162 
DR. H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER 
I have described in another place, and have shown that they cannot be relied on, as 
an accurate laboratory standard, to quite the same order of accuracy as a Clark 
cell, although as a commercial instrument they have distinct and unquestionable 
advantages over the Clark cell. 
The method of keeping the Clark cells at a constant temperature has been already 
described in my earlier papers. Briefly it consists of a water thermostat with gas 
regulator, which is capable of maintaining the temperature constant to '02° C. over 
extended periods. Whenever one of the experiments on the specific heat of water 
was performed, the bath was set to regulate as near 15° as possible, and throughout 
hardly ever varied more than '01° or '02°, unless some sudden change in the gas- 
pressure or water supply introduced a disturbance of too sudden a nature to be at 
once rectified by the regulator. The bath was supplied by a stream of water from a 
constant-level head through a spiral of copper tubing about 2 millims. diameter, and 
was heated by the gas flame, controlled by the regulator, as it passed through. 
During the winter, the water-supply in the laboratory was always between 8 C and 
10° C. at the place where the bath was located, so that there was no difficulty in 
maintaining the bath at 15b During the summer, however, the water sometimes 
reached 18° or 20°, and it became impossible to keep the bath at 15° without running 
the inflowing water through an ice tank before it entered the bath. As this entailed 
considerable trouble, the bath was allowed simply to take the temperature of the 
inflowing tap water, and rose and fell in temperature slowly with it. There was no 
sj3ecial object after all in keeping the cells at 15°, on account of the accuracy of the 
temperature coefficient, and the complete agreement of all the cells with one another 
at all the temperatures of comparison. The temperature of the bath was taken with 
a Geissler thermometer reading to ‘01°. This thermometer was reduced to the 
nitrogen scale by comparisons, with a platinum thermometer, made both by Professor 
Callendar and myself in 1896. It has seldom varied more than a few degrees 
either way from 15° since then, and as it was a somewhat old thermometer at the 
time of comparison with the platinum, it is unlikely that its readings have changed 
much since. Moreover, our later tests on the temperature coefficient made with this 
thermometer and thermometers calibrated by it, have agreed so well with the 
earlier measurements that there is no reason to doubt the correctness of its readings. 
The comparison of the E.M.F. of the different cells was made on a specially 
constructed potentiometer, but as it has already been described it will be unnecessary 
to more than mention it here. Special attention was given to having the readings 
sufficiently sensitive to the order of accuracy we attempted, and defective insulation 
was amply guarded against. For differences in E.M.F. the potentiometer read 
directly in millivolts, at the rate of '01 mv. for each millimetre of scale. A 
6000-ohm galvanometer in the circuit was sensitive to a scale distance a little less 
than 1 millim. 
In Table I. I have arranged the complete series of comparisons made on six of our 
