438 Peirce and Willson—Temperature Variation of the 
time gave the amount of heat conducted through the slab per 
second per unit of area. 
The platinoid ends of the thermal junctions were soldered 
together and placed in an oil bath, C, immersed in melting ice. 
To them was soldered a copper wire leading to the (copper) wire 
of a potentiometer. The copper ends of the couples led to a 
mercury switch, by which any one of them might be quickly 
connected with a second copper wire leading to the potentio- 
meter. 
With a comparison cell having an EK. M. F. of 2 volts and 
with about 200 ohms in the circuit, the wire of the potentio- 
meter 2™ in length gave about 1° for a difference of 1° C. in 
the temperature of the thermal junctions. The galvanometer 
was sensitive enough to determine with ease the balance within 
less than a millimeter on the bridge wire, and the temperatures 
of the junctions were certainly determined within 0-1° ©. 
Tne thermopiles were calibrated after the prism had been 
taken to pieces, by removing the junctions from the slabs to a 
frame immersed in an oil pot, which contained a standard ther- 
mometer* and was placed in a water bath which conid be brought 
to various temperatures up to nearly 100° C. That this was a 
safe proceeding appears from a paper printed last year in this 
Journal.t 
In conducting the experiments the steam was passed through 
the steam box continuously; the upper box packed with ice 
and water, and the stirring apparatus started. 
Readings on the potentiometer were made at intervals, until 
each of the thermopiles had attained a fixed temperature. This 
state was reached after from five to eight hours, and continued 
as long as the conditions were properly maintained, the indica- 
tions of the thermopiles remaining satisfactorily constant for 
hours. The reading of all the 12 thermopiles, when some 
skill had been attained, required about 3 minutes. 
A prism of this sort was heated many times while we were 
perfectingt our apparatus for the determination of absolute 
thermal conductivities, and the following is given as showing 
the nature of the consistent results thus obtained : 
Distance in millimeters from 
Interspace. the bottom of the Cold Box. Temperature. 
1 10°3 1 
2 21:0: 32°8 
3 319 48°5 
- 42°9 62°5 
5 54:0 75°8 
6 64°7 89°4 
* Our final standard was Tonnelot No. 11,142, upon which a very complete 
set of tests has been made at the International Bureau of Weights and Measures. 
+ Vol. xlviii, pp. 302-306. 
{ We are much indebted to Mr. G. W. Thompson, the mechanician of the Jef- 
ferson Physical Laboratory, for the skillful help which he gave us in this work. 
