CRYSTALLIZATION IN IGNEOUS ROCKS 467 
indicate it for the average rock of any type in order that emphasis 
may be placed upon the importance of the effusive types as evidence 
of the order of beginning of crystallization and upon the distinction 
between order of beginning of crystallization and commonly stated 
“order of crystallization” (order of cessation of crystallization). 
CONSEQUENCE UPON THEORIES OF DIFFERENTIATION 
Variation in composition in different parts of a single rock 
body, or in closely associated separate bodies, is usually discussed 
in terms of order of crystallization of the constituent minerals. 
In nearly all cases the constant order of crystallization of Rosen- 
busch is assumed, whatever the composition of the magma treated. 
For example, the more femic upper portions of a granitic batholith 
that are so often observed have been explained as due to the early 
crystallization of ferromagnesian minerals at the cool contact and 
the continued diffusion in that direction of femic material, the 
result being an enrichment near the contact in ferromagnesian 
minerals. Apart from the question as to whether diffusion could 
take place and whether it would be better to appeal to convection 
as the mechanism of transfer, there is the fundamental question 
of the truth of the assumption that the ferromagnesian minerals 
crystallize early in granites. If it is recognized that quartz and 
feldspar are the minerals of early crystallization in granitic magma, 
the explanation is out of the question. Possibly early crystals 
of quartz and feldspar settled out of the contact portion. 
A discussion of differentiation is not, however, intended here, 
and the above suggestion is made merely to show the extent to 
which views on differentiation may be altered by a recognition 
of the distinction between order of beginning of crystallization 
and order of cessation of crystallization (commonly stated ‘‘order 
of crystallization’’). 
SUMMARY 
1. The criteria for the determination of “order of crystalliza- 
tion’”’ of rock minerals as applied in thin sections are discussed and 
it is shown that only order of cessation of crystallization can be 
so determined. 
