118 WAVES OF SAND AND SNOW 



upon the formation of dew among desert sand- 

 dunes (see ante, p. 69), for I suppose the cement- 

 ing matter of the hard-set snow to be, so to 

 speak, soHd dew. 



Below the icy crust the snow in the snow-waves, 

 though not hard, was dense, having a specific 

 gravity of 0-4, which was twice that of the snow 

 in a copse near by, which was 02. At Glacier 

 House, in the Selkirk Mountains, where the atmo- 

 sphere is calm and there is no drifting, the density 

 of the snow at the surface was 01, at a 

 depth of I foot 02, and at a depth of 4 feet 

 0"354. Increased density of snow is due to 

 removal of air from' the interstices, and the above 

 figures show, not only that the minute snowflakes 

 which fall in the dry climate of Manitoba in mid- 

 winter pack tighter than the large flakes which fall 

 in the Selkirk Mountains, but also that the wind, 

 in turning over the material, gives it an additional 

 compactness greater than that imparted by the 

 pressure of a superincumbent layer of about 

 3 feet in thickness. 



Ripples in Granular Snow or " Snow-sartd.'' 



The loose particles of waves of fresh -fallen dry 

 snow remain unrippled, the surface of these waves 



