154 
IDDINGS. 
so, while at Pantelleria the iron oxide is the most variable 
and the predominant bivalent element. 
The most marked instance of discordance is between 
magnesia and alumina ; so that it may be stated in general 
that as the magnesia increases, the alumina decreases; to this 
there are numerous exceptions. Moreover, they do not vary 
to the same extent. The most notable instance of this is in 
the Crandall group of rocks. The last mentioned relation¬ 
ship shows that in general as the aluminous molecules in¬ 
crease in number, the magnesian molecules decrease; but this 
relationship is not so definite that it may be expressed in 
terms of simple mineral molecules. One would expect from 
this that as the magnesian minerals increase in abundance 
in a rock, the aluminous minerals must decrease. In a general 
way, this is undoubtedly true; but it is not correct to assume 
that the aluminous minerals are wholly feldspathic, or that 
the alumina is always present in combination as a feld¬ 
spathic molecule. Aside from any knowledge of the actual 
presence of alumina in certain ferromagnesian minerals 
that occur in rocks, it may be shown by such diagrams as that 
of the group of rocks from Electric peak that there are in¬ 
stances in which the alumina molecules exceed those of soda, 
potash, and lime combined; so that the alumina must enter 
into some other compound molecules than the feldspathic. 
Chemical independence of the essential oxides .—It has been 
shown that molten magmas which have similar chemical 
compositions crystallize into different associations of min¬ 
erals according to the conditions under which they solidify,* 
or, more correctly, through which they pass during crystal¬ 
lization ; so that the chemical constituents of a molten magma 
are capable of entering into different molecular combinations 
according to circumstances. From this it must be concluded 
* Roth (J.) Gesteinsanalysen in tabellarischer Ubersicht, etc. 4°. Ber¬ 
lin, 1861, p. xxi. 
Iddings (J. P.) Bull. Phil. Soc. Washington. 8°. Washington, 1892, 
vol. 11, p. 217; also Twelfth Annual Report U. S. Geol. Surv. 8°. Wash¬ 
ington, 1892, p. 656. 
