determining the Thermal Conductibility of Bodies. 131 



Besides which it appears to me that a rotating apparatus which 

 rubs with such velocity will itself produce heat, and hence com- 

 plicate or extinguish the phenomenon which is to be investigated. 

 The results, also, to which various experimenters have attained 

 by the formula (1) by no means agree. 



If as unity the quantity of heat is taken which heats 1 kilo- 

 gramme of water 1° C, there passes in a second through a copper 

 disc of 1 square metre superficies and 1 millim. in thickness, 

 and 1° difference in temperature between the two surfaces, 



According to Clement 0*231 



„ Thomas and Laurent 1'22 



„ Peclet with (friction of the surfaces) 19*11 



The last value, which so considerably exceeds the first two, is 

 yet, as is clear from what follows, still considerably too small. 



According to the latter method, which depends upon the ap- 

 plication of the formula (2), the process has been as follows : 

 bars of the substance to be investigated have been procured and 

 have been heated at one end until the temperature had become 

 stationary, and the temperature of the bar investigated at dif- 

 ferent parts, either by thermometers, or by the contact of a 

 thermo-electric element. This method gives greater accuracy 

 than the foregoing, but labours under the defect that it does 

 not give the value of k separately, but only the ratio between h 

 and k } by which means the value of the latter magnitude is 

 expressed in a function, namely, that of the radiation from the 

 surface, which is not known. Hence it is that the value of h is 

 variable, and depends not only on the difference from the tem- 

 perature of the space, but also on the absolute temperature of the 

 bar, as Dulong and Petit's investigations on the law of cooling 

 have shown \ it is thus evident that in this way only relative 

 values of the conducting power of the various bodies can be 

 obtained — and only this under the supposition that the bars 

 retain the same surface, and the observations be made between 

 the same limits of temperature. Principally by taking these 

 circumstances into account, have Wiedemann and Franz obtained 

 concordant results in their valuable investigation. 



Besides the above methods, others have been used which may 

 be called mixed, like that of Tyndall for the conducting power of 

 different kinds of wood, or that of Calvert and Johnson for 

 metallic alloys. For these experiments short bars of the sub- 

 stances in question were used. They were heated at one end, 

 and the heat observed which during a given time they imparted 

 to a mass of mercury or water surrounding the other. Since in 

 this case, as has been already remarked in reference to the first 

 method, the specific heat of the bars and the conductibility from 



K2 



