296 : N. L. BOWEN 
from these data that Becker’s objection was raised. There has 
been as yet no discovery of a general principle connecting the 
rate of diffusion of matter in solution with other physical constants. 
Each substance has its coefficient of diffusivity (in a given medium) 
characteristic of the substance and determinable only by experi- 
ment. The coefficient is the constant factor, k, in the equation 
expressing Fick’s law of the rate of diffusion, viz., 
dc @¢ 
an” ag? 
and is evidently equal to the number of grams which diffuse past — 
I square centimeter of any plane in unit time when the concentra- 
tion gradient normal to the plane is unity. This is a very small 
quantity for substances investigated. For most salts in aqueous 
solution it is of the order 3 X107° in cm.’ per second at the ordinary 
temperature and increases rapidly with rise of temperature; for 
molten metals it is considerably larger and of the order 3 X10-°, 
again increasing with rise of temperature.” 
In the case of the quite different type of matter, molten silicates, 
it is perhaps not probable, but nevertheless not inconceivable, 
that the coefficient of diffusivity should be fairly large at the high 
temperatures concerned in spite of the usual high viscosity. Only 
actual measurements can make us at all certain that we have a 
firm basis of fact for the discussion of diffusion in molten silicates. 
Practically no measurements have been made. Endell has demon- 
strated the fact of the interdiffusion of lime and microcline glass.’ 
Schulze has measured the rate of migration of silver ion in glass. 
Indeed, it is perhaps principally from the problems of glass manu- 
facture that we gain our impressions of diffusion in molten silicates. 
There diffusion is often excessively slow, but it should be noted 
that glasses belong for the most part among the very viscous 
silicate mixtures. On account of the lack of experimental data it 
was considered advisable to undertake some measurements of the 
1 Roberts-Austin found for the coefficient of diffusivity of gold into molten lead 
k=3.47X10-5 at 492°. Roy. Soc. London, Phil. Trans., Vol. 187A (1896), p. 383. 
See also Van Orstrand and Dewey, U.S. Geol. Survey, Prof. Paper 95-G. 1915. 
2K. Endell, Silikat-Zeitschrift, Vol. I, p. 195. 
3 Ann. phys., Vol. XL (1913), P- 335: 
