462 N. L. BOWEN 
glass. They may show phenocrysts of the ferromagnesian minerals 
and of lime-alkali feldspar but only in company with much alkali 
feldspar and quartz. Finally the accessories, instead of being 
almost universally present, commonly fail. 
The conclusion is that the order of beginning of crystallization 
in granitic magma is very different from the “‘order of crystalliza- 
tion’? commonly stated, that is, from the order of cessation of 
crystallization. Regarding the rhyolites as quenched granites the 
order of beginning indicated is: (quartz, alkali-feldspar); (lime- 
alkali feldspar, ferromagnesian minerals); accessories. (Alkali 
feldspar may often begin before the quartz, and ferromagnesian 
minerals may often begin before lime-alkali feldspar, depending 
upon their amounts.) If the course of crystallization is repre- 
sented in diagram the result will be somewhat as follows: 
t, Ce 
Granute Se ey aoe ee 
F 
more quartz ose. (a) © 
A 
t, ts 
SE 
r 
less gvartzose. (b) c 
P 
A 
FIG. 5 
The positions of the beginnings of the lines indicate the order of beginning of 
crystallization, and of the ends of the lines, order of cessation of crystallization. 
The vertical order of the lines themselves is chosen at random. 
It may be objected that the crystals which a rhyolite shows 
often form while the rhyolite is on its way to the surface, or even 
after it has reached the surface, and are therefore formed under 
different conditions of pressure and perhaps in the presence of less 
mineralizers than in the case of the granite, with the result that 
the order of crystallization in granites is reversed in rhyolites. 
The likely effect of such changes of conditions is a slight displace- 
ment of equilibrium conditions, not the drastic change implied 
