358 PROCEEDINGS OF THE AMERICAN ACADEMY. 



ature. By adjusting the resistance in A and B the potentiometer can 

 be made to correspond to the resistance of any section of the experi- 

 mental wire. For instance, suppose the galvanometer lead gi is con- 

 nected with lead no. 9, and that under these circumstances a balance 

 of the galvanometer is obtained with g^ reading nearly 100 on the 

 potentiometer ; then suppose that a balance for no. 8 is obtained at 

 nearly on the potentiometer. The difference between the potentiom- 

 eter readings, that is, nearly the whole length of the potentiometer, will 

 then correspond to the resistance of the section 8 to 9. Repeating the 

 operation with the heating current in the opposite direction, a some- 

 what different length on the potentiometer will be found to correspond 

 to the resistance of the section 8 to 9. The difference between these 

 two lengths corresponds to the change of resistance of section 8 to 9 on 

 the reversal of the current, on the same scale that the mean of the two 

 lengths corresponds to its absolute mean resistance. Hence the change 

 of temperature of any section of the wire, on reversal of the current, 

 can be calculated. Also if the mean absolute resistance be determined, 

 from a separate experiment, the absolute change of resistance and hence 

 the diff'erence in the rate of generation of heat, by the mere overcoming 

 of resistance, in section 8 to 9, on reversal of the current, can be 

 calculated. If from another experiment the relation between h, the 

 heat dissipated by radiation and convection from any part of the sur- 

 face of the wire, and the temperature of that part be known, then the 

 difference in the rate of dissipation of heat by the section on reversal of 

 the current can be calculated. There remains, for the complete deter- 

 mination of the Thomson Effect in section 8 to 9, to be found the 

 change in the temperature gradients at the points 8 and 9, due to the 

 small change of temperature on the reversal of the current. 



Let us suppose that by the method described above the change of 

 temperature on reversal of the current has been found for every section 

 of the bar, i. e. 3 to 4, 4 to 5, 5 to 6, etc. If the results are plotted a 

 curve such as c d, Fig. 3, is obtained. Let the curve a e represent 

 the distribution of temperature along the wire. The curve c d and 

 a e are drawn to different scales. When the largest ordinate of a e 

 was 235° that ofcc?was about 1°.2. The curve cd being a temper- 

 ature difference curve for corresponding points on a e, a little reflection 

 will show that the tangent to c d a,i any point, measured in degrees per 

 cm., will be the change in the tangent to a e at the corresponding point, 

 on the reversal of the current. Hence by multiplying this difference 

 tangent by the area of cross section of the bar and by the thermal con- 



