POLYMORPHISM AT HIGH PRESSURES. 179 



polymorphism forms at least two different kinds of molecules, and 

 tliat these two kinds of molecules are present in both phases, but in 

 different proportions. Furthermore, the ratio between the two kinds 

 of molecules in a single phase varies continuously with temperature. 

 It must be frankly recognized that this theory has succeeded in cor- 

 relating a formidable array of chemical facts, many of them new, but 

 nevertheless it seems to me that the picture it presents of the mech- 

 anism of a crystal involves considerable physical difficulties. It is 

 hard to see how components of varying proportion can be arranged 

 on definite space lattices, as w^e know they are. The modern concep- 

 tion of the crystal is one for which the molecule has lost its signifi- 

 cance. We know that when we build up a crystal from its elements 

 out of solution or the melt, these elements are added as entu'e mole- 

 cules; or when a crystal is taken apart, as by melting, the crystal comes 

 apart in molecules. Inside the crystal, however, the molecular bonds 

 lose their individuality and fuse together. The molecular bond makes 

 its appearance only when we try to remove a part from the crystal. 

 Certainly as a thermodynamic entity, concerned with specific heats, 

 the molecule has little significance for crystals. 



It is significant that none of Smits' results were obtained with single 

 crystals, but with aggregates of small crystals. Between the crystal 

 grains there must be transition layers more or less amorphous in char- 

 acter, and the idea of association may be applicable within these 

 layers. It is not unreasonable to suppose that the phenomena which 

 Smits finds are consistent with the idea of a varying association are 

 connected with the transition layers. At any rate it seems to me 

 that at present we should withhold acceptance of Smits' theory, in 

 spite of the chemical facts on its side, until he has shown by actual 

 X-ray photographs that in an individual crystal, of Hglo for example, 

 there is the continuously varying constitution demanded by his theory. 



This discussion is to be guided by the same idea as that underlying 

 a pre\ious discussion of the thermodynamic behavior of liquids under 

 high pressures. It turned out that the experimental facts were of an 

 unexpected complication, a complication so great that previous pic- 

 tures of the atoms as smooth spheres were powerless to provide a 

 sufficient range of possibilities. It was the purpose of that discussion 

 to show that if we made the next degree of refinement in our repre- 

 sentation of the atoms and considered them as having characteristic 

 shapes, that we had thereby opened up at least the possibility of 

 explaining all the effects. The purpose of this discussion is the same; 

 to show that if we go to the next stage of refinement and consider 



