32 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



the common one, as the following table, which gives the results of 

 all the fifteen days of successful observation of the main phenomena, 

 will show : 



There is a certain inaccuracy in the method, already described, by which 

 the A's are obtained from the E's. Each E has been treated as if each 

 of the two opposing couples producing it had the same mean temperature. 

 But observations to be described later, made by means of the junctions 

 E, F, etc., shown in Figure 3, indicate that the temperature of the warmer 

 bar exceeds that of the cooler by 0.055° at the mid section of the cool 

 couples, by 0.265° at the mid section of the middle couples, and by 

 0.145° at the midsection of the hot couples. These differences would, 

 because of the variation of" thermo-electric height" of copper and ger- 

 mau-silver with temperature, give appreciable values of E if there were 



