DOCS.

275,

276

AUGUST

1911

197

const. But his

expression

is not

correct if

permanent magnetization

is

involved.

There

is

no

solution that

would

encompass

all

cases.

(iH)

was

correct

in

our

paper.

In

Abraham's

paper

(iB)

is correct.[8]

What

was

perhaps open to

criticism in

our

paper

was

our

introduction of

angular momenta.

One

can

do that,

but

it

is not required,

and

the

treatment

becomes

simpler

when

angular

momenta

are

excluded.

The

relativistic

treatment

of

gravitation is

causing

serious

difficulties.

I

consider

it

probable

that

in its

customary

formulation

the

principle

of

the

constancy

of the

velocity

of

light

holds

only

for

spaces

of

constant

gravitational

potential.[9]

Best

regards

to

you

and

your wife[10]

from

your

A.

Einstein

My

wife also

sends her best

regards.

Lenard[11]

and

his

pals are

and

remain detestable

swine.

276. To

Michele

Besso

[Prague,

second

half of

August

1911][1]

Dear

Michele,

Here

are

the

collars and

the booklet for

Vero.[2]

From the latter

you

can see

what

an

inveterate schoolmaster

I

have

already

become. I

am

still

scribbling

about

the

Boltzmann

principle.

It

seems

that the

principle

really

does hold

exactly

if

the

entropy

is

conceived

as

the

property

of

a

state

region.[3]

For

each

state

the

entropy

becomes

-*.

For

example,

to

bring

a

suspended particle

into

a

certain

layer a, one

has

to

perform

great

work

against

the

osmotic

pressure

of

this

particle.[4]

One

can see

in this

example

that

the

Boltzmann

equation

holds

exactly.

Unfortunately,

we

lack

the

knowledge

required

to

determine the

entropy

in

a

general

way

while

taking

into consideration the

statistical

irregularities,

so

that

we

must content ourselves with

an

approximation.

It

is

also

embarrassing

that

we can only

speak

of the

entropy

of

such states

that

are

at

least

in

principle thermodynamically

realizable.

It

is

thus

always

necessary

to contrive

an

analogue

of the

semipermeable

membrane,

which, however,

in most

cases

cannot be

accomplished.

We

cannot

conceive

of

"walls"

that

will

keep

the

thermal

energy

of

a

subsystem

within

specific

limits.

For that

reason,

the

entropy

of

a

heat

content

region

cannot be

determined

either,

so

that

the calculation

of

the

temperature

fluctuation lacks

rigor.

The situation

is just

as

bad

when it

comes

to

the

energy

of

a

periodically

oscillating system.

Some

means

ought

to be

found

for

extending

the

entropy concept

of

thermodynamics

to

cases

that

cannot be

regarded

as

instances

of

thermodynamic

equilibrium.

I

hope

that

you

will

use

what

is

left

of Vero's vacation

to

make

some

proper

excursions. I

am glad

that

all

of

you are

cheerful

and

in

good

health. After several

attempts

to find

a

better shelter

for

Fanni's

child,

we

took

it

into

our own

house

today.[5]