82 CORUNDUM, ITS OCCURRENCE AND DISTRIBUTION. 
began to cool the minerals would crystallize out, not according to 
their infusibility, but according to their solubility in the molten 
magma. The more basic portions, according to the general law of 
cooling and crystallizing magmas, being the most insoluble, would 
therefore be the first to separate out. These would be the oxides con- 
taining no silica, which in the present case would be chrome spinel 
and corundum. 
Morozewicz" has shown by some very important experiments that 
molten glass of a character similar to the basic magnesian rocks would 
dissolve alumina readily, and as this molten mass began to cool 
corundum and spinel Avere the first minerals to separate out. Accord- 
ing to this, the corundum and spinel w^ould be the first to crystallize 
or solidify out from the molten mass of the peridotite as this began 
to cool, and this crystallization Avould take place first on the outer 
border of the mass, because here it would cool first. Convection cur- 
rents would then tend to bring a new supply of material carrying 
alumina into this outer zone, and when this was reached crystalli- 
zation Avould take place and the alumina would be deposited as- 
corundum. 
This is essentially the idea advanced by Becker ^ in a paper on 
fractional crystallization of rocks, and where this process has taken 
place in dikes and laccolites a concentration is observed of the earlier 
and more basic minerals at their outer boundary. 
The more basic the magma the more fluid it is apt to be, and the 
more tendency there is for this process to take place, as is shown by 
many well-knoAvn instances of the separation of ilmenite, pyrrhotite, 
etc., from gabbro magmas, as described by Vogt,^* Adams,^' and others. 
Fig. 12 is a diagrammatic vertical cross section of the writer's idea 
of the appearance of a mass of peridotite soon after its intrusion into 
a gneiss, the peridotite holding a large amount of alumina in solu- 
tion, Avhich afterwards separated out as corundum. In this figure the 
corundum zone has been greatly exaggerated in order to better illus- 
trate the cross section. The corundum would be concentrated toward 
the borders of the peridotite, and would make a sharp and nearly 
regular contact with the gneiss. With the peridotite, however, the 
contact would sometimes be sharp and regular; at other times therc^ 
Mould be an irregular line of contact and masses of the corundum 
would penetrate into the peridotite. 
The rapid erosion to Avhich the rocks in this mountain region have 
been subjected would readily Avear them doAvn to their present condi- 
tion, represented by tlie dotted lines in fig. 1'2. 
" Zeltschr. fiir Krystall., vol. 24, 1895, p. 281. ' 
"Am. .Tour. Sci., 4th ser., vol. 4, 1897, p. 259. 
' Zeitschr. fiir prak. Geol., Nos. 1, 4. and 7. 189:;. 
<< Paper read before the Gen. Min. Assoc, of the I'rov. of Quebec, Montreal, .Tan. 12, 
1894. 
