638 FRANK F. GROUT 
magma to indicate that assimilation after intrusion could yield new 
types. Successive intrusions certainly differed slightly, and it is 
likely that some parts of the earlier magma were very nearly of the 
composition of anorthosite. However, although this early magma 
is different from the later, larger intrusion, there is no sign that the 
variety in the main gabbro is due to successive intrusions. If this 
main gabbro was heterogeneous when intruded there was plenty 
of time for it to become mixed, unless the tendency of the parts 
was to become more distinct rather than to mix. It therefore 
seems that the main development of variety in the lopolith depended 
on processes of differentiation. ‘The variety of rocks described 
above shows how thorough this differentiation was. 
Processes of differentiation.—Recent experimental work is very 
conclusive in maintaining the reality of crystal settling in a magma 
and the improbability of extended diffusion as factors in differen- 
tiation. It is much less conclusive in dismissing convection and 
assimilation and the separation of immiscible fractions of magmas." 
There are some evidences at Duluth which indicate the processes 
involved. 
Crystal settling versus convection.—One of the first considerations 
in a study of crystal settling concerns the plagioclase. The mass 
was so large that at Duluth the crystals must have grown very 
slowly, for the most part too slowly to yield zoned structures. 
With such very slow cooling a plagioclase would begin to crystallize 
with a composition much more basic than would be calculated from 
an average of the magma; but as crystallization proceeded it 
might have readjusted itself to the magma, so changed in compo- 
sition as finally to be the plagioclase indicated by the composition 
of the average magma. This adjustment is supposedly interfered 
with in cases of very slow cooling, hence to be expected at Duluth, 
by the settling or floating of the early basic crystals out of reach 
of the residual liquid. The crystals should then be more basic 
plagioclase wherever the early crystals accumulated and more acid 
elsewhere.?, Let us see how the Duluth gabbro fits the case. 
«N. L. Bowen, ‘Later Stages in the Evolution of Igneous Rocks,” Jour. Geol., 
XXIII (1915), supplement to the December number. 
2F. L. Bowen, zb7d., p. 33. 
