1176 
It seems natural to assume, that these equations when they shall 
have been found, will be able to account for the different above- 
mentioned deviations from the law of equipartition of energy. In 
fact these deviations are closely connected with one another. If e.g. 
the energy of visible lightvibrations at 100° is imperceptibly small 
compared with that of infra-red rays, we cannot wonder that the 
vibrations of electrons which are in equilibrium with those light 
vibrations have an energy very small compared with that of vibra- 
tions of greater period. The thermal motion of the molecules may 
here probably be considered as a vibration of rather large period, 
although it is not a simple harmonic vibration. At a higher tempe- 
rature the small wavelengths become more predominant in the 
spectrum. It is therefore to be expected that also the vibrations of 
the electrons of short period, which at a low temperature are devoid 
of energy, at a higher temperature will obtain a measurable amount 
of energy, so that the specific heat with constant volume will increase 
with the temperature. 
The physicists occupied with these problems have noticed this 
connection between the normal spectrum and the specific heats from 
the beginning. JEANs') e.g. has applied his theory, which originally 
was meant to be an explanation of the c, of gases, to explain the 
properties of the normal spectrum; and it is not astonishing that 
vice versa the theory of PrANcK for the normal spectrum was soon 
used for the explanation of the specific heats. 
The method in which we start from a theory for the normal 
spectrum and deduce from it the value of c‚ seems to have advan- 
tages over the opposite way. For we have in the spectral formula 
of PrarcK a relation which agrees well with the observations and 
which moreover is independent of the special nature of the walls. 
I will therefore follow this method.. 
§ 2. Lhe centra of radiation. 
We may make the following two assumptions concerning the way 
in which the partition of energy of the normal spectrum is brought 
about. 
1st. We may assume that every vibrator considered separately 
has the property to transform radiation of an arbitrary partition of 
energy into the partition of energy of the normal spectrum. 
2nd. We may assume that this property only belongs to groups 
') J. H Jeans, Proc. Royal. Soc. of London 67, p. 236, anno 1900. 
Phil. Mag. (6) 2, p. 421 and 638, anno 1901. 
Proc. Phys. Soc. of London 17, p. 754, anno 1901, etc. 
