338 On certain Molecular Constants. 
of convection-currents is at least as inconsiderable as in the 
experiments which have been performed for determining the 
diffusion of soluble salts in water. It is scarcely worth while 
amusing oneself by dividing these diffusion percentages by the 
so-called " atomic weights " of the metals. A more serious 
consideration might be the result of the division of the diffused 
weight by the specific gravity of the metal. Comparing the 
numbers of the group Sn, Pb, and Zn with one another, we 
may remember that the metals are all cast, and therefore so 
far indefinite in structure. This may be especially the case 
with zinc, which cracks and thereby allows the mercury to 
rise by capillarity and so enrich itself, and generally set up 
conditions of amalgamation which I do not care to trace, for 
I do not see my way through. 
As to the comparison of the alkaline group with the 
Sn, Pb, Zn groups, such comparison must be vague, for the 
reason that the K and Na are employed as amalgams (as 
though one would study the diffusion of nitre into water by 
employing a solution of nitre containing only 2 per cent, of 
the anhydrous salt) ; whereas the Sn, Pb, and Zn are use 
with what was supposed to be a sufficient supply of pure 100- 
per-cent. metal. But this imperfection of these conditions is 
manifest if we remember that while the solid metal melts and 
dissolves downwards, the liquid mercury rises. Accordingly 
there is, after the first instant of contact, supposing the metals 
diffuse, no constant metallo-motive force in the same place. 
§ 34. I conclude therefore that the general curve of amal- 
gamation, and therefore of alloyage, and therefore perhaps 
of elementary atomic and molecular diffusion generally, is of 
the kind shown in fig. 8. In the case of Na almost the com- 
plete curve was obtained j whereas in the case of Zn the point 
of contriflexure had not been reached. The very fact that 
the K and Na curves are more complete, in this fashion, 
than the Pb, Zn, and Sn, prove to my mind that K and Na 
have a far greater diffusive energy than the heavier metals 
examined. And although in this case the percentage of metal 
actually found at a given depth was in all cases much less 
than the percentages of the heavier metals, it will be borne in 
mind that, while the latter were few r er and had acted for 
thirty-one days, the former were amalgams containing less than 
2 per cent, of the metal. Comparing K with Na, I do not 
think we can draw any conclusion beyond the rather negative 
one, that the superior diffusive faculty which seems to be the 
property of K salts in regard to water does not evidence itself, 
if it exist, when that metal and sodium are compared in respect 
to their diffusion in mercury. I am far from asserting that 
