CONTINUOUS ELECTRIC CALORIMETRY. 
97 
(23.) Reduction of Results to the Hydrogen Scale. 
A most important factor in the variation of the specific heat of water, as Rowland 
was the first to point out, is the correction to the absolute scale of the readings of the 
particular thermometer employed in the research. In the present case, the tempera¬ 
ture observations were taken with platinum-thermometers of standard wire, and were 
approximately reduced to the absolute scale by the difference-formula, 
t ■— pt = dt (t : — 100)/10,000 .(l). 
The recent observations of Messrs. Harker and Chappuis (‘Phil. Trans.,’ A, 
1900) with a constant-volume nitrogen thermometer have confirmed my conclusion 
that a formula of this type represents the deviation of the platinum-thermometer 
from the absolute scale within the limits of error of observation over the range 0° 
to 600° C. A similar conclusion follows from the comparison by Griffiths (‘ Phil. 
Trans.,’ 1893) and by Waidner and Mallory (‘ Phil. Mag.,’ July, 1899) of mercury- 
thermometers standardized at the International Bureau with platinum-thermometers 
from 0° to 25°, and from 0° to 50°. 
The value of the difference-coefficient d in this formula was assumed to be I *50 
from the mean of a number of observations taken at different times with different 
samples of the wire. The thermometers themselves were not directly tested, as the 
variation of the difference-coefficient for different specimens of the pure wire is in 
nearly all cases less than the probable error of a single determination. The 
particular sample of wire employed for thermometers E, on which most of the 
results depend, was tested by Mr. Tory (‘ Phil. Mag.,’ 1900) by comparison with the 
original standard wire from which it was drawn, and found to be identical. 
The limits of variation of the difference-coefficient for pure platinum wire, as 
tested by competent observers, assuming the normal boiling-point of sulphur to be 
444 0, 53 C., on the scale of the constant-pressure air-thermometer, is only P49 to 
1*51. The greater part of this variation is probably due to errors of observation 
and differences of annealing. A variation of (ROl in the value of d would affect 
the specific heat of water by only 1 part in 10,000 at 0° or at 100° C., and by much 
less at intermediate points of the range. If we had assumed the boiling-point of 
sulphur to be 445 0, 27 C., # as found by Harker and Chappuis with a constant- 
volume nitrogen-thermometer, the value of d would have to be increased to 1*54, 
which would increase the values of the specific heat of water by only 4 parts in 
10,000 at 0° C., and by 2 in 10,000 at 25° C. The difference in the above values of 
the boiling-point of sulphur may possibly be explained as due to a real difference in 
* Chappuis has recently (‘Phil. Mag./ 1902) accepted the results of LIolborn 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. 
VOL. CXCIX.—A. 
O 
