2 2 4 D O C U M E N T 9 4 J U L Y 1 9 2 0

same place in time in the same state of motion. Time t is conventional; but in the

static case, time t is universally given (with the exception of an additive correction),

if it is given at one location. For in the static case, time t has physical meaning to

the extent that it is chosen so as to make static or stationary processes appear re-

spectively as static or stationary. If I allow, e.g., monochromatic light to travel from

the Sun to the Earth, then the time interval t for generating 100 oscillations is

equal to the time interval t for receiving the 100 oscillations on Earth. With an-

other time choice, the process of light propagation between Sun and Earth would

not appear

stationary.[3]

For resting standard clocks that are set up at two locations, one has the equations

.

We now want to assume further that location (1) is the Sun’s surface; location (2),

however, is only so far away from the Sun that the special theory of relativity still

applies precisely enough. Then (g44)2 = 1 or s2 = t2 is true. In the conventional

unit of time ( t)2 = 1, the standard clock hence strikes exactly once ( s2 = t2 = 1).

On the Sun, however, for the same conventional time unit ( t)1 = 1, which accord-

ing to the foregoing case of the static field has direct physical meaning,

.

Hence, in the conventional unit of time, the clock makes less than one stroke (since

g44 1).

Alternatively, one can also reason like this. If the clock strikes once on the Sun,

s = 1, then the elapsed conventional time (which preserves the system’s static

character)

.

The equation you cited[4]

makes no sense to me either.

When you write back to me, please remain consistent in having the ’s be

lengths of a clock’s period measured by a special standard clock.

Best regards, yours,

A. Einstein.

s

1

g44

1

t

1

=

s

2

g44

2

t

2

=

s2 g44

1

=

g44 t1 1=

t1

1

g44

----------- 1 =

dt

sun

dt

g44

----------- =