328 ME. ALEEED HAEKEE ON THE GABBEO [Aug. 1 894, 



at once under the perfectly general principle of the degradation of 

 energy. In whatever form the elements of a given mineral exist in 

 the fluid magma, it cannot he doubted that the crystallization of 

 the mineral from the magma involves in every case a very 

 considerable evolution of heat. Hence whatever promotes crystal- 

 lization in the magma will tend to the most rapid degradation of 

 energy. When crystallization has already begun in one region of 

 the magma, this result will be attained by a determination of that 

 constituent with which the fluid is most easily saturated to that 

 region of the magma which is already on the point of saturation. 

 The region of the magma which first becomes saturated with a 

 certain constituent will therefore, as crystallization proceeds, have 

 its saturation maintained by diffusion at the expense of the 

 rest of the magma. It is evident that, as a perfectly fluid magma 

 cools down, the point of saturation, say with apatite (or with 

 phosphoric acid), will be reached first in the margin of the body of 

 magma, that being the coolest region and also, if the Soret action 

 has already set up heterogeneity, the region of greatest concen- 

 tration of the substance in question in the fluid magma. Apatite 

 begins to crystallize out at the cooling surface of the magma, and 

 diffusion maintains the saturation and crystallization in this the 

 coolest region. Ilmenite and magnetite follow, and in turn the 

 fcrromagnesian silicates. But there will evidently be a tendency 

 towards restoring temperature-equilibrium between the margin and 

 the interior, and, what is more important, with falling temperature 

 and increasing acidity the residual magma becomes so viscous that 

 diffusion is more and more checked, and finally ceases. Thus the 

 concentration towards the cooling surface is strongest for the first- 

 formed minerals, and continually feehler for those which follow, 

 according to their order of succession. 



It seems, then, that the intimate relation between the phenomena 

 of concentration and of crystallization, which has been remarked by 

 several writers, leads to a simple explanation of the concentration 

 of certain constituents in the marginal parts of a rock-mass ; and 

 from this explanation the fact that the order of concentration of 

 the several constituents is also the order of their crystallization 

 follows as a necessary corollary. Further, there is here no narrow 

 limitation of the possible degree of concentration, such as that 

 which the law of osmotic pressure imposes upon the Soret action. 

 There is no difficulty, for instance, in admitting that a pure 

 aggregate of ilmenite may segregate from a rock-magma. According 

 to ' Soret's principle,' this would imply an infinite degree of con- 

 centration, corresponding to absolute zero of temperature ! 



There is, however, one consideration that must not be passed 

 over without notice. Backstrom ' asserts that " a silicate magma 

 during its period of crystallization is certainly too viscous to permit 

 of any considerable diffusion." It does not appear on what 

 evidence this statement rests. The petrographical features which 

 most clearly point to high viscosity are connected especially with 

 1 Journ. of Geol. vol. i. (1893) p. 773. 



