Thermal Conductivities. 21 



the two surfaces of the crystal can also be found,* We thus 

 have a determination of the conductivity of the crystal in a 

 direction parallel to the axis of the disc by a method which 

 follows very closely the definition of conductivity. If the 

 saw-dust were an absolute non-conductor the method 

 would agree exactly with the definition. 



In the above the isothermal surfaces have been assumed 

 plane and perpendicular to the axis of the bar, but in the 

 experiments the temperature at five points of the outer 

 cylinder were also determined, so that each experiment 

 gives sufficient data for determining the distribution of 

 heat throughout the whole space within the outer cylinder. 

 I did not consider, however, that the accuracy obtained in 

 the observations would warrant the carrying out of the 

 calculations on these rigid lines, especially as no tables exist 

 of one of the functions which enter into the calculation .f 



The conductivity of the brass bar was determined by 

 Forbes's method, with the modifications suggested by me in 

 a previous paper.]: Two experiments are necessary, one 

 determining the outer and the other the relation between 

 the outer and inner conductivity. 



Cooling Experiments to determine the Outer 

 Conductivity. 



In these experiments one of the bars used in the crystal 

 apparatus is heated to 100° in an air bath and then allowed 

 to cool in air, the temperature being observed by thermo- 

 junctions soldered to the middle of the bar. 



Writing m for the mass of the bar cooled, s its surface, 



* The main features only of the calculations are entered into here. In the 

 actual calculation corrections are applied for the thin layer of mercury between 

 the crystal and the bar, for the variation of dvjdx within the crystal, &c. , the 

 amount of these corrections being determined by special experiments. 



+ I refer to the Bessel's function of the second kind and zero order for 

 unreal values of the argument. 



XPhil. Mag. (5) xxviii. p. 442. 



