OP ARTS AND SCIENCES. 215 



also see that, at least for this case, the Thomson effect must be pro- 

 portional to the absolute temperature. 



To look at the problem from another point of view, suppose that in 

 the homogeneous circuit there is but one abrupt change of tempera- 

 ture. That is, the temperature varies continuously from T^ — T'j, and 

 then falls abruptly to T^. In this case the curve representing the 

 change of temperature along the curve is discontinuous at T. The 

 integral of o- tlirougliout the circuit is then not necessarily zero. The 

 part of the integral corresponding to the fall from T — T^ has in this 

 case the maximum or the minimum value of which it is capable. The 

 complete effect through the circuit is represented by 



h 



To 



If (T varies with the temperature, this expression cannot equal zero. 

 If the current is in such a direction as to absorb heat at the plane of 

 abrupt variation of temperature, the part of the general integral cor- 

 responding to the abrupt change has its maximum value ; otherwise 

 it has its minimum value. There is nothing in the equation to deter- 

 mine the direction of the current. Its intensity will obviously be the 

 same in either direction, 



/" 



crdT 



T 



being a mean between its maximum and minimum values. We have 

 tried several experiments to see if the direction of the current was 

 related to the direction of the Thomson effect. It was soon perceived, 

 however, that in the method employed many complicating causes 

 masked the effect sought. From the fact that the equations show that 

 a current can take jjlace in either direction, we must conclude that 

 there are most probably equal currents in opposite directions, making 

 the resultant cui-rent equal zero. This is especially so in the perfectly- 

 symmetrical case first considered, where there are two abrupt changes 

 of temperature. We are not aware that any experiments have been 

 made on a circuit arranged in this way, and in fact it would be most 

 difficult to realize the arrangement practically. In the second case, 

 however, experiment shows that there is a current, and if there is 

 anything in the nature of the metal which prevents the plane of ab- 

 rupt variation of temperature from acting as the reservoir and refriger- 

 ator at the same time, then the equations explain the current. We 



