CONTINUOUS ELECTRIC CALORIMETRY. 
133 
calorimetric experiment without a thermal regulator. It might be objected that 
Schuster (range 18° to 20 c ) and Ludin (range 11° to 18°) found it necessary to keep 
the calorimeter always below the temperature of its surroundings, in order to avoid 
readings on a stationary or falling temperature, which are quite unreliable with 
a mercury-thermometer, owing to capillary friction. But even in this case it would 
be better to use a regulator at 25°, to avoid any risk of cooling the calorimeter below 
the dew-point, if a range of 10° is required. 
(40.) Choice of a Standard Sccde of Temperature. 
An equally important question in the definition of the practical thermal unit is the 
choice of the scale of temperature to which it should be referred. There would 
probably be little hesitation in selecting the scale of the constant-volume hydrogen- 
thermometer at 1 metre initial pressure ; but this necessitates the further considera¬ 
tion of the secondary standard by which the practical unit is to be realized, as it 
would of course be quite impossible to employ the hydrogen-thermometer at constant- 
volume directly in a calorimetric experiment. The mercury-thermometer, which is 
regarded at present as the representative of the normal scale, is a most unsatisfactory 
instrument for accurate calorimetric work. Some of its defects in this respect have 
already been incidentally mentioned, and it is impossible to regard with confidence 
results obtained by its use under conditions so different from those of the comparison 
with the ultimate standard. The only satisfactory method of standardizing a mercury- 
thermometer under the conditions of experiment, is by comparison with a platinum- 
thermometer. In the great majority of cases it would be far less trouble to discard 
the mercury-thermometer in accurate work and employ the platinum-thermometer 
itself. The results could be nominally reduced to the hydrogen scale in the manner 
described in the present paper, with an accuracy which is limited only by the gas- 
thermometer. But though nominally expressed in terms of the hydrogen scale, and 
subject to all the uncertainties of gas-thermometry, the results would have the 
advantage of being really referred to a practical scale which could be reproduced at 
any time with an accuracy depending only on that of the original observations. 
Although it is very generally admitted that the platinum-thermometer is the most 
accurate instrument for scientific work, it is also commonly assumed that the mercury- 
thermometer is much easier to work with. This is quite a mistake if an accuracy of 
the order of '00 l c C. is aimed at. If anyone wishes to realize the incalculable simplifica¬ 
tion introduced into accurate work by the employment of platinum-thermometers 
instead of mercury, even over the range 0° to 100°C., he should try to perform 
an experiment like the present. After endless labour spent in calibrating and 
standardizing a suitable series of limited scale-thermometers, he would probably find 
all his observations spoilt by uncertainties of stem exposure and capillary friction, 
and more particularly by unknown changes of zero at the higher points of the range. 
