FOYAITE-IJOLITE SERIES OF MAGNET COVE 647 



Comparison of the two shows that the curves, or rather zig- 

 zags, are much alike even though the analyses of (and con- 

 sequently the abscissal positions of the figures for) arkite are 

 different in the two. At the same time the new diagram is 

 markedly more broken, and varies less regularly and con- 

 tinuously than the former. The regularly serial character of 

 the first is thus apparently diminished, and what it was thought 

 would be an excellent example of regular differentiation seems 

 to turn out rather the contrary. 



But, as Pirsson justly remarks, the use of SiOg for the 

 abscissae is arbitrary, and, since this is one of the most impor- 

 tant rock ingredients, its variation should also be shown in a 

 manner directly comparable with those of the other oxides. 



It would seem to be undeniable that this is a lesritimate. 

 indeed a most logical, method if the differentiated mass has not 

 suffered disturbance and if circumstances permit the determi- 

 nation of the correct distances of the various differentiates from 

 the center, since the diagram then represents, not only the com- 

 positions of the various phases, but their actual relations in 

 space, both as to direction and as to relative position. 



It often happens, as apparently at Magnet Cove, that the 

 successive differentiation products are sharply separated from 

 each other, transition forms being either lacking or very small 

 in amount as compared with the main types. To correspond 

 then with the actual state of affairs, the diagram should consist 

 of steps, /. e., horizontal lines of a length equal to the breadth of 

 each zone, at the respective ordinal positions for each oxide. 



Since, however, the analyses may be assumed to represent the 

 average composition of each differentiation product, and we 

 desire to study the course and the laws of differentiation, it is 

 legitimate to represent the position of each constituent by a 

 point, and the lines connecting these will therefore express the 

 course of the differentiation of the mass of magma, even though, 

 as an actual matter of fact, all the possible gradations repre- 

 sented by the curves may not be present. Such a procedure is 

 quite in accordance with the general practice in chemical and 



