616 DOC.

594

AUGUST

1918

594. To

Friedrich

Adler

[Ahrenshoop,]

4 August 1918

Dear

Adler,

Yesterday your

manuscript

arrived.[1]

I have

already

studied the

first

chapter

and

am

thus

informed

about

all the

essentials.[2]

The basic

physical assumptions

are:[3]

a) no

Lorentz deformation

of

moving rigid bodies,

b) no

influence of motion

on

the

running

rate

of

the

clocks.

Thus the

geometrical

and kinematic elements

are entirely given, quite

apart

from

the mathematical

expression

in coordinates.

The

assumptions

(a)

and

(b)

are,

in

principle, directly verifiable,

but

not in

practice. (a)

leads

immediately

to Abraham’s

theory of

motion of

an

electron,

however,

which

is

refuted

empirically

as

soon

as

the

nonexistence of

the

deforma-

tion

of

moving

bodies

is

applied

also to

the

electron.[4]

It

must

be

considered, furthermore,

that

(a)

leads

straight

away

to

a

contra-

diction

with the

outcome

of

the

Michelson

experiment,

if

it

is assumed

that

the

law of

the

constancy

of

the

velocity

of

light

in

a

vacuum,

which

you

are

not

likely

to

question,

is

valid relative to

the

preferred

reference

system

K.

I

find

that

a

theory

can

be taken

seriously

from

the

physical

standpoint

only

when it does

justice to

the

following

observational results:

1)

Fizeau’s

experiment.

2)

Motion of electrons in

an

electromagn.

field.

3)

Aberration

law.

4)

Michelson’s

experiment.

For it

was

these

facts which

compelled

the

formulation

of

the

special

theory

of

relativity.

You have made

no

attempt

to

address these

fundamental

facts,

however.-

I

come now

to

the

formal

aspect. Basically,

on

making

arbitrary assumptions

about the

behavior of

measuring

rods and

clocks,

one can use

arbitrary

transfor-

mation

equations

without

coming

into conflict with

the

logic

or

with

experience.

If,

however,

as

corresponds

with

your

assumptions

about the behavior

of

the

measuring

rods and

clocks,

it

is

assumed

that

rigid

bodies

at rest

relative to

one

another

follow

Euclidean

geometry regarding

the

positioning

laws

[Lagerungsge-

setze],

and that static

clocks,

relative to

one

another,

run

equally quickly,

then

it

is certainly appropriate

(although logically

absolutely

not

necessary)

to

choose

the coordinates in such

a way

that

for all

legitimate

systems

the

following

is

valid: