TWO-PHASE CONVECTION IN IGNEOUS MAGMAS 4907 
gravity, and a gravitative arrangement is common. It is not to be 
expected that large segregations of a single mineral will be other 
than exceptional, because the cooling progresses in such a way that 
crystals of several minerals are likely to be growing at once and all 
settling together on the bottom as well as lodging along the walls. 
-However, the conditions are easily conceived as possible for the 
formation of magnetite and peridotite near the base, and anorthosite ~ 
near the top, of a single magma. 
The behavior of immiscible liquids may be considered at this 
time. If the separation of globules occurred in a stationary 
magma, a gravitative rise and fall would tend slowly toward the 
separation of the fractions. If the immiscible liquids separate 
during convection, the smaller fraction, if light, will separate along 
the top, if heavy, along the bottom, as a fairly distinct layer in 
logical gravitative position. This layer may crystallize before or 
after the magma from which it separated, and the first to solidify 
may be intruded by the other. Abrupt gradations and contacts 
should be the rule. A small separated layer is likely to escape from 
the general circulation and is less likely to show a fluxion structure. 
Double differentiation.—Thus it is conceived that a magma 
might differentiate into a series of bands dependent on the order of 
crystallization and settling, and at the same time give a rather abrupt 
gradation to a separated immiscible rock type—one of radically 
different composition. This is double differentiation. And the 
complexity of the sequence in some petrographic provinces is strong 
indication that two very different processes have been in operation. 
Such a suggestion might apply to the occurrence of pyrrhotite 
at Sudbury which is said to have intrusive relations in some expo- 
sures to the main norite; and the norite itself is differentiated in 
roughly gravitative fashion. 
Convection structures.—In discussing convection Pirsson makes 
the following suggestion: ‘‘ Probably at first as the liquid moved 
inward over the floor of the laccolith and became reheated, these 
crystals [formed in the cooling border zones of the magma] would 
remelt, giving rise to numerous small spots of magma of a different 
composition, which would slowly diffuse.”* It is noteworthy that 
tL. V. Pirsson, ‘‘The Igneous Rocks of the Highwood Mountains,” U. S. Geol. 
Survey Bull. 237, p. 188. 
