72 Mr. H. H. Poole on Thermal Conductivity and Specific 



interior o£ the rock is colder when the temperature is rising 

 and vice versa. The correction for the central hole amounts 

 to about 6 per cent. 



From this equation, knowing K, we can find C and hence 

 obtain the specific heat. Both rising and falling temper- 

 atures were employed, the observations with the latter being 

 obtained at the conclusion of a determination of K by 

 turning off the central and furnace currents and observing 

 6 and A as the rock cooled. For observations with rising 

 temperatures the furnace regulator was, at first, set to a 

 definite value and observations taken as the temperature 

 rose : as, however, the rate of rise becomes slower and 

 slower as the temperature rises, it was found better to 

 arrange a clock which slowly turned the torsion head of the 

 regulator so as to increase the power; by this means the 

 rate of rise was kept fairly uniform. 



The method adopted in measuring — and A was as 



follows : — At a given instant the E.M.F. of the siugle couple 

 measuring 6 was read, then almost immediately afterwards 

 the thermometer giving the temperature of the mercury key 

 which forms the cold junction ; the connexions of this key 

 were then changed, the multiple couple being connected to 

 the potentiometer and read exactly two minutes after the 

 single couple, which was then read again at the end of another 

 two minutes and so on. Thus readings of the single couple 

 were obtained at exact intervals of four minutes with 

 readings of the multiple couple at the middle of each 

 interval. 



Let SE 2 be the difference between successive readings of 

 the single couple corrected for the temperature of the key 

 and potentiometer errors, and E 4 be the reading of the 

 multiple couple also corrected for potentiometer errors, then 

 if P be the thermo-electric power corresponding to the mean 

 temperature (which approximately corresponds with the 

 temperature at which E 4 was observed) the average value 



of -=r throughout the interval 



= ■ p - and A = -£—-, since there are four junctions; 



hence we find that 



K SE, 



