915 
at low temperatures *). However I did not succeed any better for a 
long time afterwards. It is now evident that this was caused by 
starting from the supposition, which seemed obvious, that the vanish- 
ing of the resistance in approaching the absolute zero had to be 
represented by a law of exponential character’). The principal 
difficulty was to represent by one single formula the resistances at 
the temperatures of liquid hydrogen together with those at higher 
temperatures. I wished however to maintain the condition that 
both regions should be represented by one fermula, in order to 
make possible an interpolation in the interval between the boiling- 
point of hydrogen and the melting-point of oxygen, where no mea- 
sures with the gas-thermometer are available. 
2. Dependence for T small. Witx*) was the first to show the 
possibility of a different law for the resistance at the lowest tem- 
peratures in a paper in which from certain theoretical suppositions 
he deduced a decrease proportional to 7”. I verified at once that 
the hydrogen-temperatures would be much more closely represented 
by such a dependence than by exponential forms. 
GRÜNBISEN *) has pointed out that the temperature-dependence of 
the product 7c, shows a great resemblance to that of the resistance 
of the same metal, especially at low temperatures. He suggests further 
that both these quantities might be proportional to a universal 
function of 7/7,, the same for all metals, 7, being a “characteristic 
temperature” of the metal’). The data available to test both these 
rules are, however, of much lower accuracy than that wanted here. 
Thus e. g. the specific heat-formula of Degije ®) could certainly yield 
a fair representation of the resistance, but not with an accuracy in 
any way comparable to that needed for resistance-thermometry. 
It appears to me that the coincidence found by GRÜNRISEN may 
have a theoretical foundation. If such be the case the rule would 
1) Compare the formulae with 5 and 6 constants to be found in different Leiden 
Comm. and also in the dissertation of Dr. Guay, Leiden 1908. 
2) This is also the case with the formulae of OnNes, Comm. 119, Proceedings 
13, 1093 and of LrypemaAnn, Berlin Sitzungsber. 1911, 316, which do not aim at 
such a high accuracy. 
3) Berlin Sitzungsberichte 1913, 184 
4) Verh. D. Phys. Ges. 15, 186 (1913). 
5) With respect to this point the result by Meissner, Verh. D. Phys. G. 16, 
262 (1914) is interesting. According to his experiments the conduction of heat 
would show a wholly different value of 7, somewhere like half the value for 
other phenomena. 
6) Ann. d. Phys. (4) 39, 789 (1912). 
