DIFFUSION IN SILICATE MELTS 309 
the most favorable case conceivable for the establishment of a 
composition gradient as a result of a temperature gradient according 
to the Soret principle. These conditions would evidently obtain 
either when the temperature of the magma was very much above 
the crystallization temperature, or when that of the surrounding 
rock was not very much below it, the latter being the more likely 
case. The magma intruded into hot surroundings, perhaps into 
a cognate intrusive not yet cooled, is, therefore, the most favorable 
6 
Fic. 6.—Curves of cooling of an intrusive igneous sheet 20 m. thick 
D [nerves 
subject for the working of the Soret action. Yet when we realize 
that the diffusivity of mass is, according to our determinations, 
from 10,000 to 100,000 times smaller than the diffusivity of tem- 
perature in rocks, it is apparent that the temperature of any 
igneous body will fall too rapidly to allow sufficient time for the 
Soret phenomenon to manifest itself. 
This statement may perhaps be more readily appreciated if 
the Soret action is stated more definitely as a diffusion problem. 
In order to do so we may assume that the osmotic pressure is 
