LATER STAGES OF EVOLUTION OF IGNEOUS ROCKS 5 



mate sets rather an upper limit which is probably many times the 

 actual effect, for nearly all experimental investigations of the Soret 

 phenomenon agree in placing the actual effect at much less than 

 the amount calculated on the foregoing assumption. 



The establishment of this fact should not be surprising, for the 

 Soret effect does not, in reality, depend on a tendency of the solute 

 alone to become more concentrated in the cold part. There is a 

 tendency toward a greater concentration there of both solute and 

 solvent, which is of course impossible, and if both obeyed the laws 

 of ideal solutions, there could be no relative concentration of the 

 one with respect to the other/ It is because the behavior of 

 solute and solvent departs from the laws of ideal solutions, the 

 one more than the other, that the Soret phenomenon is observed. 

 The amount of the effect obtained as a result of this difference in 

 degree of departure from the laws mentioned must necessarily be 

 less than that calculated from the osmotic-pressure relation, which 

 really involves the assumption that one substance obeys these laws 

 and the other substance may be neglected, that is, is entirely unaf- 

 fected by change of temperature. 



Both theoretical considerations and experimental results indi- 

 cate, therefore, that the actual Soret effect would be even less im- 

 portant than Harker's calculation would indicate — -in short, quite 

 negligible. 



GRAVITATIVE EFFECT IN A LIQUID MAGMA 



Of the same order of magnitude is the probable effect of the 

 accumulation of the denser components of the magma solution in 

 its lower portions, resulting in a composition gradient or density 

 stratification in the liquid magma. An experiment made by 

 Morozewicz has sometimes been cited as a proof that the extent to 

 which this process may take place even in a crucible is very con- 

 siderable, but a consideration of the conditions of the experiment 

 shows that this conclusion is unwarranted.^ In the experiment 

 referred to, a sample of granite was fused or rather partly fused and, 

 after cooling, analyses were made of the glass from the top and from 



^ See Backstrom, Jour. GeoL, I (1893), 774-75. 



2 J. Morozewicz, Tscherm. Min. Pet. Mitth., XVIII (1898), 232. 



