304 N. L. BOWEN 
divergence from theory may be explained by assuming that the 
coefficient of diffusivity is not a constant but is itself a function 
of concentration and is greater for diopside-rich mixtures than for 
those poor in diopside. In those experiments with aqueous solu- 
tions, where the highest degree of correspondence with theory is 
obtained, the solutions are always kept dilute so that the medium 
into which diffusion is taking place is sensibly constant. Under 
these conditions theoretical concentrations calculated on the basis’ 
of a constant value of the coefficient of diffusivity are in marked 
accord with observed values. In the present case, however, 
there is a continual and very important change in the nature of 
the diffusion medium as time progresses, and no constancy is to be 
expected in the value of the coefficient of diffusivity. ‘There is no 
necessity, therefore, for regarding the results as showing divergence 
from Fick’s law, the results being reconcilable with theory if it is 
assumed, as mentioned above, that the diffusivity is a function of 
concentration. 
Einstein has developed for dilute solutions a theoretical relation 
between diffusivity and viscosity which makes them inversely 
proportional.t While experimental results do not entirely confirm 
his theory, they suggest its correctness if certain disturbing factors 
such as hydration could be evaluated.? At any rate, if we assume 
that the diffusivity is an inverse function of viscosity, we obtain a 
natural explanation of our experimental results. The viscosity of the 
diopside-rich mixtures is much less than that of the plagioclase- 
rich mixtures. The coefficient of diffusivity for the diopside- 
rich mixtures should be correspondingly greater, and this we have 
found to afford a natural explanation of the deviation from theo- 
retical values calculated in the ordinary way. Moreover, as one 
would expect, this deviation is more marked for the plagioclase 
liquid Ab,An, whose viscosity is very much greater than that of 
diopside liquid, and less marked for the plagioclase liquid Ab,An, 
where the viscosity contrast is not so great, while liquid Ab,Anz 
occupies an intermediate position in this particular (cf. Figs. 1, 3, 
and 5). Qualitatively, then, the experimental results suggest 
t Ann. d. Physik, Yol. XVII (1905), p. 549. 
2. W. Oholm, Medd. K. Wettenskapsakad Nobelinstitut 2, No. 26, p. 21. 
