440 Peirce and Willson—Temperature Variation of the 
piles was the same as in the larger apparatus. They were 
made, however, of much smaller wire, only 0°1™™ in thickness, 
and, owing to the difficulty of keeping our platinoid free from 
kinks, we made use of German silver. As soft solder melts at 
too low a temperature and it is not easy to use hard solder on 
such small wire, the junctions were made by electroplating the 
wires together with copper. 
The calibration was accomplished by heating the junctions 
in a large iron, triple-walled, quadruple-jacketed stove in the 
near vicinity of a platinum coil, the resistance of which was 
used as a standard of comparison. The resistance of the 
platinum coil at various temperatures was determined by plac- 
ing it (either directly or enclosed in a thin tube of glass or 
metal) in ice, steam, and the vapors of chemically pure 
anilin, napthalin, diphenylamine, mercury, etc. A curve thus 
obtained gave the relation between resistance and air ther- 
mometer temperatures. Direct measures on several thermo- 
piles were made at the same time and in the same manner, by 
immersing them in the different vapors. It was necessary to 
reduce the current flowing through the potentiometer (2 meters 
in length) so that 1°" corresponded to 2° C. A temperature 
difference of 0-1° in the junction was still easily visible. 
Various heatings were made with marble cut from two of 
those slabs which the first experiment had shown to be of 
nearly equal conductivity. It was much easier with the smaller 
size of slabs to obtain smooth surface and uniform thickness. 
The following is a specimen of the results obtained : 
Dist. in millimeters from Cold Face. Temperature. 
0-0 39:5 
10°9 78°5 
2-8 117°4 
ae i 17 
43°6 WiekeyeS) 
04°5 235°6 
65°4 PMV tks) 
76°3 310°7 
In order to obtain the temperatures on the outsides of the 
extreme slabs, a layer of asbestos paper was interposed on each 
side between the iron plates and the marble, so that the total 
fall of teinperature in the marble is some 60° less than the dif- 
ference in temperature of the hot and cold chambers. 
A curve plotted from these observations shows that the tem- 
perature gradient was practically uniform throughout the 
whole range.* A similar curve taken with the slabs in reversed | 
order gave the same result. The different slabs, though cut 
from the same block of marble, differ slightly in conductivity, 
* Tt will be seen that this agrees with results obtained by Dr. Robert Weber 
(Nature, March 7, 1895, p. 439). 
