DE. H. T. BARNES ON THE CAPACITY FOR HEAT OF WATER 
2(30 
affect the accuracy of the relative results, as regards the variation of the sj:)ecific heat 
of water. 
It is interesting to compare the absolute value of the Clark cell obtained by 
assuming Griffiths’ absolute value of the mechanical equivalent at 15°, and to 
express my mean value in terms of his experiments. By so doing the absolute 
value becomes 4 '1975 joules, which differs from Reynolds and Mooeby’s value by 
•34 per cent, or, assuming the error to be due to the Clark cell, equal to ‘17 per cent, 
on D4342, which would reduce this value to 1’4318 volt at 15°. This, however, even 
referred to the lowest of the latest absolute determinations, seems to be too low 
a value, by as much as l millivolt, to be reconciled with the most probable true value 
of the Clark cell. 
It might be thought advisable, in view of the uncertainty in the electrical units, to 
accept Rowland’s corrected values and express the present series of experiments in 
terms of his results, which would give a mean value quite sufficiently in accord with 
Reynolds and Mooeby’s mean determination. This could be done either from the 
integrated value over the range of his experiments, which would tend to eliminate 
errors in his method at the two extremes of the range, or by accepting his absolute 
value at a temperature where he could obtain the most accurate measurement. The 
present experiments over the range between 4° and 60° have already been published 
(‘ B.x4_. Report,’ 1899), and were referred to Rowland’s absolute measurement at 
20° C., but I think that the uncertainty in the thermometric standards used by 
Rowland at that time do not warrant an accuracy greater than 1 part in about 
2000, and that the mean result over the complete range of temperature referred to 
Reynolds and Mooeby’s determination is more near the truth. 
The value of the mean specific heat between 0° and 100° C., 4 - 232 joules, obtained 
by Dieteeici (‘ Wied. Ann.,’ vol. 33, p. 417, 1888) in terms of the electrical units, is 
obviously too large to be accounted for by an error in the electrical units, or to be 
reconciled with the direct determination of Reynolds and Mooeby. The curve 
obtained by Baetoli and Steacciati (‘ Beiblatter,’ vol. 15, p. 761, 1891) for the 
variation of the specific heat of water between 0° and 30° by the method of mixtures 
in terms of a thermal unit at 15° C. passes through a minimum point at 20° C., above 
which it shows a far too rapid increase in the specific heat to be reconciled with 
measurements extending as far as 100° C., unless the values pass through a maximum 
point. 
In 1895, Ludin (Dissert. Zurich and ‘Beiblatter,’ 1897) determined the variation 
of the specific heat between 0° and 100° by the method of mixtures and showed 
a minimum point at 25°, but also a maximum point at about 80°. His results are in 
good agreement with the present series of experiments over a range 0° to 25 , as 
sho wn in fig. 17 (p. 249), where I have plotted them in terms of a mean unit between 
0° and 100° C. The excessively low minimum point shown by Baetoli and 
Steacciati and by Ludin respectively, both using similar methods, suggests a 
