FORMATION OF LEUCITE IN IGNEOUS ROCKS 263 
The first problem before us, then, is the determination of the 
limits of variation in composition of all norms in which leucite does 
or does not occur in accordance with the principles laid down, and in 
holocrystalline rocks derived from which modal leucite should or 
should not be found. This problem is capable of exact mathematical 
statement and graphic expression. 
As the simplest case we should consider first only the persalic, 
peralkalic, and perpotassic magmas—that is, those composed only of 
silica, alumina, and potash, the last two present in equal amount— 
so that the norms of these and holocrystalline rocks resulting from 
their consolidation would be composed either of quartz alone, quartz 
and orthoclase, orthoclase alone, orthoclase and leucite, or leucite 
alone. Such rocks are seldom met with, and their relations would 
be very simple. Furthermore, they form a limiting value of the next 
group to be discussed, so they will be dealt with later. 
As the next simplest case we may assume that soda is present also 
in variable amount, the magmas being persalic and peralkalic, but 
the relations of the alkalies varying from perpotassic to persodic. On 
our assumption that potash has a greater affinity for silica than has 
soda, albite and leucite could not be present together in the norms, 
but these and the holocrystalline rocks derived from such magmas 
would be composed of quartz, orthoclase, albite, leucite, and nephe- 
lite, either alone or in all possible combinations, with the exception 
just stated. The general chemical composition of such magmas will 
be represented by the following equations: 
asiO,+0Al,0,+cNa,0+dK,0 =100 
Al,O,= Na,O+ K,O 
The variables here are the absolute amounts or percentages of 
silica, soda, and potash, and the ratios of silica to soda, silica to potash, 
and soda to potash. Any two of these may be selected as a basis for 
plotting, and in my paper on the rocks of the Roman comagmatic 
region the percentage of silica was used for the abscissas, and the 
ti SiO, 
Tatlo K,O 
Plates I and II have been plotted on these bases, Plate III being the 
equivalent, thus plotted, of Plate I, as will be discussed later. 
for the ordinates. Indeed, all the data represented on 
