﻿274 



DR. A. HOLMES ON THE TERTIARY 



[vol. lxxii, 



calcite and peridotite without necessarily implying the concomitant 

 formation of complementary alkali rocks. 



(c) As in Series A, the serial relationships between the rocks 

 of Series B point to a process of differentiation acting on a 

 basaltic magma. In this case, however, the magma must have been 

 well provided with water and over- saturated with silica. Given 

 a magma of the kind suggested, a process of differentiation towards 

 -andesite is not difficult to understand. By the sinking of some of 

 the crystals first formed — olivine, labradorite. and augite — the 

 residual magmas would have the composition appropriate to 

 pyroxene- and hornblende-andesites. In turn, the sinking of the 

 phenoeiysts of hornblende-andesite would leave a magma having 

 ihe composition of latite or rhyolite, according to the time during 

 which the process was able to continue. For rocks of this series 

 our knowledge of the kind of differentiation concerned in their 

 genesis has been greatly advanced by recent freezing experiments 

 made in the Carnegie Institution of Washington, and particularly 

 by those of Dr. N. L. Bowen. 1 Prof. Daly has summarized the 

 field evidence in favour of the generally accepted hypothesis that 

 andesite is an early differentiate from basaltic magma. 3 



The Mozambique lavas suggest that in the presence of water 

 not only silica is held in solution by the magma, but also that 

 calcium silicate is retained, its complete crystallization in plagio- 

 clase and, perhaps, in augite being deferred (relatively to increasing 

 silica) to a much later stage than in magmas containing carbon- 

 dioxide. The forms of the two lime-curves in the variation- 

 diagrams demonstrate the later crystallization of lime minerals in 

 Series B than in Series A, and the abundance of lime-zeolites in 

 the amygdaloids jDoints to water as the responsible agent. As 

 soon as means are devised for experimenting with artificial magmas 

 containing volatile fluxes under pressure, the suggestion here 

 advanced could be directly tested. It would only be necessary to 

 compare the course of crystallization in two melts of similar 

 composition, one charged with water, the other with carbon- 

 dioxide. 



(d) The question of the origin of the parent magma of Series B 

 is probably one that could be solved only with a complete know- 

 ledge of the preceding igneous history of Mozambique. There is 

 no reason to suppose that the excess of silica is due to the absorp- 

 tion of granite and gneiss by an ascending basaltic magma, for 

 the percentage of potash in the lavas is quite normal. Moreover, 

 as will be pointed out more particularly later, the heat-supply 

 can scarcely have been sufficient to allow a margin for the 

 occurrence of any appreciable assimilation. 



1 Am. Journ. Sci. ser. 4. vol. xxxix (1915) p. 175; 'The Crystallization of 

 Haplobasaltic, Haplodioritic, & Related Magmas ' Am. Journ. Sci. ser. 4, 

 vol. xl (1915) p. 161 ; and Journ. Geol. Chicago, Suppl. to vol. xxiii (1915). 



2 R. A. Daly, Journ. Geol. Chicago, vol. xvi (1908) p. 401 ; and ' Ig-neous 

 Rocks & their Origin ' 1914. p. 375. 



