1891.] 



On the Plasticity of an Ice Crystal. 



339 



freezing, this bar should not have bent at all. If, however, the sliding 

 layers are necessarily perpendicular to the optic axis, this bar should 

 have been free to bend on the plane of the optic axis, but not in the 

 perpendicular plane. In the experiment the plane of the total bend 

 contained the optic axis. Thus the experiment was decisive. 



In attempting to discover the manner in which the rate of the 

 molecules sliding over each other depends on the driving force, we 

 are met by the difficulty that the rate of depression depends on at least 

 three other circumstances, the temperature, the previous history of 

 the bar, and the irregularity of the stresses and strains within the bar. 

 The second is to some extent avoided by only considering the rates 

 observed immediately before and immediately after the change of 

 weight. The third is probably not very important. In the following 

 table are collected all the instances which occurred, with the attendant 

 changes of temperature. The changes of rate are not so great as the 

 square, but greater than the first power of the changes of the applied 

 force. In the table may be seen the amount of correspondence with 

 the power f . The two most glaring discrepancies are in the second 



Table Y. 



and third instances given in the table, when the power 2 is well 

 satisfied. But these discrepancies may be largely, if not entirely, 

 explained by the great change of temperature. Without elevating 

 the statement to the rank of a law, we may say that fairly close 

 agreement with the observed facts is obtained by supposing that when 

 the molecules of ice slide on each other the cube of the friction varies 



the square of the velocity. 



In attempting to pass from the rate at which the centre of a loaded 

 sinks to the coefficient of plasticity, we meet with considerable 

 difficulties, and shall have to content ourselves with a rough approxi- 



B;ion. It might well be thought that the problem of a rectangular 

 ^ " 



