CHAP. II.] REPTILES AND FISHES. 20 



The amphibia are much less sensitive to cold than are true 

 reptiles, and they accordingly extend much farther north, frogs 

 being found within the arctic circle. Their semi-aquatic life 

 also gives them facilities for dispersal, and their eggs are no doubt 

 sometimes carried by aquatic birds from one pond or stream to 

 another, Salt water is fatal to them as well as to their eggs, and 

 hence it arises that they are seldom found in those oceanic 

 islands from which mammalia are absent. Deserts and oceans 

 would probably form the most effectual barriers to their dis- 

 persal ; whereas both snakes and lizards abound in deserts, and 

 have some means of occasionally passing the ocean which frogs 

 and salamanders do not seem to possess. 



Means of Dispersal of Fishes. — The fact that the same species 

 of freshwater fish often inhabit distinct river systems, proves 

 that they have some means of dispersal over land. The many 

 authentic accounts of fish falling from the atmosphere, indicate 

 one of the means by which they may be transferred from one 

 river basin to another, viz., by hurricanes and whirlwinds, which 

 often carry up considerable quantities of waterand with it fishes 

 of small size. In volcanic countries, also, the fishes of subter- 

 ranean streams may sometinies be thrown up Ijy volcanic explo- 

 sions, as Humboldt relates happened in South America. Another 

 mode by which fishes may be distributed is by their eggs being 

 occasionally carried away by aquatic birds ; and it is stated by 

 Gmelin that geese and ducks during their migrations feed on the 

 eggs of fish, and that some of these pass through their bodies 

 with their vitality unimpaired.' Even water-beetles fiyiug from 

 one pond to another might occasionally carry with them some of 

 the smaller eggs of fishes. But it is probable that fresh-water fish 

 are also enabled to migrate by changes of level causing streams 

 to alter their course and carry their waters into adjacent basins. 

 On plateaux the sources of distinct river systems often approach 

 each other, and the same thing occurs with lateral tributaries 

 on the lowlands near their mouths. Such changes, although 

 small in extent, and occurring only at long intervals, would 



' Quoted in Lyell's Prindijlcs of Geology (11th ed. vol. ii. p. 374), from 

 Aiuam. Acad. E><mij 75. 



