258 
DR. H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER. 
reduced to 1’43325 volt at 15° C. Such a reduction is necessary to bring my 
measurements into absolute agreement with Reynolds and MoorbyCs result. This 
reduced value of the Clark cell is so nearly identical with the later absolute 
dynamometer measurements as to give a most remarkable, if not coincident, 
agreement between the electrical and mechanical units. 
If we compare the value of the mean mechanical equivalent obtained by integrating 
the values obtained by Rowland between 6° and 36°, which have been recently 
corrected to the Paris Scale by a comparison of Rowland’s thermometers with the 
Paris Scale, with the integrated value over the same range from the present 
experiments, we find the difference between, Rowland’s value, 4T834 joules, and my 
value, 4T872, in terms of the Clark cell value, 1’43420 volt, equal to ’091 per cent. 
This is a difference of only 1 part in 2000, as deduced from the comparison of the 
complete curve with Reynolds and Moorby’s result, a discrepancy which, if not 
within the limits of error of our several determinations, is relatively small considering 
the great range covered by these experiments. The reduced value of the Clark cell 
according to Rowland would be 1’43355 volt, which differs from the value according 
to Reynolds and Moorby by only ’3 millivolt. Owing to the slight difference in the 
temperature coefficient of the specific heat between Rowland’s values and my own, 
the agreement of our absolute values at any one temperature will be different at 
different temperatures. At 25° our measurements, when expressing mine in terms of 
Reynolds and Moorby’s, are almost exactly coincident; at 13° my value is lower 
than Rowland’s by 1 part in 1000, but at 6° we are in agreement again. 
Of the other direct mechanical determinations which have been made recently, we 
have the work of Miculescu* in 1892, which is deserving of some mention. 
Although his work is by no means above criticism, as was clearly pointed out by 
Schuster and Gannon in their paper, it is of interest as showing the kind of error 
which may occur between measurements by the direct method, which may be at the 
same time very carefully and accurately carried out. His value, which appears to be 
a mean value between 10° and 13°, is 4T857 joules. Rowland’s value at the same 
temperature, about 11° C., is 4T94, while my own in terms of Reylnolds and 
Moorby’s value is 4T903 joules, which, although less than Rowland’s value, is 
larger than Miculescu’s. 
Perhaps the most difficult part of the comparison of the present experiments with 
the work of other observers is in relation to the results obtained by the electrical 
method used by Griffiths and Schuster and Gannon. It is at once apparent from 
fig. 17 (p. 249) that my values are widely different to the values obtained by both 
these investigators, although expressed in the same values of the units used. The 
explanation might at once be looked for in an error in either my Clark cells or 
resistance standard ; but if it is attributed to the Clark cells used in the present 
work, then the several sets of cells made at different times and from different 
* ‘ Ann. de Chimie,’ vol. 27 (1892). 
