PROFESSOR W. 0. ROBERTS-AUSTEN ON THE DIFFUSION OF METALS. 389 
on a photographic plate. By this means a continuous record of the temperature at 
three positions in the air bath may be readily obtained, but as regards each of the 
three positions this record is intermittent, and the indications afforded by each 
individual thermo-junction form a dotted not a continuous line. The intermittence is, 
however, very rapid, and the result is three clear and distinct time-temperature 
records which enable any variation in temperature to be readily detected and 
measured. The records, which are very numerous, show that while there was no 
rapid variation of temperature, there was, however, a gradual fall in temperature 
from the beginning to the end of the experiment, and this is due to the fact that 
the burners became slightly obstructed by the products of the combustion of the 
gas, and in experiments which are now in progress a thermostat is employed. The 
occasional slight adjustment of the gas-taps by hand in accordance with the indications 
of the thermo-junctions rendered it possible to maintain a very constant temperature. 
Much care was taken to prove that the method of inserting the thermo-junctions 
down a central tube really indicated the mean differences of temperature between the 
upper, middle and lower portions of the bath, and it was shown that there was no 
object in continuing the use of a lead bath which was first adopted, as it greatly 
complicated the manipulation. 
Now that the possibility of making accurate measurements on the diffusion of 
molten metals has been demonstrated, and as it has been shown that convection 
currents are not established, even when the temperatures at the top and the bottom 
of the diffusion tubes only differ by 35° C., it will be well to attempt to do without 
this graduated temperature altogether, trusting to the difference in densities of the 
solvent and the diffusing metal. It will also be necessary to set up apparatus which 
will automatically maintain for lengthy periods any given temperature between the 
ordinary temperature and 600° C. 
At the end of a given period varying from six hours to seven days, the diffusion 
tubes were removed, cooled from below, carefully measured, and then cut into 
transverse numbered sections, the metallic contents of each of which were weighed 
and analysed. When gold is the diffusing metal, determination of its amount in 
the various sections of lead can be effected with remarkable precision by the ordinary 
method of cupellation assay, provided that check assays on samples synthetically 
prepared are made simultaneously, and the rigorous accuracy of such assays is so 
well known to the chemists who have had actual experience in the matter, that it is 
unnecessary to submit evidence on this point. The estimation of platinum or of 
rhodium in lead is a matter of greater difficulty, but details of the analysis need not 
be given. 
If the bath is unprotected below, and heated from the top, the top will be very 
much hotter than the bottom, and thus the rate of diffusion would be considerably 
greater near the top than near the bottom of each tube. The value of the results 
would also be much impaired if the bath were inadequately protected on the top, for 
