1050 
ratio of the contents of that Z-star to the contents of the “Schaale’’ 
E(Z) = E,, the total of the stars, for which the energy = &,, if 
the energy for an element g is taken = that of the centre? It is 
easy to see that this is not the case. The fraction defined above 
has a variable value beginning with O, if the surface of the shell 
touches the extreme points of the cells of which the star consists, 
increasing when the energy approaches that of the centre, and after- 
wards decreasing again to 0. The last mentioned fraction has a 
definite value when the energy is that of the star and else is zero. 
That the quantities are not of the same order either, appears from 
this that in the former case a definite distribution of state belongs 
to all kinds or shells, in the latter entirely to one “Schaale’, whereas 
reversely a definite “surface” contains much fewer mutually differing 
distributions than a definite shell. The value of the quotient _ 
‘Schaale” 
is further too dependent on comparatively accidental circumstances 
to serve as measure for the probability. 
The probability defined by us is, however, not easy to calculate, 
when as was supposed up to now, we think the elements g\ of 
the I-space arisen from cube-shaped regions g' and g" in the con- 
figuration-extension and the extension of momentum of the single 
molecule. It becomes easier when we think the space divided by 
E(pg)-surfaces and surfaces normal to them. 
If we think the F-space divided into a I'-space of the distribution 
of place and a I'-space of the distribution of momentum, we get 
as element: g'N\Xelement in F'-space. Hence the chance to a 
certain state now becomes: 
N! 
/ 
Orel 
X g'N & element in I-space 
total extension 
in which element and extension are both thought bounded by the 
surfaces H(pqg) = bh, and E(pq) = £, + dE. 
N! a’ N 
This fraction is equal to: DN xy X ratio of the spacial angles 
F ni! Ns: ete Ohe 
k N! g\N dw\N … . 
in F’-space' = De X | — |, if dw = spacial angular 
n,!n,! v dn 
Meme 7 
element for every molecule separately. 
However this probability is not applicable for the entropy in the 
state of equilibrium. When, namely, we make the probability maxi- 
mum, the obtained value P appears on calculation to depend no 
longer on ZH and v,. so that & log P cannot represent the entropy 
