1912] CLARKE— SOME GEOCHEMICAL STATISTICS. 



219 



the essential identities unobscured. Some human selections are 

 indicated in certain regions ; for example, in Norway, the high soda 

 is due to the disproportionate attention paid to the nepheline sye- 

 nites ; while in Italy the leucite rocks of Vesuvius lead to an apparent 

 excess of potash. In the mean for all Europe these differences 

 balance each other. 



The general agreement between the several averages is well 

 brought out by means of the method developed in the quantitative 

 classification of igneous rocks. For this comparison, Professor 

 Iddings has kindly made the necessary computations, with the 

 subjoined results: Duplications of averages are of course avoided. 



Average. 



Roth 



Harker 



Washington 



Loevvinson-Lessing 



Survey 



Atlantic Coast. . . 

 Yellowstone Park . 



Colorado 



California 



North Europe .... 

 Central Europe . . . 



Italy 



N. America 



S. America 



All Europe 



General mean .... 



Column. 



A 



B, K 



C 



D 



E 



F 



G 



H 



I 



L 



M 



N 



J 



P 



O 



Q 



Magmatic 

 Symbol. 



Na 



Adamellose. 



Harzose. 



Monzonose. 



Harzose. 



Tonalose. 



Andose. 



Tonalose. 



Adamellose. 



Tonalose. 



Akerose. 



Andose. 



Monzonose. 



Tonalose. 



Andose. 



Tonalose 



Tonalose. 



In brief, all the averages fall in class II., dosalane, and in orders 

 4 and 5, but near the boundary between the two. The principal 

 variations are in the rangs and sub-rangs, and are mainly due to the 

 varying proportions of the alkalies. The average rock may be 

 classed as tonalose, or, to use a more familiar term, as andesite. It 

 is evident that in order to have any statistical validity, such aver- 

 ages as are given here must represent fairly large areas, and a con- 

 siderable number of analyses. Small areas, especially those of vol- 

 canic islands, may vary widely from the mean ; for example, analyses 

 of 32 Hawaiian rocks gave in the average, only 48.55 per cent, of 

 silica. Such local variations, however, could hardly exert any note- 

 worthy influence upon the continental averages which represent 



