Present Problems of Geophysics. — Becker 1 3. 
rocks. If the separation is not molecular it is self-evident 
that it must be molar. The only molar currents readily con- 
ceivable in a body of magma are convection currents, and 
these, or even an equivalent mechanical stirring, would 
necessarily lead to fractional crystallization, a familiar pro- 
cess known even to the pupils of Aristotle, and which is 
almost unavoidable when mixed solutions solidify. This 
process is one of precipitation and is absolutely distinct from 
the differentiation (or, more properly, segregation) of rock 
magmas, in which a single liquid is supposed to separate 
into two or more distinct liquids. The general conditions 
of the order of precipitation during fractional crystallization 
in acordance with the phase rule are by no means beyond 
the reach of discussion, and the able investigations of 
Messrs. J. H. L. Vogt and J. Morozewicz have a direct bear- 
ing on this subject. 
A mystery which will assume greater importance as the 
accessible supply of coal diminishes is the origin of petro- 
leum. There is much to be said in favor of the unpopular 
hypothesis of Mendeleef, supported by experiments on cast 
iron, that liquid hydrocarbons are due to the decomposition 
of the iron carbides of the terrestrial nucleus. Such vast 
accumulations of oil as exist on the Caspian and the Cauca- 
sus seem incompatible with the hypothesis of animal or 
vegetable origin, although oils belonging to the same series 
as do the petroleums have been produced in the laboratory 
from organic materials. On the ofher hand, some meteor- 
ites contain hydrocarbons (which may themselves be due to 
the alteration of iron carbides) and there are geologists whO' 
infer that the petroleum may be derived from the mass of 
the earth itself.* If the origin of the oil is not animal or 
vegetable, the supply is very likely inexhaustible. More 
extended study of the connection between volcanic phe- 
nomena and the origin of asplialtic and other hydrocarbons 
is a desideratum. 
Ore deposits themselves form the branch of geology 
which was earliest cultivated and which will never lose its 
interest so long as mankind remains gainful. Yet much re- 
mains to be done by experiment for the theory and practice 
*H. L. Fairchild, Bull Geol. Soc. Ame.r. vol. xv. 1904, p. 253. 
