and veins and strewn with minute dark glandular dots, 8 or 9 primary veins (fairly prominent beneath) on either side of the midrib above the 3-nerved base, the secondaries prominulous and the reticulation visible to the naked eye; ¢ inflorescence in a few-branched panicle (only young ones in specimens studied) with pilose (hairs 0.5 mm. long) axis, the glabrous ¢ flowers glomerulate in the pubescent axis of the ultimate bracts, the latter about 2 mm. long with 2 basal lobes forming a shallow saucer-like cavity and the dorsal part elongating into a rounded or tri- angular-ovate lobe densely patelliform-glandular within; sepals about 0.5 mm. long, the stamens 3, with 3-locular anthers; @ inflorescence (not seen) short and bearing a few sessile crowded flowers; calyx about 3.5 mm. diameter and irregularly 3—5-lobed, tomentose on the outer Surface, the styles 2; capsule compressed-globose, bearing yellowish tomentose subulate processes about 5 mm. long. NETHERLANDS NEW GUINEA: North River near Bivak Island, Pulle 132 (4, BO, TYPE). This species, associated by Pax & K. Hoffmann with M. nova-guineensis J. J. Sm. in the section Echinocarpae, has four characters in common with the section Dimorphanthera; they are: connate stipules, 3-locular anthers, the elongate dorsal part of the bracts subtending the staminate glomerule patelliform-glandular within, and fruits with processes. The stipules were originally described as lanceolate and caducous, inferring that they are free. In the two specimens at hand the stipules are found only on the ter- 208 JOURNAL OF THE ARNOLD ARBORETUM _ [VoL. xxxiv minal buds. I have detached one of the buds and soaked it to verify the fact that its cover is a single scale rather than two. 9. stra 1 gs Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV): 3 Macaranga roemeri Pax & K. Hoffmann in Pflanzenr. 85 (IV. 147. XVII): 185. 1924. Tree to 26 m. high, 35 cm. diameter; young branchlets, young leaves, petioles, stipules, and inflorescences densely velvety-pilose or substrigose with hairs 0.8—1.2 (— 1.8 on stipules, and appressed) mm. long, glabrescent; stipules connate, caducous, 5-11 cm. long; petioles 8-16 cm. long; leaf- blades subcoriaceous, 11.5—33 cm. long, 7-22 cm. broad, entire or undulate and ovate or subrhombic-ovate or with 3 erect and narrowly acuminate lobes, with the middle lobe broadly triangular and up to 8 cm. acuminate and separated from the narrow laterals by wide sinuses, very shallowly cordate or truncate or rounded at the base, without basal glands on the upper surface, palmately 5-nerved with 7—9 ascending pairs of primary veins above the basal ones, becoming glabrescent above and dotted with scattered minute yellow glands, also sometimes minutely bullate, shortly pilose on all veins beneath and sprinkled with yellow glands, the second- ary venation and reticulation distinct; bracts enclosing é inflorescence stipule-like but not connate; panicle up to 22 cm. long, the compressed peduncle 4-7 cm. long, branches opposite, or 2 or 3 of different lengths clustered at a single node and usually subtended by lanceolate or linear acuminate bracts, the latter 0.5—-2.5 cm. long and bearing several patelli- form glands (in one instance the lowest branches subtended by a pair of lanceolate leaves 10 cm. long and 2.5 cm. wide with acumen 3 cm. long and petinie 5 mm.); the glomerules subtended by obtusish and oblong or slightly obovate bracts up to 5.5 mm. long, biauriculate at the base, patelli- form-glandular and pubescent on the upper surface; ¢ flowers with 3 granulo-glandular and pubescent sepals 1 mm. long, the stamens 4—7 with 3- and 4-locular anthers; 9 inflorescence simple, up to 15 cm. long, bearing apparently sessile flowers at 2 nodes but fruit on pedicels 5—9 mm. long, each node subtended by bracts with the petiolar bases 5 mm. long, the lower node with lanceolate acuminate bracts up to 2 cm. long, the upper node with bracts oblong or ovate and obtuse about 1 cm. long, all patelli- form-glandular and pubescent on the upper surface; calyx of the @ flower densely pubescent and minutely glandular, 1.1-1.6 cm. long, globose at the base and about 4-5 mm. across, then somewhat narrowed into a tube bearing at the apex 5 narrow tapering lobes as long as the tube (some- times lobes longer and tube very short), quickly deciduous; ovary 2- or 3-locular, the plumose papillose 2 or 3 styles with recurving apices up to 1.8 cm. long (including the 4—-4.5 mm. connate base) and dorsally pubes- cent; fruit immature, densely granulo-glandular and covered with slender pubescent processes about 2 mm. long. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 209 NETHERLANDS NEW GUINEA: frequent in rain forest, alt. 1630 m., Brass & Versteegh 11936 (6, A), Jan. 1939; in rain forest at 1700 m. alt., Brass 12276 (4, A), Jan. 1939; frequent in rather open rain forest on sides of ravines, alt. 1600 m., Brass 12365 (2, A), Jan. 1939; ee a tree on landslips in rain forest, alt. 1200 m., Brass 12782 (@, A), Feb. 1939. With only the original description as a basis for comparison, I am unable to distinguish between M. platyclada and M. roemeri, although Pax & Hoffmann have assigned their two species to different sections. The col- lections above cited sometimes have lobed and entire leaves on the same specimen, illustrating the fact that, in palmately nerved leaves, entire leaves versus lobed leaves are not distinctive characters. In the specimens cited above, the upper surface of the leaves is softly pubescent and tends to be sparsely so. Although most of the anthers were 4-locular, some 3-locular anthers were also found. 10. Macaranga a Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV): 31. 1919. Small tree 5-10 m. tall, apices of branchlets, stipules, and very young leaves ferrugineous tomentose-pilose; the stipules lanceolate, acuminate, 2-3 cm. long, free and caducous; petioles 6—28 cm. long, glabrescent; leaf- blades thinly coriaceous, 14-36 cm. long, 11-29 cm. wide, orbicular- or triangular-ovate with entire or sinuate margin, acuminate, the acumen 1—2.5 cm. long, rounded then truncate at the base and 1.5—6.5 cm. peltate, palmately 7-nerved with about 12 slightly curved primary veins above the basal ones obliquely ascending from either side of the midrib, bearing on the upper surface of the basal nerves (close to the insertion of the petiole) 3—5 flat oval glands, glabrous above, distantly and minutely dark-glandular beneath, glabrescent on the veins, the secondary venation distinct on both surfaces; the 2—3 cm. pedunculate ¢ panicles 8-12 cm. long and up to 8 cm. broad, puberulous; bracts at the base of the branches obovate, acute, carinate, 1-2 cm. long, ferrugineous-tomentose, not patelliform-glandular; bracts subtending the many-flowered glomerules about 2 mm. long, con- tracted from a broad base into an ovate blade patelliform-biglandular within; sepals 3, ovate, acute, and sparsely granulo-glandular; stamens 3—6 with 3-locular anthers; the 16—23 cm. pedunculate (peduncle scarred 2—10 cm. below the apex) @ inflorescence simple (?), bearing 3 or more fruits at the apex; ovary 2-locular; styles very broad (5 mm. long, 4 mm. broad), pubescent on the lower surface and very densely plumose on the upper, the papillae being about 2.5 mm. long and somewhat flattened; capsules immature, about 1 cm. diameter after being soaked, densely granulo-glandu- lar, bearing deciduous pubescent processes 1.5—3 mm. long. NETHERLANDS NEW GUINEA: van Gelder River, alt. about 100 m., Docters van Leeuwen 9289 (6, BO). May 1926; Albatros Bivouac, alt. abou 60 m., Docters van Leeuwen 9341 (9, BO), June 1926; Motor Bivouac, alt. Ay ary 100 m., Docters van Leeuwen 11057 (6, A, BO), Nov. 1926; rain forest, Nabire, Kanehira & Hatusima 11476 (9, A, BO), Feb. 1940 (very rare). 210 JOURNAL OF THE ARNOLD ARBORETUM [| VOL. XXXIV The only authentic material which I have of this species is the carbon rubbing of the basal part of a leaf of an isotype which matches very well those of the collection cited above. It should be noted that in the staminate material which I have before me the apex of the bracts supporting the glom- erules is very elongate and narrow rather than ovate, as in the original description. There are one or two small glands on the margin of these, whether they are patelliform I cannot say. The stamens are three and anthers three-locular; in the original description the stamens are three to six and the anthers three-locular. I regard these as variations within a species; without actual material for comparison I cannot place this material elsewhere than in this species. 11. Macaranga salomonensis sp. nov. Arbor alta cortice cinereo; ramulis glabris; stipulis ovato-ellipticis 1.5 cm. longis, 0.9 cm. latis, concavis acutis, liberis fere glabris (apicem versus pilis paucis); petiolo 5. 5-11 cm. longo glabro; laminis orbiculari-ovatis, 8.5—18 cm. longis, 5.5—11 cm. latis, breviter et late acuminatis vel acutis, basi 1.3—2.5 cm. peltatis, palmati-8-nerviis supra basales venis primariis 5—8 paribus prominulis, secundariis distinctis, novellis utrinque dense fer- rugineo-lanato-tomentellis cito glabratis, maturis supra glabris subtus floc- coso-tomentellis (fere glabris) et glanduloso-punctatis; ¢ inflorescentiis 6-17 cm. longis, pedunculo 2—7 cm. longo, glabro, ramulis 1.5—4 cm. longis floccoso-tomentellis deinde glabratis, bracteis florigeris basi 1.5 mm. con- cavis deinde ca. 2 mm. recurvis et crassiusculis semiorbicularibus intus glandulis patellaribus + confluentibus ca. 25-floris, pedicellis vix 1 mm longis pubescentibus; sepalis 3 ovatis 1-1.3 mm. longis; staminibus 5-8, antheris 4-locularibus; @ inflorescentiis ca. 8 cm. longis valde maturis, tantum una capsula terminale reliqua, 2-loculari, tomentoso-puberula, 1.7 cm. lata, 1.5 cm. alta, 1.4 cm. crassa; echinis mollibus 5—7 mm. longis, Satin be 0.9 * 0.8 & 0.7 cm. unniosts SOLOMON ISLANDS: Ysabel: rain forests, Tiratona, alt. 600 m., Brass 3327 ($6 & 9, A, TYPE), Dec. 1932 (very tall gray-barked tree, dioecious). Bougainville: rain forest, Koniguru, Buin, alt. 800 m., Kajewski 2038 (6, A), Aug. 1930 (small tree up to 13 m. high; flowers on green stems sup- ported by stiff bracts). This species can be easily recognized by the practically glabrous branch- lets, the densely tomentose new small leaves with glabrous petioles, the thickish recurved bracts (with at times confluent patelliform glands on the upper surface) subtending the glomerules of staminate flowers, and the deeply peltate leaves. 12. Macaranga stenophylla Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII) : 371. 1914. Branchlets slender with young parts at first ferrugineous-tomentose then quickly glabrate; stipules not seen; petioles 4.5—7 cm. long, tomentulose; leaf-blades elliptic-ovate or oblong or oblong-lanceolate, acuminate, 12—18 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 211 cm. long, 6—7 cm. broad, rounded at the base and 0.7—1 cm. peltate, 3-nerved and shortly palmate-nerved, glabrous above and bearing several gland-like spots between the attachment of the petiole and the basal margin, reticulate beneath, also sparsely puberulous and granulo-glandular on the veins; 4 panicles ferrugineous-tomentose; glomerules many-flowered; bracts tomen- tulose, triangular at base, then contracted and produced into a lanceolate apex patelliform-glandular within; calyx granulo-glandular and tomentu- lose; stamens 15 or 16 with 4-locular anthers. The type, Schlechter 17724, was collected at an altitude of 1000 m., on Kani Mountains, Northeast New Guinea. Pax & K. Hoffmann note that only the young é@ plant is known, but believe that a significant character is found in the narrow and peltate leaves. 13. Macaranga rufibarbis Warburg in Bot. Jahrb. 16: 21. 1893. — K. Schumann & Lauterbach in Fl. Deutsch. Schutzgeb. Stidsee 397. 1900, in part. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VIT): 373. 1914. Tree with varying amount of pubescence; young parts rusty tomentulose but very soon glabrate; stipules narrowly ovate, acuminate, 1.5 cm. long, glabrous or with a few hairs towards the apex or occasionally pubescent along the middle of the dorsal surface, caducous; petioles 2—5.5 cm. long, glabrate; leaf-blades 7-18 cm. long, 4-12 cm. broad, triangular-ovate and long-caudate with the narrow apex up to 4 cm. long, at the base rounded and narrowly (2-4 mm.) peltate or truncate and very narrowly and shallowly cordate, 2—6-maculo-glandular on the upper surface below or adjacent to the attachment of the petiole, dark granulo-glandular beneath, 3-nerved with 6-9 pairs of prominent veins above the basal ones, the secondary venation also prominent, the new leaves loosely rusty-tomentose but glabrous at maturity or with remnants of the pubescence along the midrib and around the attachment of the petiole beneath; ¢ panicle 5-12 cm. long, rusty-pilose with hairs seemingly clustered at times around the base of the branches and bracts; bracts subtending the glomerules 2 or 3 mm. long, usually triangular-ovate and obtusely acuminate, 1—2- patelliform-glandular on the upper surface, hairy beneath and granulo- glandular; sepals 3, ovate, 1.5 mm, long, pilose and sparsely granulo- glandular towards the apex; stamens 11-18 with 4-locular anthers; inflorescence 5-12.5 cm. long, simple, loosely rusty-tomentose, then glabrate; flowers clustered near the apex and also at a node about 2—5 cm below the apex; bracts ovate, about 5 mm. long, 3-4 mm. broad, patelliform- glandular on the upper surface and pilose beneath, caducous; calyx appressed-pilose then glabrate, globose-pyriform, about 4 mm. high, truncate, splitting into 2 or 3 parts and falling; ovary 2-locular, granulo- glandular and covered with hairy processes which in fruit may be 2.5-3 mm. long, the styles about 3 mm. long, slenderly papillose or short-plumose ; capsule about 1 cm. broad. NORTHEAST NEW GUINEA: Wantoat, Clemens 40925 (6, A), Feb. 5, 1940; same locality, alt. 1050-1800 m., Clemens 11198bis (2, A), Feb. 28, 1940: 212 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. XxxIv mountain bush, Sarawaket, alt. 1200-1800 m., Clemens 8308A (2, A), June 15- 18, 1938 (tree 5 inches diameter; fruit garnet or wine-red) ; Samanzing, alt. 1650—- 1800 m., Clemens 9205 (2, A), Nov. 5, 1938; Sambanga, alt. 1500-1800 m., Clemens 7740 (2, A), Nov. 25, 1937 (tree or shrub 12-15 ft., with brown fruit). Clemens 7740 is much more pubescent than the other specimens, but I believe this indicates a newer growth. Only the staminate collection has narrowly peltate leaves, but the prominent venation and the pubescence are similar in all the collections cited. Macaranga rufibarbis var. tenella (Pax & K. Hoffmann), comb. nov. Macaranga tenella Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 379. 1914. This variety is characterized by the lanceolate or linear-lanceolate leaves 9-17 cm. long, 2-6 cm. broad, with rounded or broadly cuneate and minutely cordate base, 2- or 3-maculo-glandular on the upper surface near the attachment of the petiole; bracts beneath glomerules of flowers in the @ panicle dilated at the base, then narrowed a little and above broadened into a lanceolate apex 1—2-patelliform-glandular on the upper surface, beneath pubescent and granulo-glandular; sepals 2 or 3, stamens 8-10 with 4-locular anthers. NORTHEAST NEW GUINEA: Finisterre Mt., alt. 1300 m., Schlechter 19101 (4, A), January 1909. This collection differs from those cited for the species in being some- what more glabrous. However, wherever pubescence remains on the speci- mens it is apparently the same type in all. The leaves are narrower and, although not so stated in the original description, they are maculo- glandular. The bracts in the species do not show the narrowed part above the dilated base, a character noted in the specimen cited above; there are also somewhat fewer stamens here, although one cannot say that more material wouldn’t show overlapping numbers. 14. Macaranga decipiens sp. nov. Frutex 2-3 m. altus; ramulis, stipulis, foliis novellis, et axibus inflo- rescentiarum breviter hirtellis, pilis albis ca. 0.5 mm. longis; stipulis linearibus vel lineari-lanceolatis, 3—5 mm. longis, caducis; petiolo 1.5—3.5 cm, longo; ea oblongo- paige a variabilibus, 10.5 & 4 cm., 12 3cm., 15 & 7 20 & 9 cm., 25 & 7 cm., utrinque angustatis, apice als vel ann basi rotundato-cuneatis, undulatis, supra glabris, costa et venis breviter pilosis, subtus granuloso-glandulosis, costa et venis dense rete sparsim pilosis, penninerviis, venis primariis 9-12 paribus oblique patentibus deinde prope marginem curvantibus et anastomosanti- bus, rete laxo; ¢ paniculis pauci-ramosis gracilibus, usque 5.5 cm. longis; 6 floribus plerumque apice ramulorum dense confertis; bracteis sparsim pilosis, linearibus 2—2.5 mm. longis glomerulos excedentibus, basi dilatatis, apice leviter dilatatis et intus patellari-1-glandulosis; sepalis 3 vix 0.5 mm. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 213 longis, pauce granuloso-glandulosis, staminibus 2-4, antheris 4-locularibus; 2 inflorescentia 5-10 cm. longa, simplici; bracteis petiolulatis apicem versus lineari-lanceolatis, 6- 10 mm. longis, intus patellari-glandulosis; calyce non viso; ovario 2- loculari echinis pubescentibus tecto, stylis ca. 3—5 mm. longis, plumosis; capsula immatura, echinis usque 1 mm. longis. NETHERLANDS NEW GUINEA: rain forest, Waren, 60 miles south of Manokwari, alt. 20 m., Kanehira & Hatusima 13256 (6, A, TYPE; BO), Mar. 1940 (shrub 2 m. high, 13292 (2, A, BO) (shrub 3 m. high) ; coastal plain forest, Momi, Vogelkop, alt. 30 m., Kostermans 2829 (6, BO), Aug. 1948 (tree 2 m high); Klamono, Sorong, sea level, Pleyte 664 (2, A, ee 666 (6, A, BO), Aug. 1948 (shrub 2 m. high; many growing together 1 elon caudata Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV) : 30. 1919 Shrub 2-3 m. os tree 6-8 m.) high; young parts (branchlets, stipules, petioles, and peduncles) setulose with coarse hairs 1-2 mm. long mixed with fine hairs 0.1-0.2 mm. long; stipules 3—5.5 cm. long, ovate, long- acuminate; petioles 5-14 cm. long; leaf-blades slightly denticulate, 15-30 cm. long, 10-23 cm. broad, triangular- or orbicular-ovate, 3-5 cm. caudate- acuminate, truncate and minutely cordate at the base, inconspicuously maculo-glandular on the upper surface near the attachment of the petiole, pilose or villous-pubescent on both surfaces, particularly on the nerves and veins, and densely granulo-glandular beneath, 3-5 nerved at the base with 7 or 8 other pairs of ascending primary veins; the 4-10 cm. peduncu- late and shortly branched ¢ panicle 10-20 cm. long, bracts subtending the many-flowered glomerules dilated at the base then contracted into an oblong free part puberulous outside and irregularly patelliform-glandular within; calyx 0.5 mm. long, the 3 sepals obovate, cucullate, apiculate and sparsely granulo-glandular; stamens 3 or 4 with 4-locular anthers; @ in- florescence 22—40 cm. long, the flowers crowded on short pedicels on the upper 3 or 4 cm. at two or three nodes and subtended by oblongish bracts, pedicels 1.5 mm. long in flower, becoming 4 mm. in fruit, the subtending bracts about 1 cm. long, coarsely pubescent on the lower midrib and bearing small patelliform glands above, calyx about 2 mm. high and dentate at the apex, splitting once or twice, circumscissile near the base; ovary 2-locular, styles 6-8 mm. long, pubescent on the lower surface and densely plumose above; capsule 1.2 cm. broad, 0.9 cm. high, and 0.8 cm. thick, covered with sparsely short-hairy soft processes 0.2—-0.4 mm. long, the somewhat furrowed or rugulose seeds 5 mm. high, 6 mm. broad, and 3 mm. thick. NETHERLANDS NEW GUINEA: occasional along small streams in flood plain primary rain forest, 4 km. SW. of Bernhard Camp, Idenburg River, alt. 850 m., Brass 13361 (2, A), March 1939 (open tree 6-8 m. high; leaves sub- orbicular with long slender apex; branches, etc. irritant hairy; petioles and peduncles red). The smallest leaf of the above-cited collection is, in the long apex and ascending venation, a very good match for a carbon rubbing of a leaf of 214 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. XxxIv one of the isotypes of this species. It is to be noted, however, that the Brass material is not so pubescent (pilose rather than villous) as is described for the original; also, the stipules are obtuse and mucronate rather than long-acuminate. If I have interpreted this species correctly, it is much more closely related to M. chrysotricha Lauterb. & K. Schum. than to M. urophylla Pax & K. Hoffm., the former being characterized by stiff irritant hairs in the pubescence, few stamens, and the capsules with very short processes. 16. Macaranga strigosa Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV) : 30. 1919. Shrub 1.5—2 m. tall with stout fulvous-setose branchlets, tardily gla- brescent; stipules free, obliquely elliptic, 4-7 mm. long, appressed-villous on the outer surface; petioles 5-16 cm. long, fulvous-setose with retrorse hairs; leaf-blades bullate, orbicular-ovate, long-acuminate, 25-30 cm. long, 20-27 cm. broad, cordate at the base, remotely denticulate, glabrous above, densely granulo-glandular beneath and pilose along the nerves, densely reticulate-veined; ¢ panicle villous, 11-14 cm. long, the peduncle 7-9 cm. long; bracts subtending the branches about 3 scarlet, approximate, elliptic, 1-2.5 cm. long and about equaling the length of the branches of the inflorescence; bracts subtending the glomerules 3-lobed, the middle lobe extended and patelliform-glandular within, pilose outside; sepals 3 scarcely granulo-glandular, stamens 3 or 4 with 3- and 4-locular anthers; @ raceme to 15 cm. long, glabrescent, bearing small sterile bracts above the middle, 1—2-flowered at the apex; ovary 2-locular, densely covered with processes. The type of this species, Ledermann 10939, was collected on Hunstein Mt., Northeast New Guinea, alt. about 1350 m. Although I have seen no material which matches this description well, I have described below some specimens from Papua as a variety. This variety differs in the somewhat larger stipules, the smaller bracts of the ¢ inflorescence, and the lack of sterile bracts along the peduncle of the @ inflorescence, also in foliar characters. The leaves of M. strigosa are very large, orbicular- ovate, bullate, and densely granulo-glandular on the lower surface, while the leaves of the variety are smaller, triangular-ovate, not bullate, and very sparsely granulo-glandular beneath. Macaranga strigosa, var. carrii var. nov. Arbor 4-6 m, alta; ramulis et petiolis spice et pubescentibus, setulis 1 mm. longis et pilis tenuissimis ca. 0. m. longis intermixtis setulis patentibus vel retrorsis demum caducis; mais 0.8-1.5 c ia loads. ovatis, acuminatis, concavis, subpersistentibus dorso adpresse ce petiolo 4.5-11 cm. longo; laminis deltoideo-ovatis interdum lobulatis, 8-20 cm. longis, 6-14 cm. latis, apice longe acuminatis, acumine usque 2.5 cm. longo, basi subtruncatis vel repandis et emarginatis, dentatis vel denticulatis, novellis utrinque dense villosulis, maturis utrinque venis dense ceterum sparsim pilosis, pilis 0.3-0.5 mm. longis, basi supra eglandulosis 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 215 vel maculari-biglandulosis, subtus sparsissime granuloso-glandulosis, trinerviis, venis primariis 7—9 paribus supra basales, utrinque prominulis, subtus secundariis et rete distinctis; ¢ paniculis immaturis usque 12 cm. longis, pubescentibus et granuloso-glandulosis; pedunculo 9 cm. longo; bracteis florigeris ca, 2.5 mm. longis, basi bilobatis deinde contractis, apice in laminulam late ovatam intus patellari-glandulosam abeuntibus, 4 floribus apicem versus pauci-granuloso-glandulosis; sepalis 2, 0.5—0.7 mm. longis; staminibus 2 vel 3, antheris 4-locularibus; ? inflorescentia usque 28 cm. longa, fere villosula; @ floribus apice conferte dispositis; bracteis 4.5-8 mm. longis, triangulari-ovatis, basi angustatis, intus patellari- glandulosis; calyce urceolato, 1.5—2 mm. longo, glabrato; ovario 2-loculari, granuloso-glanduloso et molliter echinato, stylis 2 basi connatis, ca. 9 mm. longis, papillosis’ vel breviter plumosis; capsula 1 cm. lata, 0.8 cm. alta, fere matura, echinis ca. 1.5 mm. longis, + caducis. PAPUA: forest, Boridi, alt. about 1400 m., Carr 13231 (2, A, TYPE; BM), Sept. 1935 (tree about 20 ft. tall); same locality, alt. 1500 m., Carr 13362 (6, A, BM), Sept. 1935 (tree about 15 ft. tall); same locality, Carr 14623 (6, BM), Oct. 1935 (tree about 25 ft. tall); on debris of an old landslip, Mt. Tafa, alt. 2400 m., Brass 4862 (2, A), Aug. 1933 (small regrowth tree; branchlets, petioles and inflorescence reddish; nerves and veins reddish on the lower surface of the leaves; fruits on very long, slender, axillary peduncles; hairs irritant). i Sena aaa Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. iV eS, This species differs from M. strigosa in having leaves narrowly (‘‘fere 1 cm.’’) peltate, the ¢ inflorescence openly branched and more pyramidal, the calyx granulo-glandular, and the anthers 4-locular. The type is Ledermann 11852 collected on Schrader Mt., Northeast New Guinea. The above is the complete description of this species given in a footnote under Af. strigosa which also has bullate leaves. This is a meagre list of characters to use for the determination of a species in a genus as large and varied as Macaranga. Unless an isotype is extant somewhere which can be studied further to elaborate additional characters, I believe the species must be rejected and its name regarded as a nomen subnudum. 18. Macaranga warburgiana Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII) : 347. 1914. Macaranga cuspidata Warburg in Bot. Jahrb. 13: 351. 1891; non Boiv. ex Baill. (1860). Shrub with glabrous branchlets; stipules free, lanceolate, acuminate, about 1 cm. long, caducous; petiole 5-17 cm. long; leaf-blades 14-25 cm. long, 10-16 cm. broad, deltoid-ovate, abruptly acuminate, truncate at the base, bearing 4 flat glands on the glabrous upper surface near the attachment of the petiole, yellowish granulo-glandular beneath, 3-nerved with 5 or 6 pairs of primary veins above the basal ones, the secondary venation distinct but not conspicuous; ¢ inflorescence granulo-glandular, 216 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. xxxIv shorter than the leaves (one in the isotype about 8.5 cm. long), bracts subtending the glomerules 1.5—-2.5 mm. long, ovate or the free apex lanceolate, acuminate or acute, patelliform-glandular within; sepals 3, about 1 mm, long; stamens 2 or 3 with 3- rarely 4-locular anthers; 2 inflorescence often longer than the leaves, on a pubescent peduncle; bracts lanceolate, granulo-glandular outside, patelliform-glandular within; calyx urceolate, truncate, glabrous, 2 mm. long; ovary pilose, 2-locular, styles papillose. NORTHEAST NEW GUINEA: Sattelberg, Warburg 20510 (4, probable ISOTYPE, A), 1889. This is a very fragmentary specimen with only a single terminal bud remaining; this bud is definitely wrapped in two stipules; further, it is to be noted that Warburg in his comment on the species stated that the stipules were not connate, hence his species could not belong to § Dimor- phanthera. I wish to emphasize this character as Pax & K. Hoffmann established a new section, § Warburgianae, for this species and one of the characters mentioned for both section and species is the connate stipules; however, Pax & K. Hoffmann do not indicate that they have seen Warburg’s specimen. For the most part the specimen is glabrous and granulo-glandular, but here and there, if one looks carefully with the aid of a lens, are a few short, fine, setulose hairs. The 9 inflorescence is described as having a pubescent peduncle. In a pocket on our sheet is a small tip, 1 cm. long, from a @ inflorescence, which is finely pubescent along the axis; it also has a few of the same type of setulose hairs as were found on the specimen. The flowers are very young. One is 2.5 mm. long with an urceolate glabrous calyx 1 mm. long, and the two styles protrude beyond the calyx about 1.5 mm. Another flower is larger with the calyx 2.5 mm. long. When the calyx is removed the styles are 4 mm. long and the ovary 1.5 mm. high. The ovary is pubescent and granulo- glandular, with four or five small spots which look like incipient processes. Either belonging to this species or very closely allied to it are two collections from Papua. They differ as follows: plants densely setulose on the new growth, later glabrate; leaves openly cordate and dentate; 6 inflorescence longer or shorter than the leaves; anthers all three- locular, but the same number of stamens. There is a great variation in leaf-size here: 21 & 18.5 cm., 10.5 & 6 cm., 9.5 K 6 cm PAPUA: secondary forest, Yodda River, he about 1350 m., Carr 13920 (6, A, BM), Dec. 1935; Isuarava, alt. about 1200 m., Carr 15481 (6, A, BM), Feb. 1936 (shrub 8 ft. tall). 19. Macaranga bifoveata J. J. Smith in Nova Guin. Bot. 8: 790, t. 139. 1912. — Pax & K. Hoffmann in Pflanzenr, 63 (IV. 147. VII): 377. 1914. Shrub or tree with slender, slightly puberulous and granulo-glandular branchlets; the broadly triangular stipules 2-3 mm. long, caducous and 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 217 somewhat pubescent; petioles 4-12 cm. long, at first granulo-glandular and puberulous, later glabrate; leaf-blades triangular-ovate, 12—22 cm. long, 7-16 cm. broad (the smaller ones elliptic, 6-15 cm. long, 3—8 cm. broad), somewhat abruptly short-acuminate, at base subtruncate or broadly cuneate or sometimes openly cordate, at apex of petiole very shortly cordulate and deeply glandular-bifoveate, denticulate, at first villous- tomentose above and tomentose beneath, gradually becoming glabrous above except for the puberulous nerves, puberulous also on the nerves and veins beneath and pale yellow granulo-glandular, 3-nerved, with 5—9 pairs of primary veins above the basal ones, secondary venation below easily seen with the naked eye; the loosely branched ¢ panicle 5-17 cm. long, pubescent and granulo-glandular becoming glabrate; peduncle 3—4.5 cm long; bracts subtending glomerules about 1.5 mm. long, spreading or recurved, semiorbicular and obtusely apiculate, cucullate with the patelli- form glands of the upper surface strongly impressed; pedicels about 0.7 mm. long; calyx 3- or 4-parted, 0.7—0.9 mm. long and puberulous; stamens 4—8 with 3- and 4-locular anthers; @ infructescence 5-10 cm. long with 2 or 3 very short branches, bracts 5—5.5 mm. long, ovate but narrowed at base as if 1 mm. petiolate, puberulous and granulo-glandular, concave, with the patelliform glands of the upper surface strongly impressed and somewhat confluent; @ calyx (fide P. & H.) fusiform, denticulate and tomentulose; ovary 2-locular; the plumose styles two, 2~3.5 mm. long; fruit 2-locular, granulo-glandular, and densely covered with short puberu- lous processes about 2.5—3 mm. long. NETHERLANDS NEW GUINEA: in primary forest on the Begowri River, alt. about 170 m., Gjellerup 209 (4, TypE, BO), June 1910 (tree 3 m. high) NORTHEAST NEW GUINEA: on the way from Ramu to the coast, Schlech- ter 14113 (4, BO), Jan. 1902; in ee forest at Kelel, alt. about 150 m., Schlechter 16211 (3, A), 16212 (one leaf, A), June 1907; margin of forest, Sattelberg, alt. about 900 m., Clemens 438 (9 , A), Oct. 1935 (tree 30 ft. high) ; Wantoat (Wantot), alt. 1050-1800 m., Clmans 11062 (2, A), Jan. 1940. The bifoveate glands at the base of the leaves and the deeply impressed somewhat confluent patelliform glands of the bracts of the inflorescences are characters which render this species easy to determine. In Clemens 438 cited above an immature ? bud 3 mm. long is oblong, practically glabrous and granulo-glandular, the calyx looks as if it would be 3- or 4-lobed, the ovary is about 1 mm. long and definitely bears processes though very small ones at this stage of development. 20. Macaranga latifolia sp. nov. Arbor 4-6 m. alta; ramulis puberulis vel glabris dense granuloso- glandulosis; stipulis sla hris ciliolatis, triangulari-ovatis 3.5-5 mm. longis; petiolo 6-19 cm. longo, praecipue apicem versus minute puberulo; laminis latissime ovatis 13—22 cm. longis, 13.5—26 cm. latis, undulato-denticulatis, apice anguste acuminatis, acumine 1.5—-3 cm. longo, basi truncatis vel 218 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. XxxIVv subtruncatis, ad petioli apicem minute auriculatis et maculari-glandulosis, supra puberulis vel glabratis, subtus puberulis et atro-granuloso-glandulosis, trinerviis, venis primariis supra basales 5 vel 6 paribus prominulis, secundariis distinctis; paniculis ¢ usque 17 cm. longis, fere glabris et dense granuloso-glandulosis; bracteis florigeris basi concavis et fere lobatis, ovatis acutiusculis vel obtusiusculis, 1-2 mm. longis, intus patellari- glandulosis; alabastris parvis 0.7 mm. longis, sepalis 3 vel 4 apice granuloso- glandulosis, staminibus 4—7, antheris 4-locularibus; inflorescentiis @ usque 7 cm. longis, simplicibus flores apicem versus gerentibus; bracteis florigeris extus puberulis, ca. 4.5 mm. longis, subrhombico-spathulatis, intus patellari- pluriglandulosis, calyce subtubulato, 2—2.5 mm. longo, puberulo, ovario 2-loculari, stylis 2-3 mm. longis, plumosis; capsula sine echinis 1 cm. alta et lata, echinis 3—4.5 mm. longis, pubescentibus. NETHERLANDS NEW GUINEA: thinned forest behind Kp. Baroe, along path, Sorong, Pleyte 412 (6 & 2, A, Type; BO), July 1941 (small tree about 4 m. high, 4 cm. diam., common; flowers light green to white; fruit green); Horna, Atasrip 41 (2, BO), Feb. 1903; in rain forest, Nabire, Kanehira & Hatusima 11673 (2, A, BO), Feb. 1940; in fringing forest, between Ayerjat and Slieber, 40 km. inland from Nabire, Kanehira & Hatusima 12655 (6, A, BO), March 1940 (plant 6 m. high). This species is unquestionably close to M. bifoveata J. J. Smith; in fact I have wondered whether the two are phases of a single species. They differ chiefly in the type of glands characteristic of the base of the leaves and of the floral bracts. In M. bifoveata the two basal glands are foveate, and those of the bracts are deeply impressed and fairly large, while in M. latifolia there are four small flat basal glands, and those of the bracts are also small; in the former, too, the bracts appear to be thicker or fleshier than in the latter. Smith’s species is more pubescent than the one here described, and the stamens have a mixture of 4- and 3-locular anthers. 21. Macaranga glandulifera sp. nov. Frutex arborescens 5—6 m. altus; ramulis novellis et petiolis dense hirtellis, pilis 1-1.5 mm. longis; stipulis triangulari-ovatis, 4-5 mm. longis, 2.5-3 mm. latis, acutiusculis, adpresse pubescentibus, rigidiusculis, subcarinatis, demum caducis: petiolo 3-12 cm. longo; laminis triangulari- ovatis, breviter acuminatis, 9-16 cm. longis, 6.5—16 cm. latis, basi truncatis et anguste cordatis, denticulatis, supra novellis pilosis, maturis costa et venis primariis tantum pilosis, saepe bullatis, basi supra maculari-pluri- glandulosis, subtus granuloso-glandulosis costa et venis dense venulis sparsim hirtellis, 5—7-nerviis 3 majoribus, venis primariis supra basales 4-8 paribus prominentibus, venulis prominulis; ¢ inflorescentia non visa; @ paniculis hirtellis, 6-17.5 cm. longis, ramis paucis et brevibus 3 cm. vel minus; bracteis oblongis basi angustatis, 7 mm. longis, extus granuloso- glandulosis et sparsim pilosulis, intus patellari-glandulosis, glandulis interdum confluentibus; calyce 3—3.8 mm. longo, subconico, 2 vel 3 lobos rumpente deinde caduco; ovario 2-loculari, granuloso-glanduloso et echinis 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 219 mollibus tecto; stylis 2, plumosis, 6.5—7 mm. longis; capsula immatura, echinis 2—4 mm. longis, pubescentibus, + caducis. NETHERLANDS NEW GUINEA: common in secondary forest, Bele River, 18 km. NE of Lake Habbema, alt. 2200 m., Brass 11393 (2, TYPE, A), Nov. 1938 (one of the lesser species in regrowths; large arborescent shrub 5-6 m., with pubescent, rugose, subpeltate, broad leaves). This species is undoubtedly closely related to M. strigosa Pax & K. Hoffm., however, the pubescence is much softer than that which I associate with the latter species, and the pistillate inflorescences are definitely branched. Those of M. strigosa are described as racemose. Unfortunately no description of the @ bracts or calyx of the @ flower is given in the original description of that species. Ls) bo Macaranga densiflora Warburg in Bot. Jahrb. 13: 350. 1891.— K. Schumann in Notizbl. Bot. Gard. Berlin 2: 128. 1898. — K. Schu- mann & Lauterbach, Fl. Deutsch. Schutzgeb. Siidsee 396. 1900. — J. J. Smith in Nova Guin. Bot. &: 238, 791. 1910, 1912* he, 12: 546. 1917. — Rechinger, Denkschr. Akad. Wiss. Wien 89: 568. 1913. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 372. 1914. — Kanehira & Hatusima in Bot. Mag. Tokyo 52: 411. 1938. Mallotus acuminata sensu K. Schumann Fl. Kaiser Wilhelms Land 77. 1889; . Arg. Macaranga involucrata var. acalyphoides sensu Warburg in Bot. Jahrb. 13: 352. 1891.—Sensu K. Schumann & Lauterbach, Fl. Deutsch. Schutzgeb. Sudsee 397. 1900.—Sensu Valeton in Bull. Dept. Agric. Ind. Neerl. 10: ; non Muell. Arg. Macaranga acuminata Ridley in Trans. Linn. Soc. II. Bot. 9: 148. 1916. Shrub or tree up to 8 m. high, the ones and petioles usually velvety-pubescent with whitish hairs 0.4-0.6 mm. long; stipules linear or linear-lanceolate, 3-6 mm. long, erect or aaa reflexed, minutely pubescent on both sides or almost glabrous within, caducous; petioles 2-12 cm. long, the blades rhombic- or triangular- or lance-ovate, (5-—) 7-18 cm. long, (2—) 4-11 cm. broad, widely cuneate, truncate or rounded then minutely cordate at the base and usually glandular-bifoveate, long- acuminate (1-3 cm.) at the apex, 3-nerved with 4~-7 pairs of primary nerves obliquely ascending above the basal ones and anastomosing near the entire or denticulate margin, glabrous or sparsely pubescent above, particularly along the midrib and primaries, somewhat crisply pilose (especially on the nerves) beneath and copiously brown-glandular; ¢ panicle slender, 5-13 cm. long and pubescent, shortly branched and usually pedunculate; bracts subtending the glomerules crowded and pubescent on both sides, irregularly 1—5-dentate with teeth sometimes patelliform-glandular within; sepals 3, granulo-glandular at the apex; stamens 3 or 4 with 4-locular anthers; @ inflorescence up to 15 cm. long with flowers and subtending bracts usually crowded at the apex; bracts 220 JOURNAL OF THE ARNOLD ARBORETUM __ [VoL. xxxiIv 0.5-2.5 cm. long, pubescent on both sides, ovate or rhombic-ovate, acuminate and dentate with teeth patelliform-glandular within; calyx short, urceolate and pubescent; ovary 1-locular, densely glandular and bearing short pubescent processes; style 3-5 mm. long and densely papillose; capsule 4.5-5 mm. diameter, subglobose with processes 1 mm. long, tending to be deciduous. NETHERLANDS NEW GUINEA: Horna (in the N.W. rigors Atasrip 42 (46, BO); South McCluer, Moeteri near Babo, Aet 20 (6, A, BO); Aéndoea River near Oeta, Aet 448 (2, A, BO); Camp VIa, Boden Klow (6 & 2, A, phot. of type of M. acuminata Ridl.): van Gelder River, Docters van Leeuwen 9249 (6, BO); seashore, Toronta, Janowsky 577 (8, BO); on the lower part of Arso River, Gjellerup 55, 55a (6, BO); Andai, Beccari 8883 (P. P. 703) (dé + 92, F); secondary forest, Nabire, Kanehira & Hatusima 11431 (6, A); common in seral growths on gravel banks in river, Hollandia, Brass 8900 (6 Me Japen Island, Mamoeri, Neth. Ind. For. Serv. 30556 (von Dijk 340) (8, BO); Soerei, Aet & Idjan 785 (2, A, BO). NORTHEAST NEW GUINEA: woods near Kubai, Schlechter 18943 (¢, A); Wau, McAdam 98 (9, LAE). BISMARCK ARCHIPELAGO: New Britain: suburb of Rabaul, Kane- hira 3987 (&, A), 3997 (2, A); igo garden, Rabaul, Kanehira 3954 (6, A); Nodup, Waterhouse 224 (6, A, NY); same locality, d’Espeissis 98 (9, BR). Duke of York ci Bradtke 42 (6, A. BR), 279 (2,A; 2 & 9, BR) SOLOMON ISLANDS: Bougainville: Kieta, Kajewski 1579 (2, A); Karngu, Buin, Kajewski 2296 (6,A In this species the striking characters are the crowded flower-clusters with sharply dentate bracts at the apices of the short branches of the ¢ inflorescence, the patelliform glands on the teeth of the bracts of the pistillate inflorescence rather than within the margin (as in M. involucrata), and the bifoveate glands at the base of the leaves. This last character is a variable one, for example: in Docters van Leeuwen 9249 there are two glands on some leaves, one or sometimes none on others; in Aet 448 the glands, if at all present, are very inconspicuous, but the 4 inflorescence is typical of this species. The pubescence is also variable, but there are here represented two collections which I believe deserve particular attention: Carr 12206 (A, BM), from Koitaki, about 450 m. alt.; and Brass 562 (A), Bisiatabu, about 450 m. alt. These specimens differ from the others cited above in having the leaves densely velvety-pubescent on the lower surface and sparsely pubescent above; bifoveate glands are lacking at the base of the leaves; the bracts of the inflorescence are densely pubescent on both surfaces and granular-glandular; whether they would be _ patelliform- glandular when more mature I cannot say; the flowers also are too im- mature to make any comparisons. Mansfeld in 1929 determined Brass 562 as M. densiflora. Until a definite study of the type of M. dalecham- pioides S. Moore can be made I prefer to regard these as probably a 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 221 densely pubescent form of M. densiflora. It is to be noted, however, that the type of Moore’s species and these two collections are from the same region. 23. Macaranga dalechampioides S. Moore in Jour. Bot. 61: Suppl. 48. I have been unable to examine the type specimen of this species. In his comment Moore compared it to M. densiflora Warb., from which it is said to differ in having more pubescent leaves and lacking glands at the base, larger and more deeply toothed inflorescence-bracts, and linear floral bracts. I have read the description carefully for distinctive characters, but the only ones mentioned which appear valuable are the linear floral bracts, the minutely pubescent ovary, and the very short style. The last two characters suggest that the flower may be very young, but of this I have no proof; as for the linear floral bracts, could it be possible that these are the stipules of the closely crowded bracts at the apex of the inflorescence? In other words, this type needs to be studied carefully with more and comparative material before its identity can be established. 24. Macaranga urophylla Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147, VII): 377. 1914. Tree to 15 m. high with ferrugineous subvillous branchlets; stipules lanceolate, 6-10 mm, long, acuminate, densely pubescent, and caducous; petioles 6—8 cm. long, pubescent; leaf-blades triangular-ovate and slightly denticulate, 12-18 cm. long, 7-13 cm. broad, caudate-acuminate at the apex, at the base 3-nerved and broadly truncate, also slightly cordate, with 2-4 flat glands on the upper surface near attachment of the petiole, reticulate-veined and densely granulo-glandular beneath, the veins villous; the pubescent and densely flowered ¢ panicle 12-17 cm. long with short remote branches, the peduncle 5—6 cm. long; all bracts small (2-3 mm. long), with the basal part suborbicular then contracted, the apex abruptly acuminate, pubescent on both sides, granulo-glandular beneath and patelli- form-glandular above; ¢ flowers pubescent and granulo-glandular near the apex, 1.5 mm. long with 3 obovate acute sepals and 7-10 stamens with 4-locular (and sometimes 3-locular) anthers; @ inflorescence slender, simple (?), few-flowered, on pubescent peduncle 9-10 cm. long; bracts subfoliaceous and petiolate, lanceolate or linear-lanceolate, 1.5 cm. long, entire, within, along the margin, patelliform-glandular; pedicels 1-2 mm. long; @ flower with cupular dentate calyx pubescent and sparsely granulo- glandular, at length splitting spathe-like; the ovary 1-locular and densely covered with short unequal processes, the plumose style 2 mm. long; capsule unknown. The type of this species was collected at Namatanai, New Ireland (Peekel 669). The description given above is a translation of the original and not drawn from the material here cited. None of these specimens agree wholly with the original description, nor do they agree too well with 222 JOURNAL OF THE ARNOLD ARBORETUM _ [VoL. xxxiv each other, but since there is not enough material at hand to estimate the amount of variation within the species, it seems best to place them here, indicating points of difference. SOLOMON ISLANDS: Bougainville: rain forest, Kupei Gold Fields, alt. 950 m., Kajewski 1648 (2, A), April 1930 (small to medium-sized tree up to 15 m. high); Kieta, sea level, Kajewski 1608 (6, A), March 1930 (small tree up to 15 m. high); Kugu-maru, Buin, alt. 150 m., Kajewski 1788 (6, A), March 1930; east side of Kamo Mountain, alt. about 100 m., Robinson & Volk NGF 804 (2, A, BR, LAE), Jan. 1945. Malaita: Quoi-mon-apu, sea level, Ka- jewski 2314 (&, A), Dec. 1930 (very small tree up to 10 m. high, the lower surface of the leaves is silky to touch and the natives use them to dry them- selves after swimming). San Cristoval: rain-forest regrowths, lowlands, Waimamura, Brass 2566 (9, A), Aug. 1932 (common shapely small tree 5 m tall, exuding a thick brown gum when cut; young parts velvety with shining gray pubescence; petioles and underside of main nerves purple). Kajewski 1648 is perhaps most like the description of M. urophylla, with the apices of the leaves very slender and ca. 2 cm. long. The @ bracts are very slender, 1.5 cm. long and scarcely more than 2 mm. wide at the base. The styles are about 9 mm. long, yet the flowers have not yet lost the calyx; this style-length is much greater than that of the original. In Kajewski 2314 the pubescence might be called subvillous, — the hairs are a little short of villous; the pubescence somewhat approaches that of the original. It is to be noted, however, that the leaves tend to be three-lobed, with the apex of the middle lobe about 4 cm. caudate; stamens five to seven. Kajewski 1608 has a short compact pubescence (almost tomentose), but the leaves show the same tendency toward lobing as in the last specimen mentioned, only in a much less degree. Kajewski 1788 and Robinson & Volk 804 look like perfectly matched ¢ and @? plants. The glands at the base of the leaves are the same type and placed the same; the @ bracts are lanceolate, 1.5 cm. long and 3—5 mm. broad at the base, gradually tapering to the apex, the stipules are ovate-lanceolate or oblong. In Brass 2566 the 2 bracts are up to 3 cm. long and 8 mm. broad. In addition to these already mentioned there are two more collections which I add here hesitantly. The stipules in Brass 3243 (92, A) are almost glabrous and membranaceous and the young leaves have a floccose tomentum; in Kajewski 2557 (4, A) the stipules are pubescent, but the upper surface of the young leaves is floccose. Both are regrowth trees. Brass 3243 was collected at Kakatio, Ysabel, alt. 900 m., and Kajewski 2557 came from Ulolo, Tutuve Mt., Guadalcanal, alt. 1200 m. Although Pax & K. Hoffmann relate the species to M. schleinitziana K. Schum., I believe the material here cited is closer to M. involucrata (Roxb.) Baillon. 25. Macaranga involucrata (Roxb.) Baillon Etud. Gén, Euph. 432. 1852. — Warburg, Bot. Jahrb. 13: 352. 1891. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 374. 1914. 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 223 Urtica involucrata Roxburgh, Hort. Beng. 67. 1814; Fl. Ind. ed. 2, 3: 592. 1832. Mecostylis acalyphoides Kurz ex Teysmann & Binnendijk in Tijdschr. Nederl. Ind. 27: 44. 1864. Macaranga amboinensis Mueller Arg. in DC. Prodr. 15 (2): 1002. 1866. Tanarius involucratus O. Ktze. Rev. Gen. 2: 620. 1891. Shrub or tree with branchlets, petioles, stipules, and inflorescences spreading-pubescent or Baberulous: stipules linear, 5-10 mm. long; petioles 3—7 cm. long; the ovate or triangular-ovate leaf- blades 8-17.5 cm. long, 5.5-15 cm. broad, 1-3 cm. acuminate, epeltate and very slightly cordate with the base rounded or truncate or sometimes very narrowly (2-4 mm.) peltate, densely pilose when young, later (except for the pilose veins) glabrous above and sparsely pilose beneath with veins more densely so, above maculo-glandular, beneath densely granulo-glandular, 3—5-nerved with 5 or 6 pairs above the basal ones; ¢ panicle 5-10 cm. long with spreading branches and usually granulo-glandular; the small (1—1.5 mm.) bracts subtending the glomerules triangular-ovate, patelliform-glandular within; flower buds just before anthesis 0.5-0.7 mm. long; sepals 3, stamens 6-10 with 4-locular anthers; ¢@ inflorescence 5-15 cm. long, mostly with ¢ flowers clustered at the apices, the bracts foliaceous and bistipulate, up to 4 cm. long, ovate, dentate, and mostly long-acuminate, within the margin patelliform-glandular, usually pubescent or puberulous on both sides and outside granulo-glandular; pedicels 1.5—3 mm. long; calyx tubular-conic, quickly circumscissile or splitting and falling; ovary 1— (sometimes 2—) locular, the plumose style 7-10 mm. long and pubescent on the dorsal surface; capsule globose, 5-6 mm. diameter, granulo- glandular and bearing hairy processes 1.5—2.5 mm. long. NETHERLANDS NEW GUINEA: on edge of rain forest, Nabire, alt. 20 m., Kanehira & Hatusima 11669 (2, A), Feb. 1940. Biak Island, on coral cliff top along coast near Soredo, alt. about 25 m., Britton 85 (¢ & 2, A), May 1945 (small tree 15-20 ft. high, common). The second specimen is atypical in having appressed pubescence, but I can find no other difference in the material at hand. The species is a native of the Moluccas; there are at hand specimens from Amboina, Ternate, Ceram, and the Key Islands. I am doubtful whether Pax & K. Hoffmann were right in raising Warburg’s var. keyensis to specific rank, as both taxa occur on the islands. Macaranga involucrata var. mallotoides (F. Muell.) comb. nov. Macaranga mallotoides F. Mueller, Fragm. 4: 139. 1864.— Pax & K. Hoff- mann in Pflanzenr. 63 (IV. 147. VII): 376. 1914. Macaranga involucrata Baillon in Benth. & F. Muell., Fl. Austral. 6: 146. 1873. —F. M. Bailey, Queensl. Fl. 5: 1451. 1902. This variety differs from the typical species in the densely flowered branches of the 4 inflorescence; bracts subtending the crowded glomerules of ¢ flowers are recurved, ovate-suborbicular, and denticulate, granulo- 224 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. xxxIVv glandular and somewhat puberulent on both sides, and on the upper sur- face at times patelliform-glandular; the bracts of the @ inflorescence are denticulate and usually acute rather than acuminate, and the style is 5-8 mm. long. NETHERLANDS hid GUINEA: forest near camp, Gelieb, Branderhorst 210 (6, BO), Nov. 190 PAPUA: gregarious in contact zone of rain and savannah forests, Gaima, Lower Fly River (east bank), Brass 8342 (4, A), 8342A (2, A), Nov. 1936 (tree 3 m. high); abundant in rain forest and invading disturbed savannah forests, Tarara, Wassi Kussa River, Brass 8694 (4, A), 8694A (@, A), Jan. 1937 (shrub or small tree 3 or 4m. high). This variety is very close to var. keyensis, the latter being more nearly glabrous with bracts undulate or only barely denticulate. ~ ai Macaranga schleinitziana K. Schumann in Bot. Jahrb. 9: 207. 1887; in K. Schumann & Hollrung, Fl. Kaiser Wilhelms Land 79. 1889.— Warburg in Bot. Jahrb. 13: 351. 1891.— K. Schumann & Lauterbach, Fl. Deutsch. Schutzgeb. Siidsee 397. 1900; Nachtr. 297. 1905. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 373. 1914. Macaranga Schletnitziana var. 8 lobulata Pax & K. Hoffmann, l.c. Shrub or small tree with slender, densely velvety-pubescent branchlets, stipules, petioles, and inflorescence-axes; the narrowly linear stipules 4-6 mm. long, caducous; petioles 2.5-12 cm. long; leaf-blades (6.5—) 11-21 m. long, (3—) 7-20 cm. broad, triangular-ovate or rhombic-ovate or somewhat 3-lobed, acuminate (apex 0.5-3 cm. long), broadly truncate or rounded and slightly cordate at the base, and above near the attach- ment of the petiole bearing 2—4 flat glands, 3-nerved with 5-8 pairs of primary veins above the basal ones, on the upper surface sparsely pubescent, beneath shortly villous and dark granulo-glandular, reticulations fairly distinct on the lower surface of the older leaves; ¢ panicle 4-15 cm. long with short but densely flowered branchlets; all bracts small, 1-3 mm. long, ovate or oblong, acute, often slightly lobed on either side at the base, within patelliform-glandular or sometimes eglandular, usually pubescent on both sides and recurved; pedicels about 1 mm. long; sepals 3, concave, about 0.7 mm. long, pubescent and granulo-glandular, stamens 7-12 with 4-locular anthers; @ inflorescence 6—25 cm. long (including the peduncle 5—15 cm. long), at times with very short branches (up to 2 cm. long); bracts foliaceous, 1-3 cm. long and ovate or rhombic-ovate, denticulate and acuminate, pubescent on both sides, within near the margin, especially toward the base, patelliform-glandular; pedicels short becoming up to 8 mm. long in fruit; calyx of the ¢ flowers cupular, truncate, 1-1.5 mm. long, densely pubescent and granulo-glandular; ovary 1—locular with pubescent processes overtopping the calyx, the plumose style 4-6 mm. long, pubescent on the back; capsule globose and granulo- 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 225 glandular, 5 mm. diameter, the processes tending to be caducous, aa toward the base (those remaining on the fruit scarcely 1 mm. NETHERLANDS NEW GUINEA: Sigar on McCluer Bay, Warburg 20697 (6 & 9, A); Rori-esi near Babo, on McCluer Bay, Aet 161 (2, A, BO). NORTHEAST NEW GUINEA: Kelel, alt. 150 m., Schlechter 16146 (@, A), June 1907; Saki, alt. 250 m., Schlechter 18289 (8, A), Sept. 1908; vicinity of Kajabit Mission, alt. 240-600 m., Clemens 40681 (6, MICH), Aug. 1939; Boana, alt. 750-1300 m., Clemens 41482 (6, A), May—Nov. 1940; Finschhafen, War- burg 20507 (9, A), March, April 1889; region of Bumi, Weinland 125 (?, BR), 1889-91; on rain-forest margin, Yalu, alt. about 15 m., White, Dadswell, & Smith NGF 1621 (2, A, BR, LAE), July 1944; Lae, White, Dadswell & Smith NGF 1676 (2, A, BR, LAE); Bulolo, Fryar NGF 3957 (2, A), Feb. 1950; Bulolo, McIntosh NGF 35734 (@, A, LAE), Jan. 1950; on beach, Salamaua, Kanehira 4018 (@, A), Feb. 1937; Constantinhafen, Schlechter 14225 (4, BO). 27. Macaranga similis Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV): 29. 191 Macaranga Brassii Mansfeld in Jour. Arnold Arb. 10: 78, 232. 1929. Small or middle-sized tree 4-7 m. high with young branchlets, stipules, petioles, leaf-blades, and inflorescences velutinous; stipules 1—1.7 cm. long, 0.4-0.6 cm. broad, lanceolate or lance-ovate or slightly obovate- oblong and carinate, not readily caducous; petioles 5—11.5 cm. long; leaf-blades triangular-ovate (sometimes somewhat 3-lobed or suborbicular- ovate), 8-21 cm. long, 6.5-19 cm. broad, subabruptly short-acuminate, cordate at the base or rarely narrowly (7 mm.) peltate, bearing 2—6 flat glands (occasionally none) on the upper surface near the attachment of the petiole, densely dark granulo-glandular beneath, 3-nerved or in peltate leaves shortly palmate-nerved, with 6—9 pairs of primary veins above the basal ones; é panicle 4-18 cm. long, the peduncle 1.5—5 cm. long; bracts subtending glomerules small (1.5-4 mm. long), somewhat 2-lobed at the base then contracted and just above slightly broadened into a lanceolate or ovate apex, sparsely granulo-glandular and pubescent beneath, on the upper surface pubescent near the margin and above patelliform-glandular; sepals 3, concave and pubescent; stamens 6-12 with 4-locular anthers; inflorescence 10-22 (—35, fide Mansfeld) cm. long with flowers and fruits on the upper 2—9 cm., bracts and flowers crowded at 1-4 nodes; bracts triangular-ovate, 0.6—2.5 cm. long, 0.3—2 cm. broad, subfoliaceous, velutinous and granulo-glandular beneath, pilose on the upper surface along the veins and on the few or many patelliform glands well within the denticulate margin; pedicel 1.5 mm. long at time of flowering, becoming 6 mm. in fruit; calyx pubescent, cupular and truncate, about 1.2—1.5 cm. long, splitting and falling; ovary 1-locular, the plumose style 6-9 mm. long, pubescent on the lower surface; fruit subglobose, about 5 mm. diameter, granulo-glandular and bearing pubescent processes about 1-2.5 mm. long; seed dark-colored, rugulose, 3.5 mm. long, 3 mm. broad. PAPUA: Laloki River, alt. about 260 m., Brass 536 (6 & 2, A), Oct. 1935; on or about rocks on Eucalyptus savannahs, also rain-forest fringes; Rona, La- 226 JOURNAL OF THE ARNOLD ARBORETUM __ [VOL. xxxIv loki River, Brass 3633 (6, A, BO, NY), March 1933 (common small shapely tree 4-5 m. high); Mafulu, White 566 (6 & @, BR), July, Aug. 1918; open country, Veiya, Carr 11707 (6, A, BM), Mar. 1935 (tree 30 ft. tall); secondary forest, Kanosia, alt. about 60 m., Carr 11060 (@, A, BM, NY), 11063 (4, A, BM, NY), Feb. 1935; secondary forest, Boridi, alt. about 1140 m., Carr 14662 (@, A, BM), 14687 (6, BM), Oct. 1935; Buna area, Lane- Poole 166A (4, BR), July 1922; Domara River, Brass 1590 (2, isotype of M. Brassii, A), May 1926 (cormmon weed tree); Fife Bay, Turner 81 (2, A, BR), Oct. 1930. In my work on this and related species I had no doubt as to the identity of M. Brassii, of which we have an isotype, but I was at a loss to interpret M. similis P. & H. because of the description of the pubescence of the leaves, given as a mixture of stellate and simple hairs. I have seen only one species of Macaranga with truly stellate hairs, M. noblei Elmer of the Philippines. In this species only the @ plant is known, although I must admit that, even though the fruit looks like a Macaranga, one should see the ¢ flowers to be sure of the genus. In M. Brassii the hairs at times appear to be clustered in little bunches or fascicles along the veins. Apart from the description of the pubescence, the rest of the characters fit M. Brassii, hence the reduction of the latter. 28. Macaranga chrysotricha Lauterbach & K. Schumann in K. Schum. & Lauterb. Fl. Deutsch. Schutzgeb. Siidsee 399. 1900. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 378. 1914. Small tree; branchlets, petioles, and peduncles closely covered with fine hairs 0.1-0.2 mm. long, also setulose with stiffish golden hairs 1-2 mm. long, tardily glabrescent; stipules 2—2.5 cm. long, ovate-lanceolate and acuminate, appressed setulose and free (in specimens seen); petioles 5-17 cm. long, the slightly denticulate leaf-blades 15-30 cm. long, 16-22 cm. broad, orbicular-ovate, entire or 3-lobed and acuminate, openly cordate at the base and inconspicuously 2—4-maculo-glandular above, pilose on both surfaces (young leaves densely hairy), finely bullate above and densely reticulate beneath, 3—7-nerved at the base with 4-6 additional pairs of openly ascending primary veins; the 7-13 cm. pedunculate ¢ inflorescence paniculate with pubescent and sparsely granulo-glandular branches; bracts subtending the glomerules dilated at the base then contracted into an ovate lobe patelliform-glandular within; sepals 2, triangular-ovate and granulo- glandular outside at the apex, stamens 2 or 3 with 4-locular anthers; the long-pedunculate ¢ inflorescence 21-30 cm. long and bearing flowers only at the apex; bracts subtending the flowers 2-3 mm. long, stipitate and obtusely ovate or subrhombic, pubescent below and patelliform-glandular above; calyx 1-2 mm. long splitting into A or 4 or sometimes 5 parts; ovary 1-locular, the style 2 or 3 mm. long, long-plumose; fruit small, 4 or 5 mm. diameter, granulo-glandular a covered with soft pubescent processes 0.4—0.6 mm. long. NORTHEAST NEW GUINEA: Sattelberg, Nyman 495 (4, G), June 1899; Kani Mt., alt. about 1000 m., Schlechter 17287 (8, A), Feb. 1908 (small tree). 1953] PERRY, PLANTAE PAPUANAE ARCHBOLDIANAE, XXI 227 28A. Macaranga chrysotricha var. glaucescens Mansfeld in Jour. Arnold Arb. 10: 78, 233. 1929. Common small tree of forest regrowths, differing from M. chrysotricha in having leaves not or scarcely bullate, the branchlets, stipules, petioles, and often the peduncles glaucous (rather than finely pubescent as in the species) and setulose with coarse hairs about 1.5 mm. long. PAPUA: rain forest, Bisiatabu, alt. about 450 m., Brass 589 (4, A, IsoTYPE), Aug. 1925 (slender tree about 6 m. tall; trunk smooth, green; branchlets giau- cous, covered with small brownish prickles; stipules large, deciduous; leaf-scars prominent); Mafulu, White 440 (2, BR), July—-Aug. 1918; in forest regrowths, Bella Vista, alt. 1450 m., Brass 5464 (6, A, BO, NY), 5473 (2, A, BO), Nov. 1933 (small tree 3 or 4 m.; peduncles, petioles, stipules, and branchlets glaucous, inflorescence on long erect peduncle; capsules small). 29. Macaranga punctata K. Schumann in Fl. Kais. Wilhelms Land 80. 1889. — K. Schumann & Lauterbach, Fl. Deutsch. Schutzgeb. Sudsee 397. 1900. — Pax & K. Hoffmann in Pflanzenr. 63 (IV. 147. VII): 362. 1914. 2 Macaranga isadenia Pax & K. Hoffmann in Pflanzenr. 63 (IV, 147. VII): 377. 1914. Macaranga ovalifolia Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV): . 1919. Macaranga pseudopeltata Pax & K. Hoffmann in Pflanzenr. 68 (IV. 147. XIV): Macaranga maluensis Pax & K. Hoffmann in Pflanzenr. 85 (IV. 147. XVII): Tree 4-12 m. tall with the tips of the branchlets somewhat crisply ferrugineous-tomentose, quickly glabrate; stipules linear-lanceolate o linear-oblong, 4-7 mm. long, stiffish, quickly caducous; petiole 2-8 cm. long; leaf-blades variable in form and size, oblong-ovate, elliptic-ovate or triangular-ovate, 11 & 4.5 cm., 13.5 K 4 cm., 19