DOCS.

439,

440

JANUARY

1918

447

Your

other

comments

are

just

as interesting

per

se

and

new

to

me.

I

am

not

so

bold

as

to

want to decide whether

they will

lead to

an

advance. In

any event,

I

wish

you

luck in

your

efforts!

With best

regards,

I

am

yours,

A.

Einstein.

440. To

Rudolf

Humm

[Berlin,]

5

Haberland

St.,

18 January 1918

Esteemed

Colleague,

You

are,

in

fact, entirely

correct.[1]

The

equations

(+)A¿#JI/

=

-k

m

dxu dxv

T

=

o----

ds ds

cannot be satisfied in

any

way.

The

equations

p(dx1/ds)2

=

...

=

p(dx4/ds)2

read

more

completely

as

(since

x4

=

it

must be

set

so

that

guv

=

-Suv

also for

m

=

n

=

4):

p(dx/dt)2/c2-v2

=

p(dy/dt)2/c2-v2

=

.

=

-p/c2-v2,

which cannot be satisified.

Changing

anything

in

the

energy

tensor

is out of

the

question.-

But

this

finding presents absolutely no

difficulty,

according

to

my conception

of

the

boundary

condition

problem.

The

supplementary

term

serves specifically

to allow

the

quasi-spherical

(or quasi-elliptic)

world to take

the

place

of

the

quasi-

Euclidean

one.

Thus

it

is

desirable

that the

equations

just

preclude any quasi-

Euclidean world.

I

presented

in

detail the

reasons

that

led

me

to this

interpretation

in

my

article

of

last

year.[2]

To

repeat,

I

only

note

that the

boundary

conditions for

a quasi-

Euclidean

world

(guv =

const. at

spatial

infinity) are

not

relative,

i.e.,

not

valid