FOYAITE-IJOLITE SERIES OF MAGNET COVE 653 



easily ascertainable as at Magnet Cove. The general conformity 

 of the two is very good evidence to show that, if plotted accord- 

 ing to their spacial {i. e., genetic) relations in the mass, the 

 analysis of the components of an igneous complex will furnish 

 regular curve. This fact is almost proof positive of the view 

 that the variation of rocks is due to differentiation of some 

 sort. 



Whether this differentiation is always as is now believed, 

 viz., that the oxides of Al, Na and K tend to segregate in one 

 direction, while those of Fe, Mg and Ca segregate in another, as 

 well as the process by which these changes are brought about, 

 are separate questions, which further investigation must settle. 

 It may be that the general course indicated by the diagrams of 

 Pirsson and myself are typical of all rock differentiation, or it 

 may be that with magmas of different character the course of 

 differentiation ma}' be radically different, and that the same 

 oxides do not always tend to go together. 



At any rate, it may be confidently expected that where a 

 mass of magma has been differentiated in situ and is of approxi- 

 mately regular shape, has not been subjected to secondary dis- 

 turbing conditions, and the exposures sufficient, we can express 

 the relations of the differentiation products and the course of 

 differentiation mathematically, as has been done in these two 

 instances. Of course, for this purpose, it is absolutely essential 

 that the analyzed specimens be representative, and that the 

 analyses be complete and accurate. Otherwise the curves will 

 be misleading or else uneven zig-zags, only rough approxima- 

 tions to the truth, and possibly not even that. There must also 

 be present at least three differentiates, as otherwise only straight 

 lines connecting the two can be drawn. 



The fact that the SiOg curve is the only one which is inflexed, 

 and that it runs very sharply up toward the acid end, leads to 

 some interesting conclusions. Since, toward the acid end, the 

 curves of (FeO), MgO and CaO drop much more rapidly than 

 those of AI3O3, Na^O and K3O, it is evident that at a short 

 distance to the right (in other words with a slight increase in 



