132 
PROFESSOR HUGH L. CALLENDAR ON 
is the fundamental interval itself. For this reason, the “ mean calorie,” which is 
equal to one-hundredth part of the quantity of heat required to raise the temperature 
of a gramme of water from 0° to 100° C., has been often proposed as the most suitable 
thermal unit. But in this case the calorimetric difficulties due to the large range of 
temperature are so exaggerated that the relation of the mean calorie to the practical 
unit employed in calorimetry cannot he determined with the same order of accuracy 
as the practical unit itself can be realized. This point is further illustrated in 
Section 45 below. An equivalent proposition is to select as the standard temperature 
for the definition of the thermal unit that temperature at which the specific heat is 
equal to its mean value over the fundamental interval. The objection to this is that 
it leaves the standard temperature uncertain. If, however, a definite temperature of 
either 15° or 20° C. were selected, we should have a definite unit, which would 
probably be within one part in a thousand of the mean calorie, or near enough for all 
practical purposes. 
On the whole, a range of 10° appears to be the most appropriate to adopt for the 
practical definition of the thermal unit. Howland’s results for the specific heat were 
calculated in this manner by taking the mean value over each interval of ten degrees. 
Griffiths also adopted a range of ten degrees, 15'’ to 25°, in his experiments, and his 
results may be held to refer to the mean of this range, which is also very close to the 
mean of the range, 18 c to 20°, of Schuster and Gannon’s experiments. For this 
reason, I have been in the habit for some years of expressing results in terms of the 
mean specific heat of water over the range 15° to 25° C., which is, within 1 part in 
20,000, the same as the value at 20° C. The mean thermal unit over this range may 
conveniently be called the “calorie at 20° C.,” although of course it cannot be 
practically realized except as the mean over a range. 
The preliminary results for the variation of the specific heat of water communicated 
to the British Association at the Dover meeting in 1899, were expressed in terms of 
the calorie at 20° C. I have thought it important to retain this unit in the present 
paper to avoid confusion, although Dr. Barnes and Mr. Griffiths* have since proposed 
units at 16° and 15° respectively, as being probably closer in magnitude to the mean 
calorie between 0° and 100° C. From a scientific point of view there is little to 
choose between these units, and the relation between them is known with a hiadi 
degree of accuracy ; but as a question of practical calorimetry, I think the unit at 
20° is undoubtedly superior. In all accurate calorimetric work it is necessary to 
employ thermal regulators, and there can be no doubt that from this point of view 
the range 10° to 20°, corresponding to the unit at 15° C., is too low. The range 20° 
to 30° would be better for temperature regulation than 15° to 25°, but I adopted the 
latter as corresponding to the mean of Howland’s, and Griffiths’, and Schuster’s 
experiments, and as agreeing more nearly with the mean range of an ordinary 
* More recently, Griffiths (‘Thermal Measurement of Energy,’ Cambridge, 1901) has proposed to 
adopt the calorie at 17‘5° C., or the mean value over the range 15°-20° C. 
