PETROLOGY: H. S. WASHINGTON 575 
when the amount of its correlate is high, and the converse, but not usually 
so when the rock magma is dominantly characterized by the abundance 
of another, non-correlated element, in which case another element, cor- 
relate with it, will appear as the co-dominant one. 
This phase of petrology, which is of a complexity probably much 
greater than is superficially apparent both as to character and cause, 
has, as yet, received comparatively little attention. J. H. L. Vogt,^ 
J. F. Kemp,2 L. de Launay,^ and W. F. Hillebrand^ have discussed the 
subject, but have confined their observations almost wholly to the dis- 
tribution of the elements according to the relative amount of silica 
(sihcity) shown by the igneous rocks. 
A few years ago I showed^ that these relations are far more complex 
than the simple one based on the silicity. Thus it is clearly 
obvious from a study of rocks and rock analyses that, among other re- 
lations, lithium, zirconium, the elements of the rare earths, glucinum 
and others, are most abundant in dominantly sodic magmas, irrespective 
of their sihcity; barium apparently in those dominantly potassic; 
titanium, vanadium, nickel and manganese in those high in iron; and 
chromium and platinum in those high in magnesium. It was also 
stated that ''soda not uncommonly tends to vary with the iron oxides, 
while potash shows similar relations to magnesia." 
The recent compilation of a second edition of A Collection of the Chemi- 
cal Analyses of Igneous Rocks, ^ embracing all those published between 
1883 and 1914 (numbering nearly 10,000), and soon to be published, 
has added greatly to the data available. This cumulative evidence 
now greatly strengthens, if it does not quite establish, the view that in 
igneous magmas potassium and magnesium on the one hand, and sodium 
and iron on the other, tend to vary together. 
The scope of this paper does not permit the presentation in full or in 
detail of the data in regard to this or other correlations which have 
been noted, but a short statement may be of interest, in advance of 
a more extended publication, as calling attention to a phase of chemical 
petrology which has been somewhat neglected, but which would seem 
to be of fundamental importance. 
The evidence is of two kinds; mineralogical and petrographical. 
Among the important mineral groups, found in igneous rocks, which 
contain the elements in question, the law enmiciated above holds good 
almost without exception. The sodic pyroxenes and amphiboles, such 
as aegirite, acmite, riebeckite, arfvedsonite, barkevikite, kaersutite, 
and basaltic hornblende, all show, along with practical absence of potas- 
sium, a strong preponderance of iron over magnesium, the latter usually 
