GEOGEAPIIICAL DISTEIBUTION. 205 



being permanently retained in fresh water in consequence of 

 geological changes are well known : thus Gottus quadricornis 

 in the large lakes of Scandinavia ; species of Gohius, Blennius, 

 and Atherina in the lakes of Northern Italy; Comephorus, 

 of the depths of the Lake of Baikal, which seems to be a 

 dwarfed Gadoid. Carcharias gayigeticus in inland lakes of the 

 Fiji Islands, is another instance of a marine fish which has 

 permanently established itself in fresh water. 



In the miocene formation of Licata in Sicily, in which 

 fish remains abound, numerous Cyprinoids are mixed with 

 littoral and pelagic forms. Sauvage found ia 450 specimens 

 from that locality, not less than 266, which were Leucisci, 

 Alburni, or Ehodei. Now, although it is quite possible that 

 in consequence of a sudden catastrophe ■ the bodies of those 

 Cyprinoids were carried by a freshwater current into, and 

 deposited on the bottom of, the sea, the surmise that they 

 lived together with the littoral fishes in the brackish water 

 of a large estuary, which was not rarely entered by pelagic 

 forms, is equally admissible. And, if confirmed by other 

 similar observations, this instance of a mixture of forms 

 which are now strictly freshwater or marine, may have an 

 important bearing on the question to what extent fishes 

 have in time changed their original habitat. 



Thus there is a constant exchange of species in progress 

 between the freshwater and marine faunae, and in not a few 

 cases it would seem almost arbitrary to refer a genus or even 

 larger group of fishes to one or the other ; yet there are cer- 

 tain groups of fishes which entirely, or with but few excep- 

 tions are, and, apparently, during the whole period of their 

 existence have been, inhabitants either of the sea or of fresh 

 water ; and as the agencies operating upon the distribution of 

 mariue fishes differ greatly from those infiuencing the dis- 

 persal of freshwater fishes, the two series must be treated 

 separately. The most obvious fact that dry land, which 



