254 
DR. H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER 
range. The best average of the values of h given in this series is a line represented 
in the form 
H/, = 11/+ -000708 (h - t), 
where t is the temperature corresponding to the measurement of Eb, and t x is the 
temperature corresponding to the value of Eb,. From further consideration of the 
changes in the value of h from the other series, this appears to represent the tem¬ 
perature change of the radiation not only for Calorimeter C, but for Calorimeter E, for 
the two determinations between 30° and 20°. 
Taking the different values of Series II., we have, on tabulating the values of the 
heat-loss, both observed and calculated, and accepting the value at 22° for H/ in the 
expression given above, the following values :— 
Temperature. 
H observed. 
H calculated. 
22-16 
•04619 
•04619 
31-40 
•05334 
•05273 
32-17 
•05282 
•05328 
41-02 
•05939 
•05954 
45-49 
•06306 
•06271 
49-68 
■06669 
•06562 
54-61 
•06867 
•06916 
59-80 
•07220 
•07285 
32-81 
•05429 
•05373 
30-54 
•05364 
•05212 
The values at 50° and 55° are not very consistent, but it will be remembered that 
the measurements at these points are not so trustworthy owing to the variation in 
the experimental conditions. 
On returning to 30°, as seen by the last two readings, the value of k has increased 
in both cases. These two values were obtained with a rise of temperature of 11 ° and 
5° respectively. 
In regarding these large variations in the heat-loss from time to time, it must be 
again emphasised that the value of the specific heat of water, owing to the method of 
treatment, in no way depends on the absolute value, but only on the constancy 
throughout the period of an experiment. 
To prove that this was so, the order of one of the experiments in Series VIII. at 
the higher points was reversed, and instead of taking the observations for the large 
flow first, as was followed for all the other experiments in this series, the observations 
for the small flow were obtained before those for the large flow. By this, any gradual 
change in the heat-loss during the time of the experiment would have produced an 
effect on the value of d in an opposite direction to the values given by the other 
experiments, and would have produced twice the error. 
