118 
JDDINGS. 
He further states that Durocher confused this idea of liqua¬ 
tion with the separation of molten magmas upon solidifica¬ 
tion into different associations of minerals, according to cir¬ 
cumstances, which view is definitely expressed by Roth in 
his “ Tabellarischen Uebersicht der Gesteins analysen ” in 
1861. In this he states that molten magmas of the same or 
very nearly the same composition separate into different asso¬ 
ciations of minerals. A consequence of this would be, Rosen- 
busch remarks, that we should not be able to determine the 
mineral composition of a rock from its chemical analysis, to 
which no petrographer at this date would consent. With 
this conclusion the writer begs leave to differ. 
Rosenbusch then defines his use of the term “ spaltung,” 
for which we will use differentiation. It is the spontaneous 
separation of a chemically homogeneous whole magma into 
two or more chemically different partial magmas. 
In order to demonstrate the existence of such a differen¬ 
tiation he arranges a number of chemical analyses of gran¬ 
ular and extrusive rocks in a table, which is finally reduced 
to a form in which it expresses the proportions of the metal 
atoms in each analysis. He then calls attention to the fact 
that the numerous different directions in which these rocks 
vary would necessitate as many different original magmas 
if the different rocks were the mixtures of extreme magmas. 
He expresses the opinion that the original magma within 
the earth from which all eruptive rocks have been derived 
must be considered to have been originally homogeneous. 
This magma must have separated into partial magmas in 
order to produce the different eruptive rocks with which we 
are acquainted; and it is evident that this process has fol¬ 
lowed chemical affinities and has not been by chance. 
The fact that certain magmas are of common occurrence 
while others are wanting must depend upon the principle 
that certain elements in molten solution mutually condition 
and exclude one another. This assumption necessitates the 
differentiation of the original magma. The occurrence as 
well as the regular recurrence of definite eruptive masses 
