54 SECTIONAL ADDRESSES. 



Now, the course of differentiation of a complex silicate melt, as suggested 

 by the relative freezing points of the various constituents, would be the 

 separation by crystallization of the more basic constituents from a residuum 

 that became progressively more acid as crystallization proceeded towards 

 complete solidification. If by any chance, as some have held, crystalliza- 

 tion of early phases in the upper portions of a magma reservoir led to a 

 settling and remelting of these constituents in the lower and hotter regions, 

 it is conceivable that an exaggerated diffusion-column might be produced, 

 with denser basic material below, and lighter acid material above. The 

 effect, however, of a crustal mass subsiding into such a differentiated but 

 wholly fluid magma would be to produce intrusions of a progressively basic 

 character. The upper acid portion would be intruded first, and the more 

 basic subsequently ; also, it may be inferred that none of these intrusions 

 would be true to type, but that they would exhibit excessive variation 

 on solidification. 



Such conditions would give rise to an igneous cycle that, so far as the 

 plutonic rocks are concerned, is exactly the reverse of what is encountered 

 in nature. In all the Tertiary plutonic centres it has been clearly estab- 

 lished that the order of intrusion of the plutonic rocks is invariably 

 from basic, or even ultra-basic, to acid. 



If we agree that differentiation is due to the separation and settling of 

 crystals and culminates in complete solidification, any intrusion drawn 

 from a basin other than the residual liquid fraction of the period must be 

 consequent upon remelting. I agree with Dr. Harker in regarding the 

 process of remelting of the already differentiated magma as the only means 

 whereby the basic to acid order of intrusion could be induced or maintained. 



It would appear that a local intercrustal reservoir established as an 

 upward extension of the primary magma, to take the place of a centrally 

 subsided crustal mass, became more or less completely differentiated before 

 it made its presence felt at higher crustal levels. The fact that the explosion 

 breccias which heralded the plutonic intrusions are associated with an acid 

 magma, or rather an acid differentiate of a basic magma, shows that 

 differentiation within the basin had almost run its course. 



It will, however, be readily understood that, without remelting, pressure 

 produced by a crustal block subsiding into a magma reservoir could only 

 cause to be exuded such a portion of the differentiated magma as still 

 retained its fluid state, and in no case could this fluid be as basic as the 

 original magma. That an acid differentiate may collect in a liquid form 

 in the upper portion of a magma column or reservoir when the rest is solid 

 has been argued by Bowen, and it is certainly probable that many acid 

 intrusions originated in this way without the intervention of remelting. 

 In this connection we may cite the case of the quartz-dolerite magma, 

 itself a product of differentiation. 



The quartz-gabbros and quartz-dolerites form some of the ring-dykes 

 and late basic cone-sheets of Mull, and some of the older ring-dykes and 

 most of the cone-sheets of Ardnamurchan. These rocks were derived 

 from a magma which as crystallization progressed clearly gave rise to an 

 acid differentiate. This acid partial magma was of strikingly different com- 

 position from the early crystalline phases, and its temperature of complete 

 consolidation was evidently far below that at which the larger and earlier 



