in Alloys of Bismuth and Tin. 565 
through them. The temperatures having again become con- 
stant, as indicated by the constant value of the current in the 
‘junction ’ circuit, the difference of temperature between the 
corresponding points of the two rods—or rather, the current 
in the compensating circuit necessary to counterbalance the 
current in the ‘junction’ circuit due to this difference of 
temperature of the junctions—is observed. 
The current through the rods is then reversed, when, by 
virtue of the Thomson effect, the temperature at any point of 
the rods changes, one rod experiencing an evolution of heat, 
the other an absorption. After an ‘interval ranging from 
30 minutes to 2 hours, according to the specimen under 
examination, the temperatures again become constant, when 
the magnitude of the compensating current in this case is 
observed. The current through the rods is then again 
reversed, and readings taken as before and so on. By sub- 
tracting the mean of two consecutive readings obtained with 
the current in the same direction from that with the current 
in the opposite direction, the mean alteration of temperature 
is obtained. Since the current is reversed in the two rods, 
this change of temperature is four times that due to the 
Thomson effect. 
The second part of any experiment consists in measuring 
the rise of temperature produced by a current in the rods. 
In this case the rods are at the same temperature through- 
out, and by means of a lead soldered to the strip of copper 
joining them, the current is sent through each in turn. The 
current is sent through one rod, and, when the temperature 
has become steady, the difference of temperature of corre- 
sponding points on the two rods measured as in the former 
experiment; this circuit is then broken, and the current 
passed through the other rod. Sufficient time having elapsed 
for the temperatures to become steady, another reading of 
the compensating current is taken; the current is then 
changed back again to the first rod, but in this case made 
to flow in the opposite direction, and readings taken as 
before. 
The mean alteration in the difference of temperature of the 
junctions produced by a change of the current from one 
rod to the other, is then twice that due to the evolution of 
heat in either rod separately. Assuming that the heat pro- 
duced is proportional to the change of temperature, the amount 
of heat evolved or absorbed in the first experiment is obtained, 
and the value of the specific heat of electricity calculated. 
For if H, is the amount of heat produced or absorbed per 
unit length per unit time when a current C, flows along a 
