PW"

H Y "r m - 1 S A M ! - 1 A u1 '|m u
Im Im bilir _v and .s'c'im r :'/ix" inf:- r cr xcc
Only Iwo can play this gam e (published as by Jam es Keys)
LA WS OE FORM
G S PENCER BROWN
- "
if- _, _ _.
TH E JULIA N PRES S , INC. Publisher s
NEW Y ORK
!
P-~
IIY H I! -1 SAME Al| l'|l()R
lruImhiIi! ,v and .s'("imIi/ic" in/crcnce L S OFFORM
()n/y Iwu ran play this game (published as by James Keys) W
i = '
} G SPENCER BROW N
. A-
,-M
\
4
1
>
i
TH E JULIAN PRESS, INC. Publishers
' NEW YORK
1
a
l'i|~tl |\lll1 ll\ll(tl in ll S A . lw
H ie .Iuli;|n l'|css, luv.
I50 lil'tli A venue
New Y or k, N.Y . llltlll
I972
liir st published in (ir czit Br itain by
(icor gc A llen and Unwin Ltd.
London, I969
A ll r ights r eser ved
including the r ight of r epr oduction
in whole or in par t in any for m .
Copyr ight 1 972 by G. S pencer Br own
Libr ar y of Congr ess Catalog Car d Num ber : 72- 80668
M anufactur ed in the United S tates of A m er ica
A NOTE ON TH E M A TH EM A TICA L A PPROA CH
The them e of this book is that a univer se com es into being
when a space is sever ed or taken apar t. The skin of a living
or ganism cuts oiT an outside fr om an inside. S o does the cir -
cum fer ence of a cir cle in a plane. By tr acing the way we r epr esent
such a sever ance, we can begin to r econstr uct, with an accur acy
and cover age that appear alm ost uncanny, the basic for m s
under lying linguistic, m athem atical, physical, and biological
science, and can begin to see how the fam iliar laws of our own
exper ience follow inexor ably fr om the or iginal act of sever ance.
The act is itself alr eady r em em ber ed, even if unconsciously,
as our r st attem pt to distinguish differ ent things in a wor ld
wher e, in the r st place, the boundar ies can be dr awn anywher e
we please. A t this stage the univer se cannot be distinguished
fr om how we act upon it, and the wor ld m ay seem like shifting
sand beneath our feet.
A lthough all for m s, and thus all univer ses, ar e possible,
and any par ticular for m is m utable, it becom es evident that
the laws r elating such for m s ar e the sam e in any univer se. It
is this sam eness, the idea that we can nd a r eality which is
independent of how the univer se actually appear s, that lends
such fascination to the study of m athem atics. That m athem atics,
in com m on with other ar t for m s, can lead us beyond or dinar y
existence, and can show us som ething of the str uctur e in which
all cr eation hangs together , is no new idea. But m athem atical
texts gener ally begin the stor y som ewher e in the m iddle, leaving
the r eader to pick up the thr ead as best he can. H er e the stor y
is tr aced fr om the beginning.
Unlike m or e super cial for m s of exper tise, m athem atics
is a way of saying less and less about m or e and m or e. A m athe-
m atical text is thus not an end in itself, but a key to a wor ld
beyond the com pass of or dinar y descr iption.
A n initial explor ation of such a wor ld is usually under taken
in the com pany of an exper ienced guide. To under take it alone,
V
Final |\ lll1 li\ il( ti in ll S A. lw
H ie .Iul|;|n l'|c\ .~., Inc.
I50 lil'tl| Avenue
New York, N.Y. ltltlll
I972
liirst published in (ircnt Britain by
(icorgc Allen and Unwin Ltd.
London, I969
All rights reserved
including the right of reproduction
in whole or in part in any form.
Copyright 1 972 by G. Spencer Brown
Library of Congress Catalog Card Number: 72-80668
Manufactured in the United States of America
A NOTE ON TH E MATH EMATICALAPPROACH
'lhe theme of this book is that a universe comes into being
when a space is severed or taken apart. The skin of a living
organism cuts oiT an outside from an inside. So does the cir-
cumference of a circle in a plane. By tracing the way we represent
such a severance, we can begin to reconstruct, with an accuracy
and coverage that appear almost uncanny, the basic forms
underlying linguistic, mathematical, physical, and biological
science, and can begin to see how the familiar laws of our own
experience follow inexorably from the original act of severance.
The act is itself already remembered, even if unconsciously,
as our rst attempt to distinguish different things in a world
where, in the rst place, the boundaries can be drawn anywhere
we please. At this stage the universe cannot be distinguished
from how we act upon it, and the world may seem like shifting
sand beneath our feet.
Although all forms, and thus all universes, are possible,
and any particular form is mutable, it becomes evident that
the laws relating such forms are the same in any universe. It
is this sameness, the idea that we can nd a reality which is
independent of how the universe actually appears, that lends
such fascination to the study ofmathematics. That mathematics,
in common with other art forms, can lead us beyond ordinary
existence, and can show us something of the structure in which
all creation hangs together, is no new idea. But mathematical
texts generally begin the story somewhere in the middle, leaving
the reader to pick up the thread as best he can. H ere the story
is traced from the beginning.
Unlike more supercial forms of expertise, mathematics
is a way of saying less and less about more and more. A mathe-
matical text is thus not an end in itself, but a key to a world
beyond the compass of ordinary description.
An initial exploration of such a world is usually undertaken
in the company of an experienced guide. To undertake it alone,
V
A NtH 'll on 'r |n~: M /\|l|l'1 M A ll('A l. /\l'l'Rt)A (Il
although possible, is per luips as tlillicult as to enter the wor ld of
m usic by attem pting, without per sonal guidance, to r ead the
scor e- sheets ofa m uster com poser , or to set out on a r st solo
ight in an aer oplane with no other pr epar ation than a study
of the pilots m anual.
A lthough the notes at the end of the text m ay to som e extent
m ake up for , they cannot effectively r eplace, such per sonal
guidance. They ar e designed to be r ead in conjunction with the
text, and it m ay in fact be helpful to r ead them r st.
The r eader who is alr eady fam iliar with logic, in either its
tr aditional or its sym bolic for m , m ay do well to begin with
A ppendix 2, r efer r ing thr ough the Index of For m s to the text
whenever necessar y.
vi
A note on the m athem atical appr oach v
Pr eface to the r st A m er ican edition ix
Pr eface xiii
Intr oduction xv
1
2
\OOO\] O\U1 - I> L J
1 0
1 1
1 2
Not
The for m 1
For m s taken out of the for m 3
The conception of calculation 8
The pr im ar y ar ithm etic 1 2
A calculus taken out of the calculus 25
The pr im ar y algebr a 23
Theor em s of the second or der 3 8
Re- uniting the two or der s 42
Com pleteness 5()
Independence 53
Equations of the second degr ee 54
Re- entr y into the for m 69
GS
A ppendix 1 . Pr oofs of S he] j'er s postulates 1 07
A ppendix 2. The calculus inter pr etedfor logic 1 1 2
Index of r efer ences 1 3 6
Index of for m s 1 3 8
A NH l'll on '|"|n~: M/\ |l|l'1 MA l l('Al. /\ |'l'Rt)A( Il
although possible, is perlmps as tlillicult as to cntcr thc world of
music by attempting, without personal guidance, to read the
score-sheets ofa muster composer, or to set out on a rst solo
llight in an aeroplane with no other preparation than a study
of the pilots manual.
Although the notes at the end of the text may to some extent
make up for, they cannot effectively replace, such personal
guidance. They are designed to be read in conjunction with the
text, and it may in fact be helpful to read them rst.
The reader who is already familiar with logic, in either its
traditional or its symbolic form, may do well to begin with
Appendix 2, referring through the Index of Forms to the text
whenever necessary.
vi
A note on the mathematical approach v
Preface to the rst American edition ix
Preface xiii
Introduction xv
1
2
\ OOO\ ] O\ U1 -I> L J
1 0
1 1
1 2
Not
Theform 1
Forms taken out of theform 3
The conception of calculation 8
The primary arithmetic 1 2
A calculus taken out of the calculus 25
The primary algebra 23
Theorems of the second order 3 8
Re-uniting the two orders 4 2
Completeness 5()
Independence 53
Equations of the second degree 54
Re-entry into theform 69
GS
Appendix 1 . Proofs of She] j'ers postulates 1 07
Appendix 2. The calculus interpretedfor logic 1 1 2
Index of references 1 3 6
Index of forms 1 3 8
A NH l'll on '|"|||~: M/\ llll'1 MA l l('Alt /\ |'l'R()A( Il
although possible, is perhaps as dillicult as to enter the world ol
music by attempting, without personal guidance, to read the
score-sheets ol'a master composer, or to set out on a rst solo
ight in an aeroplane with no other preparation than a study
of the pilots manual.
Although the notes at the end of the text may to some extent
make up for, they cannot effectively replace, such personal
guidance. They are designed to be read in conjunction with the
text, and it may in fact be helpful to read them rst.
The reader who is already familiar with logic, in either its
traditional or its symbolic form, may do well to begin with
Appendix 2, referring through the Index of Forms to the text
whenever necessary.
vi
A note on the mathematical approach v
Preface to the rst American edition ix
Preface xiii
Introduction xv
1
2
\ OOO\ ] O\ U1 -I> L J
1 0
1 1
1 2
Not
Theform 1
Forms taken out of theform 3
The conception of calculation 8
The primary arithmetic 1 2
A calculus taken out of the calculus 25
The primary algebra 23
Theorems of the second order 3 8
Re-uniting the two orders 4 2
Completeness 5()
Independence 53
Equations of the second degree 54
Re-entry into theform 69
GS
Appendix 1 . Proofs of She] j'ers postulates 1 07
Appendix 2. The calculus interpretedfor logic 1 1 2
Index of references 1 3 6
Index of forms 1 3 8
Tho obscur d this is the for m of the A ngelic land
William Blake A m er ica
lRlll*A CE TO TH E FIRS T A M ERICA N EDITION
A par t fr om the standar d univer sity logic pr oblem s, which
the calculus published in this text r ender s so easy that we need
not tr ouble our selves fur ther with them , per haps the m ost
signicant thing, fr om the m athem atical angle, that it enables
us to do is to use com plex values in the algebr a of logic. They
ar e the analogs, in or dinar y algebr a, to com plex num ber s
a + b \/1 . M y br other and I had been using their Boolean
counter par ts in pr actical engineer ing for sever al year s befor e
r ealizing what they wer e. Of cour se, being what they ar e, they
wor k per fectly well, but under standably we felt a bit guilty
about using them , just as the r st m athem aticians to use squar e
r oots of negative num ber s had felt guilty, because they too
could see no plausible way of giving them a r espectable aca-
dem ic m eaning. A ll the sam e, we wer e quite sur e ther e was a
per fectly good theor y that would suppor t them , if only we
could think of it.
The position is sim ply this. In or dinar y algebr a, com plex
values ar e accepted as a m atter of cour se, and the m or e ad-
vanced techniques would be im possible without them . In
Boolean algebr a (and thus, for exam ple, in all our r easoning
pr ocesses) we disallow them . Whitehead and Russell intr o-
duced a special r ule, which they called the Theor y of Types,
expr essly to do so. M istakenly, as it now tur ns out. S o, in this
eld, the m or e advanced techniques, although not im possible,
sim ply dont yet exist. A t the pr esent m om ent we ar e con-
str ained, in our r easoning pr ocesses, to do it the way it was
done in A r istotles day. The poet Blake m ight have had som e
insight into this, for in 1 788 he wr ote that r eason, or the r atio
of all we have alr eady known, is not the sam e that it shall be
when we know m or e.
Recalling Russells connexion with the Theor y of Types, it
ix
Tho obscur d this is the form of the Angelic land
W illiam Blake America
lRlll* ACE TOTH E FIRST AMERICAN EDITION
Apart from the standard university logic problems, which
the calculus published in this text renders so easy that we need
not trouble ourselves further with them, perhaps the most
signicant thing, from the mathematical angle, that it enables
us to do is to use complex values in the algebra of logic. They
are the analogs, in ordinary algebra, to complex numbers
a + b \ /1 . My brother and I had been using their Boolean
counterparts in practical engineering for several years before
realizing what they were. Of course, being what they are, they
work perfectly well, but understandably we felt a bit guilty
about using them, just as the rst mathematicians to use square
roots of negative numbers had felt guilty, because they too
could see no plausible way of giving them a respectable aca-
demic meaning. All the same, we were quite sure there was a
perfectly good theory that would support them, if only we
could think of it.
The position is simply this. In ordinary algebra, complex
values are accepted as a matter of course, and the more ad-
vanced techniques would be impossible without them. In
Boolean algebra (and thus, for example, in all our reasoning
processes) we disallow them. W hitehead and Russell intro-
duced a special rule, which they called the Theory of Types,
expressly to do so. Mistakenly, as it now turns out. So, in this
eld, the more advanced techniques, although not impossible,
simply dont yet exist. At the present moment we are con-
strained, in our reasoning processes, to do it the way it was
done in Aristotles day. The poet Blake might have had some
insight into this, for in 1 788 he wrote that reason, or the ratio
of all we have already known, is not the same that it shall be
when we know more.
Recalling Russells connexion with the Theory of Types, it
ix
Tho obscur d this IS the form of the Angelic land
W illiam Blake America
lRlll* ACE TOTH E FIRST AMERICAN EDITION
Apart from the standard university logic problems, W hich
the calculus published in this text renders so easy that we need
not trouble ourselves further with them, perhaps the most
signicant thing, from the mathematical angle, that it enables
us to do is to use complex values in the algebra of logic. They
are the analogs, in ordinary algebra, to complex numbers
a + b \ /1 . My brother and I had been using their Boolean
counterparts in practical engineering for several years before
realizing what they were. Of course, being what they are, they
work perfectly well, but understandably we felt a bit guilty
about using them, just as the rst mathematicians to use square
roots of negative numbers had felt guilty, because they too
could see no plausible way of giving them a respectable aca-
demic meaning. All the same, we were quite sure there was a
perfectly good theory that would support them, if only we
could think of it.
The position is simply this. In ordinary algebra, complex
values are accepted as a matter of course, and the more ad-
vanced techniques would be impossible without them. In
Boolean algebra (and thus, for example, in all our reasoning
processes) we disallow them. W hitehead and Russell intro-
duced a special rule, which they called the Theory of Types,
expressly to do so. Mistakenly, as it now turns out. So, in this
eld, the more advanced techniques, although not impossible,
simply dont yet exist. At the present moment we are con-
strained, in our reasoning processes, to do it the way it was
done in Aristotles day. The poet Blake might have had some
insight into this, for in 1 788 he wrote that reason, or the ratio
of all we have already known, is not the same that it shall be
when we know more.
Recalling Russells connexion with the Theory of Types, it
ix
ll{ll<'/\(l- I r o 'llll- I FIRS T A M ERICA N EDITION
was with som e tr epidation that I appr oached him in 1 967
with the pr oof that it was unnecessar y. To m y r elief he was
delighted. The Theor y was, he said, the m ost ar bitr ar y thing
he and Whitehead had ever had to do, not r eally a theor y but
a stopgap, and he was glad to have lived long enough to see
the m atter r esolved.
Put as sim ply as I can m ake it, the r esolution is as follows.
A ll we have to show is that the self- r efer ential par adoxes, dis-
car ded with the Theor y of Types, ar e no wor se than sim ilar
self- r efer ential par adoxes, which ar e consider ed quite accept-
able, in the or dinar y theor y of equations.
The m ost fam ous such par adox in logic is in the statem ent,
This statem ent is false.
S uppose we assum e that a statem ent falls into one of thr ee
categor ies, tr ue, false, or m eaningless, and that a m eaningful
statem ent that is not tr ue m ust be false, and one that is not
false m ust be tr ue. The statem ent under consider ation does
not appear to be m eaningless (som e philosopher s have claim ed
that it is, but it is easy to r efute this), so it m ust be tr ue or
false. If it is tr ue, it m ust be, as it says, false. But if it is false,
since this is what it says, it m ust be tr ue.
It has not hither to been noticed that we have an equally
vicious par adox in or dinar y equation theor y, because we have
car efully guar ded our selves against expr essing it this way. Let
us now do so.
We will m ake assum ptions analogous to those above. We
assum e that a num ber can be either positive, negative, or zer o.
We assum e fur ther that a nonzer o num ber that is not positive
m ust be negative, and one that is not negative m ust be posi-
tive. We now consider the equation
x2 + 1 I O.
Tr ansposing, we have
x2 I - 1 ,
X
PRlFA (Ili 'l() TH E FIRS T A M ERICA N Eom on
and dividing both sides by x gives
X21 ,
X
We can see that this (like the analogous statem ent in logic)
is self- r efer ential: the r oot- value of x that we seek m ust be put
back into the expr ession fr om which we seek it.
M er e inspection shows us that x m ust be a for m of unity, or
the equation would not balance num er ically. We have assum ed
only two for m s of unity, +1 and - 1 , so we m ay now tr y them
each in tur n. S et x = +1 . This gives
+1 =%=1
which is clear ly par adoxical. S o set x = - 1 . This tim e we have
_ _;L_
1 - _1 - +1
and it is equally par adoxical.
Of cour se, as ever ybody knows, the par adox in this case is
r esolved by intr oducing a four th class of num ber , called im agi-
nar y, so that we can say the r oots of the equation above ar e
i i, wher e i is a new kind of unity that consists of a squar e r oot
of m inus one.
What we do in Chapter 1 1 is extend the concept to Boolean
algebr as, which m eans that a valid ar gum ent m ay contain not
just thr ee classes of statem ent, but four : tr ue, false, m eaning-
less, and im aginar y. The im plications of this, in the elds of
logic, philosophy, m athem atics, and even physics, ar e pr o-
found.
What is fascinating about the im aginar y Boolean values,
once we adm it them , is the light they appar ently shed on our
concepts of m atter and tim e. It is, I guess, in the natur e of us
all to wonder why the univer se appear s just the way it does.
Why, for exam ple, does it not appear m or e sym m etr ical? Well,
xi
lRlil<'/\ ( l-I 'l'() 'l lll-I FIRST AMERICAN Eomow
was with some trepidation that I approached him in 1 967
with the proof that it was unnecessary. To my relief he was
delighted. The Theory was, he said, the most arbitrary thing
he and W hitehead had ever had to do, not really a theory but
a stopgap, and he was glad to have lived long enough to see
the matter resolved.
Put as simply as I can make it, the resolution is as follows.
All we have to show is that the self-referential paradoxes, dis-
carded with the Theory of Types, are no worse than similar
self-referential paradoxes, which are considered quite accept-
able, in the ordinary theory of equations.
The most famous such paradox in logic is in the statement,
This statement is false.
Suppose we assume that a statement falls into one of three
categories, true, false, or meaningless, and that a meaningful
statement that is not true must be false, and one that is not
false must be true. The statement under consideration does
not appear to be meaningless (some philosophers have claimed
that it is, but it is easy to refute this), so it must be true or
false. If it is true, it must be, as it says, false. But if it is false,
since this is what it says, it must be true.
It has not hitherto been noticed that we have an equally
vicious paradox in ordinary equation theory, because we have
carefully guarded ourselves against expressing it this way. Let
us now do so.
W e will make assumptions analogous to those above. W e
assume that a number can be either positive, negative, or zero.
W e assume further that a nonzero number that is not positive
must be negative, and one that is not negative must be posi-
tive. W e now consider the equation
x2 + 1 I O.
Transposing, we have
x2 I -1 ,
X
PRl FA(Il 'I () TH E FIRST AMERICAN Eomou
and dividing both sides by x gives
X21 ,
X
W e can see that this (like the analogous statement in logic)
is self-referential: the root-value of x that we seek must be put
hack into the expression from which W e seek it.
Mere inspection shows us that x must be a form of unity, or
the equation would not balance numerically. W e have assumed
only two forms of unity, +1 and -1 , so we may now try them
each in turn. Set x = +1 . This gives
+1 = %= 1
which is clearly paradoxical. So set x = -1 . This time we have
_ _;L_
1 -_1 -+1
and it is equally paradoxical.
Of course, as everybody knows, the paradox in this case is
resolved by introducing a fourth class of number, called imagi-
nary, so that we can say the roots of the equation above are
i i, where i is a new kind of unity that consists of a square root
of minus one.
W hat we do in Chapter 1 1 is extend the concept to Boolean
algebras, which means that a valid argument may contain not
just three classes of statement, but four: true, false, meaning-
less, and imaginary. The implications of this, in the elds of
logic, philosophy, mathematics, and even physics, are pro-
found.
W hat is fascinating about the imaginary Boolean values,
once we admit them, is the light they apparently shed on our
concepts of matter and time. It is, I guess, in the nature of us
all to wonder why the universe appears just the W ay it does.
W hy, for example, does it not appear more symmetrical? W ell,
xi
l|u- 1 i=i\<'i- : 'r o 'llll- I FIRS T A M ERICA N Eom on
it you will be kind enough, and patient enough, to bear with
m e thr ough the ar gum ent as it develops itself in this text, you
will I think see, even though we begin it as sym m etr ically as
we know how, that it becom es, of its own accor d, less and less
so as we pr oceed.
G S PENCER BROWN
Cam br idge, England
M aundy Thur sday 1 972
xii ,
PREFA CE
The explor ation on which this wor k r ests was begun towar ds
the end of 1 959. The subsequent r ecor d of it owes m uch, in its
ear ly stages, to the fr iendship and encour agem ent of Lor d
Russell, who was one of the few m en at the beginning who
could see a value in what I pr oposed to do. It owes equally, at
a later stage, to the gener ous help of Dr J C P M iller , Fellow
of Univer sity College and Lectur er in M athem atics in the
Univer sity of Cam br idge, who not only r ead the successive sets
of pr inter s pr oofs, but also acted as an ever - available m entor
and guide, and m ade m any suggestions to im pr ove the style
and accur acy of both text and context.
In 1 963 I accepted an invitation of M r H G Fr ost, S taff
Lectur er in Physical S ciences in the Depar tm ent of Extr a- m ur al
S tudies in the Univer sity of London, to give a cour se of lectur es
on the m athem atics of logic. The cour se was later extended and
r epeated annually at the Institute of Com puter S cience in
Gor don S quar e, and fr om it spr ang som e of the context in the
notes and appendices of this essay. I was also enabled, thr ough
the help of successive classes of pupils, to extend and shar pen
the text.
Other s helped, but cannot, alas, all be m entioned. Of these
the publisher s (including their r eader s and their technical ar tist)
wer e par ticular ly cooper ative, as wer e the pr inter s, and, befor e
this, M r s Peter Br agg under took the exacting task of pr epar ing
a typescr ipt. Finally I should m ention the fact that an or iginal
im petus to the wor k cam e fr om M r I V Idelson, Gener al
M anager of S im on- M EL Distr ibution Engineer ing, the tech-
niques her e r ecor ded being r st developed not in r espect of
questions of logic, but in r esponse to cer tain unsolved pr oblem s
in engineer ing.
Richm ond, A ugust 1 968
xiii
l |u-1 i= /\ <'i-: 'ro 'l llI-I FIRST AMliRI(fAN Eomou
it you will be kind enough, and patient enough, to bear with
me through the argument as it develops itself in this text, you
will I think see, even though we begin it as symmetrically as
we know how, that it becomes, of its own accord, less and less
so as we proceed.
G SPENCER BROW N
Cambridge, England
Maundy Thursday 1 972
xii ,
PREFACE
The exploration on which this work rests was begun towards
the end of 1 959. The subsequent record of it owes much, in its
early stages, to the friendship and encouragement of Lord
Russell, who was one of the few men at the beginning who
could see a value in what I proposed to do. It owes equally, at
zt later stage, to the generous help of Dr J C P Miller, Fellow
of University College and Lecturer in Mathematics in the
University of Cambridge, who not only read the successive sets
of printers proofs, but also acted as an ever-available mentor
and guide, and made many suggestions to improve the style
and accuracy of both text and context.
In 1 963 I accepted an invitation of Mr H G Frost, Staff
Lecturer in Physical Sciences in the Department of Extra-mural
Studies in the University of London, to give a course of lectures
on the mathematics of logic. The course was later extended and
repeated annually at the Institute of Computer Science in
Gordon Square, and from it sprang some of the context in the
notes and appendices of this essay. I was also enabled, through
the help of successive classes of pupils, to extend and sharpen
the text.
Others helped, but cannot, alas, all be mentioned. Of these
the publishers (including their readers and their technical artist)
were particularly cooperative, as were the printers, and, before
this, Mrs Peter Bragg undertook the exacting task of preparing
a typescript. Finally I should mention the fact that an original
impetus to the work came from Mr I V Idelson, General
Manager of Simon-MELDistribution Engineering, the tech-
niques here recorded being rst developed not in respect of
questions of logic, but in response to certain unsolved problems
in engineering.
Richmond, August 1 968
xiii
PRlil~A (.E
A cknowledgm ent
The author and publisher s acknowledge the kind per m ission of
M r J Lust, of the Univer sity of London S chool of Or iental and
A fr ican S tudies, to photogr aph par t of a facsim ile copy of the
1 2th centur y Fukien pr int of the Tao T Ching in the old
Palace M useum , Peking.
XIV
INTRODUCTION
A pr incipal intention of this essay is to separ ate what ar e
known as algebr as of logic fr om the subject of logic, and to
r e- align them with m athem atics.
S uch algebr as, com m only called Boolean, appear m yster ious
because accounts of their pr oper ties at pr esent r eveal nothing
of any m athem atical inter est about their ar ithm etics. Ever y
algebr a has an ar ithm etic, but Boole designedl his algebr a to
t logic, which is a possible inter pr etation of it, and cer tainly
not its ar ithm etic. Later author s have, in this r espect, copied
Boole, with the r esult that nobody hither to appear s to have
m ade any sustained attem pt to elucidate and to study the pr im ar y,
non- num er ical ar ithm etic of the algebr a in ever yday use which
now bear s Booles nam e.
When I r st began, som e seven year s ago, to see that such
a study was needed, I thus found m yself upon what was,
m athem atically speaking, untr odden gr ound. I had to explor e
it inwar ds to discover the m issing pr inciples. They ar e of gr eat
depth and beauty, as we shall pr esently see.
In r ecor ding this account of them , I have aim ed to wr ite
so that ever y special ter m shall be either dened or m ade clear
by its context. I have assum ed on the par t of the r eader no
m or e than a knowledge of the English language, of counting,
and of how num ber s ar e com m only r epr esented. I have allowed
m yself the liber ty of wr iting som ewhat m or e technically in this
intr oduction and in the notes and appendices which follow the
text, but even her e, since the subject is of such gener al inter est,
I have endeavour ed, wher e possible, to keep the account within
the r each of a non- specialist.
A ccounts of Boolean algebr as have up to now been based on
sets of postulates. We m ay take a postulate to be a statem ent
1 Geor ge Boole, The m athem atical analysis oflogic, Cam br idge, 1 847.
XV
litEi~ AcE
Acknowledgment
The author and publishers acknowledge the kind permission of
Mr J Lust, of the University of London School of Oriental and
African Studies, to photograph part of a facsimile copy of the
1 2th century Fukien print of the Tao T Ching in the old
Palace Museum, Peking.
XIV
INTRODUCTION
A principal intention of this essay is to separate what are
known as algebras of logic from the subject of logic, and to
re-align them with mathematics.
Such algebras, commonly called Boolean, appear mysterious
because accounts of their properties at present reveal nothing
of any mathematical interest about their arithmetics. Every
algebra has an arithmetic, but Boole designedl his algebra to
t logic, which is a possible interpretation of it, and certainly
not its arithmetic. Later authors have, in this respect, copied
Boole, with the result that nobody hitherto appears to have
made any sustained attempt toelucidate and to study theprimary,
non-numerical arithmetic of the algebra in everyday use which
now bears Booles name.
W hen I rst began, some seven years ago, to see that such
a study was needed, I thus found myself upon what was,
mathematically speaking, untrodden ground. I had to explore
it inwards to discover the missing principles. They are of great
depth and beauty, as we shall presently see.
In recording this account of them, I have aimed to write
so that every special term shall be either dened or made clear
by its context. I have assumed on the part of the reader no
more than a knowledge of the English language, of counting,
and of how numbers are commonly represented. I have allowed
myself the liberty of writing somewhat more technically in this
introduction and in the notes and appendices which follow the
text, but even here, since the subject is of such general interest,
I have endeavoured, where possible, to keep the account within
the reach of a non-specialist.
Accounts of Boolean algebras have up to now been based on
sets of postulates. W e may take a postulate to be a statement
1 George Boole, The mathematical analysis of logic, Cambridge, 1 84 7.
XV
INTRODUCTION
which is accepted without evidence, because it belongs to a set of
such statem ents fr om which it is possible to der ive other state-
m ents which it happens to be convenient to believe. The chief
char acter istic which has always m ar ked such statem ents has been
an alm ost total lack of any spontaneous appear ance of tr uthz.
Nobody pr etends, for exam ple, that S heffer s equations3 ar e
m athem atically evident, for their evidence is not appar ent apar t
fr om the usefulness of equations which follow fr om them . But in
the pr im ar y ar ithm etic developed in this essay, the initial equa-
tions can be seen to r epr esent two ver y sim ple laws of indication
which, whatever our views on the natur e of their self- evidence,
at least r ecom m end them selves to the ndings of com m on
sense. I am thus able to pr esent (A ppendix 1 ), appar ently for
the r st tim e, pr oofs of each of S heffer s postulates, and hence
of all Boolean postulates, as theor em s about an axiom atic
system which is seen to r est on the fundam ental gr ound of
m athem atics.
Wor king outwar ds fr om this fundam ental sour ce, the gener al
for m of m athem atical com m unication, as we under stand it
today, tends to gr ow quite natur ally under the hand that wr ites
it. We have a denite system , we nam e its par ts, and we adopt,
in m any cases, a single sym bol to r epr esent each nam e. In
doing this, for m s of expr ession ar e called inevitably out of the
need for them , and the pr oofs of theor em s, which ar e at r st
seen to be little m or e than a r elatively infor m al dir ection of
attention to the com plete r ange of possibilities, becom e m or e
and m or e r ecognizably indir ect and for m al as we pr oceed
fr om our or iginal conception. A t the half- way point the algebr a,
in all its r epr esentative com pleteness, is found to have gr own
im per ceptibly out of the ar ithm etic, so that by the tim e we have
star ted to wor k in it we ar e alr eady fully acquainted with its
for m alities and possibilities without anywher e having set out
with the intention of descr ibing them as such.
One of the m er its of this for m of pr esentation is the gr adual
building up of m athem atical notions and com m on for m s of
pr ocedur e without any appar ent br eak fr om com m on sense.
2 Cf A lfr ed Nor th Whitehead and Ber tr and Russell, Pr incipia m athe-
m atica, Vol. I, 2nd edition, Cam br idge, 1 927, p v.
3 H enr y M aur ice S heffer , Tr ans. A m er . M ath. S oc., 1 4 (1 91 3 ) 481 - 8.
xvi
INTRODUCTION
The discipline of m athem atics is seen to be a way, power ful
in com par ison with other s, of r evealing our inter nal knowledge
of the str uctur e of the wor ld, and only by the way associated
with our com m on ability to r eason and com pute.
Even so, the or der ly developm ent of m athem atical conven-
tions and for m ulations stage by stage has not been without its
pr oblem s on the r ever se side. A per son with m athem atical
tr aining, who m ay autom atically use a whole r ange of tech-
niques without questioning their or igin, can nd him self in
difculties over an ear ly par t of the pr esentation, in which it
has been necessar y to develop an idea using only such m athe-
m atical tools as have alr eady been identied. In som e of these
cases we need to der ive a concept for which the pr ocedur es and
techniques alr eady developed ar e only just adequate. The ar gu-
m ent, which is m axim ally elegant at such a point, m ay thus be
conceptually diicult to follow.
One such case, occur r ing in Chapter 2, is the der ivation of the
second of the two pr im itive equations of the calculus of indica-
tions. Ther e seem s to be such univer sal diiculty in following the
ar gum ent at this point, that I have r estated it less elegantly
in the notes on this chapter at the end of the text. When this
is done, the ar gum ent is seen to be so sim ple as to be alm ost
m athem atically tr ivial. But it m ust be r em em ber ed that, accor d-
ing to the r igor ous pr ocedur e of the text, no pr inciple m ay be
used until it has been either called into being or justied in
ter m s of other pr inciples alr eady adopted. In this par ticular
instance, we m ake the ar gum ent easy by using or dinar y sub-
stitution. But at the stage in the essay wher e it becom es neces-
sar y to for m ulate the second pr im itive equation, no pr inciple
of substitution has yet been called into being, since its use and
justication, which we nd later in the essay itself, depends in
par t upon the existence of the ver y equation we want to establish.
In A ppendix 2, I give a br ief account of som e of the sim plica-
tions which can be m ade thr ough using the pr im ar y algebr a
as an algebr a of logic. For exam ple, ther e ar e no pr im itive
pr opositions. This is because we have a basic fr eedom , not
gr anted to other algebr as of logic, of access to the ar ithm etic
whenever we please. Thus each of Whitehead and Russells ve
pr im itive im plications [2, pp 96- 7] can be equated m athem atically
xvii
INTRODUCTION
which is accepted without evidence, because it belongs to a set of
such statements from which it is possible to derive other state-
ments which it happens to be convenient to believe. The chief
characteristic whichhas always marked such statements has been
an almost total lack of any spontaneous appearance of truthz.
Nobody pretends, for example, that Sheffers equations3 are
mathematically evident, for their evidence is not apparent apart
fromthe usefulness of equations which follow fromthem. But in
the primary arithmetic developed in this essay, the initial equa-
tions can be seen to represent two very simple laws of indication
which, whatever our views on the nature of their self-evidence,
at least recommend themselves to the ndings of common
sense. I am thus able to present (Appendix 1 ), apparently for
the rst time, proofs of each of Sheffers postulates, and hence
of all Boolean postulates, as theorems about an axiomatic
system which is seen to rest on the fundamental ground of
mathematics.
W orking outwards from this fundamental source, the general
form of mathematical communication, as we understand it
today, tends to grow quite naturally under the hand that writes
it. W e have a denite system, we name its parts, and we adopt,
in many cases, a single symbol to represent each name. In
doing this, forms of expression are called inevitably out of the
need for them, and the proofs of theorems, which are at rst
seen to be little more than a relatively informal direction of
attention to the complete range of possibilities, become more
and more recognizably indirect and formal as we proceed
from our original conception. At the half-way point the algebra,
in all its representative completeness, is found to have grown
imperceptibly out of the arithmetic, so that by the time we have
started to work in it we are already fully acquainted with its
formalities and possibilities without anywhere having set out
with the intention of describing them as such.
One of the merits of this form of presentation is the gradual
building up of mathematical notions and common forms of
procedure without any apparent break from common sense.
2 Cf Alfred North W hitehead and Bertrand Russell, Principia mathe-
matica, Vol. I, 2nd edition, Cambridge, 1 927, p v.
3 H enry Maurice Sheffer, Trans. Amer. Math. Soc., 1 4 (1 91 3 ) 4 81 -8.
xvi
INTRODUCTION
The discipline of mathematics is seen to be a way, powerful
in comparison with others, of revealing our internal knowledge
of the structure of the world, and only by the way associated
with our common ability to reason and compute.
Even so, the orderly development of mathematical conven-
tions and formulations stage by stage has not been without its
problems on the reverse side. A person with mathematical
training, who may automatically use a whole range of tech-
niques without questioning their origin, can nd himself in
difculties over an early part of the presentation, in which it
has been necessary to develop an idea using only such mathe-
matical tools as have already been identied. In some of these
cases we need to derive a concept for which the procedures and
techniques already developed are only just adequate. The argu-
ment, which is maximally elegant at such a point, may thus be
conceptually diicult to follow.
One such case, occurring in Chapter 2, is the derivation of the
second of the two primitive equations of the calculus of indica-
tions. There seems to be such universal difficulty in following the
argument at this point, that I have restated it less elegantly
in the notes on this chapter at the end of the text. W hen this
is done, the argument is seen to be so simple as to be almost
mathematically trivial. But it must be remembered that, accord-
ing to the rigorous procedure of the text, no principle may be
used until it has been either called into being or justied in
terms of other principles already adopted. In this particular
instance, we make the argument easy by using ordinary sub-
stitution. But at the stage in the essay where it becomes neces-
sary to formulate the second primitive equation, no principle
of substitution has yet been called into being, since its use and
justication, which we nd later in the essay itself, depends in
part upon the existence ofthe very equation we want to establish.
In Appendix 2, I give a brief account of some of the simplica-
tions which can be made through using the primary algebra
as an algebra of logic. For example, there are no primitive
propositions. This is because we have a basic freedom, not
granted to other algebras of logic, of access to the arithmetic
whenever we please. Thus each of W hitehead and Russells ve
primitive implications [2, pp 96-7] can be equated mathematically
xvii
I.N|'R()|)UCTl()N
with a single constant. The constant, if it wer e a pr oposition,
would be the pr im itive im plication. But in fact, being ar ithm etical,
it cannot r epr esent a pr oposition.
A point of inter est in this connexion is the developm ent of
the idea of a var iable solely fr om that of the oper ative constant.
This com es fr om the fact that the algebr a r epr esents our
ability to consider the for m of an ar ithm etical equation ir r espec-
tive of the appear ance, or other wise, of this constant in cer tain
specied places. A nd since, in the pr im ar y ar ithm etic, we ar e
not pr esented, appar ently, with two kinds of constant, such as
5, 6, etc and +, X, etc, but with expr essions m ade up, appar -
ently, of sim ilar constants each with a single pr oper ty, the
conception of a var iable com es fr om consider ing the ir r elevant
pr esence or absence of this pr oper ty. This lends suppor t to the
view, suggested4 by Wittgenstein, that var iables in the calculus
of pr opositions do not in fact r epr esent the pr opositions in
an expr ession, but only the tr uth- functions of these pr opositions,
since the pr opositions them selves cannot be equated with the
m er e pr esence or absence of a given pr oper ty, while the
possibility of their being tr ue or not tr ue can.
A nother point of inter est is the clear distinction, with the
pr im ar y algebr a and its ar ithm etic, that can be dr awn between
the pr oof of a theor em and the dem onstr ation of a consequence.
The concepts of theor em and consequence, and hence of pr oof
and dem onstr ation, ar e widely confused in cur r ent liter atur e,
wher e the wor ds ar e used inter changeably. This has undoubtedly
cr eated spur ious diiculties. A s will be seen in the statem ent of
the com pleteness of the pr im ar y algebr a (theor em 1 7), what is
to be pr oved becom es str ikingly clear when the distinction is
pr oper ly m aintained. (A sim ilar confusion is appar ent, especi-
ally in the liter atur e of sym bolic logic, of the concepts of axiom
and postulate.)
It is possible to develop the pr im ar y algebr a to such an
extent that it can be used as a r estr icted (or even as a full)
algebr a of num ber s. Ther e ar e sever al ways of doing this, the
m ost convenient of which I have found is to lim it condensation
4 Ludwig Wittgenstein, Tr actatus logico- philosophicus, London, 1 922,
pr opositions 5 sq.
xviii
INTRODUCTION
in the ar ithm etic, and thus to use Va num ber of cr osses in a given
space to r epr esent either the cor r esponding num ber or its im age.
When this is done it is possible to see plainly som e at least of
the evidence for Godels and Chur chs theor em s of decision.
But with the r ehabilitation of the par adoxical equations under -
taken in Chapter 1 1 , the m eaning and application of these
theor em s now stands in need of r eview. They cer tainly appear
less destr uctive than was hither to supposed.
I aim ed in the text to car r y the developm ent only so far as
to be able to consider r easonably fully all the for m s that em er ge
at each stage. A lthough I indicate the expansion into com plex
for m s in Chapter 1 1 , I other wise tr y to lim it the developm ent
so as to r ender the account, as far as it goes, com plete.
M ost of the theor em s ar e or iginal, at least as theor em s, and
their pr oofs ther efor e new. But som e of the later algebr aic
and m ixed theor em s, occur r ing in what is at this stage fam iliar
gr ound, ar e alr eady known and have, in other for m s, been
pr oved befor e. In all of these cases I have been able to nd what
seem to be clear er , sim pler , or m or e dir ect pr oofs, and in m ost
cases the theor em s I pr ove ar e m or e gener al. For exam ple, the
near est appr oach to m y theor em 1 6 seem s to be a weaker and
less centr al theor em appar ently r st pr ovedl by Quine, as a
lem m a to a com pleteness pr oof for a pr opositional calculus.
It was only after contem plating this theor em for som e two year s
that I found the beautiful key by which it is seen to be tr ue for
all possible algebr as, Boolean or other wise.
In ar r iving at pr oofs, I have often been str uck by the appar ent
alignm ent of m athem atics with psycho- analytic theor y. In
each discipline we attem pt to nd out, by a m ixtur e of con-
tem plation, sym bolic r epr esentation, com m union, and com -
m unication, what it is we alr eady know. In m athem atics, as
in other for m s of self- analysis, we do not have to go explor ing
the physical wor ld to nd what we ar e looking for . A ny child
of ten, who can m ultiply and divide, alr eady knows, for exam ple,
5 Kur t Godel, M onatshefte fur M athem atik und Physik, 3 8 (1 93 1 )
1 7298.
6 A lonzo Chur ch, J. S ym bolic Logic, 1 (1 93 6) 401 , lOl2.
7 W V Quine, J. S ym bolic Logic, 3 (1 93 8) 3 7- 40.
xix
I.N'|'R()|)UCTI()N
with a single constant. The constant, if it were a proposition,
would be the primitiveimplication. But in fact, being arithmetical,
it cannot represent a proposition.
A point of interest in this connexion is the development of
the idea of a variable solely from that of the operative constant.
This comes from the fact that the algebra represents our
ability to consider the form of an arithmetical equation irrespec-
tive of the appearance, or otherwise, of this constant in certain
specied places. And since, in the primary arithmetic, we are
not presented, apparently, with two kinds of constant, such as
5, 6, etc and +, X, etc, but with expressions made up, appar-
ently, of similar constants each with a single property, the
conception of a variable comes from considering the irrelevant
presence or absence of this property. This lends support to the
view, suggested4 by W ittgenstein, that variables in the calculus
of propositions do not in fact represent the propositions in
an expression, but only the truth-functions ofthese propositions,
since the propositions themselves cannot be equated with the
mere presence or absence of a given property, while the
possibility of their being true or not true can.
Another point of interest is the clear distinction, with the
primary algebra and its arithmetic, that can be drawn between
the proof of a theorem and the demonstration of a consequence.
The concepts of theorem and consequence, and hence of proof
and demonstration, are widely confused in current literature,
where the words are used interchangeably. This has undoubtedly
created spurious diiculties. As will be seen in the statement of
the completeness of the primary algebra (theorem 1 7), what is
to be proved becomes strikingly clear when the distinction is
properly maintained. (A similar confusion is apparent, especi-
ally in the literature of symbolic logic, of the concepts of axiom
and postulate.)
It is possible to develop the primary algebra to such an
extent that it can be used as a restricted (or even as a full)
algebra of numbers. There are several ways of doing this, the
most convenient of which I have found is to limit condensation
4 Ludwig W ittgenstein, Tractatus logico-philosophicus, London, 1 922,
propositions 5 sq.
xviii
INTRODUCTION
in the arithmetic, and thus to use Va number of crosses in a given
space to represent either the corresponding number or its image.
W hen this is done it is possible to see plainly some at least of
the evidence for Godels and Churchs theorems of decision.
But with the rehabilitation of the paradoxical equations under-
taken in Chapter 1 1 , the meaning and application of these
theorems now stands in need of review. They certainly appear
less destructive than was hitherto supposed.
I aimed in the text to carry the development only so far as
to be able to consider reasonably fully all the forms that emerge
at each stage. Although I indicate the expansion into complex
forms in Chapter ll, I otherwise try to limit the development
so as to render the account, as far as it goes, complete.
Most of the theorems are original, at least as theorems, and
their proofs therefore new. But some of the later algebraic
and mixed theorems, occurring in what is at this stage familiar
ground, are already known and have, in other forms, been
proved before. In all of these cases I have been able to nd what
seem to be clearer, simpler, or more direct proofs, and in most
cases the theorems I prove are more general. For example, the
nearest approach to my theorem 1 6 seems to be a weaker and
less central theorem apparently rst provedl by Quine, as a.
lemma to a completeness proof for a propositional calculus.
It was only after contemplating this theoremfor some two years
that I found the beautiful key by which it is seen to be true for
all possible algebras, Boolean or otherwise.
In arriving at proofs, I have often been struck by the apparent
alignment of mathematics with psycho-analytic theory. In
each discipline we attempt to nd out, by a mixture of con-
templation, symbolic representation, communion, and com-
munication, what it is we already know. In mathematics, as
in other forms of self-analysis, we do not have to go exploring
the physical world to nd what we are looking for. Any child
of ten, who can multiply and divide, already knows, for example,
5 Kurt Godel, Monatshefte fur Mathematik und Physik, 3 8 (1 93 1 )
1 72-98.
6 Alonzo Church, J. Symbolic Logic, 1 (1 93 6) 4 0-1 , 1 01 -2.
7 W V Quine, J. Symbolic Logic, 3 (1 93 8) 3 7-4 0.
xix
|N'lR()l)U(il'l()N
that the sequence of pr im e num ber s is endless. But if he is not
shown Euclids pr oof, it is unlikely that he will ever nd out,
befor e he dies, that he knows.
This analogy suggests that we have a dir ect awar eness of
m athem atical for m as an ar chetypal str uctur e. I tr y in the nal
chapter to illustr ate the natur e of this awar eness. In any case,
questions of pur e pr obability alone would lead us to suppose
that som e degr ee of dir ect awar eness is pr esent thr oughout
m athem atics.
We m ay take it that the num ber of statem ents which m ight
or m ight not be pr ovable is unlim ited, and it is evident that,
in any lar ge enough nite sam ple, untr ue statem ents, of those
bear ing any useful degr ee of signicance, heavily outnum ber
tr ue statem ents. Thus in pr inciple, if ther e wer e no innate
sense of r ightness, a m athem atician would attem pt to pr ove
m or e false statem ents than tr ue ones. But in pr actice he seldom
attem pts to pr ove any statem ent unless he is alr eady convinced
of its tr uth. A nd since he has not yet pr oved it, his conviction
m ust ar ise, in the r st place, fr om consider ations other than
pr oof.
Thus the codication of a pr oof pr ocedur e, or of any other
dir ective pr ocess, although at r st useful, can later stand as a
thr eat to fur ther pr ogr ess. For exam ple, we m ay consider the
lar gely unconscious, but now codied, lim itation of the r eason-
ing (as distinct fr om the com putative) par ts of pr oof str uctur es
to the solution of Boolean equations of the r st degr ee. A s
we see in Chapter 1 1 , and in the notes ther eto, the solution of
equations of higher degr ee is not only possible, but has been
under taken by switching engineer s on an ad hoc basis for som e
half a centur y or m or e. S uch equations have hither to been
excluded fr om the subject m atter of or dinar y logic by the White-
head- Russell theor y of types [2, pp 3 7 sq, e.g. p 77] .
I show in the text that we can constr uct an im plicit function
of itself so that it r e- enter s its own space at either an odd or an
even depth. In the for m er case we nd the possibility of a self-
denying equation of the kind these author s descr ibe. In such
a case, the r oots of the equation so set up ar e im aginar y. But
in the latter ease we nd a self- conr m ing equation which is
XX
j;/t .
ll
INTRODUCTION
satised, for som e given congur ation of the var iables, by two
r eal r oots.
I am able, by this consider ation, to r ehabilitates the for m al
str uctur e hither to discar ded with the theor y of types. A s we
now see, the str uctur e can be identied in the m or e gener al
theor y of equations, behind which ther e alr eady exists a weight
of m athem atical exper ience.
One pr ospect of such a r ehabilitation, which could r epay
fur ther attention, com es fr om the fact that, although Boolean
equations of the r st degr ee can be fully r epr esented on a plane
sur face, those of the second degr ee cannot be so r epr esented.
In gener al, an equation of degr ee k r equir es, for its r epr esenta-
tion, a sur face of genus k 1 . DJ S pencer Br own and I found
evidence, in unpublished wor k under taken in 1 9625, suggesting
that both the four - colour theor em and Goldbachs theor em ar e
undecidable with a pr oof str uctur e conned to Booleanequations
of the r st degr ee, but decidable if we ar e pr epar ed to avail
our selves of equations of higher degr ee.
One of the m otives pr om pting the fur ther ance of the pr esent
wor k was the hope of br inging together the investigations of the
inner str uctur e of our knowledge of the univer se, as expr essed
in the m athem atical sciences, and the investigations of its
outer str uctur e, as expr essed in the physical sciences. H er e the
wor k of Einstein, S chr odinger , and other s seem s to have led
to the r ealization of an ultim ate boundar y of physical knowledge
in the for m of the m edia thr ough which we per ceive it. It
becom es appar ent that if cer tain facts about our com m on
exper ience of per ception, or what we m ight call the inside
wor ld, can be r evealed by an extended study of what we call,
in contr ast, the outside wor ld, then an equally extended study
of this inside wor ld will r eveal, in tur n, the facts r st m et with
in the wor ld outside: for what we appr oach, in either case, fr om
one side or the other , is the com m on boundar y between them .
I do not pr etend to have car r ied these r evelations ver y far ,
8 For a histor y of the ear lier essays to r ehabilitate, on a logical r ather
than on a m athem atical basis, som ething of what was discar ded, see
A br aham A Fr aenkel and Y ehoshua Bar - H illel, Foundations of set
theor y, A m ster dam , 1 958, pp 1 3 6- 95.
xxi
|N'l R()I)U(i l'l()N
that the sequence of prime numbers is endless. But if he is not
shown Euclids proof, it is unlikely that he will ever nd out,
before he dies, that he knows.
This analogy suggests that we have a direct awareness of
mathematical form as an archetypal structure. I try in the nal
chapter to illustrate the nature of this awareness. In any case,
questions of pure probability alone would lead us to suppose
that some degree of direct awareness is present throughout
mathematics.
W e may take it that the number of statements which might
or might not be provable is unlimited, and it is evident that,
in any large enough nite sample, untrue statements, of those
bearing any useful degree of signicance, heavily outnumber
true statements. Thus in principle, if there were no innate
sense of rightness, a mathematician would attempt to prove
more false statements than true ones. But in practice he seldom
attempts to prove any statement unless he is already convinced
of its truth. And since he has not yet proved it, his conviction
must arise, in the rst place, from considerations other than
proof.
Thus the codication of a proof procedure, or of any other
directive process, although at rst useful, can later stand as a
threat to further progress. For example, we may consider the
largely unconscious, but now codied, limitation of the reason-
ing (as distinct from the computative) parts of proof structures
to the solution of Boolean equations of the rst degree. As
we see in Chapter 1 1 , and in the notes thereto, the solution of
equations of higher degree is not only possible, but has been
undertaken by switching engineers on an ad hoe basis for some
half a century or more. Such equations have hitherto been
excluded fromthe subject matter of ordinary logic by the W hite-
head-Russell theory of types [2, pp 3 7 sq, e.g. p 77] .
I show in the text that we can construct an implicit function
of itself so that it re-enters its own space at either an odd or an
even depth. In the former case we nd the possibility of a self-
denying equation of the kind these authors describe. In such
a case, the roots of the equation so set up are imaginary. But
in the latter case we nd a self-conrming equation which is
XX
j;/t .
ll
INTRODUCTION
satised, for some given conguration of the variables, by two
real roots.
I am able, by this consideration, to rehabilitates the formal
structure hitherto discarded with the theory of types. As we
now see, the structure can be identied in the more general
theory of equations, behind which there already exists a weight
of mathematical experience.
One prospect of such a rehabilitation, which could repay
further attention, comes from the fact that, although Boolean
equations of the rst degree can be fully represented on a plane
surface, those of the second degree cannot be so represented.
In general, an equation of degree k requires, for its representa-
tion, a surface of genus k - 1 . DJ Spencer Brown and I found
evidence, in unpublished work undertaken in 1 962-5, suggesting
that both the four-colour theorem and Goldbachs theorem are
undecidable witha proof structure conned to Booleanequations
of the rst degree, but decidable if we are prepared to avail
ourselves of equations of higher degree.
One of the motives prompting the furtherance of the present
work was the hope of bringing together the investigations of the
inner structure of our knowledge of the universe, as expressed
in the mathematical sciences, and the investigations of its
outer structure, as expressed in the physical sciences. H ere the
work of Einstein, Schrodinger, and others seems to have led
to the realization of an ultimate boundary ofphysical knowledge
in the form of the media through which we perceive it. It
becomes apparent that if certain facts about our common
experience of perception, or what we might call the inside
world, can be revealed by an extended study of what we call,
in contrast, the outside world, then an equally extended study
of this inside world will reveal, in turn, the facts rst met with
in the world outside: for what we approach, in either case, from
one side or the other, is the common boundary between them.
I do not pretend to have carried these revelations very far,
8 For a history of the earlier essays to rehabilitate, on a logical rather
than on a mathematical basis, something of what was discarded, see
Abraham A Fraenkel and Yehoshua Bar-H illel, Foundations of set
theory, Amsterdam, 1 958, pp 1 3 6-95.
xxi
lN'l'l{()DU(T|()N
or that other s, better equipped, could not car r y them fur ther .
l hope they will. M y conscious intention in wr iting this essay
was the elucidation of an indicative calculus, and its latent
potential, becom ing m anifest only when the r ealization of this
intention was alr eady well advanced, took m e by sur pr ise.
I br eak off the account at the point wher e, as we enter the
thir d dim ension of r epr esentation with equations of degr ee
higher than unity, the connexion with the basic ideas of the
physical wor ld begins to com e m or e str ongly into view. I had
intended, befor e I began wr iting, to leave it her e, since the
latent for m s that em er ge at this, the four th depar tur e fr om the
pr im ar y for m (or the fth depar tur e, if we count fr om the
void) ar e so m any and so var ied that I could not hope to pr esent
them all, even cur sor ily, in one book.
M edawar obser ves that the standar d for m of pr esentation
r equir ed of an or dinar y scientic paper r epr esents the ver y
r ever se of what the investigator was in fact doing. In r eality,
says M edawar , the hypothesis is r st posited, and becom es the
m edium thr ough which cer tain other wise obscur e facts, later
to be collected in suppor t of it, ar e r st clear ly seen. But the
account in the paper is expected to give the im pr ession that
such facts r st suggested the hypothesis, ir r espective of whether
this im pr ession is tr uly r epr esentative.
In m athem atics we see this pr ocess in r ever se. The m athe-
m atician, m or e fr equently than he is gener ally allowed to adm it,
pr oceeds by exper im ent, inventing and tr ying out hypotheses
to see if they t the facts of r easoning and com putation with
which he is pr esented. When he has found a hypothesis which
ts, he is expected to publish an account of the wor k in the
r ever se or der , so as to deduce the facts fr om the hypothesis.
I would not r ecom m end that we should do other wise, in
either eld. By all accounts, to tell the stor y backwar ds is
convenient and saves tim e. But to pr etend that the stor y was
actually lived backwar ds can be extr em ely m ystifying.
In view of this appar ent r ever sal, Laing suggests) that what
9 P B M edawar , Is the S cientic Paper a Fr aud, The Listener , 1 2th
S eptem ber 1 963 , pp 3 77- 8.
1 R D Laing, The politics of exper ience and the bir d of par adise,
London, 1 967, pp 52 sq.
xxii
INTRODUCTION
in em pir ical science ar e called data, being in a r eal sense ar bi-
lr iir ily chosen by the natur e of the hypothesis alr eady for m ed,
could m or e honestly be called capta. By r ever se analogy, the
facts of m athem atical science, appear ing at r st to be ar bitr ar ily
chosen, and thus capta, ar e not r eally ar bitr ar y at all, but
absolutely deter m ined by the natur e and coher ence of our being.
ln this view we m ight consider the facts of m athem atics to be
the r eal data of exper ience, for only these appear to be, in the
linal analysis, inescapable.
A lthough I have under taken, to the best of m y ability, to
pr eser ve, in the text itself, what is thus inescapable, and ther eby
tim eless, and other wise to discar d what is tem por al, I am under
no illusion of having entir ely succeeded on either count. That
one can not, in such an under taking, succeed per fectly, seem s
to m e to r eside in the m anifest im per fection of the state of
par ticular existence, in any for m at all. (Cf A ppendix 2.) The
wor k of any hum an author m ust be to som e extent idio-
syncr atic, even though he m ay know his per sonal ego to be but a
fashionable gar b to suit the m ode of the pr esent r ather than the
m ean of past and futur e in which his wor k will com e to r est.
To this extent, m ode or fashion is inevitable at the expense of
m ean or m eaning, or ther e can be no connexion of what is
per ipher al, and has to be r egar ded, with what is centr al, and
has to be divined.
A m ajor aspect of the language of m athem atics is the degr ee
of its for m ality. A lthough it is tr ue that we ar e concer ned, in
m athem atics, to pr ovide a shor thand for what is actually said,
this is only half the stor y. What we aim to do, in addition, is
to pr ovide a m or e gener al for m in which the or dinar y language
of exper ience is seen to r est. A s long as we conne our selves to
the subject at hand, without extending our consider ation to
what it has in com m on with other subjects, we ar e not availing
our selves of a tr uly m athem atical m ode of pr esentation.
What is encom passed, in m athem atics, is a tr anscedence fr om
a given state of vision to a new, and hither to unappar ent, vision
beyond it. When the pr esent existence has ceased to m ake sense,
it can still com e to sense again thr ough the r ealization of its
for m .
xxiii
INTRODUCTION
Thus the subject m atter of logic, however sym bolically
tr eated, is not, in as far as it connes itself to the gr ound of
logic, a m athem atical study. It becom es so only when we ar e
able to per ceive its gr ound as a par t of a m or e gener al for m ,
in a pr ocess without end. Its m athem atical tr eatm ent is a
tr eatm ent of the for m in which our way of talking about our
or dinar y living exper ience can be seen to be cr adled. It is the
laws of this for m , r ather than those of logic, that I have attem pted
to r ecor d.
In m aking the attem pt, I found it easier to acquir e an access
to the laws them selves than to deter m ine a satisfactor y way
of com m unicating them . In gener al, the m or e univer sal the
law, the m or e it seem s to r esist expr ession in any par ticular
m ode.
S om e of the difculties appar ent in r eading, as well as in
wr iting, the ear lier par t of the text com e fr om the fact that, fr om
Chapter 5 backwar ds, we ar e extending the analysis thr ough
and beyond the point of sim plicity wher e language ceases to
act nor m ally as a cur r ency for com m unication. The point at
which this br eak fr om nor m al usage occur s is in fact the point
wher e algebr as ar e or dinar ily taken to begin. To extend them
back beyond this point dem ands a consider able unlear ning of
the cur r ent descr iptive super str uctur e which, until it is unlear ned,
can be m istaken for the r eality.
The fact that, in a book, we have to use wor ds and other
sym bols in an attem pt to expr ess what the use of wor ds and
other sym bols has hither to obscur ed, tends to m ake dem ands
of an extr aor dinar y natur e on both wr iter and r eader , and I
am conscious, on m y side, of how im per fectly I succeed in
r ising to them . But at least, in the pr ocess of under taking the
task, I have becom e awar e (as Boole him self becam e awar e)
that what I am tr ying to say has nothing to do with m e, or
anyone else, at the per sonal level. It, as it wer e, r ecor ds itself
and, whatever the faults in the r ecor d, that which is so r ecor ded
is not a m atter of opinion. The only cr edit I feel entitled to accept
in r espect of it is for the instr um ental labour of m aking a r ecor d
which m ay, if God so disposes, be ar ticulate and coher ent enough
to be under stood in its tem por al context.
London, A ugust 1 967
xxiv