168
DOCS.
173,
174 DECEMBER
1915
(I)
and
(II)
of
your
postcard
are
written
as
they
already appear
for
the
refer-
ence
system specialization
g
=
-1.[5]
First
imagine
them
written in the
general
covariant
form,
which
is
also
known,
of
course.
Certainly you
can
eliminate
the
TAo's
from these
equations
and
you
obtain
equations
that
are
only
satisfied
by
the
guv’s.
But
these
equations
are
then, just
as (I)
and
(II), generally covariant; they
therefore
do
not
require any specialization
of
the
reference system,
but
are
valid
in
any reference system
(if
they
are
valid in
one).
Specialization
to
(g
=
-1)
systems changes
nothing
in
the
consideration).
Cordial
greetings
to
your
wife,
your children, Lorentz,
both
de
Haases,
and
Fokker.
I
am
going
to
write
to
de Haas
soon.
My
congratulations
and
gravitation
paper
for
Fokker both
came
back
as
undeliverable;
I
had addressed them
to
Gravenhage.[6]
Best
regards
to
you;
write
again soon, yours,
Einstein.
174. To Paul
Ehrenfest
[Berlin,
29
December
1915]
Dear
Ehrenfest,
Your
relation
0 =
£
K
dn*
dxK-...
(IV)
may
not
be
an
identity,
but
(with
the
restr[iction]
V-g
=
1)
it
certainly is
a
general
covariant
equation.[1]
Thus
it
cannot
serve
to fix
a
preferred
system.-
I probably
disproved my “philosophical”
consideration[2]
adequately
enough
for
you.
The
field
equations
naturally
do
not
provide any unique
determination
of
the
guv's;
randomness
remains,
which
the
“philosoph.’
consideration
yields
and
which results from
the
randomness of
the
reference
system’s
continuation.-
Your
relation
(IV), or
its
generalization,[3]
remains
satisfied at
every
transfor-
mation.
Thus
it
can
pose
no
obstacle here for
the
specialization
of
the
system
according
to
g
=
-1.
Third
question.[4]
The
passage ought
to
have read: Gen.
relat.
does
not restrict
the
possibilities
with
regard
to the other
processes
more
than
special
relativity
does;
but
it
provides
the
influence
of
the
gravitational
field
on
those
processes.[5]
Fourth
question.
With
“recently” my
addendum “On
the
gen.
theory
of
rel.”[6]
was
meant,
where
I
believed
that
general
covariance demands
the
hypothesis
ETuu
=
0,
which
then turned
out to
be
unnecessary.[7]