384 PROFESSOR W. C. ROBERTS-AUSTEN ON THE DIFFUSION OF METALS. 
the gold and copper was roughly effected. I he molecular mobility of the metals has 
influenced the result, and the metals dissolved in each other, become, by a spontaneous 
process, spread or diffused uniformly ; in this case the uniformity is not materially 
disturbed when the solidification of the mass is effected. 
In view of the great interest connected with such action, tlie_ absence of direct 
experiments is remarkable, but this may perhaps be explained by the difficulty of 
conducting them. The sources of these difficulties are many. Such metals as are 
suitable for study require a more or less elevated temperature to melt them, and, 
where diffusion is concerned, small variations in temperature may he of much import¬ 
ance, for, as Graham showed, the rate of diffusion of salts in water is greatly 
increased by a small rise in temperature, the diffusibility of chloride of sodium, for 
instance, being more than doubled by a rise of 33°. It is now well-known that the 
osmotic pressure of a salt in solution is measured by the diffusion which takes place. 
A rise of temperature, therefore, which augments the osmotic pressure, must also 
increase the rate of diffusion. Graham further pointed out that the inequality of 
diffusion which various saline substances exhibit at a low temperature, becomes less 
at a high temperature, and he therefore concluded that “ it would appear to be the 
effect of a high temperature to assimilate diffusibilities ” of different salts.* 
In the case of molten metals, the necessity for working at high temperatures, which 
until quite recently could not be readily measured even with approximate accuracy, 
and the fear that the value of the results would be impaired by the action of convec¬ 
tion currents, must have deterred physicists from undertaking experiments on the 
diffusion of molten metals. Ost wald’s statement + with reference to the diffusion of 
salts, that “ to make accurate experiments on diffusion is one of the most difficult 
problems in practical physics,” may well have given rise to doubts whether any 
method which seemed to be available for conducting such investigations with molten 
metals would afford trustworthy results. The difficulties are obvious, but my long 
connection with Graham’s researches made it almost a duty to attempt to extend his 
work on liquid diffusion to metals, and, therefore, fourteen years ago the present 
investigation was undertaken, but it was abandoned because I was unable to measure 
with sufficient accuracy the temperature at which diffusion took place, and it has only 
been resumed during the past two years. 
As regards the history of the subject, I believe that a brief communication of my 
own on the mobility of gold and silver in molten lead, to the Chemical Section of the 
British Association at the meeting at Southport in September, 1883, embodied the 
results of the first experiments which were ever made with the direct object of inves¬ 
tigating the diffusion of molten metals and alloys, other than those of mercury which 
are fluid at the ordinary temperature. 1 stated that “ while molten copper and 
antimony interpenetrate but slowly, the mobility of gold and silver in molten lead is 
* Graham, 4 Collected Papers,’ p. 570, or 4 Phil. Trans. Roy. Soc.,’ 1861, p. 183-224. 
+ Ostwald, 4 Solutions,’ English Edition, 1891, p. 122. 
