2(38 K. Y AM AG AW A 



C(>iil(l be easily found. This would be :ui exceedingly Viiiuablc nietliod, 

 if one could work far from any disturbing sources. In tlie present 

 case, every precaution was taken to remove any movable [)iece of iron 

 and otber strongly magnetic substance from (he neighborhood of the 

 g'al\ aiintiieter. Also to diminish torsional set, the galvanometer- 

 mirror was suspended l)y a real s[)ider line ; nevertheless, the zero- 

 poinl moved as much as o divisions in öO minutes, lîelow is given 

 the result obtained bv this method. 



Owing to the necessity of using small wires for junctions, the 

 resistance of the circuit was considerable, anujunting t(j more than (! 

 ohms, so that the galvanometer was not so sensitive as might have 

 been desired. The difference of a degree in the two junctions gave a 

 little over 10 divisions of deflection. The thermometer used was 

 graduated to flfths of a degree centigrade. The total range of the 

 temperature in the second experiment was something below 45° C. 

 and the total range of the galvanometer reading was a little below 

 480 ; we can only be sure of one division in reading ; so the accuracy 

 of the result can only extend to something like one or two tenths of 

 a degree. 



When the temperatures of points e(jually distant from the centre 



of a sphere are e(|ual, the differential equation to be satisfied is 



la _ K / l-u _2_ lu \ 

 It ep \ Sa;- X Ix J 



u = temperatin-c 



e ^^= specific heat 



p^ density 



X = conductivity 



t = time 



X = variable radius 

 The well-known solution of this ecjualion applicaljle to the present 

 case is * 



• See Thomson's Art. on Heat, in Ency. Brit., or Thomson's Math, and Phy. Papers, Vol. 

 II. paj;o 41. In thu former V is put equal to n k, a misiJrint, which is corrected in the latter. 



