456 FRANK F. GROUT 
arises from the fact that if a crop of crystals settled across the 
intrusive bands they would destroy the banding and orientation. 
Fourthly, the alternation of bands found is so varied and extreme 
in composition—from anorthosite to peridotite—that the process 
of filter pressing can hardly yield the liquids which would be needed. 
Finally, the alternating bands, if they represent two liquids 
imperfectly mixed, should consist of a large volume of the upper 
liquid phase and a smaller amount of the phase strained off or 
filter-pressed from below. At Duluth there are found small 
volumes of granophyr and peridotite, with large volumes of anor- 
thosite and immense volumes of olivine gabbro. These would 
hardly result from filter pressing. 
It thus appears that crystal settling and filter pressing and 
heterogeneous intrusion will not explain the structures at Duluth. 
However, other modifications of the idea of heterogeneous intrusion 
may be suggested. The objections mentioned are enough to 
make them all unsatisfactory. The magmas may come from two 
reservoirs or become heterogeneous by any other process, but if 
they did they would have time to mix, and the mixing and crystal 
settling would destroy the banding. The magmas might be in- 
truded, when partly crystallized, as a great mass with “mushy” 
consistency. Banding and orientation would be satisfactorily 
explained by this idea, but in such a banded, mushy mass there 
would be no opportunity for gravitative differentiation. 
It seems necessary to believe that both differentiation and some 
sort of motion were involved in the production of the bands, and 
that these occurred after the magma reached its chamber. It may 
be best to leave the matter open as to the kind of motion that 
occurred, but the idea of convection is an attractive one. 
CONVECTION 
It is suggested that many cases of igneous banding are related 
to convection currents during crystallization differentiation. It 
is not necessary, in conceiving of this action, to regard it as a very 
thorough stirring, but rather as some degree of circulation following 
the intrusion of either homogeneous or heterogeneous magma. 
Neither is the process exclusive. Successive intrusions of hetero- 
