44 BOTANY OF SOCOTRA. 



Certainly, placed alongside typical F. cretica, this appears a very distinct 

 species. But with Anderson and Oliver I am driven to regard F. cretica as a 

 protean species with its variations more or less constant, yet so gradually 

 merging one into the other as to render specific diagnoses hardly possible. 

 This Socotran form has a strong individuality, more than in any other one of the 

 numerous forms which" have been regarded as species by authors. Its main 

 features are the persistently unifoliate leaves, which are large and fleshy, and its 

 stipules, reduced to small somewhat membranous scales, never showing a trace 

 of spines. By any one who regards F. cretica, as referred to here, an aggregation 

 of species, our plant will of necessity be considered a distinct species, and with 

 more reason than can be advanced in support of the constitution of many of the 

 frequently described species. 



Order XIX. GERANIACE^. 



A considerable order, found chiefly in temperate regions but with repre- 

 sentatives all over the globe. There are three genera in Socotra, two having 

 many species and with the distribution of the order ; the third is endemic and 

 monotypic, with extremely interesting South American alliances. 



1. GERANIUM. 



Geranium, Linn. Gen. n. 832 ; Benth. et Hook. Gen. PI. i. 272. 



A large genus widely spread in temperate regions, especially in the northern 

 hemisphere, and within the tropics occurring in mountainous districts. 



G. mascatense, Boiss. Diagn. ser. 1, i. 59, and Flor. Orient, i. 882. 



Socotra. On the hills near Galonsir. Common. B.C.S. n. 242. 



Distrib. Arabia. 



A species founded by Boissier on specimens collected by Aucher Eloy on 

 hills near Muscat. Specimens are in Kew Herbarium from north Midian 

 brought by Burton, and now it turns up in Socotra. 



It is nearly allied to G. molle, Linn. (Sp. 955), but sufficiently separated, as 

 Boissier points out, by the more deeply cut leaves with acute laciniae and by the 

 rugose non-reticulate fruit valves. Another point of difference lies in the 

 corolla. In G. molle the petals are obcordate, only slightly longer than the 

 sepals, the corolla of G. mascatense was unknown to Boissier, and as it has not 

 been described I give its characters : — " petalis purpureo-nervosis minutissime 

 unguiculatis obovatis supra basin ciliatis calyce fere duplo longioribus." The 

 petals are altogether narrower and longer than in G. molle. 



