228 



slope of South America show no such variation from their Atlantic slope 

 relatives. They would do so were the shortness and swiftness of the 

 river the cause. 



The most striking example is offered by the Percopsidse with a single 

 species on the Atlantic slope and another on the Pacific slope. Percopsis 

 on the Atlantic slope has feeble unsegmented rays in front of the dorsal 

 and in front of the anal which in Columbia on the Pacific slope are trans" 

 formed into strong spines. 



Similar structural peculiarities are to be observed in other regions. Mr. 

 A. J. Woolman informs me that all the minnows of Mexico have the 

 teeth in one row and four teeth in each row. Such uniformity is not 

 found elsewhere in America. This peculiarity being found in all the spe- 

 cies may be explained by the application of natural selection. The pecu- 

 liarity of the food is probably such as to bar the road to Mexico to all but 

 species with the teeth 4 to 4 or to kill off any that may have entered this 

 region, the minnows of Mexico having all entered from the north. 



Lake Titicaca in the high Andes of South America is inhabited by a 

 peculiar genus of Cyprinodontidie (Or^'st/as). The numerous species all 

 lack ventral fins. Perhaps no one would be willing to insist that each of 

 these species formerly possessed ventral fins and lost them through envi- 

 ronmental influences independently. The only plausible explanation seems 

 to be that they have diverged from a parent stock which had lost its ven- 

 trals before it split into a number of species. But five other genera of 

 freshwater fishes are known to me to lack ventral fins. They are: 



Ajma, a Minnow— Tennaserim. 



Tellla, a Cyprinodont — Alpine pools of the Atlas. 



Astroblepus— Rio de Palace near Popayan (in the Andes.) 



Eremophilus — Bogota (in the Andes.) 



CJianna — Ceylon! 

 A glance at this list will show that four of these six inhabit high moun- 

 tain waters. (The character of the water of the Ceylon species is not 

 stated.) 



Now, while not all mountain fishes lack ventral fins, the fact that two- 

 thirds of the fishes lacking ventrals inhabit mountain homes, half of them 

 living in the waters of the Andes seems to indicate that here, as in 

 the Pacific slope fishes, we have a case of convergence — that we have 

 to deal with a character several times produced in remote types by the 

 direct influence of the environment. Since the genera lacking ventrals 



