CONTINUOUS ELECTRIC CALORIMETRY. 
137 
with the air-thermometer, which was the most difficult part of the work. His ther¬ 
mometers have recently been compared with a mercury-thermometer standardized at 
Paris, and with a platinum-thermometer standardized by Griffiths, The result has 
been to reduce the rate of change of specific heat shown by his original calculations by 
nearly one-half, but the absolute value of the specific heat about the middle of the 
range remains almost unchanged. The accuracy of his results in terms of the 
hydrogen scale has probably been raised to 1 in 2000 on the average by this 
correction; but it must be remembered that the thermometers were compared at 
steady temperatures, and not under the actual conditions of the experiment on a 
steadily rising temperature. This may affect the validity of some of the conclusions. 
In Rowland’s experiments, the most striking difference from ours is the sharp 
minimum at 30° followed by a rapid rise. Rowland himself considered that at 30°, 
owing to the increasing magnitude and uncertainty of the radiation correction, “ there 
might be a small error in the direction of making the equivalent too great, and the 
specific heat might go on decreasing to even 40°.” It should be remembered that his 
method gives directly the total heat reckoned from the start of each experiment, and 
that the values of the specific heat over shorter ranges would be more affected by 
thermometric errors, especially near the ends of the range. If our results are com¬ 
pared with his by means of the total heat starting from the same value at 5° C., it 
will be seen that, in spite of the apparent dissimilarity in the forms of our curves for 
the specific heat, our values of the total heat do not differ from those of Rowland by 
more than 1 in 5000 over the whole range of his experiments. This is shown in the 
following table :— 
s 
Table XIII.—Values of Total Heat of Water compared with Rowland. 
Temperature, 
Centigrade. 
Total Heat, 
Callendar and Barnes. 
Total Heat, 
Rowland. 
o 
5 
5-037 
5-037 
10 
10-056 
10-058 
15 
15-065 
15-068 
20 
20-068 
20-071 
25 
25-065 
25-067 
30 
30-060 
30•057 
35 
35-052 
35-053 
(43.) The Method of Mixture. Ludin. 
The experiments of Bartoli and Stracciati (‘Boll. Mens, dell’ Acc. Gioenia,’ 18, 
Ap., 1891), by the method of mixture between 0° and 30°, gave a curve very similar 
to Rowland’s, but with a minimum at 20° C. This excessive lowering of the 
minimum may probably be attributed to constant errors inherent in their methods of 
experiment. 
VOL. CXCIX.—A. 
T 
