DOC.
225
JUNE
1916 219
It
goes
without
saying
that
just
this
necessary
difference in
the
propagation
velocities results from
your general
formulas,
when
moving
from I
to
II;
and
to
the
extent
that
an
equation
of
the
form
c1
-
c2
=
a
is
valid
both
in
systems
I
and
II,
and
likewise
also in
many
other
systems (each
time
with
another
a),
it
can
be
said,
it
expresses
the
experiment’s
result in
a
covariant form.
But this
does
not
have to
prevent
us
from
regarding
the
equation
c1
-
c2 =
0
as
different
from
c1
-
c2
#
0.
In this
sense we
shall conclude:
The
phenomena
in
the
cable do not take
place
in
the
same
way
with reference
to
coordinate
systems
I
and II.
If
you
now
try
to make this
comprehensible
in
some way
or
to visualize
it
graphically, you can
scarcely
confine
yourself
to
speaking
only
of
the
Earth,
the
cable,
and
this
“space” or
“vacuum” contained in the
latter;
you
will
be
inclined,
of
course,
to
imagine
that there
is
nothing
in and
of
itself in
the
space
or vacuum
that
would behave
differently
toward
systems
I
and II.
The
idea
certainly
suggests itself,[5]
[I
am going
to
speak
about another
one
further
down.]
and
formerly
would
surely
have seemed
very
natural
to
all
physi-
cists,
that
in
the
cable
there
is
a
medium
(the
ether)
in
which
the
waves
propagate
in such
a
way
that the
propagation
velocity
is
always
the
same
relative
to
the
medium
but that this
medium
can
be at rest with reference to
one
axial
system,
and
moving
with
reference to
the
other. If
we assume
this
point
of
view, we can
thus
say
the
experiment
had shown
us
the
relative motion
of
the Earth
against
the
ether. If
then
in this
way we
have
acknowledged
the
possibility
of
establishing
a
relative
rotation,
we
therefore
may
not
deny
from
the
outset
the
possibility
of
obtaining
changes
in such
a
translation
as well, i.e., we
may
not
hold
the
basic
principle
of relativity theory
up
as a
postulate.
We must
rather
(and
this
was
also
the actual
course
of
development)
seek
the
answer
to
the
question
in observation.
Once these observations have
instructed
us
that
no
influence
owing
to
translation
can
be
found,
we
may
through
generalization (and quite extensively
at that)
ex-
press
the
law
as
a
basic
hypothesis, whereby
we
still
grant
the
possibility,
however
(no
matter how
improbable
we
may
consider it
to
be),
that
future
observations
will
force
us
to
abandon the
hypothesis.
These observations
can
also be clothed in
another
way.
Namely,
we can
pro-
duce
standing
waves
in
the
self-contained cable and observe
the
position
of
the
nodes at
every
moment.
It
will
then
be shown
that
they
rotate in
a
circle relative
to
the Earth.
You
could,
of
course,
limit
yourself
to
establishing
the
relative
mo-
tion
of
the
nodes
against
the Earth
(or
vice
versa).
But
if
you
consider
that the
same
rotation
occurs
in
standing
waves
of
various
lengths
and various
intensities,
it
is
then
quite
obvious
(shall we say, as a
graphic summary
of
what
is
common