Vol. 50.] OF CAKKOCK PELL. 325 



probably represent the upper and lower sides of tbe intrusion as 

 originally consolidated, but, whether this be so or not, the bilateral 

 symmetry proves that gravity has not been the determining factor. 

 This disposes, for the case under consideration, of the idea of a 

 fluid magma becoming richer — in its lower strata — in the denser 

 constituents ; and also of the notion of the earlier formed crystals of 

 iron ores, etc., sinking in the still fluid magma. If these processes 

 have operated at all, they have produced effects only quite sub- 

 ordinate to the general differentiation observed. Again, Vogt has 

 suggested that any inequality in the distribution of iron compounds 

 in a molten magma, once set up, might be augmented by magnetic 

 attraction. This idea is propounded apparently with reference to a 

 central rather than a marginal concentration of iron ores in a 

 magma, and moreover it could not account for the observed con- 

 centration of other constituents, such as phosphoric acid. In view 

 of the fact that natural magnetite loses its magnetic property com- 

 pletely when heated to 557° C, it does not seem likely that magnetic 

 attractions can play any part in the equilibrium of a molten rock- 

 magma, and I shall accordingly discard this suggestion. 



The only possible causes of differentiation that remain in the 

 case under consideration are those which depend on the difference 

 of temperature between the central and marginal parts of the magma 

 while still fluid or partly fluid ; and the concentration of the iron, 

 etc., towards what were the cooling surfaces of the mass seems to 

 point directly to the influence of this factor. In what way this 

 influence took effect is a question requiring some discussion. 



Most writers who have speculated on the mode of origin of a 

 heterogeneous rock-complex by differentiation of a magma originally 

 of uniform composition, have based their conception of the nature 

 of the magma on its analogy with an ordinary saline solution. 

 Lagorio l apparently considers that one or more definite silicate- 

 compounds (H 2 2 Si0 o , etc.), which he terms ' Normalglas' act 

 as solvent for all the other constituents. It is not easy to reconcile 

 this view with the existence of a very fairly constant order of 

 crystallization for the several minerals. For instance, however 

 little phosphoric acid and however much of the iron oxides a 

 magma contains, apatite seems to crystallize out invariably before 

 magnetite or iron-bearing silicates ; and, in general, the order of 

 crystallization depends little, if at all, upon the relative amounts of 

 the several constituents contained in the magma. As an alternative 

 to Lagorio's idea of a single general solvent, we might perhaps 

 imagine that a constituent on the point of crystallizing out is then 

 the dissolved substance, the remaining fluid magma as a whole being 

 the solvent. Some such idea seems to be intended by some authors 

 who have not very clearly defined their view of an igneous rock- 

 magma as a solution. 



Xow, it follows from the theory of osmotic pressure that if 

 different parts of a simple saline solution be at different temperatures, 

 the concentration must also vary, and equilibrium will be established 

 1 Tsdherm. Min. u. Petr. Mitth. vol. viii. (1887) pp. 507, 508. 



