Physical and Chemical Behavior of Solids. 251 



process. Indeed there is evidence in the case of the metals 

 which indicates that complete recrystallization does not occur ; 

 for the density of a piece of metal which has been deformed 

 is always less than that of the same sample after annealing,* 

 which is precisely what we should expect if the metal after 

 deformation contains some small fraction in the form of sub- 

 cooled liquid. 



Summary and Conclusion. 



In all discussions of the influence of compression on solid 

 systems it is essential that we bear in mind the distinction 

 between the effects of uniform pressure and of non-uniform 

 pressure ; disregard of this distinction is responsible for many 

 of the apparently contradictory statements to be found in the 

 literature pertaining to the subject. 



Uniform pressure has a comparatively slight effect on the 

 melting point : it usually raises it, and by an amount which, in 

 the systems hitherto investigated, is seldom greater than 10°, 

 and never greater than 30°, per 1000 atmospheres. Its effect 

 on solubility is slight, and for practical purposes negligible as 

 compared with the influence of temperature upon solubility. 

 Uniform pressure tends to further those reactions which are 

 accompanied by a decrease of volume ; but it by no means 

 follows that it will cause these (or other) reactions to occur ; 

 for whether a reaction takes place or not is determined 

 primarily by its velocity under the particular conditions, and 

 such evidence as there is tends to show that reaction velocity 

 is not much affected by uniform pressure. 



The effects of non-uniform pressure greatly outweigh those 

 of uniform pressure. It alivays lowers the melting point and 

 raises the solubility, and by amounts which are many times 

 greater than the corresponding changes with uniform pressure. 

 Indeed if we make the plausible assumption that permanent 

 deformation of a crystalline aggregate is conditioned by a real 

 local melting (of those parts, which at any moment bear the 

 brunt of the load), we find the amount of pressure required to 

 cause melting at ordinary temperature to be well within the 

 bounds of probability. Such we believe to be the efficient 

 cause in producing most of the phenomena recorded as occur- 

 ring when solid systems were submitted to compression. 



From a discussion of the results from this standpoint the 

 following conclusions emerge : 



(1) That the amount of compression required to cause 

 crystalline solids to "flow" or to weld together is dependent 

 mainly on the melting temperature, heat of melting and 



* See Johnston and Adams, J. Am. Chem. Soc, xxxiv, 563, 1912, or Zs. 

 anorg. Chem., lxxvi, 274, 1912. 



