312 N. L. BOWEN 
there is no reason for the action continuing even for four years. 
It should be observed that, although the times have been computed 
for a body of a definite size, the solution is really of a general 
nature, for if the body were z times as thick, the time on the one- 
year curve would be changed to n? years and the time required 
for the Soret phenomenon would be 7? times as large. Nothing is 
gained, therefore, for the Soret effect by making the body larger 
or smaller. 
In assuming that the scale of Figure 6 represents absolute 
temperature, we have, of course, taken an impossible condition for 
any body of rock. This would mean that the surrounding rocks 
were initially at o° absolute and the magma had not yet begun to 
crystallize at 490° absolute=217°C. The assumption was made 
on account of the convenience of referring both concentration and 
temperature gradients to the same curve, but even if we make 
reasonable assumptions as to the temperature of the magma and 
of wall rock, our conclusion will not be affected. We may even 
assume that the Soret effect for silicates is many times that deduced 
‘from the theoretical (absolute temperature) relation, yet the 
outstanding fact remains that the time required to produce a 
significant amount of concentration of material by diffusion is 
enormously greater than that required for the mass to cool off. 
It may be noted that in speaking of a concentration toward 
the cool margin no mention has been made of what is concentrated. 
The reason is, of course, that it is not known. Ordinarily it is 
stated that the solute is concentrated toward the margin, but no 
distinction of solvent and solute can be applied to magmas. In 
conclusion, then, it may be stated that no concentration of any 
substance toward the cool margin could occur in appreciable. 
amount in the time available for such action in a cooling mass of 
completely molten rock. 
DIFFUSION TOWARD MARGIN DURING CRYSTALLIZATION 
There is, however, another case of diffusion of material toward 
the cool boundary for which we know the nature of the material 
that should move in that direction. Harker lays stress on the 
fact that in any cooling mass of magma there should be a time 
