1062 ME. G. A, BOULENGER ON [DeC. 10^ 



2. A Revision of the African Silurid Fishes of the Subfamily 

 Clarimcc. By G. A. Boulenger, F.R.S., V.P.Z.S. 



[Received November 11, 1907.] 



(Text-figm-es 248-257.) 



The Sihirid Fishes of the Subfamily Clariince are one of the 

 most interesting, but at the same time most difficult groups of 

 the Afi-ican fresh waters. Many of the sj^ecies of Clarias are so- 

 closely related, so difficult to deiine, and show so much individual 

 variation in characters which have been regarded as specific, that 

 I have felt compelled to take iip their study as if they had never 

 been classified before. I hope the revision which I now offer will 

 place the determination of the species on a sounder footing. 



The group is of special interest from an evolutionary point of 

 view, the series Cla7'ias — Allabenchelys — Claricdlahes — Gymnal- 

 lahes — Channallahes foi'ming what may be termed an orthogenetic 

 series * leading from a more typical Cat-fish to one that is truly 

 Eel-like, with almost every link in the chain connecting the two- 

 extremes. A few of these types are represented on the subjoined 

 text-iigures. 



The eel-shape has been reached, in this series, by an elongation 

 of the caudal part of the body concurrently with the fusion of the 

 dorsal and anal fins with the caudal, and the reduction and final 

 suppression of the paired fins, the ventrals first, followed by the 

 pectorals. At the same time the bony buckler which so efficiently 

 protects the head of the typical Clarias has been gradually 

 reduced until its complete suppression in the most elongate forms, 

 Gymnallabes and Channallabes. 



I do not believe for one moment that the more generalised 

 forms here described repi-esent the actual ancestors of the terminal 

 type, as it is not likely that they should coexist at the present day ; 

 but 1 regard the appai-ent links of the chain as side bi-anches of a 

 continuous stem, as the close allies of these extinct forms, and for 

 the purpose of the study of the lines of derivation they are just as- 

 good examples as if they were the actual ancestors, because they 

 must be so very similar to them. Even in palseontological series, 

 what we usually regard as continuous series are surely mostly made 

 up of such elements; except in the case of varieties we have 

 no evidence of any one form having turned into another, evolution 

 being after all still a hypothesis, — the only workable hypothesis, 

 built up on concurring and converging probabilities. 



It is sometimes the case that an oi-thogenetic series is sus- 

 ceptible of being interpi-eted in a reversed direction. But this 



* Tlie term ' Orthogenesis ' has been invented bj^ Haacke in 1893, and has gained 

 wide circulation through the writings of Eimer. It is intended to express deter- 

 minate evolution, as opposed to the Darwinian idea of random variation leading to 

 the formation of new types. In orthogenetic series the evolution of the organism is 

 pushed on in one direction without adaptation having to intervene, although the 

 ultimate result is an adaptation. 



