PRACTICAL STANDARDS FOR ELECTRICAL MEASUREMENTS. 245 
boiling-point of sulphur on the nitrogen scale was also made. Three independent 
sets of determinations of this point gave the following results: 
(1) Platinum thermometer K. 9, and glass cas-thermometer, 445°97, 
(2 * $ Keo; porcelain Fr 445°26. 
(3) ” ” K. 8, eB) ” 445:29, 
The mean of these, 445°-27, representing the temperature on the scale of the 
constant volume nitrogen thermometer, differs only 0°-7 from that found by 
Callendar and Griffiths for the same temperature expressed on the constant 
pressure air-scale. ; 
If, for the reduction of the platinum temperatures in our comparisons, we adopt 
the parabolic formula, and the value of 5 obtained by assuming our new number 
for the sulphur point, we find that below 100° the differences between the 
observed values on the nitrogen scale and those deduced from the platinum ther- 
mometer are very small, seldom exceeding 0°01, and that even at the highest 
temperatures the difference only amounts to a few tenths of a degree. 
APPENDIX IV. 
On the Expansion of Porcelain with Rise of Temperature. 
By T. G. Beprorp, B.A. Cambridge. 
In direct comparisons of the scales of temperature given by air and by platinum- 
yesistance thermometers at high temperatures, the expansion of the porcelain 
envelope enters as a small correction. 
In the experiments described in this paper, a direct determination of the linear 
expansion of porcelain was made at temperatures from 0° C. to 830° C. The 
method used was essentially the same as that described by Callendar (‘ Phil. Trans.’ 
1887, A. p. 167). 
On a tube of Bayeux porcelain two fine transverse marks were made at a 
distance about 91:3 cm. apart. ‘The tube was heated to as high a temperature as 
possible in a gas furnace, and was then slowly cooled by diminishing the gas 
supply. During cooling the variation in the distance between the marks was 
determined by a pair of reading microscopes which were mounted on stone blocks 
and not touched except by the screw-head during an experiment. The readings of 
the microscopes for a standard length (a glass tube kept in melting ice) were taken 
at intervals, 
The temperatures corresponding to the length measurements were deduced 
from the resistance of a platinum wire running from mark to mark in the axis of 
the tube and supported on a plate of mica. The resistances in ice and steam were 
taken after each exposure to a high temperature. The sample of platinum wire 
from which the piece used in these experiments was cut is known to have a value 
of 5, in Callendar’s formula, from 1:50 to 1:51. The value 6= 1-505 was used, and 
thus a direct determination of the resistance at the temperature of boiling sulphur 
was avoided. An error of ‘01 in 6 causes an error of less than 1° in the calculated 
value of ¢ at 1,000° C. 
Four main experiments were made ; the results were plotted and are reproduced 
on the accompanying slide. 
From 0° C. to GOC° C, the results are represented fairly well by the formula 
1,=1, (1+ 34:25 x 10-%¢ + 10°7 x 10-1?). 
Above 600° C. the points are more erratic, but still do not depart far on either 
side from the curve given by the above formula. 
A length of about 6 cm. at either end of the tube was not directly heated by 
the furnace. Hence there is an uncertainty due to the ends (greater at the higher 
temperatures), since the coefficient of expansion varies with the temperature. 
For cubical expansion the above formula gives 
%y=v (14+ 102°75 x 10-74 + 82°4 x 10-1%?), 
