390 PROFESSOR W. C. ROBERTS-AUSTEN OR THE DIFFUSION OF METALS. 
if the upper surface of the bath is cooled, convection currents will be established. The 
lead, moreover, at the surface oxidises, and forms a detachable slag, and so relatively 
increases the percentage of gold or platinum in the upper layer of lead. It was 
frequently noticed that the top layer in the tubes was slightly richer than the one 
next below it. 
In the experiments, the results of which are plotted in figs. 3 and 4 (p. 399), to which 
reference will subsequently be made, the temperature, as indicated by the top thermo- 
junction, was gradually raised to 550°, at which point it was maintained practically 
without variation during the greater part of the experiment, which lasted nearly seven 
days. The temperature, however, slowly fell during the last two days to 500°, owing 
to the choking up of the burners. The middle thermo-junction showed temperature 
25°, and the lowest one 35°, below this ; the mean temperature shown by the middle 
junction being given as the temperature of the experiment. It should be pointed out 
that these temperatures are based on measurements which assume the melting point 
of gold to be 1045°, and lead 325°. Recent work seems to show that 1045° may be 
some 1 5° too low. # 
It may be well to offer here a few general considerations respecting the phenomena 
to be observed. 
It is now held that liquid diffusion is the result of osmotic pressured A move¬ 
ment of the particles (molecules or atoms) of the dissolved substance takes place, and 
a molecular force drives them from the place wdiere they are more closely packed 
and, therefore, exert greater pressure, and impels them to positions in which they 
are more widely distributed. This movement continues until the concentration, and, 
therefore, the pressure of the diffusing metal, is constant throughout the liquid. 
Graham’s method of studying liquid diffusion consisted in filling wide-mouth phials of 
glass with the solutions of salts, which were allowed to diffuse outwards into water con¬ 
tained in capacious cylinders. This method could not well be imitated in the present 
experiments, as the manipulation, and the calculation of the results obtained by such 
a method present great difficulty. Hence the adoption in the present research of 
vertical tubes, as has already been described. In the earliest experiments made 
by me, in 1883, the little spheres of precious metal obtained from each measured 
section of lead were arranged on a card scale at measured distances each of the 
little spheres, therefore, represented the amount of gold in the section of the tube 
from which it had been derived, and their general appearance, when arranged as lias 
just been described, suggested that a trustworthy method had been secured. It 
appeared probable that the law of diffusion of salts, framed by Fick, would also apply 
* See Heycock and Neville, * Trans. Chem. Soc.,’ vol. 67, 1895, p. 160. Roberts-Austen, ‘Nature,’ 
May 9, 1895, p. 40. Le Chatelier, ‘ Comptes Rendus,’ vol. 121, 1895, p. 323. 
t Nernst, ‘ Zeitscli. fiir Physikal. Chemie,’ vol. 2, 1888, p. 613. 
X Specimens of these records were exhibited to Section B. of the British Association, at the Montreal 
Meeting, 1884, 
