A. Harker—“ Magmatice Concentration.” 547 
explanation of the concentration of iron-oxides, etc., in the cooler 
marginal part of an igneous rock-magma, but closer examination 
shows that the explanation is an inadequate one. ‘To take a single 
instance: in a large intrusion of gabbro investigated by Mr. Marr 
and myself, the amount of magnetite was found to rise from less 
than one per cent. in the centre of the mass to 25 per cent. at the 
margin. It is clearly inconceivable that the absolute temperature 
at the centre was ever twenty-five times that of the margin. Other 
equally strong examples might be cited. Soret’s principle is there- 
fore insufficient, and it is not difficult to suggest a reason for its 
failure. The principle presupposes a dilute solution: if we imagine 
the solution to be near the point of saturation, and crystallization 
to take place in the parts actually saturated concurrently with the 
concentration, we have to deal with an essentially different set of 
conditions. We must then fall back on the general law of the 
degradation of energy, as formulated, for instance, in Berthelot’s 
principle “Every chemical change accomplished without the 
intervention of any external energy tends to the production of that 
substance or set of substances which disengages the most heat.” 
Or we may conveniently adopt Becker’s extension of the law,? 
formulated thus: ‘“‘The transformations will be such as to evolve 
heat, light, etc., af'the highest possible rate.” 
Now the “heat of solution” of a salt in water is in some cases 
positive, in others negative; but we may fairly assume that ‘the 
crystallization of a mineral from an igneous rock-magma is in 
general attended by the evolution of a considerable amount of heat, 
perhaps something of the same order as the latent heat of fusion 
of the mineral by itself. Hence, in a magma near the point of 
saturation, whatever promotes crystallization will promote the most 
rapid evolution of heat. In an unequally heated magma this must 
be effected by the accumulation of the least soluble ingredients in 
the part most easily saturated ; that is, the concentration of the iron- 
oxide, etc., in the coolest region, and their crystallization there. It 
seems probable that the only practical limit to this action will be 
fixed by the growing viscosity of the magma, which, as follows 
from Vogt’s® results, will be increased by the abstraction of the 
iron-oxides and other basic ingredients. 
According to the view here offered, the Soret action, operating 
during the earlier stages, would bring about only a small part of 
the differentiation finally established. In consequence of it, and 
of the lower temperature of the reservoir towards the margin, the 
marginal portion of the magma would, as the whole cooled, be the 
first to become saturated, say with iron-oxides. At this point the 
further action to which I have drawn attention would come into 
play, and might produce much greater concentration of the iron 
in the outer part of the intrusion. 
1 Mécanique Chimique, vol. i. p. xxix. 
2 Amer. Journ. Science (8), vol. xxxi.p 191. 
3 Zeits. f. prakt. Geol. vol. i. p. 275 (1893). 
