DOC.
364
JULY
1917 355
system
is
sufficiently
defined
that
way,
but
the
origin
of
the
spatial
coordinate
system
can
still describe
an
arbitrary
world
line.
Then
surely
the
most
closely
Euclidean
system
would be
the
one
in which
no
arbitrary gravitational field
exists
apart
from those
generated by
known
masses,
that
is, no
lines
of
force
without
a
place of
origin. If,
for
ex., we move over
from
the
commonly
standard
systems
to
one
which is
uniformly
accelerated in
a
straight
line
relative to
it,
we
then obtain
a homogeneous
gravitational
field
permeating
the
entire
universe,
for which
no source
could be
found.[4]
Thus
the
system
is
fully
determined
except
for
the
uniform
velocity.
That
variations in
the
simplicity
of
describing
real
processes appear
here
as
well is
even more
foreign
a
thought to
the
pure physicist who,
in contrast to
the
astronomer,
concerns
himself
merely
with
general
differential
equations
and
not
with
concrete
integrated
equations.
In
order to
understand that
not
every
coordinate
system
reveals
the
phenomena
in
an
equally simple
manner,
also with
regard
to uniform
motion,
one
must
abandon
precisely
this
standpoint
of
the
pure physicist
and
assume
that
of the
astronomer.
On
assuming
this
point
of
view,
it
becomes
apparent
that the
one
system
with
the
fixed-star
system’s
center
of
mass
at
rest
is
preferable.
This
is just
the
idea
I
had
pointed
out
three
years ago
and which Friedrich Adler has also
had
since.
In
my
letter
to Professor
Frank at
the
time,
I
indicated
that
such
a system
is
referred
to
implicitly
when it
is
stated that the
Sun
moves
at
25
km
per
sec
against
the
Hercules constellation. For
even if,
somewhat
like
a
molecule in
a
more
comprehensive
world
system, our
fixed-star
system moves
with
reference to
its
center
of
mass
such
that
by
chance
our
Sun
happens
to
be
at
rest in
that
system,
we
should still
talk
of
such
an
independent
motion
by
the
Sun, using
as
a
reference
the
center
of
mass
of
the
larger system
to
which
the
Sun
directly
belongs.
My
reflections
related
to
the
general principle
of
relativity
were
formed
only recently; naturally,
they
could
not
have
developed
earlier
than the
general
relativity
principle
itself.
I cannot
judge,
of
course,
whether
my
considerations
are
admissible, yet
it
seems
to
me
that
a
choice
related
to rotation,
similar to
one
related
to
motion, suggests
itself.[5]
Here
we
select the
system
such
that the
sum
of all
the
momenta is
zero;
there,
such
that the
sum
of
all
the
angular momenta is
zero.
In
the
case
of
rotation,
however, we
must
take
the
entire
cosmos
into account
and cannot
disregard
even
the
remotest
masses,
which become
infinitely large
in
the
“Ptolemaic”
system,
because of their
velocity.
To
describe
the
phenomena
on
the Earth
in the
simplest way, we are
not
allowed
to
choose
the
system
in which
the Earth
is
at
rest,
because
a
whole series of
processes
on
Earth: the
deflection
of
the
winds
and
of
falling bodies,
and
the
rotation of
[Foucault’s]
pendulum
plane,
demonstrate
a
link with the fixed-star
sky
and with
the
masses
of
the
whole universe.
Regarding
the
rotation,
only
if
a
system
in
which
the
fixed-star