of Iron and German Silver » 499 



not exercise any prominent influence. That, notwithstanding 

 the great exposing-power of soot, the quantity H does not 

 assume a greater value is to be attributed to the bad conducti- 

 vity of soot for heat. 



III. Drawn German-silver rod [annealed). 

 Length . . Z= 230-4 C = 0-0944 



Diameter. . = 7*622 millims. D = 8-621 



C and D were, as before, determined specially. 

 Odd periods. 



K=8-404 H = 000318 



K = 8-405 H = 0-0031 6 



Even periods. 



K = 7-832 H = 0-00291 



K = 7-791 H = 0-00291 



K = 7-817 H = 0-00295 



Mean value from the even periods. 



K= 7-813 H = 00292 



Mean value from the odd periods. 



K = 8-404 H = 0-0031 7 



Mean of the whole. 



K = 8-108 H =0-00304 



These latter values refer to the mean temperature A of the 

 German-silver rod, 



A = 31°-25 C. 



The determination of the mean temperature to which the 

 values of K and H were referred can only be regarded as a rough 

 approximation. A step would be taken towards greater accuracy 

 if, in constructing the general equations for the propagation of 

 heat, K and H were introduced as functions of the temperature^ 

 putting K=Ko(l + «0} H = Ho( 1-1-/5^). In fact, presupposing 

 this, the stationary condition in a rod whose ends are kept at 

 constant temperatures can be determined generally*. The solu- 

 tion consists in an integration which can be effected in a detini- 

 tive form for a rod unlimited in one direction. Great difficulties, 

 however, oppose an experimental use of this solution for the de- 

 termination of a, /3, Kq, and Hq. 



A knowledge of the dependence of K and H on the tempera- 

 ture would also be desirable, in order to test the conjecture of 



* Poisson has given an approximate solution of this problem {TMorie 

 matMmatique de la Chaleur, p. 255). 



2K2 



