216 FRAGMENTS OF SCIENCE. 



plete they were. Suppose the temperature of our 

 planet to be reduced, by the subsidence of solar heat, 

 the cold of space, or any other cause, say one hundred 

 degrees. Four-and-twenty hours of such a chill would 

 bring down as snow nearly all the moisture of our 

 atmosphere. But this would not produce a glacial 

 epoch. Such an epoch would require the long-continued 

 generation of the material from which the ice of glaciers 

 is derived. Mountain snow, the nutriment of glaciers, 

 is derived from aqueous vapour raised mainly from the 

 tropical ocean by the sun. The solar fire is as neces- 

 sary a factor in the process as our lamp in the experi- 

 ment referred to a moment ago. Nothing is easier 

 than to calculate the exact amount of heat expended 

 by the sun in the production of a glacier. It would, 

 as I have elsewhere shown, 1 raise a quantity of cast 

 iron five times the weight of the glacier not only to a 

 white heat, but to its point of fusion. If, as I have 

 already urged, instead of being filled with ice, the 

 valleys of the Alps were filled with white-hot metal, of 

 quintuple the mass of the present glaciers, it is the 

 heat, and not the cold, that would arrest our attention 

 and solicit our explanation. The process of glacier 

 making is obviously one of distillation, in which the 

 fire of the sun, which generates the vapour, plays as 

 essential a part as the cold of the mountains which 

 condenses it. 2 



It was their ascription to glacier action that first 



1 * Heat a Mode of Motion,' fifth edition, chap. vi. : Forms of 

 Water, 55 and 56. 



1 In Lyell's excellent Principles of Geology,' the remark occurs 

 that * several writers have fallen into the strange error of supposing 

 that the glacial period must have been one of higher mean tempe- 

 rature than usual.' The really strange error was the forget fulness 

 of the fact that without the heat the substance necessary to the 

 production of glaciers would be wanting. 



