1887.] Mr A. Campbell on Measurement of Peltier Effect. 405 
and the whole tied between asbestos boards and copper plates. 
The whole was packed, in the usual manner, in asbestos within 
a copper box. The current was passed through the strips alter- 
nately (and for such periods as gave almost steady conditions of 
temperature), and the Joule effect found from the deflections in 
exactly the same manner as the Peltier effect in the foregoing ex- 
periments. Now the rise of temperature thus measured is directly 
proportional to the resistance of the strips, and inversely proportional 
to the heat capacity of the strips, asbestos, and thermopile ends. 
Thus, by observing also the change in resistance, we can at once 
calculate the change in heat capacity. By this method the correction 
is measured under almost exactly the same conditions as those under 
which it is to be applied. The influence of the asbestos paper 
(which was of very small mass) may be neglected. The widely 
different values of the correction found for the various metals seem to 
show that the capacity of the thermopile ends (which were of much 
smaller mass than the strips) may also be neglected. Thus the 
results may be considered as tolerably accurate measurements of the 
changes of specific heat in the various strips used. 
The resistance measurements were made with a Wheatstone’s 
bridge of the ordinary form. A thin strip of the given metal, 
soldered or brazed to two thick copper wires, was packed in asbestos 
in copper boxes. The measurements were all made at nearly steady * 
temperatures, as it was found that when this was not the case, in- 
consistent results were obtained. 
As the thermopile resistance formed a considerable fraction of the 
resistance of the galvanometer circuit, it had to be measured at 
different temperatures, and a correction introduced for this. In 
measuring this resistance, the battery contacts had to be short, other- 
wise the Peltier effects began soon to show themselves. Table XIII. 
gives some of these measurements. In this, as in all the other 
resistance and Joule effect measurements, the whole percentage in- 
crease from the value at the lowest temperature was calculated, and 
from this the mean percentage increase per degree C. from the lowest 
temperature. The resistance of the galvanometer and its connections 
was I '667 ohms. 
In order to test how far the thermopile deflections were a proper 
* I.e., not varying much more than 2° C. in half-an-hour. 
