DOC.
4 51
where the subscript
of
dN
denotes
the
time.
Taking
into
account equation
(1),
one
obtains furthermore
at
time
t
+
dt
and
the
same
region
of the
state
variables
T
Kepv)
dNt+dt
=
dNt ~
I
*dPvdPn'dt
IISTp
However,
since
dNt
=
dNt+dt' because the distribution is
stationary,
we
have
v
_
0
l~SptT"°
.
[6]
This yields
_
V
=
V
d(l°g j)
*
in

V
d(loS
e)
.
dpv
_
(/(log
e)
L
dpy
L
fpy
^v
L
fpy
~ST
dt
'
where d(log e)/dt
denotes the
change
of the function
log
e
with
respect to
time
for
an
individual
system,
taking
into
account
the
changes
with time of
the quantities
pv.
One
obtains further
v=n
dip
«
l
sf
*
*W
v\
v
_
m+ij)(E)
C
The unknown
function
^
is
the
timeindependent
integration
constant
which
may
depend
on
the variables
p1...pn,
but
can
contain
them,
according to
the
assumptions
made
in
§1,
only
in
the combination
in which
they
appear
in the
energy
E.
However,
since
i(E) = i(E*) =
const.
for all
N systems
considered,
the
expression
for
e
reduces in
our case
to
vn
v d(pv
dt I
W
V1
~^v
e
=
const.
e
=
const.
em
.
According
to
the
above
we now
have