FISHES OF SOUTH AND MIDDLE AMERICA 517 



compelled me to abandon the work when but half finished. Occasional 

 collections received from South America for identification have, how- 

 ever, kept up my interest. The entire problem presented by this fauna 

 has been reviewed and will be published in one of the reports of the 

 Hatcher expedition to Patagonia. I am permitted to publish this 

 summary of results through the courtesy of the editor of the Hatcher 

 volumes, Professor W. B. Scott, of Princeton University. 



1. Variety of Fish Life. 

 On February 23, 1866, Louis Agassiz wrote to the emperor of Brazil: 



I estimate the total number of species which I actually possess [from the 

 Amazon] at eighteen hundred, and it may be two thousand. 



To this is added a footnote in Agassiz's ' A Journey in Brazil,' p. 385'. 



To-day I can not give a more precise account of the final results of my 

 survey. Though all my collections are safely stored in the museum, every 

 practical zoologist understands that a critical examination o'f more than eighty 

 thousand specimens can not be made in less than several years. 



Agassiz secured more species from a small lake in the valley of the 

 Amazon than there are in all the fresh waters of Europe. 



The number of species he collected was overestimated by Agassiz. 

 While about half of his Amazonian collections have not, after forty 

 years, been examined, it is certain that the species not yet examined 

 will not swell his list to 1,800 species. The total number of species 

 recorded from the Amazon basin up to date is 674. 



Although Agassiz's estimate of the number of species he collected 

 is too high, the total number of species found in South America is very 

 great. About ten per cent, of all the known species of fishes have been 

 recorded from the freshwaters of South America. In 1892 I estimated 

 that three fourths of the fauna was known. Now, after examining 

 recent lists, and considering that collections have largely been made in 

 easily accessible and great highways, and that from great river basins 

 like the Purus, Tapajos, Xingu and the Uruguay and the greater part 

 of the Madeira and the Tocantins we have nothing at all, and that 

 even from the great Orinoco and Magdalena we know next to nothing; 

 I doubt very much if we even yet know so much as three fourths of 

 the fauna of the area between the Caribbean Sea and the Argentine 

 Republic. 



The tropical American fresh-water fauna, having its center of 

 greatest diversity in the middle Amazon basin, is attenuated northward 

 till it reaches the vanishing point just on the borders of the United 

 States. Southward it extends to somewhere — no one knows where — 

 south of Buenos Aires. The Patagonian fauna and North American 

 fauna are entirely different from the tropical American fauna and from 

 each other. 



The key to the great diversity of the tropical American fauna is to 

 be found in the enormous single water system, extending from 10° 



