EIGENMANN: the freshwater fishes of BRITISH GUIANA 103 



to take advantage of the temporary union, and frequently do take advantage of it. 

 Such species usually have a very wide distribution. 



In other regions where falls as high as the Kaieteur are the only means of 

 migration, they are effective barriers to the dowmvard as well as upward migration 

 of fishes. It would seem, however, from examining the distribution in the preceding 

 list of species 2, 8, 9, 10, and 20, that in some way or other fishes succeed in getting 

 down from the plateau. Some of these species, especially 2, 8, and 10, are abundant 

 above the fall, while below it they were taken at Tukeit only, or but a few miles 

 farther down the river. They are evidently fishes of the plateau which in some way 

 have made their way down. 



In contrast to these we have the species numbered 1, 4, 7, 9, 11, 12, 14, 15, 16, 

 17, 18, 18a, and 21, abundant everjTvhere below, but also found on the plateau. 



These species probably came up from below. They certainly did not ascend 

 the Kaieteur. The fact that a number of species got down may imply unknown 

 means of ascent and descent near the Kaieteur. Whether any of them ascended 

 from the side of the Amazon I cannot say. It is quite probable, as all are found 

 in the Amazon. 



The method of ascent of vertical walls by members of the genus Rivulus, one 

 of which is found on the plateau, is of great interest. One of these fishes taken in 

 Shrimp Creek jumped against the vertical face of a huge rock and clung by the 

 adhesion of its tail. From this point by another flop it made and clung to a point 

 much higher up the face of the rock. I do not know how high this genus would be 

 able to climb a vertical wall. It certainly would not be able to ascend the face 

 of the Kaieteur, but there may be smaller rivulets up one of which it might have 

 made its way. 



The fact that a very large part of the fauna consists of species of the widest 

 distribution, or of local species of genera of the widest distribution, seems to make 

 the recent settlement of the plateau by some of the species certain. 



E. Whether or not the fauna is a mixture, the genera and species of the remain- 

 ing group "c," i. e., Lithogenes villosus, Corymbophanes andersoni, Helogenes marnio- 

 ratus, and Poecilocharax bovallii should answer. 



Now Corymbophanes differs from Plecostomus in trifling characters only. The 

 genus Plecostomus is one of the widely distributed genera of South America, being 

 found on both slopes of the Andes and from Panama to Buenos Aires. Corymbo- 

 phanes may therefore be a local modification of a comparatively recent immigrant 

 to the plateau. 



Helogenes marmoratus, on the other hand, is found in the Essequibo basin 



