VELOCITY OF POLYMORPHIC CHANGES BETWEEN SOLIDS. 75 



pounded dry into a thin steel shell about inches long and 11/16 of an 

 inch in diameter, open at both ends, and perforated on the sides with 

 small holes to allow ready access of the liquid transmitting pressure 

 throughout the entire mass. Several measurements were made to 

 detect a possible restraining action of this shell, and in nearly every 

 case none could be detected, but the fact that some slight effect could 

 be detected in one or two extreme cases where the lateral holes were 

 discarded and the shell left closed on the bottom end leaves open the 

 possibility of a very slight effect in all cases. The state of subdivision 

 of the substance must also have an effect. If it is in the shape of a fine 

 powder compacted together by the hammer, the new phase must 

 spread somewhat less rapidly from grain to grain than it would in a 

 homogeneous fused mass. Some of the substances were fused into 

 place, but even these always developed cracks during use, thus intro- 

 ducing the same element of uncertainty. Furthermore, there may be 

 a slight specific effect exerted by the liquid transmitting pressure. 

 This liquid penetrates through all the crevices, and might even in 

 some cases exert a catalytic effect, although this extreme is unlikely. 

 Usually pressure was transmitted by kerosene, but mercury was used 

 for most organic substances. In the case of KCIO3 a greater reaction 

 acceleration was found with mercury for the medium than with kero- 

 sene. This may be in part a thermal conduction effect. It is also 

 likely that in some cases slight impurity may affect the transition 

 velocity, although its effect on other properties might be inappreciable. 



These considerations suggest the use to which these observations 

 can legitimately be put. They must not be used in calculating the 

 absolute rate of growth of any one phase at the expense of another, 

 they may be used cautiously and wdth great reserve in comparing the 

 ^'elocity under similar conditions of different materials, since the form 

 of apparatus was nearly constant for all these measurements, and they 

 may be advantageously employed in finding the variation of rate of 

 the same substance under different conditions of pressure and tempera- 

 ture and direction of transition. This last is the use to be made of 

 them in this paper. 



We now have to inquire just what it is that we have been measuring. 

 It is common knowledge that there are two distinct processes involved 

 in either crystallization from a melt or in the formation of new poly- 

 morphic phases. These are: first the formation of nuclei of the new 

 phase, and secondly a surface growth of the new phase, starting from 

 the nuclei. In the case of two solids, nuclei of the new phase are 

 formed only when temperature or pressure is raised or lowered a 



