372 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



Any error in the value taken for the thermal conductivity is directly 

 felt in the determination of the change in the rate of heat conducted in 

 and out of the ends of the sections, on reversal of the heating current. 

 For this reason, the value of the thermal conductivity of copjier deter- 

 mined recently by R. W. Stewart,* namely 1.12 (1 — .001 t), has been 

 applied to the Thomson Effect calculations. This value of the thermal 

 conductivity differs very much from that previously used, the temperature 

 coefficient also, being of the opposite sign. The values of the Thomson 

 P^ffect as modified by this change are given in column III. below. 



It seems to be unlikely that the decrease in the Thomson Effect with 

 rise of temperature can be accounted for by even large variations in the 

 form of the h curve or in the value of the thermal conductivity. 



A good criterion of the general accuracy of the whole scheme of the 

 experiment is the calculation of the thermal conductivity from the sev- 

 eral quantities involved. 



Consider section 4 to 12 : — The temperature gradient, 



measured on the curve, a e, Fig. 3, is 53°. 6 Cu. per cm. 

 reduced "to degrees Centigrade per cm. 



From equation (2), page 365, differentiating we have 



dt 1 



d Gu 



at no. 4, 



d X 



This must be 



d Cu 

 At no. 4, < = 41 ; hence 



1 - .0132 (.02^-1) 

 dt 1 



d C'li 



, therefore — = 

 1.024 d X 



53.6 



= 52.3. 



1.024 



Let K = the thermal conductivity at no. 4, i. e. at 41°. Then 

 section 4 to 12 is losing heat by conduction at no. 4 at the rate of 



* Phil. Trans. Roy. Soc. Lond. 1895. 



