414 Evolution and Distribution of Fishes 



Islands, Cayenne, Surinam, Brazil, Senegal, and the W. 

 African coast, in the Niger, from the Red Sea to India, 

 Singapore, Cochin China, Siam, Formosa, China with one 

 or two migrant species into the Chinese Sea, on to the N. 

 W. coast of Australia, and even to the Sandwich Islands. 



The genus Pimelodus is the next richest in species; has 

 nearly the same American distribution as the last; has 

 reached W. Africa and also the Sandwich Islands. The 

 above localities are mentioned in detail, in order exactly to 

 bring out the wide areas occupied by each genus. Now, 

 we are here dealing with ancient genera that are purely 

 freshwater except for American members of the division 

 Hexanematichthys, also Arms falcarius, A. dussumierii, and 

 A. venaticus that in reaching their eastern extension as 

 species of the genus, have become largely or wholly marine. 



Arius latiscutatus again is found in the Congo River 

 but passes out along the W. African coast to Fernando Po. 

 Broadly too it may be said that the 75-80 genera which 

 make up the family have a freshwater distribution that 

 follows closely that of Arius. 



How then it may be asked is such to be explained geo- 

 logically? For here we are dealing with what was ancient- 

 ly a freshwater group, as the associated beds and their 

 organisms testify; one also that has remained almost wholly 

 freshwater, but which shows a few of more advanced struc- 

 ture, that are becoming or have become brakish to marine 

 in habitat. The "floating raft" theory in no way solves 

 the problem, for it would be hard to imagine even a large 

 piece of land becoming separated, and yet retaining a fresh- 

 water area on it, in which freshwater fishes would survive. 



Almost as difficult would it be to imagine that eggs of 

 freshwater fishes could be carried on such miniature float- 

 ing islands, and yet survive to hatch out. The problem 

 however becomes a natural and easy one, if we again accept 

 such land connections geologically as have already been 

 outlined, and which may have existed at least from late 

 Cretaceous to early Eocene time. 



The distribution of the species in the other seven sub- 

 families, follows greatly what has been outlined for Arius. 

 But it may be noted that the relation of those known fossil 



