232 
DR. H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER 
the 5° and 8° show that the low 7 er value given by the former must be exceptional. 
Here is an error of nearly 1 part in 1000, to be accounted for by assuming either that 
for the small flows the heat-loss is not proportional to the rise in temperature, in 
which case the value of the heat-loss per degree rise increases with rise of temperature, 
and the value for the 8° rise or the value for the 5° rise must be regarded as 
exceptional, or that the error in question was due to some uncertainty at that time in 
the experimental conditions. The latter must be regarded as the most probable on 
account of the greater difficulty of measuring so small a rise of temperature to the 
same order of accuracy. Moreover, the second 15-minute interval show’s a decided 
increase, and v r ould possibly have attained the correct value given by the mean of 
the other readings if the experiment had been further continued. An error of only 
'001° on the 2° rise would account for the error in the second interval. 
Besides the observations I have just given, which were selected from a series of 
trial experiments on the flat heating-wire, a large number of the other experiments 
w r ere taken with rises of temperature ranging from 1° to 12°. These are detailed in 
the tables to be given later, and include results with the central heating-wire as w 7 ell. 
It was a matter of convenience onl} 7 that governed my choice of a rise of temperature 
for any experiment, and it sometimes happened that it w r as more convenient to change 
the mean temperature of an experiment by changing the rise of temperature in the 
v 7 ater rather than by altering the inflow temperature—for example, in obtaining a 
measure of the specific heat in the neighbourhood of the zero point, where it was 
impossible to maintain the inflowing water at a temperature lower than 0° C. 
Heat Capacity of the Calorimeter . 
Although nearly always negligible in the calculation of results, the thermal capacity 
of the calorimeter is of value in showing the size of error introduced by a change in 
temperature in the calorimeter water. To determine this, the electrical supply was 
suddenly cut off from the calorimeter at a given moment and the rate of fall in tem¬ 
perature recorded. This was done for both the limits of flow 7 used in the present 
work. The lag, on breaking the circuit of the thermometer before it commenced to 
fall, was in both cases not more than 2 or 3 seconds. If 9 be the temperature indicated 
by the outflow 7 thermometer above that of the inflow thermometer, then at an} 7 time 
after shutting off the heat supply, the value of 9 w 7 ill be approximately 9 = 
from which 
cl9/dt = hae~ at = a9. 
But C d9/dt = c/H/di, where C is the thermal capacity of the calorimeter, and H is 
the total quantity of heat carried off by the water. 
Writing JQd for dW/dt. C a9 = JQ 9, and C = JQ/a. 
The followdng set of observations w 7 as obtained for Calorimeter C :— 
