B. VERDCOURT CONTENTS. Page Abreae Ophrestia H. M. L. Forbes Pseudoeriosema Hauman...

Transcription

1 Kew Bulletin FORMERLY THE BULLETIN OF MISCELLANEOUS INFORMATION 3 PUBLISHED FOR THE ROYAL BOTANIC GARDENS, KEW Volume 24, No. 2 Studies in the Leguminosae-Papilionoideae for the 'Flora of Tropical East Africa': II* B. VERDCOURT CONTENTS Page Abreae A reappraisal of the species of the genus Abrus Adans Phaseoleae Clitoria L Glycine Willd Ophrestia H. M. L. Forbes Pseudoeriosema Hauman A. survey of the genus Teramnus P.Br. with special reference to East... Africa Erythrina L Mucuna Adans The Sphenostylis marginata E. Mey. complex Nesphostylis Verdc The genus Neorautanenia Schinz A REAPPRAISAL ABREAE OF THE SPECIES OF THE GENUS ABRUS ADANS. Breteler has recently revised the genus Abrus (in Blumea I o: (1i960)), during the course of which revision very many specimens from throughout the world were examined. Whilst carrying out routine naming of African material of this genus, Mr. J. B. Gillett reached the conclusion that the * Continued from Kew Bull. 24:' 70 (1970). All the material cited is in the Herbarium, Royal Botanic Gardens, Kew, unless otherwise stated. Specimens which have not been examined are indicated by an asterisk. A 235

2 236 KEW BULLETIN VOL. 24 specific limits had been drawn far too widely in this revision. During the preparation of an account of the genus for the 'Flora of Tropical East Africa', I reached the same conclusion and felt that the specific limits previously assigned were nearer the truth, and indeed, would have followed them had it not been for a need to comment on the revision mentioned. One of the primary practical functions of taxonomy is to give names to different plant populations so that their properties can be recorded, information tabulated and detailed phytogeographical studies made. Even if a species is genuinely polymorphic some grouping of the variants is still necessary if this primary aim is not to be lost sight of. Abrusfruticulosus sensu Breteler covers prostrate herbs, erect woody shrubs, plants with smooth pods, plants with densely tuberculate pods, plants with compressed unicolorous seeds and several Madagascan plants with spherical red and black seeds which are unquestionably much closer to Abrus precatorius L. but distinct. This aggregate contains many distinct species by any standards. Linnaeus would undoubtedly have separated most of the components. To sink wildly different components from areas thousands of miles apart which have not been seen in the field seems to me the height of folly and unlikely to serve any useful purpose. Since this present paper was first drafted Berhaut (in Adansonia 5: (1965)) has commented on Breteler's paper and also described a further species. I have not attempted a complete revision since the main object was to find names for the East African entities. What appeared to be six clearly distinct entities in East Africa are treated as three in Breteler's revision so it became essential to look at the genus on a world basis. The following key is imperfect since full material is often not available. KEY TO THE SPECIES OF ABRUS Bracts and bracteoles very short or scarcely half as long as the calyx: Madagascan plants of which ripe pods and seeds are unknown (placed here since it is possible that these will prove to have seeds similar to those of A. precatorius) : Flowers (I -3-) I5-1 8 cm. long; leaflets about I -jugate, about 2 cm. long and I cm. wide I. A. grandiflorus Flowers smaller, not exceeding I -3 cm. in length: Leaflets 4-5-jugate, dark and glabrous above, the venation con- spicuously pale, adpressed pilose beneath; inflorescences very short, about I cm. long A. parvifolius Leaflets without conspicuous contrast between venation and rest of surface above and without other characters combined: Leaflets with fine adpressed greyish indumentum 3. A. diversifoliolatus Leaflets and rhachis with longer spreading or less adpressed greyish and ferruginous indumentum. 4. A. sambiranensis Ripe seeds known: Seeds ellipsoid-globose, not compressed, usually red and black, very rarely entirely black or whitish, shiny: Leaflets and stems with fine adpressed golden indumentum; pods obliquely ribbed (Madagascar):

3 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 237 Leaflets 6-1 I-jugate... 5a. A. aureus subsp. aureus Leaflets 3-4-jugate.... 5b. A. aureus subsp. littoralis Leaflets and stems with sparse white indumentum; leaflets 8-I7- jugate; pods smoother or densely tuberculate: Pod smooth save for hairs... 6a. A. precatorius subsp. precatorius Pod covered with low tubercles... 6b. A. precatorius subsp. africanus Seeds markedly compressed, black or brown, sometimes speckled, shiny or minutely shagreened: Pods sparsely to mostly densely covered with tubercles; shrubs with terminal inflorescences: Leaflets mostly elliptic-oblong; tubercles on pod well developed: Leaflets Io-i6-jugate; indumentum of inflorescence rhachis and calyx golden... 7a. A. schimperi subsp. schimperi Leaflets 7-Io-jugate; indumentum of inflorescence rhachis and calyx greyish... 7b. A. schimperi subsp. africanus Leaflets oblong; tubercles less marked. 7c. A. schimperi subsp. oblongus Pods pubescent to hairy but not tuberculate: Pods usually thicker and woodier, acute to rounded at the apex; seeds mostly rhomboid, x 5-6 mm. (Arabia) 8. A. bottae* Pods mostly thinner, not woody, rounded at the apex save for the beak which is a continuation of the upper suture; seeds mostly ellipsoid, smaller, x 3-5 mm.: Leaflets small and narrow, x 1-3 mm., pubescent to silkypilose; pods 2 cm. long.. 9. A. fruticulosus Leaflets always broader, -8-4"5 squarish to oblong or elliptic-oblong: Leaflets 5-jugate with the lowest pair very much smaller than the upper pair; inflorescence cm. long, sessile (Madagascar).. io. A. madagascariensis Without above characters combined: Leaflets elliptic-oblong, 5-17 X 3'5-6 mm., rounded at both ends rather than truncate; leaves and stems sparsely whitish-puberulous; reticulation of leaflets raised and rather evident; pods 3'7-5'5 cm. long (S. Africa) I 1. A. laevigatus Leaflets distinctly oblong with truncate ends or even subcordate at the base; indumentum whitish or ferruginous 12. A. pulchellus aggregate (see later key on p. 247) Bracts and bracteoles as long as the calyx; corolla very dark red or purple 13. A. canescens * A. somalensis Taub. with 4-7-jugate leaves, linear-oblong, silky pubescent leaflets, 5-7 x I 5-3 mm. will probably key near here but the pod is not known.

4 238 KEW BULLETIN VOL. 24 REMARKS ON THE SPECIES I. Abrus grandiflorus Viguier in Not. Syst. 14: 172 (1952). Breteler treats this as a synonym of A. fruticulosus Wight & Arn. but from the description the very large flowers would seem to rule out that possibility. The species was described from two sheets, both of which I have examined. Although one sheet certainly has flowers in a packet which are at least I-7 cm. long, the other shows flowers as small as I"3 cm. Unfortunately the pods are extremely immature and it is impossible to tell what type of seeds this species has; it would not, however, surprise me if they eventually turn out to be globose and red and black, thus indicating relationship with A. precatorius L. and A. aureus Viguier. Without adequate material it is not possible to proceed further but on the available evidence the species seems quite distinct. 2. Abrus parvifolius (IViguier) Verdc., comb. et stat. nov. A. madagascariensis Viguier var. parvifolius Viguier in Not. Syst. 14: 174 (1952). Suffrutex scandens, ramis gracilibus saturate nigro-purpureis primum pubescentibus deinde glabrescentibus. Foliola 4-5-juga, stipulis lineari-lanceolatis 3 mm. longis; petiolus 5-8 mm. longus; rhachis 1-2 cm. longa, pilis appressis ferrugineis conspersa; petioluli 0-5 mm. longi, ferrugineo-pubescentes; foliola oblongo-elliptica, 5-12 mm. longa, mm. lata, apice rotundata et mucronulata, basi subcordata, papyracea, supra glabra, subtus pilis sericeis stricte appressis fere dense obtecta, nervis lateralibus et venulis supra conspicue pallidis. Inflorescentiae axillares, pauciflorae, circa I cm. longae, pedicellis 0-5 mm. longis. Calyx campanulatus, pilis ferrugineis et sericeis appressis obtectus. Vexillum suborbiculare, o mm. longum, 8-5 mm. latum, apice rotundatum, violaceum; alae oblongo-ellipticae, 9-5 mm. longae; carina Io.5 mm. longa. Legumen immaturum stipitatum, cm. longum, 2"9 3 mm. latum, pilis pallide ferrugineis et sericeis intermixtis appresse obtectum. Semina ignota. MADAGASCAR. Centre: between Mania and Ivato, Perrier (P, holotype-two sheets). I have examined the type of Viguier's variety and do not think that it is conspecific with his var. typicus. The small leaflets, glabrous above and with a pale reticulation of nerves, but densely adpressed silky pilose beneath, are very different from those of var. typicus which has very fine short adpressed hairs beneath. Viguier mentions that the pod is 3-4 times as long as broad but there is only one very immature pod on his type material. Although I would not normally describe a species from a single specimen in this genus, since it already has a name available, I have upgraded it, but its true affinities will remain unknown until ripe fruit is available. 3. A. diversifoliolatus Breteler in Blumea io: 610o, fig. 2/I (1960) (nom. nov., A. acutifolium Viguier substitutens). A. acutifolius Viguier in Not. Syst. 14: 173 (1952), non Miq. (1855). A. cyaneus Viguier, 1.c.: 172 (1952), pro parte, nom. confusum. A. madagascariensis Viguier var. dunensis Viguier, 1.c.: 174 (1952).

5 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 239 Without ripe fruits and a great deal more material it is not possible to formulate any useful opinion on this species. Viguier states 'foliis 8-Iofoliolatis' and Breteler says 3-5-jugate (his figure shows 3-4-jugate). The only two leaf rhachises preserved on an isotype at Kew, clearly both held 12 leaflets which are smaller than the dimensions given in the original description and measure X I-I-5 cm. It is interesting to note that the anthers are not.i1-4'5 quite uniform as pointed out by Viguier but all generic descriptions state that the anthers are uniform in Abrus. A. cyaneus Viguier is clearly composed of two elements, both of which enter into the original description. The fruits are undoubtedly those of A. precatorius L. (actually the subsp. africanus Verdc.) as suggested by Breteler but the foliage and flowers are scarcely distinct from A. diversifoliolatus. The difference in leaflet size is not significant but the pedunculate inflorescences with elongated rhachis nodes are very similar in both. A. madagascariensis Viguier var. dunensis Viguier has long pedunculate inflorescences and foliage very similar to A. diversifoliolatus; it clearly belongs here rather than to A. madagascariensis itself. 4. Abrus sambiranensis Viguier in Not. Syst. 14: 173 (1952). This, the hairiest Madagascan species, is similar in general appearance to A. pulchellus Thw. subsp. mollis (Hance) Verdc., and Breteler considers it to be a synonym of A. fruticulosus Wight & Arn. Although foliage differences are limited to the rounded leaflet apices, I do not feel it would be wise to unite it with subsp. mollis until ripe fruits are available. A sheet at Kew, Decary 2084 (Maromandia, Bejofo, 28 Apr. 1923), is not cited by Viguier who does, however, cite a Decary Abrus aureus Viguier in Not. Syst. 14: 173 (1952). I am certain that no one will follow the sinking of this into A. fruticulosus Wight & Arn. as proposed by Breteler. Viguier in his description clearly states that the seeds are closely similar to those of A. precatorius L. and material at Kew shows that they are close in size and colour. The pods are not detached pods but clearly part of the plant which makes up the sheet. Evidently it is clearly a distinct species from A. fruticulosus and, bearing in mind Breteler's specific concepts, one is surprised that he has not sunk it into A. precatorius itself. It differs, however, from A. precatorius L. subsp. africanus Verdc., its nearest ally, in the very distinct midrib, the adpressed mixed golden and white indumentum and the obliquely ribbed pod. The following sheets at Kew belong to this species and to the typical subspecies. 5a. subsp. aureus MADAGASCAR. Central Madagascar, Andrangaloaka, in forest, Aug. 1881, Parker :--climbing plant, vernacular name 'V6amaintilmy' ;without locality, Baron 817 (BM, K) & NW. Madagascar, Baron A sheet from Paris, Decary 7513, preserved at Kew (Restes de For6t au nord d'ankazobe, 13 Mar. 1930) and doubtfully determined as aureus is I think correctly named. Which of the 6 or 7 Ankazobes is meant is not clear.

6 240 KEW BULLETIN VOL. 24 5b. subsp. littoralis (Viguier) Verdc. comb. et stat. nov. A. madagascariensis Viguier var. littoralis Viguier in Not. Syst. 14: 175 (1952). I cannot agree that Viguier's var. littoralis has anything to do with his var. typicus. The latter is clearly allied to A. pulchellus Thw. even though no ripe seeds are present to confirm this. Var. littoralis clearly has red and black seeds closely similar to those of A. precatorius L. and they are not errors of mounting. The leaflets are 3-4-jugate, I X -3-2 cm.; the indumentum is *3-3.5 sparse but the minute hairs on the undersurface of the leaflets are distinctly yellowish. The pods are drawn out to an attenuate apex. Breteler's reduction of this variety to A. fruticulosus Wight & Arn. is clearly an absurdity and I have considered it a subspecies of A. aureus Viguier, but the material I have seen is scarcely adequate for a final opinion. MADAGASCAR. Est: near Ambilo, Perrier & Decary 6336* (P, syntypes). 6. Abrus precatorius L., Syst. Nat. 2: 472 (1767); Breteler in Blumea Io: 617 (1960). Even in the case of this widespread and very well-known plant I am unable to agree that the material belongs to a single taxon. Some botanists may not care to follow me in giving names to the two variants but all will agree that two entities are involved, as indeed was suggested by E. Meyer 120o years ago. A survey of this species throughout the world gives a clear impression that most of the African plants are not quite the same as the Asiatic; the differences are not specific but are usually sufficient to enable one to assign plants cultivated in the New World to their correct continent of origin. African plants have shorter muriculate pods; Indian plants have longer smooth pods in which a subsutural ridge is frequently clearly evident. There also appears to be a difference in the dehiscence of the pods, those of African plants with the valves twisting up considerably at their ends, whereas, in Indian plants, they frequently remain flatter; this is not a completely constant feature however. In Table I (p. 241) the results of this survey are tabulated. I am separating off the African plants as a distinct subspecies. The plant is one which is frequently cultivated, the brightly coloured seeds being used for necklaces, rosaries, etc. Doubtless some were taken to the New World during the slaving period. Dealing with the exceptions mentioned in the table-the plants from Brisbane Botanic Gardens clearly came from Africa. Some Polynesian specimens may have had mixed ancestry. Certain material in the Seychelles is the typical subspecies whilst other is the African subspecies; both may have been introduced or one may be indigenous. Material I have seen from Madagascar and Mauritius is of African affinity but that from Rodrigues clearly introduced from India. In the New World both races have been introduced as might be expected. It would be interesting if some botanical garden in the tropics which finds it has both races in cultivation could carry out crossing experiments. * The 6335 in Viguier's description appears to be an error.

7 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 241 6a. subsp. precatorius Unfortunately the holotype of A. precatorius preserved in the Hermann collection at the British Museum (Nat. Hist.) (Vol. 2, p. 6) bears no fruit but it is safe to assume that it is the same taxon as occurs on the island of Ceylon now (represented at Kew by an excellent Thwaites sheet). It is very unlikely that a plant of African origin would have been cultivated in Ceylon in Hermann's day. The specimen in the Linnaean Herbarium at the Linnean Society is also without fruit. 6b. subsp. africanus Verdc., subsp. nov. a subsp. precatorio leguminibus brevibus muriculatis differt. A. minor Desv., in Ann. Sci. Nat. 9: 418 (1826). A. squamulosus E. Mey., Comm. P1. Afr. Austr. 1: 126 (1836). A. tunguensis Lima* in Broteria, sir. Bot. 19: 127 (1921). A. cyaneus Viguier in Not. Syst. 14: 172 (1952), pro parte. Widespread throughout tropical Africa, Seychelles, Madagascar and Mauritius, also introduced into Australia and the New World. Type chosen as follows. KENYA. Tana River District: Karawa [Kurawa], 48 km. S. of Garsen, on black clay soil with sand admixed, scattered Terminalia, Euphorbia, Acacia with tangled shrub clumps-grewia, Lannea, Lecaniodiscus, Commiphora, Thespia, Dobera, etc., 15 m., 7 Oct. 1961, Polhill & Paulo 628 (K, holotype of subspecies; EA, isotype):-prolific twiner, much branched, corolla mauve, lighter near base and lower side of keel, filaments white; seeds light red with black ends. I agree with Breteler that A. cyaneus Viguier is based on a mixture, the detached fruits certainly belonging to the present subspecies. I have examined the types of A. tunguensis Lima and A. squamulosus E. Mey. It does not seem necessary to cite at length all the hundreds of available specimens belonging to this subspecies; reference can be made to Breteler's paper. TABLE I. Length and surface sculpture of pod in Abrus precatorius L. Country Average length (cm.) Surface sculpture Number of specimens1 Taiwan 3-6 Smooth 4 China Smooth 5 India 3"4* 3-6 Smooth2 30 Burma at least 3 (immature) Smooth I Ceylon Smooth 2 Malay Peninsula 3-' 3"7 Smooth 2 Malay Islands 3-6 Smooth 4 Indo-China 33"* Smooth 12 Philippines3 3-6 Smooth Io Philippines: Faint traces of Mindoro 2.4 roughenings I 1 Each with I to 20o 2 pods. One fruiting inflorescence on a divided sheet separated from Hohenacker 248 has short pods 2 8 cm. long with rough valves. There is no real evidence that this came from India since it is on one of those sheets containing many different elements mounted together. 3 Merrill (Sp. Blancoanae: 50o6) says 'of prehistoric introduction'. * I have not seen the description but the specific epithet is spelt this way both on the type sheets and in the 'Index Kewensis'.

8 242 KEW BULLETIN VOL. 24 Country Average length (cm.) Surface sculpture Number of specimens' New Guinea 4'3 Smooth 5 Australia 3-6 Smooth 8 Australia: Brisbane Botanic Gardens 2-6 Rough 2 Polynesia 3'54 Smooth to very slightly roughened 14 Cape Verde Is. 2'5 Muriculate I Annobon 3 Muriculate I W. Africa 2 62 Muriculate 18 Cameroun & Congo 2-9 Muriculate 14 NE. Africa 2'7 Muriculate 3 East Africa 2-86* Muriculate 62 Central Africa 2 9 Muriculate 40 Angola 2 81 Muriculate or more or less smooth4 9 South Africa 2'96 Muriculate 7 Seychelles 2'5 Muriculate 5 Seychelles 3-8 Smooth I Madagascar 2 5 Muriculate 8 Mauritius 2 2 Muriculate I Rodrigues 3'3 Smooth 2 Florida 3-1 Smooth to muriculate5 4 Bahamas Mostly smooth II Cuba 2"22 9 Smooth to muriculate5 3 Jamaica 2 Almost smooth to muriculate 4 Haiti 2 6 Almost smooth to muriculate 4 S. Domingo 2-6 Muriculate 4 Porto Rico 2'5 Muriculate 6 Virgin Is. 2-4 Muriculate 2 Antigua 2-6 Muriculate 2 St. Barthelemy 2 5 Muriculate I Saba I. 3 Muriculate I St. Kitts 2 Muriculate I Montserrat 3'5 Muriculate I Guadeloupe 2 3 Muriculate I Dominica 2 Muriculate I Martinique 2 Muriculate I Barbados 3 Muriculate 2 Grenada 2-4 Muriculate I Trinidad 2'3 Muriculate I Mexico 2'5 Muriculate 2 British Honduras 2 8 Muriculate 4 Honduras 2 9 Muriculate 3 Columbia 2 2 Muriculate I Venezuela 2 5 Muriculate 4 Guyana 2 5 Muriculate I Surinam 2 7 Muriculate I French Guiana 2-5 Muriculate 3 Brazil 2-8 Muriculate Io 1 Each with I to 20 pods. 4 Most Angolan specimens appear to have more or less smooth valves and this needs further investigation. 5 Long-fruited specimens with pods over 3-6 cm. long from these territories are of Indian origin. * Abnormal I-seeded pods only I -7 cm. long may occur. 7. Abrus schimperi Bak. in Fl. Trop. Afr. 2: 175 (1871). This is an exceedingly distinct plant having little relationship with A. fruticulosus Wight & Arn. with which Breteler has amalgamated it. The distinctly shrubby habit, terminal elongated inflorescences, and particularly the elongated woody pods narrowed at the base and nearly always strongly tuberculate, are distinctive features; in the two commonest subspecies the

9 STUDIES IN THE LEGUMINOSAE-PAPILIONOiDEAE: II 243 leaflets are elliptic with rounded sides. The larger characteristically shaped seeds are also an additional character. The northern populations have more numerous leaflets and a golden indumentum on the rhachis of the inflorescences whereas the southern populations have fewer leaflets and a grey indumentum. They have already been separated at specific level but I am considering them subspecies. In addition a new variant is described from Central Africa. All three have already been included in the main key. The material is briefly cited below. 7a. subsp. schimperi CENTRAL AFRICAN REPUBLIC. Yalinga area, Le Testu Waka, Le Testu 1291 (BM). SUDAN. Kordofan, Jebel Dair, Turner 235. Bahr el Ghazal, 'Bongo', Schweinfurth 2137 (BM, K), 2166 & s.n. (seeds wrongly mixed with 2345). Mongalla, Gondokoro, Mearns 3064 (US) and Imatong Mts., Katire, by R. Kineti, A. S. Thomas 1569 (BM, K). ETHIOPIA. Djeladjeranne, Schimper 1552 (K, holotype; BM, isotype). UGANDA. West Nile District: Metu, Chancellor 254; Madi, Grant; Terego, Hazel 633. Acholi District: Kitgum, Liebenberg 255-7b. subsp. africanus (Vatke) Verdc., comb. et stat. nov. Hoepfneria africana Vatke in Oesterr. Bot. Zeitschr. 29: 222 (1879). KENYA. Northern Frontier Province: Mathews Range, R. Kichich, Kerfoot 2423 (EA); Lorogi [Leroghi] Plateau, Kerfoot I09I (EA, K). Machakos District: Kaiti R., Katumani Farm Staff 1053 (EA); Kibwezi, Dummer 5036, Scheffler 138 (BM, K), Bally 710 & 793 in CM 8248; Kiu, Kassner 668 (BM, K); Ithaba, Bally 126O in CM 8247 (EA, K). Kitui District: Kiome, Edwards E 132 (EA); Kitui, Hildebrandt 2797 (K, isotype). Masai District: Karibani [Garabani] Hill, V. G. L. van Someren 126 & 168 (EA, K). Teita District: Voi, Napier 976 (EA, K); S. of Voi, Sagala Hill, Polhill & Paulo 957; Sasenyi [Sadenyi] Hill, S. of Maungu, Bally Kwale District: Digo, Sacleux (P). District uncertain: R. Tana, Battiscombe 212. TANZANIA. Mwanza District: Geita, S. Karumo Forest Reserve, Carmichael 826 (EA, K); Nyantukusa, Rounce 274 (EA, K); Rumara, Ilemera, Tanner Shinyanga District: Shinyanga, Bax I 13, Burtt 5115 & 2447 (BM, EA, K), Koritschoner 1756 (EA, K), Kemp I (BM) & Greenway 7398 (EA, K); Kizumbi, Carnochan 208 (BM). Pare District: N. Pares, Kisangara [Kissangara], Peter (B, K); E. S. Pares, Kisiwani [Kisuani], Greenway 2172 (EA, K). W. Usambaras: Makangala [Makangalla], Gillman 796 (EA, K); Makuyuni, Koritschoner 672 (EA, K); Boshatal, near Mashewa, Peter (B, K); Mashewa to Nembuibui, Peter 4881o (B, K). E. Usambaras: Bwiti to Nyika, Peter (B, K). Tanga District: Amboni, Holst 2825 [the only purely coastal record]. Kigoma District: Kigoma, Peter (B, K) & Pirozynski 448 (EA). Tabora District: Tabora, Peter & Friend 364 (EA); Malongwe to Tura, Peter ; Kissengi, N. of Malongwe, Bally 8295 (EA, K). Kondoa District: 19.2 km. S. of Kondoa, Polhill & Paulo I2 15. Dodoma District: Dodoma-Kondoa road, Burtt 1431 (BM, EA, K); km. 36.8, Chunya-Itigi road, Richards I9865; km. Itigi-Chunya road, Eggeling I2.8

10 244 KEW BULLETIN VOL (EA, K); W. of Dodoma, Peter (B, K); Saranda to Makutupora, km. 56o05 on the Central Railway line, Peter (B, K); Chenene, 73-6 km. N. of Dodoma, Polhill & Paulo 1256; Kilimatindi, Mulirahangi, Claus in E.A.H p.p. & Claus in E.A.H (EA). Mbeya District: Igawa- Mbeya road, Richards 7930; Igawa-Chimala road, Richards ZAMBIA. Central: 8-8 km. S. of Kafue, Angus 2867; I2-8 km. N. of Kafue, Angus 1574 (BM, K); 4-8 km. W. of R. Luangwa crossing on Great East Road, Exell, Mendonfa & Wild I197 (BM). East: Lundazi Fanshawe South: Mazabuka, White 6629; Choma, Fanshawe 5535 & van Rensburg 2738; Pemba, Lawton RHODESIA. North: Sebungwe, Lovemore 526 & Davies West: Wankie, Gwai Lutope junction, Wild 6028; Shangani, Mbazha Dam, Davies in S.R.G.H Central: Sebakwe, Eyles 41 (BM); Gatooma, Eyles 5079 & McKinstry in S.R.G.H ; Hartley, Serui Drift, Hornby 3146; 14.4 km. S. of Que Que, Biegel 1182 (K, SRGH). Some of the above records refer to plants approaching subsp. oblongus, notably Claus 1655 p.p., Gillman 796 and Greenway c. subsp. oblongus Verdc., subsp. nov. a subsp. schimperi foliolis oblongis, leguminibus obscure tuberculatis differt. Suffrutex m. altus caulibus appresse griseo-pubescentibus vel glabris. Folia o-i 5-juga, foliolis oblongis, mm. longis, 5'5-9 mm. latis, supra glabris, subtus appresse puberulis. Legumen lineari-oblongum, cm. longum, o09-i cm. latum, apice attenuatum, rostratum. Semina rhomboidea, compressa, pallide brunnea, 6 mm. longa, 5 mm. lata, 2-5 mm. crassa, minutissime rugulosa. MALAWI. Rumpi District: Near Njakwa, on rocky banks of R. Rukuru, 30 Apr. 1952, White 2539 (K, holotype of subspecies; FHO, isotype):- weak-stemmed thicket-forming shrub, 3 m. tall, leaves dark green, dull, standard dull green, suffused with violet, wings and keel violet. RHODESIA. East: Umtali, commonage, 990 m., 22 Nov. 1951, Chase 4182 (BM, K, SRGH); same locality, 1080 m., 7 Jan. 1954, Chase 5178 (BM, K, SRGH); same locality, 8 Mar. 1966, Corby 1568 (K, SRGH); Chipinga District, Sabi Valley, Rupisi, in thicket with Androstachys johnstonii, along river through escarpment hills, 18 Feb. 196o, Goodier 930 (K, SRGH):- shrub, stems slender, branches brown, leaves alternate, pinnate, pinnae oblong, flowers mauve in terminal racemes. South: Nuanetsi District, R. Malongwe, SW. Mateke Hills, stream banks, 6oo m., 5 May 1958, Drummond 5570 (K, SRGH):-slender shrub 1-8 m. tall often supported by surrounding vegetation. In foliage this is undoubtedly very similar to some forms of A. pulchellus Thw. subsp. pulchellus, e.g. Belcher 794 from Burma is practically identical, but the pod, habit and inflorescence clearly indicate close affinities with A. schimperi Bak. A. bottae Deflers, a scarcely known plant from Arabia has rather woody pods and seeds very similar to A. schimperi but the valves lack the tubercles; the leaflets are oblong but hairy above. Without ripe pods it could also easily be confused with A. precatorius.

11 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II Abrus bottae Deflers, Voy. Yemen: 132 (1889). This too, Breteler merges with A. fruticulosus Wight & Arn. and here I agree that the foliage of the type of the latter is very similar to Schweinfurth 528 accepted as a specimen of A. bottae, but the pod of the Arabian specimens is thick and woody, particularly in the case of Schweinfurth These two Schweinfurth sheets from Arabia, are by no means identical, I843 having larger leaves and indehisced woody pods; 528 has smaller leaves, seemingly thinner pod valves and three loose seeds. These seeds are of the same type as those of A. fruticulosus but larger and more rugulose. Until more material is available, I would certainly hesitate to reduce A. bottae to A.fruticulosus, even as a subspecies. The seeds of A. bottae are very similar to those of A. schimperi, and the woodiness of the pod suggests that the two are quite closely related, as was suggested by Deflers when he described the species. No material, has been seen which could represent A. somalensis Taub. based on Hildebrandt 1391 from Meid in northern Somaliland; Taubert does not describe the pod which presumably was not present on the type material. Until material has been recollected its possible identity with A. bottae cannot be confirmed and I have preferred to leave it as a dubious species. Schweinfurth 1843 is from one of the localities cited by Deflers so there is no doubt as to its identity. ARABIA. Asir: Harub, Thesiger (BM); Jabal Faifa, Vesey-FitzGerald 16290/10 (BM). Yemen: Hadie, Forskdl (BM); Gebel Mehegjaria, Badjil, Schweinfurth 528; Agara, Hodjela, Schweinfurth I Abrus fruticulosus Wight & Am., Prodr. 1: 236 (1834); Breteler in Blumea Io: 612 (I96O), pro parte. It is true, as Breteler mentions, that there are few structural differences to use for separating the species of Abrus and that size of the various organs is often the only criterion to be employed. He states (l.c.: 615) when dealing with his immense aggregate A. fruticulosus 'A. fruticulosus is a highly variable species. This variability may be connected with its very different habitats: rain forest to savannah country or even semi-desert. The type-specimens on which the synonyms are based show certain differences between each other, and differ also from the type of A.fruticulosus. These differences, however, are always restricted to size or shape, indumentum, or number of leaflets, shape or place of the inflorescence, size or shape of the pods, or the habit of the whole plant. Single or together all these characters vary and all conceivable intermediates can be found as soon as a wide range of specimens from a sufficiently large area is examined'. On this basis two uniform populations separated by some thousands of miles on different continents are sunk together because plants in a third area provide a connecting link in some character. Where a population gradually merges into another and it is geographically possible that the two interbreed then I agree that names must reflect this, but where the intermediates have no proved geographical connections nor is there any genetical evidence that the two populations and the intermediates are capable of interbreeding, then it does not correct to unite them under one name when seem. they can be clearly distinguished at a glance. There must be few widely distributed genera where intermediates are not found between quite obviously distinct entities when the whole world's species are examined. In the case of A. fruticulosus sensu stricto, the

12 246 KEW BULLETIN VOL. 24 material I have seen appears to be separable from A. pulchellus Thw. on account of its small narrow leaflets. Very little fruiting material is available at Kew and it appears that the seeds may be either smooth and shiny or distinctly shagreened. There is variation in the amount of indumentum on the leaves which might correlate with this seed character but the seeds available are partly detached in packets and not reliable. Io. Abrus madagascariensis Viguier in Not. Syst. 14: 174 (1952). I have examined the two syntypes of Viguier's var. typicus and, although the fruits are immature on the one sheet bearing them, it is obvious that the plant is closely related to A. pulchellus Thw. and thus I suppose Breteler is, according to his wide views, justified in including it in his aggregate A. fruticulosus Wight & Arn. sensu lato. From the small amount of material seen I prefer to consider it a distinct species. The leaflets are 5-jugate with the lowest pair very much smaller than the upper pair; the inflorescences are cm. long, sessile, and the pods cm. long, 8 mm. wide. It is unfortunate that the 5"5 pods are immature but it is certain that the seeds will not be red and black. Viguier's description of the seeds in his generalized description of A. madagascariensis as 'A. precatorii similia sed minora' is clearly derived from his var. littoralis which, as I have shown, belongs elsewhere. MADAGASCAR. Ouest: Ambongo, Radama bay, Perrier 529 bis & Decary 1558 (P, syntypes). Central Madagascar, without locality, Baron I I. Abrus laevigatus E. Mey., Comm. P1. Afr. Austr. I: 126 (1836). The population of Abrus occurring in Cape Province (Pondoland), Natal, Transvaal, Swaziland and southern Mozambique is very uniform in appearance and, although the foliage differences separating it are admittedly small, I consider it is best treated as a distinct species which scarcely intergrades with other species in Africa. What it might do if crossed experimentally with other species is not known. Baker (Fl. Trop. Afr. 2: 175 (1871)) considered it to be a synonym of A. pulchellus Thw. but in herbaria the species have been kept distinct and in any case laevigatus is a much older name. Unquestionably the type of A. pulchellus from Ceylon is closer to A. laevigatus than is the bulk of A. pulchellus but only one sheet has been seen from the island. 12. Abrus pulchellus Thw., Enum. P1. Zeyl.: 91 (1859). In the case of the plant with smooth pods having thin valves, compressed seeds and rather large oblong leaflets, i.e. A.fruticulosus sensu Breteler less the elements I have already removed, I agree there is considerable difficulty in reaching a decision as to their status when looked at on a world basis. In some cases there is great variation in such characters as the number of pairs of leaflets and their size; on the other hand there are distinct trends and some populations are well on the way to being worthy of treatment as separate species, e.g. A. suffruticosus Boutique. Unfortunately the type, as has been mentioned, is not representative of the species as a whole. Pods in Asiatic populations are on the average longer than in African populations. If one treats all the variants linked by intermediates as unworthy of recognition one finds always the problem of having two taxa, which appear quite distinct

13 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 247 to the eye and may have different properties, with the same name. The object of taxonomy is thus defeated. Whatever genus one revises and no matter to which botanist one talks the complaint arises that at least some of the species seem to run together. This is still often treated as though it were a surprising fact but in the light of evolutionary theory it is surely what one must expect. These problems are particularly true when dealing with a genus on a world basis and one's experience is limited to only one area of the world. The smaller the area under consideration the easier 'species' are to separate. Although there is a good deal of material of this aggregate available it is, so far as countries outside Africa are concerned, almost all devoid of detailed field-notes giving even habit and flower colour details. It was hoped that some assistance could be obtained from seed characters. In typical subsp. suffruticosus the hilum is parallel to the shortest axis of the seed, the surface of which is shagreened. In Cameroun material of subsp. tenuiflorus corresponding to A. stictosperma Berhaut the hilum is parallel to the longest axis and the seed surface is shiny brown with blackish-violet spots. The development of the rim aril is also variable, it being quite thick in subsp. cantoniensis but reduced in subsp. suffruticosus. The attention of collectors is drawn to the fact that seed material is required from all over the world to see if any characters are constant. At present intermediate states seem to occur but the available material is not adequate. A key to the subspecies I am recognizing is given below. Some botanists with access to only selected sheets might think that I was still sinking too much but a survey of all the material seems to suggest that a compromise between Breteler's super-species and the previous recognition of some half a dozen species is the best course. Indumentum KEY TO THE SUBSPECIES OF ABRUS PULCHELLUS THW. not ferruginous but sparse and greyish, adpressed (rarely the petiolules appear to be brownish but the hairs are pale) (not African): Leaflets larger, 2-4 cm. long; pod mostly well exceeding 3 cm. in length subsp. pulchellus Leaflets smaller, mostly about I cm. long and with distinctly subcordate bases.... subsp. cantoniensis Indumentum usually ferruginous on some part of the plant but leaflets frequently only grey pubescent; or if no trace of ferruginous hairs and leaflets over I cm. long then pods about 3 cm. long: Stems and leaf rhachides "5 with dense adpressed or spreading ferruginous hairs; leaflets densely hairy: Indumentum of stem and rhachides spreading; leaflets usually larger, about 3 cm. long; pod cm. long, usually about 4 cm. 3'5-4"7 (E. Asia).... subsp. mollis Indumentum of stem and rhachides usually adpressed; leaflets usually smaller, x 3-4'5 mm.; pod about 3 cm. long (Central Africa)... subsp. suffruticosus Stems and leaf rhachides with sparser, shorter, more adpressed hairs; leaflets pubescent, 1-3'5 cm. long... subsp. tenuiflorus I 2a. subsp. pulchellus Typically this has large leaflets which are glabrous but in some Philippine, Javan, Papuan and Perak material the leaflets are pubescent above. The

14 248 KEW BULLETIN VOL. 24 indumentum on the stems and rhachides is usually grey, sparse and adpressed but some Papuan specimens have spreading hairs. Previously a good deal of African material has invariably been referred to A. pulchellus but I consider the typical plant occurs only in India, China, and the Malesian Islands. In the African material the pods are on the average shorter and where the material is sufficient it appears that the rim aril is more developed in Indian material. The sizes of various pods are shown in Table 2, below. I2b. subsp. cantoniensis (Hance) Verdc., comb. et stat. nov. A. cantoniensis Hance in Journ. Bot. 6: 112 (1868). A. cantoniensis Hance var. hossei Craib in Bull. Misc. Inf. Kew 1911: 39 (IgII). This small-leafleted form from Thailand and China is distinctive enough to be retained at subspecific level. The small, compressed, ellipsoid seeds, 3 8 x 3 mm. with a thick rim aril may be a constant character. Craib's variety does not appear to be distinguishable and is mentioned since the type Hosseus 155 is included by Breteler in his list of exsiccatae but the varietal name is not listed. It is very interesting to note that T. Samson in the field-note for his 1553 considers that A. mollis and A. cantoniensis are not distinct species; this is the only comment by any field worker on members of this complex. TABLE 2. Average fruit lengths (in cm.) of Abrus pulchellus Thw. sensu lato subsp. pulchellar subsp. mollis Yunnan 5-1 China 4'3 (3'5-4'9) India 5'7 (3'5-7) (average of 12 specimens) Ceylon North Vietnam 4 Burma 5-8 5"5 ( ) South Vietnam Perak 6-5 Thailand 4'3 4"7 Thailand 5-8 Java 5-7 (5'5-6) Philippines 5'9 (5-6"5) (4-9) Sarawak 4 Papua 7'5 Papua 7 subsp. tenuiflorus subsp. cantoniensis Brazil 3-8 (3-5-4) China 3-I Africa* 3'4 (2-4"5) Thailand 2-8 (2-3-3) (average of 20 specimens) intermediates between subsp. tenuiflorus and subsp. suifruticosus subsp. suifruticosus Africa 3-0 (2-.2-4) West Africa 3-I (2-4) (average of 17 specimens) (average of 8 specimens) * A variant in West Africa has pods 5-6 cm. long-see note under I2e, p I2c. subsp. mollis (Hance) Verdc., comb. et stat. nov. A. mollis Hance in Journ. Bot. 9: 130 (1871). Typically this is a distinctive variant often closely similar to subsp. suffruticosus but with larger leaflets up to 3 cm. long and longer pods; there are, however, intermediate forms linking subsp. mollis with subsp. pulchellus and cantoniensis. There is a flowering specimen in the Kew Herbarium from the Mishmi Hills on the Assam/Tibet border collected by Griffith, unfortun-

15 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 249 ately with no accompanying data, which is very similar to subsp. suffruticosus. Material has been seen from northern Assam, Malay Peninsula, Java, North and South Vietnam, Thailand, China, Philippines, Sarawak, Borneo and from Papua. No ripe seeds have been seen. I2d. subsp. suffruticosus (Boutique) Verdc., comb. et stat. nov. A. suffruticosus Boutique in Fl. Congo Belge 6: 84 (I954) (sine descr. lat.) & in Bull. Jard. Bot. Brux. 25: 127 (1955)- A. repens Tisserant in Bull. Mus. Hist. Nat. Paris., ser. 2, 5: 332 (1933)- [A. pulchellus sensu Hepper in Fl. W. Trop. Afr., ed. 2, I: 574 (1958), pro parte, non Thw.] In the central part of its range this subspecies is distinctive, the low habit, small, very hairy leaves and small pods being fairly constant characters; the calyx is also often distinctly lobed. In the western part of its range it merges gradually with subsp. tenuiflorus. If only East Africa were considered one would undoubtedly consider the Uganda plant (tenuiflorus) quite distinct from the southern Tanzania plants (suffruticosus). Representative material is briefly cited below. Intermediate specimens are frequent in Sierra Leone and have been seen from Mali and Guinea. A. gorsei Berhaut based on 16 specimens from Senegal (holotype, Berhaut 6877 (P)) is such an intermediate. Deighton 5615, Jordan 219, Jones 188, Scott Elliot 4312, all from Sierra Leone, closely match the type. SIERRA LEONE. Yoni Mamila Chiefdom, Kasawe, King 25b (pro parte). Rokupr, Jordan 381 (intermediate). Mando, Jordan 219. N. NIGERIA. Patti Lokoja, Dalziel 14. Bauchi Plateau, Lely P.813 (FHI, K). About 7 km. NW. of Kaduna, Jackson 261 (FHI); Kaduna, near Buruku, Charter in F.H.L CENTRAL AFRICAN REPUBLIC. Oubangi, Bozoum, Tisserant 2946 (P, syntype of A. repens). CONGO (KINSHASA.) Kinshasa [Leopoldville], S. Bayuku, Devred Lubumbashi [Elisabethville], de Georgi* (BR, holotype), F. A. Rogers 26290, Somona 63 & Schmitz 2781; 12 km. NW. of Lubumbashi [Elisabethville], Gathy 431 & 591. Kundelungu Mt., Kassner 2596 (BM). Bwana Muto, Callens 3122 (BM). TANZANIA. Kigoma District: Ujiji, R. Mkuti to Msusi, Peter Ufipa District: Kalambo Falls, Richards I2803. Rungwe District: Masukulu, Stolz 765, Songea District: Mbamba Bay escarpment, Milne-Redhead & Taylor ZAMBIA. North: Kalambo Falls, Richards I 1323 & McCallum Webster 908; near Mbala [Abercorn], track to Kasakalawe, Richards 502I; old road to Cascalawa from Chemba village, Richards I2492; Kasama, Bimba, Verboom LK 28; Nimkola to R. Lunzua, Richards 4642; Nimkola Richards 4647; Kasama, Mungwi, Robinson 5oo8; Mporokoso District, Nsama, Richards 8971; 9-6 km. Mulwe to Kawambwa, Angus West: Mufulira, Cruse 512 & Exell, MendonGa & Wild 1396 (BM); Ndola, Masansa Forest Reserve, Fanshawe 4352; Ndola, Fanshawe o1015; Kitwe Sewage Works, Shepherd 37 & 53; Solwezi, Robinson 3517; 6 km. W. of Solwezi, Drummond & Rutherford- Smith; Machili, Fanshawe Barotseland: edge of R. Luampa, near

16 250 KEW BULLETIN VOL. 24 Luampa Mission, White 2105 & 2 IO5a. Locality dubious, S. of Lake Tanganyika, Clark (BM.). 'N.S. Western Rhodesia' (?), Martin 764 (BM). ANGOLA. Onschingue, Kuito, Baum 786 (BM, COI, K). Bi6, Monongue, R. Cuiriri, Gossweiler 2830 (COI). 12e. subsp. tenuiflorus (Benth.) Verdc., comb. et stat. nov. A. tenuiflorus Benth. in Martius, Fl. Bras. 15: 216 (1859). [A. pulchellus sensu Bak. in Fl. Trop. Afr. 2: 175 (1871); Boutique in Fl. Congo Belge 6: 84 (1954); Hepper in Fl. W. Trop. Afr., ed. 2, I: 574 (1958), non Thw. sensu stricto.] A. gracilis Lima in Broteria, ser. Bot. 19: 127 (1921). A. stictosperma Berhaut in M6m. Soc. Bot. Fr : 7 (1954) & in Fl. S6negal: 30 (1954). Both A. pulchellus and A. tenuiflorus were published in the same year but it has only been possible to find out a more precise date (30 July) for Bentham's name. Since pulchellus is a much better-known name I have used it for the species as a whole. I am uncertain about the true status of subsp. tenuiflorus since very little material is available of the American plant, which appears distinctive, the inflorescences being very slender and the flowers small, 8-9:5 mm. long. Some African material, e.g. Schweinfurth 2345 is extremely similar and without further evidence I have thought it best not to separate them. A name is available for those who may find it desirable to separate the African plant. In pod length the African plants are closer to the American plants than they are to Indian plants with the exception of one variant mentioned below. On the whole, African material has larger flowers, but not always and the possible significance can scarcely be assessed without much more American material. A selection of material from Africa is cited below. SENEGAL. Gorom, Berhaut 1433 (P, holotype of A. stictosperma). Sangalkam, Berhaut 5451 & 5600 (these lack ferruginous hairs but have the small pods). Basse Casamance: Bignona, Kalounayes Forest, Berhaut 6721; S6dhiou, Berhaut Without locality, Perrotet 212 (BM). GUINEA. Macenta, Baldwin SIERRA LEONE. Yoni Mamila Chiefdom, King 25B (pro parte). Bumbura, N. W. Thomas 3716 & 3757 (BM). Kenema, Deighton 392 (BM, K). Njala, Deighton 1499 (BM, K). Pendembu, Glanville 7 (BM). Without locality, Afzelius (BM). LIBERIA. Boporo District: Tawata, Baldwin I0319. Ganta, Harley. GHANA. Kumasi, Darko 638. R. Volta Forest Reserve, Morton GC Gambaga, Adams Bame Pass, Morton GC NIGERIA. Lokoja District, Parsons. Mada Hills, Hepburn 89. Ibadan, Punch 47, Ujor in F.H.I Agolo, N. W. Thomas 349. Idah, Kabba, Acharane Forest Reserve, Latilo in F.H.I Oyo, Ife, Shasha Forest Reserve, Emwiogbon in F.H.I CAMEROUN. B6tar6 Oya, Breteler lo70. Bertoua, 25 km. along road to Nanga-Eboko, Breteler 609. Plateau of Adamaoua, 8 km. S. of Ngaound~r6, Breteler 577. Bitye, Yaunde, Bates 932 (BM). Eboluwe, Bates 604 (BM).

17 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 251 EQUATORIAL GUINEA [Rio Muni]. Bebui, path to Oko, Tessmann 576 (locality not traced). CONGO (KINSHASA). Eala, Lebrun 1480, I Io8 & Louis Luki, Devred Yangambi, Tofende I., Louis (BM). Panzi, Callens 2731 (BM). Kinshasa [Leopoldville], Dinga, Germain SUDAN REPUBLIC. Bahr el Ghazal, Seriba Ghattas, Schweinfurth UGANDA. Bunyoro District: Budongo Forest, Harris 146 in F.H. II Io & Loveridge I 12. Mengo District: near Entebbe, Kitubulu Forest, Chandler 1223; Kipayo, Dummer Io99 (BM, K); Buvuma I., Bagshawe 652 (BM). TANZANIA. Uzaramo District: Pugu Hills, Vaughan 2377 (BM, EA). Lindi District: Nachingwea, Namanga Hill, Anderson 1234 (EA); Rondo [Mwera] Plateau, Schlieben 616o (BM). Tunduru District: 96 km. from Masasi, Richards (exceptionally large-flowered form with many of the characters of subsp. suffruticosus). ZAMBIA. North: Fort Roseberry, Fanshawe 851 I1; Bancroft, Fanshawe West: Mufulira, Eyles 8316; Kitwe, Fanshawe MOZAMBIQUE. Niassa: Palma, Lima 257 (PO, holotype of A. gracilis). Manica e Sofala: Cheringoma, serragi-o de Durundi, Barbosa 1661 (BM, K). ANGOLA. Cazengo, Granja de S. Luiz, Gossweiler 5870 (BM, COI). Malange, Bondo, Quela, Nolde 131 (BM). There is a form (variant 'A') in West Africa with 6-jugate leaflets, wider, longer, more densely pubescent pods, (I.7-)3.5-6 cm. long, which comes close to subsp. pulchellus and is the form considered by Berhaut to be true pulchellus. SENEGAL. Basse Casamance: Bignona, Koubalan, Berhaut 6782; Ziguinchor, Djib6lor, Berhaut PORTUGUESE GUINEA. Bissau, Peluba, Espirito Santo DAHOMEY. Tchatchou, Poisson. NIGERIA. Nupe, Onitsha, Barter 1749 (the sheet cited by Baker, l.c., as A. pulchellus). CONGO (KINSHASA). Yangambi, Ile Tofendi, Louis 12861; Yangambi, Isalowe, Louis I3. Abrus canescens Bak. in Fl. Trop. Afr. 2: I75 (1871); Breteler in Blumea Io: 609, 620 (1960). I agree with Breteler in his circumscription of this well-marked species. It has been confused at times with typical hairy forms of A. pulchellus Thw. subsp. suffruticosus (Boutique) Verdc. but is separated by the length of the bracts, very dark flower colour and the rather well-developed rim aril DUBIOUS SPECIES 14. Abrus somalensis Taub. in Engl. & Prantl, Pflanzenfam : 355 (1894) & Engl., Bot. Jahrb. 23: 193 (1896). Breteler considered this to be a synonym of A. fruticulosus and in this case the description does indicate a plant with 4-7-jugate linear-oblong, silky pubescent leaflets, 5-7 x I'5-3 mm. Unfortunately no pods were available to Taubert who stated 'verwandt mit canescens' although his floral description does not fit this. Unfortunately the type Hildebrandt 1391 appears B

18 252 KEW BULLETIN VOL. 24 to have been destroyed and no material from Somaliland has been seen by either Breteler or myself, despite the fact that the type locality Meid [Mait] has been visited by several collectors. Until the pod is available its probable affinity with A. bottae Deflers or A. schimperi Bak. (on geographical grounds) cannot be checked. 15. Abrus wittei Bak.f. in Rev. Zool. Bot. Afr. 21: 303 (1932); Boutique in Fl. Congo Belge 6: 86 (I954)- Both Boutique and Breteler have sunk A. wittei Bak. f. into A. precatorius L. but I am not at all convinced of the correctness of their reasoning. I have examined the holotype preserved at the British Museum (Nat. Hist.) and also an isotype kindly sent on loan from Brussels. There are a number of marked differences between these specimens and typical A. precatorius. A. wittei appears to be an erect shrub but no data are given and the young fruits are narrower with more tapering ends and the style is scarcely bent. The indumentum on the young attached pods is certainly similar to that often found in A. precatorius but the very short inflorescences are quite dissimilar. On the isotype there is a packet containing a seed indistinguishable from those of A. schimperi but Boutique has dismissed this as not part of the main plant mentioning that there were no pods capable of holding ripe seed. On the holotype, however, another packet contains the same sort of seeds and a ripe pod. I see no reason to doubt that it came from the same plant but there is no certainty in the matter. There is a resemblance to the plant described earlier in this paper as A. schimperi subsp. oblongus and I am leaving A. wittei as a distinct species allied to A. schimperi but differing in inflorescences and the smooth pods. Until further collections from the type locality, Kiambi, are available the plant will remain a puzzle. A hybrid origin does not seem very likely. 16. Abrus sp. Several years after this revision was first completed a rather fragmentary specimen collected by Donaldson Smith in an unfrequented part of Ethiopia was discovered amongst undetermined specimens of Ormocarpum at the British Museum (Nat. Hist.). A brief description is given but the material is not adequate enough to justify formal description. Presumably a woody shrub; stems purplish, lenticellate, densely adpressed grey pubescent; young shoots pale, with dense, more spreading fulvous pubescence. Leaflets 8-jugate, oblong-elliptic, 0o cm. long, mm. wide, rather densely adpressed pilose beneath, the costa yellowish beneath in young leaflets. Calyx 4 mm. long, undulate, sparsely pilose. Vexillum narrowly elliptic, 1-4 cm. long, 4-5 mm. wide. Very immature pod linear-oblong, 3-5 cm. long (excluding thickened persistent style), 4-5 mm. wide, gradually acutely attenuated into the style, densely adpressed yellowish-pilose. ETHIOPIA. Bale, Scec Hussen (Sheik Husein), Smith 191 (BM). 21 Sept. 1894, Donaldson Until more material becomes available it would serve no useful purpose to attach a name to the above plant which, judging by the pod, is almost certainly closely allied to A. schimperi if not an additional subspecies of it.

19 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 253 A great deal more material is required to solve the numerous problems mentioned in the above account. The attention of collectors is drawn to the need for much more material showing ripe fruit and seeds. Chemical and cytological methods might throw some light on further ways to improve the classification suggested here. PHASEOLEAE CLITORIA L. Two species of this genus occur in East Africa apart from C. laurifolia Poir. (= C. cajanifolia (Presl) Benth.) which was formerly cultivated in Tanzania at Amani; one the well-known C. ternatea L. and the other, C. rubiginosa Pers., naturalized in Zanzibar. C. ternatea L. has for the purpose of the Flora been accepted as a single polymorphic species, the numerous variants being probably simple Mendelian varieties. The plant naturalized in Zanzibar and collected there as long ago as 1868 by Sir John Kirk requires some discussion. Typical C. rubiginosa Pers. has a dense ferruginous indumentum and often ribbed pods, whereas the Zanzibar plant is glabrescent and has a smooth pod. It does in fact closely resemble a number of specimens from the Old World named C. mariana L. and it becomes a matter of interest to ascertain the probable origin of the Zanzibar plant. All the material of the C. rubiginosa-c. mariana complex at Kew was examined and the results are given in Table 3 (pp ). The conclusion has been reached that two major groups can be separated on the shape of the stipules but I do not propose to revise these groups. It seems unlikely that all the names should be kept up at specific level and it would be better to recognize subspecies and varieties placed under the two main heads C. rubiginosa and C. mariana. Glabrous forms of C. rubiginosa are so distinctive that a variety is proposed to accommodate them. In West Africa the species is also extensively naturalized and the glabrous variety predominates. It is certain that the Zanzibar population stems from a single introduction of this glabrous variant from either West Africa, tropical America or the West Indies and is not a variant of the Asiatic species despite the close resemblance. Clitoria rubiginosa Pers., Syn. 2: 303 (1807). var. glabrescens Verdc., var. nov. a C. rubiginosa typica caulibus glabrescentibus, foliis subtus glabris vel sparse pubescentibus differt; affinis C. marianae L. stipulis ovato-ellipticis brevioribus distinguenda. African sheets and the type are cited below. A number of Brazilian sheets are intermediate between the two extremes. PORTUGUESE GUINEA. Fulacunda, Espirito Santo SIERRA LEONE. Along R. Wanjee near Baba, Bakshi 176. LIBERIA. Sanokwele District: Gbau, Baldwin 94I2. NIGERIA. Olokemeji, Foster 319; Foster 16. Lagos, Meller 129. Eket, Talbot Ewa, Ahor?, Thompson 518. TANZANIA. Zanzibar: Masingini Ridge, I Feb. I929, Greenway 1290 (EA, K):-a large cream-flowered climbing plant growing in sandy soil in the bottoms of eroded areas, not very common; without locality, Kirk s.n., Toms s.n. & Vaughan 86 (BM); 'Zanzibar and Pemba', Dowson 125.

20 TABLE 3. Characters in the Clitoria mariana L. Country Stem Stipule Leaf Leaf Calyx Indumentum shape apices indumentum indumentum China sparse, oblong-lanceolate obtuse to sub- sparse sparse adpressed to ovate-oblong acute, mucronulate India sparse, oblong-lanceolate obtuse to sub- sparse sparse adpressed or to ovate-oblong acute, -4 spreading mucronulate Burma sparse, oblong-lanceolate acuminate, sparse sparse adpressed to ovate-oblong mucronulate Siam sparse, oblong-lanceolate subacute to sparse sparse adpressed to ovate-oblong acuminate, mucronulate Java fairly dense, ovate subacute, dense? adpressed mucronulate Australia sparse to oblong-ovate rounded to sparse to sparse to fairly dense, subacute, fairly dense dense adpressed mucronulate W. Africa sparse to broadly ovate or obtuse to sparse to sparse to dense, elliptic, short slightly dense dense spreading emarginate, very shortly mucronulate

21 Zanzibar sparse broadly ovate slightly glabrescent minutely or elliptic, short emarginate, puberulous scarcely but no long mucronulate hairs Florida glabrescent narrowly obtuse, glabrous minutely lanceolate mucronulate puberulous only SE. United glabrescent, narrowly obtuse, glabrous to sparse States adpressed lanceolate mucronulate sparse Mexico sparse oblong-lanceolate, acuminate, sparse sparse to spreading rarely elliptic mucronulate; glabrous rarely obtuse British dense broadly ovate- slightly velvety dense Honduras spreading elliptic, short emarginate, beneath scarcely mucronulate W. Indies glabrescent elliptic to slightly glabrescent glabrescent to dense, ovate-rhomboid, emarginate, to velvety to dense spreading, very short scarcely beneath ferruginous mucronulate Tropical glabrescent elliptic to subacute, glabrescent usually America to dense ovate-rhomboid, emarginate, to velvety dense adpressed or very short shortly spreading, mucronulate ferruginous

22 256 KEW BULLETIN VOL. 24 TOBAGO. Bacolet, in coconut plantations, 13 Oct. 1937, Sandwith i754 (K, holotype of variety):-creeper, leaflets thick, more or less glaucous beneath, corolla white, standard with purplish stripes leaving the median vein within. GLYCINE WILLD., nom. conserv.* The names of the soya bean and its allies have always given much trouble. There are two Linnaean names which refer to the soya bean, namely Phaseolus max and Dolichos soja, both dating from Merrill made a new combination for the first under Glycine in 1917 and this name has mostly been followed, e.g. by F. J. Hermann in his monograph of the genus. A few years later, however, when discussing Loureiro's Fl. Cochinch. he refutes his own new combination and uses Glycine soja (L.) Sieb. & Zucc. Even if this were correct the name Glycine max (L.) Merr.t would still be the valid one since Merrill was the first to choose between two names of equal date; it is not, however, correct since Siebold & Zuccarini's name is not based on Dolichos soja L. but is a new name, hence no combination of Dolichos soja can be made in Glycine. This is clear since Siebold & Zuccarini cite Dolichos soja L. in synonymy when dealing with the next species in their account. F. J. Hermann has used the name G. ussuriensis Regel & Maack for the plant called G. soja by Siebold & Zuccarini but I would agree with Ohwi that there seems to be no bar to the use of their name. In this discussion I willingly agree to the proposal made by Burtt about the retention of early authors' names in cases where genera are conserved from a later date with a different type (Taxon 15: 307 (1966)). Glycine max (L.) Merr., Interpr. Rumph. Herb. Amboin.: 274 (1917); Hermann in U.S. Dep. Agr. Techn. Bull. No. 1268: 39 (1962) (further synonymy given). Phaseolus max L., Sp. P1.: 725 (1753). Dolichos soja L., Sp. Pl.: 727 (1753), non Glycine soja Sieb. & Zucc. Soja max (L.) Piper in Journ. Amer. Soc. Agron. 6: 84 (1914). [Glycine soja sensu Merr. in Trans. Am. Phil. Soc. 24: 208 (1935); Paclt in Science 109: 339 (I949), non G. soja Sieb. & Zucc.] Glycine soja Sieb. & Zucc. in Abh. Acad. Muench. 4(2): I19 (1845); Ohwi, Fl. Japan: 570 (1965). G. ussuriensis Regel & Maack in Regel, Tent. Fl. Ussur.: 50 (1861); Hermann in U.S. Dep. Agr. Techn. Bull. No. 1268: 37 (1962). The above references are sufficiently and to lead to further literature. full to illustrate the above discussion * See Verdcourt in Taxon 15: 34-6 (I1966). t I definitely do not support Paclt's proposal (Regn. Veg. 60: 33 (1969)) to reject this name as a confused name. After Piper & Prain's careful detailed typification (Piper, 1.c.: 75-84) whereby a Cliffortian specimen in the Linnaean Herbarium was chosen as the lectotype, it is a mischievous act to upset it. Incidentally Piper considered G. max and G. soja to be cultivated and wild forms of one species.

23 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 257 OPHRESTIA H. M. L. FORBES Several segregate genera have been split from Glycine Willd. (conserved as suggested by me in Taxon 15: 34 (1966)) by Hauman and F. J. Hermannthese include Pseudoeriosema Hauman, Pseudoglycine F. J. Hermann and Paraglycine F. J. Hermann. The characters separating these genera from Glycine are admittedly slight, although the first-named at least has a characteristic facies, and I have partly followed these authors' delimitations in my treatment of the group for the Flora. Several problems have, however, arisen and are dealt with here. My attention was drawn to the genus Ophrestia by the finding of a Natal specimen* tentatively labelled Tephrosia oblongifolia E. Mey. at the end of the Teramnus covers in the general herbarium of the British Museum (Nat. Hist.). Someone had obviously thought the specimen was more suitably placed in Teramnus but the Io fertile anthers rule out that possibility. Nevertheless the specimen clearly belonged to the Glycininae, and on checking it was found that the suggestion that it was the same plant as Tephrosia oblongifolia was also correct. Miss Forbes transferred this to a genus Ophrestia together with three other 'new species' and stated that it differed from both Tephrosia and Glycine. Hutchinson (Gen. Fl. P1. I: 396 (1964)) has relegated Ophrestia to Tephrosia but this is clearly not correct; Harms with his usual perspicacity very accurately considered it to be 'not quite a Glycine'. Miss Forbes does not mention whether she believes the genus to belong to the Galegeae or to the Phaseoleae, neither does she select a type species, although she obviously intended Tephrosia oblongifolia to act as such. Whilst studying the Natal specimen previously mentioned it was clear to me that it was congeneric with Paraglycine F. J. Hermann. He, like myself, had not considered Ophrestia for the simple reason it was still maintained in the Galegeae in herbaria and placed next to Tephrosia itself and one scarcely expects to have to look for relatives of the genus one is dealing with three or four tribes away. The differences between the tribes in this family are often tenuous and, sometimes, as in this case, it is scarcely possible to find clear technical characters which point in one direction. The subtle character of leaf-venation is the best indicator of affinity in this case, the widely spaced lateral nerves looping back and not forming a marginal nerve in Ophrestia compared with the closely parallel nerves which reach the margin in Tephrosia. The rather strange geographical distributions-ophrestia a Natal and Transvaal genus being unknown in Tropical Africa-Paraglycine a tropical African and Asian genus not occurring in South Africa-are now explained. The following new combinations are needed as a result of the amalgamation suggested and also in order to correct several previous errors. The characters given by F. J. Hermann to separate Pseudoglycine are I feel inadequate and the genus has exactly the same facies as Ophrestia. Ophrestia H. M. L. Forbes in Bothalia 4: Ioo3 (1948). Paraglycine F. J. Hermann in U.S. Dep. Agr. Techn. Bull. No. 1268: 52 (1962), synon. nov. Pseudoglycine F. J. Hermann, op. cit.: 74 (1962), synon. nov. * Actually Gerrard I082 the type-number of Ophrestia oblongifolia (E. Mey.) Forbes var velutinosa Forbes.

24 258 KEW BULLETIN VOL. 24 Ophrestia oblongifolia (E. Mey.) H. M. L. Forbes, op. cit.: oo004 (1948). Tephrosia oblongifolia E. Mey., Comm. P1. Afr. Austr.: io8 (1836). Glycine? wilmsii Harms in Engl., Bot. Jahrb. 26: 302 (1899), synon. nov. Ophrestia retusa H. M. L. Forbes, op. cit.: Ioo5 (I948), synon. nov. 0. swazica H. M. L. Forbes, 1.c. (1948), synon. nov. 0. nervosa H. M. L. Forbes, op. cit.: ioo6 (1948), nom. illegit., synon. nov. I have been unable to keep up Miss Forbes's species, the leaf characters she gives being quite unusable and variable. I am also unable to understand her treatment of Glycine wilmsii. She cites the name as a synonym under 0. oblongifolia adding the words 'in part' but gives Wilms 383 (the holotype and only element given of Harms's species) under her own nervosa. I have not sunk the name 0. oblongifolia var. velutinosa H. M. L. Forbes since this does seem a genuine minor variant. Ophrestia unifoliolata (Bak. f.) Verdc., comb. nov. Paraglycine unifoliolata (Bak. f.) F. J. Hermann, op. cit.: 57 (1962); Torre in Consp. Fl. Angol. 3: 244 (1966). Glycine unifoliolata Bak. f., Journ. Bot. 66, suppl. I: I14 (1928). Ophrestia unicostata (F. J. Hermann) Verdc., comb. nov. Paraglycine unicostata F. J. Hermann, op. cit.: 59 (1962). Ophrestia upembae (Hauman) Verdc., comb. nov. Paraglycine upembae (Hauman) F. J. Hermann, op. cit.: 60o (1962). Glycine upembae Hauman in Bull. Jard. Bot. Brux. 25: 94 (1955) & in Fl. Congo Belge 6: 99 (I954), non rite publ. Ophrestia digitata (Harms) Verdc., comb. nov. Paraglycine digitata (Harms) F. J. Hermann, op. cit.: 6I (1962). Glycine digitata Harms in Engl., Bot. Jahrb. 28: 40o8 (I9oo). Ophrestia radicosa (A. Rich.) Verdc., comb. nov. Eriosema radicosum A. Rich., Tent. Fl. Abyss. I: 228 (1847); Bak. in Fl. Trop. Afr. 2: 226 (1871). Glycine radicosa (A. Rich.) Bak. f., Leg. Trop. Afr.: 358 (I929). Paraglycine radicosa (A. Rich.) F. J. Hermann, op. cit.: 63 (1962). var. radicosa Glycine schliebenii Harms var. rufescens Hauman in Bull. Jard. Bot. Brux. 25: 95 & in Fl. (I955), Congo Belge 6: 98 (I954), non rite publ. Paraglycine radicosa (A. Rich.) F. J. Hermann var. rufescens F. J. Hermann, op. cit.: 65 (1962). var. schliebenii (Harms) Verdc., comb. nov. Glycine schliebenii Harms in Notizbl. Bot. Gart. Berl. I I: 814 (1933); Hauman in Fl. Congo Belge 6: 99 (I954).

25 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 259 [Paraglycine radicosa var. radicosa sensu F. J. Hermann, op. cit.: 63 (1962), non (A. Rich.) F. J. Hermann, sensu stricto.] var. enneaneura (Hauman) Verdc., comb. nov. Glycine schliebenii Harms var. enneaneura Hauman in Bull. Jard. Bot. Brux. 25: 95 (1955), & in Fl. Congo Belge 6: 99 (1954), non rite publ. Paraglycine radicosa var. enneaneura (Hauman) F. J. Hermann, op. cit.: 65 (1962). Hermann arranged his three varieties as if the type of Eriosema radicosum had been the glabrescent variety but a study of the type shows that it is densely rufous hairy, as indeed Richard's description plainly indicates. I have therefore had to make certain changes. Ophrestia hedysaroides (Willd.) Verdc., comb. nov. Glycine hedysaroides Willd., Sp. Pl., ed. 4, 3: lo6o (1802); Bak. in Fl. Trop. Afr. 2: 179 (1871); Bak. f., Leg. Trop. Afr.: 360 (1929); Hauman in Fl. Congo Belge 6: 98 (1954); Hepper in Fl. W. Trop. Afr., ed. 2, I: 564 (1958); Torre in Consp. Fl. Angol. 3: 243 (1966). G. pentandra Spreng., Syst. Veg. 3: 235 (1826). Johnia willdenowii Hook. f. in Hook., Nig. Fl.: 305 (1849). Paraglycine hedysaroides (Willd.) F. J. Hermann, op. cit.: 65, fig. 17 (1962). Ophrestia laotica (Gagnep.) comb. nov. Glycine laotica Gagnep., Not. Syst. 3: 196 (1916). Paraglycine laotica (Gagnep.) F. J. Hermann, op. cit.: 67 (1962). Ophrestia pinnata (Merr.) Verdc., comb. nov. Glycine pinnata Merr. in Lingnan Sci. Journ. 14: 15 (1935). Paraglycine pinnata (Merr.) F. J. Hermann, op. cit.: 69 (1962). Ophrestia pentaphylla (Dal1.) Verdc., comb. nov. Glycine pentaphylla Dalz. in Hook., Journ. Bot. 4: 344 (1852). Paraglycine pentaphylla (Dalz.) F. J. Hermann, op. cit.: 71 (1962). Ophrestia madagascariensis (F. J. Hermann) Verdc., comb. nov. Paraglycine madagascariensis F. J. Hermann, op. cit.: 73 (i962). Ophrestia lyallii (Benth.) Verdc., comb. nov. Glycine lyallii Benth. in Journ. Linn. Soc., Bot. 8: 266 (I865). G. lantzii Baill. in Bull. Soc. Linn. Paris I: 382 (1883), emend. Pseudoglycine lyallii (Benth.) F. J. Hermann, op. cit.: 74 (1962). In addition to these species there is a further striking plant which clearly belongs to this genus and has already been mentioned by Dr. Torre. It has been known since 1937 and Dr. Torre has informed me in a letter that he refrained from describing it since no ripe fruits were available. The plant is, however, so distinctive that, although fruits are still not available on any of

26 260 KEW BULLETIN VOL. 24 the three known gatherings, I have decided to describe it whilst dealing with the genus. Ophrestia torrei Verdc., sp. nov. ab omnibus speciebus adhuc descriptis, habitu erecto, foliolis multo latioribus, inflorescentiis longioribus differt; affinis 0. oblongifoliae (E. Mey.) H. M. L. Forbes, vexillo oblongo-cuneiformi, floribus minoribus distinguenda. Paraglycine sp. nov.?; Torre in Consp. Fl. Angol. 3: 244 (1966). Herba suffruticosa erecta vel frutex usque o-9 m. alta, ramis lignosis 4-5 mm. diametro pilis brevibus patentibus vel appressis dense subvelutine obtectis. Folia 3-5-foliolata; stipulae lineari-lanceolatae, 4-7 mm. longae, mm. latae, deciduae vel persistentes; petiolus 1-2 cm. longus; rhachis cm. longa; petioluli 1-3 mm. longi; foliola oblonga usque elliptica, 2-13 cm. longa, o cm. lata, apice rotundata vel retusa, minute mucronulata, basi rotundata vel levissime subcordata, chartacea, supra obscure appresse pilosa, subtus praecipue ad nervos venulisque dense pilosa, nervis lateralibus 7-I12-jugis supra subplanis subtus valde prominentibus; rete venarum dense reticulatum subtus valde prominens. Inflorescentiae elongatae, cm. longae, pseudoracemosae, floribus in fasciculas 5-0ofloras numerosas aggregatis; pedunculi cm. longi; pedicelli 0-5 mm. longi; bracteae persistentes, lineari-lanceolatae, 3-6 mm. longae; bracteolae filiformes mm. longae. Calyx pilis saturate brunneis dense obtectus; tubus campanulatus, 2 mm. longus, limbo bilabiato lobis 3 inferis ovatolanceolatis vel anguste triangularibus 2-3 mm. longis I mm. latis lobis 2 superis in labio rotundato 3 mm. longo et lato apice breviter bifido connatis. Vexillum oblongo-cuneiforme, vel leviter panduriforme, 6-8 mm. longum, 2'5-3'5 mm. latum, apice rotundatum, basi in ungue I mm. longo sensim angustatum, albo-viride, extra "5 superne pilis appressis saturate brunneis dense obtectum. Alae 5'5-7'5 mm. longae, valde calcaratae, ungue 2 mm. longo; lamina in parte superiore brunneo-pilosa. Carinae laminae oblongoellipticae, 4-5 mm. longae, 1-8 mm. latae, leviter calcaratae, in parte inferiore brunneo-pilosae, unguibus mm. longis. Filamentum vexillare liberum basi valde dilatatum. Ovarium fusiforme, 2 mm. longum, 2-ovulatum, dense pilosum; stylus basi subcomplanatus, glaber, superne leviter incrassatus, apice uncinatus, 2'5 mm. longus, stigmate terminali. Legumen ignotum. (Fig. I, p. 261.) ZAMBIA. Mwinilunga District: near source of R. Isongailu, in Brachystegia woodland on sand, 31 Dec. 1937, Milne-Redhead 3897 (K, holotype):- shrub to 0o9 m., inflorescence green, covered with brown hairs; petals greenish-white, with brown pubescence. ANGOLA. Lunda: R. Cuango, Xa-Sengue, without exact locality, Iooo m., Apr. I937, Gossweiler 11744; Alto Cuilo, 28 May 1954, Sanjinje V.54-9* (LISC). The genus as constituted above seems compact and has a more plausible distribution. I do not feel that it can be divided into sections equivalent to the genera I have sunk, since some species of 'Paraglycine' are closer to O. oblongifolia than they are to other species of 'Paraglycine'. Hermann divides his Paraglycine into two sections Digitatae F. J. Hermann and Hedysaroides F. J. Hermann but the latter, since it contains the type of Paraglycine,

27 Iq 10 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II ; FIG. I. Ophrestia torrei. I, habit, x 2; 2, enlarged portion of leaf, abaxial surface, x 3; 3, portion of inflorescence, x 3; 4, flower, x 5; 5, calyx, x 5; 6, standard, x 5; 7, wing petal, x 5; 8, keel, x 5; 9, androecium, x 5; io, gynoecium, x 5; i1, style and stigma, face view, x 14; 12, immature fruit, x 5. All drawn from Milne-Redhead 3897.

28 262 KEW BULLETIN VOL. 24 should have been called section Paraglycine; he includes five species in each of these sections separated on the degree of hairiness of the petals, the indumentum of the inflorescence and the number of ovules. Some striking differences are to be found in the genus in the length of the petioles, the length of the peduncles and the shape of the standard. 0. hedysaroides differs from the other species in its subsessile axillary clusters of flowers and its elliptic or rounded-oblong standard, contrasting with elongate false racemes and mostly panduriform standards. It might be worth retaining Paraglycine as a section for this species. Miss Forbes mentions the spurs at the end of the blades of the wing and keel petals and these are indeed noticeable in 0. oblongifolia where the lower corners of the standard blade are also acutely produced. These are present in some of the species previously referred to Paraglycine (e.g. 0. unicostata, 0. digitata) and absent in others (e.g. 0. hedysaroides). In Pseudoglycine the standard is broadly oblong-elliptic and the other petals are scarcely spurred; in fact little use can be made of this character. The digitate leaflets of 0. digitata and the very marked costae and narrow leaflets of 0. upembae and 0. unicostata are distinctive individual characters but on the whole Ophrestia, as emended, is a natural genus. PSEUDOERIOSEMA HAUMAN In the coastal areas of northern East Africa there is a plant closely related to P. borianii (Schweinf.) Hauman but differing from it in a number of ways. The general facies of the two is so similar that I have preferred to treat it as a subspecies. I believe the two have only quite recently evolved, owing to isolation resulting from late Pleistocene climatic changes. Pseudoeriosema borianii (Schweinf.) Hauman in Bull. Jard. Bot. Brux. 25: 98 (I955)- subsp. longepedunculatum Verdc., subsp. nov. a P. borianii typico pedunculis longioribus usque 5 cm. longis, inflorescentiis longioribus et laxioribus usque 3 cm. longis differt. Caules pilis ferrugineis obtecti. Vexillum spathulatum, mm. longum, 3'5-4'5 mm. latum. Legumen dense ferrugineo-pilosum. KENYA. Kilifi District: 5-6 km. from Malindi, in abandoned cultivations, Apr. 1960, Rawlins 838 (EA, K):-trailing vine with pink flowers. N. of Mombasa to Lamu and Witu, Whyte (BM). TANZANIA. Lushoto District: Lwengera Valley, 6'4 km. E. of Korogwe, tall grass areas, river valley flats, on black cotton soil, 300 m., 20 July 1953, Drummond & Hemsley 3385 (EA, K):-climbing on low vegetation by twining, height to I'm., calyx green, standard and wings pink, keel pale pink, ripe pods brown. Handeni District: Handeni, red clay soil in Vitex formation, I o July 1960, Semsei (K, holotype of subspecies; EA, PRE, isotypes) : trailing over bushes and trees, flowers purple-mauve. Tanga District: Magunga Estate, low-lying grassland which is swampy during the rainy season, 300 m., II July 1952, Faulkner 970:--plants spreading 2-3 ft., flowers pinky-red; Pongwe to Ngomeni, grassland in open bush association on black cotton soil, I50 m., I17 May 1930, Greenway 2245 (EA, K):-very common twining herb with pink flowers; Tanga, Tongoni, grassland and coastal savanna, sea-level, 24 June 1967, Faulkner 3961: flowers scarlet

29 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 263 fading to pinky mauve. Pangani District: Madanga, Bushiri, sandy soil, track through sisal estate, 45 m., 20 July 1955, Tanner 1963:--limp climber over grass; Bushiri, in grass, May 1950, Faulkner 555 (EA, K):-very spreading plant with bright red flowers, common; Hale to Kalilanga, 250 m., 17 Aug. I918, Peter K630 (O IV 330); Mkwaja, Mkaramo, mixed grassland, 7 July 1951, van Rensburg 545 (EA, K); Mwera, Ntaru, Nseko, in thick long grass on edge of permanent swamp on heavy black fissuring clay, 90 m., 6 June 1956, Tanner 2894:-creeper, flowers mauve and red, not aromatic, sap colourless; Mwera to Pangani, 29 Apr. 1926, Peter (B):-red-flowered trailing plant; Hale to the Great Pangani Falls, 9 June 1914, Peter 49708a (B). Pseudoeriosema andongense (Bak.) Hauman in Bull. Jard. Bot. Brux. 25: 97 (1955)- subsp. bequaertii (De Wild.) Verdc. comb. et stat. nov. Glycine bequaertii De Wild. in Fedde, Rep. Sp. Nov. I I: 547 (1913). Pseudoeriosema bequaertii (De Wild.) Hauman, l.c. (I955)- I prefer to treat this plant, which occurs in the Katanga and southern Tanzania, as a subspecies of the widely distributed P. andongense. It differs in having more cuneate leaf-bases, more contracted inflorescences, due to the suppression of the primary pedicels, shorter blunter calyx-lobes and a shorter standard but the characters are not entirely constant. I have examined all the types involved in Hauman's account of this genus and consider that P. homblei (De Wild.) Hauman is no more than a variety of P. andongense with trifoliolate leaves-de Witte 3147 has some leaves trifoliolate and some unifoliolate and clearly links the two. P. moeroi'nse (De Wild.) Hauman is, as Hauman points out, very close to P. longipes (Harms) Hauman, differing in scarcely more than its longer petioles, but it is still known only from a single specimen. A SURVEY OF THE GENUS TERAMNUS P.BR. WITH SPECIAL REFERENCE TO EAST AFRICA An attempt to sort out the East African species of this genus with only limited reference to material collected elsewhere proved abortive. This led to a survey of the genus throughout the world, but this has been carried out only so far as was necessary to obtain names for the members of the genus occurring in East Africa. The genus was for long confounded with Glycine Willd. (non L.)* and, even though described long ago, the transfer of all the species belonging to it occupied nearly a century. It is unquestionably a natural genus and clearly distinguished by both flowers and fruits (and when studied for some period by foliage also) from Glycine. As is usual this naturalness is accompanied by difficulty in deciding how many species can be recognized. In truth, the genus is scarcely divisible into species which are clear cut and, without resorting to cytology and experiments on living plants, it is likely that ordinary taxonomic treatments will remain unsatisfactory. Regretfully once * See Verdcourt in Taxon 15: 34 (1966).

30 264 KEW BULLETIN VOL. 24 again I have had to use an infraspecific hierarchy of names. I have much sympathy with those who feel that if anything can be distinguished at all it might as well be maintained at specific rank. With the immense amount of material now available it has been found that most taxa are not clear cut and infraspecific units are the only way of dealing with the situation. Use has been made of the smoothness or roughness of the seeds. The rough seeds have a minutely granular or reticulate covering and traces of this reticulate covering are often visible on the corresponding endocarp of the pod. This covering can frequently be removed by rubbing and it might be thought to be due to a secretion or decomposition of the pulp rather than part of the original testa. Dr. C. R. Metcalfe informs me, however, that he believes that the epidermal cells of the testa have broken down releasing their contents which have become deposited on the surface of the seeds. Hermann (U.S. Dep. Agr. Techn. Bull. No. 1268, II (1962)) when discussing Australian species of Glycine, mentions that the 'seeds are usually of two kinds in each species, very different in appearance one smooth and often glossy, the other variously muriculate or papillose and foveolate. The latter condition is due to the true seed coat being covered by the dried persistent remains of the perisperm which otherwise remains attached to the inner wall of the pod'. More field information is required. There does seem to be a significant correlation between rough seeds and other characters, and I have employed the character. In the past, use has been made of inflorescence characters but many species with usually elongated inflorescences can have the flowers in sessile clusters; such variation can be seen on a single plant. The holotype of T. labialis has the flowers in such clusters and too much reliance must not be placed on this character. KEY TO THE GENUS TERAMNUS P.BR. Pods (4-5-)5-6 mm. wide, glabrous or with adpressed hairs; leaflets usually distinctly acuminate (E. Asia) T. flexilis Pods 2-4 mm. wide or if 4-4'5 mm. wide then densely spreading hairy: Erect subshrub; leaflets oblong-elliptic, velvety hairy beneath; pod densely covered with more or less spreading, ferruginous hairs (W. Africa) T. buettneri Climbing, twining or prostrate herbs: Stems stout, the lower rounded, up to 4 mm. in diameter, the upper distinctly angular; leaflets velvety beneath, mostly large, rhomboid or ovate-oblong, 4-10o cm. long; pods densely, mostly adpressed hairy; calyx-lobes usually distinctly more hairy than the tube 7. T. micans (sensu lato) Stems more slender and if angular then leaflets narrow, oblonglanceolate: Calyx with dark blackish-brown hairs; leaflets usually narrowly oblong.... 6c. T. uncinatus subsp. ringoetii Calyx with white or ferruginous hairs but not so obviously blackish- brown: Seeds dark, minutely but distinctly reticulate or roughened when ripe, sometimes with a delicate wrinkled membranous wing: Pod with spreading hairs:

31 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 265 Upper 2 calyx-teeth united or almost so the calyx appearing 4-fid; leaflets elliptic; wings of corolla toothed (India, Burma) T. mollis Upper 2 calyx-teeth not entirely united, calyx distinctly 5-fid; leaflets round or oblong, obtuse; wings often not so toothed (E. Africa). 3b. T. repens subsp. gracilis Pod with adpressed hairs or rarely with rather spreading hairs; calyx 5-toothed; leaflets elliptic, usually with apex acute 2e. T. labialis subsp. arabicus Seeds smooth, pale chestnut or orange-brown to dark brown: Calyx with the 2 upper teeth entirely joined; wings of corolla with no tooth; pods mm. wide, sparsely adpressed hairy (New World)... I. T. volubilis Calyx 5-toothed (or more or less 4-toothed in some Indian variants of T. labialis): Pods, stems and leaves with spreading ferruginous hairs and leaflets round or rounded-elliptic, rarely exceeding 3 cm. in length; plant prostrate (Africa).. 3a. T. repens subsp. repens Pods with adpressed hairs or if spreading then leaflets oblonglanceolate or narrowly elliptic and plant a climber: Leaflets oblong, linear-oblong or narrowly elliptic, 2-6 times as long as broad: Leaves, stems and pods with a close ferruginous indumentum, often velvety and usually well obscuring the surface; leaflets mostly over 5 cm. long; inflorescences mostly with more numerous flowers and pedicels short (America and Africa): Leaflets predominantly narrowly oblong (Africa): Indumentum of calyx dark brown; inflorescence elongated 6c. T. uncinatus subsp. ringoetii Indumentum of calyx of brown and creamy hairs; inflorescences often subsessile clusters but sometimes elongated (leaflets often rather silvery beneath)... 6b. T. uncinatus subsp. axilliflorus Leaflets narrowly elliptic with rounded sides; pods mostly with spreading indumentum (America) 6a. T. uncinatus subsp. uncinatus Leaves, stems and pods or at least the pods with sparse indumentum not entirely covering the surface or if so leaflets under 5 cm. long; inflorescence often fewflowered and pedicels longer (T. labialis): Leaflets linear-oblong; pod indumentum sparse but spreading; pod mm. wide; plant of N. Somalia 3"5 2d. T. labialis var. somalensis Not as above: Leaflets more or less glabrous above; pod mm. 2-2"5 wide (Tropical Africa). 2c. T. labialis var. abyssinicus

32 266 KEW BULLETIN VOL. 24 Leaflets glabrous or pubescent above; pod usually mm. wide... 2a. T. labialis var. labialis and unnamed variants Leaflets round, broadly elliptic or rhomboid, I-3 times as long as broad: Leaflets larger, 5-1o x 4-9 cm.; usually robust climber (Tropical Africa): Leaflets rhomboid, acuminate; pedicels 1-2 mm. long 7c. T. micans var. fagifolius Leaflets ovate, acute or acuminate; pedicels 2-5 mm. long... 2b. T. labialis var. acutus* Leaflets smaller, usually well under 5 cm. long; weak climbers or straggling: Leaflets glabrous above and pod mm. wide (Tropical Africa).... 2c. T. labialis var. abyssinicus Leaflets glabrous or pubescent above; pod usually mm. wide.... 2a. T. labialis var. labialis and unnamed variants I. Teramnus volubilis Sw., Prodr. Veg. Ind. Occ.: 105 (1788). Glycine oblonga Benth., Bot. Voy. Sulphur: 84 (1844)- This closely resembles some forms of T. labialis but the 4-fid calyx is distinctive. The leaflets are elliptic or oblong, mostly glabrous above; pods sparsely adpressed pubescent; seeds smooth. JAMAICA. St. Mary's Parish, Orcutt 4113 & 4249; Charlton, near Ewarton, Harris 6695 (BM); Ferry River, Harris I11424 (BM). Without locality, Swartz (S, holotype*; BM, UPS, isotypes). TRINIDAD. Woodbrook, Broadway 6927 (BM, K). Tobago, Gomez's Bay, Eggers BRAZIL. R. Trombetas, Spruce. Belem, Ruhlmann Icou-Lau (?), beside the R. Javari, Traill 147. Matto Grosso, between Corumbi and Cuyaba, D. Smith 104 & Moore Ioio (BM). COLOMBIA. Nare, Holton. Bolivar, Magangue, Pennell Santa Marta, H. Smith 283- ECUADOR. Guayaquil, Sinclair (K, holotype of G. oblonga); Spruce 6354 (BM, K) & Jameson 572 (BM). It is assumed that the three lines of synonymy given by Swartz under T. volubilis, which obviously refer to T. uncinatus, have become transposed (see Fawcett & Rendle, Fl. Jamaica 4: 49 (1920) for discussion). 2. Teramnus labialis (Linn.f.) Spreng., Syst. Veg. 3: 235 (1826). Glycine labialis Linn. f., Suppl. PI.: 325 (1781). The typification of this species is crucial to an understanding of the genus as a whole. Fawcett & Rendle (F1. Jam. 4(2): 49 (1920)) mention that there is a Koenig specimen in the herbarium of the British Museum (Nat. Hist.). * An unnamed variant resembling subsp. arabicus will key here also.

33 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 267 In the Linnaean herbarium there are two sheets labelled G. labiata and ; the latter bears the name tanmanni crossed out and labiata added by J. E. Smith but the former has the letters HU (Hortus Upsaliensis) at the base of the stalk of the plant. The original description also contains the letters HU and, since the specimen 901o.15 is in substantial agreement with the description, I am accepting it as the lectotype despite the slight change in the spelling of the epithet. Linnaeus filius does not mention Koenig's name under G. labialis whereas in species to either side he does. For use in connection with the photographs which are available the following is a short description of the lectotype for the guidance of those who may wish to subdivide T. labialis. Stems slender, covered with half-adpressed white hairs. Leaflets pale, rounded-elliptic, up to 2 cm. long and I '4 cm. wide, rounded or emarginate at the apex, almost glabrous above save for hairs on the midrib and a few other scattered hairs, adpressed white pubescent beneath. Inflorescences axillary clusters; pedicels up to 2 mm. long; bracteoles 0-75 mm. long. Calyx adpressed pubescent all over; corolla small. Pod linear, 4-2 cm. long, 2-5 mm. wide, adpressed pubescent; pedicel 3 mm. long. In which is the same plant the seed is smooth. In most Indian material which I have referred to typical T. labialis the pod is wider, up to 4 mm. wide, the width in the type being less owing to immaturity. The division of T. labialis into infraspecific variants is extremely difficult, and has been attempted only in order to give names to East African groupings, which appear different in the field, but are linked with each other when the species is examined on a world basis. Other variants have not been formally named. Their status will have to be decided by those working on the floras of the areas concerned. I have not seen the types of all the synonyms involved but where the pod indumentum is described there seems little doubt as to the identity. Dalzell describes his Glycine warreinsis as having a 4-fid calyx and diadelphous stamens but adpressed pod indumentum; the type has not been seen and is presumably in Calcutta. Some Siamese specimens have the leaflets densely velvety beneath, e.g. Kerr 1531 B and many also have the flowers in axillary clusters, the reticulation of these characters giving rise to some distinctive looking variants. The actual working out of infraspecific variants in this area does not come within the scope of this paper. T. angustifolius is, I feel, no more than a narrow-leafleted form which intergrades with the typical form in the Philippines. I have seen no material from Formosa. 2a. subsp. labialis var. labialis Glycine parviflora Lam., Encycl. 2: 738 (1786). G. debilis Ait., Hort. Kew. 3: 34 (I789). G. pentandra Spreng. in Schrad. Journ. Bot. 2(1 & 2): 197 (1802). G. dolichoides Desv. in Ann. Sci. Nat., Bot. 9: 415 (1826). Teramnusparviflorus (Lam.) Spreng., Syst. Veg. 3: 235 (1826).?Glycine warreinsis Dalz. in Hook., KewJourn. Bot. 3: 21o (1851i). Teramnus debilis (Ait.) Prain, Bengal P1.: 393 (1903). T. angustifolius Merrill in Philipp. Journ. Sc., Bot. 7: 83 (1912).?Glycine subonensis Hayata, Ic. P1. Formosa 9: 27 (1920). C

34 268 KEW BULLETIN VOL. 24 Leaflets rounded or elliptic, often glabrous above but pubescent above in many specimens from the Philippines. Pods (2-5-)3-4 mm. wide, glabrescent or adpressed pubescent. Seeds smooth. Some of the material cited below matches the type in that at least some inflorescences are more or less sessile clusters but every intermediate links these with other specimens with longer inflorescences and in fact the great majority of sheets of T. labialis show elongated inflorescences. Baillon (Bull. Soc. Linn. Paris I: 382 (1883)) includes Galactia sericea Pers. in the synonymy of T. labialis. A photograph of the type of the latter kindly sent from Paris shows that it is not a Teramnus -the pods are not turned up at the apex. The type is a Commerson specimen from Bourbon, No in the Jussieu Herbarium, and also bears the name Clitoria (phryne). This specimen is unfortunately not borrowable but appears to be closely similar to Scott Elliot 2465 from Fort Dauphin, Madagascar and wrongly named Teramnus labialis. I am content to leave the species in Galactia at present. INDIA. Punjab: Sikh States, Ambala, Edgeworth 154; Punjab, Drummond 24608, 24609, 24610, 24618, 24628, 24629, & Uttar Pradesh: Moradabad, Thomson 367 (BM, K). Bihar: Plains of Behar, Hooker; Gya, Clarke (BM); above Monghyr on the Ganges, Madden; Bihar and Orissa, Japla, Palamau, Haines Bengal: Beauleah, Clarke (BM). Gujarat: Ahmadabad, Saxton 526. Madhya Pradesh/Maharashtra: Raj Nandgaon, Haines Maharashtra: Chanda District, Duthie 9459; Concan, Law & Stocks (BM, K); Bombay, Gaburne; Dalzell*. Orissa: Patna, Turekela, Mooney 2415; Kalahandi, Morijuska in Kutiya-Kond Tract, Mooney Andra Pradesh: Godavani District, Samalcoltah, Heyne. Mysore: Belgaum, Ritchie 182; Kolar, Bourne Kerala: Travancore, Lawson I85. Madras, Shuter; Thomson; Barber 78; Tranquebar, Roxburgh; Rottler; Coromandel, Koenig (BM). Locality dubious: Bunda, Edgeworth; NW. Plains, Falconer; Damboul (? Ceylon), Beckett; Teynampet, Bishops Compound, Bourne; Herb. Wight 55o8H, 55 IoC; sheet of elements of Wallich Cat. No. 55o8 (BM, K); Herb. Wight 221, 738 & 877 (BM, K); Sirivaka, Bourne 3454; E. India, Koenig (BM, syntypes of Glycine debilis Ait.). BURMA. Wallich 55o8G. Meiktila, Collett 42. Upper Burma, Huk (BM). E. PAKISTAN. Dacca, Clarke 6752, 7894 (BM). CEYLON. Peradenia, Gardner 211 (BM, K). Jungle Nalande, Simpson 9230 (BM). Pahalaveva Dambulla Tank Margin, Simpson 9801 (BM). Without locality, Thwaites 1471, (BM pro parte, K). PENANG. Wallich Cat. No (BM, K) & 5519 (BM, K). THAILAND. Sriracha, Collins 274 & Chiengmai, Kerr 1531B. Bangkok, Marcan 486 (BM, K) & Kerr INDO-CHINA. Thorel. VIETNAM. Saigon, Godefroy-Lebeuf. INDONESIA. Java: W., road from Dago to Tjiburiah, Holstvoogd 278; without locality, Horsfield 125; Res. Kediri Gadoengan Korders Pard, 22835; Res. Semarang, Kedoengdjati, Korders 24925; SE. Java, Forbes I23oE (BM). N. Celebes, Gorontalo, Riedel. PHILIPPINES. Luzon: Bontoc, Vanoverbergh 716 (BM, K, isotypes of T. * Dalzell collected a series of all the variants of T. labialis; No. 8 has the upper calyx-lip entire and the pods 3 x o04 cm. It might be of hybrid origin-it certainly is not, as indeed Dalzell points out, typical T. labialis.

35 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 269 angustifolius); Ilocos, Bangui, Ramos in P.B.S (BM, K); Pefiablanca, Ramos &Eda7io46564(BM); Manila, Loher 2314 & 2315 pro parte; Herb. Mus. Petropol.? 165; La Union, Bauang, Finix in P.B.S & Elmer 5586; Bataan, Mt. Mariveles, R. Lamao, Williams 268; Rizal, Bosoboso, Ramos Mindoro: Mansalay, Bo Manaul, Sulit in P.B.S Without locality, Cuming (BM, K). NEW GUINEA. New Britain, 40 km. from Rabaul, S. Coast, Ulaveo Plantation, Kokopo, Froggatt 2. MICRONESIA. Guam: Upi, Bryan 1268; 3-2 km. S. of Agat, near beach, Rodin 783- MADAGASCAR. Central Madagascar, Parker. RODRIGUES. Balfour (BM, K). LOCALITY UNCERTAIN. Indes Orientales, Sonnerat* (P, holotype of Glycine parviflora; K, photoholo). var. A Bujacia gampsonychia E. Mey., Comm. Fl. Afr. Austr.: 127 (1836). Glycine meyeri Benth., Comm. Leg. Gen.: 62 (1837). G. gampsonychia (E. Mey.) Walp. in Linnaea 13: 533 (1839)- This is scarcely different from the forms occurring in the Philippines. The stem is covered with a spreading ferruginous indumentum and the oblong or elliptic leaflets are pubescent above and below. The material from Madagascar scarcely differs. MOZAMBIQUE. Maniamba, Malulu, Jessi Mt. (Jeci), Pedro & Pedrdgio 4020 (EA). SWAZILAND. Hlatikulu, Gollel, Karsten. SOUTH AFRICA. Transvaal: Rustenburg, Nation I904; Pretoria, Leendertz 460 & Schlechter 4148 (BM, K); Lydenberg, Wilms 366 (BM, K); Shilouvane, Junod 582 & 2390; Nelspruit, Codd & de Winter 5133 & Acocks 16664; Brits, Mogg 16679; Magaliesberg, Schlieben 7832; Kruger National Park, Pretoriuskop, van der Schiyff 1477; Houtbosh, Rehmann Natal and Pondoland: Drige. Natal: Northdene, Wood 3183; Helpmekaar, Rorkes Drift, Codd 8242; Hlabisa, Ward 2407 & 1769; Umzimkulu, Clydesdale, Tyson 1443 & 2088 (BM, K); Drackensberg, Cooper 2249; without locality, Gerrard 1573 (BM, K). Cape Province: Kentani, Pegler 1443 (BM, K); Komgha, Drige; between Omtendo and Omsamculo, Drege (BM, syntype of Bujacia gampsonychia); Cape Town, Cape Flats, Rehmann (BM) (is locality correct?). MADAGASCAR. Central Madagascar, Baron 3807, 2663, 2664, 2398 & Parker. NW. Madagascar, Baron Imeri mandroso, Decary 375 (BM, K, P). 2b. var. acutus Verdc., var. nov. a var. labiali foliolis majoribus usque 7 cm. longis 4 cm. latis apice acutis differt. [T. labialis sensu Hauman in Fl. Congo Belge 6: o103 (I1954), pro parte; & sensu Hepper in Fl. West Trop. Afr., ed. 2, I: 573 (I958), pro parte, non (L.f.) Spreng. sensu stricto.] This is a rather diffuse variety covering glabrescent West African plants

36 270 KEW BULLETIN VOL. 24 and more pubescent East African plants. It is mostly a forest variant and may be merely an ecological one. SIERRA LEONE. Hanga, N. W. Thomas LIBERIA. Vonjama District, Zigida, Baldwin GHANA. Anum. Morton GC km. N. of Kumasi, Morton GC km. S. of Kumasi, Morton A km. beyond Senya Bereku, Morton GC7945. Kpeve Agricultural Station, Irvine 1742 (mixed with Vigna sp.). 6-4 km. beyond Akisuma, Morton GC7945. Roadside, Ejura Scarp, Morton A1552. N. of Nangodi, Morton A1265. ToGo. Kpandu, Robertson 51 (BM). NIGERIA. Lagos, Millen I14 pro parte & 61. Ibadan, Jones & Keay in F.H.L CAMEROUN. Bipinde, Zenker 4156, 4809 (K, holotype of variety; BM, isotype). Yaunde, Zenker & Staudt km. S. of Fort Foureau, near Ngouama, Letouzey km. N. of Bertoua along road to Deng Deng, Breteler, de Wilde & Leeuwenberg CONGO (KINSHASA). Oubangi, between Bulinga and Banzyville, Lebrun 2025 (BM, BR, K). UGANDA. Mengo District: Bulemezi [Bulimezi], Luwero [Luwelo], Maitland 905; Kampala, Mutungo, Chandler 1485; Entebbe road, Kajansi Forest, Chandler 1240 (this sheet bears a remarkably close resemblance to the type of T. mollis from India, showing the plasticity of the genus); Mabira Forest, Mulange, Dummer 41o8 (BM, K, US). KENYA. Trans-Nzoia District: cultivated at Kitale from seed collected near Kisii, Bogdan 3982 (EA, K) & Boonman KL 37/51265 (EA, K) (probably belongs here). Plants almost identical with those from Uganda have been found in Madagascar, e.g. Baron The leaflets are usually glabrous above and the variety grades ivith typical Indian var. labialis. In the Usambaras a trailing plant with very acute leaflets up to 6 cm. long and 2-5 cm. wide, pods with spreading hairs and smooth seeds perhaps represents an allied new variant but it has the facies of subsp. arabicus. TANZANIA. Lushoto District: W. Usambaras, above Kwashemshi, Peter (B). Tanga District: Longuza [Longusa] rubber plantation, Peter (B). 2c. var. abyssinicus (A. Rich.) Verdc., comb. et stat. nov. Glycine abyssinica A. Rich., Tent. Fl. Abyss. I: 212 (1847). [Teramnus axilliflorus sensu Hauman in Fl. Congo Belge 6: o05 (1954), pro parte, non (Kotschy) Bak. f.] [T. repens sensu Hauman in Fl. Congo Belge 6: 102 (1954), pro parte, non (Taub.) Bak. f.] Much of the material of T. labialis from the highland areas of East and Central Africa is distinctive in facies and fairly well characterized by the leaflets being small, round or narrowly elliptic-oblong, glabrous above and the pods much narrower than in typical Indian var. labialis, 2-2"5 mm. wide. The seeds are smooth. A number of variants have been included under this general varietal name, some of which approach T. uncinatus, particularly its

37 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 271 subsp. axilliflorus, and no hard and fast characters can be given to distinguish them. Some of these minor variants are in their extreme forms quite distinctive but no more than populations which will at present interbreed with ease. In East Africa var. abyssinicus is clearly distinct from subsp. arabicus and almost entirely occupies a different area of distribution, the latter being mainly coastal. Without a study of the world variation one would doubtless treat them as specifically distinct. Typically var. abyssinicus has the stems with a rather adpressed pale indumentum; other specimens have a mostly spreading, pale ferruginous indumentum on the stems and the leaves, although glabrous above, are often quite densely covered with more or less adpressed, ferruginous hairs beneath. The pods have a sparse adpressed or sometimes denser, rather loosely adpressed, sometimes even velvety indumentum. No attempt has been made to sort out these variants which grade together imperceptibly. CONGO (KINSHASA). Beni-Kasindi, Lebrun Lula, Lebrun 8I92. ETHIOPIA. Adowa, Schimper 1057 (P, lectotype; BM, K, isolectotypes). Nadda, Mooney km. NW. of Harar, NE. of Lake Alemaya, Berhe 23. Tigre v Begemder, Memsach, Schimper 526 (BM, K). Saroe, Adi Gang, Pappi 191 (? 248) (BM, K). Agherimariam, Gillett Eritrea, Mt. Savour, Pappi 5374 (EA). UGANDA. Bunyoro District: Masindi, Hazel 283 (EA, K). Toro District: Semliki Flats, Sempayo-Kibuku road, Liebenberg 948. Ankole District: R. Ruizi, Jarrett 220 (EA, K). Kigezi District: Rukungiri, Purseglove 2903 (EA, K). Teso District: Kasilo, Chandler 307; Serere, Chandler 204 p.p. & 890. Mbale District: Bugisu [Bugishu], Budadiri, Chandler 507. Mengo District: Kampala, Hargreaves TI224; Kampala, Makerere Hill, Chandler & Hancock 148; W. Kampala, Kawanda, Chandler 1547; Victoria Nyanza Region, Maitland 127, 170; Banda, Dummer 430 (BM, K) & 970 (BM, K). KENYA. Uasin Gishu District: Kipkarren, Brodhurst-Hill 407 (EA, K). Machakos District: 01 Doinyo Sabuk, Birch B59/I23 (EA). Masai District: Olulunga Rest House, Glover et al (EA, K). TANZANIA. Arusha District: Mt. Meru, Peter (B). ZAMBIA. Mumbwa, Macaulay 663. Solwezi District: R. Mwafue, Milne- Redhead 731. Mwinilunga District: Mwinilunga to Kabompo, km. 72, White Petauke District: hot spring near old boma, Verboom 681. RHODESIA. Salisbury District: Henderson Research Station, S.R.G.H Umtali, Chase 8414 (K, SRGH). ANGOLA. Ambriz, near Mubango, Welwitsch A further variant has smaller, broadly elliptic leaflets very similar to typical var. labialis but still with narrow pods. Part of Teramnus repens sensu Hauman in Fl. Congo Belge 6: 1.o2 (1954) belongs here. CONGO (KINSHASA). Kivu, Lake Edward, Vithumbi, van der Ben 282; Lake Kirwa, Lebrun ETHIOPIA. Jimma area, Milan Teffari, Siegenthaler 1532 (EA, K). UGANDA. Toro District: Bwamba, Hakitengya-Buyayu road, Maitland lo092; shores of Lake Edward, G. Taylor 2647 (BM); Bwamba; Busaru- Bubandi [Bubande] road, Liebenberg 907. Mbale District: Bugisu [Bugishu], Budadiri, Chandler 489. Mengo District: Kampala, Chandler 2424 & Snowden

38 272 KEW BULLETIN VOL (BM, K); Old Entebbe, Harker 491 (EA, K); Jumba, Dummer 3041 (BM, K). TANZANIA. Moshi District: E. Kilimanjaro, Mrao [Mraw], Haarer 673 (EA, K); Kilimanjaro, Marangu, Volkens Arusha District: SW. Meru, Haarer 242 (EA, K). Buha District: Kasulu, Heru Chini, Rounce 35 (EA, K). Rungwe District: Bomala-Kitana, Stolz 915- A fairly distinctive minor variant with a more marked spreading ferruginous indumentum on the pods closely resembles forms of T. uncinatus but has the pods narrower; it may be the result of hybridization. Richards cited below has two very similar shoots mounted together one with pods spreading pubescent, the other with them adpressed hairy. CONGO (KINSHASA). Kisolo, Beni, Gille 132. Nioka, Aru, Liben 399- UGANDA. Ankole District: Mbarara, Harker 53 (EA, K). KENYA. Northern Frontier Province: Mt. Nyiro, Kerfoot 1920 (EA, K) (form with large, more broadly elliptic leaflets approaching var. acutus). Trans-Nzoia District: E. Elgon, Endebess, Irwin 137 (EA, K). S. Kavirondo District: Kisii, Napier 2993 in C.M (EA, K). TANZANIA. Mbulu District: Gt. North Road, Pienaars Heights, Polhill & Paulo 2332; Lake Manyara National Park, Endabash, Greenway & Kanuri (EA). Arusha District: Meru, near Ngongongare, Peter (B); Engare Nanyuki, Peter (B); Ngurdoto National Park, Lake Rashitany, Richards 20o42 in part (EA, K). Moshi District: Ngare Nairobi, van Rensburg 583. Pare District: S. Pares, near Mbaga, Peter 49468*. Lushoto District: W. Usambaras, near Lushoto [Wilhelmstal], Jagertal, Peter d. var. somalensis Vatke in Oesterr. Bot. Zeit. 28: 261 (1878). This plant has been synonymised with T. uncinatus subsp. axilliflorus (Kotschy) Verdc. but does not seem to have any affinities in that direction. The leaflets are very elongated with the veins leaving the midrib at almost a right angle; the indumentum is mainly white; pods with sparse spreading hairs and smooth pale seeds. It is geographically isolated and no further material has turned up. It may be a distinct species but is left at this rank for the present. SOMALI REPUBLIC (N.). Meid, Serrus Mts., Hildebrandt 1389 (B, holotype of varietyt; BM, K, isotypes). 2e. subsp. arabicus Verdc., subsp. nov. a subsp. labiali seminibus granulatis non nitidis differt. Glycine senegalensis DC., Prodr. 2: 242 (1825). The leaflets are nearly always adpressed pilose above and the pod is adpressed pilose save in a few uncommon variants. The position of several West African populations is not at all certain owing to a lack of ripe fruits. These are mentioned at the end. Subsp. arabicus has the pods about mm. 3"5 wide, thus resembling typical subsp. labialis var. labialis. It is possible that the occurrence of this subspecies in America is due to accidental introduction * Peter numbers not followed by a herbarium abbreviation are at both Berlin and Kew. t Specimen believed to have been destroyed.

39 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 273 from the coastal parts of East or West Africa. Some specimens may resemble T. uncinatusubsp. andongensis but in general T. labialis and its variants have much longer pedicels. The stem and inflorescence indumentum varies from closely adpressed and sparse to dense and spreading but intermediates render it inadvisable to further subdivide the subspecies at this stage. Since exactly parallel forms occur in both East and West Africa varietal status is indicated. ARABIA. Fatme near Unsert, Schimper 900 (Kennedya arabica Hochst. & Schimper, nomen) (K, holotype of subsp.). SENEGAL. Pays de Walo, Perrotet 222 (BM, isotype of Glycine senegalensis). MALI. Kara, Davey I93- GHANA. Labadi, Morton AI745. Achimota, Irvine Tema, Morton A989. PRINCIPE. S. Antonio, R. Papegaio, Monod I2o68 (BM). Without locality, Barter. S. TOME. Praia Melaio, Monod I2288 (BM). ANNOBON. Towards Pico de Fogo, S. of Ambo, Melville I09 (BM). SUDAN REPUBLIC. Blue Nile Province: Geziret el Fil, Andrews o01; Sennar, Karkoj, Brown 730; Kordofan: Nuba Mts., Wadelka, Andrews 220; Mulbas, Kotschy 300 (BM, K). Equatoria: Bahr el Ghazal, (?) Kitt, Schweinfurth 1264 (BM, K). UGANDA. Acholi/Bunyoro District boundary: Paraa, Harker 644 (EA) (? small leaved and atypical, lacking ripe pods). KENYA. Machakos District: Kiboko, Bogdan Lamu District: Lamu to Witu, Whyte. Kilifi District: I -6 km. W. of Garashi, Moomaw 917 (EA). Kwale District: Ukunda, Symes 179 (EA). TANZANIA. Mbulu District: Lake Manyara National Park, Endabash, Greenway & Kanuri (EA, K) & (K)*. Lushoto District: Makumba to Korogwe, Peter (B, K); near Mashewa, Peter (B); T.T.C. site, Archbold 51 ; Amani to Derema [Nderema], Grote in E.A.H (EA, K); Fanusi, Zimmermann in E.A.H (EA, K); R. Pangani near Korogwe, Peter p.p.; Amani, Kiumba, Greenway 3930 (EA, K). Tanga District: Tengeni, Greenway I943 (EA, K) & Peter (B); Muheza, Milne-Redhead & Taylor 7274 (EA, K); Mlingano Sisal Research Station, Grundy L77t, L49 (EA, K); Sawa, Faulkner 2011; Kigombe Beach, I -2 km. NE. of I Pangani, Drummond & Hemsley 3326 (EA, K); Nyika, Bariti to Tanga, Peter (B); R. Sigi, forest below Longusa, Peter 48585; Longusa to Magrotto, Mt. Mlinga, Peter (B); Sigi to Mpandeni [Pandeni], Peter (B). Pangani District: Bushiri Estate, Faulkner 658 (EA, K). Bagamoyo District: Saadani, Peter (B). Uzaramo District: Dar es Salaam, Mbagaza See, Peter Zanzibar: Massazine, Faulkner 2480; Ndgaa road, R. O. Williams 85; Fumba, Oxtoby 16; Mwera Swamp, Vaughan 1712; without locality, Hildebrandt 958 (BM, K) (a form with oblong leaves); Pemba, Chake Chake, Vaughan 497 (EA, K). MOZAMBIQUE. Niassa, between Meconta and Nampula, 23 km. from Nampula, Pedro & Pedrdgao Sul do Save, between Boane and Moamba, 34'9 km. from Moamba, Myre & Carvalho ZAMBIA. Katondwe, Fanshawe Mpika District: Luangwa Valley, M'fuwe Game Camp, Verboom 810o. * EA duplicate is subsp. labialis var. abyssinicus. t Originally from Lindi District: Lukuledi Valley.

40 274 KEW BULLETIN VOL. 24 RHODESIA. Mazoe District: cultivated at Henderson Research Station (indigenous to area), Wild Ndanga District: N. side of R. Lundi, Chipinda Pools area, Goodier 796. MADAGASCAR. Nossi B6, Hildebrandt 2887 (BM, K). COMORO IS. Johanna, Hildebrandt 1590 (BM, K) & Bojer. Mayotte, Boivin. SEYCHELLES. Mahe, Lionnet I & Dupont 2. La Digne, Horne 295 & 477- Without locality, Wright (BM). MAURITIUS. Port Louis, Bouton; Johnston. Pamplemousses, Bouton. Without locality or locality not decipherable: Bojer; Bouton; Ayres. REUNION. St. Benoit, de l'isle; without locality, Balfour. JAMAICA. Black River, Harris 9923 (BM); Ferry River, Harris oo0040 (BM, K); St. Elizabeth, Sangster 50o8 (BM); Old Harbour, Ridley. Without locality, MacFadyen. HAITI. Port-au-Prince, Ekman VIRGIN IS. St. Thomas, Eggers (BM). LEEWARD IS. Tortola: Fishlock TI53 & T276. Antigua: Donovans, Box 958 (BM); Sandersons, Box 1014 (BM). WINDWARD IS. St. Lucia: Velez 3327A. St. Vincent: Squire; H. H. & G. W. Smith Barbados: Welchman's Hall Gully, Sandwith. TRINIDAD. Tobago: Milford Main road, Freeman & Williams in T.H ; Government Farm, R. O. Williams in T.H ; Government House, R. O. Williams in T.H. I2Io8; Store Bay, Sandwith 1825; Botanic Station, Broadway 9051 (BM); Highmoor, Broadway 3060 (BM). Trinidad: Reform, Broadway 8904 (BM); Botanic Station, Broadway 9 (BM). GUYANA. Georgetown, Kortwright. There are a number of West African sheets that probably belong here- Senegal (a 'Roger Dedit' sheet 89-authentic specimen of Glycine senegalensis DC.); Senegambia (? Heudelot I I4-a drawing on the sheet indicates a 4-fid calyx but this could not be checked from 'the material available); Galam (Heudelot; Guillemin 1839); N. Nigeria: Bauchi (Lely 632); S. Tome (Moller); Annobon (Burton); Ghana: Achimota (Milne-Redhead 5167), Accra Plains (Irvine 3028; Rose Innes GC30075). These sheets also come very close to South African material and to Philippine material. H. & M. Stehl6 (Fl. Agron. des Antilles Frangaises, 3, F. des L'gumineuses et Anti-Erosion: 78 (196o)) also records it from Guadeloupe, Grand Terre and Martinique. An uncommon variant of subsp. arabicus has spreading hairs on the pods. UGANDA. Karamoja District: Namalu, Wilson 1707 (EA). Teso District: Serere, Chandler 614. TANZANIA. Mbulu District: Tarangire, Mahinda 454 (EA, K); rift wall estate W. of Lake Manyara, Frahm Leliveld 62130; Karatu, van Rensburg 196 (EA). Moshi, Haarer 595 (EA, K). Lushoto District: Magunga, near Korogwe, Peter (B, EA, K). Tanga District: Maramba, Mwele to Tanga, Peter 48489, (B, EA, K). Morogoro District: 41-6 km. NE. of Morogoro, 2.4 km. NE. of Kitulanghalo Hill, Welch 611 (EA, K). Some of these are scarcely different from T. repens subsp. gracilis and are possibly the result of hybridization. A rather well-marked variant resembling subsp. arabicus in leaf-shape, inflorescence and indumentum but having spreading hairs on the pods and

41 STUDIES IN THE LEGUMINOSAE-PAPILIONOiDEAE: II 275 some of the facies of T. uncinatus (L.) Sw. is also recognized. It might be of hybrid origin. The following three specimens have been examined. ZAMBIA. South: Mazabuka District, Monze to Magoye, km. 31, White 7256 (FHO, K). RHODESIA. North: Gokwe, on road from Gokwe to Copper Queen, Bingham 946 (K, SRGH). East: Inyanga, Cheshire, Norlindh & Weimarck 4793 (K, LD). 3. Teramnus repens (Taub.) Bak. f. in Journ. Bot. 66, Suppl. I: (1928). II15 Glycine repens Taub. in Engl., Pflanzenw. Ost-Afr. C: 220 (1895). This has a distinctive prostrate habit and almost round leaflets; the pods nearly always have a spreading indumentum, but in Zambia, specimens have been found with almost adpressed hairs (e.g. Robinson 911). There is little difficulty in maintaining this species if Africa alone is considered but in India the position is different, T. mollis being close. The calyx character is not entirely constant; T. mollis usually has the upper two teeth entirely joined but sometimes they are partially joined to form a bifid lobe. Leaf shape also varies immensely. Material from India has not been adequate to reach a decision but what I am calling T. repens in Africa certainly occurs in India. I am maintaining two subspecies based on seed characters. 3a. subsp. repens Seeds smooth, pale. BURUNDI. Bugesera, Buhuha, Liben RWANDA. Gabiro, Lebrun 9567 (BM, BR, K). UGANDA. Bunyoro District: Bulisa [Bulesa], Purseglove 1097 (EA, K); Butiaba, Mearns 2754 (US). Toro District: Ruwenzori, Muhokya, Maitland Kigezi District: Queen Elizabeth Park, Ishasha R. camp, Symes 701 (EA, K). Teso District: Bukedea, Harker 24 (EA). KENYA. Mt. Elgon foothills, cult. at Kitale, Strange 162 (EA). TANZANIA. Bukoba District: R. Kagera, Kavingo, Stuhlmann 2000 (B, holotypet; BM, fragment); Karagwe, near Mabira, Haarer 2215 (EA, K). Buha District: Kasulu, Heru Chini, Rounce 15 (EA, K). ZAMBIA. Lusaka District: 19-2 km. S. of Lusaka, Mt. Makulu Research Station, Angus 1298 (EA, K), Verboom a & van Rensburg 3093 (BM, K). 19'2 km. N. of Choma, Muckle Neuk, Robinson 91 I1. Mumbwa, Macaulay 393. RHODESIA. Marandellas, Grasslands Research Station, Corby 978. ANGOLA. Cuanza Sul, Seles, Gossweiler 9309 (BM, COI, K). Cuanza Norte, Quizenga, Gossweiler 6340 (BM, COI). Huila, Capello & Ivens 26 (some of these have roughish seeds). 3b. subsp. gracilis (Chiov.) Verdc., comb. et stat. nov. T. gracilis Chiov. in Ann. Inst. Bot. Roma 8: 432 (1908). [T. labialis var. mollis sensu Balf. f. in Trans. Roy. Soc. Edinb. 31 : 82 (1888), non (Benth.) Bak.*] * Balfour's citation of Kennedy arabica Hochst. & Steud. (nomen) as a synonym is not correct.

42 276 KEW BULLETIN VOL. 24 Seeds rough. The typical form of this is certainly very different from the type of T. mollis, particularly in leaflet shape, but further work may show that the two are not specifically distinct. East African material has distinctly more oblong leaflets than the types but there is much variation. INDIA. Bombay, Dalzell. AUSTRALIA. Cultivated at Canberra from seed collected in Tanzania, Lake Manyara, Morgan in CPI (seeds not ripe and placing not certain). SOCOTRA. Tamarid, Schweinfurth 287. Galonsir, Balfour 143, 243 (BM, K). Jebel Rughid, Smith & Lavranos 587. COMORO IS. Mayotte, Boivin. CONGO (KINSHASA). Parc National Albert, Katanda, Lebrun SUDAN REPUBLIC. Bahr el Jebel, W. channel, Sheik Tombe, H. M. & A. F. Broun ETHIOPIA. Eritrea: Dahlak Archipelago, Terracciano 739t (FI, syntype of T. gracilis); Assaorta, Henrob, Pappi 6o (FI, syntype of T. gracilis); Ingal- Ras Koral, Terracciano 16 (FI, syntype of T. gracilis); Amasen, Pianura di Sabarguma, Pappi 4057 (FI, syntype of T. gracilis). KENYA. Kwale District: S. of Mombasa to Shimoni, Whyte (BM). Mombasa District: Likoni, Thorold Kilifi District: Kilifi, Mfumbini, Jeffery K235 (EA, K); Malindi, Sabaki Bridge, Rawlins 848 (EA, K); Malindi, Tweedie 1015, 3142 & Bogdan Lamu District: Tangawanda, Rawlins 89 (EA, K); Witu area, Rawlins 63 (EA, K); Utwani Forest Reserve, Mambasasa, Greenway & Rawlins 9355 (EA, K). TANZANIA. Tanga District: Sawa, Faulkner 2183 & 3855; Tanga-Pangani road, Machui, Faulkner 1658 (EA, K) & 3519; Kange, Faulkner 817 & 2098; Kigombe to Tangata, Peter (B). Bagamoyo District: lower path, Ruvu to Saadani, Peter Uzaramo District: Ruvu Estate, Grundy L83 (EA, K) & Rykebusch 126 (EA, K); Dar es Salaam, Revell 129. Zanzibar: Mbweni, Vaughan 1055 (EA, K); Migombani, Vaughan lo99 (EA, K). LOCALITY NOT TRACED.? Togodele, Ehrenberg. 4. Teramnus mollis Benth. in Journ. Linn. Soc., Bot. 8: 265 (1865), non Glycine mollis Wight & Arn., Prodr. Fl. Ind. Or.: 209 (1834).- Wight & Arnott distinctly describe the pods of their Glycine mollis as adpressed pubescent and the Wallich collection sheets they cite support this. Bentham describes a different plant with patently pilose pods and I am treating his name as a new name with a new type and not as a transference of Wight and Arnott's epithet. This is, I feel, permissible, because after citing their name Bentham adds ex parte and this can be assumed to exclude their types which he agrees are but forms of T. labialis. The combination of long spreading hairs on the pod, roughened seeds and often only 4-fid calyx, due to the complete fusion of the two upper calyxteeth, render this plant easy to distinguish, but the status is not easy to decide. A few intermediates with T. labialis occur in the Bombay area and small-leaved specimens are very similar indeed to T. repens subsp. gracilis. The matter is complicated by the fact that the lectotype is somewhat different from the rest of the Indian material. The species needs study in t 736 cited in original paper of Chiovenda is an error.

43 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 277 India but since typical T. mollis does not occur in Africa I have not pursued the matter further. The seeds frequently have a thin membranous keel and C. B. Clarke remarks 'seeds black with a green line round'. INDIA. Uttar Pradesh: Saharanpur, Edgeworth 152; Moradabad, Thomson. Gujarat: Saurashtra, Gurnar, Raizada 21184; Gir, R. Hiran, Raizada Maharashtra: Bombay, Dalzell. Orissa: Kalahandi, Lanjigarh, Mooney Mysore: W. Ghats, Castle Rock, Bell 4324; Tambachury Ghat,? Wynadd, Beddome (BM); Belgaum, Chulu Hill, Ritchie 1043 (long-leaved variant). Andra Pradesh: R. Godavari, Kortur, Barber Madras: Courtallum, Herb Wight 222 (K, lectotype). Without definite locality or not traced: NW. India, Royle; Sikh States,? Jayadhr, Edgeworth 154; Parwur Ghat, Ritchie 1043 (long-leaved variant); Brumagiri Hills, Beddome; Ind. or, Rupel (BM). E. PAKISTAN. Dacca, Clarke 8132;? Dhaka, Clarke I6734. CEYLON. A plant grown from Ceylon seed, flowered Dec. 1847, Herb. Hooker. Without locality, Walker, Moon I819 (BM) & Thwaites 1471 (BM p.p.). BURMA. Wallich Cat. No. 55o8H. INDONESIA. Lombok, Rensch 167 (acuminate leaflets rather like T. flexilis). Java, Surokirto, Horsfield (BM, mixed with labialis). 5. Teramnus flexilis Benth. in Journ. Linn. Soc., Bot. 8: 265 (1865); Prain in Journ. Asiatic Soc. Bengal 66: 412 (1897). I am at a loss to understand Prain's remarks on this species. He claims that T. oxyphyllus (Benth.) Kurz (based on Galactia oxyphylla Benth.) is undoubtedly a Teramnus and differs only from T.flexilis in its shorter racemes. I can only assume that he examined material quite different from Bentham's type since even at a glance it is obvious that the two names refer to very different species. T. flexilis has flowers barely 5 mm. long with short deltoid pubescent calyx-lobes mm. long whereas the flowers of Galactia oxyphylla are just over I cm. long with longer lanceolate bristly calyx-lobes 4-7 mm. long; in the former the leaves are much less hairy. No pods are available of G. oxyphylla at Kew, where it is still only represented by two old sheets seen by Bentham, though doubtless some may be lurking elsewhere, but a dissection of one of the few flowers showed that all ten anthers are fertile, the style is equal in length to the ovary and the stigma is large and peltate; these characters rule it out of Teramnus. Despite the lack of fruits I feel its affinity is with 'Dolichos' lagopus Dunn, which has the above characters and also very similar pollen grains; this species has nothing to do with Dolichos L. and a further discussion of this matter will be given in my account of the generic delimitation of that genus in a later paper. T. flexilis is closely allied to T. labialis, differing only in its larger mostly more acuminate leaves and longer, broader pods. INDIA. Manipur, Jhirighat, Meebold Assam, Jenkins (K, syntype of T. flexilis). E. PAKISTAN. Dacca, Clarke Chittagong, Clarke (BM, K); Khan 594 & Hooker f & Thomson (BM, K). Sylhet, Wallich Cat. No (K, syntype of T. fiexilis, BM, K-W, isosyntypes), & Hookerf. & Thomson. BURMA. Martabania, Wallich Cat. No. 5508b pro parte. THAILAND. Chiengmai, Nfi Tang, Kerr 4473.

44 278 KEW BULLETIN VOL Teramnus uncinatus (L.) Sw., Prodr. Veg. Ind. Occ.: I05 (1788). Dolichos uncinatus L., Sp. PI., ed. 2: o1019 (1763)- This is a widespread species in tropical America and Africa, easily distinguished in the New World from the other two species occurring there by the ferruginous furry indumentum. Hauman in Fl. Congo Belge comments on the extreme closeness of' T. andongensis' and T. uncinatus and I have found that, if members of the 'T. andongensis'- T. axilliflorus complex are laid out mixed with American material it is almost impossible in the case of some intermediate material to tell from which country they originated. There is, nevertheless, a general difference in facies due chiefly to the shape of the leaflets and colour of the indumentum. I have therefore kept up three subspecies. There is, however, an unfortunate complication; although the name andongensis has long been in use for the common African taxon, an examination of the type shows that it has leaflets closely similar to typical T. uncinatus and is not representative of the African plant, which has narrow mostly oblong leaflets. I have therefore reluctantly made use of a synonym for this subspecies. T. andongensis must be considered an atypical form intermediate with the South American subspecies. 6a. subsp. uncinatus Glycine angulata Desv. in Journ. de Bot. 3: 78 (1814). Teramnus botrydium Schott in Wien Zeitschrift 4: I216 (1830). Glycine botrydium (Schott) Walp., Rep. 1: 760 (1842). G. discolor Mart. & Gal. in Bull. Acad. Brux. Io(2): 190 (1843). G. elliptica Mart. & Gal., l.c. (1843). G. uncinata (L.) Macbride in Publ. Field Mus. Nat. Hist., Bot. 13(3): 350 (1943)- In this subspecies the leaflets are mostly elliptic with rounded sides and the indumentum ferruginous. The inflorescence is, in certain plants, reduced to a subsessile head, as is also the case in the African subspecies; the pods are mostly rather longer than in African material but the indumentum may be spreading or adpressed. Numerous sheets have been seen from Mexico, El Salvador, Costa Rica, Panama, Cuba, Jamaica, Haiti, S. Domingo, Puerto Rico, Colombia, Venezuela, Peru, Brazil, Bolivia and Paraguay. True subsp. uncinatus has also been cultivated in Rhodesia (Marandellas, Corby 993) and in Kenya (Kitale, Bogdan 3862 (EA, K)). The holotype is specimen in the Linnaean Herbarium; this was sent from Jamaica to Linnaeus by Patrick Browne. 6b. subsp. axilliflorus (Kotschy) Verdc., comb. et stat. nov. Glycine axilliflora Kotschy in Sitzb. Akad. Wiss. Wien 51: 366 (1865). G. gilletii De Wild. in Ann. Mus. Congo Belge, Bot. sir. 5, 5(I): 150 (1904). G. reducta De Wild. in Rev. Zool. Afric. 12, suppl. Bot.: 16 (1924). Teramnus axilliflorus (Kotschy) Bak. f., Leg. Trop. Afr.: 364 (1929). T. gilletii (De Wild.) Bak. f., op. cit.: 365 (I929). This is extremely close to subsp. uncinatus but differs in the narrower oblong leaflets. From subsp. ringoetii it differs in the presence of a mixture of white and pale ferruginous rather than dark blackish-brown hairs on the

45 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE : II 279 calyx. The inflorescence is more usually a subsessile cluster than in either of the other subspecies and the leaves are more silvery. There are, however, numerous intermediates showing alternative combinations of characters. Some specimens show a strong resemblance to T. labialis var. abyssinicus and are scarcely distinguishable. SENEGAL. Niokolo-Koba, Berhaut SIERRA LEONE. Musaia, Deighton GHANA. Wenchi area, Banda village, Morton GC NIGERIA. Abinsi, Dalziel 589. Lake Chad area, Lassa, Royer IoI (BM). R. Benue, Talbot (BM). (The last two have the long inflorescences of subsp. ringoetii.) CONGO (KINSHASA). Kisantu, Gillet (BR, holotype of Glycine gilletii). Rutshuru, Bequaert 6232* (BR, holotype of G. reducta). SUDAN. Equatoria, Shillick District, Ajab, Binder I (W, holotype of Glycine axilliflora). UGANDA. Bunyoro, Grant. Teso District: Serere, Chandler 23 & 791. Without locality, Liebenberg 962. TANZANIA. Buha District: Kasulu, Heru Chini, Rounce 29 (EA, K). Lushoto District: Korogwe, Archbold 949. Tanga District: E. Usambaras, Tengeni, Peter Ufipa District: W. Rukwa, Kinambo, Kipompo Swamp, Michelmore 1595; N. Rukwa, Robinson I666 (this is a form very similar to T. labialis but depauperate). ZAMBIA. Fort Jameson area, Verboom 499- ANGOLA. Golungo Alto, R. Cuango, near Ndele, Welwitsch 2188 (BM, K). 6c. subsp. ringoetii (De Wild.) Verdc., comb. et stat. nov. Glycine andongensis Bak. in Fl. Trop. Afr. 2: 179 (1871). G. ringoetii De Wild. in Fedde, Rep. Sp. Nov. 12: 295 (1913)- G. lanceolifoliata De Wild. in Rev. Zool. Afric. 12, suppl. Bot.: 12 (1924). Teramnus andongensis (Bak.) Bak. f. in Journ. Bot. 66, Suppl. I: 115 (1928). T. lanceolifoliatus ('lanceolifoliolatus') (De Wild.) Bak. f., Leg. Trop. Afr.: 364 (1929). This is usually clearly distinguishable by its narrowly oblong leaflets and very dark indumentum on the calyx, but intermediates are too frequent to maintain it as a distinct species and unfortunately the type of andongensis has rounded elliptic leaflets. In some specimens it seemed that there were distinct floral differences between this and subsp. uncinatus. In the latter the calyx-lobes are often narrower and longer, the vexillum more abruptly narrowed into the claw and the wings less spathulate distally; the pods are also usually spreading hairy. A survey of all the available material shows that none of these is a constant character. That Harms agreed with uniting T. uncinatus and T. andongensis is shown by the fact that he cites Lindblom s.n. from Mt. Elgon, Kitosh as T. uncinatus and gives the distribution as tropical America and Africa (Notizbl. Bot. Gart. Berl. Io: 85 (1927)). Similarly Rossberg (Fedde, Rep. Sp. Nov. 39: 164 (1936)), records T. uncinatus from Angola; Malange, Bondo, Quela, Nolde 403 (Bt) which Torre (Consp. Fl. Angol. 3: 247 (1966)) includes as an imperfectly known species. Glycine vanderystii De Wild. which E. G. Baker (Leg. Trop. Afr.: 364 (1929))

46 280 KEW BULLETIN VOL. 24 suggests is a Teramnus close to T. lanceolifoliatus is actually Eriosema vanderystii (De Wild.) Hauman, previously better known as E. velutinum Bak. f. & Haydon. SIERRA LEONE. Jigaya, N. W. Thomas Musaia, Deighton 5704 & Near Kuruboula on path to Seredu and Loma Mts., Morton SL2519- GHANA. Pawpaw Mt., Nkwanta Krachi, Morton A3722 (intermediate between subsp. axilliflorus). NIGERIA. Bauchi Plateau, Lely 721 (elliptic-leaved variant). Jos Plateau, Batten-Poole 140. Zaria, Kujama, Afaka Forest Reserve, Olorunfemi in F.H.I CAMEROUN. Adamawa Plateau, Breteler km. WNW. of Tibati, Letouzey km. NNW. of Mokolo, near Loubam, Letouzey Kounden, Saxer km. ENE. of Bagodo, Letouzey km. NNE. of Bafia, Nyandingi, Letouzey km. WNW. of Linte, Letouzey 8ooo. CENTRAL AFRICAN REPUBLIC. 69 km. W. of Yalinga, Bamb6l6, Le Testu 3171 (BM). CONGO (KINSHASA). Lubumbashi [Elisabethville], Lubumbashi Valley, Robyns 16I0. Upper Katanga: Shinsenda, Ringoet in Hombld 466 (BR, holotype of G. ringoetii). S. of Thysville, M'vuazi, Devred I 148. About 18 km. E. of Dungu, Keay K40 (BM, K). Orientale, Aru, Froment 129. Irumu, Bequaert 2865 (BR, holotype of G. lanceolifoliata). RWANDA. Parc National Kagera, Mt. Kiburara, Lebrun UGANDA. West Nile District: Aka, behind Paida Dispensary, Chancellor 198 (EA, K). Acholi District: Rom, Liebenburg 235. Toro District: Kichuamba [Kichwamba], Kitakwenda, Bagshawe 1219 (BM). Ankole District: Ibanda, Thornton 33 (EA); Rwashamaire [Lwasamaire], Snowden I659B; Rwampara, Bugamba, Harker 258 (EA, K). Kigezi District: Nyakagyeme, Purseglove 2083 (EA, K). Teso District: Serere, Maitland I194 & Chandler 697 (EA, K). Mbale District: Bugisu [Bugishu], Sipi, A. S. Thomas 407. Masaka District: Buddu, Brown 130. Mengo District: Banda, Dummer 430; Bulemezi [Bulimuzi], Luwero [Luwelu], Maitland 897; N. Mengo, Wakyato to Ngoma, Langdale Brown 2274 (EA, K). KENYA. Trans-Nzoia District: Kitale, Bogdan 3564 (EA, K); Kitale, Mabaonde, Tweedie Elgeyo District: Marakwet, Brodhurst Hill 271. Masai District: Chyulu Hills N., Bally 366, 385 in C.M (EA, K) (variant with small leaves but calyx characteristic). TANZANIA. Bukoba District: Kabirizi, Haarer 2284 (EA, K); Bukoba, Lind 2376; Masaka-Bukoba road, 6-4 km. within the Tanganyika border, Norman 71; Bunazi, Gillman 490 (EA). Ngara District: Muganza, Rusengo, Tanner Buha District: Mbirira [Birira] to 'Manyoni See', Peter (B); Mbirira [Birira] to Kisuzi [Nisusi], Peter (B). Kigoma District: Ujiji, R. Mkuti to Msosi, Peter p.p.; Uvinza [Uvinsa], Peter (B). Ufipa District: Kasamvu, Bullock 2700 (BM, K); Sumbawanga, Chapota, Richards 8504; Memya Mt., Bullock 3681; Lake Kwela, Richards Mbeya, Davies D86, 429 (EA, K). Rungwe District: Rungwe, Davies D45 (EA, K); Kyimbila, Stolz 845 (BM, K); Mwasukulu, StolZ 669 (BM, K). Songea District: Matengo Hills, Miyau, Milne-Redhead & Taylor 8798, 8798A (EA, K). MOZAMBIQUE. Zambezia, between Muobede and Tacuana, 3 km. from Muobede, Barbosa & Carvalho 2843.

47 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 28I MALAWI. North: Mzimba, G. Jackson I290 (BM, EA, K). South: Shire Highlands, Buchanan 432; Zomba, Salubeni 284 (K, SRGH). Unlocalized, Buchanan 333 (US). ZAMBIA. North: Mbala [Abercorn], McCallum- Webster 853; Kalambo Farm, Sabi Valley, Richards 5099; Lumi marsh, Richards West: Kitwe, Fanshawe 2309, & Mutimushi 1322; Luano, Fanshawe South: Mumbwa, Macaulay 784; Mazabuka, van Rensburg KBSI585. RHODESIA. East: Nyumquarara Valley, Gilliland 1584 (BM); Chirinda, Swynnerton 442, 1437 (BM); Umtali, Inyamatshira Mt., Chase 5977 (BM, K); Melsetter, Black Mountain Inn, Corby I 105; Haroni R., near Dragon's Tooth, Hall 483 (BM, SRGH). ANGOLA. Pungo Andongo, near Luxilo, Welwitsch 2189 (LISU, holotype of G. andongensis; BM, isotype). Cazengo, Dalatondo, Gossweiler 5528 (BM, COI). Robinson 3474 (Zambia: Solwezi) is a form or perhaps even a hybrid of subsp. ringoetii with very small leaflets about 2 cm. long and 4-7 mm. wide but with the dark blackish indumentum on the calyx. 7. Teramnus micans (Bak.) Bak. f. in Journ. Bot. 66, Suppl. I: 115 (1928); Hauman in Fl. Congo Belge 6: Io3 (1954). Glycine micans Bak. in Fl. Trop. Afr. 2: 179 (1871). This is a robust climber with the stems conspicuously angled, or almost winged in the typical variety. The leaflets are elliptic-rhomboid or broadly elliptic, usually thickly hairy beneath. The lower part of the calyx-tube is frequently much less hairy than the calyx-lobes, or even glabrescent. Even here, however, there is a gradation to the large-leaved forms of T. labialis on the one hand and broader leaved forms of T. uncinatus on the other. Hauman has divided the species into three varieties and these seem quite distinctive in their typical states. 7a. var. micans Stems robust and angular. Leaflets distinctly rhomboid. Corolla usually orange but sometimes blue. PORTUGUESE GUINEA. Bafat4, Santo SIERRA LEONE. Yonibana, N. W. Thomas 4698 & 5Io01. Musaia, Deighton Sefadu, Deighton LIBERIA. Nimba, J. G. Adam (K, P, UPS) (atypical, leaves glabrous above). NIGERIA. Jos Plateau, Naraguta, Hepper CAMEROUN. Bamenda, Belo, Maitland Bertoua, Breteler 620. Bitya, Ebolowa, Bates 6oi, 602 (BM). CONGO (KINSHASA). Kasai, Kanda Kanda, Gandajika, Liben Katanga, Kapona to Kalima [Albertville], Devred Mt. Senga, Kassner 2918 BM).? Lufonzo, Kassner ZAMBIA. North: Mbala [Abercorn], Ndundu, Richards (EA, K). West: Solwezi District, R. Mwafue, Milne-Redhead 727; Lualaba, Holmes 1479; Kitwe, Fanshawe ANGOLA. Cuanza Norte: Cazengo, Camondai, Gossweiler 5177 (BM, COI),

48 282 KEW BULLETIN VOL (BM, COI) & IoI88 (BM, COI). Golungo Alto: Sobado de Mussengue, Welwitsch 2186 (LISU, holotype; BM, COI, K, isotypes). 7b. var. cyaneus (De Wild.) Hauman in Fl. Congo Belge 6: 104 (1954). Glycine cyanea De Wild. in Rev. Zool. Afric. 12, suppl. Bot.: 8 (1924). Teramnustolzii Bak. f., Leg. Trop. Afr.: 365 (1929). Stems distinctly angled but the leaflets are more regularly oblong-elliptic; the calyx is usually more densely hairy all over. The corolla is often blue but this is not a constant character. CONGO (KINSHASA). Nioka, Liben 406, Bamps 105 & De Craene 218 (BM, BR). Rutshuru, Bequaert 6126 (BR, syntype). Mboga, Bequaert 3030 (BR, syntype). UGANDA. Toro District: Musandama, Maitland IoIo; Ibanda, Ross 404 (BM); Ruwenzori, Nyamwamba [Namwamba Valley], Kilembe, G. Taylor 2543 (BM). Ankole District: Bunyaruguru, Purseglove 480; Rwashamaire [Lwasamaire], Snowden 1658 (BM, K). Mengo District: Entebbe, Maitland 26; Victoria Nyanza Region, Maitland I51; Kirerema, Dummer 349 (BM); Kyagwe, Banda, A. S. Thomas 787. Mubende District: 16o km. NW. of Kampala on Mubende road, E. Brown in Dummer Dubious, possibly in Kenya, 2nd and 3rd days out from Mumia's, Whyte. KENYA. Kavirondo, Scott Elliot 70o61. TANZANIA. Ulanga District: Ukwama, Kwiro, Haerdi 548/o (EA, K); Mahenge, Liondo, Schlieben 2056 (BM, K). Rungwe District: Kyimbila, Stolz 827 (BM, holotype; K, isotype of T. stolzii) (leaves rather less obtuse than usual. Songea District: 14'5 km. N. of Mahanje by R. Mwandi, Milne-Redhead & Taylor 10o940 (EA, K); Matengo Highlands, Umgano [Ugano], 8 May 1936, Zerny 648 (W). MALAWI. Chipata Mt., Verboom 881. Zomba District: R. Mulunguzi, Banda 19 (BM). ZAMBIA. Luangwa Valley, Upper Mtonga R., Verboom LK I117 (atypical leaves). Fort Jameson, Kalichero, Verboom 452-7c. var. fagifolius Hauman in Bull. Jard. Bot. Brux. 25: 95 Stems much more rounded than in the other (1955)" varieties, scarcely if at all winged, but robust. Leaflets elliptic-rhomboid. CAMEROUN. W. of Tibati, Mayo Tolore, Letouzey CONGO (KINSHASA). Masisi to Walikale, Lebrun 5197 (BR, holotype of variety). Ubangi, between Libenge and Gemena, Lebrun Yambuya, Bequaert Between Yangambi and Yakusu, Germain Uele, Lebo, Gerard 716 (BM, BR). SUDAN. Equatoria, Yei, Kagelo Station, Myers UGANDA. Mengo District: Kyagwe, Banda, A. S. Thomas 786 (EA, K) & Pool I 184;? Kirerema, Dummer 322 (BM, K). 8. Teramnus buettneri (Harms) Bak. f., Leg. Trop. Afr.: 365 (1929). Glycine buettneri Harms in Engl., Bot. Jahrb. 26: 302 (1899). This savanna species is characterized by its erect shrubby habit, the leaflets

49 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 283 densely felted or velvety beneath and the pod densely covered with somewhat spreading ferruginous hairs. Nevertheless, there are no real structural differences between this and other species and it is pretty clearly derived from the same stock as T. uncinatusubsp. ringoetii and T. micans, the stems often being distinctly angular, as in the latter. IVORY COAST. Cercle de Mankono, between Buonolougou and Marabadiassa, Chevalier GHANA. Ashanti, Wanki, Chipp 482. NW. Ashanti, Sampaa, Morton A3253; Enti in F.H Banda, Hale Kumasi, Vigne (EA, K). Atebubu-Kete Krachi road, Enti in F.H Krachi, Nkwanta Hills, path from Shiare to Chilinga, Morton A398. Kete Krachi-Kpedsu road, Morton in GC 716o. TOGo. Kpandu, Robertson 135 (BM). Bismarckburg, Biittner 163 (B, holotypet). NIGERIA. Ilorin, Ilesha, 9-6 km. from Ilesha on Okuta road, Okafor & Latilo in FHI CENTRAL AFRICAN REPUBLIC. Buar, Mildbraed 9440; 'Kamerun' 50 km. SE. of Buar. above Lobaje, Mildbraed Yalinga region, Le Testu 3037 (BM, K). DOUBTFUL SPECIES OR SPECIES EXCLUDED FROM TERAMNUS P.BR. Other names published in Teramnus, not previously mentioned, with suggested or certain identities are as follows:-- T. CLANDESTINUS (Wendl.) Spreng.= Glycine clandestina Wendl.* T. GRANDIFLORUS Griseb. Herpyza grandiflora (Griseb.) C. Wright T. HEDYSAROIDES - (Willd.) Spreng. = Ophrestia hedysaroides (Willd.) Verdc. T. OBCORDATUS Baill.t = Galactia obcordata (Baill.) Verdc., comb. nov. T. TENUIFLORUS (Willd.) Spreng. = Galactia tenuiflora (Willd.) Wight & Am. T. RHOMBIFOLIUS Beurl. presumed to be a synonym oft. uncinatus (L.) Sw. but not seen T. WALLICHII Kurz = Neocollettia wallichii (Kurz) Schindler UNIDENTIFIED SPECIMENS OF TERAMNUS P.BR. Beddome 2165 (BM) from India, Andhra Pradesh, Kurnool, Nallamalais Hills, is a distinctive plant and although not in flower the reflexed tips of the pods leave little doubt that it is a true Teramnus. It certainly cannot be a variant of T. labialis. The leaflets are densely adpressed silky pilose beneath and the short, broad, 3-4-seeded pods are cm. long, 4'5 mm. wide, 2.5-2'7 densely covered with mixed adpressed fulvous and silver hairs. The seeds are very finely reticulate-foveolate quite unlike those of any other Teramnus. Further material is needed. Hassler 7563 (BM) from north Paraguay, an obviously perennial native plant resembling T. volubilis, has the upper calyx-lobes joined for only two- * Burtt's proposal (Taxon 15: 307 (1966)) has been adopted; my original indication of Willd. as the correct author (op. cit: 35 (1966)) is now incorrect. I In Bull. Soc. Linn. Paris I: 382 (1883). I have examined the type collected by Boivin and agree with this collector's suggestion that the plant is a Galactia. Unfortunately only small buds are present but Io anthers can be seen. Further material is much desired. D

50 284 KEW BULLETIN VOL. 24 thirds of their length. The indumentum rules out T. uncinatus and technically it appears to be T. labialis but this is not likely. The pod has fine adpressed hairs, as in that species, but the standard is 7 mm. long. In a genus of such featureless plants only an analysis of large populations can evaluate specific characters. The present problem is beyond the terms of reference of this revision. ERYTHRINA L. Erythrina abyssinica DC., Prodr. 2: 413 (1825); Gillett in Kew Bull. 15: 426 (1962); Torre in Consp. Fl. Angol. 3: 248 (1966). subsp. abyssinica Chirocalyx abyssinica (DC.) Hochst. in Flora 29: 600 (1846). C. tomentosa Hochst., l.c. (1846). Erythrina tomentosa A. Rich., Tent. Fl. Abyss. I: 213 (1847); Bak. in Fl. Trop. Afr. 2: 184 (1871), pro parte; Bak. f., Leg. Trop. Afr.: 373 (1929); Majot-Rochet & Duvign. in Fl. Congo Belge 6: 120 (1954), pro parte. E. bequaertii De Wild. in Rev. Zool. Afric. 8, suppl. Bot.: BI5 (1920); Bak. f., Leg. Trop. Afr.: 376 (1929). E. kassneri Bak. f., Leg. Trop. Afr.: 375 (1929). E. platyphylla Bak. f., 1.c.: 376 (1929). E. tomentosa A. Rich. var. longicauda Bak. f., 1.c.: 374 (1929). E. webberi Bak. f., 1.c.: 373 (1929), pro parte. E. eggelingii Bak. f. in Journ. Bot. 76: 238 (1938). subsp. suberifera (Bak.f.) Verdc., comb. et stat. nov. E. suberifera Bak., Fl. Trop. Afr. 2: 183 (1871); Bak. f.,-leg. Trop. Afr.: 372 (1929). E. huillensis Bak., 1.c. (1871); Bak. f., 1.c. (1929). [E. tomentosa sensu Majot-Rochet & Duvign., l.c. (1954), pro parte, non A. Rich.] Majot-Rochet & Duvigneaud (1.c.) have considered E. suberifera to be completely synonymous with E. abyssinica (which they call E. tomentosa) but I would agree with J. B. Gillett (annotations on Kew folders) that there appears to be a good case to recognize two subspecies. In the north-east of the combined range the calyx-lobes are predominantly linear or filiform and in the south-west they are predominantly short, often elliptic, obovate or spatulate. As the Belgian authors have stated, mixed intermediate populations occur in the intervening areas, e.g. in Rhodesia and Congo (Kinshasa), and intermediates are so frequent as to render the use of subspecific names in that area impossible. Nevertheless, the considerable difference between the two extremes makes subspecific names useful in certain areas. An additional problem is how to treat E. sudanica Bak. f. and E. sigmoidea Hua (-E. dybowskii Hua and E. eriotricha Harms) both of which have very short calyx-lobes resembling those of subsp. suberifera but even shorter; since there are indumentum differences and apparently no merging with E. abyssinica, they are left distinct for the present. Clearly all have been derived from the same immediate ancestor.

51 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 285 Erythrina haerdii Verdc., sp. nov. E. sacleuxii Hua valde affinis, sed bracteis primariis et secundariis conspicuis o06-1-i cm. longis, foliis adultis majoribus, vexillo minore, indumento rhachidis floriferae minus tenui differt; a E. abyssinica DC. inflorescentiis longioribus, foliis adultis subtus glabris distinguenda. Arbor usque 8 m. alta floribus praecocibus; truncus haud descriptus verosimiliter crassespinosus; ramuli ferrugineo-velutini, demum glabrescentes, porcati, spinis rectis armati. Folia trifoliolata; stipulae deciduae, crassae, oblongae, I cm. longae, 2 mm. latae, leviter falcatae, hirsutae; petiolus 8-26 cm. longus, aculeatus; rhachis cm. longa; petioluli 5-10 mm. longi; foliola anguste vel late ovato-rhomboidea, cm. longa, cm. lata, apice rotundata vel subtruncata, basi truncata, rotundata vel late cuneata, primum hirsuta, mox glabra, nervis lateralibus Io- 5-jugis interdum aculeis minutis instructis venulis subtus reticulatis prominentibus. Inflorescentiae densae, circa 15 cm. longae, pilis pallide ferrugineis velutine obtectae; pedunculi cm. longi, primum velutini, demum glabrescentes; pedicelli 2-3 mm. longi; bracteae primariae et secundariae linearispathulatae, o.6-1- I cm. longae, 1-2 mm. latae, deciduae vel subpersistentes; bracteolae filiformes, o cm. longae, 0-5 mm. latae, deciduae. Calyx fusiformis, primum dense hirsutus, mox glabrescens; tubus I-3 cm. longus, prope ad basin unilateraliter fissus spatham distinctam formans; limbus in 5 lobis linearibus o-8-i-3 cm. longis o.5-i.5 mm. latis divisus. Vexillum ellipticum, cm. longum, cm. latum, breviter unguiculatum, 3"I coccineum, glabrum vel extra pilis perpaucis conspersum. Alae falcatae, 9-1o mm. longae, 4-5 mm. latae. Petala carinae subrotundatae, 5-6 mm. longae, 4-5 mm. latae. Filamenta I 9 cm. longa, stamen vexillare liberum antheris linearibus mm. longis. Ovarium anguste cylindricum, I'4 cm. I'8-2.5 longum, I3-14 ovulatum, pilis pallide ferrugineis dense obtectum; stylus filiformis cm. longus. Legumina I-I.I 8 cm. longa, moniliformia, articulis subglobosis valde compressis I cm. longis I 5-2 cm. latis, glabrescentia. Semina miniata, ellipsoidea, 8-5 mm. longa, 6-5 mm. lata et crassa, hilo nigro ornata. TANZANIA. Ulanga District: Kiberege, 300 m., Oct. 1935, Culwick I:- small tree with thorns and broad triangular leaves, flowers scarlet coming at the end of the dry season before the leaves, used as a galactogogue for foster mothers, vernacular name (Kibena) 'murongoloma'; Ulanga Valley, in the hills bordering the valley, 330 m., June 1935, Culwick I2:-tree with small thorns, and broad leaves with tiny prickles on the veins; Ifakara, 15 Aug. 1959, Haerdi 305/0 (K, holotype; EA, isotype):-tree to 8 m.; same locality, Dec. 1959, Haerdi 3o5/oB (EA, K). This species is closely allied to E. sacleuxii but has smaller flowers, larger leaflets, which are usually much blunter, more conspicuous bracts and a coarser indumentum. It is, however, equally allied to E. abyssinica and flowering material can be difficult to distinguish. If it were not for the extreme care of Mr. Haerdi I would not have ventured to describe the plant since the leaves have a prominent venation beneath very similar to those of E. schliebenii Harms. There is a possibility that E. sacleuxii and E. abyssinica hybridize. Careful collections of Erythrina from marked trees in south-eastern Tanzania are very much needed. In no other East African genus is more care needed to make quite certain that leaves produced and

52 286 KEW BULLETIN VOL. 24 collected later in the season do actually correlate with flowers which have been collected earlier. ERYTHRINA WEBBERI Bak. f., Leg. Trop. Afr.: 373 (1929). This species based on Webber 608 collected in 'British East Africa' (and, judging by the number, in the 'coastal forests') has never been satisfactorily understood. It is true that Dale & Greenway in 'Kenya Trees and Shrubs' (p. 365) cite further material, but an examination of this does not clear the matter up; they state that further flowering material will probably demonstrate that the name is a synonym of E. sacleuxii Hua. Immediately on examining the type of E. webberi preserved at Kew, I was struck with the strong resemblance shown by the inflorescences to those of E. abyssinica, both having the same rather woolly indumentum, quite different from the tomentose inflorescences of E. sacleuxii. Nevertheless, the glabrous leaflets preserved on the type-sheet are undoubtedly those of E. sacleuxii and there is no doubt in my mind that the sheet is a mixture of the two species. There is further strong evidence that this is possible. Graham who knows the coastal forests well collected both species giving them successive numbers 278 and 279; moreover he has labelled the packets of flowers in the field showing that it was possible he thought there might be confusion if he did not. I suspect that the Webber specimen was collected like the Graham specimens in the Shimba Hills. MUCUNA ADANS. Seventy years ago Prain (Journ. Asiatic Soc. Bengal 66: 404 (1897)) stated that a monographer would almost certainly find it necessary to divide Mucuna into two genera, Mucuna and Stizolobium P. Br. Since then many other authors have said much the same and several American workers (including Burkart) have kept up Stizolobium. There is much to be said in favour of maintaining two genera. In Stizolobium the seeds are compressed oblong-ovoid with a very short hilum surrounded by a conspicuous rim aril whereas in typical Mucuna the seeds are discoid, large and flat with a hilum extending a of the circumference and without an aril. There are associated characters but these have not been studied for more than a few species. The first leaves to appear above the cotyledons are opposite, simple and cordate in Stizolobium whereas in Mucuna proper the initial leaves are said to be all alternate and scale-like. The dorsifixed anthers in Mucuna gigantea (Willd.) DC. are barbate whereas in M. pruriens (L.) DC., M. glabrialata (Hauman) Verdc. and M. stans Bak. (all of which belong to Stizolobium) they are glabrous. An examination of the pollen of a few species shows that in subgenus Stizolobium the walls are thin with an easily visible open reticulation whereas in Mucuna the walls are thick and the reticulation is much closer and more difficult to see. M. poggei Taub. presumably belongs to the subgenus Stizolobium but the pods appear to be more or less indehiscent with discoid seeds having the hilum extending almost one-third of the way around the seed; there is a faint rim aril and it does to a certain extent bridge the gap between the two groups. Bearing in mind this latter species, the fact that the genus needs a general monograph and that, despite the marked differences between the two groupings, there are also great similarities (e.g. general flower structure, stiffened keel apex, inflorescence structure, presence of

53 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 287 irritant hairs, etc.), I have decided to follow the general tradition of maintaining a single genus divisible into two well-marked subgenera. Another fact, perhaps of some significance, is that other undoubtedly distinct genera have seeds very similar in structure to those of subgenus Mucuna, Dioclea being an example. The following notes are needed to validate names used in the flora treatment. Mucuna gigantea (Willd.) DC., Prodr. 2: 405 (1825). subsp. quadrialata (Bak.) Verdc., comb. et stat. nov. M. quadrialata Bak. in Fl. Trop. Afr. 2: 186 (1871); Bak. f., Leg. Trop. Afr.: 379 (1929); Brenan, Check-List of Tanganyika Trees and Shrubs: 436 (I949). M. longipedicellata Hauman in Bull. Jard. Bot. Brux. 25: 99 (1955); & in Fl. Congo Belge 6: 133 (1954), sine descr. lat. I agree with Dyer (Bull. Misc. Inf. Kew 1931: 464 (1931)) that M. quadrialata Bak. and M. gigantea (Willd.) DC. are conspecific but a survey of the available Asiatic material shows that it has mostly smaller flowers than the African material. Certain Oriental specimens with large flowers do not appear to be correctly named and unfortunately are not correlated with fruits; practically no material has been seen from India and nothing from Malabar, the type-locality. Until a better survey can be made I have preferred to treat the uniformly large-flowered African populations as a distinct subspecies. I have examined the holotype of Hauman's supposed species (Ghesquiere 3771) and an isotype and can find no differences between it and M. gigantea subsp. quadrialata. THE MUCUNA POGGEI TAUB. COMPLEX M. poggei Taub. varies greatly in the size of its flowers and I have felt unable to maintain M. pesa De Wild. as a distinct species. It seems convenient to recognize three varieties covering the area of East Africa which do not appear to be correlated with geography or habitat. I have examined the types of the entities involved. Thie three varieties occurring in East Africa can be separated as follows: Corolla 6-9 cm. long; leaflets densely silvery-grey hairy beneath var. poggei Corolla 4-5 cm. long: Leaflets densely grey hairy beneath.... var. pesa Leaflets glabrescent beneath... var. glabrescens Mucuna poggei Taub. in Engl., Bot. Jahrb. 23: 194 (1896). var. pesa (De Wild.) Verdc., comb. et stat. nov. A. pesa De Wild. in Fedde, Rep. Sp. Nov. 13: 115 (1914); Bak. f., Leg., Trop. Afr.: 381 (1929); Hauman in Fl. Congo Belge 6: 133 (I954). CONGo (KINSHASA). Katanga, Kapiri Valley, Hombld 1162 (BR, holotype of M. pesa). Lake Tanganyika shore, Kassner 3032 (BM).

54 288 KEW BULLETIN VOL. 24 UGANDA. Toro District: Ruimi [Wimi] Forest, Bagshawe 1037 (BM) (no leaves). TANZANIA. Handeni District: Tamota, Bally Med. IoI in C.M (EA). Mpwapwa, Mr. & Mrs. Hornby 702 (EA, K). Ulanga District: Ulanga Valley, Culwick 17. Iringa District: 80 km. NE. of Iringa, Image, Polhill & Paulo I686A (EA, K). Rungwe District: Tukuyu-Chimala road, Kiwira R. bridge, Greenway & Hoyle 8332 (EA, K); Usafua, Goetze 1040 (BM). Lindi District: Rondo [Muira] Plateau, Busse 2637 (EA). Niassa Province, Amaramba Division, Mandimba to Mas- MOZAMBIQUE. sangulo, 12 km. from former, Pedro & Pedrdgdo 3416 (EA). 'Zambeziland', Kirk. Macequege District: Vila de Manica, R. Zambusi, Chase 2263 (BM, K, SRGH). MALAWI. Nyambi, G. Jackson I647. Without locality, Buchanan Io66 (BM, K). RHODESIA. Inyanga District: Pungwe R. near Mozambique border, Chase 6450 (K, SRGH); Honde Valley, Phipps I 117 (K, SRGH); Nyumquarara Valley, Wild 5717 (K, SRGH) & Corby 1326 (K, SRGH). var. glabrescens (Hauman) Verdc., comb. nov. M. pesa De Wild. var. glabrescens Hauman in Bull. Jard. Bot. Brux. 25: 99 (1955); & in Fl. Congo Belge 6: 133 (1954), sine descr. lat. CONGO (KINSHASA). Parc National de l'upemba, Mt. Kia to Lake Upemba, van Meel in de Witte 5704 (BR, holotype of M. pesa var. glabrescens). TANZANIA. Buha District: Kalinzi, Verdcourt 3414A. Ufipa District: Namwele-Kisungu, Bullock Ulanga District: near Kwiro, Haerdi 534/0 (EA, K); 35 km. S. of Mahenge, Sali, Schlieben 2246 (BM, BR, K). Iringa District: Dabaga, Emson H16/36 (EA); Iringa, Emson 522 (EA). Rungwe District: between Tukuyu and Mbelezi, Igale, St. Clair- Thompson 1041 (EA, K). Also related to this complex is the Ethiopian plant, Mucuna melanocarpa A. Rich., which seems to differ in little else than its dark blackish-purple (not greenish-white) flowers. At present I prefer to treat M. poggei as a separate species from this; far more material is needed from Ethiopia. THE MUCUNA CORIACEA BAK. COMPLEX Mucuna stans Bak. is a common and well-known erect Mucuna of the savanna areas of East and Central Africa, Cameroun and Angola; it is clearly derived from climbers and scramblers of the M. coriacea complex. Some difficulty has been experienced in classifying the non-erect members of this group. Attempts to divide M. coriacea into several subspecies failed until it was decided that one of the taxa involved was best treated as a separate species; the results then obtained were more satisfactory. A key to the taxa involved follows: Corolla with wings glabrous or hairy only at base; leaves glabrescent or finely pubescent M. glabrialata Corolla larger; wings with conspicuous zone of hairs on upper part of lamina: Leaves woolly, especially beneath. M. coriacea. subsp. coriacea Leaves finely pubescent... M. coriacea subsp. irritans

55 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 289 Mucuna glabrialata (Hauman) Verdc., stat. nov. affinis M. coriaceae Bak. subsp. coriaceae, foliis pills minutis fulvis obtectis non dense hirsutis, alis apice glabris distincta; M. coriaceae Bak. subsp. irritantis (Burtt Davy) Verdc. valde similis, floribus minoribus, alis apice glabris differt. M. coriacea Bak. var. glabrialata Hauman in Bull. Jard. Bot. Brux. 25: 99 (1955); & in Fl. Congo Belge 6: 132 (1954), sine descr. lat. M. sp. I; F. White in For. Fl. N. Rhod.: 160 (1962). Herba volubilis robusta usque 3 m. alta, ramis leviter sulcatis in siccitate nigrescentibus glabris vel praecipue ad nodos pilis minutis fulvis appressis sparse obtectis. Folia 3-foliolata; stipulae triangulares vel lanceolatae, 4 mm. longae, mm. latae, extra hirsutae; stipellae filiformes, usque 4 mm. I.5 longae; petiolus 2-5-I8 cm. longus; rhachis 1-3 cm. longa; petioluli mm. longi; foliola rhomboideo-ovata, lateralia valde obliqua, cm. longa, 2-7'5 cm. lata, apice late rotundata vel levissime emarginata sed mucronulata, basi rotundata, chartacea, in siccitate nigrescentia, pilis minutis fulvis appressis obtecta; nervi laterales 5-6-jugis subtus prominentes. Inflorescentiae pseudo-racemosae, 5-15 cm. longae, multiflorae, floribus in fasciculis 3-floris dispositis; pedunculi 2-9 cm. longi; pedicelli 4-5 mm. longi; bracteae deciduae, lanceolatae, I-I-I cm. longae, 4 mm. latae, fulvopilosae; bracteolae bractearum similes, 7-9 mm. longae I.5 mm. latae. Calyx appresse fulvopilosus, tubo 5 mm. longo I cm. lato; lobi inferiores triangulares, 4'5-9 mm. longi, acuti, lobis 2 superioribus in labio integro apice obtuso connatis. Vexillum fere nigrum, rotundato-ovatum, cm. longum, cm. latum, apice rotundatum, basi in ungue brevi 7 - mm. lato productum, glabrum. Alae atro-coccineae vel fere nigrae, anguste oblongofalcatae, cm. longae, mm. latae, basi margine inferiore hirsutae. Carina atro-coccinea vel fere nigra, falciformis, cm. 2"5-3 longa, 6 mm. lata, apice rigida. Filamenta 3-3 cm. longa, antheris basifixis ellipticooblongis 2-2 mm. longis aliis dorsifixis 1-2 mm. latis omnibus glabris. Ovarium subcylindricum I cm. longum, 2 mm. latum, stylo 2 8 cm. longo ultra medium torto inferne breviter piloso; stigma minute capitatum. Legumen elongato-sigmoideum, 7-8 cm. longum, I.I-I-4 cm. latum, 4-5- seminatum, compressum, pilis rigidis pallide flavido-fulvis irritantibus circa 2 mm. longis dense obtectis. Semina ambitu elliptica vel rotundata, valde compressa, olivaceo-brunnea usque nigra, mm. longa, 7-8 mm. lata, 2-5-4'5 mm. crassa, hilo oblongo 3-5 mm. longo margine conspicue nigroarillata. (Fig. 2, p. 290.) CONGO (KINSHASA). Upper Katanga: Lubumbashi [Elisabethville], Mt. Mukuen, Schmitz 1502 (BR, holotype); Parc National Upemba, 21 Feb. 1948, de Witte 3430 (BR, K); Tshinsenda, Ringoet 502* (BR). TANZANIA. Buha District: Bagaga-Kasulu [Kassulo], in forest remnant, 1260-I350 m., 22 Nov. 1926, Peter (B):-climber with almost black flowers. Kigoma District: Uvinza [Uvinsa], W. of Malagarasi, km Io87-5 on Central Railway line, mixed bush, io6o m., I Feb. 1926, Peter 36071I (B, K):-large liane with blackish flowers. Chunya District: 152 km. from Mbeya on Itigi road, Lupa N. Forest Reserve, slope in Brachystegia- Julbernardia woodland, 1400 m., I March 1963, Boaler 857:--climber to 4 m., flower dark purple, occasional. Mbeya District: Mbosi to Mkama, 9 April 1932, Davies 627 (EA, K). Songea District: about km. S. of Gumbiro, I15

56 290 KEW BULLETIN VOL. 24 ' -7.--/?,--- -,8L::.- j ~~~z ~~ /...?:? :..:~'' ' "2 " FIG. 2. Mucuna glabrialata. I, part of stem with inflorescence to show habit, X 1; 2, part of stem with infructescence, x 1; 3, flower, x 2; 4, calyx, X 2; 5, standard, x 2; 6, wing petal, x 2; 7, keel, inside view, x 2; 8, androecium, x 2; 9-1i, stamens showing medifixed and basifixed anthers, x 4; I, gynoecium, x 2; 13, style and stigma, x 4. All drawn from Milne-Redhead & Taylor ioi 16.

57 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 291 in Brachystegia woodland on sand, 900goo m., 8 May 1956, Milne-Redhead & Taylor I o 16 (EA, K) :-strong twiner going up shrubs to 3 m., stem, petiole, rhachis and nerves yellow-green, leaves green with a fine? level reticulation above, pale green with fine level reticulation beneath, peduncles thick and pendulous, dull green, calyx dull, the colour of sunburnt skin, standard black, wings very deep maroon, almost black, keel ditto with very hard sharp beak, stamens included in keel the basal part of which is flattened dorsally. MALAWI. North: Mzimba, Mbawa Expt. Station, in Julbernardia scrub woodland, 6 April I955, Jackson I600oo:-climbing or prostrate weed, leaves with yellow veining, flowers blackish-purple. South: Liwonde district, in Uapaca-Brachystegia forest, March 1937, Clements 833 (FHO, K) :-root extract used as a black dye for cloth. ZAMBIA. North: Mbala [Abercorn] District: Mpulungu road, about 8 km. from Mpulungu, in bush by side of road, over low bushes, 900 m., 20 April 1952, Richards 1613; near Kambole Falls, woodland, 1500 m., 5 June 1957, Richards oo0026:-climber to 0o9 m., pea dark (practically over), seeds shiny yellow-straw colour with bands of green. West: Ndola District: Ndola, over shrubs and lower parts of trees in plateau woodland, 16 July 1954, Fanshawe 1380; same locality, 4 March 1954, Fanshawe I 155:-flowers dusky purple in dense pendent racemes, calyx with reddish hairs; Mufulira, 21 April 1966, Lawton 1396:-creeping legume with deep purple flowers; Kitwe, woodland, 6 June I955, Fanshawe 2319 (EA, K) :-vine twining clockwise up lower part of trees to 3 m., flowers dusky purple, young fruits flat, S-curved, appressed golden-brown woolly. South: Mumbwa District: Kafue National Park, Mumbwa-Mankoya road, I6 km. E. of Kafue Hook pontoon, in woodland (Trapnell's P-5) of Julbernardia paniculata, Uapaca kirkiana and Hyparrhenia spp., 12 June 1964, Mitchell 25/74. Locality uncertain: B.D.C.?, 28 May 1958, Verboom LK 29 (K, PRE). Mucuna coriacea Bak. in Fl. Trop. Afr. 2: 187 (1871); Bak. f., Leg. Trop. Afr.: 380 (1929); Brenan, Check-List of Tanganyika Trees and Shrubs: 433 (I949). subsp. coriacea M. rhynchosioides Taub. in Engl., Bot. Jahrb. 23: 194 (1896). The material seen is cited in condensed form: TANZANIA. Mbulu District: Pienaar's Heights, Burtt 2202 (BM, EA, K); Ufiomi, Bonga to Bereu [? Bereku], Peter (B). Kondoa District: 47 km. N. of Kondoa turn-off, Milne-Redhead & Taylor (EA, K). Ulanga District: Mahenge, Haerdi 567/0 (EA, K). Songea District: N. of Songea, by R. Luhira near Mshangano fish-ponds, Milne-Redhead & Taylor Io821 (EA, K); Matengo-Songea, Ugano, Zimmer 56 (BM). Lindi District: Rondo [Mwera] Plateau, Schlieben 6281I (BM) (verging to subsp. irritans). MOZAMBIQUE. Niassa, Massangulo, Pedro & Pedrdglio 3549 (EA); Vila Cabral, Torre I97, I98 (BM, K). Manica e Sofala: Chupanga, Kirk (K, syntype) & Stewart (K); Chupanga, near Mazzara, Kirk. Without detailed locality, 'Sambesi', Carvalho (B, holotype of M. rhynchosioidest).

58 292 KEW BULLETIN VOL. 24 MALAWI. 'N. Nyasaland', Whyte. South: Mt. Chiradzulu, Johnston (somewhat intermediate with subsp. irritans); Lower Valley of R. Shire, Meller; near Cholo, Pole Evans & Erens 541; Manganja Hills, Meller (K, syntype) [note this sheet is rather nearer to subsp. irritans so, if lectotype is selected, the Kirk specimen should be chosen]. Without locality, Lawrence 794 & Webb (BM). ZAMBIA. East: Lunkwakwa-Fort Jameson, Mutimushi 149I (only a sterile shoot seen). subsp. irritans (Burtt Davy) Verdc., comb. et stat. nov. M. irritans Burtt Davy, Flowering P1. Ferns Transvaal 2: xxvii, 414 (1932). [M. coriacea sensu Hauman in Fl. Congo Belge 6: 131 (1954), non Bak. sensu stricto] CONGO (KINSHASA). Uele, frontier with Sudan, Missa Dogo, Piedboeuf 49 (BR, K fragment). UGANDA. Bunyoro District: near Bugoma Forest, Bagshawe 1404 (BM). TANZANIA. Ufipa District: Mpui-Ilemba gap road, Richards (not typical but no more than varietally distinct). Morogoro District: Uluguru Mts., Mgeta-Bunduki, Harris 127 (EA, K). MOZAMBIQUE. Niassa, Massangulo, 13? 51' S., ' E., Gomes Sousa Zambezia, Lugela-Mocuba, Namagoa Estate, Faulkner 250 (BM, K). Manica & Sofala, 12-8 km. N. of R. Pungwe on road to Vila Gouveia, Evans & Erens 479 (K, PRE). 'Zambeziland', Kirk. MALAWI: South: Mt. Mlanje, Whyte (BM). RHODESIA. North: Darwin, Kandeya, Pachanga Store, Bingham 1438 (K, SRGH); Goromonzi, Ngomakwira, Chindamora Reserve, Wild 7569 (K, SRGH); Mazoe, Eyles 361 (BM); Shamwa,? Mainwar in Eyles 2241 (K, PRE). East: Odanzi [Odzani] R. Valley, Teague 495 (BOL, K); Melsetter, Stapleford, Nyumquarara, Wild 5685 (K, SRGH); Umtali Commonage, Chase 4955 (BM, K, SRGH); Umtali-Vumba road, near Inyanga, Corby 934 (K, SRGH); Inyanga, Phipps 1075 (EA, SRGH); near Chirinda, Swynnerton 450 (BM, K); Chipetzana, Swynnerton 1487 (BM, K); R. Nyahodi, Swynnerton 1488 (BM, K). South: Zimbabwe Ruins, Mundy 2799 (K, SRGH) & Leach 8220 (K, SRGH). TRANSVAAL. Pietersburg: Tzaneen Estate, Burtt Davy 2578 (FHO, K); Sampson (K, fragment); Houtbasloop, Mogg (K, PRE); Pietersburg, Dyer 3161 (K, PRE); Duivels Kloof, Galpin 9670 (K, PRE). Letaba: by side of old road between the prov. road and Mtataspruit, Westfalia, Scheepers 218 (K, PRE); old road between 2nd prov. road to Duivels Kloof and riverside meadow above bridge near quarry, Scheepers 123 (BM, EA, K, PRE). Nelspruit: Pretorius Kop entrance to Kruger National Park, Rodin 4138 (K, UC); Kruger National Park, Numbi, van der 2653 (K, PRE); Schioff 4-8 km. S. of Nelspruit, Leach (K, PRE). Barberton: Barberton, F. A. Rogers 23850; Barberton, Loouws Creek, van Dam in F. A. Rogers 23875; 21-"6 km. SW. of Hectorspruit, Codd 7813 (EA, K, PRE). Without exact locality, Saunders 166 in Wood 3893 (not from Natal). Burtt Davy gives the type as Zoutpansberg and Barberton, F. A. Rogers 20255, a number not represented at Kew. SWAZILAND. Piggs Peak, Lomati R., Compton (EA, K, PRE).

59 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 293 THE SPHENOSTYLIS MARGINATA E. MEY. COMPLEX There seems to be very little, if any, difference between S. marginata E. Mey. and S. erecta (Bak. f.) Bak. f. save in habit and the frequency of occurrence of acute leaflets. Most South African material is prostrate and most Central and East African material either erect or prostrate. The frequency of occurrence of various habits is shown in the following table drawn up from the Kew material. TABLE 4. Habit data for Sphenostylis marginata-s. erecta complex Country Number of erect Number of Habit not plants climbing or obvious prostrate plants Congo (Kinshasa) Io 3 Tanzania Mozambique I Malawi 8 I Zambia Rhodesia 17 I 2 South Africa I Io 9 The South African material has mostly oblong, very obtuse leaflets which can be exactly matched in Tanzania but there is a marked tendency for Central African material, particularly that which is prostrate, to have more or less acuminate leaflets. It is possible that field studies may reveal further differences but, at present, I am dividing the whole complex into three rather diffuse subspecies which are not well-separated geographically. As is usual, Zambia is the area where intermediates are most profuse. More information is needed concerning the relationship between habit and burning before any comments can be made on the possibility that the prostrate Central African plants are due to an invasion of subsp. marginata north into an area where subsp. erecta had become the common race. In most pyrophytic herbs it has been found that habit is a most variable character. A certain amount of material is cited below in condensed form. Sphenostylis marginata E. Mey., Comm. P1. Afr. Austr.: 148 (1836); I. Verdoorn in Fl. P1. Afr., t (1968). Habit climbing or prostrate: Leaflets oblong, rounded at the apex.... subsp. marginata Leaflets elliptic-lanceolate to almost orbicular, mostly subacute or acuminate at the apex.... subsp. obtusifolius Habit erect; leaflets oblong to elliptic-lanceolate, rounded to acuminate at the apex... subsp. erecta subsp. marginata SOUTH AFRICA. subsp. obtusifolia Widespread in Natal and Swaziland. (Harms) Verdc., comb. et stat. nov. S. obtusifolia Harms in Notizbl. Bot. Gart. Berl. 5: 205 ( 9'I). [S. marginata sensu Wilczek in Fl. Congo Belge 6: 278 (I954), non E. Mey. sensu stricto]

60 294 KEW BULLETIN VOL. 24 CONGO (KINSHASA). Katanga: Biano station, Symoens ZAMBIA. North: Mbala [Abercorn] District: Chilongowelo, Plain of Death, Richards 5529; Kasama District: Misamfu, Astle 1279 & Angus 2653; Mwamba's, Richards I617I; Kasama, J. M. Wright 329; Chibutubutu, R. Lukulu, 48 km. S. of Kasama on Mpika road, Richards 12540; Chisimba Falls, Exell, Mendonfa & Wild 1361 (BM); Malole Rock, Richards 12689; Kasama to Mbala [Abercorn] road, Richards I2684; Chinsali District: Shiwa Ngandu, Machipara Hill, Richards IO688; Fort Roseberry, Watmough 182; Mpika, Fanshawe West: Ndola, Symoens Ioooo. Central: Katonino (Kitinina) Hills, Kassner 2176 (B, holotypet; BM, K, isotypes). South: Mumbwa, van Rensburg 218o. RHODESIA. East: Melsetter District: Chimanimani Mts., gully above upper Haroni, Phipps 456 p.p. Umtali District: Vumba, Chase 1530 (BM, K) & Head 172 (BM); Inyanga, Mtarazi, Goodier 1035 & Chase 1627 (BM, K); R. Nyumkombu, Gilliland 1420 (BM, K). South: Bikita, Cleghorn I72. ANGOLA. Benguela: Caala, Cuima, Exell & Mendonfa 1939 (BM, COI); between Nova Lisboa and Vila Teixeira de Silva, Exell & Mendonga 1799 (COI); Ganda-Caconda, Hundt 916 (BM). Bid: Cuito, mouth of R. Campaluwe, Baum 796 (BM, COI, K). Moxico: Dilolo, R. Cassai, Dundo road, Exell & Mendon a 1495 (BM, COI). Huila: Lubango, Quilemba, Exell & Mendonfa 2525 (BM, COI). S. gossweileri Bak. f., of which I have seen only the type, comes very close to this subspecies. Material from eastern Rhodesia with I-few-flowered inflorescences may be distinct enough to recognize varietally. subsp. erecta (Bak.f.) Verdc., comb. et stat. nov. Dolichos erectus Bak. f. in Trans. Linn. Soc., ser. 2, 4: Io (1894). Sphenostylis erecta (Bak. f.) Bak. f., Leg. Trop. Afr.: 422 (1929); Wilczek in Fl. Congo Belge 6: 274 (1954); Torre in Consp. Fl. Angol. 3: 277 (1966). CONGO (KINSHASA). Katanga: Lubumbashi [Elisabethville], Quarrd 358o, F. A. Rogers io165, & Burtt Davy (BM); Baudouinville, Robyns 2254; Likasi [Jadotville], Robyns 1725 (BM, K), Hoffman 873 & Poelman 60o. Upemba Park, source of R. Lungea, Robyns Mushoshi, F. A. Rogers Mt. Morumbe, Kassner 2943 (BM, K). BURUNDI. Kitaba-Rururi, Becquet 131 (BM, K) (cited by Wilczek as S. marginata but the habit is doubtful). Murutoke-Niakassu, Peter (B) (not erect). TANZANIA. Biharamulo District: Lusahunga, Tanner 5325; Usui, Grant 206 (habit?). Ngara District: Keza, Bushubi, Tanner 5112 (EA, K) (creeper). Buha District: Mbirira [Birira]-See Deschi, Peter (B) (habit?); am See Manyoni bei Mbirira [Birira], Peter 3781o (B) (climber); Bujenzi- Kwa Bikare (olim Kwa Kijina), Peter (B) ('liegt'); Kibondo, Bullock 3'18 & Verdcourt 2872 (EA, K); Kasulu, Heru Chini, Rounce 44 (EA). Kahama District: Ushirombo, Joseph Mpanda District: Mahali Mts., Utahya, JVewbould & Jefford I684. Ufipa District: Lake Kwela, Richards Morogoro, Wallace 548 & Schlieben 4242 (BM). Chunya District: Lupa Forest Reserve, Boaler 664. Mbeya District: Mbeya, Davies D426; Mbosi, Horsburgh Porter (BM). Songea District: 32 km. E. of Songea, Milne-Redhead & Taylor

61 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 295 IO693 (EA, K); Songea area, Warne (EA). Lindi District: 1 2 km. W. of Nachingwea, Evans 15 (EA); Lake Lutamba, Schlieben 5340 (BM). MOZAMBIQUE. Between Beira and Massi Kesi (Macequege), Cecil 25. Niassa: Vila Cabral, Mt. Mandimba, Torre 264 (BM). MALAWI. North: S. Rukuru Plains, McClounie 43. Central: Lilongwe, J. M. Wright 232, Kantikana 29 (BM); Lilongwe, Chankhandwe Dambo, G. Jackson 79 & 602. South: Shire Highlands, Scott Elliot 8528 (BM, K) (habit?); Manganja Hills, Waller; Kankanje, Kirk; Cholo District, Wiehe 715; Blantyre, Banda 348 (BM, K); Mt. Mlange, Whyte (K, lectotype) & Shinn 191 (BM).? Sandawa to Msambanjati, Laurence 593; Zomba Mt., Banda 18I (BM); Ncheu District, Msasa escarpment, Dedza-Golomoti road, Exell, Mendonfa & Wild Io38A (BM). Without locality, Buchanan 63 (BM). ZAMBIA. North: Mbala [Abercorn], Bullock Io017 (EA, K), 3325, Kafuli 4I, Nash 24 (BM), Siame 6A (BM) & 709 (BM); Mpulungu-Mbala [Abercorn] road, Richards 2324; Lake Chila, Richards 4963 & Kafuli 41 (BM); Mbala [Abercorn], Simanve Farm, Richards 4966; Mbala [Abercorn] Sandpits, Richards 13109o; Mbala [Abercorn]-Kambole road, Richards 13204; Mbala [Abercorn], old Kasama road to Pans, Richards I297, 2030 & 5909; 1-6 km. from Ndundu on Kawimbe road, McCallum Webster 857; 2'4 km. from Ndundu, N'Kali side of dambo road, Richards I912; near Ndundu, Richards 2156i; R. Inono, 40 km. from Mbala [Abercorn], McCallum Webster 858B ('subclimber'); 43-2 km. S. of Mbala [Abercorn], Hutchinson & Gillett 3868; Chilongowelo, Plain of Death, Richards 1271 (partly climbing); Nkolemfumu to Kasama, Lawton 743; Kasama, J. M. Wright 307; Kawambwa, Fanshawe 3690; 8 km. from Luwingi, Richards Central: Broken Hill to Bwana Mkubwa, Allen 334; Broken Hill, F. A. Rogers 8295 & Mutimushi 952; 9 km. S. of Broken Hill, Gillett 17459; I6 km. S. of Broken Hill, Evans & Erens 1902; Lusaka, Lusaka N.H.C. 53; Chilanga, J.H.H. 163; Chilanga, Quien Sabe, C. Sandwith 88; Chilanga Fish Farm, Lusaka N.H.C. 137; Rufunsa to Lusaka, Greenway & Brenan 8074 (EA, K); 12-8 km. E. of Lusaka, King 126. East: Fort Jameson, Lunkwakwa Forest Reserve, Verboom 689; Fort Jameson, White 2437; Katete, Wright 47; Chadiza, Robson 774 (BM, K); Lundazi, White 2493; Lundazi-Chama, Tigone Dam, Robson 140 (BM, K). West: Mwinilunga District: 64 km. S. of Boma and I9-2 km. W. of R. Lunga, Milne-Redhead 91o; NE. of Matonchi Farm, Milne-Redhead 2550; Mwinilunga, Marks 8; Solwezi, White 3214, 3237 (BM, K); Mutanda Bridge, Milne-Redhead 532; Chingola, Fanshawe I 13' (EA, K); Kitwe-Nkana, Shepherd I I5a (sprawling); Mufulira, Cruse Io; 8 km. SE. of Mufulira, Cruse 384 (intermediate in habit); Ndola, Greenway & Miller 5654 (EA, K), Fanshawe 185 & Young 103 (BM); Chichele Forest Reserve, 9-6 km. W. of Ndola, Angus 369 & Fanshawe 446. South: Mumbwa, Macaulay 783 & van Rensburg 2493; Mapanza W., Robinson 943; Mapanza W., near R. Mungeke, Robinson 228; Namwala to Kalomo, Muchila's, Trapnell I loo; Kalomo, Gilges 453 (EA, PRE); Nanwala, Read 46; Pemba, F. A. Rogers 8555 & 8576; Mazabuka, Angus 91. RHODESIA. North: Shamva, Moubray in S.R.G.H (BM); Salisbury to Shamva, Gilliland 681 (BM); Miami, Rand 145 (BM); Sinoia, Rand 227 (BM); Umvukwe Hills, Mermaid's Pool, Gilliland Q627 (BM, K); Trelawney, Jack 51. Central: Selukwe, Walters 2356; Salisbury, Craster 12O; Peter 30658, Walters 2454, Rand 607 (BM), 461 (BM) & Eyles 1789; Wedza, Myres 576; Hartley, Pool Farm, Hornby East: Umtali, Commonage, Chase 700oo8;

62 296 KEW BULLETIN VOL. 24 Nyumquarara Valley, Gilliland 1585 (BM, K); Chirinda, Swynnerton 364 (BM); S. Melsetter, Swynnerton 6626 (BM). South: Zimbabwe, Chilopopo R., Rendle 407 (BM). ANGOLA. Malange: between Nova Gaia and Chassengue, Young 765 (BM). Benguela: Caala, Lepi, Calusipa, Gossweiler (BM). NESPHOSTYLIS VERDC. Nesphostylis Verdc., gen. nov. tribus Phaseolearum; Sphenostylidi Harms affinis, calycis lobis superioribus in labio truncato connatis, bracteolis magnis rotundatis persistentibus, vexillo appendicibus ornato, filamentis apice dilatatis, filamento vexillare basi dente armato valde differt. Herba perennis volubilis. Folia alterna, pinnatim 3-foliolata, petiolata, stipulae haud infra insertionem productae, persistentes; stipellae praesentes. Inflorescentiae axillares, I-florae, pedunculis apice glandulosis. Flores pedicellati; bracteolae 2, magnae, suborbiculares, calycis tubo imbricatae, subpersistentes. Calycis tubus campanulatus, lobis 2 superioribus in labio leviter bifido vel truncato connatis, 3 inferioribus ovato-triangularibus. Vexillum suborbiculare, basi auriculatum, unguiculatum, apice emarginatum, glabrum, intus medio appendicibus 2 lamelliformibus subparallelis ornatum. Alae anguste obovatae, unguiculatae, basi brevissime calcaratae, in parte inferiore leviter corrugatae. Carina obovata, unguiculata, apice leviter incurvata, rotundata. Androecium staminibus Io diadelphis, filamentis apice valde dilatatis, filamento vexillare basi dente armato. antheris 5 subdorsifixis, antheris 5 alternantibus basifixis. Discus breviter vaginifer, lobatus. Ovarium lineare, subsessile, pluri-ovulatum; stylus carti- lagineus, basi leviter incrassatus, superne in parte obtriangulare compressa dilatatus; stigma terminale. Legumina linearia, compressa, marginata, dehiscentia. Semina oblongo-ovoidea, hilo lineare ornata, arillata. Typus generis: Nesphostylis holosericea (Bak.) Verdc. Species I, Africae tropicalis incola. Nesphostylis holosericea (Bak.) Verdc., comb. nov. Vigna holosericea Bak. in Fl. Trop. Afr. 2: 200 (1871); Hiern, Cat. Afr. P1. Welw. 1: 258 (1896). V. hastifolia Bak., l.c. (1871). Sphenostylis holosericea (Bak.) Harms in Engl., Bot. Jahrb. 33: 177 (1902); Bak. f., Leg. Trop. Afr.: 419 (1929); Hepper in Fl. W. Trop. Afr., ed. 2, I: 565 (1958); Torre in Consp. Fl. Angol. 3: 275 (1966). S. kestingii Harms, op. cit.: 176 (1902). S. holosericea (Bak.) Harms var. hastifolia (Bak.) Bak. f., 1.c. (1929). S. calantha Harms in Notizbl. Bot. Gart. Berl. Ix : 818 (1933). Herba perennis, volubilis, usque I m. longa, ramis pilis patulis ferrugineis dense obtectis. Stipulae triangulares, circa 4 mm. longae; petiolus cm. longus; rhachis cm. longa; petioluli mm. longi; foliola oblonga, lanceolata vel elliptico-rhomboidea, 3-8 cm. longa, I 5-6 cm. lata, integra vel trilobata, apice acuta vel rotundata, mucronulata, basi cuneata vel rotundata, chartacea, ubique dense vel velutine pubescentia,

63 STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 297 nervis lateralibus 6-9-jugis, supra subplanis, subtus prominentibus. Inflorescentiarum pedunculi 1-5'5 cm. longi; pedicelli 2-9 mm. longi; bracteae deciduae, triangulares, 2-5 mm. longae; bracteolae 6-io mm. longae et latae, dense ferrugineo-pubescentes. Calyx velutine pubescens; tubus circa 6 mm. longus, lobis circa 5 mm. longis. Vexillum albescens vel caeruleum, demum flavescens, 2"5 cm. longum, 2-7 cm. latum, emarginatum. Alae purpureae; lamina 2-6 cm. longa, I 4 cm. lata, ungue 7 mm. longo. Carina pallida; laminae 21I cm. longae o-8-1-o cm. latae, unguibus 7-8 mm. longis. Ovarium lineare, cm. longum, stylo 8-9 mm. longo, parte superiore 3 mm. lato. 2"3 Legumina IO-12-(--I5) cm. longa, mm. lata, dense velutine griseovel ferrugineo-pubescentia. Semina saturate brunnea, mm. longa, 4 mim. lata, mm. crassa, arillo oblique incrassato 5 mm. longo ornata. (Fig. 3, P. 298.) SENEGAL. Berhaut 1799 (fide Hepper, l.c.). IVORY COAST. Cercle du Baoul&-Sud, district de Toumodi, entre Angouakoukro et Toumodi, 15 Aug. 19o9, Chevalier Vall6 du Camo&, entre le Camo6 et Yabarasso, 12 Dec. 1909, Chevalier GHANA. Ashanti, N. Agogo, Dukwesein, grass savannah, 23 Dec. 1913, Chipp 605:--blue-flowered climber. Wenchi District: Banda ravine, 21 Feb. 1953, Morton GC :-flower mauve, climber on tall grasses. TOGo. Sokod6-Basari, Kersting 497 & 518 (B, syntypes of Sphenostylis kerstingii Harmst). NIGERIA. Nupe, grassy plains, Barter 937 (K, syntype of Vigna hastifolia Bak.) :-rootstock tuberous, flowers pale blue, turning yellow as they fade, very fragrant; same locality, rates (BM). Jebba, Barter (K, syntype of Vigna hastifolia Bak.). Lokoja, in bush, 20 Oct. 1908, Dalziel 26. Lokoja, grassland, 21 Sept. 1907, Parsons LI8:-handsome mauve flowers. Gombe, in long grass, 3 Oct. 1921, Lely 655:-a climbing leguminous herb to 3 m. with very large, highly scented mauve flowers, the wings darker than the standard, keel white, a very handsome plant, even more suitable for garden cultivation than Vigna violacea, as the flowers are larger and more highly scented. Lagos, 10 Oct. 1900oo, Punch 30:-igbogula (? vernacular name). Western Lagos, Rowland:--vernacular name 'eruwa'. Abbeokuta, Irving 88. Oyo, Olokemeji Forest Reserve, NW. corner, in open savannah woodland with Daniellia and Lophira, 12 Sept. 1949, Keay in F.H.I :--herbaceous twiner with mauve flowers. Cameroons, Adamawa Division, Vogel Peak area, Gurum, at foot of pass to R. Kirimi, dry savannah with scattered cultivation and Khaya senegalensis and Parkia clappertoniana, 640 m., II 1957, Hepper 1300:- twiner several m. INov. long over tall grasses, leaves grey-green beneath, white nerves, well below the inflorescence, standard cream, tinged purple, wings mauve, keel cream. TANZANIA. Tanga District: grown at Mlingano Sisal Research Station from plants originally from Sakura Estate in Pangani District, Oct. 1961, Grundy LI38 (EA, K):-climbing twining creeper, large blue flowers, found so far only in a limited area S. of the R. Pangani and on the coast. Pangani District: Mwera, Kenge, 6 June 1908, Braun 1818 (EA) :-climber with large red flowers, vernacular names (Kiswahili) 'majani fundofundo' & (Kishambaa) 'nfundo'; Mwera sisal plantation, 28 April 1926, Peter 39926:- climber with large lilac flowers. Ulanga District: between Mahenge Plateau and the junction of the rivers Kilombero, Luwegu and Rufiji, Schauri,

64 298 KEW BULLETIN VOL. 24 r, - ' 2 :C).. ; i.,,,..?~ ' zgi I '. ' ~LL? " i-- :-, : r 7.!,? r FIG. 3. Nesphostylis holosericea. I, flowering stem to show habit, X 2; 2, portion of leaf, adaxial surface, x 4; 3, same, abaxial surface, x 4; 4, bract, X I; 5, calyx, x I; 6, standard, x I; 7, wing petal, x I; 8, keel, x I ; 9g, free vexillary stamen, x 1); io, united stamens, x i?; zi, gynoecium, longitudinal section, x I1; z2, fruit, x 5; 13, 14, seed side and face views. I, 4-z2 from Faulkner 253; 2, 3 from C. D. Adams 4866; 13, 14 from Grundy L.138.

65 A STUDIES IN THE LEGUMINOSAE-PAPILIONOIDEAE: II 299 II June 1932, Schlieben 2314 (B, holotype of S. calantha). Masasi, light woodland and grass, 450 m., 22 April 1935, Schlieben 6361 (BM). Lindi District: Nachingwea, disturbed ground by roadside, formerly miombo woodland, grey sandy soil, 420 m., 9 April 1955, Anderson 1036 (EA, K):-twining herb with perennial rootstock, corolla purple. MOZAMBIQUE. Mocuba District: Namagoa, growing among grass, 14 April 1948, Faulkner 253:-a handsome pea, growing in plantations, flowers large, pale purple, sweet-scented, opening in the morning and closing in the afternoon; same locality, in bush and plantation, 6o-120 im., Jan.-Mar. 1943, Faulkner 364 (PRE number) (Drawing 390) :-trailing and climbing plants.?o 'oi. N,?. -i- ; " , L...i.,, ".,,.t. :... ; - x.-..., "" i 8-77/2-94???.-,,,,.... :,. 6-85/ ~~~7-93/2-6766/ /0 -".).7...L..Z9 /...., 1?,:? 112/ /2 88 I 2 13/ /2 S.? :. ;-,' ". S "? /-73/ / / '3.3/.2 4. " /2,84 - r'.' 107/288I 0 6/0 52.)'3/ 60,000,000;1 i.. -6:67/ 03 ":..1: 60 6, _ 6/27,.3 3?... - \ ' '.;-, / - MAP I. Variation of petiole lengths of Neorautanenia in tropical Africa (excluding Tanzaniasee Map 2). First figure is mean length (in cms.) and second figure is ratio of petiole to rachis length. E

66 300 KEW BULLETIN VOL. 24 extending to many feet from a single root, flowers large, handsome, pale purple and scented, very common. ANGOLA. Cuanza Norte: Cazengo, Gossweiler 4553 (BM, K); Cazengo, between Cambondo and Luinha, by thickets, June 1855, Welwitsch 2258b (BM, K):-flowers about 3-8 cm. long, pods (young) cm. long, rather rare; Cazengo, Camondai, Jan. 1912, Gossweiler 5765 (BM); Golungo Alto, in bushy places at the banks of the R. Muio, Sobado de Cabanga Cacalungo, June 1856, Welwitsch 2258 (LISU, holotype; BM, K, isotypes):--whitishsilky in all parts, flowers very large, 4'5 cm. long, rose-purple. THE GENUS NEORAUTANENIA SCHINZ The genus Neorautanenia was erected by Schinz (Bull. Herb. Boiss. 7: 35 (1899)) and is equivalent to Dolichos L. section Pseudopachyrhizus Harms. As a generic unit it is compact and easily recognized but the division into species is exceedingly difficult unless a wide view is taken. C. A. Smith described a number of species from the Transvaal and also started to study material from further north. It is clear from his notes that he had decided to describe quite a number of new species. After several detailed examinations of the genus in the herbarium and over 15 years of routine naming, during which time members of the genus turned up regularly, I am convinced that only three species are clearly recognizable. For the purposes of the 'Flora of Tropical East Africa' I am not even attempting to describe named varieties although extremes can appear very different. A short erect form which flowers before the leaves, appears very distinct from a vigorous liane which flowers when the foliage is fully developed. This type of behaviour is now well known in savanna species particularly where burning plays a part in the ecology. It is of great interest to note that Story 4982 collected 272 miles NW. of Molepolole in Botswana was recorded as 'erect 2' tall' on the field note but as prostrate when cultivated at Prinshof, Pretoria. In order to investigate any possible geographical correlation between combinations of characters a large number of herbarium sheets were scored for those characters which contribute most to the difference in appearance of the various forms. The results are shown in Table 5 (p. 302). One evident correlation with geography is the ratio of length of petiole to length of rhachis and this is shown in detail in Maps I & 2* (pp. 299 & 301). The flowers are uniform in structure, differing only in dimensions; in one case only did it appear that distinct structural differences were available. Napier 2055 (Kenya) and Andrews AI456 (Sudan) have the claws of the petals extremely short and the staminal tube about half the normal length. It was discovered that this was due to the galling of the ovaries which had affected the development of the flower. Every flower on the plant can be involved, producing a deceptively different looking inflorescence. Even when only three species are recognized there are a fair number of specimens which are either completely intermediate or whose facies agree in most characters with one of the species recognized, save those which have been considered important, and hence are put into one of the other * To save space the original tables of data have been omitted and only the averages given in map form.

67 i i STUDIES IN THE LEGUMINOSAE-PAPILIONOiDEAE: II 301 4"-b......i~. "..., -_ -:... ::::::"",o.,,::::::/r:.o4: ~ 10-45/ i,.,.':"..".-' :.... MAP 2. Variation of petiole length of Neorautanenia in Tanzania. First figure is mean length (in cm.) and second figure is ratio of petiole to rachis length. species. For instance Greenway 4512, and Greenway from NE. Tanzania are basically the same as de Winter & Wiss 4227 from South West Africa. I feel that the situation in this genus can be explained either by the fact that in the south two species are just separating from the main bulk of the genus, while there is still gene flow or, more probably, that formerly quite distinct species have come into contact again recently and are hybridizing, particularly in Rhodesia. It can be seen from Table 5 that there is scarcely any- correlation between any one character or combination of characters with geography, but slight tendencies are noticeable which may show that varieties are developing. KEY TO SPECIES OF NEORAUTANENIA Petiole short and ratio of petiole to rhachis lengths under I: Leaves usually elliptic or rhomboid, entire or slightly lobed N. amboensis Leaves mostly broad and deeply divided, resembling those of Ficus carica N. ficifolius Petiole usually long and ratio of petiole to rhachis lengths over I. N. mitis Neorautanenia amboensis Schinz in Bull. Herb. Boiss. 7: 35 (24 Jan. 1899).

68 TABLE 5. Analysis of characters in Neorautan Number of Stems and foliage Mid-lobe of leaflet Habit plants I showing not not Leaflets combination densely pubescent glabrous broadened broadened, broadened, entire twining prostrate of characters velvety at apex short and long and indicated obtuse acute habit not ascertain I - - I Some I + + Some habit not ascertain habit not ascertain 4 + habit not ascertain Io + + I + divided leaflets just appearing habit not ascertain

69 habit not ascertain -Some habit not ascertai I + + Some +I I+ + Some 3 + +

70 Number of Stems and foliage Mid-lobe of leaflet Habit plants I - - _ - -_ -Petiole showing not not Leaflets combination densely pubescent glabrous broadened broadened, broadened, entire twining prostrate of charactersvelvety pubescentat apex shortand longand indicated obtuse acute habit not ascerta I I I + + Some Upper habit not ascertai habit not ascertai I I

71 t- habit not ascertain habit not ascertain S habit not ascertain I habit not ascertain + leaves absent at flowering * The numbers prefixed by U, K or T represent the geographical divisions used in the Fl. Trop. E. Afr. for

72 306 KEW BULLETIN VOL. 24 Dolichos brachypus Harms in Engl., Bot. Jahrb. 26: 323 (31 Jan. 1899). Galactia lugardii N. E. Br. in Bull. Misc. Inf. Kew 1909: 104 (1909). Dolichos seineri Harms in Notizbl. Bot. Gart. Berl. 5: 206 (191 I). Pueraria rogersii L. Bolus in Ann. Bolus Herb. I: 189 (1915)- Neorautanenia brachypus (Harms) C. A. Smith in Burtt Davy, Flowering P1. Ferns Transvaal 2: xxvii & 418 (1932). N. coriacea C. A. Smith, 1.c. & 417 (1932). N. edulis C. A. Smith, op. cit. xxviii & 4 18 (1932). N. lugardii C. A. Smith, l.c. & 417 (1932). N. rogersii (L. Bolus) C. A. Smith ll.cc. (1932). N. seineri (Harms) C. A. Smith 1.c. & 418 (1932). DISTRIBUTION. South West Africa, Transvaal, Rhodesia, Botswana. Neorautanenia ficifolius (Benth.) C. A. Smith in Burtt Davy, Flowering P1. Ferns Transvaal 2: xxviii & 417 (1932). Rhynchosiaficifolia Benth. in Harv. & Sond., Fl. Cap. 2: 251 (1862). Puerariaficifolia (Benth.) L. Bolus in Ann. Bolus Herb. I: 189 (1915). Dolichosficifolius (Benth.) Harms in Pflanzenw. Afr. 3(1): 681 (1915)- Neorautanenia deserticola C. A. Smith, 1.c. xxvii & 417 (1932). DISTRIBUTION. Transvaal, Natal, Orange Free State, Botswana. Dr. Codd and Dr. de Winter in a letter to Kew (12 Dec. 1962) concluded after an examination of the South African material that only two species could be recognized and I have followed them since there appears to be more correlation between leaf-lobing and other characters in the south than in the north where it is not really feasible to separate entire-leaved specimens from lobed-leaved specimens. Neorautanenia mitis (A. Rich.) Verdc., Common Poisonous Plants of E. Afr.: 89 (1969). Dolichos mitis A. Rich., Tent. Fl. Abyss. I: 224 (1847). Pachyrhizus? orbicularis Bak. in Fl. Trop. Afr. 2: 208 (1871). [Pachyrhizus angulatus sensu Bak. in Fl. Trop. Afr. 2: 208 (1871), quoad spec. Barter, Petherick et Schimper, non A. Rich.] Cacara? orbicularis (Bak.) Hiern, Cat. Afr. P1. Welw. I: 261 (1896). Dolichos pseudopachyrhizus Harms in Engl., Bot. Jahrb. 26: 320 (1899). D. pseudopachyrhizus Harms var. subintegrifolius Harms, 1.c. (1899). D. pseudopachyrhizus Harms var. kilimandscharii Harms, 1.c. (I899). D. ellenbeckii Harms in Engl., Bot. Jahrb. 33: 177 (1902). Pueraria hochstetteri Chiov. in Ann. Ist. Bot. Roma 8: 434 (1908); Cuf., Enum. P1. Aeth. Sperm.: 319 (1955)- Dolichos orbicularis (Bak.) Bak. f., Leg. Trop. Afr.: 451 (1929). Neorautanenia pseudopachyrhiza ('pseudopachyrrhiza')(harms) Milne-Redh. in Kew Bull. 5: 355 (195I). N. pseudopachyrhiza (Harms) Milne-Redh. var. ellenbeckii (Harms) Cuf., Enum. P1. Aeth. Sperm.: 315 (1955). N. orbicularis (Bak.) Torre in Bol. Soc. Brot., s6r. 2, 39: 216 (1965). DISTRIBUTION. Widespread in Tropical Africa.

73 STUDIES IN THE LEGUMINOSAE-PAPILIONOiDEAE: II 307 I am indebted to the Director of the Laboratoire de Phandrogamie, Paris, for sending me on loan the type of Dolichos mitis. From the description it was fairly evident what it must be. It has presumably long been ignored because it consists only of a few fruits and leaflets. I have seen authentic material of all the other names save D. ellenbeckii. It does not seem worth while to cite at length the very large number of specimens available of this common and easily recognizable plant. It should be stated that various workers who have seen this plant in the field consider my treatment wrong and would prefer to recognize several species, but the furthest I would go is to give varietal names to each of the various combinations of characters that appear commonly in Table 5. I would not even subscribe to this until some genetic information is available. Botany in Leningrad.-It is hard to explain the peculiar charm and fascination of Leningrad, the spell the city casts upon those who visit it, however briefly, calling forth feelings of affection and love. Physical beauty, a sense of living history, a scientific and cultural community of outstanding strength and vigour, human warmth and hospitality-all these and more enter into it, yet the whole is more than the sum of its parts. To the botanist, however, there is one more cogent reason for visit and return; the city is the home and birthplace of the Komarov Botanical Institute, one of the world's greatest centres of botanical research activity. Stanwyn Shetler visited Leningrad in I964 and did not escape its spell. The result is this book*, in which he attempts to present the history of the Institute and its scientific activity, from its earliest beginnings in young St. Petersburg to its 250th anniversary in The author's enthusiasm for his subject is obvious, yet he never lets it get the better of objectivity and accuracy. Bias and omissions are there, of course, as he himself readily admits. Being an American and a taxonomist, he naturally writes from an American taxonomist's viewpoint; but what matters is that a most important chapter of botanical history has been brought widely to the notice of the English-speaking world, where too long it has remained too little known. The book opens with a brief outline of the history of Leningrad and of the modern city, sketching the background to the more detailed account of the Institute itself which forms the second part. This is the bulk of the book. It is a compilation from the published sources given in the bibliography. It is good to see Russian names and titles given in their original as well as in transliterated and (where appropriate) translated form; but the Russian title to reference 8, the author's most important source, has inadvertently been omitted. A short final section attempts to assess the past and present and to look to the future. The general lay-out of the book is pleasing; it is clearly printed, free from errors and well illustrated with half-tone plates, including many portraits of past and present members of the Institute's staff. A useful chronological table is provided. Some of the footnotes are placed unfortunately far from the places in the text to which they refer, and there is a certain verbosity and repetitiveness here and there. This latter may, I fear, be more a reflection of a parallel fault in some of the sources than the * The Komarov Botanical Institute. By Stanwyn G. Shetler. Pp. xiv + 240, numerous black-andwhite photographs. Washington: Smithsonian Institution Press, I967. Price $5.95. F