76 BRIDGMAN. 



suitable amount away from the equilibrium point. Of course no hard 

 and fast bounds can be set to this region within which nuclei are not 

 formed, because at any point the formation of a nucleus is a matter of 

 chance, but within the limits of time within which laboratory experi- 

 ments can be made it is safe to say that there is a region on both sides 

 •of the equilibrium point within which new nuclei will not be formed. 

 By working only within this region, the phenomena of surface growth 

 may be observed separated from the disturbing effect of nucleus forma- 

 tion. This is an essential distinction between a polymorphic transi- 

 tion and melting. ^Yith only a very few possible exceptions a solid 

 cannot be carried any distance into the domain of a liquid without the 

 formation of liquid nuclei, so that it is impossible to observe pure sur- 

 face growth of a liquid into a solid. 



The experimental conditions were such that the measurements of 

 this paper are almost entirely concerned with the surface growth and 

 not with formation of nuclei. It will be recalled that the method was 

 to increase the pressure on one phase until a nucleus of the new phase 

 appeared; and then to observe the rate of fall of pressure. All the 

 internal evidence makes it probable that in most cases only one 

 nucleus is formed, and that what is observed is the rate of advance of 

 a sino;le surface of separation. In support of this statement it is in 

 the first place evident that as soon as a single nucleus is formed the 

 progress of the reaction makes the formation of others much more 

 difficult, and since nucleus formation is a matter of chance it is very 

 unlikely that two nuclei are formed simultaneously. In the second 

 place, the regularity of the points, always lying on smooth curves, 

 makes it very probable that the conditions are always comparable; 

 if sometimes we were measuring the rate of growth of surfaces starting 

 from three nuclei and at other times from only one, we could not expect 

 smooth curves. This is a matter of considerable importance for our 

 interpretation of the results. ^Ye have noticed that the transition 

 acceleration with decreasing pressure is in practically all cases greater 

 than with rising pressure. Now this might be either because the 

 surface of separation at a constant distance from equilibrium moves 

 more rapidly with decreasing pressure, or because more nuclei of the 

 high pressure phase are formed than of the low pressure phase, so that 

 with decreasing pressure we are observing the rate of advance of more 

 surfaces than with increasing. This particular point may be settled 

 experimentally. If during a reaction with rising pressure we arti- 

 ficially raise the pressure into the region of the reverse transition, but 

 not into the region of the formation of nuclei, the ensuing fall of 



