74 



and showed that by comparing the composition of the rock as a 

 whole with that of the matrix in which the larger crystals are 

 embedded it was possible to establish the following general law : 

 progressive crystallisation in an andesitic magma, consolidating 

 under the conditions which prevailed at Mount Pelee in 1902 and 

 in the Cheviot district in Old Red Sandstone times, produces a 

 mother-liquor richer in silica, poorer in lime, iron-oxides and 

 magnesia, and richer in soda and potash (especially potash), than 

 the original magma. He then pointed out that the Cheviot lavas 

 are traversed by dykes of quartz-felsite — a rock which agrees very 

 closely in composition with the matrix of the andesite — and that 

 this succession in time can be explained if we suppose that the 

 quartz-felsite represents the mother-liquor of the andesitic magma 

 squeezed out, or separated in any other way, from the crystals 

 after a certain amount of crystallisation had taken place in the 

 underground reservoir from which the earlier-formed lavas were 

 supplied. This view was confirmed by an examination of that 

 portion of the underground reservoir, consisting of granitic 

 rocks, which had been exposed by denudation in the central 

 mountain group. The theory of progressive crystallisation, or 

 crystallisation-differentiation, as it was now generally called, co- 

 ordinated a number of apparently disconnected facts and thereby 

 acquired a very high degree of probabiHty. He had formulated 

 the theory for the Cheviot district in 1885. (See pi. v. fig. 1.) 



Basalt was considered from the same point of view. Some 

 account was given of recent work in Hawaii, and especially at 

 the remarkable crater-lake of Kilauea. It was pointed out that 

 the dominant rock in Hawaii is an olivine-basalt in which olivine 

 was the first mineral to form ; and that by the sinking of the 

 heavier olivines in the mother-liquor many of the other and less 

 abundant varieties of rock occurring on the island could be 

 accounted for. But it was also pointed out that crystallisation 

 in basaltic magmas does not always follow the same course ; 

 that, in some cases, iron-oxides and magnesia, instead of separat- 

 ing out in olivines during the earlier stages, become concentrated 

 along with alkalies in the mother-liquor, and so may give rise, 

 on further differentiation, to soda-felsites and rocks, rich in 

 olivine and magnetite. (See pi. v. figs. 2-6.) 



The speaker then proceeded as follows : — 



In a basalt from Carnmoney Hill, near Belfast,, a 

 plagioclase felspar, not olivine, appears to have been 

 the first mineiral to form — I say appears because there 

 is at least a possibility that olivine may have been 

 formed and subsequently dissolved — then augite followed, and 

 lastly interstitial matter. The early felspars are not homoge- 

 neous as they frequently are in olivine-basalts. They are zoned 

 and the outer zones are richer in soda than the inner ones. 

 This proves that during the growth of the felspars the magma 

 was becoming richer in soda. At the same time it was evidently 



