77 6 THE JOURNAL OF GEOLOGY. 



Thus I have tried to show, that " Soret's principle " cannot be 

 applied to magmas, and consequently, if magmatic differentiation 

 were a process of molecular diffusion it could not be explained. 

 And it seems to me to be going too far to apply the laws of 

 dilute solutions to magmas before having attempted to consider 

 them simply as mixtures of liquids. 



As an illustration of the conduct of two liquids when mixed, 

 let us take aniline and water. If they are mixed at ordinary 

 temperature, when equilibrium is established two layers are 

 formed, one containing I per cent, of aniline and 99 water, the 

 other 98 aniline and 2 water. 1 But if they are mixed at ioo° the 

 two layers formed will contain 4 aniline and 96 water, and 91 

 aniline and 9 water ; at 150 the proportions are 14 aniline and 

 86 water, and 76 aniline and 24 water ; at 160 they are 25 aniline, 

 75 water, and 68 aniline, 32 water, and at 166 the two layers 

 should have the same composition, being consequently identical. 

 Therefore, above 166 aniline and water mix in all proportions, 

 but below this temperature the reciprocal dissolving capacity is 

 limited and generally a separation into two layers takes place, 

 the composition of which is a function of the temperature. 



This seems to be common for all liquid-mixtures where no 

 chemical action takes place. For all such mixtures there exists 

 a temperature, above which they mix in all proportions. It is 

 true that this temperature is known only for a few combinations 

 of liquids, but it must be regarded as certain that it exists, and if 

 not below then at the critical temperature, because here the 

 capacity of mixing in all proportions is a general property of the 

 gases. 



On the other hand, there are certain fluids, which at ordinary 

 temperature dissolve one another without limit, and for these the 

 temperature below which the dissolving capacity is limited is yet 

 to be determined, but in some cases this may not be reached 

 before the transition into the solid form takes place. For us the 

 principal question now is, can we assume that all the chemical 



1 The numbers given are obtained by interpolation in the curve of Alexejew in 

 Wiedemann's Annalen 28, table 3. (i£ 



