14 



THE JOURNAL OF BOTANY 



which characterise those of the latter species. The leaf and flower 

 also recall I. obscura Chois., hut the hirsutulous narrow pointed 

 sepals at once distinguish it. 



Hab. British East Africa. Machakos, S. L. Himie, 1896. 

 I. cEAssiPEs Hook. In Bull. Herb. Boiss. vii. 44-48, Dr. Hallier 

 has elaborated this species, extending it to include a number of 

 species previously described by himself and others. These are 

 arranged under ten varieties, and one variety is further divided into 

 subvarieties. I have not seen all the specimens cited in the above 

 arrangement, but, after carefully working through the accessible 

 material, I find myself somewhat at variance with the conclusions. 

 For instance, if I. kewittioides Hall. f. becomes I. cmssipes var. 

 heivittioides, why is not 1. fulvicaulis (Aniseia fulvictiulis Hoclist.), to 

 which 1. heivittiuides shows far more resemblance than to typical 

 I. crassipes, also included as a variety ? The same question arises 

 with other species, e.g. /. asperifolia Hall. f. ; in fact, once start 

 making these species-aggregates, and it is not easy to stop. Dr. 

 Hallier, having gone so far, should certainly have gone farther. I 

 have tried to arrange the forms in question in accordance with the 

 more generally accepted views on the limitation of species, with the 

 following results : — 



1. crassipes Hook. (/. calystegioides E. Meyer) comprises the 

 South African forms included by Hallier under var. genuina, var. 

 long epe dune ulata, var. ovata, and probably var. thunbergioides, from 

 the description. These hang together fairly well, varying in the 

 greater or less hairiness, breadth of leaf, length of peduncle, and 

 moderate to large lanceolate to ovate bracteoles. Var. iikambensis 

 Hall, f., which is I. ukambemis Vatke, in LinniBa, xliii. 510, from 

 East Tropical Africa, must, I think, be regarded as a distinct species 

 closely allied to /. crassipes, but differing in the markedly rounded 

 apex of its oblong leaves. Var. Jiewittioides Hall. f. (/. kewittioides 

 Hall f.), in Engl. Jahrb. xviii. 127 (Dec. 1893), /. andowjejise Rendle 

 and Britten in Journ. Bot. 1894, 171, an Angolan plant, is, as I 

 have already indicated, very distinct, and more nearly allied to 

 /. fidvicaidis, both in form of leaf and in the dense several-flowered 

 inflorescence. 



Hallier also suggests that another Angolan species (i. adumbrata 

 Rendle and Britten in Journ. Bot. 1894, 173) may be synonymous 

 with his var. ononoides from the Transvaal. I have seen no 

 authenticated specimens of the variety, but should prefer to retain 

 /. adumbrata as a distinct species of the affinity of /. crassipes, but 

 separated by the shorter, proportionately broader, very obtuse 

 leaves, and small slender almost linear bracteoles. 



1. Greenstockii Rendle in Journ. Bot. 1896, 35, is also indicated 

 as a synonym of the species (p. 44) ; in fact, if we exclude the two 

 Nyassaland specimens which are on p. 46 tentatively referred to 

 var. akambensis, it remains as the sole typical representative. It is, 

 however, a distinct form with dwarfed growth, short crowded 

 ascending to suberect branches, and long narrow suberect leaves ; 

 and, whether or no specifically distinct, is certainly far from typical. 



On this view of the species there are left two plants not yet 



