BY H. C. RICHARDS. 187 



the normality or otherwise of the alkaline and sub-alkaline 

 series of the rocks in question is shown, also the inter- 

 mediate nature of the third series is made quite evident. 



There is no question as to the reliability of the curves 

 on Barker's generalised variation diagrams, and they fur- 

 nish an excellent method of comparison. Harker's curves 

 are plotted from analyses of all igneous rocks whereas 

 we are concerned with volcanic rocks only, and although in 

 certain minor respects the curves for volcanic rocks would 

 differ from curves for plutonic rocks, the influence this 

 would have on these comparisons might be considered of 

 minor importance. In addition Harker's curves have not 

 been drawn up from water-free analyses recalculated to 

 100 per cent. 



There are many interesting features about the varia- 

 tion diagrams but the most outstanding thing is the 

 general paucity of alumina and the general richness of 

 iron-oxides for the rocks as a whole. Curves have been 

 drawn for lime, magnesia, soda, potash, combined soda and 

 potash, alumina, total iron-oxides as ferrous oxide, and 

 •combined alumina and total ferrous oxide. Well-shaped 

 curves with the respective rocks of each series closely con- 

 forming to its particular curve can be drawn for lime, 

 magnesia, soda, potash, combined soda and potash, and 

 combined alumina and total iron-oxide as ferrous oxide. 



In the case of alumina alone, and also of the total 

 iron-oxides as ferrous oxide, it is impossible to draw curves 

 of any value at all which have all the rocks conforming to 

 them ; but it is most noticeable that by combining these two 

 values and plotting them, excellent curves, with the rocks 

 closely conforming, can be drawn. An examination of 

 Plate VIII. shows that the alumina values are as a rule 

 markedlj^ low and the ferrous oxide values markedly high, 

 and, moreover, there is an antipathetic variation from the 

 normal curve of these values in the same rock. Normally, 

 curves for alumina and ferrous oxides are antipathetic, so 

 that one might expect to obtain reasonable curves for the 

 combined oxides. This expectation is very well fulfilled 

 {see Plate IX.), and the curves for the three rock series 

 are all very close and are closely comparable with the 



