( 811 ) 
which for the limit dv =O is equal to | 5). I would be possible 
t 
to take the constant A different for the several strips in which 
Ei can be divided; the ensemble attained in this way is also 
stationary ?). I shall prove that the ensemble with A constant (the 
energy-space ensemble) is the limit of a microcanonical ensemble. 
In order to prove this, it is necessary to consider more closely the 
velocity v. If we take a point p,....q, it is easy to indicate the 
components of v; these are: 
de 
y ZE —— 
} de . | 
v from 1—n 
de | 
Qy == — 
ps | 
therefore 
: Bee Gere de | & 
jie I= A AM ae tee 
| ( Op, dg, (©) 
This velocity can be connected with a purely geometrical quantity 
relating to the space /,—; in the point p,, q. 
The direction-coefficients of the normal on this space in the point 
in question (a, , ,) are: 
de 
a 
EN EN 
TRE 
I | dg, Og 
de 
‘ Op, 
Dl ne 
i | \0g, op). | 
Let A denote a distance on the normal, ending on the space 
Eon, , + de, we shall have 
n (Q 0 
ge=AZS\— a -- =e =Av 
1 Og. Op, 
Ar aL anes re : 
If A approaches O we find gy? ie the differential-coefficient of 
OLN 
e in the direction of the normal at the space Zo. Or therefore, 
') I shall prove that do is absolutely constant for a strip. 
*) An ensemble of this kind is produceed by cutting through the space Hon—1 a 
ayer from the ensemble of page 803. 
