SYMMETRY

By

HERMANN WEYL

,?i

LT*

E
:.-.::...

-----

....

_--_....

YMMETR

J3$?

ifr-,

Ti~~^

95-oo

f7y

SYMMETRY
Symmetry

is

bonss

one of the ideas by which man

PRESTON POLYTECHNIC

through the ages has tried to comprehend

and create order, beauty, and perfection.

somewhat vague notion

symmetry = harmony of proportions, this
book gradually develops first the geometric
concept of symmetry in its several forms
Starting from the

LIBR^^SuLEARNING RESOURCES SERVICE

This book must be returned drt of before the date last

of

as

bilateral,

translator^,

rotational,

orna-

mental and crystallographic symmetry, and

finally rises to the general abstract mathematical idea underlying all these special

-5. jun.

.41 DEC 1989

forms.

Professor

Weyl on

the one

hand displays

the great variety of applications of the prin-

symmetry in the arts, in inorganic

and organic nature, and on the other hand
ciple of

he

clarifies step

\\.

by step the philosophical-

mathematical significance of the idea of

symmetry. The latter purpose makes it
necessary for the reader to confront the

notions and theories of symmetry and relativity,

while a wealth of illustrations sup-

porting the text help to accomplish the

former.

This book is semi-popular in character.

does not shun mathematics, but detailed
treatment of most of the problems it deals
with is beyond its scope.
It

The

late

Professor

Weyl was world

famous for his contributions to mathematics and the philosophy of science. In

this book his penetrating mathematical
insight illumines and transforms the worlds
of nature and of art.

m
PI

1995

FASF RPTIIRN TO PAMDIK TNinTTflTPD

30107

001 374 963

stamped

SYMMETRY

01788
701.17

SYMMETRY
BY HERMANN WEYL

PRINCETON UNIVERSITY
PRINCETON,

NEW

JERSEY

PRESS, 1952

5"Z0O6g"S0

5bo uj

^u?

PREFACE
AND BIBLIOGRAPHICAL REMARKS
Starting from the somewhat vague notion
of symmetry = harmony of proportions,
these four lectures gradually develop

geometric concept of symmetry in

forms,

as

bilateral,

translatory,

first

its

the

several

rotational,

ornamental and crystallographic symmetry,

etc., and finally rise to the general idea underlying all these special forms, namely that of
invariance

of a

configuration

of elements

under a group of automorphic transformations.

I aim at two things: on the one hand
to display the great variety of applications

of the principle of

symmetry

in the arts, in

inorganic and organic nature, on the other

hand to clarify step by step the philosophico-

mathematical significance of the idea of symThe latter purpose makes it necesmetry.

sary to confront the notions and theories of
symmetry and relativity, while numerous
illustrations

supporting

the

text

help

to

accomplish the former.

As readers of this book I had a wider circle

mind than that of learned specialists. It
does not shun mathematics (that would defeat
in

its

purpose), but detailed treatment of most

it deals with, in particular

of the problems

complete mathematical treatment,

its

scope.

To

is

beyond

the lectures, which reproduce

modified version the Louis Clark

Lectures given by the author at
Princeton University in February 1951, two

in slightly

Vanuxem
Copyright 1952, by Princeton University Press

Second Printing i960

Third Printing 1962

appendices containing mathematical proofs

have been added.

Other books

Fourth Printing 1965

F.
Fifth Printing 196(3

Printed in the United States of America

M.

in the field, as for instance

Jaeger's classical Lectures on the principle

of symmetry and

its

applications in natural science

CONTENTS

(Amsterdam and London, 1917), or the

much smaller and more recent booklet by
Jacque Nicolle, La symetrie et ses applications
Albin Michel, 1950) cover only part

(Paris,

of the material, though in a

more detailed
Symmetry is but a side-issue in
D'Arcy Thompson's magnificent work On
growth and form (New edition, Cambridge,
Engl.,
and New York, 1948). Andreas

Translatory, rotational, and related

Speiser's

Ornamental symmetry

fashion.

Ordnung

Theorie

Gruppen

der

von

endlicher

1937) and other

same author are importhe synopsis of the aesthetic and

(3.

Aufl.

mathematical aspects of the subject. Jay

Hambidge's Dynamic symmetry (Yale University Press, 1920) has little more than the
name in common with the present book. Its
closest relative

is

symmetry

perhaps the July 1949 num-

ber on symmetry of the

Studium Generale (Vol.

2,

German

Crystals.

The

complete

tions

To

list

general mathematical

pp. 203-278: quoted

of sources for the illustra-

to be found at the end of the book.

the Princeton University Press and its
I

wish to express

warm

thanks for the

inward and outward care they have lavished

on this little volume; to the authorities of
Princeton University no less sincere thanks
for the

opportunity they gave

me

to deliver

swan song on the eve of my retirement

from the Institute for Advanced Study.
this

Hermann Weyl
Zurich

December 1951

119

APPENDICES
A. Determination of

all finite

groups

of proper rotations in 3-space

149

periodical

is

editors

83

idea of symmetry

B. Inclusion of improper rotations

155

Acknowledgments

157

Index

,61

as Studium Generale).

41

symmetries

Berlin,

publications by the
tant for

Bilateral

BILATERAL SYMMETRY

BILATERAL SYMMETRY
If

am not mistaken

word symmetry is
in two meansymmetric means

the

used in our everyday language

In

ings.

one sense

the

something

like

well-proportioned, well-bal-

anced, and symmetry denotes that sort of con-

cordance of several parts by which they integrate into a whole.

Beauty

bound up with

is

who wrote a
book on proportion and whom the ancients
praised for the harmonious perfection of his
sculptures, uses the word, and Diirer follows
him in setting down a canon of proportions
Thus

symmetry.

for the

human

figure.

by no means

is

In

acoustical

applications.

this sense the idea

restricted to spatial objects;

synonym "harmony"

the
its

Polykleitos,

points

and musical than

Ebenmass

is

more toward
geometric

its

a good

German

equivalent for the Greek symmetry; for like

this
1

carries also the connotation of

it

"middle

Diirer, Vier Biicher von menschlicher Proportion, 1528.

To

be exact, Diirer himself does not use the word

symmetry, but the "authorized" Latin translation by
his friend Joachim Camerarius (1532) bears the title

De

symmetria parlium.

To

ascribed (^epl fie\oiroiiK<i>v,

of a great

Polykleitos the statement

iv, 2)

that "the

many numbers would

is

employment

almost engender

correctness in sculpture." See also Herbert Senk,

Au

de l'expression ovunirpia dans Diodore i, 98,

5-9, in Chronique d'EgypU 26 (1951), pp. 63-66.
Vitruvius defines: "Symmetry results from proportion
Proportion is the commensuration of the
various constituent parts with the whole."
For a
sujet

more elaborate modern attempt

in

the

same

direction

George David Birkhoff, Aesthetic measure, Cambridge, Mass., Harvard University Press 1933, and the
lectures by the same author on "A mathematical

see

theory of aesthetics

and music,"

and

its

applications to poetry

Rice Institute Pamphlet, Vol.

1932), pp. 189-342.

19 (July,

measure,"

mean

the

which

toward

the

virtuous should strive in their actions accord-

Nicomachean

ing to Aristotle's

as

and

Ethics,

De temper amentis describes

that state of mind which is equally removed

which Galen

in

upon

of space

itself,

p-> />',

S:

the arbitrary point p into this

p' with respect to E.

its

that carries

mirror image

mapping

is

defined

which every
p'
Anwith
an
image
associated
point p is
around
a
perpenrotation
example:
a
other
dicular axis, say by 30, carries each point p

whenever a rule

is

established by

from both extremes: ovp-ntTpov

oirtp

knarkpov

Tcbv anpoiv airkxtt-

The image

the

of

balance

provides

natural link to the second sense in which the

word symmetry

is

bilateral symmetry,

right,

used in modern times:

the

symmetry

of

and

left

is

so conspicuous in the structure

of the higher animals, especially the

human

body.

Now

strictly

geometric and, in contrast to the

this

bilateral

symmetry

is

vague notion of symmetry discussed before,

an absolutely precise concept. A body, a
spatial configuration, is symmetric with respect to a given plane

carried into

if it is

by reflection in E. Take any line / perpendicular to E and any point p on /: there

itself

one and only one point p' on / which

has the same distance from E but lies on the
exists

other side.
if/? is

on E.

The

mapping.
around an

around

appears thus as the

itself by
symmetry

carried into

if it is

Bilateral

/.

first

concept of symmetry

case of a geometric

that

to

such

rotations.

Be-

refers

operations as reflections or

cause of their complete rotational symmetry,

the circle in the plane, the sphere in space

were considered by the Pythagoreans the

most perfect geometric figures, and Aristotle
ascribed spherical shape to the celestial bodies

because any other would detract from their

heavenly perfection. It is in this tradition

2
that a modern poet addresses the Divine
Being as

"Thou

great symmetry":

point p' coincides with p only

Reflection in

E is

that

God, Thou great symmetry,

mapping

Who

put a biting

me

lust in

From whence my sorrows

For

all the frittered

That I have
p-

define

spent in shapeless

wide or as narrow as you

meaning, is one idea by which

its

create order, beauty,

The

ways

as

through the ages has tried

FIG.

spring,

days

Give me one perfect thing.

Symmetry,

and

to

First

metry

some

in

comprehend and

will discuss bilateral

detail

may
man

perfection.

course these lectures will take

follows.

and

its

is

as

sym-

role in art as

Reflection

in B,

Anna Wickham,

quarry,

defines a

figure has rotational symmetry

axis

rotations

all

which

and thus

of space into a point p'

"Envoi," from The contemplative

Harcourt, Brace and Co., 1921.

well as organic

we

and inorganic nature.

Then

concept gradually,

shall generalize this

in the direction indicated

by our example of

rotational symmetry,

staying within the

first

confines of geometry, but then going beyond

these limits through the process of

mathe-

matical abstraction along a road that will

lead us to a mathematical idea of

finally

great generality, the Platonic idea as

behind

symmetry.

plications of
this

scheme

edge:

We

is

To

were

a certain degree

typical for all theoretic

knowl-

begin with some general but vague

principle (symmetry in the

find

it

and ap-

the special appearances

all

first

sense), then

an important case where we can give

that notion a concrete precise

meaning

(bi-

symmetry), and from that case we

lateral

gradually

again to generality, guided

mathematical construction and
abstraction than by the mirages of philosophy;

more

rise

by

and if we are lucky we end up with an idea

no less universal than the one from which we
started.

Gone may be much of its emotional

it has the same or even greater

appeal, but

unifying power in the realm of thought and

is

exact instead of vague.

I open
the discussion on bilateral symmetry by using this noble Greek sculpture
from the fourth century b.c, the statue of a

praying boy (Fig. 2), to let you feel as in a

symbol the great significance of this type of
symmetry both for life and art. One may ask
whether the aesthetic value of symmetry depends on its vital value: Did the artist discover the symmetry with which nature according to some inherent law has endowed
its

creatures,

what

nature

realizations;

and then copied and perfected

presented

but

in

imperfect

or has the aesthetic value of

symmetry an independent source?

am

inFIG. 2

clined to think with Plato that the mathe-

matical idea

is

the

common

origin of both:

the mathematical laws governing nature are

the origin of

symmetry

in

nature, the in-

tuitive realization of the idea in the creative

artist's

mind

its

origin in art; although

am

ready to admit that in the arts the fact of the

bilateral

symmetry

of the

human body

in

its

outward appearance has acted as an additional stimulus.

Of all

ancient peoples the Sumerians seem

have been particularly fond of strict bilateral or heraldic symmetry.
A typical design on the famous silver vase of King Entemena, who ruled in the city of Lagash
to

FIG. 4

around 2700

B.C.,

with spread wings

FIG. 3

MBMBga

tiS8BB8EBBEgBfflEESBM jagBSBSBBB

shows a lion-headed eagle

each of whose claws

en face,

grips a stag in side view,

frontally attacked

by a

which

in

its

upper design are replaced by goats

lower) (Fig.

turn

is

lion (the stags in the

in the

Extension of the exact sym-

3).

metry of the eagle to the other beasts obviously enforces their duplication.
later the eagle

is

Not much

given two heads facing in

symoverwhelming the
imitative principle of truth to nature.
This
heraldic design can then be followed to
Persia, Syria, later to Byzantium, and anyone

either direction, the formal principle of

metry

who

thus

lived

remember

completely

before the First

World War

will

the double-headed eagle in the

coats-of-arms

of

Russia

Czarist

and

the

Austro- Hungarian monarchy.

Look now at this Sumerian picture (Fig. 4).

The two eagle-headed men are nearly but
not quite symmetric; why not? In plane
geometry reflection

in a vertical line

can

also be brought about by rotating the plane

in

space around the axis

by 180.

If

you

look at their arms you would say these two

monsters arise from each other by such rotation; the overlappings depicting their position in space prevent the plane picture

from

having bilateral symmetry. Yet the artist

aimed at that symmetry by giving both
figures a half turn toward the observer and

by the arrangement of feet and wings:

the drooping wing is the right one in the
also

left figure,

the

left

one in the right

figure.

FIG.

FIG- 7

legs.

In

analogy

FIG

Christian
in

certain

Eucharist as on

The

designs on

seal stones are

symmetry.

the cylindrical Babylonian

frequendy ruled by heraldic

remember

seeing in the collec-

my

former colleague, the late Ernst

Herzfeld, samples where for symmetry's sake
not the head, but the lower bull-shaped part

tion of

of a

god's body,

doubled and

10

rendered in

was
two hind

profile,

given four instead of

this

may

see an

representations

of the

times one

Byzantine paten

(Fig.

5), where two symmetric Christs are facing

the disciples. But here symmetry is not
complete and has clearly more than formal
significance, for Christ on one side breaks
the bread, on the other pours the wine.
Between Sumeria and Byzantium let me
insert Persia: These enameled sphinxes (Fig.
6) are from Darius' palace in Susa built in

11

the days of

Who

B.C.

Crossing the Aegean

Marathon.

we find these floor

Megaron in Tiryns,

patterns (Fig. 7) at the

about 1200

late helladic

believes strongly in historic con-

and dependence will trace the graceof marine life, dolphin and
octopus, back to the Minoan culture of Crete,
the heraldic symmetry to oriental, in the
tinuity
ful

designs

last

instance Sumerian, influence.

thousands of years
ences at

work

we

in this

still

see the

plaque

(Fig. 8)

dome

altar

enclosure

Italy,

eleventh century a.d.

in

the

drinking from a pine well

Skipping

same

influ-

from the

of Torcello,

The peacocks
among vine leaves

an ancient Christian symbol of immortality, the structural heraldic symmetry is

are

FIG. 9

oriental.

For
FIG. 8

occidental

in contrast to the orient,

art, like life itself, is inclined to mitigate, to

even to break strict symBut seldom is asymmetry merely the

absence of symmetry. Even in asymmetric
designs one feels symmetry as the norm from

loosen, to modify,

metry.

which one deviates under the influence

of

think the

non-formal character.
from the famous Etruscan Tomb of the
Triclinium at Corneto (Fig. 9) provide a good
example.
I have already mentioned repreforces of

riders

sentations of the Eucharist with Christ dupli-

The
cated handing out bread and wine.
central group, Mary flanked by two angels,
in this

10)

in

mosaic of the Lord's Ascension (Fig.

the cathedral at Monreale, Sicily

has almost perfect sym[The band ornament's above and

below the mosaic will demand our attention
The principle of
in the second lecture.]
symmetry is somewhat less strictly observed
(twelfth century),

metry.

in

an

earlier

mosaic from San Apollinare

in

13

Mother,

may

to the left

also

Evangelist

think

on both

crucifixions as

FIG.

10

Ravenna

(Fig.
11), showing Christ surrounded by an angelic guard of honor. For
instance Mary in the Monreale mosaic raises
both hands symmetrically, in the orans gesture; here only the right hands are raised.

Asymmetry has made

next

picture

(Fig.

further inroads in the

12),

a Byzantine relief

ikon from San Marco, Venice.

It

is

a Deesis,

and, of course, the two figures praying for

mercy as the Lord is about to pronounce the

last judgment cannot be mirror images of
each other;

14

for to the right stands his

Virgin

John the

of

Baptist.

of the

11

FIG.

12

You

Mary and John

sides

FIG.

cross

the
in

examples of broken symmetry.

we touch ground here where the

geometric notion of bilateral symmetry begins to dissolve into the vague notion
of Ausgewogenheit, balanced design with which
Clearly

precise

"Symmetry," says Dagobert

On the Problem of Symmetry in
3
and binding, asymmetry
rest
"signifies
Art,
motion and loosening, the one order and law,
the other arbitrariness and accident, the one

we

started.

Frey

in

an

article

formal rigidity and constraint, the other life,

play and freedom." Wherever God or Christ
everlasting

as symbols for

are represented

truth or justice they are given in the sym-

Probably
metric frontal view, not in profile.
similar reasons public buildings and

no inner difference, no polarity

and right, as there is for instance
in the contrast of male and female, or of the
anterior and posterior ends of an animal.
It
requires an arbitrary act of choice to determine what is left and what is right. But
after it is made for one body it is determined
I must try to make this a
for every body.
between

is

left

clearer.

little

In space the distinction of

left

and right concerns the orientation of a screw.

If you speak of turning left you mean that the
sense in which you turn combined with the
upward direction from foot to head of your
body forms a left screw. The daily rotation

for

of the earth together with the direction of

houses of worship, whether they are Greek

axis

temples or Christian basilicas and cathedrals,

It is, however,
are bilaterally symmetric.
true that not infrequently the two towers of
Gothic cathedrals are different, as for instance

But in practically every case

this seems to be due to the history of the
cathedral, namely to the fact that the towers
in Chartres.

built in different periods.

were

It is

under-

standable that a later time was no longer

fied

satis-

with the design of an earlier period; hence

one may speak here of historic asymmetry.

Mirror images occur where there is a mirror,
be it a lake reflecting a landscape or a glass
mirror into which a woman looks. Nature
as well as painters
trust,

examples

The one most

at

it

in

my

make

use of this motif.

will easily

come

to

familiar to me, because

study every day,

is

your mind.
I

look

Hodlcr's Lake

While we are about

to turn

from art

to

nature, let us tarry a few minutes and first

consider what one may call the mathematical
philosophy of
3

left

and

Studium Generale,

right.

p. 276.

To

the scientific

its

North Pole is a left screw,

it is a right screw if you give the axis the
opposite direction. There are certain crystalline substances called optically active which
betray the inner asymmetry of their constitution by turning the polarization plane of
polarized light sent through them either to
the left or to the right; by this, of course, we
mean that the sense in which the plane rotates
from South

to

while the light travels in a definite direction,

combined with that

direction, forms a left

screw (or a right one, as the case

Hence when we
in a

said

may

be).

above and now repeat

terminology due to Leibniz, that

and right are

indiscernible,

we want

left

to express

that the inner structure of space does not

permit

us,

tinguish a
I

except by arbitrary choice, to disleft

from a right screw.

this fundamental notion

wish to make

more

of Silvaplana.

16

mind there

on

still

depends the entire

theory of relativity, which is but another
aspect of symmetry.
According to Euclid
one can describe the structure of space by a
number of basic relations between points,
precise, for

such as

ABC lie

it

on a straight

line,

ABCD

lie

17

AB is congruent CD.

in a plane,

Perhaps the

space
best way of describing the structure of
single
the
by
adopted:
is the one Helmholtz
notion of congruence of figures.

A mapping S

point
of space associates with every point p a
in
s
s''-p*P'>
P air of ma PP S $*
p''-p*P''
of the
p> _>
f which the one is the inverse
p
S'
p'
then
into
other, so that if S carries p

world

is

ture.

They

revealed by the general laws of naare formulated in terms of cer-

tain basic quantities

space and time.

physical

which are functions

in

We would conclude that the

structure

space

of

"contains

screw," to use a suggestive figure of speech,

a
if

FIG.

carries p'

13

back into p and vice versa,

is

spoken

mappings or transtransformation which preserves

of as a pair of one-to-one

formations.

and if we define this

Helmholtz way, that would
mean that it carries any two congruent figures
is called an autointo two congruent ones

the structure of space

structure in the

Leibniz
mathematicians.
morphism by
underlying
idea
the
this
is
recognized that
the

the

concept

geometric

of

similarity.

carries a figure into

An

one that

automorphism
in Leibniz' words is "indiscernible from it if
each of the two figures is considered by itself."
What we mean then by stating that left and
right are of the same essence is the fact that
reflection in

Space
space

is

as such

is

is

also the

currences.

18

a plane

The

studied by geometry.

But

of all physical oc-

structure

of the

physical

were not invariant throughout with

intellectual shock

as a

Ernst

Mach

tells

of the

he received when he learned

boy that a magnetic needle

a certain sense, to the

left

is

deflected in

or to the right,

if

suspended parallel to a wire through which

an

electric current

tion (Fig.

14).

is

sent in a definite direc-

Since the whole geometric

and physical configuration, including the

electric current and the south and north poles
of the

an automorphism.

medium

these laws

respect to reflection.

magnetic needle,

to all appearances,

are symmetric with respect to the plane

through the wire and the needle, the

needle should react like Buridan's ass between

laid

19

equal bundles of hay and refuse to decide

between left and right, just as scales of equal

space and

arms with equal weights neither go down on

their left nor on their right side but stay
But appearances are sometimes
horizontal.
Young Mach's dilemma was the
deceptive.

explicable

result of a too hasty

assumption concerning

the effect of reflection in

on the

electric

current and the positive and negative magnetic poles of the needle: while we know a

how geometric entities fare under

reflection, we have to learn from nature how
And this is
the physical quantities behave.
plane E
in
the
reflection
under
what we find
priori

the electric current preserves

its

the magnetic south and north

Of course

changed.

way

this

establishes the equivalence of

direction, but

poles are inter-

which reand right, is

out,

left

possible only because of the essential equality

and negative magnetism. All

doubts were dispelled when one found that
the magnetism of the needle has its origin in
molecular electric currents circulating around
of positive

the needle's direction;

reflection

in

change the sense

The

net result

clear that under

such

which they

in
is

is

it

plane

the

currents

flow.

that in all physics nothing

left

and

right.

Just as

all

points

and

left

and

right.

Position, direction, left

and

In language tinged

right are relative concepts.

with theology this issue of relativity was discussed at great length in a famous controversy

between Leibniz and Clarke, the latter a

clergyman acting as the spokesman for

Newton. 4
4

See G.

Newton with
W.

his belief in absolute

Leibniz, Pkilosophische Schriften, ed.

Gerhardt (Berlin 1875

scq.),

vu, pp. 352-440, in

particular Leibniz' third letter, 5.

20

proof

would be

it

why matter moves

in

this

in any other direction.

Leibniz
burden God with such decisions

in-

rather

than

is

to

as lack

loath

"sufficient reason."
Says he, "Under the assumption that space be something in itself it
is

impossible to give a reason

why God

should

have put the bodies (without tampering with

their

mutual distances and

just at

this particular

relative positions)

place and not some-

where else; for instance, why He should not

have arranged everything in the opposite
If,
order by turning East and West about.
than
on the other hand, space is nothing more
the spatial order and relation of things then
the two states supposed above, the actual
one and its transposition, are in no way different from each other
and therefore it
.

is

a quite inadmissible question to ask

why

one state was preferred to the other." By

pondering the problem of left and right Kant
was first led to his conception of space and
time as forms of intuition. 5
act of

this: If

God had been

hand then

this

Kant's opinion
the

first

creative

the forming of a

left

hand, even at the time when

could be compared to nothing

else, had
which can
only intuitively but never conceptually be apprehended. Leibniz contradicts: According
to him it would have made no difference if
God had created a "right" hand first rather
than a "left" one. One must follow the
it

directions in space are equivalent, so are

all

arbitrary will, for otherwise

seems to have been

has shown up indicating an intrinsic differ-

ence of

time considers motion a

of the creation of the world out of God's

the distinctive character of

left,

world's creation a step further before a differ-

ence can appear.

5

Had God,

rather than

Besides his "Kritik der rcinen Vernunft" see

es-

pecially 13 of the Prolegomena zu einer jeden kunftigen

Metaphysik.

21

making

and then a right hand,

hand and then formed
hand, He would have changed

first

left

started with a right

another right

the plan of the universe not in the

second

in the

first

but

by bringing forth a hand

act,

which was equally rather than oppositely

oriented to the first-created specimen.
Scientific thinking sides with Leibniz.
Mythical thinking has always taken the conis evinced by its usage of right
symbols for such polar opposites

trary view as

and

left as

good and

as

You need

evil.

only think of the

double meaning of the word

In

right itself.

from Michelangelo's famous Creation of Adam from the Sistine Ceiling (Fig.
15) God's right hand, on the right, touches
this detail

into

life

Adam's

left.

People shake right hands.

Latin word for
of the

But

left,

Sinister

and heraldry

side of the shield as

left

sinistrum

is

at the

its

still

is

the

that

which

and in common English only this

figurative meaning of the Latin word survives. 6
Of the two malefactors who were
one who goes with
on His right. St. Matthew,

crucified with Christ, the

Him

to paradise

is

Chapter 25, describes the

follows:

"And he

judgment

last

as

sheep on his

shall set the

hand but the goats on the left. Then

King say unto them on his right

right

shall the

hand,

Come

herit the

ye,

blessed of

my

Kingdom prepared

the foundation of the world.

shall say also

Father, in-

you from

for
.

Then he

unto them on the

left

hand,

Depart from me, ye cursed, into everlasting

fire, prepared for the devil and his angels."

am

not unaware of the strange fact that as a

terminus technicus in the language of the Roman augurs
I

sinistrum

22

had

just the opposite

meaning of

propitious.

lecture Heinrich Wolfflin

remember

once delivered

in

in paintings";

together with an article on

"The problem

of inversion

Zurich on "Right and

left

(Umkehrung)

Raphael's tapistry cartoons," you

now

find

in
it

printed in abbreviated form in his Gedanken

By a number of exzur Kunstgeschichte, 1941
.

Sistine Madonna and

Rembrandt's etching Landscape with the three

amples,

as

Raphael's

Wolfflin tries to show that right in

painting has another Stimmungswert than left.
trees,

methods of reproduction interand

right, and it seems that former
change left
less sensitive than we are
much
times were
Practically all

toward such inversion.

(Even Rembrandt

did not hesitate to bring his Descent from

the Cross as a converse etching upon the

market.)
6

15

sinister side.

same time

evil,

is

FIG.

speaks

Considering that

we do a

lot

more

reading than the people, say, of the sixteenth

century, this suggests the hypothesis that the

23

to the inversion of

by Wolfflin is connected
with our habit of reading from left to right.
As far as I remember, he himself rejected this
difference pointed out

as well as a

number

of other psychological

explanations put forward in the discussion

after his lecture.

The

printed text concludes

with the remark that the problem "obviously

has deep roots, roots which reach down to
the very foundations of our sensuous nature."
I

for

my

am

part

disinclined to take the

matter that seriously. 7

Leibniz

made

the metaphysical idea of causation,

it

the one

way

which are interchanged by inverting

and with respect to

the direction of time,

and

positive

negative electricity.

especially in the second,

it is

In these cases,
perhaps clearer

than for the pair left-right that a priori evi-

dence

is

not sufficient to settle the question;

have to be consulted.
which past and future
role
be
sure,
the
To
would indicate
our
consciousness
play in
the
past knowable
intrinsic
difference
their
unknown and
future
and unchangeable, the
taken
now and
still alterable by decisions
the empirical

facts

one would

expect that this difference has

basis in the physical laws of nature.

those laws of which

certain
7

Cf.

Bilde,"

we can

its

But

boast a reasonably

knowledge are invariant with respect

also

A.

Faistauer,

"Links und rechts im

Amicis, Jahrbuch der

1926, p. 77; Julius

v.

osterreichischen

Galerie,

Schlosser, "Intorno alia lettura

Critica 28, 1930, p. 72; Paul Oppe,

"Right and left in Raphael's cartoons," Journal of
the Warburg and Courtauld Institutes 7, 1944, p. 82.

dei quadri,"

modi

and with

may

character of time,

enter

physics through the statistical interpretation

of those laws in terms of probability

Our

and negative

of equivalence arises with respect to past and

right.

are not altered by letting time flow backward,

face of certain biological facts presently to

future,

and

Even if it is true that the exact

"wave laws" formulated by quantum physics

leaves

The same problem

left

clear that the temporal

the world.

In science the belief in the equivalence of

left and right has been upheld even in the

shocked young Mach.

it

past and future refer to the causal structure of

particles.

be mentioned which seem to suggest their

inequivalence even more strongly than does
the deviation of the magnetic needle which

time as they are with re-

spect to the interchange of

us even

and

present physical knowledge

more uncertain about

the

equivalence or non-equivalence of positive

electricity.

devise physical laws in

It

seems

difficult to

which they are not

intrinsically alike; but the negative counter-

part of the positively charged proton

still

re-

mains to be discovered.
This half-philosophical excursion was
needed as a background for the discussion of
the left-right symmetry in nature; we had
to understand that the general organization
of nature possesses that symmetry.
But one
will not expect that

ture shows
prising

to

it

any

special object of na-

Even

to perfection.

what extent

must be a reason

for this,

seek: a state of equilibrium

so, it is sur-

prevails.

it

and
is

it is

There

not far

likely to

to

be sym-

metric.
More precisely, under conditions
which determine a unique state of equilibrium the symmetry of the conditions must
carry over to the state of equilibrium. Therefore tennis balls and stars are spheres; the
earth would be a sphere too if it did not rotate
around an axis. The rotation flattens it at
the poles but the rotational or cylindrical symmetry around its axis is preserved. The fea-

ture

that

needs explanation

not the rotational

symmetry

of

is,

its

therefore,

shape but

25

24

symmetry as exhibited
by the irregular distribution of land and water
and by the minute crinkles of mountains on
the deviations from this

surface.

its

It

is

for

such reasons that in his

monograph on the left-right problem in

zoology Wilhelm Ludwig says hardly a word
about the origin of the bilateral symmetry
prevailing in the animal kingdom from the
echinoderms upward, but in great detail discusses all sorts of secondary asymmetries
superimposed upon the symmetrical ground

plan. 8

quote:

"The human body

of the other vertebrates

is

like that

basically

built

All asymmetries oc-

bilateral-symmetrically.

curring are of secondary character, and the

more important ones

inner

the

affecting

organs are chiefly conditioned by the necessity for

the intestinal tube to increase

its

surface out of proportion to the growth of

the body, which lengthening led to

an asym-

And

metric folding and rolling-up.

in the

course of phylogenetic evolution these

first

asymmetries concerning the intestinal system

appendant organs brought about

asymmetries in other organ systems." It is
well known that the heart of mammals is an
asymmetric screw, as shown by the schematic
drawing of Fig. 16.
If nature were all lawfulness then every
phenomenon would share the full symmetry
with

FIG.

16

its

of the universal laws of nature as formulated

by the theory of
that this

is

relativity.

The mere

not so proves that contingency

essential feature of the world.

fact
is

Clarke in

an
his

controversy with Leibniz admitted the latter's

principle of sufficient reason but
the sufficient reason often
of

God.
8

added that
mere will

the

think, here Leibniz the rationalist

W. Ludwig,

Rechts-links-Problem im

beim Menschen, Berlin 1932.

26

lies in

Tierreick

und

is

definitely

But

track.
to

wrong and Clarke on the right

would have been more sincere

it

deny the principle of

for all that

other

is

unreason

God

and Clarke with

this:

The

al-

responsible

in the world.

hand Leibniz was

of relativity.

reason

sufficient

together instead of making

right against

On the
Newton

his insight into the principle

The

truth as

we

see

it

today

is

laws of nature do not determine

uniquely the one world that actually

exists,

one concedes that two worlds

arising from each other by an automorphic
transformation, i.e., by a transformation
which preserves the universal laws of nature,
are to be considered the same world.
If for a lump of matter the overall symnot even

if

laws of nature

is

limited by nothing but the accident of

its

metry inherent
position

then

the

in

it

assume the form of a

will

sphere around the center P.

Thus the lowest

forms of animals, small creatures suspended

in water,

are

more or

less

spherical.

For

forms fixed to the bottom of the ocean the direction of gravity

rowing the

set of

is

an important factor, nar-

symmetry operations from

all

around the center P to all rotations

about an axis. But for animals capable of
self-motion in water, air, or on land both the
rotations

direction in which their

body moves and the direction of gravity are

postero-anterior

of decisive influence.

After determination of

the antero-posterior, the dorso-ventral,

and

thereby of the left-right axes, only the distinction

arbitrary,

between
and at

left

and

right

remains

no higher symtype can be expected.

this stage

metry than the bilateral

Factors in the phylogenetic evolution that

tend to introduce inheritable differences be-

tween left and right are likely to be held in

check by the advantage an animal derives

27

from the bilateral formation of

motion,

cilia

its organs of
or muscles and limbs: in case of

asymmetric development a screw-wise

instead of a straight-forward motion would
their

This may help to explain

our limbs obey the law of symmetry
more strictly than our inner organs. Aristonaturally result.

why

phanes

in Plato's

story of

how

Symposium

tells

a different

the transition from spherical to

symmetry came about. Originally,
he says, man was round, his back and sides
forming a circle. To humble their pride and
might Zeus cut them into two and had
Apollo turn their faces and genitals around;
and Zeus has threatened, "If they continue
bilateral

insolent
shall

will

hop around on a

The most
in

them again and they

split

single leg."

examples of symmetry
the inorganic world are the crystals.
The
striking

gaseous and the crystalline are two clear-cut

states of

matter which physics finds relatively

a mirror image of the other, like left and

A substance which is optically
right hands.
active, i.e., turns the plane of polarized light

either left or right, can be expected to crystal-

such asymmetric forms. If the laevonature one would assume that

lize in

form
the dextro-form exists likewise, and that in
the average both occur with equal freexists in

quencies.

In 1848 Pasteur

covery that

when

the sodium

made

the dis-

ammonium

salt

of optically inactive racemic acid was recrystallized from an aqueous solution at a lower
temperature the deposit consisted of two

which were mirror

images of each other. They were carefully
separated, and the acids set free from the
one and the other proved to have the same
kinds

of

tiny

crystals

chemical composition as the racemic acid,

but one was optically laevo-active, the other
The latter was found to be
dextro-active.

easy to explain; the states in between these

identical with the tartaric acid

two extremes,

fermenting grapes, the other had never be-

states,

are

and the plastic

amenable to theory.

like the fluid

somewhat

less

In the gaseous state molecules

around

random

in

move

freely

space with mutually independent

and velocities. In the

atoms oscillate about positions of equilibrium as if they were tied to
them by elastic strings. These positions of
equilibrium form a fixed regular configurapositions

crystalline state

tion in space.

how

What we mean by regular and

the visible

symmetry of

crystals derives

from the regular atomic arrangement

will be

fore

been observed in nature.

says

F.

M. Jaeger

principle of symmetry
science,

"has a

in

and

its

his

present in

"Seldom,"

lectures

On

the

applications in natural

scientific discovery

had such

far-

one had."
Quite obviously some accidents hard to
control decide whether at a spot of the solution a laevo- or dextro-crystal comes into
being; and thus in agreement with the sym-

reaching consequences as

this

metric and optically inactive character of

most of the thirty-two geometrically possible

whole and with the law of

amounts
of substance deposited
chance the
form at any moment
and
the
other
in the one

systems of crystal symmetry involve bilateral

of the process of crystallization are equal or

explained in a subsequent lecture.

symmetry, not
is

not involved

so-called

28

and dextro-form, each form being

in a laevo-

all

of

them

we have

enantiomorph

do.

While

Where

it

the possibility of

crystals

which

exist

the solution as a

very nearly equal.

On

the other

hand na-

ture, in giving us the wonderful gift of grapes

so much enjoyed by Noah, produced only one

29

of the forms, and it remained for Pasteur

to
produce the other! This is strange indeed.
It is a fact that most of the
numerous car-

bonic

compounds occur

in nature in one,

either the laevo- or the dextro-form

only.
The sense in which a snail's shell winds is an

inheritable character founded in

constitution, as

winding of the

Homo

sapiens.

is

its

the "left heart"

genetic

and the

duct in the species

This does not exclude that inintestinal

inversions occur, e.g. situs inversus of the intestines of man occurs with a frequency of
about

0.02 per cent;

later
!

of our

we

shall

come back

to that

mere

means

chemical

it is understandable that Pasteur

clung to the opinion that the production of

material, 9

compounds was

single optically active

demarkation that can at present be drawn

between the chemistry of dead and living
Pasteur tried to explain his very
matter."

of

experiment where

first

racemic

acid

action of bacteria in the atmosphere

every one of us.

the dextro-rotatory form of glucose and laevoin

rotatory form of fructose.

tion of this genotypical

A horrid manifesta-

asymmetry

is

a meta-

bolic disease called phenylketonuria, leading

to insanity, that man contracts when a small

quantity of laevo-phenylalanine

added

is

to

his food,

while the dextro-form has no such

disastrous effects.
To the asymmetric chemical constitution of living

organisms one must

attribute the success of Pasteur's method of

isolating the laevo- and dextro-forms of sub-

was

transformed by recrystallization into a mixture of laevo- and dextro-tartaric acid by the

neutral solution.

we have a
is turning the same way
Thus our body contains

the

In 1860 he wrote,
very prerogative of life.
"This is perhaps the only well-marked line

human body shows

that

by

inactive

optically

Also the deeper chemical constitution

screw, a screw that

that he

It

is

on

was wrong; the sober physical

planation

his

quite certain today

ex-

the fact that at lower temper-

lies in

ature a mixture of the two oppositely active

tartaric forms is more stable than the inIf there

active racemic form.

in principle

between

life

is

a difference

and death

it

does not

lie in the chemistry of the material substratum; this has been fairly certain ever since
Wohler in 1828 synthesized urea from purely

But even as late as 1898

a famous lecture on "Stereo-

mineral material.

F. R. Japp in
chemistry and Vitalism" before the British

Association

upheld

view in

Pasteur's

the

by means of the enzymatic action of

and the like. Thus
he found that an originally inactive solution
of some racemate became gradually laevorotatory if Penicillium glaucum was grown in it.

modified form: "Only the living organisms,

or the living intelligence with its conception

Clearly the organism selected for

ing the experiment but to

stances

bacteria, moulds, yeasts,

its

nutri-

ment that form of the tartaric acid molecule

which best suited its own asymmetric chemical constitution.

The image

of lock

and key

has been used to illustrate this specificity of

the action of organisms.

In view of the facts mentioned and in view

30

of the failure of all attempts to "activate"

can produce this result (i.e. asymDoes he really mean

Pasteur's intelligence that, by devis-

of symmetry

metric compounds)."
that

it is

its

own

great sur-

dual tartaric crystals? Japp

continues, "Only asymmetry can beget asymprise, creates the

There

is

known today one

clear instance, the

reaction of nitrocinnaminacid with

circular-polarized light generates

an

bromine where
optically active

substance.

31

The

metry."

truth of that statement

willing to admit; but

am

the combined magnetic fields of earth and

there is no symmetry in the accidental past

and present set-up of the actual world which

sun are of immediate help in this regard.

Another possibility would be to assume that
development actually started from an equal

of

little

begets the future.

There

is

however a

real difficulty:

Why

should nature produce only one of the doublets of so many enantiomorphic forms
the
origin of

which most certainly

lies

in living

enantiomorph forms, but

an unstable equilibrium which
under a slight chance disturbance tumbled

distribution of the

that this

is

over.

organisms?

Pascual Jordan points to this

fact as a support for his opinion that the
beginnings of life are not due to chance events
which, once a certain stage of evolution is

division of the fertilized egg of an animal into

reached, are apt to occur continuously

two

here

now

now

there, but rather to

an event of
quite singular and improbable character,
occurring once by accident and then starting
an avalanche byautocatalytic multiplication.
Indeed had the asymmetric protein molecules
found in plants and animals an independent
origin in

many

places at

many

times, then

their laevo- and dextro-varieties should show

nearly the same abundance. Thus it looks
as if there is some truth in the story of Adam
and Eve, if not for the origin of mankind

then for that of the primordial forms of

It

was in reference

when

value

left

and

right, at least as far as the

constitution of the organic world

But we

life.

to these biological facts

said before that if taken at their face

they suggest an intrinsic difference

between

may be

does not

lie in

is

concerned.

cells fix

on earth from

questions arise: Does the

left

the other for

its

first

so that one

median plane,

the

of the cells contains the potencies for

its left,

Secondly what

right half?

determines the plane of the

first

begin with the second question.

any animal above the protozoa

division?

The egg

of

possesses

the beginning a polar axis connecting

from

what

develops into the animal and the vegetative

This axis together with
poles of the blastula.
the point

where the

fertilizing

spermatozoon

enters the egg determines a plane,

would be quite natural

is

the plane of the

there

is

first

evidence that

to

division.
it is

and

assume that
so in

it

this

And indeed
many cases.

Present opinion seems to incline toward the

assumption that the primary polarity as well

as the subsequent bilateral symmetry come
about by external factors actualizing potenIn

or of the light received

Two

genesis.

tialities

biological laws

the phylogenetic problems of

right let us finally turn to their onto-

any universal

itself,

From
and

sure the answer to our riddle

but in the accidents of the genesis of the

organismic world
Pascual Jordan shows one
way out; one would like to find a less radical
one, for instance by reducing the asymmetry
of the inhabitants on earth to some inherent,
though accidental, asymmetry of the earth

32

But neither the earth's rotation nor

the sun.

help since

it is

inherent in the genetic constitution.

many

axis

is

instances the direction of the polar

obviously determined by the attach-

ment of the oozyte to the wall of the ovary,

and the point of entrance of the fertilizing
sperm is, as we said, at least one, and often
the most decisive, of the determining factors
for the

may

median plane.

But other agencies

be responsible

for the fixation of the

also

33

In the sea-weed Fucus

chemical gradients

that a single blastomere isolated from its

partner in the two-cell stage develops into a

determine the polar axis, and in some insects

and cephalopods the median plane appears to

whole gastrula differing from the normal one

only by its smaller size. Here are Driesch's
famous pictures. It must be admitted that

one and the other.

light or electric fields or

be fixed by ovarian influences before fertilization The underlying constitution on which

.

gists

work

sought by some bioloin an intimate preformed structure, of

these agencies

is

Driesch's disthis is not so for all species.

covery led to the distinction between the actual

and the potential destiny

of the several

which we do not yet have a clear picture.

Thus Conklin has spoken of a spongioplasmic
framework, others of a cytoskeleton, and as
there is now a strong tendency among bio-

parts of an egg.

chemists to reduce structural properties to

former, but shrinks in the course of development. Let me illustrate this basic point by

fibers, so

much

so that Joseph

Needham

Terry Lectures on Order and

his

life

(1936)

is

largely

dares the aphorism that biology

the study of fibers, one

may

in

expect them to

find that that intimate structure of the egg

Driesch himself speaks of

prospective significance (prospektive Bedeutung), as against prospective potency (prospektive Potenz); the latter is wider than the

another example taken from the determinaAccording

tion of limb-buds of amphibia.
to experiments

who

performed by R. G. Harrison,

transplanted discs of the outer wall of

framework of elongated protein

consists of a

molecules or fluid crystals.

We know a little more about our first queswhether the

tion

vides

it

into

left

first

and

mitosis of the cell diright.

fundamental character of
the hypothesis that this

is

Because of the

bilateral

symmetry

so seems plausible

However, the answer cannot be an

Even if the hypothesis should be true for the normal development we know from experiments first performed by Hans Driesch on the sea urchin
enough.

unqualified affirmation.

10

Julian S. Huxley and G. R. de Beer in their

classical

Elements

versity Press,
xrv,

of

embryology

(Cambridge

Uni-

1934) give this formulation (Chapter

Summary,

p. 438):

"In the

earliest stages, the

egg acquires a unitary organization of the gradientfield type in which quantitative differentials of one
or more kinds extend across the substance of the egg
in one or more directions. The constitution of the
egg predetermines it to be able to produce a gradientfield of a particular type; however, the localization
of the gradients

is not predetermined, but

about by agencies external to the egg."

34

is

brought

FIG.

17

Experiments on pluripotence in Echinus.

Normal gastrula and normal pluteus.
ai and bi.
Half-gastrula and half-pluteus, exand b 2
pected by Driesch.
The small but whole gastrula and
a3 and b3.
pluteus, which he actually obtained.
a2

35

body representing the buds of future

the

limbs, the antero-posterior axis

is

determined

parts of

whose nervous system are asym-

metric.

First the fertilized

egg

splits into

axes; thus at this stage the opposites of left and

right still belong to the prospective potencies

/ and a smaller P of obviously different

In the next stage they dinature (Fig. 18).
vide along two perpendicular planes into
/'
P2 respectively. ThereI" and Pi

of the discs,

after the

at

a time when transplantation

invert the dorso-ventral

and

may

still

and the medio-lateral

depends on the influence

of the surrounding tissues in which way this
potency will be actualized.
it

cell

handle Pi

P 2 comes

+P

turns about so that

into contact with either

or/";

one it contacts B, the other A. We

now have a sort of rhomboid and roughly
APi is the antero-posterior axis and PPi the

call the

dorsal-ventral one.

Only

the next division

which along a plane perpendicular

separating

symmetric
that

A and B splits A
halves A = a -f-

which determines

to the

as well as

B =

oc,

left

and

one
into

/3,

right.

is

further slight shift of the configuration destroys

this bilateral

FIG.

symmetry.

The

question arises

whether the direction of the two consecutive

shifts is a chance event which decides first

18

dorsal

between anterior and posterior and then

between left and right, or whether the constitution of the egg in its one-cell stage contains specific agents which determine the

anieriqr....p

direction of these

shifts.

The

hypothesis of

the mosaic egg favoring the second hypothesis

seems more

poster/or
d

Driesch's violent encroachment on the

normal development proves that the first
cell division may not fix left and right
of the
growing organism for good. But even in
normal development the plane of the first
division
stages

may

of

not be the median.

cell

studied for the

36

division

worm

have

The

been

likely for the species Ascaris.

There are known a number of cases of

genotypical inversion where the genetic constitutions of two species are in the same relation as the atomic constitutions of two
enantiomoi ph crystals. More frequent, how-

first

closely

Ascaris megalocephala,

ever,

is

phenotypical inversion.

edness in

more

man

is

an example.

interesting one.

the lobster type have

and functionally
and a smaller a.

Left-handgive another

Several Crustacea of

two morphologically

different claws, a bigger

Assume

that in normally

developed individuals of our species, A is the

If in a young animal you cut off
right claw.

37

the right claw, inversive regeneration

takes
place: the

claw develops into the bigger

form A while at the place of the right claw
a
small one of type a is regenerated.
One has
to infer from such and similar
experiences the
left

bipotentiality of plasma,

namely that all genwhich contain the potency of an

asymmetric character have the potency of
erative tissues

bringing forth both forms, so however

that

normal development always one form

left or the right.
Which one is
genetically determined, but abnormal
exin

develops, the

ternal

On

circumstances

may

the basis of the strange

inversive regeneration

cause inversion.

phenomenon

of

Wilhelm Ludwig de-

veloped the hypothesis that the decisive

factors in asymmetry may not
be such specific
potencies

as,
say, the development of a
"right claw of type A," but two
R and L
(right and left) agents which are
distributed
in the organism with a
certain gradient, the
concentration of one falling off from
right
to left, the other in the opposite
direction.

The

essential point is that there is

not one
but that there are two opposite
gradient
fields

R and L. Which

strength

is

stitution.

is produced in greater
determined by the genetic conIf, however, by some
damage to

the prevalent agent the

other previously
suppressed one becomes prevalent, then
inversion takes place.
Being a

mathematician

and not a biologist I report with the

utmost
caution on these matters, which seem
to me
of highly hypothetical nature.
But
that the contrast of left and right

it is

clear

connected
with the deepest problems concerning
the
is

phylogenesis as well as the ontogenesis

of

organisms.

38

TRANSLATORY, ROTATIONAL, AND

RELATED SYMMETRIES

TRANSLATORS ROTATIONAL, AND

RELATED SYMMETRIES
From bilateral we

now

shall

turn to other

Even

kinds of geometric symmetry.

cussing the bilateral type

drawing

now and

in

in dis-

could not help

then such other sym-

metries as the cylindrical or the spherical

ones.

It

seems best to

fix

the underlying gen-

eral concept with some precision beforehand,

and to that end a little mathematics is needed,

which

for

ask your patience.

have spoken

A mapping S of space as-

of transformations.

sociates with every space point p a point p'

as

its

image.

special such

mapping

identity / carrying every point

Given two mappings

S,

/;

into

is

the

itself.

T, one can perform

and
p" then the resulting mapping, which we denote by ST, carries p into
p".
A mapping may have an inverse S' such
that SS' = / and S'S = I; in other words,
if S carries
the arbitrary point p into p'
then S' carries p' back into p, and a similar condition prevails with S' performed
in the first and 6" in the second place.
For
such a one-to-one mapping S the word transformation was used in the first lecture; let the
inverse be denoted by $~~\
Of course, the
identity / is a transformation, and / itself is
one

after the other:

T carries p'

its

inverse.

if

carries p into p'

into

Reflection in a plane, the basic

operation of bilateral symmetry,

its

iteration

other words,

SS

results

it is its

own

composition of mappings

ST

in

is

inverse.
is

in

In general

not commutative;

need not be the same as TS.

instance a point

such that

the identity;

in a plane

and

Take for
S be a

let

41

horizontal translation carrying o into

0\

and

Any

say that the automorphisms form a group.

Then

ST

totality,

carries o into the point 02 (Fig- 19), but

TS

group provided the following conditions are

a rotation around o by 90.

carries

If S

into 0*

the inverse S~~\ then

is

a transformation with

S~

is

also a transforma-

and its inverse is S. The composite of

two transformations ST is a transformation
-1
(in this
again, and (ST)' equals T-I S
order!).
With this rule, although perhaps
not with its mathematical expression, you are
When you dress, it is not immaall familiar.
terial in which order you perform the operalions; and when in dressing you start with the
shirt and end up with the coat, then in undressing you observe the opposite order; first
take off the coat and the shirt comes last.
tion

19

set

of transformations

satisfied: (1) the identity /

form a

belongs to F; (2)

S belongs to T then its inverse S -1 does;

(3) if S and T belong to T then the composite

if

ST does.

FIG.

any

One way

of describing the structure

space, preferred by both

of

Newton and Helm-

is through the notion of congruence.

Congruent parts of space V,
are such as
can be occupied by the same rigid body in
two of its positions. If you move the body
from the one into the other position the particle of the body covering a point p of V will
afterwards cover a certain point p' of V, and
thus the result of the motion is a mapping
p * p' of V upon V. We can extend the
rigid body either actually or in imagination
so as to cover an arbitrarily given point p of

holtz,

and hence the congruent mapping

can be extended to the entire space.
Any such congruent transformation I call
it by that name because it evidently has an
space,

/;

have further spoken of a special kind of

transformations of space called similarity by

But I preferred the name of

them, defining them with

the geometers.

automorphisms

for

Leibniz as those transformations which leave

the structure of space unchanged.

moment

it

is

ture consists.

For the

immaterial wherein that struc-

From

the very definition

it is

an automorphism,
and if S is, so is the inverse S~ l
Moreover
the composite ST of two automorphisms S, T
is again an automorphism.
This is only anclear that the identity /

is

other

way

similar to

of saying that (1) every figure

itself,

(2) if figure

F then F is similar to F',

to F'

and F'

to

F"

and

then

F'
(3)

is

is
if

42

similar to

F is similar

similar to F".

The mathematicians have adopted

group to describe this situation

is

the word
and therefore

>

p'

is a similarity or an automorphism; you can easily convince yourselves

that this follows from the very concepts.
It
is evident moreover that the congruent transformations form a group, a subgroup of the
group of automorphisms. In more detail the

inverse />'>/>

situation

is

this.

Among

the

similarities

do not change the

dimensions of a body; we shall now call them
there are those which

congruences.

A congruence is either proper,

screw into a left and a right

one into a right, or it is improper or reflexive,
changing a left screw into a right one and
vice versa.
The proper congruences are
those transformations which a moment ago
we called congruent transformations, concarrying a

left

43

necting the positions of points of a rigid

body before and after a motion. We shall

now call them simply motions (in a nonkinematic geometric sense) and call the
improper congruences reflections, after the
most important example: reflection in a
plane, by which a body goes over into its

Thus we have

mirror image.

this step-wise

* congruences =
similarities without change of scale * motions
arrangement:

proper

similarities

congruences.

and length as AA', in other words the vector

BB' = AA'. 1 The translations form a group;
indeed the succession of the two translations

BC

AC.
do with symmetry?
provides the adequate mathematical
It
language to define it. Given a spatial configuration ft, those automorphisms of space
which leave g unchanged form a group T,

AB,

results in the translation

What

has

this to

all

The congruences

form a subgroup of the similarities, the motions form a subgroup of the group of congruences, of index

means

The

2.

addition

latter

any given improper congruence, we obtain all improper congruences

in the form BS by composing B with all posthat

if

congruences

proper

sible

is

proper congruences form one

improper ones another

half,

group; for the composite

congruences A,

and the
of the group of all

But only the

congruences.

is

Hence the

S.

AB

half,

first

of

half

a proper congruence.

congruence leaving the point

may

is

two improper
fixed

around 0; thus there

are proper and improper rotations. The
be called

rotation

and

this

group describes exactly the symmetry pos-

sessed by %.

Space

itself

metry corresponding
automorphisms, of all

to

has the
the

full

group

similarities.

symof

multiples of 360/5 (including the identity),

and then by the five reflections in the lines

what

us
1

with the
sort

of

five vertices.

These ten

as

group.

around a given center

The

simplest

translation

represented by a vector AA'; for

tion carries a point

44

into B' then

form a

type of congruences

are the translations.

tells

symmetry the pentagram

While a segment has only length, a vector has

A vector is really the same thing

length and direction.

a translation, although one uses different phrase-

ologies for vectors

rotations

21

metry of any figure in space is

subgroup of that group. Take for instance
the famous pentagram by which Dr. Faust
banned Mephistopheles the devil. It is
carried into itself by the five proper rotations
around its center 0, the angles of which are

operations form a group, and that group

20

FIG.

described by a

joining

FIG.

all

The sym-

into A'

BB' has

the

if

may

be

a transla-

and the point

same direction

and

translations. Instead of speak-

ing of the translation a which carries the point

and ininto A' one speaks of the vector a = AA';

stead of the phrase: the translation a carries A into
A' one says that A' is the end point of the vector a
laid off from A. The same vector laid off from B ends
carries B
in B' if the translation carrying A into A'
into B'.

45

Hence the natural generalization

which leads from bilateral symmetry to sympossesses.

metry

wider geometric sense consists

plane by any group
of automorphisms.
The circle in a plane
in this

in replacing reflection in a

band ornaments and the

like.

and the sphere in space around

have the symmetry described by the group

After these general mathematical considerations let us now take up some special

plane or spatial rotations respectively.

% does not extend to infinity then

groups of symmetry which are important in

The operation which defines
art or nature.

with center
of

For such reasons as these we shall

almost exclusively consider groups of congruences even if we have to do with actually
or potentially infinite configurations such as

all

If a figure

symmetry, mirror

reflection,

es-

an automorphism leaving the figure invariant

must be scale-preserving and hence a con-

sentially

gruence,

figure consists of one

Here is the simple proof. Had
we an automorphism leaving g unchanged,
but changing the scale, then either this
automorphism or its inverse would increase

straight line

(and not decrease)

dimensional line, whereas its only proper

congruences are the translations. Reflection

unless the

point only.

all linear

certain proportion a:\

greater than

and

let a,

figure g.

/3

1.

dimensions in a

where

Call that

number

is

automorphism

S,

be two different points of our

They have a

positive distance

d.

The H-times

iterated transformation S" carries

into two points <x n , /3 of our figure

a and /3
whose distance is d a n
With increasing
exponent n this distance tends to infinity.
But if our figure J is bounded, there is a
number c such that no two points of % have a
.

Hence a

c.

is

operation.

can be reflected in any of

its

points 0; this reflection carries a point P into

that point P' that has the same distance from

Such reflections
but lies on the other side.
are the only improper congruences of the one-

followed by the translation

in

OA

yields

A\ which halves the

figure which is invariant

reflection in that point

distance OA.

ornament

of

= S\ SS = S SSS = S\

distance greater than

one-dimensional

under a translation

Iterate the transformation S,

bilateral

is

shows what in the art

called "infinite rapport,"

i.e.

repetition in a regular spatial rhythm.

pattern invariant under the translation
2

is

under
moreover under the identity t = I,
_1
of t and its iteraand under the inverse /

also invariant

its

iterations

tions

r\ r 2 r
,

3
,

by the amount a then

contradic-

na

(n

0,

If

I" shifts it

shifts the line

by the amount

1, 2,

tion arises as soon as n

becomes so large that

d a" > c. The argument shows another
thing: Any finite group of automorphisms

Hence

consists exclusively of congruences.

power

it

contains an

6"

if

1,

>

1,

then

many iterations S ,S 2 ,S z
l

all

if

shift

characterize a translation

we
a

t" is

by

then the iteration or

characterized by the multiple na.
it

effects

the

All translations carrying into itself a given

pattern of infinite rapport on a straight line

con-

are in this sense multiples na of one basic

that enlarges linear dimen-

sions at the ratio a

infinitely

For

the

tained in the group would be different because they enlarge at different scales a\ a 2

translation

a.

This rhythmic

may be com-

bined with reflexive symmetry.

If so

the

46

47

centers of reflections follow each other at

half the distance 2 a.
Only these two types
of symmetry, as illustrated by Fig. 22, are

possible

for

"ornament."

a one-dimensional pattern or

(The

crosses

mark

the

centers of reflection.)

*-

a
II

FIG. 22

Of course the real band ornaments are

not strictly one-dimensional, but their symmetry as

far as

we have

described

it

now

makes use of their longitudinal dimension

only.
Here are some simple examples from
Greek art. The first (Fig. 23) which shows

FIG.

a very frequent motif, the palmette, is of

type i (translation
reflection).
The next
(Fig. 24) are without reflections (type n).
This frieze of Persian bowmen from Darius'

widest, carried out in a peculiar technique,

palace in Susa (Fig. 25) is pure translation;

but you should notice that the basic transla-

the outer contour of the basic tree-like motif,

tion covers twice the distance

man

from

because the costumes of the

alternate.

Once more

I shall

24

Monreale mosaic of the Lord's Ascension

(Fig. 10), but this time drawing your attention to the

later taken

translatory

band ornaments framing

it.

The

up by the Cosmati, displays the

symmetry only by repetition of
FIG. 25

man to
bowmen

point out the

FIG. 23

48

49

while each copy

symmetric

is

filled

by a different highly

two-dimensional

The

mosaic.

palace of the doges in Venice (Fig. 26) may

stand for translator)' symmetry in archi-

Innumerable examples could

tecture.

be

added.

As

said before,

band ornaments really

around a

consist of a two-dimensional strip

and thus have a second

central line

trans-

the design of which sugstrand

crosses the other in
gests that one
space (and thus makes part of it invisible).
plaits of

If

some

sort,

interpretation

this

become

operations

accepted,

is

possible; for

further

example,

re-

flection in the plane of the ornament would

change a strand slightly above the plane into

one below. All this can be thoroughly

analyzed in terms of group theory as is for
instance done in a section of Andreas
Spciser's book, Theorie der Gruppen von endlicher

Ordnung, quoted in the Preface.

In the organic world the translatory sym-

metry, which the zoologists call metamerism,

is

seldom as regular as bilateral symmetry

A maple shoot and a shoot
is.

frequently

(Fig. 27) may serve

In the latter case translation

of Angraecum distichum
as examples.
is

accompanied by longitudinal

Of

tion.

into infinity (nor does a

but one

slip

reflec-

course the pattern does not go on

may

say that

it is

band ornament),
potentially infinite

one direction, as in the course of

time ever new segments separated from each
Goethe
other by a bud come into being.

at least in

said of the tails of vertebrates that they allude

as

it

to the potential infinity of organic

were

The

existence.

shown
FIG.

26

As such they can have

The

further symmetries.

carried into
line

/;

let

reflection
line

itself

by

may

cem-

symmetry, the basic

segment and longitudinal reflection.

In one-dimensional time repetition at equal

/.

Or

reflection in a

the pattern

may

itself by longitudinal refleccombined with the translation by %a

slip

bilateral,

translational,

reflection in the central

from the transversal

(longitudinal

bincd with

regular

operations of which are translation by one

us distinguish this as longitudinal

perpendicular to

fairly

scolopendrid (Fig. 28),

be

pattern

be carried into
tion

reflection).

frequent

motif in band ornaments are cords, strings, or

50

in this picture, a

possesses

versal dimension.

central part of the animal

intervals

is

the musical principle of rhythm.

grows it translates, one might say,

Rea slow temporal into a spatial rhythm.

As

a shoot

This and the next picture are taken from Studium

p. 249 and p. 241 (article by W. Troll,

Generate,

"Symmetriebetrachtung

in der Biologic").

28

flection,

inversion in time, plays a far less

important part

in

music than rhythm does.

melody changes

its character to a considplayed backward, and I, who

a poor musician, find it hard to recog-

erable degree

am

if

when

nize reflection

it is

used in the construc-

But such schemes fall

3
heading of symmetry.

hardly

under the

symmetry in space. Take a

band ornament where the individual section
repeated again and again is of length a and

We

return to

no such spon-

around a circular cylinder, the circumference of which is an integral multiple

taneous effect as rhythm. All musicians

agree that underlying the emotional element
of music is a strong formal element.
It may

You then obtain a

of a, for instance 25a.
over into itself
carried
pattern which is

tion of a fugue;

be that

it

certainly has

capable of some such mathematical treatment as has proved successful

it

is

for the art of

ornaments.

If so,

we have

probably not yet discovered the appropriate

mathematical tools. This would not be so
surprising.

For

celled in the

after all, the

ornamental

Egyptians ex-

art four

thousand

years before the mathematicians discovered

group concept the proper mathematical

instrument for the treatment of ornaments
in the

and for the derivation of

metry classes. Andreas

their possible

sling

it

through the rotation around the cylinder axis

by a = 360/25 and its repetitions. The
twenty-fifth iteration

is

We

the rotation

by 360,
group

thus get a finite

or the identity.
of rotations of order 25,

one consisting of
25 operations. The cylinder may be replaced by any surface of cylindrical symi.e.

3
The reader should compare what G. D. Birkhoff
and music
has to say on the mathematics of poetry
Lecture
i, note 1.
quoted
in
publications
the
two
in

sym-

who

Speiser,

has
taken a special interest in the group-theoretic
aspect of ornaments, tried to apply combinatorial

mathematical nature
problems of music. There

principles of a

also to the formal

a chapter with this title in his book, "Die

mathematische Denkweise," (Zurich, 1932).
As an example, he analyzes Beethoven's pas-

is

toral sonata for piano,

opus 28, and he also

points to Alfred Lorenz's investigations on

the formal structure of Richard

Wagner's

chief works.

is

related,

science

common

and
has

Metrics in poetry
here,

so

penetrated

principle

in

Speiser

much

closely

maintains,

deeper.

music and

prosody

seems to be the configuration a a

b which is
theme a that is repeated
and then followed by the "envoy" b; strophe,
antistrophe, and epode in Greek choric lyrics.

often called a bar: a

52

FIG.

29

FIG.

30

53

metry, namely by one that

by

is

carried into

around a certain axis, for

instance by a vase.
Fig. 29 shows an attic
vase of the geometric period which displays
quite a number of simple ornaments of this
type.
The principle of symmetry is the same,
although the style is no longer "geometric,"
in this Rhodian pitcher (Fig. 30), Ionian
itself

rotations

all

school of the seventh century b.c.

illustrations are

Other

such capitals as these from

Egypt (Fig. 31). Any finite group of

proper rotations around a point
in a plane,
or around a given axis in space, contains a
early

primitive rotation

part

360%

consists
t

of

of the
its

identity.

whose angle

full

rotation

iterations

The

acterizes this .group.

t\

is an aliquot
by 360, and

2
,

/-',
,

order n completely char-

The

the analogous fact that

result follows

any group of

from

trans-

lations of a line, provided it contains no

operations arbitrarily near to the identity
except the identity itself, consists of the iteraFIG.

31
FIG. 32

tions va of a single translation a (v

2,

may

0,

1,

The wooden dome

once

in the

the palace of the Beys of

serve as

Bardo of Tunis,
Tunis

an example from

(Fig. 32),

interior archi-

next picture (Fig. 33) takes

you to Pisa; the Baptisterium with the tinylooking statue of John the Baptist on top is a
tecture.

The

whose exterior you can

horizontal layers each of

central building in
distinguish

rotary

six

symmetry

of a different order

make the picture still more

by adding the leaning tower with
could

leries of

54

arcades

all

n.

One

impressive
its

six gal-

having rotary symmetry

55

"

the round arcs of the friezes, octagonal central

symmetry (n = 8, a low value compared to
those embodied in the several layers of the
Pisa Baplisterium) in the small rosette and the
three towers, while bilateral symmetry rules

the structure as a whole as well as almost

every detail.
Cyclic symmetry appears in its simplest
form if the surface of fully cylindrical symmetry is a plane perpendicular to the axis.
We then can limit ourselves to the two-

it

FIG. 34

as
SV'SM^

FIG. 33

same high order and the dome itself,

the exterior of whose nave displays in columns

of the

and

friezes patterns of the lineal translatory

type of symmetry while the cupola is surrounded by a colonnade of high order rotary

symmetry.

An

entirely different spirit speaks to us

from the view, seen from the rear of the

choir, of the Romanesque cathedral in Mainz,
Germany (Fig. 34). Yet again repetition in

56

57

'

FIG. 35

Mag-

dimensional plane with a center 0.

examples of such central plane symmetry are provided by the rose windows of
Gothic cathedrals with their brilliant-colored
nificent

glasswork.

The

richest I

remember

rosette of St. Pierre in Troyes, France,

is

based on the

number

is

the

which

3 throughout.

Flowers, nature's gentlest children, are also

conspicuous for their colors and their cyclic

symmetry. Here (Fig. 35)

an iris with its triple pole.
of 5
like

58

is

most frequent among

the

following

(Fig.

is

a picture of

The symmetry

flowers.

36)

from

A page
Ernst

Haeckel's Kunstjormen der Natur seems to indithat it also occurs not infrequently

cate

among

the lower animals.

warn me

that the

But the biologists

outward appearance of

these echinoderms of the class of Ophiodea

to a certain degree deceptive;

is

their larvae

are organized according to the principle of

symmetry. No such objection attaches to the next picture from the same
source (Fig. 37), a Discomedusa of octagonal
bilateral

symmetry.

For the coelentera occupy a

place in the phylogenetic evolution where

way

cyclic has not yet given

metry.

Haeckel's

which

to bilateral

extraordinary

sym-

work,

his interest in the concrete

in

forms of

organisms finds expression in countless drawings executed in minutest detail, is a true

nature's

codex of symmetry.

Equally

re-

vealing for Haeckel, the biologist, are the

thousands and thousands of figures in

Challenger Monograph, in
for the first time 3,508

new

species of radio-

by him on the Challenger

larians discovered

Expedition,

his

which he describes

1887.

One

should

not forget

these accomplishments over the often all-too-

speculative

which

phylogenetic

constructions

this enthusiastic apostle of

in

Darwinism

indulged, and over his rather shallow materialistic

philosophy of monism, which

quite a splash in

Germany around

made

the turn

of the century.

Speaking of Medusae I cannot resist the

temptation of quoting a few lines from D'Arcy

Thompson's

classic

work on Growth and Form,

a masterpiece of English literature,

combines profound knowledge

physics,
tion

and

nality.

in

which

geometry,

and biology with humanistic erudiscientific insight of

Thompson

unusual

origi-

reports on physical ex-

periments with hanging drops which serve

60

FIG.

37

^
by analogy the formation of
medusae. "The living medusa," he says,
"has geometrical symmetry so marked and
regular as to suggest a physical or mechanical
element in the little creatures' growth and

to

illustrate

construction.

It

has,

to

begin

with,

vortex-like bell or umbrella, with

manubrium.

metrical handle or
is

traversed

its

sym-

its

The

bell

by radial canals, four or in

its edge is beset with ten-

multiples of four;

smooth or often beaded, at regular

or of graded sizes; and certain

tacles,

intervals

or

tions

No

pulsate;

to

miniature

Buds,

are

'otoliths,'

interspersed.

gins

including

structures,

sensory

also

solid

concre-

symmetrically

sooner made, then

the bell
replicas

it

be-

begins to 'ring.'
of

the

parent-

organism, are very apt to appear on the

on the manubrium or sometimes

on the edge of the bell; we seem to see one
vortex producing others before our eyes.
The development of a medusoid deserves to
be studied without prejudice from this point
Certain it is that the tiny medusoids
of view.
tentacles, or

of Obelia, for instance, are

rapidity
suggests

budded

off with a

and a complete perfection which

an automatic and all but instantane-

ous act of conformation, rather than a gradual

process of growth."
While pentagonal symmetry is frequent in
world, one does not find it
the most perfectly symmetrical creations of inorganic nature, among the crystals.
There no other rotational symmetries are
the

organic

among

possible than those of order 2, 3, 4,

Snow

crystals provide the best

and

known

6.

speci-

Fig. 38 shows
marvels of frozen water.
In my youth, when they came down from
heaven around Christmastime blanketing the

mens
some

of hexagonal symmetry.
of these

little

63

landscape, they were the delight of old and

Now only the skiers like them, while
they have become the abomination of motor-

in their

colum-

nar structures." 5

Those versed in English literature will

remember Sir Thomas Browne's quaint ac-

improper rotations
are taken into consideration, we have the two

ists.

count in

his Garden of Cyrus (1658) of

Up

to

hexag-

following
rotations

and

"quincuncial"

now we have

proper rotations only.

symmetry which
"doth neatly declare how nature Geometrizeth and observeth order in all things."

onal

paid attention to

If

possibilities

for

groups of

finite

in plane geomaround a center

etry, which correspond to the two possibilities
we encountered for ornamental symmetry on

One versed in German literature will remember how Thomas Mann in his Magic

a line: (1) the group consisting of the repetitions of a single proper rotation by an aliquot

Mountain*

part

wesen"

Hans

describes

of the

the

snow storm

"hexagonale
in

which

Castorp, nearly perishes

Un-

his hero,

when he

falls

asleep with exhaustion

and leaning against

deep dream of death and

dreams his
hour before when Hans sets out
on his unwarranted expedition on skis he
enjoys the play of the flakes "and among
these myriads of enchanting little stars,"
so
a barn

love.

An

he philosophizes, "in their hidden splendor,

too small for man's naked eye to see, there
was not one like unto another; an endless
inventiveness governed the development and
unthinkable differentiation of one and the
same basic scheme, the equilateral, equiangled hexagon. Yet each in itself this was
the uncanny, the antiorganic, the life-denying
character of them all each of them was absolutely symmetrical, icily regular in form.

They were

too regular, as substance adapted

never was to this degree the living
principle shuddered at this perfect precision,
found it deathly, the very marrow of death
to life

Hans Castorp
reason

why

felt

he understood now the

the builders of antiquity pur-

posely and secretly introduced minute varia4

New

64

from absolute symmetry

tion

young.

quote Helen Lowe- Porter's translation,

Knopf
York, 1927 and 1939.

a = 360/w

of 360; (2) the group of

combined with the

these rotations

reflections

The

axes forming angles of J^a.

in n

first

group C n and the

group is
group
D n Thus these
second the dihedral
called the cyclic

are the only possible central symmetries in

two-dimensions:

C, C2

(1)

Cz,

;D h D

Ci means no symmetry at
symmetry and nothing else.
the

all,

ings with the

after

symmetry

The

antiquity,

Florence

first

S.

(begun

D\

bilateral

Towers

of 4 prevails.

symmetry

In architecture
often

Central build-

have hexagonal symmetry.

frequent.

,D,,

of 6 are

much

less

pure central building

Maria degli Angeli in

1434),

is

an

octagon.

Pentagons are very rare. When once before

I lectured on symmetry in Vienna in 1937 I
said I knew of only one example and that a
very inconspicuous one, forming the passageway from San Michele di Murano in Venice
8

Diirer considered his canon of the human figure

as a standard from which to deviate than as a

more

standard toward which to

strive.

Vitruvius'

tem-

seem to have the same sense, and maybe

the little word "almost" in the statement ascribed to
Polykleitos and mentioned in Lecture i, note 1,
points in the same direction.

peralurae

65

hexagonal Capella Emiliana. Now,

we have the Pentagon building in
Washington. By its size and distinctive
to the

of course,

shape,

it

provides an attractive landmark for

Leonardo da Vinci engaged

bombers.

in

systematically determining the possible symmetries of a central building and how to at-

tach chapels and niches without destroying

the symmetry of the
modern terminology,

nucleus.

In abstract

his result

is

essentially

our above table of the possible finite groups

of rotations (proper and improper) in two
dimensions.

FIG. 39

So far the rotational symmetry in a plane

had always been accompanied by reflective
symmetry; I have shown you quite a number
of examples for the dihedral group D n and
none for the simpler cyclic group Cn
But
this is more or less accidental.
Here (Fig.
39) are two flowers, a geranium (i) with the
symmetry group Ds, while Vinca herbacea (n)
has the more restricted group Cs owing to
the asymmetry of its petals.
Fig. 40 shows
what is perhaps the simplest figure with rotational symmetry, the tripod (n = 3).
When
one wants to eliminate the attending reflec.

FIG.

40

symmetry, one puts little flags unto the

arms and obtains the triquetrum, an old
magic symbol. The Greeks, for instance,
tive

short

time before Hitler's hordes occupied

added concerning the swastika:
I

"In our days

it

has become the symbol of a

than the snake-girdled

terror far

symbol for the three-cornered Sicily.

(Mathematicians are familiar with it as the
seal on the cover of the Rendiconti del Circolo

Medusa's

head" and

applause

and

broke

it

as the

Matematico

Palermo.)

di

The

with four instead of three arms

is

modification
the swastika,

which need not be shown here one of the

most primeval symbols of mankind, common
possession of a

pendent

symmetry

number

civilizations.

in

Vienna

of apparently inde-

In

my

lecture

on

in the fall of 1937, a

41

Austria,

with the Medusa's head in the center

used

66

FIG.

more

audience.

It

terrible

booing

pandemonium
in

of

the

seems that the origin of the

magic power ascribed

to these patterns lies

their

startling

incomplete

rotations

without

reflections.

in

loose

symmetry
Here (Fig.

the gracefully designed staircase of the

pulpit of the Stephan's dome in Vienna; a
41)

is

triquetrum

alternates

with a swastika-like

wheel.

67

So much about rotational symmetry

dimensions.

two

in

dealing with potentially in-

If

band ornaments or with

the operation under which
invariant is not of necessity a

patterns like

finite

infinite groups,

the pattern

is

congruence but could be a similarity. A

one dimension that is not a mere

similarity in

has a fixed point

translation

dilatation s from

where a 9*
assume

1.

to

this

It

>

is

and

A
is

how

42).

1, 2,

Q,

really

is

).

remarkable

quite

exactly the widths of the consecutive

whorls of this shell follow the law of geometric

some

of

clocks perform a con-

tinuous uniform rotation, others

minute

to

integral

number

tinuous

subgroup

group of

The

minute.

all

jump from

rotations

by an

of minutes form a discon-

within

and
and its

rotations,

consider a rotation

in

the

motion during the time t\ + t 2

minute leads to a definite trans1
.

S(\),

and

for all integers n the

motion S(n) performed during

the iteration

n
:

The motion

minutes

is

2
embedded

the discontinuous group

is

group with the parameter t

One could
consisting of the motions S(t).
say that the continuous motion consists of the
endless repetition of the

motion in consecutive

same

infinitesimal

infinitely small

time

intervals of equal length.

We

could have applied this consideration

the

continuous

is

it

natural to

dilatations.

We now

similarity

i.e.

s,

envisage any proper

one which does not inter-

change left and right. If, as we assume, it is

not a mere translation, it has a fixed point
combined
and consists of a rotation about
It can
with a dilatation from the center 0.

continuous group.
We can
viewpoint to any similarity in 1,

be obtained as the stage S{\) reached after

1 minute by a continuous process S(t) of combined uniform rotation and expansion.

2,

or 3 dimensions, as a matter of fact to any

transformation s.
The continuous motion of

along a
This process carries a point t*
so-called logarithmic or equiangular spiral.

This curve, therefore, shares with straight

line and circle the important property of
going over into itself by a continuous group

contained

apply

iterations (2) as

in the

this

space-filling substance, a "fluid," can

mathematically be described by giving the

transformation
position

of

moment over
/

/'.

These

U(t,l')
which
any point of the

into

its

position

transformations

parameter group
time difference/'

68

to the rotations of a plane disc as well as to

The hands

42

two con-

result

in the continuous

progression.

FIG.

+ h)

consisting of the iterations of s

It

S(h

secutive time intervals

formation

good example of this type of symmetry

shown by the shell of Turrilella duplicate

(Fig.

expresses that the motions during

essential restriction

of the dilatations
(n

S(h)S(h)

no

of

Then the fluid is

The simple group

law

during

iteration

repeated.

is

"uniform motion."

in

a certain ratio a:\

operation generates a group 2 consisting

(2)

during equal time intervals always the

same motion

in

Indefinite

0.

is

i.e. if

carries

the

fluid at the

at the time

form a onedepends on the

/only, U(t/) = S(l' - /),

if

U(l,l')

of similarities.

The words by which James

Bernoulli had the spira mirabilis adorned on

his

tombstone

in

"Eadem mutata

the

Munster

resurgo,"

at

Basle,

are a grandilo-

69

The most

general rigid motion in three-

dimensional space

is

a screw motion

s,

com-

bination of a rotation around an axis with a

Under the influtranslation along that axis.

ence of the corresponding continuous uniform

motion any point not on the axis describes a
screw-line or helix which, of course, could
say of

itself

with the same right as the loga-

rithmic spiral: eadem resurgo.

The

which the moving point reaches

FIG. 43

quent expression of this property. Straight

line and circle are limiting cases of the logarithmic spiral, which arise when in the combination rotation-plus-dilatation one of the

two components happens to be the identity.

stages reached by the process at the

The

times
(3)

-2, -1,0,1,2,

form the group consisting of the iterations

The well-known shell of Nautilus (Fig.
(2).
43) shows this sort of symmetry to an astonishing perfection.

You

continuous logarithmic

see here not only the

but the potensequence of chambers has a

symmetry described by the discontinuous
group 2. For everybody looking at this
tially

spiral,

infinite

picture (Fig. 44) of a giant sunflower,

Hehanthus maximus, the florets will naturally
arrange themselves into logarithmic spirals,

two

70

sets of spirals of

opposite sense of coiling.

stages

Pn

at the equi-

moments (3) are equidistributed over

the helix like stairs on a winding staircase.
If the angle of rotation of the operation j- is a

distant

fraction p/v of the

full

angle 360 expressible

terms of small integers

in

Pn

point of the sequence

vertical,

and

full

y.

The

it.

lies

i>th

on the same

turnings of the screw are

necessary to get from

above

then every

/x,

leaves

Pn

to the point

Pn+V

around the shoot of a

show such a regular spiral arrangement. Goethe spoke of a spiral tendency in nature, and under the name of
phyllolaxis this phenomenon, since the days
plant often

of Charles Bonnet (1754), has been the sub-

much investigation and more speculaamong botanists. 6 One has found that

ject of
tion

the fractions n/v representing the screw-like

arrangement of leaves quite often are members of the "Fibonacci sequence"
(4)

H,

which

y2

13

H, H, H, Hs, Ai,

results

of

the

irrational

This number

Helianthus with

its

number

is

played such a role in attempts to reduce

of proportion to a mathematical

beauty

wound

is

The

cylinder on which the screw

could be replaced by a cone;

this

amounts to replacing the screw motion s by

any proper similarity rotation combined

The arrangement

with dilatation.

on a

The
disc

of scales

fir-cone falls

general form
transition
is

under this slightly more

symmetry in phyllotaxis.
from cylinder over cone to

of

obvious, illustrated by the cylindrical

This phenomenon plays also a role in

J. HamHis Dynamic symmetry contains
on pp. 146-157 detailed notes by the mathematician
6

bidge's constructions.

R. C. Archibald on the logarithmic

and the Fibonacci series.

section,

72

spiral,

with

Where one can

florets.

its

check the numbers

(4) best,

namely

for the

arrangement of scales on
curacy is not too good nor are considerable
P. G. Tait, in the
deviations too rare.
a fir-cone, the ac-

Proceedings

Royal Society of Edinburgh

of the

(1872), has tried to give a simple explanation,

H. Church

while A.
treatise

Relations

in

voluminuous

his

of phyllotaxis

to

mechanical

laws (Oxford, 1901-1903) sees in the arith-

metics of phyllotaxis an organic mystery.

am

modern

afraid

doctrine

of phyllotaxis

than

seriously

less

whole

botanists take this

their forefathers.

Apart from

reflection all

symmetries so far

considered are described by a group consisting of the iterations of one operation

case,

and that

is

In

s.

undoubtedly the most

the resulting group is finite,

one takes for s a rotation by an
= 360/n which is an aliquot part
For the two-diof the full rotation 360.
mensional plane there are no other finite
groups of proper rotations than these; witness
of Leonardo's
the first line, C h C 2 C 3
table (1).
The simplest figures which have

namely
angle a

if

formula.

leaves, a fir-cone

discoidal inflorescence of

important,

no other but
the ratio known as the aurea sectio, which has
1).

and the

scales,

its

one

from the expansion into a con-

tinued_ fraction

M(V5

%,

stem of a plant with

golden

the corresponding

symmetry are the regular

polygons: the regular triangle, the square,

The

the regular pentagon, etc.

there

is

for

every

number

a regular polygon of n sides

=
is

3,

fact
4,

5,

that

closely related

to the existence for every n of a rotational

group of order n in plane geometry. Both

Indeed, the situafacts are far from trivial.
altogether differis
in
three
dimensions
tion
infinitely
many regular
exist
ent: there do not
than five,
not
more
polyhedra in 3-space, but
often called the Platonic solids because they

73

play an eminent role in Plato's natural phi-

They

losophy.

are the regular tetrahedron,

moreover the
pentagondodecahedron, the sides of which
are twelve regular pentagons, and the icosahedron bounded by twenty regular triangles.
One might say that the existence of the first
But
three is a fairly trivial geometric fact.
the discovery of the last two is certainly one
of the most beautiful and singular discoveries
made in the whole history of mathematics.

With a

octahedron,

the

cube,

the

fair

amount

of certainty,

it

Greeks never used the word "symmetric"

In common usage
in our modern sense.
(xvfj.fiTpos
it is

means

proportionate,

while in Euclid

equivalent to our commensurable: side

and

diagonal of a square are incommensurable

quantities, durvnnerpa neykdr).

Here

page from Haeckel's

the skeletons
showing
monograph

(Fig. 45) is a

Challenger

FIG.

45

can be

traced to the colonial Greeks in southern Italy.

The

made

suggestion has been

of

crystals

abundant

in

that they ab-

dodecahedron from the

stracted the regular

pyrite,

sulphurous

mineral

But as mentioned be-

Sicily.

symmetry of 5 so characteristic for

dodecahedron contradicts the
laws of crystallography, and indeed one finds
that the pentagons bounding the dodecahedra
in which pyrite crystallizes have 4 edges of
The
equal, but one of different, length.
fore, the

the

regular

first

exact construction of the regular penta-

gondodecahedron is probably due to Theaetetus.

There is some evidence that dodecahedra were used as dice in Italy at a very
early time and had some religious significance
in

Etruscan culture.

Timaeus,

Plato, in the dialogue

the

associates

regular

pyramid,

octahedron, cube, icosahedron, with the four

elements of
this

order),

hedron he

fire,

air,

earth,

and water

(in

while in the pentagondodeca-

sees in

some

universe as a whole.

sense the

image

of the

A. Speiscr has advo-

cated the view that the construction of the

five regular solids

is

the chief goal of the

deductive system of geometry as erected by

the Greeks
ments.

74

May

and canonized
I

in

Euclid's Ele-

mention, however, that the

75

of several Radiolarians.
Nr. 2, 3, and 5 are
octahedron, icosahedron, and dodecahedron

astonishingly regular form; 4 seems to have

in

a lower symmetry.

about the three outer planets, Uranus, Neptune, and Pluto, which were discovered in
1781, 1846,

and 1930

to find the reasons

He

respectively.)

tries

the Creator had

why

order of the Platonic solids and

chosen

this

draws

parallels

between the properties of

the planets (astrological rather than astrophysical properties) and those of the corre-

sponding regular bodies.

proclaims

which

he

spatioso

numen

his

mighty

hymn

in

"Credo

credo,

in orbe," concludes his book.

share his belief in a mathematical

harmony of the universe. It has withstood

We

still

the test of ever widening experience.

we no
forms

longer seek this

like the

harmony

But
static

in

regular solids, but in dynamic

laws.

As the regular polygons are connected with

the finite groups of plane rotations, so must
the regular polyhedra be intimately related
to the finite groups of proper rotations around

FIG

From the study of

in space.
a center
plane rotations we at once obtain two types
Indeed,
of proper rotation groups in space.

46

Kepler,

in

his

Myslerium

cosmographicum,

published in 1595, long before he discovered

the three laws bearing his name today, made

an attempt

reduce the distances in the

planetary system to regular bodies which are
to

alternatingly inscribed
spheres.

Here

and circumscribed

(Fig. 46)

is

to

his construction,

by which he believed he had penetrated

deeply into the secrets of the Creator. The
six spheres correspond to the six planets,
Saturn, Jupiter, Mars, Earth, Venus, Merseparated in this order by cube,
tetrahedron, dodecahedron, octahedron, ico-

curius,

sahedron.

76

(Of course, Kepler did not know

the group C n of proper rotations in a horican be interzontal plane around a center

preted as consisting of rotations in space

around the

vertical axis

Reflec-

through 0.

tion of the horizontal plane in a line

of the

plane can be brought about in space through

a rotation around / by 180 (Umklappung)
You may remember that we mentioned this
in connection with the analysis of a Sumerian
picture

Dn

(Fig.

4).

In

way

this

in the horizontal plane

is

the group

changed

into a

group D'n of proper rotations in space; it contains the rotations around a vertical axis
by the multiples of 360/n and
through
the

Umklappungen around

through

which form

n horizontal axes

equal

angles

of

77

360/2/z with each other.

But

it

should be

observed that the group D\ as well as C2

consists of the identity and the Umklappung

around one

These two groups are

and in a complete list of

line.

therefore identical,

the

groups of proper rotations in three

dimensions, D\ should be omitted if C2 is
different

kept.

Hence we

start

our

D'2

is

(5)

3 , Z>'4 ,

This

and the Umklappungen around three

mutually perpendicular axes.

For each one of the five regular bodies we
can construct the group of those proper rotations

which carry that body into

this give rise to five

to three,

and that

Does
No, only

itself.

new groups?

for the following reason.

Inscribe a sphere into a cube

and an octahedron into the sphere such that the corners

of the octahedron lie where the sides of the

FIG

47

cube touch the sphere, namely

in the centers

of the six square sides.

47 shows the

(Fig.

two-dimensional analogue.)

In this position

cube and octahedron are polar

sense of projective geometry.

figures in the

It is clear that

every rotation which carries the cube into

itself

leaves the octahedron invariant,

also

and vice

Hence the group for the

the same as for the cube.
In
same manner pentagondodecahedron and
versa.

octahedron
the

is

icosahedron are polar figures.

polar to a regular tetrahedron

The
is

figure

a regular

tetrahedron the corners of which are the

antipodes of those of the

new groups

first.

Thus we

find

W,
and P; they are those leaving invariant the
regular tetrahedron, the cube (or octahedron), and the pentagondodecahedron (or
three

78

Cn

(n

D'n

(n

W,

=
=

1, 2, 3,

2, 3,

is

complete:

),

);

P.

the so-called four-group consisting of the

identity

of these three groups our table

>'
,

are 12, 24, 60 respectively.

simple
It can be shown by a relatively
addition
the
with
that
A)
(Appendix
analysis

T,

Ci, C2, Cs, C4,

D'2

thus:

list

icosahedron) respectively. Their orders, i.e.

the number of operations in each of them,

of proper rotations, T,

is

the

modern equivalent

to the tabula-

the Greeks.
tion of the regular polyhedra by
three,

These groups, in particular the last

subject
are an immensely attractive

for

geometric investigation.
further possibilities arise if improper
groups?
rotations are also admitted to our

What

use
This question is best answered by making
rotation,
improper
of one quite singular
point
namely reflection in 0; it carries any
found
to
respect
P into its antipode P' with
and prolonging the
by joining P with
straight line

P0

by

its

own

length:

PO =

with every
OP'. This operation Z commutes
= S Now let T be one of
rotation S,

<?

groups of proper rotations. One

our
way of including improper rotations is simply
adding to
by adjoining , more precisely by
the improper
the proper rotations 5 of T all
The
in T).
rotations of the form <S (with S
finite

thus oborder of the group T = T + Z?

Another
tained is clearly twice that of V.

way

arises
of including improper rotations

contained
this situation: Suppose T is
group T'
another
in
2
index
of
subgroup
as a
of the
one-half
that
so
rotations;
of proper
oneand
them
S,
call
in
I\
lie
V
elements of

from

half,

by

S',

the

do not.
improper

manner you

Now

replace these latter

ZS> In tms

rotations

get a group V'T

which contains

79

F while the other half of

S' of

Dn

not contained in

pungen around the

operations are

its

For instance, T
= D'n
group of index 2 of
improper.

Cn

=
;

is

a sub-

the operations

are the

Umklap-

The

n horizontal axes.

corresponding ^S' are the reflections in the

vertical planes perpendicular to these axes.

Thus DCn

the angles of

and of the

multiples of 360%?,

through

planes

vertical

around
which are

consists of the rotations

the vertical axis,

reflections in

axis

this

forming

You

with each other.

angles of 360/2tz

might say that this is the group formerly

denoted by D n
Another example, the
simplest of all: T = C x is contained in V' =
d. The one operation S' of C2 not contained
-

the rotation by 180 about the vertical

in Ci

is

axis;

<5"

is

reflection in the horizontal plane

through 0.

Hence C>Ci

sisting of the identity

is the group conand of the reflection

given plane; in other words, the group

which bilateral symmetry refers.
The two ways described are the only ones
by which improper rotations may be included
in our groups.
(For the proof see Appendix
in a

to

Hence

B.)
finite

this

groups

is

of

the complete table of

and

(proper

all

improper)

rotations:

cn

c,

D'n

D'n

(n

C/2nC/n

The

last

W,

D'n a,

(n

r,

1, 2, 3,

group

W,

T,

WT

is

made

P;

2, 3,

WT.

possible by the

T is a subgroup of index 2 of the octahedral group W.

This list will be of importance to us when in
fact that the tetrahedral

the last lecture

metry of

80

we

crystals.

group

shall consider the

sym-

ORNAMENTAL SYMMETRY

ORNAMENTAL SYMMETRY
This lecture

will

have a more systematic

character than the preceding one, in as much

special kind of
as it will be dedicated to one
complicated
most
the
symmetry,
geometric
angle.
but also the most interesting from every
ornasurface
of
the
art
In two dimensions
it
dimensions
three
in
it,
with
ments deals

characterizes the arrangement of atoms in a

shall therefore call it the ornacrystal.

We

mental or crystallographic symmetry.

Let us begin with an ornamental pattern
probably occurs
in two dimensions which

more frequently than any

other, both in art

and nature: the hexagonal pattern

used
see
is

for

here realized

it

built

The

floors

tiled

by our
cells

bees'

photograph
prisms.
consists

is

so often

bathrooms.

in

by the honeycomb

common

You
as

it

hivebees (Fig. 48).

have prismatic shape, the

taken in the direction of these

As a matter of fact, a honeycomb

of two layers of such cells, the prisms

the other
of the one layer facing one way,
How the inner ends of
the opposite way.
these two
which we

layers dovetail

is

a spatial problem
At the

shall presently take up.

moment we

are concerned with the simpler

two-dimensional question. If you pile roundwill of

shots or round beads in a heap they
themselves get arranged in the three-dimensional analogue of the
tion.

hexagonal configurais to pack

In two dimensions the task

equal circles as compactly as possible. You

circles that
start with a horizontal row of
another
drop
you
If
other.
each
touch

from above upon

between two adjacent

circle

this

row

circles

it

of

will nest

the

row,

83

and the centers

an equilateral

of the three circles will

triangle.

circle there derives a

From

this

form
upper

second horizontal row of

between those of the first row;

on (Fig. 49). The circles leave little
lacunae between them. The tangents of a
circles nesting

and

so

circle at the points

surrounding
FIG.

circles

where it touches the six

form a regular hexagon

and if you replace

hexagon
you obtain the
each circle by this
filling the
hexagons
regular configuration of

circumscribing the

circle,

whole plane.
According to the laws of capillarity a soap
film spanned into a given contour made of
thin wire assumes the shape of a minimal surface, i.e. it has smaller area than any other

48

FIG.

surface

with

bubble into

the

same contour.

which a quantum

of air

is

49

soap

blown

assumes spherical form because the sphere

encompasses the given volume with a mini-

mum

of surface.

Thus

is

it

not astonishing

that a froth of two-dimensional

equal area will arrange

pattern because among

itself in
all

bubbles of

the hexagonal

divisions of the

plane into parts of equal area that

for which the net of contours has
length.

We

is

the

one

minimum

here suppose that the problem

has been reduced to two dimensions by dealing with a horizontal layer of bubbles, say,
between two horizontal glass plates. If the

boundary (an epidermal

the biologist would say), we observe

froth of vesicles has a

layer, as

84

85

FIG.

50

that it consists of circular arcs each forming

an angle of 120 with the adjacent cell wall
and the next arc, as is required by the law
of minimal length.
After this explanation
FIG.

51

FIG. 52

one will not be surprised to find the hexagonal

pattern realized in such different structures
of maize
for instance the parenchyma
as

the retinal pigment of our eyes,

which I
the surface of many diatoms, of
specimen,
show here (Fig. 51) a beautiful
50),

(Fig.

and finally the honeycomb. As the bees,

which are all of nearly equal size, build their
cells will
cells gyrating around in them, the
form a densest packing of parallel circular
our
cylinders which in cross section appear as
hexagonal pattern of circles. As long as the

work the wax is in a semi-fluid

and thus the forces of capillarity proba-

bees are at
state,

bly

more than

within
circles

by the

hexagons (whose
show some remains of

into circumscribed

corners however
the

the pressures exerted from

bees' bodies transform the

circular

still

form).

With the parenchym

celof maize you may compare this artificial

diffusion
the
by
formed
(Fig.
52)
tissue
lular

in gelatin of

drops of a solution of potassium

87

The

ferrocyanide.

regularity leaves some-

thing to be desired; there are even places

where a pentagon is smuggled in instead of a
53 and 54) are two
other artificial tissues of hexagonal pattern
taken at random from a recent issue of Vogue
(February 1951). The siliceous skeleton of

Here

hexagon.

(Fig.

FIG. 53
FIG.

54

one of Haeckel's Radiolarians which he called

Aulonia hexagona (Fig. 55) seems to exhibit a

FIG. 55

regular hexagonal pattern spread out

not in a plane but over a sphere. But a
hexagonal net covering the sphere is im-

fairly

possible

a fundamental formula of
This formula refers to an arbi-

owing

topology.

to

trary partition of the sphere into countries

that border on each other along certain

edges.
tries,

that the

number

number A

E of edges

of coun-

and the number

of corners (where at least three countries

come

It tells

the

2.

together) satisfy the relation

Now

for

a hexagonal net

A+ C E
we would
89

E = 3A,C =

have

And

0!

2A and hence A

sure enough,

we

some

meshes of the net of Aulonia are not

of the

hexagons but pentagons.

From the densest packing of
plane

+C-E

see that

let

us

now

circles in a

pass to the densest packing

of equal spheres, of equal balls in space.

start

We

with one ball and a plane, the "hori-

zontal plane" through

packing

this ball will

("like the seeds in a

its

center.

In densest

touch on twelve others

pomegranate," as Kepler

says), six in the horizontal plane, three below,

and three above.

impossible

the

If

balls

mutual penetration
in

is

arrangement,

this

under uniform expansion around their fixed

centers, will change into rhombic dodecahedra that fill the whole space. Mark that
the individual dodecahedron is not a regular
solid
whereas in the corresponding twodimensional problem a regular hexagon re-

sulted!

The

bees' cell consists of the lower

dodecahedron with the six

prolonged as to form a
hexagonal prism with an open end. Much
has been written on this question of the
geometry of the honeycomb. The bee's
strange social habits and geometric talents
could not fail to attract the attention and
half of such a
vertical

sides

excite the
servers

and

so

admiration of their
exploiters.

"My

human

ob-

house," says

the bee in the Arabian Nights, "is constructed

according to the laws of a most severe archi-

and Euclid himself could learn from

studying the geometry of my cells."
Maraldi
tecture;

The arrangement is uniquely determined only if

one requires the centers to form a lattice. For the
definition of a lattice see p. 96; for a fuller discussion of the problem: D. Hilbert and S. Cohn1

Vossen, Anschauliche Geometrie, Berlin,

arid

H.

Minkowski,

Diophantiscke

Leipzig, 1907, pp. 105-111.

90

932, pp. 40-41

Approximationen

be the first to have carried

measurements,
and he found
out fairly exact
the cell have
bottom
rhombs
of
that the three
in

1712 seems

to

an obtuse angle a of about 110 and that the

angle /3 they form with the prism walls has
the

same

value.

He

asked

himself

the

geometric question what the angle a of the

rhomb has to be so as to coincide exactly

He finds a = /3 =
with the latter angle 0.
109 28' and thus assumes that the bees had
When prinsolved this geometric problem.
ciples of

minimum were

introduced into the

study of curves and into mechanics, the idea

was not farfetched that the value of a is deter-

mined by the most economical use of wax;

with every other angle more wax would be
needed to form cells of the same volume.
This conjecture of Reaumur was confirmed
by the Swiss mathematician Samuel Koenig.
Somehow Koenig took Maraldi's theoretical
value for the one he had actually measured,
and finding that his own theoretical value
based on the minimum principle deviated
from it by 2' (owing to an error of the tables
he used in computing \/2) he concluded
that the bees commit an error of less than
2' in solving this minimum problem, of which

beyond the reach of clasrequires the methods of

and
sical geometry
The ensuing discusLeibniz.
Newton and
Academy
was summed up
sion in the French
he says that

it lies

by Fontenelle as Secretaire perpetuel in a

famous judgment in which he denied to the
bees the geometric intelligence of a Newton

and Leibniz but concluded that in using the

highest mathematics they obeyed divine
guidance and command. In truth the cells
are not as regular as Koenig assumed, it
would be difficult to measure the angles even
within a few degrees. But more than a
91

hundred years

later

Darwin

bees' architecture as "the

known

still

spoke of the

most wonderful of

and adds: "Beyond

instincts"

this

stage of perfection in architecture natural

selection

now

(which

has replaced divine

guidance!) could not lead; for the

the hive-bee, as far as
lutely perfect in

we can

see,

comb
is

of

abso-

economizing labor and wax."

hedra gives even a better economy of surface

in relation to volume than the plane-faced

rhombic dodecahedra. I am inclined to

believe that Lord Kelvin's configuration gives
the absolute minimum; but so far as I know,
has never been proved.

this

Let us now return from the three-dimensional space to the two-dimensional plane

FIG.

57

and engage in a more systematic investigation of symmetry with double infinite rapport.
First we have to make this notion precise.
As was mentioned before, the translations,
the parallel displacements of a plane form a

group.

translation a can be completely

described by fixing the point A' into which

carries a given point A.

it

The

translation

When

one truncates the six corners of an

octahedron in a suitable symmetric fashion
one obtains a polyhedron bounded by 6
squares and 8 hexagons. This tetrakaidekahedron was known to Archimedes and rediscovered by the Russian crystallographer
Fedorow. Copies of this solid obtained by
suitable translations are capable of filling the

whole space without overlappings and gaps,

rhombic dodecahedron does (Fig.
In his Baltimore lectures Lord Kelvin
56).
showed how its faces have to be warped and

or vector BB'

AA'

BB'

if

is

is

the

same

parallel to

AA' and of the same

length.

The composition

usually

denoted

is

by

the

of translations

B and
carries A

into
b

b.

into

first

If a carries

b carries

into

C and may

is

Thus

+.

sign

the translation resulting from

carrying out a and then

point

as the translation

the

then

thus be

just as the

edges

curved

minimal

area.

to

If this

of space into equal

92

fulfill

and

is

the

condition

done the

AC

indicated by the diagonal vector

parallelogram

ABCD.

Since here

in the

AD = BC

of

partition

parallel tetrakaideka-

and DC = AB = a (Fig.
the commutative law a + b =

57),

we have

a for the

93

composition of translations, or, as one also

This addi-

says, for the addition of vectors.

tion of vectors

which two
a

-f-

forces

such that a
2a,

The

have the identity or

carries every point into

a) =

a-fa,

is

for,

l)a

vector b

(na

)+

^a

and

a
is

a, etc.
is

zero, or

Oa

the unique solution of

a fraction m/n with integral

is

Hence

it is

clear

and denominator n, as for instance

or ${3; and then also by continuity what it means for any real number X,

whether rational or irrational.

Ci, C2 are linearly independent

Two

The plane

because every vector

uniquely
x\t\

such

as

*2e2 in

The

in-

coefficients

X2 are called the coordinates of r with

respect to the basis

definite point
Ci,

can be represented
linear combination

C2.

C2)

we can

(eg,

C2).

and

it

there are infinitely

more necessary

the

is

and hence we can distinguish them

The names
only by attaching labels to them
we employ happen to be pairs of numbers
.

(x h Xi).

tion of any transformations

sociative law

(a

b)

the as-

satisfies

(b

c).

For the multiplication of vectors a, b,

one has the law
by real numbers X, ix,
.

X( M a)

and the two

(Xm)o

(X

distributive laws

+ m)o =
+ b) =

X(a

One must

+
+

(Xa)
(Xa)

(mo),

(Xb).

how

ask oneself

the coordinates

an arbitrary vector r change

one passes from one vector basis (eg, e 2 )
(xi,

of

*)

The vectors

another

(ci, e'2 ) .

sible in

terms of d,

(1)

After fixing a

ascribe to every point

OX

94

of them;

ci

fluCi

c 2,

and

02^2,

ci, c'2

as
to

are expres-

vice versa:
e'2

01261

C2

a'12 e[

+ ^22*2

a-ifa

as origin (and a vector basis

two

= x\ti ^262, and

coordinates x h x 2 by
vice versa these coordinates xi, xz determine
the position of

system" (0,

C 2 ).

eg,

way because

two-dimensional

terms of two fixed linearly

dependent vectors d,
x\,

is

many

in

and

This has to be done in a

recognize them.
systematic

distinguish

vectors

if no linear
combination of them x&j + #262 is the null
vector
unless the two real numbers x\ and

*2 are zero.

we can

a plane by which

what

numerator

to the points

nothing but to give names

o.

Xa means

you with

Besides the commutative law, the addition

as a matter of fact, the composiof vectors

if

to torture

alike,

the equation 3b

a.

had

namely

expressed in the formulas

is

sorry that

obvious

n, positive,

am

as points, unlike persons, are all completely

general rule by which the multiple na

negative,

The

inverse

its

It

0.

stand

defined for every integer

(n

form a resultant

translation a has

4a,

3a,

a-f-a,

We

which

and every

what

b unite to

forces.

null vector

a,

but the law by

else

according to the so-called paral-

lelogram of

itself,

nothing

is

The
these elements of analytic geometry.
purpose of this invention of Descartes' is

relative to the "coordinate

and
(V)

Ci

= a'n

t'i

n t'2

a'

Represent the arbitrary vector

in terms of

the one and the other basis:

?

xid

-V2C2

*iei

x'2 t'2

95

By

substituting (1) for

one

or

t'2

(1') for e t , e 2

the coordinates xi, x 2 with

that

finds

respect to the

t[,

the coordinates

connected with

basis are

first

* 2 in the second system

x[,

by the two mutually inverse "homogeneous

linear transformations"
xi

(2)
f

r,/\

(2

xx

The

+
= a n xi +
= a n x'i

ai2* 2 ,

x2

a 12 * 2 ,

*2

1,1

+ 022*2;
1,1
= a xi + a n x

a 2 i*i
2l

coordinates x vary with the vector

y;

but

the coefficients
oi 2\

Y*Ms

022/

a 'n,

<*[ 2

It is

easy to see under what

circumstances a linear transformation like

has an inverse, namely,

modul

so-called

011^22

if

and only

as

of vectors a

namely

*!

(1) addition

tor a by a number X, (3) the operation by

which two points A,B determine the vector

Ci,

basic for metric geometry.

length of an

the

arbitrary

The

square of

vector

is

quadratic form
11*1

(3)

2guX 1 X 2

+ 22*2

as

tent

of Pythagoras'

its

96

value

is

(3)

is

theorem.

The

positive-definite,

positive for

using the original basis (d,

e 2)

frame of reference throughout, you

see

x'itz,

that the vector with the coordinates *i, *2

goes into that with the coordinates x'u x 2

where

+x

metric

2 e2

x[t x

+x

2 e2,

and hence
x[

(4)

OllXl

012*2,

*J

[formulas (2) with the pairs

^21*1

x 2 ),

(*i,

(*i,

* 2)

replaced by points the homoare replaced

transformations
geneous linear
Let
ones.
non-homogeneous
throughout by
If vectors are

*2),

2)

(*i,

be the coordinates of the

in two coordinate

same arbitrary point

systems (0;

d,

c 2 ),

{0'\

e(,

c 2 ).

Then we

have

namely

any values of the

+ 022*2

interchanged].

*'

its

ground form

X2

(*i,

coordinates x h x 2 with constant coefficients gi U g i2 , g 22

This is the essential conof

c2

x t t[

AB, and concepts logically defined in terms

of them, we do affine geometry.
In affine
geometry any vector basis d, e 2 is as good
as any other.
The notion of the length |r|
of a vector J transcends affine geometry and
is

'2

passes into the Cartesian basis e lf

* 2 e 2 g es into
then the vector r = X1C1
x 2 c , and if you write this as
x ci

basis

different

is

(2) multiplication of a vec-

b,

/2

X\

If by a
algebraic expression of a rotation.
Cartesian
the
rotation around the origin

if its

use no other concepts than

those introduced so far,

xiei

we

homogeneous linear transformation (2), (2 )

which leaves the form x\ + x\ unaltered,

0.

As long

There

0.

coordinate systems are equally admissible.

Transition from one to the other is affected
by an orthogonal transformation, i.e. by a

r'

from

x\; they consist of two muvectors d, l-i of equal

perpendicular
tually
all Cartesian
geometry
metric
In
length 1.

c2

012021

x2

But with a slight modification such a transformation may also be interpreted as the

a 22/

\0 2 i,

x 2 except for*i

expression x\

/in,

*i,

exist special coordinate systems, the Cartesian

ones, in which this form assumes the simple

X\

are constants.

(2)

variables

OX =

*id

*2d,

O'X =

*'xCi

+ 44
97

and

since

OX =

(5)

in

+ O'X:
= anx'i + a x +

xi

00'

i2

any two such coordinate systems linked by

an orthogonal transformation

(i

bi

1,

2)

are, as

we

shall

Hence what

say, orthogonally equivalent.

Leonardo had done can now be formulated

where we have set 00' = Mi + b 2 t 2
The
non-homogeneous differ from homogeneous
transformations by the additional terms bi.
.

The mapping
(6)

x'i

anxi

a i2 x 2

bi

(i

1,

2)

carrying the point (x h x 2 ) into the point

(x[,
x 2 ) expresses a congruence provided
the

homogeneous part of the transformation

(4)

x'i

giving

the

vectors,

is

a (1 xi

^2*2

(*

1, 2),

corresponding mapping of the

(Here of course the

orthogonal.

coordinates refer to the same fixed coordinate

system.)
Under these circumstances we
call

non-homogeneous transforma-

also the

tion

orthogonal.

tion

by the vector

In

particular,

(b h b 2 )

is

transla-

expressed by the

transformation

ha

return

finite rotation

such that (1) any two of the groups in

his list are orthogonally inequivalent, and
(2) any finite group of orthogonal transformations is orthogonally equivalent to a
tions

group occurring

He made up

*2
to

x-i

b2

2,

advantages will presently become evident.

The symmetry of ornaments is concerned with

its

discontinuous groups of congruent mappings

If such a group A contains
of the plane.
translations

of

many

Cartesian

vector

discontinuity:

one of the
is

reflection axes.

expressed

translations na (n

we

0,

1,

replace finiteness

it

requires that there

is

no

which the numbers

an

in

basis.

is not so for the groups

here we
n
normalize the algebraic expression by introducing as the first basic vector Ci one that

lies in

0)

gives rise to

other words, there is a positive number e

such that any transformation (6) in our group

This

of rotations

to postulate

translation

operation in the group arbitrarily close to

In
the identity, except the identity itself.

algebraic expression of the operations

choice of the

of a

iteration

Therefore

one of the groups C does not depend on

the

group

terms of the

Cartesian coordinate system as a group of

orthogonal transformations.
Its expressions

98

would be absurd

from the identity

(different

it

for

finiteness,

for

The

say briefly:

of orthogonally in-

list

groups of orthogonal transThis

seems an unnecessarily informations.
stating
a simple situation; but
volved way of

by

Leonardo's table of

Z) 3

We

list.

groups in the plane,

Dh D

in his

a complete

finite

equivalent

Ci, C2, Cz,

(7)

of groups of orthogonal transforma-

list

infinitely

*1 ** *l

We now

up a

He made

language as follows:

in algebraic

lie

between

which

all

012,

1,

an,

a 22

-e and +e
these

is

1,

b
b 2;)

the identity (for

numbers are

zero).

The

translations contained in our group form a

discontinuous group

of translations.

For

such a group there are three possibilities:

Either it consists of nothing but the identity,
the null vector 0; or

all

the translations in

99

the group are iterations xt of one basic trans-

0,

lation e 9* a (#

1, 2,

or these

);

translations (vectors) form a two-dimensional

namely

lattice,

tions X\t\

of

two

The

consist of the linear

combina-

# 2 e 2 by integral coefficients

linearly independent vectors

third

case

the vectors

that

is

rapport in which

we

of double

call

x2
C2.

infinite

Here

are interested.

form what we

Ci,C2

Xj,
Ci,

lattice

Choose a point
as origin; those
points into which
goes by all the translations of the lattice form a parallelogramatic
basis.

lattice of points (Fig. 58).

integral coefficients that has

same type

is

an inverse of the
by the mathe-

called unimodular

maticians; one easily sees that a linear trans-

unimodular

if

coefficients

integral

with

formation

and

only

its

if

is

modul

ai22i equals +1 or 1.
order
to determine all possible disconIn
groups
of congruences with double
tinuous
we now proceed as follows.
rapport
infinite

fliit22

We

as origin

choose a point

the translations in

and represent

A by

our group

the lattice

which they carry the point O.

I
our group may be considof
Any operation
followed by a
around
rotation
ered as a
of points into

The

translation.
carries

the

the rotary part, then

first,

lattice

into

itself.

Moreover
and

these rotary parts form a discontinuous,

hence finite group of rotations T = {A}.

In

the terminology of the crystallographers it is

symmetry
this group which determines the
class of

groups in Leonardo's table

FIG. 58

Cn D n

(8)

(n

(7),

1, 2, 3,

).

choice of the lattice basis for a given lattice

but one whose operations carry the lattice L

This relationship between the
into itself.

arbitrary?

rotation group

To what

we

extent,

If e 1 ,C 2

is

will ask at once,

is

the

another such basis we

must have
(1)

tx

where the

fluCi
atj

fl2iC 2 ,

e'2

integers.

But

-f-

are

= and

+ 02262

also the coeffi-

cients in the inverse transformation (1')

must

be integers, otherwise
tute a lattice basis.
gets

el, c'2 would not constiFor the coordinates one

two mutually inverse linear transformawith integral coefficients

tions (2), (2')

:

(2")

A
100

T must be one of the

the ornament.

(*'
\a 2 i,

*A
022/

homogeneous

and

fa
\a
21 ,

j*\
a
22

linear transformation with

T and the lattice L imposes

certain restrictions on both of them.
As far as T is concerned, it excludes from
the table
4,

6.

all

values of n except n

Notice that n

cluded values

5 is

1,

among

2,

3,

the ex-

Since the lattice permits the

by 180, the smallest rotation leaving

part of 180,
it invariant must be an aliquot
form
of
the
or
rotation

360 divided by 2 or 4 or 6 or 8 or

We must show that the numbers from 8 on are

impossible.

A among

Take the

all lattice

= 8 and let
9*0 be one that

case of n

points

101

A
nearest to

is

Then

(Fig. 59).

the whole

which arises
octagon A = Ay, A2, A3,
from A by rotating the plane around

through

% of the

angle again and again

full

is

may

Since OAi,

points.

called quincuncial, thinking

OA*

are lattice vectors their difference, the vector

of the quincunx

A1A2, must also belong to our lattice, or the

figure,

although the

ing to

do with the number

point

be a

However

contradiction, since

A =

OB = A A

determined by

lattice point.

should

this leads to a

than

nearer to

is

A\, indeed the side A\Ai of the regular

octagon

Hence

than its radius QA\.

group T we have only these 10

smaller

is

for the

possibilities:
FIG. 59

(9)

d, C2 C 3
,

C4,

Dh D D D
2,

3,

4,

D,.

lattice

points are the corners O and the centers

(It was an arrangement
of the rectangles.
of trees in such a lozenge lattice which

Thomas Browne

consists of lattice

The

serve as lattice basis.

as

its

elementary

lattice in fact
5.)

has noth-

Shape and

fundamental rectangle or rhomb

size of the

are arbitrary.
After having found the 10 possible groups V
of rotations and the lattices L left invariant

by each of them, one has to paste together a

T with a corresponding L so as to obtain
the full group of congruent mappings.
Closer investigation has shown that, while
10 possibilities for T, there are
essentially different possibilities
17
exactly
Thus
for the full group of congruences A.
there are

It is

easy to see that for each of these groups

there actually exist lattices

left

invariant by

the operations of the group.

Clearly, for

and

& any lattice

will do,

any lattice is invariant under the identity

and the rotation by 180. But let us consider D\, which consists of the identity and
the reflection in an axis / through 0.
There
are two types of lattices left invariant by
this group: the rectangular and the diamond
for

lattices
is

II

(Fig.

60).

The

rectangular lattice

obtained by dividing the plane into equal

rectangles along lines parallel

dicular to

/.

The

and perpen-

corners of the rectangles

are the lattice points.

natural basis for

the lattice consists of the two sides

issuing

whose
FIG.

60

from

left-lower corner

lattice consists of

the plane

is

C2

is

0.

The diamond

equal rhombs into which

divided by the diagonal lines of

the rectangular lattice.

fundamental rhomb the

102

Ci,

of the fundamental rectangle

The two
left

sides of the

corner of which

^here are 17 essentially different kinds of

symmetry possible for a two-dimensional

ornament
Examples
found

double

with
for all

among

the depth

the

decorative

One can
of

inventiveness

rapport.

17 groups of symmetry are

antiquity, in particular

ornaments.

infinite

among

imagination

geometric

is

of

the Egyptian

hardly overestimate

in

reflected

Their construction

patterns

far

these

and

patterns.

from being mathe-

ornament conform the oldest piece of

To be
higher mathematics known to us.
sure, the conceptual means for a complete abmatically
tains

trivial.

The

art of

implicit

in

stract formulation

of the underlying prob-

namely the mathematical notion of a

group of transformations, was not provided
before the nineteenth century, and only on

lem,

is one able to prove that the 17

symmetries already implicitly known to the

this basis

103

Egyptian craftsmen exhaust

all possibilities.

Strangely enough the proof was carried out

now
The Arabs fumbled

only as late as 1924 by George Polya,

teaching at Stanford. 2

around much with the number 5, but they

were of course never able honestly to insert
a central symmetry of 5 in their ornamental
designs

of double

infinite

rapport.

They

various deceptive compromises,

however.
One might say that they proved experimentally the impossibility of a pentagon
in an ornament.
tried

was clear what we meant by

saying that there are no other groups of rotations associable with an invariant lattice than
the 10 groups (9), our assertion of the existence of not more than 17 different ornamental
groups calls for some explanation. For instance, if T = C\ then the group A is one consisting of translations only; but any lattice
is possible here, the fundamental parallelogram spanned by the two basic vectors of the
lattice may be of any shape and size, we have

Whereas

In arriving at the

of possibilities.

we count

all

number

these as one case only; but

need my analytic
geometry.
If we look at our plane in the
light of affine geometry it bears two strucwith what right?

tures:

(i)

Here

the metric structure by

which every

vector y has a length the square of which

expressed in terms of coordinates by a

is

positive

definite

quadratic

metric ground form;

owing

(ii)

to the fact that the

form

Cf. his paper

symmetrie

"Ueber

der Ebenc,"
60, pp. 278-282.

104

in

which

the metric ground form has a unique normalized expression x\

-f-

x\,

while in the algebraic

continuous manifold

of the

representation

of invariant lattices there remains a variable

But instead of adapting our co-

element.

ordinates to the metric by admitting Cartesian

we

coordinates only,

lattice structure first

could put the

and adapt the coordi-

nates to the lattice by choosing

with

lattice basis,

Ci,

as a

e2

the effect that the lattice

now normalized in a uniquely determined

manner when expressed in terms of the cor-

is

responding coordinates

now

xj,

*2.

Indeed the

whose coordiIn general we cannot do

nates are integers.
both at the same time: have a coordinate
system in which the metric ground form appears in the normalized form x\ -f- x 2 an d
lattice vectors are

those

>

the lattice consists of all vectors with integral

coordinates

xi.

*i,

We now

follow

the

second procedure, which turns out to be

mathematically more advantageous. I con-

one which
morphology.

sider this analysis as

importance for

all

is

of basic

As an example consider once more D%.

is rectangular and the

If the invariant lattice

lattice basis

described
identity

is

chosen in the natural manner

above,

then D\ consists of the

and the operation

the

xi

ornament endows
In the usual

die Analogie der Kristall^eitschr. j.

the metric structure

tesian coordinate systems in terms of

a lattice structure,

the plane with a vector lattice.

2

(3),

first

and thus introduces the Car-

into account

it

the choice in a continuously infinite manifold

17

procedure one takes

Kristallographie

The

xu

x2

X2.

metric ground form can be any positive

form of the special type a x x\ -+- a-ix\. If the

invariant lattice is a diamond lattice and the
sides of the fundamental diamond are chosen

105

basis, then Di contains the

and the further operation

group of linear transformations

as the lattice

any

identity

with real coefficients one

*i

The

x2

*2,

metric ground form

D",

D\

*i-

may

form of the special type a(x\

we now

But instead of Di

tive

quadratic forms

be any positive

+ *l) +

2bx\x 2

obtain two groups

which, though orthogonally, are

no longer unimodularly equivalent, the one

consisting of the two operations with the
coefficient matrices

may

construct posi-

invariant under these

left

How many

transformations. 3

of linear transformations with integral

coefficients,

finite

different,

i.e.

unimodularly inequivalent, finite groups of

linear transformations with integral coeffi-

two variables, exist then? Ten,

namely our old friends (9)? No, there are
more, since we have seen that D\, for instance,
breaks up into two inequivalent cases D\, D v
The same happens to D 2 and D 3 with the
result that there are exactly 1 3 unimodularly
cients

in

inequivalent finite groups of linear operations

(!4)

with integral

the other of

From a mathe-

coefficients.

matical standpoint

this

is

the really interest-

ing result rather than the table (9) of the

10 groups of rotations with invariant

(::>
Two

groups of homogeneous linear trans-

formations are, of course, called unimodularly

equivalent if they both represent the same
group of operations, the one in terms of one,
the other in terms of another lattice basis,

In the lattice-adapted coordinate system

T now appear

geneous linear transformations

(4)

as

homo-

with in-

each carries the

x 2 assume integral values

tegral coefficients aq\ for as

lattice into itself, x[,

whenever integral values are assigned to *i

and x^ The arbitrariness in the choice of
the lattice basis finds its expression by the
agreement to consider unimodularly equivalent

groups of linear transformations as the

same

thing.

all

the

translations
x[

xi

x2

bu

x2

b2

with integral b\,b 2 and no other translations.

This

last step offers little difficulty,

remarks that remain

to

be

made

and the

are better

based on the 1 3 finite groups T of homogeneous transformations which result from cancelling the translatory parts.

So

far only the lattice structure of the plane

has been taken into account.

Of

course,

the metric of the plane cannot be ignored

forever.

And

it

is

here that the continuous

Besides having integral coeffi-

cients the transformations of

will leave

certain positive definite quadratic form (3)

106

last step

ear transformations which contain

i.e. if they change into each other by a unimodular transformation of coordinates.

the operations of

lattices.

one has to introduce the

translational parts of the operations, and one
obtains 17 unimodularly inequivalent discontinuous groups of non-homogeneous linIn a

invariant.

But

restriction;

indeed

this
it

is

really

no additional

can be shown that for

This

is

The proof

a fundamental theorem due to H. Maschke.

is

simple enough: Take any positive quadx\, perform on it each of the

ratic form, e.g. x\

transformations

and add the forms

an invariant positive form.

of our group

thus obtained: the result

is

107

aspect of the problem comes

of the

groups Y

13

there

For each

in.

invariant

exist

Such

form

is

n*i

+ gaa*f-

2gi2*i*2

characterized by

The form

(gn, 12, 22)-

G(x)

its

is

coefficients

not uniquely

determined by T; for instance G(x)

replaced by any multiple

constant

positive

factor

quadratic forms G(x)

left

may

be

G(x) with a real

c.

All

invariant

by the

"cone" of simple nature and of 1, 2, or 3

dimensions.
For instance in the cases D"
and D\ we had the two-dimensional manifolds
of all positive forms of the types a\x\

+ 02*2

2bx&% respectively. The

metric ground form is always one in the
a{x\

-+-

x\)

manifold of invariant forms.

In the

groups

which are

varying

The

description of the ornamental

A we have now

features

of

full

clearly divided those

discrete,

over

discrete feature

is

and those capable

continuous

manifold.

exhibited by represent-

ing the group in terms of lattice-adapted

coordinates and turns out to be one of 17

To each of them
continuum of possibilities
for the metric ground form G(x), from which
the one actual metric ground form must be
picked.
The advantage of adapting the co-

definite distinct groups.

there corresponds a

ordinate system to the lattice rather than to

becomes visible in the fact that

now the variable element G(x) varies over
one simple convex continuous manifold,
the metric

nates the lattice L, which then appears as

the variable element,

ranges over a con-

tinuum that may consist of several parts, as

example of Di showed. The advantage

108

if

one passes from the

[A]

to

full

the clearcut

way

which

in

this distinction

is

carried out.

After

somewhat

these

all

matical generalities

abstract mathe-

am now

going to show

you a few pictures of surface ornaments with

double

wall papers, carpets,

all sorts

and
one

You

infinite rapport.

find

floors,

tiled

them on
parquets,

of dress material, especially prints,

so forth.

Once

one's eyes are opened,

be surprised by the numerous symmetric patterns which surround us in our

daily lives.
The greatest masters of the
will

geometric art of ornament were the Arabs.

The wealth

of stucco ornaments decorating

the walls of such buildings of Arabic origin

as

Alhambra

the

Granada

in

is

simply

overwhelming.
For the purpose of description it is good to
know what a congruent mapping in two
dimensions looks like. A proper motion maybe either a translation or a rotation around a
If such a rotation occurs in our

point 0.

symmetry group and

occurring in

it

all

the rotations around

are multiples of the rotation

by 360/n, we call
or simply rc-pole.

while in terms of the metric-adapted coordi-

the

only

ornamental group A. The splitting

into something discrete and something continuous seems to me a basic issue in all
morphology, and the morphology of ornaments and crystals establishes a paragon by

positive

operations of T form a continuous convex

and

fully revealed

truncated homogeneous group T

the

positive quadratic forms

G( x )

is

values except n

a pole of multiplicity n

We know

2,

3,

An improper congruence
in

a line

/,

or such

group,

is

called

an

/.

no other

are possible.

either a reflection

a reflection

a translation a along

spectively.

is

that

4,

If

combined with

it

occurs in our

axis or gliding axis re-

In the latter case iteration of the

congruence leads to the translation by the

109

symmetry axes. Designs 3', 3a, and 3b

reduce the multiplicity of those poles to 3;
among them 3' is without symmetry axes,
the

3a axes pass through every

in

3b

3-pole, in

only through those (one-third of the whole

number) which had been

The homogeneous groups

six-fold

are

6,

before.

C&, Cz, D%,

respectively, where Z>, D\ are the two

unimodularly inequivalent forms assumed by

D\

in

a lattice-adapted coordinate system.

There now follow some actual ornaments

of Moorish, Egyptian, and Chinese origin.
This window of a mosque in Cairo, of the
fourteenth century (Fig. 62),

is

of the hexaFIG.

62

FIG. 61

vector 2a; hence the gliding vector a must

be one-half of a lattice vector of our group.

The

first

picture (Fig. 61)

is

drawing of
which

the hexagonal lattice, the discussion of

started today's lecture.

symmetry.
2, 3,

and

It

has a very rich

There are poles of multiplicities

6,

indicated in the diagram by dots,

small triangles, and hexagons respectively.

The

vectors joining

tice vectors.

The

two 6-poles are the

lat-

lines in the figure are the

There are also gliding axes, which

have not been shown in the drawing; they
axes.

run mid-way between and parallel to the

axes.
The possible symmetry groups of the
hexagonal type are five in number and are
obtained

by

putting

one

figures 6 or 6' or 3' or 3a or

the six-poles.

Designs 6 and

of

the

simple

3b

into each of

6'

preserve the

multiplicity 6 of these poles, but 6' destroys

no

111

gonal class
trefoil

Z) 6

The elementary

figure

is

knot the various units of which are

interlaced with superb artistry.

interrupted

tracks cross

Almost un-

the design in the

in

axes.

symmetry of the geometric pattern as exby

pressed by a certain group A is reduced
expressed
the coloring to a lower symmetry
by a subgroup of A. A symmetry of the
square class D 4 is exhibited by this (Fig. 64)

back of the alcove in the Sala de

well-known design for brick pavements;

no ordithe amusing thing about it is that

three directions arising from the horizontal

by rotations through

0, 60, 120; the

lines of these tracks are gliding axes.

can easily discover

lines that are

mid-

You

ordinary

Such axes are absent from this

Azulejos ornament (Fig. 63) decorating the

FIG.

Granada. The group is 3' or

6', according to whether or not one takes
account of the colors. This is one of the
the
finer tricks of the ornamental art that

Alhambra

Camas of the

63

FIG.

FIG. 64

nary axes, only gliding axes, pass through

Of
the 4-poles (one of which is marked).

same symmetry is the Egyptian ornament

shown next (Fig. 65), as well as the two following Moorish ornaments (Fig. 66). A
monumental work on our subject is Owen
which
Jones' Grammar of ornaments, from
the

some
more

Of a
of these illustrations are taken.
special character is the Grammar of

by Daniel Sheets Dye, which

deals with the lattice work the Chinese use for
I reprothe support of their paper windows.
Chinese

lattice

(Figs. 67 and 68) two characterdesigns from that volume, one of the

duce here
istic

hexagonal and one of the

Z) 4 -type.

113

65

FIG. 68

wish

could analyze some of these ornaBut a prerequisite for such

ments in detail.
an investigation would be an
braic

of

description

groups.

What

this

the

lecture

explicit alge-

17

ornamental

aimed

at

was

more a clarification of the general mathemorphology

matical principles underlying the
of

ornaments (and

crystals)

than a group-

ornaments.
theoretic analysis of individual
me from
prevented
has
Shortness of time
and
abstract
the
sides,
justice to both

doing

the concrete.

tried to explain the basic

some
mathematical ideas, and I showed you
indiI
them
pictures: the bridge between
step
it
over
you
lead
cated, but I could not
by
FIG.

114

step.

67

115

CRYSTALS

THE GENERAL MATHEMATICAL

IDEA OF

SYMMETRY

CRYSTALS

THE GENERAL MATHEMATICAL

SYMMETRY

IDEA OF
In the last lecture

we

considered for two

dimensions the problem of making up a complete list (i) of all orthogonally inequivalent
finite groups of homogeneous orthogonal
transformations,

have invariant
larly

(ii)

of all such groups as

lattices, (iii) of all

unimoduhomo-

inequivalent finite groups of

genous transformations with integral coefficients, (iv) of all unimodularly inequivalent

groups of non-homogeneous
which contain the
translations with integral coordinates but no

discontinuous

transformations

linear

other translations.

Problem

(i)

was answered by Leonardo's

list

Dn

C,

(n

I, 2, 3,

)>

by limiting the index n in it to the values

= 1, 2, 3, 4, 6. The numbers hi, k n hm,

(ii)

hj V

of groups in these four

lists

turned out to

be
17

oo, 10, 13,

respectively.
is

doubtless

these

The most important problem

One could have posed
(iii).

same four questions

for the

one-dimen-

sional line rather than the two-dimensional

plane.

There the answer would have been

very simple, and one would have found all

Inthe numbers h u k n h ux h JV equal to 2.
,

deed in each of the cases

(i),

(ii),

(iii)

group consists either of the identity

alone, or of the identity
x'

and the

x'

the

reflection

-x.

119

But what we are intending now is not to

descend from 2 to 1, but ascend from 2 to
3 dimensions.
All finite groups of rotations
in 3 dimensions were listed at the end of the
second lecture;
C,

D'n

repeat them here:

A:

List

C 2n Cn

C,

D'n

D'2n D'n

= 2, 3,

(n

W,

T,

P;

1, 2, 3,

D'n C n

W,

T,

WT.

P;

one requires that the operations of the

group leave a lattice invariant, only axes of
rotation of multiplicity

And

by

2,

6 are admisour table re-

3, 4,

this restriction

duces to the following

List B:

C\, C?, Cz, C*, Ct;

D'2 Z/3 D\, D'6

,

Ci,

C2

>;,

D'3 D'4

C4

C3,

Ce;

D[;

D't D'2

D'6 D'3

D'2 C2
T,
It

W,

T,

D'3 C 3

D\C h Dfa;

invariant

One

easily con-

country L. Dickson

this

much

effort.

decorate surfaces with

flat

ornaments;

gone in for solid ornaments.

But they are found in nature. The arrangements of atoms in a crystal are such patterns.
The geometric forms of crystals with their
plane surfaces are an intriguing phenomenon
However the real physical symof nature.
crystal
is not so much shown by
metry of a

outward appearance as by the inner

In

lattices.

entire space.

three

di-

its

Its

macroscopic symmetry

expression in a group Y

will

of rotations.

orientations of the crystal in space

Only such

are physically indistinguishable as are carried

into each other by a rotation of this group.

which

in general propa-

mensions the numbers h u k u km, hiv have

For example,

the values

gates with different velocities in different di-

*
In

may

its

32, 70, 230.

rections

algebraic formulation our problem

be posed for any

*i, X2,

number m

of variables,

x m instead of just 2 or

3,

and

the corresponding theorems of finiteness

have been proved. The methods are of the
greatest mathematical interest.
The com-

bination

"metric plus lattice"

lies

at

the

bottom of the arithmetical theory of quadratic

forms which, inaugurated by Gauss, played
a central part in the theory of numbers

120

in

have spent

art has never

find

vinces oneself that each of these 32 groups

possesses

We

all,

physical structure of the crystalline substance.

Let us suppose that this substance fills the

W, WT.

contains 32 members.

of algebraists,

its

C2C1, C4C2, CzCa;

refined arithmetic

more

theory of so-called algebras or hypercomplex

number systems, on which the last generation

above

If

sible.

Hermite, and more recently Minkowski

and Siegel have contributed to this line of
The investigation of ornamental
research.
symmetry in m dimensions is based on the
results won by these authors and on the

let,

algebraic and the

(n

Dirich-

throughout the nineteenth century.

in

light,

the

crystalline

medium,

will

propagate with the same speed in any two

directions that arise from each other by a

So for all other

For an isotropic me-

rotation of the group T.

physical properties.

dium
but

the group V consists of

for

number

a crystal
of

it

is

rotations,

rotations,

all

made up

of a finite

sometimes

nothing but the identity.

even

of

Early in the his-

tory of crystallography the law of rational

indices was derived from the arrangement of

121

'

the plane surfaces of crystals.

It led to

the

hypothesis of the lattice-like atomic structure

This hypothesis, which explains

of crystals.

now been

the law of rational indices, has

definitely

confirmed by the Laue interference

patterns,

which are

graphs of

crystals.

More

exactly the hypothesis states that the

discontinuous group
carry

into

number

contains

itself

the

which

in

our

maximum

of linearly independent transla-

Incidentally this hypothesis can be

tions.

reduced

to

much

simpler

Atoms which go over

operation of

The

of congruences

arrangement of atoms

the

crystal

X-ray photo-

essentially

A may

requirements.

into each other

by an

be called equivalent.

equivalent atoms form a regular point-

set in the sense that the set

is

carried into

by every operation of A and that for

any two points in the set there is an operation
of A which carries the one into the other.
Speaking of the arrangements of atoms I
itself

to their

refer
fact

positions

the atoms oscillate

in

equilibrium; in

around these

posi-

Perhaps one should take a lead from

tions.

quantum mechanics and

substitute for the

atoms' exact positions their average distribution density: this density function in space
is

invariant with respect to the operations of

The group r =

A.

{Aj

of the rotational,

parts of those congruences that are

of

leaves the lattice

which

arise

of points invariant

from the origin

tions contained in A.

members

The

by the

transla-

resulting 32 pos-

T enumerated in List B correspond to the 32 existing symmetry classes of

crystals.
For the group A itself we have 230
distinct possibilities, as was mentioned above.
sibilities for

See for instance

P.

Niggli,

Geometrische Kristal-

lographie des Diskontinuums, Berlin, 1920.

122

While r

{A} describes the manifest macro-

and physical symmetry, A desymmetry hidprobably

all know on
behind
it.
You
den
photography
of
von
Laue's
what the success
The image of an object
of crystals depends.
certain
wave length will
traced by light of a

scopic spatial

fines the microscopic atomic

be

fairly accurate

only with respect to details

of considerably greater dimensions than that

wave

length, whereas details of smaller di-

mensions are leveled down. Now the wave

length of ordinary light is about a thousand
Howtimes as big as the atomic distances.
ever X-rays are light rays the
of

which

is

wave

length

8
exactly of the desirable order 10~

In this way von Laue killed

two birds with one stone: he confirmed the
lattice structure of crystals, and proved what

centimeters.

had been merely a tentative hypothesis at the

time of his discovery (1912), that X-rays
consist

of shortwave

light.

portraits of the atomic pattern

grams show are not

Even
which

so,

the

his dia-

likenesses in the literal

By observing a slit whose width is

only a few wave lengths, you obtain a somewhat contorted image of the slit made up by
In the same sense these
interference fringes.
patterns of
interference
Laue diagrams are
able
to comYet one is
the atomic lattice.
sense.

pute from such photographs the actual arrangement of atoms, the scale being set by

wave length of the illuminating X-rays.

Here are two Laue diagrams (Figs. 69 and

the

70), both of zinc-blende

paper

(1912);

the

from Laue's original

pictures

are

taken

in

such directions as to exhibit the symmetry

around an axis of order 4 and 3 respectively.

Whereas

in

the oral lecture

could show

various three-dimensional (enlarged) models

of the actual arrangement of atoms, a photo-

123

FIG. 71
FIG. 69

graph of one such model

for this printed text:

of

it

(Fig. 71)

must

represents a small part

an Anatas crystal of the chemical constitu-

TiO%\ the light balls are the

dark ones the 0-atoms.
tion

In spite of

X-ray
is

suffice

all

Ti-,

the

contortion which mars our

likenesses, the

symmetry of the

faithfully portrayed.

This

is

crystal

a special case

of the following general principle: If conditions

which uniquely determine

possess certain symmetries,

will

exhibit

the

same

Archimedes concluded a
weights balance

FIG. 70

124

in

their effect

then the effect

symmetry.
priori

scales

Thus

that equal

of equal

arms.

Indeed the whole configuration is symmetric

with respect to the midplane of the scales,
and therefore it is impossible that one mounts
while the other sinks. For the same reason

125

we may

be sure that in casting dice which

are perfect cubes, each side has the

same

%. Sometimes we are thus enabled

make predictions a priori on account of

chance,
to

symmetry

for special cases, while the general

law of equilibrium
for scales with arms of different lengths, can
only be settled by experience or by physical
principles ultimately based on experience.
As far as I see, all a priori statements in
physics have their origin in symmetry.
To this epistemological remark about
symmetry I add a second. The morphological laws of crystals are today understood
in terms of atomic dynamics: if equal atoms
exert forces upon each other that make possible a definite state of equilibrium for the
atomic ensemble, then the atoms in equicase, as for instance the

librium will of necessity arrange themselves

The nature

in a regular system of points.

composing the crystal determines

under given external conditions their metric
disposition, for which the purely morphoof the atoms

The examples

exochen.

chemistry, and

nature and nurture,

continuous; and

for the

manner

determines

the

of

its

growth, and this in turn

peculiar shape

it

under the influence of environmental

No wonder

assumes
factors.

then that crystals actually occur-

will not
in

is

It

is

and

constitution

environment.

Whether water, whose molecule has a

chemical

constitution,

is

solid,

definite

liquid,

or

characteristic

the

most convincing way.

deny that the general prob-

need of further epistemological

high time for

me now

to close the dis-

this last lecture is to show the principle of

symmetry at work in questions of physics
and mathematics of a far more fundamental
nature, and to rise from these and its previous

applications to a final general statement of

What

itself.

the theory of relativity

symmetry was

has to do with

briefly explained in the first

before one studies geometric forms

in space with regard to their

symmetry one

the structure of space itself

must examine
under the same aspect. Empty space has a
very high grade of symmetry: every point is
like

any other, and

intrinsic

physical objects usually are the

how such

cussion of geometric symmetries dwelling in

ornaments and crystals. The chief aim of

directions.

of

seen

clarification.

ring in nature display the possible types of

teristics of

for

features of crystals in a

But
lem

symmetry in that abundance of different

forms at which Hans Castorp on his Magic
Mountain marvelled. The visible characresults

we have

can be carried out

lecture:

the crystal's physical behavior, in particular

some way bound up

between discrete and

in

splitting into the discrete and the continuous

of the crystal lattice

also responsible for

is

with the distinction

the principle

is

of crystallography,

genetics cause one to suspect

that this duality, described by the biologists

as that of genotype and phenotype or of

summed up in the 230

groups of symmetry A had still left a continuous range of possibilities. The dynamics
logical investigation

126

vaporous depends on temperature. Temperature is the environmental factor kat'

difference
I

told

at a point there

between

the

is

no

several

you that Leibniz had

given the geometric notion of similarity this

philosophical twist: Similar, he said, are

two things which are indiscernible when each

considered by itself. Thus two squares in
the same plane may show many differences
when one regards their relation to each other;

is

127

one may be inclined by 34 against the sides of the other.

for instance, the sides of the

But

if

each

statement

taken by

is

any objective

itself,

made about one

the

But

automorphisms.

vealed that

among
has

physics

re-

an absolute standard length

is

will hold for the

built into the constitution of the atom, or

and

rather into that of the elementary particles,

other; in this sense they are indiscernible

hence similar. What requirements an objective statement has to meet I shall illustrate

by the meaning of the word "vertical."

Contrary to Epicurus we moderns do not

in particular the electron with

its

definite

charge and mass. This atomic standard

length becomes available for practical meas-

vertical to be

urements through the wave lengths of the

spectral lines of the light emitted by atoms.
The absolute standard thus derived from

see in

nature

consider the statement about a line that

it

it is

an objective one, because we

an abbreviation of the more complete

statement that the line has the direction of

gravity

a certain

at

point

P.

Thus the

gravitational field enters into the proposition

and moreover there

as a contingent factor,

enters into

it

an individually exhibited point

P on which we
strative act as

lay the finger by a

is

here, now, this.

Hence Epicurus'

direction of gravity

place where
Stalin lives,

demon-

expressed in words like

shattered as soon as

it
is

is

belief

we have

concerned,

followed Helmholtz in adopting as

the one basic objective relation in space that

At the beginning of the

of the group of congruent transformations, which is contained
as a subgroup in the group of all similarities.
Before going on I wish to clarify a little further
the relationship of these two groups.
For
of congruence.

there
tivity

is

we spoke

the disquieting question of the rela-

of length.

In ordinary geometry length

is

relative:

building and a small-scale model of

it

platinum-iridium

the

of

meter bar kept in the vaults of the Comite

International des Poids et Mesures in Paris.
described
I think the real situation has to be
Relative to a complete system
of reference not only the points in space but

as follows.

numbers.

I live and at the place where

and that it can also be changed

second lecture

standard

better than the conven-

is

different here at the

is

much

also all physical quantities

realized that the

In concreto, as far as geometry

tional

itself is

I,

by a redistribution of matter.
Let these brief remarks suffice here instead
of a more thorough analysis of objectivity.

128

similar; the dilatations are included

are

Two

systems

can be fixed by

of

reference

are

both of them all

universal geometric and physical laws of
nature have the same algebraic expression.
The transformations mediating between such
equally admissible systems of reference form

equally admissible

if

in

the group of physical automorphisms; the laws

of nature are invariant with respect to the

transformations of this group.

It

that a transformation of this group

is

is

a fact

uniquely

determined by that part of it that concerns the

coordinates of space points. Thus we can
speak of the physical automorphisms of space.
Their group does not include the dilatations,
because the atomic laws fix an absolute
length, but it contains the reflections because
no law of nature indicates an intrinsic difHence the
ference between left and right.

group of physical automorphisms is the group

of all proper and improper congruent mapIf we call two configurations in space
pings.

129

congruent provided they are carried over

into each other by a transformation of this

question, to deduce, as he said, the absolute

motion of bodies from their differences, the

group, then bodies which are mirror images

observable relative motions, and from the

of each other are congruent.

forces acting

think

it

is

But although

bodies.

of rigid bodies, for reasons similar to those

which induce the physicist to substitute the

thermodynamical definition of temperature

succeed in objectively distinguishing rest of

a mass point from all other possible motions,

an ordinary thermometer. Once the

group of physical automorphisms = congruent mappings has been established, one

but only motion in a straight line with uniform velocity, the so-called uniform transla-

necessary to substitute this definition of con-

gruence for that depending on the motion

for

may

define geometry as the science dealing

with

the

between

of congruence

relation

spatial figures,

and then the

geometric auto-

morphisms would be those transformations of

space which carry any two congruent figures

tion,

from

all

other motions.

the statement that

Einstein, people said yes.

conviction

is

All these considerations are deficient in one

happen not only

the world

space but in space and time;

in

spread out not as a three- but

as a four-dimensional continuum.
The symis

basis of this

happening at the moment

when they observe it. But the foundation
of this belief was long ago shattered by Olaf
Roemer's discovery that

respect: they ignore that physical occurrences

The

obviously people's habit of con-

of physical automorphisms
dilatations.

at the

same time (but at different places, say here

and on Sirius) objective meaning? Until

sidering an event as

and includes the

Again, has

two events occur

and one need not be

surprised, as Kant was, that this group of
geometric automorphisms is wider than that

into congruent figures

light

with

propagates not
velocity.

instantaneously

but

Thus one came

to realize that in the four-

finite

dimensional continuum of space-time only

the coincidence of two world points, "herenow," or their immediate vicinity has a
directiy verifiable

meaning. But whether a

scribed by Einstein.

of this four-dimensional continuum in three-dimensional layers of simultaneity and a cross-fibration of one-dimen-

ask, that

sional fibers, the world-lines of points resting

metry, relativity, or homogeneity of this four-

medium was

dimensional

correctly de-

first

Has the statement, we

two events occur at the same place

an objective

We

significance?

are inclined

stratification

in space, describe objective features of the

we

world's structure

understand position as position relative to

the earth on which we spend our life.
But

Einstein did was

to say yes;

is

it

but

rotates

is

now

and that

Newton wrote
turalis

it

we do

clear, if

principia

it

his

told

in

school that

moves about
treatise

mathematica

so,

Even our

sure that the earth rests?

youngsters are

130

upon the

he firmly believed in absolute space, i.e. in

the objectivity of the statement that two
events occur in the same place, he did not

it

in space.

Philosophiae na-

to

answer

this

became doubtful.

What

without bias he colwe have about

physical
evidence
the
lected all
four-dimensional
of
the
the real structure
space-time continuum and thus derived its
this:

group of automorphisms. It is called

the Lorentz group after the Dutch physicist
H. A. Lorentz who, as Einstein's John the
true

131

Baptist, prepared the

relativity.
this

It

way

for the gospel of

turned out that according

of simultaneity nor invariant fibers of

The

rest.

light cone, the locus of all world-points

which a

in

to

group there are neither invariant layers

light signal given at a definite

world-point

"here-now,"

0,

received,

is

and past, into

that part of the world which can still be influenced by what I do at
and the part
which cannot. This means that no effect
travels faster than light, and that the world
divides the world into future

has an objective causal structure described

by these

light cones issuing

Here

from every world

point

0.

down

the Lorentz transformations and

sketch

how

is

not the place to write

special relativity theory with

fixed causal

and

inertial structure

to
its

gave way

invariants (invariant relations, invariant quantities, etc.), and by solving it

for the more important special groups
find

its

whether these groups are known or are not

known to be the groups of automorphisms for
This is
certain fields suggested by nature.
what Felix Klein called "a geometry" in

the abstract sense.

Of
the transformations of this given group.
symmetry one speaks with respect to a subgroup 7 of the total group. Finite subgroups deserve special attention. A figure,
peculiar kind of
i.e. any point-set, has the

symmetry defined by the subgroup y

to general relativity

have become

where these structures

by their interaction
with matter. 2 I only want to point out that
it
is
the inherent symmetry of the fourdimensional continuum of space and time

of 7-

that relativity deals with.

nection between

flexible

We found that objectivity means invariancc

may

Reality

not always give a clear answer

what the actual group of

automorphisms is, and for the purpose of
some investigations it may be quite useful to
replace it by a wider group.
For instance
the question

to

in

plane geometry
such

in

parallel

origin

we may be

relations

as

under
the

is

and projective geometry.

will

One may compare my

Munich

interested only

invariant

or central projections; this

of affine

The mathematician
2

are

prepare for
recent

all

lecture

such

at

the

meeting of the Gesellschaft deutscher

Naturforscher: "50 Jahre Relativitatstheorie," Die
Naturwissenschqften 38 (1951), pp. 73-83.

geometry, Klein said,

defined by a group of transformations, and

investigates everything that is invariant under
is

goes over into itself

with respect to the group of automorphisms.

132

eventualities by posing the general problem,

how for a given group of transformations to

The two

if

it

by the transformations

great events in twentieth century-

physics are the rise of relativity theory and of

Is there also some conquantum mechanics.

quantum mechanics and

Yes indeed. Symmetry plays a

symmetry?

great role in ordering the atomic and molecular spectra, for the understanding of which

quantum physics provide

An enormous amount of empirical

the principles of

the key.

material concerning the spectral lines, their

wave lengths, and the regularities in their
arrangement had been collected before

quantum
this

physics

success

first

success;

in deriving

the law

scored

consisted

its

of the so-called Balmer series in the spectrum

of the hydrogen

atom and

in

showing how

the characteristic constant entering into that

law is related to charge and mass of the elec-

tron
h.

and Planck's famous constant of action

Since

then

the

interpretation

of

the

133

of the corresponding set of

or

and the decisive new features, the electronic spin and Pauli's strange
exclusion principle, were discovered in this

points (Pi,

turned out that, once these foundahad

been laid, symmetry could be of
tions
great help in elucidating the general char-

replaced by

P3 P

1-+3, 2-+

5,

quantum

way.

physics;

It

Approximately, an atom

is

a cloud of elec-

moving around a

trons, say of n electrons,

fixed nucleus in 0.

say approximately, since

the assumption that the nucleus

not exactly true and

even

is

is

fixed

less justified

is

than

treatment of the sun as the fixed center of our

For the mass of the sun

is 300,000 times as big as that of an individual
planet like Earth, whereas the proton, the
nucleus of the hydrogen atom, is less than
2000 times as heavy as the electron. Even
We disso it is a good approximation!
planetary system.

the

tinguish

labels 1,2,

by attaching the

electrons

if

them; they enter into

motion by the coordi-

n to

the laws ruling their

nates of their positions

Ph

Pn

with

respect to a Cartesian coordinate system with

The symmetry prevailing is

First we must have invariance

origin 0.

two-

fold.

with

will,

Thus,

Pn).

example,

for

5 electrons the laws must be

in case of n
unaffected if the points Ph P2 , P3, P\, P& are
,

6,

P2, Pi, P* [permutation

3-2, 4->l,

->

4].

The

permutations form a group of order n\ =

1
2
n, and the second kind of sym

acter of the spectra.

metry

expressed by this group of permutaQuantum mechanics represents the

is

tions.

system by a vector in a
space of many, actually of infinitely many,
state of a physical

dimensions.
either

other,

Two

states that arise

by a

from each

rotation of the

virtual

system of electrons or by one of their permutations, are connected by a linear transforma-

with that rotation or that

the profoundest and

tion associated

Hence

permutation.

most systematic part of group theory, the

theory of representations of a group by linear

comes into play

transformations,

here,

must refrain from giving you a more precise

account of this difficult subject. But here
symmetry once more has proved the clue to a
great variety and importance.

field of

From

from

art,

lography and physics

which

from

biology,
I

crystal-

finally turn to mathe-

must include

all

the

more

be-

respect to transition from one Cartesian co-

matics,

ordinate system to another; this symmetry

cause the essential concepts, especially that

of a group, were first developed from their

comes from the rotational symmetry of space

and is expressed by the group of geometric
rotations about O.
Secondly, all electrons
are alike; the distinction by their labels
,n is one not by essence, but by name
1,2,
only: two constellations of electrons that arise
from each other by an arbitrary permutation
of the electrons are indiscernible.
A permutation consists of a re-arrangement of the

labels;

it

is

really a one-to-one

the set of labels (1, 2,

134

you

spectra has accompanied the development of

mapping

of

n) into itself,

more

applications in mathematics,
larly in the

An

algebraist

particu-

theory of algebraic equations.

is

man who deals

but the only operations he

is

in

numbers,

able to perform

+, X, *-. The
numbers which arise by the four species
and 1 are the rational numbers.
from
The field F of these numbers is closed with
are the four species

respect to the four species,

ence,

i.e.

sum,

differ-

and product of two rational numbers

135

are rational numbers,

and

so

is

their quotient

from zero.
have
had
no reaThus the algebraist would
had
not the
son to step outside this domain F,
provided the divisor

different

is

demands of geometry and

forced

physics

its

coefficients a v , has n solutions, or "roots"

(as

one

;nd to embed
the rational numbers in the continuum of all
This necessity first appeared
real numbers.
when the Greeks discovered that the diagonal
and side of a square are incommensurable.
Not long afterwards Eudoxus formulated the
general principles on which to base the construction of a system of real

numbers

suitable

measurements. Then during the

Renaissance the problem of solving algebraic

for

numbers

-f-

The mystery

with real components

hi

that

= (x-

/(*)

Here

a:

is

equation

#2,

'

'

#n, so

'

>

decomposes

itself

-*)(*-

*!)(*

*)

a variable or indeterminate, and the

is

to be interpreted as stating that

the two polynomials

on

either side coincide

coefficient for coefficient.

Such relations between two indeterminate numbers x, y as the algebraist is able

to construct with

and

tion

of addi-

operations

his

can

multiplication

brought into the form R(x, y)

be

always

where the

y/ 1

they are nothing but pairs

numbers, pairs

for

(a, b)

that

of ordinary

which addition and

multiplication are so defined as to preserve

This

the familiar laws of arithmetics.

all

can indeed be done in such a

real

number

may

complex number
i

2
z'

plicitly

of

(a,

1,0).

way

that

equals 1, more ex-

Thus the equation x 2 +l =0

by any real number x

came solvable. At the beginning of
nineteenth century it was proved that
introduction of the complex numbers
not

any

be identified with the

0) and that the square

(0, 1)

be-

solvable

made

polynomial,

solvable not only this but

all

the
the

aW

had

algebraic

~2

x -f fli*"- +- aox n
for the

unknown

.v,

-f-

whatever

+a
its

n -ix

an

degree n and

is

monomials

of

(Mj

0, 1, 2,

These

with rational coefficients a M ,.

rela-

tions are the "objective relations" accessible

Given two complex numbers a,

ask what polynomials
R(x, y) with rational coefficients exist which
get annulled by substituting the value a for
From two
ior y.
the indeterminate x and
to

him.

he

will

therefore

one

may

pass to

numbers

plex

#1,

any number of given com-

The

algebraist

will ask for the automorphisms of this set 2

of numbers,

of

namely for those permutations

#n which destroy none of

#!,,

the algebraic relations /?(#i,

,# n)

'

Here R(x u
existing between them.
rational
with
is
any polynomial
indeterminates
xi,
cients of the n
which

/(*)-

sum

a finite

equations: the equation

(1)

i.e.

x,

of the type

was com-

when one recognized

pletely dissolved

real

(a,b).

function R(x, y) of the two variables

shrouded them and

first

imaginary unit

their

136

i?i,

all

equations led to the introduction of complex

to say)

into the n factors

the mathematicians to engage in the dire

business of analysing continuity

wont

is

that the polynomial f(x)

is

=
x n)

coeffi
,

xn

annulled by substituting the values

The autoxn
# for *i,
is
called the
morphisms form a group which
mathematician
Galois group, after the French
&h

137

Galois

Evariste

As

(1811-1832).

scription shows, Galois' theory

is

nothing else

but the relativity theory for the set 2, a set

which, by its discrete and finite character, is
conceptually so

much

simpler than the

infi-

nite set of points in space or space-time dealt

with by ordinary relativity theory. We stay

when
members

entirely within the confines of algebra

we assume

t? l5

in particular that the

#n

of the set

n roots of an algebraic

of

2 are defined

equation

(1

as the

),/(*)

0,

degree, with rational coefficients a v

then speaks of the Galois group of the

nth.

One

R(x h

,x n )

Galois' ideas,

solve the equation.

which

for

remained a book with seven

seals but later exerted a more and more
profound influence upon the whole development of mathematics, are contained in a
farewell letter written to a friend on the eve
of his death, which he met in a silly duel at
This letter, if judged
the age of twenty-one.

several decades

by the novelty and profundity of ideas it contains, is perhaps the most substantial piece
of writing in the whole literature of mankind.

tion

is

with rational coefficients

x1

(2)
Its

The
is taken from antiquity.
between diagonal and side of a
determined by the quadratic equa-

first

V2

square

- V2)(* + V2).

(x

rational.

made on

oras,

the thinkers of antiquity

is

evidenced by a number of passages in Plato's

dialogues.

It

was

which forced

this insight

the Greeks to couch the general doctrine of

quantities in geometric rather than algebraic

Let R(x h

terms.

xh

xi

X2

with

#2

zero.

also

is

be a polynomial of
vanishing

xz)

rational coefficients

assuming the value zero) tor

The question is whether

(i.e.

answer

is

we can show

If

that

affirmative for every

x\

#i,

/?(# 2 , #i)

the

then the

transposition

(3)

two roots are #i

an automorphism

r?x

> t?!, t? 2

as well as the identity

The proof runs as follows.

x) of one indeter-

#2-

The polynomial

R{x,

minate x vanishes for x

by x 2 - 2,

-x)

R(x,

(X*

2)

rational coefficients
x:

leaves a remainder ax

d(x)

b of

degree

(ax

the resulting equation a&i +- b

unless a

0.

-x)

0, b

R(x,

+
1

b)

with

Substitute #i for

b.

a,

Its division

t?i.

tradicts the irrational nature of

=
t?i

con-

\/2

Hence
(x

2)

(*),

and consequently R(&2, #i) = R(&2, di) =

0.
Thus the fact that the group of automorphisms contains the transposition (3)
besides the identity

is

equivalent to the irra-

tionality of y/l.

0.

\/2 and # 2

# 2 ->th

0!

is

give two examples of Galois' theory.

The

satisfying certain conditions.

But once it has been ascertained one can

learn from the structure of this group a lot
about the natural procedures by which to

ratio

mentioned a moment ago, they are irThe deep impression which this
discovery, ascribed to the school of Pythag-

As

equation j{x) = 0. It may be difficult

enough to determine the group, requiring
as it does a survey of all polynomials

x2

de-

this

= &i =

My

other example

is

Gauss' construction

of the regular 17-gon with ruler

and compass,

- V2,
138

139

which he found as a young lad of nineteen.

Up to then he had vacillated between classical philology and mathematics; this success
was instrumental in bringing about his final
In a plane

decision in favor of mathematics.

we
x

represent

+ yi

any complex number z

by the point with the

The

coordinates (x,y).

-1

tf

real Cartesian

algebraic equation

arrangement: the dialing of

iteration of its rotation

group Cie evidently has a subgroup Cs of

index 2; it is obtained by turning the dial
through 3^>

1)

(z-

+ s"- +
2

'

'

'

+* + l).

the others are the roots of the equation

(4)

" +

C 16
which

+Z+

'

"

'

of the

full

angle.

=*

\)'(z p

%, %>

By repeating this process of skipping alternate

points we find a chain of consecutive subgroups
means "contains")

has p roots which form the vertices of a regular p-gon.

z = 1 is one vertex; and since
(z

diagram,

this

by 3^6 of the whole

periphery, gives the 16 automorphisms as
permutations among the 16 roots. This

i.e.

0.

If p is a prime number, as we shall now assume, they are algebraically indiscernible

starts

DCsDCtDCzDd
with the

full

group Ci6 and ends

with the group C\ consisting of the identity

which each group is contained

in the preceding one as a subgroup of index 2.
Due to this circumstance one can determine
the roots of the equation (4) by a chain of 4
Equaconsecutive equations of degree 2.
only, a chain in

tions of

solved

degree
(as

quadratic equations, are

2,

Sumerians already knew)

the

by extraction of square roots. Hence the

problem requires, besides the

solution of our

rational operations of addition, subtraction,

and

multiplication,

the four species

division,

four consecu-

However,
and extraction of a square

tive extractions of

square roots.

root are exactly those algebraic operations

which
ruler

may

FIG. 72

roots

is

a cyclic group of order p

describe the situation for the case p

The

left

1.

17.

17-point dial (see Fig. 72) shows the

labeling of the vertices, the right 16-point

dial the

140

6 roots of (4) in a mysterious cyclic

This

is

the reason

why

the

pentagon and 17-gon,

=
and
17, may be constructed by
3, 5,
p
ruler and compass; for in each of these cases
the group of automorphisms is a cyclic group
whose order p 1 is a power of 2,
regular

and the group of automorphisms for the p

geometrically be carried out by

and compass.

3
It

=
is

21

triangle,

1,

amusing

l,

17

to observe that,

(obvious) geometric
is

22

symmetry

24

1.

whereas the

of the 17-gon

described by a cyclic group of order 17,

141

its

(hidden) algebraic symmetry, which deter-

mines

constructibility,

its

of order 16.

heptagon

Only

if

1 1

and

of 2, p

2 n,

then,

according to Gauss's analysis, the regular

by ruler and compass.

cannot be a prime
number unless the exponent n is a power of 2.
For assume that 2* is the exact power of 2
by which n is divisible, so that n = 2" m
where m is an odd number. Put 2 2 " = a;
1
But for odd m the
then 2 n + 1 = a m
number a m -\- 1 is divisible by a + 1,

p-gon

constructible

is

However, p

2"

and multiplication of two numbers of the

number

field give rise to

of the

field.

Nor

does the operation of division lead beyond

the

3 sides.

a prime number such that

is

power

is

described by one

not constructible nor are the

is

regular polygons with

P 1

is

certain that the regular

It is

Obviously addition, subtraction,

rational.

For

field.

ber of the

let

b \/2 be a numfrom zero with the

-f-

field different

components a, b and let a' =

y/2 be its "conjugate." Because 2 is
not the square \>f a rational number, the sorational

norm

called
a2

2b 2

of a, the rational
different

is

number

act

from zero, and therefore

one obtains the reciprocal - of a

as a

number

(a

IK*"'-'

a"

unless

number

next

"2

Thus
a

1),

/w

Therefore the

of the form 2"

after

3,

5,

and 17 which has a chance to be a prime

number is 2 8 -f 1 = 257. As this is actually
a prime number, the regular 257-gon is constructible by ruler and compass.
Galois' theory

may

ct

aa!

and hence a composite number with the

factor a

in the field as follows:

be put in a slightly

dif-

the field

\/2\

- bV2

a2

'

is

2b 2

'

closed with respect to

the operations of addition, subtraction, multiplication,

and

division,

with the self-under-

stood exclusion of division by zero.

now

We may

ask for the automorphisms of such a

An automorphism would

field.

be a one-to-

one mapping
a* of the numbers of the
field such that a
and a /3 go into
/3
a* -\- 0* and a* /3* respectively, for any
>

'

numbers

a,

in

/3

the field.

It

follows at

by the equation (2).

Let us consider all numbers of the
form a = a + b y/2 with rational components a, b; we call them the numbers of the

once that an automorphism changes every

{\/2}- Because of the irrationality of

such
a number is zero only if a = 0,
\/2
=
b
0.
Consequently the rational compo-

there are only two possible automorphisms,

ferent form, as

shall illustrate

field

nents
a

a,

\/2
(a

uniquely determined by

b are

+ fa y/l
+ (b - bi)

oi)

0, b

b\

\/2

number a of the
and the other carrying

the one which carries every

field

\/2} into

any number a

itself,

+b

y/2

into

its

conju-

It is evident that this

y/2.
second operation is an automorphism, and

gate a'

yields

ai

a\

a, for

number into itself and \/2 into a

number # satisfying the equation # 2 2 = 0,
thus either into y/2 or \/2.
Hence
rational

0;

one has thus determined the group of

automorphisms for the field \/2\.

all

or a

142

a i, b

bi,

provided a,b and a u bi are

field

is

perhaps the simplest algebraic

143

we can

structure

numbers.

invent.

Characteristic for

the operations of addition

These

operations

among them

elements are

Its

and

satisfy

its

structure are

multiplication.

certain

axioms,

those that guarantee a unique

and

inversion of addition, called subtraction,

a unique inversion of multiplication (provided the multiplier is different from zero),

Space

called division.

an

is

another example of

endowed with a structure. Here

elements are points, and the structure is

entity

the

established in terms of certain basic relations

between points such as: A, B, C lie on a

AB is congruent CD, and the

straight line,
like.

What we

learn from our whole discus-

sion and what has indeed become a guiding

principle

modern mathematics

in

this

is

lesson: Whenever you have to do with a structure-

endowed

entity

try

determine

to

its

group of

group of those element-wise

transformations which leave all structural reYou can expect to gain
lations undisturbed.
a deep insight into the constitution of 2 in
this way.
After that you may start to investigate symmetric configurations of elements, i.e.
configurations which are invariant under a
certain subgroup of the group of all automorphisms; and it may be advisable, before

automorphisms, the

composition

of their

If

is.

the

group

is

finite one could tabulate the composition of

The group scheme or abstract
elements.

group thus obtained

entity,

its

or table
st

u.

is

a structural

itself

structure represented by the law

of

composition for

Here the dog

bites into

its

own

tail,

is

spect to a given abstract group:

group of

its

elements,

a clear enough warning for

Indeed one may ask with re-

and maybe that

us to stop.

its

What

is

the

automorphisms, what are the

s * s' of the group into

one-to-one mappings
itself

which make

st

go over into

the arbitrary elements

s,

s't'

while

go over into /,

/'

respectively?

Symmetry

is

a vast subject, significant in

and nature. Mathematics lies at its root,

and it would be hard to find a better one on
which to demonstrate the working of the
mathematical intellect. I hope I have not
completely failed in giving you an indication
of its many ramifications, and in leading you
art

up

the ladder from intuitive

concepts to

abstract ideas.

looking for such configurations, to study the

subgroups themselves, e.g. the subgroup of
those automorphisms which leave one ele-

ment

fixed,

fixed,

and

or

finite

or leave two distinct elements

what discontinuous
exist, and so forth.

investigate

subgroups there

In the study of groups of transformations

one does well

such a group.

to stress the

This

is

mere structure of

accomplished by

taching arbitrary labels to

its

at-

elements and

then expressing in terms of these labels for any

two group elements

144

s,

what the

result u

st

145

APPENDICES

APPENDIX A
DETERMINATION OF ALL FINITE GROUPS OF PROPER
ROTATIONS IN 3-SPACE
A

(cf. p.

77).

simple proof for the completeness of the

(5) in Lecture II is based on the fact

list

established

first

by Leonhard Euler

in the

eighteenth century that every proper* rotation

which is not the identity / is rotaaround an axis, i.e. it leaves fixed not
but every point on a ceronly the origin
in 3-space

tion

tain straight line through 0, the axis

sufficient to consider the

is

sphere

of unit radius

/.

It

two-dimensional

around

instead of

the three-dimensional space; for every rotation carries 2 into itself and thus is a one-to-

one mapping of 2 into itself. Every proper

rotation 9* / has two fixed points on 2 which
are antipodes of each other, namely the points

where the axis / pierces the sphere.

Given a finite group T of proper rotations of
order N, we consider the fixed points of the
TV 1 operations of T which are different

from

We

/.

call

them

poles.

has a definite multiplicity

or

-)

The

operations

v
5"

Each pole p

(= 2

or 3 or 4

of our

group

which leave p invariant consist of the iterations of the rotation around the corresponding
axis by 360/"i and hence there are exactly v
such operations S. They form a cyclic subgroup T p of order v. One of these operations
is the identity, hence the number of opera-

p fixed amounts to v 1
For any point p on the sphere we may con-

tions 9* I leaving

sider the finite set

which p
group;

is

we

of those points q into

carried by the operations of the

call

them

points equivalent to p.

149

Because T

a group this equivalence

is

the nature of an equality,

equivalent to

then p
qz

is

if

equivalent to

q;

itself;

the point p

i.e.

among each

and

both

if

may

point of the class

inasmuch

tive p

serve as

speak of

any

S\L,

others.

UL~

group has been

this

subgroup

finite

Sh

Of how many
the

points

answer:
gests

itself,

equivalent

to

Cp of

The

consist?

is

qi

T carry p into two different points q\ = pS\,

= pS 2 since their coincidence q\ = q%
X

would imply that the operation S\S2 carries

~ =
itself, and would thus lead to S\S 2
/,
X

p into
S\

=S

But suppose

2.

multiplicity

now

that p

is

a pole of

so that v operations of the

group

carry p into itself. Then, I maintain, the

number of points q of which the class Cp consists

Proof: Since the points of the class are in-

discernible even under the given group T,

each must
Let us

first

L~ SL
X

f>

into p

L~ SL where S

150

of

Vice versa,

T carrying
X

multiplicity

this explicitly.

into

if

is

itself

and hence
is

v.

SL where 5

any

carries p
operation of

then

S = LTL~

of the form

is

an element of the group T v

Vice versa,

q.

S,.

one of the

is

Now

therefore

let ffc

operations

v
-

qn

be the

C = Cp and

1,

Then

).

all

the

operations of the table

one of the

is

fixed;

be one of the operations in T carrying

into i

S\L\,

S,Li,

SiL 2

SfLz,

OyLn

SiL*

Indeed each

are different from each other.

individual line consists of different operations.

And
line

the operations

all

must be

line since the

different

of,

from those

qt,

contained

is

and must therefore

qi,

carries
q n , say

figure in the zth

our table.

This proves the relation jV

the fact that the multiplicity v

JV.

Moreover

any one of them

p into one of the points

into

in the fifth

former carry p into q 2 and the

into the point q h =^ q 2

every operation of the group T

line of

the second

say,

latter

If

carries p into q then

carries q into q provided

into p.

carries

same

demonstrate

the operation

leaving

in the table because

equals N/v.

be of the

operations

and thus

carries p into p

the only

leaving

v different

SL carry p into

operation of the group which leaves p fixed.

For then any two different operations S h S2
of

operations

different

naturally sug-

that

correct provided /

is

let Li

of JV points,

n different points of the class

the

V.

points does the class

L~ S 2 L,

Moreover, the

U=

points of a class remain even indiscernible

T =

operations

proper rotations, the

to

ele-

and no

limited

Sy are the v

if

of all

U is an operation of V carrying p into q then

with p

While
under the group

S2

I,

L~ S L,
L- SL

the

fixed.

the points of a sphere are indiscernible

after

are

as the class contains

the points equivalent to p

all

=
=

Ti
T.

and

representa-

its

Si

if

ments leaving p fixed then

is

and q 2 are

We

other.

q\

set as a class of equivalent points;

our

Thus

equivalent to p

is

are equivalent to p then q\

equivalent

of

is

We

use

the

notation

multiplicity of a pole p;

is

we know

nv

and thus

a divisor of
vv

that

for
it is

the
the

same for every pole p in a given class C,

and it can therefore also be denoted in an
unambiguous manner by vc- The multi151

and the number

class C are connected by the
plicity v c

After these preparations

n c of poles in the

relation n c v c

us

let

sider all pairs (S, p) consisting of

now

JV.

con-

an opera-

tion S t* I of the group T and a point p left

or, what is the same, of any pole
fixed by S
p and any operation S 9* I oi the group leav-

This double description indicates a double enumeration of those pairs.

On the one hand there are JV 1 operations
S in the group that are different from /, and
each has two antipodic fixed points; hence
ing p fixed.

of the pairs equals 2(JV

the

number

On

the other hand, for each pole p there are

1 operations 5^ / in the group leaving p

vp

that JV

and hence the number

fixed,

of the pairs

In

II.

poles into classes

+i

v\

vi

But

two

= N/v

collect the

and

thus obtain the following basic equation:

2(JV-

1)

^nc(vc

n c vc

of poles.

On

tions

positive

~N
JV

" 2/

n\

JV/vi,

each

if

n%

two

1.

classes of equivalent

here

(vertical) axis of

this case

JV.

the cyclic group of rota-

is

order

JV.

we have

by JV

yields

Arrange the
order,

<

bers

v h v 2 , Vi

the

left

fs

*o/

vx

multiplicities
v2

<

2.
JV

v3

"2

v3

Not

in

v
all

can be greater than

side

would give a

% + M "KM *

of the right side.

i+
v2

ascending

three

num-

2; for

then

result that

is

contrary to the value

1>

Hence

pi

2,

^^ +^ 1
2

vz

JV

Not both numbers v 2 v s can be > 4, for

then the left sum would be < ^. Therefore
,

1,

and there are no

Leaving aside

152

JV;

of the

"1

all

taking the equation

group T consists of the identity only; then

v%

around a

What follows is a discussion of this equation.

The most trivial case is the one in which the
JV

integers

"2

1)

the relation

"1

i + i + I = 1+

into account, division

III

^+*-

can have the sum 2 only

What we find
HI. In

where the sum to the right extends over

or

poles consists of one pole of multiplicity

classes

and

v%

of equivalent poles

2.

our equation gives

this case

=i

JV

i)

We

than

1).

equals

poles p.

side

left

less

respectively)

Hence each
all

but

1,

classes of equivalent poles (Cases II

n.2

extending over

at least

is

The first fact makes it impossible for the

sum to the right to consist of one term only.
Hence there are at least two classes C. But
certainly not more than 3.
For as each vc
is at least 2, the sum to the right would at
least be 2 if it consisted of 4 or more terms.
Consequently we have either two or three

sum

I (, -

and hence the

at least 2

is

of our equation

equals the

poles.

this trivial case

we can

say

C2

2 or

3.

153

First alternative Illtf v x

Second alternative

2v z

III 2

v2

2,

APPENDIX

vx

2, v 2

INCLUSION OF IMPROPER ROTATIONS

3;

c3

n in

Case

We

IIIi.

classes of poles of multiplicity 2

sisting of n poles,

and one

two poles of multiplicity

that

have two
each con-

class consisting of
It

n.

is

easily seen

conditions are fulfilled

these

by the

dihedral group D'n and by this group only.

For the second alternative III 2 we have,
in view of v 3

>

v2

3,

the following three

possibilities:
V3

3,

N=

12;

v3

5,

N=

60,

v3

4,

X=

24;

the

finite

group F*

the other are their antipodes.

We

therefore

obtain the tetrahedral group.

The

6 equiva-

two-poles are the projections from

onto the sphere of the centers of the 6 edges.

W: One class of 6 four-poles, forming the
corners of a regular octahedron; hence the

octahedral group.

One

class

of 8

of

let A be one
,Sn be the proper operaThe latter form a subgroup F,

them and S h

tions in P*.

and T* contains one

tions and another line

0)

Si>

(2)

AS h

one

class of

12 two-poles (correspond-

ing to the centers of the edges).

Case P: One class of 12 five-poles which

must form the corners of a regular

icosa-

20 three-poles correspond to
the centers of the 20 sides, the 30 two-poles to
the centers of the 30 edges of the polyhedron.

hedron.

The

"

proper operaimproper
of
operations
line of

Sn

AS.

It contains no other operations.

For if T is
an improper operation in V* then A~ X T is
proper and hence identical with one of the

fore

T=

AS,.

first line,

say

and

Si,

there-

Consequently the order of

T* is 2n, half of its operations are proper

forming the group T, the other half are
improper.

We now distinguish two cases according to

whether the improper operation ^ is or is
not contained in F*.

choose

f r

^ and

In the

thus get

may

In the second case we

second line in the form

first

Y*

case

we

V.

also write the

three-

poles (corresponding to the centers of the

sides);

of rotations in 3-

space contains improper rotations

operations in the

which we denote by T, W, P respectively.

T: There are two classes of 4 three-poles
each.
It is clear that the poles of one class
form
a regular tetrahedron and those of
must

lent

78).

= 1+2.
If

Set

(cf. p.

(2')

Z^l,

'

<^7"n

where the Ti are proper

this case all the Tt
Si.

rotations.

are different from

Indeed were Ti

But
all

Sk then the group

T* would contain with %Ti = %Sk and

also the element G3S)jTs

the hypothesis.

in

the

Under

Z->

5*

contrary to

these circumstances

the operations
5

ijl

(3)

154

155

form a group T' of proper rotations of order

In in which T is contained as a subgroup of
Indeed the statement that the two
index 2.
lines (3) form a group is, as one easily verifies,
equivalent to the other that the lines (1) and
(2') constitute a group (namely the group
Thus T* is what was denoted in the
r*).
main text by TT, and we have thus proved

two methods mentioned there are

the only ones by which finite groups containing improper rotations may be constructed.
that the

ACKNOWLEDGMENTS
I

my debt to

wish to acknowledge especially

Miss Helen Harris in the Marquand Library

of Princeton University who helped me to

photographs of

find suitable

objects
grateful

to

allowed

me

the publishers

publications.

listed

below.

who

also

from
These publications are

Anderson photograph.
68.
Dye, Daniel Sheets,

Figs. 67,

am

generously

Alinari photographs.

Figs. 10, 11, 26.

Fig. 15.

of the art

to reproduce illustrations

their

of

many

pictured in this book.

Chinese

Harvard-Yenching

Institute

Cambridge, 1937.
Ewald,
69, 70, 71.

grammar

C9b,

Figs.

lattice,

SI 2a.

Monograph

V.
Figs.

und Rbntgenstrahlen,

P.

Figs.

P.,

44,

Kristalle

45,

125.

Springer, Berlin, 1923.

Figs.

Haeckel,

37.

36,
der

Natur,

Pis.

Ernst,
28.

10,

Kunstformen

Leipzig

und

Wien, 1899.
Haeckel, Ernst, Challenger mono-

Fig. 45.

graph.

Report on the

voyage of

H.M.S. Challenger, Vol. XVIII,

H.M.S.O., 1887.

PI. 117.

Hudnut

Fig. 54.

ment
of

ornament,

of the

Sales Co., Inc., advertise-

in Vogue,

Figs. 23, 24, 31.

scientific results

February 1951.

Jones,
Pis.

Owen, The grammar

XVI, XVII, VI.

Bernard Quaritch, London, 1868.

Kepler, Johannes, Mysterium Cosmographicum.
Tubingen, 1596.

Fig. 46.

157

Photograph by

Fig. 48.

"Symmetry," Journal

Bruno

PI. 104.

Cassirer Verlag, Berlin,

Ludwig,

18.

16,

Problem im

W.,

Springer, Berlin, 1932.

Needham, Joseph,

17.

Fig.

5.

Rechts-links-

und beim Menschen,

Tierreich

Figs. 81, 120a.

Fig.

June
Figs.

1924.
Figs.

Academy

Kunst,

Maurische

Ernst,

Kiihnel,

32.

'

Figs. 22, 58, 59, 60, 61, 64.

ler no., Paris, 1950.

tes.

Fig.

Reali-

Kitrosser.

I.

Yale

Order and

University

Press,

life,

8,

of

12.

Wulff,

Figs.

Washington

28,

No.

6,

Figs. 2, 5, 6, 7, 8, 9.

O.,

byzantinische Kunst; II,

Kunst,

of the

Vol.

Sciences,

15, 1938.

Weyl, Hermann,

523,

Verlagsgesellschaft

Altchristliche

und

Die byzantinische

514.

Akademische

Athenaion,

Berlin,

1914.

New

Haven, 1936.

New York

35.

Fig.

photograph of

Pfuhl, Ernst, Malerei und geichnung

Fig. 29.
der

Garden,

Botanical

Iris rosiflora.

Band,

III.

Griechen;

Verzeichnisse

und Abbildungen, PL I (Fig.

Bruckmann, Munich, 1923.

10).

F.

Speiser, A., Theorie der Gruppen

Figs. 62, 65.

von endlicher Ordnung, 3. Aufl., Figs. 40, 39.

Springer, Berlin, 1924.

Swindler,

Figs. 3, 4, 6, 7, 9, 25, 30.

Ancient painting, Figs.

192,
Press,

408,

125,

New

91,

253.

(p.

Mary H.,
45),

Haven, 1929.

Thomp-

Figs. 42, 43, 44, 50, 51, 52, 55, 56.

son,

127,

Yale University

On growth and form,

D'Arcy W.,

Figs. 368, 418, 448, 156, 189, 181, 322,

213.

New

Cambridge UniCambridge and New York,

edition,

versity Press,

1948.
Fig. 53.

Reprinted from Vogue Pattern Book,

Conde Nast
Figs.

27,

28,

Publications, 1951.
Troll,

39.

metriebetrachtung
Studium Generate,
Figs. (1 9

& 20)

in

der

Biologie,"

Jahrgang, Heft 4/5,

2.
,

Wilhelm, "Sym-

Berlin-Gottingen-

Heidelberg, Juli, 1949.

U.S. Weather Bureau photograph

by W. A. Bentley.

Fig. 38.

158

159

,,

INDEX
absolute vs. relative space and time, 21
abstract group, 145
active (optically active) substances, 17, 29

Adam and

Eve, 32
addition of vectors, 93

afEne geometry, 96, 132

algebra, 121, 135
Alhambra (Granada), 109, 112
analytic geometry, 94
Angraecum distichum, 51
a priori statements and symmetry, 126

Arabian Nights, 90
Archibald, R. C, 72
Archimedes, 92
Aristotle, 3
arithmetical theory of quadratic forms, 120

arrangement of atoms in a
Ascaris megalocephala, 36

crystal,

122

associative law, 95

asymmetry and symmetry

in art, 13, 16; historic

of
29;
26; the prerogative of
16;

crystals, 17,

in the

life?

of

life,

31 ;

animal kingdom,

and the origin

32

aurea sectio, 72
automorphism, 18, 42; geometric and physical
129-130; in algebra, 137; for any structureendowed entity, 144
axis, 109, 149

Balmer series, 133

band ornaments, 48-51
bar, 52
beauty and symmetry, 3, 72
De Beer, G. R., 34
bees, their geometric intelligence, 91
Beethoven (pastoral sonata), 52
bilateral symmetry, Chap. I; definition,

animals, 26; aesthetic and vital value,

human body, 8
bipotentiality of plasma, 38

6,

4;

of

28; of the

Birkhoff, G. D., 3, 53
Bonnet, Charles, 72
Browne, Sir Thomas, 64, 103
Buridan's ass, 1
Capella Emiliana (Venice), 65
Cartesian coordinates, 97
causal structure, 25, 132

161

/I

two dimensions, 65;

central symmetry, in

in

three

Chinese window lattice, 113

Church, A. H., 73
Clarke, 20
class of crystals, 122;

of equivalent

points,

150

Cohn-Vossen, S., 90
and incommensurable, 135
commensurable, 75;
commutative law (in general not valid for composition of transformations), 41

for addition of vectors,

93

complex numbers, 136

composition of mappings,

41,

of elements

in

an

abstract group, 145

congruence and congruent transformations, 18, 43,

130
CONKLIN, 34
constitution and environment, 126
construction of 17-gon by ruler and compass, 139, 141
contingency (and lawfulness), 26

1 36
discontinuous features, 108-109, 127;
group and discontinuous subgroup, 68-69
coordinates and coordinate system, 94
Corneto, 13

continuity, analysis of

continuous

vs.

Cosmati, 49

,
28; crystallographic
28; enantiomorph
groups, 120, for any number of dimensions, 121

crystals,

cube, 74
cyclic group, 65

Darwin, Charles, 92
densest packing of circles, 83,

Descartes, 95
determination of

left

and

of spheres, 90

right in animals, 33-38

dextro- and laevo-forms of crystals, 29

diamond

lattice,

102

Epicurus, 128
equiangular spiral, 69
equivalence of left and right, 19-20, 129;
of past
and future, 24;
of positive and negative electricity,
25
equivalent points, 150
Euclid, 17, 74
Eudoxus, 136
Euler, L., 149; his topological formula ("polyhedron
formula"), 89
exclusion principle, 134

Faistauer, A., 24
Faust, 45
Fedorow, 92
Fibonacci series, 72
field of numbers, 143
flowers, their symmetry, 58

fontenelle, 91
Frey, Dagobert, 16
Fugus, 34
future, see: past

Galen, 4
Galois, E., Galois group and Galois theory, 137, 142
Gauss, 120, 139, 142
general relativity theory, 132
genetic constitution, 37
genotype and phenotype, 126
gcnotypical and phenotypical inversion, 37
geometry, what is a
?, 133
geometric automorphism, 130
geranium, 66
gliding axis, 109

diatoms, 87

Goethe,

Dickson, L., 121

dihedral group, 65

group of automorphisms, 42, 144;

dilatation

and

Diodore,

'dilatory'

68;

symmetry, 68

DlRICHLET, LEJEUNE, 121

of translations,
discontinuous group, 68, 99;
99-100; see also under: continuous
distributive law, 95
dodecahedron, 74
double infinite rapport, 100
Driesch, Hans, 35
Durer, 3, 65
Dye, Daniel Sheets, 113
dynamic symmetry, 72

162

echinoderms, 60
Einstein, Albert, 130, 132
enantiomorph crystals, 28
Entemena, King
8

dimensions, 79-80
Ghartres, 16

51,

72

of permutations, 135;

of
of transformations,

dilatations,

43; abstract group, 145; finite group of proper rotations in 2 dimensions, 54,
of proper and

improper rotations, 65; such groups having an

invariant lattice, 120; the same for 3 dimensions,
79, 149-154; 80, 120, 155-156; 120; finite group of
unimodular transformations in 2 dimensions, 107,
in 3 dimensions, 120.

Haeckel, Ernst, 60, 75, 88

Hambidce, J., 72
Harrison, R. G., 35
Helianthus maximus, 70, 73

163

longitudinal reflection, 50
Lorentz, H. A., 131; Lorentz group, 131

helix, 71

helladic,

12

Helmholtz, H.,

Lorenz, Alfred, 52
Ludwig, W., 26, 38

18, 43, 128

heraldic symmetry, 8
Hermite, 121

Herzfeld, Ernst, 10
hexagonal symmetry, 63;

Mach,

lattice or pattern, 83,

110

Hilbert, D., 90
historic asymmetry, 16

Hodler, F., 16
homogeneous linear transformation, 96

Homo

mapping,

sapiens, 30

hypercomplex number system, 121

mapping, 41

left and right, 23;

genotypical and phenotypical

inversion of

quantities,

of time,
37;

24, 52;

situs inversus,

30
irrationality of square root of two,

129

Needham, Joseph, 34
negative, see under: positive

Newton,

I., 21, 130

Kelvin, Lord, 93
Kepler, 76, 90
Klein, Felix, 133
Koenig, Samuel, 91

objective statements, 128

octahedron, 74

33

Oppe, Paul, 24

structure and metric

104; Chinese window 113; rectangubasis, 100;

and diamond

lattices,

102

in paintings,

23

Leibniz, 17, 18, 20, 21, 27, 127

Leonardo da Vinci, 66, 99
linear independence of vectors, 94; linear transforma-

homogeneous, 96, and non-homogeneous, 97

logarithmic spiral, 69

164

optically active, 17

order of a group, 79

Laue, Max von, 122-124; Laue interference pattern

or diagram, 124
law of rational indices, 121
left and right: polar opposites or not?, 17, 22; concern

tion,

symmetry (and asymmetry),

ontogenesis of bilateral

laevo-, see under: dextro-

a screw, 17;

20, 27, 43

Nicomachean Ethics, 4
Nig c li, Paul, 122

Kant,

lar

Nautilus, 70

Jaeger, F. M., 29
Japp, F. R., 31
Jones, Owen, 113
Jordan, Pascual, 32

structure,

34
modul, 96
Monreale, 14
morphology, 109, 126
motion (in the geometric sense), 44
multiplication of a vector by a number, 94
music, formal elements of
, 52
mythical thinking, 22
mitosis,

47
invariant quadratic form, 106, 108;
relations, etc., 133

infinite rapport,

lattice

104

Michelangelo, 22
Minkowski, H., 90, 121

indiscernible, 17, 127

100;

18, 41

structure,

icosahedron, 74
identity, 41

lattice,

crystals,

Maraldi, 90, 91
Maschke, H. (and Maschke's theorem), 107
Medusa, 66; medusae, 63
metric ground form, 96;
structure and

honey-comb, 83, 90-91

Huxley, Julian S., 34

inverse

Ernst, 19
macroscopic and microscopic symmetry of
123
magnetism, positive and negative?, 20
Mainz cathedral, 56
Mann, Thomas, 64

ornamental symmetry in one dimension, 48;

two dimensions, 99-115
orthogonal transformation, 97
orthogonally equivalent, 99

parenchyma

in

of maize, 87

particle (and wave), 25

and future, 24, 132

Pasteur, 29-31
Pauli, W., 134
Penicillium glaucum, 30

past

165

pentagonal symmetry
pentagram, 45
permutation, 134

in organic nature,

of animals, 27; as
5;
defined by a finite group of rotations, 53

San Apollinare (Ravenna), 13

San Marco (Venice), 14
S. Maria degli Angeli (Florence), 65
St. Matthew, 22
San Michele di Murano (Venice), 65

phenotypical inversion, 37
phenylketonuria, 30
phyllotaxis,

rotational symmetry,

63

72

phylogenesis of asymmetry, 33
physical automorphism, 129
Pisa Baptisterium, 55

St. Pierre

(Troyes, France), 58

10

Planck, Max, 133

seal stones,

planetary orbits and the regular polyhedra, 76

Plato, 8, 28, 73; his Timaeus, 74; Platonic solids, 73
polyhedra, 78
polar axis, 33;
polarity, see under: equivalence
pole, 109, 149; animal and vegetative poles of blastula,
33; magnetic poles, 19

sectio aurea, 72

Polya, G., 104

Siegel, C. L., 121

42
127
Sistine Chapel, 22
situs inversus, 30
slip reflection, 50
similarity, 18,
sinister, 22,

space, 17; absolute or relative?, 21

space-time, 131

polyhedral groups, 78, 154

POLYKLEITOS, 3, 65
positive and negative electricity, 24;
netism, 20

mag-

special relativity theory, 132

spectra, 133

potentially infinite, 51

Speiser, Andreas, 50, 52, 74

probability, 25

spherical symmetry, 25, 27

projective geometry, 132

spin, 134

proper and improper congruences, 43

prospective potency and significance, 35
Pythagoras' theorem, 96

spira mirabilis, 69

'spiral'

symmetry, 70

Stephan's

dome

(Vienna), 67

structure of space, 17-18, 130;

quadratic forms, 96, 108; their arithmetical theory,

120
quantum physics, 25, 133
quincunx, 64, 103

symmetry

radiolarians, 76, 88

rectangular

lattice,

reflection, 4,

tapestry cartoons), 23

set,

and rotation by
122;

180, 9, 77;

at

47

73
129;

of length, 128;

and

right, of position, 20,

relativity

45, 127, 133, 144; theory

166

flowers, 58; bilateral, dilatory,

bilateral, dilatory, etc.

Tait, P. G., 73
tartaric acid, 29

47, especially in

theory of relativity,

double meaning of the word, 22; see also under:

roots of an algebraic equation, 137

of

symmetry, see under:

Thompson, D'Arcy, 93
Thomson, William (Lord Kelvin), 93

music, 51
left

crystals, 122;
etc.

tetrahedron, 74
tetrakaidekahedron, 92

17,

general theory of relativity, 132

rotation, 44, 97, 149

16;

127; special and

and symmetry,

of,

Rembrandt, 23
rhythm and rhythmic symmetry,
right,

3,

symmetry in the animal kingdom, 27;

symmetry of crystals, 121-123; symmetry classes of

polygon, 73; polyhedra,

relativity of direction, of left

proportions,

of

25,

bilateral

102

origin, 79; one-dimensional

regular point

= harmony

defined by a group of automorphisms, 45, 133, 134;

and equilibrium, 25;
and a priori statements,
126; the three stages of spherical, rotational and

racemic acid, 29

Raphael (Sistine Madonna,

Reaumur, 91

of space-time, 131;
132

structure-endowed entity, 144; causal

subgroup of index 2, 44
Sumerians, 8, 9
swastika, 66

27

time, absolute or relative?, 21; as the fourth dimension of the world, 1 30

Tiryns, 12
Tomb of the Triclinium, 13
Torcello, 12

167

transformation, 18, 41
translation, 44, 98

translatory symmetry, 47
transversal reflection, 50
tripod, 66

triquetrum, 66
Troll, W., 51
Turritella duplicata, 68

Umklappung, 77
unimodular transformation, 100
unimodularly equivalent, 106
vector, 44; vector calculus,

128
Vinca herbacea, 66
Vitruvius, 3, 65

vertical,

Wickham, Anna,
WOHLER, 31

Wolfflin, H., 23, 24

world = space-time, 131
X-rays, 122

168

93-94

Philosophy of Mathematics

and Natural Science

By

HERMANN WEYL

Scientists and philosophers will welcome

this revised

and augmented translation

of

Professor Weyl's "Philosophic der Mathc-

matik und Naturwissenschaft," which originally appeared in Oldenbourg's Handbuch

der Philosophie in

1927.

Mr. Weyl has

six new appendices on such subjects

Foundations of Mathematics, Ars Com-

added
as

binatoria,

Quantum

Mechanics,

Physics

and Biology, and Evolution.

Part 1, on mathematics, starts with some
principles of mathematical logic, describes
the axiomatic method,

and then digs deeper

problem of the infinite. Part 11 discusses space, time, and the relative merits
of the realistic and idealistic attitudes
into the

toward the external world;

it

also gives a

general methodology of science and closes

with an account of the development and

of the main features of our modern world

under such titles as Matter, Causality, Law, Chance, Freedom.

"A rich and fascinating work by one of
the best minds of our day."Scientific
American.
picture,

"A masterpiece

of synthesis. ... It will

serve as an introduction to this unfamiliar

but magnificently complex and intricate

world in which we live." Thought.

"A

sensitive

and informed appreciation

of the history of ideas."Journal of Philos-

ophy.

376 pages. $6.00