134 
PROFESSOR HUGH L. CALLENDAR ON 
It is no exaggeration to say that this investigation could not have been carried out 
successfully without the direct employment of platinum-thermometers. 
(41.) The Work of Regnault. 
The work of Regnault (‘ Memoires de lTnstitut,’ Paris, 1847) on the specific heat 
of water by the method of mixture at temperatures between 110° C. and 190° C., 
represents the only evidence at present available on the variation of the specific heat 
at high temperatures. It was carried out on a large scale with his usual 
experimental skill, and is undoubtedly entitled to great weight; but there are 
large discrepancies in several cases between the recorded data and the calculated 
results, and very little is known of the scale of the thermometers employed in 
the work. 
The total capacity of the calorimeter employed was about 110 kilogs. At each 
experiment 10 kilogs. of cold water was drawn off, and the same amount of hot 
water, at a known temperature, was introduced from a high-pressure boiler. The 
rise of temperature varied from 8° to 15°, according to the initial temperature of the 
hot water, and was read to the hundredth of a degree. The discrepancies in the 
observations taken at any one point under similar conditions are of the order of 
•5 per cent. The results have been recalculated from the data columns by 
J. M. Gray (‘Proc. Inst. Mech. Eng.,’ 1889), who finds in several cases disagree¬ 
ments amounting to from 2 to 4 per cent. In all these cases it appears that the 
total recorded quantity of water considerably exceeded the capacity of the calori¬ 
meter. As the result of a careful enquiry, Gray concludes that Regnault’s 
calculations are probably right, and that the apparent discrepancies in the data 
arise from deficient information or erroneous entries. 
It would appear to be hopeless at the present time to make any corrections to the 
readings of the thermometers employed for observing the temperature of the hot 
water, beyond those which Regnault himself applied. The principal source of 
uncertainty lies with the calorimetric-thermometers, as Regnault was unable to 
obtain any consistent evidence of deviation from the scale of the air-thermometer 
between 0° and 100° C., and did not apply any correction from the mercurial to the 
absolute scale. Moreover, no allowance was made for the temporary depression of 
zero to which French “ cristal " thermometers appear to be particulary liable. Some 
thermometers of this class show a depression of as much as half a degree after 
heating to 100° C. If such a thermometer is suddenly immersed in steam, its 
reading rapidly rises to a maximum, and then slowly falls for half an hour or so 
towards its steady reading. In using a mercury-thermometer in a calorimeter, 
where it is exposed to a sudden rise of temperature, the effect of this phenomenon 
is to make the observed rise of temperature too large. The maximum reading is 
always taken, with additive corrections to represent the subsequently observed rate 
