Cockcrdl — The Scales of the African Cyprinid Fishes. 147 



B. pmnauratus, sarana (large scale, with ])asal region larger), 

 scJianicus, dorsalis, diola, pleurotxnia, and the African B. perince 

 (R. Xile) and camptacanlJius. 



(n.) B. rjonhjsoma group. Like the last, hnt the polygonal nuclear 

 pattern excessively developed. Includes the Asiatic ijoniosoma, 

 maculatus (fish very like goniosoma) and hurmanicus. The African 

 B. f/arnci/i (natalensis) may go here, but is peculiar. 



(o.) B. h I o-chelli gronp. Like pinnaMr«<MS group, V^ut scales Ijroader. 

 Lateral circuli coarse in hurcJiclU, fine in others, as mahccoln. In- 

 cludes the African -B. burcheUi, t.niiurus, capensis, aspilus, (some 

 polygonal areas), guirali, nunimifer, rJioadesl (\mt radii weak) and 

 trispilus {n^aJkeri). Asiatic species are B. lateristriga (scale not so 

 l)road), ]mlawaiiensis, liampal (immense scale, not so broad) and 

 maliccola. 



(p.) B. cJtri/f^opomd (Asiatic). Scale of the ])hinniiratus type, but 

 long, with the nuclear area far apicad. 



It is not suggested that all these groups are natural, l)ut to some 

 extent at least they must be significant. Later studies will no 

 doubt suggest modifications, especially wlien tlie characters of the 

 fishes are correlated. Some generic and subgeneric names are 

 applicable as follows : 



Cheilobarbus A. Smith. Type, cnpensis (gr. o. ) 



Pseuilobnrbus A. Smith. Ty|ie burchi'Ili (gr. o. ) 



Ani'inaticJilhi/.'f Bleeker. Type upogon (gr. k. ) 



i/f m//jar/v/s Bleeker. Type ^rtc/m.s Schlegel iachlcgclii) (gr. k. ) 



Labcobdrbij.f Uiipi). Type vedgia (gr. g. ) 



Numerous other names have been jiroposed for subdivisions of 

 BcD-hus, which is here interpreted in the broadest sense, following 

 Bouleuger. No doul)t some of these represent valid genera, and it 

 is pnibable that Bctrbus should ])e restricted to the Pala?arctic 

 groups, but I am not in a position to propose a new classification. 

 According to any reasonable scheme, it appears that enough generic 

 names already exist to supply all requirements, unless perhaps a 

 name should be ]troi)osed (at least in a subgeneric sense) for 

 B. grahami of Regan. 



Day, in liis work on the species of British India, recognized 

 three subgenera: 



Barbados, with four barbels. 



Capo'i'ia, with two barliels. 



Puiitius, without barbels. 



I have examined the scales of 17 of his Barbodcx, five of his 

 CapoHa, and two (ajxjgon and Jilatnenlosiis) of his Pnnlius. On 

 the scale characters, the groups do not hang well together, and for 

 this and other reasons I believe they are to a considerable extent 

 artificial. Day calls attention to the great resemblance between 

 B. maliccola, which has four barbels, and B. filamcnlosus, which 



