of the Heat-conducting Power of Mercury. 491 



between 40° and 160° is perfectly constant. The small devia- 

 tions which in the very sensitive lower temperatures are hardly 

 avoidable, lie principally a little more above the mean of all the 

 values 7*525. 



T— t' 



The cause of the decided decrease of the values of T . from 



m — v 



160° upwards (where the horizontal line runs across the Table, 

 and the numbering ceases) is simply that there isolated air-bubbles 

 became visible, developed at so high a temperature from the up- 

 permost mercury of the tube, and having been till then absorbed. 

 These air-bubbles of course cause too high a temperature m to 



T — t' 

 be read off, and thereby lower the values of -. ; for boiling 



the mercury in the tube was not very feasible, since the thermo- 

 meter T, extending to only 200°, formed a closure to it. On 

 this account I had, before the experiments, heated the mercury 

 of the tube continuously only to 200°, in order to expel the ab- 

 sorbed air. This was therefore satisfactorily accomplished up to 

 temperatures T of 160° ; for the immediate decided fall of the 



T—t' 

 values of - f as soon as air-bubbles were seen, sufficiently in- 

 dicates that this circumstance played no important part so long 

 as the values showed no noticeable diminution. Moreover, in 

 the experiments Nos. 42-45, a remote intimation of this beha- 

 viour is perhaps given. 



Accordingly, in this apparatus the behaviour of mercury can- 

 not be tested beyond 160°. 



In order to make more apparent how variable the heat-con- 

 ducting power of mercury, in opposition to the results we have 

 gained, would be, if it corresponded to the electric conduction, 

 in the following Table the values shall be given on the hypothesis 

 of the variability of both conductivities being equal. As an 

 approximate mean of the various values given (E. Becquerel* 

 0-00104, Miillerf 0*00119, Schroder van der KolkJ 0*00086, 

 and Siemens § 0*00098) for the variability /3 with mercury, 0*001 

 shall be taken, and we will put a=/3. With this and with the 

 once for all adopted mean value t' = 17°, the quotient iu the 

 middle column, 



y/3 + 2aT + ^ / 

 1+aT 



is calculated, which only in the case of t'=0 in the above for- 

 mula contains the variability. The last column then contains, 



* Pogg. Ann. vol. lxx. p. 248. t Ibid. vol. lxxiii. p. 440. 



X Ibid. vol. ex. p. 476. § Ibid. vol. cxiii. p. 104. 



