A Study in the Natural History of Ice 301 



is more advanced, the wall may not be water-tight and air 

 penetrates easily into the space left vacant in the act of 

 fusion, and we have really air-bells. 



Let us imagine that the mass of ice consists of grains 

 which are all of equal size and have the regular form of cubes. 

 During intergranular fusion, a trough forms itself round each 

 grain which is illuminated, and it is filled with water to the 

 extent of nine-tenths of its volume, while the remaining tenth 

 is filled with air or vapour. 



If the tubes of an upper layer do not lie conformably on 

 those of the lower layer, an apparent stratification is at once 

 produced, and it would have a very regular character. But in 

 the real glacier-ice, as we meet with it in nature, the grains 

 are quite irregular in form ; and, in order that they may be 

 capable of completely filling space, they must necessarily be 

 of very different sizes. When radiation produces intergranular 

 fusion in such a mass, the apparent stratification produced 

 must be irregular. This kind of stratification, crude though it 

 may be, is always produced in the white surface layer of the 

 glacier, and may be recognised in lines on the rough exterior 

 of the glacier. 



When the winter returns, the intergranular water which 

 has accumulated during the summer begins to freeze so soon 

 as the cold has begun to affect the inner walls of the grotto ; 

 but this takes some time, during which there is neither melting 

 nor freezing in the grotto. When, however, it does begin to 

 take effect, the outer surface of the wall of the grotto is cooled 

 down below o C., and this temperature propagates itself into 

 the mass of the ice as an isothermal surface, which may be 

 considered a vertical plane. Isothermal planes corresponding 

 to lower temperatures follow in its wake, and first the inter- 

 granular water freezes at a temperature which depends on its 

 purity, leaving, however, always a relatively impure liquid 

 nucleus often of microscopic dimensions. While the inter- 

 granular water is being frozen, it attaches itself to the grain in 

 conformity with its crystalline orientation, so that the new ice 

 added is not to be distinguished from the pre-existing ice of 

 the grain. If there has been no leakage, then, when all the 



