392 
PROF. LOUIS VESSOT KING ON THE CONVECTION OF 
wire in order to avoid errors due to optical distortion. It was found possible to reset 
the cross-hairs consistently to each division of the micrometer head, i.e., to about 
10~ 5 cm. The results are tabulated in Table II. and further details are given in the 
accompanying description. 
The measurement of the lengths of wire between potential terminals offered no 
difficulties. In order to calculate the heat-loss per unit length at any temperature it 
was necessary to correct for expansion with temperature by means of the formula 
l g = l 0 (l + 06 + /36 2 ), where l e is the length at 6 ° C., l 0 that at 0° C., and a and (3 the 
coefficients of linear expansion given by Holborn and Day ( 31 ) as a — 0'08868 x 1CU 4 , 
(3 = 0'001324 x 1C)- 6 . 
Section 11. On the Calculation of Temperatures. 
From the mean temperature-coefficient of resistance c of platinum between 0° C. 
and 100° C. given by c — (R lu0 — R o )/l00R o , the temperature of the wires were calculated 
from the formula 6 p = (R/R 0 — l)/c, R being the resistance at the temperature 6 p on 
the platinum scale and R 0 that at 0° C. In the present instance the resistances of the 
platinum wires were referred to the temperature t — 17 C., so that the ratio R/R 17 
are those from which the temperatures are to be calculated. We may take very 
approximately R t = R (l (l +ct) for t = 17° C., in which case it is not difficult to prove 
that 6 p — t = (R/R t — l) ( l/c + t ) by means of which temperatures on the platinum scale 
were calculated. The temperatures of the wires were then obtained in terms of the 
true thermodynamic scale by means of Callendar’s formula, 
6° C. = 6 p + S (t/100— 1) . t/ 100 with S = 1‘54. 
Actual calculation of temperatures was avoided by the use of the conversion tables 
given by Burgess and Le Chatelier ( 32 ) calculated for S = 1’50. From an auxiliary 
table the temperature correction for a value of d other than this value is given ; it is 
seen that the correction A 6 for Ad = 0‘01 is A 6 = 0‘9 C C. at 6 = 1000 C., showing 
that the value of d can vary considerably without altering the value of 6 to an extent 
greater than the limit of accuracy of the other measurements. 
No attempt was made to work with the wires at a higher temperature than 1200° C. 
owing to the fact that the use of Callendar’s formula beyond this point is subject to 
uncertainty ( 33 ); also at high temperatures the wires soften and are liable to break 
when driven through the air at the high velocities. 
( 31 ) ‘Smithsonian Physical Tables,’ 5th edition, 1910, p. 222. 
( 32 ) Burgess and Le Chatelier, ‘ The Measurement of High Temperature,’ Table V., p. 493; also 
Table VII., p. 495. 
( 33 ) See Langmuir, ‘Journal Amer. Chem. Soc.,’ vol. 28, p. 1357, 1906. 
