_ TWO-PHASE CONVECTION IN IGNEOUS MAGMAS 499 
to disturb the banding and give a gradation rather than an alter- 
nation in composition. It may be suggested also that stoped 
blocks settling in a magma would disturb such banding. Differ- 
entiation, however, is in no way interfered with’ by the circulation 
that produces the banding. 
SUMMARY 
Several lines of evidence indicate that active convection occurred 
in many large, deep-seated magmas, and the process seems to be 
mechanically probable. In starting a current in such a mass the 
increase in density of growing crystals is probably more important 
than the development of any gas or separate liquid phases; and, 
added to the effect of simple cooling, the forces seem to be ample. 
Nearly all the field observations commonly made on igneous rocks 
may have a bearing on the question; probably first should be 
placed an alternation of bands of varying mineral composition; the 
position of the bands and the walls of the chamber; the mineral 
composition of the bands and the walls; any parallelism of grain 
and its direction; the form and size of the mass; grain variation 
near the margin; contact effects; the sequence of rocks formed in 
differentiation, continuous, double, or broken series; abrupt or 
gradual variations; intrusive relations between differentiates; 
gravitative or border position of differentiates; occurrence of one 
differentiate as a matrix for grains of another; the order of crystal- 
lization; signs of mineralizers, and their association with certain 
differentiates; globular forms. 
The idea of convection becomes of practical service to the 
geologist when related to the banded structure. By it we may find 
the position of the walls of the chamber. The complexity of some 
differentiated rock series is best explained by assuming that some 
of the series developed during convection. Knowing the order of 
crystallization we can at once decide whether a mineral like mag- 
netite is likely to be in bands near the bottom, or nearer the center, 
of the magma chamber. Finally a knowledge of the direction of 
the convection current aids greatly in estimating the extent of 
such a body of segregated magnetite. 
tN. L. Bowen, ‘‘The Later Stages of the Evolution of Igneous Rocks,” Journal 
of Geology, Supplement to Vol. XXIII (1915), p. 16, does not agree. 
