46 WAVES OF SAND AND SNOW 



accumulation had occurred. This case is illus- 

 trated in Plate IV. On the lee side of the 

 peak the loose sand had been entirely removed, 

 the converging spirals of air having cut down to 

 the solid substratum, whereas the saddles of suc- 

 ceeding ridges were still connected by a bed of 

 loose sand. I did not ascertain at the time the 

 precise variation of condition which produced this 

 changed function of the eddies on the side of 

 peaks. I can, however, give an explanation which 

 is based upon subsequent observations. Wind 

 blowing over loose sand is sometimes overcharged, 

 and drops more than it picks up. It then silts. 

 Sometimes the amount which it is picking up is 

 equal to that which it is dropping, and it then' 

 neither silts nor scours. Thirdly, it is sometimes 

 picking up more than it drops, and then it scours. 

 In the first condition, and perhaps in the second, 

 a mound of sand would be deposited where the 

 two whirls of air coming from the right and left 

 meet under the peak. In the last case erosion 

 occurs. The result is, as is partly shown in the 

 photograph, to sweep bare a more or less oval 

 space between two succeeding peaks, whilst the 

 space on either side between the succeeding saddles 

 remains covered with loose sand. On one part 

 of the sandbank the general appearance was that 



