Molecular Constants. 329 
of K or Xa, the slab instantly expands, and the film of lyco- 
podinm- powder on its surface cracks radially, exposing the 
brilliant metallic surface, which is seen to be agitated over its 
whole extent. In a few seconds the slab contracts to its ori- 
ginal size and the lycopodium cracks heal. 
Does this extension of the slab depend upon the diminu- 
tion of the cohesion of the mass of the mercury, or upon a 
surface effect '? 
§ 23. I frequently in my researches have had recourse to 
the fact, which I first described in the year 1863, that a little 
sodium added to mercury enables that metal to touch with 
positive capillarity metals which in its and their ordinary 
state are not wetted by the liquid metal. I here make use of 
the same fact. A platinum tube, 6 millim. in internal dia- 
meter and 2 centim. in height, is rubbed and soaked in some 
weak sodium amalgam, and then washed in several quan- 
tities of pure mercury. Placing such a tube vertically in the 
middle of a slab of mercury so that its lower edge is clear 
of the surface upon which the mercury slab rests, we have 
the condition shown in fig. 4. A little grain of sodium 
amalgam dropped into the platinum tube causes no imme- 
diate change: but in a time measurable by seconds, say 20 
to 30 seconds, the slab starts on its expansion and reaches 
its maximum size, apparently immediately. It seems, then, 
that since the effect is not instantaneous, it is a surface effect. 
The effect when produced is due to an alteration of the surface 
between the tube and the outer portion of the slab. By 
dipping the platinum tube further down into the slab so as to 
be within ^ of an inch of the bottom, I have found the effect 
to be distinctly delayed. 
§ 24. The fact mentioned in § 23, that the release of the 
mercury skin-tension by sodium is brought about after a time, 
short indeed, but appreciable when introduced into the central 
part of a liquid slab inside the platinum tube, points to the 
existence of a true diffusion between the metals: and this leads 
to the third part of this communication. For I have exa- 
mined already a few such cases, which I will now describe, 
because I believe the subject of elementary diffusion has been 
neglected excepting in the case of gases, and even here but 
little is really known. 
§ 25. Metallic Diffusion. — The metals potassium and sodium 
suggested themselves of course at once. They offer excep- 
tional facilities for the determination of the composition of the 
mixture, when they have diffused through mercury, because 
the mere addition of water translates the alkaline metal into 
hydrogen. The neutralization of the alkalized water, say, by 
