PROFESSOR W. C. ROBERTS-AUSTEN ON THE DIFFUSION OF METALS. 387 
The continuance of these experiments is mainly due to the interest which 
Lord Kelvin has always taken in them, and in a letter dated 4th of November, 
1883, he reminded me of the necessity for maintaining a graduated temperature 
from the top to the bottom of the diffusion tubes, and he has from time to time 
given me, on points of detail, suggestions, which I need hardly say have been 
adopted. 
Description of the Apparatus employed in the later Experiments. 
The tubes that contain the lead in which the diffusion takes place are arranged 
in a little furnace of special construction. In the earlier experiments they were, as 
has been already stated, placed in a bath of molten lead, but this was abandoned 
in favour of an air bath with double walls which can be heated above the melting 
point of lead and readily maintained at definite temperatures. The drawing, fig. 2, 
Plate 7, shows the general arrangement. The diffusion tubes are closed at the base, 
and two of such tubes are shown at TT', fig. 2, Plate 7, placed in a cylinder of iron, 
I, 3 inches in diameter and 7 inches high. The sectional plan (fig. 2, Plate 7) shows 
six of the diffusion tubes arranged symmetrically in this iron cylinder, which is enclosed 
in a second cylinder of thick copper, C, 4^ inches* in diameter and 8| inches high. 
The lower half of this copper vessel is surrounded with a layer of asbestos cloth, A. 
There is a lid composed of two discs of copper, DD', with asbestos between them. 
If, for any special reason, U-tubes should be employed, one open end of each tube 
might communicate with a hole in the double lid, and the metals, the diffusion of 
which is to be studied, are introduced through this hole. They fall by gravity to 
the base of the U-tube, and then rise by diffusion up its opposite limb. 
The heating is effected by a series of clay gas-burners BB', mounted on a ring 
EH'; the burners surrounding the upper portion of the copper cylinder. Investing 
cases of fireclay GG'G'G'", and a lid of clay H, completes the construction of the 
furnace. 
It is, of course, a matter of great importance to obtain a gas supply of constant 
volume, and this has been effected by means of a regulator, shown in fig. 3, Plate 7, 
and a delicate gauge not shown in the plate is also provided, and by its aid any 
variation in the pressure of the gas is indicated. 
The absence of a trustworthy method for the measurement of the temperature 
would have rendered it impossible to conduct these experiments, but by the aid of 
thermo-junctions such measurement can readily be effected. These thermo-junctions 
JJT ', are shown in fig. 2, Plate 7, and the method of using them will now be indicated. 
I have, however, already fully described the method elsewhere,! and the general 
* It is convenient to give these dimensions in inches, but a metric scale is also provided with the 
drawing. 
t ‘ Roy. Soc. Proc.,’ vol, 49, 1891, p. 347. ‘ Inst. Civil Engineers Proc.,’ vol. 110, 1891-2, Part iv, 
3 d 2 
