DISCUSSION OF CAUSES OF DIFFERENTIATION 409 



separately, in so complicated a matter as the differentiation of a mass 

 of magma. Our notions of the relations of solvent and solute, and of 

 solutions in general, especially as applied to rock magmas and the con- 

 ditions in which they exist, are at present so vague that we must move 

 with great caution. The matter is complicated by the fact that in such 

 large masses purely physical, as well as physico-chemical and chemical, 

 forces (such as gravit}' and convection currents), almost certainly come 

 into play. This being the case, it would perhaps be wiser to defer all 

 discussion for the present, but I can not refrain from suggesting an ex- 

 planation which has certain elements of probability. 



A liquid solution, such as a rock magma may be supposed to be, is a 

 mixture of two or more bodies, one of which is said to act as a solvent 

 for the others — that is, the solutes. The difference between the two is 

 more apparent than real, since in many cases their functions can be in- 

 terchanged. But the general idea underlying the use of the term solvent 

 is that it is that constituent which is present in excess.* Often either 

 constituent may act as the solvent, and in many such cases it is found 

 that as the solution is cooled socalled eutectic mixtures tend to form.f 



Leaving aside the consideration of these, if a more or less dilute solu- 

 tion (that is, one in which the solvent is largely in excess) is cooled suffi- 

 ciently, or frozen, as it is called, the solvent crystallizes first, the mixture 

 tending to become eutectic. Thus, if we partially freeze an aqueous solu- 

 tion of salt the ice formed is quite pure and the brine more concentrated. 



This idea has been applied to the ciystallization of magmas by Lago- 

 rio,;!; who also suggests § that the general solvent has the composition 

 (K,Na) 2 0.2Si0 2 , basing this conclusion on the results of many analyses. 

 While the exact stoichiometrical character of the solvent in general may 

 be doubted, and while many cases may be cited where such a solvent is 

 out of the question, 1 1 yet in the alkaline rocks, especially those of the 

 foyaitic series, it seems probable that the part of the magma which plays 

 the role of the solvent is composed of silica, alumina, and alkalies. Pos- 

 sibly these exist in stoichiometrical ratios, but more probably not. 



We would seem to have, then, in this a good explanation of the facts 

 at Magnet Cove, Umptek, and Ramnas. Just as in a highly cooled ves- 

 sel of salt water the ice crystallizes at the sides, bottom, and top, leaving 

 a core of more concentrated liquid at the center, so here the solvent may 

 have frozen out, collecting at the borders of the cavity in a more or less 



* Cf. Nernst. Theoretical Chemistry, 1895, p. 414. 



f For a resume of this subject compare Teall. British Petrography, 1S88, p. 39G. 

 | Lagorio : Tsch. Min. Pet. Mitth., vol. viii, 1887, p. 513. 

 § Lagorio : Op. cit., p. 508. 



|| On these points see the remarks of Zirkel (Lehrb. Pet., vol. i, 1893, p. 767) and Iddings (Bull. 

 Phil. Soc. Wash., vol. xii, 1892, p. 156). 



