GLACIERS OF THE CANADIAN ROCKIES AND SELKIRKS. 123 



these bands represent portions of the glacier that have been completely liquified 

 by pressure, allowing the air bubbles to escape, seems, to the writer, very im- 

 probable, for four reasons, (i .) Blue bands occur in the basal layers, parallel with 

 the valley floor. The thickness of the ice of an ordinary glacier is not sufficient 

 to induce general melting by its simple weight. (2.) If the granular structure of 

 the glacier is completely destroyed by melting, it cannot be reproduced by simple 

 freezing, and still the granules are best developed in the blue band areas. (3.) Oc- 

 casional granules may be found which extend from the blue bands into the ad- 

 jacent vesicular ice. (4.) Water freezing in cavities in the body of a glacier should 

 form a series of prisms, standing with their main axes at right angles to the ice 

 surfaces bounding the cavity. Such filled cavities are found in the ice but they 

 do not constitute blue bands. 



d. Ice dykes. In connection with the Lefroy Glacier chiefly, there were 

 noted in the early summer what appeared to be former crevasses, filled with ice and 

 forming ice dykes in the body of the glacier. Some of these were cut by crevasses, 

 testifying to their greater relative age and suggesting that they might persist 

 from one season to another. A few of the dykes contained granular ice, the 

 granules being moderately coarse, and were assumed to have been formed by the 

 filling in of crevasses with ice avalanched from the hanging glacier upon Mt. 

 Lefroy. Most of the dykes, however, were completely filled with a double tier 

 of ice prisms, having their bases attached to the walls of the crevasse and 

 extending horizontally out into the cavity, at approximately right angles. 

 Generally the prisms met at the centre those from the opposite face of the crevasse 

 and their inner ends interlocked. Sometimes a space was left between the oppo- 

 site tiers of prisms. Ellipsoidal shaped spaces were also found completely filled 

 with radially arranged prisms meeting at the centre. The explanation given for 

 these features is that they were formed by the freezing of water in crevasses, 

 and other cavities, in the spring, or early summer, while the glacier still retained 

 a sufficient degree of its winter's temperature. The water was supplied by the 

 early melting, or rains, and the freezing surfaces were the walls of the crevasse, 

 instead of the lower stratum of the atmosphere, as is usually the. case. Since in 

 freezing, water forms a series of parallel prisms, with their axes lying, as a rule, 

 at right angles to the surface of refrigeration, these prisms have the abnor- 

 mal horizontal position, instead of the usual vertical one. Although the upper 

 part of the dyke may have been lost by melting, there was no evidence that a 

 horizontal stratum of ice had formed across the top from freezing induced directly 

 by the atmosphere. 



Drygalski has argued that it is pressure that determines the direction 

 that the crystalline plates will assume when water is freezing, and that ttie 

 main prismatic axes will lie parallel with this pressure. In the case of 

 the ice of a lakelet or basin, after it has once been enclosed by the ice 

 cover, the under side will be subjected to an upward pressure owing to 

 the gradual expansion of the water as it is brought to the temperature of 

 freezing. But the orientation of the basal plates, parallel with the upper 



