VELOCITY OF POLYMORPHIC CHANGES BETWEEN SOLIDS. 79 



seem capable of j^ivino; information about the mobility of the molecules 

 in a crystal, or the number that must fall together in the right position 

 in order to start a nucleus. It is probable that the formation of even a 

 nucleus is a rather complicated matter. If nothing more were de- 

 manded than that two or three molecules come together in the right 

 position, it is difficult to see why the region of nucleus formation 

 should have any boundaries. The process may be something like 

 this ; two or three molecules fall together to form the beginning of the 

 nucleus. These molecules by their orientation tend to attach other 

 molecules to them in the same orientation; this tendency is greater 

 at points farther removed from equilibrium. This process of aggluti- 

 nation until a full fledged nucleus is formed, is somewhat different 

 from the ordinary surface growth. In the early stages there are dis- 

 integrating forces due to the comparatively small number of elements 

 involved and the effects of surface tension, which vanish under the 

 conditions of surface growth proper. 



One of the most important results of these measurements is the 

 establishment by an extrapolation that there is a region of no appreci- 

 able velocity. It has been stated that there may be actually some 

 velocity in this region. The question arises, therefore, as to how 

 justified the extrapolation is which entirely neglects this very small 

 effect. The answer is that practically, except in one case, there was 

 never the slightest doubt. Theoretically the possibility must be 

 i-ecognized that the velocity curves may turn gradually at the bottom 

 and so vitiate the extrapolation, but practically, on diagrams of about 

 ten times the scale of the published figures, there was no such effect. 

 The only exception was KCIO3 at low temperature, and it has already 

 been stated that for this substance the formation of nuclei was ab- 

 normally persistent over a wide range, and this effect is entirely com- 

 petent to explain the character of the curves. So even in the case of 

 this one exception, we have no reason to believe that the small reaction 

 within the region of indifference would vitiate the extrapolation. 

 One is the more strengthened in the belief that the small residual 

 effect is due to a distinct mechanism when one considers that there 

 must be corners and edges as well as surfaces of separation of the two 

 phases, and that at corners and edges just this sort of an effect would 

 be expected because of the lack of perfect homogeneity. Furthermore, 

 observations of this small residual rate were made only after pressure 

 had been carried artificially into the indifferent region. Under these 

 conditions corners and edges must have been more numerous than if 

 the reaction had been allowed by its own progress to enter this region, 



