FORMATION OF LEUCITE IN IGNEOUS ROCKS 365 
the affinity of other oxides for silica, as lime, magnesia, or ferrous 
oxide, assumed to be greater than that of potash; but these need not 
be discussed here. 
The general coincidence between fact and theory on the first as- 
sumption, and the grave discordance observed on the second, are clear 
proof that the former is in general the correct one, and that, while 
some exceptions do exist, in the vast majority of cases potash has a 
greater affinity for silica than soda has. Dismissing, therefore, any 
other assumption as to affinity for silica than the one with which we 
started, we may proceed to examine the exceptional cases mentioned 
above. These show clearly that our assumed order of affinity, while 
valid in most cases, is not a constant one, but that it can be super- 
seded by other factors, into the character of which we have now to 
inquire. 
Considering first the rocks with normative but not modal leucite, 
it is found that a large proportion of them are effusive rocks con- 
taining glass, their incomplete crystallization rendering their evi- 
dence inconclusive. It will be observed, however, that the great 
majority of these glassy rocks, while low in silica, are relatively high 
in alkalies, especially soda. ‘The list of analyses shows that they 
-are mostly nephelinites and nephelite-basalts in common parlance, 
all belonging to the lenic orders, and that most of them are dosodic. 
Their incomplete crystallization is in accord with this relatively 
alkalic character, as the researches of Lagorio and others show that 
such alkalic magmas are prone to undercooling and the formation 
of glass. From what we know of the order of crystallization which 
generally obtains in effusive rocks, and the usually late appearance of 
leucite, it is highly probable that leucite would have formed, in many 
cases at least, had the conditions permitted complete crystallization. 
We may therefore reasonably eliminate these from the list of excep- 
tions on the presumption that, had they been holocrystalline, they 
would have been modally leucitic and hence not exceptional. 
But again we find a very considerable number of perfectly holo- 
crystalline rocks with normative but not modal leucite, represented 
by uncolored circles. These, it will be observed, are apt to fall 
toward the lower part of the diagram; that is, they are poor in potash: 
and they lie for the most part well toward the right; that is, they 
