Latent Heat of Evaporation of Benzene. 11 



the same temperature the value varied but little, the changes 

 were sufficient to affect the resulting value of %q in exceptional 

 cases. In the earlier experiments, especially from Preliminary 

 to No. VII., when the difficulty of the final adjustment had 

 not been fully realized, the values of d" were unduly large, 

 and their equivalent degree-measurements are therefore some- 

 what doubtful, owing to insufficient determinations of the value 

 of the galvanometer-swings. We feel sure, however, that from 

 No. VII. onwards any errors due to this cause must be very 

 small — certainly not so great as 1 in 50 ; for the value of a 

 swing of 100 (in terms of amillim. of the bridge-wire) would, 

 throughout a group of experiments at the same temperature, 

 vary (for example) from 1*31 to 1*34 millim.* This change in 

 value would produce no appreciable effect on 2^ when the 

 difference between the initial and final swings did not exceed 

 50 or 60, as was the case in most of our later, and better, 

 experiments. 



The value of G 9i (capacity for heat of calorimeter and 

 contents) remains the same as last year with the exception 

 that 0*1 grm. of the oil was removed at the commencement of 

 July when withdrawing the platinum thermometer for purposes 

 of re-standardization. As the value of Gg l varied with Q x from 

 301 to 323, this loss was negligible. 



The quantity Xq represents the heat absorbed by the calori- 

 meter and contents, and it should be remembered that it is 

 by the measurement of a similar quantity that the majority of 

 the determinations of thermal constants have been made by 

 previous observers. We were, however, anxious to diminish 

 the importance of this term as much as possible, for we wished 

 our values of L to be independent of any thermometric errors. 

 As shown by Table II., the average value of 2^ was only about 

 t Jq of the total heat-supply. In the experiments in which 

 the value of 2</ is less than 1 (of which there are many 

 examples), we may say that our final results are independent 

 of such errors, for, as pointed out in Paper W, the value of 

 Q s is independent of temperature measurements, since it 

 depends on the ratio of the rate of rise due to the mechanical 

 supply to the rate of rise due to an electrical together with a 

 mechanical supply (Paper W, p. 331). 



The values of C 9l at 20° and 50° used in this Paper differ 

 somewhat from those given in Paper W. This difference is 

 due to the error, previously referred to, in the value of Qs at 

 those temperatures. Thus 



the value of C 9l at 20° is reduced from 307*5 to 305*2, 

 and „ „ 50° „ „ 323*1 to 322*0. 



* 1 millim. of the bridge-wire indicated a temperature-difference of 

 about 0°-009 C. 



