A TYPE OF IGNEOUS DIFFERENTIATION 637 
some movements at about the time of solidification. The pegma- 
tites and aplites below would be located not so much by gravity 
as by simple vaporous tension; the lighter separate, being more 
fluid, might penetrate cracks on any side of the magma chamber 
in advance of the main magma. The red rock at the borders of 
siliceous sediments and filling the pores of sandstones as a cement 
may similarly have escaped from the magma under the tension of 
the vapors. The position of such a red-rock zone may be deter- 
mined less by gravity than by porosity. The vaporous solution 
could escape through the pores in advance of the gabbro. In an 
extreme case this separation of red rock might even be determined 
by a porous inclusion. However, the most favorable conditions 
for the accumulation of red rock must be the combination of a large 
body of magma with plenty of water and a sandstone roof having 
a tight cover above. 
RELATION OF ROCK TYPES TO POSITION 
It is estimated that over two-thirds of the gabbro mass at 
Duluth consists of olivine gabbro, varying only slightly from the 
average. Such average rocks are scattered from top to bottom. 
On the other hand, specialized types have a more limited range. 
The peridotite occurs only near the base; the magnetite gabbro, 
equally heavy, is near the center; the anorthite ranges from the 
center toward the top and is largely in the thin earlier intrusion. 
Very locally at the base of the early gabbro there is an apatitic 
hypersthene gabbro. The occurrence of red rock, mostly near the 
top and in a sill at a higher horizon, is emphasized above. 
DIFFERENTIATION 
Introduction.—It must be granted, in regard to the Duluth lopo- 
lith, that a magma supply was available, and, as indicated by the 
earlier flows in the same region, it varied from time to time or place 
to place. The problem of its history as a lopolith begins with its 
intrusion into the present chamber. Various theories are current 
as to the processes by which a magma during such a history gives 
rise to a series of rock types instead of a single one. The roof of 
the magma for much of its area was diabase and too much like the 
