﻿part 3] 



VOLCANIC ROCKS OF MOZAMBIQUE. 



265 



series as ' alkaline,' for the high alkalies on the right are com- 

 pensated on the left hy high lime and magnesia. In each direction 

 the curves have "been continued beyond the extreme analyses bv 

 ^vhat seemed to be a natural extrapolation. At the mafic end, the 

 potash-curve meets the base at SiO., = 38 per cent. The com- 

 position of tins mafic limit of the series is that of a melilite-basalt. 

 At the felsic end the lime and magnesia curves meet the base at 

 SiO.,=70'6 per cent., and the corresponding composition of the 

 acid limit of differentiation is that of an alkali-rhyolite. Although 

 these two limiting rocks have not been found in Mozambicpie, 

 they both occur in quite similar series of rocks in British East 

 Africa, and are evidently natural members of the series, although 

 occurring only where differentiation has proceeded to its extreme 

 limits. 



In order to demonstrate the similarity of this series of Mozam- 

 bique rocks to those of other areas, variation-diagrams have been 

 made for the lavas of Abyssinia, British East Africa, Reunion, and 

 Teneriffe (fig. 8, pp. 268-69). Most of the analyses on which these 

 are based can be found in Iddings's ' Igneous Rocks,' 1 or in papers 

 to- which references are $here given, and in Dr. H. S. Washington's 

 well-known tables. 3 The curves for each constituent have a strong 

 family likeness throughout. The chief points of dissimilarity are 

 found in British East Africa, where, as Dr. Harker has already 

 pointed out, 3 alumina falls rapidly in the more siliceous rocks, and 

 iron-oxide (largely ferric) instead of gradually declining, compensates 

 for the low alumina by ascending. The magnesia-curve for the rocks 

 of British East Africa also differs from those of other localities in 

 its slow ascent towards a mafic limit. The discrepancy is, however, 

 due to subsidiary differentiation among the basic rocks, for melilite- 

 basalts and picrite-basalts occur in British East Africa, which (if 

 analysed) should provide a normal magnesia-curve. The resem- 

 blances of the curves for Abyssinia, Mozambique, and Teneriffe 

 are the most perfect, and indicate that similar processes must have 

 been at work on similar deep-seated materials in each of those 

 regions, these conditions being the controlling factors in the 

 development of a petrographic province or of a number of similar 

 provinces in widely-separated regions. Analyses of the alkali- 

 rocks of Madagascar, the Transvaal, Angola, and the Los Islands 

 were also plotted, but in none of these cases was there any resem- 

 blance to the Mozambique curves, probably on account of the 

 existence of more than one series of differentiates in those districts. 4 



Series B is based on analyses of rocks from the Monapo-River 

 and Sokoto-Hill districts. The analysis of the ground-mass of 

 hornblende-andesite is included, for it provides a good indication 

 of the course of differentiation towards the felsic end of the series. 



1 Vol. ii (1913) pp. 576-90. 



2 U.S. Geol. Sxirv. Prof. Paper 14 (1903). 



3 'Natural History of the Igneous Rocks' 1909, p. 124. 



4 Such, for example, as occur in the Christiania district ; see A. Harker, 

 op. cit. p. 123. 



