THE ORIGIN OP IGNEOUS ROCKS. 
191 
these groups, which had a quartzose, intermediate, and py- 
roxenic modification, it was necessary to assume a special 
ephemeral lake or reservoir, and therefore a special course 
of erosion. Moreover, the reservoirs were supposed to have 
been produced at successively greater depths within the 
crust of the earth, and since they differ from one another in 
composition, the material of the crust must vary in compo¬ 
sition with the depth, and also in such a manner that zones 
of like composition must have been affected in the same 
regions at widely different times in order that andesites or 
rhyolites and basalts might be erupted at various geological 
periods. 
Two years later Dutton advocated the theory (also held 
by others) that igneous rocks have be^n derived from the 
fusion of sedimentary rocks, and that their diversity is not 
due to a differentiation of a molten magma. This was based 
on the erroneous assumption that the original magma, of 
which the most ancient rocks were formed, must have had 
the composition of basalt, and that a differentiation of a 
molten magma must have been mechanical, and could not 
have taken place, but that the separation must have been 
brought about by processes of erosion and sedimentation. 
His speculation should be regarded more as a contribution 
to the physics of eruption than to the problem of the origin 
of volcanic rocks; for he admits his surprise that there should 
be as much regularity as there is to the order of succession 
of volcanic rocks upon his theory of their origin, and he also 
confesses that he knows of no cause for the increase of heat 
necessary for their production. 
The next discussion of the origin of volcanic rocks is that 
by Rosenbuscli, which, however, is confined entirely to their 
chemical characteristics. Upon purely chemical considera¬ 
tions, he is of the opinion that the variability in the compo¬ 
sition of igneous rocks of all kinds must be the result of a 
differentiation of a generally uniform magma. Moreover, 
this separation is thought to have followed chemical affini¬ 
ties. By comparing a number of chemical analyses of dif- 
