ON THE SPECIFIC HEAT OF WATER. 205 



the result of the ahove entirely independent series on the shorter periods gives 



J, 3 M = 4-1819. 



Thus the temperature errors on short periods with mercury thermometers seem to 

 disappear in the average in a satisfactory manner. The most recent figures previous 

 to our own (BARNES, 'Roy. Soc. Proc.,' A, voL 82, 1909, p. 390) yield a lower value 

 for this interval, viz., 



Jw 65 = 4-175. 



From the above values for the periods we obtain for the average values of J from 

 C. to any temperature 6 the expression 



J' = 4-2085-0-0015110+0-000026110 a -0-00000012250 3 . 



This expression gives values which agree with the observed values within 1 part 

 in 4000. 



To obtain the series of values of J at various points from the above expression 

 we have 



This gives the value of the 15 C. calorie as 



4'179. 

 BARNES' most recent value -for the 15 C. calorie is 



4-184, 



and, indeed, the only value lower than ours which has been obtained by previous 



workers is the figure 



4-174, 



which results from the mechanical method employed by JOULE in 1878. We are 

 inclined to attribute these differences from BARNES and other observers who have 

 employed the electrical method in part to the thermoid effect, to which reference has 

 already been made, and which appears to have been hitherto unnoticed. 

 We regard our fundamental figure, that is to say, 



Ju 55 = 4-182, 

 for the mean value of J from 13 C. to 55 C. as being probably correct within 



0-001. 



2. Description of the Calorimeter. In fig. 1 is shown the general arrangement of 

 the calorimeter, with the inflow and outflow arrangements installed which were 

 necessary for the continuous flow series of experiments. A is a Dewar vessel of clear 



