34 
MUSEUM BULLETIN NO. 30 . 
(2) The proportional relations of the different minerals in 
the various rock types show that these gravitative relations must 
have been effective after crystallization, not before. 
(3) Changes occurring in the composition of the separate 
minerals, in so far as these can be determined, follow the course 
prescribed by the theory. 
(4) Certain special textures and structures can be most 
easily and naturally explained by this theory. 
Spatial Relations. 
These have already been fully described, and it need only be 
added that the arrangement found follows the order of the 
specific gravity of the rocks, and hence indicates gravity as the 
controlling factor of differentiation. 
Proportional Relations of Minerals. 
One of the most striking characteristics of these rocks that 
first appears to the investigator is the remarkable uniformity of 
composition exhibited by the component minerals, which goes 
hand in hand with the widest variation in their relative propor- 
tions. This variation in the proportions of the constituent 
minerals (see page 35) can only be explained by some hypothesis 
which involves a purely mechanical means for their partial or 
complete separation from each other during or after crystalliza- 
tion, and makes the final location of the separated crystals 
largely fortuitous. The variations are too large, too rapid, too 
irregular and arbitrary to admit of any other explanation. The 
only theory of differentiation at hand which postulates such a 
mechanical separation is that emphasized by Bowen, that gravita- 
tive influence causes the sinking of crystals as they form and 
their consequent removal from their mother liquor. The dis- 
tance to which they sank would be purely fortuitous, depending 
on’ the rate of sinking and the length of time their movement 
continued, both of which factors are dependent on the change of 
viscosity of the magma during cooling. 
