DOCS.

587,

588

JULY

1918 611

we on our

part just

have

the

comment

re

the

diagram

under “C”

(page

5)

of

the

attached

typed

copy

that the

opponent[2]

is

expected

to

point

to

the

omission in

the

directive force of

the

oscillation-impeding gyroscope.

For

depending

on

this

gyroscope’s

size

and

revolving speed,

the

line

a-a

will

not coincide

exactly

with

the

meridian

but

will

include

a specific,

although

invariable and

easily

ascertain-

able

angle,

because

this

gyroscope’s

directive force also

is

transmitted

to

the

float

k

via

the

spring

tied to it.

(If

this

one

oscillation-impeding gyroscope

is

replaced by 2

of

equal

size

but

re-

volving

in

opposite directions,

this deviation

disappears,

which

case

corresponds,

by

the

way,

to

the

diagram

of

figure

4 of

our

Patent

Specification

241

637.)

In

other

words,

with

both

gyroscope

axes

forming

an

angle

with

the

meridian,

your

diagram

thus

again

resembles

the

design example

of

our

patent

specification,

the

more so

since it

suffices

for

only one gyroscope

to be

equipped

with

a

yielding

link

with the

float.

Since

it

does not bear

well

to

go

into the

details of these

relations,

it would

perhaps

be advisable

to

change

the

end of Section

C

a

little.[3]

We

reserve

for ourselves

to

convey our

patent

attorney’s opinion

on

this

to

you

as

well,

and remain in

the

meantime,

in utmost

respect, your

devoted

servants,

Anschütz

&

Co.

588. From Felix Klein

Göttingen, 15 July 1918

Esteemed

Colleague,

I have succeeded

in

discovering

the

organic

formation law for Hilbert’s

energy

vector.[1]

If

the

infinitesimal transformations

of

my January note[2]

are

not

spe-

cialized to vanish at

the

edge

of

the

integration

domain,

one

obtains,

in

addition

to

the 4-fold

integral (which

must vanish for

arbitrary

pa),

another

triple integral

taken

along

the

edge

(which,

independent

of

the

edge’s

shape,

must

correspond-

ingly

be

zero):[3]

JJJyM*

dw2dw3dw4 +

.

e4

•

dw1dw2dw3}.

And here

now

is

e1...e4,

subject

to

changes

one

might

want to

attach

owing

to

the

field

equations, or

even

the

addition

of

a

simple

tensor,

whose

divergence

vanishes

identically,

the

sought

vector.-[4]

I

hope very

much to

pave

the

way

from