THE AFRICA1«^ SPECIES OF ALLOPHTLUS 155 



of the thvrse. Students of the genus are much indebted to Dr. Radl- 

 kofer for his most careful work, and especially for his paper in Sitz. 

 Bayer. Akad. Wiss. xxxviii. 221-237 (1909) " Uber die Gattung 

 Allopliylus und die Ordnung ihrer Arten." 



The division into unifoliolate leaves and trifoliolate leaves is not 

 entirely satisfactory : in certain special instances such as A. congolanus 

 Gilg., in vv^hich, although usually trifoliolate, the lateral leaflets are 

 occasionally entirely absent. The division also of trifoliolate-leaved 

 s])ecies into those with a simple thyrse, and those with a branched 

 thyrse, is also occasiorally difficult of application, as in some species — 

 such as, for instance, A. repandus Engl. — a simple and branched thyrse 

 is often found on the same specimen. I note that Dr. Kadlkofer 

 doubts whether this latter species is really specifically distinct from 

 A. alnifolius Radlk., the former being founded on Schmidelia rcpanda 

 Ijaker, the latter on 8. alnifolia Baker. I have careful!}^ examined 

 the types ; in the latter the leaves are cuneate-obovate and blunt, 

 whilst in 8. repanda they are broadest about half way down and acute, 

 and as far as one can judge the species are quite distinct. 



Dr. Gilg has also made important contributions to our knowledge 

 of the African species ; his papers are in Englers Jalirhnch, xxiv. 286 

 (1897), where he published 17 novelties, xxx. 348, where are three, and 

 in 1914 in the Botany of the Deutschen Zentral- Africa Exjjedition^ 

 474 (1911) he published three species. In Journ. Linn. Soc. xxxvii. 

 136 (1905) I described A. pseudopaniciilatus, A. suhcoriacfus, and 

 A. lat&foliolatus from material collected by Dr. Bagshawe in Uganda, 

 and in xl. 48 (1911) A. chirindensis, from specimens collected by 

 Mr. C. F. M. Swynnerton in the Chirinda Forest. 



For the discrimination of the trifoliolate species special attention 

 must be paid to the character, consistence, absence or presence of 

 indumentum, and toothing of the leaflets, the absence or presence of 

 ])etiolules, the character of the thyrse, the size of the flowers, and the 

 size and shape of the fruit. The structure of the flower does not seem 

 to be of primary importance for taxonomic purposes. 



In the following enumeration of the African species known to me, 

 either from herbarium material or description, 1 have thought it 

 advisable to retain as far as possible Dr. Radlkofer's sequence, interpo- 

 lating the more recently described novelties and those here described 

 in their correct positions. 



Clavis specter um. 



A. Folia unifoliolata rarissime trifoliolata. 

 Thyrsi simplices. 



Thyrsi petiolos vix aequantes vel paullo superantes. 

 Ramuli hirtelli vel hirsuti. Foliola papyracea. 

 Foliola margine dentata apice abrupte acuminata . 1. hirtellus. 

 Foliola margine crenato- dentata apice longe acumi- 

 nata 2. nigericus. 



Thyrsi mediocres petiolos superantes. Kamuli 

 cinerei, glabri. Foliola chartacea, apice acuta 



vel obtusa 3. Pervillei. 



Thyrsi foliorum vix J adasquantes. Calyx ferrugineo- 



tomentosus 4. hylophilus. 



Thyrsi longi. Eamuli puberuli vel glabriusculi. 



Calyx glaber vel subglaber 5. mono'plyUvs. 



Thyrsi ramosi, rami glabri 6. yneUiodvnO:. 



> 2 



