CO >-

DO

64741

OUP- 880-5-8-74-10,000.

OSMANIA UNIVERSITY LIBRARY

Call

No.

^''l

Accession No.

Author

This book should be returned on

of

before the date last

marke4

/>eiow,

THE
EINSTEIN

THEORY

OF
RELATIVITY

Boofcs

by L

/?.

ancf H. G. Lieber

NON-EUCLIDEAN GEOMETRY
GALOIS AND THE THEORY OF GROUPS
THE EDUCATION OF
T.

C.

MITS

THE EINSTEIN THEORY OF RELATIVITY

MITS,

WITS

AND LOGIC

INFINITY

Books of drawings by H. G. Lieber

GOODBYE

MR.

MAN, HELLO MR. NEWMAN

L.

(WITH INTRODUCTION BY

R.

LIEBER)

COMEDIE INTERNATIONALE

THE
EINSTEIN

THEORY

OF
RELATIVITY
Text By

LILLIAN

R.

LIEBER

Drawings By

HUGH GRAY

LIEBER

HOLT, RINEHART

AND WINSTON

New

York / Chicago / San Francisco

Copyright, 1936, 1945, by

L.

R.

and

H. G. Lieber

All rights reserved, including the right to repro-

duce
In

this

of

Canada, Canada,

book or portions thereof in any form. Holt, Rinehart and Winston

Limited.

First Edition

October 1945 Second Printing, April 1946 Third Printing, November 1946 Fourth Printing, November 1947 Fifth Printing, May 1950 S/xtfi Printing, March 1954 Seventh Printing, November 1 957 Eighth Printing, July 1958 Ninth Printing, July 1959 Tenth Printing, April 1960 Eleventh Printing, April 1961 Twelfth Printing, April 1964 Thirteenth Printing, November 1966
Firsf Printing,

1975

85251-0115
Printed in the United States of America

To

FRANKLIN DELANO ROOSEVELT

who saved
the world from those forces

of evil which sought to destroy

Art and Science and the very

Dignity of

Man.

PREFACE
In
this

book on

the Einstein Theory of Relativity

the attempt is made to introduce just enough

mathematics
to

HELP

and

NOT to HINDER

the lay reader/ "lay" can of course apply to various domains of knowledge

perhaps then

we should

say:

the layman in Relativity.

Many

"popular" discussions of
all,

Relativity,

without any mathematics at have been written. But we doubt whether

even the best of these can possibly give to a novice an adequate idea of

what

it is

all

about.

What
in

very clear when expressed mathematical language

is

sounds "mystical"
ordinary language.

in

On

the other hand,

there are

many

discussions,

including Einstein's own papers, which are accessible to the

experts only.
vii

We
who

believe that
is

there

a class of readers

can get very little out of either of these two kinds of

discussion readers who

know enough about

mathematics to follow a simple mathematical presentation of a domain new to them, built from the ground up,
with sufficient details to bridge the gaps that exist

FOR THEM
both

in

the popular and the expert presentations.

This book is an attempt to satisfy the needs of
this

kind of reader.

viii

CONTENTS
PREFACE
Part
I

-THE

SPECIAL THEORY
3
8

I.

INTRODUCTION
The Michelson-Morley Experiment
Re-Examination of the Fundamental Ideas

II.

III.

20
31

IV. The

Remedy
Solution of the Difficulty

V. The

39 44
Con-

VI. The Result of Applying the

VII.

Remedy

The

Four-Dimensional tinuum

Space-Time

57

VIII.

Some Consequences
Relativity

of the Theory of

69
Summary

IX.

Point of Logic and a

83

The Moral
Part
II

87

-THE GENERAL THEORY

II

GUIDE TO PART
X. Introduction

91

95
101

XI. The Principle of Equivalence

XII.
XIII.

Non-Euclidean World!

107
113

The Study of Spaces

Is

XIV. What

a Tensor?

127
of a

XV. The
XVI.

Effect

on Tensors

Change

in the
1

Coordinate System

41

A Very Helpful Simplification

ix

150

XVII. Operations with TenseXVIII.

160
167 173
1

A Physical

Illustration

XIX. Mixed Tensors

XX.

Contraction and Differentiation

Little g's

78

XXI. The
XXII.
XXIII.

187
191
at Last

Our

Last Detour

The Curvature Tensor

Is

200 206
21 3

XXIV. Of What Use

the Curvature Tensor?

XXV.

The Big G's or

Einstein's

Law Law

of Gravitation of Gravitation

XXVI. Comparison
XXVII.

of Einstein's

with Newton's

219

How

Can the

Einstein

Law

of Gravitation Be

Tested?

227
Difficulties

XXVIII. Surmounting the

237
255

XXIX. "The Proof

of the

Pudding"

XXX. More About

the Path of a Planet

of Mercury

266
272

XXXI. The Perihelion

XXXII.
XXXIII.
Deflection of a Deflection of a

Ray

of Light
Light, cont.

276
283

Ray of

XXXIV. The

Third of the "Crucial"

Phenomena

289 299
303

XXXV. Summary
The Moral

Would You Like

to

Know?

310
318

THE ATOMIC

BOMB

Parti

THE SPECIAL THEORY

I.

INTRODUCTION.

In order to appreciate the fundamental importance

of Relativity,
it is

necessary to
it

know
11

how

arose.
a "revolution

Whenever
in
it is

takes place,

any domain, always preceded by

a tension,

some maladjustment producing

which ultimately causes a break, followed by a greater stability

at least for the

time being.

What was
in

the maladjustment in Physics the latter part of the 19th century, which led to the creation of
11

the "revolutionary

Relativity
briefly:

Theory?

Let us summarize
It

it

has been assumed that

space is filled with ether,* through which radio waves and

all

light

waves

are transmitted

any modern child

*This ether
is

talks quite glibly

of course

NOT the

chemical ether

which surgeons use!

ft is it

not a liquid, solid, or gas,

has never been seen

by anybody,

presence is only conjectured because of the need for some medium to transmit radio and light waves.
its

11

about "wave-lengths in connection with the radio.

Now,

if

there

is

an ether,

does it surround the earth and travel with it, or does it remain stationary
while the earth travels through it?

Various known (acts* indicate that

the ether does

NOT travel

with the earth.

the earth is moving If, then, there must be an "ether wind/'

just as a

THROUGH

the ether,

person riding on a bicycle

still air,

through feels an

air

wind blowing

in his face.

so an experiment was performed and Morley (see p. 8) Michelson by in 1887,

to detect this ether wind/and much to the surprise of everyone, no ether wind was observed.

And

This
a

unexpected result was explained by Dutch physicist, Lorentz, in 1 895, in a way which will be described
in

Chapter

II.

The search for the ether wind was then resumed by means of other kinds of experiments.!
*See the
M
article

Aberration of Light",

by A.
in

Eddington, the Encyclopedia Britannica, 14th ed.

S.

tSec the article "Relativity" by James Jeans, also in the Enc. Brit. 14th ed.

But, again and again, to the consternation of the physicists, no ether wind could be detected,'
until
it

seemed

that
1

nature was in a "conspiracy' to prevent our finding this effect!

At

this

point

Einstein took

up the problem,
that
11

and decided

a natural "conspiracy must be a natural

And
as to

operating. to answer the question

this law,

LAW

what is he proposed
published
in

one

in

his Theory of two papers, 1905 and the other

Relativity,

in

1915.*

He first found it necessary to re-examine the fundamental ideas upon which classical physics was based, and proposed certain vital changes in them.
He
then

made

VERY LIMITED NUMBER OF MOST REASONABLE ASSUMPTIONS

by means
of

from which he deduced his theory. So fruitful did his analysis prove to be
that
it

he succeeded

in:

(1) Clearing up the fundamental ideas. (2) Explaining the Michelson-Morley experiment in a much more rational way than

had previously been done.

*Both

now

published

in

one volume

including also the papers

by

Lorentz and Minkowski,

to which

we

shall refer later/

see

SOME

INTERESTING READING, page 324.

(3)

Doing away with

other outstanding difficulties

in physics. (4) Deriving a

NEW LAW OF GRAVITATION

much more adequate than the Newtonian one

(See Part II.: The General Theory) and which led to several
important predictions

which could be verified by experiment; and which have been so verified

since then.

(5) Explaining

QUITE INCIDENTALLY
a famous discrepancy in astronomy which had worried the astronomers
for

many

(This also

years is discussed in

The General Theory).

Thus, the Theory of Relativity had
a profound philosophical bearing of physics, on as well as explaining

ALL

many SPECIFIC
that

outstanding difficulties

had seemed to be entirely

UNRELATED,
and
of further increasing our of the physical world by suggesting a number of

knowledge

NEW experiments which NEW discoveries.

No
has

have led to

other physical theory

been so powerful though based on so FEW assumptions.

As we

shall see.

II.

THE MICHELSON-MORLEY EXPERIMENT*

page 4 we referred to
set themselves.

On

the problem that

Michelson and Morley

Let us

now

see

what experiment they performed and what was the startling result.
In

order to get the idea of the experiment

in

very clearly
it

mind,
first

will

be helpful

to consider the following simple problem,

which can be solved

by any boy who

has studied

elementary algebra:

Imagine a river in which there is a current flowing with

velocity v, in the direction indicated

by the

arrow:

Now
for a

which would take longer

man

to

swim
to

From

to

B and back
in

'Published

the

Philosophical Magazine, vol. 24, (1887).

or

from
if

to

C and back

to

the distances

AB and AC are

equal,

AB
and

being parallel to the current,

in still

AC perpendicular to it? Let the man's rate of swimming

water

be represented by c / then, when swimming against the

from

current,

to

his rate

would be only

whereas,

when swimming with from 8 back to A ,

his rate

the current,

would, of course, be c

v.

Therefore the time required to swim from A to fi

would be a/(c v), where a represents the distance and the time required
for the trip

AB ;

from

to
v).

A
would be

would be a/(c

Consequently, the time for the round

trip

ti

or

ti

= =

a/(c

/)

2ac/(c

+
v
2

a/(c
).

v)

Now
how
from

let us

see
trip

long the round

to

C and back

to

would take. If he headed directly toward C , the current would carry him downstream, and he would land at some point to the left of C in the figure on p. 8.
Therefore, in order to arrive at

he should head
just far

for

some point

enough upstream

to counteract the effect of the current.

In
if

other words,
the water could

be kept

still

until

he swam

at his

own

rate c

from

to

and then the current were suddenly allowed to operate, carrying him at the rate v from D to C
(without his making any further effort), then the effect would obviously be the same
as his going directly from

iojC
2

with a velocity equal to Vc' v/ as is obvious from the right triangle:

ex

\r

Consequently/
the time required for the journey from

A
2

to

would be a/Vc^- v , where a is the distance from

Similarly,
in

to

it is

going back from C to easy to see that,

A,
10

by

the same

method

of reasoning,
2

_
v
2
.

the time would again be a/Vc Hence the time for the round trip

from

to

C and back
fa

to

would be

= 2a/vV ti

_
A,
f2

y\

In

order to compare

and

more
2
.

let us write ft for

c/

Vc
2

easily,

Then we
and

get:
ti

fa

2a/3

/c

2a/3/c.

Assuming that v is less than c , 2 2 v being obviously less than and c

the

Vc

is

therefore less than c

1

and consequently ft is greater than (since the denominator

is

less than
t\

Therefore
that
IT
is,

the numerator). is greater than

fa ,

TAKES LONGER TO

SWIM UPSTREAM AND BACK

THAN TO SWIM THE SAME DISTANCE

ACROSS-STREAM

AND

BACK.

But what has all this to do with the Michelson-Morley experiment? In that experiment, to B: a ray of light was sent from

C-r

^-

HB
11

At 8 there was a mirror which reflected the light back to

so that the ray of light makes the round trip from
just as the
in

A,
back,

AioB and

swimmer did

the problem described above. since the entire apparatus shares the motion of the earth,

Now,

which is moving through space, supposedly through a stationary ether/ thus creating an ether wind
in the opposite direction, (namely, the direction indicated above), this experiment seems entirely analogous to the problem of the swimmer.

And,
and

therefore/ as before/
ti

ti

= 2a0Yc = 2a|S/c.

0)
(2)

and

is now the velocity of light, the time required for the light to C and back to to go from

Where

*2 is

(being reflected from another mirror at

If/
ti

Q.

therefore, and t> are found experimentally/

by dividing (1) the value of /? would

then

by (2), be easily obtained.

2

And

since

= c/V c

-T

2 ,

c being the known velocity of light, the value of v could be calculated.

That

is,

THE ABSOLUTE VELOCITY

would thus become known.

OF THE EARTH

Such was the plan of the experiment.

Now

what actually happened?

12

The experimental values of t\ and were found to be the SAME,

instead of
ti
\

ti

Obviously

this

being greater than ti was a most disturbing

result,

quite out of harmony with the reasoning given above.

The Dutch

physicist, Lorentz, then suggested the following explanation of Michelson's strange result:

Lorentz suggested that

matter,

owing

to

its

electrical structure,
IT IS

SHRINKS
and
this

WHEN

MOVING,
MOTION.*

contraction occurs

ONLY IN THE DIRECTION OF of shrinkage The he assumes to be in the ratio of 1/ff

AMOUNT

v (where /3 has the value c/Vc Thus a sphere of one inch radius

as before).

becomes an
with
its

ellipsoid when shortest semi-axis

1//3 inches long)

it is

moving,

(now only

*The two papers by Lorentz on this subject are included in the volume mentioned in the footnote on page 5.
In the first of these papers Lorentz mentions that the explanation proposed here occurred also to Fitzgerald.

Hence

it

is

often referred to as
1

the "Fitzgerald contraction" or the "Lorentz contraction* or

11

the "Lorentz-Fitzgerald contraction.

13

in

the direction of motion,

thus:

a erection

to the

Applying this idea Michelson-Morlcy experiment,

the distance

AB (=
2

a)

on

p. 8,

becomes a/jS , and ti becomes 2a/3/c

instead of 2a/3 /c , so that now ft t2

just as the

experiment requires.

One

might ask

how

it is

that Michelson did not

observe the shrinkage? Why did not his measurements show

that

AB was

shorter than

AC

(See the figure on p. 8)? The obvious answer is that the measuring rod itself contracts

when applied to AB, so that one is not aware of the shrinkage.

To
this

explanation

Michelson-Morley experiment the natural objection may be raised

that an explanation
for the express

of the

which

is

invented

purpose

14

of smoothing out a certain and assumes a correction

of
is

difficulty,

JUST
too

artificial

the right amount, to be satisfying.

And

Poincare, the French mathematician/

raised this very natural objection.

Consequently, Lorentz undertook to examine

his contraction
in

hypothesis

other connections, to see whether it is in harmony also with facts other than
the Michelson-Morley experiment. He then published a second paper

in

1904,

giving the result of this investigation.

To present
let us
first

this result in a clear

form

re-state the

argument

as follows:

vt

Consider a set of axes, X and Y, supposed to be fixed in the stationary ether, and another set X' and Y' , attached to the earth and moving with it,

15

with velocity v

as indicated

above

Let

and

V"

move along X, move parallel to

V.

Now
is

suppose an observer on the

measure

earth,

say Michelson,
trying to
it

the time

takes a ray of light to B , to travel from

and 8 being fixed points on the moving axis X .

both
r

At

the

moment
ray of light
starts at
f

when the

suppose coincide, and A coincides with D / and while the light has been traveling to B
the axis

that

Y and Y

has

moved

the distance vt

and B has reached the position shown in the figure on p. 1 5,

t

being the time

if

it

Then, we have x'

This
is

DB = x and AB =

takes for this to happen.

x',
(3)

vt.

only another way of expressing what was said on p. where the time for

the

first part of the journey was said to be equal to a/(c

v).*

And, as we saw there, this way of thinking of

did

the

phenomenon

NOT agree

Applying now
*Since

with the experimental facts. the contraction hypothesis

we are now designating a by x', = t , or x = ct we have x'/(c v)

But the distance the light has traveled

vf.

is x , and x

ct,

consequently x'

=x-

vt

is

equivalent to a/(c

v)

t.

16

proposed by Lorentz, x should be divided by /3, so that equation (3) becomes

r

x7/3
or
x'

- vt - vt). (x
x
other fads,

(4)

Now

when Lorentz examined

on
p.
1

as stated

5,

he found that equation (4) was quite in harmony with ail these but that he was now obliged
to introduce a further correction

facts,

expressed by the equation

(5)

where /3 , t , v , x , and c have the same meaning as before But what is t?! Surely the time measurements in the two systems are not different:

Whether the

origin

is

at

or at

should not affect the

TIME-READINGS.
In
t'

was a

other words, as Lorentz saw it, 11 sort of "artificial time

introduced only for mathematical reasons,

because
in

it

helped to give
facts.
t

results

harmony with the

had

But obviously

for

him

NO
As

PHYSICAL MEANING.
it:

Jeans, the English physicist, puts

"If the observer

could be persuaded

artificial way, wrong to begin with and then making them gain or lose permanently, the effect of his supposed artificiality

to measure time in this

setting his clocks

17

would

just

counterbalance
11

the effects of his motion

through the ether

!*

Thus, the equations finally proposed

are:

by Lorentz

x'

= =

(x

vt)

z'

Note

that
1

since the axes attached to the earth (p. are moving along the X-axis,

5)

obviously the values of y and z

(z

being the third dimension)

are the

same

as

and z

respectively.

The equations
are

just

given

known

as

THE LORENTZ TRANSFORMATION,

since they show how to transform a set of values of x , y , z , t
into a set x',
in a

y, z,

t'

coordinate system moving with constant velocity

v,

along the X-axis,

with respect to the

unprimed coordinate system.

And,

as

we

saw,

whereas the Lorentz transformation

really expressed the facts correctly, it seemed to have

NO

PHYSICAL MEANING,
article

*See the

on

Relativity in the

Encyclopedia Britannica, 14th edition.

19

and was merely

a set of empirical equations.

Let us

now

see what Einstein did.

III.

RE-EXAMINATION OF THE

FUNDAMENTAL

IDEAS.

As Einstein regarded the situation, the negative result of the Michelson-Morley experiment,
as well as of other experiments

which seemed to indicate a "conspiracy" on the part of nature

against man's efforts to obtain knowledge of the physical world (see p. 5),

these negative results,

according to Einstein, did not merely demand

explanations of a certain number
of isolated difficulties,

but the situation was so serious

that a

complete examination

of fundamental ideas

was necessary.
In

he
in

other words, felt that there was something

fundamentally and radically wrong

physics,
rather than a

mere

superficial difficulty.

And

so he undertook to re-examine

such fundamental notions as

our ideas of

LENGTH

and

TIME and MASS.

20

His exceedingly reasonable examination

is

most illuminating,

as

we

shall

now

see.

But

first

let us

remind the reader

why

length, time

and mass
that

are fundamental,

Everyone knows

VELOCITY depends upon the distance (or LENGTH)

traversed in a given

TIME,

hence the unit of velocity

DEPENDS UPON
the units of
Similarly,

LENGTH
is

and TIME.

since acceleration

the change in velocity in a unit of time, hence the unit of acceleration

DEPENDS UPON
the units of velocity and time, and therefore ultimately upon
the units of
Further,

LENGTH
is

and TIME.

since force

measured
of

by the product

mass and acceleration, the unit of force

DEPENDS UPON
the units of mass and acceleration,

and hence ultimately the units of

upon

MASS, LENGTH And so on.

In
all

and TIME,

other words,

measurements

in

physics

depend
That
is

primarily on

MASS, LENGTH
why

and TIME.

21

the system of units ordinarily used is called the "C.G.S."

where C stands

for

system,^ "centimeter"

(the unit of length), stands for "gram" (the unit of mass), and 5 stands for "second" (the unit of time)/

these being the fundamental units from which all the others are derived.

Let us

now

return to

Einstein's re-examination of

these fundamental units.

Suppose

that

two observers

wish to compare their measurements of time. If they are near each other

they can, of course/ look and compare them.

at

each other's watches

If they are far apart/ they can still compare each other's readings

BY

MEANS OF

SIGNALS,

say light signals or radio signals/ 1 that is/ any "electromagnetic wave* which can travel through space. Let us/ therefore/ imagine that

one observer/ f , is on the earth/ and the other/ 5 / on the sun/ and imagine that signals are sent
as follows:

By his own watch/ 5 sends a message to which reads "twelve o'clock/" f receives this message
say/ eight minutes later;*
*Since the sun is about from the earth,

93 000 000

miles

light travels about 186 000 miles per second, the time for a light (or radio) signal to travel from the sun to the earth/

and

is

approximately eight minutes.

22

now,
it

if

his

watch agrees with that of S n f<

will

read

12:08

when

the message arrives.

then sends back to

the message "12:08,"

and, of course,
receives this message 8 minutes namely, at 12:16. Thus 5 will conclude,

later,

from

this series of signals,

that his

watch and that of f

are in perfect agreement.

But

let us

now imagine

that the entire solar system

is moving through space, so that both the sun and the earth

moving in the direction shown in the figure:

are

without any change

in

the distance between them.

Now

let

the signals again

be

sent

as before:
his message "1 2 o'clock," but sincejf js moving awayjrom jjta IRelatter wiffnot reaclTE in 8 minutes,

S sends

but will take

to overtake

some longer time

,

Say,

minutes.

23

If
it

Fs watch
will read

is in agreement with 12:09

that of

5,

when
and

the message reaches him,

accordingly sends a return message/ reading "12:09."

Now 5
and
in
it

is

traveling toward this message/

will therefore

reach him

LESS than 8 minutes,

say, in 7 minutes.

Thus S receives
at 1 2:1 6,

Fs message

just as before.

Now

if

5 and F

are both

UNAWARE
(and, indeed,

of their motion

we
in

are

undoubtedly moving
that

ways

we

are entirely unaware of,

so that

assumption is far from being an imaginary one.) 5 will not understand

this

why Fs message reads "12:09" instead of "12:08," and will therefore conclude that Fs watch must be fast.

Of

course, this

is

only
in

an apparent error
because, as
it is

Fs we know,

watch,

really

due

to the motion,

and not
to
It

at all

any

error in

Fs

watch.

must be noted, however, that this omniscient "we"

who
what

can see exactly

is

"really" going on
exist,

in

the universe,

does not

and

that all

human observers

24

are really in the situation

in

which 5

is,

namely,
that of not knowing about the motion in question, and therefore being OBLIGED to conclude that 's watch is wrong!

And

5 sends

therefore, the message

telling him that if sets his clock

back one minute,

then their clocks will agree.

In

suppose
all

the same way, that other observers,

of

j4,B,C,ctc.,

whom
,

are at rest

WITH RESPECT

TO
S
,

5 and
all

set their clocks to agree with that of

by the same method of signals described above. They would all say then
that all their clocks are in agreement. Whether this is absolutely true or not,

they cannot tell (see above), but that is the best they can do.

Now
when

let us

see what will happen

these observers wish

To measure the length you can place it,

to measure the length of something. of an object,

say, on a piece of paper,

put a mark on the paper at one end of the object, and another mark at the other end,
then, with a ruler,
find out

how many

units of length there are

25

between the two marks.

This
is quite simple provided that the object you are measuring and the paper are at rest (relatively to you).

But suppose the object

say, a
fish

is

swimming about
its

in a

tank?

To measure

length while it is in motion, by placing two marks on the walls of the tank, one at the head, and the other at the tail,
it

to

SIMULTANEOUSLY for,
if

would obviously be necessary make these two marks

otherwise, the mark 6 is

made

at a certain time,

B
/
V-

then the
in

fish

allowed to swim

the direction indicated

at the

by

the arrow,

and then the mark

is made when it

head

at

some

later

time,

has reached

C,

then you would say that the length of the fish is the distance BC,

which would be a

fish-story

indeed!

Now

suppose

that our observers,

all in

after their

clocks are

agreement (see p. 25),

undertake to measure
the length of a train

26

which

is

moving through

their universe

with a uniform velocity. They send out orders that

at 1

2 o'clock sharp, whichever observer happens to be

place where the front end of the train,

at the

A'9

arrives at that

moment,

to

NOTE THE SPOT;

and some other observer, who happens to be at the place where the rear end of the train, B , is at that same moment, to put a mark at THAT spot.
Thus, after the train has gone,

they can, at their leisure, measure the distance between the two marks, this distance being equal to
the length of the train, since the two marks were

made
at

SIMULTANEOUSLY,
their clocks
in

namely

12

o'clock,

being

all

perfect agreement with each other.

Let us

now

talk to the

people on the

train.

Suppose, first, that they are unaware of their motion, and that, according to them, A f B f ^, t etc., are the ones who are moving,
a perfectly reasonable assumption. And suppose that there are two clocks

on the

train,

one and

at A', the other at B',

that these clocks

in

have been set

agreement with each other

of signals described above. , B , C, etc., Obviously the observers admit that the clocks at and B' wit!

by

the

method

NOT

27

arc in agreement with each other, since they "know" that the train is in motion,

and therefore the method of

signals

used on the moving train has led to an erroneous setting

of the moving clocks (see p. 25). Whereas the people on the train,

since they

"know"
it is

that

A, B , C,
claim that

etc./ are the

ones

who

are moving,

the clocks
etc.,

belonging to A, 8, C, which were set wrong.

What

is

the result of this

difference of opinion?

When
and

the clocks of

and B

say,

both read 12 o'clock,

at that instant

and B

each makes a mark at a certain spot, then A and B claim, of course,

that these

marks were

made
do not admit

simultaneously; but the people on the train

that the clocks of

and 8
set,

have been properly

and they therefore claim that the two marks were

NOT
and

made SIMULTANEOUSLY,

that, therefore,

the measurement of the

is

LENGTH

of the train

NOT correct.
train

Thus,

when the people on the make the marks

simultaneously,
as

judged by their own clocks, the distance between the two marks

28

will

NOT be the same as

that

before.

Hence we see

MOTION
prevents agreement
setting of clocks,
in

the

and, as a consequence of

this,

prevents agreement in the measurement of LENGTH!

Similarly,
as

we

shall

see on p. 79,
affects

motion also

the measurement of mass

different observers obtaining
different results

when measuring the mass

of the

same object.

And
as
all

since,
p.

we mentioned on

21,

other physical measurements

depend upon
length, mass,
it

and time,

seems

that

therefore there cannot

in

be agreement any measurements made

velocities!

by different observers who are moving with different

Now, of course, observers on the earth

partake of the various motions
to which the earth
is
its

subject
axis,

the earth turns on

it

goes around the sun, and perhaps has other motions as wefl.

Hence

it

would seem

that

observations

made by people on

the earth

29

cannot agree with those taken from

some

other location in the universe,

and are therefore

not really correct

and consequently worthless!

Thus Einstein's careful and reasonable examination
led to the realization that

Physics was suffering from

no mere
as

single ailment,

evidenced by the Michelson-Morley experiment alone, but was sick from head to foot!
find a

Did he

remedy?

HE

DID!

IV.

THE REMEDY.

So

far,

then,

we

see that

THE OLD IDEAS REGARDING THE MEASUREMENT OF LENGTH, TIME AND MASS
involved an "idealistic" notion of 11 "absolute time

which was supposed to be

the same for
all

observers,

and
a

that

Einstein introduced

more

PRACTICAL

notion of time
of

based on the actual way of

setting clocks

by means

SIGNALS.

This led to the

DISCARDING

of the idea that

31

the
is

LENGTH
is

of an object

a (act

about the object

and

independent of the person

III.)

who does the measuring/ since we have shown (Chapter

that the

measurement of length

DEPENDS UPON THE STATE OF MOTION OF THE MEASURER.

Thus two observers,

moving

relatively to each other

with uniform velocity/

DO NOT
LENGTH
is

FOR THE LENGTH OF

Hence we may
but rather a
the

GET THE SAME VALUE A GIVEN OBJECT.

say that

NOT a FACT about an

OBJECT,

RELATIONSHIP between OBJECT and the OBSERVER. And similarly for TIME and MASS
In

(Ch.

III.).

other words,
this

from
it is

point of view
say:

NOT CORRECT to
x'

vt

Michelson did* (see p. 16, equation (3) since this equation implies that the value of x'
as
is

),

a perfectly definite quantity,

*We do

Michelson made

not wish to imply that a crude error

ANY

CLASSICAL PHYSICIST

would have made the same statement, for those were the prevailing ideas
thoroughly rooted in everybody's mind, before Einstein pointed out the considerations discussed in Ch. III.

32

namely,

THE
in

length of the arm AB of the apparatus the Michelson-Morley experiment

(See the diagram on p. 1 5). Nor is it correct to assume that

(again as Michelson did)

two different observers, which would imply that both observers agree in their time measurements.
for

These ideas were contradicted by Michelson's EXPERIMENTS, which were so ingeniously devised

and so precisely performed.

And

so Einstein said that

instead of starting with such ideas,

and basing our reasoning on them,

let us rather

START WITH THE EXPERIMENTAL DATA,

and see
they
to

what relationships

will lead us,

relationships

between

the length and time measurements of different observers.

Now
must

we

what experimental data take into account here?

They

are:
(1):
It is

FACT

impossible

to measure the "ether wind,"

words, impossible to detect our motion relative to the ether.

it is

or, in other

This was clearly

shown by the

33

Michelson-Morley experiment,

by all other experiments devised to measure this motion (see p. 5).

as well as

Indeed, this is the great "conspiracy* that started all the trouble, or, as Einstein prefers to see
1

it,

and most reasonably

THIS

IS

so,

FACT.

FACT

(2):

The velocity of light is the same no matter whether the source of light
is

moving or
fully,

stationary.
this

Let us examine

statement

more

to see exactly what

it

means.

To do
it is

this,

necessary to remind the reader

of a few well-known facts: Imagine that we have two trains, one with a gun on the front end, the other with a source of sound

on the

front end,

say, a whistle. Suppose that the velocity, u , of a bullet shot from the gun,

happens to be the same

as

the velocity of the sound.

suppose that both trains moving with the same velocity, v , in the same direction. The question is: How does the velocity of a bullet
are
(fired

Now

from the

MOVING train)
34

relatively to the ground,

compare with

that

the velocity of the sound came from the whistle

on the other
in

MOVING
traveling?

train,
air,

relatively to the

medium/ the

which

it

is

Are they
No!

the

same?

The velocity of the

is

bullet,

RELATIVELY TO THE GROUND,

v

,
is

since the bullet

not only wiih

given to
but, in
it

its

now propelled forward own velocity, u ,

by

the force of the gun,

addition, has an inertia! velocity, v , which it has acquired from

the motion of the train

and which
all

is

shared

by
wave

objects on the train.

in

But

the case of the sound

is

a series of pulsations, (which alternate condensations and rarefactions of the air

in rapid succession), the first condensation formed in

travels

the neighborhood of the whistle, out with the velocity u

relatively to the medium, regardless as to whether

the train

is

moving or not.
condensation

So

that this
its

has only

own

velocity
inertia!

and does

have the due to the motion of the

NOT

velocity

train,

the velocity of the sound depending only upon the

medium

35

(that

is,

whether
it is

it is

air

or water/ etc.,

and whether

hot or cold, etc.), but not upon the motion of the source from which the sound started.

The following diagram shows the relative positions after one second/
in

both cases:

CASE
Both

I.

trains at rest.

Tram

jt.

-ft.

CASE
Both
trains

II.

moving with velocity

v.

v
UfV

u 4*.
Thus/ the bullet has
in

Case

II.,

moved
one second

v feet

in

36

from the starting point, whereas the sound has moved only u feet from the starting point/
in that

one second.
see that

Thus

we

the velocity of sound is u feet per second relatively to the starting point,

whether the source remains stationary

as in Case
I.,
II.

or follows the sound, as in Case

Expressing

it

algebraically,

ut

applies equally well for sound

in both Case I. and Case x being the distance
II.,

FROM THE STARTING

Indeed,
this fact
is

true of

POINT. ALL, WAVE

MOTION,

and one would therefore expect that it would apply also to LIGHT.

As a matter it DOES,
and
that
is

of

FACT,
is

what

meant by

FACT
Now,
it

(2)

on

p. 34.

as a result of this,

appears,
referring again to the

by

diagram on p. 36,
train

that

relatively to the

MOVING
(u

(Case

II.)

we should
for

then have,
x'

sound

v)t

x being the distance from T to the point where

the sound has arrived after time
t.

37

From which, by measuring x

then calculate v the velocity of the train.

'

and t,

we could

And,

similarly, for light

using the moving earth instead of the moving train,

we should
as a

then have,

consequence of
x'

FACT

(2)

on p 34,

(c

v)t

where
out
in

is

the velocity of light

(relatively to a stationary observer

and v

space) the velocity of the earth relatively to this stationary observer

is

and hence
the

ABSOLUTE

velocity of the earth.

Thus
But
it is

we should be

able

to determine v.
this contradicts

FACT

(1),

according to which

IMPOSSIBLE
it

to determine v.
that

Thus

APPEARS

(2) requires the velocity of light

FACT
be

RELATIVELY TO THE
to
c

MOVING EARTH
it

v (see diagram on p. 36),

whereas

FACT
(1)

(1) requires

to

be

c.*

*FACT

may be

re-stated as follows:

The velocity of

light

RELATIVE TO
must be

A MOVING
c

OBSERVER

(For example, an observer on the moving earth)

c,

and

NOT

vt
,

for otherwise,

he would be able to find v which is contrary to fact.

38

And
Or,

so the two facts

contradict each other!

stating
it

another way:

If,

in

the earth

one second, moves from E

to E'

while a ray of light, goes from the earth to L

then

FACT(1)
E'/.

be equal

requires that to c ( = 1

86,000 miles)

while

FACT
it is

EL be

(2) requires that to c / equal

Now

needless to say that

FACTS

CAN NOT CONTRADICT

EACH OTHER!
Let us therefore see how,
in

FACTS

the light of the discussion in Ch. III. (1) and (2) can be shown to be
contradictory.

NOT

V. THE

SOLUTION OF THE DIFFICULTY

We
IS

have thus seen that

facts,

according to the

the velocity of light

ALWAYS THE SAME,

39

whether the source of

is

light

stationary or

moving
light

FACT (2) on p. 34), and whether the velocity of is measured

(See
relatively to the

medium

in

which

it

travels,

or relatively to a

MOVING

observer

(See p. 37).
Let us express these
for
(acts algebraically,
f

two observers, K and K , who are moving with uniform velocity

relatively to each other,

thus:

K writes
and K'

ct

(6)
ct',

writes x'

(7)

both using

THE SAME VALUE FOR THE VELOCITY OF LIGHT,

namely, c , and each using his own measurements of length, x and x', and time, t and t', respectively.
It is

assumed

that

at the instant

when

the rays of light start on their path, and K' are at the place, and the rays of light

SAME

radiate out from that place

in all directions.

Now according to equation (6), K who unaware of his motion

,

is

through the ether

(since he cannot measure it), may claim that he is at rest, and that in time, t ,

40

K' must have

moved

to the right,

shown and that/

as

in in

the figure below/ the meantime,

the light/

which

travels out in all directions from

all

has reached

points at

the distance ct from

K,

and hence all points on the circumference

of the circle having the radius
ct.

But

W claims that he
K
moved

is

the one

who
and
has

has remained stationary/ on the contrary, that /

TO THE

LEFT/

furthermore that the light travels out f from K as a center,

instead of from K\

And

this

is

what he means

when he

says
x'

ct'.

How

can they both be right?

41

We

may be

willing

not to take sides

in their

controversy regarding the question as to

which one has moved

K' to the right or

K to

the

left

because either leads to the same

But what about the circles?

result.

,-?

They cannot possibly have both

'

K and K

as their centers!

One
This

of
is

them must be

right

and the other wrong.

another

way

of stating

the

APPARENT CONTRADICTION BETWEEN

(1)

FACTS
Now,

and (2) (see

p. 39).

at last,

we

are ready

for the explanation.

Although

K claims
when

that

at the instant

the light has reached the point C(p. 41), it has also reached the point
still,

on the other

side,

WE MUST REMEMBER THAT

when

K says
arrival of

two events happen simultaneously

(namely, the
K'
the light at

C and

A),
(see p. 28).

DOES NOT AGREE THAT THEY ARE SIMULTANEOUS

So
that

when

K says that the arrival of the lisht at (rather than at C and A)

his

C and B

ARE SIMULTANEOUS,
statement

DOES NOT CONTRADICT THAT OF K, since K and K' DO NOT MEAN THE SAME THING
42

WHEN THEY SAY

"SIMULTANEOUS-.
for
11

K's clocks at

C and A

do

not agree with K'*s clocks at C and A. Thus when the light arrives at A ,

the reading of K*s clock there is exactly the same as that of K's clock at

(K having set all clocks in his system by the method of signals described on
while n K s clock at

p. 25),

Ai
C

when when

the light arrives there/ reads a LATER TIME than his clock at the light arrived at so that K maintains that the light reaches LATER than it reaches

C,

C,
instant,

and
as

NOT at the SAME

that

K claims.

Hence we see

they are not really contradicting each other, but ihat they are merely using
different systems of clocks, such that

two

the clocks in each system agree with each other alright,

but the clocks

in

the
set

one system

have
in in

NOT

been

agreement with the clocks

the other system (see p. 28).
is,

That
If

we

take into account

the inevitable necessity of

using signals in order to set clocks which are

43

at a distance from

each other,

and

that the arrivals of the signals

at their destinations

are influenced

by

our state of motion, of which we are not aware (p. 24),

it

becomes
IS

clear that

THERE
due
in

NO

REAL CONTRADICTION HERE,

differences

but only a difference of description

to

INEVITABLE

the setting of various systems of clocks.

We
in a

now

see

general qualitative way,

is

that the situation

not
as
it

at all

mysterious or unreasonable,
to

seemed

be

at

first.

But

we must now

find out

whether these considerations,

when applied

QUANTITATIVELY,

actually agree with the experimental facts.

And now

a pleasant surprise awaits us.

VI.

THE RESULT OF APPLYING THE REMEDY.

chapter

In

the

last

we saw

that

by

starting with

two fundamental

FACTS

(p. 34),

we reached
expressed

the conclusion

in

the equations

x and
x'

= =

ct
ct'

(6) (7)

44

which are graphically represented on p. 41, and we realized that these equations
are
(as
if

NOT

contradictory,
at
first),
is

they appear to be

we remember

that there

a difference in the setting of the clocks

in

the two different systems.

shall derive,

We

now, from (6) and (7), between the measurements of the two observers, K and K
relationships
f
.

And
is

all

the mathematics

we need

for this

a little simple algebra,

such as any high school

boy knows..

From (6) and (7) we get

x
and
Therefore
x' x'

ct
ct'

=0
=
0.

ct'

X(x

ct)

(8)

where X

is

a constant.

Similarly, in the opposite direction,

x'

ct'

/x

(x

ct)

(9)

H being another constant. By adding and subtracting

(8) and (9)

we
and

get:

x'
ct'

= ax - feet = act-fex
ju)/2

(10) (11)

where a
Let us
in

(X

-f-

and b

(X

fe

ju)/2.

now

find the values of a

and

terms of v

and c

and /C (the relative velocity of the of light. velocity ,

),

45

This

is

done

in

the following

ingenious manner:*

when x = , then x = bet/ a

but x'

From (10)

(12)

at the point K'\

And
that

in this

case

is

the distance from

is,

K to

K! ,
t

the distance traversed/ in time f by K moving with velocity v

relatively to K.

Therefore x

=
this

vt.

Comparing

with (1 2),

we

get

v
Let us

be/a.

(13)

now
first:

consider the situation

from the points of view of

K and

K'.

Take

For the time

/Cgetsx'
or

ax (from (10)),

x'/a.

(14)

Hence

K says

that
11

*See Appendix in "Relativity by Einstein, Pub. by Peter Smith, N. Y. (1931).

I

46

to get the "true" value,

K should

divide his x

x, by a/

in particular,

ifx'-1,

K
is

says that

K"s

unit of length
unit.

only 1/a of a "true"

But
att'

K,

0,usins(11)

says

fex=act
and since from (10),

(15)

(1

5)

becomes
6x
ac(ax
x')/6c
f

62^2 "x ax
from which
x'

ax

a(1

6 /a

)x.

(16)

And

since 6/a

=
=

y/c from

(1 3),

(16) becomes

x
In

a(1

vVc-)x.

(17)

other words, K' says: In order to get the "true" value,

x',

K should

multiply his x
a(1

by

v /c).

In particular,

then
f('s

ifx- 1, K says
unit
is

that
2

really a(1

vr/c

) units

long.

Thus

47

each observer considers that his own measurements are the "true" ones,

and advises the other fellow

to

make

a "correction.*

And indeed, although the two observers,

may
still

K and
1

/(',

express this "correction*

forms,
of the "correction"

in different

the

MAGNITUDE

recommended by each of them MUST BE THE SAME, since it is due in both cases
to the relative motion, only that each observer attributes this

motion

to the other fellow.

Hence, from (14) and

1/a
Solving
this

(1 7)

we may
v /c
,

write

a(1

).*

equation
a

for a

we

get

= c/V?~-v*.
is

*Note

that this equation

NOT obtained

by

ALGEBRAIC SUBSTITUTION
from (14) and (17), but is obtained by considering that the CORRECTIONS advised by K and K'
in (14) and (17), respectively, must be equal in magnitude as pointed out above. Thus in (14) K says: "You must multiply your measurement by 1/a", whereas in (1 7) K' says:

"You must multiply your measurement by and since these correction factors must be equal

(1

48

Note
is

that this value of a

/?

the same as that of

on

p. 1

Substituting in (10) this value of a

and the value

fee

ay from (13),

we
or

get
x' x'

= =

0x - |8vf /3(x-rt)

(18)

which

is

the

first

of the Lorenti transformation

of the set of equations on page 19!

Furthermore,

from

(1

8) and /

\x'

= =

ct
ct'

we
or

get
ct'
t'

= = =

P(ct
/3(t

- vt) - vt/c).
vx/c ),
2

Or, since

=
t'

x/c,
/3(t

(19)

another of the equations of the Lorentz transformation!

which

is

That the remaining two equations = y and z' z also hold, y' Einstein shows as follows:

Let

K and

/C'

each have a cylinder

of radius r, when at rest relatively to each other,

and whose axes coincide with the Now, unless / = y and z = z, K and K would each claim that
r f

X (X ) axis/
7

his

own

cylinder

is

OUTSIDE

the other fellow's!

We
by

thus see that

the Lorentz transformation was derived

Einstein

(quite independently of Lorentz), as a set of empirical equations

NOT

49

devoid of physical meaning, but, on the contrary,

as a result of
a

most rational change

in

our ideas regarding the measurement of the fundamental quantities

length and time.

And
the

so, according to him,

of the equations of the Lorentz transformation,

first

namely,
x'
is

/3(x

vt)

so written

as Lorentz

because of any real shrinkage, supposed, but merely an apparent shrinkage,

to the differences
Einstein writes
t'

NOT

due

in

the measurements

made by

K and
2

/C

(see p. 45).

And

/3(t

vx/c

NOT because
it is

it is

just a

mathematical

trick

WITHOUT any MEANING

but again because
the differences
of the

(see p. 19)

the natural consequence of in the measurements

two observers.

And
that

each observer may think he is right

is

and the oiher one and yet

each one,

wrong/

by

using his

own measurements,
same form
it

arrives at the

*This shrinkage,

will

be remembered,
1 3).

occurs only

in

the direction of motion (see p.

50

when he expresses
as, for

a physical fact,

example,

when
and

K says
f

ct' , says x' they are really agreeing as to the of the propagation of light.

= =

ct

LAW

And similarly, K writes any

if

other law of nature,

and

if

we apply

the Lorentz transformation

to this law, in order to see

what form the law takes

expressed in terms of 7 the measurements made by /C ,

it is

when

we

find that
is still

the law

the same,

although it is now expressed in terms of the primed coordinate system.

Hence

Einstein says that

although no one knows what the "true" measurements should be,

yet,

each observer
in

may

use his

own measurements

WITH EQUAL RIGHT

formulating

AND EQUAL

SUCCESS

THE
or,
in

LAWS OF NATURE,

formulating the of the universe, the which remain unchanged quantities namely, in spite of the change in measurements

INVARIANTS

due

to the relative motion of

K and K

Thus,

we

can

now

appreciate

Einstein's Principle of Relativity:

"The laws by which

52

the states of physical systems

undergo change,
are not affected

whether these changes of

to the

state

be

referred

one

or the other

of
in

two systems of coordinates

uniform translatory motion.

"

"But

Perhaps some one will ask is not the principle of

it

relativity old/

and was

not known long before Einstein?

in a train

Thus a person

moving

into a station

with uniform velocity

looks at another

train

which

is

at rest,
is

and imagines
whereas
his

that the other train

is

moving

own

at rest.

And
by
is
if

he cannot

find out his mistake

making observations within his train

the same as
it

since everything there

just

would be
at rest.

his train

were really

Surely

this fact,

and other

similar ones, must have been observed

long before Einstein?"

In

other words,
to an observer

RELATIVELY

on the

train

everything seems to proceed in the same way whether his system (i.e., his train)
is

at rest or in uniform*

motion,

and he would therefore be unable

*0f
course
if

the motion

is

not uniform,

but "jerky",
things on the train would jump around and the observer on the train

would
that his

certainly

know
was not
at rest.

own

train

53

to detect the motion.

Yes, this certainly was known long before Einstein. Let us see what connection it has
as stated

with the principle of relativity by him:

Referring to the diagram

on

p.

36

we

see that

a bullet fired from a train

has the

same velocity

RELATIVELY TO THE TRAIN

whether the latter is moving or not, and therefore an observer on the train could not detect the motion of the train by making measurements on the motion of the bullet.
This kind of relativity principle
is

the

one involved

in

the question on page 53,

and

WAS

known long before

Einstein.

Now Einstein EXTENDED this

so that
it

would apply

principle to

electromagnetic

phenomena

(light or radio waves).

Thus,

according to

this

extension of

the principle of relativity, an observer cannot detect

his

motion through space

by making measurements on
the motion of

ELECTROMAGNETIC WAVES.
this

But

why should
a great
it

extension

be such

achievement

why had

not been suggested before?

54

BECAUSE
it

according to fact (2)

must be remembered that see p. 39,

whereas, the above-mentioned extension of

the principle of relativity
requires that EL should be equal to c (compare the case of the bullet on p. 36).
In other words, the extension of the principle of relativity

to electromagnetic

phenomena

seems to contradict fact (2) and therefore could not have been made before it was shown that fundamental measurements are merely "local" and hence the contradiction was
only apparent, explained on p. 42; so that the diagram shown above must be interpreted in the light of the discussion on p. 42.
as

Thus we see that whereas the principle of relativity as applied to MECHANICAL motion
(like that of the bullet)

was accepted long before Einstein,

the

SEEMINGLY IMPOSSIBLE EXTENSION

phenomena
55

of the principle to electromagnetic

was accomplished by him.

This extension of the principle, for the case in which and K' move relatively to each other with velocity, and which has been discussed here/ is called the SPECIAL theory of relativity.

UNIFORM

We

shall

see

later

how

Einstein generalized this principle

STILL FURTHER,
to the case in which

K and
that
is,

K?
a

move

relatively to each other

with an

ACCELERATION,

CHANGING

And, by means
which he called
the

velocity. of this generalization/

GENERAL

theory of relativity,

he derived

A NEW LAW OF
Newtonian
first

GRAVITATION/
than

much more adequate even

the

and
is

law, of which the latter

approximation.

But before

we

we must how the

first

can discuss see

this in detail

ideas which

we have

already presented were put into a remarkable mathematical form by a mathematician named Minkowski.
This
in

work was essential to Einstein

as

the further development of his ideas/ we shall see.

56

VII.

THE FOUR-DIMENSIONAL SPACE-TIME

CONTINUUM.

We shall now see how Minkowski* put

in a
in

Einstein's results

remarkably neat mathematical form, and how Einstein then utilized this
the further application of his Principle of Relativity,

which led to The General Theory of Relativity,

resulting in a

NEW LAW OF GRAVITATION

and leading to further important consequences and discoveries.

NEW

It is

now

clear
1

from the Lorentz transformation (D.

that

9)

measurement, x , one coordinate system depends upon BOTH x and and that
a length
in
t'

t in

another,

depends upon Hence^

also

BOTH

x and

t.

as

instead of regarding the universe being made up of

Space, on the one hand, and Time, quite independent of Space,

there
is

a closer connection

between Space and Time than we had realized.

In

other words,
in

*See collection of papers mentioned

footnote on p. 5.

57

that the universe

is

NOT a

universe of points,

with time flowing along irrespective of the points,

but rather,
this
is

UNIVERSE OF EVENTS,
at a certain

everything that happens,

happens

place

AND at a

certain time.

Thus, every event is characterized by the PLACE and TIME of its occurrence.

Now,
since
its

place

may be

designated

by
By

three numbers,

namely,
the
(using

x, y, and z co-ordinates of the place any convenient reference system),

it,

and since the time of the event needs only one number to characterize we need in all

FOUR NUMBERS TO CHARACTERIZE

just as

AN

EVENT,

we need

three numbers to characterize

a point in space.

Thus

we may

say that

we

live in a

four-dimensional world.
This
that

does

NOT

mean

we
is

but
that

Space, only another way of saying

live in

live in four-dimensional

we

A WORLD OF
rather than of

EVENTS POINTS only,

58

and

it

takes

FOUR

numbers to designate

each significant element, namely, each event.

Now
by
in a

if

an event

is

designated

the four numbers

x, y, z,

f ,

given coordinate system, the Lorentz transformation (p. 19)

shows how to

find
y', z', t',

the coordinates x', of the same event,

in

moving

another coordinate system, relatively to the first

with uniform velocity.

11

In

studying "graphs every high school freshman learns

how

to represent a point

by two
that
is,

coordinates, x and y, using the Cartesian system of coordinates,

straight lines

two

perpendicular to each other. Now, we may also use

another pair of perpendicular axes, X' and Y' (in the figure on the next page),
as before, , having the same origin, and designate the same point by x and
in this

y'

When

coordinate system. the high school boy above-mentioned

new

goes to college, and studies analytical geometry, he then learns how to find

60

/ H
.-.
fl

't^-

J
6
\

the relationship between

the primed coordinates and the original ones/

and

finds this to

be expressed
x'cosfl
x'sinfl

as follows:*

x
anc

= =

/sine

/cos0 /

(20)

where 6 is the angle through which the axes have been revolved, as shown in the figure above.

The equations (20) remind one somewhat

of the Lorentz transformation (p. 19),

since the equations of

*Seep. 310.

61

the Lorentz transformation

also

show how

to

go

from one coordinate system to another.

Let us examine the similarity between (20) and the
Lorentz transformation
a little

more closely,

selecting from the Lorentz transformation

only those equations involving x and t , and disregarding those containing y and z, since the latter remain unchanged in going from one coordinate system to the other.
Thus

we

wish to compare (20) with:

/ t

x'
t'

|8(x
/3(t

vt)
2

vx/c

).

Or,
that

if,

for simplicity,

we
by

take c

1,

taking the distance traveled

is,

light in

one second,

as the unit of distance,

we may we wish

say that
to

compare (20) with

x'
t'

= /3(x- vt)) = P(t- vx)f

x and
t ,

Let us

first

solve (21) for

so as to get them more nearly in the form of (20).

By ordinary algebraic operations,*

*And temembering we are taking c =
and
that therefore

that
1,

62

we
and

get

x
t

= =

fat
j8(t'

+ vt') + vx') }
)

'

Before

let us linger a

we 90 any further, moment

and consider equations (22): Whereas (21) represents /(speaking, and saying to K "Now you must divide x by /3,
'.

before you can get the relationship between x and x' that you expect, namely, equation (3) on p. 16;
in other words, your x' has shrunk although you don't know it."

In

(22),

speaking, tells K the same thing, namely that K must divide x by

it is

and he

/? ,

to get the "true" which is equal to

x x

vt'.

Indeed,
this is

quite in accord with the discussion in Chapter VI.,

in

which

it

was shown

that

each observer
gives the other one precisely the same advice!

Note

that the

only difference
is

between (21) and (22)

that

+
in

y becomes

going from one to the other.

63

And this is again quite in accord with our previous discussion since each observer
believes himself to be at rest, and the other fellow to be in motion, only that one says:

"You have moved to the whereas the other says: "You have moved to the
Otherwise,

right*'

(+

v),

left*' (

v).

their claims are precisely identical;

and

this is exactly what equations (21) and (22) show so clearly.

Let us
of (22)
in

now

return to the

comparison
that

and (20): Minkowski pointed out

(22),

if,

t is

replaced
t'

by

IT

(where

= V

1 ),

and

by

IT',

then (22) becomes:

x
IT

= =

/3(x'

/?(//

+ ivr') + VX')

or

/ X \ IT

= =

frt'

I/V

+ +
by
*i):

Or (by

multiplying the second equation

x
T
Finally,

/3x'
/3r'

i/3vr'

fivx'.

64

substituting* cos0 for

]8

and

sin0 for

these equations

become

\ r

= =

x'cosfl
x'sinfl

r'sinfl )

r'cosfl j

EXACTLY
In
ff

like (20)!

other words, observes a certain event

finds that

and

the four numbers necessary to characterize it (see p. 58)

are

x, y, z

and

',

observing the

SAME

event,

finds that in his

f

system

the four numbers

are x',

y, i ', r

then the form (23) of the Lorentz transformation

shows
to

that

go from one observer's coordinate system

merely necessary

to the other
it is

first coordinate system through an angle 0, in the x , r plane, without changing the origin,

to rotate the

*Since

is

greater than 1 (see p.

11)

must be an imaginary angle:

See

p.

another

25 of "Non-Euclidean Geometry/' book by H. G. and L. R. Lieber.

2

Note

that sin

cos

holds for imaginary angles as well as for real ones/

hence the above substitutions are legitimate,

thus
2

+
2

since

v ), 1/(1 r being taken equal to

/3

(-

i/5v)

= F 2

V=

(1

(see p. 62).

65

thus:

(remembering

that

And
and
then That

since

we

= y and z y took (p. 65)

f

z').

p
ifiv

tan
is,

= cos e = sin = iv.

the magnitude of the angle

depends upon v

the relative velocity of

K and K

f .

And

since, from (23),

2
2

/x
(

= =

(x
7

)
2

cos
sin

(x

+
2

2xYsin0 2x r $infl
/

cosfl

+ (rj costf + (r')

sin

cos

2
fl

then, obviously,

x
or (since

+r =
and z
z
2

(x')

+
),

(rj

y
2

= /

= (xT
66

+ (y? + (z ) +
7

(rj

Now,

it

will

be remembered

from Euclidean plane geometry,

that x*

y represents

the square of the distance

and between and similarly,

in

Euclidean three-dimensional space,

also represents the square of the distance between two points.

)C

Thus, also,

+r+

r represents
1 *

the square of the "interval between two EVENTS, in our four-dimensional world (see p. 58).

And,
just as in

plane geometry

the distance between two points remains the same

whether

we

use

the primed or the unprimed

coordinate systems (see p. 61),

that
is,

*>

= (xj
67

(yj

(although x does

NOT

and y does
So,
in

NOT
x2

equal equal /).

x',

three dimensions,

= (x7

(yj

(zj

and, similarly, as we have seen on p. 66,

the "interval" between two events, in our four-dimensional

space-time world of events, remains the same, no matter which of the two observers,

K or
That

/C',
it.

measures
is

to say,

although K and K' do not agree on some things/

as, for

their length

they
(1) (2)
Etc.

DO

example/ and time measurements, agree on other things:

p,

The statement of their LAWS (see The "interval" between events,

51)

In other words/ although length and time

are
in

no longer

INVARIANTS,

the Einstein theory, other quantities,

like the

space-time interval between two events,

ARE
in this

invariants

theory.
invariants are the quantities

These

68

which have the

SAME

value

for all observers,*

and may therefore be regarded

as the realities of the universe.

Thus, from this point of view, the things that we see or measure

NOT

are the realities, since various observers

do not
of the

get the same measurements same objects,

but rather
certain mathematical relationships

between the measurements 2 2 2 2 r ) z (Like x y

are the realities, since they are the

for all observers.*

same

We
in

shall see,

discussing
Relativity,
fruitful

The General Theory of

how

Minkowski's view-point of a four-dimensional Space-Time

World

proved to be.

VIII.

SOME CONSEQUENCES OF THE

THEORY OF
RELATIVITY.

We
if

have seen that

two observers,

K and K , move
f

relatively to each other

*Ail observers moving relatively to each other

with

UNIFORM

velocity (see p. 56).

69

their

with constant velocity, measurements of length and time

are different;

and, on page 29,

we promised
that their
In this

also to

show
different.

measurements of mass are

chapter

we

shall discuss

mass measurements,
as well as other

measurements which

depend upon
these fundamental ones.

We
in a

already

know

that

if

an object moves

direction parallel to the relative motion of K and

gives the relationship

then the Lorentz transformation

between the length and time measurements of K and K'.

We
in a

also

know

that

direction

the relative
there
is

PERPENDICULAR motion of K and K

f

to

NO difference

in

the
p. 50),

LENGTH
and,

measurements (See footnote on

in this case,

the relationship between the time measurements may be found as follows:

For this

PERPENDICULAR

direction

Michelson argued that the time would be

t2

= 2a/c

(seep. 12).

Now
is

this argument supposed to be from the point of view

of an observer

who
70

DOES take

the motion into account,

and hence already contains the "correction" factor /}/

hence,
replacing to by t'f the expression t' = 2a/3/c represents the time
in

the perpendicular direction

as

K tells

K SHOULD be written.
it
,

Whereas

in his

own

system,

would, of course, write

t

2a/c
t.

for his "true" time,

Therefore
t'

|8f

gives the relationship sought above, from the point of view of K.

From
a

this

we

see that
with velocity u
direction,

body moving

in this

will

PERPENDICULAR appear to K and K to

have

different velocities:

Thus,

Since u

d/t and
</'

u'

c/'/f'

represent the distance traversed by the object

as

where

c/

and

measured by K and and since J = </'

(there being

respectively;

NO difference

in

LENGTH
see p. 70)
f

measurements

in this direction

and
then

i/

= j8t , as shown above, = c///3t = (1//3)u.

=
u/t and a'

Similarly,

since a

u /t

71

where a and
as

a' are the

accelerations of the

body,
/(',

measured by
find that

K and

respectively,

we

In like

manner
find the relationships

we may

between various quantities in the primed and un primed systems of co-ordinates,

provided they depend upon
length

and time.

But, since there are

THREE

basic units in Physics

and since the Lorentz transformation deals with only two of them, length and time,
the question now how to get the
Einstein found
is

MASS

into the

that the best

game. approach

to this difficult problem was via the Conservation Laws of Classical Physics.

Then,

just as the

old concept of

the distance between two points

(three-dimensional)

was "stepped up" to the new one of the interval between two events
(four-dimensional), (see p. 67) so also the Conservation Laws will have to be "stepped up" into

FOUR-DIMENSIONAL SPACE-TIME.
And, when
an amazing
this is

done

vista will

come

into view!

CONSERVATION LAWS OF
CLASSICAL PHYSICS:
(1) Conservation of Mass:
this means that no mass can be created or destroyed,

72

but only transformed from one kind to another.

Thus, when a piece of wood its mass is not destroyed, for
if

is

burned,

one weighs

all

the substances into which

transformed/ together with the ash that remains, this total weight is the
it is

same

as the

weight of the original wood.

express this mathematically thus: where 2 stands for the SUM, so that

We

ASm = 2m

is the mass, and A, as usual, stands for the "change", so that = o says that the change in

TOTAL

ASm
total

mass

is

zero, which

is

the

very convenient, brief, exact form!

(2) Conservation of Momentum: this says that if there is an exchange of momentum
(the product of mass

Mass Conservation Law

in

and velocity, mv) between bodies, say, by collision, the

TOTAL
is

momentum BEFORE
after collision:

collision

the

SAME
TOTAL
A2mv =
o.

as the

(3) Conservation of Energy:

which means that Energy cannot be created or destroyed, but only transformed from one kind to another. Thus, in a motor, electrical energy is converted to mechanical energy, whereas
in a

the reverse change takes place. both cases, we take into account the part of the energy which is

dynamo
if,

And

in

transformed into heat energy, then the energy

by

friction,

TOTAL
and

BEFORE
is

the

AFTER the transformation SAME, thus: A2E = o.

a

Now,

moving body has

73

KINETIC energy,

expressible thus:

two moving, ELASTIC bodies collide, there is no loss in kinetic energy of the whole system, so that then we have

When

2 Conservation of Kinetic Energy: AS^mv (a special case of the more general Law);

whereas, for inelastic collision, where

some

of the kinetic energy

is

changed
2

into

other forms, say heat, then AZ^r/nv

5^ o.

Are you wondering what is the use of all Well, by means of these Laws, the most

this?

PRACTICAL problems can be solved,* hence we must know what happens to them
in Relativity Physics!

You
(a)

will see that

they

will lead to:

NEW Conservation
Momentum and

Laws

for

INVARIANT

Energy, which are under

the Lorentz transformation, and which reduce, for small v, J to the corresponding Classical Laws

(which shows why those Laws worked so well for so long!)

(b) the

IDENTIFICATION

of

MASS and ENERGY!

Hence mass CAN be destroyed as such and actually converted into energy!
Witness the

ATOMIC BOMB

(see p. 318).

See, for example, "Mechanics for Students of Physics and Engineering'* by Crew and Smith, Macmillan Co., pp. 238-241
1 *

Remembering

that the "correction

factor, ft,
is

is

equal to

c/V c

the velocity of!

you
i

see that,

when v
1

grit, c, thus making and hence no "correction* is necessary.

small relatively to 2 v negligible, then

74

Thus the Classical Mass Conservation Law was only an approximation and becomes

merged

into the Conservation of

Energy Lawl

Even without following the mathematics of the next few pages/ you can already appreciate the revolutionary
these results, and become imbued with the greatest respect for the human

IMPORTANCE

of

MIND

which can create

all this

and

PREDICT happenings
Here
is

previously unknown!

MAGIC
may be

for

you!

Some
others

readers

able to understand

the following "stepping

up" process now,

to
it

may

prefer to
II

come back

after reading Part

of this book:

The components of the velocity vector

in Classical

Physics, are:

cfx/c/t, c/y/c/t, c/z/c/t.

And,
these

if

we

replace x, y, z
in

become,

by xi, X2, xs, modern compact notation:

JxjJt

0-1,2,3).
momentum components
are:

Similarly, the
m.c/x,/c/t

(/= 1,2,3)

so that, for n objects, the Classical Momentum Conservation

Law

is:

=o
But (24)
is

(/= 1,2,3)
under

(24)

NOT an

invariant

the Lorentz transformation;

75

the corresponding vector which IS so invariant is:

=o

0=1,2,3,4)

(25)

where s is the interval between two events. * and it can be easily shown that c/s = c/t//3,
c/s

being, as you know,

itself invariant

under the Lorentz transformation.

Thus,

and
(i.e.

in going from 3-dimensional space -dimensional absolute time

from Classical Physics)

to 4-dimensional

SPACE-TIME,
the independent variable

we must

use
t.

s for

instead of

Now let us examine (25) which is so easily obtained from (24) when we learn to speak the

NEW LANGUAGE OF
Consider
first

SPACE-TIME!
3 components of (25):
(i

only the

first

Then
is

A{33m.c/xi/c/s}

= 1,2,3)

(26)

the

NEW

Momentum

since, for large v, it and, for small v, which

Conservation Law, holds whereas (24) does

makes
it

/3

and

c/s

NOT/ = c/f ,

(26)

BECOMES

And
and

(24), as the taking now,

should!

FOURTH

namely, m.cta/cfs or mc.dtjds (see p.

substituting
c/t//J
is

for </s,

we

get mci8 which

mc.c/Vc

_
2
2

component 233)
v or
2

of (25),

mc/Vl

v /c or mc(1

"*
.

/c )

(27)

Expanding, by the binomial theorem,

we

get
Since
2

cp + j.^ + j-^ +
,
,
,

/,

3 v

4
. .

\ -j,
2

c/s

= cVt2 2

(c/x

+ c/y + c/z
2

dividing
(c/s

by
2

c/t

and taking

(see p. 233).
2

/c/0

we
A/1

get

v anc/

c/s/c/t

1 //S

76

which/ (or small v(neglecting terms after v ),

(1
1

vr\ ~
2

I.

(28)
2
.

And,

multiplying

2 by c, we get me

^mv

Hence, approximately,
2

A{Z(roc

|/nv )}

o.

(29)

Now,
which

if

is
2

then A2/nc
is

constant, as for elastic collision, 2 o and therefore also A2(J?mv )

the Classical

Law

of the

Conservation of Kinetic Energy for

elastic collision (see p. 74); thus (29) reduces to this Classical for small v, as it should!

Law
that

Furthermore,
for

we can also see from (29) INELASTIC collision, for which

A{2|mv
hence also

2
}

^o

(see p. 74)

A2mc ^
2
2

o or

c being a constant, c

A2m ^

which says that, for inelastic collision, even when v is small,

any

loss in kinetic

energy

is

compensated

for

by an increase in mass (albeit small) a new and startling consequence for

Physics itself! this from viewpoint Thus, even in Classical Physics

CLASSICAL

NEW

we

realize that

the

Mass

of a

body

is

NOT a
its

constant but

varies with

changes

in

energy

amount of change in mass being too small to be directly observed)!

(the

Taking now (27) instead of (28), not be limited to small v/

we

shall

77

and, multiplying

we

get

A{2mc /3j =
2

by

c as before,

o for the

Conservation Law of Energy, which/ together with (25), is invariant under the Lorentz transformation, and which,
as

NEW

we saw above,

reduces to
v.

the corresponding Classical Law, For small Thus the expression for the

NEW
is:

ENERGY
(30)

of a
for

body

/nc /3, which,

gives fo

= me

showing

that

ENERGY

and

MASS are

one and the same entity

instead of being distinct, as previously thought!

Furthermore,

even a
is

SMALL MASS^m,
LARGE
ENERGY,

amount of equivalent to a 2 since the multiplying factor is c ,

the square of the enormous velocity of light! Thus even an atom is equivalent to

tremendous amount of energy. Indeed, when a method was found (see p. 318) of splitting an atom into two parts and since the sum of these two masses is
a
less

than the mass of the original atom,

that

you can see from (30)

this loss in

mass must yield

of energy

terrific

amount

(even though this process does not transform the entire mass of the original atom into energy).

Hence

the

ATOMIC BOMB!
this terrible
all

(p.

318)

Although
stunned us
let us

gadget has

into the realization

of the dangers in Science,

not forget that

78

the
is

POWER
human

behind

it

the

MIND

itself.

Let us therefore pursue our examination of

the consequences of Relativity, the products of this POWER!

REAL

In

1901 (before

Relativity),

Kaufman*, experimenting with fast moving electrons, found that the apparent mass of a moving electron
is greater than that of one at rest a result which seemed

very strange at the time! Now, however, with the aid of (26) we can see
that his result
is

perfectly intelligible,
for it quantitatively! instead of c/t,

and indeed accounts

Thus the use of
c/s

(where

c/s

dtj(3) brings in
,

the necessary correction factor, j3, for large v not via the mass but is inherent in our

NEW
in

RELATIVITY
r

LANGUAGE,
the idea of
it

which

c/x /c/s replaces

c/Xj/c/t,

velocity,

and makes

unnecessary and undesirable to think mass depending upon velocity.

in

terms of

Many
c/s

writers
c(t/j8 in

by

on Relativity replace (26) and write it:

A{Sinj8.(/Xj/(/t}

o,

putting the

correction

on the m.
course gives
K1-2,

Though
*

this of

Gcscll. Wiss. Gott. Nachr., Math.-Phys., 1901

p.

143, and 1902,

p.

291.

79

the same numerical result, it is a concession to

CLASSICAL LANGUAGE,
and
Einstein himself

does not

like this.

He

rightly prefers that since

we

are

learning a

NEW language (Relativity)

think directly in that language translating each term

we should

and not keep

into our old

CLASSICAL
11

LANGUAGE

before

We

meaning. must learn to "feel" modern and talk modern.

its

we

"feel

Let us next examine

another consequence of the Theory of Relativity:

When

radio waves are transmitted

1

through an "electromagnetic field/ an observer K may measure the electric and magnetic forces
at

any point of the

relationship

field

at a given instant.

The
is

between

these electric and magnetic forces

expressed mathematically
the well-known

by

Maxwell equations

(see page 311).

Now,

if

another observer, K'/

relatively to

moving

with uniform velocity,

makes his own measurements on the same phenomenon,

and, according to the Principle of Relativity,
uses the
in his

same Maxwell equations

primed system,

80

it is quite easy to show* that the electric force

is

NOT an INVARIANT
two observers/
is

for the

and

similarly

the magnetic force

also

NOT AN INVARIANT

although the relationship between the electric and magnetic forces expressed in the

MAXWELL EQUATIONS
has the same form for both observers; just as, on p. 68, though x does equal x' f and y does equal y

NOT NOT

still

the formula for

the square of the distance between two points has the same form
in

both systems of coordinates.

Thus
the

we have seen that SPECIAL Theory of

is

Relativity,
I

which

the subject of Part (see p. 56), has accomplished the following:

revised the fundamental physical concepts.
of

(1)

It

(2)

By the addition
namely,

ONLY ONE NEW POSTULATE,

the principle of relativity

the extension of

See
the

Einstein's

first

paper (pp. 52
in the

&

53)

in

bo>k mentioned

footnote on p. 5.

81

to

ELECTROMAGNETIC
made

phenomena*

(which extension was

by
it

possible the above-mentioned revision

of fundamental units

see p. 55),
results

explained

many
experimental

ISOLATED
As,
for

which baffled the

pre-Einsteinian physicists:

example,

the Michelson-Morley experiment, Kaufman's experiments (p. 79), and many others (p. 6).
(3)
It

ONE LAW

led to the merging into

of the two, formerly isolated, principles^ of the Conservation of Mass and the Conservation of Energy.
In Part
II

we
the

shall

see also how SPECIAL Theory served

point for

as a

starting

the

GENERAL THEORY,

*The reader may ask:

call this a postulate? not based on Facts?" The answer of course is that
(s it

"Why

a scientific postulate must

be

BASED

on

facts,

but it must 30 and hold also

facts that are

further than the

for

known

facts

still

TO

BE discovered.

So
(a

an ASSUMPTION really only most reasonable one, to be sure

that
it

is

since
if it

it

agrees with facts

now known),
in

which becomes strengthened

continues to agree with as they are discovered.

the course of time

facts

NEW

82

which, again,

by means

of only

ONE
led to
results

other assumption,

FURTHER

NEW IMPORTANT

RESULTS,

which make the theory the widest in scope

of any physical theory.

IX.

POINT OF LOGIC

AND A SUMMARY

It is

interesting here

to call attention to a logical point which is made very clear

by
In

the Special Theory of Relativity. order to do this effectively

let us first list

and number

certain statements, both old

and new,

to which

we

shall

then refer

by

NUMBER:

(1)

impossible for an observer detect his motion through space (p. 33). to
It is

(2)

The velocity of light is independent of the motion of the source

The old PRE-EINSTEINIAN postulate and length measurements
is,

(p. 34).

(3)

that time

are absolute,
that

are the

same

for all observers.

(4) Einstein's replacement of this postulate

by the operational
that

fact (see p.

31)

time and length measurements

83

are

NOT absolute,

but relative to each observer.

(5) Einstein's Principle of Relativity (p. 52).

We

have seen that

retained

(1)and(2)
are contradictory IF (3)
is

but are
(3)
is

NOT

contradictory IF

replaced by (4).

(Ch. V.)

Hence
it

may

NOT

be

true to say that

two statements

MUST

be

contradictory or without specifying the

Thus, in the presence of (3)
(1) and (2)

EITHER

NOT contradictory,

ENVIRONMENT

ARE

contradictory,

whereas, in the presence of (4), the very same statements (1) and (2)
are

NOT contradictory.*
may now
briefly

We

summarize

the Special Theory of Relativity:

(1), (2)

and (4) are the fundamental ideas in

and,
since (1)

it,

and (4)

are

embodied

in (5) 7

then (2) and (5) constitute the BASIS of the theory.

Einstein gives these
as

two

POSTULATES
whether two statements are EQUIVALENT or not

*Similarly

may also depend upon the environment (ee p. 30 of "Non-Euclidean Geometry" bv H. G. and L R. Lieber).

84

from which he then deduces

the Lorentz transformation (p. 49) which gives the relationship
of

between the length and time measurements] two observers moving relatively to each other
that

with uniform velocity,

and which shows

there
is

an intimate connection

between space and time.

This connection was then

EMPHASIZED
who showed
that

by Minkowslo,

may be regarded r plane from one set of rectangular axes to another

the Lorentz transformation
as a rotation in the

in a

four-dimensional space-time continuum

(see Chapter VII.).

fFor the relationships between other measurements, jee Chapter VIII.

86

THE
1.

MORAL

Local, "provincial" measurements are not universal,

although they may be used to obtain universal realities

if

compared with other systems

measurements taken from
certain

of

local

a different viewpoint.

By examining

RELATIONSHIPS BETWEEN

LOCAL MEASUREMENTS,
and finding those relationships which remain unchanged in going from

one local system to another, one may arrive at

the

INVARIANTS

of our universe.

2.

By emphasizing the

fact that

absolute space and time are pure mental fictions,

and
that

that the

only

PRACTICAL

notions of time

man can have

are obtainable only the Einstein Theory

that

by some method
shows
that

of signals,

'Idealism" alone, 11 is, "a priori thinking alone, cannot serve for exploring the universe.

On

the other hand,

since actual measurements

are local

and not

universal,

87

and

that

only certain

THEORETICAL RELATIONSHIPS
are universal, the Einstein Theory shows also that
practical
is

measurement alone

also not sufficient

for

exploring the universe.

In short,

a judicious combination

Of THEORY and PRACTICE, EACH GUIDING the other

a "dialectical materialism" is our most effective weapon.

PART

II

THE GENERAL THEORY

A
I.

GUIDE FOR THE READER.

The

first three chapters of Part the meaning of the term ''General Relativity/*

II

give

what it undertakes to do, and what are its basic ideas. These are easy reading and important.
II.

introduce Chapters XIII, XIV, and the fundamental mathematical ideas

XV

which

will

also easy reading

III.

be needed and important.

to

Chapters

XVI

XXII build up

the actual

streamlined mathematical machinery not difficult, but require

the kind of
care and patience and work go with learning to

that

run any The amazing

NEW

machine.
of this

POWER

new
operated,

and the

TENSOR CALCULUS, EASE with which

it is

are a genuine delight!

IV.

this

Chapters XXIII to XXVIII show how machine is used to derive the

NEW LAW OF GRAVITATION.

This law,

though

at

first

complicated
91

behind
is

its

seeming simplicity,

then

REALLY
V.

SIMPLIFIED.

Chapters XXIX to XXXIV constitute THE PROOF OF THE PUDDING!

easy reading again

and show what the machine has accomplished.

Then there are
a

SUMMARY

and

THE

MORAL

92

INTRODUCTION,
In Part
I,

on the SPECIAL Theory, it was shown that

two observers who

are

moving

relatively to each other

with

UNIFORM

velocity

can formulate
the laws of the universe

"W!TH

EQUAL RIGHT AND EQUAL SUCCESS,

11

even though
their points of

view

are different, and their actual measurements

do not

agree.

to

The things that appear alike them both

are the

the

'TACTS of INVARIANTS.

11

the universe,

The mathematical relationships which both agree on

are the

"LAWS

11

of the universe.

Since man does not

the "true laws of

know
1

God/

should any one human viewpoint be singled out as more correct than any other?

why

And
it

therefore
fitting

seems most

to call

THOSE

relationships

95

"THE

laws,"

which are

VALID

from

viewpoints, taking into consideration all experimental data

DIFFERENT

known up

to the present time.

Now,

it

must be emphasized

Theory, only that change of viewpoint was considered

that in the Special

which was due to

the relative

UNIFORM

velocity

of the different observers.

This was accomplished

Einstein
in his first

by

paper*
in

published

1905.

Subsequently, in 1916*, he published a second paper

in

which

he

GENERALIZED

the idea

to include observers

moving
with a
(that
is,

relatively to each other

CHANGING
with an
is

velocity

ACCELERATION),
it is

and
"the
It

that

why

called
11

GENERAL

Theory of
I

Relativity.

was shown

in Part

that

to

make

possible

even the SPECIAL case considered there, was not an easy task,
1

*See "The Principle of Relativity* by A. Einstein and Others,

published by Methuen

&

Co., London.

96

for

it

required

a fundamental

change

in

Physics

to

remove the
certain

APPARENT CONTRADICTION
between

EXPERIMENTAL FACTS!
Namely,
the change from the that TIME is absolute
(that
is,
it is

OLD

idea

that

the same (or

all

observers)

to the

NEW

idea that

time

is

measured
to an observer,

RELATIVELY
just as the

ordinary

space coordinates, x , y , z,
are measured relatively to
a particular set of axes.

This

SINGLE new

idea

was SUFFICIENT
to accomplish the task

undertaken

in

the Special Theory.

We

shall

now

see that

again by the addition of

ONLY ONE
called
Einstein

more

idea,
1

"THE PRINCIPLE OF EQUIVALENCE/

made
possible
the

GENERAL

Theory.

Perhaps the reader

may

ask

why

ONLY ONE
was added?

the emphasis on the fact that new idea

Are

not ideas good things?

97

And

is it

not desirable

many of them as possible? To which the answer is that

to have as

the adequateness
of a
is

new

scientific

theory

judged

(a)

By
and

its

correctness, of course,

(b)

By

its

SIMPLICITY.

No doubt everyone appreciates the need for correctness,

but perhaps
the lay reader may not realize the great importance of

SIMPLICITY!
he
iJBut,"
will say,

"surely the Einstein Theory is anything but simple!

Has

it

not the reputation

of being unintelligible to all but a very few experts?"

Of course "SIMPLE does

11

not necessarily mean

*

"simple to everyone/* but only in the sense that

*lndeed,
it

everyone

WHO

can even be simple to

will take the trouble to learn

some

mathematics.

Though this mathematics was DEVELOPED by experts, it can be UNDERSTOOD by

any earnest student.
Perhaps even the lay reader
will appreciate this

after reading this little

book.

98

if

all

known

physical (acts

are taken into consideration,

the Einstein Theory accounts for

a large
in

number

of these facts

the

SIMPLEST known way.

now
see

Let us

what

is

meant by
of Equivalence/*
it

"The Principle
and what

accomplishes.

It is impossible to refrain from the temptation

to brag about

it

a bit

in anticipation!

And

to say that

by making the General Theory

Einstein derived

possible,

A NEW LAW OF
which
is

GRAVITATION

even more adequate than

the Newtonian one, since it explains,

QUITE INCIDENTALLY,
experimental
facts

which were

left

unexplained

theory, and which had troubled

the astronomers
for a

by the older

long time.

And, furthermore, the General Theory

PREDICTED
this is of

NEW

FACTS,

which have since been verified

course
the supreme test of any theory.

But
to

let us

get to work

show

all this.

99

XL THE PRINCIPLE OF EQUIVALENCE.

Consider the following situation:

Suppose
lives in a

that a

man, Mr. K,

away

spacious box, from the earth

all

and from
there.

so that there

other bodies, is no force of gravity

And
the

suppose that

box and all its contents are moving (in the direction indicated in the drawing on
with a changing velocity,
increasing

p.

100)

32 ft. per second every second. Now Mr. K, who cannot look outside of the box, does not know all this; but, being an intelligent man,
he proceeds to study the behavior
of things around him.

We We

watch him from the outside,

us.

but he cannot see

notice that

he has

a tray in his hands.

And

of course

we know

that

the tray shares the motion of everything in the box,

101

and therefore remains

relatively at rest to

him

namely, in his hands. But he does not think of

this

it

in

way/
is

to him, everything
at rest.

actually

Suddenly he
continue to

lets

30 the

tray.

Now we know

that the tray will

move upward
also

with

CONSTANT velocity/*
and, since
is

we

know

that the

box

moving upwards with

an

ACCELERATION,
that very soon the up with the tray
floor

we expect
will catch

and

hit

it.

And,
Mr.

of course,

we
it

see

this

actually happen.

K also

sees
it

happen,
still

but explains

differently,

he says that everything was until he let go the tray,

and then the tray FELL and

hit

the floor;

and

attributes this to
11
11

"A
He

force of gravity.
to study this "force.
is

Now K begins

finds that there

an attraction

between every two bodies,

*Any moving
with

object

CONTINUES
in a

to

move

CONSTANT speed
it is

STRAIGHT LINE due

unless

to inertia,

stopped by
example.

some

external force,

like friction, (or

102

and and

its

strength

is

proportional to
1

their "gravitational

masses/

varies inversely as the

square of the distance between them.

He also makes other experiments, studying the behavior of bodies

pulled along a smooth table top,

and

finds that different

bodies offer

different degrees of resistance to

this pull,

and he concludes
is

that the resistance

proportional to the

"inertial

mass" of

body.

And

then he finds that

ANY

FALLS

object which he releases with the acceleration,

SAME

and therefore decides

that

the gravitational mass and the inertial mass of a body

are proportional to each other.
In

that all bodies

other words, he explains the fact fall with the

acceleration,

SAME
by
is

saying that the force of gravity such that

the greater the resistance to motion

which

body

has,

the harder gravity pulls it, and indeed this increased pull
is

supposed to be

JUST BIG

ENOUGH TO OVERCOME
IN

the larger resistance, and thus produce

THE SAME ACCELERATION

Now,
if

ALL

BODIES!

Mr.

K is

very intelligent

103

Newtonian
he says,

physicist,

"How

strange that these

two

distinct

properties of a

body should

always be exactly proportional to each other.

But experimental (acts show accident to be true,
continues to worry him.

this

and experiments cannot be denied."

But
it

if

On the other hand, K is an Einsteinian

relativist,

he reasons entirely differently: "There is nothing absolute about

my way
Mr.

of looking at

phenomena.

outside,

(he means us), may see this entire room moving upward with an acceleration,

and

attribute all these

happenings

to this

motion

rather than to
a force of gravity
as

am doing. His explanation and mine

I

are equally

good,

from our different viewpoints."

This
is

what Einstein called

the Principle of Equivalence.

Relativist

continues:

"let

me

try to

see things from

f

the viewpoint of

my good
though
I

neighbor, K , have never met him.

of course see

He

would

104

the floor of this

hit

room come up and

I

ANY object which

it

might release,
to

and

would therefore seem

ENTIRELY

NATURAL

him

for all objects released

at a

from a given height given time

to reach the floor together,

which of course is actually the case. Thus, instead of finding out by

long and careful
that

EXPERIMENTATION

the gravitational and inertial masses are proportional,

as

my Newtonian

ancestors did/

he would predict
that this

PRIORI
case.

MUST

be the

And

so,
facts are

although the
in either

explainable

way, K"s point of view

is

the simpler one, and throws light on happenings which

could acquire only by arduous experimentation, if were not a relativist and hence
I I

quite accustomed to give equal weight to

my

neighbor's viewpoint!

Of course as we have told the story, we know that K is really right:

f

But remember that

in

the actual world

we do

We

not have this advantage: cannot "know" which of the two

is

explanations

"really" correct.

105

But, since they are

EQUIVALENT,

we may

select the simpler one,

as Einstein did.

Thus we already see an advantage in

Einstein's Principle of Equivalence.

as

And, we
is
it

said in Chapter X.

this

only the beginning,

led to his
of Gravitation which

(or

new Law

RETAINED ALL THE MERITS OF

NEWTON'S LAW,
and
has additional
merits which Newton's Law did not have.

NEW

As we

shall see,

106

XII.

NON-EUCLIDEAN WORLD.

Granting, then, the Principle of Equivalence, according to which Mr. K may replace
11

the idea of a "force of gravity by a "fictitious force due to motion/

1 *

the next question

is:

"How

does

this

A
In

new Law

help us to derive 11 of Gravitation?

answer to which
ask the reader to recall

we
a

few simple things which

in

he learned

elementary physics

in

high school:
11

*The idea of a "fictitious force due to motion

is

familiar to

everyone

in

the following example:

Any
if

he swings

youngster knows that a pail full of water

in a vertical

plane
fall
is

WITH SUFFICIENT SPEED,

the water will not

out of the pail,

even when the

actually upside

pail

down!

And

he knows that

the centrifugal "force" is due to the motion only,

since,
if

he slows down the motion,

the water

WILL
a

fall

out

and give him

good dousing.

107

If

at
it

on a moving object an angle to this motion, will change the course of the object,
a force acts

and we say
the

that

body

has acquired an

ACCELERATION,
even though its speed may have remained unchanged! This can best be seen with the aid of
the following diagram:

If

AB
if

represents the original velocity

in

(both

magnitude and direction)

is

and

the next second

the object

moving with
,

a velocity

represented by AC due to the fact that

some

force (like the wind)

its

pulled it out of then obviously

course,

108

BC

must be the velocity which

11

had to be "added" to
to give the "resultant

AB AC,
1
'

as any aviator, or even any high school boy, knows from

the "Parallelogram of forces.

Thus

BC

is

the difference between

the two velocities,

>ACand AB.
is

And,

since

ACCELERATION

defined as

the change in velocity, each second, then BC is the acceleration,

even
equal
that

if

AB

and

AC happen

to

be

in length,

is,
if

even

the speed of the object

has merely

has remained unchanged;*

the very fact that

it

changed in DIRECTION shows that there is an ACCELERATION!

Thus,
if

an object moves

in a circle/

with uniform speed, it is moving with

an acceleration since
it is

always changing

its

direction.

Now
lives
is

on

imagine a physicist a disc which

who

revolving with constant speed!

Being a physicist, he is naturally curious about the world, and wishes to study it,

even as you and I. And, even though

*This distinction
is

we

tell

him
11

that

between "speed
1

and "velocity"

discussed on page

28.

109

he

is

moving with an acceleration

he can formulate

he, being a democrat and a relativist,

insists that

the laws of the universe

"WITH

EQUAL RIGHT AND EQUAL SUCCESS

:

and therefore claims

he
but
is
is

that

not moving at all merely in an environment

11

in

which

a "force of gravity

is

acting

(Have you ever been on

and actually
Let us
felt this

a revolving disc 11 "force ?!).

now watch him

become interested know whether
in circles:

tackle a problem:

We
He

see him

wants to

the circumferences of two circles

are in the

same

ratio as their radii.

He

draws two

circles,

a large

one and

a small

one

(concentric with the axis of revolution of the disc) and proceeds to measure
their radii

and circumferences.
larger circumference,

When

he measures the

we know,
from a study of
the Special Theory of Relativity* that he will get a different value

from the one

WE

should get

(not being on the revolving disc); but this is not the case with
his

measurements of the

radii,

since the shrinkage in length, described in the Special Theory,

"See Part

of this

book.

110

takes place only

IN THE DIRECTION OF and not in a direction which

MOTION,
is

PERPENDICULAR
(as a radius
is).

to the direction of motion

Furthermore/ when he measures the circumference of the small circle,

his

value

is

not very different from ours

is

since the speed of rotation

small

around a small

circle,
is

and the shrinkage

negligible.

therefore

And
he
are

so, finally,

it

turns out that

finds that the circumferences

NOT
is

in

the same ratio as the radii!

that

Do we
and

tell

him

he

is

wrong?

that this
that

not according to Euclid?

is

he

a fool for trying

to study Physics

on

a revolving disc?

Not

at all!

On

the contrary,
relativists,
all right,

being modern "That is quite

we

say

neighbor,

you are probably no worse than we are, you don't have to use Euclidean geometry
it

if

does not work on

a revolving disc,

for

now

there are

non-Euclidean geometries which are exactly what you need

Just as

we would not expect Plane Trigonometry to work on

for

a large portion of the earth's surface

which we need
which

Spherical Trigonometry,
in

the angle-sum of a triangle

isNOTlSO

111

as

we might

naively

demand

after

a high school course in

Euclidean plane geometry,

In short/

instead of considering the disc-world

as an accelerated system,

we
by

can, the Principle of Equivalence,

it

regard

as a system in
11

which
acting,

a "force of gravity

is

and, from the above considerations, we see that

in a

space having such a

gravitational field

Non-Euclidean geometry,
rather than Euclidean,
is

applicable.

We shall
how

now

illustrate

the geometry of

a surface or a space

may be

studied.

This will lead to

the mathematical consideration of

Einstein's

Law

of Gravitation

and

its

consequences.

XIII.

THE STUDY OF SPACES.

first

Let us consider

the familiar Euclidean plane.

Everyone knows

that

for a right triangle

on such

a plane

the Pythagorean theorem holds:

Namely,

113

that
s

Conversely,
it is

true that

IF the distance

between two points

on
is

a surface

given

by
2

(1)

y>

THEN
the surface

A
it is

MUST BE EUCLIDEAN PLANE,

obvious that
to

Furthermore,
the distance from

A
114

ALONG

THE CURVE:

is

no longer

the hypotenuse of a right triangle,

and of course

we
If,

CANNOT write
however,
take two points,

here

we

and 8,

sufficiently near together,

the curve

AB

is

so nearly

a straight line,
that

we may

actually regard

ABC as
in

a little right triangle

which the Pythagorean theorem

does hold.

115

Only that here we shall represent

by
as
c/s
is
,

its

three sides

c/x

and
in

c/y ,

done

the differential calculus, to show that the sides are small.

So
(2)

that here

we have
2

c/s

</x

c/y

Which
and
is

still

has the form of (1)

characteristic of

the Euclidean plane.

be found convenient x and y and xi xi by , respectively, so that (2) may be written

It

will

to replace

(3)

c/s

c/x L

c/xA

Now
on
a

what is the corresponding situation non-Euclidean surface,

such as,
the surface of a sphere, for

example?

Let us take

two points on

this surface,

and 8

designating the position of each

by

its

latitude

and longitude:

116

**"^^ \

TV

Let

Df be

the meridian

from which
longitude
is

measured

the Greenwich meridian.

And
and

let

DK be

a part of the equator,

the north pole.

latitude of

Then the longitude and

are, respectively,

the

number

of degrees in
,

the arcs

^F and AG

(or in the

corresponding central angles,

Similarly,

a and

0).

117

the longitude and latitude of

are, respectively,

the

number

of degrees in

the arcs

CF and BK.
is

The problem again

to find the distance

between
If

and

B.
is

the triangle /ABC

sufficiently small,

we may
en
a

consider

it

to lie

Euclidean plane which practically coincides with

the surface of the sphere
this little
in

region, and the sides of the triangle to be straight lines

(as

ABC

on page

1 1

5).

Then,
since the angle at
is

a right angle,

we have
(4)

AB*
let us

=
see

A? +

BC*-

And now
what
if

this

expression becomes
in

we change
(4)

the Cartesian coordinates

(in the

tangent Euclidean plane) to the coordinates known as

longitude and latitude

on the surface of the sphere.

Obviously

AB

has a perfectly definite length

irrespective of

118

which coordinate system we use; but>4C and BC,

the Cartesian coordinates in the tangent Euclidean plane may be transformed into

longitude and latitude on the surface of the sphere, thus:

let r

be

the radius of the latitude circle

FAC,

and R the radius of the sphere. Then

Similarly

BC

=/?$.
in (4),

Therefore, substituting

we have
2

(5)

c/s

tJc?
xi

+
,

this

And, replacing a by may be written

2

and

j8

by

(6)

c/s

i>c/x?

tfdxt

comparison of (6) and (3)

will

show

that

*any
that

high
if

an

arc,

school student knows x represents the length of is the number of and

it,

radians in

then

And

therefore

120

on the sphere, the expression (or

is

c/s

as

not quite so simple it was on the Euclidean plane.

The question naturally arises, does this distinction between a Euclidean and a non-Euclidean
always hold,

surface

and

is

this a

way
between them?

to distinguish

That
if

is,

we know

the algebraic expression which represents the distance between two points

which actually holds on a given surface,

we then immediately decide whether the surface

can
is

Euclidean or not?
it

Or does

perhaps depend upon

the coordinate system used?

To answer
let us

this,

30 back to the Euclidean plane,

and use oblique coordinates instead of the more familiar

rectangular ones
thus:

121

The coordinates of the point

now xand y
are

represented

by

which are measured

parallel to the

X and Y axes,
each other.

and are now

NOT at right angles to

Can we now
find

the distance between

and

using these oblique coordinates?

Of
for,

course

we

can,

by the well-known Law of Cosines in Trigonometry,

we

can represent

the length of a side of a triangle

122

lying opposite an obtuse angle,

by:
2

2xy cos

a.

Or,

for a

very small triangle,

2

</$

c/x

c/y

-2c/x</ycosa.

And, if we again replace x and y

by
this

xi

and X2

respectively,

becomes
2

(7)

c/s

c/x

Jxl

2 c/xrc/X2-cosa.

Here we see
even on
a

that

Euclidean plane,
2

the expression for is not as simple as

c/s
it

was before.

And,

if

we had used

polar coordinates on a Euclidean plane,

we would have obtained

or

(8)
*

Js

=
24)

c/x

+ x?

(See page

123

The reader should remembering that

verify this,

the polar coordinates of point

are
its distance, xi , from a fixed point, , makes with a fixed line (2) the angle, x2 , which

(1)

OP

OX.

Then (8)

is

obvious from

the following figure:

o
124

Hence we see
depends upon
(a)

that

the form of the expression for c/r

BOTH

the

KIND

we
and

pF SURFACE

are dealing with,

(b) the particular

COORDINATE SYSTEM.

We
a

shall

soon see

that

whereas

mere superficial inspection 2 of the expression for </s

is

not sufficient

to determine the kind of surface

we are dealing with, DEEPER examination

of this expression help us to

DOES
For

know

this.

this

deeper examination

we must know
how,
from the expression for
the sVcalled
2

c/s ,

"CURVATURE TENSOR

11

of the surface.

And this brings us to the study of tensors:

What
are tensors?

what use are they? and are they used?

Of

HOW

Let us see.

125

XIV.

WHAT
*

IS

TENSOR?

The reader

is

no doubt
1

familiar
11

with the words "scalar

and "vector.
which

A
a

scalar

is

a quantity

has magnitude only,

whereas
vector has

both magnitude and direction.

Thus,
if

we
70

say that

the temperature at a certain place

is

Fahrenheit,
is

there

obviously

NO
is

DIRECTION

to this temperature,

and hence

TEMPERATURE
But

SCALAR.

if we say that an airplane has gone

one hundred miles

east,

obviously its displacement from its original position

is

VECTOR,

whose

MAGNITUDE

is

and whose

DIRECTION

is

100 miles, EAST.

Similarly,
a

person's

AGE

is

SCALAR,

whereas

127

the
is

VELOCITY
VECTOR,*

with which an object

moves

and so on;
the reader can easily
find further

of both scalars

examples and vectors.

discuss

We

shall

now

some

quantities

which come up in our experience and which are neither scalars nor vectors,
but which are called

TENSORS.
And,
when we have
illustrated

and defined these,

we
a

shall find that

is

SCALAR
VECTOR
shall

TENSOR

whose

RANK

is

ZERO,

and
a
is

TENSOR
is

and we
a

see what

whose RANK meant by

is

ONE,

Thus

TENSOR of RANK TWO, or THREE, etc. "TENSOR" is a more inclusive term,

""speed" and "velocity
is

*A

distinction

is

often

made between
the latter a

the former Thus when

SCALAR,
is
its

VECTOR.
in

we

are interested

HOW FAST a thing

ONLY

moving,

and do not care about

DIRECTION of motion, we must then speak of its SPEED, but if we are interested ALSO in DIRECTION, we must speak of its VELOCITY. Thus the SPEED of an automobile
would be designated by
"Thirty miles an hour,"

its

but

its

VELOCITY would be
EAST."

"Thirty miles an hour

128

of which

"SCALAR
discuss

11

and

"VECTOR

11

are

SPECIAL CASES.
Before

we

the physical meaning of

a tensor of rank two,
let us

consider

the following facts about vectors.

Suppose

that

we have

any vector, AB , in a plane, and suppose that we draw a pair of rectangular axes, X and y,
thus:

X'

Drop
from

a perpendicular

BC

to the X-axis.
that
,

Then we may say

AC
for,

is

the
is

and CB
as

X-component oF AB the /-component of

from

we know

the elementary law of "The parallelogram of forces/* if a force acts on a particle

AC

and CB also

acts

on

it,

the resultant effect

as that of a force

is

the

same

AB

alone.

And
the

that

is

why
are called

>AC and

CB

"components"
course
if

of

AB.

Of

we had used

the dotted lines as axes instead, the components of AB

would now be
In

AD and

DB.

other words, the vector AB may be broken up

into
in

components

various ways,

depending upon our choice of axes.

Similarly,
if

we

use

THREE
two

axes

in

SPACE

rather than

we
as
in

plane, can break up a vector

in a

into

THREE components

shown
the diagram 1 31 .

on page

130

D
By dropping the perpendicular BD from 8 to the XY-plane,
and then drawing
the perpendiculars
to the

DC and DE

X and

axes, respectively,

we have
namely,

the three components of

AB

and, as before,
the components

depend upon

the particular choice of axes.

Let us

now

illustrate

the physical meaning

of a tensor of rank two.

Suppose we have
at

a rod

every point of which

is

there

a certain strain

due

to

some

force acting on

it,

As

a rule the strain

131

is not the same at all points, and, even at any given point, the strain is not the same in

all

directions.*
if

Now,
(that
is

the the

STRESS

is,

FORCE

at the point A causing the strain at

A)

represented
in

both

magnitude and direction

by/B

*When

an breaks object finally under a sufficiently great strain, it does not fly into bits
it

as

would do

if

the strain were the same

at all points

and

in all directions,

but breaks along certain lines where, for one reason or another, the strain is greatest.

132

and

if

we

are interested to
this force

know

the effect of the surface

upon CDEF (through A} f

we

are obviously dealing with a situation which depends

not on a
but on

SINGLE

vector,

TWO vectors:
AB ,
in

Namely, one vector,

which represents the force and another vector

(call
it

question;

whose
the

/AG), direction will indicate

ORIENTATION

of the surface
will represent

CDEF,

and whose magnitude the AREA of CDEF.

In other words, the effect of a force upon a surface on the force depends

but
size

NOT ONLY ALSO on the

can

itself

and orientation of the

surface.

Now, how
by
If

we

indicate

the orientation of a surface

a line? a line through

we draw

in the

plane

obviously
in

CDEF , we can draw

no way

this line

many

different directions,
is

and there

of choosing one of these to represent the orientation of this surface.

BUT,
if

we

PERPENDICULAR
such a line
is

take a line through A to the plane

CDEF,

UNIQUE
133

and

CAN
if

therefore

be used

to specify the orientation

of the surface

CDEF.

a vector, direction perpendicular to and of such a length that

Hence,

we draw

in a

CDEF

it represents the magnitude of the area of CDEF,

then obviously this vector

AG

indicates clearly

both the SIZE and the

of the surface

ORIENTATION

CDEF.

Thus.
the

STRESS

at

upon the surface CDEF depends upon the AB and AG , and is called
a

TWO vectors,

TENSOR
now
in

of

RANK TWO.
convenient way

Let us

find a

of representing this tensor.

And,
let us

order to

do

so,

consider the

stress,

F,

upon

a small surface, c/S ,

in

represented

the following figure

Now
is

if

OG,

perpendicular to

ABC
,

the vector which represents the size and orientation of ABC

then,

134

z
c

it is

quite easy to see (page

36)

that the

X-component

of

OG
the XZ-plane.

represents in magnitude and direction

the projection

08C

of

ABC upon
of

And
the

similarly,

7 and

Z components
,

OG

represent the projections

0/C and 0/B

respectively.

135

To show
both
in

that

OK represents OBC

magnitude and direction:

T
That
is
it

does so

in direction

obvious,

since

OK

is

perpendicular
that

to

OBC

(see p.

34).

As
we

regards the magnitude

must

now show

OK _ OBC OG ABC'
(a)

between

Now OBC = ABC ABC and OBC

x cos

of the dihedral angle

(since the area of the projection

of a given surface
is

equal to

ihe area of the given surface multiplied

by

136

the cosine of the dihedral angle

between the two

But
this

planes).

dihedral angle equals anale and since are respectively perpendicular to ABC and OBC ,

GO/C

06

OK

and cos

ZGOKisOK/OG.

Substitution of this in (a) gives the required

OBC _ OK OGABC
Now,
which
acts
if

the force

F,

is itself

a vector,

on

ABC ,
its

we

can examine

total effect

separately the effects of its three components

fx i fy i

by considering

and

fz

upon

EACH

of the three projections

OBC,OAC*ndOAB.
Let us designate these projections

by dSx

dSy end dSz

respectively.

Now,
since fx

(which
acts

is

the

upon

EACH

X-component of F) one of the three

above-mentioned projections,
let us

designate the pressure

due

to this

component alone

upon

the three projections

by
Pxx
respectively.
/

Pxy

Pxz

We

must emphasize

the significance of this notation: In the first place,

137

the reader must distinguish between

the "pressure" on a surface and the "force" acting on the M The p ressurel1 k the

surface.

FORCE PER UNIT AREA.

that
is

So
the

TOTAL FORCE MULTIPLYING

PRESSURE by
'

obtained by
of the surface.

the

the

AREA

Thus the product

PXX dSx
gives the force acting upon the projection dS, due to the action of f,

ALONE.
'

Note

the

DOUBLE
PXX
i

subscripts in
i

Pxy

PXZ
refers to the fact

The
that

first

one obviously
all
f.,

these three pressures from the component

emanate
alone;

whereas, the second subscript designates the particular projection upon which
the pressure acts.

Thus

p/.v

means

the pressure due to fr upon the projection dSy

Etc.
It

follows therefore that

fx

pxx'dSx

+ Pxy'dSy + pxz'd

And,
and

similarly,

~
>y
fx

Pyx'dSx -f Pyy'dSy
PZX dSx
'

+ Pyz'J
+p
zz

~\~

fzy dSy

'

dS

138

Hence

the

TOTAL

STRESS, F,

on the surface c/S,

F=(x+(y +
or
i

?,

PXX cfix ~T Pxy

'
'

'

cr jj/

~r PXZ

'

dSz
C/S2
.

+ Pyx + PZX
Thus
stress

uSx H~ Pyy uSy p yz dSx PZU dSy ~\- p Z2

'

'

'

'

C/Sz

we
is

see that

just a vector, with three components in

not

three-dimensional space (see p. but has NINE components

in

30)

THREE-dimensional space.
is

Such a quantity

called

TENSOR OF RANK TWO.

tensor suffice:

For the present

let this illustration of a

Later
It is
if

we

shall give a precise definition.

obvious that
dealing with a plane

we were

instead of with

three-dimensional space,
a tensor of rank

two would then have

components instead of nine, since each of the two vectors involved has only two components in a plane, and therefore, there would now be only
only

FOUR

2X2 components for the tensor

X
3 as above.

instead of 3

And, in general, if we are dealing

with

n-dimensional space,

139

a tensor of rank

two

has n components

which are therefore conveniently written

in a

SQUARE

array
1

as

was done on page

39.

Whereas,
in

n-dimensional space/

VECTOR VECTOR

has only n components:

Thus,
a
in a

PLANE
it

has
in

TWO components/
has

THREE-dimensional space
components/

THREE

and so on.

Hence,
the components of a
are therefore written
in a

VECTOR

SINGLE ROW;

instead of in a

as in the case of a

SQUARE ARRAY TENSOR of RANK TWO.

Similarly,

n-dimensional space 3 of rank THREE has n components, and so on.

in

TENSOR

To sum
In

up:

n-dimensional space,

a
a a

VECTOR has n components, TENSOR of rank TWO has n components, TENSOR of rank THREE has n components,
2
3

and so on.

The importance of
in Relativity

tensors

will

as

become clear we 90 on.

140

XV. THE EFFECT

CHANGE

IN

ON TENSORS OF THE COORDINATE

SYSTEM.

In Part

of this

book (page 61)

we had

occasion to mention

the fact that

the coordinates of the point

141

in the unprimed coordinate system can be expressed in terms of its coordinates in the

primed coordinate system

by

the relationships

w
/

Ov

= = (y
x

x co$0 y x sin0 + / cos0

of

as

is

known

to

any young student

elementary analytical geometry.

Let us

now

see
this

what effect

change

in

the coordinate system has

upon

a vector

and

its

components.

Call the vector cfs,

and
its

let o'x and c/y represent components in the UNPRIMED SYSTEM, and dx and c/y'
its

as

components in the PRIMED shown on page 143.

SYSTEM

142

Obviously
is

c/s itself

not affected by the change

of coordinate system,

but the
in

COMPONENTS

of

</s

the two systems

are
as

DIFFERENT, we have already pointed out

1

on page

30.

Now
are

if

the coordinates of point

in

x and y

one system

143

and

in the other, x' and the relationship between these four quantities
is

given

And
we
c/x

by equations (9) on p. 142. now, from these equations, by

differentiation*,

can,

find the relationships

between

and c/y and 7 c/x' and c/y

It

will

be noticed,

in

equations (9),

that

x depends upon BOTH x and y', so that any changes in x' and y'
will

BOTH
the
c/x ,

affect x.

Hence
namely
will

TOTAL

change

in

x,

depend upon
Partially

TWO causes:
in x',

(a)

upon the change

c/x' ,

namely
and
(b) Partially

upon the change

7

in y',

namely

c/y

Before writing out these changes, it will be found more convenient

to solve (9) for x and y in terms of x and y.|
7 7

*See any book on

Differential Calculus.

fAssuming of course

that the
in (9)

determinant of the coefficients is not zero.

(See the chapter on "Determinants" "Higher Algebra" by M. Bocher.)

in

144

In

other words,

to express the

NEW,
in

primed coordinates, x' and /,

original ones,

terms of the

OLD,

x and y ,

rather than the other

way around.

This will of course give us

x' ==
N (10) '
<
,

ax
ex

=
J

+ 6y + dy
,

where a,
It

fc,

c,

are functions of 0.

will

be even

better

to write (1 0) in the form:

(11) x '

x2

621X1

using xi and XL instead of x and y , 2 instead of (and of course x[ and x'

x'

and /);

and putting different subscripts on the single letter a ,

instead of using
four different letters:

d
is

The advantage
not only that
easily

of this notation

we

can
to

GENERALIZE

n dimensions

from the above

two-dimensional statements,
but,
as

we

shall

see

later,
itself

this

notation lends

to

a beautifully

CONDENSED
work with.

way

of

writing equations,

which renders them

very

EASY

to

145

Let us

now proceed

with

the differentiation of (11):

we

get

k'
(1 * 2) '

I \

C/X2

= =

ail( 'Xl

"^ ai2C^X2

521C/X1

a 2 2C/X?

The

MEANING

of the a's in (12)

should be clearly understood: Thus an is

the change in
x'\

due

to
in

UNIT

CHANGE

xi,

so that

when it is multiplied by the total change in xi , namely

c/xi ,

we

get

THE THE And

612

CHANGE CHANGE
similarly in

IN IN

*(

DUE TO

* ALONE.
CHANGE
in

a^cfa,

represents

the change in xi and therefore

PER UNIT

x2 ,

the product of a\? the total change

gives

in

and x / namely dx

THE THE
Thus
the
is

CHANGE CHANGE

IN x{ IN x2

DUE TO ALONE.
in xi

TOTAL CHANGE
by

given

just as

the total cost of

a

number of apples and oianges would be found

cost of

by multiplying the

ONE APPLE
by
the total

number

of apples,

146

and

ADDING
one

this result

to a similar

for the oranges.

And

similarly for

Jx 2

in (1 2).

We
a

may

therefore

replace an

by dx(/dx\

symbol which represents the partial change in xj per unit change in xi*, and is called
the "partial derivative of xi with respect to xi ."

Similarly,

dx{
'

3l2

cl21

9x2
T~~ OXi
/

^22

9X2

9x5 " ^ 0X2

And we may
in

therefore rewrite (1 2)

the form

dx{
f\

j dx\ *^i H
i

dx{
f\

OXi

0X2
<?X2

(13)
</X2 2

9X2 , T-'CfXi

+
,

T~
0x2

But perhaps the reader

is

getting a
is
it

little

tired of all this,

and
what

wondering
has to

do

with Relativity.

*Note
is

that a
i A A

PARTIAL
MJM
d

change
letter

always denoted by the

"d"

in contrast to

which designates a

TOTAL

change

147

To which we may give him

a partial answer

now

and hold out hope

of further information
in

the remaining chapters.

What we can

already say
is

is

that

since General Relativity

concerned with

finding the laws of the physical world

which hold good for ALL observers/ and since various observers
differ

from each other,

in that

as physicists,

only

they

use different coordinate systems, we see then

that Relativity
is

concerned

with finding out those things

which remain

INVARIANT

under transformations of
coordinate systems.

Now,

as

we saw on page 143,

is

a vector

such an

INVARIANT/

and, similarly,
tensors in general

are such

INVARIANTS,

so that the business of the physicist

really

becomes

to find out

which physical quantities

are tensors,

and are therefore

1

the "facts of the universe/ since they hold good

for all observers.

*See

p.

96.

149

Besides,
as

we promised on page 125,

1

must explain the meaning of "curvature tensor/ since it is this tensor

we

which

CHARACTERIZES
then

a space.

And

with the aid of the curvature tensor of

our four-dimensional world of events,*

in this world what paths the planets take, and in what path

we shall find out how things move

a ray of light travels

as
it

passes near the sun,

and so on.

And

of course
all

these are

things which

can be

VERIFIED BY EXPERIMENT.

XVI.
Before

VERY HELPFUL SIMPLIFICATION

further

we go any

let us write

equations (13) on page

147

more
thus:

briefly

(,4)

-**
THREE
TIME
is characterized by space-coordinates and of its occurrence
I.

*FOUR-dimensional, since
each event
its

the

(see Part

of this book, page 58)

150

A careful study of (14) will show

(a) That

(14) really contains

it

TWO

equations

(although
since, as
its

looks like only one),

give

we

M
1

possible values,

and 2

we have
c/x[

and

c/Xj

on the
in

left,

just as

we

did

(13);

The symbol 2, means that when the various values of <J ,

(b)

namely

and 2

are substituted for

cr

(keeping the
must be
Thus, for
(1

constant in any the resulting two terms

\i

one equation)

ADDED
/i

together.
<r

and

4) becomes

it dxi

dx(
dxi

-dxi

dxi

j
-dx2,

0x2

just like

the FIRST equation in (13),

and, similarly,

by

taking

ju

and again "summing on the <r's ," since that is what 2, tells us to do,

we

get

dx2
which
Thus

If

3X2 I T-'cfxi OXi

+OX
I

5X 2 J
2

-ax2

is

the

SECOND
all

equation

in (13).

we

see that
of (13).

(14) includes

A
is

still

further abbreviation

introduced

by

omitting

the symbol 2,

151

WITH THE UNDERSTANDING THAT WHENEVER A SUBSCRIPT OCCURS TWICE IN A SINGLE TERM
(as, for
in
it

example,

<r

the right-hand
will

member
that

of (14)

),

be understood

SUMMATION

ON THAT
(15) U ;
in

is to be made SUBSCRIPT.

Hence we may
d <Jx' *"

write (14) as follows:

dx
'*-<lx *'

dx.

which

we

shall

know

that the presence of the

in

TWO </s

the term on the right,

that

means

2a

is

understood.

And
since

now, finally, c/xi and cfe

are the

components of
system

c/s in

the

UNPRIMED
let us

represent them more briefly

by

A
respectively.

and

The reader must

these
with

NOT confuse

EXPONENTS 2

SUPERSCRIPTS

is not the "square of" A but the superscript serves merely the same purpose as a

thus >4

SUBSCRIPT,
namely,
to distinguish the

components

from each other.

Just

why we use SUPERSCRIPTS instead

appear

of subscripts

will

later (p. 1 72).

152

c
v

And
will

the components of

c/s

in the

PRIMED
1

coordinate system

now be

written

A' and A'\

Thus
(1

5)

becomes

And
if

so,
a certain vector

we have
is,

A"

that

a vector

whose components
2

are

and
if

in a certain

coordinate system,
to

and
a
in

we change

new coordinate system

accordance with

the transformation represented then

by (11) on page 145

(16)
the
in

tells us

what

will

components of

this

be same vector
coordinate system.

the

new (PRIMED)

Indeed, (1 5) or (16) represents the change in the components of a vector

NOT ONLY for the

but for
of coordinates:*

change given

in

(11),

ANY transformation

Thus
a point in

suppose x, are the coordinates of one coordinate system,

*Except only that
the values of (x a ) and (x/) must

be

in

one-to-one correspondence.

154

and suppose

that
xj

ft

(xi

x2/ ....)

ft

x! 2
etc.

=f>(xa )
this entire

Or, representing
set of

equations
Xp

by
Ifj.

\Xa)f

where the f s represent

any (unctions whatever,

then, obviously
'

^
~

HI

J OXi

OXi
or, since
ft

JJ J'CTX2 ~r T" ^ 0X2

afl

x{

If

=
{

r-~ OXi

C/Xi

'

C/X2

0X2

etc.

Hence

gives the

manner of transformation

of the vector

dxa

to

ANY other coordinate system

(see the only limitation

mentioned

in

the footnote on

page

54).

And

in fact

ANY set of quantities which

transforms according to (16)
is

DEFINED
or rather,

TO

BE

VECTOR,

A CONTRAVARIANT
the meaning of

VECTOR -

"CONTRAVARIANT"
155

will

appear

later (p. 1 72).

The reader must not

forget that
in

whereas the separate components the two coordinate systems are

different,

the vector

itself is

an

INVARIANT

under the

transformation of coordinates

(see page 143).

It

should be noted further that

(16) serves not only to represent a two-dimensional vector,

but

may

represent

a three- or four- or

n-dimensional vector,
since
all

that

is

necessary

is

to indicate the

number
2 and

of values that

M and
Thus,

(r

may

take.
1
, <r

if

=
JLI

we have
but
if

two-dimensional vector/ 1 , 2 , 3 , and a = 1 2 ,

(16) represents a 3-dimensional vector,

and so on.
For the case M

3 and

(7

= 1,2

(16) obviously represents

THREE EQUATIONS
the right-hand

in

which

members
terms:

each have

THREE

(JX[

ux-2

0x3

3*2
axi

-A

/!

+
i

dXo
r
}

fa' I* -A-+ ,

0x3

5x 3

156

Similarly

we may now

give

the mathematical definition of

a tensor of rank two,*

or of

any other

rank.

Thus
a contravariant tensor of rank
is

two

defined as follows:

(17)

Here, since 7 and

in
it

occur

TWICE

the term on the right,

is

understood that
must

we

SUM

for these indices

over whatever range of values they have. Thus if we are speaking of

THREE DIMENSIONAL SPACE,

we have 7 = 1 , 2 , 3 and ALSO a = 1 , 2 , 3 , and
But
5

= 1 = 1

2
,

3.
,

3;

NO SUMMATION
on the a and
j8

is

to

be performed

since neither of

them occurs
term/

TWICE
so that

in a single

any

particular values of

cv

and p

must be retained throughout

ANY ONE

equation.

For example,
for the case

1 , /3

It will be remembered (see page 128)

that

VECTOR

is

TENSOR

of

RANK ONE.

157

(1 7) gives the

equation:

,,

/\

_ ~

<9x[
'

V~~

dxs ~ All
'

r\

dx[ " dxj

'

'

"I

12

dx! 3*2
'

'

Jl

9xi 9xi

9xi 0x2

9xi 0x3

..
6x2 9xi

__

9x2 9x2

9x2 9x3

+ ^.^./31 +
9X3 9xi
It

dx dx
l.

*.

AV

??!.

dx
?.

9X3 9X2
that

9X3 9X3

will

be observed

7 and

<5

have each taken on

their

THREE

possible values:
in

1,2,3,

which resulted

NINE
a =
1

terms on the right/

whereas

and

ft

have been retained throughout.

And now
there will

since

a and

/3

may each have

be

the three values,

1,2,3,

NINE
Thus

such

EQUATIONS

in all.

(1

7) represents

nine equations each containing nine terms on the right,

if

we

are considering

three-dimensional space.

Obviously
(17)

for

two-dimensional space,

will represent

only four equations each containing

only four terms on the

right.

Whereas,
in

four dimensions,

as

we must have

in

Relativity*

*See the footnote on

p. 1 50.

158

(17)

will represent

sixteen equations each containing sixteen terms on the right.

And
in

in

general,

n-dimensional space,

a tensor of

RANK TWO,

defined by (17),
consists of

n equations, each containing n terms

in

the right-hand member.

Similarly,
a contravariant tensor of
is

RANK THREE

defined by

(18)

A'-to-l^-^-A vXp ox? dx

ff

and so on.

As
the

before,

number

of equations represented

by (18)

and the number of terms on the

right in each,

depend upon
the dimensionality of the space in question.

The reader can already appreciate somewhat

the remarkable brevity
of this notation,

But

when he

will

see

in

the next chapter

how
are

easily such sets of equations

MANIPULATED,
will

he

be

really delighted,

we

are sure of that.

159

XVII.

OPERATIONS WITH TENSORS.

For example,
take the vector (or tensor of rank one) 2 1 and having the two components

Af

in a

plane,

with reference to a given set of axes.

And
of

let

6 be another
a ,

such vector.

Then, by adding the corresponding components

Aa

and B

we

obtain a quantity

also having

two components,

namely,

A +
1

B and

A +B
2

which may be represented by

and C
respectively.

Let us

now prove

that this quantity

is

also a vector:
is

Since A*
its

a vector,
is:

law of transformation

(19)

/V

=
:

M"(see

Similarly, for B"

X

(20)

B'

ox,

^--B\

Taking corresponding components,

160

we

get, in full:

f\

AH

lAli *. /I ^
i

OXi

0X2

and
B/l

^ Xl
TT

.....

Dl D ~r
1

axi

^ Xl D2 -D . 0x2

The sum of these

gives:

Similarly,

dxi

6x2
in:

Both these

results are

included

Or
(21)

c^^-e. dx
a

Thus
a

we

see that
is

the result

VECTOR

(see p. 155).

Similarly for tensors of

higher ranks.

Furthermore, note that (21)

may be obtained

QUITE

MECHANICALLY

by adding (19) and (20)

AS

IF

each of these were

A SINGLE equation
containing only SINGLE term on the right,

161

instead of

A SET OF EQUATIONS EACH CONTAINING

SEVERAL TERMS
Thus the notation

ON

THE RIGHT.

AUTOMATICALLY
shall

takes care that

the corresponding components

be properly added.

This
in

is even more impressive the case of multiplication.

Thus, to multiply

(22)

A'

M'
^
(A, /.,,/*

by
(23)

""^^
0-g.g.C*

= 1,2)

we

write the result immediately:

(24)

(X //t/

,/3-1,2X

To convince the reader

it is quite safe to write the result so simply,

that

let us examine (24) carefully and see whether it really represents

correctly the result of multiplying (22)

by (23). By "multiplying (22) by (23)" we mean that EACH equation of (22) is to be

multiplied

by
163

EACH

equation of (23)

in
in

the

way

in

which

this

would be done

ordinary algebra.

Thus,

we

must

first

multiply

by
f \r
j

fj

We
(25)

8 et,

9xi 9xi

9x2 9xi

9xi 9xi
,
-

ox* 0x2
Similarly

we

shall set

three

more such equations/ whose left-hand members are,

respectively,

A'T,

A'-B'\ A"- B'

and whose right-hand members

resemble that of (25).

Now, we may
by
taking X

obtain (25) from (24)

1

=
,

/*

retaining these values throughout,

since no summation
[that
in
is,

is

indicated on X and
/x

neither X nor

is

repeated

any one term of (24)].

164

But since

a and

ft

each
in

OCCUR TWICE

the term on the right,

they must be allowed to take on

all

possible values, namely,

and 2

and

SUMMED,
we
replace A'B**

thus obtaining (25),

except that

by by

the simpler taking A

symbol C
1 , /*

a^

*.

Similarly,

2
/?

in

(24),

and summing on a and

as before,

we

obtain another of the equations

mentioned on page 164.

And

, /*

gives the third of these equations/ and finally A = 2 , M - 2 gives the fourth and
last.

Thus (24) actually does represent

COMPLETELY
the product of (22) and (23)!

Of

course,

in

three-dimensional space,

(22) and (23) would each represent

THREE THREE

equations, instead of two,

terms on the right, instead of two;

each containing

and the product of (22) and (23)

*Note
that either

A" B
11

or C<*

allows for

FOUR

Namely, AW or C AW or C", AW or C and AW or C

21 ,

components:
,

22

And hence we may use G* instead of A* V.

165

would then

consist of

NINE
NINE
But
is still

equations, instead of four,

terms on the right, instead of four.

each containing
this result

represented
of course,
in

by

(24)!

And,

four dimensions

(24) would represent

SIXTEEN

equations, and so on.

I :

Thus the tensor notation enables us

to multiply

WHOLE
containing
as

SETS
as

OF EQUATIONS
TERMS IN EACH,
elementary algebra!
multiply
in

MANY
we

EASILY

simple monomials

we

Furthermore, see from (24)

that

the
is

also a

PRODUCT of two tensors TENSOR (see page 157),

and, specifically, that

the product of two tensors each of ONE,

RANK

gives a tensor of

RANK TWO.
m
and n

In general,
if

two tensers of ranks

respectively,
are multiplied together,

the result
a

is

TENSOR OF RANK m

n.

This process of multiplying tensors

is

called
multiplication,

OUTER

166

to distinguish

it

from

another process

known

as

multiplication which is also important

in

INNER

Tensor Calculus,
shall

and which we

describe later (page 183).

XVIII.

PHYSICAL ILLUSTRATION.

But

first

let us discuss

a physical illustration of

ANOTHER KIND OF TENSOR, A COVARIANT TENSOR:*

Consider an object whose density
is

different in different parts of the object.

^This

is

to

be distinguished from the

1

CONTRAVARIANT tensors
discussed on pages
55ff.

167

We

may

then speak of

the density at a particular point,

Now,
but a

density

is

obviously

NOT a
And

directed quantity,
(see page 127).

SCALAR

since the density of the given object

is not uniform throughout, but varies from point to point, it will vary as we go from A to

So

that

if

we

designate

by ^

the density at

A,
.

then

^ and ^
--

-"--

dxi

dx2

represent, respectively,

the partial variation of

in

\l/

the xi and x> directions.

Thus, although
a

^
in

itself is

NOT
depend upon
quantity,

DIRECTED

quantity,

the the

CHANGE

DOES

DIRECTION
IS therefore a

and

DIRECTED
are

whose components

9xi

Now

let us

see
this

what happens to

quantity
is

when

the coordinate system

changed (see page 149).

We

are seeking to express

d\[/ r

d\l/
,
,

d$
9xi

in

terms o!

d\l/ ,

9x1

3x2
three variables,

9x2

Now

if

we have

say^^andz,
169

such that y and z depend upon x it is obvious that

the change
IF IT
in

z per unit change in

x,

CANNOT
in

BE

FOUND

DIRECTLY,

may be found by
multiplying the change

y per

unit

change

in

x
y,

by
the change in z per unit change
or,
in

expressing this in symbols:

..

c/z

cfz

Jy

(26)
In

cfx~d[ydx-

we have

our problem above, the following similar situation:

affect

A change in x( will
BOTH
and the
xi
will affect 1/7

and x2 (see

p.

145),

resulting changes in XL

and

X2

hence

Note

that here
right

we

have

TWO

terms

on the

instead of only

ONE,
in x{

as in (26),

since the change

affects

BOTH
in

xi

and x2

and these
whereas
a

in turn

BOTH

affect ty,

(26),

change in x affects y/ which in turn affects z , and

that
is all

there was to

it.

Note
since

also that

the curved
all

"d"

is

used throughout

in

(27)

the changes here

170

are

PARTIAL

changes

(see footnote on page 147). And since ^ is influenced also

by

change

in

x2/

this influence

may be
by

similarly represented

And, as before, we may combine (27) and (28) by means of the abbreviated notation:

where the occurrence of a TWICE in the single term on the right indicates a summation on a ,
as usual.

And,

finally,

writing

A^
4

for the

represented

J'in

W
C/ JC

two components

and

>A ff

for the two components,

we may
(30)

write (29) as follows:

Al-jfrA. CXM
(30) with (16)

0*^ =

If we now compare we note a

VERY IMPORTANT DIFFERENCE,

namely,
that the coefficient
is

on the

right in

(30)

the reciprocal of the coefficient on the right

in

(16),

171

so that (30) does

NOT satisfy
in

the definition of a vector given But it will be remembered that

(16).

(16)

is

the definition of

A CONTRAVARIANT VECTOR
(30) introduce to the reader
in

ONLY.

And
we

the mathematical definition of

A COVARIANT
Note
that

VECTOR.

to distinguish the two kinds of vectors, it is customary to write the indices

as SUBscripts in the

one case
in

and

as

SUPERscripts

the other.*

1 56), represent a vector in any number of dimensions, depending upon the range of values

As

before (page

(30)

may

given to

JJL

and

0"

and

for

ANY transformation

of coordinates.

A COVARIANT
is

Similarly,

TENSOR OF RANK

TWO

defined

by

and so on,

for higher ranks.

and and (31) (17).

COMPARE

CONTRAST

carefully

^Observe that the SUBscripts are used for the COvariant vectors,
in

which the

PRIMES

in

the coefficients
(see (30), p. 171).

are in the

DENOMINATORS

To remember this more easily a young student suggests the slogan

"CO, LOW, PRIMES BELOW/

172

XIX.

MIXED TENSORS.

is

Addition of covariant vectors performed in the same simple manner

1

as for contravariant vectors (see p.

61

Thus, the

SUM
A\

of

r-~7*.n a

ox\

and

Sx=
dx(
is

"

c=
Also,

3x

the operation defined on page

as
is

166

OUTER MULTIPLICATION
the same for

covariant tensors:

Thus, the

OUTER PRODUCT
A(

of

^-A a

and

is

dx a

Furthermore,
it is

also possible to multiply

COVARIANT tensor by CONTRAVARIANT one,

173

thus,

the

OUTER PRODUCT of
A(

=
|J.

A.

and

is

(32)

C/5;:=

^'-ax"'

C'-

Comparison of (32) with (31) and (17) shows that it is

NEITHER

a covariant

NOR a
It is

contra variant tensor.

called

A
the

MIXED TENSOR
generally,

of rank

TWO.

More

OUTER PRODUCT

of

and

is

rfr

***

That
if

is,

any two

tensors of ranks

and n,

respectively, are multiplied together so as to form their

the result

OUTER PRODUCL is a TENSOR

of rank

m+

175

thus, the rank of (33) and that of (34) is 2 ,

is

hence,
the rank of their outer product, (35),
is

5.

Furthermore,

suppose the tensor of rank

is

MIXED
m

tensor,

having mi indices of covariance*

and

(such that
has

mi m^ = m), and suppose the tensor of rank n

indices of contravariancef

indices of covariance*
contravariance/f"

and n 2 indices of
then,
their outer

product

will

be

MIXED

mi and

+m

tensor having indices of covariance*

m2

rt2

indices of contravariance.f

All
in

this has

already been illustrated

the special case given above:

Thus,

(33) has

ONE

index of covariance (7)

(5),

and (34) also has index of covariance

ONE

therefore their outer product, (35), has indices of covariance (7,

TWO

<5);

and

similarly,

since (33) has indices of contravariance (a/ and (34) has

TWO
*

/?)

SUBscripts.
t SUPERscripts.

176

ONE
has

index of contravariance

(K),

their outer product, (35),

THREE

indices of contravariance

(,

/3,

K) O

We

hope the reader appreciates

it

the fact that

although
it

takes

many words

to describe these processes

is

extremely

EASY

to

DO them

with the AID of the TENSOR NOTATION.

Thus the outer product of

A*
is

and B75

simply

C$

Let us remind him, however, that behind this notation,

the processes are really complicated: Thus (33) represents a whole SET of equations* terms on the right each having

MANY*
MANYj

(34) also represents a SET of equations!

And

each having

terms on the right.

And
is

their outer product, (35),

obtained by multiplying
*Namely, EIGHT
for

two-dimensional space;

TWENTY-SEVEN
SIXTY-FOUR

for three-dimensional,

for four-dimensional,

and so on. tFour for two-dimensional space,

NINE

SIXTEEN

for three-dimensional space, for four-dimensional space}

and so on.

177

EACH EACH

equation of (33)

by

one of (34), resulting in a SET of equations, (35),

containing

THIRTY-TWO

equations for two-dimensional space,

TWO HUNDRED AND

three-dimensional space,

FORTY-THREE

for

ONE THOUSAND AND TWENTY-FOUR

all

for

four-dimensional space, and so on.

And

with a

correspondingly large number of terms on the right of each equation!

And

yet
11

"any child can operate

as easily as

it

pushing a button.

XX.

CONTRACTION
and

AND

DIFFERENTIATION.

This powerful

easily operated machine,

the

TENSOR CALCULUS,
and Levi-Civita

was devised and perfected by

the mathematicians
Ricci
in

about 1900, and was known to very few people

until Einstein

Since then

it

has

made use of become

it.

widely known, and we hope that

will

this little

book

make

it

intelligible

even to laymen.

178

But what use did Einstein

make

of it?

What

is its

connection with Relativity?

are nearly ready to Fulfill the promise made on page 1 25.

We

When we have explained two more operations with tensors, namely, CONTRACTION and DIFFERENTIATION, we shall be able to derive
the promised

CURVATURE TENSOR,
of Gravitation

from which
Einstein's
is

Law

obtained.

Consider the mixed tensor (33), p. 175: suppose we replace in it

7 by a,
obtaining

f\<x

dx a dx x dxM

^~f

-\

**"

By

the summation convention (p.

52),

the left-hand
so that

member is to be summed on (36) now represents

only equations instead of eight,* each of which contains

TWO

TWO terms
since

on the

left

instead of one;

furthermore, on the RIGHT,

a occurs twice we must sum on a

for

here,

each pair of values of v and X

Now,
*Secp. 177.

180

when

happens to have a value

from X,

DIFFERENT
then

dx, dx^

__

dx

__

dxa dxx

dxx
other

BECAUSE
the x's are

NOT functions of each

p

(but only of the x"s) and therefore

there
is

NO variation of x
x,

with respect to
a

DIFFERENT
be

namely xx

Thus coefficients of
will all

A?

when A

ZERO
these terms drop out.

and

will

make
v.

BUT When A =
then

dx,

dx

a dxx dx'

dxl dx x
Thus (36) becomes
(37)

dx' a 3xx

A'

g-^

which we must sum on the right for X and M

in
-

still

To make

all this

clearer,

let us write

out explicitly

the two equations represented

l

by
l

(37):
2

'\

+ A'?

|^ UXi
d
UXi

(A\

Al

!?

+ A'f =

f (A? +
181

^l

^ + A?) + ^ (A? + Al
+
(Al
OX'2

).

Thus (37)

may be

written

more

briefly:

08)
where

c-g-e
C'^/T + ^T, C = A'\ + X?
2

and

C2
In

A[

422

other words,

by making one upper and one lower index

ALIKE
in

(33),

we have REDUCED
a tensor of rank a tensor of rank

THREE ONE.

to

The important thins to note

or
as

is

that this process of reduction

CONTRACTION,
it

is called/ leads again to

TENSOR,
it is

and
foi

obvious that

every such contraction the rank is reduced by TWO, since for every such contraction

two of the

partial derivatives in

the coefficient

cancel out (see page 181).

We
how

shall

see

later
this

important

process ot contraction

is.

if

Now, we form
the

the

OUTER PRODUCT
(p. 1

of

two

tensors,

in

way already described 182

75)

and if the result a mixed tensor,

then,

is

by
as

contracting this

mixed tensor
is

shown above,
get a tensor which
called

we
an
in

INNER PRODUCT
contrast to

their

OUTER PRODUCT.
OUTER
product of

Thus the

A*
is

and B7
(see page
1

Ck
if

77);

now,

in this result

we

replace

7 by

|8 ,

obtaining
CS, or

Da

(see pages

80

to

82),

then

D A*

is

an

INNER

product of

and B\

And now we come to DIFFERENTIATION.

We
if

must remind the reader that

uv

where // u, and v are variables,

then

dy
c/x

dv
c/x

c
j,

f?f*

c/x

Applying this principle to the differentiation of

(39)

A* = ^'A', OXff
differential calculus

*See any book on

183

with respect to

x,' ,

we

get:

Or, since

dA

dA* dxr

hence (40) becomes

( 41 \

=
dxT

From (41 ) we see that if the second term on the right were not present, then (41) would represent
a

mixed tensor
in certain

of rank two.

And,
this

special cases,

second term does vanish,

cases,

so that
in

SUCH

differentiation of a tensor

leads to another tensor

whose rank

is

one more than

the rank of the given tensor. Such a special case is the one
in

which the coefficients

dx.
in

(39)

are constants,
as in (13)

on page 147,

since the coefficients in (1 3) are the same as those in (1 1) or (10);

184

and are therefore (unctions of 6 , 6 being the angle through which the axes were rotated (page 141), and therefore a constant.
In

when the

other words, transformation of coordinates

is of the simple type described on page 141

then
ordinary differentiation of a tensor leads to a tensor.

BUT, IN
and IN
so,

GENERAL,

these coefficients are

NOT constants,

GENERAL
give a tensor evident from (41).

differentiation of a tensor

does
as
is

NOT

BUT
there
is

a process called

COVARIANT DIFFERENTIATION
which and which we

ALWAYS

leads to a tensor,

shall presently describe.

We
the
of

cannot emphasize too often

IMPORTANCE

any process which leads to a tensor, since tensors represent

the

"FACTS

11

of our universe

(see page 149).

And, besides, we shall have to employ

COVARIANT DIFFERENTIATION
in deriving

185

t/
X

the long-promised

CURVATURE TENSOR
and

EINSTEIN'S

LAW OF GRAVITATION

XXI.

THE

LITTLE

g's.

To explain covariant

differentiation

we must
in

first

refer the reader

back

to chapter XIII;

which

it

was shown that

the distance between two points, or, rather, the square of this distance,
2

namely, c/s , takes on various forms

depending upon
(a) the surface in question

and
(b) the coordinate system used.

But now,
with the aid of the remarkable notation

which

we have

since explained,
2

we
in

can include
these expressions for
c/s

ALL
the

SINGLE

expression

(42)
and, indeed, this holds

Jf = g^Jx^Jx9ff

NOT ONLY

for

ANY
or

SURFACE,

but also for

any THREE-dimensional space,

FOUR-dimensional,

187

or, in general,

any n-dimensional space!*

Thus, to

show how (42)

represents

equation (3) on page 116, we take M = 1 , 2 and ^ =

obtaining
2

1,2,

(43)

c/s

guc/xi-c/Xi
'

g2lC/X2

C/Xi

+ +

gi>c/xrc/x2
'

g22C/X>

C/X2 /

since the presence of

ju

and
in

TWICE
in

the term on the right

(42)
/*

requires

SUMMATION
may be
2

on both

and

p.f

Of

course (43)
c/s

written:

(44)

giic/x?

g, 2 c/xic/X2

and, comparing (44) with (3),

we

find that

the coefficients in (3) have the particular values:

=
ffll

=
I ff!2
,
jjfel

=
7

22

^Except only at a so-called "singular point"

of a space/
that
is,

a point at

which

matter

is

actually located

In other words, f42) holdr. for any region Pee page 1 52.

AROUND
exponent

matter.

| Note that in c/xi (as well as in c/x:)

the upper

"2"
is

is

really an

and

NOT a
NOT

SUPERSCRIPT
an

since (44)

ordinary algebraic equation and is in the

ABBREVIATED TENSOR NOTATION.

188

Similarly, in (6)

on page 120,

gu
and,
in

r\

ffi2

g2 i

, 322

=
=

/?*

(7) on

page 123,

gn

=*

1, 912

- cos a, g2i

= -cos a,

g2 2

1,

and so on.

Note
the

that gi2

and
in

g?i

have

SAME value.
indeed,
general
9?"

And

9w

in

(42) on page 187.

Of course, if, in (42), we take /*- 1,2, 3 and ^=1,2, 3, we shall get the value for c/s2
in a

THREE-dimensional space:
2

(45)

c/s

guc/xi

gi2C/xrc/x2

gisc/xrc/xs

4- g>2]dx2-dxi

Thus,

in particular,

for ordinary

Euclidean three-space, using the common rectangular coordinates, we now have:

gn

g22

gas

1 /

and
for

all

the other g's are zero,

so that (42) becomes,

THIS

PARTICULAR CASE,
2

the familiar expression

c/s

= =

c/x?

+ +

c/xl

c/xi

c/s

c/x

c/y

c/z

and

similarly for

higher dimensions.

189

Thus/ for a given space/

two-, three-, four-, or n-dimensional, for a given set of coordinates, we get a certain set of g's.

and

It is

easy to show* that

set of g's,
is

any such
(which
the

represented

by gv)
of a

constitutes

COMPONENTS
in fact, that

TENSOR,

and;

COVARIANT TENSOR OF RANK TWO,

is appropriately designated with SUBscriptsf:

and hence

TWO

Let us

now

briefly
far:

sum up

the story so

By introducing the Principle of Equivalence Einstein replaced the idea of M a force of gravity" by the concept of
a geometrical space (Chap. XII).

And
is

since a space

characterized

by

its

g's,

the
is

knowledge
things

of the g's of a space

essential to a study of

how

move

in

the space,

and hence
Einstein's

essential

to an understanding of

Law

of Gravitation.

*Seep.313. tSeep.172.

190

XXII.

OUR LAST DETOUR

As we

said before (page 185),

to derive the
Einstein Law of Gravitation, we must employ COVARIANT DIFFERENTIATION.

Now,

the

COVARIANT DERIVATIVE
known
as

of a terror

contains certain quantities

CHRISTOFFEL SYMBOLS*
which are functions of the tensor g^
discussed
in

chapter XXI,
set

and also of another

which

g^

(note the SUPERscripts here)

we

shall

now

describe:

let us limit

For simplicity, ourselves for the

moment

to

TWO-dimensional space,
is,

that

let us

take

/x

2 and v

then gMV will have

FOUR

components,

namely,
the four coefficients on the right
in

(44).
let us

And
in a

arrange these coefficients

thus:

SQUARE ARRAY,
ff21

ff22

which

is

called a

MATRIX.
921

Now

since gt2
for the

(see page

89)

*Named

mathematician, Christoffcl.

191

this

is

called a

SYMMETRIC MATRIX,
since
it

is

symmetric with respect to

the principal diagonal

(that
in
is,

the

one which

starts

the upper left-hand corner).

IF

we now

replace the double bars

on each side of the matrix

as

by SINGLE bars, shown in the following:

we
a

get what

is

known

as

DETERMINANT.*
carefully

The reader must

DISTINGUISH between
*The reader probably knows that a square array of numbers
with single bars on each side

6
3

2
is

and

called a determinant, that its value is found thus:

5X3-6X2 = 15-12Or, more generally,

at =*J
I

A determinant does not necessarily

have to have
but

TWO

may have n rows and

rows and columns, n columns,

.

and is then said to be of order n The way to find the VALUE of

a determinant of the nth order
is

described

in

any book on

college algebra.

193

a square array with

SINGLE

bars

from one with

The and has

DOUBLE bars: FORMER a DETERMINANT

is

SINGLE
way
in

VALUE

found by combining the "elements"

in a certain

as

mentioned Whereas
is

the foot-note on p. 193.

the

DOUBLE-barred

array

a set of

SEPARATE

"elements,"
in

be COMBINED They may be just

NOT to

any way.

the coefficients of the separate terms on the right in (44),

which,
as

we mentioned on page

90,
of a tensor.

are the separate

COMPONENTS

The determinant on page 193 may be designated more briefly by

g.

Oi

2)

or,

still

better, simply

by

g.

And now
in

let us

form a

new

square array
of

the following manner: DIVIDE the

of the determinant

COFACTOR*

EACH ELEMENT

on page 193 by the value of the whole determinant,

namely, by g,
thus obtaining the corresponding element of

the

NEW array.

*For readers unfamiliar with determinants this term is explained on p. 195.

194

The
is

COFACTOR

of a given element

of a determinant

found by

striking

out

the row and column containing the given element, and evaluating the

determinant which

is left over, or prefixing the sign according to a certain rule:

Thus,

in

the determinant

523
4 6
1

the cofactor of the element 5,

1

is:

=
7

1X7-8X0 = 7-0 = 7.
4
is:

Similarly, the cofactor of

2
8

3
7

= -(14 -24) = 10;

and so on.

Note

that in the

first

case

we
we

prefixed the sign , while in the second case prefixed a

rule
is:
.

The
the

prefix a

+
the

or

according as
of steps required to

NUMBER
first

go

from the
(that
is,

element
in

one

the upper left-hand corner) to the given element,

is

EVEN

or

ODD,

respectively/

thus to

go from "5" to "4" it takes one step, hence the cofactor of "4" must have

MINUS
2

prefixed before

'

8
But

all this is

explained

in

more thoroughly any book on

college algebra.

195

Let us
at the

now 90 back
bottom of
p.

to the array described

194.

This

which

new array, we shall designate by

g"

can also be shown to be

a

TENSOR,
this

and,

A CONTRAVARIANT
OF RANK TWO.
it is

time,

TENSOR

That

also

SYMMETRIC

can easily be shown by the reader.

We

can

now

give the definition

of the Christoffel

symbol
{^v, \]

which
It

we need.
by
for:

is

designated
is

and

symbol

In

other words,
:

the above-mentioned Christoffel symbol involves partial derivatives of

the coefficients in (44),

combined
the

as

shown

in

(46)

and multiplied by

components
in

of the tensor

g^

Thus,

two-dimensional space,

*There arc other Christoffel symbols, but we promised the reader

to introduce only the barest minimum of mathematics

necessary for our purpose!

196

since

a each have

the values

1 ,

we

have, for example,

(11
*

1) {
I

a 11
<J

\dxi
12

4-

9n

1-1 dxi

9n
^T

dxi

and

similarly (or the remaining

SEVEN

values of

obtained by allowing M
to take

and X
for each.

on

their

two values

Note

that in evaluating

we

SUMMED
a
to

allowing

above, {11 ; 1 on the a, take on BOTH values, 1,2,

}

BECAUSE
if

EACH TERM
and

(46) were multiplied out, would contain

this calls for

a TWICE,
1

SUMMATION
Now
that

on the a (see page

the meaning of

52).

we know

the 3-index Christoffel symbol

X},

we

are ready to define

the covariant derivative of a tensor, from which it is only a step to

the
If

new Law
is

of Gravitation.

Ar

a covariant tensor of rank

one,

its

COVARIANT DERIVATIVE
xr

with respect to
is

DEFINED

as:

(47)

dA.

^-{r,aM..
197

It

can be shown to be a
it is

TENSOR

in fact/

COVARIANT TENSOR OF RANK TWO*

and may therefore be designated by

A,T
Similarly,
if

we have
by

a contravariant tensor of rank one,

represented
its

Aa ,

COVARIANT DERIVATIVE
the

with respect to x,
is

TENSOR:

<

48 >

K.

+ln

r]

*.

Or,
starting with tensors of rank

TWO,

we have

the following three cases:

(a) starting with the

CONTRAVARIANT tensor, 4", we get the COVARIANT DERIVATIVE:

(b) from the MIXED tensor, we get the

A:,

COVARIANT

DERIVATIVE:

',*See p.

60

of
1

"The Mathematical Theory of Relativity/ by

A.
the

S.

Eddington,
Edition.

1930

198

(c) from the

we

set the

COVARIANT tensor, A. , COVARIANT DERIVATIVE:

T

Arrp

AA

-jT^ UXp

{*P/

*}

Ar

\Tp / *}

A,

And

similarly (or the

COVARIANT DERIVATIVES
of tensors of higher ranks.

Note that IN ALL CASES COVARIANT DIFFERENTIATION OF A TENSOR

leads to a

TENSOR

having

ONE MORE UNIT OF COVARIANT CHARACTER

than the given tensor.

Of

course since

the covariant derivative of a tensor

is itself

a tensor,
find

we may
which
is

ITS covariant derivative

then the
of

SECOND COVARIANT DERIVATIVE

the original tensor, and so on for

higher covariant derivatives,

Note also that when the g's happen

as, for
in

to

be

constants,

example,

the case of a Euclidean plane,

using rectangular coordinates,

in

which case

we have

(see p.

88)

so that

=
flfll

l2

, ff21

22

199

all

constants,

then obviously the Christoffel symbols here

are
all

ZERO,
is

since the derivative of a constant

zero,

and every term of the

Christoffel

symbol

has such a derivative as a factor/

so that (47)

becomes simply

~~

That

is,

in this case,

the covariant derivative

becomes

simply the ordinary derivative. But of course

this
is

NOT so

IN

GENERAL

XXIII.

THE CURVATURE TENSOR AT LAST.

Having now built up the necessary machinery, the reader will have no trouble
in following the derivation of the new Law of Gravitation.

Starting with the tensor,

A,

form

its

covariant derivative
:

with respect to x r

(49)

A =

- K!A.(seep.197).

*See page 196.

200

of

Now form the covariant AT (see page 1 99)

xp
:

derivative

with respect to

(50)

=
rp

!<rp, e)

A, -

[rp,

ej

A,.

obtaining
a

SECOND
is

covariant derivative of

Aa

which

COVARIANT TENSOR OF RANK THREE.

Substituting (49) in (50),

we
A

get

A,,,

l/

^ff

T.-*-

ffr'/

^^a ^7" UJ\p

A - A

^
i
v/JCp

<

{"/

"I

9A

dA,

or

202

If

we had
the

taken these derivatives

order,

in

REVERSE

namely, FIRST with respect to xp and THEN with respect to xr

we would

of course have obtained

the following result instead:

|f M<rT
+
which
a
is

3
)

'

ax~

{<7T,e}
f

{tp,a} A.
Vf\a

{pT'tlfc
again

,f T + !P /

(T6 /

A,

COVARIANT TENSOR OF RANK THREE

Now,
comparing (51) with (52)

we

shall find that

they are

NOT alike THROUGHOUT: Only SOME of the terms are the SAME in both,
but the remaining terms are different.

Let us see:

the FIRST term

2

in

each

is:

j ,- , V- ^-~ and T dx dx/ dxTdxp

ff

3 Ar

d~A
p

respectively.

These,

by ordinary

calculus,

203

The
is

ARE the same.* SECOND term

the same as

of (51)

the

FOURTH
in

term of (52)

since the occurrence of

(or e)

TWICE

the same term

implies a

SUMMATION
!

and it is therefore immaterial what letter is used (a or e) f

Similarly for

the

FOURTH

term of (51)

and the SECOND of (52). The SIXTH term (and the SEVENTH) is the same in both since the reversal of r and p in

*For, suppose that z

as, for

is

a (unction of

x and /

example, z

2xy,

Then

2x

+
2

2y

(treating

y as constant)

and

v v" dx-dy
THEN

(treating

as constant).

if we reverse the order of differentiation, finding FIRST the derivative with respect to y with respect to x , and

And,

we would

get

Y
\0

= 2x

(treating

as constant)

and
the

7,

dydx

(treating

y as constant)

SAME FINAL
this is true

result.

And
t

IN

GENERAL.
is

An

index which

thus easily replaceable

is

called a

"dummy"!

204

does not

alter the value of

this Christoffel

symbol:

This can easily be seen by referring to the definition of this symbol;* and remembering that the tensor gM ,
is

SYMMETRIC,
is,

that

ft,

g,M (see

page
is

89).

Similarly the last term

in

the same

both (51) and (52).

But the
are

THIRD and FIFTH

to
subtraction

terms of (51)
in

NOT equal

any of the terms

(52).
\

Hence by

we
}

get
f

{^

A*

Aa

oxp

or

(53)

And

since addition (or subtraction)

of tensors
gives a result which is itself a tensor (see page 161) the left-hand member of (53) is

A COVARIANT
is

TENSOR OF RANK THREE,

hence of course the right-hand member

also such a tensor.
But,

now,

since

An

is

an arbitrary covariant vector,

*Seepa 9 e196.

205

its

coefficient;
in

namely; the quantity must also be a tensor

square brackets,

according to the theorem on p. 31 2.

Furthermore,
this

bracketed expression

must be a
since
it

MIXED

tensor of

RANK FOUR,

on inner multiplication by A.
is

must give a result which

of rank

THREE;
it

and indeed

must be of the form

(see page 31 3).

This

AT LAST
is

the long-promised
( P a 8 e187),

CURVATURE TENSOR
and
is

known

as

THE RIEMANN-CHRISTOFFEL TENSOR.

Let us examine
so that
its
it

carefully

we may

appreciate

meaning and value.

XXIV.

OF WHAT USE

IS

THE CURVATURE

TENSOR?

In

the

first

place
that

we must remember
it

is

an abbreviated notation for

in

the expression
in

square brackets

(53) on page 205;

in
if

which,

we

substitute for the Christoffel symbols,

206

and so on, accordance with the definition on page 196,

(crp, e}

their values in

we
First

find that

we have
partial derivatives

an expression containing

and second
,

of the g's

which are themselves the coefficients

in

the expression for

c/s

(see p.

87)

How many
Obviously

components does
that

the Riemann-Christoffel tensor have?

depends upon the

dimensionality of the space

under consideration.
Thus,
a
if we are studying two-dimensional surface,

then each of the indices,

will

so that B*TP

have two possible values, would then have

sixteen components.
Similarly,
in
it

three-dimensional space

4 would have 3 or 81 components, and so on.

in

For the purposes of Relativity, which we have to deal with

FOUR-dimensional continuum

this tensor has

44

or

256 components!

We
it is

hasten to

add

that

not quite so bad as that, as we can easily see:

In
if,

the

first

place,

in this tensor,*

*That
in

is, in the expression (53) on page 205.

in

square brackets

207

we interchange r and p , the result is merely to change

Hence,

its

sign.J

of the possible 16 combinations of r and p only 6 are independent:

This
that
is

in itself

we

shall linger

so interesting here for a moment:

a,,;j ,

(wherea=

Suppose we have 16 quantities, 1 , 2, 3 7 4, and

- 1,2,3,4),

which we may arrange

as follows:

a-2t

322 332
^42

a23
a;j;}

324
331
44

ast
641

^43

And
(that

suppose
is,

that a a)3

a fta

a reversal of the
in a

two

subscripts

change of sign of the term), = an implies that an = then, since a n , and similarly for the remaining terms
results

only

in

the principal diagonal,

hence,
the above array becomes:

324

323

324

Thus there are only

SIX

distinct quantities

instead of sixteen.

Such an array

is

called

ANTISYMMETRIC.
tThc reader would do well to compare this expression with the one obtained from it by an interchange of r and
p throughout.

208

Compare
a

this

with the definition of

array

SYMMETRIC And so,

on page 193.J on page 208,

come back we now have

to
six

to the discussion

combinations of r and p

to

be used with

sixteen combinations of
giving

16

or

<r and a, 96 components

instead of 256.

Furthermore, it can be shown

that

we

can further reduce

this

number

to 20.*

Thus our curvature tensor,

for the situation in Relativity,

has only

20 components and
consider for a

NOT

256!

Now

let us

moment
of this tensor

the great in the study of spaces.

fThus
in

IMPORTANCE

an

ANTISYMMETRIC
a/3
,

matrix

we have

ap =
whereas,
in a

SYMMETRIC

matrix

we have

Note
were
it

that

if

the

first

matrix

on

p.

208

SYMMETRIC,
distinct elements,
in

would reduce to
the principal diagonal
in that case.

TEN

since the elements

would

NOT

be zero

*See A. S. Fddington's The Mathematical Theory of Relativity,

page 72 of the 1930 edition.

209

Suppose we have
a

Euclidean space

more dimensions, and suppose we use

of two, three, or

ordinary rectangular coordinates. Here the g*s are all constants.*

Hence,
since the derivative of a constant
is

zero

the Christoffel symbols will also be zero (see page 200);

and, therefore,
all

the components of the

curvature tensor
will be zero too, because every term contains

a Christoffel

symbol (see page 205).

BUT,
if

in

the components of a tensor any given coordinate system

all

are

zero,
its

obviously

components

in

any other coordinate system would also be zero

(consider this
in

the simple case on page 129) c

And

so,
2

whereas from a mere superficial inspection

of the expression for
cfe

whether Euclidean or not,! an examination of the curvature tensor (which of course is obtained from the coefficients
tell

we cannot

the space

is

in

the expression for

1
1

c/s )

*Sec page JScc page

89. 25.

210

can definitely give this information, no matter what coordinate system

is

used

in setting

up

cfs

Thus,

whether

we

use (3) on page

116

or (7) or (8) on
all

page 123,

of which represent

the square of the distance

between two points

ON A
we

EUCLIDEAN PLANE,

using various coordinate systems,

shall find that
in all

the components of B"Tp

three cases

ARE ALL ZERO.*

The same
for all
is

true

coordinate systems

and

for any number of dimensions, provided that we remain in Euclidean geometry.

*To have

a clear idea of

the meaning of the symbolism, the reader should try the simple exercise
of showing that

He

must bear

in

for (8) on B?TP = mind that here

p.

123.

911

1,

312

=
g-2i

o,

gfc^xj,

and use these values in the bracketed expression in (53) on page 205,
remembering of course
is

that the

meaning of [ap 1

1]

etc.

given

by
all

the definition on page 196;

indices,
<r

also that

p
1

etc.

have the possible values

since the space here
is

and 2,

two-dimensional; and he must not forget to

SUM

whenever an index appears TWICE IN ONE TERM.

ANY

211

Thus
(54)
is

B; rp

=
condition

NECESSARY

that a space shall

be

EUCLIDEAN.
It

can be shown that

is

this

also a

SUFFICIENT

condition.

other words, given a Euclidean space,

In

this tensor will

be zero,
is

whatever coordinate system

AND

used,.

CONVERSELY,

given this tensor equal to zero, then we know that

the space must be Euclidean.

We
how
is

shall

the

now see new Law

of Gravitation

EASILY

derived

from

this tensor.

XXV. THE BIG G'S OR EINSTEIN'S OF GRAVITATION.

In

LAW

(54) replace p

by a ,

obtaining

(55)
Since
in

0.

a appears
,

twice
left,

the term on the

must,

we

according to the usual convention,

sum on a,

213

t\

so that (55) represents

4X4
in a

SIXTEEN equations corresponding to the values of <r and r

only
four-dimensional continuum.

when er = r (55) becomes

Thus,

Bin
Similarly for
<r

Bin
r

+
=

Bin

B?u

0.

1 ,

we

get
0121

Bl22

Bi23

Bi24

=
and r

and so on,
for the

16

possible combinations of

<r

We
(56)

may

therefore write (55) in the form

G =
ffr

as
In

where each G consists of 4 shown above.

other words,

B's

by

CONTRACTING
is

B,' rp ,

which

a tensor of the

we

get a tensor of the

FOURTH rank, SECOND rank,

namely,

ffT

as explained

on page 182.

The QUITE INNOCENT-LOOKING EQUATION (56) IS EINSTEIN'S LAW OF GRAVITATION.

Perhaps tht reader
is

startled

by
But

this

sudden announcement.
look into (56)
is

let us

carefully,

and see what

behind

its

innocent simplicity,

215

and why
the
In
it

it

deserves to be called

Law

of Gravitation.

the first place must be remembered

that before contraction,

B*rt>

represented the quantity in brackets in the right-hand member

of equation (53)

on page 205.
it

Hence,

when we contracted by replacing p by a

we

can see from (53) that

Gar represents the following expression:

\

r\

(57)

{era, e}

(er, a}

{<"",

a]

{aa, a]

{err,*}

{,<*},

which,

in turn,

by

the definition of

the Christoffel symbol (page 196)

represents

an expression containing
first

and second
little g's.

partial derivatives

of the

And,

of course,

(57) takes 16 different values as <r and r each take on

their

different values,
letters in

while the other Creek

(57),

namely, a and e, are mere dummies (see page 204)

and are to be summed

(since each occurs twice in each term), as usual.

216

To get
just

clearly in

mind

what (57) means,

is

the reader

advised

to replace each Christoffe!

in

symbol

accordance with the definition on page 196, and to write out in particular

one

of the

16 expressions represented by (57)

<r

by

putting, say

and r

and allowing a and

in succession, the values 1,2, 3

to assume,

,4.

It

can easily be shown

that (56) actually represents

NOT
So
is

16 DIFFERENT equations

but only 10,

and, of these, only

that the

6 are independent.* new Law of Gravitation

first.

not quite so complicated

it

as

appears at

But

why do we call it a Law of Gravitation at all?

It

will

be remembered

that a space,

of any number of dimensions,

is its

characterized

by
2

expression for

cfs

(see page 187).

Thus
(56)
is

completely determined by

the nature of the space which, by the Principle of Equivalence

determines the path

of a freely moving object in the space.

*Se

p. 242.

217

But,

even granting the

Principle of Equivalence,
that
is,

granting the idea

that the nature of the space,
11

rather than a "force

of gravity,

determines

how

objects (or light)

move
in

in that

space

other words,

granting that the g's alone

determine the

Law
ask:

of Gravitation

one may

still

Why
Law

is

this particular

expression (56)

taken to be the
of Gravitation?

To which the answer

it is

is

that
is

the

SIMPLEST

expression which

ANALOGOUS to
Perhaps the reader
at this reply,
is

Newton's Law of Gravitation.

unpleasantly surprised

and thinks

that the choice has

been

made

rather

ARBITRARILY!
book
1

May we
in

therefore suggest to him

to read through the rest of this

order to find out

the

CONSEQUENCES
Law
predict that he will

of Einstein

choice

of the

of Gravitation.

We
and

be convinced
this

of the

WISDOM

of this choice,*
is

will

appreciate that

part of Einstein's

GENIUS!
particularly

*The reader

who

is

interested in this point

wish to look up a book called "The Law of Gravitation in Relativity" by Levinson and Zeisler, 1931.

may

218

He

will see, for

example, on page 271

that the equations giving

the path of a planet,

derived
are the

by Newton,

SAME,

to a

first

approximation,

as the Einstein equations,

so that the latter can

do

ALL
and
the

that the

Newtonian equations do,

FURTHERMORE,

term in (84) accounts for the "unusual" path of the planet Mercury,

ADDITIONAL

which the Newtonian equation (85) did not account for at all.
But

we

are anticipating the story!

Let us
a form

now

which

express Newton's Law in will show the analogy clearly.

XXVI.

COMPARISON OF

EINSTEIN'S

LAW

OF GRAVITATION WITH NEWTON'S.

Everyone knows
that,

according to Newton,* two bodies attract each other

with a force which
is

proportional

to the product of their masses, and inversely proportional to the

square of the distance between them,

thus:

*See the chapter on the "Theory of Attractive Forces*' in Ziwet and Field's Introduction to Analytical Mechanics,

219

In this

formula

we

regard the two bodies, of masses mi and m ,

>

as each concentrated at a single point

11

(its

and

"center of gravity r is then precisely

),

the distance between these two points. Mow we may consider that mi
5

is

in

surrounded by a "gravitational field* which the gravitational force at A

(see the diagram on page 221) is given by the above equation.

If we divide we get

both sides by m%
fc/m ~ __ _F_
1

o~~

ni2

r"

And,

according to Newton,
a
the acceleration with which

=
/3l2

the force

m> would move due to F acting on it.

We
(58)

may

therefore write

where the constant C now includes

since

we

are speaking of
.

the gravitational field around mi

*Thus
it is

a fact that

to support a
it is

body
over,

not necessary to
it

hold

up

all

but one needs only support it right under its center of gravity,
as
if its

entire mass
at that point.

were concentrated

220

Now,
and
it

acceleration

is

a vector quantity/*

may be

split

up

into

components:!
,

Thus take the origin to be at mi and the mass 012 at A i

then

OA =
let
2

r;
at

and

(since
in

AB represent the acceleration m is being pulled toward mi)

both magnitude and direction.

if

Now
it is

is

the x-component of a

obvious that

X
a

x
r'

Therefore

Or,

better,

*See page 127. fSee page 129.

221

to

show

that the direction of

is

to the

left.

Substituting in this equation

the value of a from (58)

we

get:

y- A
And,
similarly,

^*

y=

--

and,

in

3-dimensional space,

we would have
By

also

Z=
get:

--p

differentiation,

we

dx
But, since r
(as
is
2

=
,

+
r),

obvious from the diagram

if

on page 131
i

AB =
'

dr

then

x = r

dx

Substituting this in the

it

above equation,

becomes

dx

And,

similarly,

From these we
(59)

get:

3X,ay,az _ TV"
[-

ox

ay

az

222

This equation

may be
,

written:
,

(60)

dV
ax
is

2 "*"

^
a/

*
dz
2

2 "*"

where

a (unction such that

and

is

called the
1

*
*/

"gravitational potential

obviously (60)

is

merely another way

of expressing the field equation (59)

obtained from

Newton's Law of Gravitation.

This form of the law,
is

namely (60),

generally 'known as the Laplace equation

and

is

more

briefly
2

denoted by

v ^ where the symbol

that

merely denotesf

the second partial derivatives with respect to x , y , and z ,

respectively,
are to
as

be taken and added together, shown in (60).

from (60), then,

We see

that the gravitational field equation

obtained from

Newton's Law of Gravitation

an equation containing the second partial derivatives of the gravitational potential.
is

See footnote on page 21 9. is read "nabla", fThe symbol 2 and is read *'nabla square".

223

Whereas (56)

is

a set of equations

which also contain

nothing higher than the second partial derivatives
of the g's,

which,

by the Principle of Equivalence, replace the notion of

a gravitational potential

derived from the idea of

11

a "force

of gravity,

by

the idea of

the characteristic property of the SPACE in question (see Ch. XII). It is therefore reasonable
to accept (56) as the
gravitational field equations

which follow from the idea of

the Principle of Equivalence.

HOW
will

REASONABLE

it

is

be evident when we test it by

EXPERIMENT!
It

has

been

said (on

page 21 5)

that

each

consists of four B'S O

Hence,
if

the B's are

all

zero,
all

then the G's will

be zero;

but the converse

is

obviously

NOT true:
all

Namely,
even
it
if

the G's are

zero,

does not necessarily follow

that the B's are zero.

925

But

we know

that,

to have the

fi's all

zero

implies that

the space

is

Euclidean (see p. 213).

Thus, if the condition for Euclidean space

is

fulfilled,

namely,

Off Tp

= V/ Ba ~
automatically follows/

then
thus

G<rT

is

true in the special case of

Euclidean space.
But, more than since
this,

does
hence

NOT NECESSARILY
fi's

imply

that the

are zero,

can be true

EVEN

IF

THE SPACE

IS

NOT EUCLIDEAN,
namely, in the space around
a

body which

creates a gravitational field.

Now
but

all this

sounds very reasonable,

naturally asks:

still

one

"How
Law
be

can

this

new

of Gravitation

tested

EXPERIMENTALLY?"
226

Einstein suggested several ways in which it might be tested.

and,
as every child

now knows/

when

the experiments were

actually carried out,

his predictions

were

all fulfilled,

and caused

a great stir

not only in the scientific world, but penetrated even into

the daily news the world over.

But doubtless the reader

would

like to

know

the details of these experiments/ and just how the above-mentioned

Law
is

of Gravitation

applied to them.
is

Thai

what

we

shall

show

next.

XXVII.

HOW CAN

THE EINSTEIN

LAW OF

GRAVITATION

BE TESTED?

We

have seen that

G, r

=
new

represents Einstein's Law of Gravitation/

and consists of 6 equations

containing partial derivatives of the little g's.*
*See pages 21 5 to 21
7.

227

,u

In

order to test

this

law
substitute in
it

we must obviously

the values of the g's which

actually apply in our ohysical world;
in

other words,
first

we must know
what
is

the expression for d* which applies to our world

(see Chapter XIII).

Now,

if

we

use

the customary polar coordinates,

we know
in

that

two-dimensional
2 c/s

EUCLIDEAN
2

space

we have

c/r

+ rW.*

Similarly,
for three-dimensional

EUCLIDEAN
we have

space

the well-known:
2

c/s

c/r

+ rW +

sinW</>

The reader can

2
c/s

easily derive this from

c/x?

Jxl

+ Jx]

(on pa g c 189),

by changing

to polar coordinates with the aid of the diagram

on page 230.
*See page 123

229

where
xi

X2

xs

= = = =

x
r

= 01 = OMcos
cos

<t>

POM
r

cos

=
</>
<

r sin ^

cos

<

y
z

= LM = = PM =

OM sin
cos
0.

r sin

sin

And,
for

4-dimensional space-time

230

we have
2

fc/s

= -

Jr

rW - fWe/fl-cfy +
2

(61 a) or
2

= =

[c/s

c/x?

- xl^ 9, x3

x?sinVc/xi
4

-f c/xi

(where xi

r, x2

= 0, x =

and c
as

we

taken equal to 1), can readily see:

is

Note
in

that the general form for

four-dimensional space
Cartesian coordinates,

analogous to the 3-dimensional one on p. 189,

2

c/s

Jx

dy

</z

c/r

But 7 on page

67

we showed
in

that

order to get
in

the square of an "interval"

space-time
in this

form/
four plus signs,
to take r

with

all

we had

NOT

equal to

the time, t, BUT to take T

i

= -

ict* where

= =

the velocity of light;

2

from which
c/r

= - cW,

and the above expression becomes:

*As
in

a matter of fact,
11

"Special Relativity,

we

took r

it,

but that was because

we

also took c

1/

otherwise,

we must

take r

=
231

ict.

And,
2 2

furthermore,

since in actual fact,

c
c/t
is

always found to be
2

Sreater than (</x

c/y

c/z ),

therefore,

to

it is

make c/s come out reel instead more reasonable to write

2

of imaginary,

c/s

-c/x

-Jy

-c/z

+cW,

which

in

polar coordinates,
a).

becomes (61

The reader must

this

clearly realize that

formula

still

applies to

EUCLIDEAN
which
the
is

space-time, involved in

SPECIAL

theory of Relativity*

where we considered only

observers moving with

UNIFORM

velocity relatively to each other.

But now, in the GENERAL theory (page 96)

where we are considering

accelerated motion (page 102),

and therefore have a

NON-EUCLIDEAN
(see Chapter XII),

space-time
2

what expression
shall

for

c/s

we

use?

In

the

it is

first place reasonable to assume that

(61 b)

Js

= -

e* </x?

e* (xldxl

+
(where Xi
*See Part
,

e'c/x

X2 / Xa / X4 represent
of this book.

233

the polar coordinates r, 0,

respectively,

$, and

t,

and A

IJL

and

v are (unctions of xi only),

BECAUSE:
(A) we do not include product
of the form
or,
cfxi

terms

cfa

more generally, of the form dx c/xr where

ff

cr

(which
since

ARE

included

in

(42), p. 187)

from astronomical evidence

it

seems

that
is

our universe
(a)

ISOTROPIC and

(b)

HOMOGENEOUS:
is,

That

the distribution of matter

(the nebulae)
is

the

SAME
ALL DIRECTIONS
and

(a)

IN

(b)

FROM WHICHEVER POINT WE LOOK.

the omission of terms like

Now, how does

dx JxT where
ff

<r

represent this mathematically?

Well, obviously,
a term like c/rc/0
(or C/0-C/0 or c/rc/(/>)

would be
for 6 (or

different

</>

or r) positive or negative,
2

and, consequently,
the expression for
c/s
if

would be
in

different

we

turn

opposite directions

234

which would contradict the

experimental evidence that the universe is ISOTROPIC

And
from

of course the use of

2

the same expression for

c/s

ANY point
HOMOGENEITY.
it is

reflects the idea of

And

so

we

see that

reasonable
2

to have in (61 b)

only terms involving c[0 , c[</> , c/r in which it makes no difference

whether

we

substitute

+c/0 or

c/0, etc.

Similarly,

since in getting a measure for

c/s /

we are considering STATIC condition,

is

and not one which

from

changing

moment

to

moment,

we must
for
in

therefore not include

terms which will have different values

+dtand

-c/f/

other words, we must not include

product terms like c/rc/t,

In short

etc,

\we must not have any terms involving

cfx^-c/Xr

where

or

^T,

but only terms involving

dx -dxr where
ff

(7

T.

(B)

The
in

factors

x
,

M
, e",

are inserted

the coefficients

to allow for the fact

*Cf.(61a)and(61b).

235

that our space

is

now

NON-EUCLIDEAN.
Hence they are so chosen as to allow freedom to adjust them
to the actual physical world /since they are variables),

jnd yet
iheir
it

FORM

is

such that
to manipulate

will

in

them making the necessary adjustmentshall see.*

be easy

as

we

Now,
(61 b) can

be somewhat

simplified

by

replacing
2

e*x?

by

(x() ,

and taking

x( as a

new

coordinate,

thus getting rid of e" entirely/

and we may even drop the prime, since any change in c/xi which arises from the above substitution can be taken care of by taking X correspondingly different.
Thus (61 b) becomes, more simply,
2

(62)

c/s

= -

e\/x?

xldxl

- xrsinWx2 +
.

e'-c/x4

And we now

have to find

the values of the coefficients e and e"

in
x

terms of xi.j

^Further justification for (61 b)

may be found
R. C. Tolman's
Relativity
p.

in

Thermodynamics

& Cosmology,

239

ff.

ISee page 934.

236

We warn the reader that this a COLOSSAL UNDERTAKING,

is

we

but, in spite of this bad news, hasten to console him by

telling

him

that

terms will reduce to zero, and the whole complicated structure

many

will

we

melt down to almost nothing; can then apply the result

to the physical data with the greatest ease.

To any reader who "can't take

'

it

we

suggest that

he omit the next chapter and merely use the result

to follow

Einstein

the experimental tests of the Law of Gravitation

given from page

255

on.

BUT HE WILL MISS

LOT OF FUN!

XXVIII.

SURMOUNTING THE

DIFFICULTIES.

So

far,

we have
SLI

then, the following values:

x
/ ff22

= cr

and gar

x?

g33

= -

x?sin

x2 , g44

e*

when

r.

(see (62) on p. 236.)

Furthermore, the determinant

is

g (see page 194)

simply equal to
principal diagonal,

the product of the four elements in

its

237

since

ail

the other elements are zero:

Hence

Also 7

in this case,

and
T

g*

when a

r.

We
in

need these relationships determining e* and e in (62).

shall shall

Now we
how

see

the big G's will help us to Find the little g's

and how the little g's will help us to reduce the number of big G's 10

ONLY

THREE!
show
that

First let us

the set of quantities

is

SYMMETRIC,}:

and therefore
*See the chapter on determinants in any college algebra,
to Find out
a

how

to evaluate
"

determinant of the fourth order.

M tSee the definition of g on page 196. 193. {Seepage

239

Gar =
as

reduces to

TEN
on

equations

instead of sixteen/ v and r each take

1

their values

,2/3,4.
that
p.

To show

this,

we must remember

on and let us examine {act, a} which occurs in (57):

ffT

really represents (57)

216;

By

definition (page 196),

But,

remembering

that

the presence of in term

a and

TWICE

EACH

(after

multiplying out)

on implies that we must the reader will easily see that

SUM

a and

many
and

of the terms will cancel out

that

we

shall qet

Furthermore, v the definition of g" on page 196, the reader may also verify the fact that

by

dg

2ff

dg

where g

is

the determinant of p. 239.

And,

from elementary calculus,

*See pase 193.

240

Similarly,

/} =
we
get:

:->-

log

v-gr.

Substituting these values in (57),

<\

(63)

G =
ffT

{era, e}

{er, a}

jar, aj

We

can

now

easily see

that (63) represents

10 equations and not

In

sixteen/

(or the following reasons:

the

first

place/

{er, a]

{re, a}

(see pp. 204, 205).

Hence,

by
the
its

interchanging a and r , first term of (63) remains unchanged,

factors

two

merely change places

*Notc

that

we might

also have obtained

is

V +

_
g,

= -y, (we shall show on p. 252 that X 4 2 and therefore g on p. 239 becomes xi -sin X2) it is more reasonable to select v g f which
will

but since g

always negative

make

the Christoffel symbols,

in

and hence also the terms

the

new Law

of Gravitation,

REAL

rather than imaginary.

241

(since
as

and a are mere dummies,

explained on page 204).

And,
the second, third and fourth terms of (63) are also unchanged by

the interchange of a and r. in other words,

Thus, if we arrange the 16 quantities in

in a

JT

square array:

Gn
vJ21

Gi2
tf22

Gis
VJ23

G
\3

Gsi
G41

Gs2
642
that

633
Gl3

G 6

We
this

have
is

just

shown

SYMMETRIC
to

matrix.*

Hence (63) reduces instead of 1 6 ,

as

10 equations

we

said before.

We
how

shall

not burden the reader

with the details of

(63)

to only
that

is further reduced SIX equations.! is

But perhaps the reader

are
still

thinking

"only six" equations

no great consolation, particularly if he realizes

*See page 239. he is interested, he may look this up on page 115 of M 1 The Mathematical Theory of Relativity/

flf

by A. S. Eddington, the 1930 edition.

24*

how long each of these equations But does he realize this?

he would do well to take cr and r = 1 , and r = 1 , say <r in order to see just what
particular values of
,

is

ONE
is

of the equations

in

(63)

really like!

(don't forget to
Is

sum on the dummies

I)

just
Is

the reader wondering what we are trying to

this a subtle

do

to

him?

mental torture

by which we
alternately frighten The fact is that

and console him?

we do want
to

to frighten him sufficiently

realize

make him

the colossal amount of computation that is involved here,

and yet to keep up

his

courage too

by the knowledge
it

that

does eventually boil down

might not appreciate simple form
it.

to a really simple form.

He
the
if

final

he did not know

this

the labor that produced

With

we shall how the

apology, now proceed to indicate

further simplification

takes place.
In

we

each Christoffel symbol in (63), must substitute specific values

for the
It is

Greek

letters.

obvious then

243

that there will (a) those in

all

be

four possible types:

letters are alike:

which the values of

three

Greek
g

Thus:

ffo-

a}
0-<r ,

(b) those of the form

(c) those of the form

or

and
(d) those of the form
{err ,

p}.

Note that it is unnecessary to consider the form {rer, r} since this is the same as (or, r} (see p. 204).

Now, by

definition (page 196),

and, as usual, we must sum on a.

But since the only g's which are not zero
are those in which

the irrdices are alike (see p.

237)

and,

in that case,

flT*=1/ff~

(P- 239).

Hence
f

_
(jcr,

a} }

dS _ jL( v~ ^'l Adx

2g

9~ 4-r ~^5x
ff

d 9~ ---

\
I

(T

<y

3x,/

and therefore

which,

by elementary
{<r<r,<r}

calculus, gives

(a)

=
^

log

gro.

244

Similarly,

+ *"-' = Mi!*-" 2 Vox,

will

1dx,

-Hdx
a

Here the only values keep the outside

of

a
T

that
ra

factor
,

are those for which

a =

from being zero

and since r

(for otherwise

we

should have case (a)

we

get

or
/L\ (b)

Likewise

(c)

{CTT.T}

=
g

log grr

and
(d)
{<rr

/P }=0.

Let us

now

evaluate these various forms

for specific values:

Thus, take,

in

case (a),

cr

=
1

1
r)

Then

{11/1}

^7

lo

33n

x Butgn= -c

(Seep. 239).

hence {11

i}=l.|og(-O
calculus, gives

which,

by elementary

245

L V 4 A represents where

or

6x1
since xi

dr

r (see

page 233),

Similarly,

in,
But, since

i4

if

*-J

in taking a PARTIAL derivative with respect to one variable, all the other variables are held constant,

hence

and therefore

(22,2} =0.

And,

likewise,

{33 ,3)

= {44,41=0.
2
/

Now,
take

for case (b),

first (r

then
{11
,

2}

= -

^
x2
is

gn

= -

~
2flL> 2

~3x2

But, since X is a function of x\ and is therefore held constant

only/

while the partial derivative

with respect to x>

taken,

hence {11,2} and so on.

Let us see

=0,
specific values

how many
in all.

we

shall

have

Obviously

(a) has

specific cases,

*Se

page 234.

246

1 ,2,3,4, namely, <r which have already been evaluated above.

(b) will have 12 specific cases, since for each value of a = 1 , 2

r can have 3 of
(for

its

possible 4

values

here a

r)

(c) will also

have 12 cases,

and
(d) will have 4 but since {or , p}
this

24

cases,

[TV , p\

(see p. 204),

reduces to
in all

1 2.

Hence

there are

40

cases.

The reader should verify the fact that 31 of the 40 reduce to zero,
the

9 remaining ones being

(64)

'Remember

that

x2

Note

that

/=

oxi

~=%
or

Now,

in (63),

when we

give to the various

Greek

letters

their possible values, we find that,

since so many of the ChristofFel symbols are equal to zero, a great meny (over 200) terms drop out ! And there remain now

only FIVE equations, each with a much smaller number of terms. These are written out in full below, and,
lest

this
final

the reader think that is the promised

simplified result, hasten to add that the BEST is yet to come!

we

Just

Gii is obtained, the reader how showing to on a. and e and which terms drop out (because they contain zero factors}

how

SUM

will be found on page 31 7.

in

V,

And,
similarly (or the other G's.

Here we give the equations which result after the zero terms have been eliminated.

248

Gu=

11,1} {11,1}+
13, 3} {31,3}

12, 2} {21, 2}

14,4)141,4}

=
Similarly,

0.

,3] (83,3)

0.
/

= 2{33 1}{13,3}

+ 2{33,2}{23,3}
133, 2}

-ST {33,11-ir

0.

,1}

{14,41-(44,1}

0.

0.

249

If

we now

substitute

these equations the values given in (64),

in

we

get

424
v

"

""

xv - -

=
Similarly

0*

G =
33

sin

0-e-

(/

X')

sin

0.

=
=

0.

d X

\"

and

v"

=
j

250

and 612 becomes:

cotfl

cotfl

which is identically zero and therefore drops out, thus reducing the number of equations
to

FOUR.
also that
,

Note

633 includes 622

so that these two equations

are not

independent hence now the equations are

THREE.

And
we

now, dividing

44

by

e*~
,

and adding the

get
A'

result to

Gn

(65)

= - / = ~
dv T" dr
'

d\
or

T" or

Therefore,

by
A

integration,

(66)

= -

where

/ is a

constant of integration.

But, since
at

an

infinite distance

from matter,

our universe would be Euclidean,*

*See page 226.

251

and then, (or Cartesian coordinates, we would have:

that

is,

the coefficient of c/x? and of dx\ must be 1 under these conditions/

hence,
if

(61 b)

is

to hold also for

this special case,

as of course

we
In

it must do, should then have A =

0.

other words,
v

since,

when

then, from (66),

X also equals , / , too, must be zero.

Hence
(67)

\=-v.

Usins (65) and (67),

622

on page 250 becomes

e'(1
put

(68)
If

+n/) =
,

1.

we

and

differentiate with respect to r

we

get

or
or

dy ^~

"

dv
dr

Hence (68) becomes

(69)

+ fy =

1.

*See pages 189 and 231.

252

This equation

may now be

easily integrated/

obtaining

(70)

where 2m is a constant of integration. The constant m will later be shown

to have an important physical meaning.

Thus

we have succeeded
and e an
x

in finding

e'

*From elementary theory of

differential equations,

we

write (69):

or

or
cf

T_ = -Jr
r

1-7
Having separated the

'

we

can

now

variables, integrate both sides

thus:

log (1
or

7)

= =

log r

+ constant;

log r (1

7)

constant,

and therefore
r (1

7)

constant.

We

may

then write

r (1

we
A 1

7)

= 2m
2m
.

from which

get

253

in

terms of xi
e"

1/e

=7 = 1- 2m/r = 1
2
2 2

2/n/x,

and (62) becomes:

(71 )
ctf

= - 7-i</r2 - rV0 =
=

r sin

c/</

where, as before (p. 233),

r

xi /

x2

x3

x4

And
the

hence

new Law of Gravitation, consisting now of only

the

THREE

remaining equations:

Gn =
are

G =
33

and

G =
44

now
little

fully

determined by

the

of (71).

We

can

now proceed
it

to test this result to see whether

really applies

to the physical world

we

live in.

XXIX. "THE

PROOF OF THE PUDDING.

11

The
the

first

test

is

naturally

to see what

new Law

of Gravitation

has to say about the path of a planet.

It

was assumed by Newton that

1 *

body

"naturally

moves

along a straight line

255

if it is

not pulled out of

force acting

its
it:

course

by some
As,
a
for

on

example,

body moving on

a frictionless Euclidean plane.

Similarly,

according to Einstein
if it

moves "freely

11

on

the surface of a
it

SPHERE
1

would 30 along the

is,

"nearest thing to a straight line/

that

along the

GEODESIC
circle.

for this surface,

namely, along a great

And,
it

for other surfaces,

or spaces of higher dimensions,

would move along

the corresponding geodesic for the particular surface or space.

Now
what

our problem
is

is

to find out

the geodesic

in

our non-Euclidean physical world,

since a planet must

move

along such a geodesic.

In

order to find

the equation of a geodesic it is necessary to know

something about the

1

"Calculus of Variations/ so that we cannot go info details here.

But

we

shall give the reader

a rough idea of the plan,

together with references where

257

he may look up this matter Suppose, For example, that

further.*

we have given two points, A and B

on

Euclidean plane,it is obviously possible to

join
thus:

them by various paths,

B
Now,
which of
is

all possible paths the geodesic here? Of course the reader knows the answer:
It is

the straight line path.

But

how do we

set

up

the problem mathematically so that we may solve

similar

problems

in

other cases?

*(1) For fundamental methods see "Calculus of Variations/' by

G. A.

Bliss.

(2) For this specific

1

problem see "The Mathematical Theory of

Relativity/ by A. S. Eddington, p. 59 of the 1930 edition.

(3)

Or

see pages 128-1 34 of

1

"The Absolute

by

Differential Calculus/ T. Levi-Civita.

258

Well,

we know
that
if

from ordinary calculus

a short arc

on

any

of these paths

is represented then

by
f.

c/s ,

f
JA

represents the total length of

that entire path.

And

this of

course applies to

any one

of the paths from

to B.

How

do we now

select from

among

these

the geodesic?
This

problem is similar to one with which the reader

undoubtedly
familiar,

is

namely,
if

y
is

f(x),

find the values of

y for which 16 an "extremum" or a "stationary.

in

Such values of y are shown

the diagram on the next page at a , 6 , and c:

259

a
and/ (or
all

these/

we must have

dy/dx
or

(72)

<fr=ffo)-c/x
is

where x,
Similarly/

or

c.

to

go back to our problem on pp. 258 and 259,

the geodesic

we

are looking for

would make
rB

f JA
a stationary.

This

is

expressed

in

the calculus of variations

by

(73)

JA

c/s

analogously to (72).

To

find the

equation of the geodesic/

satisfying (73),
is

not as simple as finding

260

maximum

or

minimum

in

ordinary calculus,

and we

shall give here the result:* only

(74)

W .,

Let us consider (74):

In

the

first

place,

for ordinary three-dimensional

Euclidean space a would have

three possible values: since we have here

1,2,3,
x3 /

three coordinates xi, X2,

furthermore,

by choosing

Cartesian coordinates,

we would have
ffll

(see page
==

89):

522

ff33

and

& =
and therefore

for

/i

which involves derivatives of the g'sf

would be equal to zero, so that (74) would become

(75)

~
in

*For details, look up the references the footnote on p. 258.

fSee (46) on p. 196.

261

Now, if in (75) we replace c/s by

it

dt

becomes

(76)

-0
way
of writing

(*

= 1,2,3)

which

is

a short

the three equations:

/ 77 \

(77)

But what
the

is

PHYSICAL MEANING
we
consider an observer

of (77)?

*lf

who

has chosen his coordinates

in

such a

way

that
C/X2

C/Xl
in

C/X3

other words, an observer who

a

is

traveling with

moving object, and for whom the object

standing
still

is

therefore

with reference to

his ordinary space-coordinates, so that only time is changing for him,

then, for him (61 a)

2

becomes

c/s
2

= = =

2 c/x4
2

or or

c/s

c/t

c/s

c/t.

That
c/s

is

becomes

to say, of the nature of "time/"

for this reason

c/s is

often called

"the proper time"

since
it is

"time"
itself.

for the

moving object

262

Why, everyone knows

(or uniform

that,

motion,

v=s/t,
where v
a
it

is

the velocity with which

body moves when

goes
a distance of 5 feet in

t
If

seconds.
the motion
is

NOT

uniform,

we

can, by means of

elementary differential calculus,

express the velocity

AT AN INSTANT,
.

by
v

ds/dt
are

Or,

if

x, /, and z
s

on the and Z X, Y, axes, respectively, and vx , vy , and vz are the projections of v on the three axes,
the projections of

then

e/x ~

dy

__

di

in the

abridged notation,

v,

^
vx ,vv ,Vz'

('=1,2,

3)

where we use
instead of

xi, X2 , Xs

x, y, z,
, va

and

vi ,

v2

instead of

Furthermore,
since acceleration
is

the change in velocity per unit of time,

263

we have
Jv
or

(78)

<T

=^
1.2

tr-1 V*
*

2 *

3) */

Thus (77)

states that

the components of the acceleration

must be zero, and hence the acceleration must be zero, thus:

itself

or

From
or

this

we

set,

by

integrating,

a constant,

-r
and therefore

a constant,

by

integrating again,
s

= at+

b,

which

is

the equation of

A
In

STRAIGHT
other words,

LINE.

when the equations

namely, (74),

for a geodesic,

are applied to the special case of

THREE-DIMENSIONAL EUCLIDEAN SPACE,

they lead to the
(act that

264

in this special

case
IS

THE GEODESIC

STRAIGHT

LINE!

We
and

hope (he reader

is

DELIGHTED

NOT DISAPPOINTED
is

to get a result which

so familiar to him;

and we hope
in (74)!

it

gives him

a friendly feeling of

confidence

And
that

of course

he must

realize

(74) will work also for any non-Euclidean space,

since
it

contains

the

little g's

which characterize the space/*

and

for
cr

any dimensionality,

since

may be

given
values.

any number of
In particular,
in

our four-dimensional

non-Euclidean world,
(74) represents the path of an object moving in the presence of matter

(which merely makes the space

non-Euclidean), with no external force acting upon the object;

and hence (74)

is

THE PATH OF

PLANET
for!

which we are looking

*Seep. 190.

265

XXX.

MORE ABOUT

A
Of
a

THE PATH OF PLANET.

course (74)

is

only

expression, and does not yet apply to

GENERAL

our particular physical world, since the Christoffel symbol

involves the g's, is therefore not specific until we substitute the values of the g's

and

which apply
in

in a

specific case

the physical world.

in

Now
in

(64)
,

we have

And, by (67), A hence we know {&$ ,

v
, r

the values of \a/3 terms of X , v , r and 0.

a}

<?}

in

terms of

and

6.

Further, since e"

= 7

and 7

is

known

in

(see page 252) terms of r from (70),

f

we

therefore have [ct$

r

a}

in

terms of

and

0.

The reader must bear

that

in

mind

whereas (76), in Newtonian physics, represents only three equations, on the other hand,
(74) in Einsteinian physics is an abridged notation for

FOUR

equations,

266

as
its

<J

takes on

FOUR

Taking

first

possible values: 1,2,3,4. the value a = 2 ,

and, remembering that x2

(see page 233),

have, (or one of the equations of (74), the following:

/-rn\

we

a
2

(79)

+{^} dx
.

dxa

^
-

And
since

now, a and

/?

each occur

TWICE
the second term, must sum on these as usual, so that we must consider terms
in

we

containing, respectively,

{11,2}, {12,2}, [13,2}, {14,2}, {21,2}, {22,2}, {23,2}, {24, 2}, etc
which a always equals 2 , and a and /3 each runs its course
in

from

to 4.

But, from (64), we see that most of these are zero,

the only ones remaining being

and

{33,2}
Also, by page 204,

= -

sin0-co$0.

{21,2}

{12,2}.
267

Thus (79) becomes

fftfYk

<PO

(80)

+ -.._.-

1 dr dd

tm fl. eoi

,/</</>

..

=0.

If

in

we now choose our such a way

coordinates

that

an object begins moving

6

in

the plane

= =

7T/2

then
-j-

and cos0

di

and hence

If we now substitute all these values in we see that this equation is satisfied,

(80),

and hence
thus

ir/2

is

a solution of the equation,

showing
plane.

that

the path of the planet

is

in a

Thus from (80) we have found out that

a planet,

according to Einstein, must move in a plane,

just as in

Newtonian physics.
further,

Let us now examine (74) and see what

tell

the 3 remaining equations in

us

it

about planetary motion:

263

For

<T

1 ,

(79) becomes

Or

But since
then

we have chosen
In

7r/2

~-r

and

sin0

1 ,

hence

this

equation becomes

And
for
a-

similarly,

(79) sives
( 82 )

as

as as

and

for

cr

4,

we

get

(83)

+F f** ^ or
as as

o.

269

And now
from (81), (82), (83), and (71), tosether with (70),

we

get*
2

c/

W
(84)

___

m
A
5

==

where c and h are

constants of integration,

and u

1/r.

Thus (84) represents the path of an object moving freely,

that
is,

not constrained

by any

external force,

and
in a

is

therefore,
straight line in

sense,

analogous to a

Newtonian physics. But it must be remembered

that in Einsteinian physics,

owing to the
Principle of Equivalence (Chapter XI), an object is constrained by any external force

NOT

even when
as, for

it

is

moving

in

the presence of matter,

example,

a planet

moving

around the sun. And hence (84)

would represent the path of

*For details see page

a planet.

86

in

"The Mathematical Theory of

Relativity/'

by

A.

S.

Eddington (the 1930 edition).

270

From

this

point of view

we

are not interested in

comparing (84) with the straight line motion

in

Newtonian physics, as mentioned on page 270,

but rather with
the equations representing the path of a planet
in in

Newtonian physics,

which, of course, the planet is supposed to under the

of the sun.
it

move

GRAVITATIONAL FORCE
has
in

been shown Newtonian physics

that a

body
1

moving under a "central force/ (like a planet moving

under the influence of the sun)

moves

in

an ellipse,

with the central force (the sun) located at one of the foci.*

And

the equations of this path are:

2

e/

u
~r u
,

__ -

m
TO

jLTsi

(85)

*
where
r
is

"

the distance

from the sun to the planet, m is the mass of the sun,

*Sec Ziwct and Field: "Mechanics/*
or

any other book on mechanics.

271

is

4> is

the semi-major-axis of the ellipse, the angle swept out by the planet
t.

in

time

We

notice at once

the remarkable resemblance between

(84) and (85).

They

are indeed
for

IDENTICAL EXCEPT

the presence of the term and of course the use of

</s

3mu2,

instead of dt in (84).*

Thus
the

we

see that

Newtonian equations (85)

are really a first approximation to

the Einstein equations (84);

that
is

why
satisfactorily

they worked so
for so long.

Let us

now

see
is

how

the situation

affected

by

the additional term 3mir.

XXXI. THE PERIHELION

OF MERCURY.

Owing

to the presence of the term

3mu2
(84) is no longer an ellipse but a kind of spiral
in
is

which the path

NOT
p.

retraced

each time the planet

*See
262.

272

makes
but
in
is

complete revolution,

shifted as

shown

the following diagram:

in

which
is,

the "perihelion/* that the point in the path

nearest the sun, S, at the focus/ is at the first time around/

at at

6 C

the next time/ the next/

and so on.

273

In

a planet

other words, does not 30

round and round

in the same path, but there is a slight shift in the entire path, each time around. And the shift of the perihelion can be calculated

by means

of (84).*

This shift can also

be

experimentally, and therefore can serve

as a

MEASURED
method

of

TESTING
the Einstein theory
in actual fact.

Now
when
is

it is

obvious that

a planetary orbit

very nearly

CIRCULAR

this shift in
is

the perihelion

not observable, this is unfortunately the situation with most of the planets. There is one, however,

and

in

which

this shift

measurable, namely,
the planet

IS

MERCURY.

Lest the reader think

that the astronomers

*For details see aqain

in the

Eddington's book referred to footnote of page 270.

274

can make only crude measurements, let us say in their defense, that the discrepancy even in the case of Mercury is an arc of

ONLY ABOUT
43

SECONDS PER CENTURY!

Let us make clear what we mean by 19 "the discrepancy: when we say that the Newtonian theory requires the path of a planet to be an ellipse, it must be understood that this would hold only if
there were a SINGLE planet; the presence of other planets causes so-called "perturbations," so that

even according to Newton

there

would be
shift in

some

the

perihelion. But the amount of shift due to this cause has long been known to be 531 seconds of arc per century, whereas observation shows that the actual shift is

574 seconds,
thus leaving a shift of

43 seconds per century

in

UNACCOUNTED FOR
the

Newtonian theory.

Think of the

DELICACY
275

of the measurements

and the patient persistence

by
by

over a long period of years generations of astronomers

is

that

represented

the

above

figure!

And

this figure

was known

to astronomers

It

long before Einstein. worried them deeply

since they could not account for the presence of this shift.

And

then

the Einstein theory,

which originated in the attempt to explain

the Michelson-Morley experiment,* with the intention and

NOT AT ALL

of explaining the shift in the perihelion of Mercury,

QUITE INCIDENTALLY EXPLAINED

THIS DIFFICULTY
for the

ALSO,
in

presence of the term 3mu'

(84)

leads to the additional shift of perihelion of 42.9" |

!

XXXII.

DEFLECTION OF
in

A RAY OF

LIGHT.

We

saw

the previous chapter

that the experimental

evidence

*See Part

I.: the Special Theory of Relativity. fFor the details of the calculation which leads from (84) to this correction of perihelion shift,

see p. 88 of the

1930

edition of
1

"The Mathematical Theory of Relativity/

by

A.

S.

Eddington.

276

in

connection with
shift of

the

the perihelion of
at

Mercury

was already

hand

when Einstein's theory was proposed, and immediately served as a check of the theory.
Let us

now

consider

further experimental verification

of the theory,

but

this

time

the evidence did not precede

but was

PREDICTED BY

the theory.
This

was

in

connection with

the path of a ray of light as it passes near a large mass

like the sun.
It

will

be remembered

that

according to the Einstein theory the presence of matter in space

makes the space non-Euclidean and that the path of anything moving freely (whether it be a planet
or a ray of light)
will

be along

a geodesic

in that

space, and therefore

will

be affected by the presence

of these obstacles in space.

Whereas,
according to classical physics,
the force of gravitation could be exerted

only by one mass (say the sun) upon another mass (say a planet),
but

NOT

upon

a ray of light.

277

Here then was

a definite difference in viewpoint between the two theories,

and the facts should decide between them. For this it was necessary
to observe

what happens to a ray of light coming from a distant star

as
is
it it

passes near the sun

bent toward the sun, as predicted by Einstein, or does it continue on

in a straight line, * as required by classical physics?

Now
to

it is

make

this

obviously impossible observation

since

under ordinary circumstances, we cannot look at a star

rays are passing near the sun,
itself:

whose

on account of the brightness of the sun Not only would the star be invisible,
but the glare of the sun

would make
to

it

impossible
direction at
all.

look
so

in that
it

And
*lf,

was necessary
were considered

to wait for a total eclipse,

however,

light

to

a stream of incandescent particles instead of waves,

be

the sun

WOUL6

have

a gravitational effect upon a ray of light, even by classical theory,

the

AMOUNT of deflection
on
this basis,

calculated even
as

DOES NOT AGREE

we
shall

with experiment,

show

later (see p.

287).

278

when
but

the sun
glare
is

is

its

up in the sky hidden by the moon,

so that the

stars

become

distinctly visible during the day.

Therefore, at the next total eclipse astronomical expeditions were sent out to those parts of the world

where the eclipse could be

advantageously observed/
and,
since such an eclipse
lasts only a few seconds, they had to be prepared

to take photographs of the stars

rapidly and clearly, so that afterwards,

upon developing the

the positions of the

plates/

stars

could be compared
with their positions in the sky when the sun is present.

NOT

279

The following diagram shows

/A

F
the path of a ray of light, from a star, v4 , when the sun is in that part of the sky.

AOE ,

NOT

And,

also,
IS present,

when the sun

and the ray is deflected and becomes ACF ,

280

so that,

when viewed from F,

the
star

APPEARS to

be

at B.

Thus, if such photographs

could be successfully obtained,

AND

IF

they showed

that all the stars

in the part of the sky near the sun were really displaced (as from A to B)

AND
the

IF

MAGNITUDE

of the displacements

agreed with the values calculated by the theory,

then of course
this

would

constitute
in favor

very strong evidence

the Einstein theory.

of

Let us

now determine

the magnitude of this displacement as predicted by the Einstein theory:

We
that
in

have seen (on page 233)

1

the "Special Theory of Relativity/

in

which applies
2
c/s

EUCLIDEAN

space-time,
2

cW

(c/x

c/y

+
by

c/z )/
2 cfc ,

we now we get
if

divide

this

expression

but

Jx c/y </z "~T i ~~r i ~~r

are

281

the components of the velocity, v moving thing (see p. 263), then obviously the above quantity in brackets is v2, and the above equation becomes:
a

of

= c

Now

when
1*

the "moving thing

a light-ray,

happens to be

then v

and we

get,

FOR LIGHT, c/$ = 0.

world,

But what about our

NON-EUCLIDEAN
containing matter?
It

will

be remembered (see

p.

118)

that in studying a

non-Euclidean two-dimensional space (namely, the surface of a sphere)

in a certain

small region,

we were aided by
the Euclidean plane which practically coincided with the given surface in
that small region.

Using the same device for space of higher dimensions,

we
in

can,

studying a small region of NON-Euclidean four-dimensional space-time, such as our world is, also utilize the EUCLIDEAN 4-dimensionaI space-time which

282

practically coincides with in that small region.

it

And

hence

will
in

apply

our

FOR LIGHT even NON-EUCLIDEAN world.

And

now,
A

using this result in (71), together with the condition (or

a geodesic,

on page 261 ,

we

shall obtain

THE PATH OF

A RAY OF

LIGHT.

XXXIII.

DEFLECTION OF A LIGHT (Continued)

RAY OF

In chapters XXIX and we the condition for a geodesic

XXX

showed

that

given on page led to (74),

the

260

which, together with little g's of (71) gave us the path of a planet, (84).

And
in

now,

order to find

the path of

A RAY OF
c/s

LIGHT,

we must add

the further requirement:

= 0,
in

as

we

pointed out

Chapter XXXII.

in Substituting c/s the second equation of (84),

283

we

get

which changes the

first

equation of (84) to

which

is

the required

PATH OF
And
by
this,

A RAY OF

LIGHT.

integration*

gives, in rectangular coordinates,

+
for the equation of the curve

on

page 280.

Now,

since

a (page 280)

is

a very small angle, the asymptotes of the curve

may be

found by taking y very large by comparison with x ,

and

so,

neglecting the x terms on the right in the above formula,

it

becomes

And,
*For details see page

90

of

Eddinston's "The Mathematical Theory of Relativity/'

the

A.

S.

1930

edition.

284

using the familiar formula for the angle between two lines

(see any

book on
tan
OL

Analytics):
__

mi ~~~

iii2

where a is the desired angle, and mi and m> are the slopes of the two lines,

we

get

tana
from which
it is

4/?m

4m 2 -fl2

easy to find

sin

4m

4m2/R

And, a being
its

value

in

small, radian measure

is

equal to so that

sin

we now have
/O/A (86)

=
R

4m

+ 4mVR
a

Now,
what
in
in
is

the actual value of

the case under discussion,

which

R =
and

the radius of the sun

is its

mass?

*For the proof of this see

any book on calculus, or look up a table of

trigonometric functions.

285

Since

and

m=

R = 697,000

kilometers,
f

1.47 kilometers
neglected by
/?,

4m 2 may be

comparison with

so that (86) reduces to the

very simple equation:

4m

from which

we

easily get
1

a =
In
it

.75 seconds.

other words,

was predicted by

the Einstein theory

that,

a ray of light passing near the sun would be bent into a curve (ACF),
as shown and that,
in

the figure on p. 280,

consequently
a star at

A would APPEARtobeatB,
the reader will stop a

a displacement of an angle of 1.75 seconds!

If

moment

to consider

how small is an angle of even one DEGREE, and then consider that
one-sixtieth of that
is

an angle of one

MINUTE,
is

and again
one-sixtieth of that

tSee page 315.

286

an angle of one

SECOND,
small
is

he
a

will realize

how
1

displacement of

.75 seconds!

Furthermore, according to the Newtonian theory,* the displacement would be

only half of

that!

And

it is

this

TINY

difference

that must distinguish

between the two

After
all

theories.

the trouble that

the reader has been put to, to find out the issue,

perhaps he

is

disappointed to learn

how

small

is

the difference

between the predictions of Newton and Einstein. And perhaps he thinks that a decision based on
so small a difference

can scarcely be relied upon! But we wish to point out to him,

that,
far

from losing his respect and

faith

in scientific

method,

he should,

ON
be

THE CONTRARY,
the

all

more

filled

with

ADMIRATION AND WONDER

to think that

experimental work
IS that

in

astronomy

SO ACCURATE
278.

*See the footnote on

p.

287

these small quantities* are measured

WIFH PERFECT CONFIDENCE,

and
that

they

DO distinguish
between the two theories and decide in favor of the

DO
as

Einstein theory,
is

shown by the
British

following figures:

The
gave

expeditions, in 1919,

to Sobral

and Principe,

for this displacement:

1.98"

0.1

2"

and
1.61 "0.30",
respectively; values which have since

been

confirmed at other eclipses, as, for example,

the results of Campbell and Trumpler,

who
using

obtained,

two

different cameras,

1.72"
in

0.11" and 1.82"

0.1 5",

the 1922 expedition of the Lick Observatory.

that

So
all

by now

physicists agree that the conclusions are

beyond

question.
in

*See also the discrepancy

the shift of the perihelion of Mercury,

on page 275.

288

We
in

cannot

refrain,

closing this chapter, from reminding the reader that

191 9 was right after World War and that Einstein was then classified dS
a

I,

GERMAN

scientist,

and yet,
the British scientists, without any of the
stupid racial prejudices then

(and

alas! still)
in

rampant

the world,

went to a great deal of trouble to equip and send out expeditions

to test a theory 11 an "enemy.

by

XXXIV. THE THIRD

OF THE "CRUCIAL PHENOMENA.

11

We

have already seen that

two of the consequences from

the Einstein theory

were completely
experiment:
(1)

verified

by

One, concerning

the shift of

the perihelion of Mercury, the experimental data for which

was known long before Einstein

BUT NEVER BEFORE EXPLAINED.

And

it

must be remembered

that the Einstein theory

was

289

NOT expressly designed

explain but did
this shift,
it

to

QUITE INCIDENTALLY!
(2) The other, concerning the bending of a ray of light as it passes near the sun.
It was never suspected before Einstein that

a ray of light

when

passing

near the sun

would be bent.*
It

was

for the

first

time
theory,

PREDICTED by
was actually

this

and, to everyone's surprise,

verified

by experiment,

QUANTITATIVELY as well as QUALITATIVELY (see Chap. XXXIII). Now there was still another
consequence of this theory which could be tested experimentally,
according to Einstein. In order to appreciate
it

we must

say something about spectra.

Everyone probably knows that if you hang a triangular glass prism

in

the sunlight,

band of

different colors,

like a rainbow, will appear on the wall

where

the light strikes after it has come through the prism.

The explanation of

this

phenomenon
p.

*But, see the footnote

on

278.

290

is

quite simple,

as

may be seen

from the diagram:

When
it

beam

of white light,

SD,

strikes the prism

ABC,

does

the

NOT continue in SAME direction, D,

is

but

bent.*

Furthermore,
if it is

"composite"

light,
light,

like sunlight or

any other white

of

which

is

composed

light of different colors

(or different wave-lengths),

*This
is

bending of

a light ray

called "refraction,"
to

and has nothing

do

with

the bending discussed in Ch. XXXIII.

The reader may look up "refraction" in any book on elementary physics.

291

each constituent

bends a different amount; and when these constituents

reach the other side,

BC,

of the prism,

they are bent again,

as shown in the diagram on p. 291, so that, by the time they reach the wall, the colors are all separated out,
as

XY

shown,

the light of longest wave-length,

namely, red, being deflected least. Hence the rainbow-colored spectrum.

Now,
if

obviously, the light from 5

1

is

"monochromatic/
that
is,

light of a

SINGLE

wave-length only,

instead of "composite,"
like sunlight,

we have

instead of a "rainbow/*

a single bright line

on

XY

having a DEFINITE position, since the amount of bending,

as

we

said above,

depends upon the color or wave-length

of the light in question.

Now

may be obtained

such monochromatic light from

the incandescent vapor of a chemical element

thus sodium, when heated, burns with a light of a certain definite wave-length,
characteristic of

sodium.

292

And
This

similarly (or other elements.

is

explained as follows:

The atoms of each element

vibrate with a certain

DEFINITE period

of vibration,

characteristic of that substance,

and, in vibrating, cause a disturbance in the medium around

this

it,

disturbance being

a light-wave of definite

wave-length

corresponding to
the period of vibration, thus giving rise to
a

DEFINITE color
is

which
i

visible in

DEFINITE
so,

position in the spectrum.

And

if you look at a spectrum you can tell from the bright

lines in

it

just

what substances

are present at 5.

Now

then,

according to Einstein, since each atom has

a definite period of vibration,
it is

a sort of natural clock

as
c/s.

and should serve

a

measure

for
c/s

an "interval"

Thus take

be the interval between the beginning and end of one and eft the time this takes,
to or the "period" of vibration;

vibration,

then, using space coordinates

such that

dr=

S=

d<t>

= 0,

293

that

is,

the coordinates of an observer

for

whom

the atom

is

vibrating at

the origin of his space coordinates (in other words,

an observer traveling with the atom),

equation (71) becomes

<tf

= yd?
1

or

c/s=

Vydt,

where

7 =

2/n ---

(see p. 253).

Now,
if

is

an atom of, say, sodium vibrating near the sun,

we
for

should have to substitute

and

the mass and radius of the sun/

and, similarly, if an atom of the substance

vibrating near the earth,

is

and

would then have

to

be

the mass and radius of the earth, and so on:

Thus

7 DEPENDS upon
is

the location of the atom.

But since ds

the space-time interval between the beginning and end of a vibration,

as

judged by an observer
consequently
of the location

traveling with the atom,

c/s is

INDEPENDENT
of the atom/ then, since

294

obviously

eft

would have

to

be

DEPENDENT UPON THE LOCATION.

Thus,

though sodium from a source

in a laboratory gives rise to a line in a definite part of the spectrum, on the other hand,

sodium on the sun, which, according to the above reasoning, would have a

DIFFERENT DIFFERENT DIFFERENT

period of vibration,
light of a

and hence would emit

wave-length,
a bright line in a

would then give

part of the spectrum

from that ordinarily due to sodium.

And now

let us

see
a

HOW MUCH of
and whether
it is

change

in

the period of vibration is predicted by the Einstein theory

borne out

by

the

facts:

If Jt and dt' represent the periods of vibration near the sun and the earth,

respectively,

then

or

295

Now

Tearth is

very nearly
1

hence
cfc

&

V^un

J,

2m R

_ ~
<(

!l

m* _
f

.* m
7?

Or, using the values of and R given on page 286,

qet get

we

This result implies that

an atom of a given substance should have a

slightly

LONGER
is
is

when when

it it

period of vibration near the sun than

near the earth,

and hence a
slightly

LONGER

wave-length

and therefore
its

lines

SHIFTED

should be a little toward the

of the spectrum (see p. 292).

RED end

This was a most unexpected result! and since the amount of shift

was so

slight,

^Neglecting higher powers of A

since

is

very small

(see the values of

m and R

on

p.

286).

296

it

made

the experimental verification

very

difficult.

For several years after

Einstein

announced

this result

(1917)

experimental observations on this point

were doubtful, and this caused many physicists to doubt the validity of the
Einstein theory, in spite of its other triumphs, which we have already discussed.

BUT FIN ALLY,

in

1927,

the very careful measurements

made by Evershed
definitely settled the issue

IN

FAVOR OF THE
Adams
known
as

EINSTEIN THEORY.

similar

Furthermore, experiments were performed

by W.
on the

S.

star

the companion to Sirius, which has a relatively

LARGE MASS
thus

and

SMALL

RADIUS,

making the ratio

S7eft

+r

much
in

larger than

the case of the sun (see p. 296) and therefore easier to observe

experimentally. Here too

the verdict was definitely

IN

FAVOR OF THE
that

EINSTEIN THEORY!

So

to-day

297

all

physicists are agreed

that the Einstein theory

marks a definite step forward

for:

(1) IT

EXPLAINED PREVIOUSLY KNOWN FACTS

MORE ADEQUATELY THAN

PREVIOUS THEORIES DID
(2) IT

(seep. 103).

EXPLAINED FACTS

NOT EXPLAINED AT ALL

BY PREVIOUS THEORIES
such
(a)
as:

The Michelson-Morley experiment,*

(b) the shift in the perihelion of Mercury (see Ch. XXXI),

(c) the increase in mass of

an electron
(3) IT

when

in

motion. f

PREDICTED FACTS
The bending of a

NOT PREVIOUSLY KNOWN AT ALLlight ray passing near the sun (see Ch. XXXII). (b) The shift of lines in the spectrum (see p. 296).

(a)

when

(c)

The identity of mass and energy,

which/
in turn,

led to the

ATOMIC BOMB!
ff.)

(Seep. 318

And
by
and

all this

using

VERY FEW
*See Part

I,

"The Special Theory.

11

tSee Chap. VIII.

298

VERY REASONABLE
hypotheses (see p. 97), in the slightest degree "far-fetched" or "forced."
not

And

what greater service

can any physical theory render than this

!

We
has

trust that

the reader
this little

been led by

book

to have a sufficient insight into the issues involved,

and to appreciate
the great breadth and

fundamental importance of

THE EINSTEIN THEORY OF RELATIVITY!

XXXV. SUMMARY.
I.

In
it

the

SPECIAL

Relativity

Theory

was shown that two different observers, may, under certain

SPECIAL

conditions,

study the universe from their

different points of

view

and yet obtain

the
II.

In

SAME LAWS and the SAME the GENERAL Theory,

result

FACTS.

this

democratic hold also for

was found to

ANY two observers,

without regard to the special conditions mentioned
in
I.

299

III.

To accomplish

this

Einstein introduced the

PRINCIPLE
by which
the idea of a

OF EQUIVALENCE,
FORCE OF GRAVITY

was replaced by
the idea of the

CURVATURE OF

SPACE.

IV. The study of this curvature required the machinery of

the

TENSOR CALCULUS,

by means of which
the

CURVATURE TENSOR

was derived.

V.

This led immediately to

the

NEW LAW OF GRAVITATION

THREE CRUCIAL

which was tested by

the

PHENOMENA

and found to work

VI.

beautifully!

And READ AGAIN

pages 298 and 299
1

300

THE

MORAL

THE
Since man has been
so successful

MORAL

can

we

in science, not learn from

THE SCIENTIFIC
and

WAY OF
is

THINKING,
of,

what the human mind

HOW
is

capable

it

achieves

SUCCESS:
in science.

There

NOTHING ABSOLUTE

Absolute space and absolute time have been shown to be myths.

We

must replace these old ideas

concepts.

by more human,

OBSERVATIONAL

303

II.

But what

we observe

is

profoundly influenced by the state of the observer,

and therefore
various observers get widely different results

even

in their

measurements of

time and length!

Ill

However,
in spite of these differences, various observers may still

study the universe

WITH EQUAL RIGHT

AND EQUAL
the

SUCCESS,

and CAN AGREE on what are to be called

LAWS

of the universe.

305

IV. To accomplish

this

we need

MORE MATHEMATICS THAN EVER BEFORE,

MODERN, STREAMLINED, POWERFUL
MATHEMATICS.

307

V. Thus

a combination of

PRACTICAL REALISM (OBSERVATIONALISM)

and

IDEALISM (MATHEMATICS),

TOGETHER
have achieved SUCCESS.
VI.

And
knowing
that the laws are

MAN-MADE,
we know
that

they are subject to change

and we are thus

PREPARED FOR CHANGE,

But these changes
are
in

science

NOT
the

made

WANTONLY,

BUT CAREFULLY
by BEST

AND CAUTIOUSLY

MINDS and HONEST HEARTS, and not by any casual child who
thinks that

the world

may be changed

as easily

as rolling off a log.

309

WOULD YOU
ARE DERIVED:

LIKE

TO KNOW?
(20)

HOW THE EQUATIONS

ON PAGE

61

x
.'.

= = =

a cosfl
(x'

(x'

fc)

cos0

y' tanfl)

cos0

x'

cos?

y' sinf?.
\f

y^c-fc/^
f

COSC^

^+
}

sinfl

cosO

+
i

(*

~
a

/ = -+

x smO

'

co$0

/.

x' sinfl

+ / cos0.

HOW

THE

FAMOUS MAXWELL

EQUATIONS LOOK:

X Y Z represent the
f f

components

of the

ELECTRIC FORCE

at a point x , y , z in an electromagnetic field/

at a

given instant,

t,

M N represent the components

,

of the
at the

MAGNETIC FORCE

at

same point and the same instant.

311

III.

HOW TO JUDGE WHETHER A SET OF QUANTITIES

IS

We

TENSOR 9R NOT:

may apply

various criteria:

(1) See if it satisfies any of the definitions of tensors of

various character

and rank

given

in (1 6), (1 7), (1 8), etc.,

or in (30), (31), etc. or in (32), etc.

Or
(2) See
of
if it is

the

sum, difference, or product

two

tensors.

Or
(3) See if it satisfies the following theorem:

A QUANTITY WHICH ON INNER MULTIPLICATION

BY ANY COVARIANT VECTOR (OR ANY CONTRAVARIANT VECTOR) ALWAYS GIVES A TENSOR, IS ITSELF A TENSOR.
theorem may be quite easily proved
This
as follows:

Given
for

that

XA a

is

known

to

be

a contravariant vector,

any choice
that

of

the covariant vector

Aa /

To prove

Now
it

since

X XA a

is

a tensor:

is

a contravariant vector,

must obey (16), thus:

(X Aft) =

(XA a );
312

but

A' s

X y

hence, by substitution,
yr Af _ A
/A/9

dx,;

dXfl
'

dx y ox a

/i/sA

Ary

or
j/ S\0

/y/ VA

3x5 3x^ y\ A I "^ 6xT 3xa

TV~~

_ ^ v,

But

/A^

does not have to be zero,

hence

XI _
which

C'X^ ~

'

C/X0

ax T oxa

y A

satisfies (1 7),

thus proving that

X must be a CONTRAVARIANT TENSOR OF RANK TWO.

And
Thus if XA = B, y then X must be a tensor of the form /
similarly for other cases:

and
then

if

XA* =
Barp
a

C^

rp /

X must

be
,

a tensor of

the form

and so on.

Now
is

let us

show

that

the set of

little g's in

(42)

a tensor:

313

SCALAR i.e.

Knowing

that

c/s

is

TENSOR OF RANK ZERO

member
(1

(see p. 128), then

the right-hand

of (42)
1

is

also

TENSOR OF RANK ZERO/

c/x, is,

but

A CONTRAVARIANT
hence,
,

by

5) on p.

52,

VECTOR,

A COVARIANT
And, again, since C/XM is

by the theorem on page 312, gMI dx^ must be

TENSOR

OF RANK ONE.
a contravariant vector,

then, by the same theorem,

A COVARIANT
OF RANK TWO,
and therefore
it is

g^ must be

TENSOR

appropriate to write

it

with
as

TWO SUBscripts
of

we have been doing

in anticipation

this proof.

314

IV.

WHY MASS CAN

KILOMETERS:

BE

EXPRESSED IN

The reader may be

surprised to

see the mass expressed in kilometers! But it may seem more reasonable"
from the following considerations: In order to decide in what units
a quantity

we must
in

is expressed 14 consider its "dimensionality terms of the fundamental units of

Mass, Length, and Time: Thus the "dimensionality" of

a velocity
is

L/T;

the "dimensionality" of 2 an acceleration is i/7 ;

and so on. Now, in Newtonian physics, the force of attraction which

the sun exerts

being

=
is

upon the
2

earth

where

fcmm'/f (see p. 219), the mass of the sun,

the mass of the earth, and r the distance between them;

and
j

also,

j,

being the centripetal acceleration of the earth toward the sun

(another one of the fundamental laws of Newtonian mechanics);

hence
fcmm'
2

_ m ,. j

or

m = v r2-

i.

Therefore,

315

the "dimensionality" of
,, L

is

= V

'r

since a constant, like V, has no "dimensionality/*

And now
if

we

the time

take as a unit of time/ it takes light to go

a distance of

one kilometer/
n
of time

and
M
a

call this unit

kilometer

(thus

300/000 kilometers would equal one second/ since

in

light goes 300/000 kilometers one second)/

we may express the "dimensionality" of 3 2 L /T or simply /

then
thus

thus:

we may
in

express
kilometers.

mass also

So

far as

considerations of

"dimensionality" are concerned/ the same result holds true also for
Einsteinian physics.
If

the reader has never before

this

encountered

idea of

"dimensionality"
(which,

by

the way,

is

a very

important tool

in scientific

thinking),

he

enjoy reading a paper on "Dimensional Analysis" by

will

Dr.

A. N. Lowan,

published by the Galois Institute of Mathematics

of

Long

Island University in

Brooklyn, N. Y.

316

V,

HOW

Gu LOOKS

IN FULL:

13,3)131,31

If

this

mathematics

BORES you

BE SURE

TO READ
317

PAGES

318-323!

THE ATOMIC

BOMB
is:

We

saw on

p.

78

that the

energy

which a body has when

fo

at rest,

= me

tells us

Thus, the Theory of Relativity not only that

mass and energy are one and the same

but that, even though they are the same still/ what we consider to be

even a SMALL MASS ENORMOUS when translated

is

into

ENERGY

terms,

so that a mass as tiny as an atom has a tremendous amount of energy

the multiplying factor being c , the square of the velocity of light!
2

How

to get at this great storehouse of

energy locked up in atoms and use it to heat our homes, to drive our cars and planes, and so on and so on?

Now,

so long as

is

constant,

as for elastic collision, fo will also remain unchanged.

m,

But, for inelastic collision, and therefore Fo , will change;

this
is

and
the

the situation
IS

when

SPLIT UP, for then sum of the masses of the parts is LESS than the mass of the original atom. Thus, if one could split atoms,
the resulting loss of mass would release a tremendous amount of energy! And so various methods were devised

AN ATOM

by

scientists like

Meitner, Frisch,

Fermi, and others

318

to
It

"bombard
was
Finally

11

atoms.

shown

that

when

Uranium atoms were bombarded with neutrons* these atoms split up ("fission")
into two nearly equal parts, whose combined mass is less than the mass of the uranium atom itself,
this loss in

mass being equivalent,

as the Einstein formula shows,

to a

tremendous amount of energy,

thus released

by

the

fission!

When

Einstein

warned President Roosevelt

terrific

that such experiments might lead to

the acquisition of

new

sources of
race,

power

by

the

ENEMY of the

human

the President naturally saw the importance of having these experiments conducted

where there was some hope that they would be used to END the war and to

PREVENT
by

future wars

instead of

those

who

set out to

take over the earth for themselves alone!

Thus the

ATOMIC BOMB
in

was born

the U.S.A.

And now
has

that a practical

method

of releasing this energy

been developed,

the

MORAL

is

obvious:
it

We MUST
scientific

realize that

has

become

too dangerous to fool around with

GADGETS,
is

WITHOUT UNDERSTANDING in the MORALITY which

*

Read about these amazing experiments in "Why Smash Atoms?" by A. K. Solomon,

Harvard University Press, 1940.

319

Science, Art, Mathematics

SAM,

for short.

These are

NOT

mere

idle words.

We
and

must

ROOT OUT the

DOCTRINE
is

AND DANGEROUS that SAM amoral

FALSE
is

indifferent to

Good

and

Evil.

We

must

SERIOUSLY EXAMINE SAM FROM THIS VIEWPOINT.*

Religion has offered us
a Morality,

lc

wise guys refused to take it

but

many

have

seriously,

and have distorted

meaning!

its

And

now, we are
is

getting

ANOTHER CHANCE-

SAM
we

now

also

warning us that

MUST UNDERSTAND
is

the

MORALITY
for

which

HE now offering us. And he will not stand

our failure to accept
a source of gadgets!
it,

by regarding him merely

as

for

Even using the atomic energy "peaceful" pursuits,

*See our book

"The Education of
of this vital point.

IC

Mits"

for a further discussion

321

like heating the furnaces our homes/

IS

in

and

NOT ENOUGH, will NOT satisfy SAM.

is

For he

desperately trying
his

to prevent us from

merely picking

pockets

to get at the gadgets in them/ and is begging us to see

the

Good/

the True, and

the Beautiful

which are

in his

mind and

heart,

And, moreover,
he
is

giving
clear

new and

meanings to
1

these fine old ideas

which even the sceptical 11 "wise guys

will find irresistible.

So

DO NOT
or

BE

AN
you

ANTI-SAMITE,

SAM

will get

with his

atomic bombs,
his cyclotrons/

and

all his

new

whatnots.

He
if

is

so anxious to

HELP

us

only

we would
IT

listen

BEFORE

IS

TOO

LATE!

323

SOME

INTERESTING READING:

(1) "The Principle of Relativity" by Albert Einstein and Published by Methuen and Company, Others.

London.
(2) The
original

paper

ment:

Philosophical Magazine, Vol.

on the Michelson-Morley experi24 (1887).

D.
Carmichael.

(3) "The Theory of Relativity" by R. Pub. by John Wiley & Sons., N. Y.

(4) "The Mathematical Theory of Relativity" by A. Eddington, Cambridge University Press (1930).
(5) "Relativity" by Albert Einstein. Smith, N. Y. (1931).
(6)

S.

Published by Peter

"An

Introduction

to

the Theory
E. P.

of

L. Bolton.

Pub. by

Dutton

&

Relativity" Co., N. Y.

by

(7) Articles in the Enc. Brit., 14" ed., on: "Aberration of Light" by A. S. Eddington, and "Relativity" by
J.

Jeans.

(8) "Relativity Thermodynamics and Cosmology" by Pub. by Clarendon Press, Oxford. R, C. Tolman. (9) "The
Civita.

Absolute

Differential

Pub. by Blackie

&

Calculus" by Son, London.

T.

Levi-

(10) "Calculus of Variations" by G.

Court Pub. Co., Chicago.

A.

Bliss.

Open
Einstein.

(11) "The

Meaning of Relativity" Princeton University Press, 1945

by Albert

(12) "The Law of Gravitation in Relativity" by Levinson and Zeisler. Pub. by Univ. of Chicago Press.

324