FOUNDATIONS OF STATISTICAL

PHYSICS

47

The

only

evidence

of

personal

influences

on

Einstein's

early

work

on

statistical

physics

concerns

his

correspondence

with

Michele

Besso from

1903,

and

his conversations with

a

patent-office

colleague, Joseph

Sauter.

Einstein's

correspondence

with Besso

is

discussed

in section III

below,

and in the annotations to

one

of

the

papers.[48]

A

lengthy

discussion

with Sauter

is

mentioned

in a

letter

from

Besso

to Einstein

of

13 June 1952:

Lately

he found

again your papers on thermodynamics

of 26.VI.1902

[Doc.

3]

and

26.I.1903

[Doc. 4].

He remembers

having

discussed

one

of

them at

length

at that

time,

and

having

to

a

certain extent saved what

was

essential in

it,

in

spite

of

a

mistake that he

discovered,

this

in

the face

of

a pessimistic

attitude

on your part;

but he cannot recall which

of

the two it concerned.

Er fand letztlich wieder Deine Arbeiten über

Thermodynamik vom

26.VI.1902

[Doc.

3]

und

26.I.1903

[Doc. 4].

Er

erinnert sich über die eine,

seinerzeit,

lange

diskutiert

zu

haben,

und das Wesentliche davon, trotz

eines

damals

von

ihm entdeckten

Fehlers,

entgegen pessimistischer Einstellung von

Dir,

gewissermassen gerettet zu

haben;

kann

sich aber nicht besinnen welche

von

den beiden

es

betraf.[49]

III

As noted

above,

Einstein's

contributions to the foundations

of

statistical

physics grew

out

of

the Maxwell-Boltzmann tradition in kinetic

theory,

which

was

then also called the

"mo-

lecular

theory

of heat"

("molekulare

Wärmetheorie"),

more specifically,

the

aspect

of

the tradition that

emphasizes

the

general

statistical foundations

of

thermodynamics.

Ein-

stein's

aim

was

to derive the basic

concepts

and laws

of

thermodynamics

from mechanics

and the

probability

calculus. In

doing

so,

Einstein

improved upon

Boltzmann's

arguments

in several

important

respects,

so

that his three

papers on

the

subject

form

a

bridge, parallel

to that in Gibbs

1902,

between

Boltzmann's

work and the modern

approach

to statistical

mechanics.[50]

Among

the features that

distinguish

Einstein's

approach

from

Boltzmann's

are: (1)

his

attempt

at

generality-he tried

to

proceed

with

a

minimum

of

assumptions

about the nature

of

the

systems

under

consideration;

(2)

his

conception

of

what needs to

be derived from these

assumptions-not

only

the second law

of

thermodynamics,

but also

pp.

44 and

46).

See also Einstein

1915a,

pp.

260-261, and the editorial

note,

"Einstein

on

Brownian Motion,"

p.

218.

[48]

See Einstein 1903

(Doc.

4),

notes

4,

6,

and 7.

[49]

Einstein

replied on

13

July

1952 that he

recalled

many

discussions with Sauter about the

papers,

but could not remember the details. Sau-

ter

himself

described this

episode

in

a

radio

broadcast in

1955,

in which he recalled discuss-

ing

Einstein's

early papers during

walks home

with Einstein

after work at the

patent

office

(for

the

text

of

this

talk,

see

Flückiger

1974,

pp.

155-156).

An

error

that

may

be the

one

discov-

ered

by

Sauter

is discussed in Einstein 1903

(Doc.

4), note 7.

[50]

Another

important

transitional work

is

Ehrenfest

and

Ehrenfest 1911. Two

of

the earli-

est texts to

incorporate

the modern statistical

point

of

view

are

Wassmuth 1915 and

Hertz,

P.

1916.