THE P^VOLUTION OF PETROLOGICAL IDP:AS. 805 



deep might liave a density of 8. 28 while that from the bottom iniu-ht 

 have a density of 8.85; hut there is some d()ul)t as to whether the 

 constituents were ever nniforml\' mixed in the molten state, and if 

 not, whether sufficient time was allowed for diffusion to establish 

 homogeneity. It is certain, however, that they were unifonuly mixed 

 in the solid state, and the experiments are therefore of great interest; 

 for, if they do not prove differentiation in a molten mass they prove 

 tliat a uniform solid mass may l)ecome differentiated as it liquefies by 

 a kind of liquation process analogous to that which takes place in the 

 extraction of silver from copper. 



Professor Iddings has carefully considered the chemical composi- 

 tions of groups of rocks belonging to several different petrographical 

 provinces from the point of view of the differentiation hypothesis, 

 and has arrived at the conclusion that ""tlK^ simple oxide molecules 

 shift about independently of one another to a great extent." If this 

 conclusion be cori'ect, it is clear that the phenomena can not be 

 explained by the h^^pothesis of a differentiation solel}^ connected with 

 the formation of known minerals. But this view does not appear to 

 be accepted l)v Professor Brogger, who l)elieves "that the process of 

 differentiation must be referred to magmatic diffusion of definite 

 chemical compounds to and from the cooling surface; further, that 

 these diffusion phenomena in all probability stand in direct relation 

 to the order of crystallization of minerals in the corresponding magma, 

 and, lastly, that the order of crystallization, the nature of the dif- 

 ferentiation, and the sequence of eruptions are all closely related 

 phenomena.*" 



Differentiation dependent upon crystallization rests on a somewhat 

 firmer basis, and it was this kind of differentiation that first attracted 

 my attention. Mr. Clough, while mapping the Cheviot district, proved 

 that the widespread series of andesitic lavas is cut by a numbei' of 

 quartz-felsite dikes. Why did ([uartz-felsite succeed andesit(^ in the 

 Cheviot district? This was the question which kept contiimally recur- 

 ring to me during my examination of the rocks of the district. Now. 

 a microscopic examination of the andesites pro\'ed that the })henocrysts 

 taken together nuist have the composition of a basic rock, for they 

 were composed of lal)radorite, augite, and hypersthene, and therefore 

 the glassy l)ase present in some of the andesites nmst be allicni to 

 (luartz-felsite in composition. The se([uence esta])iished by Mr. Clough 

 could therefore ]>e explained l>v the assum])tion that t\\o (|iiai'tz-felsite 

 magma represented the mother li([U()r of the andesitic magma, after 

 tlie phenocrysts had separated out. Thus, if crystallization had ])ro- 

 gressed in the phitonic mass to tiie stage repr(\sented by the ])heno- 

 crysts of the la\a, or a bttle further, and the mother li(|uoi- had tluMi 

 been scpieezed out as one s(|uee/es watei' out of a sponge, or s<'[)ai'ate(l 

 in any other way, and I'oi-eed ui)ward into ei'acks in the o\(Mlyuig 

 series of andesitic lava flows, the tiuestion above i-el'erred to could be 

 SM 1902 20 



