36 PROCEEDINGS OP THE AMERICAN ACADEMY. 



should be a straight line within the range of temperature here considered, 

 we get as the Thomson effect coefficient, 



v = a- -r- (273 + i (90.3+ 13.0)) = - 757 X 10~ 10 . 



Attempts to determine the Thomson effect directly in precise units 

 have, apparently, been very few. Le Roux in 18G7 * gave the result of a 

 rough absolute measurement of this effect in a certain alloy of bismuth 

 and antimony, called " bismuth of Becquerel." Battelli published in 

 1886 f the results of measurements made by himself with various metals, 

 one of which was iron. Mr. R. O. King published in 1898$ the method 

 and results of a study iu absolute measure of the Thomson effect in 

 copper, his work having been done at the Jefferson Physical Laboratory 

 of Harvard University, though the general plan of his investigation came 

 from Professor Callendar at Montreal. 



Battelli, whose method seems not capable of very great accuracy, 

 though it was probably a great improvement over that of Le Roux, got 

 data which yield the following values of v for iron : 



He concludes from his own experiments: "1. That in all the sub- 

 stances studied, cadmium, iron, antimony, bismuth, pakfong [german- 

 silver?], an alloy of 10 Bi + lSb by weight, the Thomson effect is 

 proportional to the intensity of the current. 2. That in all these sub- 

 stances, with the exception of iron, the Thomson effect [the cr, not the v, 

 of this paper] is proportional to the absolute temperature." 



The value of v obtained from Battelli's data for the temperature 53° , 

 which is very near the mean temperature, 51.7° , of the range used in 

 the experiments described in this paper, is about 0.37 of the value here 

 found for v. Battelli gives no description whatever of the quality of the 

 iron which he studied. It was "ferro." His apparent discovery that the 

 v for iron increases with rise of temperature, which requires that the line 

 representing iron in the ordinary thermo-electric diagram shall be a curve 

 growing steeper at high temperatures, has probably not been generally 



* Ann. de Cliimic et de Physique, 4 me se'rie, t. X, p. 201. 



t Atta della Reale Accademia delle Scienze di Torino, 22, 48, 539. 



\ These Proceedings, 33, 19, June, 1898. 



