INTERNAL STRUCTURES OF IGNEOUS ROCKS 453 
its main support when it develops (as in Duluth) that the floor and 
roof are rocks of about the same composition as the average gabbro, 
and that the bands range from anorthosite to peridotite—with 
compositions that could hardly be synthesized from any rocks in the 
region. It is admitted that schlieren, developed from xenoliths, 
occur in some local spots, but they have no relation to the banding 
and show no extreme in composition. 
Deformation during crystallization might explain the orienta- 
tion of grains, but cannot clearly explain the banding. 
The process of differentiation has not been described so as to 
explain the banding. It is, of course, probable that a rhythmic 
variation in the process of crystallization would give a rhythmic 
alternation of rock deposited, but that furnishes no explanation of 
the fluxion structure. None of the theories of differentiation out- 
line a process that will result in a combination of gravitative arrange- 
ment, parallel banding, and parallelism of grain. 
It is well said that the necessary conditions for igneous banding 
are heterogeneous composition and differential movement (34). 
Of the suggestions listed above, those which fulfil these two 
conditions are successive intrusion, heterogeneous intrusion, and 
deformation during crystallization. It is to be noted that each of 
these involves movement. The orientation of the platelike grains 
can hardly be accomplished except by some sort of movement of 
the magma while the plates are suspended in it. Such orientation 
is seen in surface flows where it is parallel to the direction of flow, 
and is often visible in thin sections of trachytes. To be sure, the 
settling of crystals, which idea is in special favor recently, might 
be thought of as analogous to the settling of mica plates in a sedi- 
ment. Those falling on a flat bottom might adjust themselves in 
horizontal and parallel positions. On the contrary, it does not 
seem probable that such an orientation would occur in a crystalliz- 
ingmagma. ‘The settling is slow, so that other crystals might lodge 
close to the plate and prevent its rotation. Furthermore, the 
difference in specific gravities, tending to orient the plates, is less 
than the difference for mica in water, while the viscosity opposing 
the rotation is much greater. As a final argument against orienta- 
tion by settling, the relation of structures at Mt. Johnson (g) should 
