PHANEROGAMS— PROFESSOR BAYLEY BALFOUR. 45 



2. DIRACHMA. 



Dirachma, Schweinf. in Proc. Roy. Soc. Edin. xiii. (1883). 



Flores regulares. Calyx 8-partitus, lobis valvatis. Petala 8, perigyna, imbricata. Grlandulse 

 disci inconspicuae. Stamina 8, libera, petalis opposita, omnia antherifera ; antherse magnse 

 oblongse. Ovarium 8-lobum, 8-loculare, rostratum ; stylus centralis integer, obtusus ; ovula 

 in loculis solitaria, adscendentia. Capsula 8-loba in carpella 8 ventraliter debiscentia intus 

 lanata secedentia. Semina compressa, in loculis solitaria ; testa nitida ; albumen 

 sparsum. — Frutex ramosus plusminusve pubescens. Folia alterna, dentato-serrata, paullo 

 revoluta, stipulata. Pedunculi axillares, 1-flori. Flores albi. Calyx 4-bracteatus. 



An extremely interesting monotypic and endemic genus, the nearest 

 affinity of which is somewhat puzzling ; but I think its position in this family 

 is most satisfactory. We obtained several specimens of the plant, but without 

 flower, and it is owing to Schweinfurth's beautiful flowering specimens that I 

 have been able to determine its relationship. Its general habit is that of 

 Tiliacese, but its imbricate petals, and the definite stamens are features hardly 

 reconcilable with its position in that family. Sterculiaceae might almost claim it, 

 and indeed there is no character forbidding its inclusion in the family, though the 

 free perigynous stamens are exceptional. With Samydacese and allied calycifloral 

 orders there are very strongly marked affinities, but the solitary ascending ovule 

 seems to exclude it. In Geraniaceae it appears to me to find its most natural 

 neighbours in the tribes Vivianiece and Wendtiew. Technically it does not 

 exactly fit into either as at present constituted: from the former its alternate 

 leaves, solitary ovule, and bracteolate calyx, amongst other points, separate it; 

 from the latter its valvate sepals is an easily ascertained diagnostic mark. 

 But the general facies of the plant with the sum total of its technical characters 

 mark its near alliance with the small South American (Chili and Peru) mono- 

 typic genera Wendtia and Balbisia, and with the somewhat larger genus Viviania, 

 and in their vicinity I therefore place it. This, it will be observed, is a very note- 

 worthy generic affinity, and there are others of like character in the flora of the 

 island which are referred to under the respective genera (see Thamnosma, 

 Ccelocarpum). 



Such an antipodean relationship is not a peculiar feature in the Socotran 

 flora, but, as is pointed out in the Introductory Chapter, is characteristic of the 

 islands in the Indian Ocean. But I may here specially direct attention to 

 the resemblances which this plant bears to the Samydacese and allied orders; 

 for in the monotypic Mathurina, endemic in Rodriguez, with a Central 

 American affinity, we have a Turneraceous genus which, as I have elsewhere 

 pointed out, (Proc. Linn. Soc. xv. (1877), 162), has many characters of the 

 SamydaceaB. And thus two genera, each endemic in an Indian Ocean island 

 and belonging to the same alliance of plants, present the same special features 

 in the distribution of their immediate allies. 



