124 
MESSRS. W. R. BOUSFIELD AND C. ELSPETH BOUSFIELD 
and consequently the zero of the scale. Possibly a pneumatic dash-pot would serve 
for accurate experiments or even a pair of light springs one on each side of the 
dash-pot piston. 
(c) The pointer had a rather broad mark which made it difficult to see when the 
arm was exactly horizontal in order to obtain the zero correction. A finer mark or 
a fine pointer would probably obviate this. 
The zero correction was always taken at the end of an experiment. The balance 
weight was always placed at the exact graduation 376 or 377 corresponding to about 
5 amperes current and 188 watts. This exact graduation could be read with 
accuracy, but, owing to the breadth of the pointer mark, the zero corrections could 
only be read to within about +0T. The result of this was appreciable in comparing 
duplicate observations made under circumstances exactly similar except for the zero 
correction. Thus, for example, in Table III., comparing Experiments 490 and 498, 
where the weights of water employed were exactly the same, we see that the figures in 
the second set are lower throughout. This must be attributed to the zero error of the 
balance. The irregularities must mainly be attributed to the thermometer readings) 
as a difference of 0°'01 C. on an interval of 13° C. would make a difference of 10 on 
the above figures. In the specific heat measurements to be described duplicate 
experiments were always made, and if these were not sufficiently concordant the 
experiment was again repeated. 
The electrical arrangements were as shown in fig. 4 of the former paper ( loc. cit.,- 
p. 209), except that the auxiliary resistance M x and its shunt circuit were omitted. 
The value of the ampere balance readings was known from the former experiments 
to within +1 in 5000 when passing a current of about 5 amperes. To calibrate the 
watt balance (Experiment 416) the calorimeter was charged with water, and a 
current of about 5 amperes was passed through the apparatus. At the same time a 
steady flow of water was passed through the calorimeter until a definite regime was 
obtained at a definite temperature. The watt balance graduation being fixed at 376 
the current and the ampere balance were adjusted until the two balances were both 
in equilibrium together. The current was accurately known from the ampere 
balance and the resistance from the mercury thermometer heater, the shunt 
resistance R,, of the watt balance being also observed. The mean of a series of 
experiments showed that the watt balance graduation required at this point a 
correction on the reading equal to + 0'5. The zero correction, to correct for 
alterations of the zero from time to time, had to be added to this. To correct for 
the portion of the energy expended, not in the calorimeter but in the leads and 
primary circuit of the watt balance, it was always necessary to multiply the total 
energy measured by the factor M 2 /(M 2 + p). This was practically constant, the 
values of the factor being 
0 ° C. to 13° C. 13° C. to 26° C. 26° C. to 39° C. 
0'99032, 0-99043, 0‘99054. 
