98 PROFESSOR HUGH L. CALLENDAR ON 
the scales of the constant-volume and constant-pressure thermometers, as ve have little 
direct experimental knowledge of the relation of the scales at these temperatures. 
It may be equally due to systematic differences of reduction of the observations, 
as for instance in the application of the correction for the expansion of the 
envelope of the gas-thermometer. The correction applied by Chappuis,* obtained by 
extrapolation of a formula deduced from observations over the range 0° to 100° C., 
is much larger than similar corrections found by Callendar and Bedford for hard 
glass and porcelain from observations over the range 0° to 600° C. The lower value 
of the boiling-point of sulphur obtained by Callendar and Griffiths in 1890 
was confirmed by Callendar with a different instrument in 1893, and later by 
Eumorfopoulos. Since the uncertainty of the correction for the expansion of the 
envelope is so great (Callendar, ‘ Phil. Mag.,’ December, 1899), it was decided to 
adopt the older value, which has been in use for 10 years, rather than to attempt 
a special correction based on the probable scale difference of the constant-pressure 
and constant-volume thermometers. 
The uncertainty of this correction is mainly due to the difficulties of gas 
thermometry. The scale of a platinum-thermometer constructed of pure wire is 
so easily and so accurately reproducible, that it appears practically certain, as I have 
already explained at some length in a previous paper (‘ Phil. Mag.,’ December, 1899), 
that it would afford a more convenient standard of reference than the hydrogen 
thermometer for scientific purposes. By employing a standard difference-formula, 
such as (1) for reduction of platinum temperatures to the absolute scale, we should 
obtain results in sufficiently close agreement with the thermodynamical scale for all 
practical purposes, and we should be saved the trouble and confusion incidental to 
small uncertain corrections. From this point of view it would be more scientific to 
omit any further reduction to the hydrogen scale, but it may be of interest to 
indicate the nature and the probable magnitude of the correction. 
Since the parabolic difference-formula (1) was established and verified by means of 
observations with air- or nitrogen-thermometers, it would be most natural to assume 
that the scale obtained by its application coincided very closely with that of the 
nitrogen-thermometer, and to reduce the results to the hydrogen scale by the 
application of the table of corrections given by Chappuis, deduced from the 
following formula for the difference, 
t n -t h — t{t- 100) (+ 6-318 + -00889* - -001323* 2 ) X 10~ 6 . . (2). 
This formula has been applied to our results by Griffiths (‘ Thermal Measurement 
of Energy,’ Cambridge, 1901), who gives a table of the corrected values. But there 
* Chappuis has recently (‘Phil. Mag.,’ 1902) accepted the results of Holborn and Day for the 
expansion of Berlin porcelain, which reduce his value for the boiling-point of sulphur to 444° "7 C.— Added 
March 11, 1902. 
