1171 
id 
of equilibrium from its connection with the solid body, and to bring 
it in the gas space, hence c is negative. 
We can immediately integrate with respect to the gas, through 
which we get: 
/ 
(| wd |) = 1 (2 m kT)" hn vr fe be psn Esse AIEN ne wor (6) 
n LE 
in which v is the volume of the gas. 
Now we can replace the values gs: to sv (inclusive) by 
variables @',,....,@3/ Which are in linear connection with them, 
Bn’ Bn’ 
so that &, = 4 > fig:” with all positive /’s, while &)= 5— =p’, 
1 : am 4 
; OBS ts 
when p'; is the value corresponding to q';, hence Dn The quantities 
qi 
qi are evidently a criterion for the deviations of the molecules from 
their positions of equilibrium. 
As according to a known thesis dpi... d's; = dpsn4i---dgsn, 
we get: 
Le Os nl 
E = fg? Ep 
z ne ‘ ey Leet 2m 2 
kT WET SER are La 
e dpgnti +++ dq3N = ¢ +e dpy' es dan . (7) 
When we however should simply integrate on the right with 
respect to all the values of —o to + oe, and substitute the result 
in (6), as being equal to the integral with respect to the leftside 
member of (7), we should commit a serious error and arrive at an 
absurd final result. 
We had namely originally to integrate with respect to all the 
values of the q’s inside the volume occupied by the solid body. In 
this those values are naturally left out of account for which the 
energy is very great, for which a molecule is therefore pretty far 
from its position of equilibrium, as this according to the formula 
for the probability very rarely occurs. The proportionality of the 
energy with g’, however, only holds for slight departures out of 
the position of equilibrium, and no longer when a molecule has got 
so far that it can pass between the neighbouring ones. This must 
actually occasionally occur, though very seldom, and in this way 
two molecules can interchange their original positions, and each 
molecule can successively be found at all possible points of the 
solid body, and have a position of equilibrium which was before 
