INTERNAL STRUCTURES OF IGNEOUS ROCKS 455 
called to the variety sometimes shown in a series of extrusive 
lava flows, apparently derived from a single large chamber. The 
most recent statement of the case is incidental to a discussion of 
differentiation by crystallization and settling (33). It is necessary 
to introduce some modification to explain the development of the 
banded structures often seen. If differentiation took place by 
settling of crystals, and before the mass was all solid some dynamic 
process squeezed the liquid out from between the settled crystals, 
this liquid would not be the same in composition as the supernatant 
magma. ‘These two liquids might be involved in an intrusive layer 
and produce bands if not thoroughly mixed before crystallization. 
There are several difficult points in the application of this idea to 
“such banded rocks as the Duluth gabbro, though it seems clear from 
the variety of the Keweewanan lava flows that differentiation was 
well advanced before intrusion. 
First, the mechanics of the filter-pressing process in a deep 
reservoir like a batholith is not stated and is a little hard to con- 
ceive. Pressure on a magma is largely hydrostatic and not 
differential. 
Secondly, if heterogeneous liquids were intruded into so large a 
chamber there would be a great stirring and mixing effect and 
plenty of time to make the mixture more homogeneous before it 
crystallized. In general, the larger the mass the more time avail- 
able for diffusion and mixing. If banding was a result of hetero- 
geneous intrusion, the larger masses would be least banded. Asa 
matter of fact, the Duluth gabbro, one of the largest known intru- 
sions, is most strikingly banded. 
Thirdly, there is no reason to assume that the differentiation 
which caused the variation in the magma in the deep reservoir 
should suddenly cease upon intrusion into an upper horizon. In 
fact, from the gravitative arrangement it seems almost certain 
that some differentiation did take place. To be sure, if the hetero- 
geneous magma varied in specific gravity, the several parts might 
have been intruded in roughly gravitative position; but even if they 
were, there was nothing to stop the differentiation until the magma 
cooled. In so large a mass as the Duluth gabbro there would be 
plenty of time for further differentiation by settling. The difficulty 
