6o Records of the Indian Museum. [Vol. XXIV, 



Annandale, 1 when discussing the evolution of the adhesive 

 apparatus in hill-stream fishes, made the following remark 

 about the genus Garra : — " Whereas the chief factor in the case of 

 Psilorhynchus was rapid-running water in a rocky stream-bed, in 

 Discognathus the primary factor was a peculiar mode of feeding." 

 Quite recently I * also subscribed to this view, but a more detailed 

 study of the adhesive apparatus has led me to modify my previous 

 ideas. I believe that the mental disc of Garra has not primarily 

 been evolved for the " peculiar mode of feeding," which is practi- 

 cally similar in all the genera of hill-stream fishes, but for securing 

 adhesion to rocks in rapid running waters. The adhesive appar- 

 atus on the under surface of the paired fins is an additional organ 

 of adhesion in species that live in very rapid waters. In certain 

 highly evolved species of Garra, which have secondarily taken to live 

 in lakes and pools, the peculiar pad-like structure on the under 

 surface of the paired fins have disppeared though the charac- 

 teristic mental disc is still present, as it is probably of use to the 

 fish in its " peculiar mode of feeding ' ' which was acquired as a 

 direct response to a life in hill-streams. 



I am, therefore, led , to conclude that none Of the hill-stream 

 forms are ancestral forms; but that all of them are descended 

 from migrants from the slow-running streams. The modifications 

 that some of these forms exhibit are due to the physical condi- 

 tions prevailing in mountain-rapids ; and it is to this cause that 

 we must ascribe the similarity in form and structure exhibited by 

 the more advanced members of the genera dealt with in this 

 paper. 



Means of dispersal. — When -dealing with the fish of Manipur 

 it was pointed out by me that most of the new species from the 

 hill-streams had a localised distribution. This is the case with 

 almost all the hill-stream fishes and naturally it is difficult to 

 imagiue a wide range of distribution of these forms. The most 

 highly modified forms are not capable of living for a long time 

 in muddy channels, on account of the form of their bodies and 

 the structure of their jaws. In cases where a very wide range 

 has been attributed to a hill-stream species, it has always been 

 found on comparison of material from different localities that 

 several allied forms had been grouped together under the same 

 name and that most of them are capable of specific separa- 

 tion. For example Glyptosternum labiatum which was described 

 from the Mishmi Hills in Upper Assam, was recorded by Vinci- 

 guerra 8 from the Kachin Hills, Upper Burma. Regan* in 1905 

 separated the Burmese specimens from those collected in the 

 Mishmi Hills under the new name G. vinciguerrae. A remarkably 

 wide range of distribution was attributed to Garra lamta, but as 



' Annandale, Rec. Ind. Mus. XIV, p. 117 (1919). 



2 Hora, Rec. Ind. Mus. XIX, p. 213(1920). 



3 Vinciguerra, Ann. Mas. Civ. Star. Nat. Genova, (2), IX, p. 252 (1889). 

 * Regan, Ann. Mag. Nat. Hist. (7) XV, p. 184 (i9°5)- 



