to make use of the well-known fact that CALLENDAR’s formula holds 
at higher temperatures, which simply means that the resistance can 
be represented there by a quadratic form of 7. Therefore I tried 
to make formula (1) suitable for high temperatures by adding to 
the numerator a term — #7“, in which % has to be determined 
from CALLENDAR’S constant. 
This very term seems to hinder a close fit at low temperatures. 
Now Henninc') found later that at low temperatures CALLENDAR's 
formula did not deviate gradually from experiment, but that, on the 
contrary, it fits exactly down to 230° abs., deviating at once very 
markedly below this temperature. So it seemed useless to look 
for one formula suitable for both regions above and below that 
temperature, and I thus confined myself to test the unchanged 
formula (1) for temperatures below 230°. 
Some tentative computations with the experimental material of 
KAMERLINGH ONNES and Cray, and of Henninc, mentioned above, 
showed at once the suitability of the formula in question. Taking 
into account the additive resistance caused by impurities, the formula 
may be put thus: 
- 
== on —+w,.... (4) 
L--ar—1+ pr—?+ yr—3 
in which t has been substituted for 0.01 7’ to grasp easier the order 
Ww 
of magnitude of the different terms. In (4) the coefficients «, 3, y do 
no longer depend on the particular value of the resistance, but merely 
on the inherent properties of the metal experimented on. Besides, 
impurities giving only approximately an increase in resistance which 
is independent of temperature, the constants «,‚8,y will depend, 
though slightly, on the purity of the metal. 
The computations gave for Pt as the best values of the constants 
something like: 
u B Y 
0.15 0.16 0.08 
or 0.17 0.12 0.09 
I did not undertake a more precise determination with the aid 
of the data available, as the Leiden observers had announced the 
publication of a new calibration of the Pf-thermometer carried out 
TL Oils 
+. Determination of the constants. In Leiden Communications 
N°. 141a®) KAMERLINGH Onnes and Horst give the results of a very 
1) Ann. d. Phys. (4) 40, 635 (1913). 
*) These Proceedings 17, 501, 
