Measurement of the Peltier Effect. 455 



gradient towards the junction, and therefore, to compensate 

 this, the current through the fine wire would be adjusted so 

 that the temperature gradient, as indicated by insulated 

 thermo junctions, vanishes in each bar. The Joule effects 

 in the bar and leads 4 mm. thick were considered to be 

 negligible, and to have no disturbing effect on the tempera- 

 ture gradients in the bar. 



If I = the current through the bar, 



£ = the current through the heating coil, 



<? = the potential difference at terminals of heating coil. 



Then 



pr= 



ei. 



At the ordinary temperature of the air, the heat evolved 

 per cm. length of the iron lead is approximately one-third of 

 the heat absorbed at the junction for a current of 20 amperes, 

 and therefore the neglect of the Joule effect is scarcely 

 justifiable. The dimensions of the apparatus are large for 

 an experiment in a limited space, and owing to the large 

 thermal capacity of the bar a considerable interval would be 

 required before the steady state and the final adjustment of 

 the compensating current were attained. 



In the following method the two junctions of the copper 

 with the bismuth could not be isolated, and the apparatus is 

 equivalent to a duplication of Pellat's arrangement with 

 many subsequent advantages. 



If a current be sent round a circuit of two metals, then 

 the temperature difference between the junctions, arising 

 from the Peltier effects, can be made to vanish by a con- 

 tinuous supply of heat to the cooler junction. In the ideal 

 case, when the junctions are equal in all respects, the rate at 

 which heat must be supplied to the cooler junction is equal 

 to twice that at which it is evolved at the warmer junction. 

 The thermal conductivity of copper is much greater than 

 that of bismuth, and therefore it is preferable to supply the 

 heat by the passage of a current through an insulated fine 

 wire embedded in the copper close to each junction. The 

 presence of a heating coil at each junction renders it possible 

 to eliminate from the final result the electrical resistances 

 and the different thermal emissivities of the junctions. The 

 apparatus was also dimensioned so that, at each junction, the 

 Joule heat for the average current of one ampere was about 

 one-tenth of the Peltier heat. In the experiments of Jahn 

 with a Bunsen ice-calorimeter the Peltier heat was only a 

 small part of the measured quantity of heat. 



