You are on page 1of 206

THE PHILOSOPHY OF PHYSICAL SCIENCE

PREFACE
This book contains the substance of the course of lectures which I
eli!ere as Tarner Lecturer of Trinit" Colle#e Ca$bri#e in the Easter
Ter$ %&'() The lectures ha!e affore $e an o**ortunit" of
e!elo*in# $ore full" than in $" earlier books the *rinci*les of
*hiloso*hic thou#ht associate with the $oern a!ances of *h"sical
science)+
It is often sai that there is no ,*hiloso*h" of science,- but onl" the
*hiloso*hies of certain scientists) .ut in so far as we reco#ni/e an
authoritati!e bo" of o*inion which ecies what is- an what is not
acce*te as *resent+a" *h"sics- there is an ascertainable *resent+
a" *hiloso*h" of *h"sical science) It is the *hiloso*h" to which those
who follow the acce*te *ractice of science stan co$$itte b" their
practice) It is i$*licit in the $ethos b" which the" a!ance science-
so$eti$es without full" unerstanin# wh" the" e$*lo" the$- an in
the *roceure which the" acce*t as #i!in# assurance of truth- often
without e0a$inin# what kin of assurance it can #i!e)
There shoul be no conflict between the clai$ that a *hiloso*h" is
scientificall" #roune an the clai$ that it is- so far as it #oes- a true
*hiloso*h") .ut in a s*eciali/e work of this kin the *ri$ar" ob1ect
$ust be to ascertain an iscuss the *hiloso*h" which- whether true
or not- is the *resent *hiloso*h" of *h"sical science in the sense
state abo!e) Those of us who belie!e that science- notwithstanin#
continual failures an rea1ust$ents- is slowl" rawin# nearer to the
truth- are content that *hiloso*hic truth shoul be reache b" the sa$e
$etho of *ro#ressi!e a!ance)
In orer to $ake sure of our scientific founations it is foun necessar"
to enter rather ee*l" into the *rinci*les of relati!it" theor" an
2uantu$ theor") Since the intention is to #i!e- not $erel" an
e0*osition- but a 1ustification of the !iews to which the" lea- so$e
*arts of the book introuce $atters of consierable technical ifficult")
3enerall" I ha!e abstaine fro$ $athe$atical for$ulae4 this- howe!er-
is not wholl" out of consieration for the #eneral reaer- but because
those whose $ins are too $uch i$$erse in $athe$atical for$ulae
are likel" to $iss what we are here seekin#)
The iscussion- althou#h relatin# to the sa$e sub1ect $atter- is $ainl"
on ifferent lines fro$ that #i!en ele!en "ears a#o in The Nature of the
Physical World) The startin# *oint in the *resent treat$ent is
knowle#e) The title of the earlier book $i#ht ha!e been e0*ane
into ,the nature of the *h"sical uni!erse- with a**lications to the theor"
of *h"sical knowle#e,4 the corres*onin# title of the *resent book
woul be ,the nature of *h"sical knowle#e- with a**lications to the
theor" of the *h"sical uni!erse,) The chan#e of e$*hasis $akes for a
$ore lo#ical se2uence of ieas4 but *ri$aril" it reflects a chan#e which
has occurre in *h"sical science itself) It is si#nificant of this chan#e
that the contrast between the scientific table an the fa$iliar table- with
which The Nature of the Physical World o*ens- ha beco$e a contrast
between the scientific stor" an the fa$iliar stor" of e0*erience at the
be#innin# of New Pathways in Science) The first was- I belie!e- the
natural for$ of e0*ression accorin# to the scientific outlook of %&5(4
the secon ha beco$e $ore natural si0 "ears later)
Neither the scientific a!ances of the last ecae nor the "ears of
reflection ha!e altere the #eneral tren of $" *hiloso*h") I sa" ,$"
*hiloso*h",- not as clai$in# authorshi* of ieas which are wiel"
iffuse in $oern thou#ht- but because the ulti$ate selection an
s"nthesis $ust be a *ersonal res*onsibilit") If it were necessar" to #i!e
a short na$e to this *hiloso*h"- I shoul hesitate between ,Selecti!e
sub1ecti!is$, an ,Structuralis$,) The for$er na$e refers to the
as*ect $ost *ro$inent in the first ei#ht cha*ters4 the latter refers to a
$ore $athe$atical conce*tion which o$inates the rest of the book)
.oth can now be carrie $uch farther than in The Nature of the
Physical World) The o$ain of sub1ecti!it" has been e0tene as a
conse2uence of our better unerstanin# of 2uantu$ $echanics4 an
the conce*tion of structure has been $ae $ore *recise b" the
connection now reco#ni/e between the founations of *h"sics an
the $athe$atical Theor" of 3rou*s)
6ith this ,*hiloso*h" of *h"sical science, as a nucleus- I enea!or in
the last two cha*ters to e!elo* the outline of a #eneral *hiloso*hical
outlook which a scientist can acce*t without inconsistenc") I a$ not
a$on# those who think that in the search for truth all as*ects of
hu$an e0*erience are to be i#nore sa!e those which are followe u*
in *h"sical science) .ut I fin no ishar$on" between a *hiloso*h"
which e$braces the wier si#nificance of hu$an e0*erience an the
s*eciali/e *hiloso*h" of *h"sical science- e!en thou#h the latter
relates to a s"ste$ of thou#ht of recent #rowth whose stabilit" is "et to
be teste)
A) S) E)
CA7.RI83E- A*ril %&'&)
CHAPTER I
SCIENTIFIC EPISTE7OLO3Y
I
.ET6EEN *h"sics an *hiloso*h" there lies a ebatable territor"
which I shall call scientific epistemology) E*iste$olo#" is that branch
of *hiloso*h" which treats of the nature of knowle#e) It will not be
enie that a si#nificant *art of the whole fiel of knowle#e is that
which has co$e to us b" the $ethos of *h"sical science) This *art
takes the for$ of a etaile escri*tion of a worl + the so+calle
*h"sical uni!erse) I #i!e the na$e ,scientific e*iste$olo#", to the sub+
branch of e*iste$olo#" which eals with the nature of this *art of our
knowle#e- an therefore inirectl" with the nature an status of the
*h"sical uni!erse to which it for$all" relates)
There are two $atters of efinition which it is esirable to $ake clear
at the outset)
So$e writers restrict the ter$ ,knowle#e, to thin#s of which we are
2uite certain4 others reco#ni/e knowle#e of !ar"in# e#rees of
uncertaint") This is one of the co$$on a$bi#uities of s*eech as to
which no one is entitle to ictate- an an author can onl" state which
usa#e he has hi$self chosen to follow) If ,to know, $eans ,to be 2uite
certain of,- the ter$ is of little use to those who wish to be uno#$atic)
I therefore *refer the broaer $eanin#4 an $" own usa#e will
reco#ni/e uncertain knowle#e) An"thin# which woul be knowle#e if
we were assure of its truth- is still counte as knowle#e 9uncertain or
false knowle#e: if we are not assure)
It will not be necessar" for us to for$ulate a #eneral efinition of
knowle#e) Our *roceure will be to s*ecif" a *articular collection of
$ore or less wiel" acce*te knowle#e- an then to $ake an
e*iste$olo#ical stu" of its nature) Es*eciall"- thou#h not e0clusi!el"-
we ha!e to consier the knowle#e ac2uire b" the $ethos of
*h"sical science) For bre!it" I will call this physical knowledge) In
*rinci*le we $i#ht ientif" *h"sical knowle#e with the contents of
certain enc"clo*aeic works- such as the Handbuch der Physik- which
between the$ co!er the !arious branches of *h"sical science) .ut
there are ob!ious ob1ections to a sla!ish acce*tance of a *articular
authorit"4 an I will therefore efine *h"sical knowle#e to be that
which a ri#ht+thinkin# *erson woul toa" acce*t as 1ustifie b"
*h"sical science)
It shoul not be o!erlooke that *h"sical knowle#e inclues a !ast
a$ount of $iscellaneous infor$ation which woul be out of *lace in
scientific te0tbooks) For e0a$*le- the result of a $easure$ent of
wei#ht is *h"sical knowle#e- whether it is $ae for the *ur*ose of
eciin# a scientific issue or for eciin# the a$ount of a traes$an;s
bill) The conition is that it shall be *asse as scientificall" correct 9b"
the ri#ht+thinkin# *erson:- not that it shall be scientificall" i$*ortant) It
shoul also be notice that the ter$ is intene to refer to *h"sical
science as it stans toa") 6e are not #oin# to occu*" oursel!es with
s*eculations as to *ossible future e!elo*$ents) 6e are to take stock
of the results which the $ethos of *h"sical science ha!e "iele u*
to now- an see what kin of knowle#e we ha!e been ac2uirin#)
I ha!e sai that I o not re#ar the ter$ ,knowle#e, as i$*l"in#
assurance of truth) .ut in consierin# a *articular bo" of knowle#e-
it $a" be assu$e that an effort has been $ae to a$it to that bo"
onl" the $ore trustworth" knowle#e4 so that usuall" a reasonable
e#ree of certaint" or *robabilit" is attributable to the knowle#e which
we shall ha!e occasion to iscuss) .ut the assess$ent of certaint" of
knowle#e is to be re#are as se*arate fro$ the stu" of the nature
of knowle#e)
The other $atter of efinition is the ter$ ,*h"sical uni!erse,) Ph"sical
knowle#e 9as acce*te an for$ulate toa": has the for$ of a
escri*tion of a worl) 6e define the *h"sical uni!erse to be the worl
so escribe) Effecti!el" therefore the *h"sical uni!erse is efine as
the the$e of a s*ecifie bo" of knowle#e- 1ust as 7r) Pickwick $i#ht
be efine as the hero of a s*ecifie no!el)
A #reat a!anta#e of this efinition is that it oes not *re1u#e the
2uestion whether the *h"sical uni!erse + or 7r) Pickwick + reall" e0ists)
That is left o*en for iscussion if we can a#ree on a efinition of ,reall"
e0ists,- which for $ost *ersons is a *arrot+*hrase whose $eanin# the"
ha!e not trouble to consier) The few who ha!e atte$*te to #i!e it a
efinite $eanin# o not alwa"s a#ree on the $eanin#) ." efinin# the
*h"sical uni!erse an the *h"sical ob1ects which constitute it as the
the$e of a s*ecifie bo" of knowle#e- an not as thin#s *ossessin#
a *ro*ert" of e0istence elusi!e of efinition- we free the founations of
*h"sics fro$ sus*icion of $eta*h"sical conta$ination)
This t"*e of efinition is characteristic of the e*iste$olo#ical a**roach-
which takes knowle#e as the startin# *oint rather than an e0istent
entit" of which we ha!e so$ehow to obtain knowle#e) .ut in efinin#
scientificall" a ter$ alrea" in co$$on use- we $ust be careful to
a!oi abuse of lan#ua#e) To 1ustif" the abo!e efinition of the *h"sical
uni!erse- we ou#ht to show that it is not in conflict with what the
orinar" $an 9in which ter$ I o not inclue *hiloso*hers:
unerstans b" the *h"sical uni!erse) This 1ustification is eferre to *)
%<&)
II
The nature of *h"sical knowle#e an of the worl which it *rofesses
to escribe has lon# been a battle#roun for ri!al schools of
*hiloso*hers) .ut *h"sicists can scarcel" be enie a hearin# on a
sub1ect which concerns the$ so inti$atel") A stuent of *h"sical
science shoul be in a *osition to throw so$e li#ht on the nature of the
knowle#e obtainable b" the $ethos which he *ractises) Recentl" a
nu$ber of books ha!e been written b" authors whose 2ualifications
are *urel" scientific- in which scientific e*iste$olo#" is e!elo*e an
use as an a**roach to the wier *roble$s of *hiloso*h") I o not
think that this ,intrusion, into *hiloso*h" is a $atter for sur*rise or
caustic co$$ent)
One often fins an i$*ression that it is an inno!ation for scientists to
inul#e in *hiloso*h"4 but this in incorrect) I ha!e notice that so$e of
the recent books are *lentifull" s*rinkle with 2uotations fro$
scientists of the nineteenth centur" which- whether the" fortif" the
ar#u$ent or not- *ro!e at an" rate that our *reecessors share the
co$$on foible of holin# stron# *hiloso*hic !iews + an e0*ressin#
the$) So$e were out of their e*th- then as now) .ut so$e were
*rofoun thinkers + Cliffor- =arl Pearson- Poincare- an others +
whose writin#s ha!e an honore *lace in the e!elo*$ent of scientific
*hiloso*h")
It is- howe!er- i$*ortant to reco#ni/e that about twent"+fi!e "ears a#o
the in!asion of *hiloso*h" b" *h"sics assu$e a ifferent character)
>* till then traffic with *hiloso*h" ha been a lu0ur" for those scientists
whose is*osition ha**ene to turn that wa") I can fin no inication
that the scientific researches of Pearson an Poincare were in an"
wa" ins*ire or #uie b" their *articular *hiloso*hical outlook) The"
ha no o**ortunit" to *ut their *hiloso*h" into *ractice) Con!ersel"-
their *hiloso*hical conclusions were the outco$e of #eneral scientific
trainin#- an were not to an" e0tent e*enent on fa$iliarit" with
reconite in!esti#ations an theories) To a!ance science an to
*hiloso*hi/e on science were essentiall" istinct acti!ities) In the new
$o!e$ent scientific e*iste$olo#" is $uch $ore inti$atel" associate
with science) For e!elo*in# the $oern theories of $atter an
raiation a efinite e*iste$olo#ical outlook has beco$e a necessit"4
an it is the irect source of the $ost far+reachin# scientific a!ances)
6e ha!e isco!ere that it is actually an aid in the search for
knowledge to understand the nature of the knowledge which we seek)
." $akin# *ractical a**lication of our e*iste$olo#ical conclusions we
sub1ect the$ to the sa$e kin of obser!ational control as *h"sical
h"*otheses) If our e*iste$olo#" is at fault- it will lea to an i$*asse in
the scientific e!elo*$ents *roceein# fro$ it4 that warns us that our
*hiloso*hical insi#ht has not been ee* enou#h- an we $ust cast
about to fin what has been o!erlooke) In this wa" scientific
a!ances which result fro$ e*iste$olo#ical insi#ht ha!e in turn
eucate our e*iste$olo#ical insi#ht) .etween science an scientific
e*iste$olo#" there has been a #i!e an take b" which both ha!e
#reatl" benefite)
In the !iew of scientists at least- this obser!ational control #i!es to
$oern scientific e*iste$olo#" a securit" which *hiloso*h" has not
usuall" been able to attain) It introuces also the sa$e kin of
*ro#ressi!e e!elo*$ent which is characteristic of science- but not
hitherto of *hiloso*h") 6e are not $akin# a series of shots at ulti$ate
truth- which $a" hit or $iss) 6hat we clai$ for the *resent s"ste$ of
scientific *hiloso*h" is that it is an a!ance on that which went before-
an that it is a founation for the a!ances which will co$e after it)
In science the obser!ational test is !aluable- not onl" for controllin#
*h"sical h"*otheses 9for which it is inee the onl" *ossible
#uarantee:- but also for etectin# fallacies of ar#u$ent an
unwarrante assu$*tions) It is the latter kin of control that an
obser!ational test a**lies to scientific e*iste$olo#") This $a" see$
su*erfluous to those who ne!er reason incorrectl") .ut *erha*s e!en
the $ost confient *hiloso*her will a$it that there are so$e of his
o**onents to who$ such control woul be salutar") I ha!e little oubt
that e!er" one of the *hiloso*hical conclusions in this book has been
antici*ate b" one of the schools of *hiloso*h" + an e$*haticall"
cone$ne b" another) .ut to those who reco#ni/e the$ as fa$iliar
truis$s or as lon#+cone$ne fallacies- I woul *oint out that the" are
now *ut forwar with alto#ether new sanctions which ou#ht to be
reckone with)
Theoretical *h"sicists- throu#h the inesca*able e$ans of their own
sub1ect- ha!e been force to beco$e e*iste$olo#ists- 1ust as *ure
$athe$aticians ha!e been force to beco$e lo#icians) The in!asion
of the e*iste$olo#ical branch of *hiloso*h" b" *h"sics is e0actl"
*arallel to the in!asion of the lo#ical branch of *hiloso*h" b"
$athe$atics) Pure $athe$aticians- ha!in# learnt b" e0*erience that
the ob!ious is ifficult to *ro!e + an not alwa"s true + foun it
necessar" to el!e into the founations of their own *rocesses of
reasonin#4 in so oin# the" e!elo*e a *owerful techni2ue which has
been welco$e for the a!ance$ent of lo#ic #enerall") A si$ilar
*ressure of necessit" has cause *h"sicists to enter into e*iste$olo#"-
rather a#ainst their will) 7ost of us- as *lain $en of science- be#in with
an a!ersion to the *hiloso*hic t"*e of in2uir" into the nature of thin#s)
6hether we are *ersuae that the nature of *h"sical ob1ects is
ob!ious to co$$onsense- or whether we are *ersuae that it is
inscrutable be"on hu$an unerstan- we are incline to is$iss the
in2uir" as un*ractical an futile) .ut $oern *h"sics has not been able
to $aintain this aloofness) There can be little oubt that its a!ances-
thou#h a**l"in# *ri$aril" to the restricte fiel of scientific
e*iste$olo#"- ha!e a wier bearin#- an offer an effecti!e contribution
to the *hiloso*hical outlook as a whole)
For$all" we $a" still reco#ni/e a istinction between science- as
treatin# the content of knowle#e- an scientific e*iste$olo#"- as
treatin# the nature of knowle#e of the *h"sical uni!erse) .ut it is no
lon#er a *ractical *artition4 an to confor$ to the *resent situation
scientific e*iste$olo#" shoul be inclue in science) 6e o not
is*ute that it $ust also be inclue in *hiloso*h") It is a fiel in which
*hiloso*h" an *h"sics o!erla*)
III
So lon# as a scientific writer on *hiloso*h" confines hi$self to
scientific e*iste$olo#"- he is not outsie the borers of his own
sub1ect) .ut $ost authors ha!e felt that the" coul usefull" a!ance
farther an consier the #eneral *hiloso*hical bearin# of the new
conce*tions) This !entureso$eness has been stron#l" critici/e4 but it
see$s to $e that the critics ha!e faile to #ras* the situation)
It is recore that Archbisho* 8a!ison- in con!ersation with Einstein-
aske hi$ what effect he thou#ht the theor" of relati!it" woul ha!e on
reli#ion) Einstein answere? ,None) Relati!it" is a *urel" scientific
theor"- an has nothin# to o with reli#ion), In those a"s one ha to
beco$e e0*ert in o#in# *ersons who were *ersuae that the fourth
i$ension was the oor to s*iritualis$- an the hast" e!asion is not
sur*risin#) .ut those who 2uote an a**lau the re$ark as thou#h it
were one of Einstein;s $ost $e$orable utterances o!erlook a #larin#
fallac" in it) Natural selection is a *urel" scientific theor") If in the earl"
a"s of 8arwinis$ the then Archbisho* ha aske what effect the
theor" of natural selection woul ha!e on reli#ion- ou#ht the answer to
ha!e been ,None) The 8arwinian theor" is a *urel" scientific theor"-
an has nothin# to o with reli#ion,@
The co$*art$ents into which hu$an thou#ht is i!ie are not so
water+ti#ht that funa$ental *ro#ress in one is a $atter of inifference
to the rest) The #reat chan#e in theoretical *h"sics which be#an in the
earl" "ears of the *resent centur" is a *urel" scientific e!elo*$ent4
but it $ust affect the #eneral current of hu$an thou#ht- as at earlier
ti$es the Co*ernican an the Newtonian s"ste$s ha!e one) This
alone woul see$ to 1ustif" the scientific authors in takin# a broa !iew
of their task) It see$s to $e unreasonable to $aintain that the workin#
out of these wier i$*lications of the new conce*tion of the *h"sical
uni!erse shoul be left entirel" to those who o not unerstan it)
Not so !er" lon# a#o the sub1ect now calle *h"sics was known as
,natural *hiloso*h",) The *h"sicist is b" ori#in a *hiloso*her who has
s*eciali/e in a *articular irection) .ut he is not the onl" !icti$ of
s*eciali/ation) ." the breakin# awa" of *h"sics the $ain bo" of
*hiloso*h" suffere an a$*utation) In *ractice- if not in theor"-
acae$ic *hiloso*h" has also beco$e s*eciali/e- an is no lon#er
co+e0tensi!e with the s"ste$ of thou#ht an knowle#e b" which we
orient oursel!es towars our $oral an $aterial en!iron$ent) To a
$an;s *hiloso*h" in the broaest sense + to his religio vitae + natural
*hiloso*h"- uner the na$e of science- has continue to be a
*owerful- *erha*s e!en a *reo$inant- contributor) It woul be ifficult
to *oint to an" e!elo*$ent in acae$ic *hiloso*h" which has ha so
#reat an influence on $an;s outlook as the #rowth of the scientific
theor" of e!olution) In the last twent" "ears it has been the turn of
*h"sics to reassert itself as natural *hiloso*h"4 an I belie!e that the
new contribution of *h"sical science- if full" #ras*e- is not less
si#nificant than the octrine of e!olution)
6e $a" efine rather $ore closel" the status of a scientist who writes
on the *hiloso*hical outco$e of $oern *h"sical theories) I o not
think that an" school of *hiloso*hers is *re*are to wash its hans of
the *h"sical uni!erse an lea!e the *h"sicists to $ake what the" like
of it) It see$s therefore to be a#ree that scientific e*iste$olo#" is still
an inte#ral *art of *hiloso*h") Those whose work lies in the
e*iste$olo#ical e!elo*$ents of $oern *h"sics $ust therefore be
counte as s*ecialists in one of the e*art$ents into which *hiloso*h"
is i!ie + a e*art$ent not far fro$ the heart of the sub1ect) In their
iscussion of *hiloso*h" as a whole the" are likel" to is*la" the faults
of a s*ecialist who fins hi$self outsie his own #roo!e4 but the" are
not co$$on intruers) The e!ils of s*eciali/ation woul- I think- be still
$ore *ronounce if the" $ae no atte$*t to correlate with the rest of
*hiloso*h" the *ro#ress that has been $ae in their own e*art$ent)
Scientific e*iste$olo#" is the $ain the$e of these lectures) 6e shall
consier it *ri$aril" fro$ the scientific as*ect) .ut we shall also at
ti$es enea!or to !iew it in its #eneral settin# as a re#ion of o!erla* of
*h"sics an *hiloso*h"- an trace its conse2uences in both fiels)
IA
For the truth of the conclusions of *h"sical science- obser!ation is the
su*re$e Court of A**eal) It oes not follow that e!er" ite$ which we
confientl" acce*t as *h"sical knowle#e has actuall" been certifie
b" the Court4 our confience is that it woul be certifie b" the Court if
it were sub$itte) .ut it oes follow that e!er" ite$ of *h"sical
knowle#e is of a for$ which $i#ht be sub$itte to the Court) It $ust
be such that we can s*ecif" 9althou#h it $a" be i$*racticable to carr"
out: an obser!ational *roceure which woul ecie whether it is true
or not) Clearl" a state$ent cannot be teste b" obser!ation unless it is
an assertion about the results of obser!ation) E!er" ite$ of *h"sical
knowle#e $ust therefore be an assertion of what has been or woul
be the result of carr"in# out a s*ecifie obser!ational *roceure)
I o not think that an"one + least of all- those who are critical of the
$oern tenencies of *h"sics + will isa#ree with the first a0io$ of
scientific e*iste$olo#"- na$el" that the knowle#e obtaine b" the
$ethos of *h"sical science is li$ite to obser!ational knowle#e in
the sense e0*laine abo!e) 6e o not en" that knowle#e which is
not of an obser!ational nature $a" e0ist- e)#) the theor" of nu$bers in
*ure $athe$atics4 an non+co$$ittall" we $a" allow the *ossibilit" of
other for$s of insi#ht of the hu$an $in into a worl outsie itself) .ut
such knowle#e is be"on the borers of *h"sical science- an
therefore oes not enter into the escri*tion of the worl introuce in
the for$ulation of *h"sical knowle#e) To a wier s"nthesis of
knowle#e- of which *h"sical knowle#e is onl" a *art- we $a"
*erha*s correlate a ,worl, of which the *h"sical uni!erse is onl" a
*artial as*ect) .ut at this sta#e of our in2uir" we li$it the iscussion to
*h"sical knowle#e- an therefore to a *h"sical uni!erse fro$ which-
b" efinition- all characteristics which are not the sub1ect of *h"sical
knowle#e are e0clue)
A istinction is co$$onl" $ae between obser!ational an theoretical
knowle#e4 but in *ractice the ter$s are use so loosel" as to e*ri!e
the classification of all real si#nificance) The whole e!elo*$ent of
*h"sical science has been a *rocess of co$binin# theor" an
obser!ation4 an in #eneral e!er" ite$ of *h"sical knowle#e + or at
least e!er" ite$ to which attention is orinaril" irecte + has a *artl"
obser!ational an *artl" theoretical basis) The istinction- so far as it
can be $ae- has reference to the $oe of obtainin# the knowle#e +
to the nature of the e!ience for its truth) It oes not concern the
knowle#e itself + what it we inten to assert) Thus our a0io$ that all
*h"sical knowle#e is of an obser!ational nature is not to be
unerstoo as e0cluin# theoretical knowle#e) I know the *osition of
Bu*iter last ni#ht) That is knowle#e of an obser!ational nature4 it is
*ossible to etail the obser!ational *roceure which "iels the
2uantities 9ri#ht ascension an eclination: which e0*ress $"
knowle#e of the *lanet;s *osition) As a $atter of fact I i not follow
this *roceure- nor i I learn the *osition fro$ an"one who ha
followe the *roceure4 I looke it u* in the Nautical Al$anac) That
#a!e $e the result of a co$*utation accorin# to *lanetar" theor")
Present+a" *h"sics acce*ts that theor" an all its conse2uences4 that
is to sa"- it a$its the calculate *osition as a foreknowle#e of the
results which woul be obtaine b" carr"in# out the reco#ni/e
obser!ational *roceure) Of $" two *ieces of knowle#e- na$el"
knowle#e of the results of a $athe$atical co$*utation an
foreknowle#e of the results of an obser!ational *roceure- it is the
latter which I assert when I clai$ to know the *osition of Bu*iter) If- on
sub$ission to the Court of A**eal- $" foreknowle#e of the result of
the obser!ational *roceure *ro!es to be incorrect- I shall ha!e to
a$it that I was $istaken an i not know the *osition of Bu*iter4 it
will be no use $" ur#in# that $" knowle#e of the result of the
$athe$atical co$*utation was correct)
It is the essence of acce*tance of a theor" that we a#ree to obliterate
the istinction between knowle#e eri!e fro$ it an knowle#e
eri!e fro$ actual obser!ation) It $a" see$ one+sie that the
obliteration of the istinction shoul rener all *h"sical knowle#e
obser!ational in nature) .ut not e!en the $ost e0tre$e worshi**er of
theor" has *ro*ose the re!erse + that in acce*tin# the results of an
obser!ational research as trustworth" we ele!ate the$ to the status of
theoretical conclusions) The one+sieness is ue to our acce*tance
of obser!ation- not theor"- as the su*re$e Court of A**eal)
A
6e ha!e seen that e!er" ite$ of *h"sical knowle#e- whether eri!e
fro$ obser!ation or theor" or fro$ a co$bination of both- is an
assertion of what has been or woul be the result of carr"in# out a
s*ecifie obser!ational *roceure) 3enerall" it is an assertion of what
would be the result if an obser!ation were $ae4 for this reason it is
$ore accurate to escribe *h"sical knowle#e as hypothetico-
observational) Occasionall" the h"*othetical for$ can be ro**e + the
obser!ation has been $ae an the result obtaine + but the
*ro*ortion of knowle#e to which this a**lies is s$all- an $ostl"
uninterestin#) I a$ not en"in# the i$*ortance of actual obser!ation
as a source of knowle#e4 but as a constituent of scientific knowle#e
it is al$ost ne#li#ible) 6hene!er in the *rocess of reucin#
obser!ations a ,correction, is a**lie- obser!ational knowle#e of an
actual e0*eri$ent is re*lace b" h"*othetico+obser!ational knowle#e
of what woul ha!e been the result of an e0*eri$ent uner $ore ieal
conitions)
Consier- for e0a$*le- our knowle#e that the istance of the $oon is
about 5CD-DDD $iles) The e0act $eanin# of this assertion $ust be
ascertaine b" reference to the efinition of istance in *h"sics an
astrono$" 9Cha*ter A:4 but- accuratel" enou#h for *resent *ur*oses-
what we clai$ to know is that 5CD-DDD E %FGD "ar+sticks *lace en
to en woul reach fro$ here to the $oon) This is h"*othetico+
obser!ational knowle#e4 for certainl" no one has carrie out the
e0*eri$ent) It is true that actual obser!ations were e$*lo"e in
arri!in# at the fi#ure 5CD-DDD $iles4 but a*art fro$ theor" we shoul
not know that the resultin# 2uantit" was the istance of the $oon)
There are a !ariet" of *ractical $ethos of finin# the istance4 one of
the $ost accurate in!ol!es inter alia swin#in# *enulu$s in ifferent
latitues on the earth) Althou#h it woul be true to assert that 5CD-DDD
$iles is the result of an actual obser!ational *roceure of swin#in#
*enulu$s- etc)- that is not what we inten to assert when we sa" that
the istance of the $oon is 5CD-DDD $iles) ." e$*lo"in# acce*te
theor" we ha!e been able to substitute for the actual obser!ational
*roceure a h"*othetical obser!ational *roceure which woul "iel
the sa$e result if it were carrie out) The #ain is that h"*othetico+
obser!ational knowle#e can be s"ste$ati/e an #athere into a
coherent whole- whereas actual obser!ational knowle#e is s*oraic
an esultor")
One cannot hel* feelin# a $is#i!in# that h"*othetico+obser!ational
knowle#e is not entirel" satisfactor" fro$ a lo#ical stan*oint) 6hat
e0actl" is the status of conitional knowle#e if the conition is not
fulfille@ Can an" sense at all be attribute to a state$ent that if
so$ethin#- which we know i not ha**en- ha ha**ene- then
certain other thin#s woul ha!e ha**ene@ Yet I cannot hel* *ri/in#
$" knowle#e that 5CD-DDD E %FGD "ar+sticks would reach fro$ here
to the $oon- althou#h there is no *ros*ect that the" e!er will o so)
AI
Scientific stu" of the facts of obser!ation has le us to $ake a
nu$ber of #enerali/ations which we call laws of nature) 3enerali/ation
is the $ost cons*icuous source of the h"*othetico+obser!ational
character of *h"sical knowle#e- since it fla#rantl" o!erste*s actual
obser!ation an asserts knowle#e of what woul be obser!e on an"
occasion if the necessar" *roceure were carrie out)
I think it is so$eti$es $aintaine that a law of nature is a
s"ste$ati/ation- not a #enerali/ation- of knowle#e) Ieall" it is
*ossible to acce*t a s"ste$ati/ation of e0istin# obser!ational
knowle#e- without *re1u#in# whether an" future obser!ations will
confor$ to the s"ste$) To a *erson holin# this !iew it shoul be a
co$*lete sur*rise e!er" ti$e a new obser!ation is foun to obe" the
law) For e0a$*le- .oe;s law of *lanetar" istances can be re#are
as a s"ste$ati/e state$ent concernin# the istances of the si0
*lanets known in his ti$e- an not e0*ecte to a**l" to *lanets
isco!ere subse2uentl") This $a" be the ri#ht attitue to ao*t
towars *articular laws that ha!e been enunciate4 but it certainl"
cannot be a**lie #enerall" throu#hout *h"sics) 6e $ust not i$a#ine
that .oe;s s"ste$ati/ation woul continue to be *ossible in a *h"sics
*ur#e of #enerali/ations) >nless we acce*t certain *rior
#enerali/ation- e)#) that li#ht tra!els in strai#ht lines- the istances of
the *lanets cannot be eter$ine4 an .oe;s law then ro*s out-
because there is no $aterial for it to s"ste$ati/e) The fact is that
#enerali/ation fro$ obser!ation has- whether a!ertentl" or not- been
*ractise in *h"sical science fro$ the !er" be#innin#4 an we $ust
re#ar it as no less *art of the scientific $etho than obser!ation itself)
An with the #enerali/ations there has entere into the bo" of
scientific knowle#e a h"*othetico+obser!ational ele$ent- which has
been concee the ri#ht to re$ain)
Our $ain conclusion is that- notwithstanin# the i!ersit" of $etho-
*h"sical knowle#e re$ains ho$o#eneous in its nature4 it is
knowle#e of what woul be the result of an obser!ational *roceure if
it were carrie out- incluin# as a s*ecial case the result of an"
obser!ational *roceure that has been carrie out)
In the *ro#ress of *h"sics ini!iual facts ha!e beco$e lar#el"
$er#e in #enerali/ations) 6oul it be true to sa" that co$*lete
knowle#e of *h"sics woul consist wholl" of such #enerali/ations@
The answer is ifferent accorin# as we refer to *h"sics in a narrow
sense 9incluin# che$istr"- but not astrono$" or other obser!ational-
as istin#uishe fro$ e0*eri$ental- sciences: or to *h"sical science
#enerall") In the narrow sense *h"sics is- I think- concerne solel" with
#enerali/ations) The *h"sicist is not intereste in s*ecial facts e0ce*t
as $aterial for #enerali/ation) If he stuies a *articular lu$* of iron- it
is as a sa$*le e0hibitin# the #eneral *ro*erties of iron) The
astrono$er on the other han is intereste in the *articular lu$* of
$atter on which we ha**en to li!e- whether or not it is a sa$*le of
*lanets #enerall") He is curious about the e0istence of !e#etation on
7ars- when the ne0t bri#ht co$et will a**ear- how closel" a $inor
*lanet a**roache the earth- an so on) It $a" be sai that this is 1ust
an a$ateurish interest which the $ore serious+wine *h"sicist has
out#row4 an astrono$er $ust- of course- ascertain the constants of the
earth- as a *h"sicist $ust ascertain the constants of his #al!ano$eter-
but he has no business to be interested in the$) Astrono$ers will
scarcel" a#ree4 but let that *ass) It is sufficient to sa" that these
s*ecial facts are knowle#e ac2uire b" the $ethos of *h"sical
science- an $ust not be ne#lecte in scientific e*iste$olo#" which
we ha!e efine as the stu" of the nature of the knowle#e that has
co$e to us in this wa"4 nor are the" ne#li#ible in the uni!erse of which
that knowle#e for$s a escri*tion)
6e $ust therefore re$e$ber that not all our knowle#e of the
*h"sical uni!erse is co$*rise in knowle#e of the laws of nature) The
warnin# is not so su*erfluous as it see$s) I ha!e often foun an
i$*ression that to e0*lain awa" the laws of nature as wholl" sub1ecti!e
is the sa$e thin# as to e0*lain awa" the *h"sical uni!erse as wholl"
sub1ecti!e) Such a !iew is alto#ether unfoune)
CHAPTER II
SELECTIAE S>.BECTIAIS7
I
Let us su**ose that an ichth"olo#ist is e0*lorin# the life of the ocean)
He casts a net into the water an brin#s u* a fish" assort$ent)
Sur!e"in# his catch- he *rocees in the usual $anner of a scientist to
s"ste$ati/e what it re!eals) He arri!es at two #enerali/ations?
9%: No sea+creature is less than two inches lon#)
95: All sea+creatures ha!e #ills)
These are both true of his catch- an he assu$es tentati!el" that the"
will re$ain true howe!er often he re*eats it)
In a**l"in# this analo#"- the catch stans for the bo" of knowle#e
which constitutes *h"sical science- an the net for the sensor" an
intellectual e2ui*$ent which we use in obtainin# it) The castin# of the
net corres*ons to obser!ation4 for knowle#e which has not been or
coul not be obtaine b" obser!ation is not a$itte into *h"sical
science)
An onlooker $a" ob1ect that the first #enerali/ation is wron#) ,There
are *lent" of sea+creatures uner two inches lon#- onl" "our net is not
aa*te to catch the$), The ichth"olo#ist is$isses this ob1ection
conte$*tuousl") ,An"thin# uncatchable b" $" net is ipso facto outsie
the sco*e of ichth"olo#ical knowle#e- an is not *art of the kin#o$
of fishes which has been efine as the the$e of ichth"olo#ical
knowle#e) In short- what $" net can;t catch isn;t fish), Or + to translate
the analo#" + ,If "ou are not si$*l" #uessin#- "ou are clai$in# a
knowle#e of the *h"sical uni!erse isco!ere in so$e other wa" than
b" the $ethos of *h"sical science- an a$ittel" un!erifiable b"
such $ethos) You are a $eta*h"sician) .ahH,
The is*ute arises- as $an" is*utes o- because the *rota#onists are
talkin# about ifferent thin#s) The onlooker has in $in an ob1ecti!e
kin#o$ of fishes) The ichth"olo#ist is not concerne as to whether
the fishes he is talkin# about for$ an ob1ecti!e or sub1ecti!e class4 the
*ro*ert" that $atters is that the" are catchable) His #enerali/ation is
*erfectl" true of the class of creatures he is talkin# about + a selecte
class *erha*s- but he woul not be intereste in $akin#
#enerali/ations about an" other class) 8ro**in# analo#"- if we take
obser!ation as the basis of *h"sical science- an insist that its
assertions $ust be !erifiable b" obser!ation- we i$*ose a selecti!e
test on the knowle#e which is a$itte as *h"sical) The selection is
sub1ecti!e- because it e*ens on the sensor" an intellectual
e2ui*$ent which is our $eans of ac2uirin# obser!ational knowle#e) It
is to such sub1ecti!el"+selecte knowle#e- an to the uni!erse which
it is for$ulate to escribe- that the #enerali/ations of *h"sics + the so+
calle laws of nature + a**l")
It is onl" with the recent e!elo*$ent of e*iste$olo#ical $ethos in
*h"sics that we ha!e co$e to realise the far+reachin# effect of this
sub1ecti!e selection of its sub1ect $atter) 6e $a" at first- like the
onlooker- be incline to think that *h"sics has $isse its wa"- an has
not reache the *urel" ob1ecti!e worl which- we take it for #rante- it
was tr"in# to escribe) Its #enerali/ations- if the" refer to an ob1ecti!e
worl- are or $a" be renere fallacious throu#h the selection) .ut
that a$ounts to cone$nin# obser!ationall" #roune science as a
failure because a *urel" ob1ecti!e worl is not to be reache b"
obser!ation)
Clearl" an abanon$ent of the obser!ational $etho of *h"sical
science is out of the 2uestion) Obser!ationall" #roune science has
been b" no $eans a failure4 thou#h we $a" ha!e $isunerstoo the
*recise nature of its success) Those who are issatisfie with an"thin#
but a *urel" ob1ecti!e uni!erse $a" turn to the $eta*h"sicians- who
are not cra$*e b" the self+i$*ose orinance that e!er" assertion
$ust be ca*able of sub$ission to obser!ation as the final Court of
A**eal) .ut we- as *h"sicists- shall continue to stu" the uni!erse
re!eale b" obser!ation an to $ake our #enerali/ations about it4
althou#h we now know that the uni!erse so reache cannot be wholl"
ob1ecti!e) Of course- the #reat $ass of *h"sicists- who *a" no
attention to e*iste$olo#"- woul ha!e #one on oin# this in an" case)
Shoul we then i#nore the onlooker with his su##estion of selection@ I
think not4 thou#h we cannot acce*t his re$e") Su**ose that a $ore
tactful onlooker $akes a rather ifferent su##estion? ,I reali/e that "ou
are ri#ht in refusin# our frien;s h"*othesis of uncatchable fish- which
cannot be !erifie b" an" tests "ou an I woul consier !ali) ."
kee*in# to "our own $etho of stu"- "ou ha!e reache a
#enerali/ation of the hi#hest i$*ortance + to fish$on#ers- who woul
not be intereste in #enerali/ations about uncatchable fish) Since
these #enerali/ations are so i$*ortant- I woul like to hel* "ou) You
arri!e at "our #enerali/ation in the traitional wa" b" e0a$inin# the
fish) 7a" I *oint out that "ou coul ha!e arri!e $ore easil" at the
sa$e #enerali/ation b" e0a$inin# the net an the $etho of usin# it@,
The first onlooker is a $eta*h"sician who es*ises *h"sics on account
of its li$itations4 the secon onlooker is an e*iste$olo#ist who can
hel* *h"sics because of its li$itations) It is 1ust because of the li$ite +
so$e $i#ht sa"- the *er!erte + ai$ of *h"sics that such hel* is
*ossible) The traitional $etho of s"ste$atic e0a$ination of the ata
furnishe b" obser!ation is not the onl" wa" of reachin# the
#enerali/ations !alue in *h"sical science) So$e at least of these
#enerali/ations can also be foun b" e0a$inin# the sensor" an
intellectual e2ui*$ent use in obser!ation) E*iste$olo#" thus
*resents *h"sics with a new $etho of achie!in# its ai$s) The
e!elo*$ent of relati!it" theor"- an the transfor$ation of 2uantu$
theor" fro$ an e$*irical to a rational theor" are the outco$e of the
new $etho4 an in it is our #reat ho*e of further funa$ental
a!ances)
II
6e return to our fish to illustrate another *oint of #reat i$*ortance) No
su##estion was offere as to the secon #enerali/ation + that all sea+
creatures ha!e #ills + an- so far as we can see- it coul not ha!e been
euce fro$ an e0a$ination of the net an its $oe of use) If the
ichth"olo#ist e0tens his in!esti#ations- $akin# further catches-
*erha*s in ifferent waters- he $a" an" a" brin# u* a sea+creature
without #ills an u*set his secon #enerali/ation) If this ha**ens- he
will naturall" be#in to istrust the securit" of his first #enerali/ation) His
fear is neeless4 for the net can ne!er brin# u* an"thin# that it is not
aa*te to catch)
Generaliations that can be reached epistemologically have a security
which is denied to those that can only be reached empirically!
It has been custo$ar" in scientific *hiloso*h" to insist that the laws of
nature ha!e no co$*ulsor" character4 the" are unifor$ities which ha!e
been foun to occur hitherto in our li$ite e0*erience- but we ha!e no
ri#ht to assert that th" will occur in!ariabl" an uni!ersall") This was a
!er" *ro*er *hiloso*h" to ao*t as re#ars e$*irical #enerali/ations +
it bein# unerstoo- of course- that no one woul be so foolish as to
a**l" the *hiloso*h" in *ractice) Scientists- assure b" their
*hiloso*h" that the" ha no ri#ht to e0*ectations- continue to cherish
inefensible e0*ectations- an inter*rete their obser!ations in
accorance with the$)
Atte$*ts ha!e been $ae b" the theor" of *robabilit" to 1ustif" our
e0*ectation that if an occurrence 9whose cause is unknown: has
ha**ene re#ularl" hitherto it will continue to ha**en on the ne0t
occasion4 but I think that all that has e$er#e is an anal"sis an
a0io$ati/ation of our e0*ectation- not a efence of it)
The situation is chan#e when we reco#ni/e that so$e laws of nature
$a" ha!e an e*iste$olo#ical ori#in) These are co$*ulsor"4 an when
their e*iste$olo#ical ori#in is establishe- we ha!e a ri#ht to our
e0*ectation that the" will be obe"e in!ariabl" an uni!ersall") The
*rocess of obser!in#- of which the" are a conse2uence- is
ine*enent of ti$e or *lace)
.ut- it $a" be ob1ecte- can we be sure that the *rocess of obser!in#
%

is unaffecte b" ti$e or *lace@ Strictl" s*eakin#- no) .ut if it is affecte
+ if *osition in ti$e an s*ace or an" other circu$stance *re!ents the
obser!ational *roceure fro$ bein# carrie out *recisel" accorin# to
the reco#ni/e s*ecification + we can 9an o: call the resultin#
obser!ation a ,ba obser!ation,) Those who resent the iea of
co$*ulsion in scientific law $a" *erha*s be $ollifie b" the
concession that- althou#h it can no lon#er be acce*te as a *rinci*le
of scientific *hiloso*h" that the laws of nature are unco$*ulsor"- there
is no co$*ulsion that our actual obser!ations shall satisf" the$- for
9unfortunatel": there is no co$*ulsion that our obser!ations shall be
good obser!ations)
6hat about the re$ainin# laws of nature- not of an e*iste$olo#ical
ori#in- an therefore- so far as we know- non+co$*ulsor"@ 7ust the"
continue to $ar the sche$e as a source of inefensible e0*ectations-
which ne!ertheless are foun to be fulfille in *ractice@ .efore
worr"in# about the$- it will be well to wait till we see what is left of the
s"ste$ of natural law after the *art which can be accounte for
e*iste$olo#icall" has been re$o!e) There $a" not be an"thin# left to
worr" about)
The introuction of e*iste$olo#ical anal"sis in $oern *h"sical theor"
has not onl" been a *owerful source of scientific *ro#ress- but has
#i!en a new kin of securit" to its conclusions) Or- I shoul rather sa"-
it has *ut a new kin of securit" within reach) 6hether the *resent
conclusions are secure is a 2uestion of hu$an fallibilit"- fro$ which the
e*iste$olo#ist is no $ore e0e$*t than the classical theorist or the
1
The stanar s*ecification of the *roceure of obser!in# $ust be sufficientl" etaile to
secure a uni2ue result of the obser!ation) It is the ut" of the obser!er to secure that all
attenant circu$stances which can affect the result- e)#)- te$*erature- absence of
$a#netic fiel- etc)- are in accorance with s*ecification) E*iste$olo#ical laws #o!ernin#
the results of the obser!ation are such as are inferable solel" fro$ the fact that the
*roceure was as s*ecifie) The contin#enc" referre to in this *ara#ra*h is e0e$*lifie
b" the fact that it is i$*ossible to $ake a reall" ,#oo, obser!ation of len#th in a stron#
$a#netic fiel- because the stanar s*ecification of the *roceure of eter$inin# len#th
re2uires us to eli$inate $a#netic fiels 9*) (D:)
*ractical obser!er) 6hilst not for#ettin# that the actual results achie!e
$ust e*en on the insi#ht an accurac" of those who use the
e2ui*$ent- I woul e$*hasi/e that we ha!e now the e2ui*$ent to *ut
theoretical *h"sics on a surer footin# than it for$erl" as*ire to)
III
"uis custodiet ipsos custodes# 6ho will obser!e the obser!ers@ The
answer is + the e*iste$olo#ist) He watches the$ to see what the"
reall" obser!e- which is often 2uite ifferent fro$ what the" sa" the"
obser!e) He e0a$ines their *roceure an the essential li$itations of
the e2ui*$ent the" brin# to their task- an b" so oin# beco$es
aware beforehan of li$itations to which the results the" obtain will
ha!e to confor$) The"- on the other han- onl" isco!er these
li$itations when the" co$e to e0a$ine their results- an- unaware of
their sub1ecti!e ori#in- hail the$ as laws of nature)
It $a" be ar#ue that- in acce*tin# the ai of e*iste$olo#"- *h"sical
science continues to be wholl" an inference fro$ obser!ation4 for the
e*iste$olo#ist too is an obser!er) The astrono$er obser!es stars4 the
e*iste$olo#ist obser!es obser!ers) .oth are seekin# a knowle#e
which rests on obser!ation)
I a$ sorr" I $ust offen the obser!ationalists b" re1ectin# this so* to
traitional !iews4 but the analo#" between obser!in# stars an
obser!in# obser!ers will not hol #oo) The co$$on state$ent that
*h"sical science rests on obser!ation- an that its #enerali/ations are
#enerali/ations about obser!ational ata- is not 2uite the whole truth) It
rests on good obser!ation- an its #enerali/ations are about good
obser!ational ata) Scientific e*iste$olo#"- which is concerne with
the nature of the knowle#e containe in *h"sical science- has
therefore to e0a$ine the *roceure of #oo obser!ation) The *ro*er
counter*art to the e*iste$olo#ist who obser!es #oo obser!ers is the
astrono$er who obser!es #oo stars)
This 2ualification of obser!ers as ,#oo,- which is the first *oint
attene to in *ractice- see$s often to ha!e been o!erlooke in
*hiloso*h") In s*eakin# of obser!ation- there is often a failure to
istin#uish the s*ecial kin of obser!ational acti!it" conte$*late in
*h"sical science fro$ iniscri$inatel" ,takin# notice,) The istinction is
stron#l" selecti!e4 an it inicates one wa" in which the sub1ecti!e
selection- to which we ha!e referre- is introuce into the uni!erse
escribe b" *h"sics) If astrono$ers were si$ilarl" allowe to
istin#uish #oo stars an ba stars- astrono$" woul oubtless be
enriche b" so$e re$arkable new laws + a**l"in#- of course- onl" to
the #oo stars which obe" the laws so *rescribe)
6hether an obser!ation is #oo or ba e*ens on what it *rofesses
to re*resent) A ba eter$ination of the $eltin#+*oint of sul*hur $a"
be an e0cellent eter$ination of the $eltin#+*oint of a $i0ture of
sul*hur an irt) The ter$s use to escribe an obser!ation + to state
what it is an obser!ation of + i$*l" b" their efinition a stanar
*roceure to be followe in $akin# it4 the obser!er *rofesses to follow
this *roceure- or a *roceure which he takes the libert" of substitutin#
for it in the belief that it will assurel" #i!e the sa$e result) If- throu#h
ina!ertence or *ractical ifficult"- the *rescribe conitions of the
*roceure are not carrie out- the obser!ation is a ba obser!ation-
an the obser!er in this instance is a ba obser!er) E2uall" fro$ the
*oint of !iew of *h"sical science he is a ba obser!er if his belief that
his $etho can be substitute for the stanar *roceure is $istaken4
thou#h in this case he will *ass the bla$e on to the theorist who
a!ise hi$ wron#l")
The e*iste$olo#ist accorin#l" oes not stu" the obser!ers as
or#anis$s whose acti!ities $ust be ascertaine e$*iricall" in the
sa$e wa" that a naturalist stuies the habits of ani$als) He has to
*ick out the #oo obser!ers + those whose acti!ities follow a
con!entional *lan of *roceure) 6hat the e*iste$olo#ist $ust #et at is
this *lan) 6ithout it- he oes not know which obser!ers to stu" an
which to i#nore4 with it- he nee not actuall" watch the #oo obser!ers
who- he knows alrea"- are $erel" followin# its instructions- since
otherwise the" woul not be #oo)
The *lan $ust be sou#ht for in the $in of the obser!er- or in the
$ins of those fro$ who$ he has eri!e his instructions) The
e*iste$olo#ist is an obser!er onl" in the sense that he obser!es what
is in the $in) .ut that is a *eantic escri*tion of the wa" in which we
isco!er a *lan concei!e in an"one;s $in) 6e learn the obser!er;s
*lan b" listenin# to his own accor of it an cross+2uestionin# hi$)
IA
6e $a" istin#uish knowle#e of the *h"sical uni!erse eri!e b"
stu" of the results of obser!ation as a posteriori knowle#e- an
knowle#e eri!e b" e*iste$olo#ical stu" of the *roceure of
obser!ation as a priori knowle#e) A !aluer $a" arri!e at the
#enerali/ation a posteriori that no article in a certain house is worth
$ore than si0*ence4 the sa$e #enerali/ation $i#ht also ha!e been
reache a priori b" noticin# that the owner furnishe it fro$
6oolworth;s) The obser!er is calle u*on to su**l" the furniture of the
$ansion of science) The *riorist b" watchin# his $etho of obtainin#
the furniture $a" antici*ate so$e of the conclusions which the
*osteriorist will reach b" ins*ectin# the furniture)
I think that I a$ here usin# the ter$ ,a priori knowle#e, with its
reco#ni/e $eanin# + knowle#e which we ha!e of the *h"sical
uni!erse *rior to actual obser!ation of it) At an" rate e*iste$olo#ical
knowle#e is near enou#h akin to a priori knowle#e to arouse the
sa$e o**osition fro$ *h"sicists of the traitional school) $ priori
knowle#e has isre*utable associations in science4 an I a$ not
#oin# to tr" to conceal the fa$il" skeleton b" 2uibblin# about na$es)
To be 2uite e0*licit + e*iste$olo#ical or a priori knowle#e is *rior to
the carr"in# out of the obser!ations- but not *rior to the e!elo*$ent
of a *lan of obser!ation) As *h"sical knowle#e- it is necessaril" an
assertion of the results of obser!ations i$a#ine to be carrie out) To
the 2uestion whether it can be re#are as ine*enent of
obser!ational e0*erience alto#ether- we $ust- I think- answer no) A
*erson without obser!ational e0*erience at all- an without that
inirect knowle#e of obser!ational e0*erience which he $i#ht #ain b"
co$$unication with his fellows- coul not *ossibl" attach $eanin# to
the ter$s in which e*iste$olo#ical knowle#e- like other *h"sical
knowle#e- is e0*resse4 an it woul be i$*ossible to *ut it into an"
other for$ which woul ha!e a $eanin# for hi$)
6e $ust #rant then that the euction of a law of nature fro$
e*iste$olo#ical consierations i$*lies anteceent obser!ational
e0*erience) .ut it $ust be e$*hasi/e that the relation of the law of
nature to the obser!ations which for$ this anteceent e0*erience is
alto#ether ifferent fro$ its relation to the obser!ations which it
#o!erns) A stran#er to our >ni!ersit"- obser!in# that uner#rauates
were insie their Colle#es before $ini#ht- $i#ht belie!e that he ha
isco!ere a law of hu$an nature + that there is so$ethin# in the
nature of the uner#rauate which i$*els hi$ to seek the *rotection of
the colle#e walls before the stroke of twel!e) 6e $ust unecei!e hi$-
an *oint out that the law has a 2uite ifferent one + the Colle#e
authorities) Shoul he conclue then that the law is alto#ether
ine*enent of uner#rauate nature@ Not necessaril") Careful
research woul re!eal that the law e*ens on consierable
anteceent e0*erience of uner#rauate nature) 6e cannot sa" that
the twel!e o;clock rule is not base on uner#rauate nature4 but it is
not base on it in the wa" the stran#er assu$e)
In *h"sical science a priori conclusions ha!e lon# been anathe$a4 an
we $ust e0*ect o**osition fro$ those who ahere to traition) It has
co$e to be acce*te as a scientific *rinci*le that we can ha!e no a
priori knowle#e of the uni!erse) A#ree? *ro!ie that b" ,uni!erse, is
here $eant ,ob1ecti!e uni!erse,- as was unoubtel" intene when
the *rinci*le was fra$e) .ut- as a**lie to a uni!erse efine as the
the$e of *h"sical knowle#e rather than b" its intrinsic
characteristics
5
- the *rinci*le cancels itself) If we cannot ha!e a priori
knowle#e of the uni!erse- we cannot ha!e a priori knowle#e that it is
ob1ecti!e4 an therefore we cannot ha!e a priori knowle#e that we
2
6hen we ha!e efine a thin#- we ha!e a priori knowle#e that it has the characteristics
s*ecifie in the efinition) The *rinci*le $ust e!ientl" be unerstoo to e0ce*t this a
priori knowle#e)
cannot ha!e a priori knowle#e of it)
The re+introuction of a priori *h"sical knowle#e is 1ustifie b" the
isco!er" that the uni!erse which *h"sical science escribes is
*artiall" sub1ecti!e) Historicall" the ar#u$ent is the other wa" about) It
has been foun that certain knowle#e can be obtaine b" a priori
$ethos4 it is unoubtel" knowle#e of the uni!erse of *h"sics-
because it is *recisel" the knowle#e which *h"sicists ha!e sou#ht-
an in so$e cases foun- b" the usual a posteriori $ethos) .ut-
a$ittel"- such a priori knowle#e woul be i$*ossible if the uni!erse
to which it relates were wholl" ob1ecti!e) This starts us on an in2uir" in
which we are able to trace the $oe of entr" of a sub1ecti!e ele$ent
into the uni!erse of *h"sics)
It see$s a**ro*riate to call the *hiloso*hical outlook that we ha!e
here reache selective sub%ectivism) ,Selecti!e, is to be inter*rete
broal") I o not wish to assert that the influence of the *roceure of
obser!in# on the knowle#e obtaine is confine to si$*le selection-
like *assin# throu#h a net) .ut the ter$ will ser!e to re$in us that the
sub1ecti!e an the ob1ecti!e can be co$bine in other wa"s than b"
$ere aition) In $athe$atics a !er" #eneral t"*e of such co$bination
is that of o*erator an o*eran- selecti!e o*erators bein# a *articular
case)
Selection i$*lies so$ethin# to select fro$) It see$s *er$issible to
conclue that the $aterial on which the selection is *erfor$e is
ob1ecti!e) The onl" wa" to satisf" oursel!es of this is to e0a$ine
carefull" the wa"s in which sub1ecti!it" can cree* into *h"sical
knowle#e throu#h the *roceure of obser!in#) So far as I can see-
selection or o*erations $athe$aticall" akin to it co!er the whole ran#e
of *ossibilit"4 that is to sa"- the whole sub1ecti!it" is co$*rise in
o*erations of a selecti!e t"*e) The sub1ecti!it" bein# confine to the
o*erators- the ulti$ate o*eran $ust be free fro$ sub1ecti!it")
I see no reason to oubt the fore#oin# ar#u$ent- but it e*ens on a
!i#ilance of scrutin" which I cannot #uarantee as conclusi!e)
,Ob1ecti!e, is essentiall" a ne#ati!e characteristic 9non+sub1ecti!e: of
knowle#e- althou#h we re#ar it as a *ositi!e characteristic of the
thin# to which the knowle#e refers4 an it is alwa"s $ore ifficult to
e$onstrate a ne#ati!e than a *ositi!e conclusion) I acce*t an
ob1ecti!e ele$ent in *h"sical knowle#e on- I think- reasonabl" stron#
#rouns- but not with the sa$e assurance as the sub1ecti!e ele$ent
which is easil" e$onstrable)
Selecti!e sub1ecti!is$- which is the $oern scientific *hiloso*h"- has
little affinit" with .erkelian sub1ecti!is$- which- if I unerstan ri#htl"-
enies all ob1ecti!it" to the e0ternal worl) In our !iew the *h"sical
uni!erse is neither wholl" sub1ecti!e nor wholl" ob1ecti!e + nor a si$*le
$i0ture of sub1ecti!e an ob1ecti!e entities or attributes)
CHAPTER III
>NO.SERAA.LES
I
The funa$ental ieas of $oern *h"sics- in so far as the" iffer fro$
the classical ieas of the nineteenth centur"- are containe in two
co$*rehensi!e theories + relati!it" theor" an 2uantu$ theor")
Relati!it" theor" arri!e in two sta#es- na$el" Einstein;s s*ecial theor"
in %&D<- an his #eneral theor" in %&%<4 to these shoul be ae
6e"l;s theor" or relati!it" of #au#e in %&%(- which is now an essential
*art of the relati!istic conce*tion) Iuantu$ theor" be#an in a *a*er b"
Planck in %&D%4 it is thus so$ewhat the oler of the two theories- but it
has been $uch lon#er in reachin# $aturit"- 6hereas relati!it" theor"
fro$ the be#innin# was associate with a new *hiloso*hical outlook-
2uantu$ theor" u* to %&5< contribute nothin# to *hiloso*h" e0ce*t
bewiler$ent) Heisenber# introuce an i$*ortant new iea in %&5<4
an in the ne0t "ear or two- throu#h the ai of a nu$ber of
contributors- the theor" reache its *resent for$- #enerall" calle
,wa!e $echanics,) It cease to be entirel" a collection of e$*irical
$a#ic4 an- althou#h still rather obscure- it contains certain coherent
lines of thou#ht which ha!e *hiloso*hical i$*lications not less
i$*ortant than those of relati!it" theor")
Two broa i!isions of the sub1ect $atter of *h"sics are reco#ni/e-
na$el" microscopic physics which eals with s"ste$s on an ato$ic
scale- an molar physics which eals with s"ste$s on a scale
a**reciable to our #ross senses- co$*risin# !ast nu$bers of
$icrosco*ic constituents) 6e #enerall" sa" that relati!it" theor"
a**lies to $olar s"ste$s an 2uantu$ theor" to $icrosco*ic s"ste$s)
That oes not $ean that Nature is i!ie a#ainst herself) In acce*tin#
relati!it" *rinci*les or 2uantu$ *rinci*les we acce*t the$ for the whole
of *h"sics4 but the" $a" ha!e a $ore irect *ractical a**lication to
one branch than to the other) The 2uantu$ ,*rinci*le of uncertaint", is
*resu$abl" !ali also for $olar s"ste$s4 but it woul be ifficult to
*rouce a $olar e0a$*le in which the uncertaint" is etectable) The
,s*ecial *rinci*le of relati!it",- which asserts the *ri$iti!e e2ui!alence
of the s*ace+ti$e fra$es corres*onin# to ifferent !elocities- is !ali
also in $icrosco*ic *h"sics4 but it has no irect a**lication to the
interior of an ato$ or of a nucleus since- as a reference fra$e for
internal structure- a s*ace+ti$e fra$e in which the ato$ or nucleus as
a whole is in $otion is not e2ui!alent to one in which it is at rest)
Relati!it" an 2uantu$ principles are !ali throu#hout *h"sics4 but the
collections of theore$s an for$ulae which are co$$onl" sai to
constitute relati!it" theor" an 2uantu$ theor"- res*ecti!el"- ahere
fairl" closel" to the i!ision between $olar an $icrosco*ic *h"sics)
The relation between the $icrosco*ic 92uantu$: laws of *h"sics an
the $olar laws was $ae clear b" Niles .ohr in his ,corres*onence
*rinci*le,) The $olar laws are a si$*lifie for$ to which the
$icrosco*ic laws con!er#e when the nu$ber of *articles or 2uanta
consiere is !er" lar#e) This $eans that ieall" the $icrosco*ic laws
alone are sufficient to co!er the whole fiel of *h"sics- the $olar laws
bein# $erel" a con!enient aa*tation of the$ to a s*ecial- but
fre2uentl" occurrin#- *roble$) 6e ha!e so often to eal with
collections of !er" lar#e nu$bers of *articles that it is useful to ha!e a
state$ent in co$*act for$ of the outco$e of the $icrosco*ic laws as
a**lie to such s"ste$s- takin# a!anta#e of the si$*lifications which
occur when a!era#in# is *er$issible) 7olar law is this conense an
bowleri/e !ersion of $icrosco*ic law)
Accorin#l"- in a lo#ical orer of *resentation- $icrosco*ic law shoul
*recee $olar law) .ut *ractical e0*erience has *resente the
*roble$ the other wa" roun- since our sensor" or#ans are
the$sel!es $olar s"ste$s) Thus scientific in!esti#ation ca$e across
the $olar laws first4 an these ha!e been wele into a lo#icall"
co$*lete sche$e in relati!it" theor") 7icrosco*ic theor" is intrinsicall"
$ore ifficult- an has ha a later start4 few of the $ore si#nificant
*heno$ena associate with 2uantu$ theor" were known before %&DD)
In a sur!e" of *h"sical knowle#e at the *resent e*och- it is
i$*ossible to a!oi #i!in# $olar law an a!entitious *ro$inence
9which will *resu$abl" not be *er$anent:- since we are ac2uainte
with a co$*lete s"ste$ of $olar laws- but are still *ainfull" tr"in# to
*erfect the *artiall" unra!elle sche$e of $icrosco*ic laws)
A stu" of $ob+*s"cholo#" woul be a !er" unsatisfactor" founation
for a theor" of the hu$an $in) The $olar law- or $ob+law- of *h"sics
is an e2uall" unsatisfactor" introuction to the theor" of ini!iual or
ato$ic beha!ior) Thus no sooner ha we see$e to reach a
co$*rehension of the wa"s of Nature in $olar law than an entirel"
new conce*tion of what she is about be#an to a**ear in $icrosco*ic
law) Let $e sa" at once that the analo#" of the ini!iual an the $ob
is i$*erfect) It is one of the interestin# features of our *ro#ress that we
ha!e foun it to be i$*erfect) This is because the ini!iual *article or
entit" in *h"sics is a $uch $ore elusi!e conce*tion than the oler
ato$ists reali/e) .ut it re$ains true that $uch that is !ital to a *ro*er
unerstanin# of the *h"sical uni!erse is orinaril" lost to obser!ation
throu#h the e0cessi!e s$oothin# effect of our #ross senses4 an the
!er" *attern of natural law *ro!es to be ifferent fro$ the conce*tion
we ha for$e fro$ our first ac2uaintance with it in its li$itin# for$ for
lar#e nu$bers)
II
6e fre2uentl" s*eak of ,the *rinci*le of relati!it",4 but it is ifficult to
fin a ri#orous an authoritati!e enunciation of the *rinci*le) I ha!e
*ublishe three fairl" lon# e0*ositions of relati!it"- but- if $" $e$or"
oes not ecei!e $e- none of the$ atte$*ts a definition of the #eneral
*rinci*le of relati!it")
'
I think others ha!e been e2uall" reticent) For $"
own *art I ha!e re#are relati!it" as a new outlook whose
conse2uences $ust #rauall" unfol the$sel!es- rather than as a
*articular a0io$ or h"*othesis to be translate once for all into efinite
for$ulation)
Perha*s the nearest a**roach to a for$ulation of the *rinci*le is the
3
The ,s*ecial, *rinci*le is a co$*arati!el" si$*le iea)
state$ent that we obser!e onl" relations between *h"sical entities)
This re2uires a certain a$ount of critical e0*lanation which I shall not
enter on now) For- as I ha!e sai- I woul rather ientif" relati!it" with
the outlook which leas to this conclusion) I woul e$*hasi/e the
2uestion ,6hat is it we reall" obser!e@, rather than the answer ,6e
onl" obser!e relations between *h"sical entities,) .ecause as soon as
we ask the 2uestion- the classical sche$e of *h"sics is a *uncture
bubble4 an we start on a *ath of re!olution of which *erha*s the en
is not "et in si#ht) It is co$$on to escribe the state of theoretical
*h"sics in the last thirt" "ears as a succession of re!olutions4 but it is
all one re!olutionar" $o!e$ent which was starte b" this si$*le
2uestion) Heisenber# re*eate the 2uestion in %&5<? ,6hat is it we
reall" obser!e in an ato$@, The result was the new 2uantu$
$echanics)
Our first e*iste$olo#ical conclusion was that *h"sical knowle#e is of
an obser!ational nature- in the sense that each ite$ is an assertion of
the result of an obser!ation- actual or h"*othetical) Clearl" the ne0t
ste* $ust be to in2uire *recisel" what is co$*rise in the ter$
,obser!ation,) The obser!er of obser!ers + the e*iste$olo#ist + $ust
set to work to fin out what obser!ation reall" assures us of) In raisin#
the 2uestion ,6hat is it that we reall" obser!e@, relati!it" theor"
su$$one e*iste$olo#" to the ai of science)
This secon ste* is so co$*rehensi!e that it $ust be taken in sta#es)
At each sta#e we reach onl" a *artial truth4 but- after the $anner of
science- we are $ore concerne to a**reciate an e!elo* the new
insi#ht re!eale to us in the *artial truth than to stri!e *re$aturel" after
a final answer) Pro#ress so far has consiste- not so $uch in
s*ecif"in# what is actuall" obser!e- as in eli$inatin# what is efinitel"
unobser!e an unobser!able) Accorin#l" in this cha*ter we shall
consier es*eciall" the situation create b" the isco!er" than certain
2uantities- *ro$inent in *re+relati!it" *h"sics- are ,unobser!ables,)
As etecti!e stor" writers are fon of *ointin# out- it is notoriousl"
ifficult to e0tract fro$ a witness the actual e!iential facts4 he cannot
hel* wra**in# the$ u* in a #loss of his own) The habit of rou#h an
rea" inference co$es to us earl" b" instinct or eucation4 an the
,*lain $an of science, $i0es his facts with crue an unreliable
inferences- like an" other witness- when he has to state what he has
obser!e) Relati!it" theor" $ae the first serious atte$*t to insist on
ealin# with the facts the$sel!es) Pre!iousl" scientists *rofesse
*rofoun res*ect for the ,har facts of obser!ation,4 but it ha not
occurre to the$ to ascertain what the" were)
8i!iin# *h"sicists into three classes + relati!it" *h"sicists- 2uantu$
*h"sicists- e0*eri$ental *h"sicists + the relati!it" *h"sicist stuies the
har facts of obser!ation) The 2uantu$ *h"sicist follows the sa$e
*rinci*le as far as he can4 but owin# to the $ore intricate an $ore
re$ote nature of his sub1ect- the ai$ of constructin# a theor" which
shall e$brace onl" the obser!able facts re*resents his ieal rather
than his achie!e$ent) As for the e0*eri$ental *h"sicist- I will onl" sa"
that because a $an works in a laborator" it oes not follow that he is
not an incorri#ible $eta*h"sician)
III
I will be#in the iscussion of unobser!ables b" re$inin# "ou that the
etection of e!en a sin#le unobser!able- which has #ate+crashe into
*h"sical knowle#e- $a" ha!e far+reachin# conse2uences) The
founation of the s*ecial theor" of relati!it"- an the be#innin# of the
$oern re!olution in *h"sics- was the reali/ation that ,!elocit" of the
aether, is unobser!able) If we sa" that the $oon is istant 5CD-DDD
$iles- we are assertin# what woul be the result of a certain
obser!ational *roceure if it were carrie out4 but if we sa" that in a
certain re#ion the aether has a !elocit" of (D k$) *er sec) relati!e to
the earth- we are not assertin# the result of an" obser!ational
*roceure- actual or h"*othetical)
Let $e first re$o!e a co$$on $isunerstanin#) I o not $ean that
the in#enuit" of e0*eri$ental *h"sicists has as "et been insufficient to
e!ise an obser!ational *roceure which will $easure the !elocit" of
the aether) It is ne!er the task of the e0*eri$enter to e!ise the
obser!ational *roceure which is the ulti$ate test of the truth of a
scientific assertion) That $ust be inicate una$bi#uousl" in the
assertion itself- ha!in# re#ar to the efinitions of the ter$s e$*lo"e
in it4 otherwise it is inca*able of sub$ission to the Court of A**eal- an
is ina$issible as an ite$ of *h"sical knowle#e) 6here the in#enuit"
of the e0*eri$enter is often re2uire is in e!isin# a *roceure
e2ui!alent to- but $ore *racticable than- the *roceure referre to in
the assertion) The isco!er"- which beca$e the founation of the
s*ecial theor" of relati!it"- was that the assertion that the aether has a
relati!e !elocit" of (D k$) *er sec) turns out on scrutin" not to s*ecif"
an" obser!ational *roceure at all)
C
The in#enuit" of the e0*eri$enter
is not calle u*on4 for he cannot be aske to e!ise a *racticable
*roceure e2ui!alent to a *roceure which has ne!er been s*ecifie)
If we shut our e"es to the lo#ical incoherenc" in the efinition of aether
!elocit"- it is *ossible to treat its unobser!abilit" as an orinar" *h"sical
h"*othesis- su##este an confir$e b" obser!ation so far as it has
"et been *ossible to test its conse2uences) Since aether is not $atter-
it cannot be assu$e a priori that the usual attributes of $atter +
ensit"- ri#iit"- $o$entu$- etc) + are also attributes of the aether)
Accorin#l" the h"*othesis to be teste is that !elocit"- althou#h a
well+known attribute of $atter- is not one of the attributes of the aether)
Put in this wa"- it is not a truth that coul ha!e been foreseen a priori4 it
is a $ill" sur*risin#- but clearl" *ossible- conclusion euce a
posteriori fro$ the null result of e0*eri$ents esi#ne to etect effects
which woul be e0*ecte if there e0iste a lu$iniferous aether with the
t"*e of structure to which !elocit" coul be attribute)
This attitue is *o*ular with those who islike the e*iste$olo#ical
in2uir" associate with the new e!elo*$ents of *h"sics) It is so eas"
to cut short an ar#u$ent one oes not want to unerstan b" sa"in#? ,I
a$ not intereste in "our reasons- but I a$ 2uite willin# to tr" an"
conclusion "ou $a" ha!e reache as a h"*othesis to be teste b"
4
On the unerstanin# that the ,aether, referre to is 7a0well;s electro$a#netic aether-
efine as ha!in# the *ro*erties s*ecifie b" 7a0well;s e2uations)
obser!ation) Then- if it is confir$e- it will take rank with the other
confir$e h"*otheses of *h"sics- an we shall not nee "our
ar#u$ents), ." this kin of short+circuitin#- the $ore ifficult
consierations are cut out of the sub1ect4 an we can e$bark at once
on the strai#htforwar $athe$atical euction of the conse2uences of
the h"*othesis with a !iew to obser!ational test) Thus the new wine is
*ut into the ol bottles) It oes not burst the bottles4 but it loses $ost of
its in!i#oratin# + $" o**onents woul *erha*s sa"- its into0icatin# +
2ualities)
Let us tr" to reca*ture the in!i#oration) 6e can at least see that the
etection- whether b" lo#ical scrutin" or b" e0*eri$ental test- of one
unobser!able $as2uerain# as an obser!able in the classical sche$e
of *h"sics- an the i$*ortant conse2uences which ha!e resulte fro$
the etection- shoul be followe u* b" a s"ste$atic search for other
i$*ostors) Se!eral others ha!e been foun + in each case with !er"
far+reachin# results) The best known is Heisenber#;s isco!er" that a
co$bination of e0act *osition with e0act !elocit" is not obser!able4 this
constitutes the fa$ous ,uncertaint" *rinci*le,)
As a further e0a$*le- it was *ointe out ten "ears a#o that when we
are ealin# with *articles- such as electrons- which are
inistin#uishable fro$ one another obser!ationall"- the orinar" co+
orinate J0
5
+0
%
of one *article relati!e to another is not an obser!able4
the obser!able in this case is a t"*e of 2uantit" *re!iousl" unfa$iliar in
anal"sis- na$el" a ,si#nless co+orinate, J) >* to the *resent-
2uantu$ *h"sicists ha!e chosen to i#nore this i$*osture4 an the
$oern te0t+books still ahere to the erroneous theor" of a s"ste$ of
two such *articles- which assu$es the obser!able to be ) The" ha!e
thereb" $isse the o*enin# for a $uch neee a!ance)
I ha!e $entione this last e0a$*le because it is a clear case in which
unobser!abilit" is a $atter of e*iste$olo#ical *rinci*le- not of *h"sical
h"*othesis) For si$*licit"- consier *articles in one i$ension onl"-
sa" east an west) If we ha!e a #reen ball an a re ball- we can
obser!e that the #reen ball is- sa"- < inches west of the re ball)
Accorin#l"- for *ur*oses of escri*tion- we introuce an obser!able
2uantit" which states the istance of the #reen ball fro$ the re ball
$easure towars the west4 a ne#ati!e !alue of will inicate that the
#reen ball is to the east) .ut su**ose instea that we ha!e two balls
e0actl" alike in color- an with no istinction at all that we can obser!e)
In such a s"ste$ there is no obser!able corres*onin# to ) 6e can
obser!e that the balls are < inches a*art in the east+west line- an we
can introuce an obser!able which states the istance a*art) .ut-
unlike - is a si#nless 2uantit")
It is a natural $istake to a**l" the orinar" theor" of the obser!able
beha!ior of *articles 9*article $echanics- as we call it: to *rotons an
electrons- o!erlookin# that at an earl" sta#e in that theor"- na$el" in
introucin# an efinin# a relati!e co+orinate - it was taken for
#rante that the *articles coul be istin#uishe obser!ationall") This
$echanics beco$es ina**licable when is unobser!able) For *rotons
an electrons we ha!e a $oifie $echanics with as the obser!able)
This funa$ental ifference in the $echanics $ust be followe u*
$athe$aticall"4 an althou#h the *roble$ is rather ifficult- I think it is
ri#orousl" eucible that the ifference is e2ui!alent to a force
between the *articles which is actuall" the well+known Coulo$b force)
That is to sa"- the electrostatic 9Coulo$b: force between electrons an
*rotons is not an ,e0tra, arisin# we know not wh"- but is si$*l" a ter$
which ha ro**e out in the orinar" eri!ation of the e2uations
throu#h the o!ersi#ht of takin# instea of as the obser!able- an
ha therefore to be re+inserte e$*iricall")
Those unfa$iliar with wa!e+$echanics $a" be astonishe that there
shoul be a ifference between the $echanics of istin#uishable
*articles an the $echanics of inistin#uishable *articles) .ut it ou#ht
not to sur*rise 2uantu$ *h"sicists- since it is uni!ersall" a$itte that
there is a ifference in their statistics- which is no less $"sterious)
Inee I ha!e ne!er been able to unerstan wh" those who are well
aware of the i$*ortant conse2uences of inistin#uishabilit" in lar#e
asse$blies o not trouble to e0a$ine its *recise conse2uences in
s$aller s"ste$s) 6hether we consier the well+known effect on the
statistics of lar#e asse$blies or the less well+known effect on the
$echanics of a s"ste$ of two *articles- the conclusions a**ear
increible unless we bear in $in the sub1ecti!it" of the worl
escribe b" *h"sics an of all that it is sai to contain) It is naturall"
ob1ecte that the *articles cannot be affecte b" our inabilit" to
istin#uish the$- an it is absur to su**ose that the" $oif" their
beha!ior on that account) That woul be true if we were referrin# to
wholl" ob1ecti!e *articles an wholl" ob1ecti!e beha!ior) .ut our
#enerali/ations about their beha!ior + the laws of $echanics + escribe
*ro*erties i$*ose b" our *roceure of obser!ation- as the
#enerali/ations about catchable fish were i$*ose b" the structure of
the net) The ob1ecti!e *articles are unconcerne with our inabilit" to
istin#uish the$4 but the" are e2uall" unconcerne with the beha!ior
which we attribute to the$ *artl" as a conse2uence of our failure to
istin#uish the$) It is this obser!able beha!ior- an not the ob1ecti!e
beha!ior- that we are concerne with)
Returnin# to the 2uestion of *h"sical h"*othesis !ersus
e*iste$olo#ical *rinci*le- it is concei!able that a $an- unwillin# to
a**l" his $in to an"thin# e0ce*t $athe$atical for$ulae- $i#ht treat
our assertion that the obser!able is K- not L- as a su##este
h"*othesis which $ust stan or fall b" a co$*arison of the euce
conse2uences with e0*eri$ent) In for$ it rese$bles a *h"sical
h"*othesis- an its conse2uences can be followe u* in the sa$e wa")
.ut in this case the obser!ational test is *erfunctor" + like the
e0*eri$ental !erification of *ro*ositions of Eucli) A iscorance $i#ht
*oint to an error in eucin# the obser!ational conse2uences of the
assertion- or it $i#ht si#nif" that electrons are after all not entirel"
inistin#uishable4 but it woul not *ersuae us to contraict oursel!es
b" assertin# that- when A is obser!ationall" inistin#uishable fro$ .- it
is *ossible to obser!e that A is to the west- not east- of .)
It woul be an e0a##eration to sa" that the unobser!abilit" of aether
!elocit" is e2uall" ob!ious e*iste$olo#icall" + that the i$*ossibilit" of
obser!in# it lea*s to the e"e as soon as we consier how we shoul
set about the obser!ation) That is because reference to the aether
in!ol!es us in a $a/e of half+for#otten efinitions throu#h which it is
ifficult to threa our wa" without losin# oursel!es in a ust+stor$ of
!erbal contro!ers") .ut the aether has few friens nowaa"s- an we
attach $ore i$*ortance to a closel" relate unobser!able- na$el"
,istant si$ultaneit",) The unobser!abilit" of istant si$ultaneit" is
essentiall" the sa$e *rinci*le as the unobser!abilit" of aether !elocit"-
but it is free fro$ the *ossibl" a$bi#uous *hraseolo#" associate with
the ol aether h"*otheses) The unobser!abilit" of istant si$ultaneit"
is foun to be a *urel" e*iste$olo#ical conclusion)
The classical outlook took it for #rante that- in the histor" of a bo"
an"where in s*ace- there $ust occur an instant ientifie in an
absolute wa" with the instant ,now, which we oursel!es are this
$o$ent e0*eriencin#) It was also taken for #rante that the *roceure-
necessar" to ecie obser!ationall" which instants ha!e this relation of
absolute si$ultaneit"- woul be ob!ious to co$$onsense) .ut if
si$ultaneit" at istant *laces is to be e$*lo"e as a scientific ter$- we
cannot tolerate !a#ueness of efinition- an $ust insist on *recise
instructions as to the obser!ational *roceure intene) It is foun that
atte$*ts to for$ulate the instructions alwa"s en in a !icious circle)
For e0a$*le- the instruction $a" be to correlate instants at ifferent
*laces b" li#ht si#nals or raio si#nals- $akin# correction for the ti$e
of transit4 but when we in2uire how we are to eter$ine the latter
correction- the instructions are to $easure the ti$e of transit with
clocks alrea" a1uste to show si$ultaneit") It oes not re2uire a
7ichelson+7orle" e0*eri$ent to *ro!e to us that there is a !icious
circle in this efinition + thou#h it is *robable that the flaw woul ha!e
lon# continue to esca*e our notice if the result of the 7ichelson+
7orle" e0*eri$ent ha not insti#ate a scrutin")
The hint that a 2uantit" is *ossibl" an unobser!able has so$eti$es
been #i!en b" obser!ation4 that is to sa"- when atte$*ts were $ae to
$easure it- it *ro!e to be une0*ectel" elusi!e) .ut our efinite
knowle#e that it is unobser!able oes not co$e fro$ the failure of
atte$*ts to obser!e it4 it co$es fro$ a scrutin" of its efinition- which
is foun to contain a self+contraiction or !icious circle or other lo#ical
flaw) The efinition s*ecifies so$ethin# which souns like an
obser!ational *roceure4 but when we e0a$ine the $eanin# of the
ter$s 9which often in!ol!es tracin# a lon# chain of efinitions:- we fin
that the s*ecification oes not $ake sense) Since the iscri$ination of
unobser!ables e*ens on a stu" of the *roceure of obtainin#
obser!ational knowle#e- or alle#e obser!ational knowle#e- an not
on a stu" of the results of carr"in# out the *roceure- it co$es uner
scientific e*iste$olo#"4 an a *rinci*le of unobser!abilit"- such as the
s*ecial relati!it" *rinci*le- the uncertaint" *rinci*le- or the $oifie
$echanics of inistin#uishable *articles- is an e*iste$olo#ical
*rinci*le) Such *rinci*les ha!e an alto#ether ifferent status fro$
*h"sical h"*otheses- thou#h the" lea to the sa$e kin of *ractical
conse2uences)
6hen an unobser!able is introuce into a state$ent which *rofesses
to be an e0*ression of *h"sical knowle#e- the state$ent is usuall"
renere $eanin#less4 as an ite$ of *h"sical knowle#e it $ust assert
the result of a s*ecifie obser!ational *roceure- an the intrusion of a
ter$ which is without obser!ational $eanin# causes a hiatus in the
s*ecification) .ut e0ce*tionall" it $a" ha**en that the unobser!able is
in!ol!e in such a wa" that the truth of the state$ent is ine*enent of
the !alue ascribe to it) It then oes not !itiate the state$ent4 because-
althou#h *art of the obser!ational instructions *ro!es illusor"- it oes
not reall" $atter what result we su**ose this *art of the *roceure to
ha!e #i!en) For e0a$*le- it is an ite$ of *h"sical knowle#e that a
bo" which has been $o!e four "ars to the north an then three
"ars to the east will be fi!e "ars fro$ the startin# *oint) This a**lies
to $easure$ents on other *lanets besies the earth) Reasonabl"
inter*rete- it a**lies e!en on a non+rotatin# *lanet- althou#h ,north, is
then an unobser!able4 for althou#h the ter$s ,north, an ,east, are
use in e0*ressin# the knowle#e- its truth is ine*enent of the result
of the obser!ational *roceure b" which the first of these two
rectan#ular irections is lai own)
Thus there are two wa"s of ealin# with the unobser!ables which ha!e
been ina!ertentl" a$itte in classical *h"sics) One wa" is to
refor$ulate our knowle#e in such a wa" as to root the$ out
alto#ether) The other wa" is to sterili/e the$4 the" can be allowe to
re$ain *ro!ie that the assertions which contain reference to the$
re$ain true whate!er !alue we ascribe to the$ + whate!er result we
su**ose the illusor" obser!ational *roceure to ha!e #i!en) Althou#h
cu$brous fro$ a *hiloso*hical stan*oint- the latter $etho is
#enerall" the $ost con!enient in the *ractical e!elo*$ent of *h"sical
science) It in!ol!es less interference with the traitional for$ of
e0*ression of our knowle#e) 6e can $ore easil" trace the
conse2uences of the unobser!abilit") The *ossible assertions which
in!ol!e reference to unobser!ables are !er" $uch restricte in for$-
because the" $ust *ossess an ,in!ariance, which kee*s the$ true
howe!er we chan#e the su**ose !alue of the unobser!able)
Orinaril" such a restriction woul a$ount to a *h"sical h"*othesis + a
h"*othesis that the actual beha!ior of thin#s is in accorance with the
restriction) .ut in the *resent case the restriction is at root a tautolo#"4
for it woul be 2uite $eanin#less to $ake an assertion which i not
confor$ to the restriction)
IA
The reaer $a" ha!e notice that the e0a$*les we ha!e #i!en of the
a**lication of e*iste$olo#ical consierations to *h"sics are not 2uite
what he was le to e0*ect in Cha*ter II) 6e there conte$*late
#enerali/ations 9laws of nature: arisin# fro$ the selecti!e effect of the
*roceure of obser!ation) Here our e0a$ination of the *roceure
a**ears to ha!e le to a ifferent kin of isco!er"- na$el" that certain
2uantities incor*orate in the current sche$e of *h"sics are
unobser!able) ." e!elo*in# the conse2uences of this unobser!abilit"-
we can euce laws of nature which ha *re!iousl" been isco!ere
or su##este e$*iricall"- an thereb" transfer the$ fro$ an a
posteriori to an a priori basis4 but there is see$in#l" little as "et to
su**ort the outlook which I ha!e calle selecti!e sub1ecti!is$) I will tr"
to show later that the i!er#ence is onl" a**arent) 7eanwhile it $a"
be note that an a**arent i!er#ence was to be e0*ecte4 because
the *hiloso*hical in2uir" in Cha*ter II a**roache the sub1ect fro$ the
stan*oint of obser!abilit"- an the scientific in2uir" in this cha*ter has
a**roache it fro$ the stan*oint of unobser!abilit"- so that there is
so$e wa" to tra!el before the" $eet)
>nobser!abilit" of a 2uantit" arises fro$ a lo#ical contraiction in the
efinition which *rofesses to s*ecif" the *roceure for obser!in# it) I
$ust e$*hasi/e that this is not a 2uestion of ca*tious criticis$ of the
worin# of the ol efinitions #enerall" 2uote as authoritati!e) 6e o
not cone$n a 2uantit" as unobser!able- until e!er" effort has been
$ae- b" reworin# the efinition if necessar"- to re$o!e
contraictions) To $ake it clear that the criticis$ is not $erel" !erbal- I
will refer a#ain to the two unobser!ables that we ha!e alrea"
iscusse)
Take first the unobser!abilit" of the co+orinate ifference L of two
inistin#uishable *articles) Here there is no 2uestion of a$enin# the
efinition of L4 for it is inis*ensable in its *resent for$ in the stu" of
istin#uishable *articles) The lo#ical contraiction arises in a**l"in# it
to inistin#uishable *articles- o!erlookin# that it *resu**oses the
*articles to be istin#uishable obser!ationall")
The unobser!abilit" of istant si$ultaneit" raises $ore ifficult
consierations- because the conce*t has e0iste fro$ ti$e
i$$e$orial- an it has been taken for #rante that the *ractical
obser!er woul know how to eter$ine it without *recise instructions)
On tr"in# to for$ulate the *recise instructions- we ha!e foun that the"
contain a !icious circle- *resu**osin# a knowle#e of so$ethin# which
in turn *resu**oses a knowle#e of istant si$ultaneit") .ut we ha!e
to $eet the ob1ection that the instructions which we 9the relati!ists:
knock own are those which we ha!e oursel!es rawn u*4 an if the
instructions ha been rawn u* b" $ore sensible *eo*le- the" woul
not ha!e containe the !icious circle) Our re*l" is that the $ore
sensible *eo*le ha!e now ha thirt" "ears in which to co$e forwar4
but no one has *rouce instructions free fro$ a !icious circle) 6e are
willin# to take reasonable trouble to fin out a $eanin#- howe!er
i$*erfectl" e0*resse- but it is not $ere ca*tiousness if we refuse to
hol u* the *ro#ress of *h"sics b" enless search for a $eanin#
where there is no reason to su**ose that a $eanin# e0ists)
As a $atter of fact the *athetic faith of those who talke about istant
si$ultaneit" that so$eone $i#ht one a" be cle!er enou#h to fin out
what the" $eant- has co$e rather near to bein# 1ustifie) In
cos$olo#ical in!esti#ations it has been foun that if the istribution of
the #ala0ies throu#hout s*ace is unifor$ 9or nearl" unifor$: there is a
natural s"ste$ 9or a**ro0i$ate s"ste$: of ti$e+reckonin# a**ro*riate
to the uni!erse as a whole) The worl+wie instants in this reckonin#
$a" reasonabl" be taken to efine a istant si$ultaneit") .ut it woul
be far+fetche to ientif" this with the istant si$ultaneit" referre to in
the Newtonian s"ste$ of *h"sics) I o not belie!e that the classical
*h"sicist- in his references to si$ultaneit"- ha a *re$onition of a
relation contin#ent on the e0istence an law of istribution of $illiars
of #ala0ies- unsus*ecte in the cos$olo#" of his ti$e)
A
A feature of the a!ance of theoretical *h"sics has been a *ro#ressi!e
i$inution in the nu$ber of its funa$ental h"*otheses)
Althou#h we co$$onl" istin#uish between fundamental *h"sical
h"*otheses an casual h"*otheses $ae to e0*lain *articular
*heno$ena or to fill #a*s in our obser!ational knowle#e of the
ob1ects aroun us- it is ifficult to for$ulate a ri#orous istinction)
Howe!er- oubt selo$ arises in *ractice4 an without enorsin# the
current classification 9which I shall later *ro*ose to re*lace b" a $ore
si#nificant one: I use it as a classification which is reco#ni/e de facto)
In the o$ain thus set a*art we fin the sa$e #roun bein# co!ere
b" a continuall" i$inishin# nu$ber of h"*otheses)
The i$inution has co$e about in a nu$ber of wa"s) Firstl"- the
abanon$ent of the ieal of a $echanical e0*lanation of e!er"thin#
has eli$inate a #reat eal of ile h"*othesis) The *ro*erties of the
funa$ental entities of *h"sics are now state in the for$ of
$athe$atical e2uations- instea of bein# ,e0*laine, b" h"*othetical
$echanis$s) 7athe$atical for$ulation is !er" econo$ic of
h"*othesis) Sub1ect to a certain reser!ation- it enables us to state a
conclusion which oes not #o be"on the ascertaine facts4 it is no
$ore than a s"ste$ati/e state$ent of what is obser!e) The
reser!ation is that- whereas the ascertaine facts 1ustif" the
$athe$atical for$ula within a li$ite e#ree of a**ro0i$ation- uner
li$ite conitions- an in a li$ite nu$ber of instances- the
$athe$atical for$ula o$its reference to these li$itations) If we ha to
a**en to the for$ula a scheule of the instances in which it ha been
foun to be true- its usefulness woul e!a*orate) In one sense
Einstein;s 9or Newton;s: law of #ra!itation is not a h"*othesis4 it is a
su$$ar" state$ent of what we ha!e obser!e- within certain li$its of
a**ro0i$ation) It beco$es a h"*othesis when we assert that it is e0act
an uni!ersal) Since $athe$atical for$ulation is now ao*te
throu#hout the funa$ental *art of *h"sics- the onl" funa$ental
h"*otheses re2uire are h"*otheses of #enerali/ation in this sense)
Another *owerful factor in reucin# the nu$ber of h"*otheses has
been the #rowin# unification of *h"sics) .ranches) for$erl" treate
ine*enentl"- ha!e been unite4 an it has been foun that their
se*arate sets of h"*otheses were an unnecessar" u*lication) A
notable e0a$*le is the ientification of li#ht with electro$a#netic
wa!es- which eli$inate at one stroke all the h"*otheses of o*tics- the
h"*otheses of electro$a#netis$ bein# ae2uate to co!er the whole
sub1ect) E!en if we count the ientification of li#ht with electro$a#netic
wa!es as a new *h"sical h"*othesis- the substitution of this sole
h"*othesis for the s*eculati!e aether theories of the nineteenth
centur" is a substantial reuction) .ut the ientit" of li#ht with
electro$a#netic wa!es cannot be counte as an internal h"*othesis of
*h"sics- since it is alto#ether outsie the *ro!ince of *h"sics to
consier how the sti$ulation of the o*tic ner!e b" electro$a#netic
wa!es awakens in consciousness the sensation calle li#ht)
This eli$ination of h"*othesis ha $ae consierable *ro#ress before
the introuction of e*iste$olo#ical $ethos) It has alwa"s been the
ai$ of scientific in2uir" to trace a co$$on cause unerl"in# i!erse
*heno$ena4 an the nor$al *ro#ress of *h"sics has been towars a
unification which will e0hibit the whole orerin# of the uni!erse as the
result of a few si$*le causes) 6e $a" co$*are it with #eo$etr" which
reuces a #reat !ariet" of theore$s to a few ele$entar" a0io$s) If the
analo#" with #eo$etr" were to hol #oo- there woul be a li$it to the
eli$ination of h"*othesis- for a #eo$etr" without an" a0io$s at all is
unthinkable) .ut an e2uall" *ossible analo#" is with the theor" of
nu$bers) There we ha!e also a #reat !ariet" of theore$s- isclosin#
*ro*erties of nu$bers 2uite unforeseeable to orinar" intelli#ence4 an
"et throu#hout the whose sub1ect there is nothin# that can be calle an
a0io$) 6e shall fin reason to belie!e that this is in closer analo#"
with the s"ste$ of funa$ental laws of *h"sics)
6ith the co$in# of relati!it" theor" "et a thir $etho of reucin# the
nu$ber of h"*otheses cre*t in- na$el" the re*lace$ent of *h"sical
h"*otheses b" e*iste$olo#ical *rinci*les) 6e ha!e alrea" notices the
wa" in which an e*iste$olo#ical conclusion can *la" the sa$e *art as
a *h"sical h"*othesis so far as obser!ational conse2uences are
concerne)
6e ha!e seen that laws an *ro*erties which ha!e an e*iste$olo#ical
ori#in are co$*ulsor" an uni!ersal) It $a" be ae that- in so$e
cases at least- the" are e0act) For the unobser!abilit" of certain
2uantities + which is the $ost co$$on for$ of state$ent of an
e*iste$olo#ical *rinci*le + is trace to a lo#ical contraiction in their
efinitions4 an the conse2uences 9in so far as the" are reache b"
lo#ical euction alone- an not b" co$bination with $ore or less
uncertain an ine0act h"*otheses: are 2uite efinite) The *er!asion of
funa$ental *h"sics b" e*iste$olo#" has therefore #reatl" chan#e
its character- an brou#ht e0actitue within reach) So lon# as the
$ethos were wholl" a posteriori- there was no warrant for re#arin#
the euce laws of nature as better than a**ro0i$ations)
To a!oi $isa**rehension it is best to state here 9*re$aturel": that
althou#h we now reco#ni/e laws which we can confientl" assert are
e0act- the sub1ect+$atter of these e0act laws is *robabilit") There is
therefore not a corres*onin# *recision in the laws of obser!ational
*heno$ena 9as istin#uishe fro$ the laws of probability of the
*heno$ena:4 an- notwithstanin# its newl" ac2uire e0actness- the
s"ste$ of funa$ental *h"sical laws is ineter$inistic)
AI
6e ha!e seen that in the $oern theories of *h"sics e*iste$olo#ical
results *erfor$ *art of the ut" which was for$erl" allotte to *h"sical
h"*otheses + which inee is still often ascribe to *h"sical
h"*otheses b" those who o not look far enou#h into its ori#in) .ut it is
not eas" to #i!e a strai#htforwar e0a$*le of a h"*othesis of the oler
*h"sics which has isa**eare throu#h this re*lace$ent) This is
because the sche$e of theoretical *h"sics is !er" $uch interlocke) A
sin#le h"*othesis is not $eant to stan b" itself4 it *resu**oses that
other h"*otheses in the sche$e ha!e been acce*te) Newton;s law of
#ra!itation oes not account for the orbits of the *lanets or the fall of
an a**le unless his laws of $otion are also acce*te) Thus we cannot
e0*ect a one+to+one corres*onence between *h"sical h"*otheses in
the ol *h"sics an e*iste$olo#ical *rinci*les in the new sche$e) An
e*iste$olo#ical *rinci*le- such as the s*ecial *rinci*le of relati!it"- cuts
athwart the whole sche$e of h"*othesis) The h"*otheses re2uire to
su**le$ent it are less e0tensi!e than the s"ste$ of h"*otheses
*re!iousl" acce*te4 but the chan#e is not a si$*le o$ission of one or
$ore h"*otheses lea!in# the rest intact)
The best I can o is $ake a co$*arison of the h"*othetical ele$ents
in Newton;s an Einstein;s laws of #ra!itation) To facilitate co$*arison
I i!ie Newton;s h"*othesis into three ste*s of increasin#
s*eciali/ation?
9%h: There is a uni!ersal law of #ra!itation)
95h: It is e0*ressible b" a ifferential e2uation of the secon orer)
9'h: The secon orer e2uation 9in e$*t" s*ace: is
2
JD)
I $a" re*eat that the h"*othetical ele$ent is the #eneralit" an
e0actness of these state$ents) If we ha!e in $in a certain li$ite
e#ree of #eneralit" an accurac"- we $a" substitute ,e$*irical truth,
for ,h"*othesis, in 95h: an 9'h:)
Anal"sin# Einstein;s law si$ilarl"- we ha!e?
9%e: There is a uni!ersal law of #ra!itation)
95h: It is e0*ressible b" a ifferential e2uation of the secon orer)
9'e: The secon orer e2uation 9in e$*t" s*ace: is 3

J#

)
The first an thir ste*s- $arke e 9e*iste$olo#ical: are ri#orousl"
eucible fro$ an e0a$ination of the obser!ational *roceure followe
in obtainin# the $easure$ents which are ee$e to establish the law
of #ra!itation) The" in!ol!e no *h"sical h"*othesis at all) .ut the thir
ste* is left sus*ene in the air until the secon ste* is taken- an for
this we still a**eal to *h"sical h"*othesis) 6e $ust therefore take as a
$easure of the reuction of h"*othesis the eli$ination of 9%h: an
9'h:- lea!in# 95h: unchan#e) .ut to this $ust be ae a further
reuction of h"*othesis connecte with the laws of $otion) In
Newtonian *h"sics the laws of $otion are aitional h"*otheses4 but in
the relati!it" sche$e the" are eucible as $athe$atical
conse2uences of Einstein;s law of #ra!itation)
I ha!e little oubt that the ste* 95h: can also be trace to an
e*iste$olo#ical ori#in4 but to in!esti#ate this it is necessar" to enlar#e
the sco*e of the iscussion so as to co!er !irtuall" the whole of e0tra+
nuclear *h"sics- an not $erel" $echanics) This woul $er#e the
in2uir" in the #eneral *roble$ of esti$atin# how $uch h"*othesis
re$ains in the funa$ental laws of *h"sics after the e*iste$olo#ical
*ur#e has been carrie out in its entiret") This will be consiere in
Cha*ter IA)
CHAPTER IA
THE SCOPE OF EPISTE7OLO3ICAL 7ETHO8
I
A lar#e *art of the o$ain of *h"sical science is ae2uatel" co!ere
b" classical *h"sics) >suall" the $ore recent a!ances are #rafte on
to this oler knowle#e- an a**ear as corrections to it) 6e ha!e seen
in the last cha*ter that e*iste$olo#ical scrutin" has re!eale
i$*ostures an e0*ose lo#ical fallacies in the efinitions of 2uantities
in the classical sche$e) ." *resentin# the results in this wa"- we show
e*iste$olo#" in a rather ne#ati!e as*ect + a!ancin# *h"sics b"
re$o!in# errors which were blockin# its *ath)
Althou#h a co$*arison with classical *h"sics is the si$*lest- an
#enerall" the $ost useful- wa" of e0hibitin# the new a!ances- we
shoul also enea!or to #ras* the *ositi!e as*ect of e*iste$olo#icall"
#roune theor" as a self+containe e!elo*$ent of *h"sics- which- if
*ursue fro$ the be#innin#- will be uni$*ee b" the t"*e of error that
we ha!e been consierin#)
Re#are in this wa"- the characteristic of e*iste$olo#ical *h"sics is
that it irectl" in!esti#ates knowle#e- whereas classical *h"sics
in!esti#ate or enea!ore to in!esti#ate an entit" 9the e0ternal worl:
which the knowle#e is sai to escribe) Accorin#l" the $oern
*h"sicist has e!ise a techni2ue a**ro*riate to the in!esti#ation of
knowle#e of the kin a$itte in *h"sics4 whereas the classical
*h"sicist e!ise a techni2ue a**ro*riate to the in!esti#ation of an
entit" such as he concei!e the e0ternal worl to be) If fro$ the
be#innin# we reali/e that it is obser!ational knowle#e that is bein#
anal"/e + that the $athe$atical s"$bols re*resent ele$ents of
knowle#e- not entities of the e0ternal worl + unobser!ables cannot
be introuce e0ce*t b" eliberate intent as au0iliar" 2uantities in the
$athe$atics) The $oern *h"sicist is often re*roache for assu$in#
that because he has no knowle#e of a thin# it is non+e0istent) .ut this
is a $isconce*tion4 there is no nee to $ake an" assu$*tion about
thin#s of which we ha!e no knowle#e irect or inirect- since the"
cannot a**ear in an anal"sis of our knowle#e)
This ifference is $ost strikin#l" e0hibite in $oern 2uantu$ theor")
Accorin# to the classical conce*tion of $icrosco*ic *h"sics- our task
was to isco!er a s"ste$ of e2uations which connects the *ositions-
$otions- etc) of the *articles at one instant- with the *ositions- $otions-
etc) at a later instant) This *roble$ has *ro!e alto#ether bafflin#4 we
ha!e no reason to belie!e that an" eter$inate solution e0ists- an the
search has been frankl" abanone) 7oern 2uantu$ theor" has
substitute another task- na$el" to isco!er the e2uations which
connect knowle#e of the *ositions- $otions- etc) at one instant with
knowle#e of the *ositions- $otions- etc) at a later instant) The
solution of this *roble$ a**ears to be well within our *ower)
The $athe$atical s"$bolis$ escribes our knowle#e- an the
$athe$atical e2uations trace the chan#e of this knowle#e with ti$e)
Our knowle#e of *h"sical 2uantities is alwa"s $ore or less ine0act4
but the theor" of *robabilit" enables us to #i!e an e0act s*ecification of
ine0act knowle#e- incluin# a s*ecification of its ine0actitue) The
introuction of *robabilit" into *h"sical theories e$*hasi/es the fact
that it is knowle#e that is bein# treate) For *robabilit" is an attribute
of our knowle#e of an e!ent4 it oes not belon# to the e!ent itself-
which $ust certainl" occur or not occur)
6a!e $echanics in!esti#ates the wa" in which *robabilit" reistributes
itself as ti$e ela*ses4 it anal"ses it into wa!es an eter$ines the
laws of *ro*a#ation of those wa!es) 3enerall" the wa!es ten to
iffuse4 that is to sa"- our knowle#e of the *osition 9or of an" other
characteristic: of a s"ste$ beco$es !a#uer the lon#er the ti$e
ela*se since an obser!ation was $ae) A suen accession to
knowle#e + our beco$in# aware of the result of a new obser!ation + is
a iscontinuit" in the ,worl, of *robabilit"+wa!es4 the *robabilit" is
reconcentrate- an the *ro*a#ation starts a#ain fro$ the new
istribution) There are e0ce*tional for$s of *robabilit" istribution of
certain of the attributes of $icrosco*ic s"ste$s which o not iffuse- or
iffuse !er" slowl"4 so that our knowle#e of these attributes oes not
so ra*il" #row out of ate) Particular attention is la!ishe on these
,stea" states, an on the e2uations eter$inin# the$- since the"
*ro!ie a basis for lon#+ran#e *reictions)
The state$ent often $ae- that in $oern theor" the electron is not a
*article but a wa!e- is $isleain#) The ,wa!e, re*resents our
knowle#e of the electron) The state$ent is- howe!er- an ine0act wa"
of e$*hasi/in# that the knowle#e- not the entit" itself- is the irect
ob1ect of our stu"4 an it $a" *erha*s be e0cuse b" the fact that the
ter$inolo#" of 2uantu$ theor" is now in such utter confusion that it is
well+ni#h i$*ossible to $ake clear state$ents in it) The ter$ ,electron,
has at least three ifferent $eanin#s
<
in co$$on use in 2uantu$
theor"- in aition to its loose a**lication to the *robabilit" wa!e itself)
6a!e $echanics shows us i$$eiatel" wh" the istinction between
obser!ables an unobser!ables is so essential) A ,#oo, obser!ation
of a 2uantit"- althou#h it oes not eter$ine the 2uantit" *recisel"-
narrows own the ran#e in which it is likel" to lie) It creates a
conensation in the *robabilit" istribution of the 2uantit" or- as we
usuall" sa"- for$s a wa!e *acket in it) The $etho of wa!e $echanics
is to in!esti#ate the wa!e e2uations which #o!ern the *ro*a#ation of
wa!es fro$ such a source) .ut if the 2uantit" is unobser!able- these
wa!e *ackets cannot be for$e) A stu" of the *ro*a#ation of wa!es
which there is no $eans of *roucin# can ha!e no a**lication to
*h"sics4 an a theor" which *rofesses to euce obser!ationall"
!erifiable results b" such anal"sis is e!ientl" !itiate b" a $is+
ientification)
II
I e0*ect I shall be accuse of e0a##eratin# the e*iste$olo#ical
5
Na$el"- the *article re*resente b" a 8irac wa!e+function- the *article introuce in
secon 2uanti/ation- an the *article re*resente b" the internal 9relati!e: wa!e+function
of a h"ro#en ato$)
ele$ent in $oern *h"sical theor"- an before #oin# farther I will tr" to
e0a$ine this criticis$)
Fro$ the ti$e of Newton until recentl" the e*iste$olo#" of science
was stationar"4 for two hunre "ears the e0tension an orerin# of
our knowle#e of the *h"sical uni!erse continue without $oif"in# it)
6e ha!e seen that the *h"sicist is b" ori#in a *hiloso*her who has
s*eciali/e in a *articular irection4 but for hi$ e*iste$olo#" ha
beco$e ancient histor"- an he ha lon# cease to concern hi$self
with it) 3enerall" he *rie hi$self on bein# a *lain $atter+of+fact
*erson + which was his wa" of escribin# a $an who acce*te the
naM!e realis$ of Newtonian e*iste$olo#") If he inul#e in *hiloso*h"
at all- it was as a hobb" ke*t a*art fro$ the serious occu*ation of
a!ancin# science)
Thus althou#h scientific e*iste$olo#" has alwa"s been *art of the
o$ain of *h"sics- the *h"sicist ha left it so lon# unculti!ate that-
when at last he turne attention to it- his ri#ht+of+wa" was 2uestione)
The re+entr" into this ne#lecte fiel was the be#innin# of the $oern
re!olution of *h"sics- the first result bein# the theor" of relati!it") .ut
we $ust not look u*on e*iste$olo#" as a lon#+estran#e relati!e who
has une0*ectel" be2ueathe us a fortune in the *rinci*le of relati!it")
The sensible wa" to treat a rich relati!e is to in!ite hi$ to re1oin the
fa$il" circle- so that "ou can touch hi$ for a lot $ore)
The 2uestion $a" be raise- How far oes #eneral o*inion a$on#
leain# *h"sicists toa" reco#ni/e this reunion@ It is ifficult to
ascertain) 7" i$*ression is that the #eneral attitue $i#ht be
escribe as #ru#in# acce*tance) A**eal to new e*iste$olo#ical
consierations $a" be allowe in e$er#enc"- but it is not *er$itte to
beco$e *art of the routine of scientific a!ance) There is a #eneral
reco#nition that i$*ortant a!ances ha!e resulte fro$ critical
e0a$ination of the nature of our obser!ational knowle#e) I think too
that the leain# authorities woul a#ree with $" brief account of the
$etho of 2uantu$ theor" in the last section + that it *rocees b" a
irect anal"sis of knowle#e of a s"ste$- instea of b" an anal"sis of
the s"ste$ itself + an the" woul acknowle#e that this chan#e of
$etho is res*onsible for all recent *ro#ress) The" see$ to be aware
of the e*iste$olo#ical ele$ent introuce in the re!olution of *h"sics4
an the" ha!e e0*erience of the *ractical !alue of a rational
e*iste$olo#ical outlook) .ut there is an unaccountable reluctance to
e!elo* scientific e*iste$olo#" s"ste$aticall") Althou#h *articular
*rinci*les ha!e secure reco#nition an are inee worke to eath-
there see$s to be no reali/ation that it woul be *rofitable thorou#hl"
to e0*lore the e*iste$olo#ical $etho so as to e!elo* it to the ut$ost
a!anta#e)
There are $an" new *roble$s connecte with the nucleus- raiation-
cos$olo#"- etc)- which it is a$itte that the *resent 2uantu$ theor"
cannot co!er without so$e funa$ental a!ance) One woul ha!e
thou#ht that we ha learnt b" now how to set about esca*e fro$ this
ealock) Another a**eal shoul be $ae to our rich relation
e*iste$olo#" who has rescue us on for$er occasions4 another ste*
forwar shoul be taken in answerin# the funa$ental 2uestion- 6hat
o we reall" obser!e@ This wa" of a!ance is still o*en4 we ha halte
onl" because the floo of new insi#ht re!eale b" the earlier ste*s was
for the $o$ent $ore than we coul bear) 6hate!er $" own scientific
work in this irection $a" si#nif"- it at least shows where the o*enin#s
lie- an that a!ance throu#h these o*enin#s is b" no $eans
i$*racticable)
I can scarcel" su**ose that 2uantu$ *h"sicists are unaware of the
errors of ientification of obser!ables which ha!e been re*eatel"
*ointe out in the last ten "ears4 but the" *refer to stick to the errors +
*resu$abl" because the" re#ar the$ as a lesser e!il than a further
encroach$ent of e*iste$olo#") As one of the$ has nai!el" *ut it?
,&bservable is a !er" elusi!e conce*tion- an if we *ursue the
criticis$ to the en- we shall ha!e to oubt a lot of thin#s we o not in
the least want to oubt),
It a**ears then that- althou#h the e*iste$olo#ical character of $oern
*h"sical theories is reco#ni/e an at ti$es stron#l" e$*hasi/e-
there is not as "et a reall" effecti!e union of scientific e*iste$olo#"
with science) I ha!e been referrin# to the attitue of those who
s*eciali/e on the funa$ental *roble$s) If we turn to the $uch lar#er
circle of *h"sicists who are occu*ie- not with e!elo*in#- but with
a**l"in# the results of the new theories- it is still $ore ifficult to sa"
where the" stan) To what e0tent can the !er" #eneral acce*tance of
the new theories be re#are as an acce*tance of their
e*iste$olo#ical outlook@ It is still- I think- an unfa$iliar iea a$on#
scientists that scientific *hiloso*h" shoul ha!e an" relation to
scientific *ractice) If one writes of the heat+eath of the uni!erse-
re$arkin# that it is inesca*able accorin# to the secon law of
ther$o"na$ics- so$e critic is sure to *rotest that this is an entire
$isconce*tion of scientific law4 a scientific law- he woul sa"- is no
$ore than an e$*irical #enerali/ation !ali o!er the ran#e of s*ace
an ti$e an circu$stances for which it has been !erifie- an it is
unscientific to e0tra*olate such a #enerali/ation to an unknown istant
future) Yet that sa$e critic- if he were refereein# a *a*er on so$e new
*roble$- such as a *ossible ori#in of cos$ic ra"s in #ala0ies be"on
the ran#e of our telesco*es- woul assurel" look to see whether the
*ro*ose e0*lanation was consistent with the secon law of
ther$o"na$ics- an the *a*er woul ha!e s$all chance of
acce*tance if it was not)
6hen Einstein;s theor" arri!e- which not onl" *ro*oune a new
e*iste$olo#" but a**lie it to eter$ine the law of #ra!itation an
other *ractical conse2uences- *h"sicists were *u//le how to classif"
it) So$e ar#ue that it was *hiloso*h"- alias $eta*h"sics- an $ust be
re1ecte out of han) Others concee that the for$ulae a**eare to
a#ree with obser!ation an acco$*lishe a !aluable s"ste$ati/ation
of knowle#e- but belie!e that a ,#enuinel" *h"sical, inter*retation of
its $eanin# woul in ti$e su**lant the e*iste$olo#ical 1ar#on which at
*resent en!elo*s it) Fewer reali/e that the new e*iste$olo#ical
outlook is the !er" heart of the theor"- su**lantin# a fallacious s"ste$
of thou#ht which was barrin# *ro#ress) E!en now we often fin
authors- who are b" no $eans i#norant of the reasons for the chan#e
of thou#ht- *ro*ounin# theories for which the" clai$ the a!anta#e
that the" in!ol!e onl" Newtonian conce*tions) As thou#h it coul be an
a!anta#e to incor*orate a fallacious an obsolete !iew of the nature
of obser!ational knowle#eH
This !a#ueness an inconsistenc" of the attitue of $ost *h"sicists is
lar#el" ue to a tenenc" to treat the $athe$atical e!elo*$ent of a
theor" as the onl" *art which eser!es serious attention) .ut in
*h"sics e!er"thin# e*ens on the insi#ht with which the ieas are
hanle before the" reach the $athe$atical sta#e)
The conse2uence of this tenenc" is that a theor" is !er" co$$onl"
ientifie with its leain# $athe$atical for$ulae) 6e continuall" fin
s*ecial relati!it" theor" ientifie with the Lorent/ transfor$ation-
#eneral relati!it" with the transfor$ation to #enerali/e co+orinates-
2uantu$ theor" with the wa!e e2uation or the co$$utation relations)
It cannot be too stron#l" ur#e that neither relati!it" theor" nor
2uantu$ theor" are su$$e u* in fool+*roof for$ulae for use on all
occasions) A relati!ist is not a $an who e$*lo"s Lorent/+in!ariant
for$ulae 9which were introuce so$e "ears before the relati!it"
theor" a**eare:- but one who unerstans in what circu$stances
for$ulae ou#ht to ha!e Lorent/+in!ariance4 nor is he a $an who
transfor$s e2uations into #enerali/e co+orinates 9a *ractice at least
a centur" ol:- but one who unerstans in what circu$stances a
s*ecial s"ste$ of co+orinates woul be ina**licable) In 2uantu$
*roble$s allowance $ust be $ae for the backwar state of the
theor"4 an the worl is still awaitin# a 2uantist who unerstans in
what circu$stances the stanar wa!e e2uation an the co$$utation
relations are a**licable + as istinct fro$ one who $erel" a**lies the$
an ho*es for the best)
It is clear that no coherent *hiloso*h" can be $ae out of a half+an+
half reco#nition of the *lace of e*iste$olo#" in science) 6hat reall"
concerns our in2uir" is that the leaers of *h"sics ha!e so far
co$$itte the$sel!es in acce*tin# its ai that its co$*lete
assi$ilation is onl" a 2uestion of ti$e)
III
I o not see how an"one who acce*ts the theor" of relati!it" can
is*ute that there has been so$e re*lace$ent of *h"sical h"*otheses
b" e*iste$olo#ical *rinci*les4 nor o I think that those who acce*t the
theor" with unerstanin# will be incline to is*ute it) The $ore
contro!ersial 2uestion is- How far can this re*lace$ent e0ten@ Here
$" conclusion- base on *urel" scientific in!esti#ation- is $uch $ore
rastic than that of $ost of $" collea#ues) I belie!e that the whole
s"ste$ of funa$ental h"*otheses can be re*lace b" e*iste$olo#ical
*rinci*les) Or- to *ut it e2ui!alentl"- all the laws of nature that are
usuall" classe as funa$ental can be foreseen wholl" fro$
e*iste$olo#ical consierations) The" corres*on to a priori
knowle#e- an are therefore wholly sub%ective)
I a$ sorr" to ha!e to *ut in the forefront what will #enerall" be
re#are as an ini!iual scientific conclusion4 but this cannot be
a!oie) I think I can see a clear *hiloso*h" e$er#in# fro$ the
conclusion that the s"ste$ of funa$ental laws is wholl" sub1ecti!e) I
cannot see an" coherent *hiloso*h" e$er#in# fro$ the conclusion that
so$e are sub1ecti!e an so$e ob1ecti!e) I$$eiatel" I start on that
line I a$ beset with ob1ections an *er*le0ities which I o not know
how to $eet) I o not cone$n it on that account4 *erha*s with a #reat
eal $ore thou#ht a wa" of *ro#ress coul be seen) .ut there is no
inuce$ent to s*en $" ti$e tr"in# to o!erco$e the ifficulties of a
*hiloso*h" associate with scientific beliefs which I o not share) No
one can conte$*late enterin# on a ifficult research base on
*re$ises which he has reason to belie!e erroneous) You will fin
*lent" of *hiloso*hies of ob1ecti!e natural law4 "ou will fin here a
*hiloso*h" of sub1ecti!e natural law) If e!er a *hiloso*h" of $i0e
sub1ecti!e+ob1ecti!e natural law is e!elo*e- it will not be b" $e- for I
a$ con!ince that there is no scientific su**ort for such a *hiloso*h")
7" conclusion- ri#ht or wron#- has a *urel" scientific basis) That it
leas to a si$*le an rational *hiloso*h" is *erha*s an ar#u$ent in its
fa!or4 but that is an after+thou#ht which i not wei#h with $e in
reachin# the scientific conclusion)
I i not set out with an" *reconcei!e iea of the sco*e of the
e*iste$olo#ical $etho4 an the conclusion that the whole of the
funa$ental laws of nature can be euce fro$ e*iste$olo#ical
consierations was the result of trial) Ha!in# ha lon# associations
with the theor" of relati!it"- in which the $etho first showe its *ower-
an seein# fro$ ti$e to ti$e further o**ortunities for a**l"in# it- I
foun it was co!erin# $ore an $ore of the #roun of funa$ental
*h"sics until at len#th the conclusion beca$e irresistible)
One feature of this e!ience nees to be e$*hasi/e) A law of nature
is nowaa"s e0*resse b" a $athe$atical e2uation) Our knowle#e of
the law can onl" be sai to be co$*lete if we know- not onl" the
al#ebraic for$ of the e2uation- but the !alues of the *ara$eters
occurrin# in it) .ut it is custo$ar" to confine the ter$ ,law of nature, to
the al#ebraic for$- the *ara$eters bein# referre to se*aratel" as
,constants of nature,) For e0a$*le- the Newtonian theor" of #ra!itation
introuces a law- na$el" the in!erse s2uare law- an a constant-
na$el" the #ra!itational constant4 si$ilarl" for Einstein;s theor") In the
co$*arison of Newton;s an Einstein;s laws on *) CF- I ha!e o$itte
all reference to the #ra!itational constant) .ut in the $ore far+reachin#
in!esti#ations to which I a$ now referrin#- the constants as well as the
al#ebraic for$s are inclue) 7" conclusion is that not onl" the laws of
nature but the constants of nature can be euce fro$
e*iste$olo#ical consierations- so that we can ha!e a priori
knowle#e of the$)
Treatin# the sche$e of natural law as a whole- as it is set out in the
funa$ental e2uations of *h"sics- four constants of nature which are
*ure nu$bers
G
are in!ol!e) These I fin to be *reictable a priori) I
focus attention on these- because it is a $ore strin#ent test of the
*ower of the e*iste$olo#ical $etho to *ro!ie a nu$ber 9!erifiable in
so$e cases to about % *art in %DDD: than to *ro!ie for$s of law) I
6
For$e b" eli$inatin# our three arbitrar" units 9centi$eter- #ra$- an secon: fro$ the
se!en constants of nature orinaril" reco#ni/e 9New Pathwa"s in Science- *) 5'5:
think that the classical *h"sicist ha an inner feelin# that the in!erse
s2uare law was a natural for$ of weakenin# of an effect b" istance-
which $i#ht be e0*ecte a priori to a**l" to #ra!itation + thou#h it
woul- of course- be contrar" to his *rinci*les to acknowle#e an" a
priori e0*ectation) .ut he coul scarcel" ha!e ha an inner inti$ation-
acknowle#e or unacknowle#e- of the $ost likel" stren#th of such
a force)
Now it is to be re$e$bere that whate!er is accounte for
e*iste$olo#icall" is ipso facto sub1ecti!e4 it is e$olishe as *art of
the ob1ecti!e worl) After a consierable nu$ber of results ha!e been
foun- there co$es a ti$e when we *ause to rest on our *icks an
1u#e- if we can- the e0tent of the re#ion $arke out for e$olition)
Orinar" #enerali/ation woul ha!e su##este $uch earlier that the
re#ion was co+e0tensi!e with the funa$ental laws 9incluin#
constants: of *h"sics) .ut I coul not brin# $"self to belie!e the
#enerali/ation until the last of the four constants ha succu$be) I
was hel back b" the co$$on feelin#- which I now see was not
*hiloso*hicall" well+#roune- that it was necessar" to lea!e at least
one ob1ecti!e *e# on which to han# the sub1ecti!e cloaks) Certainl" an
ob1ecti!e *e# is necessar"- but we nee not su**ose that it has
is#uise itself to rese$ble the cloaks)
One of the four natural constants is a !er" lar#e nu$ber calle the
cos$ical nu$ber) It is *erha*s $ost si$*l" escribe as ,the nu$ber
of *articles in the uni!erse,- thou#h it also a**ears in *h"sics in other
$ore *ractical wa"s) 6e are incline to sa" that if an"thin# is
unforeseeable a priori- it is the nu$ber of *articles in the uni!erse)
This see$s the inner$ost citael of ob1ecti!it") .ut- after the work on
the other constants- the e*iste$olo#ical ori#in of this nu$ber was
co$*arati!el" eas" to trace)
Fro$ a *hiloso*hical stan*oint all the earlier work is subsiiar" to the
attack on the cos$ical nu$ber- which is the real turnin# *oint in our
thou#ht) So lon# as its ob1ecti!it" is outstanin#- e!en if it is the onl"
*urel" ob1ecti!e fact in *h"sics- we watch the e0*osure of sub1ecti!e
influences with e2uani$it") >nification we o not fear4 an with one
assurel" ob1ecti!e fact to *oint to- there is no an#er of nullification)
.ut when we fin that the cos$ical nu$ber is sub1ecti!e + that the
influence of the sensor" e2ui*$ent with which we obser!e- an the
intellectual e2ui*$ent with which we for$ulate the results of
obser!ation as knowle#e- is so far+reachin# that b" itself it ecies
the nu$ber of *articles into which the $atter of the uni!erse a**ears
to be i!ie + not onl" o we lose the su**ort on which we were
rel"in#- but there is no heart left in us to o**ose the e!ourin# floo of
sub1ecti!it" an" lon#er)
I shall accorin#l" eal at so$e len#th with the sub1ecti!it" of the
cos$ical nu$ber in Cha*ter EI) The actual eri!ation fro$
e*iste$olo#ical consierations is- of course- $uch too technical to be
#i!en there) .ut in an" case a $athe$atical e$onstration carries no
con!iction if its result is ,ob!iousl" i$*ossible,4 it onl" leas those who
are sufficientl" intereste to look for a hole in the e$onstration) I shall
instea *oint out the wa" in which sub1ecti!it" has obtaine a foothol
in a re#ion of thou#ht fro$ which one woul ha!e su**ose it to be
ri#orousl" e0clue4 so that e!en the e!aluation of the nu$ber of
*articles in the uni!erse will be seen to be not ob!iousl" i$*ossible)
I o not inclue ato$ic nuclei in this iscussion of the laws of *h"sics-
because the *resent state of nuclear theor" is co$*arable to the state
of 2uantu$ theor" before %&5<- an *ro!ies no basis for
*hiloso*hical inference) The fillin# of this #a* oes not see$ !er"
ur#ent) H"ro#en- it shoul be re$e$bere- contains no nuclei 9other
than *rotons:- so that it is full" co!ere b" the *resent iscussion)
F

Hitherto it has not occurre to an"one to a!ocate a *hiloso*h"
iealist as re#ars h"ro#en an realist as re#ars o0"#en4 I think I
$a" assu$e si$ilarl" that- in a stu" of the sub1ecti!e or ob1ecti!e
nature of the uni!erse- its che$ical co$*osition is rather irrele!ant)
I $ust $ention here a *oint which ulti$atel" turns out to be of #reat
7
Incluin# the non+Coulo$bian force which $anifests itself in the close encounters of
*rotons) Since this force also *la"s a lar#e *art in nuclear theor"- the nucleus is to this
e0tent inclue in the iscussion)
i$*ortance) The ,law of chance, is not usuall" counte as a
funa$ental law of *h"sics- an I o not inclue it a$on# the laws that
can be foreseen wholl" fro$ e*iste$olo#ical consierations) .ut
accorin# to the $oern s"ste$ of *h"sics all our *reictions of
*heno$ena are *reictions of what will *robabl" ha**en- an are
base on an assu$*tion of non+correlation of the beha!ior of
ini!iual *articles which is eri!e fro$ the law of chance) 6ithout an
a**eal to the law of chance *h"sics is unable to $ake an" *reiction
of the future) The law of chance $i#ht therefore be clai$e to be the
$ost funa$ental an inis*ensable of all *h"sical laws) The reason
wh" it is o$itte is that- fro$ the orinar" *oint of !iew- rano$ness is
a ne#ation of law4 an it see$s unnecessar" to la" own a law sa"in#
that there is no law) .ut the orinar" !iew takes it for #rante that the
*h"sical uni!erse- an the *articles into which we anal"/e it- are
wholl" ob1ecti!e4 an the status of the law of chance 9or non+
correlation: re2uires reconsieration when a**lie to a *artl"
sub1ecti!e uni!erse) It is i$*ossible to treat this *oint full" until a late
sta#e of the iscussion) The !iew finall" ao*te will be foun on **)
%(D- 5%() If in the $eanti$e the reaer fins $" ar#u$ent tenin#
a**arentl" to a $ore an $ore increible conclusion- he $a" await a
later twist that will soften it into so$ethin# which will- I think- not too
#rossl" affront his co$$onsense)
IA
6ithout further a*olo#" I shall now assu$e the reaer;s assent to the
*ro*osition that all the funa$ental laws an constants of *h"sics can
be euce una$bi#uousl" fro$ a priori consierations- an are
therefore wholl" sub1ecti!e) At least the onus of *roof woul see$ to
rest on those who clai$ ob1ecti!it" for an" law- thereb" isturbin# the
ho$o#eneit" of the sche$e without a**arent necessit")
Let us #o back to the analo#" of the fish 9*) %G: to illustrate the *osition
now reache) 6hen the ichth"olo#ist re1ecte the onlooker;s
su##estion of an ob1ecti!e kin#o$ of fishes as too $eta*h"sical- an
e0*laine that his *ur*ose was to isco!er laws 9i)e) #enerali/ations:
which were true for catchable fish- I e0*ect the onlooker went awa"
$utterin#? ,I bet he oes not #et !er" far with his ichth"olo#" of
catchable fish) I woner what his theor" of the re*rouction of
catchable fish will be like) It is all !er" well to is$iss bab" fishes as
$eta*h"sical s*eculation4 but the" see$ to $e to co$e into the
*roble$),
I think that there is so$ethin# in this ob1ection) It *erha*s unerrates
the *ower of the $athe$atician to hanle selecte $aterial
intelli#entl") .ut if "our *ur*ose is to eter$ine laws of ob1ecti!e ori#in
co$in# throu#h to us in a for$ $oifie b" sub1ecti!e selection- I o
not think the best wa" is to su**ress all theories about the ob1ecti!e
worl + at an" rate as workin# h"*otheses) .ut at first si#ht the
*ro#ress of *h"sics see$s to contraict this4 for it was 1ust when
h"*othesis about the ob1ecti!e worl was abanone- an we turne
to a irect stu" of *h"sical knowle#e- that *ro#ress beca$e
astonishin#l" ra*i)
The e0*lanation is si$*le) All this *ro#ress relates to sub1ecti!e law) It
all relates to unifor$ities i$*ose on the results of obser!ation b" the
*roceure of obser!ation) As for the t"*e of unifor$it" illustrate b" the
ichth"olo#ist;s secon #enerali/ation that all sea+creatures *ossess
#ills + unifor$ities intrinsic in the worl aroun us + we ha!e not e!en
$ae a be#innin#) The onlooker was ri#ht) No *ro#ress at all has
been $ae with the kin of biolo#ical stu" that he ha in $in)
I ha!e allue 9*) C': to a ifficult" in efinin# the istinction between
,funa$ental, an ,casual, h"*otheses in *h"sics) The sa$e ifficult"-
e0*resse in a sli#htl" ifferent for$- arises in $akin# a ri#i
istinction between ,laws of nature, an ,s*ecial facts,)
In classical *h"sics this ifficult" oes not occur) Followin# La*lace- it
is assu$e that fro$ the co$*lete state of the uni!erse at an" one
instant the co$*lete state at an" other instant- *ast or future- is
calculable) The funa$ental laws of nature are then efine to be the
laws which- taken all to#ether- furnish a sufficient set of rules for the
calculation) To co$*lete our knowle#e of the uni!erse we $ust know-
besies the rules- the initial ata to which the" are to be a**lie)
These ata are the s*ecial facts)
It is *ossible that we $a" isco!er a rule or re#ularit" a**l"in# to the
s*ecial facts) If so- we shoul *robabl" not en" it the title of a law of
nature) .ut it can be istin#uishe fro$ the funa$ental laws of
nature- because it is no *art of the sche$e of *reiction) It is 1ust a
*attern of the s*ecial facts #ratuitousl" incor*orate in the esi#n of
the uni!erse)
The istinction can be e0*resse !er" succinctl" in $athe$atical
lan#ua#e) The ifferential e2uations eter$inin# the *ro#ress of the
uni!erse are the funa$ental laws of nature- an the bounar"
conitions are the s*ecial facts)
.ut this $oe of istinction is *ossible onl" in a eter$inistic uni!erse)
In the current ineter$inistic s"ste$ of *h"sics- there is no
corres*onin# e$arcation between the laws an the s*ecial facts of
nature) The *resent s"ste$ of funa$ental laws oes not furnish a
co$*lete set of rules for the calculation of the future) It is not e!en *art
of such a set- for it is concerne onl" with the calculation of
*robabilities4 an if e!er the search for a sche$e of efinite *reiction
is renewe- it will be necessar" to start a#ain fro$ the be#innin# on
ifferent lines) The *art *la"e b" the s*ecial facts is also altere) The
s*ecial facts- which istin#uish the actual uni!erse fro$ all other
*ossible uni!erses obe"in# the sa$e laws- are not #i!en once for all at
so$e *ast e*och- but are bein# born continuall" as the uni!erse
follows its un*reictable course) 7oreo!er- in the ifferential e2uations
of 2uantu$ theor" the bounar" conitions are not the ob1ecti!e facts
but the knowle#e we ha**en to *ossess about the$)
The si$*le e$arcation in classical theor" between funa$ental laws
of nature an s*ecial facts is associate with eter$inis$ an cannot
be carrie o!er into the $oern theor") .ut- a**roachin# the 2uestion
fro$ the *oint of !iew of sub1ecti!it"- a new line of e$arcation
a**ears) 6e ha!e foun that the su**osel" funa$ental laws are
wholl" sub1ecti!e) It is onl" reasonable that the *art of our knowle#e
which is wholl" sub1ecti!e shoul be of a reco#nisabl" ifferent t"*e
fro$ that which in!ol!es the ob1ecti!e characteristics of the uni!erse) It
a**ears that this ifference was not o!erlooke b" the earlier
*h"sicists4 an we fin the re#ion to be anne0e to *ure sub1ecti!it"
alrea" $arke out uner another na$e- !i/) ,funa$ental,)
The s*ecial facts- on the other han- cannot be inferre fro$
e*iste$olo#ical consierations an are not wholl" sub1ecti!e) It is the
essence of our conce*tion of a s*ecial fact that it $i#ht 2uite well ha!e
been otherwise + that there is no a priori reason wh" it shoul be what
it is) It is true that $an" ha!e hel the !iew that the laws of nature
$i#ht 2uite well ha!e been otherwise4 but the" woul scarcel" assert
that this is an inse*arable *art of the conce*tion of a law of nature)
E!er"one reco#nises that it is in so$e sense takin# a #reater libert"
with the uni!erse to i$a#ine the laws of nature to ha!e been ifferent
that to i$a#ine the s*ecial facts to ha!e been ifferent)
Results eucible b" the a priori e*iste$olo#ical $etho are
co$*ulsor"- an it is therefore i$*ossible that the $etho shoul be
e0tene to *reict the s*ecial facts- which ,$i#ht 2uite well ha!e
been otherwise,) I a$ afrai that before I finish I shall ha!e *ersuae
the ocile reaer to belie!e so $an" ,i$*ossible, thin#s that the wor
will $ake little i$*ression on hi$- an he will not 1ib at i$*ossibilit"
when I want hi$ to) Let $e then *ut the *oint rather ifferentl") If b" an
a!ance of e*iste$olo#ical theor" we succee in *reictin# one of the
so+calle s*ecial facts in a wholl" a priori wa"- we shall at once a$en
the classification? ,Clearl" we were $istaken in su**osin# that it was a
s*ecial fact) Now that we see $ore clearl" into its ori#in- we reali/e
that there is a law of nature which co$*els it to be so),
The cos$ical nu$ber affors a #oo e0a$*le of such a chan#e of
!iew) Re#are as the nu$ber of *articles in the uni!erse- it has
#enerall" been looke u*on as a s*ecial fact) A uni!erse- it is hel-
coul be $ae with an" nu$ber of *articles4 an- so far as *h"sics is
concerne- we $ust 1ust acce*t the nu$ber allotte to our uni!erse as
an accient or as a whi$ of the Creator) .ut the e*iste$olo#ical
in!esti#ation chan#es our iea of its nature) A uni!erse cannot be
$ae with a ifferent nu$ber of ele$entar" *articles + consistentl"
with the sche$e of efinitions b" which the ,nu$ber of *articles, is
assi#ne to a s"ste$ in wa!e+$echanics) 6e $ust therefore no
lon#er look u*on it as a s*ecial fact about the uni!erse- but as a
*ara$eter occurrin# in the laws of nature- an- as such- *art of the
laws of nature)
A
I $ust tr" to counteract the i$*ression that the ob1ecti!e ele$ent in the
*h"sical uni!erse- ha!in#- as it were- been ri!en into a corner before
the a!ancin# tie of sub1ecti!it"- now nees onl" *erfunctor"
reco#nition) That i$*ression is #i!en in *h"sics- because *h"sics 9in
the narrow sense: is not intereste in s*ecial facts e0ce*t as ata to
be su*ersee b" #enerali/ations) Other *h"sical sciences- such as
astrono$"- are not so e0clusi!e- an to so$e e0tent restore the
*ers*ecti!e) .ut the *h"sical uni!erse- as it affects us a" b" a"- is
not 1ust a bunle of laws of nature4 an the s*ecial facts are as
i$*ortant to us as the laws) Thus- althou#h it is onl" throu#h the
s*ecial facts that we iscern an"thin# of the ob1ecti!e uni!erse- it is a
b" no $eans e$*t" !iew) 7oreo!er- it is not as in the eter$inistic
a"s when the s*ecial facts were collecte into a sin#le instant) 6ithin
the li$its of the uncertaint" *rinci*le the" are e!er+chan#in# as the
$o$ents *ass b")
The s*ecial facts are *artl" sub1ecti!e an *artl" ob1ecti!e- e*enin#
*artl" on our *roceure in obtainin# obser!ational knowle#e an
*artl" on what there is to obser!e) To se*arate the sub1ecti!e or
ob1ecti!e ele$ents co$*letel"- we $ust consier laws4 since a law or
re#ularit" $a" ori#inate wholl" in our *roceure of obser!ation or
wholl" in the ob1ecti!e worl) It $a" be 2uestione whether we coul
e!er isolate an ob1ecti!e law as co$*letel" as a sub1ecti!e law- since it
woul ha!e to be *resente to us !ia our sub1ecti!e for$s of thou#ht4
but at least we coul etect a re#ularit" an reco#ni/e that its ori#in
was ob1ecti!e- e!en if we coul onl" escribe it in sub1ecti!e ter$s)
6e are in an#er of fallin# into a confusion re#arin# laws of nature +
a confusion between what the" are an what we ori#inall" intene
the$ to be) To a!oi a$bi#uit" I will iscri$inate 9te$*oraril": between
,laws of nature, an ,laws of Nature,) Law of Nature will ha!e the
$eanin# that the ter$ was ori#inall" intene to bear + a law
e$anatin# fro$ the worl+*rinci*le outsie us- which we often
*ersonif" as Nature) Law of nature will $ean as heretofore a re#ularit"
which we ha!e foun in our obser!ational knowle#e- irres*ecti!e of
its source) In short a law of nature is whate!er woul be esi#nate b"
that na$e in current *h"sical *ractice)
It will be seen that a law of Nature is a law of the ob1ecti!e uni!erse)
.ut all reco#ni/e laws of nature are sub1ecti!e) 6e ha!e thus
reache the !erbal *arao0 that no known law of nature is a law of
Nature) Effecti!el" the ter$s ha!e beco$e $utuall" e0clusi!e)
It is true that we ha!e left an o*enin#) A law of Nature is a law of
nature if it woul be 9not necessaril" if it alrea" is: acce*te as such
in *h"sics) This brin#s $e to a further 2uestion- Ha!e we an" reason
to belie!e that if a law of Nature + a #enerali/ation about the ob1ecti!e
worl + were to beco$e known to us- it woul be acce*te b" current
*h"sics as a law of nature@ I think it woul onl" be acce*te if it
confor$e to the *attern of *h"sical law that we are accusto$e to)
.ut this *attern is the *attern of sub1ecti!e law) 6e shall tr" later to
show b" e*iste$olo#ical stu" how the *attern has #rown out of the
sub1ecti!e as*ect of *h"sical knowle#e) The *attern is the !er"
hall$ark of sub1ecti!it") An" e0*ectation we $a" ha!e for$e that the
ob1ecti!e laws of Nature- when the" are isco!ere- will confor$ to the
sa$e *attern is 2uite unreasonable)
6e $ust not atte$*t to la" own in a!ance the t"*e of re#ularit"
which we call b" antici*ation a law of Nature) To assu$e that we know
beforehan the *attern of ob1ecti!e law woul be an assertion of a
priori knowle#e of the ob1ecti!e uni!erse- re*uiate b" all schools of
scientific thou#ht) Not all our s"ste$ati/ation of knowle#e is of the
,e0act, t"*e e$*lo"e in *h"sical science4 an in other sciences law
has a broaer inter*retation) Hitherto it is onl" in the *urel" sub1ecti!e
*art of our knowle#e that we ha!e foun the e0act t"*e of law
obe"e)
Perha*s it will be ar#ue that althou#h ob1ecti!e law- when it is
isco!ere- $a" *ro!e to be of an unfa$iliar *attern- *h"sicists will
$oif" their ieas to acco$$oate it) The *attern of *h"sical law has
not re$aine i$$utable4 an the *attern reco#ni/e toa" woul not
ha!e been acce*te in classical *h"sics) If *ro#ress in the ob1ecti!e
*art of our knowle#e $akes it necessar" to enlar#e the *attern- it will
not be the first ti$e *h"sics has uner#one re!olution) That is one
*ossibilit"4 but there is an alternati!e) 6oul the enlar#e sub1ect
necessaril" retain the na$e of *h"sics@ The for$er chan#es were
co$*ulsor"4 we i not ro* classical *h"sics solel" because its fiel
was too li$ite- but because we isco!ere its efects) .ut here the
*ro*ose chan#e is not force on us b" a efect in the treat$ent of the
sub1ecti!e knowle#e at *resent co!ere- but is for enlar#e$ent onl")
It $i#ht well be consiere $ore a**ro*riate to confine the na$e
*h"sics to the o$ain it at *resent occu*ies- an treat the new
e!elo*$ent as ,outsie *h"sics,) If so- the laws of Nature will ne!er
be the sub1ect $atter of *h"sics)
This souns like a 2uibble o!er na$es4 but it #i!es a hint that $a" be
reall" i$*ortant) The su##estion is that when we succee in $akin#
*ro#ress with the stu" of the ob1ecti!e worl- the result will be !er"
ifferent fro$ *resent+a" *h"sics- an that there is no *articular
reason to e0*ect that it will be calle *h"sics) 6e ha!e s*oken of this
as a e!elo*$ent in the future4 but $a" it not ha!e occurre alrea"@
It see$s to $e that the ,enlar#e, *h"sics which is to inclue the
ob1ecti!e as well as the sub1ecti!e is 1ust science4 an the ob1ecti!e-
which has no reason to confor$ to the *attern of s"ste$ati/ation that
istin#uishes *resent+a" *h"sics- is to be foun in the non+*h"sical
*art of science) 6e shoul look for it in the *art of biolo#" 9if an":
which is not co!ere b" bio*h"sics4 in the *art of *s"cholo#" which is
not co!ere b" *s"cho*h"sics4 an *erha*s in the *art of theolo#"
which is not co!ere b" theo*h"sics) The *urel" ob1ecti!e sources of
the ob1ecti!e ele$ent in our obser!ational knowle#e ha!e alrea"
been na$e4 the" are life' consciousness' spirit!
6e reach then the *osition of iealist- as o**ose to $aterialist-
*hiloso*h") The *urel" ob1ecti!e worl is the s*iritual worl4 an the
$aterial worl is sub1ecti!e in the sense of selecti!e sub1ecti!is$)
CHAPTER A
EPISTE7OLO3Y AN8 RELATIAITY THEORY
I
The !ocabular" of *h"sics inclues a nu$ber of ter$s such as len#th-
ener#"- te$*erature- *otential- refracti!e ine0- etc)- which we call
*h"sical 2uantities) Relati!it" theor" insists that all *h"sical 2uantities
shall be efine in a wa" which will enable us to reco#ni/e the$ in
*ractical e0*erience) A efinition of *otential $ust s*ecif" a wa" of
eter$inin# *otentials) A efinition of len#th $ust s*ecif" a wa" of
$easurin# len#ths)
This e$an is si$*l" a reco#nition that if the theorist an the
e0*eri$enter are to co+o*erate the" $ust s*eak a co$$on lan#ua#e)
If we call in an e0*eri$enter to test the truth of our state$ents- his first
2uestion $ust be ,How a$ I to reco#ni/e the thin# "ou are talkin#
about@, The answer we #i!e hi$ is its efinition) If he !erifies the truth
of the state$ent- his certificate a**lies onl" so lon# as the wors $ean
what we tol him the" i) To kee* in reser!e so$e other efinition +
so$e non+obser!ational $eanin# + of the wors in the state$ent- to be
use after we ha!e wan#le the certificate- is ishonest) It is not less
ishonest if we oursel!es belie!e that the two efinitions refer to the
sa$e thin#4 for the belief was not sub$itte to e0*eri$ental test- an
has not been certifie)
Pro#ress ha been $ae towars this t"*e of efinition in *re+relati!it"
a"s) At one ti$e $ass was efine as 2uantit" of $atter4 but the
e0*eri$enter was #i!en no inication how he was to reco#ni/e the
sa$e ,2uantit" of $atter, when *resente in ifferent for$s- e)#) wool
an lea) Conse2uentl"- althou#h it was not realise at the ti$e- no
state$ent about $ass 9unless restricte to one kin of $atter: was
e!er !erifie e0*eri$entall") .ut later a efinition of $ass in ter$s of
obser!able inertial *ro*erties was substitute4 an- with this chan#e
$eanin#- obser!ational test of the state$ents beca$e *ossible) It has
co$e to be the acce*te *ractice in introucin# new *h"sical
2uantities that the" shall be re#are as efine b" the series of
$easurin# o*erations an calculations of which the" are the result)
Those who associate with the result a $ental *icture of so$e entit"
is*ortin# itself in a $eta*h"sical real$ of e0istence o so at their own
risk4 *h"sics can acce*t no res*onsibilit" for this e$bellish$ent)
The inno!ation $ae b" Einstein in his relati!it" theor" was that the
*h"sical 2uantities in!ol!e in the $easure$ent of s*ace an ti$e
were brou#ht uner this rule) The refor$ was clearl" necessar"4 for the
e0*eri$enter is calle u*on to certif" the truth of our conclusions about
istances an inter!als of ti$e- 1ust as $uch as he is calle on to
certif" the truth of our conclusions about te$*eratures or $a#netic
fiels) A efinition of len#th which s*ecifies a wa" of eter$inin#
len#ths obser!ationall" is inee the $ost ur#ent re2uire$ent of all4
for when we co$e to e0a$ine what is actuall" $easure in an" kin of
e0*eri$ent- it is nearl" alwa"s a len#th or s*atial $easure + the len#th
of a threa of $ercur" in a ther$o$eter- the shift of a bri#ht s*ot on a
#al!ano$eter scale- the is*lace$ent of a ark line in a s*ectro#ra$-
etc))
It is stran#e that this inno!ation shoul ha!e arouse an o**osition-
which is not e!en "et e0tinct) There still flourishes an unreasonin#
insistence that ter$s which ha!e reference to $easure$ent of s*ace
$ust not be connecte with obser!ation in the sa$e wa" that ter$s
ha!in# reference to $echanical- o*tical- electro$a#netic- ther$al an
other $easure$ents are) There is a fa$ous *assa#e in the writin#s of
Henri Poincare- of *re+relati!it" ate- which is often 2uote in this
connection?
If Lobatchewsk";s #eo$etr" is true- the *aralla0 of a !er" istant star will be finite)
If Rie$ann;s is true- it will be ne#ati!e) These are results which see$ within the
reach of e0*eri$ent- an it is ho*e that astrono$ical obser!ations $a" enable us
to ecie between the two #eo$etries) .ut what we call a strai#ht line in
astrono$" is si$*l" the *ath of a ra" of li#ht) If- therefore- we were to isco!er
ne#ati!e *aralla0es- or to *ro!e that all *aralla0es are hi#her than a certain li$it-
we shoul ha!e a choice between two conclusions? we coul #i!e u* Eucliean
#eo$etr"- or $oif" the laws of o*tics- an su**ose that li#ht is not ri#orousl"
*ro*a#ate in a strai#ht line) It is neeless to a that e!er" one woul look u*on
this solution as the $ore a!anta#eous) Eucliean #eo$etr"- therefore- has
nothin# to fear fro$ fresh e0*eri$ents)
Those who 2uote this *assa#e ha!e usuall" $isse the $oral) Clearl"
the $oral is that the efinition of the *aralla0 or istance of a star $ust
not be left to *ure $athe$aticians whose assertions ha!e nothin# to
fear fro$ fresh e0*eri$ents) It was true enou#h in Poincare;s ti$e that
theorists talke about istances which i not $ean an"thin# in
*articular- so that "ou were at libert" to choose whether the" obe"e
Eucliean or non+Eucliean #eo$etr") .ut the e0*eri$enter went his
own wa" an $easure istances which $eant so$ethin# !er"
*articular + to the se!enth or the ei#hth si#nificant fi#ure) Theorist an
e0*eri$enter i not s*eak the sa$e lan#ua#e) Relati!it" theor"
institute the ob!ious refor$- an those ha**"+#o+luck" a"s are o!er)
Now- if a theorist reaches a conclusion about stellar or #alactic
istances- he oes so with a *ro*er fear of fresh e0*eri$ents) I $ust
a$it ha!in# *ut forwar conclusions of this kin- an I tre$ble e!er"
ti$e a new e0*eri$ental result is about to be announce) Not that I
shall necessaril" belie!e it)
II
The efinition of len#th or istance an the corres*onin# efinition of
ti$e+e0tension are *articularl" i$*ortant- because in #eneral the
efinitions of other *h"sical 2uantities *resu**ose that len#th an
ti$e+e0tension ha!e been efine- an an" a$bi#uit" of their $eanin#
woul s*rea throu#h the whole su*erstructure) If- instea of len#th
bein# efine obser!ationall"- its efinition were left to the *ure
$athe$atician- all the other *h"sical 2uantities woul be infecte with
the !irus of *ure $athe$atics)
Practical *h"sicists ha!e lon# been occu*ie with the accurate
eter$ination of len#ths- an the *rinci*les which the" stri!e to follow
were settle before the theor" of relati!it" arose) This branch of
*ractical *h"sics is calle $etrolo#") 6hen therefore it beca$e
necessar" to ao*t for$all" an obser!ational efinition of len#th- there
coul be no 2uestion of settin# u* a ri!al *roceure) The efinition
$ust #i!e instruction as to a *roceure of $easure$ent of len#ths) To
the $etrolo#ist these instructions a$ounte si$*l" to ,Carr" on,)
It is not unco$$on for writers to treat the efinition of len#th or ti$e+
inter!al as thou#h the $eanin#s of these ter$s were freel" at their
is*osal) .ut it is scarcel" le#iti$ate to ao*t such an attitue towars
ter$s in current use) A writer $a" *erha*s clai$ to ha!e fulfille his
obli#ations if he $akes clear the $eanin# the ter$ will ha!e in his own
writin#s- irres*ecti!e of the wa" it has been e$*lo"e hitherto) .ut the
co$$on !iew is that it is re*rehensible to use the ter$ ,white, to
escribe the a**earance $ore usuall" known as ,black,4 an the
*ractice- recentl" introuce b" writers on kine$atical cos$olo#"- of
#i!in# to len#th an ti$e a $eanin# which no .ureau of Stanars
woul acce*t- thou#h *erha*s not a$ountin# to $oral obli2uit"- is a
source of unnecessar" confusion)
In all orthoo0 *h"sical theor"- the $etrolo#ical *ractice + or $ore
strictl" the *rinci*le which it atte$*ts to carr" out + su**lies the
theoretical efinition) Thus it is secure that- when the e0*eri$enter
checks the theorist- both are referrin# to the sa$e thin#)
Accorin#l"- b" len#th in relati!it" theor" we $ean what the $etrolo#ist
$eans- not what the *ure #eo$eter $eans) In acce*tin# relati!it"
*rinci*les- the *h"sicist *uts asie his *ara$our *ure $athe$atics-
is$isses their #o+between $eta*h"sics- an enters into honorable
$arria#e with $etrolo#") I a$ afrai those who re*resent the brie are
incline to sus*ect that he is not entirel" off with his first lo!e) So$e
writin#s on relati!it" look a bit $athe$atical) Since I a$ not entirel"
con!ince of the innocence of so$e of $" collea#ues- I $ust on this
*oint answer onl" for $"self) I eclare that the sus*icions are
#rounless) If I so$eti$es e$*lo" *ure $athe$atics- it is onl" as a
ru#e4 $" e!otion is fi0e on the *h"sical thou#ht which lies behin
the $athe$atics) 7athe$atics is a useful !ehicle for e0*ression an
$ani*ulation4 but the heart of the theor" is elsewhere?
Eu*helia ser!es to #race $" $easure
.ut Chloe is $" real fla$e)
The crucial *art of the efinition of len#th is the s*ecification of a
stanar which shall be a!ailable for co$*arison at an" *lace an at
an" ti$e) 7etrolo#ists o not look u*on a *articular bar of $etal- such
as the Paris $eter- as an ulti$ate stanar4 the $ere fact that the"
feel an0iet" as to its *er$anence shows that the" ha!e in $in a $ore
ieal stanar with which it $i#ht be co$*are) 6hat is neee is a
*h"sical structure- not necessaril" *er$anent- but uni2uel"
reproducible) A calcite cr"stal whose len#th contains %D
(
lattice
inter!als will su##est the sort of stanar re2uire) If that were
s*ecifie as the stanar of len#th- it coul be re*rouce in the
farthest #ala0" or at the re$otest e*och of ti$e)
Let us consier fro$ a #eneral *oint of !iew the *roble$ of s*ecif"in#
a re*roucible stanar of len#th) Ob!iousl" we $ust not e$*lo"
len#ths in the s*ecification4 for that woul be a !icious circle) Nor can
we use an" of the other ,i$ensional, *h"sical 2uantities- for their
efinitions *resu**ose that stanars of len#th- ti$e- an $ass are
alrea" efine) The 2uantitati!e *art of the s*ecification $ust
therefore consist of uni$ensional 2uantities- i)e) *ure nu$bers) For
e0a$*le- the stanar *ro*ose abo!e was s*ecifie b" the nu$ber
of lattice cells) 6e can- if we like- #o farther an s*ecif" the che$ical
co$*osition of the cr"stal b" *ure nu$bers- na$el" the ato$ic
nu$bers of the ele$ents concerne in it)
A *urel" nu$erical escri*tion of $aterial structure is elaborate in
2uantu$ theor") The structure is escribe as consistin# of a certain
nu$ber of nuclei an electrons whose arran#e$ent is s*ecifie b"
2uantu$ nu$bers) Such a structure is necessaril" uni2ue fro$ the
obser!ational *oint of !iew4 for if two s*eci$ens showe obser!able
ifferences- it woul be taken as a *roof that the e0istin# theor" of the
structure was inco$*lete an aitional 2uantu$ nu$bers woul be
introuce to istin#uish the$)
Thus the #eneral answer to our *roble$ is that an" structure which is
*racticall" re*roucible fro$ a 2uantu$ s*ecification will ser!e as
stanar) All such stanars are e2ui!alent- bein# in efinite nu$erical
ratios to the unit of len#th h(mc which a**ears in the funa$ental
e2uations of 2uantu$ theor")
The stanar of ti$e+e0tension is efine si$ilarl") A s*atial e0tension
of the 2uantu$+s*ecifie structure *ro!ies the stanar of len#th4 a
ti$e+*erioicit" of the sa$e structure *ro!ies the stanar of ti$e+
e0tension) The *arallelis$ is closest if we e$*lo" a cr"stal4 for- !iewin#
the structure in four i$ensions- *erioicit" is a lattice structure in
ti$e4 an our two stanars are res*ecti!el" a s*ecifie nu$ber of
s*ace+lattice cells an a s*ecifie nu$ber of ti$e+lattice cells of the
cr"stal structure)
It is *erha*s not su*erfluous to a that no 2uestion arises as to
whether the stanar of len#th efine in this wa" is really constant at
all ti$es an *laces) The 2uestion i$*lies that we ha!e in $in so$e
$ore ulti$ate stanar 9in!este with ,realit",: b" which to efine the
elin2uencies of the *h"sical stanar) The conce*tion of *h"sical
2uantities ha!in# to confor$ to so$e *articular role allotte in a!ance
in a !a#uel" i$a#ine real$ of realit"- is not reco#ni/e in *h"sical
science4 2uantities such as len#th an ti$e+e0tension are introuce
solel" for the *ur*ose of succinct escri*tion of obser!ational
$easure$ents actual or h"*othetical)
III
6e notice that relati!it" theor" has to #o outsie its own borers to
obtain the efinition of len#th- without which it cannot be#in) It is the
$icrosco*ic structure of $atter which introuces a efinite scale of
thin#s) Since we ha!e se*arate $olar *h"sics fro$ $icrosco*ic
*h"sics *ri$aril" out of consieration of the #rossness of our sensor"
e2ui*$ent- it woul be unreasonable to e0*ect to fin it co$*lete in
itself) 6e can onl" $ake it lo#icall" co$*lete as far as the *oint where
its roots stretch own into *h"sics as a whole) Nor is $icrosco*ic
theor"- when se*arate fro$ the rest- self+sufficient) The *h"sical
2uantities *ertainin# to ato$s- electrons- *hotons- which we talk about
in $icrosco*ic theor"- $ust also be #i!en efinitions which will enable
the e0*eri$enter to $easure the$) .ut he oes not $easure the$-
an is not i$a#ine e!en in h"*othetical e0*eri$ents to $easure
the$- with ato$s- electrons- an *hotons4 the state$ents of
$icrosco*ic *h"sics are not assertions of the results of such
uni$a#inable e0*eri$ents) The $easure$ents are $ae with $eter
scales- $icro$eters- s*ectro#ra*hs + ulti$atel" with our own #ross
sense+or#ans) 7olar *h"sics alwa"s has the last wor in obser!ation-
for the obser!er hi$self is $olar)
The secret of the union of $olar an $icrosco*ic *h"sics + of relati!it"
theor" an 2uantu$ theor" + is ,the full circle,) The" are not so $uch
branches forkin# fro$ one root as se$i+circles 1oine at both ens)
3enerall" we enter on the circle at the 1unction now uner iscussion-
where relati!it" theor" takes its stanar of len#th fro$ 2uantu$
theor") .ut relati!it" theor"- which has $ae #reater *ro#ress alon# its
arc than 2uantu$ theor" alon# its arc- is alrea" e0*lorin# the other
1unction- where the cos$ical constant an $atters of that kin are
in!ol!e) At this 1unction the roots of 2uantu$ theor" *enetrate into
relati!it" theor"- as at the other 1unction the roots of relati!it" theor"
*enetrate into 2uantu$ theor") Onl" in co$bine relati!it"+2uantu$
theor" 9not to be confuse with current ,relati!istic 2uantu$ theor",
which has un1ustifiabl" usur*e the na$e: can we e0hibit the circle as
a whole)
The failure to $ake relati!it" theor" co$*letel" ine*enent of
2uantu$ theor" has one *ractical a!anta#e) It has secure that the
stanar of len#th in relati!it" is the sa$e as in 2uantu$ theor") The
sa$e ulti$ate stanar is also reco#ni/e b" the $etrolo#ists- who
seek to reali/e it in the wa!elen#th of ca$iu$ li#ht or the #ratin#+
s*ace of calcite) Thus *ractical $etrolo#ists- relati!it" *h"sicists an
2uantu$ *h"sicists all $ean the sa$e thin# when the" s*eak of len#th
or istance or ti$e+inter!al) There is co$*lete unifor$it" + e0ce*t for a
newco$er- the kine$atical cos$olo#ist- who consiers that e!er" one
is out of ste* e0ce*t hi$self)
It is often su##este that so$e of the constants of nature- e)#) the
!elocit" of li#ht or the #ra!itational constant- !ar" with ti$e) >nless the
stanars of len#th an ti$e+e0tension ha!e been carefull" efine-
such iscussions are $eanin#less4 an $uch that has been written on
the sub1ect is iscounte b" the fact that the writers are e!ientl"
unaware of the nature of the efinition of these stanars) An"one who
su##ests !ariation of a funa$ental constant has before hi$ a hea!"
task of reconstruction of theor" an reinter*retation of obser!ational
$easure$ents before he can reach an" obser!ational confir$ation or
contraiction of his su##estion) 7eanwhile I think that *ro#ress of the
e*iste$olo#ical $etho has assure us that the constants of nature
9a*art fro$ our arbitrar" units: are nu$bers introuce b" our
sub1ecti!e outlook- whose !alues can be calculate a priori an stan
for all ti$e) For this reason $" *ersonal conclusion is that there is no
$ore an#er that the !elocit" of li#ht or the constant of #ra!itation will
chan#e with ti$e than that the circu$ference+ia$eter ratio will
chan#e with ti$e)
Let us e0a$ine $ore closel" what is i$*lie in the su##estion that the
!elocit" of li#ht in vacuo chan#es with the ti$e) An i$$eiate
conse2uence is that the ratio of the wa!elen#th to the *erio T of
an" s*ectral line- sa" a h"ro#en line- chan#es with ti$e) Now for all
e*ochs the stanar of ti$e is a ti$e+*erio in so$e 2uantu$+
s*ecifie structure- an the stanar of len#th is a s*ace+e0tension in
so$e 2uantu$+s*ecifie structure) 6e $a" take this structure to be a
h"ro#en ato$ in the 2uantu$+s*ecifie state in which it e$its the line
consiere) It follows that either the ratio of the *erio of the e$itte
li#ht to the ti$e+*erio intrinsic in the e$ittin# ato$ !aries with ti$e- or
the ratio of the len#th of the e$itte wa!es to the s*atial scale of
structure of the e$ittin# ato$ !aries with ti$e) I o not think those who
*ro*ose the !ariabilit" of the !elocit" of li#ht reali/e that- if their wors
ha!e an" $eanin#- the" i$*l" that the *erio of the li#ht has no
constant relation to + is therefore not eter$ine b" + an"
corres*onin# *erioicit" in its source4 or alternati!el"- that the
wa!elen#th of the li#ht has no constant connection with the linear
scale of its source) If this were true- it woul in!ol!e a conce*tion of
ato$ic structure so far re$o!e fro$ that of *resent+a" 2uantu$
theor" that scarcel" an"thin# in our *resent knowle#e woul sur!i!e)
IA
Thus far we ha!e been consierin# how to efine len#th
una$bi#uousl"- but ha!e not attene to the co$*lications which $a"
arise when e0tre$e accurac" is re2uire) 6e ha!e been concerne to
circu$!ent the clerkes in Chaucer;s tale?
Ye conne b" ar#u$entes $ake a *lace
A $"le broo of twent" foot of s*ace)
These ar#u$ents still a**ear in scientific 1ournals- *articularl" in
reference to len#ths an ti$es at a re$ote cos$olo#ical e*och) It has-
for e0a$*le- been *ro*ose to e0ten the rather unco$fortabl"
cra$*e ti$e+scale of the uni!erse b" $akin# a lo#arith$ic
transfor$ation of our ti$e+reckonin#?
Lat see now if this *lace $a" suff"se
Or $ake it rou$ with s*eche- as is "oure #"se)
6e now *ass on to consier 2uestions of e0tre$e accurac") Our
2uantu$+s*ecifie stanar is re*roucible at the re$otest ti$es an
*laces- an therefore satisfies the ut$ost e$ans of cos$olo#ical
theor") Ne!ertheless- it has certain li$itations) I will $ention the two
$ost i$*ortant)
First- the 2uantu$+s*ecifie stanar oes not *ro!ie an e0act
efinition of len#th in stron# electric or $a#netic fiels) This is because
it is not strictl" re*roucible in such fiels4 a structure in an
electro$a#netic fiel cannot ha!e *recisel" the sa$e 2uantu$
s*ecification as a structure in fiel+free conitions) 6e a**eal in !ain
to the *ractical $etrolo#ist to arbitrate on this ifficult"4 he $erel"
*oints out that- in usin# a stanar of len#th- it is an ele$entar"
*recaution to #et ri of electric an $a#netic fiels) .ut it is ile to tell
a $an who is stu"in# *heno$ena in a $a#netic fiel that he $ust #et
ri of the fiel before $akin# an" $easure$ents) Let us su**ose that
he wishes to $easure the cur!ature of tracks of char#e *articles in a
$a#netic fiel) He will *robabl" not be so ill+a!ise as to *ut the
stanar into the fiel4 he will- for e0a$*le- take a *hoto#ra*h of the
tracks an $easure the photograph with the stanar) He $ust then
use theoretical for$ulae to euce the cur!ature of the tracks in the
fiel fro$ $easure$ents of the *hoto#ra*hs outsie the fiel) .ut how
is he to test whether his theoretical for$ulae are ri#ht@ The inirect
*roceure is onl" 1ustifie if it is known to #i!e the sa$e results as a
irect $easure$ent woul ha!e one4 but in this case- owin# to the
non+re*roucibilit" of the stanar- no irect *roceure e0ists4 an it
cannot be clai$e that the inirect *roceure is known to #i!e the
sa$e results as a non-e)istent irect *roceure)
Since there are no *receent con!entions to follow- a theorist
concerne with e0act e2uations in intense fiels is free to introuce his
own efinition of len#th- *ro!ie onl" that it con!er#es to the
acce*te efinition as the fiel tens to /ero) So $uch a!anta#e has
been taken of this freeo$- that at least a o/en ifferent ,unifie
theories, of the #ra!itational an electro$a#netic fiels ha!e been *ut
forwar each i$*l"in# a sli#htl" ifferent efinition of len#th) The" are
all of the$ ri#ht + ri#ht if the efinition of len#th is a1uste to fit the$)
The" will all of the$ be ,confir$e b" obser!ation,- because the
inferences fro$ $easure$ents $ae outsie the fiel- or the
corrections to the $easure$ents 9if an": $ae insie the fiel- will be
eter$ine fro$ the theor" uner test?
There are nine an si0t" wa"s of constructin# tribal la"s-
An e!er" sin#le one of the$ is ri#htH
.ut- it $a" be sai- will not 2uantu$ theor" ulti$atel" be able to
calculate *recisel" how $uch a cr"stal stanar e0*ans or contracts
when *lace in a $a#netic fiel- or how a wa!elen#th is $oifie@ 6e
shall then $erel" ha!e to a**l" the corrections for the chan#e of the
stanar) I a$ afrai it is not 2uite so si$*le) Not e!en 2uantu$ theor"
can calculate a 2uantit" which has not been efine) >noubtel"
2uantu$ theor" will fin us a correction4 but that onl" $eans that
2uantu$ theor"- like the unifie theories- has introuce 9or will
introuce: its own con!entional efinition) 8oubtless a efinition which
co$$ens itself to the 2uantu$ *h"sicists will *re!ail in the en b"
force ma%eure4 an inee it is a**ro*riate to lea!e the $atter in their
hans- since it was 2uantu$ theor" that su**lie our ori#inal
efinition) .ut it shoul be borne in $in that- althou#h it is clearl"
necessar" b" so$e con!ention to e0ten the usual *h"sical
ter$inolo#" to intense fiels- whate!er con!ention we ao*t- the
istances will be *seuo+istances 9an si$ilarl" for all eri!e
*h"sical 2uantities- incluin# the $easure of the fiel itself:- since the"
lack the $ost funa$ental characteristic of the $etrolo#ical
conce*tion of len#th- na$el" the corres*onence between si$ilarit" of
len#th an si$ilarit" of *h"sical structure)
A
The secon li$itation is that the stanar $ust be short) A lon#
stanar will not work e0ce*t in !er" s*ecial circu$stances)
Su**ose we tr" to $easure the ia$eter of the earth with a lon#
cr"stal stanar stuck throu#h it like a knittin# neele throu#h an
oran#e) It is well known that the earth is straine out of sha*e b" the
tie+raisin# forces of the sun an $oon4 the lon# cr"stal will likewise
be straine) A *ractical $etrolo#ist woul insist on re$o!in# the sun
an $oon 9an earth: before elicate $easure$ents were atte$*te-
since it is an ele$entar" *recaution that the stanar $ust not be
sub1ecte to strain) 6e $a" e0*ress the ob1ection $ore for$all" b"
*ointin# out that the strain $eans that the structure of the cr"stal no
lon#er has the s*ecification *rescribe in the efinition of the stanar)
6e cannot alwa"s re$o!e the boies that are causin# the strain) If we
are $easurin# u* the solar s"ste$- we cannot be#in the *roceein#s
b" clearin# awa" the sun) Thus- in #eneral- we ha!e to be content with
short stanars which are *ro*ortionatel" less affecte b" strain) 6ith
the short stanar we can onl" $easure short istances irectl") To a
first a**ro0i$ation we can eter$ine lar#e istances b" $easurin#
the$ in short sections- an su$$in# or inte#ratin# the results4
(
but to
a hi#her a**ro0i$ation this $etho also leas to a$bi#uous results)
This a$bi#uit" is known as the non-integrability of displacement)
It is worth noticin# that there is a irect obser!ational test which woul
show that the lon# cr"stal ro no lon#er ha the *rescribe structure)
If- for e0a$*le- a sin#l" refractin# cr"stal were chosen- the strain
woul be inicate b" the a**earance of ouble refraction) The *oint
of this re$ark is that the non+re*roucibilt" of a lon# stanar- an our
conse2uent inabilit" to efine non+infinitesi$al istances accuratel"- is
a fact which $i#ht ha!e been isco!ere b" irect obser!ation- instea
of bein# euce fro$ our knowle#e of tie+raisin# forces in
#ra!itational theor") This is i$*ortant- because *resentl" we shall use
the ineter$inac" of lon# istances as a founation for #ra!itational
theor"- instea of eucin# it as a conse2uence of the theor")
For the sur!e" of a lar#e re#ion we ha!e to e!elo* a s"ste$ of
$etrical escri*tion in which onl" the infinitesi$al istances are
obser!ational ata) This is a technical *roble$ of ifferential #eo$etr"-
which we nee not iscuss here) For si$*licit" I ha!e o$itte
reference to ti$e4 but si$ilar consierations a**l" to the four+
i$ensional s*ace+ti$e worl)
The failure to efine lon# istances obser!ationall"- or in $athe$atical
8
That is to sa"- we efine a lar#e istance as the result of inte#ratin# short istances
9*ro!ie that the result is una$bi#uous: instea of efinin# it as the result of co$*arison
with a lon# stanar)
lan#ua#e the non+inte#rabilit" of is*lace$ent- is the founation of
Einstein;s theor" of #ra!itation) Accorin# to the usual outlook
#ra!itation is the cause of the trouble4 #ra!itation *rouces the strains
which rener lon# stanars useless) .ut Einstein;s outlook is $ore
nearl" that the ,trouble, + the non+inte#rabilit" of is*lace$ent + is the
cause of #ra!itation) I $ean that in Einstein;s theor" the orinar"
$anifestations of #ra!itation are euce as $athe$atical
conse2uences of the non+inte#rabilit" of is*lace$ent) I cannot enter
here into the etails- which re2uire a lar#e treatise4 but the #ist of it is
that Einstein showe how to s*ecif" the non+inte#rabilit" 2uantitati!el"-
an use the nu$bers thus introuce + the fa$ous #

+ as a $easure
of the influence which isturbs the ieal conitions in which
is*lace$ents woul be inte#rable) ,3ra!itational fiel, is the na$e
which we ha!e #i!en to this influence) As $i#ht be e0*ecte- this
s"ste$atic s*ecification of the #ra!itational fiel has been foun to be
$ore *recise than the casual s*ecification of it b" one of its effects
which ha**ene to strike Newton;s attention when he sat uner an
a**le tree)
Einstein;s s*ecification is $ore accurate than Newton;s4 but that the
two refer to the sa$e thin# is seen when we recall that it was the
strain- *rouce b" the two ens of the lon# stanar tr"in# to fall with
ifferent accelerations towars the sun or $oon- which !itiate it as a
stanar an frustrate our effort to $easure irectl" an inte#rate
len#th) 6e nee therefore not be sur*rise that fro$ Einstein;s
s*ecification the $ore orinar" $anifestations of #ra!itation in fallin#
boies can be euce)
This is a *articularl" #oo e0a$*le of the wa" in which e*iste$olo#ical
stu" has brou#ht about a #reat a!ance in science4 an it is worth
while to recall the *rinci*al ste*s) If *h"sics is to escribe what we
reall" obser!e- we $ust o!erhaul the efinitions of the ter$s e$*lo"e
in it so that the" e0*licitl" refer to obser!ational facts an not to
$eta*h"sical con1ectures) Len#th an ti$e inter!al in *articular nee
to be carefull" efine- since the" are the basis of nearl" all other
*h"sical efinitions) To a!oi circular efinitions it is essential that the
stanars of len#th an ti$e inter!al shoul be the e0tensions of
structures co$*letel" s*ecifie b" *ure nu$bers) 6ith such structures
as stanar we obtain a efinition of infinitesi$al inter!als 9in the
absence of an electro$a#netic fiel:- but we o not obtain an e0act
efinition of lon# inter!als) Thus- in orer that *h"sics $a" e0*ress
*urel" obser!ational knowle#e- it is necessar" to e!elo* a s"ste$ of
escri*tion of the location of e!ents base wholl" on infinitesi$al
istances an ti$e inter!als4 we thereb" a!oi reference to lon#
inter!als which ha!e no e0act obser!ational efinition) This s"ste$ of
location- e*enin# on infinitesi$al inter!als- is the founation of
#eneral relati!it" theor") In relati!it" theor" a lon# istance is in #eneral
an a**ro0i$ate conce*tion onl"4 it is inca*able of e0act efinition)
&
As soon as we reali/e that the efinition of len#th oes not co!er lon#
istances an so oes not i$*l" inte#rabilit" of is*lace$ent-
inte#rabilit" beco$es a s*ecial h"*othesis which re2uires efenin#)
One oes not acce*t h"*otheses #ratuitousl") Proceein# fro$ this
rational basis of s*ace+ti$e $easure$ent we fin that the
*heno$enon of #ra!itation a**ears auto$aticall" + unless we
eliberatel" introuce a h"*othesis of inte#rabilit" to e0clue it + an in
this wa" we are le i$$eiatel" to Einstein;s theor" of #ra!itation)
AI
I ha!e been continuall" e$*hasi/in# the sub1ecti!it" of the uni!erse
escribe in *h"sical science) .ut- "ou $a" ask- was it not the boast
of the theor" of relati!it" that it *enetrate be"on the relati!e
9sub1ecti!e: as*ect of *heno$ena an ealt with the absolute@ For
e0a$*le- it showe that the usual se*aration of s*ace an ti$e is
sub1ecti!e- bein# e*enent on the obser!er;s $otion- an it
substitute a four+i$ensional s*ace+ti$e ine*enent of the
obser!er) It $a" see$ ifficult to reconcile this !iew of Einstein;s
theor" as liftin# the !eil of relati!it" which hies the absolute fro$ us-
with $" *resent account of $oern *h"sics as ac2uiescin# in- an
9
The breakown of the orinar" efinition lea!es the ter$ at the is*osal of in!esti#ators-
an !arious technical efinitions of lon# istances ha!e been *ro*ose) .ut these
technical uses of the ter$ are irrele!ant here)
$akin# the best of- a *artiall" sub1ecti!e uni!erse)
It is necessar" to re$e$ber that there has been thirt" "ears; *ro#ress)
Relati!it" be#an like a new broo$- swee*in# awa" all the sub1ecti!it" it
foun) .ut- as we ha!e a!ance- other influences of sub1ecti!it" ha!e
been etecte which are not so easil" eli$inate) Probabilit"- in
*articular- is frankl" sub1ecti!e- bein# relati!e to the knowle#e which
we ha**en to *ossess) Instea of bein# swe*t awa"- it has been
e0alte b" wa!e $echanics into the $ain the$e of *h"sical law)
The sub1ecti!it" referre to in these lectures is that which arises fro$
the sensor" an intellectual e2ui*$ent of the obser!er) 6ithout
!ar"in# this e2ui*$ent- he can !ar" in *osition- !elocit"- an
acceleration) Such !ariations will *rouce sub1ecti!e chan#es in the
a**earance of the uni!erse to hi$4 in *articular the chan#es
e*enin# on his !elocit" an acceleration are $ore subtle than was
reali/e in classical theor") Relati!it" theor" allows us to re$o!e 9if we
wish: the sub1ecti!e effects of these personal characteristics of the
obser!er4 but it oes not re$o!e the sub1ecti!e effects of generic
characteristics co$$on to all ,#oo, obser!ers + althou#h it has
hel*e to brin# the$ to li#ht)
Confinin# attention to the *ersonal- as istin#uishe fro$ the #eneric-
sub1ecti!it"- let us see *recisel" what is $eant b" re$o!in# this
sub1ecti!it") There oes not see$ to be $uch ifficult" in concei!in#
the uni!erse as a three+i$ensional structure !iewe fro$ no
*articular *osition4 an I su**ose we can- after a fashion- concei!e it
without an" stanar of rest or of non+acceleration) It is *erha*s rather
unfortunate that it is- or see$s to be- so eas" to concei!e4 because the
conce*tion is liable to be $ischie!ous fro$ the obser!ational *oint of
!iew) Since *h"sical knowle#e $ust in all cases be an assertion of
the results of obser!ation 9actual or h"*othetical:- we cannot a!oi
settin# u* a u$$" obser!er4 an the obser!ations which he is
su**ose to $ake are sub1ecti!el" affecte b" his *osition- !elocit"
an acceleration) The nearest we can #et to a non+sub1ecti!e- but
ne!ertheless obser!ational- !iew is to ha!e before us the re*orts of all
*ossible u$$" obser!ers- an *ass in our $ins so ra*il" fro$ one
to another that we ientif" oursel!es- as it were- with all the u$$"
obser!ers at once) To achie!e this we see$ to nee a re!ol!in# brain)
Nature not ha!in# enowe us with re!ol!in# brains- we a**eal to the
$athe$atician to hel* us) He has in!ente a transfor$ation *rocess
which enables us to *ass !er" 2uickl" fro$ one u$$" obser!er;s
account to another;s) The knowle#e is e0*resse in ter$s of tensors
which ha!e a fi0e s"ste$ of interlockin# assi#ne to the$4 so that
when one tensor is altere all the other tensors are altere- each in a
eter$inate wa") ." assi#nin# each *h"sical 2uantit" to an
a**ro*riate class of tensor- we can arran#e that- when one 2uantit" is
chan#e to corres*on to the chan#e fro$ u$$" obser!er A to
u$$" obser!er .- all the other 2uantities chan#e auto$aticall" an
correctl") 6e ha!e onl" to let one ite$ of knowle#e run throu#h its
chan#es + to turn one hanle + to #et in succession the co$*lete
obser!ational knowle#e of all the u$$" obser!ers)
The $athe$atician #oes one ste* farther4 he eli$inates the turnin# of
the hanle) He concei!es a tensor s"$bol as containin# in itself all its
*ossible chan#es4 so that when he looks at a tensor e2uation- he sees
all its ter$s chan#in# in s"nchroni/e rotation) This is nothin# out of
the wa" for a $athe$atician4 his s"$bols co$$onl" stan for
unknown 2uantities- an functions of unknown 2uantities4 the" are
e!er"thin# at once until he chooses to s*ecif" the unknown 2uantit")
An so he writes own the e0*ressions which are s"$bolicall" the
knowle#e of all the u$$" obser!ers at once + until he chooses to
s*ecif" a *articular u$$" obser!er)
.ut- after all- this e!ice is onl" a translation into s"$bolis$ of what
we ha!e calle a re!ol!in# brain) A tensor $a" be sai to s"$boli/e
absolute knowle#e4 but that is because it stans for the sub1ecti!e
knowle#e of all *ossible sub1ects at once)
This a**lies to *ersonal sub1ecti!it") To re$o!e the #eneric
sub1ecti!it"- ue sa" to our intellectual e2ui*$ent- we shoul ha!e
si$ilarl" to s"$boli/e the knowle#e as it woul be a**rehene b" all
*ossible t"*es of intellect at once) .ut this coul scarcel" be
acco$*lishe b" a $athe$atical transfor$ation theor") An what
woul be the result if it were acco$*lishe@ Accorin# to Cha*ter IA- if
we re$o!e all sub1ecti!it" we re$o!e all the funa$ental laws of
nature an all the constants of nature) .ut- after all- these sub1ecti!e
laws an facts ha**en to be i$*ortant to bein#s who are not e2ui**e
with re!ol!in# brains an !ariable intellects) An if the *h"sicist oes
not take char#e of the$- no one else is 2ualifie to o so)
E!en in relati!it" theor"- which eals with the absolute 9in a so$ewhat
li$ite sense:- we continuall" hark back to the relati!e to e0a$ine how
our results will a**ear in the e0*erience of an ini!iual obser!er) 6e
are not so ea#er now as we were twent" "ears a#o to eli$inate the
obser!er fro$ our worl !iew) So$eti$es it $a" be esirable to banish
hi$ an his sub1ecti!e istortion of thin#s for a ti$e- but we are boun
to brin# hi$ back in the en4 for he stans for + oursel!es)
CHAPTER AI
EPISTE7OLO3Y AN8 I>ANT>7 THEORY
I
I $ust still kee* ha$$erin# at the 2uestion- 6hat o we reall"
obser!e@ Relati!it" theor" has returne one answer + we onl" obser!e
relations) Iuantu$ theor" returns another answer + we onl" obser!e
probabilities)
Consiere e*iste$olo#icall"- *robabilit" is a !er" *eculiar conce*tion4
for a *recise an certain knowle#e of a *robabilit" is construe as a
rou#h an uncertain knowle#e of so$ethin# of which it is the
*robabilit") This see$s to conflict with our co$fortable assurance that
knowle#e is knowle#e- an facts are facts) Probabilit" is co$$onl"
re#are as the antithesis of fact4 we sa" ,This is onl" a *robabilit"-
an $ust not be taken as a fact,) .ut- if the answer of 2uantu$ theor"
is ri#ht- the ,har facts of obser!ation, are *robabilities) 6hat we
$ean is that the result of an obser!ation- thou#h unoubtel" a fact in
itself- is onl" !aluable scientificall" because it infor$s us of the
*robabilit" of so$e other fact) These seconar" facts- known to us
onl" throu#h *robabilit"- for$ the $aterial to which the #enerali/ations
of *h"sics refer)
6e $a" assu$e that current theor" is ri#ht in its conclusion that our
$easure$ents eter$ine onl" the *robabilities of the 2uantities an
entities which fi#ure in classical *h"sics4 so that the *robabilit"
conce*tion necessaril" intrues in a co$*arison of the newer with the
classical ieas) .ut it is *erha*s not ob!ious that a *robabilit"
inter*retation is essential when we are seekin# to e!elo* the new
outlook on its own ine*enent basis) The results of obser!ational
$easure$ent can be construe as a ha/" an uncertain knowle#e of
the entities of classical *h"sics4 but oes not that $ean that we shoul
abanon the classical entities- an introuce $ore funa$ental
entities of which obser!ation #i!es us *recise an certain knowle#e@
The su##estion is that in the new *h"sics the so+calle *robabilities
are actuall" the real entities + the ele$ental stuff of the *h"sical
uni!erse) 6e ha!e *recise knowle#e of the$4 an it woul see$
retro#ressi!e to *ostulate other entities behin the$ of which our
knowle#e $ust alwa"s be uncertain)
I think that this iea is at the back of a rather co$$on su##estion that
a *ro*er refor$ulation of our ele$entar" conce*ts woul banish the
*resent ineter$inis$ fro$ the s"ste$ of *h"sics) The iea is that the
ineter$inis$ re!eale b" the new *h"sics is not intrinsic in the
uni!erse- but a**ears onl" in our atte$*t to connect it with the
obsolete uni!erse of classical *h"sics) Probabilit" woul then be
$erel" the funnel throu#h which the new wine is *oure into ol
bottles)
.ut the su##estion o!erlooks the essential feature of the
ineter$inis$ of the *resent s"ste$ of *h"sics- na$el" that the
2uantities which it can *reict onl" with uncertaint" are 2uantities
which- when the time comes- we shall be able to obser!e with hi#h
*recision) The fault is therefore not in our ha!in# chosen conce*ts
ina**ro*riate to obser!ational knowle#e) For e0a$*le- Heisenber#;s
*rinci*le tells us that the *osition an !elocit" of an electron at an"
$o$ent can onl" be known with a $utuall" relate uncertaint"4 an-
takin# the $ost fa!orable co$bination- the *osition of the electron one
secon later is uncertain to about C centi$eters) This is the uncertaint"
of the *reiction fro$ the best *ossible knowle#e we can ha!e at the
ti$e) .ut one secon later the *osition can be obser!e with an
uncertaint" of no $ore than a fraction of a $illi$eter) It has often been
ar#ue that the i$*ossibilit" of knowin# si$ultaneousl" the e0act
*osition an e0act !elocit" onl" shows that *osition an !elocit" are
unsuitable conce*tions to use in e0*ressin# our knowle#e) I ha!e no
s*ecial attach$ent to these conce*tions4 an I will #rant- if "ou like-
that our knowle#e of the uni!erse at the *resent $o$ent can be
re#are as *erfectl" eter$inate 9the su**ose ineter$inac" bein#
introuce in translatin# it into an ina**ro*riate fra$e of conce*tion:)
.ut that oes not re$o!e the ,ineter$inis$, 9which is istinct fro$
the ,ineter$inac",:- na$el" that this knowle#e- howe!er e0*resse-
is inae2uate to *reict 2uantities which- ine*enentl" of our fra$e of
conce*tion- can be irectl" obser!e when the ti$e co$es)
Returnin# to the $ore #eneral as*ect of the *robabilit" conce*tion- we
fin that it cannot be #ot ri of b" an" transfor$ation of outlook) It is
not *ossible to transfor$ the current s"ste$ of *h"sics- which b" its
e2uations links *robabilities in the future with *robabilities in the
*resent- into a s"ste$ which links orinar" *h"sical 2uantities in the
future with orinar" *h"sical 2uantities in the *resent- without alterin#
its obser!able content) The bar to such a transfor$ation is that
*robabilit" is not an ,orinar" *h"sical 2uantit",) At first si#ht it a**ears
to be one4 we obtain knowle#e of it fro$ obser!ation- or fro$ a
$i0ture of obser!ation an euction- as we obtain knowle#e of other
*h"sical 2uantities) .ut it is ifferentiate fro$ the$ b" a *eculiar
irre!ersibilit" of its relation to obser!ation) The result of an obser!ation
eter$ines efinitel" a *robabilit" istribution of so$e 2uantit"- or a
$oification of a *re!iousl" e0istin# *robabilit" istribution4 but the
connection is not re!ersible- an a *robabilit" istribution oes not
eter$ine efinitel" the result of an obser!ation) For an orinar"
*h"sical 2uantit" there is no ifference between $akin# a new
eter$ination an !erif"in# a *reicte !alue4 but for a *robabilit" the
*roceures are istinct)
Thus we $a" e0*an the answer of 2uantu$ theor" that ,we onl"
obser!e *robabilities, into the for$? The s"nthesis of knowle#e which
constitutes theoretical *h"sics is connecte with obser!ation b" an
irre!ersible relation of the for$al t"*e fa$iliar to us in the conce*t of
*robabilit")
6e shall later 9*) &G: ha!e to consier the e*iste$olo#ical reasons
which ha!e $ae it necessar" for theoretical *h"sics to *rocee in this
wa"- rather than b" aherin# to the for$ulation of a uni!erse whose
contents ha!e a re!ersible relation to obser!ation) .ut for the *resent
we shall si$*l" acce*t the $oern theor" as the result of a posteriori
e0a$ination of obser!ational knowle#e- an tr" to unerstan the
nature of the irre!ersibilit" is*la"e in it)
II
The followin# e0a$*le will hel* to $ake clear the irre!ersibilit"
associate with *robabilit") 6e ha!e two si$ilar ba#s A an .4 A
contains two white balls an one re- . contains two re balls an one
white) 6e raw a ball fro$ one of the ba#s- an fin that it is white) It
can be euce that the chances are 5 to % that the ba# is A) Si$ilarl"
the rawin# of a re ball inicates that the chances are 5 to % that the
ba# is .) Now su**ose that we are hane one of the ba#s with the
infor$ation that the chances are 5 to % that it is the ba# A4 what will be
the result of rawin# a ball@ Re!ersibilit" woul re2uire the answer to
be efinitel" a white ball4 for if a re ball is rawn it shows that the
chances are 5 to % that the ba# is . + contrar" to the infor$ation
state) .ut this answer is- of course- 2uite wron#4 the correct answer is
that the chances are < to C in fa!or of white)
Let us call the *robabilit" that the ba# is A the $-ness of the ba#) A
certain obser!ational *roceure a**lie to the ba# 9rawin# a ball: can
be use to $easure the A+ness) Of the two *ossible results ) an y of
the *roceure- ) inicates an A+ness 5N'- an y an A+ness %N') .ut if
the *roceure is a**lie to a ba# whose A+ness is known to be 5N'- the
result is not necessaril" )) This is in contrast to an orinar" *h"sical
2uantit") If we were eter$inin# the wei#ht instea of the A+ness- an
fro$ the results of reain#s ) an y we coul infer wei#hts % #$) an 5
#$)- res*ecti!el"- it is a0io$atic that a wei#ht alrea" known to be %
#$) woul #i!e the reain# )- not y)
Another contrast $a" be notice) Su**ose that fro$ the result of an
obser!ation we ha!e eter$ine the !alue ) of a *h"sical 2uantit") If
we re*eat the obser!ation an obtain the sa$e result- we take it as
confir$in# the !alue )) .ut it is not so with *robabilit") ." rawin# a
ball we eter$ine the A+ness of the ba# to be 5N') If we re*eat the
rawin# an obtain the sa$e result 9na$el" white:- instea of takin#
this as a confir$ation- we alter the !alue 5N' to CN<H The A+ness
inicate b" rawin# two white balls in succession is CN<)
To show that the sa$e irre!ersibilit" a**lies to *robabilit" as it is
actuall" use in $oern *h"sical theories- we $a" contrast the
*robabilit" wa!es treate in 2uantu$ theor" with soun wa!es)
Accorin# to wa!e $echanics- an obser!ation eter$ines or *rouces
a concentrate wa!e *acket in the *robabilit" istribution) This wa!e
*acket iffuses accorin# to laws e$boie in the e2uations of the
theor"4 an we can calculate the for$ into which the wa!e *acket will
ha!e s*rea one unit of ti$e later) *ut the theory does not assert that
this is the form of wave packet which would be produced by an
observation made one unit of time later! On the other han- if fro$ the
obser!ationall" eter$ine for$ of a soun wa!e at one instant we
calculate the for$ into which it will ha!e s*rea one unit of ti$e later-
the whole *oint of the theor" is that we obtain the for$ which woul be
eter$ine b" obser!ations $ae one unit of ti$e later)
%D
A*art therefore fro$ an" other i$*lications of *robabilit"- we $a"
istin#uish it for$all" as a na$e #i!en to so$ethin# which has a
relation to obser!ational knowle#e ifferent fro$ that of orinar"
*h"sical 2uantities + an irre!ersible instea of a re!ersible relation) This
absolute ifference will re$ain howe!er we chan#e the no$enclature4
an it will be seen that the ho*e of #ettin# back to so$ethin# like the
classical s"ste$ of *h"sics b" rena$in# the *robabilities as entities
cannot be fulfille)
The entire s"ste$ of laws of *h"sics at *resent reco#ni/e is
concerne with *robabilit"- which- we ha!e seen- si#nifies so$ethin#
that has an irre!ersible relation to obser!ation) As a $eans of
calculatin# future *robabilities the laws for$ a co$*letel" eter$inistic
s"ste$4 but as a $eans of calculatin# future obser!ational knowle#e
the s"ste$ of law is ineter$inistic) The irre!ersibilit" ensures that-
10
The *arallelis$ is obscure b" the fact that we s*eak of the for$ of the soun wa!e as
,obser!ationall" eter$ine,- an the for$ of the *robabilit" wa!e as ,obser!ationall"
*rouce,) This ifference of *hrasin# is itself a reco#nition of the istinction between a
re!ersible an an irre!ersible relation to obser!ation)
thou#h we *ut efinite obser!ational knowle#e into the eter$inistic
$achine- we cannot take efinite obser!ational knowle#e out of it)
Hence- as re#ars obser!ational content- the s"ste$ of $oern
*h"sics is ineter$inistic)
6e ha!e sai 9*) %D: that e!er" ite$ of *h"sical knowle#e $ust be an
assertion of what woul be the result of carr"in# out a s*ecifie
obser!ational *roceure) It is now necessar" to a that it is a
2ualifie assertion clai$in# onl" a certain e#ree of *robabilit") Strictl"
s*eakin#- the Court of A**eal shoul be aske to ecie- not whether
the assertion is true- but whether it has the e#ree of *robabilit"
clai$e for it) >suall"- howe!er- we sub$it to obser!ational test
assertions which clai$ to ha!e so hi#h a *robabilit" as to be
,*racticall" certain,) Low *robabilities can onl" be #i!en a statistical
test) This $eans that the ini!iual assertion of low *robabilit" is
re*lace b" a statistical conse2uence which has !er" hi#h *robabilit" +
hi#h enou#h to be rate as *racticall" certain + an the latter is teste
b" obser!ation) Thus the Court of A**eal is onl" calle u*on to 1u#e
assertions clai$e as *racticall" certain)
Re$e$berin# that e!er" ite$ of *h"sical knowle#e $ust assert the
result of an obser!ational *roceure- we ha!e to in2uire what is the
obser!ational *roceure whose result is asserte when we ha!e
*h"sical knowle#e that the *robabilit" of a state e!ent is %N') The
onl" obser!ational *roceure in an" wa" associate with this
knowle#e is the statistical test abo!e+$entione- which consists of
eter$inin# the fre2uenc" in a lar#e class of e!ents of a si$ilar kin to
the *articular e!ent state) 6e ha!e alrea" cone$ne the
i$$oralit" of clai$in# obser!ational knowle#e of one thin# when we
ha!e actuall" teste so$ethin# ifferent) The statistical assertion $ust
therefore be re#are- not as euce fro$ the assertion of low
*robabilit"- but as e0*lainin# what a low *robabilit" actuall" $eans4
that is to sa"- the assertion of low *robabilit" $ust be unerstoo to
assert the result of the statistical test- althou#h !erball" it a**ears to
refer to a sin#le e!ent) 6hate!er si#nificance *robabilit" $a" ha!e in
other e*art$ents of thou#ht- in *h"sical science *robabilit" is
essentiall" a statistical conce*tion4 that is to sa"- it is defined as a
fre2uenc" in a class of e!ents)
The ter$ ,*robabilit", is often use- without statistical i$*lications- in
reference to the stren#th of our e0*ectation or belief) 6hen use in
this sense it cannot for$ *art of a scientific assertion- since the
assertion woul be renere thereb" inca*able of obser!ational test)
.ut it can still be use to 2ualif" the assertion as a whole + to escribe
our confience or lack of confience that the assertion woul be
confir$e if sub$itte to obser!ational test) It is i$*ortant to
istin#uish this non+2uantitati!e use of the ter$ fro$ its technical
scientific use as an obser!ationall" eter$inable 2uantit")
%%
III
Probabilit"- which was first introuce into theoretical *h"sics in
connection with ther$o"na$ics an the kinetic theor" of #ases- has
been of increasin# i$*ortance in later e!elo*$ents- an is now fir$l"
e$bee as one of the $ost funa$ental conce*ts) 6e ha!e seen
that- owin# to its irre!ersible relation to obser!ation- it is
istin#uishable fro$ other *h"sical 2uantities) 6e cannot eli$inate this
irre!ersibilit" b" an" chan#e of our conce*tual fra$e of thou#ht) If we
shoul ecie that it ou#ht to be roote out of *h"sics- the onl" wa" is
to abanon the *resent s"ste$ of *h"sics an buil u* a new one fro$
the be#innin#) This brin#s us to a consieration of the a priori as*ect
of the *roble$) If we consier the wa" in which obser!ational
knowle#e is obtaine an for$ulate- at what *oint oes it beco$e a
$atter of necessit" or of e0*eienc" to e!iate into *robabilit"@ The
answer is not so ele$entar" as we $i#ht at first e0*ect) It see$s as
thou#h it ou#ht not to be ifficult to for$ulate obser!ational knowle#e
as a *recise escri*tion of a uni!erse without usin# the *robabilit"
conce*tion4 an it is b" no $eans eas" to *ut one;s fin#er on the flaw
in such an atte$*t)
11
For other as*ects of the *roble$ of *robabilit" in *h"sics- see New Pathways in
Science- Cha*ter AI)
It is first necessar" to recall the i$*licit unerstanin# that b"
,obser!ation, is $eant good obser!ation) To efine a *h"sical 2uantit"
we $ust s*ecif" a *roceure which will #i!e a #oo $easure$ent of it)
.ut we ha!e now to introuce a new *oint) ,3oo, is not here to be
taken to $ean ,*erfect,) ." good obser!ation we e$*haticall" o not
$ean perfect obser!ation)
The trouble is not that in *ractice all our $easure$ents are $ore or
less i$*erfect) It woul certainl" be untrue to sa" that the basis of
current *h"sical knowle#e is *erfect obser!ation) .ut this is onl" a
for$al criticis$- which we $a" consier to ha!e been ae2uatel"
countere b" the orinar" theor" of errors of obser!ation) If the
*robabilit" conce*t in *h"sics ha been confine to the theor" of errors
of obser!ation- it woul not ha!e been of an" #reat concern to us here)
.ut it has *enetrate $uch $ore ee*l" into the roots of *h"sics)
The serious ifficult" a**ears when we consier what the efinition of
a *erfect obser!ation i$*lies) 6e re2uire- not onl" *erfect a**liances
an *erfect skill- but *erfect conitions + free fro$ isturbin#
influences) To co$*lete the efinition the *erfect conitions $ust be
s*ecifie in ter$s of obser!ation) It is not sufficient to sa" that all
isturbin# influences $ust be eli$inate4 we cannot ecie whether an
outsie influence is a isturbance or *art of the stanar conitions for
a *erfect obser!ation- unless the stanar conitions of *erfect
obser!ation ha!e been efine)
A #oo e0*eri$enter rearran#es thin#s in the nei#hborhoo of the
s"ste$ he is stu"in#) He surrouns it with a ther$ostat4 he shiels it
fro$ raio+acti!e substances4 he counteracts the earth;s $a#netic
fiel) These are his efforts to secure stanar good conitions) A #oo
obser!er is + a bit fuss") A *erfect obser!er woul be intolerable) For
his stanar perfect conitions he woul want to rearran#e the stars
an i$*ro!e the uni!erse out of reco#nition)
The o thin# is that- ha!in# $ae his *erfect arran#e$ents- the
*erfect obser!er often fails to acco$*lish thin#s which to the #oo
obser!er are 2uite ele$entar") Here is a si$*le test) He is aske to *ut
a #ra$ of h"ro#en at DC into a s*herical !essel of < c$) raius- an
$easure the *ressure) The #oo e0*eri$enter will o this without
ifficult"4 but the *erfect e0*eri$enter- ha!in# trie se!eral ti$es an
obtaine wiel" ifferent answers each ti$e- #i!es it u* an
announces that the *ressure is 2uite ineter$inate) The reason is that
in $akin# the interior surface of the !essel not $erel" a #oo s*here
but a *erfect s*here- he has re$o!e those useful little rou#hnesses
which issi*ate an" an#ular $o$entu$ the #as $a" *ossess when it
is *ut in) Conse2uentl" the #as in each e0*eri$ent was left with an
arbitrar" rotation- an the $easure *ressures iffere accorin#l") In
stri!in# after *erfection the obser!er onl" achie!e ineter$inac")
Aarious contretemps of this kin await the *erfect obser!er who is
workin# alone4 but if another *erfect obser!er 1oins hi$- the result is
chaos) Each of the$- in orer to secure the *erfect conitions for his
own e0*eri$ent- will insist on re$o!in# the a**aratus set u* b" the
other) The obser!er of len#ths s$ooths out the uni!erse- so that no
as"$$etrical influence $a" istort his stanars4 whereu*on the
obser!er of an#ular irections co$*lains that his lan$arks ha!e been
re$o!e- an the uni!erse $ae so s"$$etrical that there is no
reco#ni/able *oint to $easure fro$)
One *erfect obser!er is a nuisance) Two *erfect obser!ers $ake a
fi#ht) Three *erfect obser!ers sen us fleein# for refu#e to the conce*t
of *robabilit")
6hilst one *erfect obser!er is en#a#e in the *ri$ar" obser!ation
#i!in# the $ain 2uantit" or reain#- an ar$" of *erfect obser!ers $ust
be testin# the conitions uner which the obser!ation is bein# $ae
so that the *ro*er corrections $a" be a**lie if the" are foun to be
i$*erfect) Possibl" the" $i#ht be able to o this without interferin#
with the *ri$ar" obser!er4 but- bein# *erfect obser!ers- the" too will
re2uire *erfect conitions for their $easure$ents- an these
conitions $ust be checke b" a further batch of *erfect obser!ers)
The result is a *erfect anarch" + obser!ers fi#htin# to re$o!e each
other;s a**aratus- interferin# with each other;s e0*eri$ents- tr"in# to
$ake e!er" *article in the uni!erse res*on to half a o/en tests at
once) To a!oi this fiasco of *erfection- we $ust be content to
co$*ro$ise- an trust a little to luck as well as $easure$ent) An so
we #et a s"ste$ of *h"sics in which luck 9*robabilit": an
$easure$ent are associate to#ether)
It is well known that the interference of ifferent kins of $easure$ent
is the source of Heisenber#;s uncertaint" *rinci*le- which is the
e*iste$olo#ical #atewa" b" which the *robabilit" conce*tion enters
2uantu$ theor")
IA
The al$ost uni!ersal *ractice of usin# the ter$ ,an obser!ation, for
what woul be $ore *ro*erl" escribe as ,an ite$ of obser!ational
knowle#e), is liable to cause confusion in *hiloso*hical iscussions)
Consier- for e0a$*le- an obser!ation of the a**arent $a#nitue of a
star) If we are aske to state the *recise nature of this obser!ation- we
#i!e an account of the *hoto$etric *roceure b" which the result- sa"
%%
$)
C5- is reache) .ut this result is not in itself an ite$ of
obser!ational knowle#e- or at least not an ite$ of scientific !alue) 6e
knew beforehan that a$on# the $"rias of stars so$e woul al$ost
certainl" ha!e the $a#nitue %%
$)
C5) The !aluable ite$ of
obser!ational knowle#e is that a star whose identity has been
recorded has the $a#nitue %%
$)
C5) The obser!ational *roceure b"
which this knowle#e is eri!e inclues the obser!ations necessar"
to ientif" the star as well as those which $easure stellar bri#htness)
In The Nature of the Physical World I introuce the ter$ ,*ointer
reain#, to escribe the #eneral nature of e0act obser!ation) 6hate!er
2uantit" we sa" we are ,obser!in#,- the actual *roceure nearl"
alwa"s ens in reain# the *osition of so$e kin of inicator on a
#rauate scale or its e2ui!alent) The *ointer reain# is an
obser!ation in the strict sense4 but it oes not b" itself constitute an
ite$ of obser!ational knowle#e- which is *robabl" the $ore usual
$eanin# of the wor) In a rather well+known e0a$*le in The Nature of
the Physical World- in referrin# to our obser!ational knowle#e that
,the $ass of the ele*hant is two tons,- I ientifie ,two tons, with the
reain# of the *ointer when the ele*hant was *lace on a wei#hin#
$achine4 but the knowle#e that the $ass two tons is the $ass of the
ele*hant is not ac2uire b" *a"in# attention solel" to the $o!e$ent of
the *ointer)
7ore #enerall" we $ust reco#ni/e that an ite$ of obser!ational
knowle#e in!ol!es- besies a *ri$ar" *ointer reain#- seconar"
*ointer reain#s ientif"in# the circu$stances in which the *ri$ar"
*ointer reain# occurre) It $ust be a$itte that e!en an isolate
*ointer reain# is an ite$ of knowle#e of a sort4 but it is not with such
ite$s that the scientific $etho eals) For scientific knowle#e the
association with other *ointer reain#s is an essential conition4 an
we $a" therefore escribe *h"sical knowle#e as a knowle#e of the
associations of *ointer reain#s)
The seconar" *ointer reain#s are ,co+orinates, 9in a #enerali/e
sense: of the *ri$ar" *ointer reain#) 6hen- for e0a$*le- we
eter$ine the intensit" of a $a#netic fiel- we associate with it the
ti$e an the co+orinates in s*ace of the *oint to which the
eter$ination a**lies) The $a#netic intensit" is then the *ri$ar"
*ointer reain#- an the co+orinates in s*ace an ti$e are the
seconar" *ointer reain#s) .ut the chain of *ointer reain#s oes not
sto* here) Tertiar" *ointer reain#s are re2uire to ientif" the s"ste$
of co+orinates use- an to eter$ine its $etric4 but these tertiar"
reain#s are co$$on to all ite$s of knowle#e referre to the
coorinate fra$e- an 9unlike the seconar" *ointer reain#s: are not
eter$ine afresh for e!er" *ri$ar" reain#) There is consierable
econo$" in usin# a s"ste$atic *lan of ientification such as a s*ace+
ti$e co+orinate s"ste$4 for otherwise a lon# re#ression of *ointer
reain#s $a" be necessar" to connect the *ri$ar" *ointer reain#
with other ele$ents in our *h"sical knowle#e)
In The Nature of the Physical World it is e$*hasi/e that *h"sical
knowle#e is concerne with the connection of *ointer reain#s rather
than with the *ointer reain#s the$sel!es4 an it is conclue that the
connecti!it" of *ointer reain#s- as e0*resse b" the laws of *h"sics-
su**lies the co$$on back#roun which realistic *roble$s alwa"s
e$an + the back#roun escribe b" the tertiar" *ointer reain#s
which are not eter$ine afresh for each ini!iual ite$ of knowle#e)
.ut- if I $a" !enture to critici/e the author of that book- he oes not
see$ to ha!e a**reciate the ifficult" which arises throu#h the
interference of *ointer reain#s with one another when we
conte$*late such an unli$ite $ulti*licit" of *ointer reain#s) It is true
that the interference is ne#li#ible in $olar *h"sics 9to which the
iscussion in The Nature of the Physical World was li$ite:) .ut in a
funa$ental iscussion of this kin it is not le#iti$ate to se*arate
$olar *h"sics fro$ $icrosco*ic *h"sics4 for we ha!e seen 9*) FG: that
neither branch is lo#icall" co$*lete in itself)
Our efinition of the *h"sical uni!erse is that it is the worl which
*h"sical knowle#e is for$ulate to escribe) The interference of
obser!ations creates a ifficult" which $ust be $et in one of two wa"s)
Either we $ust take the co$*lete escri*tion of the *h"sical uni!erse
to e$bo" $ore than the totalit" of our *ossible knowle#e of it4 so
that- whiche!er of two interferin# obser!ations we choose to $ake-
there will be a *lace for it in the escri*tion) Or we $ust ao*t a
fle0ible uni!erse containin# nothin# which is not re*resente b" our
actual knowle#e 9or in theoretical iscussions b" the su**osel"
actual knowle#e furnishe as ata of the *roble$ concerne:) In the
first alternati!e we cannot consistentl" su**ose all the ite$s of the
co$*lete escri*tion to be re*resente b" actual *ointer reain#s4 an
it is therefore not true to sa" that its structure is a connecti!it" of
*ointer reain#s) The secon alternati!e is ao*te in wa!e
$echanics- which acce*ts as leain# features of the *h"sical uni!erse
the *robabilit" wa!es create b" actual obser!ation of the *h"sical
2uantities with which the" are associate) Clearl" there is no $ore
than a for$al istinction between the stu" of a uni!erse fle0ible
accorin# to the knowle#e we ha**en to ha!e of it an a irect stu"
of the knowle#e itself) Either alternati!e brin#s us back to the
conclusion that the co$$on back#roun is re2uire to connect one
ite$ of knowle#e with the rest of knowle#e- rather than one ele$ent
of an e0ternal uni!erse with the rest of the uni!erse)
A
6e ha!e reache a *oint at which it is esirable to take stock of our
*osition) The followin# su$$ar" will recall the *rinci*al conclusions
that we ha!e so far reache?
9%: Ph"sical knowle#e 9b" efinition: inclues onl" knowle#e
ca*able of obser!ational test4 an ite$ of *h"sical knowle#e $ust
therefore assert the result of a s*ecifie obser!ational *roceure)
95: The efinitions of the ter$s use in e0*ressin# *h"sical knowle#e
$ust be such as to secure that 9%: is satisfie) In *articular the
efinition of a *h"sical 2uantit" $ust s*ecif" una$bi#uousl" a $etho
of $easurin# it)
9': Strict aherence to 95: in!ol!es a nu$ber of $oifications of the
conce*tions an *ractice of classical *h"sics4 an inee there still
sur!i!e #larin# !iolations of it in current 2uantu$ theor") The *oints 9C:
to 9&: below arise when the efinitions are scrutini/e fro$ this *oint of
!iew)
9C: The first efinitions re2uire are those of len#th an ti$e+inter!al-
since the efinitions of other *h"sical 2uantities *resu**ose these)
The stanars of len#th an ti$e $ust be structures s*ecifie b" *ure
nu$bers onl" 9since no other 2uantitati!e ter$s are a!ailable at this
earl" sta#e:) This $eans that the stanars $ust be re*roucible fro$
a 2uantu$ s*ecification)
9<: Onl" short stanars- suitable for $easurin# infinitesi$al
is*lace$ents in s*ace an ti$e- are *ro!ie b" such s*ecifications4
an it $ust not be assu$e that the infinitesi$al is*lace$ents so
$easure are inte#rable)
9G: Owin# to the interference of e0act obser!ations with one another-
an atte$*t to efine obser!ationall" the e0act conitions uner which
the $easure$ent of a *h"sical 2uantit" is intene to be carrie out
breaks own) It is therefore necessar" to lea!e the $inor etails to
chance)
9F: In this wa" the *robabilit" conce*tion is incor*orate in the
funa$ental efinitions) It introuces an irre!ersible relation between
obser!ation an for$ulate obser!ational knowle#e) This
irre!ersibilit" $akes the e0istin# s"ste$ of *h"sics ineter$inistic-
consiere as a s"ste$ of *reiction of what can be obser!e at a
future ti$e)
9(: Certain 2uantities use in the for$ulation of *h"sical knowle#e in
classical *h"sics are foun to ha!e no efinition satisf"in# 95:) These
are unobser!ables- e)#) absolute si$ultaneit" at a istance)
9&: Other 2uantities- conitionall" obser!able- ha!e been e$*lo"e in
conitions in which the" are unobser!able) For e0a$*le- the efinition
of relati!e co+orinates *resu**oses that the *articles are
istin#uishable- but orinar" relati!e co+orinates are still use
erroneousl" in *roble$s concernin# inistin#uishable *articles)
9%D: The conclusions 9C: to 9&: are reache b" consierin# the wa" in
which *h"sical knowle#e is obtaine an for$ulate) 6e refer to
the$ as e*iste$olo#ical or a priori conclusions- to istin#uish the$
fro$ a posteriori conclusions eri!e fro$ a stu" of the results of
obser!ations which ha!e been obtaine an for$ulate in this wa")
9%%: Althou#h e*iste$olo#ical conclusions are of the nature of truis$s-
the" ha!e far+reachin# conse2uences in *h"sics) Thus the
unobser!abilit" of absolute si$ultaneit" 9(: leas to the s*ecial theor"
of relati!it"4 the non+inte#rabilit" of is*lace$ent 9<: leas to Einstein;s
theor" of #ra!itation4 the introuction of the *robabilit" conce*tion in a
funa$ental wa" 9F: leas to the $etho of wa!e $echanics)
9%5: In the $oifie theories which result- e*iste$olo#ical *rinci*les
*la" a *art which was for$erl" taken b" *h"sical h"*otheses- i)e)
#enerali/ations su##este b" an a posteriori stu" of the results of
obser!ation)
9%': Current relati!it" theor" an 2uantu$ theor"- as usuall" acce*te-
ha!e not "et taken full a!anta#e of this e*iste$olo#ical $etho) It
a**ears than when the e*iste$olo#ical scrutin" of efinitions is
s"ste$aticall" a**lie- an its conse2uences are followe u*
$athe$aticall"- we are able to eter$ine all the ,funa$ental, laws of
nature 9incluin# the *urel" nu$erical constants of nature: without an"
*h"sical h"*othesis)
9%C: This $eans that the funa$ental laws an constants of *h"sics
are wholl" sub1ecti!e- bein# the $ark of the obser!er;s sensor" an
intellectual e2ui*$ent on the knowle#e obtaine throu#h such
e2ui*$ent4 for we coul not ha!e this kin of a priori knowle#e of
laws #o!ernin# an ob1ecti!e uni!erse)
9%<: It is not su##este that the *h"sical uni!erse is wholl" sub1ecti!e)
Ph"sical knowle#e co$*rises- besies ,laws of nature,- a !ast
a$ount of s*ecial infor$ation about the *articular ob1ects surrounin#
us) This infor$ation is oubtless *artl" ob1ecti!e as well as *artl"
sub1ecti!e)
9%G: The sub1ecti!e laws are a conse2uence of the conce*tual fra$e
of thou#ht into which our obser!ational knowle#e is force b" our
$etho of for$ulatin# it- an can be isco!ere a priori b" scrutini/in#
the fra$e of thou#h as well as a posteriori b" e0a$inin# the actual
knowle#e which has been force into it)
9%F: The characteristic for$ of the funa$ental laws of *h"sics is the
sta$* of sub1ecti!it") If there are also laws of ob1ecti!e ori#in- the" $a"
be e0*ecte to be of a ifferent t"*e) It see$s *robable that where!er
effects of ob1ecti!e #o!ernance ha!e a**eare the" ha!e been
re#are as an inication that the sub1ect is ,outsie *h"sics,- e)#)
conscious !olition- or *ossibl" life)
9%(: E*iste$olo#ical laws 9if correctl" euce: are co$*ulsor"-
uni!ersal- an e0act) Since the funa$ental laws of *h"sics are
e*iste$olo#ical- the" ha!e this character + contrar" to the !iew usuall"
a!ocate in scientific *hiloso*h"- which has assu$e that the" are
$erel" e$*irical re#ularities)
The ne0t four cha*ters will be e!ote to a $ore intensi!e stu" of the
conce*tual fra$e of thou#ht referre to in 9%G:) This will show $ore
irectl" the wa" in which the sub1ecti!e ele$ent enters into *h"sical
science- an hel* to 1ustif" the na$e ,Selecti!e Sub1ecti!is$, which
we ha!e #i!en to the scientific s"ste$ of *hiloso*h")
CHAPTER AII
8ISCOAERY OR 7AN>FACT>RE@
I
About 5FD "ears a#o a historical e0*eri$ent was *erfor$e in this
colle#e 9Trinit" Colle#e: which was consiere to e$onstrate the
co$*osite nature of white li#ht) The account of the isco!er"- in the
leain# te0t+book of $" uner#rauate a"s- is?
It was still su**ose that e!er" refraction of li#ht actuall" *rouce color- instea
of $erel" se*aratin# the colors alrea" e0istin# in orinar" white li#ht- but in %GGG
Newton $ae the i$*ortant isco!er" of the actual e0istence of colors of all kins
in solar li#ht- which he showe to be no other than a co$*oun of the !arious
colors- $i0e in certain *ro*ortions with each other an ca*able of bein#
se*arate b" refraction of an" kin) 9Preston- Theor" of Li#ht- 5n) e)- *) &):
It see$s a si$*le $atter to e$onstrate that the white li#ht of the sun
is reall" a $i0ture of li#ht of !arious colors) .ut su**ose that- instea
of *ro!in# it to a ocile stuent- we ha!e to *ro!e it to a s*iritualist who
watches e!er"thin# we o with the sa$e sus*icion that we shoul feel
it our ut" to ao*t in in!esti#atin# his clai$s) 6e be#in b" takin# a
s*ectrosco*e + a *ris$ s*ectrosco*e woul be $ore closel"
re$iniscent of Newton- but we ha**en to ha!e #ot hol of a #ratin#
s*ectrosco*e an it is not worth while to chan#e it) 6e let a ra" of
sunli#ht fall on one en of the instru$ent- an in!ite the s*iritualist to
a**l" his e"e to the other) He is astonishe to see a brilliant #reen li#ht
which- we tell hi$- the s*ectrosco*e has se*arate out fro$ the other
colors *resent in white sunli#ht) Sus*ectin# tricker"- he e0a$ines each
*art of the instru$ent) He *ounces on a #a#et scratche with
thousans of fine *arallel lines) Triu$*hantl" he reali/es how it works)
The li#ht falls obli2uel" so that the *arallel lines reflect it- not
si$ultaneousl"- but one after the other) A sin#le incient *ulse is thus
$ulti*lie b" reflection into thousans of *ulses followin# at re#ular
inter!als) E!ientl" this has been arran#e to *rouce the *articular
*erioicit" which our e"es reco#ni/e as #reen color) The clai$ that the
#reen color 9i)e) #reen *erioicit": alrea" e0iste in the sunli#ht was
false4 we ha hien in the instru$ent a e!ice- which we ho*e he
woul not isco!er- for introucin# the #reen *erioicit") The s*iritualist
#oes awa" confient that he has e0*ose a clu$s" frau)
." usin# a #ratin#- instea of a *ris$ which acts $ore $"steriousl"-
we #a!e the show awa") As $entione in the abo!e 2uotation- it was
the *re!ailin# !iew before Newton that the *ris$ actuall" *rouces the
color4 so that the essential *art of Newton;s e$onstration was a
series of e0*eri$ents belie!e to *ro!e that the *ris$ oes not
*rouce the color but se*arates it) These we were #oin# to show the
s*iritualist4 but it is no #oo showin# the$ now) These further
e0*eri$ents work as well with a #ratin# as with a *ris$4 an whate!er
the" *ro!e for the *ris$ the" *ro!e for the #ratin#) It is useless to
a**eal to the$ as su**ortin# a conclusion which- in the case of the
#ratin#- we see to be untrue)
I think it not unlikel" that e!en an e0*ert $i#ht fall into this tra* toa" +
such is the #la$our of a historic e0*eri$ent) He reall" knows better4
but one oes not alwa"s recall one;s knowle#e when it is wante) The
*osition was $ae *lain b" Ra"lei#h an Schuster- an is inee *art
of the usual o*tical teachin#) 6hite li#ht- such as sunli#ht- is a 2uite
irre#ular isturbance with no tenenc" to *erioicit") .ut
$athe$aticall" we can anal"se an" isturbance- howe!er irre#ular-
into the su$ of *erioic Fourier co$*onents4 an we can- if we like-
think of the isturbance as $ae u* of these co$*onents) 6hether
the s*ectrosco*e sorts out a *articular *erioicit" or impresses it- is
1ust a $atter of e0*ression) The iea of ,sortin#, is a**ro*riate-
because the s*ectrosco*e woul fail to i$*ress the *articular
*erioicit" on li#ht in which the corres*onin# Fourier co$*onent
ha**ene to be $issin#4 an in fact the solar s*ectru$ shows ark
lines where the white li#ht fails to take the i$*ress of the
corres*onin# *erioicit"- owin# to certain Fourier co$*onents ha!in#
been sifte out of the li#ht before it reaches us) .ut the iea of
,i$*ressin#, the *erioicit" is also a**ro*riate4 for we shoul not
e0*ect an i$*ress to take on unsuitable $aterial- an the Fourier
anal"sis $a" be re#are as the $athe$atician;s *reli$inar" test to
see if the $aterial will bear the i$*ress) It is *articularl" a**ro*riate
when a #ratin# is use- since ,i$*ressin# the *erioicit", is then a
*lain ele$entar" state$ent of the modus operandi)
The $istake was not in sa"in# that a #reen co$*onent alrea" e0ists
in the sunli#ht- for that is at an" rate a le#iti$ate wa" of thinkin#- but in
clai$in# that we coul ecie e0*eri$entall" between two e2uall"
*er$issible for$s of escri*tion) An- b" our o!ersi#ht- it ha**ene
that the for$ of escri*tion we cone$ne was rather $ore natural
an a**ro*riate than the one we unertook to efen)
The reali/ation that natural white li#ht is a 2uite irre#ular isturbance-
into which re#ularit" is introuce b" our $etho of s*ectrosco*ic
e0a$ination of it- was the first si#n of an uneasiness a$on# *h"sicists
as to whether in our e0*eri$ents we $a" not interfere so $uch as to
estro" what we were seekin# to in!esti#ate) The uneasiness has
beco$e $ore acute in $oern ato$ic *h"sics- since we ha!e no tool
fine enou#h to *robe an ato$ without #rossl" isturbin# it)
The 2uestion I a$ #oin# to raise is + how $uch o we isco!er an
how $uch o we $anufacture b" our e0*eri$ents@ 6hen the late
Lor Rutherfor showe us the ato$ic nucleus- i he find it or i he
make it@ It will not affect our a$iration of his achie!e$ent either wa"
+ onl" we shoul rather like to know which he i) The 2uestion is one
that scarcel" a$its of a efinite answer) It turns on a $atter of
e0*ression- like the 2uestion whether the s*ectrosco*e fins or
whether it $akes the #reen color which it shows us) .ut since $ost
*eo*le are *robabl" uner the i$*ression that Rutherfor foun the
ato$ic nucleus- I will $ake $"self a!ocate for the !iew that he $ae
it)
II
The tenenc" of writers on 2uantu$ theor" has been *erha*s to #o
farther that I o in e$*hasi/in# the *h"sical interference of our
e0*eri$ents with the ob1ects which we stu") It is sai that the
e0*eri$ent *uts the ato$s or the raiation into the state whose
characteristics we $easure) I shall call this Procrustean treat$ent)
Procrustes- "ou will re$e$ber- stretche or cho**e own his #uests
to fit the be he ha constructe) .ut *erha*s "ou ha!e not hear the
rest of the stor") He $easure the$ u* before the" left ne0t $ornin#-
an wrote a learne *a*er ,On the >nifor$it" of Stature of Tra!ellers,
for the Anthro*olo#ical Societ" of Attica)
The *h"sical !iolence- howe!er- is not reall" the essential *oint) Ieall"
the e0*eri$enter $i#ht wait until the conitions of his e0*eri$ent
ha**ene naturall"- as those en#a#e in the obser!ational sciences
are force to o) 6e #rossl" interfere with the irre#ularit" of white
sunli#ht b" *assin# it throu#h a s*ectrosco*e4 but sunli#ht $a"
occasionall" fall throu#h a cre!ice on to a natural cr"stal an for$ a
s*ectru$ without our hel*) The stanar conitions- which turn
ai$less $easure$ent into a #oo $easure$ent of a efinite *h"sical
2uantit" useful for scientific inuction- $a" so$eti$es occur without
hu$an interference) .ut- so far as *h"sical theor" is concerne- it
$akes no ifference whether we create or whether we select the
conitions which we stu") 6hether the interference of the obser!er is
*h"sical or selecti!e- it is none the less $arke in the resultin#
conclusions) The kin of obser!ation on which *h"sical theor" is base
is not a casual takin# notice of thin#s aroun us- nor a #eneral runnin#
roun with a $easurin# ro) >ner co!er of the ter$ ,#oo,
obser!ation the be of Procrustes is artfull" conceale)
To what len#th can this interference be carrie@ I o not think that an"
li$it can be set a priori) It is *ertinent to re$e$ber that the conce*t of
substance has isa**eare fro$ funa$ental *h"sics4 what we
ulti$atel" co$e own to is form) 6a!esH 6a!esHH 6a!esHHH Or for a
chan#e + if we turn to relati!it" theor" + cur!atureH Ener#" which- since
it is conser!e- $i#ht be looke u*on as the $oern successor of
substance- is in relati!it" theor" a cur!ature of s*ace+ti$e- an in
2uantu$ theor" a *erioicit" of wa!es) I o not su##est that either the
cur!ature or the wa!es are to be taken in a literal ob1ecti!e sense4 but
the two #reat theories- in their efforts to reuce what is known about
ener#" to a co$*rehensible *icture- both fin what the" re2uire in a
conce*tion of ,for$,)
Substance 9if it ha been *ossible to retain it as a *h"sical conce*tion:
$i#ht ha!e offere so$e resistance to the obser!er;s interference4 but
for$ *la"s into his hans) Su**ose an artist *uts forwar the fantastic
theor" that the for$ of a hu$an hea e0ists in a rou#h+sha*e block of
$arble) All our rational instinct is rouse a#ainst such an
anthro*o$or*hic s*eculation) It is inconcei!able that Nature shoul
ha!e *lace such a for$ insie the block) .ut the artist *rocees to
!erif" his theor" e0*eri$entall" + with 2uite rui$entar" a**aratus too)
7erel" usin# a chisel to se*arate the for$ for our ins*ection- he
triu$*hantl" *ro!es his theor") 6as it in this wa" that Rutherfor
renere concrete the nucleus which his scientific i$a#ination ha
create@
8o not be $isle b" thinkin# of the nucleus as a sort of billiar ball)
Think of it rather as a s"ste$ of wa!es) It is true that the ter$
,nucleus, is not strictl" a**licable to the wa!es 9cf) the electron- *) <%:?
but it is e2uall" unri#orous to s*eak of the nucleus as ha!in# been
,isco!ere,) The isco!er" es not #o be"on the wa!es which
re*resent the knowle#e we ha!e of the nucleus)
8oes the scul*tor;s *roceure iffer in an" essential wa" fro$ that of
the *h"sicist@ The latter has a conce*tion of a har$onic wa!e for$
which he sees in the $ost unlikel" *laces + in irre#ular white li#ht- for
e0a$*le) 6ith a #ratin# instea of a chisel- he se*arates it fro$ the
rest of the white li#ht an *resents it for our ins*ection) Bust as the
scul*tor se*arates the rou#h block of $arble into a bust an a hea* of
chi*s- so the *h"sicist se*arates the irre#ular wa!e isturbance into a
si$*le har$onic #reen wa!e an a scra*+hea* of other co$*onents)
In Fourier an other reco#ni/e $ethos of anal"sis- *h"sics allows
an *ractices the s*littin# of for$ into co$*onents) It allows us to
select a for$ which we oursel!es ha!e *rescribe- an treat the rest
as conta$ination which we can re$o!e- if we can e!ise the
necessar" a**aratus- so as to e0hibit our selecte for$ b" itself) In
e!er" *h"sical laborator" we see in#eniousl" e!ise tools for
e0ecutin# the work of scul*ture- accorin# to the esi#ns of the
theoretical *h"sicist) So$eti$es the tool sli*s an car!es off an o+
sha*e for$ which we ha not e0*ecte) Then we ha!e a new
e0*eri$ental isco!er")
It is ifficult to see where- if at all- a line can be rawn) The 2uestion
oes not $erel" concern li#ht wa!es- since in $oern *h"sics for$-
*articularl" wa!e for$- is at the root of e!er"thin#) If no line can be
rawn- we ha!e the alar$in# thou#ht that the *h"sical anal"st is an
artist in is#uise- wea!in# his i$a#ination into e!er"thin# + an
unfortunatel" not wholl" e!oi of the technical skill to reali/e his
i$a#ination in concrete for$)
An illustration $a" show that a serious *ractical 2uestion is raise)
Bust now nuclear *h"sicists are writin# a #reat eal about h"*othetical
*articles calle neutrinos su**ose to account for certain *eculiar
facts obser!e in +ra" isinte#ration) 6e can *erha*s best escribe
the neutrinos as little bits of s*in+ener#" that ha!e #ot etache) I a$
not $uch i$*resse b" the neutrino theor") In an orinar" wa" I $i#ht
sa" that I o not belie!e in neutrinos
%5
) .ut I ha!e to reflect that a
*h"sicist $a" be an artist- an "ou ne!er know where "ou are with
artists) 7" ol+fashione kin of isbelief in neutrinos is scarcel"
enou#h) 8are I sa" that e0*eri$ental *h"sicists will not ha!e sufficient
in#enuit" to make neutrinos@ 6hate!er I $a" think- I a$ not #oin# to
be lure into a wa#er a#ainst the skill of e0*eri$enters uner the
i$*ression that it is a wa#er a#ainst the truth of a theor") If the"
succee in $akin# neutrinos- *erha*s e!en in e!elo*in# inustrial
a**lications of the$- I su**ose I shall ha!e to belie!e + thou#h I $a"
12
8oubtless until a truer unerstanin# of the s*in *roble$ is reache- it is better to $ake
shift with neutrinos than to i#nore the ifficult" which the" are intene to $eet) I ha!e no
ob1ection to neutrinos as a te$*orar" e0*eient- but I woul not e0*ect the$ to sur!i!e +
e0ce*t that- as su##este in this *ara#ra*h- sur!i!al $a" not be wholl" a 2uestion of
intrinsic $erit)
feel that the" ha!e not been *la"in# 2uite fair)
The 2uestion is raise whether the e0*eri$enter reall" *ro!ies such
an effecti!e control on the i$a#ination of the theorist as is usuall"
su**ose) Certainl" he is an incorru*tible watch+o# who will not allow
an"thin# to *ass which is not obser!ationall" true) .ut there are two
wa"s of oin# that + as Procrustes reali/e) One is to e0*ose the
falsit" of an assertion) The other is to alter thin#s a bit so as to $ake
the assertion true) An it is a$itte that our e0*eri$ents do alter
thin#s)
I ha!e been actin# as a!ocate for an e0tre$e !iew- *resu$in# that
"our natural *re1uices are all the other wa") I $ust now tr" to reco!er
the *oise of a 1u#e) I o not think that as yet the anal"tical
i$a#ination of the $athe$atical *h"sicist has e!elo*e into the
unfettere i$a#ination of the artist) He *la"s the #a$e accorin# to
certain rules which- arbitrar" as the" $a" see$ at first si#ht- e0*ress
an e*iste$olo#ical *rinci*le that #oes ee* into the roots of hu$an
thou#ht) This we shall iscuss *resentl") .ut ha!e we a #uarantee that
the rules are for all ti$e@ The bo" who outra#eousl" breaks the rules
of a #a$e $a" be suitabl" *unishe b" his co$*anions- or he $a" be
co$$e$orate as the founer of Ru#b" football) The $an who $akes
neutrinos will not be *unishe if he has o!erste**e the rules4 he will
be acclai$e for freein# *h"sics fro$ an obstruction to its useful
e!elo*$ent)
Howe!er- our concern is with the characteristics of *resent+a"
*h"sics an not with what it $a" beco$e in future) 6e shall now enter
on a !er" e0tensi!e sub1ect of iscussion- na$el" the nature an ori#in
of the rules which istin#uish the $ethos of the *h"sicist fro$ the
free i$a#ination of the artist)
CHAPTER AIII
THE CONCEPT OF ANALYSIS
I
In introucin# sub1ecti!e selection 9*) %G:- I attribute it to ,the sensor"
an intellectual e2ui*$ent, use in obtainin# obser!ational knowle#e)
The inclusion of intellectual e2ui*$ent $a" ha!e see$e sur*risin#) It
is eas" to see that our sensor" e2ui*$ent has a selecti!e effect + that
the nature an e0tent of our knowle#e of an e0ternal worl $ust be
lar#el" conitione b" its lines of co$$unication with consciousness-
*ro!ie b" our sense or#ans) It is not so ob!ious that within the $in
there is an" further selection at work on the $aterial thrust u*on it b"
the sense or#ans)
It $ust- I think- be a#ree that all that co$es into consciousness as
the result of sti$ulation of the sense or#ans is knowle#e of a kin) It
is not *ossible to *ercei!e without knowin# that we *ercei!e4 an
*erce*tion i$*lies ,knowle#e of our *erce*tions as such,) .ut we are
here concerne with *h"sical knowle#e- which is an abbre!iation for
knowle#e ac2uire b" the $ethos of *h"sical science 9*) 5:)
Intros*ecti!e e0a$ination of our *erce*tions as such is no *art of the
$etho of *h"sical science) An intellectual acti!it" be#ins when we
relate our *erce*tions to one another) The result of this acti!it" is a
s"nthesis of *erce*tion- an a for$ulation of knowle#e of a ifferent
t"*e fro$ knowle#e of ini!iual *erce*tions as such) =nowle#e of
the relateness of sensor" *erce*tions- e)#) the soun of thuner
followin# the flash of li#htnin#- is the be#innin# of science) The
rui$ents of the $etho of *h"sical science are- of course- e$*lo"e
lon# before an" s"ste$atic stu" of what is #oin# on aroun us is
atte$*te4 an e!en the $ost unso*histicate a**rehension of
*heno$ena in!ol!es co$$onsense as well as sense + that is to sa"-
an intellectual as well as a *urel" sensor" acti!it" of consciousness)
6e ha!e alrea" *ai attention to this intellectual s*eciali/ation of
*h"sical knowle#e- in *ointin# out that the ac2uire$ent of
obser!ational knowle#e $eans so$ethin# $ore than obser!ation in
the ele$entar" sense of takin# notice) 6e ha!e seen that- for refine
scientific e!elo*$ents- what is re2uire is #oo 9thou#h not *erfect:
obser!ation of efine 2uantities) There is a bi# #a* between this an
the *assi!e rece*tion of sensor" i$*ressions4 an in this #a* the
selecti!e influences of our intellectual e2ui*$ent ha!e their
o**ortunit") If we consier the se2uence- ob1ecti!e e!ent + *erce*tion +
*h"sical knowle#e- there is a ouble siftin#- firstl" b" our sensor"
e2ui*$ent- seconl" b" our intellectual e2ui*$ent) In the *resent
e*iste$olo#ical treat$ent we start fro$ knowle#e- so that the orer is
re!erse- an it is the intellectual siftin# that first co$es uner
consieration)
In anal"sin# this intellectual acti!it"- I shall $ake use of the *hrase
,for$ of thou#ht,4 or- when the for$ is in so$e e#ree elaborate-
,fra$e of thou#ht,) This $a" be re#are as a *reeter$ine for$ or
fra$e into which the knowle#e we ac2uire obser!ationall" is fitte)
For e0a$*le- a !er" ee*+roote for$ of thou#ht is that which
for$ulates the knowle#e ac2uire b" obser!ation as a escri*tion of
a worl) E!er" ite$ of *h"sical knowle#e is fitte into this for$ of
thou#ht- an ee$e to be a escri*ti!e fact about a uni!erse) So
*re!alent is this for$ that knowle#e which is not concerne with the
relateness of sensor" *erce*tions is often force into it- an treate
as a escri*ti!e fact about a non+$aterial worl + a s*iritual worl) I
on not think the reasons for or a#ainst e$*lo"in# this for$ of thou#h
are stron#er in the one case than in the other) 6hate!er we ha!e to
a**rehen $ust be a**rehene in a wa" for which our intellectual
e2ui*$ent has $ae *ro!ision)
The e*iste$olo#ical $etho of in!esti#ation leas us to stu" the
nature of the fra$e of thou#ht- an so be forewarne of its i$*ress on
the knowle#e that will be force into it) 6e $a" foresee a priori
certain characteristics which an" knowle#e containe in the fra$e will
ha!e- si$*l" because it is containe in the fra$e )These
characteristics will be isco!ere a posteriori b" *h"sicists who e$*lo"
that fra$e of thou#ht- when the" co$e to e0a$ine the knowle#e the"
ha!e force into it) Procrustes a#ainH
These foreseeable characteristics are not b" an" $eans tri!ial4 the"
are laws or nu$erical constants which *h"sicists ha!e been at #reat
*ains to eter$ine b" obser!ation an e0*eri$ent) As an e0a$*le we
$a" take the law of increase of $ass with !elocit"- which has been the
sub1ect of $an" fa$ous e0*eri$ents) It is now reali/e that this law
auto$aticall" results fro$ the en#raine for$ of thou#ht which
se*arates the four+fol orer of e!ents into a three+fol orer of s*ace
an an orer of ti$e) 6hen knowle#e is for$ulate in a fra$e which
co$*els us to se*arate a ti$e i$ension fro$ the four+fol orer to
which it belon#s- a co$*onent calle the $ass is corres*onin#l"
se*arate fro$ the four+fol !ector to which it belon#s4 an it re2uires
no !er" *rofoun stu" of the conitions of se*aration to see how the
se*arate co$*onent is relate to the rest of the !ector which
*rescribes the !elocit") It is this relation which is reisco!ere when we
eter$ine e0*eri$entall" the chan#e of $ass with !elocit")
In one sense the outlook of relati!it" theor" has e$anci*ate us fro$
the fra$e of thou#ht which se*arates a ti$e i$ension fro$ the rest of
the four+fol orer4 an the law of !ariation of $ass with !elocit"
shoul ha!e isa**eare fro$ *h"sics- since it refers to conce*tions
associate with a iscare fra$e) The fact which corres*ons to it in
the new fra$e of thou#ht is an ob!ious truis$ which oes not re2uire
se*arate $ention) .ut a #lance at the literature of $oern *h"sics
shows that the law has not isa**eare4 an its i$*ortance is 1ust the
sa$e as when the hi#hest skill of the e0*eri$enter was bein# a**lie
to eter$ine it e$*iricall") The real *osition is that b" the outlook of
relati!it" we ,see throu#h, the for$ of thou#ht- but we o not actuall"
iscar it + e0ce*t te$*oraril" in s*eciali/e researches where its
istorte !iew woul be a hinrance) The law of chan#e of $ass with
!elocit" therefore retains its *lace as a scientific conclusion4 an it is
b" no $eans a tri!ial conclusion) The test of tri!ialit" $ust be whether
the result a**eare tri!ial before we unerstoo its real nature) E!en
the *rouction of a rabbit fro$ a hat is a tri!ial *heno$enon if "ou
know how it is one)
It is not *ossible to s*ecif" a *articular fra$e of thou#ht as the fra$e
*re!ailin# consistentl" throu#hout *resent+a" *h"sics) 6e $ust in
an" case istin#uish between the fra$e of thou#ht which corres*ons
to the out*osts of $oern theor" an the fra$e which furnishes $ost
of our current !ocabular") The latter is $ore or less the sa$e as that
which corres*ons to fa$iliar a**rehension of thin#s aroun us) .ut
e!en fa$iliar a**rehension oes not ahere consistentl" to an" one
fra$e of thou#ht) For e0a$*le- lookin# own fro$ the to* of a sk"+
scra*er- we see a nu$ber of tin" ob1ects walkin# about in the street
below) The inference that the" are ob1ects of nor$al hu$an stature-
$ae to a**ear s$all b" istance- is not a $atter of i$$eiate
a**rehension4 it is a consiere inter*retation of what we a**rehen)
.ut for ob1ects near to us- the scientific fra$e has beco$e the fa$iliar
fra$e) 6hen a $an walks awa" fro$ us in a roo$- we o not ,see,
hi$ #ettin# s$aller) 6e see- or think we see- an ob1ect of constant
si/e chan#in# its istance fro$ us4 an it is onl" b" an effort of
intros*ection that we con!ince oursel!es that the !isual i$a#e is
beco$in# s$aller)
Since it is *art of the art of *h"sics to e$*lo" the co$$on for$s of
thou#ht as ser!ants when the" cease to be our $asters- it is scarcel"
accurate to s*eak of the$ as ha!in# been iscare) It is better to sa"
that the a!ance of *h"sics has e$anci*ate us fro$ so$e of the
co$$on for$s of thou#ht) 6e use the$- but we are not ecei!e b"
the$)
II
The $oern theories of *h"sics ha!e e$anci*ate us fro$ certain
traitional for$s of thou#ht) That is wh" the" see$ so re!olutionar") Is
this the en of the a!ance- or o there re$ain in our outlook other
for$s of thou#ht obstructin# *ro#ress fro$ which future *h"sicists will
succee in freein# the$sel!es@ An if so- can the e$anci*ation be
continue inefinitel"- or is it a**roachin# a li$it in which the sur!i!in#
for$s will be the bare necessities of thou#ht@
6e shall e0a$ine so$e of the for$s of thou#ht which still re$ain
unchallen#e in the scientific outlook) 6e are sus*icious of the *hrase
,necessities of thou#ht,4 for scientific thou#ht has #rown accusto$e
to oin# without $an" of its alle#e necessities) .ut- whether
necessities or not- the for$s we are about to iscuss ha!e a hol on
us which see$s inco$*arabl" stron#er than an" we ha!e hitherto
thrown off)
For a scientific outlook I think the $ost funa$ental of all for$s of
thou#ht is the concept of analysis) This $eans the conce*tion of a
whole as i!isible into *arts- such that the co+e0istence of the *arts
constitutes the e0istence of the whole) In a for$al efinition I shoul
not use the ter$ ,e0istence,- since it refers to a conce*t which is
*robabl" less ele$entar" than the conce*t of anal"sis) .ut a for$al
efinition is not re2uire in referrin# to a for$ of "our own thou#hts) 7"
escri*tion is sufficient for "ou to reco#ni/e the for$ I $ean- an that
is all that is neee)
The *oint which I $ust e$*hasi/e is that I a$ referrin# to the
conce*tion of a set of parts- not to the ini!iual conce*tion of a part)
In the conce*t of anal"sis a *art is alwa"s a $e$ber of a co$*lete set
of *arts- an its si#nificance is boun u* with the s"ste$ of anal"sis in
which it occurs) 6e coul- if necessar"- e0*ress this relation of a *art
to a s"ste$ of anal"sis b" usin# the ter$ ,co$*onent,4 but that
*erha*s sa!ors too $uch of $athe$atical ter$inolo#" for the *resent
!er" #eneral a**lication)
At first si#ht $" insistence that a *art $ust alwa"s be associate with
anal"sis into a co$*lete set of *arts see$s an ile for$alit") 6e can
reco#ni/e that the hea is *art of the bo" without referrin# to a
s"ste$atic anato$ical classification of the *arts of the bo") To fulfil
the for$al re2uire$ent of the conce*t of anal"sis- we can sa" that the
hea is a *art of the bo" associate with a s"ste$ of anal"sis which
i!ies the bo" into two *arts- na$el" the hea an the rest of the
bo") .ut since that a**lies to an" arbitraril" selecte *art of an"thin#-
the reference to a s"ste$ of anal"sis beco$es a tautolo#")
To e0*lain wh" we ha!e to start fro$ the notion of a co$*lete set of
*arts rather than fro$ the a**arentl" si$*ler notion of a sin#le *art- I
$ust ask a 2uestion) Is the bun#+hole of a barrel *art of the barrel@
Think well before "ou answer4 because the whole structure of
theoretical *h"sics is tre$blin# in the balance)
Su**ose that the answer is Yes) Then the actual barrel is to be
re#are as consistin# of an un*erforate barrel to#ether with a hole +
a close wooen structure to#ether with a $inus 2uantit" of woo) It is
not a 2uestion whether this e0*resses an absolute truth) The 2uestion
is whether it is a for$ of thou#ht which we shall *er$it oursel!es to
e$*lo") In this for$ of thou#ht one of the *arts- na$el" the
un*erforate barrel- is $ore than the whole) Eucli in his innocence
thou#ht that ,the whole is #reater than the *art,4 but Eucli was not
ac2uainte with $oern *h"sics)
Our answer has $ae the ter$ ,*art, $eanin#less in itself) 6hate!er
A an . $a" stan for- A is alwa"s *art of .4 since our for$ of thou#ht
a$its the i!ision of . into two *arts- na$el" A an .+A) The ter$
,*art, can therefore onl" be usefull" e$*lo"e for the *arts which are
associate with a s"ste$ of anal"sis- an the whole si#nificance of a
*art is associate with the s"ste$ of anal"sis in which it a**ears) It
con!e"s nothin# to sa" that A is a *art of .4 but it con!e"s so$ethin#
to sa" that A is one of the *arts a**earin# in a s*ecifie s"ste$ of
anal"sis a**lie to .)
Ne0t su**ose that we ahere to Eucli;s a0io$ an ecie that the
bun#+hole of a barrel is not *art of the barrel) The ob1ection to this is
that it has lon# cease to be the for$ of thou#ht e$*lo"e in *h"sics)
It is- I think- reall" a co$*oun association of two conce*ts- the
conce*t of anal"sis an the conce*t of substance) The conce*t of
substance introuces a clear istinction of *ositi!e an ne#ati!e4 so
that we can ha!e a li$ite for$ of the conce*t of anal"sis- which we
$a" call substance+anal"sis- in which the s"ste$s of anal"sis are
restricte to those which furnish a co$*lete set of positive *arts) 6hen
the anal"sis is not associate with substance 9or with a structurall"
e2ui!alent conce*t:- when for e0a$*le it is associate with wa!e for$-
the restriction cannot be i$*ose) In o*tics arkness is consiere to
be constitute of two interferin# li#ht wa!es4 li#ht $a" be a ,*art, of
arkness) In Fourier anal"sis the co$*onents *artiall" cancel one
another in the $anner of *ositi!e an ne#ati!e 2uantities) Thus-
althou#h there $a" be cases in *h"sics in which anal"sis is a**lie to
entities which b" efinition are essentiall" *ositi!e an the restriction of
substance+anal"sis a**lies- we now look on it as an inciental
restriction in a *articular a**lication an not as *art of the funa$ental
conce*t of anal"sis)
That the #eneral for$ of the conce*t of anal"sis is the for$ acce*te
in *h"sical science is shown conclusi!el" b" the e0a$*le of the
*ositron) A *ositron is a hole fro$ which an electron has been
re$o!e4 it is a bun#+hole which woul be e!ene u* with its
surrounin#s if an electron were inserte) .ut it woul be out of the
2uestion nowaa"s to efine ,*art, in such a wa" that electrons are
*arts of a *h"sical s"ste$ but *ositrons are not)
You will see that the *h"sicist allows hi$self e!en #reater libert" than
the scul*tor 9*) %%%:) The scul*tor re$o!es $aterial to obtain the for$
he esires) The *h"sicist #oes further an as $aterial if necessar" +
an o*eration which he escribes as re$o!in# ne#ati!e $aterial) He
fills u* a bun#+hole- sa"in# that he is re$o!in# a *ositron) .ut he still
clai$s that he is onl" re!ealin# + sortin# out + so$ethin# that was
alrea" there)
Once a#ain I woul re$in "ou that ob1ecti!e truth is not the *oint at
issue) 6e $ust not $ake the $istake- illustrate at the be#innin# of
Cha*ter AII- of tr"in# to ecie b" crucial e0*eri$ental test between
what are onl" two ifferent for$s of e0*ression) I are sa" "ou ha!e a
batter" of ar#u$ents to *ro!e to $e b" irrefutable lo#ic that the bun#+
hole is not *art of the barrel) .ut this is 2uite irrele!ant4 it will onl" show
that "ou o not use 9e0ce*t b" ina!ertence: the ter$ ,*art, with 2uite
as $uch #eneralit" of $eanin# as the *h"sicist oes)
Our *ur*ose is to e0*ose- not necessaril" to 1ustif"- the fra$e of
thou#ht unerl"in# the e0*ression of our *h"sical knowle#e) Partiall"
at least we e$anci*ate oursel!es fro$ a fra$e of thou#ht as soon as
we reali/e that it is onl" a fra$e of thou#ht an not an ob1ecti!e truth
we are acce*tin#) An" *ower for $ischief it $a" ha!e is sterili/e so
lon# as it is ke*t e0*ose) I woul not like to sa" that the conce*t of
anal"sis is a necessit" of thou#ht- thou#h it a**ears to be a necessit"
of an" for$ of scientific thou#ht) .ut- whether it is a necessar" for$ or
not- it has o$inate the e!elo*$ent of *resent+a" *h"sics- an we
ha!e to follow u* its influence on the sche$e of escri*tion of
*heno$ena which has resulte)
III
It is clear that the conce*t of anal"sis as a**lie in *h"sics $ust ha!e
been s*eciali/e accorin# to so$e #uiin# *rinci*le4 otherwise there
woul not be the sa$e #eneral a#ree$ent as to the *roucts of
anal"sis of the *h"sical worl- na$el" $olecules- ato$s- *rotons-
electrons- *hotons- etc)) There is another en#raine for$ of thou#ht
which has selecte the s"ste$ of anal"sis to be a**lie in *h"sics) I
will call this s*eciali/ation of the conce*t of anal"sis the atomic
concept- or for #reater *recision the concept of identical structural
units)
The new conce*tion is- not $erel" that the whole is anal"sable into a
co$*lete set of *arts- but that it is anal"sable into *arts which
rese$ble one another) It is at the o**osite *ole fro$ the anal"sis- sa"-
of a hu$an bein# into soul an bo"- in which the two *arts belon# to
alto#ether ifferent cate#ories of entities) I will #o farther- an sa" that
the ai$ of the anal"sis e$*lo"e in *h"sics is to resol!e the uni!erse
into structural units which are precisely like one another)
it $a" be ob1ecte that the structural units reco#ni/e in *resent+a"
*h"sics- thou#h rese$blin# one another to a certain e0tent- are not
*recisel" alike) The Fourier co$*onents of white li#ht- thou#h all
si$*le har$onic trains of wa!es- iffer in wa!elen#th + a ifference
which we obser!e as ifference of color) .ut this ifference is not
intrinsic) It e*ens on the relation of the obser!er to the structural
unit4 if he recees fro$ the source of li#ht- #reen li#ht turns to re)
Intrinsicall" the constituents of li#ht + the wa!e trains or the *hotons +
are all *recisel" alike4 it is onl" in their relations to the obser!er- or to
e0ternal ob1ects #enerall"- that the" iffer) That is the essence of the
relati!it" theor") All the !ariet" in the worl- all that is obser!able-
co$es fro$ the !ariet" of relations between entities) Therefore- when
we reach the consieration of the intrinsic nature or structure of the
entities that are relate- there is nothin# left but sa$eness + in so far
as that nature or structure co$es within the sco*e of *h"sical
knowle#e an is *art of the uni!erse which *h"sical knowle#e
escribes)
3rantin# that the ele$entar" units foun in our anal"sis of the uni!erse
are *recisel" alike intrinsicall"- the 2uestion re$ains whether this is
because we ha!e to o with an ob1ecti!e uni!erse built of such units-
or whether it is because our for$ of thou#ht is such as to reco#ni/e
onl" s"ste$s of anal"sis which shall "iel *arts *recisel" like one
another) Our *re!ious iscussion has co$$itte us to the latter as the
true e0*lanation) 6e ha!e clai$e to be able to eter$ine b" a priori
reasonin# the *ro*erties of the ele$entar" *articles reco#ni/e in
*h"sics + *ro*erties confir$e b" obser!ation) This woul be
i$*ossible if the" were ob1ecti!e units) Accorin#l" we account for this
a priori knowle#e as *urel" sub1ecti!e- re!ealin# onl" the i$*ress of
the e2ui*$ent throu#h which we obtain knowle#e of the uni!erse an
eucible fro$ a stu" of the e2ui*$ent) 6e now sa" $ore e0*licitl"
that it is the i$*ress of our fra$e of thou#ht on the knowle#e force
into the fra$e)
6e ha!e 1ust seen that the conce*t of ientical structural units is
i$*licit in the relati!it" outlook- which attributes !ariet" to relations an
not to intrinsic ifferences in the relata4 but I su**ose it woul be too
$uch to clai$ that the relati!it" outlook is en#raine in us + that our
$ins are so constitute that we cannot hel* $oulin# our thou#hts in
the Einsteinian wa") I want to show therefore that the conce*t of
ientical structural units e0*resses a !er" ele$entar" an instincti!e
habit of thou#ht- which has unconsciousl" irecte the course of
scientific e!elo*$ent) .riefl"- it is the habit of thou#ht which re#ars
!ariet" alwa"s as a challen#e to further anal"sis4 so that the ultimate
en+*rouct of anal"sis can onl" be sa$eness) 6e kee* on $oif"in#
our s"ste$ of anal"sis until it is such as to "iel the sa$eness which
we insist on- re1ectin# earlier atte$*ts 9earlier *h"sical theories: as
insufficientl" *rofoun) The sa$eness of the ulti$ate entities of the
*h"sical uni!erse is a foreseeable conse2uence of forcin# our
knowle#e into this for$ of thou#ht) That it is reall" en#raine in us
can be seen fro$ the followin# e0a$*le)
Anal"sis of $atter- as usuall" *resente in *resent+a" theor"- reaches
a consierable e#ree of ho$o#eneit" of the ulti$ate *arts- but oes
not 2uite attain the ieal) 6e fin *rotons e0actl" like one another4 we
also fin electrons- like one another but ifferin# fro$ *rotons) Thus
the *h"sicist reco#ni/e two !arieties of ele$entar" units4 an
nowaa"s it is ifficult to restrain hi$ fro$ ain# se!eral others) 6h"
oes a *roton iffer fro$ an electron@ The answer su##este b"
relati!it" theor" is that th" are actuall" si$ilar units of structure- an
the ifference arises in their relations to the #eneral istribution of
$atter which for$s the uni!erse) The one is relate ri#ht+hanel" an
the other left+hanel") This accounts for the ifference of char#e4 an
the ifference of $ass is also 9in a $ore co$*licate wa": a ifference
of relation to the e0ternal $atter without which there woul be no
$eans of eter$inin# $ass obser!ationall") There is no reasonable
oubt that this answer is correct4 but what interests us here is not the
scientific answer resultin# fro$ the a**lication of relati!it" theor"- but
the wa" in which we instincti!el" tr" to account for the ifference) 6e
cannot allow oursel!es to think of the ifference between a *roton an
an electron as an irreucible ualis$ + like the ifference between soul
an bo") 9I use the best co$*arison I can fin4 but the for$ of
thou#ht- which insists on #ettin# behin + on e0*lainin# + !ariet"- is so
uni!ersal that e!en the ualis$ of soul an bo" is challen#e b" it:)
No sooner o we isco!er a ifference between *rotons an electrons
than we be#in to woner what $akes the$ ifferent) 6hen this
2uestion arises- we alwa"s fall back on structure) 6e tr" to e0*lain the
ifference as a ifference of structure- the structure of the *roton bein#
*resu$abl" the $ore co$*licate) .ut if *rotons an electrons
*ossess structure- the" cannot be the ulti$ate units of which structure
is built) Therefore the *resent !ariet" of the en+*roucts of *h"sical
anal"sis is an inication that we ha!e not "et touche botto$4 an we
$ust *ush our in!esti#ations farther- till we reach ientical units which
will not challen#e us to further anal"sis) The inference- as it ha**ens-
is fallacious- because the ifference between *rotons an electrons is
in the e0ternal relations an is not intrinsic) .ut a fallacious inference is
infor$ati!e as to our back#roun of thou#ht4 an the thou#ht which
insists on intruin# is that thin#s which iffer o so because the" ha!e
ifferent structure) The ifference resies in the structure an not in
the units out of which structure is built)
I conclue therefore that our en#raine for$ of thou#ht is such that we
shall not rest satisfie until we are able to re*resent all *h"sical
*heno$ena as an inter*la" of a !ast nu$ber of structural units
intrinsicall" alike) All the i!ersit" of *heno$ena will then be seen to
corres*on to ifferent for$ of relateness of these units or- as we
shoul usuall" sa"- ifferent confi#urations) There is nothin# in the
e0ternal worl which ictates this anal"sis into si$ilar units- 1ust as
there is nothin# in the irre#ular !ibrations of white li#ht which ictates
our anal"sis of it into $onochro$atic wa!e trains) The ictation co$es
fro$ our own wa" of thou#ht which will not acce*t as final an" other
for$ of solution of the *roble$ *resente b" sensor" e0*erience)
In current 2uantu$ theor" the anal"sis a**roaches- but has not "et
reache- this ieal) For that reason 2uantu$ *h"sicists are still
unsatisfie that the" ha!e #ot to the botto$ of the relationshi*s of the
!arious kins of *article that the" reco#ni/e- an of the connection
between #ra!itation- electro$a#netis$ an 2uanti/ation) For $" own
*art I think that the account #i!en in $ost books on 2uantu$ theor" b"
no $eans re*resents the full e0tent of our *resent knowle#e of this
*roble$) If $ore attention is *ai to the relati!it" sie of the *roble$-
the $ain outline of the e0tension of *h"sical theor" fro$ the *resent
haltin#+*oint of 2uantu$ theor" to the ulti$ate structural units is fairl"
clear) A #eneral account of the e!elo*$ent of a rational s"ste$ of
*h"sics- startin# fro$ the structural units- is #i!en on **) %G5+%G&) For
fuller etails of the ste*s b" which fro$ this be#innin# we are able to
euce the acce*te laws an constants of nature- reference $ust be
$ae to $" $athe$atical treatise)
%'
IA
It is usuall" i$*lie in the conce*t of anal"sis that the *arts are self+
sufficient) A *art can- without !iolence to thou#ht- be concei!e as
e0istin# without the other *arts which abut it) Or- to *ut it $ore
ri#orousl"- we can concei!e a whole which- when sub1ecte to the
s"ste$ of anal"sis we are e$*lo"in# woul "iel no $ore than this
one *art) The theoretical *h"sicist e$*lo"s this conce*tion of self+
sufficienc" when- in orer to in!esti#ate the structure of an ato$- he
re$o!es fro$ consieration the whole uni!erse e0ce*t this one ato$)
.ut here there arises a conflict of conce*tion) If a *art- e)#) an ato$-
were 1ust what it woul be without the rest of the uni!erse- an the rest
of the uni!erse were 1ust what it woul be without this one ato$- our
boies 9which for$ *art of the rest of the uni!erse: are 1ust what the"
woul be without this ato$- an we can therefore ha!e no sensor"
e0*erience in an" wa" connecte with or e$anatin# fro$ the ato$)
The conce*tion of *er$anentl" self+sufficient *arts of the *h"sical
uni!erse is self+contraictor"4 for such *arts are necessaril" outsie
obser!ational knowle#e- an therefore not *art of the uni!erse which
obser!ational knowle#e is for$ulate to escribe)
The $oel structure of an ato$ is inco$*lete unless it contains so$e
13
Relati!it" Theor" of Protons an Electrons 9Ca$bri#e- %&'G:
*ro!ision b" which we can beco$e aware of what is ha**enin# in the
ato$) In short- *h"sics ha!in# taken the worl to *ieces- has the 1ob of
ce$entin# it to#ether a#ain) The ce$ent is calle interaction)
One of the $ost re$arkable achie!e$ents of current 2uantu$ theor"
is the wa" it has sur$ounte the ifficult" of #i!in# to the *arts of the
uni!erse a kin of self+sufficienc" which oes not cut the$ off fro$
interaction with the rest of the uni!erse) To each t"*e of ato$ is
assi#ne a set of ele$entar" states 9ei#enstates:- each state
corres*onin# to a ifferent structure) It is these states- rather than the
ato$s the$sel!es- that are the en *roucts of our anal"sis) The ato$
itself is a co$bination of its states4 or- as we #enerall" sa"- it has
!arious *robabilities of bein# in its ifferent states) Si$ilarl" the
ulti$ate structural unit 9ientifie on *) %G' with a ,si$*le e0istence
s"$bol,: is an electron or *roton in an ele$entar" state + not- as it is
usuall" obser!e- in a co$bination of ele$entar" states) 6hen an
ato$ is isturbe b" other *articles- its ele$entar" states are not
isturbe4 their structure re$ains the sa$e is when the ato$ is
alto#ether isolate fro$ its surrounin#s) The onl" thin# isturbe is
the istribution of *robabilit" between the !arious ele$entar" states)
Thus the anal"tical *arts of the uni!erse are self+sufficient as re#ars
structure4 but our obser!ational knowle#e is concerne with the
istribution of *robabilit" a$on# the$- an in re#ar to the *robabilit"
istribution the" are interactin#)
The $ore closel" we stu" the $etho of 2uantu$ anal"sis- the $ore
we a**reciate the neatness of the wa" in which it o!erco$es the
conflict of thou#ht which re2uires the *arts "iele b" anal"sis to be
conce*tuall" ine*enent- but intere*enent in actual obser!ation)
The fact that it is *ossible to anal"se the uni!erse into co$*letel"
ine*enent *arts an then to a an interaction between the *arts
without in an" wa" $oif"in# the anal"sis- is less $"sterious when we
reali/e that an interaction can be wholl" sub1ecti!e) E!en if the *arts
the$sel!es are wholl" ob1ecti!e- an ha!e no *h"sical influence on
each other;s beha!ior- a sub1ecti!e interaction $a" a**ear in our
knowle#e of the$) 6e ha!e seen that the en+*roucts of our
anal"sis $ust be ientical structural units- which are therefore
inistin#uishable fro$ each other obser!ationall"- so that the" can be
interchan#e without affectin# obser!ation) Con!ersel" the s"ste$
inferable fro$ obser!ation + the knowable s"ste$ + is less
*articulari/e than the ob1ecti!e s"ste$- for the ini!iual *articles in
the knowable s"ste$ are left unientifie) 6e can onl" sa" that a
*article of the knowable s"ste$ has e2ual *robabilit" of bein# an" one
of the ob1ecti!e *articles) In co$*arin# the beha!ior of the knowable
s"ste$ with the beha!ior of the ob1ecti!e s"ste$- account $ust be
taken of the statistical effect of this inistin#uishabilit") The effect is
e2ui!alent to that which woul be *rouce b" *h"sical forces of
interaction) For e0a$*le- a *article $a" a**ear to e!iate fro$ its
e0*ecte *osition because it has been acte on b" a force or because-
owin# to the obser!ational inistin#uishabilit"- another *article has
been $istaken for it) An e0a$*le of this *urel" sub1ecti!e interaction
was #i!en on *) 'G)
There is now stron# reason to belie!e that all interaction forces in
*h"sics arise fro$ the inistin#uishabilit" of the ulti$ate *articles)
Interaction has therefore a sub1ecti!e ori#in) 6e ha!e alrea"
concee a *artial sub1ecti!it" to the ulti$ate *articles- but the
interaction ue to inistin#uishabilit" is ine*enent of this) It is not an
i$*erfection of our anal"sis that it fails to se*arate the uni!erse into
co$*letel" ine*enent *arts- an lea!es a certain a$ount of
interaction between the$4 it is rather the *erfection of the anal"sis
which brin#s about this result) 6e ha!e alrea" notice 9*) &F: that
there is a kin of iscontinuit" between ,#oo, an ,*erfect, in *h"sics)
,Perfect, is not so $uch the su*erlati!e of ,#oo, as a ,#oo, which
has o!erreache itself an efeate its own ai$s) If the ai$ of anal"sis
is to se*arate- it $ust sto* short of the ulti$ate structural units4
because when the *arts beco$e so si$*le that th" are
inistin#uishable- their inistin#uishabilit" confuses the$ in our
obser!ational knowle#e an- in a sense- unoes the se*aration which
the anal"sis has effecte)
A
,Substance, is one of the $ost o$inant conce*ts in our fa$iliar
outlook on the worl of sensor" e0*erience- an it is one with which
science fins itself continuall" at war) 6e ha!e alrea" touche on one
as*ect of it + that it is essentiall" *ositi!e- as contraste with for$
which is inifferentl" *ositi!e an ne#ati!e) Another attribute of
substance is its *er$anence or se$i+*er$anence4 an in this res*ect
*h"sics has ri itself of the conce*t of substance onl" to re*lace it b"
so$ethin# e2uall" *er$anent) Inirectl" therefore substance still
o$inates our for$ of thou#ht + a watere+own substance- of which
no attribute sur!i!es but its *er$anence)
To accor with this for$ of thou#ht- the anal"sis of the uni!erse into
*arts is re2uire to be- not a transitor" *artition- but a se*aration into
*arts which ha!e so$e e#ree of *er$anence) Per$anence is
for$ulate scientificall" in laws of conser!ation + conser!ation of $ass-
of ener#"- of $o$entu$- of electric char#e) In con1unction with the
ato$ic conce*t- the re2uire$ent of *er$anence leas us to reco#ni/e-
as the ulti$ate ele$entar" *articles- units 9*rotons an electrons:
which are nor$all"- an *robabl" alto#ether- inestructible) Further- in
wa!e $echanics which eals e0*licitl" with *robabilit"- we ha!e an
anal"sis into ei#enstates- i)e) stea" istributions of *robabilit" which
ha!e a consierable e#ree of *er$anence)
Owin# to the ifference of the natural ti$e+scale concerne-
*er$anence has a ifferent e*iste$olo#ical si#nificance in $olar
*h"sics fro$ *er$anence in $icrosco*ic *h"sics) In the ti$e+scale of
ato$ic flu0 a hunreth of a secon is !irtuall" an eternit") A
characteristic $ust be ,e!erlastin#, b" this stanar if it is to a**ear at
all in the ti$e+scale of orinar" hu$an *erce*tion) There is therefore a
clear reason for selectin# the *er$anent an isre#arin# the
transient features of $icrosco*ic s"ste$s) Classical- as well as
$oern- statistical $echanics is base on this consieration- which is
*robabl" the olest e*iste$olo#ical *rinci*le e0*licitl" acce*te in
*h"sics) .ut *er$anence in $olar *h"sics refers to a $uch lon#er
*erio of *ersistence- an there is not the sa$e reason for
concentratin# attention on characteristics which *osses it) That our
sub1ecti!e for$ulation of *h"sical knowle#e shoul i$*ose a
selection in fa!or of *ersistence u* to a hunreth of a secon or so is
the natural result of the coarse+#raineness of our ti$e+*erce*tion) If
there is an" selection in fa!or of *ersistence u* to a"s an centuries-
it $ust rest on other #rouns)
I ha!e e$*hasi/e the selecti!e effect of the $in;s insistence on
*er$anence in $" earlier writin#s- in which I was concerne onl" with
$olar *h"sics)
%C
It was the first hint of selecti!e sub1ecti!is$ that I
ca$e across) Lookin# back- I fin it curious that I was first con!ince
of the sub1ecti!e ori#in of so$e of the laws of nature b" a
consieration of $olar law- an was incline to re#ar the $icrosco*ic
laws 9at that ti$e onl" i$l" foreshaowe: as likel" to be ob1ecti!e4
because the a**lication to $olar law raises a ifficult 2uestion which
oes not a**ear in the a**lication to $icrosco*ic law) Let us consier
this ifference)
6e re#ar the $in as e$anin# b" its ,necessities of thou#ht,
certain 2ualities in the *arts which $ake u* the *h"sical uni!erse) The
$in i$*oses its e$ans b" refusin# to a$it an" s"ste$ of anal"sis
into *arts which oes not "iel *arts with the re2uire 2ualities) The
funa$ental laws of *h"sics are si$*l" a $athe$atical for$ulation of
the 2ualities of the *arts into which our anal"sis has i!ie the
uni!erse4 an it has been our contention that the" are all i$*ose b"
the hu$an $in in this wa"- an are therefore wholl" sub1ecti!e) It
woul be fatal to this !iew if it were foun that the ob1ecti!e uni!erse
,*la"s u*, to our anal"sis + that it e0hibits an intrinsic tenenc" to
se*arate into these *arts- as thou#h antici*atin# the $in;s e$an)
6e $ust therefore e0a$ine sus*iciousl" an" *heno$ena in which the
*arts see$ s*ontaneousl" to *resent the$sel!es se*aratel"- without
ha!in# to be u# out b" anal"sis)
In e0a$inin# $icrosco*ic *heno$ena- we ha!e to bear in $in the
14
S*ace- Ti$e- an 3ra!itation- *) %&G4 The Nature of the Ph"sical 6orl- *) 5C%
Procrustean $ethos of the e0*eri$enter which contri!e to su**l"
what our fra$e of thou#ht e$ans) Like the scul*tor- he reners
!isible the *arts or co$binations of *arts which our anal"tical
i$a#ination creates4 or at least his sortin# an $anufacturin#
o*erations *rouce effects which hu$our our belief that the *arts are
there) .ut in $olar *h"sics e0*eri$ental interference is too li$ite to
$atter) Our a**aratus cannot *rouce *lanets e0ecutin# orbits
*rescribe to orer in the wa" that it *rouces $onochro$atic li#ht
wa!es e0ecutin# !ibrations *rescribe to orer) If therefore we fin in
$olar *h"sics an"thin# which see$s to bolster u* our ao*te s"ste$
of anal"sis- it threatens our theor" $ore seriousl")
The *heno$enon re2uirin# e0a$ination fro$ this *oint of !iew is the
*resence of $ore or less *er$anent soli ob1ects in the worl of
fa$iliar a**rehension) Althou#h the *ersistence of $aterial for$s is
not an e0act e2ui!alent of the scientific *rinci*le of conser!ation of
$ass- there is a fairl" close association) Nor$all" a consierable
chan#e of $ass is associate with a *erce*tible chan#e of $aterial
for$) The *er$anent ob1ects aroun us #i!e- in a rou#h wa"- a
continuous *ractical e$onstration of the conser!ation of $ass) This
was scarcel" to be e0*ecte4 for a priori knowle#e onl" forewarns us
that the conser!ation of $ass $ust occur- not that it will be shoute at
us) There will be a conser!ation of so$ethin#- but not necessaril" of
so$ethin# a**rehensible in sensation)
The worl of fa$iliar *erce*tion- consistin# lar#el" of ob1ects with
so$e e#ree of *er$anence- to this e0tent fits s*ontaneousl" into our
for$ of thou#ht) For this it see$s an ae2uate e0*lanation that without
so$e e#ree of har$on" between thou#ht an sensation our
continue e0istence woul be i$*ossible) To in2uire how the har$on"
has co$e about + whether our sensor" e0*erience *uts it into our
heas to think as we o- or whether the e!olution of $an;s senses has
been #uie b" natural selection in such a wa" as not to conflict too
#rossl" with his necessities of thou#ht + $a" be like in2uirin# whether
the hen co$es first or the e##4 an it is *erha*s not !er" i$*ortant to
ecie) 6e $i#ht well lea!e o*en the 2uestion whether the for$s of
thou#ht which o$inate our outlook are ac2uire or innate) .ut I a$
incline to belie!e that the ulti$ate root is efinitel" $ental + a
*reis*osition inse*arable fro$ consciousness) It $ust be
re$e$bere that $ere sensation oes not eter$ine what we
orinaril" call the fa$iliar worl of sensor" e0*erience- in which the
ob1ects of $ore or less *er$anent for$ an si/e occur) That in!ol!es a
co$bination of sense with co$$on+sense) Of our !arious senses onl"
si#ht an touch ha!e an" res*onsibilit" for the fa$iliar conce*tion of
an e0ternal worl of *er$anent soli ob1ects) The *ri$iti!e for$s of
si#ht an touch + a #eneral sensiti!eness to li#ht an arkness- an a
sensiti!eness of fle0ible tentacles + *ro!ie little $aterial for a conce*t
of *er$anence) Fro$ these be#innin#s an elaborate sensor" s"ste$
has been e!ol!e in such awa" as to *ut !i!il" before us a worl
confor$able to the $in;s re2uire$ent of *er$anence)
It is clear that our sensor" e2ui*$ent $ust ha!e a selecti!e effect on
the knowle#e ac2uire throu#h it) The for$ of thou#ht which e0hibits
obser!ational knowle#e as a escri*tion of an e0ternal worl-
re*resents that worl as containin# ner!es an brains throu#h which
obser!ational knowle#e is ac2uire b" $ins) The selection of the
*arts or co$binations of *arts of the uni!erse to *erfor$ this function
of trans$ission- eter$ines the relati!e *ro$inence of the !arious
*arts an co$binations of *arts in our sensor" e0*erience) It is the ai$
of *h"sics to eli$inate this a!entitious *ro$inence- so that ulti$atel"
it oes not concern the scientific escri*tion of the uni!erse4 for
e0a$*le- the scientific escri*tion oes not reco#ni/e an" break
between !isible an ultra+!iolet raiation) .ut the *ro$inence in our
fa$iliar outlook ac2uire b" a *art throu#h its close relation to the
$echanis$ of sensation has $uch the sa$e effect in $olar *h"sics as
the sortin# out b" e0*eri$ental interference has in $icrosco*ic
*h"sics4 either $etho of isolation #i!es the *art a !i!iness in our
e0*erience which at first si#ht see$s out of kee*in# with the !iew that
it is $erel" the *rouct of a con!entional s"ste$ of anal"sis)
I conclue that it is not necessaril" a is*roof of the a priori character
of a *h"sical law to fin it closel" illustrate b" a *ro$inent feature of
the worl of fa$iliar a**rehension) Fa$iliar a**rehension is sub1ect to
the sa$e necessities of thou#ht as those which- b" $ore s"ste$atic
a**lication- "iel the scientific escri*tion of the uni!erse4 so that a
*artial con#ruence is not une0*ecte)
AI
The followin# sur!e" of our *osition e$*hasi/es $ore es*eciall" the
*oint we ha!e been consierin# in the last section?
9%: ." consieration of certain ee*l" roote for$s of thou#ht we can
foresee the funa$ental laws an constants which occur in the
*h"sical escri*tion of the uni!erse- the escri*tion ha!in# been
e!elo*e uner the #uiance of those for$s of thou#ht) .ut we
cannot foresee what will be the corres*onence between ele$ents in
this a priori *h"sical escri*tion an ele$ents in our fa$iliar
a**rehension of the uni!erse)
95: The corres*onence $i#ht be so re$ote that the a priori theor"
woul see$ al$ost irrele!ant to obser!ation) .ut actuall" the
corres*onence is fairl" ele$entar") 6e o not ha!e to search unul"
far in our fa$iliar e0*erience before we co$e across the thin#s which
obe" the laws *rescribe b" the a priori theor") 6e can al$ost see
*rotons an electrons in a 6ilson cha$ber4 we can al$ost see $ass
bein# conser!e) 6e o not actuall" see these thin#s4 but what we o
see has a !er" close relation to the$)
9': 6e coul force obser!ational knowle#e- whate!er it $i#ht ha**en
to be- into a *reeter$ine fra$e of thou#ht) The si#nificance of 95: is
that obser!ational knowle#e see$s to show a *reis*osition to fit into
the fra$e of thou#h without o!er$uch forcin#) This *reis*osition-
howe!er- shoul not be e0a##erate) The !er" wie rift now e0istin#
between the fa$iliar worl an the worl escribe in $oern scientific
theories is a $easure of the a$ount of forcin# that has been foun
necessar")
9C: Fro$ this *oint of !iew ,seein#, electrons an *rotons is not so
si#nificant as ,seein#, conser!ation of $ass) Electrons an *rotons
are sorte out b" e0*eri$ental interference4 but the *erce*tion of
ob1ects which illustrate conser!ation of $ass occurs without artificial
conitions an is a**arentl" a s*ontaneous testi$on" of sensation to
the a**ro*riateness of the a priori anal"sis)
9<: The e0istence of certain lines of sensor" co$$unication- relatin#
sensations in consciousness to selecte entities or conitions in the
*h"sical worl- is a selecti!e factor in our knowle#e) This selection is
alto#ether outsie our *resent control- but it is conitione b" the fact
that life woul be i$*ossible without so$e e#ree of har$on"
between the results of the selection an our en#raine for$s of
thou#ht) Conse2uentl" our *erce*tual reco#nition an abstraction of
certain ele$ents 9*er$anent *h"sical ob1ects: out of the web of
interconnecteness which $akes u* the *h"sical uni!erse follows with
a rue a**ro0i$ation the sa$e lines as the scientific anal"sis of the
*h"sical uni!erse base on the sa$e for$s of thou#ht)
The *ri$iti!e for$s of thou#ht which continue to o$inate *h"sics in
s*ite of the $oern re!olution are?
9%: The for$ which for$ulates knowle#e obtaine throu#h sensor"
e0*erience as a escri*tion of a uni!erse) It is throu#h this that the
*h"sical uni!erse is introuce an efine)
95: The conce*t of anal"sis- which re*resents the uni!erse as a
coe0istence of a nu$ber of *arts) As use in *h"sics the conce*t is
not li$ite to ,substance+anal"sis, which re2uires all the *arts to be
*ositi!e) In the $ore #eneral conce*tion of ,for$+anal"sis, the *arts
are inifferentl" *ositi!e an ne#ati!e4 an it is a conse2uence of this
#eneralit" that the si#nificance of a *art cannot be etache fro$ the
s"ste$ of anal"sis of which it is the result)
9': The ato$ic conce*t- which re2uires the s"ste$ of anal"sis to be
such that the ulti$ate *arts are ientical structural units4 so that all
!ariet" ori#inates in the structure an not in the ele$ents out of which
the structure is built)
9C: The conce*t of *er$anence 9a $oifie for$ of the conce*t of
substance: which re2uires the ulti$ate *arts to ha!e so$e e#ree of
*er$anence) This also leas us to #i!e s*ecial reco#nition to
*er$anent or se$i+*er$anent co$binations of *arts- an to
characteristics which re$ain *er$anent in the !icissitues of
*heno$ena)
9<: A conce*t of self+sufficienc" of the *arts 9eri!e *resu$abl" fro$
the conce*t of e0istence:) This to so$e e0tent conflicts with the
fore#oin# conce*tions) ." a co$*ro$ise the *arts are re#are as
intrinsically self+sufficient- but interactin# in our knowledge which is
concerne with *robabilit") This takes a!anta#e of the irre!ersible
relation between obser!ation an for$ulate knowle#e introuce b"
the *robabilit" conce*t 9*) &%:) 6e $i#ht- in fact- euce the
irre!ersibilit" 9an hence the nee for the *robabilit" conce*t: as an
e*iste$olo#ical conse2uence of the fra$e of thou#ht which re2uires
the ele$entar" *h"sical s"ste$s to be isolable an "et obser!able)
This list $a" not be e0hausti!e- but it see$s to e$brace the for$s
chiefl" res*onsible for our *resent outlook) It is i$*ortant that the"
shoul be brou#ht into the o*en- when we are consierin# how $uch
*h"sical science is eter$ine b" the a priori for$ of knowle#e an
how $uch b" an ob1ecti!e source of that knowle#e) Ha!in#- as far as
we can- trace the *ri$iti!e sources of the scientific fra$e of thou#ht-
we now turn to consier the fra$e which- b" so*histicate intellectual
effort- has been e!elo*e out of the$) The fra$e escribe in the
ne0t cha*ter re*resents the *resent frontier of a!ance) E!en the
$athe$atical *h"sicist oes not $aintain so a!ance a le!el of
thou#ht habituall"4 an it is usual to return to $ore fa$iliar $oes of
for$ulation to a**reciate the fruits of the a!ance)
CHAPTER IE
THE CONCEPT OF STR>CT>RE
I
Theoretical *h"sics toa" is hi#hl" $athe$atical) 6here oes the
$athe$atics co$e fro$@ I cannot acce*t Beans;s !iew that
$athe$atical conce*tions a**ear in *h"sics because it eals with a
uni!erse create b" a Pure 7athe$atician4 $" o*inion of *ure
$athe$aticians- thou#h res*ectful- is not so e0alte as that) An
unbiase consieration of hu$an e0*erience as a whole oes not
su##est that either the e0*erience itself or the truth re!eale in it is of
such a nature as to resol!e itself s*ontaneousl" into $athe$atical
conce*tions) The $athe$atics is not there till we *ut it there) The
2uestion to be iscusse in this cha*ter is- At what *oint oes the
$athe$atician contri!e to #et a #ri* on $aterial which intrinsicall"
oes not see$ *articularl" fitte for his $ani*ulations@
The $athe$atician will naturall" be#in b" introucin# a nu$ber of
s"$bols) Contrar" to the *o*ular belief- this oes not of itself rener a
sub1ect $athe$atical) If in a *ublic lecture I use the co$$on
abbre!iation No! for a nu$ber- nobo" *rotests4 but if I abbre!iate it as
N- it will be re*orte that ,at this *oint the lecturer e!iate into hi#her
$athe$atics,) 8isre#arin# such *re1uices- we $ust reco#ni/e that
the allocation of s"$bols $' *' +' O) to !arious entities or 2ualities is
$erel" an abbre!iate no$enclature which in!ol!es no $athe$atical
conce*tions)
The ne0t ste* is to introuce so$e kin of relation or co$*arison
between A an .) If we e0a$ine the $ental *rocess of co$*arin# two
ob1ects- I think we shall catch oursel!es i$a#inin# a series of ob1ects
inter$eiate between the$) 6e can best reali/e how the" iffer b"
consierin# what we shoul ha!e to o to chan#e one continuousl"
into the other) If the iea of #rauall" $oif"in# one into the other is
too far+fetche- we si$*l" ecie that the two ob1ects are so utterl"
unlike that a co$*arison woul be $eanin#less) It will therefore be
useful to introuce the conce*tion of an o*eration which chan#es one
ob1ect or 2ualit" into another) For e0a$*le- the conce*tion of an
o*eration of e0*ansion is useful when we ha!e to co$*are ob1ects of
ifferent si/e) Accorin#l" alon#sie our ori#inal A- .- C- O) we ha!e a
new set of s"$bols P- I- R- O) stanin# for the o*erations which
chan#e A into .- A into C- . into C- etc))
.ut we are still in the sta#e of no$enclature- an $athe$atics see$s
as far off as e!er) To continue- we $ust tr" to co$*are the o*erations
P- I- R- O with one another) Accorin# to our for$er conclusion this
leas us to i$a#ine an o*eration of chan#in# the o*eration P into the
o*eration I) Thus we ha!e a new set of o*erations 9or h"*er+
o*erations: E- Y- P- O) which chan#e P into I- P into R- I into R- O)
An so we #o on in an or#" of notation- introucin# $ore an $ore
s"$bols- but ne!er #ettin# be"on notation)
It is eas" to introuce $athe$atical notation4 the ifficult" is to turn it to
useful account?
Let 0 enote beaut"- " $anners well+bre- / fortune 9this last is
essential:-
Let L stan for lo!e + our *hiloso*her sai +
Then L is a function of 0- "- an /
Of the kin that is known as *otential)
Now inte#rate L with res*ect to t
9t stanin# for ti$e an *ersuasion:
Then between *ro*er li$its- ;tis eas" to see
The efinite inte#ral 7arria#e $ust be
9A !er" concise e$onstration:)
%<
At the start there is no essential ifference between this e0a$*le of
$athe$atical notation- an the A- .- C- O))- P- I- R- OO- E- Y- P- O))
15
Prof) 6) B) 7) Rankine- Son#s an Fables- %(FC)
that we ha!e been iscussin#) 6e $ust fin what it is that turns the
latter into a *owerful calculus for scientific *ur*oses- whereas the
for$er has no *ractical outco$e + as the *oe$ #oes on to relate)
To introuce $athe$atics we $ust so$ehow *ut a sto* to the infinite
re#ression of s"$bols) Such a ter$ination will be reache if we fin
that the E- Y- P- O)) are not new o*erations- but are alrea" containe
in the first set of o*erations P- I- R- O) that we introuce4 that is to
sa"- if we fin that the sa$e o*eration which chan#es one entit" into
another will also chan#e one o*eration into another)
As an e0a$*le- consier the o*erations of u*licatin#- tri*licatin#-
2uaru*licatin#- etc) If these are taken as P- I- R- O we ha!e ne0t to
consier- sa"- the o*eration Y which chan#es u*licatin# into
2uaru*licatin#) Iuaru*lication consists of two o*erations of
u*lication- i)e) of u*licatin# u*lication) Thus the o*eration Y is
u*licatin#- an has alrea" been introuce as P) 7ore #enerall"- if
the set P- I- R- O) enotes all *ossible o*erations of $ulti*lication-
fractional as well as inte#ral- the o*erations of chan#in# P into I- P
into R- I into R- etc) are also o*erations of $ulti*lication- an
therefore no new s"$bols are re2uire)
As another e0a$*le- su**ose that the initial entities A- .- C- O) are
*oints on a s*here) The o*eration
of chan#in# one *oint on a s*here into another is a rotation of the
s*here4 thus the o*erations P- I- R- O) are rotations) If P an I are
rotations throu#h e2ual an#les in ifferent *lanes- the one *lane is
chan#e into the other- an therefore P into I- b" another rotation- sa"
R) If P an I are rotations throu#h une2ual an#les- one can be
chan#e in to the other b" a co$bination of the o*erations of rotation
an $ulti*lication) 3rou*in# to#ether all *ossible o*erations of rotation
an $ulti*lication- no further o*erations are introuce in co$*arin#
one rotation with another)
6e see therefore that there e0ist ,ter$inable sets of o*erations, which
o not lea to a re#ression of no$enclature of e!er+increasin#
co$*le0it") It is onl" throu#h such ter$inable sets that $athe$atical
thou#ht can be introuce) To the e0tent to which the !arious *ortions
of our e0*erience can be relate to one another in ter$s of these
o*erations the" for$ $aterial for $athe$atical treat$ent) The full
e!elo*$ent of the iea- here briefl" inicate- is containe in the
Theory of Groups)
%G
II
A ter$inable set of o*erations- or as it is technicall" calle a group-
has a structure which can be escribe $athe$aticall") The fact that
the o*eration which chan#es P into I is alwa"s another $e$ber R of
the #rou* furnishes a set of trian#ular connections as the #rounwork
of the structure) These trian#ular connections can interlace a #reat
!ariet" of *atterns4 an it is the *attern of the interlacin# which
constitutes the absolute structure) 3rou*s are ifferentiate fro$ one
another b" their abstract structure) The $athe$atical escri*tion of the
#rou* s*ecifies onl" the *attern of interlacin#- an *a"s no attention to
the *h"sical nature of the o*erations which "iel this *attern) 6e $a"
therefore ha!e 2uite ifferent sets of o*erations with the sa$e #rou*+
structure- an therefore e2ui!alent so far as $athe$atical escri*tion
is concerne)
One of the $ost i$*ortant #rou*s in *h"sics is the #rou* of rotations in
si0 i$ensions) There are fifteen ine*enent *lanes of rotation in si0+
i$ensional s*ace 9corres*onin# to the three ine*enent *lanes of
rotation in three+i$ensional s*ace:4 an since we ha!e alwa"s to a
the o*eration of ,lea!in# thin#s as the" are,- which is an e) officio
$e$ber of e!er" #rou*- we ha!e si0teen ele$ents with which to for$
a #rou*+structure) A efinite interlockin# *attern is constitute b" the
association of these ele$ents 9other than the e) officio ele$ent: in si0
sets of fi!e 9*entas:- each ele$ent bein# a $e$ber of two *entas)
Interlacin# with it is an association of the ele$ents in trias- the trias
16
An ele$entar" account of the theor" of #rou*s- an of the *art it *la"s in the
founations of theoretical *h"sics- is #i!en in New Pathways in Science- Ch) EII)
the$sel!es bein# associate in con1u#ate *airs) Each of the fifteen
ele$ents *la"s an e2ui!alent *art in the *attern)
Rotation in si0 i$ensions is onl" one of $an" sets of o*erations
which "iel this *articular #rou*+*attern) For e0a$*le- if we *lace four
ifferent coins on the table- the o*erations of interchan#in# the$ in
*airs- with or without turnin# one *air the other wa" u*- for$ a #rou*
with this structure)
%F
The sa$e *attern of relations turns u* in the
#eo$etr" of =u$$er;s Iuartic Surface- in the theor" of Theta
Functions- an + $ost i$*ortant of all for our *ur*oses + in the
s*ecification of an ele$entar" *article 9*roton or electron: in an
ele$entar" state- incluin# the s*ecification of its char#e an s*in)
Pro*erl" to reali/e the conce*tion of #rou*+structure- we $ust think of
the *attern of interwea!in# as abstracte alto#ether fro$ the *articular
entities an relations that furnish the *attern) In *articular- we can #i!e
an e0act $athe$atical escri*tion of the *attern- althou#h
$athe$atics $a" be 2uite ina**ro*riate to escribe what we know of
the nature of the entities an o*erations concerne in it) In this wa"
$athe$atics #ets a footin# in knowle#e which intrinsicall" is not of a
kin su##estin# $athe$atical conce*tions) Its function is to eluciate
the #rou*+structure of the ele$ents of that knowle#e) It is$isses the
ini!iual ele$ents b" assi#nin# to the$ s"$bols- lea!in# it to non+
$athe$atical thou#ht to e0*ress the knowle#e- if an"- that we $a"
ha!e of what the s"$bols stan for)
6e shall refer to this abstraction as the $athe$atical conce*t of
structure- or briefl" as the conce*t of structure) Since the structure-
abstracte fro$ whate!er *ossesses the structure- can be e0actl"
s*ecifie b" $athe$atical for$ulae- our knowle#e of structure is
co$$unicable- whereas $uch of our knowle#e is inco$$unicable) I
cannot con!e" to "ou the !i!i knowle#e which I ha!e of $" own
sensations an e$otions) There is no wa" of co$*arin# $" sensation
of the taste of $utton with "our sensation of the taste of $utton4 I can
onl" know what it tastes like to $e- an "ou can onl" know what it
17
New Pathwa"s in Science- *) 5GF) Letters are there substitute for coins)
tastes like to "ou) .ut if we are both lookin# at a lansca*e- althou#h
there is no wa" of co$*arin# our !isual sensations as such- we can
co$*are the structures of our res*ecti!e !isual i$*ressions of the
lansca*e) It is *ossible for a #rou* of sensations in $" $in to ha!e
the sa$e structure as a #rou* of sensations in "our $in) It is *ossible
also that a #rou* of entities which are not sensations in an"one;s $in-
associate to#ether b" relations of which we can for$ no conce*tion-
$a" ha!e this sa$e structure) 6e can therefore ha!e structural
knowle#e of that which is outsie e!er"one;s $in) This knowle#e
will consist of the sa$e kin of assertions as those which are $ae
about the *h"sical uni!erse in the $oern theories of $athe$atical
*h"sics) For strict e0*ression of *h"sical knowle#e a $athe$atical
for$ is essential- because that is the onl" wa" in which we can confine
its assertions to structural knowle#e) E!er" *ath to knowle#e of
what lies beneath the structure is then blocke b" an i$*enetrable
$athe$atical s"$bol)
Ph"sical science consists of *urel" structural knowle#e- so that we
know onl" the structure of the uni!erse which it escribes) This is not a
con1ecture as to the nature of *h"sical knowle#e4 it is *recisel" what
*h"sical knowle#e as for$ulate in *resent+a" theor" states itself to
be) In funa$ental in!esti#ations the conce*tion of #rou*+structure
a**ears 2uite e0*licitl" as the startin#+*oint4 an nowhere in the
subse2uent e!elo*$ent o we a$it $aterial not eri!e fro$ #rou*+
structure)
The fact that structural knowle#e can be etache fro$ knowle#e of
the entities for$in# the structure- #ets o!er the ifficult" of
unerstanin# how it is *ossible to concei!e a knowle#e of an"thin#
which is not *art of our own $ins) So lon# as the knowle#e is
confine to assertions of structure- it is not tie own to an" *articular
real$ of content) It will be re$e$bere that we ha!e se*arate the
2uestion of the nature of knowle#e fro$ the 2uestion of assurance of
its truth) 6e are not here consierin# how it is *ossible to be assure
of the truth of knowle#e relatin# to so$ethin# outsie our $ins4 we
are occu*ie with the *rior 2uestion how it is *ossible to $ake an"
kin of assertion about thin#s outsie our $ins- which 9whether true
or false: has a efinable $eanin#)
III
I woner if "ou hesitate before acce*tin# $" state$ent 9*) %'&: that
2uaru*licatin# is a u*licatin# of the o*eration of u*licatin#) If I ha
sai ,four ti$es is twice twice, "ou woul ha!e a$itte it
unhesitatin#l"4 but it su##ests itself that u*licatin# or twicin# as
a**lie to an o*eration $i#ht well $ean oin# it o!er a#ain for a
check- an it is rather #ratuitous to assu$e that the secon o*eration
is necessaril" one on the en *rouct of the first)
Outsie $athe$atics the state$ent that ,two an two is four, is rather
too swee*in#4 but we $a" #o so far as to assert that- if ,two an two,
is a nu$ber at all- that nu$ber is four) In other wors- if u*licatin#-
tri*licatin#- etc) are unerstoo to for$ a #rou*- i)e) a ter$inable set of
o*erations- so that when a**lie to each other the" "iel other
o*erations of the set- then the $e$ber of the set obtaine b"
u*licatin# u*licatin# is 2uaru*licatin#)
Su**ose- howe!er- we acce*t the other $eanin#- so that when the
o*eration of u*licatin# is a**lie to u*licatin# it "iels a new kin of
o*eration- ifferent fro$ an" of the ori#inal set- which we $a" escribe
as a ,checke u*lication,) Let us tr" another u*lication) The twice
*erfor$e $ulti*lication b" two is itself to be u*licate4 an this can
ha!e no other $eanin# than that the $ulti*lication b" two is *erfor$e
four ti$es) Thus- if not at the first ste*- at an" rate at the secon ste*-
we reach the #rou* conce*t of u*licatin# which confor$s to the rule
that four ti$es is twice twice) This is in accorance with what we ha!e
alrea" notice + that $athe$atical thou#ht oes not be#in to take
char#e until the secon ste*- when we reach relations between
relations or o*erations on o*erations)
In orer to for$ulate this *oint e0*licitl" we shall istin#uish between a
structural conce*t an $ore #eneral kins of conce*t) A structural
conce*t is obtaine fro$ a corres*onin# #eneral conce*t b"
eli$inatin# fro$ our conce*tion e!er"thin# which is not essential to the
*art it *la"s in a #rou*+structure) It is an ele$ent in a s*ecifie *attern
without an" *ro*erties e0ce*t its connection with the *attern) Its
*ro*erties are those of a $athe$atical s"$bol- which consist solel" of
its associations 9or $ore strictl" the associations of its associations:
with other s"$bols) The corres*onin# #eneral conce*t- if an"- is our
conce*tion of what the s"$bol re*resents in our orinar" non+
$athe$atical for$ of thou#ht) A #eneral conce*t lacks the *recision of
a $athe$atical conce*t- an is often ifficult to *in own to an"thin#
efinite) E0ce*t as a**lie to sensations- e$otions- etc) of which we
can be irectl" aware- it is oubtful if the #eneral conce*t is $ore than
a self+ece*tion which *ersuaes us that we ha!e an a**rehension of
so$ethin# which we cannot a**rehen) Ne!ertheless- such conce*ts
$ust be reckone with as *art of our en#raine for$ of thou#ht)
The conce*ts referre to in Cha*ter AIII were #eneral conce*ts
occurrin# in our orinar" for$ of thou#ht) It is now *ossible to a that
in e$*lo"in# the$ to furnish the fra$e of thou#ht in which our
scientific knowle#e is containe- we ha!e #rauall" eli$inate their
#eneral as*ects- until now we reco#ni/e onl" the corres*onin#
structural conce*ts) Corres*onin#l" the resultin# fra$e of thou#ht
has beco$e a $athe$atical fra$e- an the knowle#e containe in it
is $athe$atical knowle#e + a knowle#e of #rou*+structure) ."
introucin# the $athe$atical theor" of structure $oern *h"sics is
able to carr" out in a *recise $anner the #eneral *rinci*les escribe
in the last cha*ter) For e0a$*le- we there insiste that the si#nificance
of a *art cannot be issociate fro$ the s"ste$ of anal"sis to which it
belon#s) As a structural conce*t the *art is a s"$bol ha!in# no
*ro*erties e0ce*t as a constituent of the #rou*+structure of a set of
*arts)
To show how these ieas are a**lie- let us consier the conce*t of
s*ace) Takin# first the #eneral conce*t- we usuall" re#ar infinite
Eucliean s*ace as the si$*lest kin of s*ace to concei!e) One woul
ha!e thou#ht that the infinitue woul be rather a serious obstacle to
conce*tion4 but $ost *eo*le $ana#e to *ersuae the$sel!es that
the" ha!e o!erco$e the ifficult"- an e!en *rofess the$sel!es utterl"
unable to concei!e a s*ace without infinitue) .ut- whate!er the truth
about the #eneral conce*t- the structural conce*t of Eucliean s*ace
is e0ce*tionall" ifficult) Since I want to #i!e here a co$*arati!el" eas"
illustration- I will consier unifor$ s*herical s*ace which has a $uch
si$*ler structural conce*t)
An" *oint in s*herical s*ace can be chan#e into an" other *oint b" a
rotation of the s*here) Thus to the *oints or ele$ents of s*herical
s*ace A- .- C- O) there corres*on o*erators P- I- R- O) which are
the rotations of the s*here4 an the #rou* of the o*erators is si$*l"
the #rou* of rotations in the *ro*er nu$ber of i$ensions 9in this case
four i$ensions:) Re#arin# ,s*ace, as a structural conce*t- all that
we know about s*herical s*ace is that it has the #rou*+structure of this
#rou* of rotations) 6hen we introuce s*herical s*ace in *h"sics we
refer to so$ethin# + we know not what + which has this structure)
E2uall"- if we refer to Eucliean s*ace we refer to so$ethin# + we
know not what + with a s*ecifiable #rou*+structure- thou#ht it re2uires
rather $ore a!ance $athe$atical conce*tions to for$ulate the
s*ecification) Si$ilarl" the s*ace of irre#ular cur!ature which a**ears
in Einstein;s theor" is so$ethin# with a #rou*+structure re2uirin# rather
$ore elaborate s*ecification)
The #eneral conce*t- which atte$*ts to escribe s*ace as it a**ears
in fa$iliar a**rehension + what it looks like- what it feels like- its
ne#ati!eness as co$*are with $atter- its ,thereness, + is an
e$bellish$ent of the bare structural escri*tion) So far as *h"sical
knowle#e is concerne- this e$bellish$ent is an unauthori/e
aition) Philoso*hicall" it is all to the #oo if we fin a ifficult" in
concei!in# in non+$athe$atical for$s of thou#ht the kins of s*ace
which $oern *h"sics has introuce4 for we are thereb" iscoura#e
fro$ $akin# such e$bellish$ents)
IA
The $athe$atical theor" of structure is the answer of $oern *h"sics
to a 2uestion which has *rofounl" !e0e *hiloso*hers)
.ut- if I ne!er know irectl" e!ents in the e0ternal worl- but onl" their
alle#e effects on $" brain- an if I ne!er know $" brain e0ce*t in
ter$s of its alle#e effects on $" brain- I can onl" reiterate in
bewiler$ent $" ori#inal 2uestions? ,6hat sort of thin# is it that I
know@, an ,6here is it@,
%(
6hat sort of thin# is it that I know@ The answer is structure) To be 2uite
*recise- it is structure of the kin efine an in!esti#ate in the
$athe$atical theor" of #rou*s)
It is ri#ht that the i$*ortance an ifficult" of the 2uestion shoul be
e$*hasi/e) .ut I think that $an" *ro$inent *hiloso*hers- uner the
i$*ression that the" ha!e set the *h"sicists an insoluble conunru$-
$ake it an e0cuse to turn their backs on the e0ternal worl of *h"sics
an welter in a barren realis$ which is a ne#ation of all that *h"sical
science has acco$*lishe in unra!ellin# the co$*le0it" of sensor"
e0*erience) The $athe$atical *h"sicist- howe!er- welco$es the
2uestion as one fallin# es*eciall" within his *ro!ince- in which his
s*eciali/e knowle#e $a" be of ser!ice to the #eneral a!ance$ent
of *hiloso*h")
The *hrase ,if I ne!er know $" brain e0ce*t in ter$s of its alle#e
effects on $" brain, !i!il"- if not alto#ether accuratel"-
%&
escribes the
conitions uner which we labor) .ut it is not !er" alar$in# to the
*h"sicist- whose sub1ect abouns with this kin of c"clic e*enence)
6e onl" know an electric force b" its effects on an electric char#e4 an
we onl" know electric char#es in ter$s of the electric forces the"
*rouce) It has lon# been e!ient that this is no bar to knowle#e4 but
it is onl" recentl" that the s"ste$atic $etho of for$ulatin# such
knowle#e in ter$s of #rou*+structure has beco$e a reco#ni/e
18
C) E) 7) Boa- Aristotelian Societ"- Su**) !ol) IE- *) %'F) Iuote b" L) S) Stebbin#-
Philoso*h" an the Ph"sicists- *) GC)
19
A $ore accurate for$ woul be? ,if I ne!er know an" brain e0ce*t in ter$s of its alle#e
effects on a brain),
*roceure in *h"sical theor")
The bewiler$ent of the *hiloso*hers e!ientl" arises fro$ a belief
that- if we start fro$ /ero- an" knowle#e of the e0ternal worl $ust
be#in with the assu$*tion that a sensation $akes us aware of
so$ethin# in the e0ternal worl + so$ethin# ifferin# fro$ the
sensation itself because it is non+$ental) .ut knowle#e of the
*h"sical uni!erse oes not be#in in that wa") One sensation 9i!orce
fro$ knowle#e alrea" obtaine b" other sensations: tells us nothin#4
it oes not e!en hint at an"thin# outsie the consciousness in which it
occurs) The startin# *oint
5D
of *h"sical science is knowle#e of the
#rou*+structure of a set of sensations in a consciousness) 6hen these
fra#$ents of structure- contribute at !arious ti$es an b" !arious
ini!iuals- ha!e been collate an re*resente accorin# to the for$s
of thou#ht that we ha!e iscusse- an when the #a*s ha!e been
fille b" an inferre structure e*enin# on the re#ularities isco!ere
in the irectl" known *ortions- we obtain the structure known as the
*h"sical uni!erse)
After this #eneral s"nthesis of structure- we are in a *osition to
escribe an" *articular *ortion of the structure in the ter$s in which
*h"sical knowle#e is orinaril" e0*resse) This will *ro!ie an
alternati!e 9*h"sical: escri*tion of the ori#inal sensations) Since the"
are ele$ents of a structure of sensations- an this structure has been
incor*orate in the structure which constitutes the *h"sical uni!erse-
we can escribe the$ in *h"sical ter$s) Our *h"siolo#ical knowle#e
is *robabl" insufficient to s*ecif" the e0act *h"sical e!ent which is also
a sensation in so$eone;s $in4 but a**ro0i$atel" enou#h for $ost
*ur*oses we $a" take it to be a set of electrical i$*ulses occurrin# at
the brain+ter$inal of a bunle of ner!es)
It is i$*ortant to notice that the inter*retation of sensor" e0*erience-
like the inter*retation of a ci*her- inclues two istinct *roble$s)
,Inter*retin# a ci*her, $a" $ean the *roceure of isco!erin# the
20
I $ean the lo#ical startin# *oint- not the historical startin# *oint- of a sub1ect which has
#rown out of crue be#innin#s)
coe- or it $a" $ean ecoin# a *articular $essa#e with the coe
alrea" known) In the sa$e wa"- the *roceure of inter*retin# our
sensations as infor$ation about an e0ternal worl $a" refer to the
*roble$- which stans at the be#innin# of *h"sics- of associatin# the
fra#$ents of structure in consciousness with the structure of an
e0ternal uni!erse4 or it $a" refer to the *articular infor$ation
obtainable fro$ each new sensation when we a**l" our accu$ulate
*h"sical an *h"siolo#ical knowle#e) In re#ar to the initial *roble$-
a sin#le sensation is no $ore infor$ati!e that a sin#le letter in a ci*her
of which we ha!e not the ke") .ut- after the initial *roble$ has been
sol!e- we are able to inter*ret sensations ini!iuall" as a ci*her is
ecoe letter b" letter) A sensation of noise infor$s $e of an
electrical isturbance of a *articular ner!e+ter$inal + which- of course-
oes not $ean that it infor$s $e that this is the correct *h"sical
escri*tion of what has occurre) The escri*tion is *ro!ie
beforehan b" the solution of the initial *roble$- so that it is rea" for
use when the sensation infor$s $e that an e!ent has occurre to
which it is a**licable)
The isturbance at the ner!e ter$inal is #enerall" the result of a lon#
chain of causation in the *h"sical worl) In fa$iliar thou#ht we usuall"
lea* to the far en of the chain of causation- an sa" that the
sensation is cause b" an ob1ect at so$e istance fro$ the seat of the
sensation) In the case of the !isual sensation cause b" a s*iral
nebula- the ob1ect is not onl" re$ote in s*ace but $a" be $illions of
"ears istant in ti$e) Causation bri#es the #a* in s*ace an ti$e- but
the *h"sical e!ent at the seat of sensation 9*ro!isionall" ientifie with
an electrical isturbance of a ner!e ter$inal: is not the cause of the
sensation4 it is the sensation) 7ore *recisel"- the *h"sical e!ent is the
structural conce*t of that of which the sensation is the #eneral
conce*t)
Thus- when "ou tell $e that "ou hear a noise- the infor$ation i$*arte
is re*resente in $" knowle#e b" 9a: a #eneral conce*t of a hear+
noise- i)e) a conce*t of so$ethin# of si$ilar nature to $" own
awareness of noises- an 9b: a structural conce*t of a hear+noise- i)e)
a *art of the structure of the *h"sical uni!erse which we escribe as
an electricall" isturbe ter$inal of an auitor" ner!e) Of these two
conce*ts of a hear+noise- the one refers to what it is in itself- the
other refers to what it is as a constituent of the structure known as the
*h"sical uni!erse)
A
The reco#nition that *h"sical knowle#e is structural knowle#e
abolishes all ualis$ of consciousness an $atter) 8ualis$ e*ens
on the belief that we fin in the e0ternal worl so$ethin# of a nature
inco$$ensurable with what we fin in consciousness4 but all that
*h"sical science re!eals to us in the e0ternal worl is #rou*+structure-
an #rou*+structure is also to be foun in consciousness) 6hen we
take a structure of sensations in a *articular consciousness an
escribe it in *h"sical ter$s as *art of the structure of an e0ternal
worl- it is still a structure of sensations) It woul be entirel" *ointless
to in!ent so$ethin# else for it to be a structure of) Or- to *ut it another
wa"- there is no *oint in in!entin# non+*h"sical re*licas of certain
*ortions of the structure of the e0ternal worl an transferrin# to the
re*licas the non+structural 2ualities of which we are aware in
sensation) The *ortions of the e0ternal uni!erse of which we ha!e
aitional knowle#e b" irect awareness a$ount to a !er" s$all
fraction of the whole4 of the rest we know onl" the structure- an not
what it is a structure of)
Let us enote b" E the entit" of which the *h"sical uni!erse is the
structure-
5%
an istin#uish the s$all *art E
s
known to be of sensor"
nature fro$ the re$ainer E
u
of which we ha!e no irect awareness) It
$a" be su##este that there re$ains a ualis$ of E
s
an E
u

e2ui!alent to the ol ualis$ of consciousness an $atter4 but this is- I
think- a lo#ical confusion- in!ol!in# a switch o!er fro$ the
e*iste$olo#ical !iew of the uni!erse as the the$e of knowle#e to an
e0istential !iew of the uni!erse as so$ethin# of which we ha!e to
21
I usuall" call E the ,e0ternal worl,- the ,*h"sical worl, bein# li$ite to the structure of
the e0ternal worl)
obtain knowle#e) Structurall" E
u
is no ifferent fro$ E
s
- an to #i!e
$eanin# to the su**ose ualis$ we ha!e to i$a#ine a
su**le$entar" non+structural knowle#e of E
u
re!ealin# its unlikeness
to E
s
) 6e ha!e to su**ose that a irect awareness of E
u
- if we coul
*ossess it- woul show that it was not of sensor" nature) .ut the
su**osition is nonsense4 or if we ha the su**ose irect awareness
of E
u
- it woul ipso facto be a sensation in our consciousness) Thus we
cannot #i!e $eanin# to the ualis$ without $akin# a su**osition
which eli$inates the ualis$)
Althou#h the state$ent that the uni!erse is of the nature of ,a thou#ht
or sensation in a uni!ersal 7in, is o*en to criticis$- it oes at least
a!oi this lo#ical confusion) It is- I think- true in the sense that it is a
lo#ical conse2uence of the for$ of thou#ht which for$ulates our
knowle#e as a escri*tion of a uni!erse) .ut it re2uires $ore #uare
e0*ression if it is to be acce*te as a truth transcenin# for$s of
thou#ht)
To su$ u*) The *h"sical uni!erse is a structure) Of the E of which it is
the structure- we onl" know that E inclues sensations in
consciousness) To the 2uestion? 6hat is E when it is not a sensation in
an" consciousness known to us@ the ri#ht answer is *robabl" that the
2uestion is a $eanin#less one + that a structure oes not necessaril"
i$*l" an E of which it is the structure) In other wors- the 2uestion
takes us to a *oint where the for$ of thou#ht in which it ori#inates
ceases to be useful) The for$ of thou#ht can onl" be *reser!e b" still
attributin# to E a sensor" nature + a sensation in a consciousness
unknown to us) 6hat interests us is not the *ositi!e conclusion- but the
fact that in no circu$stances are we re2uire to conte$*late an E of
non+sensor" nature)
The fact that the conce*t of structure affors an esca*e fro$ ualis$
has been reco#ni/e es*eciall" in the *hiloso*h" of .ertran Russell)
Althou#h I ha!e 2uote it in three earlier books- I feel obli#e to 2uote
a#ain a *assa#e fro$ Russell;s ,ntroduction to -athematical
Philosophy 9%&%&: which has #reatl" influence $" own thou#ht?
,There has been a #reat eal of s*eculation in traitional *hiloso*h"
which $i#ht ha!e been a!oie if the i$*ortance of structure- an the
ifficult" of #ettin# behin it- ha been reali/e) For e0a$*le- it is often
sai that s*ace an ti$e are sub1ecti!e- but the" ha!e ob1ecti!e
counter*arts4 or that *heno$ena are sub1ecti!e- but are cause b"
thin#s in the$sel!es- which $ust ha!e ifferences inter se
corres*onin# with the ifferences in the *heno$ena to which the"
#i!e rise) 6here such h"*otheses are $ae- it is #enerall" su**ose
that we can know !er" little about the ob1ecti!e counter*arts) In actual
fact- howe!er- if the h"*otheses as state were correct- the ob1ecti!e
counter*arts woul for$ a worl ha!in# the sa$e structure as the
*heno$enal worlO) In short- e!er" *ro*osition ha!in# a
co$$unicable si#nificance $ust be true of both worls or of neither4
the onl" ifference $ust lie in 1ust that essence of ini!iualit" which
alwa"s elues wors an baffles escri*tion- but which for that !er"
reason is irrele!ant to science),
This was written ine*enentl" of the new scientific theories- which
were then in an earl" sta#e4 but it illu$inate the *hiloso*hic tren
which was be#innin# to a**ear in the$) It is interestin# to co$*are the
scientific *osition in %&%& with the *osition in %&'&) In %&%& it was a fair
inference that *h"sical knowle#e $ust be knowle#e of structure-
althou#h in the for$ in which it was then *resente it i not look
$uch like it) In #eneral the structural knowle#e i not a**ear in
*h"sics e0*licitl"4 it was thou#ht of as the kernel of truth which woul
outlast the chan#in# theories which enhulle it) In the inter!enin#
"ears the i$*ortance of i##in# out the structure fro$ its inessential
tra**in#s beca$e reco#ni/e- an it was notice that in the Theor" of
3rou*s in *ure $athe$atics the necessar" techni2ue ha been
e!elo*e) 7oreo!er- the iea of structure- which ha *re!iousl" been
rather !a#ue- was foun ca*able of e0act $athe$atical efinition)
Conse2uentl" toa" it is not $erel" a truth hien in our *h"sical
knowle#e but *h"sical knowle#e in its current for$ that we
reco#ni/e as structural)
CHAPTER E
THE CONCEPT OF EEISTENCE
I
I fin a ifficult" in unerstanin# books on *hiloso*h" because the"
talk a #reat eal about ,e0istence,- an I o not know what the" $ean)
E0istence see$s to be a rather i$*ortant *ro*ert"- because I #ather
that one of the $ain sources of i!ision between ifferent schools of
*hiloso*h" is the 2uestion whether certain thin#s e0ist or not) .ut I
cannot e!en be#in to unerstan these issues- because I can fin no
e0*lanation of the ter$ ,e0ist,)
The wor ,e0istence, is- of course- fa$iliar in e!er"a" s*eech4 but it
oes not e0*ress a unifor$ iea + a uni!ersall" a#ree *rinci*le
accorin# to which thin#s can be i!ie into e0istin# an non+
e0istin#) 8ifference of o*inion as to whether a thin# e0ists or not
so$eti$es arises because the thin# itself is i$*erfectl" efine- or
because the e0act i$*lications of the efinition ha!e not been
#ras*e4 thus the ,real e0istence, of electrons- aether- s*ace- color-
$a" be affir$e or enie because ifferent *ersons use these ter$s
with so$ewhat ifferent i$*lications) .ut a$bi#uit" of efinition is not
alwa"s res*onsible for the ifference of !iew) Let us take so$ethin#
fa$iliar- sa" an o!erraft at a bank) No one can fail to unerstan
*recisel" what that $eans) Is an o!erraft so$ethin# which e0ists@ If
the 2uestion were *ut to the !ote- I think so$e woul sa" that its
e0istence $ust be acce*te as a #ri$ realit"- an others woul
consier it illo#ical to concee e0istence to what is intrinsicall" a
ne#ation) .ut what i!ies the two *arties is no $ore than a 2uestion
of wors) It woul be absur to i!ie $ankin into two sects- the one
belie!in# in the e0istence of o!errafts an the other en"in# their
e0istence) The i!ision is a 2uestion of classification- not of belief) If
"ou tell $e "our own answer- I shall not learn an"thin# new about the
nature or *ro*erties of an o!erraft4 but I shall learn so$ethin# about
"our usa#e of the ter$ ,e0ists, + what cate#or" of thin#s "ou inten it
to co!er)
It is a *ri$iti!e for$ of thou#ht that thin#s either e0ist or o not e0ist4
an the conce*t of a cate#or" of thin#s *ossessin# e0istence results
for$ forcin# our knowle#e into a corres*onin# fra$e of thou#ht)
E!er"one oes this instincti!el"4 but there are borerline cases in
which all o not e$*lo" the sa$e criteria- as the e0a$*le of the
o!erraft shows) A *hiloso*her is not boun b" traitional or instincti!e
con!entions to the sa$e e0tent as a la"$an4 an when he si$ilarl"
e0*resses his knowle#e in this *ri$iti!e fra$e of thou#ht- it is
i$*ossible to #uess what classificator" s"ste$ he will ao*t) It woul
be rather sur*risin# if all *hiloso*hers ao*te the sa$e s"ste$) In
an" case I o not see wh" such a $"ster" shoul be $ae of it- nor
how an arbitrar" ecision as to the classification to be ao*te has
co$e to be transfor$e into a fer!i *hiloso*hical belief)
I o not want to $ake swee*in# char#es on the basis of a !er" li$ite
reain# of *hiloso*h") I a$ aware that in the $ore reconite works the
$eanin# of the ter$ is so$eti$es iscusse) .ut- after all-
*hiloso*hers o occasionall" write for the la"$an4 an so$e of the$
seek to re*el the scientific in!aer in lan#ua#e which he is su**ose
to unerstan) 6hat I co$*lain of is that these writers o not see$ to
reali/e that the ter$ ,e0ist,- if the" o not e0*lain the $eanin# the"
attach to it- $ust necessaril" be as bewilerin# to the scientist- as- for
e0a$*le- the ter$ ,cur!ature of s*ace,- if left une0*laine- woul be to
the *hiloso*her) An I think it is not an unfair inference fro$ this
o$ission that the" the$sel!es attach $ore i$*ortance to the wor
than to its $eanin#)
It is not e!er" sentence containin# the !erb ,to e0ist, that troubles $e)
The ter$ is often use in an intelli#ible wa") For $e 9an- it a**ears-
also for $" ictionar": ,e0ists, is a rather e$*hatic for$ of ,is,) ,A
thou#ht e0ists in so$ebo";s $in-, i)e) a thou#ht is in so$ebo";s
$in + I can unerstan that) ,A state of war e0ists in Ruritania-, i)e) a
state of war is in Ruritania + not !er" #oo En#lish- but intelli#ible) .ut
when a *hiloso*her sa"s ,Fa$iliar chairs an tables e0ist,- i)e) fa$iliar
chairs an tables areO)- I wait for hi$ to conclue) Yes@ 6hat were
"ou #oin# to sa" the" are@ .ut he ne!er finishes the sentence)
Philoso*h" see$s to $e full of half+finishe sentences4 an I o not
know what to $ake of it)
S*eech is often elli*tical- an I o not $in unfinishe sentences if I
know how the" are $eant to be finishe) ,A horrible noise e0ists,
*resu$abl" is intene to be co$*lete in so$e such for$ as ,A
horrible noise is + isturbin# $e,) .ut that is not how the *hiloso*her
intens $e to co$*lete his unfinishe state$ent ,noises actuall" e0ist,
+ an I reall" ha!e no iea what co$*letion he oes inten) I $"self-
when I a$ not inti$iate b" the e0istence
55
of critics eter$ine to
$ake nonsense of $" wors if it is *ossible to o so- often sa" that
ato$s an electrons e0ist) I $ean- of course- that the" e0ist + or are +
in the *h"sical worl- that bein# the the$e of iscussion in the conte0t)
6e nee not e0a$ine the *recise elli*sis b" which a $athe$atician
sa"s that the root of an e2uation e0ists- when he $eans that the
e2uation has a root4 it is sufficient to sa" that he has no iea of *uttin#
forwar a clai$ to inclue the root of a $athe$atical e2uation in the
cate#or" of thin#s which *hiloso*hers s*eak of as ,reall" e0istin#,)
In the *recein# cha*ters I ha!e iscusse a nu$ber of thin#s which
e0ist in the *h"sical uni!erse4 that is to sa"- the" are in- or are *arts of-
the *h"sical uni!erse) 6e ha!e seen that ,to e0ist in,- e!en in the
e2ui!alent e0*ression ,to be *art of,- is not free fro$ a$bi#uit"- an is
$ae efinite onl" b" the con!entions iscusse in connection with the
conce*t of anal"sis) The 2uestion whether the *h"sical uni!erse itself
e0ists has not arisen) I ha!e- in fact- a!oie sa"in# that it e0ists +
which woul be an unfinishe sentence) Orinaril" it woul be
unnecessar" to be so *articular) The e0istence or non+e0istence of
thin#s is a *ri$iti!e for$ of thou#ht4 an- if I ha use the ter$- it
woul $ean no $ore than that I was forcin# our obser!ational
knowle#e into such a fra$e-
5'
as it is force into se!eral other fra$es
22
No4 "ou ha!e not cau#ht $e this ti$e) The critics inti$iate $e 1ust as $uch- whether
*hiloso*h" concees to the$ ,real e0istence, or not)
23
If we wish the assertion to $ean $ore than the e0*ression of a *ri$iti!e for$ of
that we ha!e iscusse) =nowin#- howe!er- that as *hiloso*hers we
$ust seek to #et behin these for$s of thou#ht- I ha!e thou#ht it best
in this book to a!oi introucin# it e!en te$*oraril")
II
It is an a!anta#e of the e*iste$olo#ical a**roach that the 2uestion of
attributin# a $"sterious *ro*ert" calle ,e0istence, to the *h"sical
uni!erse ne!er arises) Set $e re$in "ou a#ain of the *osition) Our
startin# *oint is a *articular bo" of knowle#e) 6e ha!e no nee to
efine knowle#e + to iscuss the e0act sco*e of the ter$) 6hat is
re2uire is a s*ecification of the *articular collection of knowle#e or
alle#e knowle#e which is to be the the$e of iscussion) .roal" this
is taken to be whate!er is recei!e as knowle#e within the o$ain of
*h"sical science accorin# to the $ost u*+to+ate conclusions) In
accorance with a su**ose necessit" of thou#ht this knowle#e has
been for$ulate as a escri*tion of a *h"sical uni!erse) That is how
the *h"sical uni!erse co$es into the iscussion) That- I think- is all that
is neee to tell "ou 1ust what the *h"sical uni!erse is4 an "ou woul
not know an" $ore about it if I ae the unfinishe sentence ,the
*h"sical uni!erse is an entit" which isO),- or e!en if I were so
heteroo0 as to sa" ,the *h"sical uni!erse is an entit" which isn;tO),)
I ha!e referre also to an ob1ecti!e uni!erse which cannot be ientifie
with the uni!erse of which the abo!e+$entione bo" of knowle#e
for$s a escri*tion) The latter uni!erse is- as we ha!e seen- *artiall"
sub1ecti!e an *artiall" ob1ecti!e) So$e will *erha*s sa" that until
recentl" ,*h"sical uni!erse, was alwa"s unerstoo to $ean the
ob1ecti!e uni!erse- an the ter$ shoul still be use in that sense) It
woul then be necessar" to raw a istinction between the *h"sical
uni!erse an the uni!erse of *h"sics- i)e) the uni!erse escribe in
*h"sics) 6hen a ter$ has been associate with se!eral conce*tions
which are foun to be conflictin#- it is alwa"s a $oot *oint which of
the$ shoul ha!e efinin# force) 8oubtless the ter$ ,*h"sical
uni!erse, was for$erl" intene to refer to so$ethin# which
thou#ht- we sa" ,reall" e0ists,)
*ossesses- alon# with other characteristics- *ure ob1ecti!it"4 but to test
whether ob1ecti!it" is *art of its efinition- I $ust ask- ,6ill "ou stick to
this efinition whate!er ha**ens@, Su**ose- for e0a$*le- it shoul
turn out that there is nothin# *urel" ob1ecti!e in e0*erience e0ce*t
3o4 will "ou a#ree that when "ou sai ,*h"sical uni!erse, "ou were
reall" all the ti$e referrin# to ,3o,@ I o not think "ou will) .ut that
$eans that the ulti$ate efinition- which "ou are *re*are to ahere to
in all circu$stances- is eter$ine b" other consierations) Ob1ecti!it"
is not a efinin# *ro*ert"- but a *ro*ert" which we ha 9wron#l"- as it
ha**ens: e0*ecte the thin# efine b" other *ro*erties to *ossess)
That bein# so- we $ust e0a$ine o*en+$inel" whether the *h"sical
uni!erse *ossesses ob1ecti!it"- an not tr" to s$u##le in the ob1ecti!it"
as *art of its efinition)
Ha!in# re1ecte a efinition *ostulatin# ob1ecti!it"- we co$e back to
the e*iste$olo#ical efinition that I ha!e been followin#) The *h"sical
uni!erse is the worl which *h"sical knowle#e is for$ulate to
escribe4 an there is no ifference between the *h"sical uni!erse an
the uni!erse of *h"sics)
It woul be a serious ob1ection to this efinition if it coul be sai that it
in!ol!es usin# the ter$ in a sense ifferent fro$ that in which it is use
in orinar" s*eech) Orinar" s*eech oes not concern itself $uch with
the uni!erse4 but the sa$e consieration a**lies to *arts of the
uni!erse- na$el" *h"sical ob1ects) 8oes the scientist $ean b" a
,*h"sical ob1ect, what the *lain $an woul take it to $ean@ For
e0a$*le- when we #i!e a scientific escri*tion of a chair accorin# to
the $ost $oern *h"sical theories- are we escribin# the ob1ect which
in e!er"a" life is calle a chair@
So$e of the *ure *hiloso*hers en" that the scientific escri*tion
a**lies to the ob1ects which in orinar" s*eech are calle *h"sical
ob1ects) Their o*inion is !oice b" Prof) Stebbin#? ,He Qthe *h"sicistR
has ne!er been concerne with chairs- an it lies be"on his
co$*etence to infor$ us that the chairs we sit u*on are abstract),
5C

24
L) S) Stebbin#- Philoso*h" an the Ph"sicists- *) 5F(
Ph"sicists are not concerne with chairsH Are we reall" e0*ecte to
take this sittin# own@
Let us first notice that the *hrase ,chairs we sit u*on, as nothin# to
the ter$ ,chair,) For what sits on the chair is the bo"4 an if we ha!e
to iscri$inate the scientific chair- i)e) the ob1ect- not reall" a chair-
which the *h"sicist escribes- fro$ the fa$iliar chair- we $ust also
iscri$inate the scientific bo"- i)e) the ob1ect- not reall" a bo"- which
the *h"sicist escribes- fro$ the fa$iliar bo") So when we sit on a
chair- the fa$iliar bo" sits on a fa$iliar chair- an the scientific bo"
sits on a scientific chair) An if there is an abstract bo" it oubtless
*erfor$s an abstraction of sittin# on an abstract chair)
I o not ob1ect to the *hiloso*her conte$*latin# a construct of sensor"
2ualities which is not to be ientifie with the ob1ect escribe in
*h"sics) .ut when he clai$s that it is this *hiloso*hical chair an not
the scientific chair that the orinar" $an refers to- he is self+ecei!e)
For if he were ri#ht- wh" is it that a Trans*ort Co$*an"- wishin# to
i$*ro!e its arran#e$ents for seatin#- consults a *h"sicist who is not
concerne with the chairs we sit u*on- instea of a *hiloso*her who
is@
If the *h"sicist is not concerne with chairs- the astro*h"sicist is not
concerne with stars) There is one Professor of Astro*h"sics- Prof)
8in#le- who has not been afrai to reco#ni/e this lo#ical conclusion?
He Q.ertran RussellR has $isse the essential *oint that *h"sics is
not concerne at all with *lanets),
5<
Prof) 8in#le- like Prof) Stebbin#-
has ro**e the outlook which eter$ines the fa$iliar use of wors-
an has stra"e into a worl where $en look at thin#s in the wa"
*hiloso*hers woul ha!e the$ look- an lan#ua#e is i!erte to
escribe those thin#s which *hiloso*hers consier $ost worth
attention)
Twinkle- twinkle little star
How I woner what "ou areH
25
H) 8in#le- Throu#h Science to Philoso*h"- *) &')
.ut the chil is not wonerin# whether the star is ,a function of sense
ata, 9Russell: or ,a co$$onsense #rou*in# of e0*eriences, 98in#le:)
He is wonerin# how bi# it is an how far awa"- what kee*s it fro$
ro**in# own- whether it is $ae of #ol- whether it is lit b" electricit")
6hen he woners what a star is- it is 8in#le the astro*h"sicist- not
8in#le the *hiloso*her- who can #i!e hi$ the infor$ation for which he
thirsts) One 2uestion leas to another4 an in the reconite treatises of
*h"sics we are still askin#- an now an a#ain answerin#- the
unceasin# flow of 2uestions) 6hen a *h"sicist tells us b" his scientific
escri*tion what a star is- he is still answerin# the chil;s 2uestion4 but
the chil is a little oler)
It is true that the chil + or the orinar" $an + oes not know that what
he is reall" wonerin# about is the #rou* structure of the star) .ut
when we ole out to hi$ bit b" bit- in lan#ua#e aa*te to the sta#e he
has reache- *articulars about the #rou* structure- he reco#ni/es that
the infor$ation is an answer to his half+for$e 2uestions) An his
curiosit" is not satisfie until he has e0tracte in this wa" e!er"thin#
we can tell hi$ about the #rou* structure- or until he fis we ha!e
beco$e so unintelli#ible that it is useless to 2uestion us an" lon#er)
The *h"sicist hi$self $a" be *artl" to bla$e for the sus*icion that he
is talkin# about so$ethin# ifferent fro$ what the orinar" $an woul
$ean b" the *h"sical uni!erse an *h"sical ob1ects4 for he has not
alwa"s been scru*ulous in his a**ro*riation of fa$iliar wors) .ut
there has been no $isa**ro*riation in this case)
The *h"sical uni!erse as escribe in this book $a" see$ re$ote
fro$ the uni!erse orinaril" conte$*late because of the e$*hasis
that has been lai on its sub1ecti!it") .ut the sus*icion that the ter$ is
bein# $isuse arises fro$ a $isunerstanin#) It has been $" s*ecial
task in these lectures to stu" the sub1ecti!e ele$ent in the *h"sical
uni!erse- so that the ob1ecti!e ele$ent has been ke*t out of the
li$eli#ht4 but- as I ha!e *ointe out- the ob1ecti!e ele$ent bulks lar#el"
in the uns"ste$ati/e *art of our knowle#e which also for$s *art of
the escri*tion of the *h"sical uni!erse) 6hen we *ut off the blinkers
of the s*ecialist- an !iew the two ele$ents to#ether in *ro*er
*ers*ecti!e- we shall fin that the" for$ a uni!erse not unacce*table
as an answer to the ele$entar" 2uestions which arise out of fa$iliar
e0*erience- as well as to the $ore reconite scientific 2uestions)
I ho*e it is now sufficientl" clear that I re*uiate an" $eta*h"sical
conce*t of ,real e0istence,4 an I $a" without an#er introuce a
structural concept of e)istence which has a $athe$aticall" efine
sense) It is a *ri$iti!e for$ of thou#ht that thin#s either e0ist or o not
e0ist) I su**ose that e!er"one catches hi$self thinkin# that wa"-
thou#h he woul fin it i$*ossible to cr"stalli/e the conce*tion of
e0istence referre to) Let us set asie the ha/" #eneral conce*t- an
consier onl" the structure of the conce*t) Its !er" si$*le structure is
re*resente b" a s"$bol which contains in itself two *ossibilities +
e0istence an non+e0istence) In $athe$atical lan#ua#e it is a s"$bol
B with two ei#en!alues- which are $ost con!enientl" taken to be %-
stanin# for e0istence- an D- stanin# for non+e0istence) The s"$bol
B $ust satisf" the e2uation B
5
+BJD- since that is a 2uaratic e2uation
which has 1ust the two solutions BJ% an BJD) Another wa" of writin#
the sa$e e2uation is B
5
JB) 6e call a s"$bol which is e2ual to its own
s2uare an ie$*otent s"$bol)
The structural conce*t of e0istence is re*resente b" an ie$*otent
s"$bol)
In #eneral it re2uires $ore that one ele$ent to for$ a structure4 an
e0istence is the onl" e0a$*le of a structure *ossesse b" a sin#le
ele$ent) It will be re$e$bere that structure first a**ears when the
o*eration E which turns the o*eration P into the o*eration I is not a
new kin of o*eration but one of the set of o*erations alrea" efine)
6hen there is onl" one o*eration B to consier- this conition for
structure e#enerates into ,the o*eration which turns the o*eration B
into the o*eration B is the o*eration B,) That is what the ie$*otent
conition B)BJB asserts) Thus- if we re*resent the ulti$ate ele$ents of
our anal"sis b" ie$*otent s"$bols- we e0*ress the for$ of thou#ht
that- a*art fro$ its structural association with other ele$ents- all that
can be sai about an ele$ent is that it e0ists- or alternati!el" that it
oes not e0ist)
The entit" re*resente b" the si$*le e0istence s"$bol B is like a *oint
in that it ,has no *arts an no $a#nitue,4 for if it ha *arts- it woul be
*ossible to concei!e one *art e0istin# without the other- an each *art
woul re2uire an ine*enent e0istence s"$bol) The e0istence of the
whole- bein# e2ui!alent to the co+e0istence of the *arts- woul then
e*en on a co$bination of their se*arate e0istence s"$bols- an not
on a si$*le s"$bol B) The entit" in *h"sics which- like the *oint in *ure
#eo$etr"- has no *arts is escribe as an elementary particle) At
*resent our ele$entar" *article has ,no $a#nitue,4 since $a#nitue
is relati!e an we ha!e introuce nothin# to relate it to) The
$a#nitues 9$ass 7- char#e e- an ran#e in nuclear *heno$ena:
which we attribute to an ele$entar" *article belon# to it- not
intrinsicall"- but on account of its relations to the rest of the uni!erse)
Obser!ation can onl" isclose the relations between entities4 an the
$ost ele$entar" relation we can consier is the relation between two
ele$entar" *articles with si$*le e0istence s"$bols B
%
an B
5
-
res*ecti!el") This relation onl" e0ists if both the *articles e0ist) 6e
therefore assi#n to it the ouble e0istence s"$bol B
%
0 B
5
- which will
ha!e the e0istence ei#en!alue % if both B
%
an B
5
ha!e the ei#en!alue -
an the non+e0istence ei#en!alue D if either or both ha!e the
ei#en!alue D)
A relation between two relations will e0ist onl" if both relations e0ist-
an it shoul accorin#l" be assi#ne a 2uaru*le e0istence s"$bol)
.ut that wa" leas to an or#" of e!er+e0*anin# notation)
5G
Our #oal is
a structure in which the relations of relations are re*resente b" the
sa$e set of s"$bols as the relations the$sel!es- so that the
conitions for a $athe$atical escri*tion in ter$s of #rou* theor" are
fulfille) The relations of relations will therefore ha!e the ouble
e0istence s"$bols of the si$*le relations with which the" are
26
Iuaru*le e0istence s"$bols are howe!er i$*ortant at a later sta#e because a
$easure$ent in!ol!es four entities 9*) %G(:)
ientifie)
It is to be re$e$bere that the ele$entar" *articles are not actual
ata) The atu$ is our obser!ational knowle#e- which- since it is of
co$$unicable character- is necessaril" a knowle#e of #rou*
structure4 an the ele$entar" *article is a *rouct of anal"sis of this
#rou* structure) The constituents of the #rou* are the o*erators which
we ha!e calle P- I- RO) 6e want now to e0*ress s"$bolicall" the
fact that P- I- R- O) are relations) The structural conce*t of a relation
is that it is so$ethin# whose e0istence is contin#ent on the e0istence
of two entities either of which coul e0ist or not e0ist) Thus- in
re*resentin# P- I- R- O b" ouble e0istence s"$bols- we are sa"in#
no $ore about the$ than is necessar" in orer that the" $a" be
concei!e as relations) Hitherto we ha!e talke !a#uel" about a #rou*
structure of P- I- R- O - but we can now s*ecif" the *articular
$athe$atical #rou* in!ol!e- na$el" the #rou* of the ouble
e0istence s"$bols) On in!esti#ation this is foun to be the sa$e #rou*
as the #rou* of rotations in si0+i$ensional s*ace- to which reference
has alrea" been $ae 9*) %CD:)
6h" si0 i$ensions@ E!en if we inclue ti$e- the continuu$ of
location re!eale in obser!ational e0*erience has onl" four
i$ensions) .ut we are ealin# with *articles- not #eo$etrical *oints4
an the ouble e0istence s"$bol re*resents a relation between two
*articles $ore co$*le0 than 9but incluin#: the *urel" #eo$etrical
relation between the two *oints that the" occu*") Not onl" is the
#reater co$*le0it" irectl" *reicte b" this $etho- but further
e!elo*$ent of the theor" shows how it will $anifest itself in
obser!ational e0*erience) The aitional co$*lication corres*ons to
the *lane of s*in an si#n of char#e of an ele$entar" *article + which
ha!e no counter*arts in a #eo$etrical *oint)
The ne0t ste* in the theoretical e!elo*$ent is conitione b" the fact
that the ele$ents of structure in the uni!erse are e0ceein#l"
nu$erous) 6e ha!e been conte$*latin# a structure which oes not
,e0ist, unless e!er" one of its ele$ents e0ists) .ut- when the nu$ber
of ele$ents is !er" lar#e- we ha!e a so$ewhat ifferent conce*tion of
the e0istence of the structure- accorin# to which the *resence of two
or three ele$ents $ore or less is not worth troublin# about) Since no
one of its *articles is now essential to the e0istence of the structure-
we $ust #i!e the structure an e0istence s"$bol ine*enent of the
e0istence s"$bols of its ini!iual *articles)
To e0*ress this chan#e of our for$ of thou#ht- it $ust be notice that it
u*licates the e0istence of each ele$ent4 as a contributor to the
structure it is concei!e to e0ist continuall"- but as an ine*enent
entit" it $a" or $a" not be *resent) Let us e0*ress $athe$aticall" the
nature of the ine*enent e0istence that we attribute to it) 6e $ust
first u*licate its orinar" e0istence s"$bol- obtainin# 5B) 6e then
abstract the *art which re*resents its e0istence as a contributor to the
structure4 this is not a *otentialit" of e0istence or non+e0istence for
which s"$bolic re*resentation is re2uire- but unconitional e0istence
which we ha!e a#ree to re*resent b" the nu$ber %) The re$ainer
5B+% re*resents the ine*enent e0istence of the ele$ent) 6e thus
obtain the conce*tion of an ine*enentl" e0istin# *article-
re*resente b" an independent e)istence symbol =J5B+%) The
ei#en!alues of = are % for e0istence an +% for non+e0istence) The
absence of the *article is now- not a $ere ne#ation 9D:- but a hole 9+%:
occurrin# in- or ae to- the structure)
A*art fro$ its e0istence or non+e0istence- the onl" characteristic
*ossesse b" the ele$entar" *article is its relation to the whole
structure) In our new *oint of !iew the relation co$es first) That is to
sa"- we classif" the ifferent *ossible relations of an ini!iual *article
to the whole structure- an then to each *ossible relation we assi#n an
ine*enent e0istence s"$bol = which inicates whether a *article
ha!in# that relation e0ists or not) The $ore usual ter$inolo#" is to call
the relation to the whole structure a state- an to escribe the state as
occupied or unoccupied) 6e $a" therefore call = an occupation
symbol instea of an ine*enent e0istence s"$bol)
In this $oe of re*resentation the ,whole structure, *la"s a si$ilar *art
to the #eoi in #eoes"- fro$ which the actual earth is obtaine b"
ain# or subtractin# $aterial at !arious *oints) Ha!in# re#ar to this
analo#" I *ro*ose to call the structure that is conte$*late as
continuall" e0istin#- the uranoi) This will istin#uish it fro$ other
structures we $a" ha!e occasion to consier) Each ine*enent
*article is ine*enent onl" because it has contribute half its
,e0istence, to the uranoi) This i$*lies that the nu$ber of *articles
conte$*late 9but not necessaril" e0istin#: has been fi0e at the start-
an the uranoi constructe accorin#l") 6e shall consier later how
this nu$ber is eter$ine)
IA
For the further ste*s in the e!elo*$ent of theoretical *h"sics fro$
this e*iste$olo#ical founation reference $ust be $ae to $"
$athe$atical treatise)
5F
I ha!e trie here- firstl" to e0hibit in etail the
*rinci*les b" which $athe$atics first #ets a #ri* on the sub1ect- so that
we $a" unerstan the e0act relation of the s"$bolic $athe$atical
fra$e of thou#ht to our non+$athe$atical conce*tion of thin#s4 an
seconl" to carr" the e!elo*$ent far enou#h to show that the
$athe$atical $aterial obtaine in this wa" is not tri!ial) Neither at this
sta#e nor at a later sta#e is there an"thin# arbitrar" in the course of
the e!elo*$ent- *ro!ie we a$it that it $ust be such as to e0*ress
knowle#e accorin# to certain for$s of thou#ht which we reco#ni/e
as en#raine in our outlook)
I $ust- howe!er- $ention a conce*tion which arises in later
e!elo*$ents- because there will be occasion to use it in the ne0t
cha*ter) E!er" state is ieall" istinct- an it is associate with a
ifferent s"$bol = which can enote onl" occu*ation or non+
occu*ation4 but in *ractice we so$eti$es i#nore the $ore $inute
istinctions an run a #reat nu$ber of states into one) Thus we often
ha!e to o with conense states- for$e b" runnin# to#ether n
ele$entar" states- an ca*able therefore of bein# occu*ie b" an"
nu$ber u* to n *articles) To escribe the state of occu*ation of a
27
Relati!it" Theor" of Protons an Electrons 9%&'G:- es*eciall" Cha*ter @@
conense state we $ust associate with it a carinal s"$bol =;- whose
ei#en!alues are the inte#ers fro$ +n to n an re*resent the
e0cess of the nu$ber of occu*ie states o!er n)
5(
The e2uation
satisfie b" =; is
=;9%+=;
5
: 9%+=;
5
N5
5
: 9%+=;
5
N'
5
:O) 9%+=;
5
Nn
5
:JD
since its roots are the re2uire ei#en!alues) >suall" n is so lar#e that it
is consiere infinite- an the left+han sie of the e2uation is then an
infinite *rouct which is well known to be e2ual to sin=;) Thus the
characteristic e2uation
5&
of =; is sin=;JD- which is satisfie b" e!er"
inte#er- *ositi!e an ne#ati!e- incluin# /ero) This is the for$ in which
the e2uation is use in current 2uantu$ theor"4 but it is to be
re$e$bere that it is an a**ro0i$ation- an in the strict for$ of the
e2uation there is a hi#hest inte#er)
6e shall #ain so$e iea of the $a#nitue of the work that re$ains
before the *resent e!elo*$ent of theor" on a *urel" e*iste$olo#ical
basis $er#es into current *h"sical theor"- if we reali/e that we ha!e
not "et introuce measurement4 so that the orinar" *h"sical
2uantities which result fro$ $easure$ent ha!e not "et a**eare) In
Cha*ter A we s*ent $uch ti$e o!er the efinitions of len#th an ti$e
inter!al which are the basis of all other *h"sical $easure$ents) This
s"ste$ of e0*ression of our knowle#e b" $easure *h"sical
2uantities has "et to be linke on to the $ore *ri$iti!e e0*ression in
ter$s of #rou* structure)
6e are sai to ,obser!e, a relation between two entities4 but a
,$easure$ent, consists in co$*arin# such a relation with a stanar)
Thus a $easure$ent of len#th is a co$*arison of a relation of
e0tension between two entities in the s"ste$ uner obser!ation with a
relation of e0tension between two entities $arkin# the ens of our
28
6e ha!e taken n e!en) A**ro*riate $oifications are introuce if n is o) Half the
e0istence of each *article- i)e) Sn alto#ether- is re#are as inclue in the uranoi- an
the e0cess o!er Sn is therefore the ine*enent e0istence associate with the conense
state)
29
In the sa$e sense in which B
5
JB is the characteristic e2uation of B)
ao*te stanar) A $easure therefore in!ol!es four entities- an at its
first a**earance is associate wit a 2uaru*le e0istence s"$bol) It is-
howe!er- conce*tuall" transferre to the relation that is co$*are with
the stanar4 an so$e $easures 9e)#) a $easure of $ass: are e!en
transferre to a sin#le entit" chosen fro$ the four) The for$al
treat$ent of this transference leas to !er" e0tensi!e ra$ifications of
*h"sical theor") Here we shall onl" note that the 2uaru*le e0istence
s"$bols- fro$ which we turne asie near the be#innin# of this
iscussion- *la" an i$*ortant *art in the later sta#es because of their
irect association with the *rocess of $easure$ent) This- howe!er- is
not the be#innin# of an infinite re#ression) The conce*tual fra$e in
which *h"sical knowle#e is e0*resse is such as to in!ol!e si$*le-
ouble- an 2uaru*le e0istence s"$bols corres*onin# res*ecti!el"
to entities- relations- an $easures4 but it has no funa$ental use for
octu*le an hi#her s"$bols)
Fro$ the association of $easure$ent with four entities we are le-
without further in!esti#ation- to e0*ect that the nu$ber C will in so$e
wa" $ake itself e!ient in the worl+*icture which e$boies the
results of our $easure$ents) It is the see fro$ which s*rin# those
ol" assorte *ure nu$bers which we call constants of nature 9*) <(:)
This conclusion in itself tells us !er" little- an #i!es no warrant for
nu$erolo#ical s*eculations) I belie!e that the nu$ber C introuce in
this wa" is actuall" res*onsible for the four i$ensions of s*ace+ti$e-
but only indirectly) In an actual calculation the nu$ber of i$ensions of
s*ace+ti$e is reache b" the route
9C)':N9%)5: +% +% J C
an it is a coincience that the nu$ber we en with is the nu$ber we
start with) 7an" threas $ust be wo!en to#ether before we $ake
an"thin# out of this s$all be#innin#)
One cannot $ake bricks without straw) The fore#oin# iscussion will
*erha*s ser!e to show where I #et the straw for the bricks which I
$ake + or ro* + in the co$*lete theor")
CHAPTER EI
THE PHYSICAL >NIAERSE
I
I belie!e there are %<- FCF- F5C- %'G- 5F<- DD5- <FF- GD<- G<'- &G%-
%(%- <<<- CG(- DCC- F%F- &%C- <5F- %%G- FD&- 'GG- 5'%- C5<- DFG- %(<-
G'%- D'%- 5&G *rotons in the uni!erse an the sa$e nu$ber of
electrons)
In this total a *ositron is counte as $inus one electron4 so that the
creation an annihilation of electrons an *ositrons in *airs- which is
continuall" ha**enin#- oes not affect the total) How $esotrons are to
be counte 9if at all: in the total cannot be state until we know $ore
about these *articles) Neutrons an nuclei are- of course- counte
accorin# to the nu$ber of *rotons an electrons co$*osin# the$)
It ou#ht not to be necessar" to re+state what has been well known
since %&5D- that each nucleus is co$*ose of a efinite nu$ber of
*rotons an electrons ascertainable fro$ its ato$ic wei#ht an ato$ic
nu$ber) .ut a few "ears a#o there was a cra/e for en"in# this- which
s*rea so wiel" that the $ost har$less reference to it still e!okes
criticis$ + as thou#h it betra"e i#norance of recent ieas of nuclear
structure) The state$ent refers to the co$*osition- not the structure- of
the nucleus) Onl" those *ossesse of a sin#ularl" naM!e i$a#ination
coul e!er ha!e su**ose it $eant that a close ins*ection of the
nucleus woul re!eal electrons stickin# in it like currants in a *uin#)
A nucleus is co$*ose of *rotons an electrons in the sa$e wa" that
an o$elette is co$*ose of e##s4 that is to sa"- when the o$elette
a**ears on the table- there are fewer e##s in the larer) The *roton+
electron co$*osition assi#ne to the !arious nuclei is a$*l" confir$e
b" trans$utation e0*eri$ents which a**l" irectl" the ,o$elette,
criterion of co$*osition) I think the cra/e for a $eta*h"sical
*ronounce$ent- that e##s an electrons cease to e0ist when the" are
scra$ble- has now ie awa"4 but it was in an" case an irrele!anc")
Co$in# back to the nu$ber of *rotons an electrons- I ha!e state $"
belief) One belie!es with !ar"in# e#rees of confience) 7" belief that
I know the e0act nu$ber of *rotons an electrons in the uni!erse oes
not rank a$on# $" stron#est scientific con!ictions- but I shoul
escribe it as a fair a!era#e sort of belief) I a$- howe!er- stron#l"
con!ince that- if I ha!e #ot the nu$ber wron#- it is 1ust a sill" $istake-
which woul s*eeil" be correcte if there were $ore workers in this
fiel) In short- to know the e0act nu$ber of *articles in the uni!erse is
a *erfectl" le#iti$ate as*iration of the *h"sicist)
C"nical *ersons will sa" that it is rather a safe calculation- because no
one will e!er count the *articles an show that $" reckonin# is- sa"- %C
out) I a$ #oin# so far to 1ustif" the c"nics as to a$it that- if I ha
thou#h there was the least chance that an"one woul count the
*articles- I woul ne!er ha!e *ublishe $" calculation) .ut $" reason
is not the one "ou sus*ect) The reason is that in the calculation I
e$*lo"e a t"*e of anal"sis a**ro*riate onl" to uncountable *articles4
so that if an"one *ersuae $e that he coul actuall" count the
*rotons an electrons- he woul *ersuae $e that $" calculation was
on a wron# basis- an I shoul withraw it without waitin# to hear
whether his count isa#ree)
Let us see wh" *rotons an electrons are uncountable) It is not $erel"
because there are so $an" of the$) Iuantu$ *h"sicists tell us that an
electron is not efinitel" in one *lace but is s$eare o!er a *robabilit"
istribution4 also that electrons are inistin#uishable fro$ one another)
That is not !er" *ro$isin# $aterial for countin#) There is nothin# to
re$e$ber about the electron "ou last counte + neither its *osition nor
an" istin#uishin# $ark) So how can "ou know whether the ne0t "ou
notice is a new one or one alrea" counte@ ." the uncertaint"
*rinci*le- the $ore closel" "ou *in own its *osition at one instant- the
$ore uncertain "ou are of its !elocit" an where it will turn u* ne0t)
6hen "ou retire to rest- as a !ariant to countin# shee* in a #reen fiel-
*erha*s "ou $a" like to tr" countin# electrons in a *robabilit"
istribution)
The nature of electrons $akes it i$*ossible to count the$ e0ce*t in
!er" s*ecial cases4 an the sa$e a**lies to *rotons) Ne!ertheless-
*h"sicists tell us confientl" the a**ro0i$ate nu$ber of electrons
9about G)%D
5'
: in a #ra$ of h"ro#en) Ob!iousl" the" ha!e not counte
the$) That in itself oes not call for criticis$- because we reco#ni/e
that it is le#iti$ate to obtain such a result in an inirect wa") .ankers
eter$ine 9or use to eter$ine: the nu$ber of so!erei#ns b"
wei#hin#- knowin# that this woul #i!e the sa$e result as the $ore
laborious *rocess of countin#) .ut can it be sai that the inirect
*roceure use b" *h"sicists to eter$ine the nu$ber of electrons in
a #ra$ of h"ro#en #i!es the sa$e result as woul be obtaine b"
actuall" countin# the$@ Clearl" not4 because we ha!e 1ust seen that
the" cannot be counte + actuall" countin# the$ woul ha!e #i!en no
result at all)
E!er" ite$ of *h"sical knowle#e is an assertion of the result of an
obser!ational *roceure actual or h"*othetical) 6hen knowle#e is
clai$e of the nu$ber of *rotons an electrons in a #ra$ of h"ro#en-
the obser!ational *roceure referre to cannot be countin#) It $ust be
knowle#e of the result of so$e other *roceure b" which an inte#ral
nu$ber is affi0e to a s"ste$) 6e call this nu$ber the nu$ber of
*articles4 but it is the nu$ber reckone accorin# to a ,2uantu$
arith$etic, not base on the sa$e conce*tions as the P"tha#orean
arith$etic of countin#)
Perha*s we ou#ht to be an#r" with the 2uantu$ *h"sicist for eluin#
us) .ut it is i$*ossible not to a$ire the e!astatin# beaut" of
2uantu$ arith$etic- an the in#enuit" of the trick b" which it assi#ns
nu$ber to the uncountable)
The na$e ,arith$etic,- like ,#eo$etr", or e!en ,h"ro"na$ics,- $a"
be a**lie to a branch of *ure $athe$atics with its own efinitions
an a0io$s unconnecte with an"thin# in the *h"sical uni!erse4 but in
scientific conte0t these ter$s are to be unerstoo in their ori#inal
*ractical $eanin# as sciences ha!in# for their sub1ect $atter the
nu$berin# of *h"sical ob1ects- the $easure$ent of the worl an the
$o!e$ent of $aterial fluis) As branches of *h"sical science the" are
e$brace in the #eneral tren towars unification4 an as relati!it"
theor" has unifie #eo$etr" an $echanics- so 2uantu$ theor" has
e!en $ore auaciousl" unifie arith$etic an wa!e $echanics) 6e
ha!e alrea" seen 9*) F<: that it is necessar" to turn to 2uantu$ theor"
for a s*ecification of the $aterial stanar of len#th b" *ure nu$bers
onl") Relati!it" theor" is able to e0*ress knowle#e in ter$s of nu$ber-
but onl" because it has borrowe its stanar of len#th fro$ 2uantu$
theor"- an with it the 2uantu$ association of nu$bers with *h"sical
s"ste$s)
The ke" to the unification which has turne arith$etic into wa!e
$echanics is the carinal s"$bol =; which we introuce on *) %<F) =;
is a s"$bol satisf"in# sin =;JD- an its ei#en!alues are the whole set
of inte#ers *ositi!e an ne#ati!e) 6a!e $echanics thus brin#s the
inte#ers 9which for$ the whole $aterial of our orinar" arith$etic: into
its *ur!iew as the ei#en!alues of one of its s"$bolic o*erators)
Introuce in this wa" the inte#ers are conce*ts unassociate with the
*roceure of countin#) Countin#- if it is introuce at all- is efine in
ter$s of the ei#en!alues of =;- i)e) the inte#ers- instea of !ice !ersa)
The ste* which we $ake fro$ ' to C in countin# is a transition + a
2uantu$ 1u$* + of a *h"sical characteristic of a s"ste$ fro$ one
ei#en!alue to another) The transition of a s"ste$ fro$ a state of
threeness to a state of fourness is onl" one of $an" kins of 2uantu$
1u$* which the s"ste$ can uner#o- an is not ifferentiate fro$
other 2uantu$ 1u$*s in #eneral wa!e+$echanical theor")
The $aterial- or o*eran- on which the s"$bolic o*erators work is
calle the wa!e function of the s"ste$) 6hen su**lie with certain
$aterial- =; reuces to the nu$ber C4 we then sa" that the nu$ber of
*articles in the s"ste$ re*resente b" the o*eran is C4 an si$ilarl"
for other nu$bers) It shoul be unerstoo that this is not a s*ecial
s"ste$ of inter*retation in!ente for the s"$bol =;4 it is a brin#in# into
line of ,nu$ber of *articles, with the other *h"sical 2uantities
escri*ti!e of the s"ste$- which in wa!e $echanics ha!e each their
a**ro*riate s"$bols reucible to ifferent nu$bers 9not usuall"
inte#ers: accorin# to the wa!e function su**lie to the$ to o*erate
on) Often the $aterial su**lie + the wa!e function + is such that the
o*eratin# s"$bol is not reucible to an" nu$ber4 for that s"ste$ the
*h"sical 2uantit" re*resente b" the s"$bol has not a efinite !alue-
but there are $ethos of calculatin# an ,e0*ectation !alue, + a !alue
associate with so$e e#ree of uncertaint") This $a" ha**en also in
the case of the nu$ber of *articles- if the knowle#e re*resente b"
the wa!e function is sufficient onl" to #i!e a *robable esti$ate of the
nu$ber)
The fact that =;- if it reuces to a nu$ber at all- is alwa"s an inte#er
iscri$inates it fro$ $ost of the other s"$bolic o*erators) This
conition secures that we ne!er catch oursel!es sa"in# that the
nu$ber of *articles in a s"ste$ is *recisel" ') It is true that there are
other o*erators associate with a *h"sical s"ste$ which ha!e onl"
inte#ral ei#en!alues4 but that is because *articles are not the onl"
thin#s we can count + or rather for which we can e!ise a substitute for
countin#) It will be re$e$bere that ,ato$icit", e0tens to raiation
9*hotons: an to an#ular $o$entu$- as well as to $aterial *articles)
In swallowin# u* arith$etic- 2uantu$ theor" has a little o!erreache
itself) To co$*rise the whole of the inte#ers the carinal o*erator $ust
satisf" sin =;JD) .ut we ha!e notice that- althou#h this is the
e2uation usuall" e$*lo"e- it is onl" an a**ro0i$ation4 an if we use
the e0act e2uation the series of inte#ers re*resente b" =; sto*s at a
rather hi#h nu$ber which we will call N) There is therefore a istinction
between 2uantu$ arith$etic an P"tha#orean arith$etic) There is no
infinit" in 2uantu$ arith$etic- an the nu$bers sto* at a hi#hest
nu$ber N) So that the *rinci*le b" which the nu$ber calle ,the
nu$ber of *articles in a s"ste$, is assi#ne to the s"ste$ $akes it
i$*ossible to assi#n nu$bers hi#her than N- which o not occur in
2uantu$ arith$etic)
The cos$ical nu$ber N takes the *lace of infinit" in relati!istic
2uantu$ arith$etic in $uch the sa$e wa" that the !elocit" of li#ht
takes the *lace of infinite !elocit" in ele$entar" relati!it" theor") So far-
I ha!e not sai an"thin# as to the wa" in which N is eter$ine4 I ha!e
been concerne onl" with the settin# in which it a**ears) .ut now that
we ha!e ra##e to li#ht the nature of the e!ice b" which nu$ber is
assi#ne to the uncountable- I thin# it will be clear that in clai$in# to
eter$ine a priori the nu$ber of ele$entar" *articles in the uni!erse
we are not usur*in# a *rero#ati!e which has usuall" been ascribe to
the Creator of the uni!erse)
A further wor $a" be sai as to the uncountabilit" of electrons) It $a"
be ob1ecte that we actuall" o count electrons in a 6ilson cha$ber-
where their tracks are $ae !isible b" an in#enious e!ice) .ut how
$an" electrons are there in a 6ilson cha$ber@ So$ethin# of the
orer %D
5D
) An how $an" o we count@ A o/en or so) 6e count u* to
about twel!e an then sto*4 not because we are tire- but because
there is no wa" of continuin#) I o not think that conflicts with $"
state$ent that countin#- which I take to $ean s"ste$atic enu$eration-
is ina**licable to electrons)
In #eneral the onl" countable *articles are those which *ossess
e0ce*tionall" hi#h !elocities) It is true that b" e0a$inin# these we can
eri!e a ratio of $ass to nu$ber- which- if assu$e to hol also for
uncountable *articles- enables us to euce the nu$ber fro$ the
$ass) .ut I a$ not en"in# that the *h"sicists ha!e foun a rational
an consistent wa" of e0tenin# the efinition of nu$ber to
uncountable s"ste$s) The *oint is that the" have e0tene the
efinition) So that when I- obeientl" followin# their efinition- s*eak of
the nu$ber of *articles in the uni!erse- "ou $ust not think that I $ean
that there are N iscrete entities- *ut there b" the Creator- rea" to be
enu$erate)
II
The theoretical calculation of the cos$ical nu$ber N e*ens on the
fact that a $easure$ent in!ol!es four entities an is therefore
associate with a 2uaru*le e0istence s"$bol) Fro$ this it a**ears
that the cos$ical nu$ber $ust be the total nu$ber of ine*enent
2uaru*le wa!e functions- which is foun to be 5 0 %'G 0 5
5<G
) This is
the nu$ber of *rotons an electrons) The nu$ber of *rotons is %'G 0
5
5<G
- which is the nu$ber #i!en in full at the be#innin# of this cha*ter)
The followin# is- I think- a le#iti$ate wa" of re#arin# the construction
of this nu$ber) The nu$ber %'G is characteristic of the #rou* structure
of the 2uaru*le e0istence s"$bols4 an for that reason it also turns
u* in the theor" of the other nu$erical constants of nature 9the fine+
structure constant an the $ass+ratio:) The structure *attern is an
interlacin# of the relations of %'G ele$ents- which in the *resent
a**lication are ientifie with %'G conense states) The carinal
s"$bol =; associate with each conense state re*resents an
arith$etic in which the hi#hest inte#er is 5
5<G
) Finall" the nu$ber is
ouble because we start with half the *articles 9or half the ,e0istence,
of the *articles: incor*orate in the uranoi4 so that the ne#ati!e
inte#ers re*resentin# subtraction of *articles fro$ the uranoi $ust be
taken into account as well as the *ositi!e inte#ers re*resentin#
aition)
'D
The nu$ber 5)%'G)5
5<G
is associate with 2uaru*le e0istence
s"$bols) The corres*onin# nu$bers associate with ouble an
si$*le s"$bols are 5)%D)5
%G
an 5)')5
C
) The latter nu$ber is &G) 6e
foresaw 9*) %G&: that the nu$ber C- associate with $easure$ent-
woul be incor*orate in so$e for$ in our worl+*icture of the results
of $easure$ent- thou#h we reali/e that it $i#ht a**ear rather
is#uise) 6e now fin that one of its is#uises is the raisin# of the
nu$ber of *articles in the uni!erse fro$ &G to ')%C<)%D
F&
)
I ha!e tol "ou what I belie!e to be the true stor" of the cos$ical
nu$ber N) 6hat shoul we conclue fro$ it@
To *ut it cruel"- we ha!e ebunke N) It is not an enu$eration of a
30
Fro$ this *oint of !iew the ,hi#hest inte#er, in 2uantu$ arith$etic is 5
5<G
) To for$ N we
then a to#ether 5 0 %'G such nu$bers b" orinar" arith$etic)
crow of iscrete *articles constitutin# the ob1ecti!e uni!erse) Since it
is $erel" a nu$ber foiste on us b" 2uantu$ theor"- bein# associate
in an a *riori wa" with its $ethos of anal"sis- is it an" lon#er of
interest@ I o not think its scientific interest is at all affecte)
Intrinsicall" the nu$ber of *articles in the uni!erse- e!en if it were
#enuine- woul be a $atter of rather tri!ial curiosit") The nu$ber is
scientificall" i$*ortant because it kee*s cro**in# u* in $ore *rosaic
*roble$s) It fi0es the ratio of the electrical to the #ra!itational force
between a *roton an electron + a 2uantit" which *ractical *h"sicists
ha!e been at #reat *ains to eter$ine) Actuall" their eter$inations
!erif" the !alue we ha!e foun to about one *art in <DD) It also fi0es
the s*ee of recession of the istant nebulae which *rouces ,the
e0*ansion of the uni!erse,) The astrono$ical ata are rather rou#h-
but the" confir$ the calculate !alue of N to within 5< *er cent) It also
fi0es the ran#e of the *eculiar forces which #o!ern the e2uilibriu$ of
the ato$ic nucleus) The a#ree$ent with the e0*eri$ental !alue is
within % *er cent)
I ha!e sin#le out N for attention because- of all the knowle#e
co$*rise in funa$ental *h"sics- knowle#e of the nu$ber of
ele$entar" *articles ha see$e least likel" to be tainte with
sub1ecti!it") It was therefore *articularl" suitable for a test case) .ut the
sa$e sub1ecti!it" a**ears e!er"where- an is usuall" not so ifficult to
iscern) The whole sche$e of *h"sical law is ebunke- if "ou like to
*ut it that wa") .ut ebunkin# the laws of o*tics will not *ut out the
sun;s li#ht4 ebunkin# the law of #ra!itation will not *re!ent us fro$
fallin# own stairs4 ebunkin# the laws of ballistics will not *ut a sto* to
war) E!en if the $"ster" is torn fro$ the$- the laws of our se$i+
sub1ecti!e uni!erse are !ali in that uni!erse- an in the technical
isco!eries an in!entions of science will continue to bear fruit for
#oo or e!il)
The bi# nu$ber- #i!en at the be#innin# of this cha*ter- has so$ehow
#ot into the sche$e of *h"sics) 6ho was res*onsible for *uttin# it
there@ There are a nu$ber of sus*ects) 6e naturall" e0a$ine first
those who constructe the $etho of wa!e $echanics4 but I think the"
clear the$sel!es) Sus*icion rests $ore hea!il" on the $an who first
$ae electrons + we re1ect his *lea that he onl" foun the$) .ut he too
$ust be ac2uitte) In the en there see$s to be no *ossible !erict
but ,Natural Causes,) It is 1ust the *ri$iti!e for$s of thou#ht workin#
the$sel!es out- an clai$in# e!er"one who takes *art in the
e!elo*$ent of *h"sics as their tool) To the ar#u$ent that- e!en if
there has been no cul*able ne#li#ence- there $ust ha!e been a little
carelessness in lettin# a nu$ber like that sli* in- we can answer that
nu$bers- once the" #et into a sub1ect- ha!e a wa" of breein#- an
this nu$ber of (D i#its is- as it were- onl" the #ranchil of the
nu$ber &G)
A sli#ht reenin# of the li#ht of the istant #ala0ies was the first clue
to the cos$ical nu$ber) That is the traitional a posteriori $etho of
science) .ut to the obser!er of obser!ers- the e0act !alue of the
cos$ical nu$ber is i$*licit in his first #li$*se of an e0*eri$ental
*h"sicist?
I lifte u* $ine e"es a#ain- an looke- an behol a $an with a
$easurin# line in his han)
'%
III
Ei#hteen "ears a#o I was res*onsible for a re$ark which has often
been 2uote?
'5
It is one thin# for the hu$an $in to e0tract fro$ the
*heno$ena of nature the laws which it has itself *ut into
the$4 it $a" be a far harer thin# to e0tract laws o!er which it
has ha no control) It is e!en *ossible that laws which ha!e
not their ori#in in the $in $a" be irrational- an we can
ne!er succee in for$ulatin# the$)
This see$s to be co$in# true- thou#ht not in the wa" that then
31
Pechariah- ii- %)
32
Space' Time' and Gravitation- *) 5DD
su##este itself) I ha in $in the *heno$ena of 2uanta an ato$ic
*h"sics- which at that ti$e co$*letel" baffle our efforts to for$ulate a
rational s"ste$ of law) It was alrea" a**arent that the *rinci*al laws
of $olar *h"sics were $in+$ae + the result of the sensor" an
intellectual e2ui*$ent throu#h which we eri!e our obser!ational
knowle#e + an were not laws of #o!ernance of the ob1ecti!e
uni!erse) The su##estion was that in 2uantu$ theor" we for the first
ti$e ca$e u* a#ainst the true laws of #o!ernance of the ob1ecti!e
uni!erse) If so- the task was *resu$abl" $uch $ore ifficult than
$erel" reisco!erin# our own fra$e of thou#ht)
Since then $icrosco*ic *h"sics has $ae #reat *ro#ress- an its laws
ha!e turne out to be co$*rehensible to the $in4 but- as I ha!e
enea!ore to show- it also turns out that the" ha!e been i$*ose b"
the $in + b" our for$s of thou#ht + in the sa$e wa" that the $olar
laws are i$*ose) 7eanwhile a new situation in re#ar to laws of
ob1ecti!e ori#in has arisen- because the s"ste$ of *h"sics is no lon#er
eter$inistic) The totalit" of $in+$ae law oes not i$*ose
eter$inis$) It is in the uneter$ine beha!ior- for which roo$ is left
within the co$*lete sche$e of *h"sical law at *resent reco#ni/e- that
the #o!ernin# laws 9if an": of the ob1ecti!e uni!erse $ust a**ear)
Ei#hteen "earns ha!e therefore not brou#ht us an" nearer to a
for$ulation of the ob1ecti!e laws of #o!ernance4 the onl" ifference is
that what I then escribe as *ossibl" irrational beha!ior is now
escribe as uneter$ine beha!ior)
In current *h"sical theor" the uneter$ine ele$ent in the beha!ior of
a s"ste$ is treate as a $atter of chance) If there were serious
e!iations fro$ the law of chance- obser!ation an theor" woul not
a#ree) 6e $a" therefore sa" that it is a h"*othesis in *h"sics-
su**orte b" obser!ation- that there are no ob1ecti!e laws of
#o!ernance + unless chance is escribe as a law)
Ne!ertheless- if we take a wier !iew than that of *h"sics- I think it
woul be $isleain# to re#ar chance as the characteristic feature of
the ob1ecti!e worl) The enial of ob1ecti!e laws of #o!ernance is not
so $uch a h"*othesis of *h"sics as a li$itation of its sub1ect $atter)
8e!iations fro$ chance occur- but the" are re#are as $anifestations
of so$ethin# outsie *h"sics- na$el" consciousness or 9$ore
ebatabl": life) There is in a hu$an bein# so$e *ortion of the brain-
*erha*s a $ere s*eck of brain+$atter- *erha*s an e0tensi!e re#ion- in
which the *h"sical effects of his !olitions be#in- an fro$ which the"
are *ro*a#ate to the ner!es an $uscles which translate the !olition
into action) 6e will call this *ortion of the brain+$atter ,conscious
$atter,) It $ust be e0actl" like inor#anic $atter in its obeience to the
funa$ental laws of *h"sics which- bein# of e*iste$olo#ical ori#in- are
co$*ulsor" for all $atter4 but it cannot be ientical in all res*ects with
inor#anic $atter- for that woul reuce the bo" to an auto$aton
actin# ine*enentl" of consciousness) The ifference $ust
necessaril" lie in the uneter$ine *art of the beha!ior4 the *art of the
beha!ior which is uneter$ine b" the funa$ental laws of *h"sics
$ust in conscious $atter be #o!erne b" ob1ecti!e law or irection
instea of bein# wholl" a fiel of chance)
The ter$ ,law of chance, tens to $islea- because it is a**lie to
what is $erel" an absence of law in the usual sense of the ter$) It is
clearer to escribe the conitions b" reference to correlation) The
h"*othesis of current *h"sical theor"- which is confir$e b"
obser!ation of inor#anic *heno$ena- is that there is no correlation of
the uneter$ine beha!ior of the ini!iual *articles)
Accorin#l" the istinction between orinar" $atter an conscious
$atter is that in orinar" $ater there is no correlation in the
uneter$ine *arts of the beha!iors of the *articles- whereas in
conscious $atter correlation $a" occur) Such correlation is looke
u*on as an interference with the orinar" course of nature- ue to the
association of consciousness with the $atter4 in other wors- it is the
*h"sical as*ect of a !olition) This oes not $ean that- in orer to
e0ecute a !olition- consciousness $ust irect each ini!iual *article
in such a wa" that correlation occurs) The *articles are $erel" a
re*resentation of our knowle#e in the fra$e of thou#ht corres*onin#
to the conce*t of anal"sis an the ato$ic conce*t) 6hen we a**l" the
s"ste$ of anal"sis which #i!es this re*resentation- we cannot foresee
whether the resultin# *articles will ha!e correlate or uncorrelate
beha!ior4 that e*ens entirel" on the ob1ecti!e characteristics of
whate!er it is that we are anal"sin#) 6hen non+correlation is assu$e-
as is custo$ar" in *h"sics- it is assu$e as a h"*othesis) .ut- without
$akin# an" h"*othesis- we can sa" that correlation an non+
correlation are re*resentations in our fra$e of thou#ht of ifferent
ob1ecti!e characteristics4 an since non+correlation a$ittel"
re*resents the ob1ecti!e characteristic of s"ste$s to which the orinar"
for$ulae of *h"sics a**l"- correlation $ust re*resent another ob1ecti!e
characteristic which + since it is not characteristic of s"ste$s to which
the for$ulae of *h"sics a**l" + is re#are b" us as so$ethin#
,outsie *h"sics,)
In the iscussion of freewill *ro!oke b" the $oern *h"sical theories-
it has- I think- #enerall" been assu$e that- since the orinar" laws of
inor#anic $atter lea!e its beha!ior uneter$ine within a certain
narrow ran#e- there can be no scientific ob1ection to allowin# a !olition
of consciousness to ecie the e0act beha!ior within the li$its of the
aforesai ran#e) I will call this h"*othesis A) For an" s"ste$ on a
$olar scale the *er$itte ran#e is e0ceein#l" s$all4 an !er" far+
fetche su**ositions are necessar" to enable !olition- workin# in so
s$all a ran#e- to *rouce lar#e $uscular $o!e$ents) To obtain a
wier ran#e we $ust a$it correlation of the beha!ior of the *articles)
This is the theor" we ha!e been iscussin#- an will be calle
h"*othesis .) In for$er writin#s I ha!e a!ocate h"*othesis . $ainl"
on the #roun of the inae2uac" of h"*othesis A4 but in the *resent
$oe of a**roach h"*othesis . *resents itself as the ob!ious an
natural solution)
Althou#h leain# to the sa$e conclusion- $" earlier iscussions
''

were $arre b" a failure to reco#ni/e that h"*othesis A is nonsense4
so that I was $ore a*olo#etic than I nee ha!e been for #oin# be"on
it) There is no half+wa" house between rano$ an correlate
beha!ior) Either the beha!ior is wholl" a $atter of chance- in which
33
The Nature of the Physical World- **) '%D+'%<) New Pathways in Science- *) (()
case the *recise beha!ior within the Heisenber# li$its of uncertaint"
e*ens on chance an not on !olition) Or it is not wholl" a $atter of
chance- in which case the Heisenber# li$its- which are calculate on
the assu$*tion of non+correlation- are irrele!ant) If we a**l" the law of
chance to the tossin# of a coin- the nu$ber of heas in %DDD throws is
uneter$ine within the li$its- sa"- C<D to <<D) .ut if a coin+tossin#
$achine is use which *icks u* an throws the coin not entirel" at
rano$- the non+chance ele$ent is not a factor eciin# which
nu$ber between C<D an <<D will turn u*4 a correlation- or s"ste$atic
tenenc" in tossin#- $a" *rouce an" nu$ber of heas fro$ D to
%DDD)
The fallac" of h"*othesis A was that it assu$e the beha!ior to be
restricte b" the orinar" laws of *h"sics incluin# the h"*othesis of
non+correlation or ,law of chance,- an then to be further restricte 9or
ecie: b" a non+chance factor 9!olition:) .ut we cannot su**ose the
beha!ior to be restricte b" chance an non+chance 9non+correlation
an correlation: si$ultaneousl") The a**licabilit" of the law of chance
is a h"*othesis4 the a$ission that the beha!ior is not #o!erne solel"
b" chance enies the h"*othesis) So if we a$it !olition at all- we
$ust not for#et first to re$o!e the h"*othesis of chance if we ha!e
been a**l"in# it4 is *articular we $ust ro* the Heisenber# li$its
which a**l" onl" to non+correlate beha!ior) If !olition o*erates on the
s"ste$- it oes so without re#ar to the Heisenber# li$its) Its onl"
li$its are those i$*ose b" the funa$ental e*iste$olo#ical laws)
Our !olitions are not entirel" unconse2uential4 so that there $ust be
laws of so$e kin a**l"in# to the$ an connectin# the$ with other
constituents of consciousness- thou#h such laws are not e0*ecte to
be of the $athe$aticall" e0act t"*e characteristic of sub1ecti!e law)
Pri$aril" the s*here of ob1ecti!e law is the inter*la" of thou#hts-
e$otions- $e$ories an !olitions in consciousness) In controllin#
!olitions ob1ecti!e law controls also the correlations which are the
*h"sical counter*arts of !olitions)
Our *hiloso*h" has le to the !iew that in so far as we can se*arate
the sub1ecti!e an ob1ecti!e ele$ents in our e0*erience- the sub1ecti!e
is to be ientifie with the *h"sical an the ob1ecti!e with the
conscious an s*iritual as*ects of e0*erience )To this we now a- as
a hel*ful analo#" *ro!ie it is not *resse too far- that conscious
*ur*ose is the ,$atter, an chance the ,e$*t" s*ace, of the ob1ecti!e
worl) In the *h"sical uni!erse $atter occu*ies onl" a s$all re#ion
co$*are with the e$*t" s*ace4 but- ri#htl" or wron#l"- we look on it
as the $ore si#nificant *art) In the sa$e wa" we look on
consciousness as the si#nificant *art of the ob1ecti!e uni!erse- thou#h
it a**ears to occur onl" in isolate centers in a back#roun of chaos)
IA
I a$ about to turn fro$ the scientific to the *hiloso*hical settin# of
scientific e*iste$olo#") This is accorin#l" a suitable *lace at which to
$ake a co$*arison with the $ost co$$onl" acce*te !iew of
scientific *hiloso*h") The followin# state$ent is fairl" t"*ical?
That science is concerne with the rational correlation of
e0*erience rather than with the isco!er" of fra#$ents of
absolute truth about an e0ternal worl is a !iew which is now
wiel" acce*te)
'C
I think that the a!era#e *h"sicist- in so far as he hols an"
*hiloso*hical !iew at all about his science- woul assent) The *hrase
,rational correlation of e0*erience, has a sa!or of orthoo0" which
$akes it a safe #a$bit for a**lause) The re*uiation of $ore
a!enturous ai$s #i!es a co$fortable feelin# of $oest" + all the $ore
a#reeable if we fanc" that so$eone else is bein# tol off) For $" own
*art I acce*t the state$ent- *ro!ie that ,science, is unerstoo to
$ean ,*h"sics,) It has taken $e nearl" twent" "ears to acce*t is4 but
b" stea" $astication urin# that *erio I ha!e $ana#e to swallow it
all own bit b" bit) Conse2uentl" I a$ rather flabber#aste b" the li#ht+
hearte wa" in which this *ronounce$ent- carr"in# the $ost *rofoun
i$*lications both for *hiloso*h" an for *h"sics- is co$$onl" $ae
34
>nsi#ne re!iew- Phil! -ag!' !ol) 5<- *) (%C- %&'()
an acce*te)
I ha!e no serious 2uarrel with the a!era#e *h"sicist o!er his
*hiloso*hical cree + e0ce*t that he for#ets all about it in *ractice) 7"
*u//le is wh" a belief that *h"sics is concerne with the correlation of
e0*erience an not with absolute truth about the e0ternal worl shoul
usuall" be acco$*anie b" a stea" refusal to treat theoretical *h"sics
as a escri*tion of correlations of e0*erience an an insistence on
treatin# it as a escri*tion of the contents of an absolute ob1ecti!e
worl) If I a$ in an" wa" heteroo0- it is because it see$s to $e a
conse2uence of acce*tin# the belief- that we shall #et nearer to
whate!er truth is to be foun in *h"sics b" seekin# an e$*lo"in#
conce*tions suitable for the e0*ression of correlations of e0*erience
instea of conce*tions suitable for the escri*tion of an absolute
worl)
The state$ent e!ientl" $eans that the $ethos of *h"sics are
inca*able of isco!erin# fra#$ents of absolute truth about an e0ternal
worl4 for we shoul ha!e no ri#ht to withhol fro$ $ankin the
absolute truth about the e0ternal worl if it were within our reach) If the
laboratories- built an enowe at #reat e0*ense- coul assist in the
isco!er" of absolute truth about the e0ternal worl- it woul be
re*rehensible to iscoura#e their use for this *ur*ose) .ut the
assertion that the $ethos of *h"sics cannot re!eal absolute
9ob1ecti!e: truth or e!er fra#$ents of absolute truth- concees $" $ain
*oint that the knowle#e obtaine b" the$ is wholl" sub1ecti!e) Inee
it concees it far too reail"4 for the assertion is one that ou#ht onl" to
be $ae after *rolon#e in!esti#ation) As I ha!e *ointe out- sciences
other than *h"sics an che$istr" are not so li$ite in their sco*e) The
isco!er" of un$istakeable si#ns of intelli#ent life on another *lanet
woul be haile as an e*och+$akin# astrono$ical achie!e$ent4 it can
scarcel" be enie that it woul be the isco!er" of a fra#$ent of
absolute truth about the worl e0ternal to us)
=ee*in# to *h"sics- the co$$onl" acce*te scientific *hiloso*h" is
that it is not concerne with the isco!er" of absolute truth about the
e0ternal worl- an its laws are not fra#$ents of absolute truth about
the e0ternal worl- or- as I ha!e *ut it- the" are not laws of the
ob1ecti!e worl) 6hat then are the"- an how is it that we fin the$ in
our correlations of e0*erience@ >ntil we can see- b" an e0a$ination of
the *roceure of correlation of our obser!ational e0*erience- how
these hi#hl" co$*le0 laws can ha!e #ot into it sub1ecti!el"- it see$s
*re$ature to acce*t a *hiloso*h" which cuts us off fro$ all other
*ossible e0*lanations of their ori#in) This is the e0a$ination that we
ha!e been conuctin#)
The en of our 1ourne" is rather a bathos after so $uch toil) Instea of
stru##lin# u* to a lonel" *eak- we ha!e reache an enca$*$ent of
belie!ers- who tell us ,That is what we ha!e been assertin# for "ears,)
Presu$abl" the" will welco$e with o*en ar$s the toilworn tra!ellers
who ha!e at last foun a restin# *lace in the true faith) All the sa$e I
a$ a bit ubious about that welco$e) Perha*s the assertion- like $an"
a reli#ious cree- was intene onl" to be recite an a**laue)
An"one who believes it is a bit of a heretic)
CHAPTER EII
THE .E3INNIN3S OF =NO6LE83E
I
6e turn now to consier the relation between *h"sical knowle#e an
hu$an e0*erience as a whole) Two $oes of a**roach are o*en to
us?
9%: 6e $a" e!elo* our #eneral *hiloso*h" fro$ the be#innin#-
trustin# that the e0*erience #aine in an intensi!e stu" of one branch
$a" assist us to a ri#ht ecision on so$e of the 2uestions o!er which
*hiloso*hers ha!e been i!ie)
95: 6e $a" in2uire which of the e0istin# s"ste$s of *hiloso*h"
har$oni/es best with the conclusions reache in scientific
e*iste$olo#")
The first course brin#s us into the #a$e as players- the secon as
umpires)
If we choose the secon course- we 9as scientists: naturall" take the
!iew that a *hiloso*hical s"ste$ which is in isaccor with the results
of scientific e*iste$olo#" $ust be re1ecte) It is $ore likel" that there
is an error in the assu$*tions or lo#ic of the #eneral *hiloso*h" than
that the e*iste$olo#ical *rinci*les- whose conse2uences ha!e been
teste in innu$erable *ractical a**lications- are unsoun) .ut- lea!in#
asie this clai$ to a1uicate on ulti$ate truth- we $a" well
concentrate on a nearer ai$) If science is the stu" of the rational
correlation of e0*erience- the enea!or of the scientific *hiloso*her
$ust be to e0ten this rational correlation fro$ a li$ite fiel of
e0*erience to the whole of e0*erience) His task is to *ro!ie a #eneral
*hiloso*h" which a scientist can acce*t without throwing over his
scientific beliefs) If our scientific enli#hten$ent has not "et reache the
sta#e at which we are ri*e to recei!e *ure *hiloso*hical truth- it is
none the less ur#ent to inte#rate our thou#ht into a consistent
*hiloso*h" which #oes as far towars the truth as the *resent
li$itations of science *er$it)
The scientific intruer feels hi$self uner a isa!anta#e if he follows
the first course) Ostensibl" he is treatin# *roble$s which ha!e been
the life+stu" of hunres better e2ui**e on $ost *oints than hi$self)
His one a!anta#e- which 1ustifies the intrusion- is ke*t in the
back#roun- na$el" that scientific e*iste$olo#" #i!es hi$ a
foreknowle#e of certain of the conclusions to which the ar#u$ent
$ust lea- or at least of certain conclusions which it $ust contri!e to
a!oi) Consierations of caution therefore stron#l" incline $e to the
secon course4 but consierations of clearness co$*el $e to ao*t
the first) There see$s no esca*e fro$ the rule that to $ake clear a
s"ste$ of thou#ht one $ust be#in at the be#innin#) 8esire for clarit"
so$eti$es re2uires us to $ake clear thin#s which it woul be safer to
lea!e !a#ue4 it e0*oses to attack out*osts of our thou#ht which $a"
not be essential to the $ain *osition) 7" *ur*ose in this book is to
$ake certain s*ecific contributions to *hiloso*h" rather than to set
forth a co$*lete *hiloso*hic s"ste$4 but contributions cannot be left
sus*ene in vacuo or 9still worse: in an at$os*here hostile to the
scientific thou#ht in which the" ori#inate- an I therefore feel obli#e to
sketch in a *ossible back#roun for the$ which will- I ho*e- $ake their
*lace in #eneral *hiloso*h" better unerstoo)
It is- I think- ina!isable to tr" to escribe a scientificall" #roune
*hiloso*h" b" the labels of the oler *hiloso*hical s"ste$s) To acce*t
such a label woul $ake the scientist a *art" to contro!ersies in which
he has no interest- e!en if he oes not cone$n the$ as alto#ether
$eanin#less) .ut if it were necessar" to choose a leaer fro$ a$on#
the oler *hiloso*hers- there can be no oubt that our choice woul be
=ant) 6e o not acce*t the =antian label4 but- as a $atter of
acknowle#e$ent- it is ri#ht to sa" that =ant antici*ate to a
re$arkable e0tent the ieas to which we are now bein# i$*elle b"
the $oern e!elo*$ents of *h"sics)
Reference $a" also be $ae to another #eneral *hiloso*hical s"ste$-
na$el" lo#ical *ositi!is$) Our insistence that *h"sical 2uantities are to
be efine in such a wa" that the assertions of *h"sics a$it of
obser!ational !erification- $a" su##est an affinit" with lo#ical
*ositi!is$) The $eanin# of a scientific state$ent is to be ascertaine
b" reference to the ste*s which woul be taken to !erif" it) This will be
reco#ni/e as a tenet of lo#ical *ositi!is$ + onl" it is there e0tene to
all state$ents) 6hen it is li$ite- as here- to ite$s of *h"sical
knowle#e- it is in no sense a *hiloso*hical tenet4 it is onl" a brin#in#
into line of the lan#ua#e of theoretical an of e0*eri$ental *h"sics- so
that we $a" not clai$ the su**ort of obser!ation for assertions which
ha!e no obser!ational founation) If it were a #eneral characteristic of
knowle#e- it woul not be so useful to us in iscri$inatin# *h"sical
knowle#e fro$ other kins of knowle#e) 6e are therefore not
*articularl" *reis*ose to fa!or the $ore #eneral assertion of lo#ical
*ositi!is$ that the $eanin# of all non+tautolo#ical state$ents is to be
ascertaine in the sa$e wa"- na$el" b" reference to the *roceure of
!erif"in# the$)
II
The co$*arison with lo#ical *ositi!is$ for$s a useful o*enin# to an
in2uir" into the nature of other t"*es of knowle#e) If I sa" to "ou ,I a$
!er" tire,- "ou know what I $ean- because "ou ha!e felt tire
"ourself) You $a" atte$*t !erification of the state$ent b" lookin# out
for confir$ator" s"$*to$s in $" beha!ior4 but e!en if the s"$*to$s
*ro!ie an infallible test- the $eanin# of the state$ent is not
ascertainable b" reference to the$) The state$ent means ,I a$ !er"
tire,4 it oes not mean ,I a$ #oin# to "awn,)
It $ust be a$itte that the knowle#e con!e"e b" the state$ent is
li$ite + $uch $ore li$ite than at first a**ears) You know what I
$ean because I a**eal to a feelin# which "ou ha!e e0*erience
"ourself) .ut "ou know onl" "our own feelin# of tireness4 "ou cannot
know $ine) Your unerstanin# of $" $eanin# 9if "ou o unerstan
it: is a sympathetic understanding) S"$*athetic knowle#e- if we
ecie to a$it it as knowle#e- $ust be istin#uishe both fro$
irect knowle#e- such as we ha!e of our feelin#s- an structural
knowle#e- such as we ha!e of the *h"sical uni!erse)
Is it ri#ht to count s"$*athetic unerstanin# as knowle#e@ So far as
*h"sical science is concerne- the answer is uni$*ortant4 for we ha!e
seen that onl" a structural knowle#e of the sensations in
consciousness- whether our own or an"one else;s- is use in *h"sics4
an this can be freel" co$$unicate without a**eal to s"$*athetic
unerstanin# of the sensations) .ut if we are to see *h"sical science
in its ri#ht relation to other branches of hu$an thou#ht- it is necessar"
to co$e to a ecision about s"$*athetic knowle#e)
One *ossibilit" is to en" alto#ether the !aliit" of s"$*athetic
unerstanin#- treatin# it as a wa" of *ersuain# oursel!es that we
unerstan so$ethin# which we on;t) If so- the state$ent ,I a$ !er"
tire, $ust be entirel" without $eanin# for "ou4 for it oes not $ean
that I a$ e0*eriencin# your feelin# of tireness- nor oes it refer to the
*h"sical s"$*to$s of $" tireness which woul ha!e a $eanin# for
"ou) Its $eanin# + for the one *erson for who$ it has a $eanin# + then
see$s to fall within the rule of lo#ical *ositi!is$- na$el" that it is to be
ascertaine b" reference to the *roceure of !erif"in# it) Su**ose that
I beco$e uncertain? Is it a feelin# of tireness or a feelin# of re*letion
which is $akin# $e isincline for acti!it"@ The !erification woul- I
su**ose- consist in callin# u* fro$ $e$or" an ao*te stanar
feelin# of tireness an co$*arin# $" *resent feelin# with the
stanar) The !erification of this knowle#e of tireness is essentiall"
the sa$e as the !erification of knowle#e of len#th- e0ce*t that I alone
can *erfor$ the !erification)
.ut- before co$in# to a ecision- we shoul notice that a !er" si$ilar
ifficult" arises in re#ar to memory of our own feelin#s) 6hen I
e0a$ine the totalit" of $" knowle#e- I fin that *art of it consists of
irect awareness of $" feelin#s- but a #reat eal $ore consists of
$e$or" of $" feelin#s) The $e$or" is so$ethin# of which I a$
irectl" aware4 but- as an ob1ect of irect awareness- it is 2uite ifferent
fro$ the feelin# itself) No one can $istake the $e$or" of a toothache
for a toothache)
It is therefore incorrect to sa" that I ha!e irect awareness of $" own
sensations- unless the ter$ ,sensation, is confine to sensations
occurrin# at the *reset $o$ent + a li$itation contrar" to co$$on
usa#e) As re#ars for$er sensations- the ,I, that is asse$blin# its
knowle#e has lost the irect awareness that it once ha- an knows
the$ onl" b" $e$or") Thus- lea!in# asie a transitor" knowle#e of
what I feel at the $o$ent- the stable *art of $" knowle#e is a
knowle#e of those uller i$*ressions calle $e$ories of sensations-
or- as it is *erha*s clearer to call the$- re$e$bere+sensations) I a$
irectl" aware of re$e$bere+sensations4 but it is a uni!ersal for$ of
thou#ht that a re$e$bere+sensation is to be re#are- not as a
constituent of knowle#e i$*ortant in itself- but as an inirect
a**rehension of a *ast sensation of which I ha!e not irect
awareness) It is- in short- a s"$*athetic knowle#e of the *ast
sensation)
The ifference between this s"$*athetic unerstanin# of our own
*ast feelin#s an s"$*athetic unerstanin# of other *eo*le;s feelin#s
is lessene as we #o back to istant $e$ories) 7" $e$ories of
feelin#s #o back to those of a s$all bo" who see$s $uch $ore of a
stran#er to $e now than $an" of $" *resent ac2uaintances) I oubt
whether I reall" know what thin#s felt like- taste like- looke like to
hi$ an" better than I know what the" feel- taste- an look like to a $an
I ha!e 1ust been talkin# with)
If we en" all s"$*athetic knowle#e- our outlook beco$es not $erel"
soli*sistic but ultra+soli*sistic) 6e conte$*late a worl in which onl"
the $o$entar" self e0ists4 for we en" all knowle#e of a self
*recein# it) The earlier self- of which $e$or" tells us- is is$isse as
a construct of the su**ose s"$*athetic unerstanin# which
*ersuaes us we ha!e a knowle#e which we ha!e not) On the other
han- if we a$it s"$*athetic unerstanin# e!en of the li$ite kin
re2uire for the inter*retation of $e$ories- we a$it a thir kin of
knowle#e which is neither structural knowle#e nor irect awareness)
It oes not follow that our s"$*athetic unerstanin# of other *ersons;
feelin#s $ust also be a$itte to be #enuine knowle#e4 but a leain#
ob1ection to its *ossible #enuineness is o!errien)
Lookin# at the 2uestion broal"- I o not think we can en" a *lace in
the su$ of hu$an knowle#e to knowle#e a**rehene onl" b"
s"$*athetic unerstanin#) Since we reco#ni/e that there is a
sub1ecti!e ele$ent in knowle#e- it is necessar" to inicate clearl" the
sub1ect+*artner whose knowle#e we are consierin#) The sub1ect+
*artner in ,hu$an knowle#e, is not to be ientifie with the sub1ect+
*artner in ,$" knowle#e,4 his *recise s*ecification $ust e*en on
our conce*tion of hu$an nature) If we re#ar the s"$*athetic facult"
as a $orbi e0crescence- we are oubtless 1ustifie in re1ectin# the
knowle#e which it *retens to a**rehen4 as the blin $en in 6ells;s
stor"- who re#are si#ht as a $orbi irritation of the brain ue to the
isease conition of two soft e*ressions in the face- were 1ustifie
accorin# to their li#hts in re1ectin# the !isual knowle#e of the
stran#er) .ut how can I efine hu$an nature 9as istinct fro$ $"
*articular nature: unless I concee the !aliit" of $" su**ose
unerstanin# of other *ersons; $ins@ 6ithout the s"$*athetic
facult" which enables $e to reco#ni/e $"self- not as an ini!iual mei
generis- but as an ele$ent of a social co$*le0- the conce*tion of
,hu$an knowle#e, coul not arise4 an it woul therefore see$
illo#ical to re1ect this facult" in efinin# the e0tent of hu$an
knowle#e)
No one belie!es in soli*sis$- an !er" few e!en assert that the" o)
Those who are obsesse b" the wor ,e0istence, co$e so$ehow to
the conclusion that other consciousnesses besies their own e0ist4
that is to sa"- other consciousnesses can be the sub1ect of that
$"sterious sentence which the" ne!er finish) Those who ao*t the
e*iste$olo#ical a**roach take for their sub1ect $atter a knowle#e
which e$boies the e0*eriences of other ini!iuals on the sa$e
footin# as their own e0*erience) For$all" this is non+co$$ittal4 it is not
necessar" to assi#n reasons for choosin# a *articular the$e of stu")
.ut unoubtel" the choice is eter$ine b" a con!iction- akin to
reli#ious con!iction- that this co+o*erati!e knowle#e is the $ost worth
while) This con!iction is inconsistent with a soli*sistic outlook)
It woul be $eanin#less to attribute consciousness to another $an
without knowin# at all what we are attributin# to hi$) .ut
consciousness is not a structural conce*t escribable b" *urel"
structural knowle#e4 not is the consciousness that we attribute to
another $an an"thin# of which we ha!e irect awareness- since it is
not our own consciousness) It follows that- if our reco#nition of
conscious bein#s other than oursel!es has an" $eanin# at all- their
consciousness $ust be so$ethin# of which we ha!e a knowle#e
which is neither structural knowle#e nor irect awareness4 an an"
escri*tion of it $ust be e0*resse in ter$s of the thir kin of
knowle#e which we ha!e calle s"$*athetic unerstanin#) 6e
co$$onl" efine the consciousness of another *erson as so$ethin#
which has a #eneral kin of likeness to our own consciousness) .ut
we can scarcel" cou*le a #eneral likeness with co$*lete issi$ilarit"
of etail + with the assertion that there is nothin# in another $an;s
consciousness that is not co$*letel" $isre*resente b" its su**ose
counter*art in our own) It see$s necessar" therefore to cou*le with
s"$*athetic unerstanin# of another consciousness so$e $easure
of s"$*athetic unerstanin# of the feelin#s belon#in# to it)
Our conclusion that s"$*athetic knowle#e $ust be a$itte 9as the
onl" alternati!e to soli*sis$: oes not i$*l" that the knowle#e of
another *erson;s feelin#s which we co$$onl" su**ose oursel!es to
*ossess is to be acce*te without 2uestion) E0*erience of color+
blinness teaches us that one $an;s color sensations $a" not be
co$*arable with another;s) It see$s i$*ossible to attach an" $eanin#
to the 2uestion whether "our sensation of re is like $" sensation of
re) I hesitate to sa" that it is e2uall" $eanin#less to sa" that "our
sensation of re is $ore like $" sensation of re than $" sensation of
a $usical note + thou#h I confess I o not see a $eanin#) 6hen we #o
be"on sensation an consier si$ilar 2uestions or state$ents
co!erin#- let us sa" the followin# kins of feelin# + the *assa#e of ti$e-
a #uilt" conscience- the taste of su#ar- bein# in lo!e- toothache-
a$use$ent at a 1oke + our reason re!olts alto#ether) A consciousness
in which there ha been a #eneral *ost of these feelin#s coul not be
a$itte to be a consciousness at all)
Ha**il" for the *h"sicist he has a o$ain ine*enent of s"$*athetic
knowle#e- an he can lea!e to others- better e2ui**e- the task of
sortin# out the resiuu$ of truth in our co$$on conce*tion of what
another $an;s $in woul be like if we coul #et insie it) 6e
therefore onl" #o far enou#h to show what are the essentials of a
#eneral *hiloso*hical outlook which will not *lace us in the ile$$a of
either 9a: en"in# that there is an" knowle#e other than *h"sical
knowle#e or 9b: rela*sin# into the soli*sis$ which we re*uiate at the
!er" be#innin# of *h"sical science)
III
Let us now consier the co$$on root fro$ which scientific an all
other knowle#e $ust arise) The onl" sub1ect *resente to $e for
stu" is the content of $" consciousness) Accorin# to the usual
escri*tion- this is a hetero#eneous collection of sensations- e$otions-
conce*tions- $e$ories- etc) The raw $aterials of knowle#e an the
$anufacture *roucts of intellectual acti!it" e0ist sie b" sie in this
collection) 6e wish to *ick out the raw $aterial + the *ri$iti!e ata-
uns*oile b" the inter!ention of habitual for$s of thou#ht)
It $ust- I think- be reco#ni/e that this is an unattainable ieal) Our
facult" of sensor" *erce*tion is $oifie b" trainin#4 an it is
i$*ossible to concei!e it i!este alto#ether of the trainin# force on it
b" the conitions of life an aa*tation to en!iron$ent) I o not think
that sensation- as we know it- coul e0ist without an acti!it" of $in
which concentrates- co$*ares- an istin#uishes) 6hat we call a
sensation can ne!er be *urel" sensor"- .ut that is a 2uestion better
left to *s"cholo#ists) In an" case the *ractical ifficult" e0ists) 6e shall
*robe own towars the roots of knowle#e4 but the $ost *ri$iti!e
ata we can reach will not be wholl" ine*enent of the *ri$iti!e for$s
of thou#ht) 6e 1ust cannot hel* bein# brain"- an $ust tr" to $ake the
best of it)
It is inee one of the *ri$iti!e for$s of thou#ht- na$el" the conce*t
of anal"sis- which *resents the unit" of consciousness to us in the
for$ of a collection of sensations- e$otions- etc)) The anal"sis of
consciousness into *arts *resents the sa$e *roble$s as the anal"sis
of the *h"sical uni!erse into *arts) ." what criterion is the acce*te
s"ste$ of anal"sis istin#uishe fro$ other *ossible s"ste$s of
anal"sis@ Is the ieal *art self+sufficient- so that without contraiction it
can be thou#ht of as isolate fro$ the rest@
I shall take the !iew that consciousness is a whole which we anal"se
into *arts- an not a nu$ber of iscrete units 9sensations- e$otions-
thou#hts- etc): which are s*oken of collecti!el" as a consciousness)
Also I take it that our orinar" anal"sis is rather crue- an that there is
o!erla**in# an interaction of the *arts) 6hat we call a sin#le
sensation is not strictl" se*arable fro$ the en!iron$ent of e$otion-
$e$or"- conce*tual for$- etc)- in which it occurs)
At this $eetin# *oint of all branches of knowle#e- we ha!e to
istin#uish the branch which leas to knowle#e of the *h"sical
uni!erse) The raw $aterial for this knowle#e is containe in the *arts
of consciousness calle sensations or sensor" i$*ressions) The two
na$es ha!e not 2uite the sa$e si#nificance? ,sensation, i$*utes onl"
those characteristics of which we are irectl" aware- whereas ,sensor"
i$*ression, refers to a *ostulate connection with *h"sical sti$uli
trans$itte throu#h the sense or#ans) At this sta#e- while we are still
seekin# a route to the *h"sical uni!erse- an thence to our boies an
sense or#ans- the ter$ ,sensor" i$*ression, is *re$ature) There is
an#er of a !icious circle if we efine the *h"sical uni!erse b" its
#eneral connection with the structure of sensation in consciousness-
an then use efine *ortions of the *h"sical uni!erse 9sense or#ans:
to eter$ine what *art of our consciousness the na$e ,sensation,
refers to) The 2uestion therefore arises whether the istinction
between sensations an the rest of consciousness is one of which we
are irectl" aware- or whether it is i$*orte later when we ha!e learnt
about sensor" i$*ressions) I think the answer is that b" irect
knowle#e we can $ake an initial classification istin#uishin#
sensations fro$ the other contents of consciousness4 but this is
elaborate into- an ulti$atel" su*ersee b"- a $ore *recise
classification which ientifies sensations with sensor" i$*ressions)
This 2uestion is !er" $uch si$*lifie b" the fact than- althou#h all our
senses $a" be use for e0*lorin# the *h"sical uni!erse- $ost of the$
are reunant an $erel" corroborate the infor$ation which can be
obtaine b" others) It is therefore unnecessar" to know at this sta#e
the e0act sco*e of the ter$ ,sensation,) It is sufficient if we can
istin#uish b" irect awareness a *articular class of sensations- which
b" itself is sufficient to re!eal all that is known of the *h"sical uni!erse)
Ieall" all our knowle#e of the *h"sical uni!erse coul ha!e been
reache b" !isual sensation alone + in fact- b" the si$*lest for$ of
!isual sensation- colorless an non+stereosco*ic)
'<
6e can therefore
re#ar an ite$ of *h"sical knowle#e as an assertion of what has
been or woul be *ercei!e !isuall") The tests of *h"siolo#" an
e0*eri$ental *s"cholo#"- use to eter$ine sense i$*ressions fro$
other feelin#s- can accorin#l" be escribe in !isual ter$s) In this
wa" we efine the sco*e of sense i$*ressions without $akin# use of
*reli$inar" atte$*ts to efine b" irect awareness the full sco*e of
sensation)
The istinction between sensations an other feelin#s is not so self+
e!ient a classification as we so$eti$es think) One borer+line case is
*articularl" i$*ortant) Is our feelin# of the *assa#e of ti$e a
sensation@ 6e cannot e!ice a irect scientific test which woul be
an" $ore conclusi!e than our intros*ecti!e 1u#e$ent is) .ut #eneral
scientific consierations fa!or the !iew that our feelin# of the #oin# on
of ti$e is a sensor" i$*ression4 that is to sa"- it is as closel"
connecte with sti$uli fro$ the *h"sical worl as the sensation of li#ht
is) Bust as certain *h"sical isturbances enterin# the brain cells !ia the
o*tic ner!es occasion the sensation of li#ht- so a chan#e of entro*"-
35
New Pathways in Science- *) %')
either in the brain cells #enerall" or in s*ecial cells- occasions the
sensation of ti$e succession- the $o$ent of #reater entro*" bein# felt
to be the later) I ha!e treate this *roble$ rather full" in earlier writin#s
an nee not a $ore here)
'G
Alon#sie the sensations of which I a$ irectl" aware- I a$it also two
kins of sensation of which I a$ not irectl" aware- 9%: the sensations
which I re$e$ber to ha!e ha in the *ast- an 95: the sensations
which other *eo*le tell $e the" ha!e or ha!e ha) It is an a0io$ of
*h"sical science that- as raw $aterial for knowle#e- these are all on
the sa$e footin#)
The reco#nition that certain $e$ories are to be treate as a
knowle#e of *ast sensations is essential for *h"sical science4
because- as we shall see later- the first ste* towars structural
knowle#e is a co$*arison of sensations in one consciousness) The
atu$ of *h"sical science is not awareness of a sensation- but
awareness that a sensation is like- or ifferent fro$- a sensation which
we for$erl" ha) 3rantin# this- the sensations of one *erson alone
*ro!ie sufficient $aterial for structural anal"sis4 an it woul be
*ossible to e!elo* fro$ it a scientific theor" which- e0ce*t that it is
*resente in an e#ocentric fra$e of thou#ht- woul a#ree with orinar"
*h"sical theor") .ut since the anal"sis woul ne!er take us outsie a
sin#le consciousness- it woul #i!e no inication of a worl e0ternal to
that consciousness) The e0ternalit" of the *h"sical worl results fro$
the fact that it is $ae u* of structures foun in ifferent
consciousnesses)
Thus the reco#nition of sensations other than our own- thou#h not
re2uire until a rather later sta#e of the iscussion- is essential to the
eri!ation of an e0ternal *h"sical uni!erse) Our irect awareness of
certain aural an !isual sensations 9wors hear an rea: is
*ostulate to be an inirect knowle#e of 2uite ifferent sensations
9escribe b" the wors hear an rea: occurrin# elsewhere than in
our own consciousness) Soli*sis$ woul en" this4 an it is b"
36
The Nature of the Physical World- *) %DD)
acce*tin# this *ostulate that *h"sics eclares itself anti+soli*sistic)
IA
In $an" lan#ua#es two !erbs are re2uire to co!er the $eanin#s of
the En#lish !erb ,know,) 6hen we sa" that we know our own feelin#s-
the $eanin# usuall" is kennen 9connaitre:- whereas we ha!e in this
book been $ainl" concerne with ,know, in the sense of wissen
9savoir:) It is necessar" to e0a$ine rather carefull" the nature of our
irect awareness in orer to $ake it clear that it *ro!ies ata of
knowle#e in the sense of Wissenschaft)
To the 2uestion ,6hat are we $ost irectl" aware of@, the co$$on
answer woul be ,Feelin#s- an other *arts of the content of
consciousness,) .ut this is an iio$ of s*eech) The feelin# is itself an
awareness) 6hat we $a" call sentient awareness has no #ra$$atical
ob1ect e0ce*t itself) 7" consciousness is $" awareness- an the *arts
of $" consciousness + feelin#s- e$otions- etc) + are *arts of $"
awareness4 an it is a $ere oit" of lan#ua#e which leas us to
reiterate the awareness in such *hrases as ,awareness of feelin#,)
Our *ur*ose now is to show that awareness is sa*ient as well as
sentient4 an sa*ient awareness has a #ra$$atical ob1ect- na$el" an
ite$ of knowle#e)
Consier the state$ent ,I a$ aware that I feel *ain,) This $eans that I
know that I feel *ain in the sa$e sense that I know an" other fact- e)#)
that the sun has risen) ,Aware, is use here onl" to istin#uish the wa"
I ha!e obtaine the knowle#e) 97" knowle#e that the sun has risen
is not at all a $atter of irect awareness + since it ha**ens to ha!e
been fo##" all a"): .ut it is necessar" to notice that what I a$ irectl"
aware of is a &certain fact- not that the for$ of wors ,I feel *ain, is a
correct escri*tion of the fact) The inter!ention of a for$ of wors
creates an awkwar ifficult" in iscussin# the ele$ents of knowle#e4
the $ore accurate the escri*tion- the $ore e0tensi!el" oes it raw
on our #eneral knowle#e- an so istract attention fro$ the *articular
ele$ent of knowle#e on which we want to focus iscussion) An
ine0act escri*tion is not an ieal refu#e fro$ this ile$$a4 so let us
tr" another wa")
Su**ose that I suenl" sa" ,Ouch,) That will con!e" to "ou e0actl"
what was $eant to be con!e"e b" the for$er state$ent ,I feel *ain,)
It has the #reat a!anta#e that it oes not hint at an" *s"cholo#ical
theor" of what has ha**ene4 it oes not ra# in knowle#e not wholl"
eri!e fro$ irect awareness- as an" atte$*t at *recise escri*tion
woul o) Nor$all" it is an in!oluntar" re$ark4 but it is a *it" not to use
eliberatel" an e0*ression which con!e"s e0actl" what we $ean to
con!e" an no $ore) A t"*ical ele$ent of knowle#e ac2uire b"
irect awareness is that which we con!e" to another *erson b" the
e1aculation ,ouch,)
>noubtel" an ite$ of knowle#e is con!e"e) 6hen the entist- in
the course of his e0*lorations- asks ,8oes it hurt@, an I answer
,Ouch,- he obtains efinite infor$ation) It is clear that I $"self ha that
infor$ation before the entist4 an inee it was an ite$ of knowle#e
which I *articularl" wishe to con!e" to hi$) It is also clear that the
knowle#e ca$e to $e b" irect awareness)
This- I think- lea!es no oubt that awareness is not onl" sentient- but is
a $eans of ac2uirin# ite$s of knowle#e 9Wissenschaft:) A confusion
arises when such knowle#e is *ut into wors- because the choice of
e0act worin# e*ens on #eneral knowle#e which is not as a rule
ac2uire b" irect awareness) It is onl" e0ce*tionall" that we can #i!e
!erbal e0*ression to the knowle#e ac2uire b" irect awareness
without aulteratin# it) >suall" the !erbal e0*ression $ust be re#are
as a *ointer + *ointin# out the knowle#e- but not for$in# *art of it)
CHAPTER EIII
THE SYNTHESIS OF =NO6LE83E
I
In consierin# the *ri$iti!e ata of knowle#e furnishe b" irect
awareness- it is necessar" to bear in $in that the escri*tion of the
atu$ is not *art of the atu$) In orer that "ou $a" know what atu$
I a$ referrin# to- I ha!e to use a for$ of wors as a *ointer4 but e!en if
9as $a" occasionall" ha**en: the for$ of wors is an accurate
e0*ression of a truth about the atu$- it is a truth reache b"
subse2uent in!esti#ation an not #i!en to us as a *ri$iti!e atu$)
A stran#er in a lan- where his resources of lan#ua#e fail hi$- will
o*en co$$unication b" *ointin#) In this iscussion of the ori#ins of
knowle#e we are in a like *osition an ha!e to o a #oo eal of
*ointin#) .ut as *ointin# in a literal sense is i$*ossible- we ha!e to
*oint with wors an *hrases) This use of lan#ua#e for *ointin# $ust
be istin#uishe fro$ its use for e0*licit escri*tion which cannot
be#in until a later sta#e) Lo#ical inference is not a**licable to it- for
inferences can onl" be $ae fro$ ata4 an a *ointer is not a atu$)
6e o not re1ect lo#ical thou#ht- but we insist that it shall be a**lie to
the real ata)
Thus- in the funa$ental *roble$s which co$e at the be#innin# of
*hiloso*h" the for$ of wors is- in #eneral- the last thin# to which one
shoul *a" attention) Either the worin# re*resents the *hiloso*hical
!iews of the *rehistoric in!entors of lan#ua#e4 or it *re$aturel"
assu$es a truth which it is our business to fin out b" in!esti#ation)
Consierin# a#ain the state$ent ,I a$ aware that I feel *ain,- "ou
know what it $eans because at ti$es "ou "ourself ha!e such an
awareness) It ser!es its function as a *ointer4 an- if "ou are satisfie
that the s*eaker is not l"in#- "ou can acce*t it 9with the $eanin# that
"ou ha!e reco#ni/e: as a atu$ of knowle#e)
.ut *resu$abl" "ou o not acce*t as a atu$ of knowle#e the
*hiloso*h" e$boie in the for$ of the state$ent + that there is a
sentient ,I, who feels- an a sa*ient ,I, aware that the sentient ,I,
feels- with *erha*s an infinite re#ression of ,I,s each aware that the
one ne0t in orer is aware of so$ethin#) E!en if "ou ha**en to a#ree
with that *hiloso*h"- "ou reali/e that it has nothin# to o with the
knowle#e that was bein# co$$unicate in the state$ent) A $an can
be aware of *ain without bein# a *hiloso*her)
Let us consier wh" the escri*tion 9thou#h not the atu$: introuces
two ,I,s- which we fin it ifficult to ientif" co$*letel") It is a
conse2uence of the non+soli*sistic outlook that the knower oes not
usuall" coincie with the feeler) Other *eo*le;s sensations are as
i$*ortant as our own4 an the usual for$ of knowle#e woul be ,I
know that so+an+so feels *ain,) 6hen the e0ce*tional case occurs-
the for$ $ust not be altere4 for it woul be a soli*sis$ to #i!e our
own sensations an" kin of *riorit" or istincti!eness in knowle#e)
Thus the escri*tion $ust inicate the *ossessor of the knowle#e
an the *ossessor of the feelin# se*aratel"- e!en when both the
knowle#e an the feelin# are *arts of the sa$e consciousness + *arts
which to a consierable e0tent o!erla*) An" atte$*t to ar#ue fro$ the
worin# that the two *ossessors cannot be co$*letel" ientical is
rule out because it $istakes the function of the for$ of wors which is
to point) 6hat is *ointe at- i)e) the atu$- is that the knowin# an the
feelin# are *arts of one consciousness istin#uishe fro$ other
consciousnesses b" the !erbal+*ointer ,I,)
It $a" be notice that ,I know that, is an idempotent phrase 9*) %G5:?
I know that I know that J I know that)
Iteration $akes no ifference to its *ointer+!alue) That the two *hrases
$ean 9i)e) *oint to: *recisel" the sa$e thin# is seen when we e0a$ine
the a**arent alternati!e ,I o not know that I know that,- which is
clearl" nonsense)
'F
If we re*resent ,A knows that, b" the s"$bol B
A
-
37
It is to be re$e$bere that ,know, oes not $ean ,know with certaint", 9*) %:)
the state$ent B
A
B
.
is nor$all" irreucible4 but in the s*ecial case AJ.-
we ha!e B
A
B
A
JB
A
) The iteration can be re*eate an" nu$ber of ti$es4
thus B
A
B
A
B
A
) ) ) ) B
A
JB
A
)
In our orinar" lan#ua#e a feelin# is associate with a knowle#e-
na$el" a knowle#e that the feelin# e0ists) There is no a$bi#uit" in
co$*letin# this ,unfinishe sentence,4 the feelin# e0ists in- or is *art
of- a consciousness) For a soli*sist this is a truis$- since feelin# is the
na$e #i!en to a *art of consciousness4 an there is onl" one
consciousness + his own + for it to be *art of) .ut when we a$it $ore
than one consciousness- we $ake the knowle#e $ore
co$*rehensi!e than the feelin# b" ain# a *ointer inicatin# the
*articular consciousness in which the feelin# e0ists or of which it is a
*art)
II
Since knowle#e of the *h"sical worl is eri!e fro$ sensations- let
us take for iscussion a *articular sensation- for e0a$*le- the
sensation escribe as I+*ercei!e+the+soun+of+the+3reenwich+ti$e+
si#nal) E!ientl" the escri*tion contains infor$ation which is no *art
of the sensation- an is not itself a $atter of irect a**rehension) 6e
$ust now ask- Is an" *art of the escri*tion a $atter of irect
a**rehension@ In *articular- ha!e we a irect awareness that the
sensation is a sub1ect+ob1ect relation- as the for$ of escri*tion
i$*lies@ I o not think that we ha!e) 6e can- if we like- e0*eri$ent
with the h"*othesis that a sensation is- or can be re*resente as- a
relation 9*ercei!in#: between a sub1ect 9,I,: an an ob1ect 9a
,sensu$,:4 but that is !er" ifferent fro$ assertin# that we are irectl"
aware that it is such a relation) That the e0*eri$ent is unsuccessful is-
I think- shown b" the barrenness of realist *hiloso*h") The ob1ect+en
of the relation is a cul+e+sac) .ut let us e0a$ine $ore closel" the
sub1ect+en of the relation)
Hitherto the ter$ ,I, has been for us a *ointer+wor- use to *oint to a
*articular consciousness of which the sensation for$s *art)
E2ui!alentl" it is a label attache to the consciousness to sa!e the
trouble of *ointin# e!er" ti$e we $ention it) 6hen b" the conce*t of
anal"sis we se*arate the consciousness into a nu$ber of sensations-
e$otions- etc)- we attach to each *art the label ,I, + or- in eference to
the #ra$$arians- ,$",) The $oifie label oes not enote *arts4 it
oes not *ostulate an owner istinct fro$ the consciousness- who
owns all the *arts an therefore the whole consciousness)
Ne!ertheless the function of ,I, as a label oes not e0haust the
si#nificance co$$onl" attache to ,I,) A$on# the contents of $"
consciousness is a self+consciousness) In the lan#ua#e of sub1ect+
ob1ect relations we sa" ,I a$ aware of ;I;,) 6ithout enorsin# this
escri*tion of self+consciousness as a sub1ect+ob1ect relation- we
reco#ni/e it as a *ointer an a$it the *ri$iti!e atu$ to which it
*oints) The 2uestion then is- 6hat aitional si#nificance is #i!en to ,I,
in connection with this atu$ of self+consciousness@
6e $ust re$e$ber that the conce*t of anal"sis is a for$ of thou#ht4
an althou#h its a**lication to consciousness ser!es certain useful
*ur*oses- there is no #uarantee that a si$*le *uttin# to#ether of the
anal"tical *arts without binin# $aterial will re*rouce the whole) E!en
in the *h"sical uni!erse- where the anal"sis is a**lie $ore
s"ste$aticall"- an #reater *recautions ha!e been taken to secure
non+o!erla**in# an *er$anent self+sufficienc" of the *arts- the
ele$entar" *arts are not strictl" se*arable) Still less is a sin#le
sensation strictl" se*arable fro$ the en!iron$ent of e$otion- $e$or"
an intellectual acti!it" in which it occurs4 nor is it strictl" se*arable
fro$ the !olition which irects attention to it an the thou#ht which
e$boies sa*ient knowle#e of it) Thus the consciousness to which a
*articular sensation belon#s concerns it- not onl" as a label- but as an
en!iron$ent)
I ha!e knowle#e of a certain sensation- an I ha!e the further
knowle#e that it is or was $" sensation) If I a$ a non+soli*sist- the
secon state$ent co$bines two ata) One atu$ refers to the
classification of sensations as belon#in# to a nu$ber of ifferent
consciousnesses- an isa**ears if all the sensations of which I a$it
ha!in# knowle#e are in one consciousness) .ut the other atu$ is
concerne with a *ositi!e as*ect of ,$",- not arri!e at b" contrast
with ,his,- an re$ains !ali e!en for a soli*sist) It is that the sensation
is not a self+sufficient ele$ent of awareness ine*enent of other
ele$ents of awareness- but is one of the *arts into which b" a
so$ewhat crue issection we ha!e i!ie an awareness which is
*resente to us as a whole) The ,I, which is the su**ose ob1ect of
self+consciousness is the correlati!e of ,$", in this secon as*ect + the
uniting ,$", + in the sa$e wa" that the ,I, which is the su**ose
sub1ect of !erbs of awareness is the correlati!e of ,$", as a label + the
contrasting ,$",) The no$inati!e- ob1ecti!e an *ossessi!e cases are
to be isre#are- since the rules of s"nta0 ha!e not been esi#ne
for *ointer lan#ua#e) The ata *ointe at are res*ecti!el" a contrast
with sensations belon#in# to another consciousness- an a unit" of
conscious awareness which *re!ents it fro$ bein# full" re*resente as
an a##re#ation of self+sufficient *arts)
6e $a"- I think- ientif" self+consciousness with awareness of this
unit" of consciousness) In one sense self+consciousness can be
counte as a ,*art, of consciousness- 1ust as the interaction between
ele$entar" *articles can be counte as a *art of the *h"sical uni!erse)
.ut it is not ho$o#eneous with the other *arts4 an in the stricter
sense- in which the $eanin# of a ,*art, cannot be issociate fro$ the
s"ste$ of anal"sis of which it is a *rouct- self+consciousness is not
an anal"tical *art but a resiuu$ which has elue the anal"sis)
In the sub1ect+ob1ect escri*tion of self+consciousness ,I a$ aware of
;I,,- the secon ,I, stans for the unit" of consciousness) 8istin#uishin#
it as I
5
- I
5
is what is left if "ou i$a#ine $e without an" of the feelin#s-
thou#hts- etc)- in!entorie b" the conce*t of anal"sis) These
in!entorie contents can be !arie without $oif"in# the essential ,I,
associate with the$) It $a" *erha*s be ob1ecte that this escri*tion
of I
5
*recisel" fits the ,I, who was fast aslee* a few hours a#o + which
see$s to lea to the reductio ad absurdum that it is in slee* that the
essential ,I, e$er#es fro$ the swar$ of thou#hts an e$otions that
orinaril" obscure it) .ut that is like ar#uin# that the essential 2ualities
of #lue are best is*la"e when it oes not conta$inate itself b"
stickin# an"thin#) To obtain the I5- of which we are aware in self+
consciousness- thou#hts an feelin#s $ust be abstracte- not
eli$inate) The unit" of consciousness is $anifeste because there
are *arts for it to unite)
To su$ u*? ,I, is first a label or *ointer+wor attache to a *articular
consciousness- an conse2uentiall" to the sensations- e$otions- etc)
into which the consciousness is i!ie b" the conce*t of anal"sis4
an seconl"- as associate with self+consciousness- it is *art of a
!erbal for$ ,I a$ aware of ;I;, use to *oint to a resiuu$ of
awareness which elues the conce*t of anal"sis) The *hrase *oints to
the atu$ 9of which we ha!e i$$eiate knowle#e: that our whole
awareness is not full" re*resente b" the *arts into which we
custo$aril" i!ie it4 in other wors- it is a unit" an not an
asse$bla#e of *arts) It a**ears to be no $ore than lin#uistic custo$
that ,I, is $ae in the first case the sub1ect an in the secon case the
ob1ect of the !erb ,to be aware,) 6hen we tr" to #et behin the
worin#- we fin nothin# to su**ort the !iew than awareness is a
sub1ect+ob1ect relation or e!en a sub1ect+intransiti!e relation)
III
Let us now turn to the su**ose ob1ect+en of the relation) For the
*ur*oses of *h"sics the onl" !alue of the irect awareness escribe
as I+*ercei!e+the+soun+of+the+3reenwich+ti$e+si#nal is that it can be
co$*are with an in so$e cases reco#ni/e to be the sa$e as
another irect awareness which I re$e$ber to ha!e ha) The t"*ical
atu$ for *h"sics is therefore I+ha!e+a+sensation+which+I+ha!e+ha+
on+a+for$er+occasion) Pro!ie that a $eans can be foun of
escribin# the for$er occasion in a wa" which will enable other
*ersons to ientif" it in their own e0*erience 9without which the
infor$ation woul be of no !alue:- the atu$ re*resents
co$$unicable knowle#e) It is unnecessar" to assu$e that the
*erson to who$ the knowle#e is co$$unicate has an" s"$*athetic
unerstanin# of $" aural sensations4 he $a" be a stone+eaf $an
unable to i$a#ine what a sense of soun woul be like)
The theor" of structure escribe in Cha*ter IE inicates the wa" in
which this co$$unicable t"*e of knowle#e is elaborate an $ae
entirel" ine*enent of the non+co$$unicable ini!iual sensations)
The ,for$er occasion, is ientifie b" its association with other
sensations or #rou*s of sensations in the sa$e consciousness- which
in turn are co$*are with an foun to be the sa$e as earlier or later
sensations) Finall"- out of these co$*arisons we e0tract a *attern of
interlockin#- which can be escribe $athe$aticall" an re*resents
structural knowle#e of the sensor" content of the consciousness
stuie)
In the case of !isual sensations- the structure is $ore self+e!ient)
6ithout referrin# to $e$or" of for$er sensations we can etect a
*attern in what we see at an" $o$ent) It is $ainl" throu#h !isual
sensation that our orinar" conce*tion of the *h"sical worl is for$e)
.ut 1ust because it lens itself so easil" to structural in!esti#ation- it
has been $ulle o!er b" our earl" ine0*ert #ro*in#s4 an it is not so
eas" to $ake a clean start in se*aratin# the $athe$atical essence of
the structure fro$ the for$ of awareness in which it is containe) Our
habit of !isuali/in# structure $akes it $ore ifficult for us to reali/e the
essential abstractness of structure)
As a result of co$$unicatin# structural knowle#e we soon fin that
the structural contents of ifferent consciousnesses are not wholl"
ine*enent) The *roble$ then arises- How are we to re*resent this
intere*enence@ 6e $a" be#in with the si$*le case in which the
sa$e structure is foun in nearl" all consciousnesses with which we
can co$$unicate- for e0a$*le- the structure of the !isual sensation
which arises when we look at a constellation in the starr" hea!ens) 6e
re1ect the iea that the occurrence of this hi#hl" s*eciali/e structure
in so $an" consciousnesses is a coincience- an thereb" co$$it
oursel!es to the h"*othesis that the $an" si$ilar structures are
re*rouctions of one ori#inal structure) This is the #er$ of the iea of
causation) In the lan#ua#e of causation we attribute the si$ilar
structures in the ifferent consciousness to a co$$on cause
containin# the sa$e structure)
A *ossible h"*othesis is that this is an effect of hereit") Nor$al
consciousnesses $i#ht contain this *articular structure for the sa$e
reason that nor$al boies contain another *articular structure calle a
li!er) This h"*otheses- howe!er- is refute b" the occurrence of no!ae
9new stars:) These are chan#es of the structure of the !isual
sensation- occurrin# si$ultaneousl" in all consciousnesses- for which
our co$$on ancestor e!ientl" cannot be hel res*onsible) The
co$$on cause cannot be locate in an" one of the consciousnesses
without soli*sis$- nor can it be locate in an ancestral consciousness4
therefore it $ust be locate outsie an" of the reco#ni/e for$s of
consciousness) This real$ outsie ini!iual consciousness- where
the co$$on causes of the sensor" structures in ifferent
consciousnesses are locate- is calle ,the e0ternal worl,)
." reco#ni/in# other consciousnesses as coe2ual with our own we
ha alrea" co$$itte oursel!es to the acce*tance of a real$ outsie
the ini!iual consciousness) Ne!ertheless it is a new ste* of #reat
$a#nitue when- b" the isco!er" of si$ilar structures co$$on to all
nor$al consciousnesses- we introuce an e0ternal worl containin#
the ori#inal structure of which the" are re*rouctions) Since the
e0ternal worl is introuce as a rece*tacle of structure- our
knowle#e of it is li$ite to structural knowle#e4 an *h"sical science
is the stu" of this structural knowle#e) .ut- shoul occasion arise-
the function of the e0ternal worl can be enlar#e so as to co$*rise
$ore than our *h"sical knowle#e) If we fin reason to be issatisfie
with a *urel" *h"sical worl e0ternal to oursel!es- there is roo$ for a
s*iritual inter*retation of the ,so$ethin#, of which the *h"sical
uni!erse is onl" the abstract structure)
6e o not- to be#in with- *ut forwar an" theor" as to how the ori#inal
structure in the e0ternal worl co$es to be re*rouce as a structure
of sensations in consciousness4 we $erel" reco#ni/e that- rulin# out
coincience- the occurrence of the sa$e structure in $an"
consciousnesses is a si#n that an ori#inal structure e0ists in a real$
outsie those consciousnesses) Thus the scene of the #ran s"nthesis
is transferre to an e0ternal real$- where the scra*s of structure which
are the ori#inals of the sensation+structures in our own an other
consciousnesses stan like *ieces of a 1i#+saw *u//le waitin# to be
fitte to#ether) This e0tre$el" intricate s"nthesis is a task which
*h"sical science has been slowl" acco$*lishin# throu#hout the a#es)
7istakes ha!e often been $ae) In *articular- the earlier theories
atte$*te to wea!e into the s"nthesis knowle#e which is not *urel"
structural4 an it is onl" in recent "ears that *h"sical theor" has
beco$e in for$- as well as in fact- a theor" of $athe$atical #rou*+
structure) .ut 2uite earl" on in the s"nthesis it was *ossible to iscern
so$e of the ste*s b" which structures in the e0ternal worl are
transferre fro$ their ori#inal location to consciousness) That is to sa"-
b" fittin# to#ether the *ieces of structure we obtain a co$*rehensi!e
structure- which contains not onl" the ori#inal *ieces but a $echanis$
for *ro*a#atin# structure)
In the *ro#ress of this s"nthesis we ha!e learne to set asie the
crue instincti!e !iew that ,seein#, is a sort of stabbin# o*eration
which collects infor$ation as a *ark+kee*er collects litter) The structure
of our !isual sensation of a constellation is re*rouce $an" ti$es
o!er in the e0ternal worl + in a set of $aterial ob1ects- in li#ht wa!es-
on the retina- in o*tic ner!es- in brain+cells) The re*rouction in actual
sensation follows at the en of this se2uence) 6hen our *h"sical
knowle#e has reache this sta#e we are entitle to substitute the
ter$ ,sensor" i$*ression, for sensation) .esies our irect awareness
of the sensation- we now ha!e inirect knowle#e that it is associate
with the ner!es an sense or#ans introuce in the s"nthesis of
structural knowle#e) This ,theor" of sensation, has- of course- been
freel" use in the e!elo*$ent of *h"sical science) It coul ha!e been
introuce at an" sta#e as a reasonable h"*othesis to be teste b"
e0*eri$ent) .ut it is not the lo#ical startin# *oint of an e0*loration of
the founations of science4 an in an e0a$ination of the nature of the
knowle#e co$*rise in *h"sical science- we $ust #o back farther to
a atu$ ine*enent of theories of sensation- na$el" that the sa$e
structures of sensation occur in $ore than one consciousness $uch
$ore fre2uentl" than can be e0*laine b" coincience)
IA
As a contrast to the $etho of *h"sical science- let us see how realist
*hiloso*h" atte$*ts to treat the ob1ect+en of the relation) It see$s to
$e to *ro!ie an illustration of the isastrous influence which the
!erbal for$s- i$*ose on us b" the un*hiloso*hical sha*ers of
lan#ua#e- are liable to ha!e on our thou#ht)
The followin# is a t"*ical introuction to realist *hiloso*h"?
It is clear that whene!er I ha!e an" kin of e0*erience- whether I
a$ rea$in#- thinkin#- ha!in# hallucinations- or $erel" *ercei!in#-
so$ethin# is rea$t- thou#ht- hallucinate- or *ercei!e- an that
$" $in has so$e relation to this so$ethin#)
'(
The ar#u$ent #oes on to *oint out that this ,so$ethin#, can ha!e
ifferent relations to the $in4 for that which is *ercei!e can also be
re$e$bere or i$a#ine) It is ar#ue that what is in the $in coul
not ha!e this !ariet" of relationshi* to the $in4 the ,so$ethin#, is
therefore not *art of the $in) The conclusion is?
It is a characteristic at once co$$on an *eculiar to all $ental acts
that the" shoul be aware of so$ethin# other than the$sel!es) To
sa" of an act that it is $ental is- inee- to sa" of it that it is an
awareness of so$ethin# other than itself) This conclusion entails
the corollar" that the ,so$ethin# other, of which there is
awareness- is unaffecte b" the $in;s awareness of it) As
e0*erience- in other wors- it is *recisel" what it woul be- if it
were not bein# e0*erience)
39
This woul be a *raiseworth" etecti!e effort if our ob1ect were to
38
C)E)7) Boa- Guide to Philosophy- *) GG) Boa is not necessaril" statin# his own !iew)
39
,bid- *) FC)
isco!er the *hiloso*hical !iews of the *ioneers of s*eech + those
ori#inall" res*onsible for the wa" in which we strin# wors into *hrases
an sentences) .ut wh" this shoul be resurrecte to ser!e as a basis
of twentieth+centur" *hiloso*h" *asses $" co$*rehension)
There is no ifference of $eanin# between ,rea$in#, an ,rea$in#
a rea$,- or between ,thinkin#, an ,thinkin# a thou#ht,) At first si#ht
,rea$in# a rea$, see$s a *ur*oseless reiteration) .ut if it is
esire to enter into *articulars- lan#ua#e *ro!ies no wa" of attachin#
the$ to a !erb4 I a$ not allowe to sa" that I was fallin#+o!er+a+
*reci*icel"+rea$in#) I ha!e to #i!e the !erb an ob1ect- e!en if it is onl"
a u$$" ob1ect- an attach to the ob1ect the *articulars which I wish to
a) I therefore state as a escri*tion of $" rea$ *articulars which
$i#ht e2uall" well ha!e been #i!en as *articulars of $" rea$in#- if
the for$s of lan#ua#e ha allowe) The realist triu$*hantl" *rouces
this u$$" ob1ect- an sa"s ,You a$it then that something is
rea$t- na$el" the rea$ which "ou ha!e so !i!il" escribe,) I
a$it nothin# of the kin) All I a$it is that the rules of lan#ua#e
co$*el $e to talk as if I a$itte it)
Si$ilarl" ,li!in# a life, is the sa$e as ,li!in#,) 3ra$$aticall" a life is
so$ethin# that is li!e4 but in actual $eanin# $" life an $" li!in# are
the sa$e) It is a t"rann" of lan#ua#e which ecrees that circu$stantial
etails of a $an;s life can be #i!en- but not of his li!in# which is
consiere to be an unanal"/able acti!it") It will be seen that there is
#reat o**ortunit" for a ialectic *hiloso*h"- uner *retence of clearin#
u* a confusion to introuce one) Thus it is *ointe out that ,sensation,
can either $ean ,sensin#, or that which is ,sense,4 an it is
su##este that the two $eanin#s ha!e been confuse in certain
*hiloso*hies) .ut there are not two $eanin#s to be confuse + onl"
two #ra$$atical for$s with the sa$e $eanin#) An it is the critics who
confuse the$sel!es b" introucin# a sensu$- i)e) so$ethin# sense
which is istinct fro$ the sensin#- in orer to *ro!ie a secon
$eanin#)
The !iew that acti!it" 9e0*resse b" !erbs an #eruns: is of a few
si$*le kins an that !ariet" resies in *assi!it" 9e0*resse b" nouns:
has *urel" lin#uistic ori#in) The *aucit" of !erb for$s is fa$iliar to
$athe$aticians as a ifficult" of orinar" s*eech easil" sur$ounte in
their own s"$bolic lan#ua#e) Thus it is *ossible to s*eak of
u*licatin#- tri*licatin#- ses2uiu*licatin#- etc)- but this $oe of
e0*ressin# !ariet" of o*erations is soon abanone4 we use instea
one !erb+for$ ,$ulti*l"in#, an transfer all the !ariet" to noun+for$s
calle nu$bers) Then *erha*s it will be sai ,It is clear that whene!er
an"thin# is $ulti*lie it $ust be $ulti*lie b" so$ethin#- an this
so$ethin#- e)#) two- is not itself a $ulti*l"in# but an ine*enent
entit"- e0actl" the sa$e when it is a $ulti*lier as when it is not a
$ulti*lier,) The ar#u$ent woul not ha!e arisen if we ha stuck to the
ter$s u*licatin#- tri*licatin#- etc)4 for one oes not u*licate by
an"thin#)
The lack of !erb+for$s an of *hrases for 2ualif"in# !erb+for$s $akes
it ifficult to escribe consciousness as what we know it to be + an
e0tre$el" !arie acti!it") In custo$ar" lan#ua#e the !ariet" of our
intellectual acti!it" is onl" to be escribe as a !ariet" of thou#hts not
as a !ariet" of our thinkin#) This $akes no ifference to the *h"sicist-
who is concerne onl" with structure- since the structure of the thinkin#
is also the structure of the thou#hts) .ut it leas $an" *hiloso*hers to
*lace all the !ariet" in sensa outsie consciousness- an to restrict
consciousness to a few unanal"sable acti!ities + *ercei!in#-
concei!in#- re$e$berin#- e$otin# the !ariet" outsie itself) .ut it is
not an essential characteristic of acti!it" that it shall be inca*able of
sub+classification) 3esticulatin#- for e0a$*le- is an acti!it" of $an"
!arieties + shru##in# the shoulers- wa!in# the ar$s- shakin# the
hea- etc)) 6e can escribe the !ariet" irectl"- without e0*anin#
,#esticulatin#, into ,#esticulatin# a #esture,- an *roceein# to classif"
the #estures) An- so I su**ose- realists will not insist that whene!er
we #esticulate so$ethin# $ust be #esticulate- an that the
so$ethin# is unaffecte b" our #esticulation of it- bein# inee
*recisel" what it woul be if it were not bein# #esticulate) Yet I
so$eti$es woner how a realist woul re#ar the #esture known as
,cockin# a snook,) It woul see$ clear that so$ethin# $ust be
cocke4 an I fear the onl" lo#ical conclusion is that there is a real$ of
e0istence containin# uncocke snooks which are e0actl" what the"
woul be if the" were bein# cocke + but *erha*s that is too an#erous
a thou#ht to *ursue when *hiloso*hers are tr"in# to e0*ress what the"
think of one anotherH
In the introuction to realis$ that I ha!e 2uote 9*) 5%%:- the
conce*tion of a sensu$ istinct fro$ the sensin# a**ears to ha!e a
*urel" lin#uistic ori#in4 but the i$*ortant conclusion that the sensu$ is
so$ethin# e0ternal to consciousness is base on the e0istence of a
nu$ber of ifferent wa"s in which it can be relate to consciousness)
A$ittin#- for the sake of ar#u$ent- an ob1ect of *erce*tion- sa" a
*atch of blue color- which is not the *erce*tion itself- I think the
nu$ber of wa"s in which it can be $entall" a**rehene has been
e0a##erate) Sur!e"in# the content of $" consciousness- I $a" fin a
*ercei!in# of blue color or an i$a#inin# of blue color) The ifference is
un$istakeable- an is intrinsic in the *erce*tion or i$a#inin#4 but the
h"*othesis of the realists is that it is the sa$e ob1ect or sensu$ than is
bein# a**rehene in two ifferent wa"s) I ha!e onl" these two kins
of awareness of blue color4 but in the content of $" consciousness I
$a" fin also thou#hts about blue color which are not an awareness of
it- thou#h the" $a" acco$*an" a *erce*tion or i$a#inin# of it) 6hen
the thou#hts 9intellectual knowle#e: an the awareness are lu$*e
to#ether- other !arieties of classification are introuce) ." intellectual
knowle#e hallucinations are istin#uishe fro$ *erce*tions- thou#h
intrinsicall" the" are ientical) Si$ilarl" re$e$berin#s are
istin#uishe fro$ casual i$a#inin#s)
As a clue to the relation between *ercei!in# an i$a#inin# it is
noteworth" that 9nor$all" at least: a new ele$entar" sensation cannot
be i$a#ine until it has first been *ercei!e) 6e can in!ent in
i$a#ination new co$binations of sensations- but we cannot in!ent
entirel" new tastes- colors- *ains- etc)) It woul see$ that the first ti$e
we *ercei!e a new taste- our consciousness beco$es $oifie in
such a wa" that thereafter an i$a#inin# of the taste is *ossible) 6e
orinaril" sa" that a $e$or" of the taste is store u* in it) I o not see
how this can be reconcile with the realist !iew that i$a#inin# an
*ercei!in# are ine*enent relations of consciousness to a sensu$
outsie consciousness)
In the *assa#e that I ha!e 2uote it is reco#ni/e that- if *ercei!in# is
*urel" a relation between the $in an an e0ternal ob1ect- the ob1ect is
not $oifie b" our *ercei!in# of it) It is not clear whether it is also
reco#ni/e that the $in is not $oifie) If the $in is $oifie b" the
act of *ercei!in#- it is incorrect to escribe *ercei!in# as a ,relation,4
an the ar#u$ent base on the e0istence of $ore than one kin of
relation falls to the #roun) On the other han- if neither the $in nor
the sensu$ is $oifie b" the act of *ercei!in#- how is it that it is not
until after the *erce*tion that a new kin of relation of the $in to the
sensu$ beco$es *ossible- na$el" re$e$berin# or i$a#inin#@
A
The occurrence of ientical- or closel" relate- structures of sensation
in ifferent consciousnesses *ro!ies the lo#ical startin#+*oint of
*h"sical science) This e!elo*s naturall" into a #eneral in!esti#ation of
the correlations of sensor" e0*erience4 but b" the ti$e we reach this
wier *roble$ the $ain line of treat$ent has alrea" been settle) The
corres*onences of structure *oint to a co$$on cause e0ternal to the
ini!iual consciousness) The correlatin# $eiu$ is therefore
concei!e as an e0ternal worl- in which influences e$anatin# fro$
!arious foci are *ro*a#ate to the *oints at which the ifferent
consciousnesses are locate) In elaboratin# this conce*tion we ha!e
to consier the *ro*a#ation of influences fro$ one *art to another of
the e0ternal worl- not onl" as a $eans of con!e"in# $essa#es to
consciousness- but as continuall" reistributin# the characteristics of
the worl- an thereb" brin#in# its !arious *arts into causal connection
in ti$e an s*ace) Thus we *ass on to the $ain task of *h"sics- which
is to for$ulate a s"ste$ of escri*tion of the e0ternal worl an a
s"ste$ of laws a**licable to the entities $entione in the escri*tion-
which shall be in e!er" res*ect accorant with the actual correlations
of sensor" e0*erience) ." ,accorant with sensor" e0*erience, we
$ean that those *ortions of its structure which are ele$ents of a
sensation structure in a consciousness ha!e a unifor$
corres*onence with the sensations actuall" e0*erience in that
consciousness)
I ha!e e$*hasi/e two features of the knowle#e of the e0ternal worl
reache in this wa"? 9a: it is *artl" sub1ecti!e- an 9b: it is structural
knowle#e) To so$e e0tent these are alternati!es4 that is to sa"- if we
e0hibit *h"sical knowle#e in the *urel" structural for$ *ro!ie b" the
Theor" of 3rou*s- we eli$inate a lar#er *art of the sub1ecti!e ele$ent
which a**ears in the $ore co$$on*lace for$ulations) I o not re#ar
e!en the #rou*+structure as wholl" ob1ecti!e4 it is contin#ent on the
ee*+roote for$s of thou#ht e0a$ine in Cha*ters AIII- IE- an E)
.ut the so+calle funa$ental laws an constants of *h"sics are not
features that can be *ointe out in the ulti$ate #rou*+structure4 the"
are introuce b" aa*tin# the knowle#e to a for$ of thou#ht less
re$ote fro$ our fa$iliar outlook) As has been *ointe out 9*) %%F:- the
fra$e of thou#ht which corres*ons to the out*osts of scientific
a!ance is not that in which we assess the results of the a!ance)
It woul be illo#ical to attribute the si$ilarit" of the structures in
ifferent consciousnesses to a co$$on cause without allowin# to the
co$$on cause a status full" as ob1ecti!e as the structures
the$sel!es) I therefore take it as a0io$atic that the e0ternal worl
$ust ha!e ob1ecti!e content) .ut accorin# to our conclusions- the
laws of *h"sics are a *ro*ert" of the fra$e of thou#ht in which we
re*resent our knowle#e of the ob1ecti!e content- an thus far *h"sics
has been unable to isco!er an" laws a**l"in# to the ob1ecti!e content
itself) This raises the 2uestion- How is it that we are able to $ake
successful *reictions of *heno$ena without knowin# an" law
controllin# the ob1ecti!e content of the uni!erse an therefore without
knowin# how the ob1ecti!e content is #oin# to beha!e@
Althou#h it is rather the fashion for scientific writers to sa" that *h"sics
is not concerne with ob1ecti!e truth- it woul be unsafe to take the$
at their wor) A**arentl" the state$ent is intene to closure
iscussion- rather than to assert a *rinci*le whose far+reachin#
i$*lications in!ite in!esti#ation) Our own conclusion is less swee*in#l"
e0*resse4 but it is $eant seriousl"- an we $ust e0a$ine the
ifficulties to which it see$s to lea)
7uch of the ifficult" isa**ears if we kee* in $in that *ure
sub1ecti!it" is confine to the laws + the re#ularities + of the *h"sical
worl) The !ariet" of a**earances aroun us is *ri$aril" an ob1ecti!e
!ariet") That a sub1ecti!e istortion is introuce in our a**rehension
of thin#s is no $ore than *h"sicists ha!e been accusto$e to a$it)
6e ha!e trie to carr" farther than hitherto the sortin# of the ob1ecti!e
an sub1ecti!e ele$ents + with *erha*s sur*risin# results) .ut we
a$it an ob1ecti!e ele$ent in the s*ecial facts which constitute a lar#e
*art of our knowle#e of the uni!erse aroun us)
So$e conclusions which are of the nature of ,s*ecial facts, ha!e
ne!ertheless a rather wie #eneralit") It is a s*ecial fact that $ost of
s*ace is nearl" e$*t"- the $atter bein# a##re#ate in relati!el" s$all
islans) No one has su##este that this shoul be ranke as a
funa$ental law of *h"sics4 we are inee is*ose to think that it is a
latel" e!elo*e feature- the *ri$orial istribution of the $atter
ha!in# been a continuous nebula) Yet for so$e *ur*oses the nor$al
e$*tiness of s*ace has $uch the sa$e i$*ortance as a law of
*h"sics) In astrono$" we often eke out our !er" li$ite obser!ational
knowle#e of the istribution of $atter b" assu$in# it as a h"*othesis
+ a casual- not a funa$ental- h"*othesis)
The reco#nition of ob1ecti!it" in the s*ecial facts- althou#h alla"in# one
kin of ifficult"- $akes still $ore *ertinent the 2uestion how we are
able to $ake *reictions without knowin# an" law controllin# the
ob1ecti!e content of the uni!erse) It is not as thou#h we coul is$iss
the ob1ecti!e content fro$ consieration4 for the *reictions are
*reictions of s*ecial facts which in!ol!e the ob1ecti!e content)
The fact is that b" the funa$ental e*iste$olo#ical laws alone it is not
*ossible to $ake an" shar* *reictions) In actual *reictions the" are
cou*le with the law of chance) 6e ha!e seen 9Cha*ter AI: that the
$oerns s"ste$ of *h"sics a$its onl" *reictions of *robabilit") In
eucin# a *robable result- the uncertaint" is narrowe own to the
Heisenber# li$its b" assu$in# that the uneter$ine *art of the
beha!ior of the ini!iual *articles concerne is uncorrelate) This
*rinci*le of non+correlation is essential in all *reictions efinite
enou#h to be the sub1ect of obser!ational test)
6e ha!e reache the conclusion 9*) %(%: that the non+correlation of
ini!iual beha!ior- in s*ite of its rather wie #eneralit"- is a s*ecial
fact) It is a s*ecial fact that $atter is nor$all" unassociate with
consciousness- 1ust as it is a s*ecial fact that s*ace is nor$all" e$*t"
or nearl" e$*t") Ph"sics woul not ha!e taken the for$ it has taken if
it were the rule- rather than the e0ce*tion- for $atter to be uner the
influence of conscious !olition4 but e2uall" *h"sics woul not ha!e
taken the for$ it has taken if the $atter encountere in nor$al
e0*erience ha been istribute continuousl" as it is in the interior of a
star)
It is often *ointe out that the *ri$ar" ifference of outlook between
the scientist an the sa!a#e is that the sa!a#e attributes all that he
fins $"sterious in nature to the acti!it" of e$ons or other s*irits) For
the sa!a#e an" *h"sical ob1ect $a" be *ossesse of e$onic !olition-
an it is i$*ossible to count on its beha!ior e0ce*t in so far as the
irectin# e$on $a" be a$enable to *ra"er an *ro*itiation) Ph"sical
science has $ae a *lace for itself b" #reatl" li$itin# the s*here of
e$onic acti!it"- so that there is an e0tensi!e real$ of e0*erience in
which beha!ior can be counte on an scientific *reiction is *ossible)
3reat as $a" be the *ractical effects of this chan#e- it is a $atter of
etail 9s*ecial fact: rather than of *rinci*le) 8e$onic acti!it" 9!olition:
re$ains- thou#h it is li$ite to certain centers in $an an the hi#her
ani$als) Pra"er an *ro*itiation $a" still influence the course of
*h"sical *heno$ena when irecte to these centers) 6e now think it
luicrous to i$a#ine that rocks- sea an sk" are ani$ate b" !olitions
such as we are aware of in oursel!es) It woul be thou#ht e!en $ore
luicrous to i$a#ine that the !olitionless beha!ior of rocks- sea an
sk" e0tens also to oursel!es- were it not that we ha!e scarcel" "et
reco!ere fro$ the re*ressions of 5<D "ears of eter$inistic *h"sics)
Accorin#l" we o not re#ar the *rinci*le of non+correlation as one of
the funa$ental laws of *h"sics) Non+correlation usuall" a**lies4 but
correlation occurs e0ce*tionall"- an the result is an une0*ecteness
of beha!ior which is reco#ni/e b" us as a *h"sical $anifestation of
conscious !olition) In sa"in# that the beha!ior is une0*ecte- we $ean
une0*ecte fro$ the *oint of !iew of *h"sics- which su**lies the #a*
left b" our i#norance of the s*rin#s of ob1ecti!e beha!ior b" assu$in#
non+correlation) Actuall" the !olitional beha!ior $a" be full" e0*ecte +
it $a" be an answer to our own re2uest + but this e0*ectation takes
into account knowle#e of the ob1ecti!e worl not co$*rise in
*h"sical science an not reucible to the acce*te *attern of *h"sical
law) In so far as the co$*arati!e rarit" of correlation can be consiere
a law- it is a law of istribution of consciousness rather than a law of
the *h"sical worl)
In $" earlier references to the s"ste$ of funa$ental laws of *h"sics-
I shoul not ha!e !enture *ere$*toril" to e0clue the law of chance if
current o*inion were is*ose to inclue it) .ut I think the $a1orit"
woul be a#ainst incluin# it- althou#h their reason is ifferent fro$ $"
own) The co$$on !iew is that where we ha!e a #reat nu$ber of
ini!iual s"ste$s it is i$*robable that there will be an" correlation of
their beha!ior unless there is so$e s*ecific cause *roucin#
correlation) .ut a s*ecific cause of correlation woul be escribe in
*h"sics as an interaction- an as such it shoul be *ro!ie for in the
orinar" s"ste$ of laws) On this !iew the law of chance is *urel"
ne#ati!e + assertin# that there are no further correlations than those
*ro!ie for in the s"ste$ of laws alrea" for$ulate) In short- the law
of chance or non+correlation is not one of the funa$ental laws of
*h"sics but the wor ,Finis, ae when the list is co$*lete)
This ar#u$ent is base on the s"nthetic !iew of worl+structure- which
starts with ini!iual *articles an co$bines the$ to for$ the ob1ects
*erce*tible to our #ross senses) Althou#h no reall" ri#orous *roof has
been foun- it see$s a reasonable conclusion that such a $oe of
construction is sufficient to ensure non+correlation4 an there is
therefore no nee to inclue the *rinci*le of non+correlation as an
aitional h"*othesis) .ut non+correlation a**ears in a ifferent li#ht
accorin# to the anal"tic !iew of worl+structure- which starts with the
#ross ob1ects an anal"ses the$ into the structural ele$ents which we
call ini!iual *articles) Accorin# to the characteristics of the ob1ect
anal"/e we $a" obtain structural ele$ents with correlate or non+
correlate beha!ior) 6hen we fin non+correlation- the ine*enence
is not an ini!iual *ossession of each *article4 it is characteristic onl"
of the *articular co$bination that is stuie) The anal"tic !iew
therefore oes not auto$aticall" i$*ose the *rinci*le of non+
correlation) This is unlike the s"nthetic !iew in which ine*enence +
inifference to what other *articles are oin# + is assu$e to be an
unconitional characteristic of each *article- so that correlate
beha!ior of the *articles in an a##re#ate s"ste$ woul be contrar" to
nature)
6e acce*t the anal"tic !iew which *ro!ies no a priori reason for non+
correlation) .ut- as alrea" e0*laine- we o not a$it a #eneral
*rinci*le or law of non+correlation) Instea we a$it a *rinci*le of rarity
of correlation as a s*ecial fact about the worl in its *resent conition)
AI
Throu#hout this iscussion we ha!e ahere to the e*iste$olo#ical
$oe of a**roach) For us knowledge has been the one thin# that
counts) The s*iritual ele$ent in $an a**ears in our sur!e" as
so$ethin# which knows + so$ethin# to be loote for the treasure of
knowle#e it contains) 6ith the ruthlessness of a collector we carr" off
the treasure to our $useu$- there to be s"ste$aticall" arran#e an
is*la"e)
I ha!e little e0cuse for e0tenin# $" sur!e" be"on the li$its inicate
b" the ter$ ,knowle#e,) .ut I woul not like to lea!e an i$*ression
that the escri*tion of the hu$an s*irit as ,so$ethin# which knows,
can be acce*te as the whole truth about its nature) It is not 2uite so
narrow a escri*tion as ,the obser!er, + the title bestowe b" *h"sical
science) Consciousness has other functions besies those of a rather
inefficient $easurin# $achine4 an knowle#e $a" attain to other
truths besies those which correlate sensor" i$*ressions) Yet-
a$ittin# the wiest e0tension of the fiel of knowle#e- its *ursuit is
onl" one of the acti!ities *ro*er to our self+fulfil$ent) The instinct to
a$ass- *erfect an #lorif" knowle#e oes not stan alone4 it is akin to
other instincts which clai$ the sa$e acce*tance- *roceein# alike
fro$ a $"stic source wellin# u* in our nature)
E!en in science we reali/e that knowle#e is not the onl" thin# that
counts) 6e allow oursel!es to s*eak of the s*irit of science) The rise of
*olitical s"ste$s hostile to science alar$s us- not so $uch because of
the check to the out*ut of knowle#e- but because of the su**ression
an *er!ersion of the s*irit of science) 8ee*er than an" ,for$ of
thou#ht, is a faith that creati!e acti!it" si#nifies $ore than the thin# it
creates) In this faith- the cru$blin# of har+won knowle#e in the
successi!e re!olutions of science is not the continual tra#e" that it
see$s)
In the a#e of reason- faith "et re$ains su*re$e4 for reason is one of
the articles of faith)
The *roble$ of knowle#e is an outer shell unerneath which lies
another *hiloso*hical *roble$ + the *roble$ of !alues) It cannot be
*retene that the unerstanin# an e0*erience #aine in the *ursuit
of scientific e*iste$olo#" is of $uch a!ail here4 but that is no reason
for tr"in# to *ersuae oursel!es that the *roble$ oes not e0ist) A
scientist shoul reco#ni/e in his *hiloso*h" + as he alrea" reco#ni/es
in his *ro*a#ana + that for the ulti$ate 1ustification of his acti!it" it is
necessar" to look- arra" fro$ the knowle#e of itself- to a stri!in# in
$an;s nature- not to be 1ustifie of science or reason- for it is itself the
1ustification of science- of reason- of art- of conuct) Of the relation of
$"sticis$ an science I ha!e written elsewhere)
The an#er of a broa !iew is that it is often a shallow !iew) 6e $a"
clai$ for the e*iste$olo#ical outlook- li$ite thou#h it $a" be- that it
#i!es to the scientist a !iew broaer than his traitional !iew without
sacrificin# e*th) That it has been beneficial to the technical *ro#ress
of *h"sical science is unoubte) At the sa$e ti$e it #i!es a 1uster
conce*tion of the si#nificance of *h"sical knowle#e in relation to
*hiloso*hic thou#ht + a *ers*ecti!e which neither e0a##erates nor
unerrates the *h"sical as*ect of the worl which for$s a settin# for
the conscious e0*erience of $ankin) In *articular- the reali/ation that
*h"sical knowle#e is concerne onl" with structure *oints the wa" b"
which the conce*tion of $an as an ele$ent in a $oral an s*iritual
orer can be o!etaile into the conce*tion of $an as the *la"thin# of
the forces of the $aterial worl)
+oDo+

You might also like