﻿264 



DK. A. HOLMES ON THE TERTIARY 



[vol. lxxii, 



of exploration renders possible. The ' amygdaloidal basalts' of the 

 second column are those in which siliceous amygdales are abundant. 

 Normal basalts are not tabulated, for they appear, in various degrees 

 of abundance, to be common to both series. The table brings out 

 the fact that the association in Mozambique of two contrasted 

 series of lavas is not exceptional, although it is the southernmost 

 locality in Eastern Africa in which that association has yet been 

 noted. The weighty problems that centre about the origin of 

 the Mozambicpie lavas thus become common to all the African 

 occurrences on the north. 



The alkali series of Mozambicpie can be closely matched by the 

 lavas of Eeunion, and by those of the Mid-Atlantic islands. 1 

 In Angola, near Cambambe, traclrytes containing katoforite and 

 cossyrite are associated with augite-andesite. 3 In none of these 

 localities are amygdaloidal basalts like those of Mozambicpie known 

 to occur. 



XIX. Variation-Diagrams. 



In the preceding sections, ten analyses have been recorded. 

 These analyses fall naturally into two series (belonging respectively 

 to Groups A & B) characterized in each case by a well-marked 

 serial relationship. The variation-diagram adopted for a graphical 

 comparison of the analyses is that in which silica percentages are 

 taken as abscissa?, and the other oxides plotted as ordinates. 3 In 

 order to render the comparison as faithful as possible, the total of 

 the two iron-oxides is expressed as ferrous oxide, and to this 

 manganous oxide is added. The analyses are then recalculated to 100, 

 water and carbon dioxide being omitted (figs. 6 & 7, pp. 266-67). 1 



It will be noticed that the two series of curves are distinctly 

 different, and that no analysis belonging to one series would fit 

 the curves connecting the other series. 



Series A is based on analyses of the rocks from the Sanhuti 

 district. The curves were drawn before the tephritic pumice of 

 the Monapo was analysed, but when the analysis was made it fell 

 naturally into its place between those of basalt and phonolite, 

 and very little modification of the curves was found necessary. 

 This seems to indicate that the lava in cuiestion belongs to the 

 same series as the Sanhuti rocks, and that it is not immediately 

 related to the amygdaloids of the district in which it was found. 



The series is characterized throughout by a high percentage of 

 iron-oxides, and by an excess of soda over potash with a marked 

 parallelism of the two alkali curves. Lime and magnesia occupy 

 a low position on the right (that is, beyond 55 per cent, of silica), 

 but rise rapidly on the left, in antipathy to the alkalies which fall 

 towards the left. It is, therefore, difficult to regard the whole 



1 G. T. Prior, Min. Mag. vol. xiii (1903) p. 254. 



2 A. Holmes & H. F. Harwood, Min. Mag-, vol. xviii (1916) p. 58. 



3 A. Harker, 'Natural History of the Igneous Rocks' 1909, p. 11S. 



4 To avoid confusion, the values for titanium dioxide are not plotted. 



