DOCS.
399,
400
NOVEMBER
1917
407
399. To Paul Ehrenfest
[Berlin,]
12 November
1917
Dear
Ehrenfest,
I
really
was
heartily
pleased
about
your invitation,[1]
and
you
can
believe
me
that
nothing
is
more appealing
to
me
than
a
trip
to
my
dear Dutch
friends,
with
whom
I
share such
close
and kindred
feelings
in
everything. Yet,
with
the
prevailing
travel
conditions[2]
and
my
weak
health,
I
may
not
think
of it.
I
am
bound
to
a
strict
diet and
am
never
really
safe
from
a
painful
attack.[3]
In
the
last half
year
it
has
admittedly
been
improving steadily;
however,
I must follow
a clinically regular bourgeois lifestyle.
Scientifically
there
is
nothing of
note to
report.
Your
objection
to
my
quantum
paper
of
1914 is
thoroughly
justified;[4]
I became
aware
of
the
same
recently upon
studying
your paper
of
1916.[5]
I
do
believe,
though,
that the
matter
can
very
probably
be corrected in
this
way:
The
equation
S
=
lgZ
is
initially only proven
for
purely
thermal
changes.
It
was
wrong,
now,
to
conclude
the
invariability
of
Z
according
to
the
adiabatic
hypothesis.[6]
Instead
I
avail
myself
of
the circumstance
that
I
can
choose
the
system’s
external conditions and
that,
depending
on
this
choice,
other
processes
are
“purely
thermal.”
Hence,
for
ex., a
rise in
temperature
with
a
constant volume
is
a “purely
thermal”
process
or a
rise in
temperature
at constant
pressure, depending
on
the
choice
of these conditions. Therefore
any
state of
the
system
is
attainable
through
“purely
thermal”
changes.
That
is
why
the
equation
S
=
lgZ
applies generally
if
it
is valid
for
purely
therm.
processes.
Cordial
regards, yours,
Einstein.
400. To
Rudolf Förster
Berlin, 5
Haberland
St.,
16
November
1917
Esteemed
Sir,
From
your
letter[1] I
see
that
I
am
dealing
with
a man
of
extraordinary
theo-
retical talent. It
would be
a
pity
if
you
did not have
enough
leisure to
think about
these
fine
problems.
Now I
would
like to
reply immediately
to
your
individual
points
in
the
order
you
followed.
1)
The introduction
of
yi
in
place
of
guv
is,
without
a
doubt,
possible
and
is
also
suitable
for
reducing
the
number of functions
sought,
from
10
to
6.
But it
is
questionable
whether
the
expressions
for
Buv
obtained in
this
way really
are