736 JOHN JOHNSTON 



per cent,^ or about 0.04 cc. per gram. On this basis^ the rise of 

 melting-point of a pure sihcate produced by a pressure of i ,000 atm. 

 would range up to 1 5°, with a possibility of a somewhat greater rise 

 in exceptional cases (specifically, whenever the latent heat is small, 

 unless the change of volume is at the same time very small). 

 Accordingly, the change of melting-point would be of the order of 

 6° for the first mile in depth, an amount which, in all probability, 

 would with each succeeding mile decrease steadily at a steadily 

 decreasing rate, but would not become zero. 



Now the question arises — could the temperature gradient down- 

 ward overtake the pressure gradient? It would seem that this 

 possibility is not to be excluded, for, according to the best evidence 

 (likewise very scanty), the increase in temperature for one mile in 

 depth is about 40° C. ; consequently, if we assume that this gradient 

 and the melting-curve gradient remain constant, the temperature 

 would at a depth of 35 miles be 1,400° C, the approximate tempera- 

 ture at which many pure silicates exposed to the pressure existing 

 at that depth would melt. 



The foregoing reasoning applies only to a pure silicate melting 

 to a liquid of its own composition; but to reason from the behavior 

 of a single pure substance to its behavior in contact with a complex 

 solution is of doubtful utility. For in the latter case the relative 

 solubility of the substance in the solution under the particular con- 

 ditions obtaining is the decisive factor which determines whether 

 it shall separate, though of course it will not separate unless it 

 would normally be solid (crystalline) at the prevailing temperature 

 and pressure; and, so far as experience goes, pressure has very 

 little influence on solubility in condensed systems (that is, systems 

 without a vapor phase). The foregoing serves to emphasize the 

 fact that any conclusions as to what will happen to rocks at high 

 temperatures and pressures must still be considered to be subject 

 to very serious limitations. 



Reversible transformation points solid ^ solid (that is, enantio- 

 tropic inversions) are altogether analogous to melting-points, and 



' See Day, Sosman, and Hostetter, Am. Jour. Set., XXXVII (1914), i; cf. the 

 results cited by R. A. Daly, Am. Jour. Sci., XV (1903), 276. 



41.3AF 41. 3X100 



