308 N. L. BOWEN 
diffusivities obtained for the silicates are, therefore, much smaller 
than those of salts in solution and those of molten metals, but 
much greater than those of solid metals. Some of the higher 
values obtained for the silicates are comparable with those of 
certain relatively viscous organic liquids. The diffusivity constant 
of glycerine in propyl alcohol at 17°C. is, for example, approxi- 
mately o. 2. 
APPLICATION OF RESULTS 
It is not at all likely that the diffusivities of substances in 
mutual solution in rock magmas can be significantly greater than 
those determined for the plagioclase-diopside mixtures, and in 
many viscous magmas they would no doubt be considerably less. 
For the purpose of applying the results to diffusion problems in 
petrogenesis a value has been taken very close to the highest, viz., 
0.25, which is at the same time convenient in calculations. The 
Soret action is one of the diffusion phenomena that has been con- 
sidered of possible importance in magmas. It has been found in 
the laboratory that if a tube containing a solution is heated at one 
end and cooled at the other there is usually a concentration of the 
solute toward the cold end which depends upon the difference of 
temperature, the relative concentrations being, for many cases, 
inversely as the absolute temperatures. In cooling magmas the 
margin must be regarded as having a lower temperature than the 
interior, and there should presumably be a tendency toward a 
greater concentration of some substance or substances at the cooler 
margin. This introduces the possibility of composition differences 
in different parts of an entirely liquid magma, the differences being 
brought about by: diffusion. If cooled entirely by conduction, 
the temperature of a magma brought into contact with cold country 
rock should at the border quickly assume a value midway between 
that of the magma and that of the country rock. For a long 
period thereafter cooling at the margin is very slow (see Fig. 6).” 
We may imagine that the original temperatures of the magma and 
‘that of the surrounding rocks are such that during this long period 
of maintenance midway between them the magma is still above its 
temperature of beginning of crystallization. Here we would have 
t See also Lane, Aun. Rept. Geol. Surv. Michigan for rgo9, Fig. 18, p. 152. 
