386 PROFESSOR W. C. ROBERTS-AUSTEN ON THE DIFFUSION OF METALS. 
these earlier experiments was to show that samples of lead (which were removed by 
sucking them from the upper part of the tubes into stems of tobacco pipes) always 
revealed the presence of weighable quantities of gold after a lapse of the first three 
hours, while a sample withdrawn at the end of a second period of three hours did not 
indicate the presence of a commensurate amount of the precious metal. This fact 
either pointed to defects in the method, to the transmission of gold by convection 
currents, or to the very rapid diffusion of gold when minute quantities of this metal 
are present in lead. I believe that the latter will ultimately prove to afford the true 
explanation of the facts observed. 
The vertical tubes were then replaced by U-tubes of wrought iron, each limb of 
which was 23u millims. long and 10 millims. internal diameter. The tubes were 
filled with lead, and heated externally in a bath of molten lead, in an appliance 
which will be described immediately, and the precious metal, in the form of a rich 
alloy with lead, was inserted into one limb of the tube. Experiments proved that 
the gold falling by gravity became rapidly and uniformly distributed through the 
column of lead in that side of the tube, a.b., Fig. 1, into which it was introduced, and 
Fig. 1. 
in the rounded part of the tube, b, c, at the base of the U. After a given number 
of days the tubes were cautiously withdrawn from the bath and cooled from the 
bottom so as to solidify the alloy. The tube was then carefully divided by transverse 
saw cuts into measured sections, which were numbered consecutively, and the alloy 
could then be readily melted from each section and weighed, after which the amount 
of precious metal it contained was determined by analysis. 
It was found, however, that the use of U-tubes greatly increases the difficulties 
of calculation, as there is great uncertainty in any assumption as to the distribution 
of the diffusing metal during the experiment at points near the bend. It 
was found that in all these experiments, the gold or platinum was very evenly 
distributed by gravity through the limb into which the metal was first introduced. 
Idle calculations were finally made on the assumption that the concentration at 
the bend was, throughout the experiment, equal to the mean of its initial and final 
concentration. I, nevertheless, determined to sacrifice a long series of results, as 
they are less trustworthy than those obtained later by the use of single tubes. 
