DIFFUSION IN SILICATE MELTS . By 
fact that rocks are normally millimeter-grained rather than centi- 
meter- or meter-grained even in large masses is, however, a tribute 
to the slowness of diffusion in magmas. The fact that certain 
conclusions are reached above on the assumption that diffusion acts 
alone should not be taken as indicating that the writer believes 
that no other processes could occur. To account for the extremely 
coarse grain of many pegmatites, for example, it seems necessary 
to assume circulation of solutions, and in many other cases cited 
circulation (convection) would be inevitable. 
SUMMARY 
The rate of diffusion in certain silicate melts has been deter- 
mined experimentally by permitting diffusion against gravity 
of a heavy liquid into a lighter liquid. The concentration curves 
found are not coincident with any theoretical curves calculated on 
the basis of a constant value of the diffusivity, but can be inter- 
preted on the assumption that the diffusivity varies with concen- 
tration and is less for concentrations corresponding to more viscous 
liquids than for those corresponding to less viscous liquids. Taking 
as representative of the “average diffusivity”? the amount of 
material which penetrates into the upper layer, the following 
values of the average diffusivity (k) were found: for diopside into 
Ab,An,, k=o.015; for diopside into Ab,An,, kR=o.14 to 0.3, 
depending on the proportions; and for diopside into Ab,An., 
k=o.2, all in cm.” per day. 
The value 0.25 (close to the maximum experimental value) 
is taken as probably representing a fair estimate of diffusivity in 
magmas, and with this as a basis it is shown that such phenomena 
as the formation of border phases about large bodies of igneous 
rock by diffusion cannot be considered possible in the time available 
for such action in a cooling magma. On the other hand, the forma- 
tion of reaction rims about inclusions may be attributed to diffusion, 
though for very wide rims a considerable period of time will be 
required. 
