688 DOCS.

645,

646

NOVEMBER

1918

5)

Now

Bouvpuv

again

has

the

desired

vector

character,

which

thus

brings

the

proof

to

an

end.-

I

chose

the

designations

Auv

and

B°uv

in order to allow

the

close

relation

with

formulas

(8), (9),

and

(10)

of Hilbert’s first

note[5]

to

stand

out.-Hilbert’s

results

agree

with

mine

so

long

as

K

+ L

are

substituted

for his

H.[6]

However,

he

provides

these

results

in his note

even

before

he

specifies

H in this

way.

There

are

strong

doubts about

whether

the

results

are

correct in

this

generalization–

whether Hilbert

had not

originally developed

them

only

for

K

+

L

and

had

just

put

them

in

the

wrong place

later

during

revision.

Of

course,

I

am

eager

to

hear

what

you say

to these considerations.

My

form

of proof is admittedly

not

pretty,

because

the

execution

of

(4)

requires

too much

mechanical

computation.

But

a

scoundrel

offers

more

than

he has.

Very

truly

yours,

Klein.

646. To

Felix Klein

[Berlin,

8

November

1918]

Highly

esteemed

Colleague,

Thank

you very

much for

the

transparent

proof,

which

I

understood

complete-

ly.[1]

The

fact

that

it

cannot

be

performed

without

calculation does not

detract

from

your

entire

analyses,

of

course,

since

you

do

not

make

use

of

the

vector

character

of Ea.-

In

the

whole

theory,

one

thing

still

disturbs

me

formally, namely,

that

Tuv

must

necessarily

be

symmetric

but

not

tuv,

even

though

both

must

appear

equiv-

alently

in the conservation

law.

Maybe

this

incongruency

will

disappear

when

“matter”

is

included,

not

only

superficially

as

it has been

up

to

now

but

really,

in

the

theory.

The

logically so

fine

approach by Weyl,

which would achieve

this,

unfortunately

does not

seem

to

me

to

provide

the

solution,

for

physical

reasons.[2]

With thanks

and

my

best

regards, yours

very truly,

A.

Einstein.