Volume I

David Pines, Series Editor

Anderson, P.W, Basic Norions of C o d w e d Mater Physics

Bethe H. and Jackiw, R., Inte ee awnturn Mechanics, Third E&tim
Feynman, R., Photon-Hdron lnternctions
Feynman, R., Quantum Elect~odynamics
Fewman, R., Statistical Mechanics
Feynman, R., The Theory of F u d w t l t d Processes
Negele, 1. W. and Orland, H., Quantum Many-Pa~tickSystems
Nozi&res,R, Theoy of Interncting Fermi Systems
Farisi, G., Statistic$ Field Theory
Pines, D., The Many-Body Probkm
Quigg, C., Gauge Theories of the Strong, Weak, and Ekcnomagnetic Interactions
Schwinger, l., Particks, Sources, and Fields, Volume I
Schwinger, J., Particles, Sources, and Fields, Volume II
Schwinger, J., Particks , Sources, and FieLls , Volume III
aOURCES, AND

ULIAN SCHWINGER
late, University of California at Los Angeles

P E R S E U S BOOKS
Reading, Marsaehusetts
Many of the designations used by manufacturers and sellers to distinguish their prod-
ucts are claimed as trademarks. m e r e chose designations appear in. this book and
Perseus Books was aware of a uademark claim, the designations have been printed in
initial capital letters.

Library of Congess Camlog Card Number: 98-87896

Copyright 43 1998, 1989, 1970 by Perseus Books Publishing, L.LC.

Ail rights reserved. Na part of this publication may be reproduced, stored in a

retrieval system, or transmitted, in any fclrm or by any mans, electranic, mechanicai,
photocopying, recording, or othenvise, without: the prior mitten permission of the
publisher. Printed in the United States of America.

Perseus Books is it member of the Perseus b o k s Group

Cover design by Suzanne Heiser

1 2 3 4 5 6 7 8 9 10-EB-0201009998
First printing, September 1998
Editor's Foreword

Perseus Books's Frontiers in Physics series has, since 1961, made it possible for
leading physicists to communicate in coherent fashion their views of recent
developments in the most exciting and active fields of physics-without having
to devote the time and energy required to prepare a formal review or mono-
graph. Indeed, throughout its nearly forty-year existence, the series has empha-
sized informality in both style and content, as well as pedagogical clarity. Over
time, it was expected that these informal accounts would be replaced by more
formal counterparts-textbooks or monographsas the cutting-edge topics
h e y treated gradually became integrated into the body of physics knowledge
and reader interest dwindled. However, this has not proven to be the case for a
nllrnher of the volumes in the series: Many works have remained in print on an
on-demand basis, while others have such intrinsic value that the physics com-
munity has urged us to extend their life span.
The Advanced Book Classics series has been designed to meet this demand. It
will keep in print those volumes in Frontiers in Physics or its sister series, Lecture
Notes and Supplements in Physics, that continue to provide a unique account of
a topic of lasting interest. And through a sizable printing, these classics will be
made available at a comparatively modest cost to the reader.
These lecture notes by Julian Schwinger, one of the most distinguished the-
ore tical physicists of this century, provide both beginning graduate students and
experienced researchers with an invaluable introduction to the author's per-
spective on quantum electrodynamics and high-energy particle physics. Based
on tectures delivered during the period 1966 to 1973, in which Schwinger
developed a point of view (the physical source concept) and a technique that
emphasized the unity of particle physics, electrodynamics, gravitational theory,
and many-body theory, the notes serve as both a textbook on source theory and
an informal historical record of the author's approach to many of the central
problems in physics. I am most pleased that Advanced Book Classics will make
these volumes readily accessible to a new generation of readers.

Bavid Pines
Aspen, Colorado
July 1998
This page intentionally left blank
 Vita

$dim Schwinger
UdverSity Professw, University of CafoAa, md Prafessor of Physics at the
U~versityof Califoda, Angeles since 1872, was bosn in Nevv York City m
F e b m q 12, 1918, Profwsor S fxis Ph.D. in physics fm
Cdvmbia U~versityin 1939, He has itlm raived honotq doctorat= irr
from four iastitutions: hrdue U~versity(19611, H m a d U~vergity(19621,
Brmdeis University (19731, and Gustavus Adolphus &Bege (1975). fn addition
to teach8 at the U~versityof Califoda, Profesmr SGhwinger h trruCyht at
hrdue U~vmsity (B41-%-$31,and at H m m d U~versity (1945-72). Dr.
Sch~ngerwas a Rmewch Asmiate at the U~versityof C a E f e a , Berkeley,
and a Staff Member of the Mlllsachwetris Institute of Twbolow hdiation
Laboratoq. In, 1965 h o f a s ~ r rwipient ( ~ t f iEchwd
F e p m n m d Sin; Itiro Tomon in Physia for wark in
qumtum d e e t r d p

the C. L, M q r Matwe of Li&t Awad

Scimw Awud for Physics (1964); a Hurnboldtt

Awad (1981); the a di Castiaone de S i ~ (1986);
a the M o ~ A.
e
Fmst Sips Xi. Awad (1956); and the Amerlcm Academy of Acbevement
Awmd (1987).

vii
This page intentionally left blank
 Special Pre

Isme Newon used his newly hvented method of f l d o u s (the cdculus) to

mmpare the impfiations of the inverse square lrtw of paGtation with Kqler's
empihcai laws of p l m e w motion. Yet, when the t i m c m @ to ~ t thee
Brincipia, he resorted entkely to gwmetfial demonstrations. Should we wnelude
that cafculus is saperflwus?
S~arce&----to wEch the w n c q t of renom is foreign-and
renornaked operamfielil t h e a ~have both been foland the =me a s w a s
problems (wh_ieh &sappoints some pmple who would prefer
that wurw thmq prduct: new-md wrong-answers), Should we waclude
that source t h w q is thus superfluous?
Both questioas m e ~ the
t same respnse: the shpler, mare intuitive foma-
tion, is preferable.
'This &ition of P m i c l a , SOUTC~S,
crnd Fields is more extmsive &m the
o ~ @ n atwa t volumes of 1970 and 19'73. It n w e o n t h s foux atSditimd seetims
that f ~ s hthe chapter entitld, ""EeetrOaryn&~csH," "ese s a t i o n s
e t t e n , h 1973, but rem~nedin pwtidl_yt w d f o m for fiftan y w s , I am
indebteb to Mr. Ronafd B o b , who aged to d&pher my fa&ng
and mmpletd the typmfipt. Pargcular attention should wted to Section
5-9, where, h a context sommhat lwger than &ex-trsd
Trzetwen saurw and operator field t h w q fin*
~e~ first acqu;zintma ~ t sour= h &mq should wnsdt
the Appendix in Volume I. This Appm& contkns srrgestions for he&&g
m's s w ay though the sometha &uttered pages,
This page intentionally left blank
 and it is a textbook, It is the r m r d of a hi@y
personal reation to the hi& e n e w peicle physics. The iopdients
were: fmstration with the matthematid mbiguities and physical rmoteness of
aperator field thmq, dissatisfa~gonwith the overly mirthematid attitude md
spwulath philomphy of the supposedly mare physied $mat& thmv, outrage
at the pretension of cunmt dgebra to be a fundmentd descfipGon rather thm at
low m e r p phenomenolow.
The result was a point of Gew and a tecwqae that empha

phenomena. The physid sour= mncqt, upon w ~ c h

ticd prwursor in operam field thmq. But it was not
w M e tackng a Haward grduate mm=, &at I.
how the phenomenolo@caf soww wncept w d d be f r d from
tmcture md used as the- basis fm a mmpletely hdqendent
development, with much domr ties to experiment,
e rmnsfmction of efec rapidly, at U C M that
,and during a repelition at W=, iastead, devotd
to the n m approwh, Developments in pion physics that
(11966-1x71, in wEch the new most sumssfay &ppl,i&dtwn-
*& me, if no one else, of the
wncqtuaE c l ~ t that
s in mathematical shpgc=itymd
y its use bestowd. The lack of appraiation of tbese faets
by others w w dqrashg, but undersbndable. a i y a d e t d d prsenation of the
ideas and methds of sour= theory could chmge t b t situa~on.The writin& of
 As a textbook, this voXum is intended for use by my s
n~melativisticqamtum mwhdcs, who fishes to fern
m ~ h ~ cI sthi& . it of the utmost h p o r t a n ~that such acquaint== with the
Berating ideas of murw t h w q w a r befare expasure to one of the current
s f i h d a ~ e hm
s w w & him past the elastic bait. In the Preface to a volume on
authar spe&s of the desirabiEty that the student have ai
(opmator) field thmq. X echo that ~ s t f ucall,
l but
ce to hcXude S-matk thwq.
f have m d e no attempt to supply the traditiond
my m who dlegdly first did wbt, whea. Perhaps I
to the distoaions fierent ia the shplistie asswia-
thn of ideas m& xllethods with s p ~ i f i hdividuals.
c But there is a mre iagportant
reason, general &tique of eistiaf: attitudes is mwntid in motivating this
new vi , it would have b n too distrrtcthg if mnstanr reference: ta
&Mqu@s for wGch obsolesmnce is intended had ammpanied the development
proach. The expert comes ready nade ~ t ~pinlons
h about what has
done. To the student d l that matters is what is new ts b md X
hope that he will fmd much in these pages.
X m gatefd to Wss Cuanane and Miss Jeri Ingersan who, at
different pfiods, devotd1y aided the: burden of t p h g the
t. The book would never have been wmpleted (I hold the world's
r w ~ r dfor the lagegt number af unfhishd first chapters) without the patience
and understandkg of my wife. Xt is thmefare apprap~telyddicated t s the.
C,G.S, system.

Belmonl, Massachusetts
Oetobet 1969
 Contents

l --I Unitary Transformations

1-3 Galitmm Relatiuity
1-3 Einsreinian Relatiuity
1-4 Oitique o f Particle Theories

Spin O Particles. Weak Source

Spin O Particles. Strong Souree
Spin I Particles. The Photon
Spin 2 Partictes. The Crauiton
Particleg with Arbitrary Integer Spin
Spin $ Particles. Fermi-Dirac Slatistics
More About Spin iPanicles. Neutrinos
Particles of Inleger + $ Spin
Unqication o f All Spins a d Staristics

3-1 The Field Concept. Spill O hrticles

3-2 The Field Concept. Spin f Particles
3-3 Some Other Spin Values
 Mtcltiqinor Fiel&
Action
Inuariance Transfarmabiaprs and Flues. Charge
Inuariance
Meebtanica
me Electromagnefir:Field.
Charge Quantriath. MW$
Prtmitr'ue EIectrom@@dicfnteraetiom
and Source M d e k
Extended Sources. Soft Photons
Ifiteraclion S k e l e f ~Seatteri~g
~. Cross Sections
Spin iProcesses
Sources m filcrrterers
H-hrticfes
Imtability and Mgltiparticle E x c h ~ g e
The Grauitational FiekJ

Appendix: How to Read Volume I

If you ~@~'fr_iiorin"em,
beat "m,
This page intentionally left blank
Sources,
S
This page intentionally left blank
 PARTICLES

The concept of the p&icle has undergone drastic changes and generalisations
in the c a m e of $he hisb6cal development that l& to the! atom, to the nu~leus,
sad then 4x1 subnuclear phenomena, This htts aka been a progesrjion from
es~eotialynonrel&ivistic behaGor to an ulkra-relativistic dometin, It is interest-
ing b appmciafe how mueh of the kinema.tical particle a t t ~ b u t e is
s implied by
the mumeb stmretuxe!of %herel8btivity goup of krm~fi3rnEfrtio~1~ among equiva-
lent coordinate sy-stem%, In preparatioxr Ear this dimurnion we fimt review some
propt?r&iesof qusntum nneehanical unitary t r d o

1-1 UNITARY TRANSFORMATIONS

Quantum mechanics is a ~Mlbolioexprwisn of the laws of micramopic m@asurt?-

med. Stabs, situ~tiogsof optimum infornation, am represented by vmtom in
a earnpiex s p w [left vectors ( 4; and physic~lpropdies by
Knear H e r ~ t i a nomratom
. on A]. The freedom in physical
de~riptiancome~pondsto the fretzciom of matlternaticaf repr6wntat;ian wswiahd
with unitary operators. These are defined through the Hermitisn adjoint
operation p by
trtu = uut = I (1-1.1)

b t us tramform df vectom and operators weording to

- v-'xu.
- - .C- - -

( =( U = ) X =
Then d l numerical and adjoint relations among vectors and operatarn are
umhanged, We rredfy that
( 2 )= a , (2lXIF) = (a'lxlb'). (1-1.4)
The adjoint relation~hip,
($1 = Id)'
i~
trsn~fomedinto

shows that the Hemitian operator A is mapped into the Hermitian operator a.
1
 A eomplete get of states (a'[ f o m s a b a i s or coordinate system ilz the state
space, Any veetor 1 ) is represenled by its components mlative ta this bmis,
(dl ). Amther basis is produeed by a unitary transformation:

and the gven vector has a eammwnding new mt of csmpnents,

Thwe numberg are albmatively dmribed m eompanent~,rel&tiveto the initial

basis, of the new w c b r U1 ), An anabgous relatian for matrix ekments of
operator8 k
(1-1.10)
If two suecwive transformations rtre w d o m e d on the bwis, the net change in
tlre componenfs of a vector is given by

1%is produeed in one s$ep by the unitary operator U%&, in vvhich the multi-
plication order reflee& the sequenee of tr~nsformations,The appasih sequence
is repremnhd by U1U2, and the two are eomgamd by defining the unitary
aprator that h needed to convert the mead equence into the first,

An infinikaimaf unitaq %ransfornnationis a transfarmation in. thr?:in-

finiksinral neighbarhood of the identity. It is represerrbd by

where C is an bfiaik8imaj Ifemitian operator, When two m& tramformatians

are compared we find that
Ut121 E 1 4- i@1121, (1-1.15)
where
at121 -s -Gall (l/i)lGt, @zl (1-1.16)
intrduces the commutator of Gl and Gz. Ttte effect of an infinitesimal uai%&q
tr&nsformationon an aperator is given by

where
sx = (r/z">[x,
q,
An equivale~tfarm ia
X = u-lxu = X - &X.
If we compare alternalive evafuatiaw of

When pre8enkd in b r m of daubfe commutatorr~,

this is reaogniad as the Jscobi id~ntity.

Now let us conisider a group af unitary tram formation^ i L h n real, con-
.
tinuous ptitmmeters k, a -- 1, . . , n, which m d m i m f e collectiively as X. If
U(Xr,%) are typi~aloperators of the group, it is requird Lhet

where

are the parameters of another element of &hegroup. For unifary operabm the
existence of the irtvem ancl, of fhe identity is awumd. An i n f i k h i m l $mm
formation of the group with parametrtm &X, i~ comtmeM from

*ere the n finite Hemitim opclrators Qa are cailed the &c3mratomof the &roup.
One is free Lo redefine Cbt: generators by real nowindrtr Einmrr frm&~msbdiom,
with eorrmponding redefinitions of the prametem. On subjmtbg the in-
finitttsimid transformation operator U(6X)Lo an mbitrav unitav tradomatioa
of the group, we mu&$obtain another infirzibsinnal frawfarma$iorr. TKi irr
expresged by
( hE
~ ( k ) - ' ~ ~ v= ) @.a(&)@&, (1-1.26)
b
where the numbem uaa(X)sre real, We shall also use a matrix m t a % i oin
~ Ghe
a-dimensional parameter ~paef3trnd write

The unitafy transformation is presenM e~lbmativefyad3

The two Bets of matrices are related by

where T degnatea matrix transposition. Note that this is equivalent to

Hermitian conjugation when applied to the u matricee, since they are real.
The correspondence established between the unitary operator U(X) and the
matrices U@),d(X) is maintained under multiplication. Thus,

and
~(~~)U(XI)GU(X~)-'U(~~>-'
= U(X~)[G~(X~)IU(A~)"
= @m2)4(X1). (1-1.32)
Since the unit operator corresponds to the unit matrix, we write

where
Ba = -ga
T T*
= ga .
This gives
[G, Gbl = gbG = -G4b
and, if the irnsginary elements of the matrix gb are designated ss

we get the explicit commutation relations of the group generators:

We see, incidentally, that

In view of the multiplicative correspondence between U(X) and u(X)[&(X)],the

matrices ga[la]also obey these commutation relations,

The latter are a set of quadratic restrictions that must be obeyed by the numbere
g* the so-called group structure constants:

This cyclic structure also follows immediately from the cyclic form of the
 [[Gat @blF Gel + [W&,
Gel, G,] + [[G,, G,],
= 0. (l-1-41)
Gb]
The ~tmctureeomfants speeify the compsition propedies of inhibsimal
par~mekm. Let blXPIXah the parametrsrzl of Ithe infinihsimal fmnsformation
that conoeetrs the two mquences in which Lhe tramformation8 labled by
and b2X, can be applid, Accordjng to the commuhtions relations of tho ~ o u p
gener~tom,they %regiven by

In the discussion ta Eollow, group composition propedies am supplicsd by

georxretfied considerations. It is important to recopize that the associated
unitary g~oupm ~ not y be an exshet image of the underfying geometrieai group,
This is tt eoasequence of an intPinsic arbitmriness 5f any quantum mechanical
description whereby all stak8 can be change$ by a commm phase facbr, whieh
is the unitary transformatka generakd by the unit operator. Consider, for
example, the commutative (Abelian) group of translations in a tw&dimensional
apace. Let the parameters of the two independent infinitmimal displacements
be writLea &zl, 8z2 and the eomi3ponding He tian diqlwement operators be
denoted 'by p 1, p2? SO Ghat

The fact that succeSsive displaeements are insensitive to the order in which. they
are t?;pptiedshould imply the eontmutalivily of p1 and pa, But d l that h requi~ed
of the commu"ttoor is that if generate a unitary transformation without physical
eomequences, Accordingly,

with a suitabEr: normaliaation of the dkplz~cementoperators, Now the eEect of

%heunitary transformation U"(6x)on the operators p1.2 is given by

whioh shows that the displacement operators also sew@as coordind o p r a b r s

We recognizie the propertim 5f the q,p phase space wsoeirtted with. a single
qurtntum degee of freedom. Translations in this two4imemional @pace are
describd by a three-parabmeter unitary group. This is explicit in the f o m
G X p liq m q lip + &PI. (1-1.47)
The onr respondence between. the unitary operators U(&) and the finite
matrices %(X), &(h) doe8 nst nece~sarilyincl& the mitary ehsraekr of the
Istfer. [Note that if the %(X) m;:unitary, or real odhagonaf, matrices, we have
%(h) == B(%), md then the Eiernnitian, or ima&ary mtkymnzetnisal r n a t ~ m
g. = ia.] Since the structure of the g matrices csn be altered by changing the
generator basis in the parameter @p-, it is useful to have a basis-independent
ith which to judge the pomibi1iC;y of exhibiti I;ian g matfices
%(X)rn%tfiw. If the mt of rr 42 operabrs i~ by tht!linearly
cornbimtiom XG, the g nnatficm ux1,dergo the 8ame liniear %rang-
tber with a similafity transfomation p r d u ~ e dby the non-
sine;ular m a t k X. Bw&usethe trmt: of nnatdx produels is unchanged by the
Xathr $randomation we somider the real quadratic form

which must be positivedefinite if the g, are transformable into linearly inde

rnatri~es. The failure. af that propedy implies Lhat m&
s and unitav %(X) nnatfiees do not exist. This i s the ~3itu&
tim, in the ~xannplewe have just di~cussiedl,of the gene.mbmp, p, I. There the
qudmtia f o m ia identicdly lie ce the unit operator is repremnbd by Lbe
nu11 r n a t ~ while
, the ma;tri i a M with q and p haw only single non-
diagonal entries in such a way that all matrix products vanish. The positive
defixlibnm of the real rsymmet~ematxrix

ry but alaaa sufficient for &heg, to bc? equivdexrt to linearly

i n ni tm . Since the elements of the r nnad~xare un-
hanged by the aimilafity transformation

A red symmet~aalpsitivedefinitr? matrix can a l ~ ~ abe

y swritten as the square
of another mch matnk, which we designate as r Then

which makes exphcit the similarigy transformation that introduces u n i t a r ~ ~

(arthogod) %(X) matrims and Hermitian (anti~ymnetrical)g matrims, In the
new bmis, r is a m~ltiple~f %heunit matrix,
Let tm suppom that the o p r a b r s G4 pomes a finih4innemitbnaI linesrly
hdewndent Hermitian mtrix reaiizatition. That inrplbs the e r i ~ b n wof the
m1 synnmetfic pogitivedefinib m a t ~ r
1-2 Ealilsan relativity 7

The invariance of the* numbers under unitary transformations on. the operator8
leads again to the form
rf= u ( x ) r ' ~ ( x ) ~ , (1- 1.54)
with the implication that the %(h)can be pre~entedas unitary matrices. The
corresponding Hesmitian matrices g,, whi~hare linearly independent if we ex-
elude the uninteresting possibility that the group has an Abelian poup as a
factor, arc? an example of a finite-dimensiod realization of the G,, m the %(X)
provide a finite-dimensional unitary realization of the U(&). Conversely, if the
matrix Tab is not positive-definite, no s u ~ finite
h unitary [Hermitian] realization
of the U(&)[&)can exist. A finite-dimensisnd realization of the group meana
that a finite number of staks can be found which are transformed among
themselves by all, operations of the goup. frz general, the aetioxl of zt unitary
operator on tz s t a b introduces new states, and the repeti-l;ion. of the operation
continues the prmess of producing additional states, This can terminae with a
finite number of states only if thztf repetition eventually cea~esto provide new
operatom, that ia, if the group paramebr sp~ect;is cio~ed. The distinction be-
tween a closed and an open group manifold is most familiar in that between
rotation8 and Lranslaf ions.
XE the matrices g, are Hermitisn, the stmetum constants gabc are anti-
symmetrical in a and c, M well as a and b, which implies antisymmetry in ;b
and c. This cornpieh antisymmetry can only be realized with n 2 3, For
n 3, EL, suitable normalization brings the im.a@nav structure constants to the
unique farm
gabe ==z i e ~ b c t (1-1.55)
where Ecrbe i~ the totally antisymmetrical symbol specified by El23 = 4-1. The
resulting group comrnufatian relations

[G., Gal =i C
e
e.acCc,

are familiar in the Lheo~esof three-dimensional anmlar momentum and isohpie

spin, The Lhree-dimensional g matrices satisfy these commutation relatians,
and r is a multipie of the unit matrix,

Space-tiw coordinghs Ebpwar in qu~zntummechmies &B an abstraction of the

roles of the maerascopie measuremenL appara,lus. All evidence eonfirms the
equivaleace af two coordinab ;systems %hatdiEer in any or aIl of the follopring
ways: a tretnslation of dhe ~patialoPi@n, a translation of Lhcs time ori@n, a
rotation of the spa% &X@@,t3r eonstant relative veloeity between the two ~y@Lf?rn~.
These trangformations cortstitute the relativity group, or rather the ~ubgroup
sf transformations that are eontinuousXy eonnectd with the identity, When Ezfl
particles move slourly in comparison with the speed of light, the time coordinrate
has an absolute significance md is affeckd only by displacement of if8 origin,
This is Galilean re1ativit;y. It is oharachrieed by infiniksirnal coordinate $ran&
formations r, t -+ F, f, where

and
6r = 6e + 6w X r + 6vE.
Note that the sign conventions are appropriak to the significance of 60, my, as
the displacement of the origin of the spatial coordinate frame to which a dven
point is referred. If it is the point that is moved by Sr, its new position relative
+
to the fixed reference system is r 6r. m e graup camposition properties of
this 10-parameter group are gpecified by compafing the wquence of tramforma-
tions

with those in the opposite order. The result of pedoming the .tramformation
sequence 1, 2, l l, 2-l, or, equivalently, for infinitesimal transformations,
I-', 2-l, 1, 2, i s

where

The irlfinitesimal unitary transformtzeisn, U = 1 + iG, that is induced by

an infinitesimal coordinate transformation is given by

The quantum unit of action, h = 1.0515 X 1 0 erg~,,c, ~i s henceforth

~ replaoed
by unity on adopting suitable abrnie units, The generatam P and J are con-
ventionally odled the tinear and angular momentum operators, while H ia the
merw or Hamiltonian operzktor. The generatom af infini-imal velocify changes
have had no verbaf desigattCion. But, in %hisroeket conscious age, the finite!
veiocity tmnsforrnations have eome ta be ~ a l l d"bm~tS."Perhapg one s h d d
tern N the booster. We need to specify group composition praperties for the
new sealsrr p%r&meter6 9 . A genera1 bilinear form for Bfxzlp= - 6 f 2 1 1 9 is

where K, I;, M are constants. The Jaeobi identity, iltpplicsd Co thrm mt8 of
infinitesimd treznsformatians, implies that

One ewily verifies that the eae6cienf of M vanisha identie~lly,bu& thorn of K

md L do not. Henee the latter eoeEelt.nts must be aero, and we have, simply,

Not as a proof but as a mnemonic, we mention, that 6 0 be and 60 &v are

-
really pseudosedars, while 6e 6v is a scalar,
The full set of generatar earnmutation. relations is

where we have adopted the summation convention for repeaM indices; here,
the index m = 1, 2, 3,
The eommuttllor of two generator8 ean be inkvreted, in two dbrnative
wntys, m the effect of an infinitssimal unitav transfomatian upon an operabr:

The eommutaltors involving the anwlar momentum apratior, for emmpIe, can
be w ~ t k nans
.
sUJ= (X/.i)fJ, J * 401 = 6c;~X J,
6,P = fI/z'>[P,3 Gccr] == 6 0 X P, (1-2.13)
BUN = (I/i)[N, J 6 4 -- 6 0 X N,
which state the rmponse of a vector to infinitesinsaf rofstions, and
&,B= (1/$7[H, -
J 401 = 0, (1-2. r4)
which charaeterires H as a rotational scalsr. Analogous linear momentum and
translettioa, rmponse equations are

The response of the anwlar momentum to tramlation is in m ~ o r with

d the nstum
of angular momentum as the moment of the linear momentum, and indicabs
the exisknee of a pasition, vwhr operafor R. such that
(l/$)[R,P . 6e] = 8e
or
1% Pll = a'Skl*
We therefore write

where S is( a translationally independent contribution to fhe ~tn@ar marneaturn.

This is inbrnat a n d a r momentum, or spin. The carrect rotratianal re~ponmaf
P is w u m d by this con~truction,and that of R will follow if the components of
R are mutually commutative, and if S commubs with R weEl ss with F. To
produce the proper rotational hbavior of S, these: operators must the
obey the snljgufar momentum eonnmut~tionrelations, which; we c m abo w r i b a ~ a
S X S = HS. (1-2. a0)
The translational respan% of N indieahs that it can be idenkifid d t ; h -MR,
together with a translationally invariant term. Since a boost ia3: a tmnslatiorz
that grows linearly with time, wt., infer that

All contmutstors among J, P, FiT are reproduced by these eowtmc-tisw.

If I ) is tt dynamically powible ~ t a t eo is U1 ), where U rep
tivlfy transflrrnation, since the veebr ) b s the same eomponenb ars dwe3
I ) in the transformed description. This means that the relativity group genera-
tors are constants of the motion, Such dso is the conclusion of %hecommutation
rdations invol\ring R, if it is uadersfood th&t R, P, S are not, explicit funotion
of t. Thus,
Of courBe, H is not; an. explicit function of 1 for an kolahd dynamical syskm.
The c a n ~ w a t i o nof Pil also appears as

which eleslrly identifies the part3lmede;r M as the i n v a ~ h i emass of the sysknn,

The position vector R moves with constant velocity:

The stmcture of H is determind somewhat by the various eon~errrationlaws.

We note thrtt
- = - 1 [P,H f = - aH
dP
-- - - - ~ [ R , H I =aH--
dt 2' d~ -p ' dt a ~ = 1 M " (1-2.26)
and

The consequence is

vvhieh gives the decomposition inta energy of motion of the whale rs;vstem and
internal enerw, The latter will generally involw indernd dynamkal variables,
which commute with R and P, combined in sueh a way that Hint is invariant
under the rotation generahd by the inbrrral angular monnenlum S.
An dementary particle is a aystem tk'ifiout inkrnd enere, or a t lea& one
for which inkrnal enerw is effectively ined under the limikd physical eir-
cumstances under consid64ration. Let us consider n elementary gafiieles, each
described as above by variables re, p,, Sa and mms m,, a = 1, . . ,n. The.
operators msoeiated with digerat pafiieles commuh, The krretie traxrsforma-
tion generators of the whole sysknt are then obtained additively as

= (pat -- m.~,) = Pt -- MR,

where

and
12 Psnl~les Chap, 1

The oprators for the total sy~teznhave the required propeeiesi. Note that the
inbmal vafiables introduced here are not l i n e d y indqendent :

which is dm conveyed by the eonnmukation relation

If the v ~ r i a u spafZ,ieles are dynamically kolabd, the energy aperator is also

dditive, More generally, we d e s e ~ b einkracting systems by

where the inbrnsl enerw of the system is

and V is a malar fwction; of the inkrnal coordinaks r, - R, p, - (m,/M)P,

I,,and possibly others.
The number of particles being degcrihd cannot be a dynamical variable
apart, from rather speial cireuxxrstanee~,Let there be several digerent t y p s of
pa&icles, with m m e s m,, Then

where N , is the nunrbr of parvticles of Qpe a. Since there am generally no

rational relations among the mttssm of diffefent particles, Che invariability of M
implie8 the comtancy of N , for e a h t y p , An exception oecum; for unstable
pa&icles, as in the a-instabilitgr of nuclei (a-particle kinetic energies can be
suficiently small to ualidak the Galilean regime), Here the mass of the unsl;able
nucleus very c10mly equals the sum of a-particle and residual nuclear masses.
The i;eneral charackrization of inkraeting sys&ms enables one to give a
simple description of the behavior of a particle that is influenced by a macro-
seapie, canlrotlslbfc?enviranmttnt. Since a clwical t h a q of such interactions
underlies the measurement of free paFticle properties, a Le& of mlf-consishncy
is &so involved. Let the Hamittonian operator of a system of par-t;icles be;
divided into two parts: H,, comprising all terms containing the variables of a
given particle; H-,, being all other terns, describing the residual system after
the partiele of interesk has been removed, We rtssume, for simplicity, that the
interaction terms in H, are no more than linear in the velocity and in the spin of
the particle, but we da not inelude bilinear or spin-orbit coupling terms. AC
1-2 Oatilsan relativity 13

though the interaction is not necessa,I.ily electromagnetic, we use a notation

designed to facilitate that identificafion :

It is understood that the noncornmuting opemtors p and A(t.t) arc?symmetrized

in rnultiplieatisn to produce a Merrnitian product. The explicit time dependence
appears as an effective replacement for the actud dependence on the variables
of the external system, as indicated by

The equations of motion are

and

where

In deriving the lmt equation we have omitted such commuta;tors as [(e/e)A, ecp].
This is La be validated, not as a clmsical approximation, but through the
negligibility of the dynamical reaction back an the externat system. We also
note the commutation relation indicated by

We do speeialiae to electromagnetism on equatiag F with a multiple of H,

which is the identification, of the inkrnal magnetic dipole moment,

Analogous electric dipole moments have never been observed. One value of the
gyromagnetic ratio g has a special property. In a homogeneous magnetie field
the veloeiLy and the spill vectors precess about the field axis. The two p r e ~ w i o n
mtes are equal if g =.I 2. The observed values of g are very sligfitly in excess of
2 for the electron f2(1.001tM)] and Lhe rnuon [2(1.001166)], but differ widely
for other padiele~.
 We shall find it interesting to consihr a spidess charged parLic1e that move8
in the magnetic field of a distant stationary magnetic charge. Let the ~oordinate
origin be plseed at the position, of this charge (we now use the letter g to denah
its strength in Gaussian units), so that

This system is characteri~edby the equation of motion

where a symmetriaed produet is understood, together with the connmuta;tioxl

relations
[nit (MV)II = i & k t t (1-2.47)
and

The last equation is in~onsistentunless the particle can be controlled to remain

distant from the magnetic charge, in. the weak sense of r > Q, That; follows from
the Jacobi identity:
O = [[vt, v2], v,] $ cyel. perm. = - 1.
m3c r"

The equation of moments, which uses symnnetrized multiplieadion, is

But since the Hamiltonian is no more than quadratic in the momenla, sym-
metrized multiplication enables one to write
d 1 dH
f (r) = [f (P), H ] = Vf(r) -- = v Vj(r),
C~P
and thereby recognize the consewed angular monnenlurn veetor
~=rXmv-eQL. (l-2,52)
C r
One easily verifies that it is the rotation generator. Them is an impadant con-
mquenee of that fact.: Consider the coordinate wave funetion repre~entixlgrt
p&rticular sLaLe, (rt[ ), and perform. a coordinak system rotation about the axis
provided by r, An infinitesimal rotation ifs given by

and the corresponding generator is simply

1-3 Einstginian relativity 15

The response to a finite roti-ttion is therefore

and the known limitation to single or double valuedness for rotation throul,Th 21r
radians implies that eg/e is either an integer or an inkger plus i.
As a discussion of magnetic charge and its implications this is quite in-
complete. However, the special opraLor system given in Eqs. (1-2.471, (1-2.481,
and (1-2.52) will soon be encountered again in a very difierrznl physical context.

The new feature associated with. the finiteness of c, the speed af light, is the
abandonment of absolute simultaneity, It is replaced, for infiniksimal trarrs-
formations, by

where &c0 is the displacement of the origin for the variable cl. We now designate
the space-time coordinates collectively by zp = et, r, where z0 = --X@ = et
and zk =r= a = r k . The infinitesimal coordinate transformations of the Einstein-
ian relralividy g o u p are

where
ady= - &oVIL.
The six independent parameters of this four-dimensional roLation are relakd ta
6w and 6v lay
li~k=
t E ~ ~ ~ h O~ k = w Bv~
~/c. , (1-3.4)
The composition properties of the 10-parameter group are specifid by

The generators of the unitary tramformation, induced by an infinihsimal

eoodinah transformation are comprised in

The eonrespondence with the GaliEean generators is

and
c ~ =o H +M C ~ .
16 PartZelss Chap, t

we 8hall reco@zle &ho&lythe neeemity far the &M% henera o ~ &b$wmn $he
relativktic and nomlati~&icdomaim (to ug@;the aanventiond l%b& for the
two relati~tie~). As to the compo@i&ioa for tht? wd&r pararnekr 8p, it i8
interesting thaL no vanear s c a k S1121~"= - - Q l a l l ~can. km fomad in flre four*
dimensianal MinkowsE spMe hm.the wectom b l ,pep and the '&mm B1s2WP".
ALccordbgly,
4IZXP == Ot (1-3.9)
and the full, mt of eommut~tomfor the geaemtom is

where QC(* i~the metric bmor specified by

The coxulmuta%amcan &im ?Mpremnkd, as

indic&tingthe respom of veetom and bnsam to infixuSdc3.siml Lorentz robtiom

(comp~~ing three-dimemional rotations and bmts), and

which give8 the traoslaLiontnl rwpom of these apratom, m e a w ~ t b nin

%hrm-dinoemionalnotation, d l the~ecornmatabm reprduct? the G a m n
with t w exeepdions
~ :

In Gdilean relativity, then, J/cZ is neglected snd H is negIeeM mhtive to

Me2, giving the effective replrteement of the operator Pale by the number M.
Eencefo&h we d o p t a b ~ unihc in, which c = X.
There is one obviow reali~atiorrof aff the commutation relatiom for the 10
geger&~om* It ig
P
I > = Q , J"" )=O, (1-3,162
g fatal ilrtva~anceof the s t m e t u m l ~vmuuna.
which d m ~ b the Any
by PO > 0. The scalar
formed from the framlatiomlly invafian* PB,

is invariant under dl operatiom of the Lomndz waup (we are g tbs

i ~ ody
1-3 Elnstelnian relativity 17

trawf~rmstionsthat are continuously connected with the icfetatity, the p r o p r

orthochronous Lorentr group). According as M' is positive, zero, or negative,
the four-vector P@is timelike, null, or space-like. With a, timelike momenhm,

is an invariant prope&y. For M = 0, too,

remains valid under Lorentz tran~formations. But the time component of s

~ p a e l i k eveetor can. be given either aigxl by appropriately choosing the eo-
ordinste system. Thus M 2 < 0 i8 of no interest for physics. The nonnegative
quantity (M2)'/' ig the mass of the system.
Another tramlslionafly invariand object is the peudovectarr

where
*JP@ ~ L C ~ ~ J . ~

forms the tensor dual to J"" with the aid of the totally antisynmretricaf ternor
srpcified by
e0128 E
-M (1-3.21)
Thk invwrietnm prowrty fsllows from the tr~wlationaIresponse of J,x, aad the
antisymmetry of c'"',

We also n o h that
&W" = 0.
The sealiar
W 2= WPWr 2 0
is invariant under aEE Lorentg tr8nsformations. As indicated, the vecfor W',
being orthogonal to PC,eannot be time-like. The @ommudation,relations among
the components of WIrare

The behavim under wordinate disglaeements that is pregexlkd in the

equations
(l/i)[J, P 6 4 = 6r X P, (l/i)[N, P be] = -- drPo, (1-3.26)
again indicates &heexistemer: of a position operator R, obeying
(Qne muat stifie %beimpukrc!to introduce a tirne operator complementauy to PO,
That w d d contrdict Lhe physical nature of the enerw s ~ c t m m . ) A par-
ticular realization of J and N, in which additional displacement independent
quantities do nof occur, is
J -- R X P, N .- PX@--- PR, (1-3.29)
where aymrrretrizd multiplication, is used for %henoncommuting operrthrs R
and P',
(l/i)[R, P'] = aP"/aP = PIP'. (1-3.30)
The eomeet three-dirnexlgiona1 rotational hhavior of slX the operators con-
~ d e r e dis ob%%ined
if
IRXRs0. (1-3.3 1)
The other charaehri~tiecommutator of Einsteinian relativity here reads

lit is obyed without fureher ado, since

i[P"Rk, P0Rl] =. RkP1 - RIPk. (1-3.33)
The simplicity of this result, despite the presence of symmetriz;ed gradueds,
depndg on the fact that cammutztforg of R ~ t functions
h of P introduce no
fudher commutator^ and are nee~marilycscnceled by the Hiermitian symmetriza,
%ion, I n this si$uation the informftt;iszlabout %heenergy operator Lhat can. be
deflved from the eanewation of N,

+
is already contained in the relation PO = (p2 M 2 ) l i Z .
Now let an inkrnaf a n w l ~ momentum
r be added:

As such, S m u ~ commuh
t with R and P while itself obying the a n v l a r mo-
mentum commutation mles. It r s
i also necessary to supplement P4 irt order to
generab the spin h r m of the commutator

A suitable form is
N = -- P'R + a(P")S X P. (1-3'37)
The ealeulettion of N X , N involves
-~P@RX [a(S X P)]-- i[a(S X P)] X P'R +
a2i(s X P) X (S X P)
+
= ( d a / d P ' ) ~X ( S X P ) 2P0& - a2[pZf5
- P X (S X P)], (1-3.38)
and the r e q u i d resulL is obtained with

We cone1ude that
a(Po) = (PO + M)-',
the alternative choice with ( P o - M)-' being singular a t P = 0, P' = M.
The 6nsl form is
J=RXP+S,
N = PZO - PR + + M SXP,
whieh incidentally shows that S' is B Lorenta invariant. It is worth pointing out
the converse, that operators with the slat& properties of R and S can be eon-
structed from the Lorentz generaton (zO= 0):

The exceptionat position of M = O is evident here,

The components of the pseudovector Wfi are given by

or
w 0 =P+,
(l-3,451
W = P@S- + M P x ( S X P ) = M S + ~ , + P~ P - S .
The last relation can dso be written as

There is a eonnee-t;ionamong the several i ~ v a ~ a n t s :

W 2 = M'S'.
This discussion refers generally to M 2 > 0. We next consider the limit ss
M' -t 0 for fixed .'S The resulting relation
can be $ven the cova~antkm

where
X = P *SIP0
is a Lorents invariant. This quantity is the component of the spin along the
direction of motion, or the helicity of the particle. In view of its invariance, a
phy~icalsylshnn need exhim only an@value of helieify, or, if ~pwemfieeLion
parity has a meaning for the interactions of that system, the pseudoscalar h osn
have two values, fs. The photon, with s = 1, illustrates the latter situation,
white for the neucfinos, with g = 3, X = +s and -a refer ta emntiaUy dZ~?ren%
p&&icfes, f f only one helieity vatue is meanh@ul, or with s 2 1, even if h = trl
+
and --s are both resliaed, not all of the 28 1 spin magnetic quantum number
ststes exist. Accordingly, the operator S cesses to be defined (with two excep
tions) in the limit M -+0, and we must introduce m w vaGabfes for this eir-
cumstance.
f n. order to deleb S we define the new position vector

wbieh i~ such that

k XP= R X P+S -- PPeS/(P0)2. (1-3.52)
Then
J=~ZXP+X(P/P@), N=Pz@-POR, (1-3.53)
and to complete the verification that only X appears explicitly, we give the
contmutabr

This is the operator system that we anticipated in discussing magnetic charge.

The co~esps~dence is

and the restriction r > O is here validated as the Lorentz invariant energy
property P@> 0. The absence of certain helicity values is now manife
the noneommutativity of componenh of R, This htrimie ~oxzlocalityof m
lew particles is de~cribedby the unm&ainty principle

or, for a nnomentum s b b with some d e g e of dimtionality,

indicating that rm average wavelen@h, roughly mk the scde of c m
spec%eability. Incidentally, when the explieit constructions of J snd N am
inw*d in the fomulm for M R and MS, thme exprwions do v&nbh, m dom
M&*
I n the situation we have just discussed, W' = 0. There is another logical
possibrtity. With X = P SIP@assuming any accessible finite value, let S
' -+ a,
as M 1 -4 0 to produoe the limit

The chetr"acf;eristimof S Gve t h w opemtors the follovving p r o p d i a :

and
[h, T2]= 0.
The h v a ~ a n t
1["2 ,
,
can be assbed any pagitive v a b . The componenb of T act to change h by
&l,and t'hk Pvithout Itimi*. We now have

The commutation relations among the oomponents of W,, which am mtisfid

t ~ v i d l ywhen T == 0, hem demand that
( l / i ) ( p o4
, = T X P, (l/i)(hP + T) X (XP + T) = P-, (1-3.63)
and t h w are valid stakmenk aboiut T. We continue to urn the wt~ifionveclor
R =R -. S X p/(PO)' f 1-3.M)

but we must be; camful h no& tbaL M 0,

"IKe now find, ss %hecounhmsrt of MS -.,T, ths&

MR = T X P/(P')'
' --t m with vsnishing M. However, M R = 0. The e
and R htion
mhfiom tN X N = J continue to be obeyed despite the bLrodue1ion of the T
brm, since
QX +
(T X P / ( P ~ ) ~ )(T X P/(PO)~)X = 0. (1-3.69)
This involves the commutator

wfiieh is also used to verify that J generates the rotations of I", The significant
obmnta;tion is that X has eeawd to be ft Lorentz i n v a ~ a n t :

This faet, together with the unbounded nature of the h spectrum, ranginf; over
all inkgers or all integem ++, indicates that physically acwssible states would
exist for which (bk12is arbitrarily large.
We suggest the following verbal principle for massless particles: A zero
mass padicle is not completely XoealizabXe, but a finite degree of localizability
exists. The principle has the: following valid consequences. There is na aginless
zero mass partieb, for the eomrnutative position veetor R would be available
The same reasoning exelude s = 3 massless pa&ie1w for which space reflection
parity is meaningful. And the systems we have just discussed, with W 2> 0,
are condemned wholesale by the existenw of state8 that are unlwalized without
limit. There is a simple pattern for the kxtown or strongly caxljectured maesless
garlieles; their spins are given by 8 = Z4, cr = -1, 0,+X.
The conwpt of elementary particle in rellitivistie mechanics remains an
operational one, Lhat under the eonditiorzs of physicafi exeifa$ion available, if is
consistent to @sign a unique vazlue ta m-, spin, and ather ehttracterisfic
invariant a(ltributes of the system. Far a set of n noninkraeti~gp & ~ i ~ the le~,
Lorentz generators of the whole system are giwn by the additive forms

The operators R and S for the total system must be obtained from the con-
atmctim (1-3.43); one is not likely to produce them by an a pfiori definition,
Consider, for example,
1-3 Einstainian relativity 23

We approach the topic of interacting particles by giving first a relativistic

generalisation of the nonrelativistic treatment of a particle moving in a macro-
scopic environment, Xn order to make covftriance more explicit we define a
proper time derivative:

with the usual symmetrization understood in the last term. Thus, for a single-
isolated prartiele, we have

With attention restricted do a homogeneous electromakgnetic field, the covariant

generalilizlations of Eqs. (1-2.39) and (31-2.40) are stated as

The constraints

are compatible with the equations of motion, a t least to term8 linear in the field
strengths; this involves only the commutation relations for a free particle,
Why bid we not begin with a general theory of inkrating particles, specified
by variables r,, p,, S, a = l, . , . , n, and then proceed to follow the mation of
one particle under the infiuence of the others, as in the nonrelativistic discussion?
Quite simply, because no such &;eneraltheory exists, Apa& from the obviously
formidable algebraic task of stating the relativistic conditions on in_leraction
brms (smdl, deviations from nonrelativistic behavior pose no problem), the
atternpt founders on the failure of the assumption that there is a fixed number
of parlEticles, The relation between reiati~sticand nonrelativistic energy can
be exhibited as

1x1 fhe nonrelativistic limit where changes in H are small compared to each ma,
the conservation of P' generally demands, first, the conservation of each N,,
and then, Ghat of H. But if the kinetic and interaction energies contained in H
b o r n eotnp&mblewith in&edud m, valuw, one ean no longer oonclude th&
the Na mmsin c o ~ s b n t . It is tbe charak6stic featwe of mlativbtic padiele
dynamim that pafticles can be @matedsnd wanihilahd in high e n e r e en-
@ounkm.

1 4 CRITIQUE QF: PARTICLE THEORIES

expriment abundantly confim, that the concept of
le objeet k unknable under pronounced relatietic
Them have Bwn two extreme reaction5 frO t h i situation,
b k r a ~ t i o n c~n.&%ions.
s ~
They rwpond fd) the f&lum of a detailed space-time description in. particle
language by: (1) bisting on the pomibilily of a detailed ~pacie-timdescription
but in h r m of a concept more fundamentat &an pa&icle; (2) rejecting the
possibility of a detaifd space-time deseriptian by denying that any concep%
underlies that of particle. We shall give brief derseripliom of these attitudes.
2. More fundamental *an parlticles carfiers of physied prop&ies an? the
volume elemenb of thrm-dimensional space itmlf, If the sped of light l i d b
wery means of commrunication, disjoint wlumw at the Bame time are phy~icatly
indewdent and &odd contribute additively to the tot81 emrm and momen-
tum. Using an evident Iimiking prmedure, we v v ~ k

where TOO(%), ~"(z)are functions of the dynamical variables a t time z0 that

eonvey the physiGal situation in the infiniksimi neighbarhood of the p i n t x,
The d y n a ~ e avafimtbfes,
l ss operator funetione, of space and time eoordinsb~,
sm oprstor fiefds, and the approach we: am describing e m be callled operator
field theory. As the above notation suaesb, covariance can be made mpli~it
by idenfifying the vohme element (dz) urith the time oompnent of a direckd
element of sres on s plane space-like surfaee in four-dimensional space. This

which i n t e p b are independent of the surface tr according to the consemation

of P". Qn h t h g the null diRerenee of two such i n t e g a l a an e q ~ v a l e avolume
t

we recognize the suficieney of the load condition

1-4 Critique of particle theories 26

The conservation of the six other Lorentz generators, regarded aa moments of

momenta,
/.
J ~=' ~ U A ( Z L - T ~x'T'~),
' (14.5)

is assured if
TP'(z) = T'"(x).
The three-dimensional form of these operators is

The tensor transformation response of the stress tensor Tp'(z) to an in-

finitesimal Lorentz transformation is given by

The possibility of producing the new operators

by the associated unitary transformation implies the commutation relations

Integrations over a space-like surface, employing the stated properties of Tp',

will reproduce all commutators for the 10 Lorentz generators if one uses the
following integration theorem for a system that is closed in space-like directions:

The commutators of quantum mechanics express the mutual interference of

measurements on the two properties involved. The physical independence of
volumes in space-like relation thus requires that

When the coordinate system is so chosen that xO= xO', an everywhere-valid

expression for such commutators must involve b(x - X') or a finite number of
derivatives of this function. For the energy and momentum densities, which are
the Tpvcomponents used to construct the Lorentz generators, the implied form
of the equal time eonnmutation relations is

The brms involving two or more derivatives are such thizt l-hey do not con-
ffibule wbem intepations are pedormeb to eonstmct one of $he Loresnde genera-
fors, We have indicated only the minimum n u m b r of d e ~ v a t i v erequired;
~
more generd possibilitks are introduee-d by appropriitb generaliaa;tion of f , 8,
snd h, The% three functions are symmetrical wikhin each pair of indices, as
iZlu~tr~tedby
f n r r f p p - f nm1plr = fna.rrq (1-414)
~fhilef and h are antisymmetrieaf under an. exchange of the pairs, as in

Anokher relation is
-aOf m n * p q ( z ) Bmnrpg(Z) Q ~ ~ * ~ ~ ( c c ) . (1-4.16)
There is a simple example of a system for which none of the additional
derivative k m s appear, We begin with the energy and rnornentum densi%y
expressions thrtt are identified with the classical electromagnetio field:

Thrt attempt to r~producethe e n e r g density cammutator

sueeeeds with. the commutation. relations

if the mornen$um density spertator is interpreted as a synnmetrized P ~ ~ u G1& n .

a r ~ v i n gst the desirt?dEorm we have used the formal ddLa funetion property
The commutators among the momentum density components also eoxltain no
higher derivative terns, but to reproduce the required structure it is necessary
to impose the following conditions,
V-E(z>=O, V-W(Z)=O, (1-4.21)
which are compatible with the commutation relations. The commutators
between energy and momentum densities then follow the anticipakd pattern and
supply familiar expressions for the stress components Fkr;in particular,

Although we have hewn wiLh a suggestion from classica,l physics, this

discussion is a self-contained verification of a Lorent~invariant quantal system.
Other properties are now derived from the stmcture of the Lorentz; 15c;nerators.
From PO Byededuce the equations of motion of the field operaton, which are
the homogeneous; Maxwell equations. The Lorenta transformation behavior of
the field strengths is that of the antisymmetrical tensor FP.. As an example,
consider

(1-4.23)
Then, since
(lli)[Ek(x),T0@(2')1= cttmar 6(x - xP)Hm(z'), (1-4@24)
we get the infinilesinral response
~ E ( x )= -&V (zOv+ xaO)E(z)- &V X H(%). (1-4.25)
We add brief comments about more realistic system^, in ~ ~ b i the
c h electro-
magnetic field interacts with other dynamical va~ables,If we are to maintain
the geometrical transformation properties of F,., the added terms in Toomust
not alter the computation just performed. That excludes from (Ex(z) , To'(z')]
any additional single derivative of a delta funetion, giving the general form,

f d is no longer necessarily tme that E is divergeneeles~,and, on. w ~ L i n g

V E = j', (1-4.27)
3% Particles Ch~p.3

Among the consequences of these relations that are produced by integr,ztiot~l;;

over X' are: the inhomogeneous Maxwell equations with P = ( j', j) identified
as the electric current vector, the local charge consewation law,

and a Lorentz; transformation response, affirming the four-veedor status of j',

Ex~mplesexist of interacting syskms for \vhieh the very singular terms of
the energy density commutator do not appear, but there are evere restrictions
in the choice of dynamical variables. Basically, only scalar, vector, and simple
apinor fields are permitted. At issue here is the consistency of the operator field
hypothesis, that meaning, albeit ideali~ed,attaches to the physical properties
awoeiakd with a sharply defined geometrical volume. To examine this queslion
we consider various weighM averages of the e n e r a density st a dven time,

and ecmstruet

This is the basis for an uneeftainty pfineiple stattlment about the aceuracy with
whieh values of T 1and !F2can be assigned in a even state. We firgt consider an
does nod enter, in comquence of the antir~ymmetryin.
application where f m"nnppa
the two sets of indices. Let T 1and ZT2 be partitions of Lhe total e n e w operator,
so that
@I(x)4-@ 2 ( ~ zz=) 1. (1-4.32)
Since derivatives of v1 and of v2 diaer only by a minus sign, we find, simply,

Now choose vl (X) to be a, unit step function, defining a semi-infinib redan which
shares a, surface \dth the complementary volume defined by etz(x). With clS an
element of area directed from the latter volume, we get

This gives a eorreet acmunt of the rate a t which the enerw in each p r t i a l
volume changes, owirtg to the enerw flux acroas the common surface. Irxei-
dentally, if the domain8 defined by v, and v2 had been regarded as disjoint but
approaching contact in a limit, the value abtsinied for the Il.igh&band side would
have b e n zero, while, if they had initiaEXy overlapped the eventual boundary
and then the cornman. volume had approached zero, the limiti~~g value of the
1-4 Critique af particle thaoriss 2r)

right-hand side would have been twice the stated one. Thus, an alternative
evaluation uses the average of the two limiting definitions. Ar~stherchoice of
weigh%functions is
@ (X) .l @(X), VZ(~) xkv(~), (1-4.35)
which gives

When @(X) is a unit step function that defines precisely a finite volume, the
operahrs T t , T 2are the asmciated energy and its first moments* But no mean-
ing can then be assigned to the products d,, d,v d,u, which calls seriously ints
question. the eonsiskncy of any operator field theory for which. $nanlpS1 (2) @ 0.
This gives a pPivileged position to that limited clslsl; of fundamental field variables
for which f does vanish,
The impact of this result is only slightly weakened by the follojving property
of physical systems that have vanishing fmnlpq.The funetion

cannot be zero, and it is correspondingly impossible to specify simultsneousfy,

within ally finite precision, the total enerlSy. and a, component of total momentum
that are associated with a shsrply defined volume. [The term physieal system
occurs here as a reminder that the vacuum state, with a11 its atkndant proper-
ties, must be compatible with the assumed eharacteristi~sof the system. I n
particular, the zero enerm and momentum invariantly assigned to the vacuum
state 1 ) require that
(TO0)= 0, TO^) = O (1.--4.38)
and also
(T") = 0.
There is some freedom to adjust the definitions of the TB"by additive constants,
but, as inspection of the [ ~ ~ ' ( z )To"(z')]
, structure will confirm, it is limited to
a multiple of g"^". A nontrivial requirement is thus given by

I n the example of the electromagnetic field, tvith Tkr = Too,it is impossible t o

satisfy 4(T0') = O since TO' = +(E' +-
H%)is a positive-definite operator; the
uncoupled electromagnetic field is not a physieal system. It is a t least coneeiv-
able that the vacuum. properties so cireumseribe the possible dynl~micalvariables
and their interactions that the real world is selected.] We consider the corn-
mutator of the energy density f'unctional
with its time derivative
(dx)v ( X ) TO' (g),

which gives the vacuum expectation value

(dx)ak~,v(x) (gh*")a p aCIu(x)
= ~(TP'T). (1-4.43)
We recognize the necessarily positive expeetaticm value of the enerw in the
nonvacuum state I" ), Tbc numbers thus form a positive-definite
matrix, but there is no guarantee that these numbers are bounded, fL is clear
from this discussictn, however, that a statement h u t momentum drtnsity is also
one concerning the time derivative of energy density, and this additional
dynamical specifiability may be unnecessary to the self-consistency of the theory.
The particle, in operator field theor?., is a derived dynamical concept. To
eonst~lletfrom a few fundamental field variables a rel%tivelylarge number of
Btable or qumi-stable excitations-particles-is the ambition of this viewpoint.
A elawifieatian of particle spectra is produced m follows. Let X(%) be an algebraic
earnbination of the fundamental field variables, so devised that it has an ele-
me&ary response ta Lorenfz transformations. This includes the requirements
that, st x = 0, the rotational behavior corresponds to a definite intrinsic angulsr
momentum or spin, while translaLional msponse is pa~ametriz;edby fbe eo-
ordinates zp,
F(z)t &I E (1JfJ apxcIz>- (1-4.44)

The finite unitary operator presentation of the latter is

where P;E:=2= P""z@and X is evaluated at the coordinrtte origin. The state x(z)[ )
is produced from the vacuum state by a localized excitation. To study the
pfdSele aspects of this excitcltion tve examine its spaert-time proipltgation
characteristics through the correlation %vithan analogous excitation having a
digerent localization :

The unitary operator fhnt describes the displacement from z' to z can be
exhibited in terms of i t s eigenvatues and the associated non~egativeHermitim
projection operators,

where
( d p ) = dpo d p , dpz dp3.
The values of fl that contribute to the integral, thwe for which F(p) f 0, must
conform to the physical sp~etral.requirements,

With a given threcdilnensional momentum, (dp012 = d 3 f S pso that

The di&rentjal dcJ, is an in~rztriantm o m e n t m space m s u r e on the hyper-

surface --P2 = M 2 . This gives

is a real, nonnegative function. The state X / ) selects from P(p) the subspace
with the angular momentum propexrlies implied by the rotational behavior of X ,
snd f ( p ) j.I. 0 at - P 2 = M 2 a~sertsthe existence of an excitation with those
physical parameters. Merely for simplicity, we only consider a scalar field X ,
which limits f ( p ) to dependence on the scalar --p2.
There are three qualitatively difierenl possibilities that can be realietxi in
f(p) = f(MZ).
a. An isolated mass value appears in the spectrum,

For a given spaLial momentum, the time dependence of the field correlation
function contains the isolated frequency 'p = +(p2 +m2) This excitation
i s s stable pareticle. We n o b thaL should ~ ( sobey
) a finite-degree differentid
equation,
) o,
ul(-a"~(~=

and f(di2) is eomposed entirely from delta functions.

b. There is e. pronounced inerease in f(M2) above a smooth background, which
is cenbred a t M = m and hss a mass width measured By T" << m. For a specified
momentum, the time dependence that is associated with this port;ion of the
mass spectrum is given by f t ='2 -- zO')
32 Psrticlss Chap. Z

in which is the energy computed from mass m. Owing to destructive inter-

krence, the amplitude of this oscillation will drop substantially below its initial
value when
(P~/~)(I/P)* (1-4*37)
This is an unstable particle decaying into several other particles with a proper
lifetime -l/I".
c. The function f ( M 2 ) varies smoothly. In ths't part of the spectrum several
gartieles are present, with no tendency to hcome associated in a single unstable
particle .
Some aspects af field equal time commutation relations can be extracted
from the comelatiorr funetion. Thus,

where
exp[ip. (X -- X')]. (1-4.59)

If the field ~ ( zis) a fundamental dynamical variable, its equal time eornmutrztion
relations have a, kinemaitical bwia, I"cs typical af a sedar field thaL

and the fatter implies the sum rufe

Imagine no\' that the field X ( % ) is uncoupled from all others, and then.
obeys a linear digerentirml equation that gives f ( M 2 ) = b ( M 2 - m:), Supipom
that when the physical couplings are restored a stttble particle still exists. Its
mass will be shifted by the interaction, m. -+ m, and f ( B f 2 ) will have multi-
particle contributions in addition to the discrete mass term: f 6 ( d f a - m').
The sum wle thus requires that f < 1. If we are not inkrested in the details
of the particular excitation used to generate the psrticfe, but wish only to
describe the physical particle itself, 11-ediseard the eonfinu~usmass cantribution
to ( X ( Z ) X ( Z ~ )and
, comespondingly adjust the scale of the correlation; function
by removing the factor f , This example has supplied the designation for the
general proceduw that transfers attention from the fundamental dynamieal
geld variables to the derived phenomenalogical ptarticle level. It is called
renormalisation,
More elaborate field earrelation functions give information about the
inkraetion of particles. Consider, for example,
~vfierethe various fields are the algebraic combinations of the fundamental
dynamical variables that contsin the pecrticles, a, b, c, d, respectively, in their
+
excitation spectra. Xn order to refer to the pa&icular rertcticm. c d -4a b +
(we do not consider here properties like electric charge that i n t r d u w the
distinction between partielee and antipsdiclw), the regions in. which the co-
ordinates are placed mmt be appropriably &own. The points x and z q i e far
in the future of z" and xf"', while z is widdy separabd spatially from z' as ia zrr
from X'". Under these circumstances the renarmalization prowdupes that
isolate the physical particles can be applied independently, and the resulting
function of particle properties supplies the amplitude for the physical reetetion.
NOWrecall that the dynamical variables of quantum mechanics fa11 naturally
into two sets .tr.hich, at a given Lime, exhibit commutstivity or antieommuta-
tivity, respectively, between a, pair of variables referring to diEerent degree8 of
freedom. I t is points in space-like relation that play the Xathr role in operator
field theory. -If x and s h r e suaieiently separated spatially that detailed corn-
posite structure is not involved,

tvhere n,, nb are the respective number of anticommuting fundamental field

variables used in constructing the operatom ~ ~ ( $ ~1~, ( z ' ) )Only
. the even or
oddnegs of the inkgers is signifiesnt, and the sign fwtor is 3-1 except when both
n, and nb are odd. Should a .and b refer to the same pa&icle type, the field
operators amciated with a pair of spatially separated poinb eammub or anti-
commute according as the number of anticommuting fwdamental variablw is
even or odd. The two-particle state produced from x(z)x($")l ) by renormalisa-
tion is correspondingly symmetfical or antisymrnetricaf in the pa&icle variables,
qpropriak to particles $hat obey Bose-Einstein or Fermi-Dirac ~tatissfies,
respectively, The same statir~tics-labeledproflrties enter in rel~tingreaeliom
in which the rofes af some initial and final pa~iclesare inferchanged, Thus, the
field correlation function

+ +
can be used ta derive the rewtian amplitude for b d -+ a e, If the two
eamelslion functions we have mention& %re known throu&oul the mulLiple
spaee-time domain, they are known for the re@onswhem xt aand 2" are in
like relation. But there they are equal, or diEer by a minus sign, depending
upon the statistics af particks b and c. According.Ey, the two function8 are
space-time extrapolations or continuations of each o$her, The impfied connec-
tions among digereat reaction amplitudes are usually referred to as eroming
relations.
Dynamics is explicit in oprrtlor field theoq, It is conveyd by the nonlinear
s t m ~ t u r eof the field equations obyed by Ghe fundamelltal dynamical va~ables,
They, in turn, imply equations carmect;ing the vafious &Id eorrelatiort funckions
84 Particles Chap, l

from which the latter can be constructed, in principle. Two radically different
situations occur, in practice. In the first, the interaeticlns are suaciently weak
that the particles of interest appear in the excitation spectra, of the fundamentstl
variables themwlves. This is the ~zssumedsituation in quantum electradynamics
where the particles are the photon and the electron (or muon). The equations
obeyed by the field correlation functions can be solved by ~ r t u r b a t i v eor itera-
tive methods bawd on the mallness of the eharaeteristi~coupling constant
a = 11137.Q.8. The result@may be presented in, twa different ways, a t thct
unrenormrtlized field stage, or at the renormalized particle leveicl, The fieid
version contains divergent integrals, the renormalized stabmenti3 are finih an;d
in exceptional agreement with experiment, The fairly rapid convergence of the
renormalised expressions means that experiments do not probe to very high
moment%or very short distances. The underlying hypothesis of operator field
theory concerning the conceptual possibility of descriptions at arbitrarily small
distances remains unksted by the available evidence. This hywthesis is in-
volved in the unrenormalized, field results, but whether &henonexi~$enceof the
expressions signifies the failure of the hypothesis, or merely the inadequacy of
the perturba'cive ealculational methods that are used, is premntly unknown, It
may be that opemtor field theory is unnecessafily dognatie about the physical
silpificanee of arbitrarily small volume elements. Totafly new concepts might
enter a t very large momenta, without altering the practical succesws that have
been obtained.
The other situation is that of &rang interaction physics. Here %hehypothesis
that whole families of particles are the dynamical manifestation of a feu?funda-
mental field variables excludes the possibilit;y. that the excitation spectrum of
the latter contains the known. par%icles. These objects must b generakd by
combinations of the basic variables, Being iporant of the underlying dynamics
of the fundamental dynamical variables and lacking the computational methods
that could @v@the consequences of that dynamics, if it were known, one musk
fall back on speculations concerning the composib field structure of the known
partieles. And such speculations become interfwind in the more immediate
problems that are presented a t the phenomencrlo@cd level. Is it not possible
to separate particle phenomenolqy from speculations about particle stmcture?

2. If the particle is the ultimate structure, no detailed description is passible in

regions of intense inbraction where the eharzteteristic additivi* of independent
particicle conhibutions eea~esto be valid. All that can bc3 done is to compare Lhe
state of noninteracting parcrtieles afker a collision with the s t a b of the generally
diffe~entnumber of noninteracting pa&iele~prbr to the collision. That relation
is an abject of calculation in o p e r w r field theory. With the present viewpoint
it is the fundamental quantiLy which, through its postulated prope&ies, gives
the?eomplete statement of the miera~~opic dynamics of particles. When we write
1-44 Critique af particle theories 36

we introduce a transformation from one complete set of noninteracting particle

states to an analag~usset. Hence the operator S is unitary,
8% S 88" 1, (1-4.66)
It is invariably referred to as the S(cattering) matrix, perhaps because one is
unable to handle more than a small set of matrix elements. Thew matrix ele-
ments are explicit in the probability amplitude

The absolute square f (i is interpreted as the probability that the transition

2" i will oceur during an unlimited time interval. The abhorrence of any
vestige of a detailed .temporal description restricts the obs62med padieles to
stable ones. In the same vein, all refexrenee to spatial desehption is rejeeted and
only the momentum speeificatioa of states, together with spin and other proper-
ties, is admitt&,
Xf the particles h a g p n not to inkract, the appropriate S is the unit operator.
Thus, more interesting than S is S -- 1, which o b y s the unitary restriction in
the form

Since a collision must respect the overall conservation of energy and momentum,
we write

thereby defining the transition matrix T. Only the momenta have been made
expfieit here, and the delta function is four-dimensionsf,

The resulting farm of the unitarity condition is nonlinear, relating matrix ele-
ments of i(T - T?) to products of T' and T matrix elements. The probabilities
of transitions must have a Lorentz invariant significance. Xt is therefore asserted
that the S, and the T,matrix: elements must be invariant functions of their
arguments, This is partieuIarty simple %.hen slf parlicles are spinfess since it
requires that the ' I '
matrix elements be a funetion only of the independent
sealars (we ignore the possibility of pseudoscafars) that can be formed from the
N = n -t- nhomrnenta, which are individualy subject to the particle mass
relations, -p: = m:. The number of such scalar combinations is 3N -- 10,
where the subtracted number counts the Lorentz trrtnsformatioh parameters,
+
Thus, in two-pa&icle reactions, where N = 2 2, there are only two inde-
pendent scalar variables, c~rre~ponding Co energy and scattefing anlSfe in the
cenkr of mass reference frame.
The constructive principle of S-matrix theory is the gostufab of analyticity.
It is =sunned that the physical reaction amplitudes, in their dependence on the
scalar variables, are boundary values of analytic functions of corresponding
complex variables. Since analytic functions are specified by the nature and
location of their sinwlaridies, the dietemination of the I8tter encompasses all
the physics that is admitted in S-matrix theory. Here are the words of some
enthusiasts: "One of the most remarketble discove~esin elementary particle
physics has been that of the existence of the complex plane,'' ". . . the theory of
functions of complex variables plays the role not of a mathematical tool, but of
tb fundamental description of nature inseparable from physics . . . .' T r o m the
viewpoint of analytic functions, the elements of T and are distinguished tts
boundary values of the same analytic function that refer to opposite sides of the
real axis for the relevant complex variables, The resulting discontinuity state-
ment is the condribution of the unitarity condition toward determining the
aingularities of the transition matrix. But the postulate of analytieity also widens
Lhe scope of the unikarity condition to include the so-called crossed reactions.
The conversion,of an initial or incoming particle into s final or outgoing particle
is formlly exprewd by the substiLution pp + ---p@, as judged by the contribu-
tion to the net energ5r-momentum balanee. This is to be achieved by rtnalyt;.rc
continuation, and the unitarity condit;ions for the various reactions that are in.
crossing relation give singularity information. a b w t me analyLic function in
various domains of its complex va~ables.As to whether this kind of information
8uEces: "The S-matrix is a Lorentz-invariant analytie function of all momentum
vafirtbles with only thow sixrmlarities required by unit~rity."
There is na explicit statement of dynamics in S-matrix theory. And the
p~sibilityof regardiw some partieles as fundamental and deriving o t b m r;ks
b u n d states is rejected as an urraeceptabfe stmeturing of the padicle concept,
distinguishing elementary and composite particles. To prevent just such a
distinction being made, it is proposed that, no matter which particles one uses
to construct composites, the same total particle spectrum emerges. This view
of dynrtmi~slself-consistency is usually rehrred to as the "bootstrap" hypothesis
The discussion of Section 1-3 indicates that S-matrix Gheory is too dog-
mslt3c in digmiming all reference to microscopic space-time description, Whether
or not one wishes to recognize it, the! structure of the Lorentrr;group itself gives
meaning to spatial loealizability and temporal development, outside redons of
intense inkrzletion. The very nature of a collision involves a memure of spsee-
time causal control, and the existence of even a limited microscopic space-time
dewriflion indicaks Lhat cacusality is not likely to be restfieled to macroseopis
circumstances. It is widely recognized that the intuitive physical property of
causality in spaee-time must underfie the abstract mathematical asartion of
analytieiQ, Should one not be able to exhibit and exploit causdity as s eou-
structive principle, thereby relegating analytieity to a secondary, derived role?
And as for the b8sic hypobbesis of S-matrix tl.leory, that the particle is the ulti-
mate unandyzable entity, we again ask: Is it not possible to separate particle
phenomenology from speculations about partide structure?
 SOURCES

The critical eornmentrs of the last section set the stage for the introduction of a
new theory of particles. I t is a phenomenologieal theory, dmignd to describe
the observd particle@,be %heystable or unstable. No speculations about the
inner stmeture of pa~ictesare introduced, bu"che road to a conceivable more
fundamental theory is left open. No abstract definition of particle is devised;
rather, the theory u8es symbolic idealization8 of the reali&ie procedures that
give physical meaning to the particle wncepL. The theztry is thereby firmly
grounded in %p&ee-time,the amna within which the experimenter manipulates
hia tools, but Lbe question of an ultimate limitation to miero~eopiespace-time
dmchption is left open, with the decision re~mvedto experiment. Correspond-
ingly, no operator fields are ursed. The compbmentary momentum-space
description plays an important role, but the possibiliw of ultimate limitations
on this space is not exciudd, and there is no appeal to andytieity in momentum
space. The constmetive principles of the new thwry are intuitive one&-
causality and uniformity in space-time. What emerges is a thcsory inkrnedict-te
in position. between operator field theor;)r and 8-matrix t'fimry, which rejects the
dogmas af each, and gains thereby a caleuXationaE erne and intuitiveness that
make it a worthy coatender ts displace the mrlier formulations.
The range of the term "particle" h.m been systematicalfy extended by
expe~merrtaIdiscovery. From the stable electron. and proton, to the Iong-lived
neutron, do the rapidly decaying ?r and A, to the highly unstable p and N*,
there has k e n a progression to more energetic and sfiorbr-lived excitations* It
Is now the normal situat.ion that a particle must be creawd in order ta study it.
And, in a general sense, that h also true of the very high energy stable particles
produced in accelerators. One ean regard aff such creation acts 8s cofli~ions.
The emence of such. a collision is that it occupies a finite, and to some degree
controllable, space-time reeon wherein other particles combine to transmit to a
particular one thee properties that call it into exisknce and uniquely character-
izie it. If i s psrt of the experimenter's creed that a new resonance not be admitted
Lo full status as ra particle until it h a been ob%rved wifh the same charae-t;eiristies
in a number of diflerent reactions. Thus, if a pa&icle is defined by the collisions
that er-ettb it, the details of s sgecifie reaction are not; relevant and one ean
idealize the role of the other particles in the coIlision, recognizing th& their
funetion is solely to ~uppjtythe needed baiance of properlies-they eonstitute
the source for the particle of inderest. What survives in. the idealization is a
37
38 Sources Chap, 2

general specification of the regioxr in space-time where the source is effective,

lvith some numerical measure given by a function $(X), and a statement of its
ability to produce various momenta, as measured by a functian S(p). The ttvo
source functions estlnut be independent but must convey the quantum mechani-
cal eomplementarity bettveen these descriptions-the more detail that one
possesses, the less is permitted t a the other.
We have spoken of particle creation, but equally important is particle
detection. This is invariably achieved by transmuting the parLieXe% properties
into other more easily handled forms, I n iil. general sense, the pa&icle is an-
nihilated by the process of detecting it, Were too are collisions with their
controllable space-time aspects which, in principle, involve the same mechanisms
that create the particle, The receiving radio set unavoidably radiates, the
r-meson created in nucleon collisions is captured in nuclei. Long-lived p a ~ i c l e s
may decay, and thus be detected, by mechanisms too weak to be useful in
creating them, but this option can be overridden a t the choice of the experi-
menter-the neutron is not generally observed through its 6-deeay. The callision
procgsses used to detect a particle can be ideaiiaed as sinks wherein the particle's
properties are handed 0x1, in zt, tvay that permits some memure of spsce-time zznd
momentum descriptiox~,but sink and source are clearly digererrt aspects of the
same idealizrttioxr arid we unite them under the general, heading of '4~ouree.p'
We now proceed to give the source concept a quantitative frannekvark, beginning
with the simple situation of stable, spinless particles.

2-'l SPIRI Q PARTICLES. WEAK SOURCE

The elementary acts ta be represented as the effect of a source are the creation
of a single particle tvhere none existed previously, and the annihilation of that
single particle. Since the actual presence of other particles in realistic collisions
is abstrackly portrayed by the source, the s t a k s appearing in corresponding
quantum mechanical amplitudes, (1,\0-)~ , /Q-), the vacuum
and ( o + ~ I , ) ~are:
state before the operation af the source K; (Q+[, the vacuum s h t e subsc;quent
to the operation of the source: (sink) K; and (l,j, II,), describing a single particle
state in lvhich the momentum is specified rr-ithin a small volume element (dp).
The connection of this discrete labeling with, the continuous variable specific&-
Lion of momentum states is

The individual creation and annihilation acts are not analyzed; the source is
defined as a measure of the whole process, as suggested by (~veanticipate a
particular variable factor)
2-1 Spin O particles, Wwk source 31)

Subscripts appear here, kmporarily, to distinguish sources efieative in emission

or absorption. The designation "weak source" means that the definitions are
appropriate %?henprobabiliw amplitudes referring to creation or annihilation of
several particles are relatively negligible, We procecf t;o make these definition8
mare precise.
The states ( p ! , Ip) refer to a particular time, or, more covariantly, a space-
like surface, f f the origin of the space-time coodinate frame is diqIaced by P,
eorrmpanding staLrss are produced by the unitarry transformation:

Since these states;play the analogous role in the new eoordinah ssystern, they are
assaciated with a space-like surfaee that is displaced by XQn the initi~lco-
ordinate system. But all that is significant in the probability amplitudes
( p l ~ - ) R and ( ~ + j ~ is) ~the relation between the space-like surface and the
space-time localization of the murees, for the vacuum staks are invariant.
Equivalent to the displacement of the surfaee by X" is the rigid dilsplaeement of
the source by -Xp. This is expressed by K --r X , where
R(x) K ( z -4-
E X), (2-1.4)
or
X(z)=K(z), Zfi=z@-X'. (2-1.5)
Now,
= eiPXK,(p), R.(p) = @ - ' P X K a ( ~ ) , (2- 1.6)
which shows clearly that the relation between the complementaryr coordinate
and momentum source desriptioas is given by Fourier transformation:

The space-time coordinates in these exponential functions are referred to an

origin located in the space-like surface, but we shall not usually make this
explicit.
We consider next the behavior under hornagenctous Lorentl; tr;amformations.
The respans af the single particle states to infinitesimal transformations of this
nature is given by
+
= i ( p l ( G o *J Sve N),
&[p)= --ifsw J g + Gv N)lp),
where [Eq. (1-3.29)]

In exhibiting N we have set z0 = 0,sinee this is the origin of time, and used an
alternative form of the symmetrised product of r with The coordinate
SQ Sources Chap. 2

opersltor in the momentum dese~ptionis represenkd by

(2-1.10)
and therefore

with an analogous formula for (p0)"21p). H ~ v i n ganticip~tedthe square root

fmtsr, we transcribe this as

vvith a similar statement for K,@). The implied infinitclsinnd change of H,(%)
or
Ka(x) if3
6K(z) = 160 r X V +
6~ * (ra, + Z'V)]K(Z)
= 6%" (~,K(z)~ (2-1.13)
where
6z" = 6 d r x , . (2-1. f 4)
This result,
fZ(z) = K ( z t 8 4 , (2-1.15)
or
R ( z ) = K ( ~ ) ,~ " = z ~ - 8 z ~ , (2-1.16)
when conzbined with the displacement response, assrrrb that the soume functions
of qinless pafticfes, K ( z ) , behave .dsr scaler functions under the tramformalions
of the Larente poup.
An important corollary is that the ehoiee of K(z) as a real flanetion would
have a Lolrent~-invariandmeaning. That is in, sharp earrtrast with the non-
relativistic situation, where N -+ -mr and'p + p2/2n. Then, if we consider
on1y boosts for simplicity,

and

The implied form for finite transformations is

R,(r, t ) = exp[v . (- imr f tV)]Ke(r, t )
= exp(--imv +
r -- i&mv2t)~(r vt, t ) (2-1,19)

R.(F, 1) = exp[-i(mv. r - fmv2t)]Ke(r, t ) ,

F = r - vt,
Evidently, a red emission or absorgtion source \.~louldhave no Galilean invariant
rne8ning. Incidentally, ia carryiw out the evduation af (2-1.19) we have used
2-1 Spin O particles. Weak source 41

a simple example of the formula ( p = - i d / a g ) :

The precise relationship betiveen the ernissioai and absorptioxr abilities of n

source is disclosed by combiriing the arthogonality af the vacuum srtd single
particle states, prior to the intervention. of sources, with the completeness of the
various particle-specified states that refer to the final situation. Tltus,

in which the additional terms are negtigitale urrder weak source conditions.
Furthermore, it suffices to use source-free values for the factors of ( 0 + [ 1 , ) ~ and
(~-jl,~ = )( ~
I ~ # \ o - ) ~namely
*~

expressirrg tlre ixlvarianee of the vacuum state, and

apart from phase factors \\.hick serve only to ensure dllat, in the resulting
relation,
) ~-- ( l p l ~ - ) ~ * ,
( o + ! ~ P= (2-1.25)
both single particle states are referred to the same space-like surface. Tliis
connection bettveen ereation and arlnihifatioxr probability amplitudes earl also
be presented as
i(o+i = [i(lpl~-)K]*. (2-1 .xi)
Thus, with a permissible choice cf arbitrary phases, the source functioxls K,(s)
sad K,(%) are reciprocal complex conjugates. The simplest possibility, and thc
one with r~hiehr,ve begin, is

a real function, We. now unite the various detaiis and state our explicit defini-

@P) - 1
K , = ( ~ U , ) " ~ K ( ~ ) ,dw, = -
( 2 ~ 1 32p0
and

The experimenter's bbasic tool is a beam of particles. A very weak beam of

~pinfeljsparticles has the following causal reprrasentation. We begin .rvith tbe
vacuum slab. Then rt weak soume IK2(x), occupying a finice apace-time region,
goes into action. It most often does nothing, with the associated probability
4 Sources Chap, 2

-
amplitude ( 0 + 1 0 - ) ~ 2 1, and occasionally produces a single particle, as
characterized by ( l , l 0 - ) ~ 1 . After the emission source has ceased to operate, the
re~ultingvacuum or single particle state persish unehawed until we e n k r the
space-tim region of an absorptian source K l ( s ) . Its effect in detecting a single
particle is described by (0+j1,)~1, and we thus return to the vacuum state. The
eompleee pmeess is represented by

where vacuum state sulbsc~ghdegignating causal aecyuence are slws?ys relative

the ixrdieakd sources. The individual vacuum amplitudes h&vethe form

On, making explicit the single particle creation and annihilation probability
amplitudes, we get

where rt, refers to the weak 8ouree limitation. The functions K l ( z ) and K2(x>
are the disjoint parts of the tots1 source in this situation, which is given by

There should be nothing in the overall description to distinpisk one component

part of the source from anodher, aside from reference to the space--time region
that it occupies. This is space-time uniformity. It implies that ( o + ( o - ) ~
depends only upon K, and in the manner made explicit by the hilinear structure
in K l and K 2 . Accordingly, we write

The displacement invariant funetion b+(z - z'), as the kernel of a quadratic

form, can be chosen symmetrical with no loss in generality,

The two equivalenL ~ontributionsof the type &;K2 then aupply the ~trueture
of h+ for a caustzl arrangement:

[We recall that @' is the energy-momentum veetor of a particle, so that =

+(p2 + m2)1'2.] From these characteristics of h+(%- g'), we deduce that
2-1 Spin O particla. Week sour~s 43

The explicit constructions of A+(z - s" may appear to refer only to c~lusal
or time-like relations between the points z aad d.But in fact the_ygive meaning
ta this funetion. everywhere, The only possible dificulty would be that, when z
rand sf are in spsee-like relation, where causality hits no invariant meaning,
different values might be obtained depcfing Upon the choice of coordinate
system. This does not happen. Since dw, and eke"'-"' are invariant structuresp
there is no harm in choosing a coordinate system for which z0 = so', and

is indepndent of the ambimous sign, for the integsal depend8 only upon
(X - X')%= (z - z ' ) ~ .As a result, there is no longer any indication in (2-1.35)
of the initial causd arrangement of sources, and that stmcturr? is applicable to
an arbitrary disposition of ~ourcm. This space-time extrapolatim must meet a
mvem k~t, however, We are nomr able to compute the probability that, despite
the intemention of the sources, the vacuum state persists, It is

(dcc)(dz")K(x) h ( I/i)b+(z - z")K(z'), (2-1.40)

where

is vdid e-rrerywltere,and the reference to the real part is uranecesssry since it is

implicit in the symmetry of the quadra;tic form, But probability considerations
also demand that

The challenge is successful1y met, for

There is one conceivable modification of b+(z - X') that would appear to

retain the necessary physicd charaeterislics, It is the addition to &+(X - z')
of a red funetion, which difiers horn zero only when (z - i~ space-like
intenrd. That would contribute nei$her to the causal exehange of partiGIes
between saurces nor the computation of the vacuum persistence probability.
The hypothesis af space-time uniformity, forbEdding the existence oT speeigl
relatianshi~pebtween: source-s, excludes that possibaity, In .(ih.b ~ o n k x t OM
,
can give the- ukformity hypothe~isa more precise, if rather ab~traet,form by
considering the four-dimensional Euclidean space that is a t i t a d d to the
Minkowski space through the ttomplex transformation
 Sourcsa Chap. 2

There is no andowe in Euelide~nspace to the Minkowski distinction between

lime-like and space-like inkmal~. AecordingIy, special space-time stmctures
would l x rejected if one insiskd that the invarianf vaeuum amplitude Chat
dmeribea a compfek physied process continue to be meaningful and invaxciant
on, mapping the Minkowski spaee onto the EucXidean space, This is the Euelidesn
postulate.
We recagniz-e that the Euclike~npa~tulateis a n8tural one by noting Ghat
6 4 % - z') has the require$ properties; there is an assrociatd Euelidean in-
variant function, AE(s - 2') which. exist6 &most everywhere (z # &). X t i s
obtained from the i n k g a l reprewntation

by the 8ubsdilut;ion
ilzD - -4 lz4 - z:I,
\vhieh requires that the ordering of the real numbers zO,zO'is mapped into the
same ordering of the real numbers Q, X:, We remove a factor of i in &fining

An explicitly Euclidean, invariant form appears on using the inkgraf relation

in which Ihe notation, ignoring any distinction betwwn contravariant and

eovariant eomponenls, ennphasi~esthe Ertctidean; stwcture. With Lhs recopi-
tion that AB($ -- r*)is a Euelidean invariant function, degenden%only upon

can r e t m to (2-1.48) and cltome the Euelidean coordinab syrstem ta get

the real positive expression

whieh i s one among a variety of single-parameter intepal representations. This

2-1 Spin O particles. Weak eourca 46

one immediately supplies the two limiting forms

Mote aho the simple inequality

Even better, since it reproduces the eorrecf limiting forms, is the inequality

One can; connect the Minkowski and Euelidean. descriptians by equating the
source strengths associated with corresponding voluple elements

while maintaining the reality of the souree function. This gives

and the right-hand side is a real number, which, is less than unity.
The physical vacuum amplitude can also be reeo.verc3.d from the Euelidean
version, by the earnpitex substitutions
2 4 -r iz0, p* -+ -ipop (2- X. 58)
provided they are understood to mean the limit of complex rotations M the
angle approaches lr/2 from smaller values,

z', p4 -r exp

Sueh cautiaxr is necessay since the resulting M i n k o ~ ~ sstructure@

k have
singularities: (z - 2'j2 = Oj the light cone singularity in coordinste space;
+
p2 m2 = 0, the partiole mass shell singularity in momentum space. We find
tfiat
(p,)2 + m' -+ p2 + m 2 - P:(I + 2 i e ) = ppr f m2 - ie,
(2-1 *W)

in which, despite various scale changes, E retains its meaning as a, prameter thaf
46 Sources Chap, 2

appromhes Eero through positive vafue@i. The resulting four-dimensional

mpmntation of A+(x - is

(2-1 .fix)
where

intmduces the Cauchy principal value for integrals. The contour integral
evaluatbn

reproduces Eq. (2-1.45). A limiting form in coordinebk spa= is

Asymflotic h s appmpriate to large 8eparwLion will be statecE for space-like

intervals, [(z -- z') = E > 0,and for time-like intervals, [-- (z - z ' ) ~ ] " ~
=
l' > 0,although they are eonneeted by the substitutions Ii! ++ il':

We turn next to the mom general situation in which Kfx) is a complex

function. Nour the sources eRwtive in ernimion and sborption art? reoiproeally
e ~ l e conjugate
x funelions, If we did no mom than introduce that featwe into
tht? previous dictussion, the single-particle term in the constmctim of the
eomplek vacuum amplitude would bmme

But this is dearly incomplete, far the implied soume stnrc%ure,Xinear in

2-1 Spin O particles. Weak source 47

dso requires the eontributictn of the causal term

referring to the ernis~ionand subsequent absorption of another kind of particle.

What is the mass of this particle? If the two masses were unequal, the stmcture
of the new h+(%- z') function in the vacuum amplitude

would still be given by (2-1.37) and (2-X.38), but differcent masses would appear
in the two causal forms, Then we could no longer conclude that B+(z - 2')
had a unique edrapolation into spaee-like regions. I t is the principle of space-
time uniformity that demands equal masses for the two kinds of particle, which
are identified as particle snd antipartick. The Euclide&n postulate produet38
the same conclusion through. the absence of an invariant distinc-t;ion btween
24 - Z: > O and 2 4 - 2; < Q, which permits only one mass pammeter to
appear.
In view of these remarh the definitions that relah sources to single-particle
production and annihilation prObabi1iw amplitudes mu^& be extended ta

where & distinwish parCicle and ranfipadicle, and

Note csrefulf y the distinction between

Accordingly, we have

Thus, the explicit appearance of the p or --p Fourier transform, representing

the energy-momenhm balance, distinpi~hesenniasion or srbsorptcion, respec-
tivety, while K and K* identify particle and antipadicle in emimion, but csn-
versely in absorption. The function of K is to ereate particles and annihilate
antiparticles, while K* creates antiparticles and annihilates particles.
In analogy with the way that sources act to increase or decrease the amount
of energy in the system, we can conceive of K and K* aeting to increase and
decrease, respectively, the quantity of s. property which must assume opphsite
values for particle and antiparticle. This is a familiar ohamebri~tieof eleotric
48 Sources Chap, 2

charge, and we rwognize that some chargelike p m p e ~ ydways dbtinguighw

pahicle and antipadiclt?. The formal countemad of them remarks &ms from
the invariane of the vetcuum transformation funetion under phwe transforma-
tions of the c o q l e x Bources,

If we examine the response of the probability amplitudes (1,*10_)' to the=

p h s e transfomstions, combined with a rigid displacement of the source, by
X@,we get
( ~ ~ * p--,-ef) i@eipX(lp+lO-)X,
~ (2-1 '77)
whieh makes explicit the mehanieal and "charge' attributes of the single-
parLiele ststes.
An dterrtative presentation is obtained by replacing the complex souree
K ( z ) with two real murees, a c c d i n g fa
K ( z ) = 2-1'2[K,i~(z)
- iK,,,(z)I, +
K t ( z ) = ~ - " ' [ K , ~ , ( z ) iKll)(z)b
(2- 1.75)
This @ves

We have now exhi*bit& twa indepadent murces, with their wsoeiated particles,
But the fact that these particles have the same mass (and spin) implies that the
dwompo~itioncan be done in an infinib variety af ways, camesponding to the
p h w transformation8 of eomplex sources, which naw appear as two-dimensional
Euelidean rotations :

The Xratkr can dso be wfithn in matrix notation m

and

i8identified as the e h ~ r g matrix.

e Nob that it is imaginary and antir~ymmetrieal.
Its eigenvalul~tsare k l, and the complex sourees K(%),K*(%) are the colme-
 Spin O partirzles, Weak source 19

sgonding eigenvectors. The real sources

do not produce single-particle states of definite charge. They refer to the eom-
plementary property of charge symmetry-the states turn into themselves or
their negatives, respectiveIy, when positive and negative eharges are inter-
changed. A matrix presentation of this transformation is
K(s) -+ rqK(z1,
where the real matrix

has the property

TqQ :T"": -Qrq.
The symbol C is often used for this charge refleetion operation,
When two-component matrix notation is used, the vacuum amplitude has
the same formal expression as tvith a single real source,

This remains true of its Euclidean counterpart:

which ettn also be written in terms of complex sources,

Euelidean. transformations decompose into two conneeted pieces, distin-

wished as proper and improper trsnsforrn&ions. In, contrast, the full Lorent~
group contains four connected pieces, owing to the discontinuous causal dislinc-
tion between z0 > O and z0 < O. The wider invariance introduced by the
Euclidmn postulate thus enables one to perform some discontinuous LarenL~,
transfornnations through the intermediary of continuous Euelidean trans-
formations, The mosL important example of that is

a proper transformation, which is rz time refleeticm transformation in Minkawski

space, The formal invariance of the vacuum amplitude under the transformation

is an immediate consequence of the symmetry

fSO Sources Chap. 2

but it is the Euclidean postuiaite that suppliea the general btzsis for this in-
v a ~ a n c e , The reflmtion of the time coordinate inverts the causal order of
sourem and inbrchangrzs creation and annihilation. This is evident from the
momentum form of the source transformation,
K(P) + K(-P)?
and thus
K,,*K:-, K,--K:+
or

According to its csnstmction W a time refleetion (I')and a space reflection

(F-arity), which also hm the egect of inhrchanging parCicle and %ntipa;rticlea
charge reflwtion (C)-this process is often known rts the TCP opration, but
one is Hkely to encounter any other permutation of the three constituents.
Perhaps it should be called the Shell g a m .

2-2 SPtM O PARTtCLES. $TRQNG SOURCE

The experimenter's beam cont2tin~many particles at a @ven time, which exisf
under conditions of effective noninteraction since they are widely s p a c d on the
sede set by nniera~opicinteraction distances. A beam of dectricaIEy charged
W i e l e s is sn exception to this, in grineiple, but in practice the disturbance by
long-range inkractions ean be made sufficiently smdf by controlling the barn.
densi$y. We give a fheoretieaX Lranscripfion of this situation by exploiting the
dimetiondity that sroureea possess as an aspect of the complemenla~tybetwen
the K($)and K ( p ) repremntation~. A Source that i8 spakially diEused and
suitably phmed (to use antenna language) ean p r d u w a highly directional
h a m , s f i a ~ l ylimiting the possible lacations of a detection Bouree if eAt"f:ctive
caupling i s to be achieved. We visualige an arbitrav number of such paim of
directional weak emiisaion and absorption sotrmes, operating side by side with
negligibk cross coupling. If roughly the slsbmrt clbusal amangement is us& far all
the pairs3 of gourees, ~ \ n ' ( 3have produced a situation in which, a t mme time inter-
mediate htween the emiaion slxld absorption regons, an arbitrary number of
padiclczs can exist in eireumgtanees of noni~terae%ion,owing to the 8p%%ial
wpslr&tionsamong them.
e r sources and de~ignateby &(g), a -- 1, 2,
We first ~ o n ~ i d real . .. , the
individual weak saurce~that correspond one self-contained emiwion aad
absorption process. The physi~alinclependence of &ese various aets, which
has been achieved through our control of the sources, is expressed by imply
mutliplying the individual probability amplitudes to produce the v w u m
amplitude for the complete anangement :
(d.)(dg')K,(z)d+(z - xF)Ka(rF)
2-2 Spin O particz-lss. Strong murm @l

The sourees Ka(sif are di~jointp&& of the tot81 8ouree

The principle of space-time uniformity requires that no specific distinctions

among the campomds of K ( s ) be admitM. I n short, the vacuum amplitude
must depend only upon R($). This b r d i s e d by imrpording the property

which asseds fhe &bf~enceof coupling between diEerea6 singk particle exchange
re@ons. Then, since the individual sources are weak,

The slam form ztpplies to two-component real sourees, and for empXex source8
if bwomea

We accept t h e wponential vaeuum awliLude ~&rueturcss as dmriptive of

any arrangement of sources, with arbitrary gtrengLh, subject only to the festric-
tion thaf the psrficle~have na efiwtive interrtolion. To tat the consbte~cyof
this sssedion we corrsrider a- gimple causal arrangement, e x p r w d by

in which. \cte maintain. $he convention that Kl refer8 60 p h y ~ i ~ d that occur

&et8
a f b r the cornpletlon of those repwanted by Kz. For the situafion of red
sources WIEShave

where, tzceording to the causal disposition of the ~OUPGW,

The ems& arrangement also ennsbles us do analyz;e the eomplete proem into an
initial multiparticle emission act, represented by the probability amplitude
((E)l 0 _ ) ~ 2 , and a subsequent absorption proeess, described by (O+l { n ) ) " ~ ~
where {n) indicates the collection of physical sttributes that distinguish the
various n-partiele states. The resulting causal snalysis of the vacuum amplitude
is
( ~ , l o - > ~= C (O+ W n))Kl({E)IQ-)Kz* (2-2.10)
In 1

To display, in this form, the explicit structure

(o+Io-)~ = (o+[o-)~'exp (O+~O_)~., (2-2.1 1)

we have only to introduce the mpansion of the exponential funetion

This supplies the required identifications

where the multiparticle label is realized by the collection of integers {B,). The
avidenf interpretation of n, is a particle o e ~ ~ p a f inumber
o~l a~aciftt;edwith the
indicated physiesl properties. This is confirmed by the response of the multi-
+
particle states to the source translation K(,) + K(% X), nvhich gives
Io-)~
((3 eiPX((n)Io-)~, (Q+/ (n))K + (@+l{R) (2-2.14)
The btal energy-momentum thus oblained,

displays the additive contributions of the particles present in the state under
consideration.
The probability amplitudes must meet the following total probability or
complekness test :

and indeed
2-2 Spin O particles. Strong source 63

No& how the vacuum amplitude has been used in two distinct ways. Through
the consideration of a causal arrangement, relative multipartick annplitudcs are
obtained :

and the assumed complekness of the multiparticle states leads to

Then the vacuum amplitude is applied directly 8s a probability amplitude, with

consistent results.
The extension to a pair of real sources, or the equivalent complex source, is
immediate, The summation over momenta in (2-2.12) is norv sutpplemenbd by
zz summatian over the two kinds of particles and the results are analogous, as in

where q == &l is the charge label that distinguishes partiele and antiparticle.
A combined source trainslation snd phase transformation changes these states as
follows :
( {n)1 ~ eiQveiPx( $4 Io-)~?
0 ~ ) --+ (O+I jnj )K + (O+I {R)) R e - i Q p e - i P X
(2-2.21)
where
Q = C
PQ
%,,g, = C
PQ
%p,$ (2-2.22)

exhibit; the h t a l charge and energy-momentum attributes of the mulLiparticle

s t a k labeled (R).
The momentum labcling of individual particle states is naf the only pas-
sibility. A spherical or angular momentum specification is introduced by the
transformation
K, = C
(dn)'I2 Y~,(P)KP@Z~P
Im
where

and jr, Vz, are standard symbols for spherical Bessel funetions and spherical
harmonies. The discrete angular momentum quantum numbers
 cation *
G h the infiGhiaaaf ~ I i m
b d gXw da. Thus

md one has anly to ohan@ the labis in. (2-2.13) ia obdsin tfie souroe mpm
wnbtion of tbe ntsw mdtipartielie &W. The p n e d i s a t i o ~fr, wmplex
is &km immdab. A m d n g to the uthd and@depnde~liwsf
which is em(a"w), the sowee robtiorr indie~Mby

&pIiays the bbI mam&k quangurn numbr of the multipartide

The TCP rttfation btwwn e ~ s and

b a h v t i a n now appam tw

wEch 8 b M h r F-1 mwem; o t h e k w 8 refemnce %a@hawmfietion is

dd&*
There is an aial de~c~ption,
which is abbined fmm

whae XJ. iS the projwW cmrdinak w b r in the plane pt3vn&calar b the:

and p indim* appmp~atemimufhal angle8 abut thw &g, The
TCP opmtian &km the fom

The mehawe of a m y e&nnaturdly b

aXm. Dimt power wriw mpamion @ves
2-2 spin O particlea, Strong murca Bg

(d.)(dd)K (x;>&+(z- z')KZ(g')

where
~em(sfi+(zi- ~2)= C
n! perm.
A+(zl - g * * A+(% -- )
.
:
g (2-2.37)
defines the pemsnent, a determinant without minus signs. Clearly displayed
here are the e h t i v e sources for n-particle emkion and absorption, together
with the funofion thaf represent8 the noninbrac$ing propagation of the n
peiebs, The fatbr ia syrnmetfigd among tfirt spwetime m r d i n a b ~of the
particles which, together with the unrestricted occupation numbers a, = 0,1,
2, . . . , proclaim8 that we are dwribing pa~iclestbspt Bo~eEinstein
stathties.
The f a ~ l i acharaokristics
r of this statisdies are dso appwent in the answer
to $he follawiq qumtion. Whad are the vduw of the general prob&bility
This is to ask how the effeetivenegs of a Bource in
emitting or absorbing particles is influenced by the prior presence of particles.
We cornider the f o l l a ~ g gc ~ w %~i6uation-
l A- strong source Kg(%)aeLs first to
creab pa&icles, vvhich are subaeqwntly idueneed by the probe souree Ko(z>,
afkr which the particles are annihilated by the detmkian source ,Kl(z) :

The causal situation is &henrzlbr~ativelyconveyed by

fmm which the d e ~ ~ probability

ed amplihdes, refer*g to the probts wwee
KaI can b ofitfaind. We fimt eo~sidera weak proh &ad accordingly relain
o d y linear b r m in KO, Then, aince
MS Sourccas Chap. 2

which refers very immediately to the rnonornial structure of t h w probability

amplitudes, we get

which generalize the initial definitions (2-1.28) while retaining the weak ssuree
limitation. I n particular, the probability for the emission of another partide,

show8 the additional stimulated emission, that is characteristic of B,E. statisf ics,
As a preliminary to picking out the pneraX transition amplitudes, we
construct the probability amplitude ( { R ) +l {n)- ) K , which has the same initial
a d final configuration and, in. that sense, is a generdi~afiontof $he vacuum
amplitude. This object is extracted by retaining only equal powers of K:, and
K z , in the expansion

We are going to introduce a useful simplification here by recognizing that for

sufficiently small do,,higlrer p w e m in this series are neglgble. XL is only the
dependence on the probe source that is at issue:

We also remark that

which applies to each momentum cell independently. For the process of inter@@%,
then,
c
P
+
(iK:,iKop iKgPiKs,) -+ +
[l iKgpn,iKop]
P

and we conclude that

({n)+i{n)-)K= exp
where

The last term is rso bvritten in order to maintain the symmetry in x and 3'. The
 Spin O particles, Strong source fiT

explicit causal forms are

Note that the probability amplitude ({R)+l {nJ-)K reduces to unity for K = 0,
which means that the initial and final multiparticle staks are r e k m d to the
same time or spaelike surface, as is appropriate to a reasonabIy loealiged probe,
To find the probability amplitudes in which initial and final ~ t a h are
s no8
the same, we do not return to the general eonstrmeLion given in (2-2.40), but
use (in)+l In) directly, in the manner of the vacuum amplitude. Thus,
coxzsider the ectusal sou=@arrangement
K(z1 = (4 4- K 2 ( ~ 1 ,
tvhich implies Chat

The coupling between the component sources is now given by

The explicit terms indicated for a given momentum describe the mverd processes
in which, respectively, no ehange in particle number oecurs, an additiond
particle is emitted, an incident particle is absorbed. Higher p w e m , containixlg
more complicated procemes involving several particles, are relatively negligible
for suficiently small do,; the probability for emitting two particles into the
momentum range do,, for example, is - ( d ~ , ) ~ But . we must not let this
apparently innoeent sirnplificaLion pms without comment;. The infinikgimd
chzbracter of h, will be vitiated if there is an inordinate sensikivity to p produced
by the appearance of very large coordinate intervals (e*'X). T o put it more
physically, we recall that we are dealing with a beam of parlieleg interwting
with a probe source. What we have done is eorreet if the probe is placed well in
the interior of the beam where there is no ~ignificantposition dependence. It
trill fail if the probe is outside or near the boundaries of the beam. This is a
68 Sources Chap. 2

reminder that underlying any momentum description is s n appreciation of the

spacetirne causal situation.
Having understood i b limitations, we now use (2-2.52) ta identify the
individual probability amplitudes for processes in which, independently for any
number of momentum cells, a single particle has been added or removed. The
result is indicated by

where the products refer to the various particles that are (e)mitted or (a)bsorbed.
+
The two statements are equivalent, if it is &mitt& that ( { n 11, + I {n f 1,) -)K
and ( { n )+l{n)J K differ negligibly, owing to the infinitesimal nature of do,.
Now we face the consistency test associated with the alternative uses of the
probabiiity amplitude. From the completeness of the fins1 or initial multi-
particle sfates we deduce
~ ~ I ~ W + I {= n
~ I - [ )l +~ (np
I ~ + ~ ) I K+~ ,I I~K ~ I ~ I
whereas, by direct calculation,
I(In)+lh)JKI= exp [- 1(dx)( d d ) K ( z )Re ( l / i ) A ,
ln +(X -z3~(2')]
(2-2.55)
and
Re ( 1 /i)A l, +(X - X') +
= ~ e / d o , ( 2 r c ~~ ) e ' ~ " - ~ * ' . (2-2.56)

The test hm been passed successfully.

The extension to complex sources and charged particles is ~traightfon~ard,
and generally the introduction of the charge label q, supplementing the mo-
mentum index p, suffices to produce the required result. We o d y remark on the
following detail. In constructing the probsbility amplitude ( { n )+l { n ) we
are led, as in (2-2.46), to the factor
exP [gGPQ%~K~P,](2-2.57)
and the result
((n) +l{n)
[I ,,+(X
= exp i (dz)( d z ' ) K * ( z ) ~ , - &)K(x!)] (2-2.58)

Now, however, the propagation function Al.)+($ - 23 has the following

meaning :
2-2 Spin O particles, Strong scccrcs fig

which is no longer neeemarily symmetried in r and x'. There is still a TCP

symmetry, in. ~vhichzfi -+-9 is c~rnbiil~ed
~vithcharge reflection,

The explicit causal stmeture of this function is given by

It should be mentioned that the propaga;tion funetion is symmet~ealin z and 2'

if the incident beam is neutral at every momentum, a,+ = B,-. Then, and
only then, can one introduce the, real sources Ktl,, KZ,, and associate two
independent particle types ~viththem,
We have seen causality and space-time uniformity ~vorllcingas ereative
principles. The physical requirement of completeness, or unitsrity, has then
been verified; it is not an. independent principle,
We shall nou7examine this relationship in more detail. But first we return
to the vaeuum persisknce amplitude far real Baurces and consider it8 e~rnpfex
conjugate :
o_)"* = (o-Io+)~ = exp (dx>(~Z?K(Z>A-(Z -xf)~(zP)

To ~ v a euniform presentation of the tvrpopropagaLioni funcfcions A*, \VC?. define,

everywhere, the positive and negative frequency functions A' as

which are connected by

TXle tiro propagation funGtions Lhien. appear as

We note the everytvhere-valid ~ l a t i o n

The momentum inbgral defived from (2-5.45) 'by complex conjugation,

f e d to the same Euclideam. function EMbefore,

by means of the gubg%ituthn

Tben we have

whieh is the same Euelidesn form that is obtained from (o+~o-)~,as the strong
wurce genera;liga$ion of (2-1.57). The Euclidean version of the veteuum ampli-
tude is a red number lying in the i n t e n d htween 0 and 1. The vacuum.
amplitude (o-Io+)~ i s regained through the substitution

Pot

which, incidentally, supplies the invariant rqresentation

Another canneckion between the reeiproealfy complex contjugak vacuum

amplitudes come8 from the t?xi~%ence
of the common Euclidean tranmription.
Pmceeding &rough the Euelidesn h m as an intermedimy, we have

(I. + i15)pa -4 -(f - if.)^^,

snd, in respon~e,
A+(z - 2') -+---A-@ --- z'), (0+
2-2 Spin O particles. Strong souroe 61

To verify this directly, we note that

with the usual djuslable scale for r. One must also remember that the ordering
of variables is retained throughout the transhmation. Limits of integration do
not change, therefore, and
(2-2.77)
while
(2-2.78)
which confirm in alternative ways the stated transformations.
We shall now use the causal structure of the theory to give a complete
derivation of the unitstrity praperty. This is done within a very limited physical
context, of course, but it is clearly a general procedure. For our present purposes
we replace the causal labling K l , K z by K[-,, K,+,. (While this may seem .to
be still another use of the overburdened signs, it will turn out to be consistent
with the notation.) Let the time T be located between the regions defined
by the two component sources. Introduce a new time coordinate for z0 > T by
reflection at T,

sncl then transform this time interval in, the manner just diacussed;

The immediate effect is to replace z', which is earlier than T to the same
extent, %hatthe original time exceeded T . The transform& Kt-) souree also
occupies a refiected position, earlier than T. Before this ontjration is p r f o r m d ,
the vacuum amplitude has the following composition:

where the appearance of the last term indicates the causal arrangement. When
the transformation is carried out, the quadratic K(+, term remains unawarc! of
what happens letttw, the quadratic K(-, term responds in the known manner
[h+ -+--A-] without reference to the other terms, and the last term changes
only by a minus sign arising from

the influence of the em"' factor, which is not compensated by the refleetion that
is al80 being used. f n the latter the inkpation limit8 are reversed to maintain
82 Sources Chap. 2

a positive measure. The result is

~ exp [-+i/(dz)(dz1)K(-)(Z)A-(X - d ) K ( - ) ( X ' )

( O + ~ O - )+

+ / (dz)(dz')( - ~ ) K ( - ) ( z ) A ( + ) (-z z 1 ) i ~ ( + ) ( x ' ) ] (2-2.83)

which does not depend upon T, that being any time after both sources have
ceased operating. The physical meaning of this combination follows from

exp [/(dz)(dzl)(-~)K,-)(Z)A(+)(X - zt)iK(+)(z')]

[F
= e . ~ ( i ~ ( - ) p )(W+)$]
*

for, on using the fact that

({n>Io-)~' = (0-1 (2-2.85)
we get
(0-10+)'(-) exp [/(dz)(dz')( - ~ ) K ( - ~ ( Z ) A ( + ) ( Z - z1)iK(+)(x')]( o + I o - ) ~ C+)

= c (0-1 { n ) ) K ( - ) ( { nIOJK(+)
(nl
)
( O - I O - ) ~ ( - ) ~ ~ ( +(2-2.86)
).
As the notation indicates, the picture has become that of a system evolving in
time from the initial vacuum state under the influence of the source K(+)(%)
and then traced back to the initial state in the presence of the source K(-)(%).
It is not the physical system that goes back in time, of course. What is reversed
is the causal order of the states that are being compared. If the two sources are
identical, we must regain the initial state; that is,
K ( - ) ( z ) = K ( + ) ( z )= K ( x ) (2-2.87)
implies
( O - I O - ) ~=
~~ (O-l{n))K({n)10JK = 1, (2-2.88)
inI

a statement of completeness, or unitarity. According to the exponential struc-

ture (2-2.83))this is true if

where the last terms appear in that form to produce the necessary symmetry in
Xand X'. We recognize the identity (2-2.67).
2-2 Spin b particles. Strong sourea 63.

The full @tatemeatof u n i l a ~ t yemergee on using the sources Kt*) to

generate arbitrary multipsrticle states. We write

where K z and .Kz#act prior to the source K, snd intfoduce the earresponding
cauwl analysm :
({a)10JK'+' = C ( {n)+l

What we mu8t verify is that all refemnce ts the source K(%) disappeam from
(2-2.862, l e ~ ~ on1
n gy

for that is the u ~ b r i t ywertion about the eEeet 01 the source K:

Beyond the condition we have almady used, Eq. (2-2.591, w h d is required is

But, under the given causal circumstances, L\- --+ --id(-) and A+ -+i~'+',
which completes the verification.
The probability amplitude (o-Io-)~[-I*~(+~ is a1w useful for the direct
eornputation of various expectation vaium. L&

for emmpb, which mplztees the unit evaluation for X = O by

This is the expectation value of eiPX for the states produced from the vaeuum
by the action af the sourn K, Sin= only the relative dhplawmenl of the tvvo
muram is sipif cant, we have

&,(eiPx -- I ) ~ K ( ~ )=
/ ' exp [T (e' -- 1)
On considering infirritmiml digplacements, we learn that

or, with an obvious identification,

The total number af psrtieles i~ iindiercled by

N = E n,. (2-2. la0)
P

Thus, the werage total number of particles creakd and the vasuum persisLence
prab~bilityare &implyreletted, according to (2-2.17) :

of Auetuations is fwilitated by writing (the vafious indiees arc;

The discu~~ion
omitted)

The simplest example is

which we can also inkrpret as

One consequence af the latkr is

Statements about tbe total nunrbttr of particles aw derived directly by

eansidering %hesourea
2-2 Spin O particles, Strong murcs 86

Aceordin&to the hmcture of the rdative prob~bilityamplitudes, we have

(2-2,107)
One digerentktion with re~pectto X, at X = 1, @ves

The coefficient of kX in the summation (2-2.107) is the probability of emitting

N particles, without further identification. Comparison of the power series
expamion supgliw iL ax
p(N,@lK = - (#jN @--(NI* (2-2. f 09)
N!
The fluctuation, formula (2-2.105) is a f a ~ l i a rchsbrackrigt-tic of thig Paissont
&tribution, All such p m p e ~ i e sare deriyd from

by differetntliation with respect to X:

The generalization of this discussion to the amplitude ({R)-1 { R ) -)Kt-l*Rt+}

only requires introducing the function A +(z - 2') and its pabrtners :

The causal relatioxls among thege functions are the same m in the v ~ e u u m
situation although b\$)(z - s'), for example, is no longer an exclusively positive
.frequency function, I n deriving expectation values we mu& note the$ the
amplitude ((B') +l {n)-)K responds to the translation K(%)-r K(z X) with +
the factsr
@x~ti(P(%') - )XI, (2-2- 1113)
since?both initial and final s l a k s are now mlevant, Some results are
(2-2.1 14)
and
(fib%;.) - (nL)(nk.) = bPPf((nk - ~ ~ ) ) ( 2+
1 11).
~ (2-2.115)
IW3 Souross Chap, 2

The treatman$ of oomplex rsourees and charged par2;iclw k quife a ~ a l ~ g a w .

The vacuum amplitude deaaribing the time cycle ir~
~ ~exp
( o _ I o - ) ~ ~ - ' ~= +' )K:-,
(dz)(hf (~)a-(z- z ' ) ~ , - ) ( z ~ )
(dz')K?+,(z)d+ -- zf)K(+)(z'f
( d & ) ~ ; -(z)A'+'
) (z- z f ) K t+)(a;')
( d z ' ) ~ ~ - , ( z ) ~ ~-
+ 'z')K;+,(z')
(z

wGch r d u e e ~.to unily for Kt,, (a;) =. ;K(+,fz):), By choa~iingthe, somew

Kf-,(z) = K(z), K{+,(z) = h e i @ ~ (fz X), (2-2.117)

with X real, we obtain

We ea0 also intmdum the total numbr of pa~itivelyand negatively chargd

padiales,
N.+== -C- Q), N- - Q), X(2-2.119)
and rmxpreBs the expseation value formula as

Accordingly,

P
~,+I" ,M-): =a

P
IK@-I~, (2-2.121)
whib individual prcibrabilitiw are given by

A simplified formula, designed to answer questions about electric oharge only, is

(eg4@); = exp[(ei@ - 1)(N+) + (e-ip - 1)(N-)j, (2-2.123)
from which wcz deivci:fhe.individua1 prababilitiers:
2-3 Spin 1 particles. The photon 67

on. using a familiar Berne1 function. expansion. The introduction of the gropag*
tion function (2-2.59), with its atkndant stmcturers, generalbw (2-2, X 16) do
the probability amplitude ({B)-1 {n)- ) K ( - ~ ~ K ~ + ~ .
2-49 SPIN 1 PARTICLES. THE FNOTQN
Before developing tbe general source regresent&tion,for padic1es of arbitrav
spin, we shall give an dementary discussion of some: gpecial examples whiah are
of great physical importance, The exponentid form that has b n mhbiished
for the vacuum amplitude, within. Ghe eontext of spinless pa&iclm, embodie~
the physicaf pas~ibiIityof producing any number of independent w t s af sin&
particle emimion and abwrption, These gpace-time properties &re independent
of the spcific spin of the particle. The latkr ean only influence the mare detailed
stmcture of the source. It is clear that, if spin O p~r$ialf3~~ are described by a;
scalar-source, sources transforming as vecbrs and tctnsar~of va~orrsranks must;
refer to pttrticles of unit and higher spin. A vector source, designated m J P ( % ) ,
is the obvious candidate to de8cfibe unit spin particles. T h m m certain
obstacles, however. This source ha8 four component^, iin contratst wiLh %hethree
independent sources one should associate with f i e three spin pawibilities that;
are accessible to a nongero mass particle, Thics pre~umabliymeans that J'(z) is
a mixture of a unit spin source with a souree of spinless particles, corresponding
to the possibility of forming a scalar function by differentiation, a,Jg(z). And,
independently, W ob~ervethat should we do no more than replace the resll
scalar murw K(x) by the real vector s o w e P(&
(o+Io-)~ = exp fdz) (d~"J@(z)A+(z - z"rJ,(x")
we should eneaunder a serious p h y s i ~ ddificufty, for

is aat marantmd to be l= than. unity, since

can wsume either sign.

Both diEculties are overcome by the following invariant slructure, which is
appropriate ta a particle af mass m $ 0:

In wduaLing the vacuum pemhknee probability, now encounter

68 Sources Chap. 2

Since this is an invariant combination, it can be inspected conveniently in the

rest frame of the time-like vector v,where
f l rest frame: = 0, = m. (2-3.6)
The components of the symmetrical tensor that appears in (2-3.5) are then
given by
p=v=O:O,
+
~ P D (1/m2)ppp~: p = k, V = 0: 0,
p = k, V = I : bat.
(2-3.7)

The result is simply 151*,which is positive, and which contains three independent
source components, transforming among themselves under spatial rotation, as
is appropriate to unit spin.
We note that (l/m)@ is a unit time-like vector, which can be supplemented
by three orthogonal space-like vectors, 4x, obeying
pp4x = 0, 4?eppxr = 8xxn. (2-3.8)
They give a dyadic construction of the metric tensor,

The symmetry of g@' indicates that complex conjugation of the three e$, produces
a unitary transformation on the set. With the definition

the vacuum persistence probability appears as

We now consider a causal source arrangement,

which implies
Jp(z) = J';(2)+ JS (4,
(o+~o-)' = (o+~o-)'l exp [/doput (P) *(gpv+ m - 2 ~ p ~ v ) i J ; ( ~(o+Io-)
)] "

= (o+~o-)'l exp [Ci l : p ~ i r 2 p x ]

PA
(o+Io-)~~.

This standard structure identifies the multiparticle states

where n,x = 0, 1, 2, ... again indicates B.E. statistics. The consistency be-
tween the two uses of the vacuum amplitude is obvious.
2-3 Spin 1 particlas. The photon 69

One can choose the unit spacolike vectors 6~to be real. The orthogondity
requirement
0
P epk = P 0epx (2-3.15)
displays the role of p in providing a reference direction. If epx is perpendicular
to p, the time component e$, vanishes. Let epl be such a real unit vector,

Then

is another one, and the set is completed by

0
eps = (p"/m)(p/lpl), epa = Ipl/me (2-3.18)
We note, incidentally, that

Alternative, complex, choices are suggested by angular momentum con-

siderations. The response of the vector JP(z) to the homogeneous infinitesimal
Lorentz transformation
?EP = xp -+-
6&*x, (2-3.20)

For a three-dimensional rotation, this becomes

and, equivalently,

Now let us consider a rotation about the direction of the momentum,

60 = ~V(P//PI).
We realize a single-particle state of helicity X :
110 Sources Chap. 2

Zero helieity is aehievd with e parallel to p. Accodingly we relabel ego,

(2-3,28)
e ~ X = rh= l correspond to the complex combination8
The helicity ~ t a t with

which are so chown that

+l
--e& X 6 0 = i lfe:kt
(6o* S)& (2-3.30)
k1=-1
give8 dhe ~tandardunit spin m a t ~ xelements,
A clwifica%ionof murees and pa&iele ataks in, relation to btal anwlar
momentum ean be introdued. As a preliminary shp, we emulab the eero spin
procedure and define
X
( d ~ ~ ) " ~ ~ ' ( p )( d ~ ) " ' y t ~ ( ~ ) J > t m ,
ilm
(2-3.31)
where

Nothing more need be done for the time component J@,which is s %h

sianaf scalar fumtian. But the fhree components of J refer to $he uni%spin,
which mast be coupled appmpriately with the orbital a n d a r marneaturn to
produce total anmlar momen%umstates, This is raccomplisbed by the following
inkmduction of s veedor o&honormml system, mpIaeing the scdar spherical
h a m o ~ mt,
e

mid the reder is warn4 not 4x1 confum the letter m, u d in subseript~to dtcrtnoh
a rnaeetic quantum. numhr, with m, appearing e1cpXieit;lg in i f s mle ws particle
m m , The above ~tmetureis mch that
2-3 Spin 1 particles. The photon Tt

We also note the relation

(2-3.36)
On combining the various contributions, we get the required form:
dadp(p)' (g, + m-'p,p.)JY (p) =
pDimx
Jpjm hJp0jmh, (2-3.371

where h = 1, 2,s di~tinpishesthe three excitation8 with total snlgular mamea-

%urnquantum xlumbrs j, m,
These sources esn. be exhibited expjicidly. The vector orthor~omality
praperty enables us to evaluate

for example, and this ean be converted into

where nolv
L = X X (lJi)V,
(It is unfortunate that the cornbination of two well-established no.Latioaal
conventions produces things like ji.) Incidentally this type of aourct: van&he?3
for ji = 0. Similarly, we find that

which also vanishes for j = 0, and

It is seen that the sources designated as X = 1 , 2 depend only upon J(z), and
that in the form V X J(z), while far the Lhird @pe of source we have, effectively,

There is no dii-ficulty in implementing the same generalizations thak were

discussed for ~ e r ospin particles--charged particles, multiparticle initial and
"112 Sources Chap, 2

final stabs, eyelie time development-but the details are too s i ~ l a do

r merit
reptition. We turn in~teadto sn important spcial situation, thc? limit of sero
mam, M realigd by the phokon.
It & evident from Eq. (S3.4) that the sero mass8 limit does not mist unfm
aPJ@(z) vanishes. One ~ g he t t e

and identify K(%),in the limit m -4, m the gouree of a nassfess gero spin
particle. The latter would be oompletely independent of the residual photon
muree, however, and since m .= 0, s = O particles sre unknwn experimentally,
in %ny event, we only eansider photons in stating the source dessc~pfiorm:
(o+Io-)~= exp
(2-3.45)
W"(%)
=: 0.

We uge the symbol D+ Lo indic%tethe redrietion tr, gem mass,

The murce associakd ~ t a hpadicular psrt;ielc? is an abstraction of the
realistic procmw that ereate or annihilate %hepadicle. It retdns what d l w ~ h
mctchanjsms have in wrnrnon and ignorw the ~pecifieehari%cteristicsof indi"ViduEtl
mechmisms, Any generail rest~etionon m u r ~ e sthat is implied by speeid h&-
tures of the particle must be eammon ta all nnechebnisms and fhus sfaks a gemfa1
law of physics. I n the situation of the photon, we have dc?ldue&, from its zero
m a ~the~ ,nc~ceseityof a reslrietian on the v e c b ~ asourct?,
l It m u ~be
t divergene
less, which is the local statement af ai commafion. law. We? are in xla doubt;
about the identi%yof this consewed physical propedy. It is elee%~c charge,
The loss of one degrw of excitation far masslem psrticbs is evident in the
vsfious ways of labelins particle staks. Thug, aa nz Q in (2-3.191, under the
rmtriction
) 0,
P ~ J @ ( P= (2-allats)
we a e v e a t
Jp3 S 0, (2-3.47)
and the two remaining sourses Jg1,2~ f e to r the two transverse linear pol%Gzg-
tioas accessible to photons. With helicity labeliw we have, equivalently,

and J P k l reprwnt the two circular pola~aations,Turning to a n ~ f s monsen-

r
dam s h b , we have analagously, from (2-3-43),

Sincej = Q dws not appear in the twa other soum Lype~,t h k is the counkvarl
af the abmnce of rtero helicity.
We have arrived a t the ratrietion fo two pola~zationor helicify s$alw for
Lhe photon by a limiting procedure that began with massiw unit, spin padides.
Now let W obtaixz this result directly, by trsiag the photon source desc&ption
@venin (2-3.45). The consideration of a causal arrangement,
J""(3;") J!(z) -I- J$(r), (2-3.W)
impfies
(O+IO-) = (O+IO-) Jg exp (o+[o-)". (2-3.51)
The dysdie represenlalion for g,, given in Eq. (2-3.9) is not ap;propriak here
dnw p"" is now a null vector,
pg = 0. (2-3.52)
Let FP be obkained from p"" by mversing the rnotion of the * o h ,

+
Then pl" 'p and p" -- pp am,rmpectively, fa timelike and spmelike vector.
They are ~uppbmentedby two orthrtgonal uniL spae-like veebm @gx,

to give the d y d i ~
coxlatrwtion

I- (P' + p'l)~ (PP' F3-pp)_ (P' -- )'P (P' - p')

gm
+ C @'PX
X
@V*
P&

We now urn the photon gaurce reskriation,

which k the desired partide exchrzngs form. If is also impGcsd by (2-3.54) that
the two have aero time component and, as spatial vecton, are perpendicular
to p. This k s self-contained desefiption of the k m tramvem8 witatima that
art3 pemithd to photons.
It bm been recomized earlier that a
s concept that is invarirtnt under proper orthoehronous Lorentr transformations.
One &odd be able fo make mare evident thL wpwt of tbe photon helioity
@ t a b .A d we should like to under~bndwhy the M e i t y slates fasve appeared
psired afthough no overt reference fa e?gatiai refieetion h= been made, b t us
begin with the remark that the conservation eonditiop imposed on JP($) is
sadisfied identiaally if
J@(x)= dPMp"z), (%a.58)
where
MP"(x) = -MY"(a;).
We introduce the eoncepd of the dual to an anti~ymmetricsl

in which P"& is the totally antisymmetrical tensor that is n o d i a e d by

The opration of forming the dual hss the foXXa~ngrepegition proprfy,

The dual t e m r is used Go m i b

The 1aGbr property indieabs that eaeh of them objects hm only three inde
pendent components, as i l l ~ s t r ~ by
%d

T h m coxrrpnents %re complex numhm, of eoum, and

The deeompo~idion@ven in (2-3.63), and indicated by

ia an invarisnf one far as eontinuouta change8 of coardiniFtte aysbms are con-

memiond nsbtion, we have t k 3 expucit 0~)mtm~tian8

The eEmtiveness of dhese gources in mdiating a phofon of heficity X is measured

s~cor&ngto (2-3.m). The n w of makhfag the value! of X k % hf f. 8haw~

that the =t;l 1 a b h OD.the Bources da indeed mfer Lo %hemiqare heli~iti.esBf the
photons that are ennitM ar abwrbd by %he~e compnent rnww~,
2-3 Spin 7 panictrs, The photon 76

Why can one not modify this phobn dwriptim by odtting J',x(z>, a y ,
and themby prduee a theory with only positive hdicity phsbna~tFor tbe same
reamn that one cannot have a theory in which only p~itivelycharged p~&icltls
occur; it would violak %hepfinciple of hipwe-time uniformity, To d k u m fhb
point in more detail, consider the cantribution to %hev ~ u u r nanaplitude
ated with the emi~sionand submquent abhiavtian of a p 4 t i v e helicity photon

where the c a u d 1abds 1, 2 have been displaced for &reah1~" elafi*y. We are
jwgified in writing g,, &nee the e s q u i v h t palariakbtion vector eummation
rduces to the appropriafe positive helieity hms, in vi&ue of (2-3.71). The
compbte sortme coupling should b linear in

and in
J;s(~=
) J;l(z)t + 5;1(2)~ (2-3.73)

J'+312)' J$~(z):t- J'+s(z):.

E (2-8.74)
One might try Lo resist the iderenccs that there is another eaupling involving
(g')l and J'/ ,($)g by introducing the ~paoe-timeextrapalation of (2-3.72)
with an addil;ionsl factar,
1, z0 > z",
q(zO- zO')=
0, x0 < zO',
which is d e ~ i m dta elirninak muree arrangementhi whem dhe m d ralw of
J;, and df are revemd. This step function d ~ have. g an invsnisnt meaning
when x and z h r e in timelike or nuli relabion, but it is not invdan6
like intervals, and its introducfion would vialab the principle a f 8
r t gpreence af the &di%io
uniformity, We cannot avoid r e ~ ~ p i ~ i%he
coupling term
(2-3.X)

and the particles dehieribd here mwt also be of n;wo m- if a unique splleetinre
extrapoltbtion is fo be acbievd. T h w antipa&iclr?s am t;he negative helicity
photons,
J'+s(~)'= Jtl(z), (2-3.77)
and the additional krm ean be rewfi6bn as
(dz)(dar')JI", (z)T (2-3,78)

Furthermore, the analogow structures involving $vl,JLl and JeltJ h equal

zero since one ar fhe okher factor in the pda~a&fionvecfor gummatian will
v a ~ s h .The rmdt is fa recanstitub the real soume
76 Sources Chap. 2

appeafing in the coupling

which ia mogniaed as the @ a w lp 8 ~ i c bexchange krm in

AB a comllary of &is dicscumion we no& that the soure=

afifl give an equivsled dmcription of photon ernimioa and absorpfion. This

new mume is ~ p r m n M
by
*JR(Z)= a, *MC~(. (2-3.83)
Tbe nature of the tmndormation is afso indieaM by
*JP& = (d@p) 1/2(~/1~1)
JW, X &@
: * (2-3.M)
which makw expli~itthaG Lhe pola~ratianv ~ t o r shave been r o b b d thmutgh
the angle ~ 1 about
2 the photon direction of mofion. If fhe mation angle k p,
the tr&~sformatiorr.
hcomes
+
JP(%) 4 Jg(x) cos (a *Jfl(z)sin 9. (2-3.S)
The subsfieuLion of *P for JP &so has a Elimple eEect upon the an&ar momen-
tum labled BWTCW, J p a i d . W@fimt remark that
J(z) = V X M(%) + a. *M(z) -+V X M(z) - ip' *M(z), (2-3.86)
whew %hefatter ~ub~ti%ution indicaks the eBectivcr; value in the in@pab that;
compBe J p ~ i , h * Sifilarly, we have
(i[pa)vX J(z) = V X *M(%) ($/pa)(VV -v')M(s)+ -
4V +
*M(z) i p a ~ ( z ) , (2-3-87)
which uses the eflective value --vZ --t 'p = (P@)' and the pmperty L V = 0.
The ~ubatitutiianJfi 4 * P , which is equivalent to M -+ *M,*M -., -M,
interehangea these vectorial structures and tnrmforms the two souroes w o r d -
ing to
J p ~ j n r l -* JlOjma, Jpojm% --JpOjml* (2-3,s)
The m m genersl sub&itution (2-3.85) gives the rotatian

In tha general spacetime form of the vseuum amplitude, Eq. (2-3.451,

rsoumw need not ernif and 8 k r b photans and indeed m&ybe incapable of doh&
2-3 Spin 1 psrticlcts, The photon 7'7

so if they vary too slowly in time, I t i~ Che principb of space-time uniformity

which thus asserts the physical unity betwen collision meehanisns &at do
liberaLe enough energy to create a particle and those otherw2se analogous
mehenisms that h~tppenta have an insufficient energy supply. To illu~tratethe
new physical information that is obtained in this manner we consider photon
sourctts Lhat vary very slowly in time. The wrty is preipared for this limi$ by
writing

where, as a consequence of (2-1.451,

rOD

This struetum indieaks that $he scale of significant T variation ia set by Ix - X'[.
If the sources vary little, in the time intervals that are msociakd with the
distances characte~sticof the instantaneous digt~bution,one can i p o r e the r
dependence in J@(x,z0 & 47) and evaluate

&P)+(X - X', - 0 sin polx - X'\

7)"" d~
PO

This ~ v e the
s foXIowirrg form to the vacuum amplitude:
(0+j0-}~= eup (2-3.93)
where

One reeognism here the aeeumulakd phmf?;change of s state that has a time
vafiable energy, E(xa). When a stesdy-state regime is established, we are led to
asgociste with it the energy value

whieh is a s&aternentof the Coulomb and Ampbrim laws of charge and current
interactions* This shows haw the principle of space-time uniformity provides
the logical connection befvveen the properLies of photons and the chsmekristics
of quasi-stationaq charge distributions.
There is one suLtLleLy hem we should not overllook. One cannot produce a
complettlly arbitrary statie charge distribution. The local conwrvatioa condition
78 Sour- Chap. 2

aBJ@= O impXim comrvaCian of the htal ehsrge

pm~4d the sourn is wnlind fo some finite ~patialre@on. Being wro in the
i&tbl muurn s h b , the bbl ch~rgemmains Piera, We m y picture initially
mmpe p ~ i t i v and
e negative charge distribution8being separaM, maved
abut anb then m o m b h d . But them is another way of viming the
in4duction of a charge &t~butionintto an emp%yredon, It requirw rmomil;-
iaig m ly than is wusl tha$ a physied dme~ptionrefern only to the
fi~b re@on which h under the exp~menbr'smntrol. The iniCial
and final vacuurn ~tatesp&ain to s boundd three-dimemionaf re@an, W e
g- on, oubide the walls- We thus &pp~ci&%e that an arbitrary charge di*
Critouticrn can b produced by the t r a a ~ p dof charge acrom the boundary, into
the mdon of inter&, and that thil charge distribution can be dismanlled
uldirxrtabIy by withdradng id across the boundaq.

2-4 SPIN 2 PARTICLES, THE ORAVITON

Next ia complexity afkr scalar and v w b r mumas is the red ~ymmetriealtensor
80uree
Tp"(z)= TF@(g). (2-4.1)
X t has k n eommnentF3, But they include the 3 + 1; component vector saurce
and the sealsr wurce
P(%) gPyT@*(jc). (2-4.2)
When thew are removed, the residual multiplicity af five is the anticip~tLedone
for spin 2 pl.ticIerr of noneem mass, m. T o c&rvout this p r o p m we exploit
our @ x p ~ e n e*th
@ unit spin p&&ielesand wride direcffy the physically am88hbq
~$mc%urrt for the vscuum pc?rsi~bnwprobability, It is
do,F'v(p)*~r.(p)g.k(p) Pk(p)
where
pp.(p) = + (l/m?)p,p.t pYB&.(p) 0, 38 (2-4.4)
and
T@'(p) - 3yrpg@@(p)Twfp)
= TL*(p) (2-4.5)
0by8
B ~ ~ ( P ) T ' ' ( P ) Qv
S (2-4.Q
In Lhe rr?rst frame of the momentum p,&,(p) prajeets ontcl thrm4imenaional
spa=, m detailed in (2-3.1). Aeeardingly, the only so- compnent~thaf
contribute in (2-4.3) are the six T k l , whieh have a vanishing diagonal sum,in
view of (2-4.6), Here is the fivefold multiplicity asso~iatdwith spin 2,
2-4 Spin 2 prrrttctm, The gravtton 78

An alkrmtive writing of (2-4.3) is given by

P v ( p )*F~~(P)P~x(P~~'~(P)
= TLp(p)*~,~,~&(p)T'~(p)~ (2-4.7)
in, which

Some properttiw of tfie latter are

The projmlion chaxacbr of q , , , , ( p ) Xeada to the dyadie repreaemfation.

and are five in numbr. The sources for specifie slates are then identified as

When hdiciv states an? uwd in the veetor & d i e mnstrueLion

= f (GheLh6 + e(lph6e;h)- B(- 1)'6-~h# C (-

X1
1) "e:kle;-hl (2-4.161

obeys the relation

C
X
(--I)~G-~X S -26+x',;l_l + @So= 0. (2-4.17)

The spin. 2 hctlici-t;ysLaks art3 then identified as

The detdleui ~tna6tureof the vaeuum probability amplitude that I d 8 to

%heprczbability (2-4.3) is wrfitten out as
with

In order that this expression continue to exirsf in the limit aa m + 0, we must

have

where Jp(s)and H(%)are independenk sorrrce~at m == 0, The partiadar linear

combination of the two scalar 8aureeg K(%), T(z) echosen to eli
coupling between them. This is evident in the limitjng form

We Bee before us %heinvsrianL decomgositian. that &hefive heliciey &ate@,

accw~ibleto a mamive pa~icle: of spin 2, undergo as m -+ 0, falling into the thrm
poups *2, &L, 0.
The massless particle of helicity f 2 vvil be identified as the graviton. Its
source da~riptionis given by

We insert a treatment of gravitons that begins with this characterization. The

causal source &mangemen%

8ves the asud factoriealioa of the vacuum amplitude:

where each component murce o b v s

%Tpv
(p) = Q. (2-4.27)
Thig eource restdction snd the dyadic form (S3.55) are combin& to abtain the
eflective rep1&cement
3($'gP" +gl@g'@- g"gP') --t chh?e$$:,
X&'
(24.28)
where, on using helieity s t a b , h = &l,

The three independent tensor%mntrtined here: rare

and

which represent the two helieiLy states of the p v i t o n .

The graviton is unknown, as yet, to exprimental science. Nevertheless,
we shall accept it and ib conjectured p m p r t i e ~as the promr starting point for
%hetheory of pvitational phelraamens, ju& as the photon with it9 attribubs
initciates the ttteory of e h e h r n ~ m e t i ephenomena;. The ~ ~ d e n efore the
existence of the gaviton is indirect, but imprmsive, To indicah its naturcct W
present the following psrable: "The laws of quantum mechanics and relativity
have b e n wctll wdsblighed, but the intermtion pmpertiea of electfie eholrgm are
known only under quasi-static conditions. Two physickb, Max Stone and
Ichira Ido, point out that all such properties would follow from the postulated
exisfence of rt certain particle, on using Lhe principtm of source thmw. They
gnedict that the particle will one day be discavered. Othem dismka this ~ J U Q ~ W -
%ionas untvlbrr~nfed~ p c d a t i o n .The is~ueremains undecidd. "
The postubbd existence of thei gmviton le& first snd foremost ta the
s o u m re~triction$T@"(z)= 0, which, as in Ihe phobn discwssian, &at= the
aislence of a geneml physical fw. It is a con~ma-tionlaw,
con~%aney of the vector
(24.32)

The noerttion already indicates that them is only one conceivable identification
of this v e e t a ~ a lproperty-it is enerw-momentum. Unlike photon ~ources,
which hme a unique measure through the electric charge inhrpretation, grraGton
gources are eonfranted with an independent sale ori@naling in Lhe mechanical
significance of T,.. We provide an empirical conversion factor by writing

Again in contrast with eleetrie charge, energy or is intrinsically positive.

The establishment of a g r t ~ t o nsource dk%fibutionin m iaitial vacuum sifurt-
tion oa;n only be realized through Lhe t r a m p & of enerm and momentum infm
the mdon of i n b r e ~ t ,through the Esoundaries th& deEdt this domah. We
~onsidera slowly vaqing distfibution of graviton m m e g and deduce, as the
anabgue of (2-3.94), the e n e r e

where the maehanicd memure of paviton sources is used.

In the folloPi.ing a~tronomieal &pplications m are eoncemed vv3fh the
inkr&ction betweerr twa bodim, oxre of which (the "Sun"")m dimemions that
are effectively negligible and is charaekfizd by the single saurce oomponent
T0'(x), such th&t
(24.35)

, the second
The interaction energy between the Sun, staLioned at fBe o ~ G nand
test body with source distribution l,,(x, zO)is given by
K
E ~ ~ ~ , (= ) GM
z '--
where

Whm a body that move8 ~ $ & yhas momentum p@,

where cr i s an inv&ant measure of the mms distribution, For s ~ t a t i o n s vbody

of m w m, with dimensions th& are xlegEi&ble eompared %a32, the distance from
the od$n, we get

Thh is idb?RCifi~bfem $he NewLanian potentid e n e r e of attracting nzw~e8,

where-
~ / 8 a= G = 6.67 X 1 0 cma/&
~ ~ seca
= (1.62 X l ~ am)2.
- ~ ~ (2-4.M)
The second ver~ianrefers to &ornic units, in whieh & = C = It.
We shall now use elennenta~eon~iderationsto reprduee the four O ~ B ~ W L L .
tionral hsts of the Eimteinisn mdificatim of Nedonian theo~)..
1. A slowly nnoGng atom of mass m hafs the total enerw nz -- (GM/R)min the
neighborhwd of the body with mass M , The enerw relewd in an inbm8l
tran~formationis thus reduced by the halor 1 - (GMlR). T h i is~ the ~ 8 v i t a -
dlorr%lred &hi&.
2-4 Spin 2 paaiclesr. Ths graviton 83

2. Let the test body be a light beam for whieh t = opZ= O. The interaction
enerw with the 8un thus exceeds it%Newlonian value (replacing mass with total
enerw) by a faetor of tw, That is afso the increase of the deBwtion angle of
light over the Nefftonian wlue, whieh is Einstein's result For a direct cdcrtla-
tion we compare the acquired transverse momentum Mrith fhc: longitudinal
momentum of the h a m , which pmses act a distanee p from the Sun. The deflee-

3, The same inkraction reduces the speed of light by the factor l - 2(GJ(r/R),
e energy of a phobn is pi (l - 2GMIR) and differentiation with respect
d n ~ the
fo p gives the velocity. This &et has b e n oherved by mmuring time delltys
in radar echoes from the inner pl%neb, We eonsider the superior eortjunetion of
a planet, with the line of aight from the earth passing at distanee p from the
Sun. Then the abntieipated additional time delay for the echo is

whem z, and z, are the distances, from the point of closest approach to the SW,
to the earth and the planet, re~pectively. The coefficient in the differential
relation

hm h e n verified with fair accuracy.

4, The mosC subtle and inkrwting test is, of eourw, the perihelion precegsion.
of planetary orbib, We fimt consider the carreetion to the Newtonian ptential
enerw
k" = -GMm/R (2-4.44)

that is produed by the motion of the plan&. fiar small speeds,

we have

where
8rC 8ources Chap. 2

Thee eEects eorrect the Nwtonian p t e n t i d to

There is a earnparable relativistic madificwtion of the kinetic energ, given by

And, finally, there is the contribution to the energy density to@that is assoeisted
with the gravitafisnal inLeration between the planet and the Sun. This is
not locslilred on eilher mass, but is distributed in space in a way that can be
esl~ufibtedwiLh suEaientP preeisisn from the Newtonian field stren@h:

The inkraetion s n e r e density is proportional Lo the mutual h r m in fhe sward

field strength. XL is ~ v e by
n

as one verifies by inkgration:

The energy of inkraction between, the mass M and this distribaM m a s is

Ail sddilional interaction terms are exhibited in

This can be simplified by using the nonrelativistic enerw rdcbtisn, E = T -4- V,

which enabbs T to be eliminated in fslvor of V. An additional eonstsnt multiple
of V' does not produce wrihetion preewion; it only changes slightly the scale
of the orbit. It is the V' term that gives the significant deviation from the
2-6 Particlss with arbitrary integer spin 85

Ne~vtonianpote~ltietl,and perihelion precession. The resulting eEeetive potential

is
Yefr. = V - 3 ~ ' l n . (2-4.55)
Now, the equation of ;rtn orbit can be wriLten

and is the angular momentum per unit planetaq mass (it is often defignated
by h). Here we have

and

We see %batthe essential deviation from Ketvtonian behaviar is a reduetion of

the angle seale by the faetor 1 - 3G2MZ/&:, which requires that p increase by
more than 2n between sueeessive perihelions. This perihelion precession angle is

which is mactly Einstein" result. [He gave it in krms of the semimajor

axis a, the period T", and the eccentricity e, The connection is G2tI/L1 =
2 ~ ( a / T ) (-
1 ez)-"'.l

2-5 PARTICLES VVliTCI ARBITRARY INTEGER SPIAl

I n discussing unit spin and spin 2 particles, it was natural to replam the s ~ a l a r
soltree of spinlms particles by vector and knsor sources, The response of rc vee-
tor source, for example, to a homogeneous infinite~irnalLorentz transformation
is given by (2-3.221, which we write as

Cleetrljr exhibited here is ~h four-dimensionsl version of orbital and spin angular

momenta. This is a particular infinitesimal il1ustral;ion of the general, linear
response of a, multicomponent object to Lorentz transformations,
S(2) = L(i)S(z), (2-5.2)
where
= EPpxY -
2Z?C rp, IF,l;t,,E"x = (2-5.3)
86 Sources Chap. 2

details the typical inhomogeneous Lorentz transformation. One can always

choose the elements of a suitably multicomponent source to be real, and this
property is maintained by a real transformation matrix L(1). Corresponding
to the composition property of successive Lorentz transformations,
-
= - 2' = 1 2 v A-
~ Ae2', (2-5.4)
name1y -
.F" = (I~~z)'Az'- (€1' + 11Pve2"), (2-5.5)
(11l2)"~= IlP.I2'~,
\S-ehave
3 )=L ~ S , S(a) = L(I,)s(x) (2-5.6)
and -
S(?) = L(l1)L(l2)S(z) = L(llZ2)S(~). (2-5.7)
It is in this sense that the finite matrices L(1) give a matrix representation of
the homogeneous Lorentz group.
For infinitesimal transformations
l', = 6':+ 8wY,. Qw,,, = -6wrr, (2-5.8)
wve wvrite
L(1) = 1 + 3i 6dVs,,, (2-5.9)
and conclude that the matrices S, = -S, which are imaginary when L(1) is
real, obey the commutation relations that express the composition properties
of the six-parameter homogeneous group [cf. Eq. (1-3.10)] :

The complete infinitesimal response of S(x) is

Comparison with (2-5.1) gives a (4 X 4)-dimensional example of the imaginary

spin matrices S,,,,
(spl)A~= (l/%? - 6bp,>.
(a)gwK (2-5.12)
As the follo\ving illustrations show,

the matrices at1 are antisymmetrical and therefore Hermitian, while the see are
symmetrical and skewv-Hermitian. It was preordained that not all the S,,
matrices could be Hermitian, for the discussion of Section 1-1 shows that the
open structure of the Lorentz group precludes any finite-dimensional realiza-
tion of the group. This injunction is not applicable to the attached Euclidean
2-6 Parti~Ieswith arbitrary integer spin 87

group, and indeed the correspondence (zr = iz')

80k = -isx4, ( S ( 4 P (25.14)
do= give Euclidean spin matrices,
(QII~)XI (l/i)(6~paYK ~ X Y ~ ~ ( X ) I (2-5.15)
that &red l imaginary, antisymmetrical, and Hermitian.
Such diEerentid operator-matrix .realizations of the Lowntz generators &re
in striking contrast with. the operator &ructures given in (11-3.42) or (1-3.72),
for there the finite skew-Rermitian matrices sok are replaced by Hefmitian
sprators that are functions af momentum. Horv do we reeonciie these very
differrdnt hms? There mu& be a eonneetlng transformation that preserves the
commutation, relations but does not mai~ltainHermi-ticity and therefore is not
a unitary transformation, A. suggeslion of ~vhatis required comes from tho
follawing Lransformstioxl, ~vhich is appropriate d s t-2 sforvly moving particle
(p' -- m):
~ - slm)f = -rnr & fr, p sj
expfk(p * s/m)](-- mr) e x p t (p +

illtustmting how the momenlrxm dependence is removed, a t the expense of in-

troducing ske\lrrHermitiars. operator^. The follol~ingis the sn,zlogous stn;temenl
for arbitrav momentum,

@xp[=tr@(p
* s/lpt)I e x p [ &(p
~ s//pl)] = -rp0 & is,
(2-5.18)
\{?here
sinh B = lpllrn, cash @ = (2-5.19)
Its vmificatian proceeds m s t simply by eonsider-ing cornponeas parallel and
pewendicullar to p. The former reduces to the defining digerentiai equation
d@(lpl)ldIpl= l/pO,
and the lat%erto
88 Soureas Chap, 2

which describes the behavior of s under the "rotaition" "specified, by 8, Inciden-

tally, if a particle moves ~vithmomentum p, aXong the third axis, the t r a w
formation to its rest frame is
z3 = g3 GO&8 - z0sinh 8,
Z@ = --re3 sinh B -f- xo cosh 8,
The appearance of &isk in the role of
?tk = 8%

is emily understood. The commutation relations for the sp, are simplified by
introducing the linear combinations indicated by

and their cyclic permutations. All four-dimnsional commutation relations are

sumrnsrizled by the three-dimensiontzX commutation properties of the tiro in-
dependent rznmlar momenta,
s(E) X Stl) &(l) sC2) X S(21 = is(2)
I (2-5.25)
Conversely, Ire have the construction (8'2 = 83, . . .)
~ = ~ " ' + ~ ' g ' , m=i(s"'-s'2'), (2-5'26)
where the use of conventional Hermitian matrix representations for s" gives *''
Hermitian s matrices and skew-Hermitian R rnstI"ice~. What ive have encoun-
tered in (2-5.18) is the special situation in \'hieh sC2' = 0, or 8"' = 0. To
deal with the more general possibility presented in (2-5.262, we musk eonsider
s ta be the resultant of other spins. And, since it is often convenient to build
the latter out of still more elementary spins, we present the following general
theorem, in wrhich the h, are any commuting objects that obey

and

a superposition of independent *ins. The theorem is

2-5 Prrrtictes with arbitrary integer spin 89

If is verified as before by using the individual relations

The general matrix construction

clearly satisfies all the relevant commutation relations, including

We note that the skew-Herrnitian character of n is maintained with the use of

Wernitian matrices for the X,, as well as with the numbers =f= l.
We have not described explicitly the symmetrization between r and p0
since it would repeat the spin O discussion. The infinitesimal Lorentz trans-
formation response of the state (p! is now written as

where spin operators have been replaced by spin matrices, as symbolid by

(PIS = S(PL (2-5.35)
in which the matrices act upon unwritten indices in (pl. Next, we perform the
transformation (there are constant factors to be specified)
(p0)"2(~lo-)S B(P)S(P)~ (2-5.38)
and derive
d8(p) = [da (p X $ + is) + 6. (p0 $ + in)] ~ ( p ) (2-5.37)
On writing
S(p) = (dz)e-'~'~(z), (2-5.38)
this becomes
&!$(X) = - X& + &,)S(X),
3 GC#'(X~~~ (2-5-39)
where the matrices S,, are those given in (2-5.32). The analogous construction
of (0+lp)' is [cf. Eq. (2-1.25)]
( P ~ ) " ~ ( ~ + I PS(P)*B(P),
W )~ (2-5.40)
which involves the Hermitian character of B(p). The infinitesimal Lorentz
transformation behavior of S(z)* is like that of S@), but with the matrices
90 Sources Chap, 2

--6:. replacing the ..s, The use of matrix representstions with imaginary .8,
is re-quired to h consistent with rest S(%),as we have mentioned before,
The compact notation used in \vriLing (2-5.36) obscures an essential point,
We are desc~binga particle of definite spixl, but embed it in a larger system
when we employ constructions like (2-5.26). tlceordingly there must h premnt
atl the left of B(p)S(p), say, an explicit election of the states of interc;st, We
shall illustrate this, and a t Ghe mme time @ve a simple example of the eonncc-
t b n bt\vef?nfhe pre~entmatrix approach and the earlier pro~edures,by Gfioosing

The re~ultantof the two spins of 4 is either s = f or s = 0, f f \re trish to

describe unit spin particlea we must sdect them from the larger system. The
familiar s = f triplet spin funetians ean Ibr; trfiften as

An stlkmative vergion, \\"hieh &ISOinvolves the reality and sy mmeky of t h w

functions, is
i I (d(2)I f T s * IQ( 1 1 ), (2-5.43)
where
g. = 2-"2(-gl + ig2), G e i l = 2-"'(@1 + icp),
(2-5.44)
age; == Q.

The three spaee vt;?etorathus defined sre orthonsrmal in the sense

We also n o k the singlet funetion:

which exploits the antisymmetry of the Pauli matrix cz, as (2-5.43) depends
upon $he symmetry of the three ~ 2 ~ The
k . latter proprty is also expressd by

A corresponding decamposilion of the four-component mume into singlet and

&ripjetfunctions is conveniently written las
p 2-112
s q ~ ~ ~ . ~ 2=) (- ) (X) #
Jp(p)g21a'2'), (2-3.48)
where
C F =
~ -g@ = 1. (2-5.49)
We now examine the unit spin parliele source structum:

in whieh appropriate factors have h e n supplied. A eonsishnt use of the matrix

2-5 Partfelss with arbitrrrw integer spin 91

notation gives
= 4 t r I@
(dop)-1'2~p~ e:b(p)#Jyl (P)~(P)]
= &:Jp:~,(p),
where
b(p) "- =p l---P@
pi'lpll*
We first note that
e',; = tr [c.e?b(p)crPb(p)]

has the follorving property:

PP$! m 0, (2-5.54)
sinee
pPa, == m(cosh 6 3- sinh Ba * p/lpl)
= m[b(p)-1]2, (2-5.55)
and
trak = 0, (2-5.56)
Next, we consider

in which the form of the second factor depends upon the HermiCian naturt? of
b(p) and the a;, The following identity expresses the role of the four matrices
2-v2g, LLBan orthonormal basis for 2 X 2 matrices:
(trapX)(tra,%f)= Ctr @X) (tr o V ) --- (tr X)(tr )'l
+

= 2lt.r (X Y ) - (tr X)(tr Y ) ]

= det ( X - Y ) - ded ( X Y). + (2-5.58)
The rnultipli~stionproperties of determinants, and the remark Chat

whieh follows from (2-5.561, shot%?that dl reference to b ( p ) diwppears from

(2-5-57'), giving
eg:epphp = f tr (G et. eh')
b ~ ~ f . (2-5.60)
And, finally, let us consider

= 4 tr (qpb2)$ +[tr (gp@') - (tr 8)(tr Q')].

tr (cpbZ) (2-5.61)
The individual traees here are
92 Sources Chap, 2

and
*[tr (rN@') - (tr bJl)(trG')] = gfiV, (2-5.63)
tvhich gives

We have nmr reproduced all the covariant p r a ~ r t i e sof the three polarization
vectors for unit spin, When the third axis in (2-5.44) is ideatifid with fhe
direction of the momen't;urn vector p, the explicik expresions obtained from
(2-5.53) are just the heficity l a b 1 4 vectom (2-3.28,29). LneidentalXy, an
using the singlet rather than the tripjet functions, we get the farm

which is the anticipated scalar combination.

AS the basis for s corresponding treatment of arbitrary integer spin, we
consider the spin combinations

where the individual makriees set upon the appropriate index of %hesource
...pQ',"
&l1). ....p(p). (2-5.67)

Since all the matrices ,'2sl:@ a = l, . . . , R, appear OD the same footing, we

impose tt perrmis~iblesymmetry. reskriclion by requiring that (2-5.67) be un-
changed by any permutation of the a indiees, in which'':c and @hzbZ'are regarded
as a unit. Thus, for n == 2, we have
8,yf C%E1)"(2)
I *B
12) : , S e a l ) (11 ( 2 ) ( 2 ) .
"1 02 @l (2-5.438)
The simplest procedure is to replace each four-valued index pair @L1'@:'' by a
four-vecdor index in the manner detailed for unit spin, This gives the equive
lent souree
S F l - *.@@(P),
which is unehangd by any permutation af the ar indices*
The knsor of rank 7% a8 intmduced here describes a larger system than s,
particle of definite spin, Part of the nrscesstzry reduction is produced by the
projection faetors &,(p) that appear separately for each. v e ~ t o rindex in the
coupled source structure

urhere the secand form refers to the rest frame of the momentum f . The num-
ber of independent components possemed by the symmetfical threedimensisnal
tensor Sh,...ks, + +
namely i(n I) (n 21, agrees with the number of states ex-
hibited by a 8ymmetricaX collection of n unit spins. The total spin quantum
2-5 Partitles with arbitrary integer spin 93

numbrtr ranges from s = rtr through s = n - 2, ?t -- 4, . . . , terminntirtrg a t f.

+ +
or Q as n is odd or even, and (2s 1) == $(n 1)(11 3). A eombinntion of-+-
two urlit spins into a, nuH resultant earrcsporlds ta Eormirkg the trace of :I,pair
of indices, as in Skkka*.*k,. T O remove this passibility srld themby %lee$ o ~ f y
s == n l we must make S k , . . . k , traceless. The subtraction of tire appropriate
number of restrictions gives the independent component courlt

as expected for s --- n.

The resuit of subtracting successive traces is indicated by thc symmetrical
fo m

where zZ= (gk)' 8nd i t i~ required that

Xn view of the btal symmetry of the tensor, this property guarantees the
vanishing of the trace for any pair of indiees. The problem thus p a s d is a
familiar one. The polynomial of degrcse 1~ given iu (2-5'72) is a solution of
Laplace's equation according to (2-5.73). With z2 set equal to unity, it is a
spherical harmonic of degree n. T o identify the coefltieients Gnmt it S U % C ~ S to
consider the single nonvanishing component S33...3= 1. With 'g = I, = p,
we encounter the poIy nomiaI

which mu& be proportional to Legendre" ppolynomirtf, P,(p). Hence,

The reference to the rest frame is removed in

S*&...p l ( p ) ~ *@ ~* z L n = S
@ P n ( p ) ~ @* ~zP.

This generalizes the construction glven for n = 2, Eq. (2-4.5), and produces
symmtfic hnsom of rank n that obey
~ ' ' ' ~ ~ ~ ) ~ f i ~ ~ ~ Q? (2-5.77)
- ~ ~ f i , ( ~ )

provided sueh tensors are used in the egeetively three-dimensional context of

Eq. (2-5.70).
 The etmefure of the coupling between gources can now be p r e ~ n b dalter-
natively aa

The form of the prajeetion Lensor TX is given by

where, for exampfe,

2 Y = flPp,(p)zY
and
r* =X - $?l[(% ~
* * ) 1 ~ ~ ~ *
The ddition theomm of spheric4 harmonics provide8 the fabatorisation

atthough we here use the s ~ r m b ~

Y ,lx to designah mlid harmonics. We infer
the d y ~ d i ccorntruelion

The solid h a r m o ~ c sare being used in a somewhrcf symhlio way. They e m

be removed by introdueiag the generating funetion

whem, in. two-eomporrend mabtrix notation,

For ~implieily,we ilntradu~ethe abbreviation

 F a r t i ~ l s swith arbitrary integer spin 96

and obtain, for arbitrary n,

and, for n = 1,

Accordingly, we constmct the polarization vectors for spin n, from those h-

longing to unit spin, by

The known re~ultsfor n =. 2, given in (2-4.18)) erre immediately reproduced,

+
One can verify the orthonornzality of the 2n I polarization tensors by multi-
plying one such expression with the complex conjugate of another, in vrrhieh.
zMiig replaced by a/dx,. This gives

from which. we infer that

rl.--r*,
eph ept...p,phG ab~~,, (2-5,921
It may be coneluded that the source effective for emission into the specific
particle state Isbeled pX is

The complete description of multiparticle emission and absorlption processes

for these B. E. particles is contained in the vacuum amplitude
(o+jo_)' = exp [iW(S)). (2-5.94)
I n order La present the structure of W(S) ss compactly as possible, we use the
four-dimensional momentcm apace version of &+(X - X') given in (2-1.61) and
obtain

The tensor f I ( p ) retains the algebmic form represented by (2-5.79) and is an

even polynomial in p of degree 2n, The corresponding coordinate spaw struc-
tures are illustrated, for n == 1 and 2, by Eqs, (2-3.4) and (2-4.201, reswetively.
All the generalizations discua~dt3arlier in the context of special examples can
be developed for the arbitrary integer spin situation.
No reference hass been made to parity as an independent ~peeificationof
particle staks. That is because the particle sdates we have construehd are
88 Sources Chep. 2

rtutornatically endowed with a defitnite psrity. The gmmed~caloprsration that

reverses the positive sense of the three spatial axes is reprwented by the unitary
operator R,. Its effect upon the individual particle operators r, p, B izs I?;iven by

The transformed sin@@pa&icle state

refers to the spatial momentum. --p. Only for p = 0 can one exhibif an sigen-
vector of R,, a state of definik parity. As in the discuion of continuous
Lorente transformations, what is relevant to the probability amplitude ( 1 ~ 1 0 ) ~
is the relettionship between the description of the padicle state and the c h a m -
terization of the source. The tr%nsforaned padiele sLateeis repmsented by a
correspondingly transformed source which illustrabs the gentjral linear response

The reflection matrix r, is required to be mal if real sources are u s d . It acts

upon the spin indices to egect the geometried transformation

or, in view of (2-5.26),

The corresponding action of re upon S,;~I..

.,!jl;n).,!r, is the interchange of the
oA1' and ck2'labels, spa& from the option of an additional minus sign, whieh
is compatible widh the simple geometrical properly
rs2 = l. (2-5.101)
The pt3rmutation of a single pair of spin indices afiets ogpasitely the singlet and
tP.iplet combinations, comesponding to the opposite behavior, under spafisl
rdection, of the time and space components of a four-vector, We have ex-
pressed it this way, sinee if Xeaves free the choice of overall sign in the refiee-
%ion response, which is the alternative between a vector, and a p~eudoor axial
vector. The behavior of the tensor S,,...,s is that implied by the several vector
indices, together with the overall i factor. The concept of parity refers to the
rest frame where the surviving source components are Sx,...rs, which aet as s
unit under spatial refi.~?etion.When standard =tor behavim is eansider~d,
the parity is ( - T ) n . This gives a sequence of integer spin particlw with deer-
..
nating parities, as symbolieed by of, I-, 2+, . . The other aequence is 0-,
l+,2-, . . . .
Although the only known or conjectured massless particles of integer spin
have already been discussed, we shaH nev&heless pre%nt a unifirtd treatment
2-5 Partletss with arbitrary integer spin 97

of aH integer spin massless particles. As in the special examples, it is clear that

the limit m + O in (2-5.95) cannot be performed unless

is valid ad m = 0. Were we to c a r e out the limiting process in the manner sX-

feady illustrat;t?b, we would be tracing Lhe decomposition of the 2% J- I spin
states into the helicity pairs X = &n, & (n - l), . . . , =tz l, and X = 0. This
time, however, we shall directly extract X = An. The invariant form of such. a
source coupling is ...v
~ i ~ * * * ~ ~l .(. .pgm,@
) * nPrr8iE
@ n(~), (2-5.103)
where the projection tensor fX has s structure indicated by.

The produets formed from x b n d yp are four-dimensional. Any u s of the vec-

tor pp, as in (2-5.79), vcpould give no contribution in, view of the source restric-
tion (2-5.102). We now exploit that fcact to replace the tensor TX with another
that is equivalent to it in the context of (2-5.103). This is accomplished by the
following substitution, applied to both z p and y6",

in which 'F is any null vector with"p 00,such that p p f 0. The absence of
any change when p*%, E=:O assures the ewivalence of the two structures for the
application of interest. The new version of II: is given by

where, for example,

g Y = s"Pp,(l-",P)xP
and

Tn the discussion of the exchange of a massless particle, p@ is dso a null vector

and qp,(p, p) projects onto the subspace orthagonal to and pp :

+
Considered in the rest frame of the time-like veetor pp $P", the orLhogoasl
veehr pp -- p@ has only spatial components, doubling the parlicle's momentum,
and we recognize that. the subspace dected by 8,, is the two-dimensional
Euelidean plane perpendicdar to the momentum of the particle.
If only hcslieitiw X = fn are to be represented in the source eoupliing
(2-5.103), the tensor n must be irreducible with respect to forming traces in
98 Sources Chap. 2

the t~vodimensionalEuclidean space,

This is equivalent to asserting that, as a function of the X variables or of the

y variables in the plane, (2-5.106) is a solution of Laplace's equation, which is
homogeneous of degree n. The required t~vodimensionalharmonic function is

[(X X)(Y . (2-5.111)

where
P =X . y/[(x.X ) ( y y)]'I2 = COS 4 (2-5.11 2)
and T n ( p ) is the Tchebichef polynomial
Tn(p) = cos (ncos-' p ) = cos n4. (2-5.113)
From the coefficients of this polynomial we learn that
( - l ) m n (n - m - l ) !
m54n: dnm - J (2-5.114)
m! 4m ( n - 2 m ) !
and, in particular,
n 2 2, dnl = -4n.
The value of -3 obtained for n = 2 is in agreement with (2-4.24).
The identity
COS n4 = +[e'n~e-'n~'+ e - i n ~ e " W ' 1, + = ~ p - ~ ' , (2-5.116)

provides the relevant addition theorem. It implies the dyadic construction

~ ~ ~ u ePl.,-*n V L . . . ~ .
n ~ l ~ ~ ~ ~ n . v=
PA ~ P A 3 (2-5.1 17)
A-f n
where
-
e ~ ~ ~ " ". x Pxfln
l = (+X .X)(112)n.nf
z
n f inp
e . (2-5.1 18)
The phases are so chosen that, for n = 1,
l 1 2 .l*l *<P
e'pilxfi= ((ti.2) z e (2-5.119)
reproduces the conventions of (2-3.29). We now have

and the explicit construction

ePl"'Pn -
pin - eyhl -..e2il,
which generalizes the n = 2 result, Eq. (2-4.31).
2-6 Spin particles. Fermi-Birae staristics $9

The massless particle of helicity 3 is represented by the space-time source

structure

where

and

Ordinary matter possesses no conserved physical properties that could be

identified with the ones described by the laeal conservation law (2-5.1231, or
indeed for any n 2 3. The inability to construct their saurees strongly sfirms
the empirical absence of the particles. Bud perhaps one should not rejeet totally
the possibility of eventually encountering such properties, and the associated
padicles, under cireumstanees that are presently unattainable.

2-43 SPIM g PARTICLES. FERMI-DIRAC STATISTICS

There are two simple alternatives for constructing a spin 4 particle description,
in the sense of Eq. (2-5.261, namely

The two possibilities are interchanged by a reflection of Lhe smtial coordinates.

This indicates the convenience of a, more symmetricail treatment in. which both
t a b part. It is sIso advantqeous to replace the complex sources upon which
the 2 X 2 Pauli matrices aet by equivalent real sources, These remarks point
to the utility af a spin -$ particle description that employs four real sources, In
order to retain the symbol cr for use in the new context, we designate the initial
2 X 2 matrices as ~ k and
, use 7; for an independent set. Real, antisymmetrical
matrices can be constructed from the .irk by rcplaeing any explicit i by the
algebraically equivalent real arltisymrnetrical m&rix hi. Thus,

which are indeed real, antisymmet~cal4 X 4 matrices. They preserve the

algebraic properties af spin matrices :
4(qk,cl] = Eikll i g 1 2 ' ~ 2= l,
i~3 (2-6.3)
and we identify
sk = $C&. (2-6.4)
Initialy, in the role of n k we have ih*rk, where X is now a 2 X 2 matrix that
commutes with the T & , and has &l as eigenvalues, When the transformation
1W OOsour~ftil Chap. 2

T E -+ Q& is introduced, the m t ~ c e that

s could be used for i k are j u ~ fthree in
number :
i ~ ~ = i h~ $ipg~
~ ~ T ~ T Q , (2-6.5)
They are the analogues of igk with the T and 7' matrices inbrchanged.
The two sc?h of three anticommuting mnatfices arc? mutually commutative.
These six antisymmetrieal mtri.ces, Q,p&, and the ten symmetrical matrices 1,
orpg, provide s basis for a11 4 X 4 matrices. Since the three px, are on the same
footing, we arbitrarily identify h with p2 and writet

where the

&rereal, symmetrical matrices. We note their algc?br&e pmpedies:

the last statement b i n g the ma-lization of Eq, (2-5.33). Since space reflecLion
induces n -, --n without ehanghg s, it is r e p r e m u by a matGx that eom-
mutes with ar and anticommutes with pz, The only matricea with those eharac-
teristics arc: pl and ps. We choose the lathr arbibrarily and multiply this anti-
symmetrical matrix by i to get the real space reflection m a t h

which obeys

The spme refleetion matrix appears in another role on considering the real
matrix ms~ciated'VVith an infinitesimal Lorexltz; tr&nsfamatianfef. Eq. (2-5.9)f:

According to the symmetv properties of the matrices, transposition hrt.8 the

fallowing @Beat,
L T = 1 - i 6 ~ - -4 6 ~v - & r r , (2-6.12)
whereas
L-' = 1 -- i 6 0 - fo+ & v - +a. (2-6.13)
We express %his,through the aclion. of re, &S

The validity of this statement for the finite tran6farmations of the groper
orthoclrronous group is assured by $he composition prverty of succeiclsive
2-6 Spin ) particles. Fermi-Dirac statistics 101

transformations,
( L ~ L ~ ) ~ T .=
L L~TLTT.LIL~
~L~ = L;~.L, = 7.. (2-6.16)
The relation (2-6.15) also holds for the space-reflection transformation, since
r.Tr. = 1 (2-6.17)
combines the antisymmetry of r. with the iterative property (2-6.10). The
appearance of the matrix r, in (2-6.15) exhibits it in its fundamental metric
role. It is the analogue of the metric tensor in
F,g,,lPx = g,x
or, using matrix notation,
lTgl = g,
for (f)g, which attributes opposite signs to time and space components, is also
the space-reflection matrix for vectors.
Another aspect of the infinitesimal transformation matrices (2-6.11, 12),
in relation to the real symmetrical matrices a k and
a0 = 1, (2-6.20)
is given by
LTaL = a! - 8w X a! - 6va0, LTaOL= a0 - 6v a, (2-6.21)
which are united in
LTapL = (6: + 8d',)av.
This is the response of a vector to homogeneous infinitesimal Lorentz trans-
formations. The repetition of such transformations yields the finite transfor-
mation law
L * ~ L= rVay,
which is also valid for the improper space-reflection transformation generated
by L = 7.. Note that the symmetry of the a' and the antisymmetry of r,, as
well as their reality, is maintained by the Lorentz transformations.
We now consider the coupling between sources associated with single-
particle exchange, where the individual emission and absorption acts are repre-
sented by (2-5.36) and (2-5.42), with

The spin 3 particle has been placed in a larger framework, as evidenced by the
existence of the three matrices pk that commute with a. Two of the four com-
ponents must be rejected by interposing a spin-independent projection matrix
between the two B(p) factors that are associated with the individual acts. The
possibilities afforded by the three pr are really only two in number, depending
upon whether the p matrix used commutes or anticommutes with a. In the
 Chap. 2

first situation, we have

while the =@andone is iflustraM by

Spin 3 particle saurces will be designated m ~ ( z ) ?(p)

, or m m explicitly qr(z),
ar(p). The space-time extrapolation of the source coupling takes two alter-
natirts forms:

where, s s we have verified on wverd occasions, the use of tbe propagstion

function d+(z -- z') is required to maintain space-time uniformity, or the
EucXidean postulate. The= are examples of the qnadrati~stmeture

As the irreducible kernel of a quacffatic form, .Kgt (zt "i)~houldrwpond as a

f

unit to the act of transposition, intercharrdng 5. and c', and X'. This is not
true of the first possibility, (2-6.271, since 1 and p2 k h a v e opgosibly under
transposikion, Accordingly, the projeckion factor f -l-p2 is spurious since only
one of the krms contributw ta the quadradic farm. The second kernel cfm
act M a unit under the general transposition:
[(mps - #(l/i)aL)~+(z'-- z)lT = -(mp3 - olh(l/i)a,)h+(z - z').
(2-43.30)
f t is antisymmetrical!
Onr? might try ta convert this kernel to a symmetrical structure, without
upgetting the spin description, by invoking partricles and etntiparticlt3.s. This
requires an additioml twwvslued source index, and permits insertion into
the kernel of the antisymmetritt~leharge m s t q.~ The resulting kernel iis gym-
metrical but indefinite, sinee q is eonveded into -q by a charge refieetion.
That is in fiat contradiction with the physical requirement on the vacuum per-
sisknce probabiliky, which demands %]h& the ima@naq part of the quadratic
f o m be positive,
I,iWlz = ,-zrrnW 1 (2-6.31)
2-6 Spin # particlss, Fermi-Dirac stetisties 163

The conclusion is unavoidable that spin -& presents ta totally new situation.
Only one coume is open. Insbad of trying to modify the symmetfy cfismckr-
isties of the kernel to suit the algebraic properLies of the source, we must adapt
the algebraic prope&ies of the source to the antisymmetry of the kernel. The
comparison of the two equivalent versiom af (2-6.29) with the andisymmetq
property
&ts(ztIX ) = -Ktlt(xp X') (24.32)
will cease to be a paradox and become sn identity if

+
We sre thus forced by the charaetedsties of spin to abmdoxl the ordinary
numerical, commutative sources of Bose-Einstein stfttisties and introduce s
new kind of source and a new statistics. It will be verified shostfy that this is
Fermi-Dirac statisties.
The symmetry aspects of this discussion have been faeilitakd by the use of
matrices with definik symmetv, the symnnetricd the antkytnmetricd p%.
In later developments, however, uniformity of algebraic properties and Lorentz
trtansfarmadion behavior are more significant. f t is alllgebraieal~llyawkward that
the anticommuting ar commute with ;'ro the representation of a Lorenta trans-
formation on the 'a ws LTapL is not a similarity tramformation, and aga"
doe8 not have tensor transformation properties. To improve latter situa-
tion one must replace LT with h-'. That is accomplished by the relation
(2-6.14) which gives the new vector transformation form

lit is convenient to define im~dnary-matrices

7' = ir;la"
that; obey
L-Xr"L= P,?",
together with.
L-l?@rgL= l @ g i p h ? E ~ h ,
and so forth. The algebraic property r: = -1, along with 'a = 1, shows that

and

which afso gives the identification

The r matrices do not have a common symmetry, The: definition (2-6.35)

hplies that
fpT --i@~r;I$
104 Saurcs Chap. 2

This restates the antisymmetry of rO,which commutes with r. = iyO, and

shows that the ~k are symmetrical, skew-Hermitian matrices since they anti-
commute with the space-reflection matrix,

Algebraic relations among the r k are obtained as

* (yk, 7 1 ) = -4 (at, W)= -&I. (2-6.44)
The various characteristics of the r, contained in (2-6.38), (2-6.43), and (2-6.44)
are united in
B{YP?v)= - ~ H v * (2-6.45)
This unified algebraic statement i s maintained by Lorenta transformations,
according to
&-l+ jr', 7 , )L = -l~,lvhgab = -g". (2-6.46)
The Y matrices also give unified expression to

We first write a = ( 1 / 2 i ) a X a as
1
crr = [at,ad = &PI, 711, (2-6 -48)
and then note that
gok = iak = ~YOY~. (2-6.49)
These matrices are united in
@-P* = ~+EY,'IYVI,
which transforms as an antisymmetrical tensor,
L,-'aPYL = 1",lvhaKx.
The symmetry properties of the imaginary a,, are given by

which affirms that the akl are antisymmetrical and Hermitian, while the a o k
are symmetrical and skew-Hermitian.
The process of multiplying different r matrices together terminates with
This matrk i s real, and since

Alkmative factadsations of r5are

whiekr allso supplies the identification

iYs = pz. fZ-@*F;a)
The Lorentz trszrrsformation behavior of rSfollows from (2-6.54) as
L-'v,L = 1°,1'.12,13k~"gkrS
= (det orS, (2-8.61 )
which c h a r w h ~ z e rS
s m a pseudoscalar. It is invariant for proper transform
tionss, det I = +I, and mvems 8ign for improper transformra;tions, or reBec-
tions, det I = --I. The latter property f ~ l l o wdirectly.
~ from the anticannnnute~.
tivity of 7 and Y &S ~peeifi~slly
Q noted in (2-6.55). Let us also obsenre the
pseudo or axial vector efiarz9cter of ~ T P Y ~ ,
L-'irfirSL= (det I)P,irPv,. (2-6.62)
The components of irprs eomp~sethe four ways of mulLiplying together three
diAFerelat r matricea
The I6 independent elements of this CXiRord-Dirae etlgebra am organizd
through their Lorents tramformation bhaviar into the five

+ + -+
for which the count is l -f- 4 6 4 1 = 16.. Clsmly reXat& but &mtinccf
is the organization by symmetq properties. As suwwted by the emstruction

we consider ror, where r refers to any of the sets exhibited in (2-6.63). Then,
= ---rT~O
= -ya~;lrT~I, (2-6.6s)
sad the various equivabnces begween transposition md Bpwe reflection sfiow
that them mstriees have a definite symmetry. Indwd, the 16 m t ~ e e given
s by
106 sources Chap. 2

+
comprise the 4 $ 6 = 10 symmetrical matrices ror,, Y%, sad the 1 4 $- 1 = 6
antisymmetrical matrices r ', rOir,r, r O r ,. All the matrices are Hermitian.
The vacuum amplitude far an arbitrary spin +,four component spinsr
source 42) will be stated with the matriees p8 and a",in (2-G.%), replmd by
the appropriate r matrices:

and the source8 arc: totally anticommuting real objects,

which constitute the elements of a, Grmsmann or exterior algebra. Let US

analyrte the causal source arrangement

It is important to notice that even combinations of the totally snticommuting

sources arc! commutstive objects, and that the Lwa terms involving ql and 9%
are equd sinee the anticommutativity of the sources matches the antisymmetv
of the kernel. Acc~rdingEy,we get

and therefore

The matrix factor that occurs here is ~ u s (2-6.26)

t in a nebf- notation,

where, it will be recalled,

cosh B = sinh B = Ipl/m. (2-6.75)
+
The projection matrix +(l TO) is constructed from any ttvo orthonormal
eigeaveedars vh,
f 'vr = V ~ U =
~ ' 6hy, (2-43.76)
2-6 Spin f partictea. Fermi-Dirrc awtistics 107

in the dyadic f o m
+(l + = F v*u!.

A multipXicity cheek is provided by the trsce of this matrix equation,

where the relevant null trace of ra expresses its antisymmetry. A more general
remark follows on nsting that

A specific ehoiee of the v& can be made t;ts eigenvectors of a cornponenf of a,

@%Y6 3 ,
daVo = @Vu, Cr & 1, (2-6.80)
We dso ~ltrantthe% eigenveetors do be refa&$ by standard spin operations:

Other statements, expressing the use of imaginary ro and ar matrices,

You: = --v:? @,v,* = -av:, -*(ex f wz)~$= v&,* (2-6.82)
are satisfied by
* -- Z"QY&Y@*
v-, (24.83)
Since v: is an eigenvector of Y @belonging to the eigenvslue -- 1, there are cor-
responding orthogonaliky properties,
= Ot
**
v62;lUf Q, (2-6.M)
On inserting the eigenvector construction for &(l + Y') in (2-6.74), we get

where

which involves the anticommutativity of Y'Y with Y' and the eigenvector
significance of v, relative to 7
'. The same properties are used in verifying the
artlraonormzzlity of these vectars in the form
* 0Up.'
ZLpuY = V:@,? = &@.P.

No%?,according to Eg. (2-6,59),

IOS Gouross Chap, 2

which, combined with the hyperbolic relatiom,

This form shows the utility of defining the v. with respect to p ss s spin refer-
ence direction. Then @ = p/lpl can be replseed by the eigenvalue a, which i~
now a helicity 1Plbel. On employing the relation (s6.83) these veebrs borne,
simply,

They are altso emneekd with Lheir complex conjugates by

Q
,-: = ior5up.
When the falowiog vemion of (2-6.851,

is combined with the orthonormality statements (2-6-87], we recognise that

this non-Hermitian makrix has the algebraic projection p r o p e ~ y
2
- m - ?P.
2m
This is equivalent to
(m - Yp)(m4- ?p) == 0,
which is directly ve~fiable,since

We also learn that up, and u;@rO

obey

Let W return to the source coupling (2-6.73) and write

the consistency of the two definitions conveys the Hermitisn nature of 7'.
These are the precise definitions of single particle emission and absorption
sourem, nrhich have k e n built up from vari~usfachrs, Thus B(p) i s contain4
2-63 Spin 4 partiicfss, Fermi-Dirrec statistics 109

in uOzr,,. In the rest frame of the parkicle, U,, reduces to v,, which is an eigen-
veetor of TO and therefore of the space-~flectionmatrix F, = g?', Thus the
single-particle states have a definite, i f imaginary, parity. Incidentally we did
not prejudge this question by using the same matrix, in. defining r, and the
+
projection factar +(l ps). I t is now clear that the latter also performs a
parity selection, and that the refieetion matri.x must be defined accordingly.
The particle sources v,, and v;fl, as linear functions of the qr(z),are also
totally anticommutative,
O;rpa, qpfua' {%t
*
~ p ~ c =
* *
r ' {(7)1~#, ~ p ' r r r ) Q. (2-6. X W )
In particular,
(VP.)~ = 0, (v;@)% = 0- (2-C5.101)
The commutafivity of even source functions is used to write

All this is quite the same as with, B. E. statistics. But; now the power series
contains just two terms: a,, = 0, 1, for, on reversing the multiplication order
of two elements, we see that

and the whole series Germintlttes at n,, -- 1. In this limitation. to a maximum

value of unity for what art3 clearly particle oeeupation numbers we have a
statement of the Exclusion Principle, which, is a chartzcterislic feature of F. D.
statistics,
The catlsa;l situation is conveyed by the causal analysis of the vacuum
amplilude,

It is indeed possible to factor the coupling terms in the desired way, but strict
account must be kepG of the minus signs that are involved, This is facilitated
by the falEo6ng procedure, which we illustrate with two pa&icle sla$es, fabeled
a and b,

By always displacing sourees Lhrough an even number of factors, one avoids the
explicit appearance of minus signs, In this way we a r ~ v eaL a facbri~atiaa
where the emission sources are multiplied in some order, r e d from left to fight,
110 Sources Chap, 2

while the absorption sources appear in the same order, but read from right t o
left, It is given general expression by the following identification. of multi-
particle states :
(in)IO-Y = (O+IO-)~ n (i,.)%g
in which nT symbolizes the opposite multiplication sense from and any
standard sequence can be used for the denurnerably infinite number of particle
states. As in the B. E. discussion, the pa&icle occupation number interpreta-
tion of n,, is supported by the response to source translation, 4%)
-+ rt(z +
X),
which gives

where

shows the additive contributions of the various particles that are present.
The completeness requirement on the muttiparticfe states is stated alter-
natively as
(2-6.109)
where
(0-1 (72.1)" ((C4IO->'? (2-6.110)
s a d by

with

We have been at pains to bvrite these more carefully than in the B. E. situation,
since we are now dealing with functions of antieornmuting numbers, No pre-
~ a u t i a n sare needed for the vscuum amplitude, which is an even function, and
we present the two completeness statements as

where we have omitted the compensating factors of z" and -i. The comparison
of the two forms suggests a rule of complex conjugation for F, D. Boureeg thaG
we shafl find is a, consistent one: complex conjugation also inverts the sense of
multiplication, as illustrated by
Then, the single statement of completeness is conveyed by

which essentially reverses the faetorizzztion procedure of the causal analysis.

We must confirm this implication of completeness with a direct computa-
( O + ~ O - ) ~ / ' . It is importsnt to recognise that the complex conjugation
rule for F. D, sources implies that the product of two red sources is imaginary,

z 'real
Consequently, q ( z ) ~ ' ~ (is ) sinee YO is imaginary. This is another aspect
of the matching of the statistics to the spin. Sirlee the matrices (I/i)rF are real,
the only eomplex quantity in W is &.+(X - 29,and

(2-6.118)
The relation
(2-6.1 X 9)
then gives
q(-p)ro(m -- r p ) ? ( p )
Up C )l(p)* y~ U ~ @ ~ ; U ~ O V ( ~ )
Q

= Re C s;,np.. (2-6.120)
P@
The injunction symbolized by Re is redundant, since

which makes essential use of the complex conjugation rule. This resull,

is the verification of eompiteteness.

The Euclidean postulate was introduced as a sharpened vergion of the
principle of spseetime uniformity. It has new and interesting implications for
spin particles, if it is interpreted to mean that the Euclidean transcription
may contain no indication of the original Minkowski space. All reference to
112 Sources Chap, 2

Miakotvski space doe8 disappear in the Euelidean descfiption of i n t e p spin

pa&icles, but spin htroduces a new situation. In dbussing unif spin, for
example, we ohsewed that the Harnnitian, real, symmetrical m t ~ c e eb4 a .= igolr;
are conveded to Eerdtian, inn+nary, a~tbymnnetricafm a t ~ c m thus , uniting
them with the srl, by me&nsof the tnrnsformation assooisted with J4 = Q@.
N o b that it i~the sqmre root of the spetcerefieetion m a t ~ x or, itie negative,
that m k r a thi8 transfomation, To pedarm an an%lagausaperation, on the
real symmetrical matrices err = ~ O Y ~unifying
, them with the ima$nary anti-
aynrmeLrieal ~g -- *&%, we mu@% find a suitable unitary transformation, one
that cornmules ~ t hthe. lslthr set. The only poasibif ties a;vailable far the uni-

~ real, and cannot ehange fhe reality of the ~ 4 Ac-

But all t h e ~ em $ t r i ~ eare .
cordingIy, an inspection af the redity, or symmetw of the c,, p, v ==: 1, . . . , 4 ,
leave8 no doubt about which Euelidean axis is related fa the Minkovrtski time
=is, This is a viof%tioaelf the EueXidean postulate.
We have slready remarked that the symmetry of m ~ t ~ ~ean e t gbe rever~ed,
witbout d t e ~ n gtheir spaeatinne charackr, by u ~ i n gan independent a n t b m -
metrical nnatrh

which acts upon stn dditiowl %WO-valued source index. Its introduction en-
ables us to form e, eomplex unitary matrix by multiplying the real r. = g'/' by
Lhe i n r a a a q q and then taking the square root, in a n a l o e ~ %thehunif &pin
procedure, The explicit tr~nsfsmationis
e(ri/4)g~' -(r i/41q7'
E @&v@ t (2-6. f 25)
and indeed

are all i m a g i n ~etntisymmetGea1

~, matrices. The detailed tfansfomatim from
Minkowski Lo Euclidean source is @yen by

Whm this trantgfismaLion is performed in, the vacuum amplitude, one en-
oounters the following rnatrk (note that a is symmetFioa1) :

are given by (r4= iro)

at = r0vX, ad = pyoe
2-43 Spin 4 patticteat, Fermf -l)iirsc atetistics 113

They arc3 all real, symmetrical matrices that obey

and

The regulting Euclidean tmn~criptionof the spin vacuum amplifrxde is

which is a real struetum when real Euclidean Isources are ued.

The implieation of the Euclidean poshfah, that every spin -& ptbdicle
possesses a charge-Xike attribute, is entilrely compatible wifh the empirical situ-
a%ir>rr,Although we must give speeial attention Lo the massless neuCrinos, it is
a gcneral inference from the data thaL every hmnion (F. B. particle), including
electrically neutral ones, has its antiparticle counte~parL,while no electdcally
neutral bo~on(B, E. particle) shows sueh duplexity, The charge label q = &l.
is added h the spin 3 state8 by edztrging Q, and up@Lo bc? eigenvecLors af the
charge matrix with the eigenvalue g. Since the charge matrk is ima$;inal.y, com-
plex conjugation introduces ---g., and some correspondingly modified statements
arB
U: -, -,= GY6~,,, (2-6. f 34)
and

The relatd particle source definitions are

qpgg =
1/23 *
(2mdo,) ~ , . , ~ O v ( p ) , = (2mdw,) "'s(p) *you,,,. (24.138)
This dbcus&on. of the Euclidean postulate brinp %heTCP operation t;o
mind, Thmu& the attached Euclidean p u p wa produce the tramformtion
zi" = ---g^ (2-6. 137')
&S
~ ( 3=
) rstq(~)p (2-6. t 38)
where
e(ri/2)elle(rii2)c.4 =
e(w~/Zlszge(ri/l)s~, e ( r i f 2 1 @ i i e ( ~ i / ~ ) * 4 --iT5
=
( 2 4 .f 39)
de-l&ifs the ewivslent raLations through the angle r in turo peqrtndieular
planes. This matrix is antisymmetricd, imaenary, and obeys

The invariance of Che vacuum amplitude is vtsrified direekiy on uging Lhe mlstion
This is accomplished, hors-ever, a t the expense of replacing the real by tan
imaginary q, sinee that is the nature of rat* If we insist on s r e d T , as in the
transfarmation
s(Z> = Y 5 1 ( z ) , (2-6.142)
W turns into --W, But this sign change can be compensated by reversing the
multiplication order of all sources, which is in sceord with the representation of
causal sequence by mu1tipEieativc?-position.
The eflect, on the individual emission and absorption, sources, of the sub-
stitution
?(P) +Y~v(-P) (2-6.143)

= (2m do,) "2q(p)*70~,-@(-icr) (2-6.144)

or
?pug + -i@vF -@ -g, (2-6.145)
and
v;uq iflvP -@ -g* (2-6. f 46)
The resulting correspondence between multiparticle emission and absorption
processes is

where
= E, -@ -qt

and the source transformation that constitutes part of the FCP operation pro-
duces the required reversal of multipliestion order,

2-7 MORE ABOUT SPIPJ $ PARTICLES. NEUTRIMQS

A8 a preliminary to discussing the angular momentum specification of particle
states .u;e review the addition of orbital angular momentum with spin 8. StaLes
+
of total angular momentum quantum number j -- 1! =i= are selected from the
subspace with orbital quantum number I by the Hermitian projection operaton

They obey
f j j j j CMrj=I
i
2-7 More about spin g prr2icles. hlautrlnor 115

and have a trace appropriate to the multiplicity,

We define ordhoxrormaf spin-angle functions

which are given explicitly by

The two functions are also connected by s n ~ p e r a b r :

where n is the unit vector that supplies the angle variables of the spherical
harmonics. The fofiawing properties of a * n are involwd: i t commutes with
the total angular momentum veetor, but dters %heorbital angufar momentum.
by uniw; it does not change the ortkonormaIity of the sph-angle funclions-it
has uniL square. All thia shows that the left- snd righ6hand sides of (2-7.6)
are the same, to within phase constants that cannot depend upon m. Xt then
sufliees to set n parallel to the third axis and choose m == l$. The unly surviving
+
harmonic, Yto = (21 1/4?r)"~,selects v+, and (2-7.6) is confirmed.
The ~trvctureof the soufce coupling produced by single-particle exchange
is (causal subscripts are omitted for simplicity)

restating Eqs. (2-6.73,74). We intraduce the preliminary transformation

( 2 dwP)
~ "2?(p) = C
1m
( d ~ ) Yim(~)rlp@tm, (2-7.8)
where

also carries an unwritten index, expressing that of the multicomponent ?(X).

+
The projection matrix &(l ro) makes a. selection of these components, and
+
4(1 rO)irS = i ~ & +-( lTO) makes a complementary selection. This will be
+
indicated by adding subscripts and - to qpar,. The residual spin multiplicity
116 Sources Chap. 2

is coupled with the Ylm(p)to produce the spin-angle functions, as in

C (dQ)'l2Ytrn(~)~*p0~m
Im
= C (d~)~"~ljrntl*potjm?
ljm
(2-7.10)

where we rely on context to distinguish the orbital magnetic quantum number

m, which assumes integer values, from the total angular momentum magnetic
quantum number m, which is an integer ++.
The specific combination that
appears in (2-7.7), for given j, m, is
COS 36 C ( d Q ) " 2 ~ l j m ~ + p ~- sin 3 0 ~(p/lpl) C ( d ~ ) " ~ z ~ j m q - ~ o t ~ m
ljm
I I
= C ( d ~ ) ~ ' COB
~ Z t 3 & + P ~ l jm - sin +6q-p~ijrn], (2-7.11)
E

and its complex conjugate, where

= 2j - 1,
indicates the orbital angular momentum change that is produced by cr p/lpl.
The orthonormality of the Zum in the subspace selected by +(l 7') gives the+
d t i n g form of (2-7.7) :

where

and the charge label is left unwritten.

On combining the various transformations, these single particle suurces are
exhibited as
np0l jmq = ( d ~ ) + p ~ Z j m q ( ~ ) * ~ ~ ~ ( ~ ) ,(2-7.15)
with

wherein the Zljmqare constructed as in (2-7.5) from the eigenvectors v,,, and
the spherical harmonics refer to the angles of the unit coordinate vector. The
comparison of (2-7.13) and (2-7.15) with the left-hand member of (2-6.73)
supplies the identification

and the antisymmetry of r0G+(x - X ) extends this to

2-7 More abaut spin 4 particles. Nsu.trlnes 117

Unlike linear momentum states in general, angular momentum states per-

mit a specification. of spaee-refiectim parity. The response of the parficfe
sources to
X) -+~ Y ~ ? ( Z-I)
~,

Imvolves the transformation behavior

This follows h m the homogeneous nature of spherical harmonic&,

and the significance of Z l j m g as an eigenvector of 7' with the eigenvelue +l.

While i~$Zrjmqassigns the eigenvalue -1to YO, this sign change is compensated
( =- - The result is

which exhibits space parity as a product of two factors, the intrinsic parity i,
and the variable orbital parity (- 1)'. The label 1 in $palimQ should be under-
stood as (-l)', the exact parity quantum number, for both orbital angular
momenta, 1 and 1, are present in this funetion. In the spin 3 situation the two
states with comrnon values of j, m can be distinl~;uishedby their different parity
values. For spinless particles, aeeording to (2-2.24), parity also appears as the
orbital parity (--I)', multiplied into an intrinsic parity which is +l for s. scalar,
--1. for a pseudoscalar source, Eere, p a ~ t yis superfluous as a label, being corn-
ptetely determined by the angular momentum quantum number. With unit
spin particles, however, parity is insuffieienf to identify d l three at8tes of speci-
fied total angular momentum. In addition to an intrinsic p a ~ t yfactor, -4for
a vector, +L far an axial vector, the state dese~bedin (2-3.39) h@ the ofiitaf
parity (-l)j, representing I = j, while the two states of (2-3.41,42) have the
orbital parity --(--l)', whieh is common to 1 = j & 1. But for the massless
photon there are just two types of 8ta;t.e~ of a given aagufar momentum wanLurn
number j > 1. The photon state with source has parity --(--I)~, and
that created by has parity (-- l)j. The two kinds of sources are conven-
tionafly cdled magnetie and electric multipole moments, respeetivdy.
Before i n v e s t i g a t i ~the e k t of the TGI, operation on anwlar momentum
states, we examine the reality properties of J . p o r j m g ( z ) . Let us ant note that
Z$ma = (-l)'+"miy~l 5 -m -pp (2-7.23)
which uses the spheric& hsrrnonie property

and the complex conjugation behavior of v,,, being Eq. (2-6.134) with'p = m.
On foming the complex eanjugate of the additional minus aigns that
are produ~edby the expficit appearance of z' may be compe988tc3:d through the
space-time reflectTon zp --+ -9,

The canneetion which this relxttion es%abEghesbetween the two causal forms of
6+(z -- a") is conveyed by the invari~neeprope&y

in agreement with (2-6.141). The eEecL, in Eq. (2-7.151, of the substitution

(&)h" ) i(- 1)
5qf~=
jmP(-~)*To~ ~p*a E J-m -Q, (2-7,28)
and then
jmg --B -i( -l ) E + j + m rlpar j -m --g- (2-7.29)
This gives the detailed canrespondence bebeen single-padicrle ernisgion and
abmrptian aets. The multipadicle correspondence is analogous, wi&hthe re-
versal of multipfication order etppmring as an aspect of the TCP transfarmation,
A, spaee-time description of the multip~rticleexch~ngebet\veen sources is
produced by f he power series expansion:

where the discrete indices an sources and prop~gationf"unc.tionsare regarded 8s

combined with the explicit spacetime coordinah~. In contrmt with %heper-
manene of (2-2.37), the symmetry of which conveys the commutativity of B, E.
SOtXrceg, the antisymmetriczal drtteminant
det,,, r"@+(zi-- x$) = e jl . . . j,~'~+(zl - zj,) a r"@+(z, - 25,)
nl perm.
(2-7.31)
expresBeS the anticommuta;tivity of F. D. sources, We aee hercl: the simlple and
necessary connection between the symmetry propedies that clnaraebrige the
two statisfcics and the elemenlary algebraic propedie~that; distinguish fhe two
kinds of soureek Appropriahty symmetfized prorfuels of individual praprtga-
tion functions give the space-time representation of the noninteraeting multi-
parLi~Iesituation.
Let us diaeuss now those generalizations in which the t e r m i d vacuum
states sre replaced by multiparticle ~t%tf?s.A causal situation is considered,
containing emiwion gaurce gz, probe source q0, detection source q1 :
2-7 More about spin 3 partiefes. Neutrinos l 19

The vacuum amplitude is given by

(O+lO-)L (0+l0-)'l+'~ exp i (dz)(dxf)ql(X)? Oc+(x - xf)qo(zf)
[/
+ i / ( d z )( d x 3 ~ o ( z ) 7 ~ @-+ bx')~a(xO](o+[O-)'O

where the index r represents any set of single-particle labels, say pcrq. The
causal analysis of this vacuum amplitude is

from which the detailed effect of the probe source can be inferred. To describe
a weak probe one must interpret the product isr({n)10-)?. If the single-pebrticle
state or mode r is initially occupied, n, = 1, the result is zero, ( v , ) 2 = 0. This
is the Exclusion Principle, forbidding the introduction of an additional particle
into an already occupied node. Otherwise,

where n,, counts the number of occupied modes that precede r in the standard
sequence, which is the number of source factors in ((n) 10-)"hat iqt must be
moved through in order to place it in proper position. Similarly,

and we get the weak source results

( (n + I t ) +l(n)-)Q(- X)n<ri~rt

{ { n - lp)+l(n}-)q Si (-I)n<riqF.
To construct the probability amplitude ((n)+[{n) -)V, one must retain only
equal powers of v f , and q 2 , in the expansion of (2-7-33),
exp [F(isrdrlo, + iv;,inzr)] + IJI[1 f i q ~ r i q o h & i ~ 2 ~ 1(2-7.38)
.
In contrast with the B. E. situation, the series terminates with the indicated
product. On referring to (2-7.35, 36) we see that

where the factor n, indicates the absence of the term n, = 0. The effective sub-
stitution is, then,

= exp [F iYwn.iq~,] (2-7.40)

920 8aure8s Chap. 2

The linear relation between ? ( X ) and emission and absorption sources for any
type of mode specification can be w i t t e n 88
(dz)~ ( z ) r ~ ~ . , ( z ) .(2-7.41
Thus,
$,.@(z) = (2mdw,) "2u,,,e'P", (2-7.42)
and Eq. (2-7.16) supplies another example. The related canslmction of the
propagation function, is that illustrated in (2-X17, 18) :

We now ge.t;
((n)+1 in)->L
exp
with

The form of the second term assures the antisymmetry of TOG~,,+(~- S').
Explicit causal stmctures are

The eannpa~on*tb (2-2.49) emphtzsiaes the essential role of the sta%isticsin

stimulating (B. E.) or suppreming (F. D.) additional particte emission.
We return t;o the vacuum amplidude exprm~ion(2-7.33) and ob~emethJ,
in gexrer&l,

X [l + i q Z i q o r i s ~ l i s ~ r l(2-7.47)
,
which converts (2-7.33) into

(2-7.48)
A typical term of the product n, in (2-7.48) appears as

where the two sets of modes labled a and e are disjunct, sinee the individual
mode facton are linear in ql1 and q f . If B nonvanishing t e r n is to result in
2-7 More about spin 3. particlsrr, Neutrinos l21

(2-7.48), it is necessary that n, --. n, = 0. Then the a(bsorbd) modes art:

tho~e,occupied in the initial atate,

th& sre not oecupied finally, while the e(mitted) modes are those oeeupietd
find1y,

~phichwere initially unoccupied. The outcome is

(in + l.)+l{n + L)-)" ({%l+{a)-)" G

[(-l)n<'i~rl nT
a
[(-~)'<~iq:l,

(2-7.52)
whieh i~ the generdizlation of (2-7.37).
In order to test $ha?eompledenctss of the multiparticEr? state8 in this general
context, we multiply (2-7.52) on the left by its complex conjugate and prrtst;nl
the ~ m l int a farm that reinstates (In) m an arbitrary initial 8taLe:

The summation. over a11 Gnsl states is represented by

keeping in mind the antieommutativity af sources. But, according to

(2-7.4, $51,
(2-7.55)

and on utilizing the analowe of (2-6.122) for a, general mode specification:

e (2-7.M) equals unity. To derive?(2-7.56)

This confims that the bfl-hand ~ i d of
direcfly from the propagation func6ion eonstmction (2-7.43)we no%@ that, as a
statement about individual elements,

is valid for df ;e - z', But irl(z)rl(z') is red, and fherefor~

(0+/0-)7' = exp (dz)(dz')q(z)~'$,(z)$,(z') *r"v(z')

as antieipatd.
The reduetion of unitarity to causality for spin i$ particles imitates the pat-
tern alresdy established with spinless particles. We follow the development of
the syskm. from the initial V ~ ~ U Ustate,
H ~ under the influence of the source
qls)(~),and then trace it back to tlne initial strtte, using the source g(,,(z), The
stru~tureof the propagation function 6+(z - z') governs %hisevolution and
we get, as the analogue of (2-2.83),

G-(z - 5') = (m - ~ ' ( ~ / i ) a , ) ~

-_z')
(,s
(2-7.6 l)
G"'(z - z') = (m -- Y@(I/~)~,)A'&'(~ --
Some relations among these matrix funetions are:

and

When arbitrary mode functions are used,

G'+'(% - zp)= C
r
G[-)(X- X')
+,(z)y~.~(~')*r~, = --C$r(~)*h(z')~o.
T

(2-7 -64)
The various functions are dso eanneeted by the identity
 According to the causal analysis

a cheGk of completeness s r unitarity is prformed by verifying that (2-7.M)

reduces to unity on identifying ll(,)(z) and ?(+)(X). This i~ just the content of
the identity (2-7.65), combined with the third statement of (2--7.62). The
generalization to the amplitude ((R) - (B) -)sc->eT(+) invalves the replacement
of C+(%-- X') by +(z - X'). It is expressed by retaining the same set of
relations but based on the new definitions,

As in the spin O discussion, tfle general unitsrirty proof uses the sources q f f ) ta
generate arbitrary terminal st&s. The complotc3 removd of reference to a
subsequently ~letingG O U T C ~~ ( 2requires
) the additional relations

w7hich are comect statements under the assigned eausd circumstances.

The eharge property that is required by the Euctidean postulate played no
role in the initid treatment of spin particles, This raises a question eoneern-
ixlg the possible existence of other kinds of spin g parLictes for which the ebarge
matrix dam make an explicit appearance. We shall take for p a n k d in this
discussion that Ghe charge attribuk remains ulnalkred during the travel of %be
particle between emission and &sorption sources. That permit8 a, dependence
of the propagation funetion upon the charge matrix g, but not upon the charge-
reflection n a t r k r,. The generd f o m of such a pmpagstion function is

which lacks only the matrices g,, = --CF.,. Since TO and ror, are antisym-
metrical and r0up symmetricd, the antisymmetrical matrix q eennot multiply
them. But p must be used to reverse the antisymmetry of Y @ Y ' ~ Y ~ . This struc-
ture shauld fit into the mode function pattern detailed in (2-7.633, M) since the
latter refern only to the combination of individual emission and absorption wcta
An essential aspect ia the positiveness property

where! ~ ( 2 is
) an arbitrary complex-valued numerieaf funetion. X t a implieation
124 Sources Chap. 2

for (2-7.69) is the following Hermitian matrix positiveness requirement asso-

ciated with any particle momentum F,

which incidentally asserts the reality of ml,mz, a, and X, When a particte of

mass m > 0 is viewed in its rest frame, this condition reads

The three matrices ipT5, Y', ror5anticommute and are of unit square, from
which we infer the numerical requirements
ma i [ ( m a ~ ) m
~: + + m:]'" 2 0. (2-7.73)
In addition to the conclusion that a is positive, we note that the zero value must
be attained if a projection matrix is to be produced, and accordingly
m2a2(1 - h2) = m; + m:. (2-7.74)
Throughout the open interval h2 < 1 , it is permissible to normaliee a by
a2(1 - X') = 1,
with the consequence
ml2 + m: = m2.
This is represented as
a = cosh 8, QX = sinh 8; ml = m cos ( p , m2 = m sin cp. (2-7.77)
It is then easily seen that
y OG+(,
,t) = ,-c 1/~
l/ 2 ) ~ ? 5 ~ )( @ ~ Q T s

X [r"(m - r*(i/i)a,)a+(x - Z t ) j e ( 1 1 ~ ) e 4 ~ 6l el 2( 1 ~ 1 7 (2-7.78)

~.

In view of the antisymmetry of and the symmetry of ((v5,these matrix fa*

tors can be consistently transferred to the two source functions in W. Thus,
despite the initial appearance of q and of T5, they disappear after suitable re-
definition of the source and we restore the structure of (2-6.67,68).
The remaining possibility is
X*= 1, r n ~= m2 = 0.
If we now chwse
a= 4,
there emerges
2-7 More about spin + particles, Msiutrinos 725

The first version ixrdicrttes that we have regained (2-6.2q, with the objection
to the antisymmetry of p2 =. iyti removed by the pmsence of the additional
antisymmetrical matrix Xq. The second version is related ta the standard form
of raG+ by symmetrical matrix factors, which could be transferred to the
sources. These are sinwlar projection matrices, however, and the new sources
will be subject to the restrictive condition

A source constraint states a universal characteristic of all realistic mechanisms

%hateoxltribute to the creation or annihilation of the given particle, Whe$her
+
the iateraetioa mechanims of spin. particks might be compatible with such
a restriction cannot be examined at this point, save for one exceptional class of
spin $- particles-the neutrinos.
Only one k n d of neutrino interaction has been observed, processes in whi~h.
thr?y are created or annihilated in company with a charged lepton (electron crr
muon), The neutrino astsoeiated with an electron is a different particle from that
agmciiated, with a muon. The masses of the neutrinos are small on the scale SM?L
by their leptonic partner, but they are not known to be zero in the same sense
as are the photon and graviton masses, where inverse masses must ex~eedvery
Iarge macrosqic distances. Both neutrinos are observed with a unique helieity,
which is detemined only by the electric charge of its partner and reverses sign
vvith the latter, The spin 4 possibility just discuswd provide8 a natural frane-
work for the representation of these properties. (Indeed, it was in essence
proposed long before the experimental disclosure of the two neutrinos, and con-
erLiLuted a prediction of that fact.) To avoid confusion with electric ebargc?,let
the ehrzrge prope&y carried by the neutrinos be designakd by 1, the leptanic
chrtrge. The alternatives contained in k2 = 1 give two kinds of particles which
are distinguished by the corresponding projee(tion factor:

On considering a neutrino of momentum ,'p the other matrix factor in (2-7.81)

becomes
-YOY'~, ='p - i r , ~p.
(2-7.M)
which usea %heeffective equivalence between i Y t j and &l %h&is enforeed by
(2-7.83). Wben the neutrino has an energy that is large in comparison with
its mass, which need not be zero, nor the same for the two neutfinos, a unique
helieity is selected :
(2-7.85)
The conservation of leptonie charge requires that the efeetrically charged lepton
accompanying a given neutrino carry a Ieptonie charge opposite to that of the
128 Sources Chap. 2

neutrino :
l + b. lept. = 0.
We now put forward the natural hypothesis that one role of Ieptonic charge is
to distinguish, and label the two leptons with a common electric charge q:
Ech. tept. = TB* (2-7.87)
Its consequence is the empirical equivalence between neutrino helicity and the
accompanying electric charge,

In the interest of compIeteness we shall exhibit the sources for specific

neutrino states, under the simplifying assumption of zero neutrino mass. One
uses the dyadic construction

where
~ ~ = 6 r~ ~ 1 ~ ~ ~ ~ ~ l ~
and
*
Up1 = Up-l.
These eigenvectors also obey

The sources for this kind of neutrino are

and those for the second type are obtained by the reflection

where r~ is the real symmetrical leptonic chargereflection matrix. The TCP

substitution
d P ) W?(-P)+

interchanges the neutrino emission and antineutrino absorption sources (2-7.93)

in the folfowing manner:

Additional minus signs appear for the other neutrino type.

Finally, here is a brief comment on the Euclidean transcription of the
vacuum amplitude associated with (2-7.81) (=i=l replace Xq). It is
 Particles of integer 4- 4 spin 127

extends the set of red symmetricsl, anticommuting matrices of unit square.

Unlike the Minkowski form, las is mtisymmetrical, and correspondingly mti-
commutes with the a,. The explicit appearance of the imaginary matrix I
means that the individual Euc-lidean forms are not reai. But if the mmsw of
the two neut~nosare the same (zero?), complex eonjug&fioninkrchanges the
two Euclidean structures, Then one can regard the two n e u t ~ n osources as ob-
tained by prclljeetion. from one general source and the complete Eucfidean
vacuum ampli%udeis red, being (2-7.97) without the factor +(l 2a5>.

2-8 PARTICLES OF fNTEGER 4- 3 SPIN

One can d e s c ~ b espin particl~sby combining the four-veclor treatment of
unit pia with the four-component spinor mpect of spin +. The resulting vector-
spinor source qf(z) has l 6 eomponents, apart from additional charge multi-
plicity. ?'fie reduction of this larger system fo Lhe one of inkrest is partly
produced by the projection. matrices approp~sdeto the constituents; pp,(p) for
spin l, m - ~p for pin 4. On, eonside~ngthe murce coupling wssoeiated with
~ingle-padicleexchange, in the rwf frsme of the particle, this procedure s u p
plies the effective source +(l +
T @ ) ~whieh
~, hrw six eomponents. The final
reduction to the four components characteristic of spin # is accompfislted by
fhe projection mabix (2-?.Q, with I = 1, j =" g. In the conkxt of three-
component vectors and two-component spinors it is represenkd by

The projector character of this matrix is equivalent ta the properl;y

and its specifie identification is confirmed by evaluating the trace over the six-
dimensional space. The resulting form of the source eouplimg in the rest frame is

where
rl:*(1 + r0)(6k& W *@k@1)'11 = q:B(1 + 7')~kt (2-8.3)

"ii rc ~rlk &Q~@ZVZ (2-8.4)

obeys
Qk-?k = 0, (2-8.5)
evhich make8 explicit the rejection of the spin & composite s p k m ,
The remoml of the rmt frame specification is facilitatd by wfiting

and the straightfomard pneralizittion of (2-8.3) (war%from a &tor of 2m) is

128 Sour~et Chap. 2

The second term is given a somewhat simpler form by noting, successively,

+
The first rearrangement restates the commutativity of 1 ra with e r ; the
second invoke8 the prope&ie~of and the Iwt uses %heprewnee of the factor
+-
m -I-. ? p to substitute m for rp. An sltc?mat;ive form replaces % (I/m)p,
with --(ilm)$hpk. The implied expression for the vacuum amplitude

1vriCten in four4imensional. momentum spaw for conciseness, iis given by

The kernel af this quadratic form is antisymmetrical, under transposition of the

mat^ and vector indices combined wifh pp --* - p p , which dernazxlb a eor-
responding smticomnnutativity of the sources or F. D, statistics.
To identify psrticlet emission and absorption mwces, ~pffieihcsllythose re-
ferring Lo the four heXicity staks msociated with % given momentum, we con-
sider a causal arrangement and marnine the coupling term in i W :

in which we bsve returned to the vemion given in (2-g.?), and u& $he dya&c
construction for gtr,(p). The introduction of the dyadic spinor realization for
(m - r p ) / 2 m gives the form

which ~ervesto idenkify the four eigc3nvectom g$&,X =. 3, , . . , -8, In order to

exhibit them explicitly m use the follo~~ingproprty,
\vhich is to be undembod irl the limited conkxt, tr = & X , This formula is
easily checked in the rest frame, where the lefbhand side reduees to et .v@ -
Its genera! validity is mured, acearding to (2-6-86), if

It will be rrzcopiet3d that this is equivalent to the constru~tion(2-6*53),with

the a matrices replaced by the algebraicslfly indistinguishable set i ~ ~ a ,
The explicit, forms of the helicity Xabefed eigenvectors are

which are standard combinations of states for unit and anwiar momentum,
Their orthonormality propdies arc given by

The resulting souree identification is

to which a ehafge lahf can, be added. The complex conjugation pmprties of

the eonstifuents supply the relation
= (-l)ta/z~+'ir,u;h. (2-8. l S)
The TCP substihtion
q"(p) + ~JV~(-P) (2-8.29)
therefore: inducm
VPX i(--f )
(812)-X
,* ?.ph -+ -i(- l ) t 3 1 f ) - X q p - ~ , (2-8.20)
which can be suppXemenM by a charge index, in the usual way,
The treatment of dess spin 8 pafti~lesgenerafly fol1cbtc.s the unit spin
pattern. On writing
ars"(x) = mq(x),
one recopizes that, a m -+ 0, helieity rt=# decouples from hetiGity &g, which
is represented by the spin source, ?(2) - +i~,q"(z). Mmsless particles of
helieity =tr# are der;cribd by
(dg)(dx') (-7 X(l/i)ax)D+(%- z')
?'(z)r0[scu
- *% ( - ~ ~ ( 1 / i l a h ) ~ + (-z Z~)?~]~@(Z'),
where
a&q"(z) = 0.
1343 Sources Chap 2

The coupling term in z"W for tt causal alrrangement is

One ewn replace evevwhere by jmt the two krms of the dyadie that refer
to helicity k X :

SlnGe
PP.Y(P)= a, m 0. (2-8.26)
We also use
rO(-rp)= 2p0&(1 -- irsa p/Ip
= 2~"
*
@~,~eup,t, (2-8.27)
U'=&
where
--. @ " ) . U ~ ~ P = (a6-E- Q ' ) u ~ ~ ~ (2-8.28)
and the algebraic ps~perties

The out~omeis the replaeemexlt of (2-&.g) with

in which +(l. -+ XB) ~eleCLsonly the staks of helicity &#. The two m d e fun*
tionis are
= ei+l@~k, (2-8.3 I)
and the corresponding sources are given by

As in the spin fj- n e u t ~ n odisewsion, one can introduce an. rtdditioml decorn-
position in which helieity is coupled to charge in a unique way.
Preparatory to generafi~ingthis approach to all parliclw of spin s n -f- 9, =I..

n = f , 2, . . . , we return to the r e ~frame

t spin projection
Vk == Ptk $@k@l171 (2-8.33)
and remark that
ilk S @ p n t p . ~ ~ @ ~ ~ t , (2-8.34)
where
a k p , t l l -= *(8kt8,, -4- akrl8ta) m "=.kp6tQ (2-8.35)
is the rest frame vemion of the n = 2 projmtion tensor that is defined generally
2-8 Pattieies of intsgsr + 5 spin 'l35

by (2-5.79). The properties of this tensor assure that

Here is the generdization of this resdframe treatment to symmetric tensar-

spinor sources :

Although it is evident that the i j k , ...kR are traceless, this property can be re-
garded as a consequence of
@klBkl.-.k, =z 0, (2-8.38)
according to
0 2= @kflk,qk,kzkg*.k, == qkkks..sk,. (2-8.39)
I<eeping in mind the restriction to tu-o-component spinors, we see that the
count of independent components is

+
consistent with the description of spin s == n ij. The numerical ftletor in
(2-8.37) ean be derived by noting that the latter should be an identity if qr,..,~~
is replaced by ql, ...l . I n that circumstance the projection tensor reduces to
+
the symmetrized unit matrix appropriate to rz 1 indices. There are two
classes of terms; those that select p .= q, which are R! in number, and those
with p = L j , q = ki, being the remaining %(R!) terms. The firs6 set is mufti-
plied by ( @ p ) 2 = 3, and the second set by 2, since

and olfir,...r. = 0. Accordingly,

I
apnkl,..knp,~I...2,ct@q~ll...l, f3(n!) -4- 2n(n!)]qk
(n 4-l>!

as stated in (2-8.37). Alternativeiy, one ean verify that the trace of the pro-
jection matrix that is defined on the spaee of n three-eamponent vector8 and
two-component spinors h a the required value of 2(n +) 1 : -+ +

since the trace of the projection matrix that refers to 7t + I thme-vector indices
+ +
equals 2(n 1) I.
132 Sour~aa Chap. 2

+
A particle of spin s = n $- can be describd by the symmtricd tensor-
spinor source $ 1 . " ' e ( 2 ) . The four-dimensional momentum space version of W i.s

This is not t o be taken fiterally, however, far an wlgebraie simplifiestion should

be pedormed hfore the space-time extrapotation embodied in the fourdimen-
-+
sional vemion is carried out, In the i n i t s cau~alsituation, ~vhere( m rp)/2m
is a projection rnrttrix decling ?p .= m, two powers of the momenlum, appearing
in the form ?'p.(m + rp)phrh, can be removed. Thus, the matrix polynomial
+
in p thffiL occurs in (2-8.44) is of degree 2% I = 2s. That ig iifluslrakd for
n =. l, s = #, by (2-8.10).
f n. the dbect applickttion of (2-8.4) Lo a, c a m 1 arrangement, the introdue-
tion of the dyadic construetiom for the spinor and tensor projection matrices
must supply the tensor-spinor dyadic

(2-8.45)
where

is consisknk ~k-ith,the proprties of this structure. It is not diEcuft to pick out

the bmm of highest hejicity, X = s,

which apwwrs on the Iefbhand sidti?of (2-8.45)with the coegeknt

The other heliaity functions are produced nnos"cimp1y fmm this one, by rota-
tion, ss effectively rewliad in. the algebraic construction

where $,&(E) is defined a8 in (2-5.87) but with n replaced by s, The rwults for
= 8 that are &vent irr (2-8.15) are immdiakly repmduced in this w8y. The
i;:
2-43 Particles of integer + f spin 133

sourees for the helicity labeled stzt;hs of thew F. I). psdieres arts identified as

T o close this seetisn, we consider the masslms particle8 of heiieity f (a +), +

n = X, 2 . . . (although no example comes to mind). It seem evident that the
mecewary tensor-spinor restriction

must be accompanied by the corresponding projection tensor, which is de-

scribed by (2-5.104), and inded the general form is

one verifiw Ghat (2-8.22) is reproduced, while becoming aware of the egtrivalmt
form

The coupling hrnn in 2"W for a causal arrangment is obtained from (2-8.52) as

where the introduction of the new projection tensor, defined in (2-5.106), is

justified by the properties

The dyadie eonstruetion (2-5. f f 7), combined with (2-5.1.21), conveds the
tensor-matrix of (2-8.55) into (the tensor indices are raised, for clarity)

where, utilizing (2-8,291 and (2-8,27),

134 Saurcm Chap, 2

The f m b r 4(1 f a') locks the spin 4 hdicity to the olfiw, and we re-co@;nige
the genemEsaLion oE (s8.31) for s = n 4 : +
vy.-.v,
= epkn (2-8.59)
with the w m c b t d wmce definitions
r/z ; P ~ * . . P * ,
$p& ( 2 ~ 'h p ) %p& ~rl...r,(~)$
h = As:

2 4 UNIFICkTIQM OF ALL SPINS AND STAT4STlCS

The proeedure~we have M b w d for descfibing the v&dow spin possibilities
exploit elementav anmlar manentum pmp&ies. The spin s (n = 2,3) . . . ET:

em be csmpoundd Irom n unit spins.. And it s u E c e ~to add s shgle spin of

+
8 to produce the aquence s n +,a == l, 2, . . . . But all a p b possibilities
=-b

c m be con8tmckd by combining the fundamental spin system a sufficient

+
ven for intc3rger i pia, odd for ixlhger % spin, Aceordlngfy,
m replace the tmmr or multivector, expressing the composition of unit spins,
and the r@la.t;etdkmor-spinor, by the; umiver~~1 muf"tispinor t h h i g appropriate
to the compo~itimof a numbr of spin 3 canstituen$~. A muftbpinsr saurce
will be explicitly written as Srl...rm(~), but the indices will often be suppressed.
All connponenL aping are on %hesame footing and %dditioaalsymmetq mquire
men& can be i m p o d on the multbpi-nor. The mast import@ntof them is %he
r e q ~ e m e nof
t tattit1 symmetq :

where . . . ar, any prmutation of 1. . .n. .

The mul%kpimor refers Ito a larger system than desird and prajwtion
m ~ t f i are
~ e req.tl_ired,
~ even ets in the simplest ~ituakiann = 1, s = a,
Indwd,
+
it s ~ m tos use m& spin prajectisn m ~ t ~ con e se ~ pino
h or index to abtain
$he mquird rduetion t;o the phyftieal ~sysbnrof spin s,

for a, aymmet~ealmultispinor. In the re86 frame of a mwive p&icXe, th&

projection matrix ia
?B

where? a desigasbs the spinor index on which the corrmpc>n&ngm t r h acts.

Its @Beeti% .to reduce dbe r a g e of each, spinor index .t.o two values. A symmebrical
funcLion of n twwva1ued indice8 has s nunnbr of intctependen* components
2-8 Unification af all spins and statistics f 35

as anticiggted, The spin vafues obtained in this way are

Only 8= 0 is ~ s u i n g . For that it suffices to consider n === 2 and choose the

An &ntisymmr?trieaI function of a, pair of h c t i v d y two-valud indices has

anly one independent component.
The general expression of these remarks is given by the foIb.eviw vaeuum
amplitude,

where, wing four-dimensionaf momentum space, we have

The h m e l of this quadratic lorm ha9 a definite wsynrrrnetq under matrix tr%ns-
position combined with the substitution p"" + --fl,

Aecorifingly, if the dgebraic propertim of the source are to mahh the symmetv
properlim of the kernel, we must have
n even, s == integer: [S(x), S(&))]= 0, B. E. stsdistics,
(2-9.10)
n odd, s = integer 3- 9: (S(%),&(X')) == 0, F. S). &&%&ties,
which is the general stakemen%of Lhe connection betwrmn spin and ~tstistics.
This proof will be eompjete, however, only when we h w e shmn that any a t
tempL to revem thme natural connections does violence to Lhe completeness of
the multiparticle stabs,
Let a8 eon~ider-the causal srrangement
B(x) = Sl(2) 4- 82(~),
which implim
(o+~o-)~= (O+IO-)'Z exp d w , i ~(p~)' [?'(m - 7p)IaiS2(p)
a
(2-9.12)
Ollr usirtg (2-6.93) far each spinor index, we have

which, in general, must be projected onto the spsoe of symmetrical spinors.

Employing he1iciQ labdeli spin functians, for definiteness, we recognize that
136 Sources Chap. 2

the highest helicity contained in (2-9.13), X = *n, is represented by the

function
n
%P. = n (up+).,
a=l
(2-9.14)
and the whole set is generated by

I n the special situation of the antisymmetrical spinor with n = 2, the single

eigenvector is
up = 2-1'2[(~~+)~(~~-)2 - (uP-)~(uP+)~~ (2-9.16)
The orthonomality of the helicity functions, in the form

is derived from (2-9.15)as

With the definitions

Eq.(2-9.12) becomes

which uses the fact that even functions of the sources are commutative for
either statistics. The causal analysis
= C @+l
(~+lo--)~
l* 1
ns'( n)l~-)~' (2-9.21)
leads to the identifications
2-9 Unification of all spf ns and statisgitiicra 137

where opposite multiplieatian order is used in the two pmductcs. It is only

through the implieit, alvbraic prope&ies of the sources,
B.E.: fSPx,SP3~f=0,
(2-9.23)
F* D.: (&x, &lx?> = o,
that the two statistics are distinguished. In particular, the algebraic property
F. D.: =O (2-9.24)
leads to the ckraehris%icF. I).limitation, n,k = 0, 1.
The two expremions of completeness,

Become, respectively,
1, (2-9.26)
and

Then the single stakment of eomplekness is given by

For a direct computation of / ( o + / o - ) ~

we/ ~
return to (2-9.8) and note that
ccomphx conjugation interchanges S(p) and S ( - p ) whife reversing their muE
tiplkation order. Therefare,

according to the Hermitian nature of each ro(n- rp) matrix. This reality
propefiy gve:s
~r-here,as a. statement about integrals,

indicates the restrietion to +

= f (p2 m2)u2. The t~v\.oterms are inter-
changed by the substitution pp --+----@, under ttihict.1 the integrand of (2-9.31)
remains unaltered. AecordingXy, with pp designating a physical momentum,
p' > 0, n-e get (all this is the four-dimensional momentum space equivalent
of an often repeated space-time computation)

in conformity -with the requirement of completeness.

N o ~ plet us examine how this consistency would be aEechd if we intervened
in (2-9.5) to reverm the natural connection between spin and staki~tics,by in-
jecting an anti?aymmelricaImsttrix

which a c b on an independent index and thus preserves the spin classification

that has been achieved. The identification of multiperticlie s t a b s from rt causal
arrangement proceeds analogously, with the helieity vectors up& exbnded to
u,hql q ==; &I, The result is given by the follotving replacement in (2-9.22):

where the product of the additional p h a ~constants reprodueeg q, I n the

difect consideration of completeness, hotvever, these phme conatants disiap-
pem along lvith the factors of i, and the outcome is just (2-9.29) tt-ikh the q
index added,
(2-9.36)

Turning to the vacuum amplitude itwlf, we obseme thak the realiky property
(2--9.30) persists 1viL1.1 the Hermitian matrix q inserkd, and that matrix t3sur-
vivtls in (2-9"33)to give

The clem contradiction .cvi%h(2-9.36) completes the unified proof of the can-
nection between spin. and statistics.
2-9 Unification of all spins and statistim *r39

The TCP opemtion, is defined for every spin by the substitution

combined with reversing the multiplication order of all sources, The egect of
the substitution on W comes down to the minus sign induced on eseh r@ by
the r, transformation, and thus W is multiplied by (- 1)". The reversd in the
sense of multiplication introduces a plus or minus sim, in accordance with the
st~tistics. Through the connection between spin and statistics, W, (and the
vBeuurn amplibde), is left invariant; under the compie%eTCP operation.
To study the eEect of TCP on individual emis~ion~ n absorption
d sources)
we first notice the generalization of the spin complex conjugatioa property
(2-6.92), which depends ugan the multiplicative eornpasifion of the u,~,

Then we find that

to which a charge index can h tzdded in the h o w n way. The eorrwponding

multipsrticle transformation is

where
= n, -x.
We have been discussing particle rcspectg in which unification is achieved,
the specific nature of the system being implicit in the pa&iculsr value of n,
the number of multhginor indices. But when we turn to the Euelidean pos-
tulate in the context of multispinor sources, %hefundamental diBerence betwen
+
the sttztistics, or between, integer and integer 4 spin, beeome~explicil. The

where p is a n n e t ~ xto be specified, replaces %hekernel of (2-9.8) with

The r@m&trices, which mirror the indefinite Minkowski metric, must be re-
moved in the tran~form&ionto the Euelidean description. This is ~.ecom-
ptished, for n even, by the symmetrical matrh

which is such %ha$

IrK) Souram Chap, 2

while the matrices

we &l1red,antisymmetriettl matrices that obey

in which we eon%hueto designate tfie Lrmsfomed m a t ~ c e sas 7,. When thia

is combined ~ %thehtransformation af momentum integrale,

we get the camespondence

The possibility of producing the transformation (2-9.46) is contingent on

the lefehand side being a symmetrical matrix. For n odd, it is an sntisym-
m e t ~ c a matfix,
l If tbe latter is to reprme_nt the Euclidean metric, it must be
unaltered by Euclidean transformations and is therefore in the nature of a
+
charge matrix q. The Euclidean postulate requires that every integer 4 spin
particle carry a chargelike attribute. Were (2-9.45) applied unaltered with n
odd, we would get pTp = i, which does not eliminste the r@ matrices. The
appmpfia"tedefinifion of p for odd n is

snd now

The tr&n~formedh matrice8 are

W£

which continue to be real, antisymmetrical, snd governed algebrsjcdly by

(2-9.48). Thus the Euelidean correspondence for odd n ia
2-9 Unification of all spins and statistics 141

Incidentally, for n = 1 the connection with the real, symmetrical a, matrices

of (2-6.129) is
a, = -iqT,, (2-9.55)

where these r, matrices are the transformed ones of (2-9.53).

The space-reflection transformation is defined generally by
S(a)=r,S(r), 2O=z0, %=-xr, (2-9.56)
with
n
r8 = ( z t ) (ir3. (2-9.57)
a==l

Some properties of this real matrix are given by

which distinguish integer from integer + 4 spin, and the generally valid

The uniform selection of p' = f l in the rest frame gives the definite parity
(f)P, which is real for integer spin. With n = 2, the alternatives of anti-
symmetrical and symmetrical spinors give the spin-parity properties 0-, I" and
0+, l+,corresponding to the sign option in (2-9.57). Otherwise, with the gen-
eral use of symmetrical spinors, integer spin particles fall into the two sequences
of parity (&)(-l)'.
No rest frame is available for massless particles. In this circumstance, the
kernel of (2-9.12), referring to causal conditions, becomes

Now it is the values of the individual helieity matrices U p/lpf and the assso-
ciated Yg matrices that specify a particle state. For a systematic classification of
almost all helicities, using symmetrical spinor sources, it suffices to identify the
value of every ir6 matrix and thereby of the individual helieity matrices. This
is sccomplished by inserting the following symmetrical real projection matrix:

Then we have
142 Sources Chap, 2

The limitation to a pair of helicity states is confirmed by waluating the trace- of

the lefthsnd side in (2-9.62), for whieh one can use the full $"-dimensional
multispinor space: 4'(1/2')(1/2"-') = 2. The list of all helicities obtained in
this wrty, X = f+, &l,&g, . . . , only lacks X = 0. For that, one can choose
n = 2, replace the r, projection factor by -$(l- i ? , , i ~ ~ and~ )use
, an anti-
symmetrical spinor. The emission and absorption sourees are identified as

Although this discussion applies to n both even and odd, the necessary
+
existence of a charge propedy in the latter situation, of h an integer 9,in-
vites ra further classification in which the helieity is tied to the charge value,
This is produced by replacing (2-9.61) with the symmetrical real projection
matrix (the common =t sign gives two alfernatives)

For a given value of q the trace of the complete projection matrix now equals
4"(X f21Z)(X/2") = 1. Thew tare only two states, labeXed by q = & l , and the
helieity is
== ( ~ ) 4 $ @ , (2-9, M)
where the sign option refers to the alternatives of (2-9.65). Pn each situaion

This treatment is k3ss general than the earlier neutrino discussion for R = 1,
sine@fph%t did not require the msumption. of zero masg.
We &all close this section by examining the connection h t w w n the mufti-
spinor description and the tensor treatment of integer spin particles, in the sim-
plest situation of a second rank spinor Silt. It is convenient to regard the
l a t k r as matrix, and to correspondingly rewrite the structure of W as
2-9 143
Unificaticn of crlt spins and M a t i ~ t f ~ ~

The general e~ntisymmetriealand symmetrical matrix can be prewnted nt,

,
re~pectively
+
2S.(p) = iroS,( p ) -tiy,r0Sz(p) ~ S ~ ~ T @ S # ( P ) ?
(2-9,70)
+
2 ~ . ( p )= rpr@S,(p) ioB'r0SPu(p),
in which the individual matrices are real, As a useful algebraic rearrangement,
\re note that
tr [(-v12 - ~ ~ ) S ( - ~ ) ~- ~ ( r n ] = -(pZ + m2)tr S(--p)TraS(p)70
u rp)S(p)yO
+ m tr ( S ( - P ) ~ ~ @S(p)yOl) [~P,
+ + tr ( [ r ~s ,( - P ) ~ ~ ~ I [ Y P~ ,( ~ 1 7 ~ 1 )
(2-9.71)
\%-here,for the tiro symmetries,

The evaluation is redueed to computing the traces of n&rices formed by

multiplying linear combinations of the Dirae matrices. These 16 mrallriees are
orthogonsl in the senw of the produet defined by the trace. Their normafiza-
Giaxrs v a v in sign \r.ith the Xfermitian or ske~v-Herrnitiannature of %liematrix,
as dictated by the space-time metric, Thus, the algebr&ieproperties of the Y,
imply that
4 tr r,r, = 4 tr YrYt;YuYS = --(firu, (2-9.73)
~vhile
PEA g,rt@rh S r x Q v a *
E (2-9.74)
The results are

with
K($) = 2"'(mS2(z) + a,P(x)), (2-9.78)
 Chap. 2

and

The K and J stmetures are the anticipated onm for spin O and spin X.
There sre additional terms, however, ~vhiehmodify the vacuum smpfitude by
the typicat. factor (S stands for SE,8 2 , S,, S,,)

This is an equivalent description. The additional phase fetctor daes not change
the vtacwm persisknee probability nor does it contribute to the coupling be-
tween sources in a causal arrangement. And it has no implication for the ob-
sem&blewpeets of the energy asociakd wikh a quwi-static source distribution,
for they refer to the effect of relative displacement of two disjoint parts.
Physical eonsideratiom that arc? sensitive to such souree overlap terms can ap-
pear only in the fudl-ter development and specialisation, of the general souree
f ormlzlism,
Far m == 0,unit helicity parlieles should be selmted by inserting the pro-
jection matrix Its aetion upon the second-rank spinor is given by the
matrix trans~riplion

The two terms in the symmetrical spinor of (2-9.70) commute and anticorn-
mub, respeetivefy, with Y ~ .Only the XatLer is retained by the projection nxat~x,
which eEecti-vely sets &(p) egual to zero. As we recognize from (2-9-78), Lhe
divmgence of the veebr source JP(%)then vanishes identically and the photon
deseription is regained, It would not have suffi~edto merely let m -+ O in
(2-9.78), since it is dsa necessary that (I/m)dJF+ O, We have remarked that
$he antisymmetricat spinor should be supplid andogously with a rSprojection
faetor that digem from (2-9.80) in the relative sign of the two Lerms. This
selects terns in S, that commute with rS, which is uniquely the axial vector
contribution of (2-9.70). Nsw, ho~\rever,it is sufficient .t;a set m = 0 in the
eEe~tivesoume (2-9.76). It seems to be a specific property of the second-rank
spinor repremtaticm that the source of massless spin 0 pa&ides acquires the
special form of the divergence of a vector.
 FIELDS

&-3 THE FIELD CONCEW. SPIN O PARTICLES

Sources are intrdueed tia give an idealized description of the creation. and the
dcrtection of particles. But the puwose of this activity is to study the proprties
of the particles, snd this takes place in some region inkmmedictte htti-een the
locations of the terminal acts of creation and debction, Thus one n e d s a con-
venient measurn of the strength of the excithion that is produced in s w i o n
that may be far from its sources. Ttre natural way to obtain such a memure is
by investigating the eflect on a probe or test source that is introduced into the
region of inkrest. Accordingly, con8irleeng spin O particles and their real scalar
sources, a represented by

we examine the efict of adding an additional weak source 6K(x), L t is given by

where

This combination of source and propstgation, function, measuring the eEwt of

preexisting sources on a weak k s t source, is the$eEd of the sources, 1%is defined
in an analogous tvay for any type aE particle, as indicatd by

or, equivalently, by the four-dimensional momentum inkegral

We now recognize that it obeys a simple inhomogeneous differential equdion.

That is mast evident from the second expression since the application of the
differential operator that produces +
m2, when wting on exp[ip(z - X')],
146
146 Fields Chap. 3

cancels the denominator and leaves the four-dimensional delta function

AIternativeIy, one uses Eq. (3-1.5) and notes that

+
since p 2 mZ = 0 in these integrals, while the discontinuity of the time de-
rivative across x0 = zO',

is equivalent to the presence of the four-dimensional delta function in Eq. (3-1.8).

The differential equation that it obeys identifies A+($ - X') m a Green's
+
function of the differential operator --a2 m'. It is the particular solution that
has only positive frequencies for X' > zO' (e-ipO'O, > 0) and only negative
frequencies for xO < xO' (eipO"O).This boundary condition is more simply stated
by considering the associated Euctidean Green's function. The latter obeys the
differential equation
[- + m 2 ~ ~-E23( =~ ~ E ( X- X'), (3-1.11)
where
(d.) 6(x - X') C-, (dx) b(x - x')]g (3-1.12)
or (xq = isa)
( I / i ) 6(2 - X') c-, SE(x - X') (3-1.13)
restates the correspondence

Unlike the Minkowski situation, the two fundamental solutions of the Eudidean
differential equation are sharply distinguished by their asymptotic behavior:
-e*"R. Thus the requirement of boundedness, for X # X', uniquely selects one
solution, the one that is produced automaticaIly by the Fourier integral solu-
tion of (3-1-11],

and A+(x - z') i s recovered by the previously explored procedures. The al-
ternative methods of imposing boundary conditions can also be applied directly
3-1 The field concept. Spin 0 particles 147

to the differential equation that describes the field of an arbitrary source,

Other kinds of fields and Green's functions are introduced on considering

the time cycle description that is associated with an initial vacuum state:
( ~ - I O - ) * ( - ) * ~ ( +It) . is characterized by

Now the test source response is written,

where the minus sign of the second term recalls the opposite sense of time de-
velopment that is involved. The two fields encountered here are

(3-1.19)
Let us examine these fields for the particular situation in which
K(-)(x) = K(+&) = K ( 4 . (3-1.20)
Then

and, on using the relations

> x0': - i ~ ' - - ' (-

~ X'),
(3-1.23)
( X - 2'1,
in which

we see that
+(-)(X> = #(+)(X)= +mt.(x) (3-1.25)
where
0mt.(4= / ( d x ' ) ~ret .(X -x')K(~) (3-1.26)
548 Fields Chap, J

As the last property proclaims, this is a retarded Grwn's function. It is

real, since complex conjugation interchanges the functions and A'-'. The
three Gren's functions A+, A-, bt. refer La the same inhomogeneous differ-
ential equation sinee A'+' are solutions of the homogeneous equation
(--a2 + rn2)h'&'(z- x') = 0,
and therefore
(-a2 + m 2 ) ~ , . t . ( z=) K(z)*
The retarded field of an arbitrary source ean be found by solving tthia equation
\vi%hthe boundary condition that the field be aero prior to the intcjrvexrtion of
the sources. 1%is remarkable that this classic boundary conditbn requires the
device of the closed Lime path for its appearance. fneidentsfly, the f o m awumed
by &Wfar the cimumstanees stated in (3-lam),namely

Is a reminder that MI = O for &-,(S) = lYc+t(z),and this property will persi~t

if the equality of the soureea K(*,(z) i s maintained by the test source.
To return do the general situation Gven in Eq. (3-f.19), iL is seen that thwe
fields obtjty the differenlid equations

The solution is eharaeterized by the following boundary conditions, Before any

saurees are operative, & " ( + f ( ~ ) contabs only negative frequeneie~and +[-,(x)
only posithe frequencies; after all sources have eeased to function, @(+)(2) and
(6(-)(z) are equal. The initial boundary conditions are msde explicit by writing

And, if it is observed that

a+(z--~r)+e-[~-~1)=a,,,.(x-z1)+dsdv,(2-~'), (3-1.33)
3-1 Tha field concept. Spin O partictes lrC9

where

one gets

which m ~ k e expliciG
s the final boundary condition.
Still other kinds af fields md Green" functions appar on replacing the
vacuum state with a general multiparticle state. Rather than USE? any specific
one, we consider a pararnetriaed mixture, as in
C
tn!
(m+l @
- ) K ~Ia ( 1 = exp[iW@(K)l, (3-1 -36)
where

and 19, is an arbitrary time-like vector with > O* OR reviewing the discussion
of ({B)+l{n) J K , pa&icuI~rlyEq. (2-2.46), we recognize that this probabgity
amplitude is linear in eaeh occupation number n,, which is merely =$ace$ by
an average value in (3-1-36), namely :

(3-1-39)
with
A6(2 - g') = 6+(1:- X') f- doP (n8 )8[e'p'z-"t' -+- G-'"~- "'1. (3-1 ,401

By expanding exp[iWg(K)Jinto the form (3-1.36) the individual ({R)+l {E)

can be recovered. The Geld
(3-1.41)
which defined by
(3-1.42)

obeys the same differential equation,

since A@fs-- 2') is anather green'^ function of the diEer~ntiaIoperator

--a2 + m2.
 To marnine tbe boundary eonditianrs that ebwracferke this Green's function,
we? write5

where

are relabd by
ah-'(~ - X@) = AB(4-1(Ze 2). (3-1 .del
On noting that
(%,)a + 1 = gP(np)sp (3- 1.47)
we can as=& the formal connections
z 2') = ~I;)(z
A ~ ' (- - 2' - ip), ab-)(~- g') = a;+'(% - 2' + ia),
(3-1 +48)
and khese are d l combined in

which i s a, real function. In analogy with the dransfarma$ianfrom the Euclidean

to the Minkowski description, if is useEul to introduce an extrapolation to a
real time-like displaeemerrt,
i @ p --+ X@, (3- 1.50)
where
.XQ > 0. (3-1.51)
For the restriekd domain specified by
0 < z0 - zO" .XXTD
or
--X0 < z0 - zO' < 0,
which are united in
/2@ -- zO'/< XO,
the relations (3-1.48) become statements about the propwation function

namely

This is an &=&ion of periodicity, which is the boundaq con&%ionfor t h w

fmnetions,
3-1 Ths f iefd concept. Spirt O plilrti~lsr l81

If one wishes to verify that the periodicity condition. does produce the
de~iredsolution of fhe Green" function diflferential equation, it is convenient ta
adopt the rest frame of the time-like vector X@,with X' = T,and satisfy the
periodicity requirement in z0by using Fourier series while retaining the Fourier
intepal. treatment of the spatial coordinates. That gives the Green" funetion
repreentation

The Fourier series that occurs here is not unfamiliar:

i
12
T
- i 0 -

(pO)z - (211~n/T)Z 2298

}
--
i2)ip0~e-ip'lr0-."/
@(l

-
+

EI ~ ) i ; p o ~ @--(X
1/ 2 ) i p 0 T ipOlrO-so' i
e
IP I ~ ~ O T
(3-1 -58)
and the substitution inverse to (3-1.50) in the rest frame,

followed by removal of the reference to the rest frame, does indeed produes
Aa(z - 3'). The same results are obtained directly from the digerential equrt-
tion for +(z) by imposing the periodicity boundary condition

To extend this discussion. to the time cycle function,

we must dso cornider the function that replaces d,1,j -(2 - 3') :
= z0 > sop:
-ihk--'(X - g'),
Am@(%-- 2" (3-1.62)
zo < 2" : -ia, ( X - %'C').
f +f

The designation th& we have given it exploits the following form& property
of the averaged oecupw;tion numbers,
@,)-a == -((%>a -C- I>,
and thus
(4-1
L'%-@ (z - XI) = --rhk-)(z -- zt), *'-"
-e (x - 57 = -A# (4-1( X - X/).
(3-2.64)
The required generalization of (8- I. 17) is
 Chap. 3

The fields defined by

~ W B ( R , -M,+,>
), = (dg) SK,-)(z)~iat-lI2)
(3-1.66)
are
- zf)K[-, (z'),
(dx')~;-~(l;'(z

(dxf) ~ . a ' ( x - zf)Kt+,(xt).

(3-1.67)
I n the particular situett;ion
K(-I(%)== Kt+,fz) .=: KQx),
t h e ~ efields become
@@r-r(z)
a tS,@rt,(z)= atmt,(z),
since sX1.the eausall relations among the var;ious bc func-diom are hdependexrt
of p, tas is

according eta (3-1.49). Removing the rmt~ation.(3- l,GB), the ssme reason8
o p r s k to produce

One esn also vv~b

or an equivalent expression using A'-' functions.

Ln c;bbinixrg these regullts directly from the diEerentid equatiom

continuity between the tm funetions is required after the sourcm have clt3wsed
operalion. Psior to the aotion of any source, the two fidds are connected by

whieh is the 8pgropriak form of the periodicity eoxrditian.

Some of this diausjsion retains its form1 appearance when charged partieks
are comiderd and de~cribedby a pair of real sourem, sixree the field i~ iseor-
rwpondingly gene;r&lized. But the extension to terminal mdtipar;iele state8 is
most conveniently prformed when complex soure= are used. Aecardiady, we
prewn6 $h& treatment in, 80nne detail. Begnning ~ %thehvseuum smpfitude,
 Ths Cietd eonospt. Spin Q) parclclas 163

(o+lo-)" = exp[iW(K)],
(3-1 35)
( ~ z ) ( ~ L . ~ ) K * ( + ) A--, ( xx')K(zF),

the introductiorl of probe sources defines tlvo fields, according to

They are

One must not be misled by the notatiori and conclude that these fields stre in
complex conjugate relation. That is s correct rts%rtiort about the differential
equations they obey,

but t;hese equations are to be solved with the same boundary coxrditiocls-th&
of ouLgoing waves in time, by ~vhichiu meant positive frequencies in. the futufe
and negative frequencies in the past of the source,
Let us examine the structure of these fields in the two mympbtic time
regions. If the fields are evaluated at a time after the sources have ceased
operation, that causal circumstance is expressed by replacing P+(z - z') witfi
h ( - X ' Thus (z > K suggests the causal arrangement):

and
z > K:

according to the definitions (2-1.72). In the other situation, the evaluation

of fields prior to the functioning of the source, A+(% - S') iis r e p b e d by
i ~ ( - ) ( z- z') :
z < K:

X < K: K *(X')
= (dup)"2e-'Pzi~~+
P
IM Eiefdd Chap, 3

The two eausal evaluation8 of the fields are assseisted with particle emimian and
absorption processes, respectively. They assign the field (h,) 'j'e'~' to an in-
dividual emission act and the field (dw,)v2e-'p' to an individual absorption act.
As in. the interpretation of complex sources these field8 produce definite charge
chsngw. Depending upon the causal situation, +(z) de~cribesenrrit;td positively
charged padieles or absorbed negatively cha-ed padicles, while +*(g) rcpre~enh
emitilt;ed negatively charged patticleg or absorbed positively charged par-t;icle~.
The time cycle vacuum amplitude is

) ( d z ' ) ~ t(z)bL+'
, (z - zf)K,+,(X')

Fielh are defined by

&W(Kf-,, = ,(z) 4- &~,+,(z) +:+,(X)
(dz)[s~:+,(z)4,,
- a ~,(z) +,-,(X)
t - a~,-, +L,(X)].
(g) (3-1
The ernlicit forms are

(3- 11 '84)
and

If is seen that the field struedure already given in Eq. (3-1.19) is duplicated
here, and the earlier discussion can be applied, enlarged by the substitutions
K + K", cf, -+ $*. In particular, when

which implie8 the analogous complex coxrjugate cqurtdions, we have

where the d m retarded fields are complex conjugates since A,,t;.(x - X') is a
real function. One implieation of this property is that any small deviation af
3-1 The f isld eoneapt. Spin O particles 186

W ( K [ - ) ,KC+,)from zero is real,

GW(K(-,, K,+,> = (dx)l (~K:+,(Z>- Q K ; - , ( ~ &.t.(z)
))
+ +,(g) -- GK<-,(X)) +,et
= &W(K,-,,K,+,)*. (3-1 239)
ThairL is a special example of the following statement, rvhich comes immediakly
from the interpretation of the dime cycle vacuum amplitude that is given ex-
plicitly in Eq, (2-2.86):

The replacement of' the vacuum state with a general rnuZdigarticIe s t a k can
be parametrizcd, as in (3-1-36), with tbc tveight funetion

The averaged oceupatioxl numbers are, analogouslly,

and d+(z - x f ) becomes

according to (2-2.59). We note again that this function is not syrnrnetrieskl in

z and x" but there is a symmetry in \\.hich. positive and negative charges are also
interchanged, The latter is accomplished by the parameter transformation
and
a 4 ---a,
das(z X@)E --- z).
ACt--aa(~' (3- 2.94)

The functions &;$'(X - X'), defined as alivays by

Other relations that falllow fmrn
+ 1 = e"'BP(gp+)a~,
+1 = (@P--)agP

,at*} -- 2') = ea~L~',;;)(z

(Z - Z' - $S),
A~;'(z. - z t ) = e-olhb$'(z - 2' + ip).
The periodicity propedy is eomespondingly modified to

h,fi(z -.- X') ) &e~4B(1:- X' - - X)

= e-Qa8(z - 2' +X), (3-1.rw)
Although we shall not; @ve the details, this boundary condition on the propa-ga-
- 5'1, in conjunction with its driaerential ewation, dow
Lion funetrion her&(%
reproduce the original funetion.
The counterpart of the multipiarticle replacement of h+($ - z') with
&,@(S - 2') is
~ - ( x- --t - g)* = h-, me(%- X'). (3-1.1QI)
The time cycle function.

Nob that if we wished to write the last term in the aldernative way that uses
the h'*' function, it is necessary to change a into --a,according to Eq. (3-1.97).
The fields defined in the manner of (3-1.83) are now given by
3-2 Tha field concept. Spin 4 particles 1S7

and

This i~ the structure already premntt3d in (3-1.67) with all d h e l i o n s carrying

the additional index a or -a, which arc, interchrtnged m K -+ K', t$ + +*, By
adding the index ar to the fields, the slakments (3-1.89) and (3-1.71) beome
applic&ble,togelher with the resulb of the substitution K --+ K*, 4 -+4". The
index a can be introduced everywhere. in (3-1,72), buL it must be mpfmed by
--a in A~$)(z- z')~when sources and fields are =signed an asterisk. Finally,
we mfe that $he field boundary condigions to be imposed on the &Rerentid
equt3ctioa~are

3-4 THE FIELD CONCEPT. SPtN 4 PARTICLES

The vacuum amplitude for this system is

where
G+(z - z') = (m - ? @ ( I / ~ ) ~ ~ ) A + (-
Z z
'). (3-2.2)
Let us obeme innmdiately that t h e algebraic prope~iesof the Y" matrices imply :
+ n)C+(z - 2')
(r@(~/i)a, = (-a2 + m2)a+(z - 2') = a($ - g'). (3-2.3)
This identifies G+(% --x') arj. a Green" function of the D i m mat^ diffaential
operator. Aoeording to the stmcture af &+(X --- z ' ) ~it iei the one that obeys out-
going wave time b o u n d ~eondi$ions.
~ The field definition fo be used here is

The presenee of the antisymmetrical matrix ro compensates the anticommuta-

tivity of the probe source Ciq(x) with +(S), which is formed from and i~
of the
same nature as the totally antiaommtr-Gingsaurces r l ( ~ ) ,
as follo5r.s directly from the definition, or by using the n~ltisymmetryproperty

The solution of the Green's function differential equrttion (3-2.3) is given

by the equivalent four4imensional momentum integrals

where has s different scale in the trvo expressions. Alternatively, the Green's
function is constructed from solutions of the homogeneous Dirac equation,
z0 > xO': iC'+'(z - z'),
G+@ - z f ) = z0 < zO': iG'-'(z - z'),

= C
P@@
(X)$,@, W )*TO,
G'-'(~ -- g') =

= -C +,.s(z) '+p.p(z')rO,
PSQ
and [Eq. (2-7.42)]
+p@4(2)= (2mdo,)'izzl,,,e'ps. (3-2.1 l)
A charge label appean since this is a general attribute of spin particles. The
inhomogeneous term of the differential equation (3-2.3) is equivalent to the
time diseontirtui$y

This requirement is obeyed by the explicit momentum integrals of (3-2. IQ),

and is aIw expressed by

which is a statement of completeness for the navefunctions of positive and

negative frequency, $,,(X) and $(,z) *.
3-42 f he f istd cancept, Spin 5 psrtictsa 159

The evduatian of the fields in esussf situations-after the source has ceased
functioning, or prior ta itts introduction-is given by

The x < structure also emerges directly from (3-2.6) as

The field that follows the action s E s source describes the previously emitted
partielea, and associtstes the tvsve function rt,,,(z) vvith an individual emission
act; the field that preeedes the action of a source describes subsequently absorbed
ptthieles and associates the tvave function \t,,(z)* with an individual ab-
sorption aet.
It will be noticed that positive and negatively charged particles have been
given a uniform treatment, That, is becaum we used reat sources, and assimed
the task of selecting la specific charge t a the multicompsnc?xrQIup,, or jjpcrq(s).
This is natumjt, since, u n l i h the spin 0 situation, spin already demzzxzds the
presence of the faur-component U,,. One ean, however, also follow the pmocedure
of praelecting the charge by using complex sources, From the pair of four-
component real. sources ) ) ( X ) ) ?<2) (X), \v"i"~anst~tct
2-312 I ( -1 2 +
1 ( ~*) = 2-112 ( 1 ( 3 ) ( 2 ) i9~2)(2))-
(3-2.17)
Then the vacuum amplitude is represented by

which uses the definition

9(s) = rl(z)"y0,
and fields are defined through
 Chap, 3

They obey the differential equations

+(z)a,T = -a,$(z),
and the obmwation that

To confom ~ t the
h new notation, we mow ~ t c ;

and
apip.
= %;@ro.
We also fib
m~8$

which restabs the property (3-2.7). The fields in the two causal situottions are
obtained W

= E +P@(%)r0iv@@-,
Pip.

with
3-2 The field concept, Spin f particles 181

and

= 22
P@
+Pu(s)iT;u+,

where
P , S:,- =

The specifications of the particle sources follow from the earlier discussion by
identifying $(z) and $(X)?' with the projections of the eightcomponent field
o n h the positive and negative charge space, respectively, To use the fields
#(X:) and $(x) is surely the most familisr and the most pptrlar waty of appfyhg
the Dirac equation. Neverthdess, we regard the asymmetry of the forms
(3-2.30,32), in contrast with (3-2;. 14), as justifying, in general, the employment
of the msl sources and the multiieomponent fields thab are defined in charge and
spin space, rather than the pairs of complex sources and their sssoeiated fields.
The time, cycle vacuurn amplitude

The fields defined by the diflerentiaf expression

6w(~[-), tlf+)) =
are given as

Since G-(z - 2" obeys the same inhamogeneaus diflFerential equation as

@+(z - z r ) , according to [Eq, (2-7.65)]
these fielh ssfjlsfy the digerential equations

An etlte-mative prment;ation of the solution &

I n t r d u c d here are the red Green" functions

1% is evident from these forms t h h , prior to the introduction of %heBOWCW,

+,+,(z) contains only negative frequencires and -,
(s)only positiw frequencies,
and that Jlt+,fs) = $(-,(2) afkr the source8 have ceased operation. Them are
the b o u n d a ~condition8 that accompany the diRerexltial equa%ions(3-2.39).
ZncidexztaXly, ans a glance at, Eqg. (S1.27) and ($1.34) will eoxlfirm, the retarded
and dvanced Greenp@ functions are also con~;;-lrueted
W

f n the ~peehlsi$uation
~f(-j(z)"'-- 8 [ + ) ( ~ X
) 9f~) (3-2-45)
one evidently h=
J.c-1(2) z=z 1L(+j(x) =.;. J/rct.(z>, (3-2,46)
with
(3-2.47")

appearing as that; red solution of the differenbial equa%ion

3-2 Tha f ietd concept. Spin 8 parOZctass 163

whiel.1 vanishes before the sauree earnrss into action, The form of W far smdX
deviations from the situation (3-2.45) is
(&~tt.,(.) - a~,-,(z)) u"$m,.(z) (3-2.49)
When the multiparticle mixture given in (3-1.91)is applied to w F, D.
-stem, where n,, = 0, 1, the a v e r w d occupation numbers are

On referring to Eq. (3-7.45), W see that G+(% - z" iis replaced by

(3-2.51)
and tlsis Green" function. appears in
( d r )(dx')q (X)?'G.& (X - z') q(zf) (3-2,521
to determine
C ((4+I -)'pa&( ( R ) ) = exp[ilva8(s)l. (3-2.53)
Far sirnplicitJI, no para-meter hsls bwn introduced to distinguish tfie w i o u s spin
stales, The funetions defined by

are, explicitly,

and
+ (1 ~ ~ p q ~ a ~ ) *$psq(zf )l?',
+ p ~ g ~ ~ )

(-4-1 ' (-1

[r0Go@ (Z z)lT = -7 0Go@ (Z -- X'), (3-2.56)
\\-hieh expresses the necessary antisymmetry of Y ' G , ~ ( ~- X') under the com-
plete transposition of space-time coodinabs and discrete indices. The relation
leb fields Chap. 3

t~herethe symbol q now indicates &hertntirjymmelricd charge matrix, The cor-

responding Green" function property is
Gafl(2- 5') = -eaQG,B(x - -X)
== -e-aqQa4(.~ - ZI + X). (3-2.59)
If cr is mt equd to Bern, or, more generally, is accommodated in by a co-
ordinab dependent rdefinition, the boundary condition on ag(r
- 2') ~ppf-t~rs
ass a s i w ebange in respan= to a coordinate displacement by X"; this property
might be -lied sntiperiadicity. Concerning the t i m cyele generalization of
d h w rmulLs, we shall only remark that the mulidipartieXe repbcement for
O,(x 3') i s El, -@(z - z'). This statement is equivalent to the relations
---.

-,
(--J"""
-@(S - 2') = -- G:~)(Z - 27, G%'
- 2') = -GoB(4-1
(X - X'),
-@(X

(3-2. (30)
and they follow from the aver&geoccupation number property
(npp)--a -.B 1 - (@pq)a~ (3-2.61)

Spin 1. According to Eq. (2-3.4), unit spin particles of mass m s" 0 am de-
seribed. by

+ (I /m2)a,~p(~)a,(~- x')~EJ'(z')]. (3-3.1)

The consideration of a, test source defines a vector field:

(dzf)A,($ - ~')aa'(z~).
(3-3.3)
The divergence of the vector field is

and this defived scslar field vanishes outside $he region oecupied by the murce.
The differential equation that is infemed from (3-3,3),
3-3 Some other spin values 166

on using the relation (3-3.4). Another version of this differential field equation is
+
a,G'"'(x) m2@(x) = J"(X), (3-3.7)
where
G,v(x) = - G ~ ~ ( x=) a , 4 , ( ~ )- a,+,(x). (3-3.S)
The differential equations that relate the vector field to its source conversely
determine the field when appropriate boundary conditions are added. The
divergence of the vector equation regains the relation (3-3.4), and thereby the
form (3-3.5). The soiution of the latter with the outgoing wave boundary condi-
tion is just our starting point of Eq. (3-3.3). In this and other examples of B. E.
systems, different boundary conditions can also be used, in straightfonvard
generalization of the spin 0 discussion.
Spin 2. Massive particles of spin 2 are described by [Eq. (2-4.20)]

+ (2/m2)asTp'(x)a+(x- xf)aX'T,x(xf)
+ ( l / m 4 ) a f i a , T ~ v ( x ) ~-+ (x')a:a'x~"~(x~)
x
- 5(T(x) - ( l / m 2 ) a , a , T v ( ~ ) )
x A + ( ~ - - x ~ ) ( ~ ( x ~ ) - ( l / m ~ ) a : a ~ ~ ~ ~ ( x(3-3.9)
'))],
in which
T ( x ) = g,,Tpv(x). (3-3.10)
The symmetrical tensor field that is introduced through

6W ( T ) = / ( d x ) ~ T " ' ( X ) ~ , ~ ( , ~ ) (3-3.1 l )

is obtained as

- ( l / m 2 ) a ./ ( d z r ) ~ + ( -
z xt)aA'~,~(x')
+ (l/m4)a,av/(dxr)a+(x - ~ ~ ) a : a : ~ ~ ~ ( d )
- *(g,,. - (1/m2)a,,av)/(d~')A+(z
- x t ) ( ~ ( . r r-) ( ~ / m ~ ) a # i ~ . ~ ( ( z ' ) .
(3-3.12)
The divergence of this tensor field is the vector

which vanishes in source-free regions. That is also true of the scalar field
and of the combination

The digerential equation derivd from Eq. (3-3.12)

On regwi~gthe vecbr and scalar wume combinations by the equivalent field

B % ~ G L u ~thi8
, dserential equation, hc~nnes

Or again, on using the infarmtian supplied by ifs trace,

the latbr can be pn;serrled W

Oflrer vemions of this digerenfilal field eqtrafbla are

Convemly, the diRerenfiaf field equafiow, gupplemenkd by the outgoing

b u n d a v esndition, have the unique solution dven in (S3.12).
The explicit gpacethe relstion betwen field and 8omce ia stjll q ~ man-
b
ageable for spin 2, but bwonnw inerwingly mwieldiy for higher spin valum.
TKm i& to some exbnd by using faur4imemiowl momen%umnobkion, as
we first 3lmtmGe for spin 2. The appropriah ~pe~ialbation of tbe gene& ex-
prwioxl (2-5.95) is
 Some other spin values 167

where
S,W(P) = g,. + (llm2)P,Pv.
Some properties of this tensor are:

The field defined bv

(3-3.27)
appears as
1
b . @ ) = p2 + - ig (P)
[ ~ P K ( P ) ~ V A - &~I*(P)SK~(P)IT"'(P).
(3-3.28)
The derived vector and scalar fields are indicated by
1
=) 2 [ P ~ ~ P A ( -
P'~PU(P P ) fpUgKr
( P ) ]F'(P) (3-3.29)
and
1 1
O(P) = ;;;i[;;;iPKPI- 38~k(P)]T"(P)- (3-3.30)

The algebraic combination

its trace,
( p 2 f m2)+(p) - P%'+~A(P)= - - ~ S ~ ~ ( P ) T " ( P ) , (3-3.32)
and the additional combination

then lead directly to the equation

which is the momentum space transcription of (3-3.17).

Spin 3. I n three examples, of spins 0, 1, 2, the use of the structure indicated

generally by Eq. (2-5.95) has produced fields that obey second-order differentia1
equations, with an inhomogeneous term that is just the source function. This
ceases to be true for higher spin values, in the sense that derivatives of the source
function cannot be completely eliminated. There is, however, the possibility of
restoring the simpler situation by using the freedom to modify W by adding
real terms in which sources are multiplied together a t the same space-time point.
As we have noted in Section 2-9, such contact terms contribute neither to tlte
va~uurnprsisknce probability nor b the coupling of causally disposed sources.
It is only through the additional con~idemtioninvolving the structure of field
equations that a reason for their presence and specific appearance can be ad-
due&. The maintenance of the inharnogene~usfield equation form that we now
regard as standard nlso requires the introduction of certain auxiliary fields,
ivhieb vanish outside the regions oceupkd by sources.
The discussion of spin 3, which is intended to illustrate these remarks, \\-ill
be: facilitaM by first examining the sirnpfer situation of m-less pa&icles. We
have not done this for spins 1 and 2, since the* physically important examples
will receive individud and exknded treatments. h t it be remarked, however,
&at the appropriate field equations are & h i n d merely by settiw m = O in,
my, Eq8. (3i-3.6) and (3-3.19). Also, the fiysically neeessaq soume restrictions,
af vmisbiw divergences, are algebraic consequencm of the field equations, ars we
em recognize from the m -. O limit of Eqs. (3-3.4) and (3-3.15). But more of
thk fater. The massless spin 3 situ%tionis represenkd by [Eq. (2-5.122)I

and the follolving restriction on the sj~rnmetrical-t;hird-rank tensor:

The I&der implies a lack of uniqueness for the field defined by

ssinee any additional term containing p&, p,, or p, ss Erzctors will not eontribuk
in (3-3-38), oiving to the souree restriction (3-3.37). Hence, the genwal form of
the field is

in rvhich the cy~licallyrelated sets of terms are required by the total symmetry
of the knsor +h,,. The new symmetrical tensor h,+)is defermind by the
murce restriction. In order to urn the latkr, we first note thzzit
3-3 Some other @pinvalues 16s

Multiplication of Eq. (3-3.39) with p9then introduces just the combination that
is evltluaLed tts *(#h --- p&+), and we get

An equation for +(p) is produced by combining

p hp pp v@ h p . = ~ P ~ P @ P ~ ~ ~ ~ P (3-3.43)
with
P~P'P~#FV pkp@p'+rp. - p2pk+* + (3-3.44)
namely
(p2)'+ = p2pk+k -- $pkp'pB+hpr. (3-3.45)
The momentum space version of the field equation obeyed by ~ b ~ ~is~ f p )
now obtained as

The construction of +(p) given in (3-3.45) is derived directly from this equation
by multiplication with phpppv. It might seem that we have failed to meet the
objective of providing second-order differential field equations. Three dePiva-
Livers act upon Q, and the scalar field obeys a fourth-order diEerenti%lequa;t;ian.
But notice that the field equs.tion involves only this combination of fields:

The combination given in (3-3.47) is an %cceplablercsdefimition of (lixpv(p),which,

means that +(p) can be trmsformed away. Thus, our final set of field equations
is (3-3.46), with g, = 0,or equivalently,

where dots represent the terms that are genemted by cyclic permutation from
the gven ones. The following is an algebraic consequence of this equation,

it is consistent with the vanishing divergence of the source, but doe8 not imply it.
Xow let us consider a mwsive particle of spin 3, fir& using maltere$ the
description given by Eq. (2-5.95) :
I70 Fiaids Chap. 3

where, according to (2-5.79) with c31 == 5;

nX,.,kt,p.t = g h h ~ j l p , ~ g v- +
v ~ + [ g ~ , g ~ . ~ g r ~gppyhhgg,ppfi
,~ + ~ ~ h ~ ~ , ~(3-3.52)
g ~ ~ h ~ l

in whielt the necessary symmetriaation in h f , p', v' h= xlsL been made explieid.
Some propertim aE this tensor are

4- Phhfigpfrt 4-
PhP,'
7 8 . 8 ~ f

The field defined by

(dP>
-- &sApp(-p)
( 2 ~ ) ~
qh&v(p) (3-3.55)
is obtained as

The presence of the symmetrical source function enforces the symmetrisrttion in

X', p', vf. To avoid doing this explicitly for X, p, v, we intrsduee an auxiliary
sxmmetrical tensor sb', and display the field ss

The combination suggested by the lefehand side of (3-3.46) (with + = 0) is

37-23 Some other spin V ~ B I U171
~S

Quite? apart. from the explicit appearance of numerous derivatives acting on the
source, the eoegeient of Sx does not have the value given. in. Eq, (3-3.46). Thak
is rectified, however, by considering

for then

- sbp[grh'8,ptgpp, -~g,vg~~fgp~w~]~x"'pf + c, (3-3.82)

with

And, from the equation for we get

As a differential equation, (3-3.62) contains third derivatives of the auxiliisv

scalar field 21;. If we reduced this .to second derivatives by regasding the gradient
of the scalar fidd as the appropriate auxiliary field, the meoxrd-order differeatis1
equation that the veetor field obeys eontains second derivatives of the source
function.
This is the background for the addition to W of a specific contact eau-
pling term:

The ~ s o c i a t e dsupplement to the field is given by

21-r. p2 + ?a2 l
4hld.Y
cont. -P~PRP, + gpvpr p8." 2;i;;li ~ r y ~ h x

(3-3.66)
and its contribution to the field equation is
 Chap. 3

where

The addition of these terms to the righGhand side of Eq, (3-3.62) removes the
third der-jivativea of 2, and replacm them by fimt derivatives of @. And %her@
is an additional contribution in (3-3,M), which can be d d e d Lo the lefbhaxld
side as
l
eont ,
=f - P ~ ) P ~ S ~ . (3-3.69)
-2% 2 P 2z+-ZOmflp2(m2

When Z; is replac& by (P in that equation, all source terms disappea~ This

successful realieafion of our proganz is displayed in the pair of diEerentid
equations :

N o b that the malar field effectively vanishes as m -+ Q, and (3-3.4fi) is re-

cover&, When supplemertted by outgoing w a w boundary eondi$ions, %het;
unique solution of the set (3-3.70) is the field 4&,, ~ v e in
n Eq. (3-3.56).

Spin 4. According to Eq. (2-8.10), this system is described by

The field defined by

m4
(dsp)
61'(- p)r0$@(~) (8-3.72)
IS, therefore,

Two derivd fields are

P'$@
1
= 2 ; i b - Yp)p, + *YP(~'YYf p.)lrlV
3-3 Some ather spin values 173

and

which. supply the Iineatr combinations

The urn of either of Lhe identities

+
(m -4- Y P ) ( ~ Y , p,) = (my, - p,) (m - Y p ) ,
(3-3.78)
(my, -C- 4- ?P) == (m - ~p)(m?;- ~n,),
prrsduces the equation

pp,Ip. - S(mr. +
1
(rp 4- m)& = 4- &r, qv +- 3 PP)~?',

(3-3.79)
from which we immediately obtain

Like the field equation, for spin +,this is (the momenlum space e q ~ v a l e nof)
l a
firr~t-orderdigerentiaf,eqtr%%ion ~viththe source appearing m the inhomogeneous
hrm. The solution of the equation under outgoing wave boundaq conditions i~
the field given by (3-3.7'3). As is indicsted by the m -.sl O limit of (3-3-76}, the
necessafy source resl.riction for massless particles,

is an snlgebraic consequence of the m = O field equation. There is a marka ably

compact way of prewrrting Eq, (3-3.80), whieh is reached by commuting ,?'
with (m -- rp) and noting that

With the examples of spins 2,3, and # b&m us we can recognize the possi-
biliky of simple dgt3braic redefinilions of the seurces Lhat preseme the general
stmcturtt, of the field equations, but introduce or modify contaet brms in Lhe
expression for W , Thus, let
which has the inverse

On introducing this redefinition into the expression for W ( T ) ,say Eq. (3-3.241,
the additional g,. terms supply only contact contributions. The explicit state-
men&is

where W(2")has the same funetiond form as W(T). The implied field trans-
formation inferred from

Since field and souree are frtznsformed linearly, and Ioeally in spacetime, the
daerenlisl field equation mainfains its general farm, buL with changed caeE-
eierrts. This is illu~krakdfor a = 4, where

and similarly for the sources, by the diAFerentia1 equation

Xf this is applied to massless particlm, the diRerenlia1 equation demands the

vanishing divergence, not of the source TL,,but of the combination
T;, -- *@&PT'= TpP. (3-3.92)
Concerning spin 3 tw rernark only that; the redefidion hdieated by

urhich maintains the general structure of the fidd equations but & a n p the
eontael germs, cannot be aged, to Iremove the latbr, As we ohmwed, second-
~rderdiffewntial equations lacking in source derivstive~are not obtained if
canttbet term8 are o ~ t ; k d ,
3-3 Some other spin values 176

Returning to spin 9, we note that the lineaf souree transformation

produces contact terms :

The field Lransfomation implirzd by

ean be written as

or
+L = $g + av,rpr(..
where

The transformed version of the field equation (3-3.80) is

(?P $ -- a(pPyY$:-/- -!~-~ ( a '-ma ' ' ~ p ) ~ ' $ := (3-3.100)
in whieh
a' = 3u(a - 1) + 1, 4(3a2- 2a
a" = + 1). (3-3.101)
Eatice that there are just two situations in which

The fimt gives the original field equation, and the second produces

Another simple choice is

e=O, a=*, (3-3.105)
whieh gives
(rp -4- -4- *%Gm - Y ~ ) Y=Yg;.~ (3-3.106)
For m = Q, the field equations imply the vanishing divergenea of a quantity
that is the transformed statement of Lhe ori@nd 8 a u r ~
funetion.
%.
Spin Again, we turn first to the massless situstion in order to get the simplest
indication of the field equation structure. The appropriate specialisation of the
176 Fields

general form given in (2-8.52) is

- f~"(-~)~'~~(-rp)r~4;~9(~)
-- f s ( - - ~ ) ~ ~ ( - ~ p ) s ( p )(3-3.107)
l,
where
?(P) a Q ~ W V ~ ~ ( P ) (3-3. XQS)
and
12,1.l"""(~)
== 0. (%3. 109)
The field defind by
(dp) (3-3 * 110)
&g"'( --P)$&P(P)

is not unique, 8.lnce $he source is restricted by (3-3, tW), ftAs general form is

where the vector-spinor It, is to be dehrmined &rough the ssume reatrictian,

We shall not detaiX the extraction of the field equation, exwpG to rean&rk%h&,
as in the spin 3 diseumion, higher derivatives geem fx, appear but c ~ be
n removed
thraugh a pernnigsible redefiniticm of Ghe field, namely

The final form of the field equafion is

The alge'lzraie eoxlsequenee of %hisequation,

is eonskbnt with the souree restriction, but dws not imply it.
By f o l l a ~ n gthe ins6ruetions given in SecCion 2-8, we a ~ v ade the f o l l o ~ n g
scrum dwripdion for a,masgive. padiele of q i n #:

Le%us also @ t a bhere the dditiorrd contact brms that are req&ed Lo bfing
the field equations into first-order fsrm without source derivatives. They are
derived from

The field equations are

wrhich reduce to (3-3.113) as m --, 0..The diligent r e d e r a k a is deairoui3 of

confirming the statemeal of Eq. (3-3.117) could solve thew field equations in
conjunction with outgoing wave boundary conditions and reproduce the field
that is derived from (3-3.115) and (3-3.116), together with the following
expression for the auxiliary field,

3-4 MULTISPINOR FIELDS

Tht! multispinor description provides a unifid approach do a11 spin values, wirth
the first-order digerential equation af spin set.ting the pattern af field equations.
I n order to reafiae t h i ~sfftndard, ho~sever,contael Germs must be introduced in
d l situations save that of spin +. Consider, for example, tho descriptian of unit
spin thad L provided by $he symmetric spinor of the second r&nk2as con-
tained in.

The appropriate contact term has already been inlrodueed, and the soume
differ8 in normaXiz;%ianfrom that used in (2-9.8), as indieat4 generally by
q(p) = (2m) 2'"-"S(p). (3-4.2)
The field definition is

wnd

Now we observe Ghat

(rzp + n)J.(p)= Z;;E [m - ?'P + m + ~,PI?(P),

I

and addition gives

f*(r";" t-r'2>?1,t-. 4rl.l~)
= ==(P)* (34.Ci)
W d t h n in coordina;tte spme, this is the first-order digerentid field equtttion

Another consequence of the equation, pair (3-4.51,

is also contali~edin (3-4.Q, ss revealed by multiplication with rlp - Y z p

and the use of
Crr ~ > ( ~ 2 ~ 1
2e (3-4.9)
The simple algebraic property just recorded provides izxl e1ementtaw snd
generally useful eantrol over our procedures. The eigenvttlue relEtLZLOnship~,
TIP = f r,p, are invariant statements of the restframe possibilities, r! = f 720,
where the plus si= se1ecl-s the appropriak aubspsee for the dwcfiption. of the
particle, Correr~pondinl;Sty,setting ~~p ==: ? % p in (3-4.6) reduces the l%tterto
+
the form of the spin Dirac equation, while the choice r2p .- -rip effectively
removes the caordinizh de~vativmand. supplies a field %h&vankhes outside
%hesource, The ~trueturegiven in (3-4.1) b c o m e ~mare obvious, for, with
TzP = TIP,

which puts into evid~neethe contact term and rrormli~ationthat are needed
do attain (3-4.8).
The effectiveness of such considerations becomes clearer on turning t a the
nnulti~pharof rank 3, where the choice h p = Yzp = Yap leads to the wad=-
tion of
 Multispinor fields 179

This indicates the proper starting point:

(m - TIP)(^ - 72p)(m- 7 3 ~+
p2 + m2 - ie ) 3m

and displays the field

l
- f(YlP+ 72P+ 7 3 ~ )*(P),(3-4.12)

+(P> = + 3m - f C 7.p] n(p) (3-4.13)

that is defined bv

There are only two general alternatives open to the ?,p. Either they are all
equal, or one of them has a sign opposite to the other two. These situations are
characterized by

and, for example,

An expression that covers all contingencies is

it can be verified directly. Note that the field statement of Eq. (3-4.16)can also
be presented as the unrestricted equation

Then, with the definitions illustrated by

which is antisymmetrical in the indicated pair of indices, we arrive a t the system

of first-order differential equations,
ll&0 FOsids Chap. 3

where the latter illuslrahs the set of three equation8 th& are relatd by eyelic
permu% a%ion.
It is aIso inbre~tingto eliminate the three auxiliary fidda and pm~entthe-
field equation in the form

f i r e the dolls indicah the two analogous exprmianr, p r d u c d by ~ G L permu-

~ c
&atian. To v e ~ that y thist ~ingleequstGion pernits the reeo~%mction of the
orie;inal field, it suBees to examine it in h a situatiorms:

wbieh do inded eantain the results of (3-4.15) ~ n (3-4.16).

d There i~
another
way ta convey (3--4,21), whieh foll~wsfrom the obmrvatioa that

I d ;isJ the smonderder differential equation

or, slbm&tively,

The Iakter may be earnpared with the second-oder di8erent;hl eqamiLio.on for
unit spin :
[ P ~ + , ~ - - - { P ~ + ~ ~ c Y ~ P ) ~ I (ImJ -. =+C.,,, ( ~ ~
and for spin i:
(pZ+. m2)$ = (m - 7p)v.
The dk~usaionof the fou&h-rank multispinor begns wi& the hi,lg&r&ie
~bhrn@rz_t

But now there is ambiguity in giving (Yl p ) 2 s more general inkrpretation; shall
it be --p2, or Q ~ . p ~ @ pIn? fsct, we shall use s ~pecificlinear combination
of the two, 80 ~hogenthat p W e mof moment&in 3/ are held h the minimume
The actual expression is

The following indicates the options availabte to the ?,p :

Y1p = Y2p = Y;@ = Y4p:

The particular structure adopted in (3-4.29) is designed to simplify the field,

for the situation of (3-4.31), by eliminating a possible p2 term. The three
examples of field equations are synthesized in

Another unifying statement, with its obvious generalisations to other index

amangements, is

Using the definitions of antisymmetrical funetions that sre illustrated by

182 Fields Chap. 3

one then wribs %hesyskm of firsborder differential equations:

where the lwL Llvo are repre~entafsi71.eof gets of ~uehequtatiom.

The elimination of $he auxiliary fielda produces the sinhyle (multkomponent)
field equation

kvhere the summation over a l paim of indices is indicated by s rc?preseatative

term. The various dternsktives are iflustraZIed by

which m ~ t a kthe field expreaiow of Eqs. (3-4.30,31,32), Another form of the

equation is

in whieh $he last sunnmafiion is extended over $istinet pairs, ar < 6,"a p',
a tit: art6 @ p', with ~ 1 repetilious
0 counting. This is equivalenf to s faurt;hh
order diR6rendial equaf ion,
Turning ts fifth-rank multispinors, we first note that
Ambigfuitieshave bee-n msolved in stating the field as

'The 8vaila;ble options are indicated by

The following art? generaEly valid statements:

where repetitioue counting of double pairs is avoided, and

in which the stmnmations m%rked X' are extendd over all index values other
.than a and 8.
The auxiliary fields
and, for example,

then lead to the field equations

To complete the system we need a. differential equation for the + g a ~ I I . ~relating

~~lr
them to the $rap1. Now

while

snd thus no immediate connection exists, owing to the inc?:s@ap&blefact that

+g h. It suggests, however, the introduction of another set of auxiliary
fields, illwtrated by
X
Xr23jg4~11= f 12m2) rlpg(r2p - yap)*(y4~--- ?BP)?,
p (3-&52)
which obey
(TIP -k Yl~$ladSlfla'~"
5m)~[a@l[a"~~ 0, (3-4.53)
aad enable one to write

The full system of firstorder differential equstions is given by Eqs. (3-4.49),

(3-4.53), and (3-4.54). The auxilisry fields can be eliminated and a single equa-
tion for J/ constructed in various equivalent forms, but they have heorne too
p n d e r o u ~to be worth recording,
These procedures ean be extended to higher-rank spinors. Without exhibib
ing the solution of the general problem, we do want to incorporate all available
results into the larger framework approprittte to a multispinor of rank n. Gen-
eralized from Eqs. (3-4.6), (3-4.20), (3-4.36), and (5-4.49),the first two equa-
tions of the system are

C'
a'<B6
+(Y~'P = 0,
Y@'~)$[aflj[a~@~l
(3-455)
which imply

The next in the sequence, ~vhicfnis as far s s it shall be developed here, are

where the k r m s left unwritten are those referring to three index pairs, This set
has the property that

'~,t'hichis appliceble even to n = 3.

The totollly symmetric muitispinor of rank n == 1, 2, 3, . . . gives a descrip-
.
tion for particles of spin s =. 412 = $ , l ,#, . . . It has alfeady b e n noted t h a t
the list is eompleti-ed with s = 0 on considering n = 2 and replacing the gym-
melt~calspinor by at1 antisymmetrical one. AI1 the equations for n = 2 continue
to apply with tbc ehaxlged symmetry, But this is a, completely pneraf, remark.
The system of equations have been diricwsing involves operations on the
fields that are entirely symmetrical among the apinor indices. Aecodinhr;ly, t h e
186 FIatds Chap. 3

field will acquire whatever symmetry characteristics the souree 72 posssesSct)s.

The significant description of the particle is given in the subspaee Y! = * 0
referring to the particle rest frame. The specific value of the particle spin
wit1 therefore bf? determind by the symmetry that is common to the source
and the field. If there is total symmetry, the spin is s ==.: *E; if there is one
arrtisymmetried index p&, and tatal syrnmetfy among the rest, the spixr is
s = +(n -- 2) = In -- 1; and so forth. By appropriately choosing the sym-
metry, then, the third-rank spinor equations can be applied Lo s =I: -# or 3; the
fou~h-rankequations to a =. 2, 1, or 0; and so on.

A syrnbolie transcription of our dis~ussionis contained in the following set of

equations :
(3-5.1)

That is, beginning with the quadr&ic source expression for W, ~vherer is the
approprirzk repregentation of the metric, fields X are defined through the con-
sideration of an in6n;itmimal test souree, The noalacal space-lime relation
between field and source that is conveyed by G is then converted into a local
differential one, which is symbolized by the operator F. Alternative expressions
for W are

and

Of particular importance is the linear combination

(Srx - gxrFx),
To this point, the field X has been a derived quantity, a eorrvenient short-
hand for M, It now acquires independent status, in the fallowing ~ense,Forego
the knowletfge of any connection bedtvwn X and S, and subject them to inde-
pendent vetriation in (3-5.7). This gives

But 6W is ~ ~ S Y and
X , the additional term should vanish. Indeed it does, since
I"x = S. This means that, consider& as rz funetions1 of X for pwscribd S, the
expression (3-5.7) has vanishing first variations or is stationary at the unique
field eonfigumtion selected by the field equations, in conjunction. with Boundary.
conditions, The quantity W is thereby invested with the adtributes of action,
producing the field equations through the principle of stationary action, We
hall no\%v
iillustrate these general remarh in the context of specific spin values.
Spin 6. The field equation is

and the action expression. (3-5.7) reads

There is a more symmetrical form that contains only first derjvatives of the
field. No addit;ional surface integral term is assigned to the partial inkgration.
This can best be appreciated with the aid of the assoeiakd Euelidertn description
where fields decrease exponentially at large distances from the source, sinee that
is the characteristic of AE(x - X", m 7C O. Even for massless pa&ieles the
(z - behavior is sufficient to suppress infinitely remote surface integral
eorttributions. Accordingly, rre ~vrite

urhere the field dependent quantity

g (4(z)) = -&[ae$(z)a~+(z)4- @(@(z))
l2 (3-5.12)
will be called the Lapange funclion of the sytstem.
We shall note here that the field equations can a180 be presented m the first-
order set :

The elirninalion of the vector field 4, gives the second-order dieerential equa%ion
(-8' + ?n2)&(z)= K ( r ) - a,Kp(z), (3-5.1 4)
wyhich exhibits the same kind of efieetive sealrtr source that Etas already been
encountered in. Eq. (2-9-76)- Tbe corresponding action expression, is

f f the first equation of (3-5.13) is regarded as a definition, (6, loses its independent
position and we recover the action %pression (3-5. XI), ~viththe effective scaler
source indicated in (3-5-14), and the additional contact term - ~ ( d z ) + l < ' ~ , .
188 Fields Chap. 3

Spin 1. The field equation

leads to the action expression

with the Lagrange function

To obtain i t one uses the rearrangements

and

But the latter could have been

and this produces a different Lagrange function:

the restriction to first derivatives does not assure the uniqueness of the Lagrange
function. Generally, two Lagrange functions that refer to the same system are
connected by a relation of the form

since the local divergence term does not contribute to the integrated action
expression. I n this situation,

converts (3-5.23) into (3-5.19).

A system of first-order equations is
a,$, - a,4, - G,, = nr,, = -nr,,, a,G"' + m2+w= J ~ . (3-5.26)
They imply the action

with the Lagrange function

d: = -3Gpc(a,+, - ad,,) + +(3GrVG,,,- m2g4,,). (3-5.28)
On regarding G,, as defined by the first equation of (3-5.26) we recover (3-5.19)
with the effective vector source J' + d,MM', analogous to (2-9.78), and the
added contact term - ~ ( & ) & i l i p v ~ f p v .
Spin 2, The fieId equations are

and om possibility for the Lagrangc? function, in the aetiorr

3- $1,
(dz)ITfiV~pv (3-5.30)

with f&e option of replacing the last tern by

or by any tveighbd average of the; two.

Fimf-order difirentiliaf equations ean be iutrodueed in two different ways,
One such see is associated with the action expremion

where K&,, ancl G&,,are antkymmetrieal in X and v, and

Gx = GXyP. (3-5.35)
These field equations are
- apG, + ma(+pp
ahcpPh + +g,.+) = F., - (3-5.361
with the left side uxldergtood to be symmettiged in p and v, and

The effective source of the lattw equalion:

Kip, == *(Kx,, + Kgkv -- KgFx K ~ s ,-4-
, Rpgk,),
K& = KhFp, (3-5,s)
incileate~the rearrangements that are =quire& to produce the form (3-5.37)
from Lhat ykM& by the stationary acLion principle. 1%is also uwfui to remark,
in conneetian with (3-5.36), that

If iS demoted to a dc?rived qu~ntity,one abtains a Lwange funetion %h&$

is an equally weightt3d average of the tvvo possibilities d e ~ c h b din Eqs, (3-5.31,
+
32). Playing the role of source in this aetion is T,. ahR;.h, where symmetrisa-
tisn in p and Y h understood, and there is an added con_tatedkm.
 The se~ondsystem of equalions is represented by

"
where H,,r is symmetrical in p and v , ~vhile
H H & = X"',. (3-5.42)
The field ewatians are
ahH,,h - avHL+ m2(+,v+ +gr&) = Tpv-- +g,.T, (3-5.43)
vvith the left side (specifically aVHJ syrnrnetrized in p and v, md
afl+rvf a&hB m W H,v&= L:.,, (&5*4)
in which

When &,h ia considered to be defined by (3-5.M),the Lagfange funetion turm

into the one that uses the term given in (3-5.32). The effective sour@@
is

and there are EldditionaiX,contact terms.

Spin 3. It b s been. seen that diRerent vemions of +he Lagrange fuxrctian. ean be
introduced, We shall be content, however, ta record only one in this rejatively
complicated situation. It is
(3-5.47)

where the amiliaq function is to rweive indepndent variation in Lhe sta-

tionary. action principle.

Spin t. Little need be said here. The field equation

trP(r/i)a,4- mIJ/-(s)= v @ ) (3-5.49)
implies the action
(3-5.50)
3-45 Action t91

with the Lapange funetiorr

There is an apparent asbitr~riness,for the derivative coufd bi? transferred from

the riglilthand field to the left-hand field, with a minus sign. "She t ~ v oalterns-
tives are identied, however,

since the rorpare symmetrical matrices and +(z) anticommutes with a,$(t).
Spin 4. Vector-spinor field equations are
lyA(l/i)ah4- m]$,#- (l/i)a&yp$P- T,(l/i)ap+P 4- ?,[R% - ~ ~ ( l / % ] a k ] r ~=$ @
V@,
(3-5.53)
and they are embodied in the actiorl
(dz)[9'r0$, + $1
with the Lagrange funetion

The second and third terns effect an explicit symmetrization between the appli-
t to the left,
cation of the derivatives to the ~ g h and

This is _automar;Cicfor the first and fast k r m s of (3-5.55).

Spin 8. The symmetrical tensor-spinor field provides the following aetion
description :
(dz)I.l"r"+#. + C], (3-5.57)
c = -+($@yro[r"(l/i)ak + m]+r. - 2vyr0((l/i)8,rk#hv
- 2#ppr0rp(l/i)a"k. + 2pvror,[m - ru(l/i)d.]th+h.
+ + r @ r ~ ( l / i ) a ' ++~~~* ' v ~ r , ( l l i ) a ~-$ *$ro[rh(l/i)ah + m]$)
+ am(#rOq - *rO+) - f*rO[rA(l/i)ah - 3m]!P. (3-5.58)
The auxiliary spinor IP is to be varied independently in applying the principle
of s t a t i o n ~ vaction.
Spins Q,*, 1, g, 2, #, ....
Under this heading are collected the multispinor
descriptions, Thus, the following ig applicable to sipins 0 or 1 :
It shaufd not b fargotkn thst these even-rank spinors are commutative quttnti-
ties (B. E. statistics), matching the symmetry of ?!?g and the antisymmetry
of Y : Y ~ ( Y ~f 79. The second-order differential equation (3-4.26) can also be
adopted as the bmis of an &ion princble. Let us write it as

Then we comtmct the aetion

where the fachr 112m is supplied to make the two action expressions direetly
compar@ble;otherwi*, it esn. be abs~crrkdinto a common scale factor for field
~tndsource. The value of the W that is implied by (3-5-62) can be pre~entedm

which a s s e d that the two actions diger in eontent only by a contact term. The
Iatler removes the contact term that was added in Eq. (S4.1). Another such
remark is based on. the eornmutativity of the s y m m e t ~ c dmathx i ~ ~ ~ 2 "
with ?!?g, as well as with the matrices (ror@),. On decomposing $ and into
component8 with the aid of the projection matrices +(l the aetion
expression (3-5.62) completely separates into two independent pshs. Thus,
it is possible to use a reduced form crf the action. principle which contains only
one of these field components and its assaeiaLed source* The latter aetion should
be multiplied by two in order to retain the same scale for sourees and fields,
That the above procedure only changes eontatzt terms is vel.ifid by considering
3-5 Action 193

An analogous description can be introduwd for spin +,incidentally, by

considering the second-order differential equation

and the action

+
(dz)[r(z)7@16(z) C (#(2))1,

Again we observe that

diEerv from the original action only by a contact term. Next, supposc? that one
of the two projection matrices

multiplied by the faetor 2, is introduced into (3-5.66) so that it multiplies YO

everywhere on the right. Then, since

a,consistent projectioxi of field and source onto a subspace has bwn brought
about. The signifieanee of this new aetion is given by

only the contact term has been altered, and the same physi~alsystem is h i n g
+
e r second-order formulation,af spin has any pra&icaf.
described. But ~ ~ h e f hthis
merit will not be diseu~edhere. One remark is in order, however, The r6-
dependent contact term in (3-5.70) i s imaginary (roir,is antisymmetrical and
real) and should be subtracted from the second-order action to: prBerve the!
detailed physical equivalence of the ttvo descriptions, Sinee thia subtractive
term is given contact form through the use of the source v, and not J, it empha-
s the ~eeond-orderfomulatian could not be adopLed m the fundamental
s i ~ ethat
+
spin description.
7M Fields Chap. 3

Spins -4- or m represented by

where
r= 7 = 7:y;yg
:
and

In dditian ta $, the three 8uxiliary functions $[23 . . . are varied independent1y

in the action principle. A. second-oder formulation is provided by

in which
C r:(l/i)a,ln*
c = [m - i$ (3-5.75)
For this situation, we have

where the last term can dsa be wfitkn [cf. Eq. (3-4.19)l

It is of eontaict type sinee the are c~IlfinedtO the interior aE the s a w .

And only the contaet terms lare cbnged if r in Eq, (3-5.74) is evemwhere rrmlti-
plied on the right by onc?!of the mall-iees I =tr: rl[(i~~)~.T h i S~( ? C O ~ ~ - O ] P &formu-
~
lation is physi~alXyequivdent to the first-order dti?h&ption.
We shall bypass the dbcumion of foudh- and fifth-mnk spinors La p r w n t
the action principle formulation for spinors of rank n,a f least in ids earlier stages.
This incompfek Lagrange function, witten vvithout the symmet~sationof
&rivatives that is wed in (%5.73), for emmple, is

All repetitions sre to be rejected in performing the summations over pain, of

indice@.
It is intere~tingto sfudy the explicit connecll;ons between, the multkphor
wtion formalations and those employing tenmm or kngor-~pinam. Only the
to&, seeond-rank PnufLispinors will be cowidered. Sginom of the acond rank
srt? wefutly kreated as m t ~ c e for
s this purpose. Some % m m ~ ~ p t i ofo ws i p
a e a n d eombinations are

snd the action. (3-5.59) is correspondingly rewfitten eils

The symmetricd field and ssource spinors of unit spin me dven general form by
which has ~ f i p p ~ r ebefom,
d in other notation, as the wcond line of Eq, (2-9.70j).
The commutators and trace evaluations sLaM in Section 2-9 apply here and
$ve immdiakfy :
4- +M""%,
(da;)fJ""br, + cl,
(3-5.82)
+
e = -tGp'(a,#r - a.+F) - t+@a'a'c,. ~@@'G,.- +m'##p.
This .is the firsborder form (3-5.27,28), with the derivatives symmetr.rzed in
application; to vtt~-t;orand knsor fields;, Similarly, the 8ntisymrnetricttl
spinam af zero spin are presented rias

and one obtains

-do be compared with Eqs, (3-5.15, 16).

The nnla;trirr Lranscriptioft

implies thad a brm of the matrix 3, that commutes (anticommuties) with i l ~is@
an eigenvector of 1"iiY52 with the eigenvalue -1 ($-l), Both (?igenvalu@&re
reprwented in the expressiom (3-531) and (3-5.83). In applying the second-
orcler aetisn form (3-5.621, it is permissible to use projected vemions af the
field and source, as illustrated for spins X and 0, re~gectirrely,by
+ rryO+,,
~ r n - l / ~= +
2rn-Iizt = r@ro(J, apM,.) (3-5.86)
and
2n-'l2$ = ir9r0@, 2rn-"'t = ir&~'(K -- a,Kr), (3-5.87)
where the f structurefs are the appropriate projections of (3-5.61). With thee
rduced fields and sources understood, the aetion (3-5.62) becamm (a factor
of 2 is supplied)

The trace evaluations give directly the action expressions (3-5.18, 19) and
(3-5.11, 12) with the effective vector and scalar sources, explicit in (3-5.86,87),
th&t have been s t ~ t e dpreviously. One can also mske the opposite choices in
these projections, and we record those action form which are, for spin 1 and 0,
 Actian 19c7

and

The effective sources that appear here are M,. - (1/2mZ)(arJ. -- aJ,) and
K, -- (11m2)d,K, respectively.
Wr; ahall close this section with a few varied comment8, First we recall that,
although we have not illustrated it here, the possibility exists of redefining
sources and fields by linear transformations which change the detailed ~ t m c t u m
of the field equatians and, therefore, of the Lapange. functions. Then it is noted
that all discurnion has been concerlled with the vacuum smplitude (O+~O_)~.
The shift of atbntion to other transformation functions k convittyed by a cfiange
of boundary conditions in the action principk. Let us be mplicit &bout the time
cycle transformation function (o-/o-)'-~'+. Here the action separates into
two snalogous terms, with opposite signs, that are associated wilh the two smsw
of time development:

The accompanying boundary conditions require negative frequencies for X+ and

positive frequencies for X - at times before the sources come into oprstion, and
the equality of X+ and X- sfter the sources have sbut down. The latter has the
following consequence. While the inkgrations in (3-5.91) can 'be regarded m
exbnded over d l space-time, there is eomplele cancellation of those re$ons that
me aubmquerrd to the functioning of the sources. Accordingity, the inbgr~tion,
domain can be made ~lenni-infinite,bounded by any space-like sudace that has
only source-free space in its future* The final comment is concerned 4 t h the
rdationship b t w e m the Lagrangtl funetlong of massive and massless particies.
X t is known that, as m 4 0, the states of a padiele with spin s efe~ompom
~ , &(s - l), . . . . Thus the desc~ptionof all
i n b those of vafious h e i i ~ i t k &g,
m ~ s i e padiclw
~s with helicitiw of magnitude ss, in ideger gteps, ahould be
contained in the a~eountof a rnamive particle .ivith spin 8, "T"hiahas received
some discumion in krms of sources. We m n t to t r a e the mrraponding field
decomposition in two importsnt examples. Let the vector field and souree of a
unit spin particle ivith m= m be expremd by

where the latter is 1Eq. (2-3.44). Then the action (3-5.18, 19) becomes
798 Fields Chap, 3

In the limit m + Q, this action separaks into two pafis:

which, as shall, be discussed in greahr detail laCer, deseribs the photon, and

rtlferring to a massless gatPticb of zero spin (helieitd. Notice that the Lagr&nge
funetion of the spinless particle comes entirely from the mms term of the original
Lagrange function, and would have been overlooked had one merely get m = O
in. (3-5.19).
The other example is spin 2, where we express the tensor field and its
source by
= h,. -- (2-lt2/m)(a,A. + d.A,) + 6 - " 2 [ ( 2 / m 2 ) d p a4-~ @p&], (3-5.96)
and
a,TF9 = m 2 - ' l Z ~ & , alrJ@= m ( 3 1 ' Z ~- 2 - " 2 ~ ) , (3-5.97)
the latkr being Eq. (2-4.21). These structures are such that

When (3-5.96) is inserbd into the actiorz (3-5.30,31), and the limit nz 0
performed, three independent parts are obtained, Two of them restate the unit
and zero hellicity actions, Eqs. (3-5.94) and (3-5.95). The third one is

As we shall also discuss later in more detail, it is the (or a) gravifon action
expression. This time the photon Lagrange function emerges completely from
the mass term of the spin 2 particle, as indicated by

but the scalar field action has contributions from b t h parts af the original
Lagrange function. Xt is an interesting unification of the actions repre~enting
massless particles of various helieities to connect them with one action expression
for a massive particle, Also impXied are the relationships between different spins
necessary to amive ad a common description for a given helieify.
 lnvarianee transformations and fluxss. Ghsrga 199

3-4 IMVARIAFJCE TRANSFORMATIONS AMD FLUXES. CHARGE

The vacuum amplitude (o+/o-)' is unaltered by a rigid translation or rota-
tion of the sources; for charged particles it remiins unchanged by B uni-
v m d phase transformation of the wurces, Physical information is obtained
through the relative tmnsformation of diflferend pa&s of the source distfibutisn.
Relakive tramfation gives information about energy-momentum, relative rota-
tion about angufar momentum, and relative phase displacement .tc?acbesabout
charge. When the source distribution is causally arranged, with the dbjoint
pieces treated a%units, one acquires knowledge of int-egral physical qu%ntities-
b t a l enerm, total charge. At the next stage one considers transformatiam that
vary arbitrarily in space-tinrle, thereby supplying more locdized data about the
various physical quantities. Fields are the instrument far eonve~ringthese data.
To illustrate these remarks, let US consider spinless charged gttrtieles of
mass m, as desel.ibed by the action

where K ( z ) and +(z) are now two-component objects in an appropriate Eu-

elidean charge space, Invariance under the constant phase transformation af
the source, -
K ( z ) = eiY9K(z), (3-6.2)
follows from the existence of the compensating field transformation

since all the Euclidean products in (3-6.1) are unchanged by s common rotation.
Next, led p become an arbitrary funetion of position. For simpfielty we consider
an infinitesimal phase transformatisn and write this generali~ationof (3-6.2) as

The msociated action variation is

But again we can introduce a compensating field transformation,

which leaves K& and $2 unaltered, and fails to keep W invariant only beeause
space-time derivatives now act upon $ @ ( X ) :

Thus, through this method of caIculatian it is found that

2 Fields Chep. 3

where
j@(.) = a"+(.)ig#(.).
The eonsiskney of the two evaluations implies, with the aid of an irrtepation
by pards, that
+
gPjp(z)= (z)iqK(z). (&G, 10)
This is verified directly, on using the field equations. When the righbfrand side
iis zero, which is true in source-frw regions, we recogni~ethe local stakment of a
con~rv&tion law. If the charge matrix Q is diagonrtliged and ~ompfexsources
intrdtreed, the action expression. and %heproprties of jfi became

and
jfi = z"(a"4*4 - 4*a"@), d p j f i= i(@*K- K*+), (3-6.12)
where

~ t hthe
, remind= th& #* is naC the complex conjug%teof 9. S t a b m e n t ~anaEo-
PUB to all thew apply to the vacuum time eyele action with agproprbte afge
braic sims in, 6t.F" Lo indicah fhe sense of time flow.
Using this more general, framework, we now re-examine the causal situation
+
with K = K Z ITz, where the phme of K 2 i~ changed by a eo&ant and that
of XI i s held fixed. For the infixliksimd dransformstion b i n g consider&, we
know that

This wei&ted average of the charge values rewmbXes an. expectation value.
I n d e d it is one, if we consihr the time cyele function, with the phase of K,+,
displaced, a f k r which $he two sourea are identifid with K,

this is an infinitesimal version of f he Xefbhand side of Eq, (2-2.123), for example.

T o apply (3-6.8) to this c a w l situation, we Bepar&&.Kl and K 2by a space-like
surface, which is otherwise! arbitrav, and then make 69(5) vanbh in the futwe
of this su~aec:and be the conrJtant 6p in its past. The step function deriva&ive!,
3-6 Invari~ncetransformation, and ffuxes, Charge 2@9

-a, GP(%), confines integration to the surface and

which is the identification

f n the time cyele degmbtion, inkgrations are exbnded up to s spwelikle

@udaee,which foilows the working of the sources. Then the choice Su?-(x) == 0,
So?"+($) = &pgives

which identifies
dgIr;iIr= (Q):*
In the latter situation jP(z)is computed from the real or mutually conjugatpl
mtarded fields, and is a real function.
f t is evident that j"(z) provides a space-time account of t h distribution
~ and
flow of charge-it is the charge Aux veclor or current vector, We sh&llevaluate
it for a single-pahicle state. On referring to Eqs. (3-1.79-81), it is seen that the
fields in the region between emission source K Z and absorption Bource .Kl, W*
eiakd with a positively chsrged particle of momentum p, are

and if only this ex~itationoccurs,

The source factors identify the emission and abwrption seLs [compare
Eq, (3-6,19)1; the cument associakd with the particle is ZpP clup. We can con-
firm this by verifying that the total eharge is unity. But first it must b recsg
nizled that the uniform value of the current is an idealization, which applies in
the interior of the parfiele beam but fails ss one nears the edge@.Of coume, the
momentum is not s p c i f i d with arbitrary precision, as in (3-6.22), but within
a cell of small but finite dimensions, having invariant measure bw,. Thrta, the
correct description is given by

To compute the tots1 charge one can integrate the eharge density jyx) over all
202 Fields Chap. 3

space a t a given time:

Q(lp+) = /(dx)lLap d% ~ X P ( ~X)P

and
4*(x) -
Similarly, for a negatively charged particle,
(do,) '12e'pzi~zp-, 4(x) = (dwp)112e-ipzi~*
IP-, (3-6.26)

jW(x)= -2pp dwp(iK~p-)(iK2p-), (3-6.27)

with an analogous verification that the total charge of the particle is -1. The
retarded fields of the time cycle description that are associated with a given
momentum, and positive or negative charge, are [cf. Eq. (3-1.84)J

respectively, with their complex conjugates. The corresponding real currents

are then given by
jr(x) = &2pr (3-6.30)

This is the contribution to the current expectation value attributed to a par-

ticular momentum and charge, being the current per particle multiplied by the
expected number of particles of the given type emitted by the source. Inci-
dentally, the consideration of a single momentum should not obscure the
presence of interference terms, in these quadratic field structures, when several
particles of different momenta are present. Such interference terms disappear
on performing the spatial integrations that evaluate the total charge.
The non-conservation equations, (3-6.10) or (3-6.12), connect the particles
and charges observed after the operation of a source to the activity in the in-
terior of the source. Let us test this in the physical circumstance of the time
cycle description, with K(-, = K(+, = K. Integration over a region that
contains the source gives

where the surface integral refers to any space-like surface that is subsequent to
the source region. On any surface that precedes the source, the retarded fields
and the current vanish. The explicit form of the right-hand side is
3-6 lnvslriancs transformations and ftuxes;, Charge 203

(I/i)[bmt.(z --. X') - i?rrCt.(zL z)] ";=- - z" -

(I/i)[b,,t.(z --.

= aft)(z -- 2') -- a(-'(% - (3-6.33)

and

~vhirthis the expected rcsult.

We have not, yet remarked on the ambiguity in the current vector that is
defined in general by (3--6.8). To a particular jP(ctr) vector can be added any
exprcs"siort of the form d,tnp"(x),tvherc nzp" is an antisymmetrical tensor, for

This supplement to the charge density, V n(x), ~vheren k = m'*, adds a two-
dimcnsianal surface ixrtegral to the charge associated with a three-dimensional
volume:
(3-6.36)

Thc calculation of tatnl cllarge is not affeckd, therefore, nor is the value of the
flux vector assigned to a uniform ~ituationsince this is also fixed by total charge
corisidcrstiorrs. Wlxy can one not ignore the ambiguity and just accept the
currexrt expression that is naturally associated with %beLagzlrtge funetion?
One reason is that alternative Lsgrange functions can produce: diaerent currents.
This is illustrated by the unit spin, situation.
Tlre seco~ld-orderL ~ r a n g efunetiorr (3-5.19) and the first-order Lagrange
fune tion (3-5.28) imply, respecdively,

and

In the abrsense of the source M,,, these current expressions are equivzttent. But,
when we use the Lsgrange function ($5.231, there results

where Q1piq+" is indeed an arttisymrnetrieal tensor. The alternative current

expression is sowwhat simpler, since d,rpY vanishes outside sour= regions, and
the tofizl charge can be caleuXated as

Let us apply this to the region betweert tbe %M-o causally sep&raM souree8
.l !,
JS, where the fidd is [Eq. (3-3.3)]

The eigenvecfor propedies

and the narnzaiigation

C*
@pxqe,pxq =1
enBure that the contribution to jp of a speeifie particle state is

The flux per padicte, 2pp hp, imlearfy of univemal appliwvdity. Aa we have
mentioned, it is fixed by Lhe normalization condition, in the manner m d e pre-
c k by (3-6.25).
One might think that the arnbiwiw of current exprmions is an! asp& of
second-order Lapange funclions, wit)l their ogtisns in arrandng $WO cferivatives,
&nd would disskppear if firs&arder Lavange functions m r e adopted. T o dispel
this iXlusian it suffices ta wmidtjr pin 2 charged padicXes, where two 6mbordr;r
fwms art;?avaiXlzble. From the Lagrane function (3-5,s) we obtain,

with the latter form applicable in soureefrtze: regions, whik (3-5.41) gives

The application of the phase transformation procedure to the spin action

(%$.m,51) produce^
jp(,) = &~.(z)y@r~@(z), (%6*47)
bgekher wi&
a,jP(z) = +(z)r0i9?(z), ($8.48)
which is also s consequence of the field equations. The field in the interval
23-43 Invariance transformations and fliuxas. Chargs 266

between the causally separated sources and is [Eq. (3-2.14)]

+(X) = C [@p.p(~)itllp.p
P@@
+ +pop(~)*itl:prgl,

where, it is recalled,
= (2m d~~)"~u,,,e'~',
* 0
UpuqY UP@Q = 1,
The contribution to 3'"" ttsscrcinted with a single-particle s t a b is

which identifies the ffux per particle:

This expected result expresses the evaluation

whieh ha8 %befotlot~ingderivation, bawd an the normalisation of IEq* (3-6.W)

fef, Eq. (2-6.97)j:
O = u;u:7°(rp + m,7') g,,,

= -2pp +2m~~~,r~r~u,.,. (3-6.84)

As an afternative to computing charges by surface integration of jp, one

can consider the volume integrtl of %jP, extended over the region occupied by
qg. The field asociated \~\-ith
the latLer gouree does not contribub l o &his
calculation,

since the mLisymmetrieaX m a t ~ xQ; removes the mateh between the antieom-
mutativity of the sources and the antisymmetry of the kernel ?'G+(% - X').
According1y,

which exhibits the charges p = & l assigned to the sin&@-particlestabs. Al-

though this interpretation is quiLs etear, it may be helm1 Lo give EL mar@formal
e d tfre analogue of the rdsttian (%ti.lS), or
discussion, b ~ ~ upm
 Chap. 3

Ta andyze the lefbhand side :

(3-6.58)
we have only Its note %h&

and the charge vdues are identified as

I L has been entpha~sizdthat the arbitmsinw in m~iminget charge flux

vector is not just the po~sibilityof adding any p3,mpVkrm, butt is inberent in the
existence of akrnative Lagrango funetion dacr+t.ions of the given physied
8ystRm. This is Lme of .spin EM well. A third-rank spinor that is antisymrnetrical
in a, pair of indices give8 a spin 4 description. The current deduced from
(s5.73) is
C
jr = +#~(j ~ g ) q + +C a<@
+L.Y+(Y: - T;Y&~,B~ + 3+[231~~:&~231f + g .
(3-6.61)
Outside all the sauree~the auxilietry fields ?C/raal v a ~ s hand

An appeapriate expremian for the third-rank spinor is ~vritten.as

It is anthymmetriesf in tI and Sz, and the propesties (&6.62), m ilfu~tratedby

with matrix notation regtored for the third spinor index, are satisfied since
frr^(l/i)a,
-l- = 0. (3-6.65)
The current that is defived from the first Lerm of (3-6.61) by inserting (3-6.63) is
3-6 tnvariancs transformations and f Iuxes, Charge 207

With the aid of the identity

m,, = -- -2 -l
9~@"'.~'k
3 2m 2

The o;,term is the only such structure that does not involve coordinate dcrivn-.
tives. Ana;Xogous but more elaborate comparisons can be medc between alterne-
Live descriptions for particles with spins 8, 5, , . . .
The technique of vsriabfe phase transformation. hag been used t o give a
more detailed space-time dewription for the average charge distribution, It
also supplies such information about charge fiuctualions, We shall illustrate this
for spinless particles, confining the discussion to the simplest measure of ffuctua-
tions. Consider, then, the time cycle vacuum amplitude with the sources
K,+,(z) = e " " ~(X > t K(-I(Z~== K(%)? (%6,70)
which is

An infiniteeimal v a ~ a t i o nof the phme constant u, gives

This equation, with cp = 0, has been discussed Now let us differentiate once,
before setting u, = 0,with the consequence that

where

is a charge fluctuation flux vector,

In, evaluating this vector from
j?+,(z) == P4t-r-I(X)~~~P(+I[X), (3-6.76)
one of the 4(+, fields is taken a t (P == 0, and beeornes + r e t , ( ~ ) . For the other we
 Eq, (3-1.19), appropri&teto the vwuurn initial &ak, and obt&in
(S6.77)
with

In k r w of the~efield8 we have

which urn the charge prop&y

= 1.
It is helpful Go decompm +(g) into r e d and im&nary p&&:

whem, weordjing to the mlation

the maf,compnent is

In the circumstances to whieh (3-6.79) refers, which are made explieit by the
apprance af the funetion eret,(g)# the advan& field vani~hw. And the
contfibution ts 4 of $+fat. caned#, XeaGxlg the re81 fom

Since the lstter is a solution of the homogeneous field egustion, one finds that

A oalculstion of the total eharge fiuetustion can be performed by integrating

over the muree :

) C K , , I ~ . (3-6.87)
(dzf)~(z)h'+'(z-- z f ) ~ ( z 8=
(h)
P@
The latter is the expected total number of particles emitted by the source,
and thia fluctuation formula
((Q - (~))')f
= (N+ f ~ -)f (3-6.88)
is contained in the more general statement (2-2.123). In effect, jt,,,, is a partiele
flux vwtor.
 Invariance transformations and fiuxfslo. Mechanical propertisrrr 205)

3-7 INVARIANCE TRAMSFORMIATIIQNS ARID FLUXES.

MECWAMICAL PRBPERTliES
The wLion for spinless particles is invariant under zc rigid source translation,
R(%)= K(%+ X). (3-7.1)
This is expressed, in (3-Gel), by the existence of the compensating field
kransformation
* ( X ) === $(a: -I- X ) . (3-7.2)
The infinitesimal versions of these transfomations are

The following are the proposed eneralizations of tberse expressions when 6XY
becomes an arbitrary function of position, 6 z v ( ( z ,

The distinction between these farms is necessary do maintain the invariance of

the sourec-t t e r n under the eonnpensating sauree and fieid vftriiations:

The response of the Lagrange function to the field variation of (3-7.4) is

only the first term on the right would appear for tz rigid translation. An equive
bnQgresentation is
62 == d,(6zp&) - tPya,&X,, (3-7.7)
where
tPw(z)= aP4(z)av+(s) gpve+ (+(z)) = $"(X). (3-7 -8)
Let us also xlote the relation

mrhere the fins1 form involves the use af the field equation. The chsne;e induced
in the action by the source variation of (3-7.4) is computed alternatively ss
270 FSelbs Chap. 3

and

The compsrisoa of the two versiorls implies t;hat

whicl~can be verified directly. This is the local stahment of a vwtorial con-

servation law when the right-hand side is zero, whieb is true in soume-frm
~giona,
+
In the causal circumstance indicated by K = K Z K Z ,B rigid infinitesimd
displacement of K 2 induces
( ~ + l o - )=~ C (O+I {E) j /o-)~'
+C(o+iIa))Ri[lt i b X * P , j ( { n J / ~ - ) ~ ~(3-7.13)
,
where
P,( (4) = C
P(I
Pp@,,, (3-7.14)
which can be writbn

Similarly, in the time cycle situation the displacement of K(+, and its subsequent
identification with Kt-t = K gives

To compare (3-7.15) \vith (3-7.111, we separate K t and K z by a space-like

surface and let 6%" vanish in the future of this surface and be the eonstant &X'
in its past, That &ves
(3-7.17)
which i~the identification

The analogous equation of the time cycle description is

where ESY(z), kvhi~his computed from the real retarded fields, is also red.
The distribution and Aow of energy a~ldmomentum is described by t"(z).
It is the enerm-momentum flux vector, or stress tensor. Let us evaflxak it for
the state of a single particle, chosen to be neutral, for simplioity. In the region
betgveen the causally separated sources, the field that is associated with at
particle of mometrtum p is
3-7 lnvarian~atransformations rand fiuxaa. Mechanical preperties 211

Xf only this excitation is considered,

which obeys the conservation law i),tM" =; 0. In any consideration involving

time or spsee averages, where the corresponding components of p" set $he
scaIe, the oscillatory terms of (3-7.21) can be ignored, The firs%tc3mm exhibits
the anticipated ftactars: the representation of the emission and abmrption acts
by iK2,, iK:,, the particle flux faetor 2pp dw,, and the measure of the quantity
being transported-the energy-momenturn vector p". In the analogous d i e
cussion using retarded fields, we have

The disappearance of all interference terms through integration can be

vePified, in the causal sifuation for example, by using Eq. (3-7.12) to obtain

where, in the region occupied by the emission source Kzl

The first parL of Lhe: field, associated with KZibelf, does not corrlribute:

since the gradient of the symmetrical function &+(X -- z F )is tan antisymmetrical
function. Accordingly, with a slight rearrangement we get
(dz)(dz')iK l (s)(l/ i ) a ' ~ ' (~X ' -- zt)iK2(z')

which exhibits %bephysical property p-hat is carried by individual particles

between the initial emig~ionand the final abssqtion acts. Perhaps we should
ktho note that
in1 )KliKtpi~zp(inl = + 1,)
@+I i% i l)({% l,]+ Io-)~',
f 3-7.28)
which shows how (3-7.18) is used to produce the general enerm-momentum
252 Fields Chap. 3

evatjtutltion (for neutral particles)

The analogue of (3-1.27) for the time cycle description is the expetation value

Enerw-mame~tumAuctu%tionscan aXm be given spacetime localization, in

for charge, but we shall not discuss the details here.
Spinlem pta&icles have Ptn al6emative characterization in the arcfian (3-5.15),
whiah scalar and veedar fields and source^. The derivdives of the field
variation (3-7.4),
a(d,Qt) azY+v(afiqb)-4- (a@@)d,cfix', (3-7.31)
provide a model. for the response of vector fief& to coordinab-dependent
dispfacements:
b#,(z) m ~ ~ v ( ~ ) ~ ,4,-+ B ( ~ ) &"(g)* (S7.32)
The eamegponding muree variation is that required to maintain the invaritance
of i ( d z ) ~ f i +namely
~,

Notice Ghat the distinction between the two forms die~ppearsfor a rigid rota-
tion, where
(3, 62, = ---a, = SW&@. (~7.a)
The reBpome of the Lsrsnge funetion

is
+ ace#. -
e = --+Pad (~7.35)

616 = Zixvdpd:-#- (p+'- dic"dYQt- cfiYa'tb)ah 6%

=.= a,(b~"&)- tc""d, (3-7.36)
with
pp +
vaV+ ---- f14V+ = tPp, (3-7.37)
%nd
(3-7.38)
We aIm have
+ ~K ~ ( ~ -
a,pv -.. -KaY# J, ( a , ~ @ )3- F ay+p),
+ ~ (3-7.39)
which e m be given oLher forms with the aid of %hefield equstionfs. The only
fieldaependent term on the rlghbhartd side can be w ~ t h ss
n -av@, nultiplied
by the effective source K - 8,KP. In the absence of the veetoriaf so- K,,
the two vensiom af the ~ t r e mbnsor coincide,
The Lapange functions (3-5.19,28) for unit spin ps&ielm contain vector
field@,and their derivativw in the cud combination. Using %heh t b r ss the,
3-7 Invariance transformations and fluxes. Mechanicat properties 213

model for an antisymmetrical .tensor field GFv,we infer by dihrentiation of

fS7.32) that

When tensor field sources are introduced, their infinikgimal variation is

But fimt we consider

= -&cppc,.- *m2@+,, G.= a,& - a.+@, (3-7.42)
which gives
6& = 6zv4,d: - GFrGvhd, 62, - m2+p$va, 8%.
- tpva, SzV
= iaV(6xVdG) (3-7.43)
and
(3-7.44)
Here
1'' = G@"'~ + m'+@< + grv&
==: IF@, (3-7.45)
and we note &hat
l = Gp'GPv-4- + 46:
= -m2$'+),. (3-7.46)
The direct evaluation af the source vagation efifect i s

and the eompa~sorrs h a ~that

?~
+
apt@"= J,GMY ( a ) , J @ ) t ~ ~ ,
or, more immediately,
a,tr""= ---Jl"av+, 4- ar"(JP9"),
where the t a t term does not eontributc:to the volume inkepakions %hatevaluate
the total energy and momentum emitted by the source,
The alternative Lagrange function (3-5,23) digem from (3-5.19) by the
divergence of a vector [Eqs. (3-5.%, 2511, and Chat remains true of the variations
6C. Such additiond terms da not contribute to &W, But let us note that the
relsfion
(dx)dh[f (%)aFd z v ]

with the last b r m annulled by the re~trietian

implies3 arm arbitra~newin the stresrs k m r , which is indicaectid by Iheeombina;tion

Thia arbitrarinem &appeam# however, if we in;sigt upon a propeAy whi~h,thus

f%r,has emerg4 automaticaHy-the ~;ymmefryaf the bmor P". Xnt, order that
tfre 8yntmetry prowrfy continue to apply in (3-7.52) without recourm to iliiffer-
c?nti&lidentiliw, we dennand
f ""(z) = f k"(z). (3-7.59
But $he eombinatian of (3-7.51) ftnd (S7.33)has the folla~r?tg
comequeDce:

Phy8ical im$icatians of the gymmetry of the s t r w tansor will be gowider4

later, in connection vv;rth.the di~cusgianaf anmlsr momentum.
Now we turn to the fiapange function

and obseme that it, Gves the ~ t r m

hnsor

1L remains tme, bowever, that

According $0 the field equtalion

and the field-dqendent kerns QD the righaand side of (3-7.57) can be

+
expressed in terms of the effective vector source J, d'M,.. The specialisation
MP, == 0 idenfifim fhe two vemi~nsof the stress t
Another qum1ian &ouL uniwenttss presenb iksdf, The generdisation to
arbitrary displacements could have emulahd more eiosefy rigid zota$ion km-
havior, ss tha;fc ($7.32) would h&veh n
3-7 lnvsrianea transformations and fluxes. Mechani~sfgropeftias 218

with the associated vector source variation

Tadeed, the distinction between field and source variations disappears com-
pletely were we to adopt

How does this freedom to choose the form of the displacementinduced v%riatiorts
r%Eectthe identification of the stress tensor, defined generally by

and presumably made unique by the requirement of symmetry? The various

choices for 6 4 p differ by terms proportional to

which vanish for rigid transfatioxls and rotations. This tensor is a, memure of
the dilation produeed by the displacement, and includes the scalar measure
c3, 62". The same dilation tensor appears in (3-7.M), in consequence of the
symmetry of the stress tensor. Consider, for definiteness, the effect of the addi-
tional field variation

on the action expression (3-5.18, 19). Sinee

contains second derivatives of the coordinak dispIacemen.t;s,s partis1 integration

is required fo attain the form of (3-7,64). This introduces second derivatives of
the fields, and the field equations will be called upon to eliminate them. The
initial choice of field variations was sueh as La obviate the need for any applica-
tion of the field equations. The dilation dependent variations can be treated m
a unit, and the stationary action principle invoked.

which shows thttt the stress tensor is changed, by the addition of

This is also what is requird in order to maintain consistency with the dimet
25(3 Fisjtds Chap, 3

evduatisn aft the zaetion v a ~ s t i o n :

We conclude that sitress temom do have tt degree of arbitraniaess eomesponding

to the frwdom in sssigning the effect of general coardinadt? displmements, but
t;hisl arbitrarines~is confined to the inkrior of the ~fsources,In the soureefrm
reeons where the &=ss tensor controls the flux of: energy and anamenturn iL
would %em to be unique, wcor&ng h %hemlm we have laid d a m for ilts
evalaation.
Rmrrangement that require the use of the field equatbw c m d s o be
sccomplished by a different choice of field varialion. We iftuslrab %hisin the
spin O situation. with
6<t, = GxPd,+ +
*+ay 6xY, (s7.71)
which gives

Tfiis stress bnsor is

tPp= ap$av + -- +a2(+2)I. (8-7 '74)
The field equation, asse&s that

an$ the ori$ttnd stress tenmr reappears, with an add4 brm. that vanish=
outside sources. But the direct w e of (3-7.74) yields

which, in source-free regions, is the result obtained in (s7.9) through the use
of the field equations.
The? question of uniqueness intrudes agrtin in this exrample. The rearrange
ment that eonneets (3-7.72) and (3-7.73) mighf have been haadled digereatly :

This wmld eh&&& the identifies%ionof %he@itresstensor by t h added

~ brm

hnsor eon&;t.ibutionhas a divergence %ha$

Not@that this gymmetrical z~tre~s
3-7 tnvarlnnce transfarmstionar and fiuxslr. Mgchaniesl proplttties ,237

Fu~hermore,it malres no contribution to the t o t d energy and momeiltum, in

vidue of the inbeation theorem (1-4, f l),

This is true for angubr mamntum also. The d d e d term i s p r e ~ e n t din the
form of (3-7.52) by choosing
' = &[B"3p
fh +
+hydfl-- 28'ya"+2. (3-7.81)
Observe that this expression is symmetrical in p and v, but does not have the
antisymmetv that is stakd in (3-7.51)- The annulment of the last krm in
(3-7.50) comes about, instend, through the differential identity (it is (3-7.79)
again)
akarfk" 0. (3-7,52)
The rejection of stress knsor terms that invalve digerential iderttities is thus
an essentid aspect of the computation rules. But we art? noiv going to sese that
stress tensors are quite analogous to charge flux vectors, Arty current vector
+
'j can be replaced by jw a,mp",with arbitrary &ntis~;mmt?kicaf ntF"". Qmmight
ag;ree ts rejeet such additional terms in studying a given Lstgritnp function, but
the existence of different Lagrange funetions for the same system, leading to
current expressions that differ in just this wayl shows that the arbitrariness is
intrinsic.
The arbitrariness in symmetrical stress tensors i s expressed by the possi-
bilify of repfacing tp"(x) \\-ith

where

and the symmetry restriction

nZ~v""IX mm"",x' .+ rtzx",l^" 8
assures that
a , a , a , ~ a ~=~ ~o.~
It also guarantees that,

and it will be verified later that the total angular momentum is equalXy un-
aflected by the additional term. In the simple example provided by (%Ten),

As in the discussion of currents, the exisknee of two sets af firsborder field

equations for spin 2 particles provides s valuable proving vound for unique
RWS questions,
2l8 Fidds Chap. 3

The tensor variation structure (3-7.40) is compatible with both antisym-

metry and symmetry of the tensor. Accordingly, the symmetrical tensor field
of spin 2 particles can be assigned the displacement variation

Two significant derivative combinations of this tensor are given in (3-5.37) and
(3-5.44). For the first, antisymmetrical combination we deduce

while the symmetrical structure obeys

Notice that second derivatives of the coordinate displacements appear here.

First, we consider the action expression (3-5.33), simplified by setting the third-
rank tensor source equal to zero, so that
GxrV = ax+Kr -aApx (3-7.92)
and
+
d: = -;tGhflYG~r, +GAGx- *m2(+pv4p,- 4'). (3-7.93)

Infinitesimal coordinate displacements induce

6s = a,(6zVd:) - trv'a, 6s' - ~ " ~ a 62,,
~ a , (3-7.94)
where
~ P V=' p '
- GFK'G',~+ 4@"'GKvx- GNG' - Gx(GX" + G'")
+ 2m24'"" - 2m24@" + gP'd: (3-7.95)
and
F ~ P V= - + +'G,vv. (3-7.96)

To extract the stress tensor we use the following identity:

A partial integration in the action variation shows that

1'' = 1'"' - ax[FA"'+ F"" - F""],,,,, (3-7.98)

where ( p ) indicates symmetrization in the indices p and v. According to the

evaluation
3-7 lnvclrian~etransformations and fluxas. NIs~hsnlcalpropertim 219

this gtress tensor must obey

When we confine our aLtention tr, sauree-free regions w considerable simpli-

fication. occurs since all vector and scalar field combinations rranigh in these
circumsdances:

and further reduetion c m be acconnplished with the aid of the field equations
ahck" = m2+pp, axc;pAp .= 0. (3-7.102)
Thus,
ah(Gh'*+'.) = m2$pK+vK+ f G""C~'~ (3-7.103)
gives

from which we obtain

to = --m2@'+,..
In discussing the alternative action expression (3-5.40), we shall p c e e d
directly to source-free eondi.tions by omitting all scalar and vector fields in the
tagrange function, This i s jmtified, despite the varistion that is going to Be
perform&, since all such fields occur in pairs, one faetor of whieh continues to
vanish after the infinitesimal coordinak displacement has been etppfieid. Then,
with
&'X a p + v ~ 4- ~ V + & X ---- d ~ + p . ,
"LI;: (3-7.106)
we have
e = ~H~'%H,.,- +mzyp6&. (3-7.107)
and
+ +
1% = --+(H@~'H.~' H P h % ~ ')) +@'a: 2dh(~'"+.. -- H X " + ~ ~ ) ~ ~ ~ l p
(3-7.108)
in whieh we have also uwd the field equation

The comparison of the two stress tensors noiv econfirms that

220 Flsida Chap, 3

satisfies the symmety requirements laid dawn for t h i t~e m r . A ~ h o r k defiv*

r
dion of the rmuft is b on direct eonsideration of the diEerence in the L*
grange funcliorrs,
CH - C@= $a""'a,+.k. (3-7.1 12)
Concerning the expre~sionfar the displacement induced &an&@of soure@
and field, we note that the availability of dditional b m ~ that involve the
general dilation Gensor (3-7.635) is strongly spindependent, if dditionai co-
ordinate derivatives are esefrewc?d, Matrices &ha$a ~ I tupon objects carrying
pin s arc! limitd Lo the e o n s h ~ t i o nof teaorr~with rank _I 2s. Af least unit
q i n is requird in order to r e d i ~ esmotld-rank Lensora, and only the gealar con-
traction, of the dilatbn bnsor is av&la;ble ta pins O and +, The spin O example
is illustraw in 63-7-71). We no\v discurn apin +. On referring ta Eq. (2-5.11),
whieh describs Lhe regponsr?,of sny m u m to coordinah transformwtions or the
equivalent rigid trandaLion and rotation of the sourn, \ve are led, for the spin 19
source 7 f ( ~ )to, the generalisalion

The spin Lerm only contains 3, 6%" - d, 6q,. There i s no symmetrical hnsor
%hater-tn be deviged from matrices, other tbwn g&" multiplid by the unit matrix.
A specific mui%ipleof d, &X' appe~m,analogous to the seatar source reBponss!
of (3--7,4). The corr~spondingfield va;ri~tion,~vhicbis &sign4 to leave intact
$(h)?ro+fis
+
&\d.(%) == &'(~)ar$(~) iQo"@+(z)&8~,(2), (3-7.1 14)
The spirr. 4 Lagrange funetion

has the following irnmediab responm to the variation (3-7.1142,

But the symxrretv of the second de~viativepicks out the matrix comGnalion

and, since the are symmetrical matrices, the last term of 642 vanishes.
Farthermore,
[rk,+opv] -~Bh~7r),
=i(ghF ~ (3-7,118)
and we get
66: == a,(lEa;'C) - tPVd, (3-E .f 19)
tvith
+
F' = 1'. = ++ro$[rp(l/i)ap $- rp(l/i)a@]+ fve. (3-7.120)
The associated sealsr i s
t = -m4+r0$ - fVro+, (3-7.121)
3-7 lnvarlancs trrrnsformationa and fluxes. proper?les
lVIschanicr@

since the field equation implies that

and this stress hnsor must obey the divergene@equation

as one can verify directly. The various terms of 6A? show explicitly how the
8csllar nature of the Lagra~gt?_ funetion l e d s to the cancellation.of the rotational
sfrueture, leaving the dilationd part and thereby producillg a symmetrical
stress .tensor,
In the re@on b t ~ v e e ntwo causally septtrated sourees, and qz, the field
cllahd with a particle of specifid momentum, spin, and charge is
fEq*34.49)f
$(z) = $ p . p ( z ) i ? ~ p ~ p4- $ ~ ~ q ( ~ ) * i q : p . ~ . (3-7.124)
Thr?eorre~pondingstrem b n m r contribution is given by

aecoding to (3-6.53). The expecked enere-momentum flux per particle is

evident here.
When pafiicles of spins Q and l are described by spinors of the second mnk,
+
it is natura! to follow the example af spin and write the displacement induced
fieid v s ~ a t i o nm
8lt. = &X%,$ + +
i#(a"f &$;")$a, 6zp. (3-7 126)
Its eKeet an the Lagrange function

ia quib andogous to the spin 4 situation, apart from the 6zy tern, for
now [u = rfr:]
6 8 = azPa,e f ++r&i (?:av r:ap)J.a,h,+
+
- ++r+(r:+~:' r k i ~ : ' ) ~ aaz..
~a, (3-7.128)
The identity (8-7.97) prduees ithe symmetriea2 stress hnsor

to kvhieh the null terms involving (?@ckY)

can be added. This is useful for the
matrix dr4tnseripdion of (3-7.129),
where the double commutator can be replaced by the double antieommutator
without agecting the value of this term.
foms will be e x a ~ m d ,
The relar;ion between this and earlier stress ~F~EISOX"
wing the gero spin example. Inserting the m a ~ fieldx (3-5.83) one finds

Lagrange funetion, given by (3-S.%), has been replac4

where the (z~lpropx"iatf3
by the Lagange function of Eq. (3-5.16) and the necessary divergence krm,
Observe how the derivatives of the vector field cancel, leaving

which agrees with (3-"1.8) and (3-7.37) in regions where the veetor sowee
vanishes,
The generalization of the field variation (3-7.126) to an arbitrav multi-
spinor is

which can 8180 be applied to the vmious auxiliary fields #raBlt J/iaB)lavt .. . .
source-free space all auxiliary fields vanish, and this property is not aBectc3d
Xrct,
by arbitrary displacements since the transformation law (3--7.133) refers only
to the 6eld under considerrttion. Accordingly, anly the first term of %heLagrange
funcLion (3-5.78) coxrtributes, and the resulk obtained for the stress tenmr is an,
immediate generalizatian of (3-7.X29), but wifh 6: = 0 befits the sourceless
circumstances:

We shall elaborate fibis for n = 3, using the spinor, antisymmetrieal; in a pair

of indices, that gives an alternative spin 4 description. The eal~ulationr e
sembles the, one already performed far the current, and the re~u1Lis

where
+ rp(l/i)ap]q
tg = +qr"+(rp(l/i)a' (3-7.1 36)
is the stress tensor of the simple sginor or Dirae equation for source-free condi-
tions. The additional terms can be exhibikd in the form of Eq. (S7.831, with
3-7 Invariance transformetions end flux=. NIechanfoet properties 223

The relative rotation, of sourcm supplies anwlar momentum information,

To d i m u ~this, consider the displaeernent
62' = dwA'(z)z*, 6 d t ( z b-dow(z), (3-7.138)
the aceomp~ayintgdilation

The latter v&&%, of eourmJ when Gu~,(s)ia a constant, dea~ribixlga rigid

rohtion, The implied actian variation is

where
jhhr = Z~th' -- %'tkpe

The dhmative evalualion refers directly to the regpom of the souree, In

order to deal uniformly with all spins ws? use the multispinor de~cription,for
which

Comparison of the two computations givw

This re8uIt is not independe~tof the divergence ewation for the ntress bngor,

since
ghjx"~ ghaxtkv -- xpaxthfi + pp - p, (3-7.145)

It is hem that we recognize the significance of the symmetry prope-rty possmd

by the stress tensor. The quantity jkv, which i s formed merely by taking first
moments of the sGrf?ss knsor, is eonsewd in ~ource-freeregion@, As the m&&-
tional analogue of P', jh" evidently provides a spsce-time sceount of the distri-
bution and flux of angular momentum, in s fourdimensional mme.
As in eadier discussions, the rigid rotation of aource q 2 relative to the-
cau~allydetached source is introduced by letting 6c~,,(z)vanish in the future
of an inhrmdiate spa-like surfam, after being the congtant 4, in its pasLa
This gives
(3-7. f 46)

AR analogous e-qustion applie~ta the time cycle demriptim, wbwe an@has the
impler physical inbmreta%ionin h r m ol an evetation value

Surface? inbgration is replaed by v ~ l u n eintegration, 6 t h the aid of (3-7,143),

%hecombination of orbib1 and spin 8 n ~ l a momenta,

r is e~denChere, The
eig~vaienfm~dth other fornnulatiaas?;is illwtraM, for fhe vwbrial source-
and field of unit spin p&ides, by [cf. Eq. (3-7.49)j

Wer ~ h dnow
l confirm that the evaluation of total ~ 1 ~ @ momentum
ar is an-
afleeM by the mbitrariness of the @%re88
&mar. AL~e~mpanying the redefinition

The verifieatioxr that the additional ~ c a x l d - d e r i v i ekrm gives no eontfibu-

&ionto l&&Pfi'is identical to that of Eq. (3-7.87)--only the oyclic symmetry
prapr,rLy (3-7.85) is invokect. The lafter also impbw this slakanent of anti-
synnme%qin X and a:
_
[ m h ~ , a ~ m b . a r ~+ _ m a ~ . h v l = o8 (3-1.152)
from which fdlows the vanishing of the s u d ~ eintevd for the Xwd term
of (3-.7.151).
Ps&icle s t a w %ha% are labled by three4imensioztrtl angular momexttum
quisb~tumnurnbr~,ar~therthan by Xinew momentum, have been exhibited for
spirms 0, *, 1. T b i r use i~
suEciently similar to the charge and enerw-momentum
&aewiom that we Eskdl not enbr into detail8 here,
In ontrwk do $he fidd translations and rotations of phygieal inkre&,
$da$iorrr~ would =em da be a d y a device thaL assists in the identification of
merw-momentum fluers. There is, however, a 8ubset of these tran~formations
that plays a more physical role in (the spc;ie,ialcireumtance of
18 is the ~ o u oZp hatropic dilations that is characteris& by

These conditions are vemy. re~krietive.We fimd nob the seafar relation

The &vergenee of $he kawr equ~tio~1

then gives
3-7 tnvsrlance transformationsand fluxes, Mschanlcai propcrrtiba 226

and et fu&her divergence asserlks thad

But even more is obtained by applying the operator a' to (3-7.153), namely

Thus &p(x) is Einnihd La a finear funetion of the ~oordinahg,

%ndrota-
The corresponding form of 6xp, apart from infinite~imddran~Xa?;tiong
tions, is the quadratic=function

Tagether with tran~lationsand rotations, them fransfarmations form s

group of 15 parsmeter&. It has the structure of the rotation group in 4 2 +
dimensions, in the senw that the homogeneous Larentg group is the rotation
-+
coup in 3 I, dimensions, Perhaps the quiekest way to recogni~ethis is through
the introduction of homogeneous coordinahs,

that are defined in the five-dimensional space of the null 'sphere'

The ten-parameter space-time translation-rotation group now appeam er*s the

subgaup of homogeneous transformations on: the null sphere that leaves q15 y~ +
invsrian t :

the?! one-ptzrannebr &a transfomations, uniform scale change8 in ~spaeetime,

are the "rotations'

and the transformations parametrized by Cibv hold g5 - fixed,

The quadratic form of (3-7.161) also admits refieetiow, including 116 -+ --g&,
which hezs the following effect upon the xp coordintaks:

an inwersion in the origin. A sequence of two inversions, first p e r f o r d at the:

origin anci then at the point with (coordinaks &bp,produces the infinibgimd
tramformation of (3-7.159) with &a= 0.
226 Fietdzii Chap, 3

The isotropic diltllions are known as conformal transformalions, Their

physiestl relevanw emerges on considering the part of the Lagrange function
response t o displacements that is given by

which, singles out the racalar- t as the significant quantity. Inspection, of the
examples with spins 0,#, 1, 2, shows that the scalar E, evaluated a t source-free
points and with m -+ Q, either vanishes or is a second-derivative structure of the
form implied by the arbitrariness of the stress tensor,

Since d l physically interesting possibilities are included in these examples, we

shall forego the luxury of a general proof, Rut what is invofved should be clear.
When X/'VL is infinik there is no standard sf length in the action. If the scale
of all coordinates is changed uniformly,

the form of the action can be maintairled by a corresponding scde change of

sources and fields that is determined by the number of derivatives in the La-
grace function when the minimum number of fields is used, This is illustrated
for spins O gnd. % by

Considered in. infirritesimal farm these settling lakvs specify a definite multiple of
4, Gzr-an example is (3-7.71) far scalar fields. The invariance of the action,
stating a local property of the Lsgrange function, requires that the scalar t
vanish everywhere or, more generally, be the divergence of a vector. As one can
verify in the simpler examples, this vector employs the gradient vector for its
construction and (3-7.168) emerges ss a generally vaIid statement for a, suitable
choice of field variations. Then, since the general $ @ ( X ) , Eq. (3-7.158), has
vanishing seeond derivatives,

and the action is invariant for the whole 15-parameter group that ineoflporates
confomsl transformations,
As aIm-ays, invariance of the action impties conserved physical quantities
with space-lime distributions and Auxes, The procedure is standard; the con-
stants 6a and lib, are replaeed with arbitrary coordinate-dependent functions.
We shall assume, for simplicity, that t = 0, If, instead, (3-7.168) is applicable,
additional terms appear in the various Awes but nothing basic is altered. The
response of the action to the generalizations of the conformal transformations is
3-8 Tha slectromagnstic field, Magnetic charge 227

with
C@ = tpyz,, C" - ~"Yz').
= 1fik(2~%2' (3-7.173)
The tensor c"^"is not symmetrical:
-
C ~ v == -22hjhpu,
and the implied scalar is
c = @pp8v 2zhek.
JR source-free regions, the Ioeal conservation laws are

and the existence of conserved total quantities is indicated, for Lhe time cycle
description, by
(3-7.177)

The physical eontetlt of these conservation statements will be discussed in the

corltext of the most familiar massless particle.

3-43 THE ELECTRQMAGI\1ETIC FIELD. MAGNETIC CHARGE

Although frequent. reference has been made t a the m --+0 limit for unit spin
particles, it is important to give ark independent discussion of the field associated
with the massfess, unit helicity particle-the photon, The slarting point is
Eq, (2-3,45), written as

I n defining a field A,(z) tlzrougl.1 the effect of a test; source 6Jp(z),

(dz) GJP( X ) A, (X), (3-8.2)

strict account must be tnkezt of the source restriction, which demands that

one shouId not identify the eoefieiexlts of 6Jp(ts). The correct conclusion, is that
they differ by any expression that leads to a vanishing integral in consequence
of the restrickion (3-5.3). The general form of such an expression is
 Chap. 3

and therefore
(3-8.6)

The aspect of A,(x) that is governed by the arbiCrary scalar function h(s)
is p k k d out by forming the divergewe of (3-8.6). This gives
a.kV(z) = a2x(z), (3-8.7)
and the application of the differential operator --aZ to (3-8.6) then provides
us with the mcond-or-der difflerential equation

We reeogniee Eq. (3-3.6), with m == 0. Since the arbitwriness af X(z) persistfs

in this &Rerentis1 field equation, the latkr musk be! unsffectd by any rdefini-
tion of A, in the form
A,(z) + A,(%) 3- dpX(z)t (3-8.9)
which is known ss s gauge trsnsfarmation (mare frequently, as a guagc? (sic]
transformation), This gauge invariance is emphmimd by ~vritingthe field
equation in the equivalent form [earnpare Eqs, (3-3.7,8)]

since %heddition of ra- gradiend term to a vector does not alkr the: curl of the
vector. Thst the divergeneetess nature of J P is built into the field equations is
8ho emphasizled, for
8,Jr = d,iZ,PV E 0, (3-8.12)
owing Co the sntisymmetry of F@". The curl construction of F,, is given, another
form in the differentid equa;tions

The tensor dud to F,, ia &fined by fEq. (2-3.60)]

'pY
= +e@,gh~Ek,

%?herethe totally sntisymnretrietal knsor of the f o u ~ hrank is normalizted to

c0123
- +l. (3-8.15)
Using this coneept, we express the differentid equations (9-8,
f 3) as
d y *F@'(%)= 0. (3-8.16)
The pair of equations, (S8.10)and (3-8-16), me MwxweII's equations for the
Lensor of ejectromagnetic field strengths, F,,. We recall the identifications of
3-43 The stearamagnstio field, Magnetic-charge 229

The alkmative evaluations of W ( J )thftL l 4 to the action expression. are

The eleetrom8gnetie aetlon, fareshadow& in (3-5.94), is

wi%hthe Lwaxrge function

The Lagrange function is explicitly gauge invariant, and so is the action beclzuse
of the &iflerentialconsemation property of J@. This time we have bemn with a,
c o n ~ w a t i o nlaw and inferred an invariance OE the action.
In =king expressions for the murce and field vadsltians that are &soci&bd
with a r b i b eoordinak
~ displacements, id is natural to mabintain fhe conserve
%ionprope&y of J P and the gauge inrrafittn~eof FP,. The vecbr souree tram-
formation law (3-7.33,47) has the required characteristics since

iassures the continued vanishing of d,Jfl; the amoeiated knsor transformation

(3-7.M),
&FP.= +
6 ~ ~ a h F , . Fk.a, + F,,9, 6zk, (3-8.23)
involve8 only gauge invariant quantities, And the vector field prescription

which w&sbased initially on the tramformation prapedies of the psdient of s

malar function, is maintained under a gauge traasform8tion. The direet evlzllua-
tisn of the imducd &@&ion variation is

while %heLagmnge function regponse is

23Q Fields Chap. 3

with

and

The comparison of the two evaluations j v e s

Same explicit stress tensor expremions are

to' = &(E2 H'), + tok (E X HIr. (3-8.30)

A direct approach to the eonfofmal conservation. laws will naw be m d e .

Multiplication of (3-8.29) by z, gives (I = 0)
d,cp = J,FP"z,, (3-8.3 f )
and similarly
J , F ~ ~ ( ~ Z "-~ g"z') = fd,tgk)(2z%' -- g"z2)
E a,eE"V, (3-8.32)
The nstum of the eamfiponding integrd mnsewrttian b w s mogt closely re-
wmbles t h ~ oft

which, being an explicit function of time, is a statement about how something

distribution in this instance, The strongest
moves; it is the eentroid of the e ~ e r w
asmrtiorr of this type that is implied by conformal invtadancl: is ~ontainedin

Thus, witb the weighting factor provided by the enerw density, the werage
value of x2 varies quadratically in time, with unit coefficient of (~'1%.The in-
krpretation in Germs of the motioa of the particles &at carry the energy ig
clear: photons move a t the speed of light. The coefficient of z0 and the constant
term supply information abouC the initial. correlation between position and
velocity and tbe initial average value of X'. This view of i(dx)eO is consistent
with its ~ignificaneein terms of the momentum distribution:
3-8 The electromagnetic field. Magnetic charge 23t

The field strengths F,, and the vector potential A, are placed on the same
footing in the following action principle:

where
= -%Fpv(apAp- aVA,) + $p%,, (3-8.38)
is explicitly gauge invariant. The field equations now read
= P,
avFNv dpAp - a,A, = F,, + M,,, (3-8.39)
or
a V ~ p=
v p , a, * p v = *JP, (3-8.40)
where
*JP = -a,, *MPv, 8, *JP E 0, (3-8.41)
and *MMvis the tensor dual to N"'. The stress tensor is symmetrical,

and
avrv= (JP+ a A ~ , , ) ~ p+vJ,M,~ + *J, *M". (3-8.43)
On setting Mpv = O and identifying the field strength tensor with the curl of
the vector potential, the previous results are recovered.
If the photon source function Jp(z) has the interpretation of an electric
current, according to the first set of the Maxwetl equations (3-8.40), is *JP($),
as realized in (3-8.41), a magnetic current? The answer is negative. It is con-
sistent with this, but hardly decisive, that the total vdue of the apparent
magnetic charge is zero,

provided MN"as the kind of spatial localizability that attaches to the source
concept. The essential remark is that, through a redefinition of the field strength,
the magnetic current is transformed into an equivalent electric current. Indeed,
the equations (3-8.40, 41) are also given by

which contains the effective electric current already exhibited in (3-8.43). But
this short-lived possibility does raise a fundamental question concerning the
existence of real magnetic charge, distributed and flowing in a manner that,
explicitly or in context, differs from (3-8.41).
To study this question, we go back to the beginning, to the source. Is it
possible to distinguish two fundamentally different kinds of photon sources?
But the two kinds must also be closeIy related, for the structure of the Maxwell
232 Fialds Chap. 3

eqwtions is rehined under the substitution

or, more generally with arbitrav angle p,

JP + *JC^sin p,
-+ J P cos p F*' 4 FP'+ *FPr@inp,
COS p
(3-8,471
*JP ---JPsin p + *JP COSp, sin p + *F@*cos p.
*F@' -+ -Pu
The consib3tency of the field strength sub&tutions involve8 the r e p t i t i o ~
p r o p e ~ yof the dual,
**F~Y = -?-JwY, (S8.48)
All this brinm to mind the discussion of Smtion 2-3, It vvm recopiad there thaG
n~Lhiftgint~nsieis albred if aX1 photon. p o i a ~ ~ a t i ovwbm
n are rotaM though
the angle */g, thus repising the initial polarisation veotor set e,h by ( p @= Ipl)

This suaests that the de~ireddisLinetion and relation b"Ewn two frintds of
soumes is realized if the memum of their eaecfiveness in emitting a given phuh~t,
1abeIed pk, udililieg e , ~for one kind of sourn and *e,x for the other. Thzrt is
indicated by (red pala~zationvectors are u ~ e dto reduce the number of @tars
in our ctyc;s)
$,X = (d@,)"2[@p~ $(p) -4-*@,h *J(p)]. (S8.W)
Ths equivarfcsnce of the desefiptions that are eonnecM by the @@wee tram
formation of (3-8.47) then expreBses the freadant to rotate bath s y s b m of
pols~sationvectors -tXlroughthe common angle p.
Again we consider a c a u ~ dsiluakion, brtf now with compnmt sour- JC;,
V1; and J t , V;. The coupling betmen the exnission and &~orptionsowem
that a single photon, mediwks is conveyed by

The intrinsic equivdence of the two mts of polzlrizadiun v e e b r ~is contained in

the dyadic relation

ancl the eommponding brmrs of (3-8.51) can. h &ven h v a h n t fom in the

kno-run way. But what of the couplillg btlCween different kin& of sourem? We
fin& note %hatthe summation in
3-43 Tha ~Iectromsgneticfield, Magnetic charge 233

can be extended to include the third unit vector prardlel to p, thereby introducing
the unit dyadie:

The coupling illustrated by JI(-p) X 'Jz(p) pipQ is very three-dimensional

in. appearance, Nevedheless, in the physicd circumtarrees under examination,
tbis is a Lorents scalar. That can be dirtj~tlyverified, It is mom rewarding,
however, to w r i k this term in an explicitly covariant form.
We begin by remarking that (causal subsc~ptsare o m i t w )

where f,(p) ha8 onXy a time component such that

~ P % ( P ) = 1. (3-8.56)
The decisive observation is that (3-8.55) remains true in its fourdimensional
form with any vector f,(p) that obeys

under the eausal conditions that mquire the photon enerw-momentum relation

Also relev&nt, but holding without regard to eausal arrangemen&, are the
current conservation statements

The fottowing identity, which is valid for arbiLr%rypp, is b w d on these con-

servation laws :

Of the two k r m s on the righbhand side, the second vanishes when gP is a photon
mamentum ohying (3-8.582, and the first is indepnden-d of the gpcific choiee
af vecbr &(p) that obeys the restriction (3-8.57). This is in, fact the proof of
covarianee for cmsaf eireurnstanees, But same explanation is caitled for, A elarss
of functions Ghat obey (3-8.57) is given by

where n, is an arbitrary constant vector, ff n, points along the time axis, fa2
example, we h w e the situation of Eq, (3-8.56). That eharae.t*rtrizatioaof f,(p)
is not eavariant; after a Lorentz, transformation is perfomed, n, will have non-
vanishing ~patialcompsnent~,although it is still a, timelike vector. 1%ias here
that the arbitrariness of n, enters, for we can replace the time-like vector by ons
with only a kennporab component. It is &rough sueh coupling of the ehoice of n,
234 Fields Chap. 3

to $be choice of coordinate sysbm that any discrimination among coordinate

systems is avoided and covarimce achieved.
The spaee-time transcription of (3-8.57) is

If, ss in (%8.fil),p(3s - z') is prapa&ianal to a constant vector n",

the digerential equation

is eRectiwly onc? in et single vrtriable, For the gituation of (3-8-56), with only
no $ 0,we get
aojO(z- g') = a(xa - X@') &(X - X)). (~8.65)
The soluGion is not unique. Two alternative solutions that correspond to re-
tarded and advane& bounday conditions are

where

represents the Heaviside step function (the capital of the Greek letter is H,
as the eapitrtl of i5 i s D, in the Chalcidian alphabet). Another choice aligns the
vector n" tvith the third spatid axis, for example, Then

and dternativt: soIutions are

fa(2 - 2') d ( ~ ' - z") 6(g1 - 2:) 6(za - zj)[?(za - X & ) , -?(%g -- za)l
(3-8.69)
We specifieagly note the equally weighted linear combination

in, which

More gener~lly,it is compalible with the diRerential e q ~ n * b fS8.62)

n to imp-
%hesymmetry restfiction
-F(z1 - 2) = fV(z- d ) . (S8.72)
The four-dimensional replacement in (3-8.55) will be used Co perform the
of dhc: murm oouplings inferrd under o
spiace-time ex%rapola%io.rs
3-8 field.
Tha ~Is~tromagnstic Magnetic charge 235

stances, If one of the f" functions in (3-8.66) were adopted, an, additional causal
elernenL, which i~ arbitrary and physically irrelevan%,would be injected into
the description, ]in contrast, the kind of funetion illustrated in. (3-8.69) is
bmporatly inert;, and its arbitrary aspecls are confined to spatial directions.
Since causa1i"ty is a fundamental widing principle, we reject the use of funetions
such m those in. (3-8.66). Without being commitkd to the specific examples of
(3-8.69, '?Q), we do insist that f"(x - X') have a, spsce-like direction and be
loealiaed in its time-like coordinate excursions.
The desired space-time exLrapoltzLrion is glxven by

Xn verifying that this properly represents the initial carnal situation, we en-
counter the Fourier transforms

and

whieh are involved in reproducing the last two terms of (3-8.51)' The latkr
are interchanged by the substitution:
JP (P) + *J,(p), *JP(p)-+ -$#(p), (3-8.77)
To test W for this symmetry property it is convenient to introduce four-dimen-
sional momentum xlotatiart :

The effect, on the last term, of the substitution (3-8.771, combined with
p,+ --p, and p ++ X, is

and invariance requires that

-~P(-P) == ;F;(yz).
I n a eauaal s3ituatian, however, fp(p)oeeur8 anly in Lhe earnbination i p y p ( p )= 1,
and this additions1 synrmetv praprty is nok .Bud if the wlatiae
htween the two source f e y p i s tit3 be maintain4 generally, &hecondikion,
(3--8.80), which is dso (3-8,72), must be i m p d . 1%foUow8 fhat W(J V) pre-
=mm it6 form under the general murce tran~formationof (3-8.47).
Two kinds of h s t sourcm define two Ends of fields:

which are, in-

There are alm two independent kinds d gauge arbitrarin~m~
corporsM in the fie!d exprewiona

(dz') (dze')D+(z- z')f '(z' - z") '( 3 : ~ .(g") -- ak'~,(z''1)

+ 4, *h($),
where @@""as been used to form dud tensors. The following identity should
be aobd [it i ~ the
3 eonGenf of Eq. (3*-8.13)J

Obmme alsa thellt, for mampfe,

3-43 The slactromagnetic fisfd. Magnetic ehsrga 237

where the Iwt form involves the inhrehange of the indiees h: and V, The fwa
gauge inv&nisrttfields

then obey
dvF""(x) JP($), 3, *Fp"(%) *J'(x), (3-8.89)

But only if *Fp"(;e)is the dual of F""(%) can we, procbim these to be the general
f o m of Maxwellk eequa;ttions, with electric and magnetic eurrent~,
A direct proeedurr; for &is purpose is to evaluate the curls of the two vechr
patentials A,($), *R,(s) and compare the results in (3-8,88), Here is another
identity that is valid for any antisymetrieal hnsor Q@,,
+
a, *G,& a, *G&,+ ax *Q@,= - % Y x x a , ~ a . (3-8.90)
1xr consequence,

with a similar expression. involving *A,(s), and the use of the differentbl
equations

gives

The necessary dual relationship is exbibikd here. Notice that the gauge in-
variant field stren&ha are also independent of the arbitrary veetor $,' It is
e~identthat thme kasors obey Maxwell's equations, The converBe is also %me;
the solution of the MaxwelX equations with outgoing wave boundav eonditiom
is just (8-8.93). To verify this the identity (3-8.90) is applied, hthe form
238 Fields Chap. 3

which. produtees
-a2Fpv= a , ~ , -- 3 , -~*(a,~ *J, -- a. *JJ. (3-8.96)
The dwired ~olutioaiis that stakd, with its dual, in Eq. (8-8.93).
Apast from the characteristic freedom of ga;uge tr~nsforrnalions,the fm
vector potentiebb ean be exhibikd in kms of the field slren@hs. Fir&, let ua
sbmrve that

(3-8.97)
h= the Foarier transform

The conmquenf v~nishingof a,(%) is exploited to derive from Eq. (3-8.m) tba&

which also uses the diWFerential equation obeyed by f"(z - zf). Thus,

and a gauge tr~nsformationan A,(%) change8 X(%) appropsatety to mainldn.

these relations. We da not mean Lo suggest, by the wsy, &at the? X(%) of Eqe~,
(3-3.83) and (3-8.1W) am the same function. A common, symbol is used since
bot;h funefians embody fhe eharacferislic (arbitrar;lness af the veceor potential*
By introducing the compensating gauge transformation, the X(%) of Eq. (3-8.1QO)
can always be reduced to zero. The rmult is et, p%rticuIsrgauge in which

and we note: %ha%

the gauge conditions (3-8.102,104) are not independent state-
menb, but are implied by the constructions of Eqs, (3-8.101) and (3-8.103).
 3-9 CHARGE C1UANTIZIATlQN. MASS RIORMAILIZATION
Preparatory to exhibiting various action expressions, we note some integral
identifie~that ineorporah the field equations. Thus, from the Maxwell equac
dion8 fS8.89) we infer %ha&

while the equation8 of (3-8.88) lead to

Xn 8rriving at the last expressions the following property of the dual is used:

which ~ltlsoimplies that

' P *FIIYm --F"E""F,,.
The qua&atic W(J *J) expression eiln be presented as

or in various equivalent foms, ineluding

The latdtjlr passea the action propedy. But one must etppreciak the context,
describing the independent field variables. In (3-9.Q, for emmpte, the fieMs
A,, F,, are subject to independen%variation, while the a;ymbol *A, stands far
the functionat of the field s6renGt.l. Gensor stakd in (3-8.103),
* (dz')fp(z - z') *F~.(Z'). (3-9.8)
This is verified by performing %heindicated operations, whieh e;ive
2 Fields Chap, 3

Proeeding from the dual La Che last equa;tian, the =and mf of M%xwellt~
equationis,
a, *F,,(z) = *P(z~, (3-s.xo>
is generahd by diRerentiation. The con~tructionof A,(z) M l m s M in
Eqs. (3-8.99,100). The use of (3-9.7) is anslogous, with *A, snd *FpF,, as inde-
pendent fields while A, i s defined as a functional of the *F,, by (3-8+101),

The ~pgymme%ry in-vot~edlin e q l o y i q either A, or *A, as i~dependentfields is

overcome with yet a fhird aetion expresgi~n:

which is explicitly invariant under the rotational transformstion of (5-8.47),

and usea, $he field stren@hs EMindependent variables. The equation produced
by the fstalionav slction principle can be presenhd as

wing momentum rJpaGe for coxnpaetn~?ss,

where

Then, since K@($)is divergeneeless,

PVK'(P) = 6,
we learn, successively, that
~P(P)K*(P)
=0
and
K"d = 0,
which also uses the positiveness of -f@(p)f,(p) that expreases the choice of np
and %herebyof ip(p)in (3-8.61) m a pace-like vwbr. A ;sfimilartreatment af
%hedual Lo Eq. (3-9. 13) suppliw

and both sets of Manuell's equations have been derived in s symmetrical way
from the i4~;ftian
expres~ion(3-9.12).
By this time the bypodhetical alert reder of li~tiessddeatian, hencefadh
aefanycnicdly known as Harold, can no longer reatrain h
exchawe ensue^.
3-9 Chargs quantiration. Mass normalization 243

H.: You showed in the previous section that the apparent magnetic ~fZ&rg@
given in (3-8.41) could be transformed aws~y. It was intirnahd that a different
kind of mapetic current would be forthcoming. Yet the action pdnciple of
($9.6) and the Geld equations (3-9.9) etre identieral in form to (3-8.57) and
(3-8.391, with

and indeed
-
How then can you claim that true magnetic charge is rrow k i n g disc
S.: Mistake me no$, goad Sagredo, er, Haratd. The fundion, (3-9.a) does
differ-in context-from the source function of (3-8.41), far it lateks that depw
of facalizability which is characteristic of sources. Consider, for example, the
choice of $,(s - z') with only the spatial companenf

giving the nonvanishing tensor component

Unlike the spatially limited magnetie charge distribution *J0(z),*Mo3(z)be-

comes independent of after one ha8 passed through the charge dist~bution,
moving po;;jitivety along the third axis, This limiting value is

and the sudace integral

whieh need not be zero, That is in contrast with *he null value of (s8.441,
which ~vasbmed on the sptstial loedizttbility of "dd,,fz), Had we umd the add
fC^function of (3-8."i"), the explicit form of *Kfo3(z)would be digerenl, bud xlat
the value of the surbce i h g r a 1 that produces the totd magnetic charge. Ths,
it is through the special properties of the class off functions that we make the
transition from mere semblance to the redity of magnetic charge. At the setme
time this poses a fundamental problem since the detailed description would seem
to depend upon the arbitrary choice of the f function, for whieh there is no
physical basis. Surmounting that formidable difficulty is the task to which we
now rtddress ourselves,
Let us introduce into (3-8-81), which is the diEerential statement of the de-
pendelice of W on the source functions, those expressions for bJp(z)and 6 *JP(%)
242 Fields Chap, 3

that convey the efft;clof an arbitrary coordinate-dependent displacement:

up(%) - Jv(~)a,azP(z)
= ~,(~z~(z)J@(x))
= -a,[ ~x~(z)J-
~ ( szV(z)
z ) JP(Z)] 26)
(g9,
and
6 "J@(z)= -a,f&rc""(s)*JP(.) - 6zV(.) *JP(%)]. (3-9.27)
The conservation requirement8 (3-8.82) are identically satisfied, This inser-
tion gives

dlternative expressions for the two terms sre exhibited in writing

Two elementary statemnls concernjng explicit f p dependence emerge from these

forms. If electric and magnetk currents are proportional with a wziversd con-
stant, the f p term vft~fi~hes,
as it should since tbis ia a rotated vemion of pure
electric charge; when electric currents are causally separabd from magnetic:
eurrents the fB term vanishes, weording to the restriction s n the etas8 of fC"
functions that confines it to space-like vectors connecting points in space-like
relistion.
It is the situation of electric and magnetic charge coexisting with different
space-time distributions that poses the problem of nonphysical P dependence.
To make this very explicit, supposef""(z - z') is chosen a-s in (3--8.701, a spatid
veetar of fixed direction with its sawort, its nonvani~hingdomain, confined to a
fine of that; direction. Those points in the two source distributions that can be
conmcted by this line canlfibub to 6W, When the direction of the line is varied
continuously, bW and W itself also vary, continuously, thereby denying to W
any physical meaning. Is Lhis the death h e l i of mslgnetic ehargt?? No. Them
is a subt;le possibility concealed bere, I t depends upon the precise fact that not
W but expfiWjis the physically significant quantity. If, in altering the direction
of f p continuously, W were indeed to change, but change discontinuously-by
multipIes of 2~-the exponential would remain unaltered and the mathennatied
arbitrariness of p should be wri%houtphysical consequence. This is impossible,
of course, when, as is assumed above, the sources are eantimuously distribrtkd
objects, Instead, they must h w e a panuJar atmcture, gving values of the p
inbgral that differ by finite amounts according a&rs the f@ line does or does not
penetrate the kernels of that structure. And, since the magnitude of the integral
it3 ~ l s measured
o by the product of electric and magnetic charge, this combination
cannot be arbitrav but must be re~trictedto certain discrete vdues. Theae are?
remarksfile conclusions-charge is eomplebly locdized and q u a n t i ~ din magni-
tude, The sweeping nature of such inferences should b-tt, emphssi~ed. We are
encountering =strictions on the structure of photon sources that are required
for the consi~~tency of a theory of electric and magnetic charges. Sourees are
introduced as idealiaations of realistic physical mechanisms, idealizsllions thaL
dispense with individual characteristics but respct all general laws. Xn uncover-
ing fundamental restrictions on sources, we are revealing general laws af nature,
Sueh was the argument when the divergenceless nature of the vector pftoton
soufee, demanded by the null photon mass, was interpreted as the rtssertion of s
general eonsewation law, that of electric charge,
A realization of electric and magnetic currents in hrms of the motion of
point charges is given by

(instead of "8, syrnbola such as g are aho used hut we wish to emphasize? the
symmetry between electric and msgncsLic quantilies). The eau~almotion of the
points is conveyed by the restrictions

The conservation. properties hold individually, according to the ealeufation

since the point % p ( & ) is infinitely remote from zp a t the teminalis of the integra-
tion, The evident identification of the e, and *e, as charges attached Lcl the
individual moving points is consistent with the evaluation of the total charges,
ars in
do, dsP(s)6(1: -- ~ ( 8 ) )= e,, (3-9.33)

where the inkgation sweeps the whole four-dimensional domain with dcr, dzp(s)
acting abs volume element.
We cttnnot sirnply insert theae expressions into W(J V ) ,however, The
latter was devised for continuously distributed sources and should not be applied
to s collection of point charges without reexamination of the physieab sipificance
244 Fields Chap. 3

of W . But it is useful, and serves as an intermediate stage in the development,

to modify the known results in a manner that is without effect for continuously
distributed sources but makes the consideration of point charges mathemdieally
meaningful. This is achieved by introducing an arbitrarily small space-like
vector Ap and constructing the provisional action
+ * J ~ ( x*A.(%
W ( X )= / ( ~ x ) v ( x ) A , ( x f X ) ) *U
(apAv(~* X ) - avAp(z * X ) ) + ~ F p v ( ~ ) F pfv X)],
- 3FpV(z> (~
(3-9.34)
where the appearance of f X signifies the procedure of equal averaging for ex-
pressions containing + X p and -Xp. This action continues to be stationary for
field variations about the solutions of the Mamell equations:

which uses the possibility of performing a displacement to transfer fX' from

one field factor to the other. To evaluate W(A) we use
* + *Jp(x)*A,,(z * X)].
W(X) = & / ( ~ X ) [ J ~ ( X ) A . X( X
) (3-9.36)
The point charge construction of the currents gives

where

is symmetrical in a and b, and

Wo(X)= #(c: + *e.')/dr ds' D+ (%.(S) - zo(d) * X). (3-939)

come is concentrated in Wo(X). I n the neighborhood of 8 - 8' -

The mathematical existence problem which the X device is designed to over-
0, D+ would
be singular without the addition of the space-like K to its argument. This
difficulty is re8tficM to the real part of B+, however,

(cf. Eq. (2-l.M)], as contrast& with

Thus
-

cos px

h= nsturesl ulpper frequency limits if the motion of the particle: ia without di*
csntixluity, and the limit X' -+ 0 can, be introduced directly i ~ t Q
(S9.42). To
discuss
w,(X) = Re W@(&)

far suficiently smdl kC"and s comesponbingly eloiose to a', it S ~ G to~ consider

S
uniform motion. Let W use the wst frame, for simplicity, and identify ds wifb.
dz: in thst frame of reference, while choosing AB to be a spsti%lvvector. Then,

Doe8 wa(X) have a pfiysical sipificance"2t does nod. This quantity is wo-
Gi~kdwith a single poinl; eh8rgf: or partide. 8inse the pa&icles that cornprim
a source have prescribed motions they are being idealized as very m m i v ~
parlicles, which are uninflueneed by the effe~tsthey praduee. The desc~ption
of their indi~duafmechanical propertie8 lsgically precedes the discussion of
inler~ctions, The nature of this description can be infemed from the rwultp,
s their vdues in single-particle states: P"= 2 h , p p p ' .
concerning stress h n ~ o rand
As we have? expldnd, &is is a simgXificatian valid in. the intt3rior of a barn
whesc? fhe variation of momentum rand the associated firtih spatial exkensian
can km neglwbd. To reinshh these, we identify pp with the p a d i e d of a @me
function ~zndintroduce s variable weight funetion,
tp"(2) = p(z)aYrp(~)d"p(s);
the m- rest~ction,
dpcparp $ m2 = O
recalls the momentum sipifieance of aFa. N o b that

and the foeal rneehanicd consemation laws are satisfied by the consemation of
padicle A l u ,
ar(~d)"v)Q. (3-9.48)
This interpretation &o supplies the value sf the intepal :
(3-9.49)

Within this picture of prescribed motion it is cornistent to take

Indeed, the consewation law is satisfied,

6(2 -- S(&)) = 0, (3-9.51)

and
(3-9.52)

I n transferring these results to the connection bekvveerz action and stress Lensor,

one must not forget the meaning of 6s,(s). It arose m a generalization, of the
~ g i displtaeements
d given ta soureefri, which were inbnded to simulak $he &h-
placement of a referexlee aurface and are therefore in the opposite mnse. Thus,
when tranglating h t o the motion of point particles a minus sign must be s&ed:

It is now necessary to generalize the identification of ds with dzo, in the rest

frame, to the invariant proper time definitian
-(ds)2 = dZY dzy. (S9.55)
Its consequence for a vafiedt molian,
--ds 6ds = dzVdsz,,
eonvert;s (3-9.54) into
3-9 Chrrgs quantixation, Mass normatization 247

and supgies the action expression for a single particle, labeled a, performing a
preserihd motion,
(3-9.58)

The phenomenological orientation of source theory has the folfotving eorol-

lary, Physical parameters identified under restrickd physical circumstances do
not change their meaning when a wider elsss of phenomena is eonsidered. The
mass parameter m, is determined from the response of the particle to weak,
slowly varying, prescribed forces as in beam deflection experiments. When
eleclrornagxtetic interactions among several particles are considered, this param-
eter i s not assigned a different value. It has already been fixed, normalized, by
experiment. Thug the single-particle term (3-9,44) mast not be added to
(3-9.58), thereby changing the value of m,. There is no question here of assign?
ing some fraction of the total mass to an electromagnetic origin, What is a t
issue is the consistency bet\veen the various levels of dyrlamicat description
through kvkich one passes in the course of the evolution of the theory. The
prescribed forces of the most elementary level become assigned to the motion
of particles a t the next stage, but in neither one is there any reference to individ-
ual particle structure and the phenoxnenoEagicaf parameter m, must be common
to both. The eonclusiorl is that the real zt*,(X) krms, which contribute neither
to the vacuum persistence probability nor to the couplings among sources, must
h struck out. Here, then, is the action to be assaciakd rr-ith a point charge
realization of photon sources:
W = Lim [W(X) -- w,(X)].
h-90 Q

Consider again the eRect of a source displacement, now pictured through

the motion of point charges, We use (3-9,28), but tvitl-i attention to the X dis-
placement and the minus sign required to translak cfizY(;lr)irfto 82:(5):

x f,(z.(s) - I&(s~)
f X) dZbr'S"
---
ds"
-C
a
4W,(X)

The antisymmetrical product of two vector displacernexlts defines a two-dimen-

sional element of area,
4xf dz; - 6~:: dx: = dcz", (3-9.61)
and the antisymmetrical product of three displacementa produces a thme-
dimensional volume element, or the equivalent directed sudace eXement for
the coordinates zz - z,;
d *C:' dzb, = d ~ t b . (3-9.62)
 Chap, 3

The corresponding pre~entationof (3-9.W)

4 d@:'[e,F,,(z. f X) S 'e, *F,,(z. f X)]

- *e,es)
1

-C @a de:afN(x. - ~b f X) - C 6wa(X)
tab a &"+d

is no Ionger limikd to infinilesimal displaeemenfs; the integrhion~extend over

the gt?ometrical domains defined by the initial and find p b i c l e trajectofies,
Given the various three-dimmsiond 8z;zdwes thezt occur in (&9.63), all the
individual fIr intepals can be made to vanish by 8pprop~stechoice of the $"
support, which n d not be restricted to straip;ht lines, For any other election
of j'p that Il;lves nonvanishing values to one or more of the inkaals, those valuw
must be confin& to irtbgral multiples of 27r. Consider a pair of pafiieles a and b,
f'ar which the three dimensional surface cr fhat i8 t r a e d out by X: - XI: is
eEectivefy displacctd by &Xp. We desimatti! these surfmes by @(&h) and wfite
the condition guarankeing physieal uniquenem

where n is an integer. In order to ensure thabonphysieal etennmb do not inter-

vene during the limiting process ?P--+ 0, we demand that this hold for aImosG
d l X@. The scale of p iis fixed by the diEerenLiaI equation (3-8-62), or the equha-
lent integral statement
(3-9 "65)

referring Lo 8ny surface that encloses the origin. The diserekne8s required by
ko a f i ~ k
(3-9,64) implies thak the suppod offp on any such sadace i~ ca~lfin~d.
number of points. And, in virtue of the qnnnnetry properLy (3-8,72),

fhat number must be an even integer, 2v. We may visualize $his number of
filaments drawn out from the o ~ $ nin a way $had assims to e a ~ hfilament its
image in the origin, Let the contribution to the sudam integral (3-9.65) th& i s
~~aciaLt?d with an individual paint a,a! .- 1, , , . , 2 ~ be
, designs64 r, so that

The basic 8ilua;dian far (3-9.M) is that @ ( X ) , for example, incfudes a single point
a,while @(-X) contains no support; point of p, Then

and the addition of sueh, sxpresaiong repmsents any other possibility. Xn. par-
3-9 G h s r ~ quentization.
s Mass normalixat4on 249

tieulnr, the summation over all a = 1, . . . , 2v gives

or, making explicit that the paints of support oeeur in pairs with equa! values
of r, and n,,

we get the charge quantization condition

Xote that the kveight faetom r, take the rational form

If all 2v points are equivalent, r, = 1/ ( 2 ~ 1 and

, the integer nab is an inkgral
multiple of v, The simplest possibility, v = 1, is illustrated in the $" funetion
of (3-8.70).
With the suceess in removing the arbitrary aspects of 6W through the re~o11;-
nition that only exp[iWf is significant, we can present (3-9.63) effectively as
4 dfl~'[eaFL".(~.
f X) -k *e, *F,,(x. f X)) - a
&W.(&)

This might seem to pose anotiier problem, hosvever, Although WE: retain the
symbol $W, it is no longer the change of a quantity W and the question of
uniqueness arises. Consider rz corltinuous deformation of the trajectories that
finally retur~isthem to the initial eonfigun-ttion, thereby defining a surface
exlelotjing s three-dimensiond voIume. As the covariant generafizatiorr of the
three-dimensional reletioxl

(3-9.75)
and similarly
4 do" *F,, = -- (3-9.76)

The net change of W an completing this circuit is, therefore,

tvhere @,(&X) indicates the three-dimensional volume, associated with padicfe a,

~vhicahis subjected to tire alternative space-like displacemtfnts &XL". The integrals
250 Fields Chap, 3

of (3-9.77) record the amounts of electric and magnetic charge within the various
volumes, Here the basic situation occurs when particle b lies within the volume
@@(X), for example, but is outside of a,(-X). The associated contribution to
d W is *(e, *eb - *e,eb), a multiple of 27r according to (3-9.69). This aammation
of the single-vafuedness of expfz'Wfwas inevitable; it was only of some interest
to see how the charge quantization condition brought it about.
The charge quantization demanded by magnetic charge provides a most
satisfying explanation for one of the more striking empirical regularities in
nature. Uespik the widest variation in ather proper-t,ies possessed by partieles,
the magnitude of the unit of pure eleetric charge is universaf. It is measured by
the fine structure constant
a == e2/4a 1=: 1/137.036. (3-9.75)
If we Msume that the smallest magnetie charge magnitude, "eo, eorresponds t o
the smallest; integer in (3-9-72), the latter becomes

and

This is very large indeed, being the equivalent of the electrie charge 2(f 37)e.
However, one might think, if only for a moment, that this great asyntnniely
could be apparent since there is the freedom to redefine; dl eleetrie and magnetic
charges by the rotation of (3-8.47) :
+
eh = e, cos rp *G, sin p, 'eh = -e, sin p f *e. cos p. (3-9.81)
Of course, there are invariants of this rotation in the two-dimensional charge
space, including
+
6 * e t , e, *eh - *e,eb, (3-9-82)
which correspond geometrically to lengths and angles Between two-dimensional
vectors. 14fso relevant is the inequality
(e, *eb - *e,eb12 2 (ez + *ez)(ei +
Now consider the following invariant slatemend. For all known p~;t"ticles,
(ei f * e , 2 ) / 4 ~is small compared to unity. Comparisoll of the inequality (3-9.83)
with the charge qusntization condition (3-9.71) then shows that the integers
n a b must all be zero. The corresponding points with coordinates e,, *e, are
confined to a single line, which thus acquires an sbsoluk significance. It is
conventional to identify that line with the axis of pure e1eet;rie charge. The
complete l-eduction of the line to equally spaced points demands the existence
+
of a. second class of particles for which (e.2 *e:)/4r is large compsred to unity.
Among such particles there is no necessity for an. absolute charge line although,
3-9 Charge qusnfization. Mass normalization 261

if the integers of the charge quantixation condition assume only moderate values,
the charge points will clusbr near a, line, which is the conventional axis of pure
magnetic eharge..
It is remarkable that we have been led to the existence of two types of
charged particles that are characterieed internally by relatively weak and reIa-
dively strong forces, for this corresponds to the empirical distinction between
leptons snd hadrons, respectively. Certainly hadrons-mesons and baryons-
are? not magnetically charged particles, nor do their interactions possess a
strength as great as (3-9.80). Rather, we view them as magrretica1l.y neutral.
eornposites of particles that carry both eleet~icand magnetic charges, with the
observed strong interactions of hadrons emergir~gas residuals of the considerably
stronger magnetic forees, lvhich thus far have successfully prevented the experi-
mental recognitioxr of free magnetic charge, I t is essential far this explanation
that a magnetically neutral composite appear as an ordinary electrical particle,
If we have a group of particles wit11 charges e,, "c, such that

the comparison with a reference pttdicle of charges eo, *eo gives

l
--
4%- a
X(e, *eo -- *gago) = 2 E n.~
a
(3-9.85)

This is the required charge relation,

The automatic appearance of conventional electrical behavior for a mag-

netically neutral composite is significant because the individual electric charges
on parficles that carry both. ty pes of chargwdual charged particles-can assume
uncanventional values, We make the specific assumption that the smallest
magnetic charge, "eo, is found on a dual charged particle with accompanying
electric charge eo f O [the value of eo \vas irrelevant in (3-9.85)f. For any other
set of dual charges, e;, "eh, refererice tn the unit of pure electric charge s h o w
that *e(; is a multiple of *eo,
*eh =: *%l (3-9.88)
md the application of the charge quantization condition to the p8ir of d u d
charged particles gives
T h i ~exhibits eo and e W independent units in 1% two-dimensional lattice that
produces all possible electric ehargea. Since m nnefilsures magnetic eharge, in
units of *eo,we &gabrmognize that a, magnetically neutrd eamposik is m t ~ c b d
ta t: m a charge unit. It also follows thak electric charge digemnms, for a eommon
value of magnetic charge, me confind to multiples of e.
The discurnion of electrical pa&icles and of dual ehargd gadicles ntafurally
~ u ~ e the s bcon~ideratianof purely magnetic partier@. The unit of pum mag-
negic charge, *e, must be an inkgral multiple of the sma1lesL magnetic @barge,
;ELS in (s9.88). We write this s p i f i c relation in terms of an inhger N,

The ~bnalopeof (3-9.79)1 connecting the unit, of pure eleetric charge with the
~mPtflmtm ~ p e t i echarge, is the following connection between the unit of pure
mwnetic charge and Ithe smaliiegt elwtFic charge:

From our various tzssunrpdions, which are grounded in the symmetry. hlwws
electric and magnetic charge, we have inferred $hat the charge units on 8 d u d
char@ padicle are the same fracLion, 1/N, of the uniks of pure e l e c t ~ cand
magnetic c h a ~ e . Among the possibilities, 2, 3, . . . , which value hw nature
elected for the integer N ?
But fir& we musk digess $0 discuss the relation b t s v a n the tati is ties of
eompwitc3:petrticlm and their constituents. One approach uses the spin-statiskics
wnneetiam. A compasitc?b m e d of an odd number of particles with inbger 4 +
spin (F.D. 8%&ti~tiesf h= a regultant spin anguIar momentum thatt is also
inbger "f- &. This eamposite particle o b y s F.D. sta%istics,I f there are an even
+
n u m b r of eonstituent particle8 with inbger 3 spin, the compsite p
inbgral spin and is a B.E. particle, It is as though a F.D, (B.E.) particle @&fie@
a, nninus (plus) sip and these ;9iws are mdtiplied to give the statistics of s
composite stm~ture. This is more than s mnemonic, far the $us and d n u s
signs identify the dgebmie propedies of the individual m u m 8 &at are m m -
pfict-d to produce the eEeetive sourge of the composite sysbm. Now, as we have
mentioned, there are two varieties of hdrong; mesons, which are B.E. parkiclm,
and baryons, w h i ~ h&reXi".f). particles, If bLh dyps of hadrons are to km con-
~ t m c t e dm mametically neutral eornposites of dual ehargd pa&ieles, the lattctr
eannot all be BB. particles, The simplest msumptioa is that they are all F.D.
partides; ~n wen number of such constituentss p d w m a B.E. particle, an d d
number builds a F.D. particle.
Cm the dud charged particles exhibit only one 8trength of magnetic charg~?
N o b %hatboth sign8 of the magaetic charge, linked to sign ehange~in e l m t ~ e
charge, will occur. This is the antiparticle concept, with both charge^ involved
in order to maintain the structure of the two @etaof Maxwell's equaliom, which.
ha-ve the field stren@h bnsor in common. If the only values of magnetic ebarge
are fl/N) "e and -(l/N) "e, they must be combined to produce a neutral
composite, and sueh p a h of F.D, d u d charged padicles are B.E. particles;
bavons cannot be manufactured in this way. Eence there must be at lemt two
digerent eharge magnitudes, According to the magnetic analowe of the elmtrie
lattice cowtruelion (%9.90), the mametie charges on dual charged padicles
with the same electrie charge must differ by a multiple of *e, the unit of pure
magnetic charge. It would seem to be a rertsonable charachrieation of dual
charged pa~iclesto describe them as carving charges that artit smaller in magni-
tude than the uniLs of pure charge. If that is granted, just two values of mag-
netic eharge are admitted, With a conventiond sign choice, they are --(X/N) *e
and [CN - l )f N ] *e. The possibje values of electric cha~grjwe analogoug:
-(l/N)e and [ ( N ---- 1)/Nje. Either electric charge can. be assigned to either
efioiee of magnetic charge, giving four dual charge combinations, although them
may be duplieatioxls of these assignmenb.
In ezddiLion to neutralizing a magnetic charge by its negative, which builds
a maon, we can now balance .the mapetic charge f ( N - IL)/RT] *.a against N - l
units of the magnetic eharge -( l / N ) *e. This is a composite of N F,D, particle@,
.
and N = 2,3, . . must be odd if the resuit is fo be a F.D. baryon. The simples-1;
possibility, which we adopt, i s N -. 3, Thus, bsryons are v i w d ~ZSe o x n p ~ ~ i h
of three entiLies that bear the magnetic charges, in. *e units, af 3, ---*,
-9, We
learn, incidexllally, from *e = 3 *ea, that

It remains undeeidd whether the two magnetic charges of -* refer to dupli-

cates of the same particle, or to BiRerend particles with s common value of
magnetic charge. To this we can only offer the observation that, withoul
reference to antiparticles, the magnetic charge average over all distinct dual
charged particles will not be zero in the first possibility, but does vanish in the
second one wbercr eharge -g has twiee the multiplicity of charge 8. We accept
the situation of greater synrmetq, and extend it to electric eharge as well. Thus,
whether we speak of electric charge in units of e or msgneLic & a r e in unik
of "e, there are three options with values g, -9, -4, I t is natural to regard
these nine possibilities as differ& slates of a fundzbrnerztal dual eharged particle.
To emphasize its basic dyadic eharwter in regard to charge, this pa&icIe is
called the dyon,
Although the hypothetical picture of magnetic charge ills the bmis of hadroaie
behavior is still quite incomplete, we haye alredy far outrun. our ability to test
it, particularly rsince et quantitat;ive phenommologiical analysis of the properties
ol hadrons is not yet before us3. We must turn away from these heady s p e d a -
P54 Fields Chap, 3

tiom and bgin the study of ordinaq elwtgeal pa&ides in dynsmied eontexte.
Hawever, Harold finds ffimwIf compelle-d h eomment.
W.: You were quite pemuaeive wncerniw the imporlanm of svoi&ng
spculakive wumgtioxls about the stmctwe of partiefw, and yet you have just
e n k ~ i n ad very bold spe~ula$ioxlinded. Is %hisnot incomishnt?
S.: The final goal of rt phenomexlollogieaf theoq is fo mtabllish contact wigh
an underlying fundamental,Lheov. My injuxlction was agaiwk the conhsion of
phenomexlolo@cal theory with fundabmental bhwv. The organiaatio~and
$hearet,icaf simplification of ex~fimentaldab should not involve impEici"c
stmaturd assumptians. But, quite? independen%$ of Lhst develiopment, o m
may de;vim speealative candidztks for evmtual contact with %hephenomenolagt
~ a$heory,
l fifltim&te~uccessssboulld be spedttd through the la@cal aeparstion.
of these two phwes.

&'IQ PRIMITIVE ELECTRQMAGMETIC INTERACTIONS AND SOURCE MODELS

The comemed nature of the photan eleed~esource JP(%) sets the patbrn for
any realkation of ~ u e hsource8 by an deetfie current metar wociated with
~pecifictype of particle. The electric currenf-a that we h&vealredy comidered
for v a ~ o u sspin choices fail ta meet this standard since they are conserved only
oubido soume rM0n.s. Let us rope& that discussion for spinlem pahicfes, using
%heslightly diRerent procedure thiat is b on (;fie ae%ionexpremion

The eortsideration. of infrinitessimd, variable p h a s tramfornations of the sources:

and of the compensating field tsansfarmst;ions

with

The eompafism of the two evaluations implies that

Natjea .that we have written q eveqwbere, replacing the ehtsrge matrix q of

&heear&ertreatment, in order to memure charge in %hephygieaf unit e,
3-1 0 Electromagnatie interactions- s ~ u r models
~e 265

The observation that jp is not conserved in the inkrior of sourees means

only that the physi~~kl d-~?~~riptionbegins with the creation of the charge-bea~xlg
particle and ignores the pre-exisbnee of that, charge? if nod Lhrs particle, in the
aouree. We must find a way to insert f he fact that charge is transmi-t;ted, nod
created. I t will be seen that this requires the indroduetion of an, electromagnetic
model of sources, which is simplified to the point of retaining only the charge
con~rvationproperty, but still has arbitrary elements. One prace-dure e r e ~ t m
a conserved electric current by smputt.ttingthe nonconserved part, in a wrty %ha$
retains the ari@nal current in the regions that are cmally separated from the
emission and absorption acts, where the current is eonsemed. This is accom-
plished by the construction

where
d&fF(rz:- X') = &(s - X')
defines a, no$ unfamiliar class of functions. When the support of f@(z- 2') is
restricted to space-like intervals, the subtracted term in (3-10-8) vani~hesad
any time for which the sources are esusaly inoperative, To keep uniformity of
treatment between jCI(z) and JP(z),we shall relate the canwrved vector, now
designated J:a,,, (X), to an arbitrary vectorial. function J p ( z )by

The vector potential Ap(z) must multiply the total current, in the action
expression. That can be rearranged to give

in which
A',(z) = A,(z) - a, - xj) A ,(x",
(dzk>fp(z (3-10.12)

and where, for convenience, we have accepted the symmetry restriction.

which ha8 no apparent pbyslcaf sigrrifiesnee here. Note that the construction. of
A:(%) from A,(s) is a gauge transformation, such that the new vector potential
i a charaekriaed by
(dzt)f"(z - z f )A:(.') = 0. (3-10.14)

This is a unique characterization, for, if the general gauge transformation

26B Fields Chap. 3

is designed to make X,(%) satisfy (3-10.14), we get

which produces thc?construction of Eq. (3-10.12).

When, two digereat kinds of pa&iclw lkre @ansideredundm physical condi-
tions af noninter&ation,the vmuum gmplitudm are multiplied and the actiom
atxtded, Thus, for noninteracti~gphotons and spinless pa&icle;s,

An interation. Ibetwczen photons and charged pet.rticles is introduced by r e

placing J&,, with the total current. We call this inforaetion primitive beeaustz
it is not the final gtatement of d l inhraedioms, but rather charackrizes a, first
elementary stage, which implie8 and is supplemenbd by further, increasingly
elaborate levels of descniption. Pmeisely in wb;af sonso it is the first of a s e ~ e of
s
dynarnlcal s k p s will be discussed later. The action expression th& e h m l e r k a
this first stage is

where we have chosen to incarparate into the Lagsange funetion %heinter-

action k r m
jr (x)A:(z) = +'(z)i@q+(z)A,(z). (3-10.29)
Although the Lagrange function here employs the vector pokntial of a specific
gauge, it is a g ~ u g einvariant combination that remains unchanged under the
unified gauge and phase transformation

This is a con~quemwof replacing a,+, with transfolemrttion behavior

by %hega,uge eovdsnL combination

(a, - ieqA,(z))+(z) --t ege'"'" (a, - i e p ~(z))+(z).
, (5-10.~)
The field, equations deduaed from the sLsLionay wtion principle by varying
@ and rf, are, re~petiv?rfy,
where the gauge covariant wmbination stays intact since the sign rever~alOf
the derivative on partial integatian is matehed by the antisymmetry of the
charge matrix g. In performing the variation of A: we must not violate the
gauge restriction on the veetar pokntial,

Thus, the correct conelusion from

where ?(S) is arbitrary as far ss the action principle is concerned, But tha
divergence of this equation gives
a,dP(x) +-a,ji"(z) = ?(X), (3-10.27)
and .tve recognize the MaxweEl equation

Ta connect the use of f@(z- z') in defining s specific gauge with the concept
of electromagneticsource modele, we perform the fdlowillg phase transformation
on 4 and cap, without the accompanying gauge transformation:
e-ieqA(zj
+ @-ie@Acz)
S
+P(z) 4
c35P f d ? (3- 10‘B)
where

and A,(rt;) is the veefor potential in an a r b i t r a ~gauge. 'This transformation

does two things. I t replaces A: in b: by
~ : ( z )4- apA(z) = A,(%), (3- 10.31)
which is the inverse of the gauge transformation (3-10.12), and the transfer81
of the uncornpensakd phase factor to the saurecs replaces them by
,n(x)=eie~b'z)~(2), K , A ( ~ ) = ~ ~ ~ ~ (~ z( +) ) K (3-10.32)
With the introduction of the arbitrary veetor ptential A,(z), we return ta the
uut; of Jta,,(z), The additional label will be omilted, however, for one can
understand from the cantext %.hether JP($) is sn arbitrary vector, since the
vector potential is limit4 to a particular gauge, sr is a conserved vector, since
the vector potential admits gauge transformations. The new sction expression is

The gauge invariance of the Lagrange function is now matched by that of all
the souret, terms, since A, -+ A, +- 4X
), induces
A(.) + A(z) "4- &(X) (3-10.34)
and
KA + eiegh(l) (z), K: (z) -t eiegh'"' K A, ) (3-10.35)

While the charged partide field equations that are implied by the action
(3-10.33) continue to be given by (3-10.23) with the sources K", K:, the eleetro-
magnetic field equation. presents ab different aspect. In contrast with the action
of Eq, (3-10,XS), &A, is arbiLrae md the charged padiele sources are furrc-
tionds of the vector pakntial. The implieation of the latter fiaet is indictzkd by
z ) (dz)4 (z)iepxA( X ) &A(2)
( d z ) + ( z ) ~ ~ (=

(3-10.36)
Thus we ROW get
4-jP(lz)
= JP(%)
$,PP(z) - (X') f &(z')ieq~: (%')l.
(dz8)fp(z- z') [4(zt)iep~"
(3-10-37)
1%iss just the Maxwellt equa;tion of (3-10.28), since

but this time we have m8de explicit a, contribulion to the electric cument that is
associated directly with the ehargd pareiele source,
Cowider Lhs fallowing fictitious source problem: A point charge e moves
uniformly with four-vector velocity %l^,

until at a given location., which we aidapt as $he o ~ g nit, g;oes out of exisknee.
Whad is the description of the phobns emitbd or absorbed by this act"i'he
curred vector is given by

it obeys the nonconfiewaition equation

where &his@f, function is time-like,

3-1 0 EIectromagnetic interactiona- source models 259

and has the momentum representation [cf. Eq. (3-8,61)]

if' ( p ) = -npi - nE"/pn,
dse-i'pn - pn 3C5 0. (3-10.44)

Recall the description of the emission and absorption of an arbitrary number of

particles, here photons, by a given source distribution, J B ( z ) . The factor in the
vacuum amplitude that couples J P to the creation and detection sources, J ;
and J:, respectively, is

[/
exp i (dx)(dx')J';( X ) D+($ - zt)J,(z) + i/ (dx)(dz')J p ( x )D+(x - x ' ) J ~(X')]
,

where A.,(x) combines the field associated with J $ and the initial photons with
that having analogous reference to the final photons. Xn view of the causzal.
arrangement of sources, wherever A,(x) is of interest in (3-10.45), it is a solution
of the source-free Maxwell equations or, in momentum space,

If we insert the current of (3-10.42) into (3-10.45) it becomes

But, observe that

which shows the equivdence, for the purpose of evaluating (3-10.471, of the
time-like jp function with the space-Iike

The latter is also an odd function of p without restriction, unlike (3-10.44)

which mirrors the asymmetry of the coordinate function in (3-10.43). We
recognize in (3-10.47) precisely the exponential factor that is associated with a
single charged particle emission act, as in
/ (dz')4(x')lCA (X') = (dx')+(X') exp [- iep/ ( d z ) f p(X - X') A, (X)] K ( z r ) ,

where x' serves as the reference point at which charge eq disappears in the source
and emerges on the particle of interest.
The members of the class of fC"functions given in (3-10.49) differ only in
the choice of the tirne-like unit vector nC",which represents the motion of the
280 Fields Chap. 3

charge in the source model. When fhe coordinate system identifies with the
time axis, f (p) has only spatial components that are independent of p@,

and

where

There is one choice off@that avoids the reference to an external unit veetor by
devising the latter from relevant physical. paramekm. It requires an exknsion.
of the structure sf f @fa include akebraic funetions of derivatives that act upon
the source funetion K(2). We indicate this repfacement in (3-10.50) and
describe its meaning by writing

(d2)fC1(z- S', P)A,(s) K ( P ) . (3-10.55)

When K f P ) reprwentx the emission ar absorption of particle^, the timelike

vector P@ repIa@ apart from a scale faetor. Thia gives

where the Imt form is the analowe of (%10.44), one that is equiv81ent for the
ea;tleulation of phofan processes*
The discussion of spinless particlm is pa&ieufarly. simple. A rsysbnr without(
intrinsic angulm momentum ewn anEy exhibit scalar properties in ita rmt frame.
In the electromagnetic conkxd this permits manopole momeat-ch
forbids multiple moments, More generftfly, a particle of spin s, in its mg&frame,
@an possess multipole momenk ta the rnrtximunl order 28, That is, a spin 3
pa&iele can have arbitrary dipale moments; a, particle of u ~ spin t can have
arbitrary dipole and quadrupale msmenk; and so forlh. A sufi~ientlygeneral
eument expre~ionfor spin $ is

This way of writing the eoeBeient in the term thwt hras the form d , d ' antieipab~
the _identification of g as the wrclmsgnetie ratio, the? magnetic moment in tbe
unit &e/2m relative do the spin angular momentum fEq. (1-2.4)). Th8t be-
comes clearer on. wing the i;dlen.tity (3-6.67), applicable in source-free regions, to
remite (3-1 0.57) ara

The magnetic moment of a system i%

which here hcomes

ma&ng explicit the roles of orbital angulztr momentum, spin. a n p l a r momentum,

and the g faetor. The dipole moments permitt& to a spin -& pafticle inelude an
electric dipole moment. It would supplemen&the second term of (SXQ.53)by 9t
s d u d spin tensor
gimilap exprwion of arbitrary coefficient that u ~ e the

No such progerty has yet been detected, however. Since .the second te-m of
the cument is identictitlly divergenceless, we still have [a, factor of e is i m h d
campared to Eq. (3-6-48)]
a,jr(z) = ~ . ( z ) ~ ~(g).
ieq~ (3-X 0.62)
The currend (f.E-10.57)is ineorporai;ted in the fot10~ngslcti~nexprwion,
an8logous do (3- 10,Is),

g(+,A,) = -&Fp'F,, - ++ro[rB(--$a, - epA,) 4- m]$ (3-10.63)

4- +F" 2m (-&g - l)$+~%.&*
The omitted electric dipole interaction term is sirniitar to the last one, with either
of the andisynrmetrieal tensorrs replaced, by ita drxgl. The Lap8nge funcC.ion is
invariant under the gauge transformation

and this propedy san be utilized, as in fhe spin O discug~ion,ta m m v e referenm

to a specific gauge whirs introducing an eleckrom~elicmodel for the particle
8ource :
(z) = eicpb(i)q(~). (3- 10.65)
2#2 fields Chap. 3

The fi& equations impXied by the action (3-10.63) are

ther witb the Maxwell equation8 employing the appropriate ~onmmQ?d

ewrenk.
The appe8ranw of the gauge @ovarian$derivative
aations (%$Q. 18) and (3-10.63) h completely general. f f
hgween the twa ways in whieh d e c t r i ~cutmen& b v e k n introdurnd. The f i r ~ t
one cowiders tche imfinih~imalresponm to a variable phme transformation. For
s typieal p&icle field ~ ( z this
) ig

where the hetor of e in (3-X0.a) provT_des the ntppropriab elwtroma~etic

memure far the cutmeat. This Enematied definition ia not unique. The d y n a ~ -
ad definition of electric current imitate8 the role of the phobrt wwee. In.
padiculsr, the reBponse of the action Lo the field variation BA, d, 8X is
.E.;

Thua, the identity of the two eoncepls is impo~edby imisfing that the aetion b
invariant under the unifid gaugephae tr&nsfomationwith

The replawment of desvatives on chargebeafing field8 by gauge e~vmiant

derivatives &coontpli~hes.t;hia for the whole poup of gauge drawfomna%ions,
Pvhich ia Ablian in struetm. And the possibility of adding independently gaup
hvas~bntbrms, M in (3-10,63), conveys the a r b i t r q aapeab of %heIcinennatieaI
t dt?fini$ian. It is generally believed &a;t them is something padieularly
and a;a%uralabout the ctjtectromametie coupling produed by wing ody
the gauge covariant aubgtitution, and there is gmfh in thb. But i-d magt not be
forgotbn that ttlkrnative de~criptionse G ~ for t the same spin vdue, and by
haowing a common procedure we rsr~veat diEeren%electronrametic prope&iw.
Thus, the third-rank apinor dmariplion of @pin+,b m d on the L a g ~ a g efun*
tion ($5.73) with gaum cova~antderi ,&vm the eument of Eq. (3-6.6X),
&p& from the f m b r of e, and the orrdi% g value, m eora~nedin
(W.68,69), is 8. If the very striking nem+quality, +g S 1, that is abwpvd
for the electron and $he muon has atay single moral, it ia the apmiall relevance
of &hesimple Dh&cspinor equer,Cionfor the description of them p&ielw.
 To illugtrate the direct use of a gauge i n v a ~ a nLavarrge
t function far intro-
dueing primitive electromagnelie inkraetions, we shall discuss charged partielw
of unit spin. Such a L a p a w e function, generdized from (3-5.28), is

which finally us= an abbreviation. for the gauge covari~atderivative,

Notice that we have devised two independently gauge invariant 6errns. Tha
a r b i t r a ~coefficient8 a and b will be related to mabgnefie moment and deetGe
quadrupole moment. VVe shall not consider Che two additional couplings pro-
dueed by replacing FE^" with its dual. They would describe e l e c t ~ cdipole and
magnetic quadrupole moments. The parti~Xefield equations derived from the
action principle are
D,$, - By+, -- C,, - (blmZ)(~,kieqCA,
-- F.hieqC\,) = M,,,
+
DvGpY m2+@- aFpiep.+, = JP, (3-10.73)
&ndthe electric cument vector, in source-free regions, is

Lf we are ixrtereskd in the intrinsic electramagnetie properties of the par-

kiele, and not; those induced by the electromagnetic field, it, S U E t~
G ~implify
~~
(3-10.n) with the aid of the uncoupled particle field equations:

ths lrtst of which, i8 an innpodant but not independent statement, This giva

and the implied coupling with an elechnzagnelic potential in ccsmpbtely soume-

fm regions is eonvqed by
&)f A,(ar+'ieq+v) - ft. ---. a -t- b)iF,p(sbPz"eq+')

The identity
+ (blm2)ah~,.(aE^dieg4v~1. (3-10.77)

(ap+%ieq$') = (dp#'iep$" )+ ap(+'ieq+') (3-10.78)

haws also that the field ~Lrengthderivative in (3-10.77) should ba sym-
metriad in the indices X and v.
2 Fields Chap, 3

For a slowly moving partich, of charge f e, the three field components

dominate, and are eonvenienay combined in the vector 9. The spin matrix
vector s is represented by the rotation

We use the spin matrices to present this speeiztlizatioxl. of (3-10.77) as

where the d y a d i ~"ErEis symmetrlrzed, and we have dao picked out the term^ that
deseribe the propagating particle in a crausd arrangement. WiLh Lke coupling
of the scalar potential A' to the charge fe serving as s reminder of the nomali-
gation, the linear coupling of the spin vector to the magnetic field identiifies
the g value :
g== 1 . - a + b , (3-10.81)
while the quadratic spin term @;iveg the quadrupole momexrt Q, in the unit,
(&e)/m2,as
& = 2b. (S10.82)
The idividual results obtained for g values when oaly the gauge covariant
derivative is used (s = fr, g = 2, 8; s 1, g = I), are given unifQrmXgby the
=I.

geneml nnulti~pinarLagrenge function (s5.78). The current %ha% the latker

implies in source and field-free @paceis

The vanishing of all auxiliary fields under such circumstances, as expregmd by

hplies the set of field equations

Hence, the rearrangement used for spin ean be applied to each of the n t e r m
that compose (3-I0.83), giving

She@the padiek: spin vector is

S = *C@,,
 Extended soureas. Soft photons 265

the g value is immediately identified as

and all other multipole moments are %em, Note that the actud spin value
eaters only through the inequality s C: i n , and

where the equality sign applies to totally synnmet~calspinors, Incidentally, a

very similar unified treatment applies to all gpixtar-symmetricaE tensor fields
used to describe integer++ spin values, As one can recognise from the examples
of spin 8 and l,# Lagrange function^, Eqs. (3-5.55) and (S-5.581, the gauge co-
v%riand electromagnetic interaction implies a current vector Lhat, in source
and field-free space, is

The sarne spin. 3 rearrangement, combined with. projection of o on the total

spin and the observation that (r3, for example, is unity when SS === e5, give8
directly
ys == 1. (3-10.91)

3-11 EXTENDED SOURCES, SOFT PHOTONS

Complementary to the pt-inciple of space-time uniformity is a principle of uni-
formity for phenomena that differ only in the values of energy-momentum that
are engaged. The source concept was inkoduced as an ideali~ationof collisi~ns
in. which precisely the right balantte of enerw-momentum or, invrtriantly ex-
pressed, msss is transferred to create a specific particle, But the sarne laws of
physics are operative when less mass, or more mass, is transferred. Long ago,
in Section 2-43, we used an extrapolation to quasi-statie source distribulions,
which are incapable of emitting particles, in order ta connect the properties of
photons with the Coulomb-Amp&ri%nlaws of charge and current interactions.
Perhaps in our recent preoccupation with the very familiar equations of Mm-
well, we may have forgotkn the initial logical ba&s far that contact. And now,
through our concern tt-ilh the electric currents that are assoeiztted with the
rtpwation of charged particle sources, we are moving in the opposib direction.
The physical situation is quik simple. The creation of a ebarged pvticXe gen-
erally involves the transfer af that charge from other particles having different
states of motion. Accelerhed charges rsdiate. Hence, unless precise eontrol is
exercised over the energy-momentum balance, the charged particle has t% nanzem
probability of being accompanied by photons. If we were to take too narrow ail
view of the source concept and decline to extend it to this mrxltipztrticle emission
act;, we would divorce the dynamical, significance of ehacrge from its kinematiesl
aspects.
26Q Fields Chap, 3

The emission or absorption of photons is not a foedized process. The

photon that accompanies the creation of a charged particle cannd be zts~igned
$0 the agency of that particle, nor to the charge8 in the Borne, but involve@
intedererrce between both effects, This is implicit in the dditive congtruction
of the electric current from con-t;ributionsof the pfcrtiicles and the source, It i s
im%ructiveto examine such phenomena in some detail, We bedn by evaluating
the probability amplitude for the ernig~ionof one photon of momentum k p and
potafiztztian X, accompanying the creatian of one spinless particle of rnomentunn
p" and charge fe. The physieal conLext Ghat underlies Lhe use of the primitive
interaction to compute this probability amplitude is that, between creation and
detection, particle and photon propagate under conditions of nsninteraetion,
Aeeordingly, it is useful to review the description of that situation when the
two particles are produced by independent sources. This is contained in the
vacuum amplitude
( O + ~ O - ) . ' ~ = (O+\O-) J ( ~ + j ~ - ) K (3-1 1-1)
the k r m involvimg one emission and one absorption source of each kind (we
place K" = O im. these considerations),

where we have used p as an aI.ternative to zl" for assistance in diskinwishing

between the two kinds of particles. (And let us hope that no confusion. results
from speafring of particle, in the singular, when we mean charged particle.)
The application of the primitive inbraetion will retain the noninteraction GO&
b x t but replace the independent sources Jg((), K 2 ( z ) by a joint soume,
J$(6)K2(5) which we now exhibit.
The restriction to the single action of a photon detection source can be intr*
dueect. by considering
(3-1 1.4)
Qr
(3-11.5)

for one csn identify $be probe source 6J" with JT. Since the field A,(4) is to be
evslunkd for $,(E) = Q, it is given by
apart from an irrelevant gauge term. The process in which m are interested
involves the @%us&coupling of three sources: J"; KK1, snd &. Tbe emission
source K2 is u s 4 to inject into the system the mornmtunn P p that, is redized
m two particles,
P" = F-+-k@, (3- l l.7)
where

Thus

This sowce is aperating in the extended sense, and we shall urn the designation
'extended source' to distinguish its mode of action from that of KIPwhich detects
the partide by absorbing mass m. A souree uLili~edin that way, performing
only its initial mission, is a 'simple source.' Now, the current of Eq. (3-11.6) is a
quadratic functional af the particle source and therefore ~ontainsa porkion
fiz(atthat is bilinear in K 1 and K 2 . ~ h igives
i a factor on the righehand side
of (3-1 1.5) thaL a'Iredy has; the required three sources, All ather te
different processes than the one af intertlst, whkk is displayed m

The relevant c u m n t stmcture, obtained front Eqs. (3-10.37) and (3--IQ.@),

with K@== 0,is

The omission of %nother f erm involving K l ( f "ieq&z (t") expresses the caustll
tzrrangement, The field + 2 ( ~ i~
) related to iits source by

or, in momentum space,

+
The fact that P2 m 2 # O [Eq. (3-11.9)) means that the field 4z(z) has no
propagation efiaraete~sdics,and is localised in the neighborhood of the source
Kz(z), Thus the field cba(z) will have no overlrtp with a sufieiently remote
detee-tdion. source Kl(x), which is the assumed causal situation. The term
'virtual particle' is used to extrapolate ordinary particle concepts to such
sihations where the energy-momentum balance is not suitable to the creEtlion
of a 'real' particle. With our new terminology we can characterize the content
268 Fields Chap. 3

of (3-11.11, 12) by saying that the extended source may emit a virtual particle
which quickly is transformed or decays into a real particle and a (real) photon,
or it may emit both final particles in one act, although the photon originates
a t a different point than the particle.
The precise meaning of these phrases is conveyed, on comparing (3-11.1 1, 12)
with (3-11.3), by

where the first derivative refers to the X' coordinates. An equivalent momentum
version, which also introduces (3-11.14), is

In this form it is easy to verify the conservation property

left.
k * ~ z ~ ( W n ( P ) = 0, (3-11.17)
which is valid for p2 + m2 = 0 and arbitrary k2:

An important simplification appears when one considers "soft" photons,

those for which energy and momentum are negligibly small compared to the
values associated with the particle. Then (3-11.16) can be written as

in which we have also introduced the form (3-10.44) for j,(k). The interpreta-
tion is clear. From the viewpoint of the soft photon, the charge eq has made an
instantaneous transition from uniform motion with velocity n, to uniform
motion with velocity p,/m. This is expressed by the photon emission source

which is the transform of the conserved electric current

Notice how the two contributions, one associated with the particle source, the
other with the particle, are fitted together in an equivalent photon source. This
is an illustration of the self-consistency that is demanded of the source concept.
The source is introduced as an idealization of realistic dynamical processes.
3-? 1 Extsndd raurcssr. Soft photons 2@

The dynsmieal theory that ia erected on this foundation must, under appropriate
m~t~cfiong, validate iLs ~tartingp i n t , Thus we learn, not s u ~ ~ s i w l %ha$
y,
the aimpie photon murce dewnption becomes wficable to w realistie syr~tenn
when there i8 ne@igible re~.(tionassociated d t h the rsmimian or abmrption
procem*
We should a h recognize the phpical significance of the cavarianf f , fune-
tion &ven in (3-10.56), which we now \$?ribas

where the Ifitkr version refers to soft photons. The eEwtive phof;on source
vanishes; the ehotrge hw not changed vejoeity and doe8 not r d i a h , This i~
the most natural Csf 80ur~emdels, in which the ernittd particle dekrminm
the velocity of the charge in, the murce and thereby supp the a~comptlnying
radiation. That mppre~ionis not limiM to mft photons, however, I f we imrf
the unapproximated version of ifp(k, P) in (3-11.16), it becomes (kg = 0):

The prabability amplitude for the emission of the two particltts labX1ed kX, pq
requires, beyond (3-11.23), the additional factors (dwk)'I2 and (dw,)li2, together
with the explicit slection of charge h e ttnd the photon pofari%a&ionX- The?
latter is produced by scalar multiplication with the polarisstion vector et:, and

There is anothm point that ean be illustrakd by the eBective wurce

(3-11 1.16). Equiv~lentta a pmicle source mde1 charaeterizr4 by fH(k) is the
mignmenf of elect~calpmpertim only ta the padiele, cornbind with the use
of v m b r paknLiEEX~in a apeeifie gauge such that [Eq. (3-1Q.14)]

The veebr pfentirtl thsf represents the emitM pfiobn i s proportional to the
polarisation vector cif, and the gauge condition (3-11.25) demands that
f r ( k > & ~= 0, (3- 1X . 26)
which 8~pPfemnts(3-11.24). Thus, with the Ghoice af I,(&) that is display&
in (3-1 0.49) we have
n,dr = 0, (3-1 1.27)
and this becomes c$& = O in the appropriate coordinate frame. The significant
obwmwtion i8 that, an mdtiplying (3-1 l . 16) by one of t h w plarizr&tionvmbrs,
270 Fields Chap, 3

Xt is pomible to remove the limitation to single photon emission, at lemb

wktln atkntion is confined to soft photona* Since there is still only one papticle
deketion source, we change tttczties and use

in which 6K(z) -4 K l ( x ) and +(z) is related to the aowee Kz(s)by the field
eqtltttiarrs
-(a, - iepA,(z))@@(z)+ m2+(g) = K;'(%), (a, - ieA,(z))@(zf = +&(z).
(3-1I.W)
The elimination af &, gives the ~econd-orderdifferentid equation

Since both p;a&iele sources already appear in (&X1.29), the clws of proeeams wile
wish -t;o aelect are exhibited by

where the notation emphasizes the dependence of .titre parti~fefield +z(z) upon
the veetar pokntial tf?(l)that represents the emitted photons in mlation ts
their deteetion souree JVfE ).

Let us fir& recovw the known ~inglephoton, result; in this new way. For
this we need the part of @i1(x) that is linear in the vector potential. The field
equation (3- l 11.3 I) retains just that amaunt of informa"cion when; it is simplifie$ to

We get wh& is requird in (3-1 1.32) by multtiplying this field equation by

Qil ( z )and integrating :

Tfie first term on the right represents thtl rdiationless enzission. of the particle,
and the geeond one reproduces (3-11.1 l). The nth hrna of the power series
expansion of @ $ ( X ) in A'(t) describes %photon emission processes. If we agree
ta consider only soft photons, all such processes can be combined in. t l ~C O R R P ~ L G ~
farnub which, as we would now expect, is equivale-nt to w photon sotlrcjr?
dmeriptian.
3-f1 Extended saurces, Soft phatone 273

The dihrential equation (3-1 1.31) is formally solved by

(dzt)A$(z,X') exp (dglfF(z" --)A,([) Kz(z" ), (3-1 X ,352

where the Green's function A$(%, z') obey8

[-- (a - i e p (2))
~ + m2]A:(z, X') = 6(2 -- z'). (3-11.36)
We introduce the following transformation :

LZ$(Z? 2') = exp (3-1 1.37)

in which the integration path is a straight fine canneeding x and x@,,as parame-
t;~zedby

This tr%nsformationinduces a gauge transformation on A,, replacing it with

and giving the new Green" function equation

The identity

produces the gauge invariant caastrucdion

This vector pokntial has two ather ~ignificantpropertiers, fn regions; far from
the eleetrornagnetie sour-ce J v( t),

and, general1y,
(Z - z')PA;(z) = 0.
Hence, if we were to begin a construction of &$(x, z') ss a power series in A:,
reprwenting photon fields far from their d e b e t i ~ nSource, the initial hrm
272 FIelzls Chap. 3

would be obtained from

(-aa + m )A+ (z,
2 A"
X') = 6(2 - zt) + 2epAL (l/i)apb+(z
(3) -- z') + * .
(3- 11-45)
But &+(z - z'), being an invariant funetion, depends only upon (z- z '
)
,
and its gradient is a multiple of the veetor (z -- .
'
)
z We learn that h$(%, z
')
has no term linear in AL.
More can be mid; if the field strengths %retreahd as harnogenmus, is
appmpriab to goft photons, of negligible momenta. Then

which implies the tran~laLiona1invariance of the Green's function,

A$(z, z') = h$(% - zf),
and the digerential equation (3-1 1.36) becomes

The anll;ujtar momentum structure of the linear field streneh, trsrm assures it;s
commutstivity with a2; it also eommutea with the qusdnttic combination of
coordinate8:
[tF@r(g,a,- %,a,), ~ z ~ F :=~ Pz ~F] ~ ~ S ~
= (3-1 1.50)
since
= F,.F""~, (3-1 13 1 )
is an. antisymmetrical function of p and v. AI1 this, and the rotationd invariance
of 6(2), shows that the differential equation. (3-11.48) can be sinnplifid Lo

We shall not stop now to solve the above equation, I t suffices to know that
~ T (--z X') is an even function of field strengths, for this means that the field
depndence of bhe latter funetion can be neglected relative to its partner in
(3-1 1-37), since, earnpared La veetor potentids, field strengths contain an
additions1 photon momentum factor.
Introducing the8e soft photon simplifications, wrearrive ttt

The straight lim inkgral that occurs here bgins a t %bandmoves, in 8 dimtion
ned by the vector (z - g':")@%ward
, an effeeCiv~!lyinfinikly dista;nd point,
3-1 f Extandad sourosar. Soft photons 273

~incethe photon emieion processes are localized near the extended s o m e K%,
And, if the eoupling betwwn the padicle sources iis to be appreciable, the g*
me$~caldisplacement (;e - x')" muat coincide cIowly in direction with that of
&hemomentum vee-t;or of the exchanged parkiele. Accordi~gly,

&Eer~only in~anwquenlidlyf ram (3-l X .21), through %heexplicit appearance

of z k the transition p i n t ;it is umd implicitly as the origin in (3-1 L.21), since
the variation of 2' over K z is not signifie8nt in the soft pfroton conkxt, This is
%he antieipakd source dweription of multi-soft; phobn emimion proc
Notice that the eRecl;ive photan source c h a r ~ k r i z e fshe probability amplitudes
for dditiond phobn emission, relative ta that of the rdia%ionlemprocess,
which is s u p l i d by the ~ignificzllnceof K 2when if acts a simple p a ~ i e l eemie
sion s0urc-e.
The time has wnne Lo face up to a eharaeterisfcic feature of soft photons.
With ra continual diminution of the e n e r a migned to a soft photon in tt given
experimental amangement, one e~entuallyreaches a, point where it is no longer
pmible to decide whether the phohn has or hm not been emitM. X tl is a some-
what complementary spsm-time obsemalion that, with increasing wavelength,
one eventually loms the pomibilily of isohling the soft photon emission procm
sineft the diswsition of sumoundixlg matbr hecomes relevant, Thus, more than
%heu ~ u aarnount
l of det&ilconcerning the aperinnendal arrangement is mquird.
This is emphasized by using $he photon sour= (3-11.55) eompub the Bverage
number of phohms enniCM along with a given parl,icle. That number is

To we the emenee of the situation, id suaces to eonrsider ra slowly moving particle,

m + fmv2, p = m*., /v (3-1Z'$7)
and a coordinste system in which nphw only a time oomponent, no = 1. Then
the eompoxlents of the vecbr combination in (3-11.56) are, approximately,
6x1writing

where dQ is the solid an&e within, which the phohn moves, we get

This photon eaerw intepal doe8 not exist matbematica;lly, divergng both at
the upper and lower limits. But clearly there are physieal r e s t ~ e t i o nrtf~ bath
en&, When one reaches energies sL which the photon eeaws t;o be soft, the
evaluation ( S l 1 . W ) no longer applies, and a lower limit is 'by the minimum
delectable photon enerw of the experimentd amangement. Onee upon a time,
the m~themsticaldivergence at zero energy was taken literally, and this soft
photon phenomenon boame known as the 'infrared catastrophe.' As s. eatas-
trophe, it nzhs rathw low on the scde. Consider %he&Berence that is impfied
in the vdue of ( N ) , depending upon whether the softmt photon considered has
s wavelength of visible light, --10-' cm, or has a wavelength comparable to the
nominal radius of the universe, cm. Since v2 < 1, that difference is

If the radius of the universe is reglmed by a typical laboratory. dimension, this

difference drops to -10'"".
To ilfwtrak the discussion of spin rraluecs other than gero, we shdf consider
spin 9, using the Lapange funetion of (3-10.63). The current of (3-11.1 1) is
repIaced by
+e
j?,(t) = @1(t)r07'&2(0 Z;;t (+g - ~)a*l$l(~l~~@"*&2fE)1

where
[?(l/.;)la + dtbz(z>= a%(%)
is ~olved,in momentum space, by

The comparison with the exchange of one particle and one photon under non-
interaction conditions,

supplies. the effective two-padiele source that represents the emission. of the
3-1 1 Extended sourc~s. Soft photons 276

extended particle source :

The momentum space equivalent is

5 )v p 1
eff.
= [rpeq + Ge (h- l)gp~k.q]4, ( P ) - f '(k)ieqv,(P)
(3-1 1.67)
Using the latter form, we observe that
m -rP
eff.
- l ] eqn2( P ) (3-1 1.68)
and, on writing
rk = r P + m - ( ~ +pm),
we get
m -YP
P =-(YP+~) eqs2( P ) (3-1 1.70)
eff.

But this is to be used in the context of Eq. (3-11.65) where the field t,bl(x)
represents particles far from their detection source, and the Dirac differential
operator in (3-11.71) produces the required null result. Alternatively, we can
use the momentum form (3-11.70) and recall that
(X) = E irl:,.,(2m
P"!?
dup)" 2 ~ - ' p z U *P ~ ~ P (3-1 1.72)
where
U~.,YO(Y~ + m ) = 0. (3-1 1.73)
Let us also note the photon analogue of (3-11.72),
A: ( E ) = U (dwk)" *
&X, (3- l 1.74)
kX
with

since both factors are useful in producing a simplification of (3-11.67).

A relevant algebraic property is
r"(m - r P ) = ?'(m - r p ) - YpYk
= 2p~' flvik, + + [ ( r p+ m)rP+ P], (3-11.76)
where both terms in the square bracket can be omitted for our purposes. Simi-
6 Fields

XarXy, we note &hat

-o@'ik;,rk = (rprk + k@)rk [kfi~kj,
s = ~ (3-1 1.77)
since"k QO,and

where noneonlfibuting k m s have been isolated in brackets. The result is

where one can alm use the substitution

1%is evideat %h&,in the Emit of sof* photsns, there is an ef-Teedivephobn source
which is identicsl with the one encounkred for zero spin. This i s ta, be expc?cLr?d,
for every spin value. The suecwsive multipole moment efXecb involve ixlerettsing
powers of the photon momentum, and all become negligible compsred to the
charge aeeeleration rsdiabtion for suficientfy soft, photons, But the particular
choice of f@(k;) fhst removes the acceleration radiation no longer suppresses
photon emission completely, since the spin-dependent effects of magnetie dipole
moment8 remain in (3-1 1.79), and no ~pmializationof g can, annul bofh term^.
We have illustrated the e x k n d d source concept in the eonbxt of emission.
1%can all be repeated w i m the exkaded; source acb h absorb ab padicle and a,
photon. But the= inverse procet3ses are also d a t e d by the TCP operation,
concerning which nothing bm been said reeenfly. The eEw&of the Eutllide&n
basd coordinste transformstion

on sources and fields is given by

and, far spin 5,

q(rr?)=rsv(z), $(Zc)==Y&#(z)*
The field4epeadent gource
3-1 2 fntsrsctian skefston, Scattering cross sactfons 277

has %hesitme trtansfomstion behwvior zlis ~ ( x if)

which finally provides a physical basis for the synnmetq propedy that, thus
far, hw been adopted for convenience. The puirely elcetrom%meticpart of the
action retainis its form undw this tran~formwtion,

while the particle contribution, including the interacfion brm, reverses ~ i p :

But the compIc?& statement of the W P opration ineludes the revergal of sll
factam. The anticommutativity of the sources and fields msociaM with the
spin +,F", D, particle provides the addition& minus siw needed fo produce the
anticipratd invariance of the action under the TCP transfammation.
The TCP operation inverts the causal order, and inbrchanges emission
and abgarption processes. On applying the transformation. to (3- l l .M), one
quickly ve~fiesthat the whole ~truetureis mainlaind, and it is therefore only
neceBsary to change the eeusal labels. The same rem~rkapplies to the momerr-
turn version (3-1 l .67), of course-, except that we folIo~vthe practice of r a v e ~ i n g
the sips of all momenta when absorption proeews are being deseribd, which
the transformation automatieaIIy supplies. What has been shown in the spin 4
framework is of general validity,

3-?2 tMTERACTLtOll5 SKELETON, SCA-ERINQ CROSS SECTIONS

A @ven primitive interaction implies a fet of coupled field equations. Here is
the example of the photon and the charged spin 3 particle, writkn* for sim-
plicity, with $g = 1:

In view of the nonlinearity of thia system, the construction of the fields in Cerm
of the sources wil be given by doubly infinite poxyermfie5, That is atss the
nature of the action tvhrtn the fields are eliminated snd W is expremed as a f u n e
tionsl of the sources, The successive terms af this series, W,,, with n particle
and v phohn saurces, represent increasingly colnplieated physical praces~es
whieh are thus mbo~t~ledged to occur, but will not be given &heirfinal dweription
rztt t h i ~first level of dynamical evolution. That is the meaning of an inbraetion
~keteton. At later s&agesof the dynamiml development, proceses already
present in skeletal form are provided with more complete descriptions, and
mme additional processes are recognized. I n thiss ~mtion,we propose to carry.
%hedigeu~sionof the simplest terms in the interaction skeleton to the p i n t of
displaying their observational implications.
There are t~\-oasymmetrical ways t;o eliminate the fields. In the fimt, one
introduces the formal solution of the parti~lefield equation:
( d s f ) @ $ ( z ,z f ) q A(X'),
(3- 521.2)
[r(--G- (z) ) + m]@$($,X') = 6(2 -- S'),
which gives the partial action expression

The stationary requirement an vasi%tionsof P1, reemem the Mamvell equation

of (3-12.11, Gcth #(S) given by (3-12.2)wa highly nonlinear equw;tiorr for the
metar potential. One can still exercise the option of removing A, from the
particle source by sdopting the special gauge of the A: potentisls.
The latkr procedure is psrtieuldy mconnmended rf we hUow the econd
course and eliminafe the vwtor p t e n t i d , replseing it with

where
j&ne,(z) jP(z) - (dz"lf""(Z - x")aj"(~~) (3-12.5)

and the gBuge condition betermina X(z) as

Another way sf p r e n t i n g this pohntial is [JPis now an arbitrary vetor]

( d z ' ) ~ / + (-
z zt),,[J'(z') + jP(z')], (S12.7)

where, writ&& in msmentum spsee h r convenience,

D/+(k),, = (g,. -- i k ~ s ( k ) ) @ ' ~ D +(g*.
( k ) - fk(k)ik.)
= (@@P - ikJ,(k) -- f,(k)ik. -- Skdh(klfk(k))D+(k) (3-12.81
31-12 tntersction skeleton. Scattering eross sections 279

i s the Green" function of the second-order Nfax%vellequation that sati~fiegthe

gauge condition
fp(k)~:fk),. = 0. (3- 12.9)
The second pta&ial action exprwsion ean be written as

or in the equivalent form that uses the nonconserved currents and ~ $ ( z- z)',.
The nonlinear field equation for J/ that is derivd from this tzetion is that of
(3-12,1), ~ t A,h replwed by (3-12.4) or (3-12.7).
Wbieh of them asymmetric hrms it is mos&convenient to consider depends
upan the process of intermt. Suppose, for example, that no photons arts in
evidence. Then one ean e t J P = 0 in (3-12.10) and marnine the nonlinear
prope&ies of the pa&iele fiftld, If the causal situation is such that interaetion~
aeew far from the padicle emission, and dekction saurees, which is part of the
amangemat of s s e ~ t b r i n geqefiment, the p Wmi in jtoa8.-causally tied
to the sourcecan be i ~ o r e d .The inkrsetiot-t.tern of (3-12.10) contains few
particle field8 and therctby ett least four sowee factors, When we consider
prmesws that involve only four sources, as in particleparticle scattering, fhe
stationary aspecCs of the =Lion principle permit us to identi.fy $ with the field

XR omiLLing further k r m ~of 8 mare cornpleb solution of the egeetive field

equation,

which are a t Xeast cubic in the source, what is thereby lacking in W fim no let35
tf-trtn six powers of the source since firstorder effects of the field change are:
annulled Lhrough the stationaw action prope&y, Thus we have ideIldifieib

where
jr(,) = IC(~)~~?@~.~PJ.(Z)
and JI(1;) is the field given in (3-12.11). Analogous r e u l b hold for any other
spin value. With spinless p&rticlr~?s, far example,
2 Fidhs Chag 3

and
(dz')A+(% -- X') K(%'). (3-12.16)

at involve only t~vop&&iclesources but any n u m b r of pftatoxl

demfibd most cmvenienlfy by the action (S12.3). The s&
fionary action principle wrxnits the identification of A, with the field of the
photon. mure@6, the o ion of jf', which is a t Xewt qadratia in %hepadicle
so-, chanGng thorn b r m in~ W thort contain no t h m four paticle murem,
Thw the whde m6 of 8keleLd inbrraction hrms is &yen by

The refe~eneeto the vector pokntid in the p&icle sowee has been dropped,
d t h the undersLanding that (3-12.17) vvill be applied ta pmes3ses in which
) umd as a ~imple
~ ( 2is pafiicle source, all partielephohn interaatiom occunlng
far from any of the sowees, To eixhlbit the individud W%,,we must expand in
power series the A@dependence of G$(%, S') and extract the term containing v
vmtor pobntials. For this p u r p e it is useful ta r e h k the Grwn% function
equatiorr of (S12.2)
+
(--ria+ m)@$(z,z8)= 6(2 - g') e P ~ ~ ( z ) ~ $ ( z , z '(3-12.19)
),
which is e o n v e ~ dCX7 an i n % v ~ equfttion
I by the formal mlution
G:(., 2') = C+(% - z') $- (dt)C+(z - 8 e q r(6)~ ~ $ (S').
t, (3-12.20)

The d m i r d p m r series expansion can now be constructed by suece~ives u b

~ ~ t i h t i ainn this equsfion. Such manipulations are facilihtd, however, by
sdopting a matrix notation in h i e h the e~ordin&les s and x' join the &scmk
spinor and charge indice8 as continuous row and column labels. Thw, we
transcrih (3- 12.m) into
+
G$ = C , c + ~ ~ ~ A G $ (3-12,21)
and wrik %hefomal solutian of this m a t ~ xequation a%

A compact statement of the expansion is, therefore,

to (3-12.17) and mite out the suceemive W2,. In doing this

one recognises &hateach parficle source is mullipli~edby a prop~gationfmction
3-9 2 Interaction rikarlaton. Scattering cross seetions 281

G+ to form the field + of (3-12.11) :

' e g r ~(s)G+ (z
(dz)(d~')#(~}r - z ' ) e q A~ (2' )J.(z') ,
(3-az.24)
(dz)(dz') ( d z " ) $ ( z } ~ ' e ~($1
~G+(z
~i - z')eq~A(2')
X @+(X" X " ) ~ Q Y A (Z")#(Z"}.

The spin O analogue of Eq, (3-12.17) is

where the? Green" function difjterenttidequation (3-1 1.36) is presenkd ss

The equivalent inbgral equation is of the following symbolic appearance,

which fiw the formal solution

b$ = [l -- A + ( ~ ~ +( ~A pA) -- e Z A 2 ) ] - ' ~ +
= A+ + ~ + ( e q ( p A+ A p ) -- e2A2)&+
+ h + ( e p ( ~+~Ap) -- +
e 2 ~ Z ) ~ + ( e P ( pAAp ) -- e2A2)4+ +- .
g *

(3-1 2.28)
The successive powers of A@are not p w ~ n k dwite4 so neatly as with spin 3.
The first two terms of the series W z , ara

in which, it has been expedient to retain the symbol

which means th& the careful orctering of factors can be ignord if the vector
potential has a vanishing four-dimensional divergence, as is the sjitualion Eor
(Slt2.18). Bokntirtls having this property am said to be in. the Lorcsmtz gauge.
 The imnndiak applications of the interwtion skelefon for which we have
b e n preparing refer to scatbring processes. Let us fherefore review the general
eonneetion between the source description and Lhe tr~nsitianprobabilitiw that
describe the @fleetsof inkractions among partictfts. The causetf ~ikuationis this.
Emission sources, generally referring to different kin& of psrtioies, act to produce
a multipartiele state sf parti~fesin a physicdly noninbracling condition, owing
to their initial apatiaf separation. Afkr zt sufficient Gme lapse, some of these
padiclm approach each ather, inhraef, and then separate to be eventudly
mnihilabd along with their noninkracting earnpanions by suitable detection
sources. The causd tknalysis of the arrangement is @yen by

1 (R" ) desefibe the traxfaitions

where the individud probability ampIi%udes({%l
induced by the psdicle inkraetions, and

represent the nonintersating muldipartiele ~tates. The Eakhr &relabled by the

numbers of parlicle~in the wrious single-paAicle mdes, d the prodwts also
range over all the differen%kinds of ptzrtiele~,of bo$h statistics. Aa a geaerating
function of the probability amplitudes, (3-1 2.32) ia more u~efullypresexlbd in
this v e ~ i o n ~

since the vaAous power8 of the emimion and detection sozlrces Bewe to direetlig
identify initid and finat stake^,
The vaeuum prob8blliLy ampfitude is detemined by the action
(3-12.35)

in which we have specifically exhibikd in symbolic f o m the qusdratie ~trueture

thst represents noninteracting psrtielef;. All relevant types of particles are
includrsd, so &at S is being used as a supersource. By r e w i n g fram both sides
of (t2-12.34) the expwssion thaL Mers to noninbracting psrtieles we arrive at
- I]
[exp(iW' (S1~8%))
3-1 2 Interaction skeleton* Scrrttering cross sections

where
tF""(Sl,Sz)=Wf(Sr+Sz)-Wf(Sx)-Mcl"(S2). (3-1Z.37)
The factor ~ X ~ [ ~ J S ~ represents
? G S ~ ] the exchange of those particles that happen
not to interact. And higher powers in the expansion of ezspfz'Wt]iindiertlt; the
possibility of repating independently in disjoint space-time regions all con-
figurations of interacting particles. Thus, the irfeducible interaction pmceslsrjs,
those that do not contain ndditiond noninteraeting parlicles and cannot be
analysed into two or more disconxreekd processes, are obtaind from

Invariance8 of the action imply selection mles for the tramidion probabili$ies,
Ri@d translations or constant p h a e transfomations of all souxces, far example,
which do not change Wy(Sx,S2), must leave the righthand side of (3/-12.38)
unaltered. The emission and absorption s o m e prducts are multiplied by
reeipro~alphase canstrtnts, re1a;ted to momentum and charge in these examples.
The individual transitiolrz. probabifitim must vanish if the phase constants do
not cancel, expressing the neeessaw eonservatisn of momentum or eharge in the
interaction process. The fwtor that imposes momentum conservation,

will emerge fmm a space-time integration over the inksactition ~ g i o n .We make
this csxplicit by writing

thereby defining the elements of the transition matrix. T. Thc inbgral is no&a
four-dimensional delta function since the integration domain is not infinite, T o
sppreciate this we must recall that the precise specification of individual mo-
mexlta used here is an idealization that holds well ~vithina partiele barn, but
faits near the bounda~es. Where the initial and final beams overlap to &vct
esusail definition to the inkeraction region, (3-12.40) is applieabfe, and limiting
the integration to that finib volume is sn ~pproxirnateway of recopizing %he
realities of the situation, It is probability that is physically significant, and we
are actually concerned with
(dz) exp [i X (nh - tl,)p.z] (dz)(dzf) exp [i (nb - n.)p.(z - X')]
The 5 integration ean now be identified as a delta function, and the X intepal
memures the Wtal inbraetion volume V, within the uaeedainfiea attached to
the bounday liayers. The proporttionality of the transition probabili-ty to the
volume of the four-dimensional inhraction region ixldicabs that the impadant
quantity is the cwffieient of propontionality, the transition prab~bilityper unit
faw-dimensional volume, or, per unit time in a unit three-dimensional volume.
This ratio is

which suppliw the physical interpretation of the transition matrix.

Led us b & n the svcific discussions of skeletal inhracdiona with the scatkr-
ing af spinless pa;rtictes, as describd by (3-12.13, 15, f 6). The field #(x) is
requird in the inderaction region, which is eausaHy intermdiah bcl.t~sreenthe
e ~ s s i o nssoumt? K2(z) and the dekction source K l ( x ) . The b t a l field is the
@uperpositionof pads related to fLfie8e wurees,
*(X) = cBl(x) + +zCzZt (3-1 2 . 4 )
where
(dz')~'-'(g - zf)KE(z'),
(3-12.45)

and the particular forms of A+($ -- z') disclose the caussl situation, The prscessj
we %reeoneerned with invdves the action of two emis~ion,sources and two
absowtion sourca. Thus, when (3-1 2.13) is considered, with the c

we m u ~ retain
t only tham eontribufiom having the required overall character-
istic,, &s conveyed by the causaX indices. Those term8 are

811 others having tso many or h a few emission or detection indices.

In earfier discussions of chargd spinless particle8 we have worked with
complex sources- But experien~ewith spin 4, for example, has shown the gnzahr
mnvenience of retaining red multiconnpontend source8 and making the apprw
pr"i8te complex projections for specific charge valum* Henceforth we shall write

where .the &WO complex ch%rgeeigenvectars are

=1 - (D-
* = 2-"'(1, i). (3-12.49)
3-1 2 Interaction skeleton. Seattatring cross ssetions 2815

Thew vectors have the properties of orthonormality,

snd, relative to the eharge matrix

they obey
pP
p; = 'p;eqr, pp,. = ~'@p.*
The eReet of complex conjugation is given by

Using this notation, we present the fief& of (3-12'45) as

where
+Pg (z)= ( d o p )' l 2(pqei~'
is the field associated with the specific particle Iabeled pqr, which enters the inbr-
&&ion re$on after its creation by the source K%,,. Sinnilarly, +P,(z)* is the
h f d of the particle labeled pq which, after leaving the interaction region, is
annihilated by the detection source K:,.
The charge structure of the various partid currents that compo~e(3-12.46)
is of importance. In j$,(z), far example, the charge frtctar assoeiakd bvitlh two
incident particles of charges qhand g" 'is

As we sec, it vxnishes unless p' +

p" -- 0; anb zero c h a r p is brou&t into the
interaction region. A similar restriction to opposite charges applies to jt;,(x).
When we consider fl2(x), the charge factor associated ~ r i t hcharge p" enkring
the interaction region and eharge qf leaving it is

and the necessary equality of p' and p" implies thet no eharge accumulates in
the interaction region. These am different ways of satisfying charge eanstirva-
tion in the scattering process. The seeond term of Eq. (3-12.47) does not con-
tribute to the sewttedng of particles with like ch~rgeand m examine that
process first.
 The form of the current j";z(z) is

in whi& we can recognize the cument, q 2 f h,, %h$ B

tsingle undegected pwtide. When only contributions from incid~ntpardicles of
$he =me ahrsrge are ret&inc?d,

where

exhibiLs the momentum form of the plnofon pmpagcttion funcl;ion, L)+(k) ==

(k2)-l, snd produces the space-time integration that enforces energy-momentum
clon8ervation. In pieking out the d e ~ k e dT matrix ekemen%we m u ~ take t inb
acoount tha;t %hesowee f a ~ b r gidentifying a particular pair of incident p a i e l w ,
iKZ,,,iKz,hp, and a particular pair of scathred particles, iKf,,,iK~,;,, can
each be produced in two ways comeapnding %athe s y n n m e t ~of thme pmduefrJ3
in p,, p; snd in p,, p:. Thus the transition matrix element will have those
symmefries, which is a statement of B. E, statigtics, The mstsix elemenk is

W&& L explicitly symmetried in p%, p: and d s a has the mquird pl, p: csym-
metq &meovemjl momentum conservation implie8 th&t

The exprimental memure of the eEectivenetss of s given ~ c % t % e faet,

i ~ ga
o h m & in beam amtzngennents, is sxt arcs or cmss section, It ex-
p r e e s the at which the dwigna;ted process ocaum per unit tinne and p r
uniid spatial volume, mlg%iveta the incident padiele flux and the den~ityof the
scatbrer~,in. the u m d 8iLuation af fixed brget, The controllable fwbm
refehng Lo the initid particle8 can be dven s general foPm fhitt permits fhtt
cro8s smtiorr eoneept b be applie?d to aolliding b e a m as well as ~ h % i o m v
targets. Le% be the particle flax vmtors of two asuch barn. An invahnt
3-1 2 interaction skeleton. Seatterrlnft cross sections 287

measure: of their relative ffux is suggested by the requirement that it mwt vanissfr
when the vectors are: proportional, and the beams run with the same velocity.
This definition, is
2 2 11%
F ~ ( g a s b ) -- t (3-1 2.64)
which does produce a real positive quantity since the Aux vectors are time-Xike.
If we write this out in terms of particle density so and particle velocity v = #/so,
the flux definition hcomes

and. ~vhenone of the beams is a stationary target (vb = 0) it reducm to

the magnitude of the incident flux multiplied by the target density.

Sinee the padicle flux associated with a single particle in a, small mamenfurur
cell is
SI( = 2pr dwpr (3-12.67)
the version, in ~vhiehwe shall apply (3-12.64) is

which introduces the masses of the particles, Other f o m ~can be used, par-
ticul~rlyone involving the total mass M, the invarisnG measure of %het o t d
moxnenturn,
= -(pn-tpb) = m: -i- - Zpapa, (3-12.69)
namely
+
F = dw. d&b2[MZ-- (m, rnb)']li'[M' -- (m. -- rnb)'Izi2. (3-12.70)
The following ratio, probability of a transition per uni-t;four-dimensional volume
[(3-It 2.43)j divided by invariant flux f (3-1 2,"i")], defines w, differential crass
aeetion. 1%is digerential since the final pahieles are gpecified within small
ranges of momenta, ets Ennited by momentum conservstion. 1nLel~;rations over
these diRerential elements supply variow diflerential erass sections of lmser
degrees of specification, le~dingfinally to a total cross section, althoagh the
latter mw,y not misf if very slight deflections e a r q a dispraporltionwtf? weight.
We shall uw the symbol da" generally for all diBerential eross sections, relying on
the explicitly stakd differentials La indicate its precise nature.
Energy-momenturn conservation in a two-particle scalteGxlirrg prmess fixes
$he energies of the scatbred particleg and leaves frw only two pa,ram&ers that
give the direction of the line alorrg which both pa&icleg move, in the rest frame
of the total. mamentum. We may as well inkgrate immediw,Le?lyover the distri-
butions af those variables that mmme precise vsltlw. Let us consider sny pair
where the m o d vemian refers to the rest frsme of P, in urbich P' == M, The
magaitude of the relstive momentum
P S Pa -B& (3- 2 2.72)
is @ven by
-
--
1
- [M' --
2M
(m. + m b ) 2 ] " 2 [ ~-2 (m. - (3-12.73)
h carqixlig out the enerm inbsa;tioxl thst selectis this value one must d k

where dft is the element of wlid angle for the relative momenturn. The immdiale
resuit is

1
X-.--""-.
1
3Sa2 M a P f 2 - (ma + mb)'~'''[~~ - (m. -- ma)'] l" da, (3-12.75)
which rdtrees the &fierentid aspeet to the angles that 8pesif.y %hedirectian of
her ee m We nob that the same squareroot kinematics1 m-
f a c b r ~aceur in the final 8tah ixlbpation (3-12.75) and in, %h(?:
incident flux
(S12.70). Thew relatively camplicaM factors will e~ncelfor a purely elstic
seatbring proew where initial and find perticks are the same.
The tramition xnat~xelement (3-12.62) provides a eimple application of
$he cross section. definitiorr, giving dirwtly

In the latter form, B is the defiection anlgle, and the full equivalence of the anglers
3-1 2 lntarraetion skeleton. Scattering cross sectlons 289

8 and W - B apremes the indistinguishability of the B. E. padicles, The

rduction is wrformed in the rest frame by noting that each of the four partiefe
energiea equids +M and this gives, for example,

Of psrfcieular interest are the vePy high and very low energy limits:

Note that at sn~a;llseatkrictg angles the latter reduees $0 the Rutherford differ-
ential cross metion for the scwtkring of dislin~ishablepwfticles,

is the relative kinetic energy of the particles.

When padiele~of opposite charge scatter, they are distinaishable by th&
chrcrgw rand only one kind of term emerges from the analowe of (g12.M) thaf
has the source factors replaced by iK:,,,iK:,i -,and iKPPPQiKIP; An Bddi-
-@.

Liond minus ~ i g nis also needed. Qf eaurcse, each process a p p s r s twice owing to
the combined wmntetry: plc--"pi, p2++ p;, q -+ --q. But now the second
k r m of (3-12.47) comes into play, widh

where mch current contributes two eqllia;l terms Lo a given process, correspondiag
to the symmetries expressed by p1 C-' p:, q -+ -q and p a t ) p;, q -+ --p. The
implied %ransitionmatrix element is
290 Fields Chap. 3

Notice the simple connection between the matrix elements (3-12.62) and
(3-12.83); they are interchanged by either of the substitutions

Correspondences of this type between different transitions have become

b o w n ss crossing relations. Their origin is not far to seek, Emission and absorp-
tion proceljses are united in the field +(z). The formal substitution pp -+ -pB
interchanges the? physical e.ETects that identify emission and absovtion acts.
And the numerical eharaclerizations provided by the individual fields (~,,(z)
respond appropriately :
"bp,(.> #p-e(a)*- (3- 12.85)
Given the transition matrix element for one process, the substitution gener-
ates another oxre in which an initial particle of properties p, q is replaced by
a final particle of properties p, --g, or conversely. Of course, this must be
done a t both ends of the reaction if one i s to retain s s c a t k ~ n gprocess, When
particles of opposib charge are present, the outcome cam, be a synnnnet~of s,
given matrix element, as illustrated by the invariance of (3-12,83) under either
of the substitutions
P P $6 --1->2* (3- f 2-86)
Note that we are considering individual applications of a transformation that,
used wholesale, is the TCP operation.
The square bracket factor of (3- 12.83) has the foliowing evaluation :

- 2m2 1 - M" 4m2

2-
- 4m2 sin2 (Bf2) I - cos 8
--p

M2 11/12

The second term, (4m2/M2) cos 8, is relatively negligible both at high energies
and at low ener$es. This provides a simple eonne~tionbetween the cross wetions
far unlike and like eharges, one thsd becomes accurate asymptotically a t bath
extremes of the mass scale and constitutes a reasonable interpolation between
these limits:
h" COS

The scattering of photons by spinless charged particles is contained in

(3-12.29), dong with other processes. The part we w a d is extracbd by writing,
as in (3-12.441,
(a(s) @I(%)""I"cbz(x)t
E (3-12.89)
3-1 2 Interaction skeleton. Scattering cross stmtions 291

together with its photon analogue

where
A'(4 = At (X) + A$(z), (3-12.90)

/
A:(z) = i (d~')~''D'-~(s - z')JlV(z'),
(3-12.91)

/
A$(z) = i (dz')f'~'+'(z - X')J2.(zf),

and then retaining those terms that have one photon and one particle emission

-
source along with one photon and one particle detection source. They are
w12 - zf)eq2~~2(z')
/(dz) (dzl)rl( z ) [ e q 2 ~( x~ ) A + ( ~
+ eq2pA~(z)A+(x- z1)eq2pA (x')l#z(z')
I

- /(dx)4l(x)2e2~i(x)A2(z)02(z), (3-12.92)
in which we have adopted the simplification that. expresses the use of the Lorents
gauge for the vector potential. Let us recall that

Of the two terms that do not refer to polarization vectors, one vanishes because
the source is divergenceless and the other, a gradient in coordinate space, can
be removed by a gauge transformation. Thus, it is in a special class of gauges
that we write

where
112 ikz
A$x(x) = (dud
p
ek~e ,
and they are Lorentz gauges, since

Now select the coefficient of iJk*,X,iK,*,,and iJk2~,iKp2,(we have finally omitted

causal labels on the sources since they are abundantly evident in the other
indices). The resulting space-time integrals produce the Fourier transforms that
convey the momentum specification of the scattering process. Here is an
example:
P
w h i ~ haltf~oU S ~ Sthe fact that the differential operators pp rand p" a& dirmtly
upon momentum eigenfunctions* The transitian matrix element is obfaimd aa
I r , ~ , l p , , ) = (dur, d ~ pdut,
, do,,)"22e24:~,~,.e~,~,, (3-12.98)

which uws the kinematical simplifieiztions

Other aspects of the kinematics are these, 7'he totat momentum is

F ?E= P I -4- kl E pz 4- h2, (3-12.101)
and therfsfare
-plkl = -p2k2 = 4(B12 - m2) (3-112.102)
while
-plkz = - p 2 k z = $(MZ- m2) + ktkz. (3-12.103)
Invariant expremions for the particle energies in the center of mass system, the
rest frame of P,are

Written in terms of the center of mass sezltteltxing angle 8, we also have

We should also nok tbft crossing symmetries exhibited by the transition matrix
element (3-X2,98,99). Sinee the sign sf q is irrelevant, there is invariance,
spcseificalfy of h,,
under the substilution

for which the equdity of klplwith k2pz, and of k,pz with Iz2pf,is decisive.
Concerning tbe photons, the use of linear polarizations with Feal potari~atioa
wetars implies the transfomation
3-1 2 tnti;sraction ~ksleton, S~attsringcross setlens ;193

end the transition matrix element sfiwld be invariant under the interchangtt.

This induces the exchange of p and v in Vp,, which indeed 8how6 the requir&
invariance.
The tensor br,, also has the fallowing imporlant proprlies:

They bring about the neeemary con~rvationof the egwtive soure@ for the
emission of the final photon and the absorption of the incident one:

The summation of the transition probability over bath polagsations of &he

scattered photon ean now be performed with the aid of (3-12-93),

If the ineidctnt photon beam i~3unpoIarizd, both polarizations appearing with:

equd probability, the nceessary average can aIso be performed by me&m
of (S12.93) :

A straightforward algebraic reduction gives

Let us again emphaize the relative 8impXicity of the kinematical factam in

the cross m~tionfor elastic scatkring, even for pslrtieles of uneqml mas. The
ratio of the inhgral in (3-12.Z) to the invari8nt Aux of (3-12*70) producw
&hefaetar

which sugplie~the unit for a, diRerential seatkring cross swtion, Then, ailnee
2eZ = ~ K O Iwe
, get directly

- m2
M %+ m2 -I- cos B (3-12.117)
M2 - m2 p

l + ~ z + r ncas
~B
vvhich u w the center of mms wwthring angle evaluation for plkz. The dif"Ter-
entiwl and totd crass wctions for the extreme mergetic limits are

the latter Ling the Thornson cross sections.

The conserved nature of the effective sources thaL emit the find photon and
tabsosb the initial photon implies the gauge invsrianee of %hetrazlgitian prab*
bilities, This perIxlits oae Lo exploit whakver simplifications cm be iarkaducftd
by spwial gauge cheiees. The question of gauge in eonneclion wiLh the polerriza~
tion vectors is implicit in the choice of E@ which, in some coordinate frame,
haa3 ita spatid component8 reversed dative to those of kC. This k e ~ p ~ ~ s l e d
wi%hthe aid of a unit time-like vecbr n@EE.S

and tho polarizalion vectors have the two arthoganality properti~,

which are incorporated in the summation

We now return to (3-12.98,99) and observe that the identification of R+# with
either py/m or p$/m produces fhe simplification

One can wrify directly that the same resuft for the surnm&tioaand average over
polarisations is obtained in this way. Applied to the final pkotons, (3-12.121)
ghes
3-1 2 Interaction skeleton. Scattering czross tstlclionas 298

and, then

If we sef 12 = p t/rn or p2/m and insert the relation

we regain (3-12.113, 115).

The $auges we have just described are particularjy useful when the particle
is at rest initially, or finally, Another choice of the vector lz ia P / M , and this is
most convenient in. the eenter of mass frame, f n ail these examples the coordinak
system is ~ h ~tos identify
e ~ npwith the time axis, Let us u8e the cenkr of ma;iss
description lo study the polarizafion dependence of the scattering cross section.
The trajectories of the incident and scattered particles define a plane, We first
choose linear polarization vectors that are either perpendicular la the plane, or
lie in it at right angles to the appropriate phston momentum. The differential
w s s sections for the various polarization as~ignmrsntscan be read off from
(3-12.98,99), m gimplified by the special choiw of fbe several vectors, including
the eenter of mws momentum relation
O = p, 4- kl = P 2 + kg,

which gives

The cross section. vanishes if one polarization vector lie8 in. the scatkring plancl
white the other is perpendicular to the plane, When bath veetars are perpen-
dicular to the scattering plane,

and when they are both in the scattering plane,

--
+ + cos 6
- m2
M2 m2
- m"
cos B
.l+
~ ~ $ m 2
 Fields; Chap. 3

The average of the Lwo, appropriate to an initially unpolarized barn, is

(3-12.117), which gives the latbr stmcture greater physical meaning, 1%isr
interesting to o k m e that photons s e a t k r d through the angIe d e k r ~ x l e dby

mid be complebly polarized perpendicular to the scattering plane.

The diBerenti,iaf erosss wction~referring to circular pla~azztionor helici$y
sr%aksean be produced from the linear polarizalian results. The eireular polafiea-
%ionvecbrs are linear eombinzttiom of thaw p8rallel and pepndicular 4x1 the
seatbring planet relatively shifM in phw by =&W [(2-3.29)]. Since the complex
conjugate po1ariza;t;ion vector repremnb the outefoing photon, the probability
amplitude with the same helieity initially and finally equals half the sum of the
two linear polarization amplitudes, and, thst for opporsite helieitiea~is half the
digerenee. The diff'erential cross sectism corresponding to no change in h&ici.ty,
or to a helicity reversal, are therefore, respectively,

The geometrical facton thst appear here, cos' +B and sin' 48, are familiar ss
probabilities, for unit anwXiar momentum with magnetic quantum number +l
in tl given direetion, that a measurement m d e in a direction at the relative
angie 8 will yidd magnetic quantum numbem +l and -1, re~pectively, 'Shere
ia also a dynamical wei&ting fpbctor that is unity at low energies, M rrr m, and
suppresses helieity changes a t very high energies. The tots1 differential cross
seetion, vvhiGh is independenl of the initial heiicity, is the gum of the partial
erass geetions in (3-12.131) :

It is equivalent h (3-12.117). At the scattering angle determined by

tan $6 == Mint, (3-12.1133)
the two partial cross sections are equai, leading to zero average angulsr ma-
mentum in the dimetion of the scatter& photon. This ie, of eourBe, the
same as the one given in (3-12.130) at which the scattered photon is linearly
polarised.
Other processes involving two particles and two photons are contained in
WzZ. When we SeXect terms with tvva partielet e ~ m i o soureas
n and two photon
deketion s m e s we are considering pa&icle-a~%iparticle andhilati~ni n h t;uro
3-7 2 lntsraction sksteton. Seatasring cross sections 2337

photo-; two photon emimion sourees bgether with tmvo pafiicle deteictisa
~ ~ u r e indicah
es the inverm process, the crestion of a padieleantigartick! pair
through collision of two phobns, The h t k r , for example, is deacl.ibed by

and we extract the coefficient of iK,',,iK:; _,

snd iJxgk,iJi;~; to get

where

Notice that the symmetry klXz k ; ~ ;is explicit, as is plp ++ p; --q when one
m 8 the kinemadical reIatiom

--plkz = -pikg = +M[+M -- (*M% - n t 2 ] l J 2cos B],

(3-12. f 37)
-pgk$ = - p f k z = +M[*M + ( * M ~- m2)"' cos 81,
where 8 is the ande htween particle and photon, relative mamenta. The lo-
cation of the: tbrmhold for the reaetian rat M = 2m i8 app8rent in the square
root that gives the mapitude of the cenhr of mass particle momen$um, a11
parfiele and photon ener@esb i n g equal to $M. Using red plariealion vectors,
the matfix elemeat (3-12,135,136) is obtained from (%12.98,99) by the
craasing tr&mformrti.t;iorr
p2 -+ --p$, klXl -+ -kbA&. (3-12.138)
Since the final pa&icles naw differ from the initial ones, the ratio of the
kinemadicrtf square root facdam appear8 mpfieilly in the differential cross sec-
%ion. We shall give it for v a ~ o u polarizst;lom
s assipments, first using tke linear
po1ar;igatiaxzs that are parallel or wrpendieular to the plane of the remtion:

the cross section vanishes when the two polarization vectors are st right angles.
Provided the threshold energy is exceeded by at least the factor 2'j2, M 2 > gmZ,
there is an angle at which the diEerential crass section for parallel polahga-
tions vanishes,

The cross ~ection,appropriate to unpalarizd photons, an average of the four

psmibilitie~,is

* (3-12.141)

The behaviar near the threshold, and a t high ener$es, is given by

The transition amplitudes for circularly polarized phobns are a g ~ simple

n
linear combinations of those referring to linear polanaaliom. The photons are
opposikfy direct& in the ceder of mms frame and as~imingthem the same
helieity, for example, means %hatthe photons have oppo~ikeanwlar momentum
doag the common line of motion. f t follows that $he tr&mitian amplitude for
equal hellieities is half the sum of the twa linear palariaation amplitudes, and
that for opposik heIieiGies is half their differenee:

The dominant reaction thus shifts horn equal helicifies near the thre~bsldtZ)
opposite helieilies at vew high ener~es,The crossing poixlL aceurg at
3-1 2 Interaction skelston, Scattering cross se~tionar S
19

wlrieh is a restatement of (%12,1.10). Note th& the geomc;tricr;tlfactors, l and

( g ) sin' 8, also refer to angular momentum. The first affirms the equivaienee of
all directions in s state of zero ane;ular momentum, and the second gives the
probsbility, for angular momentum quantum number 2, that connecb mal$netic
quantum number f2 in one direction with mapetie qu8ntum sera in another
direction inclined at the angle B.
The transition matrix element for the inverse process of particleantiga&icle
annihilation into Lwo photons has the same appearance as (3-12.135, B@), with
the causal lsbels reversed and complex conjugate polarization vectors eubsti-
tuted. Since the roles of initial and final padicles have h e n reversed, the
kinematicat syuare root factor becomes inverted, but all dse is the same. With
helicity labeling of the final photans,

and again the predominant helicity relationship of the photons change8 in going
from low to high enerdes, I n the annihila.t;ionaf slow particles, M c=i gm, there
i s no relative angula~momentum for the photons to carry away and equal
bdicities for the oppositely moving photons dominates. A t very high energies
the photons sustain the mrttximum. angular mmentum along their common line
of motion. Nevertheless both: differential cross sections are isotropic, and the
tatal eross sections are

The variation of the cross section with inverse relative s p e d v, when. v f<1,
means only that the rate of the annihilation process per unit volume is pmpor-
tional to the product of the beam densities. The eomputzlLion of the tatd eroes
section for reactions in which the final state contains two identical particles,
such as the B. E, phatons in this mnihilation process, needs one note of caution.
I n summing individually over the find tstaLes of both particles, which is here tlke
summation over heficities and the integration over all dirrsctions of motion,
every physically distilnet slate of the Lwo padicles is counted twice. That trap
has been avoided in stating the cross swtions of 63-12.1463).
 3-33 SPIN -& PROCESSES
Let W b @ n with tbe scatkhnf~of rspin particle8 that have like charges, As
in the @pia0 discussion, the relevant part of withod reference to specific
charge v%lue~,it3

but s d y the first &mmapplies to pwr2;ieles of tbe sarne charge, with

The form abtrallned for W4@,

%ndogousts (3-12*M), is

The q fabl, which is common .t;a df spixrors, It% ben, omi_l;l&. Note t2b0 that
the. id%idorder of the tatally antieom.muta%ivemure@f a e b r ~has h e n r e
amwd wiLbout the inkwention of minus signs:

The tmmition matrk element will be defined relative to %heor&r of murce~

&at appeam explicitly in (3-13.5). f3ut am must, of course, tske into account
$hat a pa&icular p d u c t of dehtion sources occurs twice in Lhe s u m a t i o n
with B rel~%iwminw ~ i powing
, to the F. D, antisymmetq under the e x e h a ~ e
plc8 ++ p:@:. A similar remark sppfies to the emission sources and the permu-
W o n p,~g ++ p&;. Thus, the required matrix element, whieh shows the anti-
that are charack~gticof F. X). a&&@, is
 Even when one is not interested in specific spin values in the inifiai and finaf
states, perhaps the simplest general procedure is to evaluate the higerentiaf
cross sections for the various helicie assignments, That is already suggeskd by
the photon pola;riz;ation eonsidergfions of the preceding metion, urbere the oud-
come of poIarization summations and averages required some algebraic reduction
to attain the result that was produced directly By considering the various helici-
ties (or Xinear polarizations). This simplification lvas particularly m a r k d in
high energy photon scattering where the helicity strongly preferred not to
change, The same tendency appears in the present situation, which we may as
well call electron-electron scattering since that is the outstanding realization.
The general construction of the spinor up,warj giwn in (2-6.90) as

where the v, sse v @eigenvectors with eigenvalue 3-1. When they are chosen to
be eigenve~torsof U pllp/ as well, identibing cr with the helicity, we get

where the latter is the high energy limit in which helicity becomes linked to the
eigenvalue of f n the cenbr of mass frame where X aX pa&icle energies equal
+M, consider the following high energy evaluations:

since irs has no diagonal matrix elements in the 7

' = +I subspace, and, using
the relation,
TOY = iY5a, (3-13.1 11)

We see that (~~,,,r~r'u,,,) vanishes if ol = --@a. The helicity does not change
in these products a t high energy because rar' commutes ~vithvs.
Accepting the restriction ol= a2,a: = g;, we find that the product a g
pearing in the first term of (3-13.7) is

There is a basic identity, expressing the completeness of the four mateces, I, a,

which is presented in Eq, (2-5.58). When the general lmratri~esX and Y are
 Chap, 3

appropriately chosen as dyadie products, it tells ua that

which is antisymmetrical in the indices crz, G; and in g,,g;. But a word of eau-
$ion about notation is cdled for here. Although we have M-rittenvoz, 8&y, one
must not forget the implicit reference to the direction of the momentum p2
slfong which the spin is projected to give the component a2,m a t we have just
referred to as antisymmetry in g 2 , a; is, properly speaking, antisymmetry in
~ ~ p 2h i . The combinittion (3-13.14) does not vanish when the helicities
2 and
crz and cr; are equal,.
We consider fimt the situation of equal, initial helicities. Then precisely
the combination that is evaluated by the identity (3-1 3.14) appears in (3-13.13)
and, owing to the antisymmetry just mentioned; the two terms of ($13.7) are
combined into one, with the factor

To evaluate the spinor products that appear in (3-3.14) we choose p2 = --p;

as reference direetion-%he x-axis-and express the choice of equd belicities, or
apposite spin projections along this direction, by standard t ~ v ocomponent

The spinm v,, and u,;are in the same relation, but ~vithrespect to the direction
af p, == --pi, wrhieh is rotated by the sngb 8 about the y-axis, for example.
That is expressed by

which, as a combination of matrix elements, is also the determinant of the uni-

modular rotation matrix. Since $he various factors of 2m and df cancel in the
matrix element, leaving 2eZ,the result i s immediate :
3-7 3 Spin 3 procssses 303

For the situation of opposite initial helicities, the combination Lhat appears
in (3-3-18), apart from a minus sign, is

The two contributions of (3-13.7) are now wsocia-ted with different final states,
which do not interfere in dihrential cross sections, The helicity labels are,
respectively, cl = crz, c; = o; = --gz and ox = a; = --c%,a: = ~ 2 Alter- .
naLiveIy expressed, unit angular momentum along the initial direction of motion
can lead Lo either of the magnetic quantum numbers, +l or -1, along Lhe
common direction taken by the scattered particles, Leaving aside the multiplier
of 2, the factors contributed by (8-13.20) in the two situations are

(y:e(l i2)isrvU+) ( v : e ( 1 / 2 ) i @ @ ~
v+) = cosZ*8,
(3-13.21)
a1 = G;, = Cr2:
*
(v-e
(1/2)illa
%+)
* (1/2)iBo.
(U-e @v+) = sin2 48,
if one is careful in translating the heiieity specifications into spin projections
along the two felevant directions. The diRcrential eross section produced by
adding the noninterfering eontribu%ionsis

The trigonometric factor in square brackets can also be tvritten as

The dieerential cross seetion appropriate to initially unpolttrjzd electrons (one

unpolarized beam will do) is the equally weighted average af the more specific
cross sections given in (S13.111)) and (3-13.22, 23). ThaL is

which has two remarkable katures. It is s perfect square, as though only tt

single process contributed, and it is identical with the spin O high energy differ-
ential cross section of (3-12."7).
This interesting equivalence of different spin values is restricted to very high
energies, That is most evident in the scattering of low energy particles, M = 2%.
Then the spinors g,, essentially reduce to the rot= $1 eigenveetorsv.. Helicity
ceases to be the m o ~ u&ul
t degcription, the referral of $1 spins fo a common.
direction in space taking its place. fnded, with that chaiee,

and the individual scattering praeosmss t ~ k pl&ce

e withouL ehsnge of mrwnetic
quantum numhr. In this low m e r limit, ~ spin and orbitd motion are dy-
nsmially i n d e ~ n d e n t in
, contrmt ta very high enerw conditions where they
are tightly linkd. When the initial, and therefore the final, mapetie quantum
numbers are equal, both te af (SL3.7) cantfibuk, with rever~edsign; for
ogpo~iteinitiaX mapetie quantum numbers one or the other of the two t e r w is
effective depending upon the msignment of oppalsite mgnetic quarrtum numbem
in the find state. The spin-averagd differential cross section, is, themfore,

The latter f o m comesponds to the alternative of averaging over the three

symmetrical spin states, vvfiieh are anbisyrnmetrieaf in spratid coordinaks, and
the single anfisymmetricstl spin state, which is symmetrical in spatial coordinab~.
This F. D, re~ulLdiBFers, of coure, from %ha& of (3-12.79), where only the sym-
metrical spatial combination appeam.
States clmsifid by the total spin of the particles are useful at high e n e r ~ ~ ~ r
too, provided the reft3rt;n;ce direction, of the magnetic qu%ntumnumber dif?Fers
for the initial and the final parLicles in con,fornnity with the altered direction of
motion, For unit anwlar momentum there is also a mp~ration.of the stabs with
unif magnitude af the magnetic quantum number from the one of zero ma6;netic
quantum number, reminiscent of that for a unit spin particle ita m m vanishes,
eEeetively a high energy limit. We have already noted that the transitions of
the initial unit magnetic quantum state tstke it only into fins1 states of magnelic
quantum number f l. indeed the weight factors, cos' f B and sin' a@, that
apgc?;ar in (3-13.22) are just the probabilities for unit anwlar momentum %;hat;
connee.t;magnetic quantum number +l in. the initial direction with +l and --X
in the final direction. T h v occurred within a pho-d-onconkxt in Eq. (3-12.131).
The unit spin state of aero magnetic q a n t u m number is the symmeticstf. eom-
bination of the two ways that realize equal helicity, B == B' = &l, The aero
spin state is the c o r r ~ p n d i n gantisymmetrical combination. Since the helieitim
are maintained in individual seataring aeb, and reversing the sic of all heliciticts
is without effect, there are no trrtnsitiom between sGzbtes of diEerent spin and
the differential cross section (3-13.19) applies to either spin state af zero mag-
netic quanhm number.
Not mueh more effort is required to obtain scattering Gross sections for
arbitrary energy, using the helicity classification of states. Now helicity changes
in scattering do occur, as exhibikd in. the general evaluation of

However, the vector structure

still requires equality of the helicities. We shall merely list relative corrtri-
budions of the various processes tE.lat appear additively in the spin-averaged
differential cross section, They are classified by initial and fins1 magnetic
quantum numbers that refer to earresponding directions of mof ion, and aeeoPd-
ing to ~vhetherbelicity trr\.nsitions have occurred,

O + 0,no:
0 -+ 0, yes: 1P
0 --+& l , yes:

X + 1, both-:

1 "4 - 1 , both:

For the last two processes, the constant --4 is the contribution asso~ialkedwidh
helicity changes, Adding these terms and supplying the appropriate factor gives

which interpolates between the high and low energy famh (3-13-24) and
(3-13.26) respectively. While resembling the zero spin result in (3-12.77), it
differs in detail, except tat high energies.
3 Fletds Chap, 3

To discws electron-positron scattering we return ta &13.1) and cornider

both .terms, with

where the two equal contributions Ghat refer La a spcifie pair of appsidely
charged parLicles h w e already bwn ~0lle~Cf3d.The transition matrix eLemenf
that is defined by %hecoefficient of

Q r * 0
(U:;.; -97 Y @Pl@lP)(%p..g'I7 ~ f i ~ p ; . ;-P)
4-
(p1 +~ $ 1 ~

The: crossing re1a;tion btween this matrix element; and the one for like charge
n abmrption prw
scaLkring again follow fmm the unification of e ~ m i o and
in %hefield $(X), as convey4 by the formal substitutions

When ~lppliedts (3- .13.7), the h~ubstitutions

produee (3-13.34), with the additional &nu@ sign coming from the reamange
men& rrecaBaq to realize (3-13.33), the stand&rd mul%iplicationorder of the
sources [i' is omitted here] :

since the two soufees asaociatd with charge ---g must revem Iheir relative
pasition.
3-1 3 Spin processes 3W

Ia order ta treat both terms of (3-13.34) in the same way, we use the relation
of (2-6,134),
up* -'I = i o ~ ~ u ~ . ~ ~ (3-13.38)
which gives

where the now mahhing charge labels have been omitted. The effect of the
additional r5 factor is ilfustrated by

The first statement depends upon the opposite motion of the two particles in
Lhe cenkr of mass frtzme, There is an irreconcilable conflict btween the nu-
merical fector, demanding the equality of the helicities --@l and o:, and the
spinor product, which requims equal magnetic quantum numbers and therefore
opposite values of the helicities --g2 and er;. The situation is snalogous to that
for a spin 1 padiele, where the time component of the vector field rranishea in
the rest frame.
As we recogmizr?.from (3-13,.i10), high energy collisions with the same initial
helicities (a2=. g;) are dominaLed by the first term of (3-13.34), which diRers
from its andague in like particle seateering only in sign; therefore [(3-13-29)]

The high energy evaluation of (3-13.39), in the psincipal circumstsnee a2=. ---c~.;,
= I-bj, is
6 1

according to (3-13.B) and the nuU property noted in (3-13.40). The firs* b r m
of (3-23.34) contributes only to the process ~viiitbcpl = crz, and, reedling that
(PI-4- p : ) g = -AfZ, (3-13.43)
 Chap. 3

R-eget

The average of (3-13.41) and (3-13.44), the diEerential cross mction for un-
p o l a ~ m dpadicles, is

This, Loo, is identical with the high energy spin O differential erom section. As
in (3- 12.88), the simple f &etor cos4 +B relates high energy electron-positron
scatkring to high energy electron-electron scattering,
Owing to the disparity of the denorninatom in. (3-13.34) a t low en.er@e~,
only the first brm is significant for d4 2%and

A general fomula, incorporakixlg both limiting cross mctions, (3-13.45) and

(3-13.4@),can be derived, S d t h electron-electron seatte~ng,by eonside~ng
all the helicity transition8 that are powible a t inbrme&mte energieg, Bud the
crossing relations make it unnecesB%ryto do tEs, The connection% between
h&vidud tran~itionampli%udesatso apply Lo %hehelicity sum of the Squares
of t h m amplitudw. Thus one can be@n with the electran-electron. scatbring
result exhibikd in (3- 113.30) and derive the requird electron-positron form.
The crossling tr%asfomationpi t., -pi implies

or, Lrans1ate-d into the g8rameters M and 8,

retains its meaning. As a quick illu~trationof the procedure leg us make t h a e

substitutions in the bigh e n e r a limit of electron-electron scattering, in&-
G&& by

= cos' 4s
3-1 3 Spin 4 procaraes %H

The relation between. the two high enerm differential erass sectiom h- bcomc?
quite transparent.
When the substitutions (3-13.48) are introduced in (3-13.30), the general
dectromi-electron. differential cross section for unpolarisd barns, the result i s
the eorre8pondin.g electron-positron differential eross section (the kinematics1
factor 1/M2does not take part in these transformations, of course):

Although written in a slightly different way Chat exhibits the dominant, low
energy and high energy behaviar, the fimt, square bracket, term is idenbiezlbl with
the spin 0 diEerentia1 eros9 section for scatfclering of opposite charges. The latkr
was only stated implicitly in (3-12.87), one half of which is the entry in the
square bracket of (3-13.51). The two additional terms in this equation %re
relatively negligible at both low and high energies, but they can be of quantita-
tive significance at intermediate energies,
In order ta illustralcr the scatter;ing of dikrent kinds tlE charged partictes,
we shall also consider the interatetion Between at gpin O partisle and a spin g
particle, The appropriate electric current vector is the sum of tbow associated
with the two types of particles 8xf.d the interaction term in is

We can write the transition matrix element directly:

where all primed yuankities refer to the spin O par$icle. This will also extend to
the masses, m and mf9auf the spin i$ and spin O particle, respectively, A, simplifica-
tion. can be introduced with the aid of the h t a l momentum,
P == PI 4-~ "$32l3-P$, (3-13.54)
for
?(pH df-5) = 2 W - - (3- 13,55)
snd
u ; ~ ~ , Y ' Yf( ~p6)up,,,
~ +
= ~~;,.,Y'(YP m ) ~ ~ 2 @ 2
= -2u,*,.,(~ -- my0)up2.2, (3- 13-56)
in ~vf-rlchthe last form refers La the center of mass frame. The introduction of
310 Fields

the heIicity consGruetion (3-13.0) gives

lvhere the kinematics of the situation, state that the eleetron enerw and mo-
mentum magnitude, whieb remain unalkred by the collision in the eenbr of
m m frame, are

At high energies, where the: individual p%&iclemass- are negXi@bXe, the electron
he:licit;y is maintained in scatlefing,

and
n2 cos2 4,
--------.
dc
M >> m, m': din sin4 48
When, a t low energies, the electron. spin is referred to a fix& direetion no change
in magnetic quantum number aecurs on sealeering, and

The general result, s u m m d over final spins and averaged over the initial
spins (the latter process is unneeessaq here) is

It is interesting to consider the two limiting siGuatisns in which the mass of one
particle becomes v e q l&rge,not only compared to the other m s g , but Ls the
total eaergy of the second particle, ff m' iis that l a r e m%,it is m m eon-
venient to introduce
df - m' + -4 0, (3- 13.63)

where the electron mew in the center of rnam fiame i e indistinguiehable from
the enerw in the coordinate system where the h e ~ v ypadicle remaiiols zr;t re&.
The limiting process givm
spin 9:

The analogous limit in whieh it is the spin .5; particle Chat has become v e v heavy
has the same form as (3-X3.M), kvithout the trigonometric faetor in the numer-
ator :
&a
spin 0:

fbo now superfZuous prime on m has been. omitted. The two diflerential cross
sections have been identified with the moving particle, the very massive one
acting only a%a stationary charge. The possibility of applying t.t, gource dt3;scrip
tion to this circumstance will be developed in the next section.
But first let us examine some processes involving photons and spin -& pa&i-
des, as eonLained in the expregsion (3-12.a) for Wa2. Electron-positron anni-
hilation into ttwa photons is described by
(dz)(dz')~., I ( X ) G+(z - zr)eq"/A1(z')J.z (X'),
(z)roeqr~ (3-13.66)

and the coefficient of i ~h l ~; ; ~

h;iq,;,;
; -,iq gives

The B. E, symmetry in kih; is evident, and the F. D. antisymmetry in

p ~ @ z pp;@;
, --p can be verified (recall that rr = --~~r,r@) with the aid of the
kixlennatical relations
JVe shall find it more coxrvenient to tvrite the dyrramical factor of (s13.67) as

In the center of mass flrarne the energies of all electrons and photons equal
*M,and
--2p2kl = *Jf[dl -- ( A f 2 - 4m2)'12 cos 81,
(3-13.70)
-2p2k; = $fii[Jf + (MZ - 4m2)'iZcos 61.
Particularly simple is the annihilation of sloif--lymoving particles, df c=t 2m,
for ~vhieh(3-13.69) &comes

Multiplied by 2m, this reduces to (?Or = irs@)

(e2/m)@'v%,.[c- e*, i. ee'*]a- kv, = 2e2igf6,.8ee X et* . klm, (3-1 3.72)

\%-herethe unneeded causal Xabels have been dropped. Only terms with an even
number of Ys fachrs survive here, and tl-e have used the fact that e X ef must
be directed along k. The equality of the helieities staks that the two magnetic
quantum numbers are opposite, in the antisymmetrical way impIied by Lhe
faetor G'. Accordingly, only the singlet state of zero htef spin can, undergo two-
photon annihilation, for slo\vly moving particles, The corresponding aero sngutar
momentum state of the two photons, a linear combination of the two equai
helicity states, is identified by the perpendicular polarization vectors of the
%\Wphotons, When we recall that

the differential cross section per unit solid angle for a given pair of perpendicu-
larly $arized photuns, with the particles in the singlet state, is oMained as
(uz/m (l/v). To compute the total annihilation cross section for unpolarized
particles we must supply the follo~vingadditional factors: 2, for the number of
polarizations available to one photon, the other polarizatisn b i n $ fixed by the
requirement of perpendicularity; 4, the Aatistical -tveigbf of the singlet sts;te;
2w, the total solid angle aecessibte to either photon \vi%houtduplication of the
fins1 states. This gives

which is half the analogous spin O annihilation cross weetion,

I d is inkresting to observe that the folXotving effective interaction term,

will directly produce the traxlsition matrix element expressed in (3-13.72), ~ v h e ~ r

evaluated for the same low energy collision. Its space-time locality, in contrast
with the nanlaeal structure of (3-13.@6),is a specific reffectioxl of those limited
energetic circumstarsees h hi eh prevent any more detailed space-time character-
ization of the process.
I n tz high energy evafuaticln of (3-13.69) the mass ns ixl the rlumerat~r~vould
be neglected. We shall see that this is justified if one excludes very small forw:itrd
or beckward scattering angles,
sin 6 p> m-/df. (8-13.77)
Then, since the resultixtg matrix commutes with ? g t the helicities are maintained,
--g; == cr2, and annihilation occurs only in urlit magnetic quantum number
states. One must be careful not to confuse the latter statement, tv1.tich refers
to the spinors a,;,; -,and upZCz,, with the properties of the spinors U;;
U,,,, where the magnetic quantum numbers are opposite, since the helicities
and-.;
-a; and 6 2 are equal. Written in s simplified notation and multiplied by
2m/e2, the high energy version of (3-13.69) is (--cf =r a):

f L is convenient to use the photon msmentum k as the spill reference direction-

the z-axis. Then the arthogonal particle spirkors describing magnetic quantum
numbers of &4 in the p direction can be ~vrittenas
v, = v+ cos fiB + v- sill $4, U?.. = --D: sill 18 + u* cos i 8 . (3-13.79)
The photon polarizzltiorr vectors appear in the combinrttioxls

tvl~iclrhave the effect of rnisirkg and Iotvering particle magnetic quantum num-
314 Fields Chap. 3

bers by unity;
*(g. + i.,)a- = v+, v:*(@, - i@@,
= V;,
(3-1 3.81)
4 -+ =, v:+(@, + iq,) = v*,
all other combinations being zero. Thus wet can ewily work out the values af
(3-1 3.78) for any choice of photon helicities. With X = --Xt = +l, for example,
tvs get

-- 2cot 38, (3-13.82)

and similarly

t,r;Izile null results are obtained far equal helicities, h = X L & l . As in the
spin 0 discu~~ion, high energy annihilation photons carry only the mmimum
angular momentum, & 2, along their line of mation. Again there i s an elementary
interpretation for the geometrical. factors of (3-f3.82,&3) which appear ixr
transition probabilities ns sin2 B cos4 48 and sin2 B sin4 +B. These are the spin 2
probabilities that connect rnagneti~quantum number +l in a given directioxl
wilh magnetic quantum numbers +2 and - 2 in another direetion at, the relative
angle 8. The transition amplitude factor i/sin2 B also appears for spin 0, in
conjunction with the geometrical factor sin2 B, \vhich produces an isotropic
differential cross seetion. No\\-, hoi~ever,the spin averaged differential cross
section is

and this alteration in angular distribution is attributable entirely to the change

of the initial state from zero to unit mbignedie quantum number,
It is the singularity of this diEerentx%tl.cross seetion a t angles B = 0 and n
that denies universal validity in angle to the high energy evaluatim. These
singularities are spurious, and trace back to the failure of the higEz energy
approximation
3-1 3 Spin 4 processes Slti

at @ == O and r , re~pectively,A suEciently more %cur& version i~

which is d w wfuX in the form of the product

making explicit the origin of Lht: angle reestriction. (3-13."1"1) If ls-auld m m thaL
one had only to replwe (3-13.M) 1~ifE.1

leading to the total annihilation cross sect;ion.

and this is correct. But there is more here than mels the eye.
When the improvemends of (gL3.86)are: introduced in the denominators of
(3-13.78) and thereby in (3-13.82,83), the resal$ is

which. is not the same as (3-13.88). Xn fmt, something h= b e n onziLM and

that, is the eonlributictn of the m-kmms in. (3-13.69), which ara not negli@ble for
sin B -- m / M , Them terms mtieommuk with YB; the helilici%yraver= ar
B; = cr2, and only initial sLraLes of zero magnetic quantum n u m b r are signifi-
cant. Multiplied by 2m/e2, this contribution to (3-13.69) is (g' = o)

Since this nneGbanis~nris unimportant under high enerw con.diti~m,except for

small v d u e of sin 8, we need not distinguish Between %bedireeLions associated
~vithfhe vectors &p and &k. A particular choice of spinom is
and only photons with the same helicidy can be emitted:

That give8 the follo~vingsupplement to the spin averaged differential cross

~ection,

and its addition to (3-13.90) effeekivefy produces (3-13.88). Node th8L the
differential cross section for forkvard and baekrvard emission comes entirely
from this last process, The value of that cross section per unit solid angle,
$(a2/m2),differs only by a factor of 2 from the result of the low energy calcula-
tion, tvhen. the kinernatical factor 2/v is replaced by its high energy value of
unity [Eq, (3-1 3,73)].
The general evaluation of (3-13.69) involves little that has not, been en-
countered af;high energies, apart from the frevenl appearance of the parameter

The helicity eonstmction of the spinors in (3-13.69), dogether with the factor
3m/ez, gives it as

Here is a list indicating the various possibilities:

 Spin 4 processes 317

The only transition not considered in the high energy discussion is tbe one with
zero initial magnetic quantum and fins1 magnetic quantum number of &2. 11;
has the anticipated geometrical factor, sin' 8. The immediately obtained form
of the differential cross section for unpolarized particles is

Explicit here is the contribution of the only pfoeesses that oecur at sin 8 = 0,
thoslc with zero initial and find magnetie quantum numbers:

2 I a2
(sin B = 0 ) = - - (I f
4 rn2~ ~~h
They are also the snIy ones that survive at low energ;)r. It is the funetion 1+
that produces the variation by a factor of 2 in proceeding from Iou. energy
( K = 0) to high energy (K = 1). Another presentation of the differential cross
section is

The laat term can be neglected Ett high energi:es, a d we recover (3-13.88). The
total annihilation cross section is evaluated as

which reduces to the limiting forms (3-13.74) and (3-13.89) in the appropriate
circumtanw.
Apart frclm changing the kinemaka1 faetor K'"-" inta K, the same digerential
cross section (3-13.101) rtpplies to the inverse process of electrcm-positron
creation in the collision of two photons, Factor8 associated with the summaGiarr
M8 Fields Chap. 3

over final potariaations md the averaging af initial onw do not change ~inee
both particles, electron and photon, have two possible po1aril;stions. But there
is a dtifference in the evrtfuation af the tokal er0863 metioa, far electron and
positron are distinct particles and the full solid angle of 43r applies. This gives
the tab1 pair ere&ion cross scletion

The differential crass section for electron-photon scattefing can. be derivd

from the electron-po~itronabnnihibtion differential cross seetion by means of
Lhe crossing substitutions
~ "-z PI, k+ -kz, (3-13.108)
The corresl)onding LransEormations on the parameters M and 8 an: indicateci
by the combinations

from which we de~vcr-

In the l i d of high energies t h e ~ eeomespandenees ~ i w l i f yto

If we apply them to the first version of (3-13.84), we get (the kinematics1 factor
l / M Z is not involved)

mat Is the sig~ficnurceof the d n u s &m?

3-3 3 Spin prmesm

Consider the individual transition m&trix elements, which are mdtiplm of

cot +B and tan They change from red to i m e n a r y values in wpm fo
%hesubstitutions (3-.13,108). Sin@ it is the absoluk square of the mtk e l e
m a s that @ve probeibifitiw, the additional factors of i are bout efllw1, bat
i2 rz=: -1 makeg a ~puriowappearance when the c r o ~ ~ i nmbstitutions
g am
applied directly to Lhe difperential crom section. The gener&liLy of this @gee%,
fos c r o ~ i n gsub~titutionsinvolving a single spin 3 parliele, can be
with the aid of a technique that we have nat u e d t h w far-the evalaatian of
pola~zstionmnmations of transition probEEf3ilitim by meam of the 8pinar

The crossing substitution on spinors, a,, +-B u: _,, which is effectively produced
by the aegative complex conjtxgsltc?of (3-13.1 l@),
gives

whereas the formal repfacement of p with --p ia (3-13.110) pro due^ %heneg*
five af this rwdL. We did not eneounhr this phenomenon in retbting elmtron-
electron sctatb~ngto electron-positron scattering since two spin. # ~ubstitutiom
are used there.
The high enerm limit of the efttctron-photon diEerenti8l crow sc?ction. far
unp1sriz;ed particle8 is, therefore,

where %he%WOferms corre~pondCo callisions with equal and with apposik s i w

of the iaiLid helicitiea, respectively; the dectron and phobn helicitim are main-
bined in scatkrlng. The apparent s i r r m h ~ t yat 8 = W is removed if we use
(13-13.88) and the high e n e r u commponderzee

The result can be present& as

and %hecome~padingtotd cross sation is

To get the electron-photon diffemntial cross seietion a t arbitrary enerrS;c;s we

make- tha appropriate subsahtbns in (3-13.101) and remove the Gnexnatiaaf,
2 Fields Chap, 3

factor 1 / ainee
~ naw inifial and final particles are identical. Thia give8 directly

The Imf tQ?rmdoes not cantribute EtL hi& enerdcts, where (3-13, P 14) is reg&ined,
nor a%low energies where fhe Thornson cross ~ ) e e t i emerges,
~n

The total cross section for electron-photon sedteringt: is

3-14 SOURCES AS SCAnERERS

The photon wurees that appear with inereming powers in Lhe interactioxls
'I;tTzl, ]FV221.. . can also be umd in the exbnded gense fo give an idwEz;ed
description of charged particles. As we have already sugge~%ed, this Bimplifid
treatment is appropride for a particle that is sufieiently heavy fo be uniaftu-
eneed by its ecattering partner. Conaider then s point charge of stre~@hZe
"Ghat ia stationed a t the orilyin,
for which the potentials are

Beginning with spinle~aparticles and the inkrackion W 2I we have, rc3premnting

a scatkring proww,

When dealing with an immobile matterer, et11 reference to momentum camerva-

Lion ie last, but enerw tzon~mettiongurviviE?~.The defirrition af the transition
matfix i ~ai eome~pondindysimplified version of the general definition (5-12.40),
mta-ining only the time in@& factor, and (3-12.43) similarly degenerak~to
a statement of the tramition pmbabilily p r unit time that digplays an the
ri&&hsnd side a single fwtor of 2z and the one delta function thati establishe~
enerm consenrslion. In the premnt situation, then, the tr&mition m a t ~ xifs

giving the transition probability per unit time ess

A diflerentid cm88 ~ c t i o nin angle is pmduet?d(on dividing this by the incidenk

particle Bux, 21p( do,,, and integrating over the well-defined final energy,
p; = p'; p:
322 Fields Chap 3

QP

which does indwd agree with (3-13.65), apart from the us@of m the charge
af the stationary scatterer.
+
The similar consideration for spin begins with

wbich is expressed by
1,,u,,) = 2m(dw,, d ~ , , ) " ~

and the transitim probrabiility per uniL time is

When heli~itystates am used,

where %hefactor slin f i e second entry reproduces the algebraic s i p s that are
r?xhibi;t;ed in (3-13.17). For either ehoice of a2 the summation over (zl dvet3;
the differential eraas section

as contained in (3-13.M). It is quite clear, in (3-14,13), that the electron retasixlg

i b helicity a t high energy while the spin remains inert in spaee ait low energy.
Xf IFzx describes scattering by the fixed char@, what do TVz2,WZ3,. . .repre-
%at? Consider, for example,

Only the field $ ( X ) refers to propagating particles and thert3fore Wgz dso
describes an electron scatkring process, as do all the &her Wz,, Thus, the
expansion in powers of the static potential A@is no longer a classification into
sue~essivelymore complicated prooesses, but represents successive latpprlroxirn*
tions ta the complete treatment of the rnation of the padiele in the Coulomb
field of the point source. The inkraction slreletan here acquires more substance,
and thereby indicates one aspect of the dynamical scheme that is generally
lacking a t the first dynamical level, namely, the? possibility for unlimited repeti-
tion of a particular interaction mechanism.
Sinee all po\vers of A' contribute to the scattering proeess one should like
t o avoid that power series expansion and work direetly with the appropriate
, or G$(x, 2'). Unfortunately, the ability to solve
Green's function, ~ $ ( z X')
the Green" function equations in a reasonably closed form is restrickd to the
nonrelativistic limit, in the physically interesting situation of a point source and
the Coulomb potential. The latter has a simple connection with the diflerential
equation (3-11.36) for A$, whieh here assumes the three-dimensional form

when one introduces the time Fouricr transform in this time translationally
invariant situation:

The transition to the nonrelativisti~limit is conveyed by $he altered meaning

of energy,
(p0)2-m2d2mE, pa--+m, (3-1 4.18)
and the term quadratic in the scalar potential is neglected. It is this omission
that produces the essential difference between Lhe two regimes, whieh other~rise
are connected by the reciprocal correlation of (3-14.18). T h u ~if, we begin with
the nonrelativistic farm of the differential cross section and i~ztrodueethe
inverse of thc substitutions (:3-14.18), we get

in agreemexlt -with (3-14.9). This is, furthermore, the exaet consequence af

(3-14, X), with the (A O) term struck out, since the nonrela.Civistic solution has
the special property that tlll higher polvers of A', or Z, lead only to a multiplice-
tive phase that disappears on forming the transition probability. Accordingly,
the first sig~lificantdeviation from (3-14.9) sriwl.;from tlte last, quadratic term
in, W Z 2 ,Eq. (3-22.29). It produces the following modification in the transi-
tion matrix:
6(1,,,1T/1,,,) = (dw,, dw,,)'/2(~a)z (dx) exp[z"(pg - PI) xl
l~herc
1
( d x ) exgfi(p2 - pl) . x] ---- ==; 417
[XI"
3 Fietds Chap 3

The earreetion. to the transformation matrix is indicftkd by the substitution

which chaagm the digerential cross =%ion far pin O m g ~ t k into

~q

The fmfar Zip implies that the comec%iondiminishes the cross smtion for pa&i~lm
of like eharge and in~remwit for the serttbfing of agpossite charge#. f n dealing
only once with the egmt of the quadratic inkr~etianbrm and ignoring %he
phase f a e m tbti reprment the consquenee of r e p a b d linear interwtisns, we
have obtained only the first term of a rnultipli~ativepower eEieb4 in Zaq. Gs
thi~3first tern displetys, there must a l ~ be
o at leaat one faetor of %hepaft;icfe @p&,
&nmthe corrmtion i8 8 dfttivi~lrtiephenomenon.
T b comegpanding discussion of spin. scatteriag proc&s mmewhat mom
indhctly since both the deaird relettivisfie correction; and the repetition of the
effectiva nonrelativistic inbraetion am combined in TTzzl Eq. (3-14.15). Taken
as i t ~ & @ n%be
d ~Xathr
, implies the following %r%mition matfix modification,

rvhere the time ink?gra%ianhm i n t r d u o d the Fourier trawfom of the electron

Green's funetion, which. appsr8 as

In the nonrelativistic limit, r@--+ 1, 7'7 = i y 5 s is negleokd, p' --P m, snd

@@l2 - m' + 2mE. This indieate8 the stmeture of the terms that sre to be
rqarded sfs already included in the phme faebr, The nonreiati~sticprewnce
of 2m in the Green's funetion implies that zpOis its reletivistic counterpart.
Indeed, the Adition of the relations
supplies the replacement
+ PO -- *(PI%f P?), (3-14.n)
witbin the context of the spin m a t ~ xelement u a d irr (3-14.24). Aceardingly,
the mtuaf earneelion contained in (3-t4.N) is (but see below)

where

The symmetry of this integral in p1 and p2 idicates Lhat V is direct& along

p1 -$- pz and we- therefore write:

with

Before diseuming this integd, let us obsente that, when hefieity staks are used,

which meam that the earreetion is confixlied to tramitions in, whieh the helicity
does not change. NOW the last b r m of (3-14.311, with %heeomfsnt, fachr
+(pI -- p2)%,can be identified ae altering the phase associated with helicity-
presehng trawitions. Put another w%y, this term is ima&n%qand, to %he
accuracy with which we are wol;king, it does not inkrfere Bvith the prheipal
eont~butisiantto the scattering matrix taad can ne:glwbd, &long with the
imrtginav parts of the other term, The remaining real hrms of X, the only
significant ones for the cross section modification. we
and the two equal inkgals illu8trated by

where the return t;o eoardixlah space fiss b e n advsntagww, Also utilizt=d is
the thre-dime?r;tsian.almomenturn i n b ~ a l
X ed((p@j2-m71j9zt
(dp) * (3-14.35)
) +
( 2 ~ p2 ~ mn -- (p')a - is 444
The mdifiesltim in the transition amplitude for scattering without helicity
change isJ conveyed by
1/2
cos 98 1
sin *@(l- sin l@)
-4-
ah2 *8 sxn2 46 cos

and the corresponding differential cm88 section for unpotaxlised padicfes is

sin *@(l- sin 38)

In this resuit, anid in, the structure of the coordinate intqral of (3-14.34), we
recognize s mechanism thst is common to spin O and spin $. The sina c o r n -
%iaah m ia speeifict to spin %.Another way of writing the last, rsquwre bra~kee,
faclor emphmizes the msoeiatiorm of the correction krm with helicity-prwrv-
ing $ramitions :

Before con6ixrGng the dkeu8sion, W@ shsfi evaluste the imaginaq pad of

8, wjbeh played no rob in the crass section calculation. On r e c o ~ z i n g&at
infegrates the vector n over the unit sphere. The unit vectsrs nt,na g p c i f y the;
directions of pl, p ~ .The inkgra-f e m b ~ r . ~ d t e n

+
where f (al nZ) and f (Bl - nz) are perpendicular vecton of length cos f B
and sin .$8, reapeetively, Basing a choice of spfie~calcoordinaLes on .them @vea

1 cos 98 (ess *B - p)
2;; dp (1 - F eos 4@)z - (1 - sin%)Bcos2 p
ge cos +e - p - I
dp 1 -- p cos #B -- 21log - .
sin 88
(3-14.43)

Thus, the complete walualion of S is

The complex structure of S, expressing a relative phme shift be;Lwmn

kelicity-presewing and hdicity-ch@n&?;irag transitiorrs, hrts a physical inalplica6ian
that cali best be appreeia$ed by relinquishing the helicity desoription. With
an. unspecifid choice of the U@ spinors, we pre~c3nL$he transition m a t k ;ast

(IPI"IPITIIP1'2'1) = 2m(d@~l
d@P*)fi2 MU.,, (3-14.45)
where
M = = f+2'go"-nl Xng
and

The computation of the btttl differential cross srtction for an arbitrzlry idtid
spin invofve~
Accordingly, if f/g ifs a conrplm number, there ia an explicit dependan@@
on the
initial spin, proGdd it has a nonvanhhing expectation value in the dirwtion
pwn&c61ar to the plane of scattering. A state of definite helicity dws not
have that properLy; it require8 a linear combination of the helicity stabs,
b i p r o c a l to the dependence of the dtiEe:rential eram section on the i ~ t i a l
pin Is the appearance of plssizatian in the parfieleg far an inithlly
unplsriaed beam. At a given scattering angle, the average final spin is

where v is the unit vmgor normal ta the s3catb~ngplane,

ng X B% = in Clv,
and

In the special situstions exprwsed by f == =trig gin 6,the wla&ation is complete:

p = &l, Notice that the game polw~~etti.on p&r&mekrexprwBw the relative
de~nde-nmof the scatbring differential cram =&ion on the i ~ t i spin:
d

This effect can be demomtraM experimentetEly by a dauble s e a t k ~ n g&mange-

men&,6 t h the fim-t,a3cafbAng act pola~eingthe padielw and %hemeand one
defecting &at pllitri~ation. If m dmipate them m a and b, the insadion of the
polarisation produmd by the fimt Mwfion into the cross section for the seeond
scatkfing the reXB;Live factor

The obmrvationatt s i p that neither p, nor pb is zero thus eomm from a de-
pendenee of the final inbn~ityupon the mlative orientrttion of the two aatfefing
plsnes. In pa&iculsr, if thf? two phnm are the same geometrically but deflectiaw
in opgosik seases am cornpar&, .= &v,, the ratio of the inbwity for &flee-
%ion8in the 8atme seme to -tlaat for the opposilti! s e w is

This it3 r &tan one when the individad 8eatkfing a & ~are iiientjeal,
p, == pb. The preference for sucemive deflection in the same m m will be
b i d with any choice of individual seatte~ngaaglm if, W in the pment
di~ussion,the pohfisation paramettt3.r hadJ a definife! sign at aU angfes:
3-3 4 Ssurees as scatterers 329

We turn now to examples of the class of phenomena in which both simple

and extended photon sources are involved. These are ehared particle inter-
ation8 witb fixed charges, in which phoGons are emitted or absorbed. The
simplest illustrations are contained in W22, They nre s i d e photon emimion
during sealt;ering in a Coulomb field and the creation of a pair of charged
psrtieles by a photon passing through a Coulomb field. The relevant part of
ITzz for a spin O particle that emits a photon during a collision ~vitha mwsive
pa&icle of charge Ze is

where A$(%)indicates the vector potential associRted with the chwge Ze. W~iing
the form of the latter that is stated in (3-14.2), the transition matrix element
appears as

where

and npis the unit time-like vector that has the single component no = 1 in the
rest frame of the charge 2%. Enerw conservation takes the form

iL is used in verifying that fb eEectivc: pkobn emission source is conserved,

for this is the algebraic pmperty

First let us observe the simplifieations that appear far soft photons, where
the photon momentum k is negligible compared to p1 - p2 and = P.:
This @ves

whieh is the trntnsitioo matrix element for scattering in the Coulomb Eidd,
multiptied by the probabiliLy amplitude for photon emission by the source that
represents the instantaneous trangition of the eharge ep from velocity p z l m to
velocity pl/m. This conforms vvith expetation. We should remark, h o r n @ ,
fsr future reference, that the connplek negbet of the pboton meehaaied proper-
536 Fields Chap. 3

ties a t sufficiently low frequencies is justified for finite particle deflection angles,
but require8 m r e careful eonrsideration when the deflecdian angle is very s m d .
Glomly rebted fo the soft photon situation, but &skinet fram it, is the low
enerw or nonrelativistic Iimit. Here, the photon momentum is negligible but
any frastion of the initial kimtie enera can be r a d i a w as the photon energy

Using the gauge eE,h, = 0, the trsnsition matrix element simplifies to

and &hediEerential Gross section for specified polariaetLion, emisgion dimctions,

and photon enerw is (ulineee~~ary labels are omitted)

The successive operations of summing over polarizations and then integrating

over photon emission dirwdions duce it to

and the fufther integration over alt pa.pticlc?se&tering angles gves a cross ection
fsr the photon energy distribution:

It is also int;eratiag to consider the digerential emss per unit solid angle dQ
l;fitzL is i n h g r a t d over JI photon energies, from the minimum debetable erreray
&Einto the maximum energy aet by the initial kinetic energy T = pg/2m. This is

?\?hiehuses the inbgration variable

The i n b g a l can be evalusltd in general, moat simpXy by f t z c b ~ n gthe de-

nominetor into 1 - z, 1 $ z, 1 -- xe", 1 - X@-'@, but we shall only present
the reeulL for the circumstance
k$i, <C T , (3- f 4-69)
where

- (n - 9)tan 98 -

This is not v~tlidfor arbitrarily small angles, however. XR ~ o n t r a with

t the
apparent singutari%ya t B = 0, (Sl4.67) there yields

The apptichility of (3-14.70) requires that 8 >> kki,/T. Still another c a v a t

must be mentioned. As in the discurnion of nonradliative wabtering, Wattis only
%hefirs%of an infinik series of proeesseb that contribute to the emission af a
photon during deflection by the Coubmb field, But unlike elastic ~cft.Lb~ng
in a Coulomb field, these ndditional processes do alter the ems8 section, par-
ticularly at low emclrdrss and large Z. We do nat intend to go into this matter
bere, however,
Let us return to the transition matrix element (3-14.57) and no& the fal-
lowing expression for a digerential eross seetian that atill refers La the dedailed
energy distribution :

where fhe ~urnnnabionover the polarization of the emitkd phoGon is given by

Here is another, invariant m-a,y to write the digerentiaf eross seelion:

although wc? have not troubled to iatroduee the invariant equivalent of the
initial pa&icle momentum. The four-clirnensionaI delta function slaks that
382 Fields Chap, 3

and, in the rest frame of ?P,which is the coordinate sysCern of physical inbrest,
b(nkz) &(P! 4- k! -- p:), +
k: = (PI kl -- p,)2, (3-14.76)
which regsin~(3-14.72). But the expression (G14.74) has a sugestive eharaete?r,
for process= rmenbling psrtiele-ptrobn 8esthring are being considered. Of
course, the incident photons are virtual, since kg > 0. Nevertheleas, this point
of view haa prmlical advantwgw a t high ener@es. Viewed in a wifable co-
ar&natcr;system, a, major fra,cdion of the digerential cross section can h evaluaM
in terms of the propedies of r e d pholons.
fn the physical coofdinak system, the incidcnt pa&iele is considerd to
move along the %-axiswith velocity et ru 4-1, so that

How think of the coordinate system in. which the particle is at rest inifiablly and
the- charge 2% moves dong the z-axis with. vdocity --D. In &is frame the vector
n@has the components [(O, z,y, z)]

The requirement nk2 = 0,which asserts tbe static nature of the field in the
pbysicd, or &attached, wordinate syslern, becomes in the particle rest frame

and therefore
kg = k$ + kgl[(l - v2)/u2],
where
kg = kK + k:,.
Thus, in circumstances for which

and k i is sufficiently small, it would seem that the virtual photons dould be
approxirnaM by red ones.
There is one app~retltdifficulty, hawever, Playing the role of pol8riaation
rrecbr for the incident photon, is the vecbr np,which is indeed such that nlcz = 0.
But we should etxgecb that the pola~zl~fion vmtor of a red photon is, or can be
cho~en,without time component or component along the propagation direction,
which i s here the negative z-axis. This suggesGs performing whrat should be a
gauge tran~formation:
n' -+ nC - (?/kg)%, (3-14.83)
w h i ~ his comtrucbd bhave vanishing time component in the pa&iele rest frame.
The z or loxl@tudinal component of the new vec-tor is then
and, provided
r k ~ l k>
;> I/T,
the transformed veetor w i U be predominrtntiy the multiple -?kT/k; of the
tramverse; unit vector kT/kr, which acts as the ineidrtnt phokon p o h r l h o n
vector, But all this is contingent on. the magnitude of the addidiond term intro-
duced by the transformation (3-14.83)1 which is p r o p o r t i d to

Now,
k z ~ g= kip, - *ki, kzpl = klp2 + +ki (3-14.87)
and therefore

whieh indicates that the substitution of real photons for the viPtu~t1pElotOns will
be justified if suitable upper limits are placed on kg 2;: k:. A suggestion of the
magnitude of this upper limit is obtained by comparing, in $he gauge ef, &,pz = 0,
the pa&icle r e ~ frame
t values

and

namely
k~ < m. (3-14.91)
We shag eonfine the discurnion to the diEerential crag@section that $ive8
the mew spectrum of the d t L t n l photons in the Z eoordinete system. Since
we are ROW f i d y establi~hdin two &Rerent eoodinate systems moving reh-
tive to eaeh other at pmctieally the spectd of light, a few notational distinctions
are needed. The Z h m e photon e n e r a will be denoted by
K = --nkl= r(k! + vkl,) r: rk!(l -- eos 8), (3-14.92)
where B is tbe photon scattering mgfe in the paPticfe frame, The kinematics of
tbe phaton s c a t k r i q process in that refewnce frame, as derivd from
0 = (p, + kz -- k112 + 'm = -2m(k$ -- k:) f 2kyk$(1 - cos 8), (3-14.93)
334 Fialds Chap. 3

is expressed by
k? = k: (3-14.94)
1 + (k$/m)(l - cos B)
Some derived relations involving K and the Z frame particle energies
E2 = rm, E, = E2 - K (3-14.95)
are
-K= (k!/m)(l - cos B)
E2 +
1 (k:/m)(l - cos B)
% =k! g P
E1
(3-14.96)

and

The latter shows that the incident photon energies k: that can produce a scat-
tered photon of energy K (two different coordinate systems are used here) will
be restricted by

Also useful is the differential relation

E2 dK k:
- - = - sin B dB.
El El m
Considered in the rest frame of the incident particle, the differential cross
section for photon-particle scattering is

where the factor in the denominator arises from division by the photon flux
multiplied by the particle density, 2k: dok22mdo,,. The polarization summation
and average is

f C le:,r, . et2~,12= 1 F [l - (%er,r,)2)

X I As
= f (l + cos2 B), (3-14.101)

which also follows from (3-12.124), with n = p2/m. The final momentum inte-
gration can be performed with the aid of the kinematical relation
and

which eould also be produeed by transformation from (3-12.117), the center

of mass expressian. for the digerenlial craras swtion. The form in which we
shall use this digerential cross section is obtained from the relations of (3-14.96,
97, 99) EM
[mkido] = 2ra -
2 dK 1
E% 2 (3-14.lM)

The real photon contribution to the differential eroas section (3-14.74) is

or, with
(dlez) = r dk$ dk2,dki,

If we change the seale of the incident photon energy in the follo~\ingmanner,

so that z range8 from X to m, this r e d s

Acoording to the restriction (3-14.85), the kg integml should be stopped a t a

2 , that ia large
louw limit that is s fraction of ( k g / ~=) ~( m 2 K ~ / 2 E 1 E 2 )one
compared with l/rZ, say l/?, but a negligible error i s introduced by extending
the inkgral down to zero. The value of the integrerl, un&r the condjifions
and &hen

The suggestion implicit in (3-14.91), h, S m, is that no significant inter-

actions oceur for larger values of the Cr~nsvemmomenhm, We propose to
e x a ~ n Chip,
e que8ti:on.
A quick indication of the quan$itaLively eorrect resullt is obtained i f one
merely aempk %hatthe eEee.t;irre replacement for &, ia indepndent of' K .
Then it sufiees Lo conrzpgre (3-14.112) with the Merenth1 crass swtion Bppr*
p ~ a l eto soft photons. This diwwsion tsktls place entirely in the &attached
physical coordinate system. The paliafization summtioo, in $he digerential
e-mm swtion derived frsm (3-24.61) i s

This expres~ioncon$ains the only reference ta the direction of Lhe emitttld pholon,
md we shall integrate over all solid angles. Removing the factor l / K Z ,that
iategal is

where R is now the uniL propagation vwtor of $he photon, and high enerw, @oft
photon simplificatiom have not yet been irtfrducd. We first obgeme t h ~ t

To inhgrak the km containing G

m denominabrs, it is ub3ef~ltO
(EE-J E2 --- E )
3-1 4 Sources as seattrrrarsl 337

as one can verify immediakly. Then we have

The v integration can also be performed, sf course, bud it is preferable to leave

if as id stands.
The soft photon differential cross section is

Lvhich still needs to be integrated over the deflection angle 8, But now we must
recall the M-arningthat the soft photon sirnplifications need to be qualified for
very small angles. I n contrztst to the singularity of (s14.118) at fl = 0, the
+
minimum value attained by (pl k l - p2)2, ~vhichoccurs for scattering and
emission in the fortvard directiorr, is

It ~vouldbe more accurate to replace (pl - by

and we recognize a characteristic aspect of (3-14.1 09). f f is unnecessary to

incorporate these refinements, hawever, They atre the content of Lhe real phooton
compubtion, The soft photon evaluation need only be applied at angles such that
rn2~
2E sin H@>> -t C3-14,121)
2_E;"lE2
[shere (3-1 4. X 18) can be used without correction* Ernployirlg the variable

y "=: (Elm)sin $B, (3- 14.122)

338 fieldra Chap. 3

we begin the integration at a conservative upper limit to the real photon dis-
n , << m, and thus
c u ~ ~ i okmaiX
Ymin r== (kmax12m) K< 1.
This gives

(3-14.124)
where the y integral simplifies to
I
dv(l + v2) log f. -X
---
u2

and
16 z2a3dK
- (3-14. f 26)
d@ --m

3 m2 K
virtual
Whexl the real photon contribution (3-14.1 12) is considered under soft pho-
ton conditions, E l --. El2, the addigion of the two part^ precisely cancels
fog (mp,,,) +-
H, and the inference is that, $enemlly,

The inference is comeet, as we shall verify by repeating the virtual photon

calculation without using soft photon simplifieations.
But Haraid interrupts,
H. : f have not forgotten that you decided to omit all reference to historical
nlatters, but your use of il, parametric device for combining denomintztors
prompts me to ask about a smafl historical point. The technique of introducing
parameters to unite denominator products into a single denominator is in-
variably ascribed h Feynman in the literature, 1%it not true, however, that
the usual intent of that device, to replace space-time integrations by invariant
parametric integrafs, \vas earlier exploited by you in a related exponential
version, and that the elementary identity combining two denominators,
(3-14.1 16) in fwt, appears quite explicitly in ra paper of yours, published in the
same issue that contains Feynman's contribution?
X. :Yes.
In the physical coordinak system, at high energy, radiation processes occur
predominantly near the forward or longitudinal direetion. We express this
through a dceompasition into longitudinptl and transverse components, sts illus-
whi& ase d t t e n in terms of Ihe dransverse momeRfunn 8uppfie;d by the *B
tionav charge
= h 4- PIT,
h 7 (3-14. 12921
as
--klp,=(Ea/zK)oz, - - ~ I ~ I = ( E ~ / ~ I ) ( E ~(3-14.1301 /~K)~~,

The virtual photon contribution to the differential cmss section that is produced
by (3-14.72,73) is

~.---"---..%.=--

virtual r2 E2 K
(S14.f 32)
U B ~ ebpprop~~k
R~ variable tra~glalions,we hwe

and, carnbining denominators with the aid of the v psrannehr,

The iatraduetion of &hetvariable

y = kT/2m
 Flelds Chap, 3

which doe8 inded differ from (3-14.124) only by the faehr E1/g2that i~ nededt.
Co combine properly with the general real photon contribution (3-14.112) and
prduce (3-14.127).
To give sn analogous discussion for spin 1 psrtioles requires, first, the explicit
f o m of the electron-photon diRerentital cross wction in the r a t frame of the
incident efeetron, That is waitstble Lo us through transformafion of (3-13.1 l?),
%hee n k r of nnw e r a s section, but, there is some i n k r s t in a direct derivation.
The transition matrix element is

whieh urns purely spatial polarization vectors and a simplified aodation. The
m a t ~ xfaetor in square brackerts reduces to

where the latter exploits the fact that uz i s an eigenvector of ?%with eigenvalue
-4-1, and introduces the aotafion nl, z far the unit propagation vwfmirs of the
photons. C o n ~ i d e e ~real
g polarization vecGors for simplicitcy, the tr%nsition
nnad~xelement becomes
(d%, d o k , do,,dwt,)1122e2u:[-el + e2 + +?,(g e p nl X e2
X el))uz. (S14.139)
+a*e2aeal
If the term and the spinor8 are omitbd we ge6 the comespondirtg spin O
expremion. The summation of Lhe tr%nsi.tionprobability over final spins can
be pe$armed with the aid of (3-13.11Q), giving the pino or faehr

where the kinematics of the mXIision dekrmine

The matrix prduet is reducd by omitting a11 krms that eontain a rli fa~tOr,
since uz is a eigenvector, or s a faetor, the latter expressing the averaging
over aft initial spins, A quite sho& edcufation then gives the foffo~4ngfar
 X [l - cos B(el = e2)' + el e2n1°ezag etg - el X -
ez alet X ozJ,
(3-14.142)
where
II~ = cos 8, (S14.143)
The appamnt dependence of the ~ m n d
term on the pa)$sri%stionvectors dis-
appwm on. invoking the identity
C@Z X X h3 * E@l ez) m21
= (et X e2)%
cos @ -- el X e2 * nlel X et nz
= --el . e2n1* e2aS elf (&l4.144)
and we g@$

replacing (et e2)' in the spin O cross section. When summed and averaged over
photon polarizstions the differential cmss section that appears in plme of
(Sli4.103)3s

For the purpom o f evaluating the pbobn ennission emss section, this is
wdm

acwrding to Lhe ~ecandrelatiun af (3-14.W) ~ n the d spin O rwult (3-14.104).

The e o m ~ p n d i n gmo&fication of (3-14.109, 1f 1) is

-
log 2BiEt
mK -
 The virtual photon contribution i n h r r d frbm fi2hsts Lhe fobwing high
enerw appearance :

whieh is to h aurnmed over the photon polakaation and the find electran spin,
and averaged over the initid spin. As in the dimurnion of pfxoton-elechn
s~afhringtat small angles, transitions with electron heliciw ehsngw %res i p
fiea ant, The cafeulatian can. h prformed dvtitnbgeau~lyby methods tha$
have dresldy k n illustraled, using photon helicify states and the photon
e ~ s s i o ndirection for reference, and expreming the efeetmn helieity states wilh
fhe aid of suitable rota;tion mal;rices. We shall only give the re~u1GsBere, which
are classified with respect to helieity change:

For soft photons, helicity ehsnges are fehtively negligible and the spin O &me-
tare is reproduced. The sXighLIy different integral assacigM wi.t;h KeXieity
ebanges is
1

clu

dv(1 -- vZ) log (1 -- v2)

The individual, eram section~sare

da
virtual, no

virtual, yea
which add La

&
virtual

This virtual phohn part and the real fiaton contribution of (3-14.148) cambine:
ta give the final high ernerw f a m of the diRerentia1 cross =%ion for phobn
r an elecfran defleetd in a Coulomb field,
e ~ m i a r by

The proeegs that converts a pfrofon into a pair af oppositely charged par-
tiele~,in the neighbarhood of a tafstianary charge, is m l ~ by a croming
transfomations to the reiaction jurzlt eonsiderd. T h m %ramforrnatiorrsare

In order to recan~truetthe absoXuk squard transition matrix element, we

take the known differential, era88 section for phohn emis8ian, d ~ ~ refer~ng
~ ~ j ~ ~ .
CQ definitr: spins and pola~zations,and f a m

where the second high eneru version also emphmizes that we arc?inlerf;sM only
in the energy specification of the particles. Under fhe croming transformation,
the kinematieal factor do,, dok,h,, beoomes dw,, bp; dwk,. The differential
er088 =etion refeming to an inciaent photon beant woufd require division by
2K dwr,, K = kg, and thus

ap8l.t from the spurious minus sign tkat accompanies this formsl substilution,
with spin 8 particles. When cross sectians involving summatioas over final
helieitiea and averages over initial ones are used, appropriate correc%ionsmust
be made far the different weight faclom. With spin 3 partieia this if9 '1;18f re-
quired, in thew reactions involving one initid pa&iclc?:and fwo final particles,
ainee b t h elwtron and photon hsve t ~ r ohelicity staks. For spin O padiolm,
howwer, the photon emission crass section, summed aver the phobn p~1ariza-
Lion, contains an additionztl faetor of 2 relative ta the photon sbmq%ianer088
section, where photon polarisations are averaged. The implied pair produetion
cross sections are
spin 6: cFo = - 8 z- ~ -~ Q- E ~ E ~
3 m2 K 2 K (3-14.159)
 1abls are omit$&, tomther with charge indices, since
the paditioning of the phabn enere,
K =E + Eft (3-14. XW)
doe8 not de*nd u p n the ~peifiecharge wimmentts.

In u8ing exbndd photon. sources 4x1 reprmnt heavy eh&-& padielw, tbe

the pdnf t h ~ new t kinds of p&&iele~am eneounkrd. They are idealixd

vemione; of cornpiLe s y ~ t e m ,S i n ~ eh y b o g e ~ ea b m ~
are the mast familiar
examplr?,they will be bmed H-pafiielee.
We first conerider a sbtic mume d i ~ t ~ b u t iJ"(x),
a n and the
p h ~ t i a fAl@(x)in Bame convenient p g e . The time tramlwtiona1 iinva~ance
of Green" funetiom, A+ for emmple, is conveyed by

where

The B. E. slynnme;try af this zero spin Green's function &a

in which rnat~xtramposition is applied to the eharge irrdice~. Eigemfuxlctions,

solutions of the homogeneous Grwn" funetion equa&ion,aw~ktin eonstmcfing
the G m n % function. They are Xabld by enerm and charge values, supple
mentd by. odher qumtnturn numbrs which air@usually relahd to a n e l m mo-
menturn. The homogenmus equation and ifs complex canjugab are

Notice that the joint sign reversal of p" and p' interohanges the forms of the
two differential op.etra;tors. Hence the eigexlfumtions can be so chomrr that

where a'* iindieaw a rf3X~hdSt?t af quantum rrumbrg, If G' iaeludes a magnetic

quantum number, for example, am refern to %henega$ive of that quantum
numbw. If i~ pomible to chwse the eiigenfuncfiona in a way tha&identifiw at*
with a', While not usuauy convenient for individual problem that Gbaiae
simpXifiw gener8f disewion~. To avoid canfugion, we &all a h undem%and
that p" is s positive quantity unless there is a specific indication otheraise.
 Another consequence of the equation pair (3-15.4) is the formal i n h p a l
relation
(dx)4,op,t,f(x) * (p0' $ p'" -- 2 e q f '~( X ) ) +,eoppopf ( X ) = 0,
(3-1 5.6)
is incorporated in the statement af arthonormdity :
~~rhich
(dx) +pop,r,p(x) '(p" + p"' - 2epfA@(X))+ , o ~ ~ , ~ ~ ,=~ ~ ( x ) E ~ ~ o ~ ? ~ ~ ,

(3-15.7)
f n the &Beme of the staLie source, this properQ is obeyed by the Imotvn eigen-
funcfGions mmciatd with small momentum cells,

according to the orthonomality of the % and the clarification of the spatial

nmmslization given in Eqs. (3-6.26,23). Here the momentum veebr p f l a y s
the role of the quantum numbers a, which aXso specify the energy value. There
is an analogue of (3-15.6) in which the bamogeneaus equation obeyed by
is replaced b y the inhomogeneous Green" function equation,
+ , o * ~ , t , ~ f ( ~ )

Sufficiently near a particular energy eigenvatue supposed to be isolated,

however slightly, fmm all: the others, the Green's function is dominakd by the
eomwponding eigenfuncdions, and (3-15.9) implies that,

The analogous behavior in the neighbarhood of --p"7 demanded by (3-15.3), is

W :

&+(X, X', P@) -- C +p@tgr@p(x)

q"O"
*+p@#qfcar ( X ' )
pot + (3-15.11)

A representation of the Green's function that ia valid near any part of the
physical enerw speclrum, or its negative, is given by

in vvhick we have dso exhibited the appropriatr? urn of the parameter e -+ +O

in order to satisfy the time baundrtyy. condition--pasitive (negative) hquenciesl
346 Fistda Chap, 3

for positive (negative) time differerrccs. This is verified directly:

Wheu the eigexrfuactioxrs (3-15.8) are inserted in (3-15,12), the 'Emojvn form of
the free particle propagation function is recovered. If \re nolv allow to
assume both positive and negative values, the Green" funetion can be premnted
more compactly as

~vl~ere
( f ) signals the extended meaning of and

Not to be confused with the infinitesimal parameter e is c(p''), stating the

algebraic sign of p". Another \\.ay of writing this function, in ~vhichthe scalc
of c -+ +O hass been changed, is

The time-dependexlt versioxl of (3-15. f 5) is

We urc intcjtrested, in this section, only in tltst podion of the energy sptlcLrum
which is inaceemible to a free particle: < m, Such s$atf?scan exist, localized
in the neighborhood of the source, if &hereis force of attraction btween the
particle and source, of sufficient strength and range, Tn the familittr situation
of the long-rang& Coulomb interitctiotr between oppo~ihlysigned charges, no
minimum strength is required, %ndan unIimidd number of sueh bound s t ~ k s
exists. These are the H-particles, Wllat are the emissiorr arrd sbsoqtion sources
for H-particles?
The insertiorr of the Green's funetion (3-15.18) into the souree coupling t e r n

TXre time dependent quantitim

are sources associated rvith the particular H-particle label4 by pat, p', with of
appearing as an %dditionalindex snalogous Lo pin. Thsf the= sourcw mfer
only to time conveys the immobjlity of the very massiw W-particles, The
repeated operatjon of these sources will inject any number of parLIeles info
bound statea. 8ince no a~countis being 8ven of the inter&cfctionsamong the
particles, we ~haffbe concerned only with the propertie8 of rr single particle,
bound to the murect anid forming an H-padiele. Nti3verthelesa, id i s desirable to
verify that probabili6-y requiremenk are satisfied in the dynsmieally simpfifid
many-pa&icle situation.
The usual consideration of tc causal arrangement of emission and absorptioxl
source8 leads h

and the eausaf arrangement restricts the energy summation in (3-15.23) to the
physical, positive values. The mulLiparLicle Aabs produced by the caueat
analysis of the vacuum amplitude have the usual canstmction in krms of gauree
producls, and probability normalilration implies that

The direct verification of this property employs the relatioll

For spin $ particles, the transform Green" ffunctian intmdtrcd by ivriting

b~nathe F. D.. symrnetq
[r"G,(xf, X, -pO)lT = -yOG+(x, X', (3-15.28)
and o h y s

Eigenfunctions are defind by %hehomogeneous equstiom

The tmroequations are related by the Hemitisn chsracter of the matrices

If, imkad, atkndion is paid ta the i m a d n a ~naGtxre of the nntktriees Y", we
ider the eigenfunctian connection (for suitable choices of a')

The eigenfunctions are normttEiz;edin accordanet:with the inbgal property.

earnbind with the symmetry requirement of (%15.28), leads Lo the Greenp@

function construction (pa' > 0) :

OX", more compactly,

1
~ ~ o e =( - ~~ - ~~ " l )( = (3-15.37)
- ~ ~ )
p@'(1 - ie) - *

The r?;xpEciltime depndenee of the Green's function irj given by

in which
~~o.(zO - so')= - G - ~ O ~ ( Z O ' - xO)
= n(xO- zO')v(po')i~b$?(zO- zO')
- *(X'' - X ~ ) ~ ( - ~ ~ ' (X') ~ A 2").
~ ~ O ~(3-15.39) -
Time dependent H-particle sources are defined by
s p ~ , q , a (xO)
t = 1( d ~ ) $ ~ o( x~)~* ~ Oa *t ( x x ~ )
-qta'(x0)*,
and

With a causal arrangement of sources, the vacuum amplitude becomes

(0+10-)*= (O+IO-)TL exp [ P qa
,i q & o , q t a ~ i v ~ p ~(0
~+1s0-
~)'
a"t ] (3-15.42)

where we have introduced

The completeness of the multiparticle states, which have the usual source
product representation, implies that (p0' > 0)

Direct calculation from (3-15.41), with the aid of the relation

gives

since, in accordance with F. D. statistics,

Now, let the static source that represents a heavy charged particle be
supplemented by a simple photon source. The terms in FV that contain one such
3M) Fields Chap. 3

photon source deseribe processes in tvhicE.1,through transitions between difierent

H-particles, a single photon ifs emitted or absorbd. Using two photon sources,
we describe transitions that result in two photons b i n g ernitted or absorbed,
rand also transitions in which 8 photon is scattered with, or kvithottt, an secom-
panying H-particle transition. And so on. I t is convenient to use the character-
igstion af electromagnetic prwesses in which all interactions refer directly .to
the ehaqed particles, with the electromagnetic m d e l of the partieIe source
transfornled into a gauge re~trictionoxr the veetor pokntial. We recall the
space-like ehaice [Eq. (3- 10.49)f

The static source defines a coordinate syskm in which nc"can be chown to have
only s Lime component. Then fc"(k) h= only spatial components, \vhich are
proportional to the vector k, and the gduge condition reads:

This gauge i8 called the r&distiongauge, since the proprty of transvemalidy

to the momentum vector is characteristie of the polarization vechrs asmeiated
witit photons. Xt has atso, but less appropriately, been termed the Coulomb
gauge. As i s most evident from the three-dimensional form of the srscorrd-order
MaxweX1. differential equations,

tile scalar potential A0(z) in the rdiation gauge i s necessarily given by the
instantaneous Coulomb potential of the charge distributiort,

But the coxlverse is not true. If it is required that AQ(z)shell be the instan-
taneous Coulomb potential, presumably the intent of a Coulomb gauge, the
inference is that the time derivative of "C" A(z) must vanish, No restriction is
thereby placed on any static compnent of the veetor potential, A(z).
I t is $he radiation gauge la, ~vhicfithe sta;tie potential AP(x) refer^, The
vector potential A(x) can be used to represent the field of nuclear magnelic
dipale moments, leading to the hyperfine structure af H-prtrticles. Xn the
follotving, however, attention will be confined to the static chmge density and
its scalar potential. This avoids notatiorlial conflicts with the pokntials that
are associated with the aimple photori sources. The latkr are only needed far
from %heemission or detection sources, There, they reduce to the vechr poten-
ia the transverse or divergenceless part of J(z) and, indad,

The solution of (3-15.52) in regions that am musally inbmediatk? between

emission and debction sources is, of course,
A(t.1 = C
kX
IAkX(z)iJzth 4- i~:th~kk(z)'l, (3-1 5.55)
with
ALh(x)= ( d ~ kliZet*eiL'
) , V.Akh(z)=O. (3-15.56)
We now distinguish A+(%, X?), which contains the static scalar potential
A0(x),from A$($, 2'). The latter also describes the effect of the vector potential
A(x) that represents photons. The differential equation for the Green's function
&$(X, X') can be presented as (p = --$V)

(-- (ia, - ecl~'(x))2- $ m' - i e ] ~ : ( z ,X')

= 6(2 - z') + (2epp A (z) -- @'A(%)')A$($, 2') (3-15.57)
The use of the Green's function A+(%,z') converts this into an integral equation,

which can be solved by an iteratian procedure. AIL this is entirely analagous to

the discussion of (12-12.27) and as there, the suceesslve interactian krm,
..
W%,,WZ1,. are most ~ ~ m p ~ eexpws~ed
tly with the aid of the particle field,
(3-25.59)
Thus,

and so forth.
The particle field +(X) is related to H-particle sources by
At a time tfr& is eausally inbrmdiab htween the actions of emission md
absorption sources, this becomes (p" > 0)

Aewrdingly, the transition m a t ~ xdemtjnt far a pmcm in which the H-p&&iele

labled po"af' transforms into the H-particle denoted by p@'o' (charge specifics-
tians &reomittftd 8ince only one siw of charge, gay qF8d f be bound to the 8t&tio
murce), vvitht the tr: ion of a photan, kk, is
(xltp . e:he-R'x#p@erorfi
(dx)dp@ta. (X).

(&XS*M)
The transition probability per unit time is given by

8ince enr?ra cammation precisely specifie~the photon enerm,

> = - PO', (3-15.66)
there is no n e d tA, mfer to the enera distribution, We @hall e%w out the
pola~z;ationsummation and the inteqa;tion over alf photon emimion directiow,
using, for simpfici(;y.of illwtrstion, the nonreI~ttivisdic~i%a&tiaxl.Eere the
phofon momentum, but not i b enerw, it3 negligibje, and $he p a ~ i c l eeigen-
functions are relrttd ta the con~entionallynomaE~ednonrela&iviaticwave-
functiom +*E"t (X) by

pot = m E': + cftPalaf(X> 2 -

t
czm)1 f z $&#a#Wt (3-25.67)

in virtue of the normali~ationcondition (3-15.7). Using ~ t s n d ~matrix

rd element
notatian, this dves

Since polaizatian v ~ b xealiae

m oxlly two of the three o&honoms'l unit vecbrs,
the palarisafion gumm~tionand integration aver emkion directions produce
the f~ljowingexpression for the tr%nsitionprobability per unit time;

which also wea the m a t ~ connr;.ctian

x between the parkicle veloeify and posiifion
vectors, This is the probability per unit time for spoa$antsous pbsfon emimion
in a transition between the specified H-particle states and, as such, is Einstein's
A-coefieient. A less gpecific A-caeEcient that refers only to energy is summed
over a' tand averaged over a"".
When a photon is incident on an H-particle in the state p'"a", a transition
to the state po'a' can occur if

The transition matrix element is

ttrhich implies the transition probability per unit time:

After integration over the sharply selected photon energy, this can be expressed
as the Einstein B-eoeEeient, which relates the transition probability per
unit time to the photon energy density per unit angular frequency range:
k0(2k0 dwr/dka). Averaging over the incident photon polarization and direction
of motion @yes the nonrefati-visticexpression

The simplicity of the ratio, for a given pair of H-padieles,

is a reminder that, apart from the kinmadittal factors involved in the definitions,
the transition probabilities for single photon emission and absorption, are inkr-
changed by the photon crossing transformation, k p 4 -kp, The emission and
tllbsorption rates are equal when the definitions refer to single photons of definite
polarization. And, as we learned long ago in the simpler coxrtexl of a probe
source, if n photans of the appropriate frequency are present initially, the
+
absorption rate is multiplied by n and the emission rate by n 1. The latter
represents the combination of stimulated and spontaneous ernimion processes.
The analogous discussion for spin $- particles begins with the Greenpg
function difierential equation

and the equivalent inkgral equation

The succe~aivephoton interaction krm8 arc: exhibibd, with the aid of the
padicle field
(3-1 5.77)

as; [compare Eq, (%12,26)f

sad so on, The particle fi& eotn be expre~sdin brms of H-p&&icleB Q U ~ C I ~ B ,

and, at a time internedide between %heoperations of emimiaa and rzb~orption,

gaurces, is given. by
$(X, s') = C '
[$p@larG*(~)e-ipa
0
4-
iqtp~'a8ap
*e
itl:p0~pfa8#p0tqfar(~) 1.
pQfqglo6 (3-1 5.86)
The tran~itionmatrix element for single phohn emission is

One can eombine the eigenfuncdion digerendid equ4ltions in %hc:mamer af

(3-6.67) to produce the inbgral identity

In a non~lafivistielimit, where the Jtpofa<~) m approximab eigenveebm of

ra,the last tern, containing the matrix TOY , relatively negligible. It
= i Y a ~ is
would be inconsistent to retain the W * k X e* contribution while repla~ing
@-a.z by unity, d a r e it multiplies p * e*. The next term of an expansion i s

and one mcopiaes the orbital contribution to the magnetic moment, wbiGh
ad& Ga the @pinmagnetic mantent in the mannrsr rc:presenM by g = 2. If we
neglecl thia mametic dipale radiatim, and the rdated eXeet~equadwpofe
rdiation, which is tke other b r m on the right-hand side aE (3-15.83), them
rmains only the rd4ttion of the electric dipXe moment, eq'x. This i s radiation
associded vvith aceebmbd charges, and is indepndent of spin. Inded, witEr
the sirnpiifieations irnpIjed by retaining only the firf;t two te
and replacing spin 4 eigenfunctions by nonrelativistic Brave functions, in &c-
cordarrce with the normalization (3-15.33), we have

This coincides with the correspondixlg spin 0 limit, (3-15'68). A similar con-
sideration, relai.Led by the photon crossing transformation, applies to the tab-
ssrptian process,
We shdl discuss orlily in the context of photon scatbring. Using the
spin 0 structure (3-15,6t), we insert the causal field decomposition

and isolate the terms of interest:

The general transition matrix element is

where the dyadic V has the components

i112~12&+(~,
xl, - kg)p;e-"l "'j~,o~~,~~(x'f
(3-15.88)

The A+ symbols are the transform Green's function A+(x, X', p'), with 'p
assig~ledone of the values
p"+k~=p"'+k~, po'--k$=p'"-k~. (3-15.89)
Far simplicity we consider only the namrelativistic limit, where the photon
momenta are neglected, as are the terms in &+(X, X', )'p having denominators
p'f $ p' N 2m, in contrast with the denominrttors''p -'p = E' - E. This
gives (using a slightly simplified notation)
3W Fields Ghsp. 3

If we restrict our attention to elastic scattering (E' = E", ky = k i = ko), the

msult is la digerential cross section far deflection of the phobn into the solid
angle dSZ :

from .inrhich less spwific erass sections are obtained by summing over aband
8veraGng over a''? by summing over XI and avr?r&ng over X2, and by i n b u ~ t i n g
over all solid angles,
At ghobn enerrgies %hatare Xsrp in comparison with H-pa~iclt,binding
~ , the 1mt tern of (3-15.91) survive^ and, with a' = a", one recog-
e n e r ~ e only
nizehs the Thornmn cross section, which describe8 the scaftr;et;riagof (on a, rek-
L i ~ t i eseale) low enerw photons by a free particle of charge rf=eand mass m,
Another limiting conneckion with Thom~onscattering should appear a t very
low frequencie~,smsll in eompaeson with the mergy inkrvals beI;wwadigerenf
H-grarticlea. m e n a, photon of essentially zero frequency is sca,&&redelmtically
by a, part4eular B-padicle, the dynarnid connectians \;vith other W-pasticleg
are not in evidmee, and the seatlering shoutd be descr;ibetd: by the Thornson
fornuLa appropriak to the H-par$icXc?icharge and masts. Sinee the latter hafj
been ideafi~edas infinite, the elsstic scattering erass ection should vani~has
k@ -+ 0. This implies a set of relations, known as sum rules, which can be
va~ouslygreesented. The immediate form implied by (3-15.91) is

(3- f 5.92)
Here is another:

and yef a third, intermediate f m , obtained by repfacing only one of Lhe

momentum. matrix elemen* fetctorar by the corresponding coordinate matrix
elemertt,
(Bfa'lpklEa)-- -..-*E: -- E")(E"c4'(zkfEa),
(EalpkfEkf')== &(E - Ef)(Ealza:
is

(3-15.95)
The 1st version shows the mathematical origin of the Bum rules; they are
matrix elements of the cammutation relation

The elementary ori@n of the sum rules does not detract fiorn their gig-
nificanee its conditions of consistency for the phenomenological particle des~rip-
%ionof eomposik systems. That is emphasized by removing the idealization
of infinite mass to obtain the necessary result involving the charge (Z" - I)@
and mass M of the H-particle, viewed as a, composite of the two particles with
charge and mass assignment8 $ivm by -e, m (electron) and Ze, &g - m (nu-
cleus). Tfre gcattering amplitude fhat appears in (S15.91) describes the pm-
e e s ~ in
e ~which the electron absorbs the incident photon and emits the seatbred
photcllm. To this will now be added the reprwentatian of the processes in which
the nucleus alone perEorms these acts, and of those in which both par-tieles are
involved. Altfiough we have not developed the relevant general famalism, the
necessary modifications here are quih clear. The ma;trix product terms of
(3-15.91) describe two successive interactions with the electric current, to
which both particles now make conlributians:

where

is the rdative momentum in the center of m s s frame. In addition, there is a

eoxltributian in which the scattering takes place in, one wt that is associahd
with an individual particle; it is extended by

I n carrying out the reduction of Ghe matrix product, the relation betuveen
relative momentum and relative velocity is now given by the reduced mass,
m(M -- m ) / J f . Removing the factor of e2, we find that what replaces the
amplitude of (3-15.91) for s, realistic H-particle as ka --t O is, apart from the
gola~zittionvector product,

Tbis is just what is demanded by the phenomenological H-particle description.

Let us idroduee these realistic modifications in (3-15.91), while retaining
sn arbitrary value for k'. The sum mle (3-15.92), together with another sum
rule fhat expresses the null value of the earnmutator Exk, Q], ~ a be
n used to
rewrite the er088 ~ e ~ f i oassnl

Here, -ed is the internal eleetric dipole moment of the system in which df
position ve~torsrefer do the cenhr of mass veetor

which relaks d to the relative position vector

In this version, the fsw frequency behavior characteristic of the H-particle is

explicit, while the disclosure of the constituents a t high frequencies is assured
by the sum ruEe.es, which here produce the earnbination

If is the amplitudes for individual scaLLhring by the two particles that are
add&, and not their cross sections, since this simplified treatment neglects the
photon, monnenbrn and thereby amumes that the photon wavelenglh is large
eompard t;o the particle separation. That restriction, is easiXy removed by
in~ertingthe relative phase factore and, with increasing frequency, the coherence
between the two scattering amplitudes disappeam,
It is also possible to derive (3-15.101) directly, by using s digerent gauge
whieh is specifiedty adapted to the long wavelength regime.. If the electric field
of the photons is homogeneous over the inkrior of the H-pahielc! and procems
involving the magnetic field are negligible, s suitable choice of potentiwfs is
A'(X,Z@)= -x*E(R,zO), A(x,zO)=O, (3-15.2M)
where
E(R, zO)= ( d ~ "2ik'[ekh
i) exp(& R -ik'z')i~~~~
kX
- i~:t&e:&exp(-ik * R + ikOzOf]. (%I 5. 107)
The seafar wkndial ~ouplesto the c h 8 ~ density,
~e

Transitionis btwwn diBerenL H-padi~1-warf3: e x c i w by the inhm&i d i p ~ l ~

moment derm, ernd this contdbution Lo photon scattering repraduees the sum-
analion k r m of (3-15.101). The exkmaE digole nnomen6 (%5 - J)aR affects
only the motioa sf the given H-pa&icle. The scatkring amplikde $hat it pra-
is gi-ven by the diagonal matrix element of the a p r a h r

The reference to the rest frame, the state of zero momentum, redurn8 %histO

which eompleks the derivation of (3-15.1011) since there is no joint effect of

the Bihrent kinds af dipole mannenb,
The analogue of (3-15.86) for a spin 3 part;icfe is

and the dyadie thaL replaces (3-15.88) h= %hecomponents

(dx) (dx')#,o~.~(x)*~'[r~e-''~ "G+(x, X',''p +k : ) ~
te"2*''

Again, we only consider the nonrelativistic limit. But this time the k r m in
the GreenP8funetion with denominabtors +
p@ = 2m cannot be neglected.
We shrall need the explicit sbtemenl of completeness for the eigenfunetiam. It
em be: inferred by eamparilrg Lhe high enerp;y limit of the Gseen's funetion
@+(X, X', P O ) ,
Lim (-rap'~+(x,X', = 6(x - X'), (3-15.113)
pO--toc
3 fietds Chap. 3

with the constructian (3-15.35), which gives

We exploit this relation by writing the Green's function as

D+(x, X', p') = +,o~,~,~ (x)$pot,r.pfn')*ro
,ka"

and then introducing the nonrelativistic simplification''p + 'p z 2m. The

correction to (pof - p')-' is negligible. But the last term of (3-15.115) becomes
-(2m)-'rO &(X - X'), and this supplies the following addition to 2Vkl:

it is the Thomson term. Introducing the nonrelativistic equivalence of Y@T

.t;o plm, the complete structure of the spin, O rmufd [Eq. (3-15.W)] is realjlfi~id,
as one would expect. The situation is simpler when the gauge (3-15.106) is used.
No matrices appear and the nonreIativistie reduetion of -t;he eigenfunctions to
wwe funciC,isns can be performed directly, with the justifiable neglect; of the
1/2m term in the Green" function, The immediate result is (3-13.101) (withoud
the 1/34 term, of course, since we have been using the souree description of
the charge Ze).
We notv have before us some simple physical situatioons in which the incorn-
pfeteness of the skeletal description of phoeon interactions bwome~evident.
Spontaneous emission is described as proceeding at a con~tantrak, even though.
the initial H-particle supply woufd be exhzrcuste-d after a su&eient lsp,se of time.
Under conditions of exact 'resonance,' k' +
Et = E, the photon scattering
cross section is predicted to be; infinite, whieh is ailways unacceptable m the
angwer to a physical question. In the next section we shall identify the sign
nifieant phenomena thaL are omitted in the skefetal de~eription,and remedy
these difficulties.

3-16 INSTABILITY AND MULTIPARTICLE EXCHAFUBE

Although the need to describe unstable gerticfes as naturally m stable ones
i s one of the motivatians in devising the theory of sourcw, %heH-pmticle pro-
3-1 6 Instability and multiparticle exchange 361

vides our first encounter with unsthle particles. The distinction between.
stable and unstable particlm is a matter of time sede. Within suitably restrickd
time intervab, the mechanism producing particle instability is ineffective and
the stable partiele description is applicable, provided, of course, that enough
time is still available for the accurate dekrmination of the characteristic particle
properties. Ot;herwige, no single-particle description is mestningfut. The H-
pa&icles supply examples of sLPLble and unstable parlicles, The particle of
minimum enerw is st&le. Those of greater energy are capable of emitting one
or more photons, thereby transforming themselves eventually into the tllbsolukly
stable variety. The initial description of H-paPficles wunned their stability,
and i s applicab-le over a rerstricM time scale, The descrip%ionis false for very
long time inhrvals because it mserts that weak H-particle sources emit and
absorb single H-particles that propagate unaltered between these acts, But,
given enough time, an unstable H-particle will transform itself into another
H-particle and a photon. These two particles are also c%pableof recombining
to form a single H-particle, Thus, a description of the coupFing b&ween weak,
causally arranged H-particle sources that does not refer to the real exisknee af
two or more particles propagating between them is physically incompbte. I t is
the inclusion of such multipart;icXe excbantgeis between sources and the con-
sideration of some of the physical consctquenees that wit1 occupy us in this
section.
The first task is the identification of effective sources for the emission and
the absorption of an H-particle and a photon, This is analogous to the discussion
of Section 3-1 1. The description of s noninteracting photon and H-padicle ia
given by (using the apin 0 example)

Comparison with the vacuum amplitude term describing single photon emission,
as contained in (3-15.W), gives
2== 8 ( -
~ (3-16.2)
eff.

and the same form applies, with appropriate causal labels, Co the a b s o ~ t i o nof $I
photon and an E-particle. Since this eRective photon souree is meant Lo bg/
multiplied by a vector potential in the radiation gauge, its appearance is simpli-
fied in comparison with the structure of (3-1 1.15). On replacing J";Zt ) K z( X )
and J2r(Et)K2(~" iin (S16.1) by these egective combinations, we obtain a
desc~ptionof the causal coupling between H-particle ssourcw that is merfiabd
by the exchange of an H-particle and a photon, under physical conditiom of
noninteraction. But, to be consistent with the use of the radiation gauge, we
must first ehange the ten~orthat eouptes the vechr photon soumes, in relation
ta the exchange of a gart;icular photon,

The polarization veetor summation is a spatial dyrtdic,

which extracts the tramverse parts of the mulfiplying currents. Thus, the
coupling krnn in the vacuum amplitude is

An understanding of the causal situation is r e q u i d before the necessary

eontml can be exerci~ed. As we see in. Eqs. (3-16.2) and (%16.5), the distribu-
tion of tbe two-partick emission and abgorption sourees is charscterized by the
&lds +a(%') and (iil(z), Suppose the E-partide emission mume supplies the
correct enerw to create a partide that is capable of spontaneous tmnsition to
otlker H-pa&icles of lesser energy, with aeeoqanying photon emission. Thier
process occurs a t a s k d y rate throughout the subsrfquent history of the particle;
it has no csffeetive locdiz;ation in time. Therefore, in order to exert a causal,
tempore1 contrd over the aet of two-particle emission and the subfequent
absorption, we must use the H-particle sources in the exbnded sense. Then the
Bources emit and absorb virtual H-particfw, which cannot exist far from their
sources and, consequently, are transmut?ed into or are produced by a real photon
and a real H-particle near these sources. This is what gives us the &iliity to
influence vvhere (when) the acts take plaee. The quantitlative equivalent of
these remarks is contained in the exprmsion for the field,

which has no propagation chafaeteristics if the sources K,op,t.t(Po) vanish for

PO = As indicated in this formula, we shell use the symbol P' to denote
fhe e n e r a that is injected by an extended H-partiek? soume, converted i n b a
real EX-particle and tt, real photan, ~ n dfinally absorbed by an =.beadd 3%
particle dehction csclurce.
Through the use of exterrded sourceg, %hen,we emure th& the fields
and have supports Lhat are eausa2ly relaLE?c;I. This pernits us to we the causal
3-1 6 fnstebility and multiparticle exchange 363

forms of the propagation funetions in (3-16.5),

where, as indicated, only positive values of''P appear. The resulting form of the
vacuum amplitude coupling term is

with

~vhichare the elements af a positive Hermitian matrix,

The additional coupling between K-particle sources csn be expressed ss a
modification of the propagation funetion ~ , o . ( z @ - E''), ~rhichnow beeomes
%I matrix, def ned generally by

where
5,,,,., pa#.u..(~O -)'2 = &-porraFr, - z').
-pofar(~o' (3-16.11)
The emission and absorption sources of the vacuum amplitude term (3-16.8)
occur in the combination

We recognize in the central factor the prop~ga4ionfunction - zO'),

evaluated under the causal restriction z0 > zO'. The space-time extrapolation
W Fields Chap. 3

of this structure is performed under the guidance of the symmetry property

(3-16.11). I t suffices to define r,otap,
po..a.t (P') for negative values of P':
r-pae,anre-pa~a~(-PO)
rpa~a~,pa~~a~~(PO)p
(3-16.13)
and the symmetry propedy is then satisfied by

The effective limitations on P' must also be removed if this propagation function
is to be meaningful for arbitrary sources. The P' integral can be defined to
simulate the initial consideration of extended sources, by excludi~ljlgneighbor-
hoods of the values p'' and p"'. If this is done symmetrically about these
values and then the limit of arbitrarily small excluded in&xlr8!% considered,
with a speeial provision for''p = p'",

the result is to use the principal value of the singular P' integral. I n contrast to
other reeips $hat m~igncompbx values fo sinwlar intepala, this proeduro
h- the mtisfactory feature of preserving $he essential wociation of eonnplex
numhrs with the propagation function d p e ( ~ ' - z").
For S more expli~ittmt of these extrapolations, we examine how &hesimple
propagation function is modified, by choosing p'ra' = p""af', and considering
x0 > zol,g?@" 3:

The physjcally inkresting regime be&= afker a time Iapm of many periods,
p"(z' -- zO')>> 1. Then the integral is dominated by the immediate neighbor-
hood of the singularity a t PO = and one can introduce a simplificstion by
replacing r (Pa)with

The principal value i n t e p ~is

l egectively computed as
 3-1 6 Instability and rnultipctrtl~ll~~
exehangs 366

The result is

which intrduces an amplitude that diminishes in time, without alhring the

time varying phase. This is in acmrd with the phenoxnenolo8cal viewpoint of
source theov. The H-particle energies that have been t?~~eur&tely. identifid
over the finite time intervals r,a...(zo - xO') << 1 do not ehange their values
when the time scale is enfarged. We are not eoncernd here with e x a m i ~ n g
how the theoretical understanding of the energy spectrum changes EH we move
to another level of dynamical description. For our pre~en1purposes the numhrs
''p are given, whether by theory or by experiment is immaterial.
The unit value of the absolute square of exp[-ipO'(zO - X@')] represents
the cedainty with wlnieh stable particle will be found in the same e n e r o state
afkr any lapse of ttime, The square of the amplitude fmtor in (sI6,fZ.O)
scribtls the changing probability &at an undable EX-parLicle (r,ol,f > 0) shall
still exid after the time interval x' - x" -. f,

There is an initial deercse, at a r a k dven by r,o,ai.But this result become8

unsatisfacbv at larger time values. The persistence probability of the H-
padicle, according tx, (3-16.21), rewhes Eero at s finite time; it then inereme8
and eventually becomes larger than unit;y. The probability formulas is evidently
linnihd in. physieal applicability to smdl values of r,ap,lt.
Wlka~Lis still missing in the physieal %count is this: We b g a n with an
ex6ended H-part;icfe source emitting a virtual H-particle that quicHy trane
formed into ta real E-particle and ab photon. This situation endured until both
p8rtiafes reached the neighborhood of %heexknded detection. souretr;where they
recombined to f o m a virtual E-particle th& is absorbed, But, @ven enough
time, the recombination do form a vir-lual H-pargide can occur far from detaction
sources with this excitation rapidly deesmgosing back into real, part;ieles, The
cycle can h repeated mfl~nytimes before the virtual H-pafiicle is findly absorbed
by the debction. source, Qthewiw expressed, the fields appearing in the coupling
term (S16.5) originate, nod only directly in the ssurcw, but also indirectly
through other, efleetive sources which sre assoeiakd with the virtual H-particle8
that form far from the sources Ghrough the propagation of resl particles,
The qualitative description. in the last sentence is @ven a quanlitaiive
meaning by the following integral equation for the field &,...(zO) :

where %hem t r i x funadion TX describes the mechanism whereby, for the tmf
dime, zt virtuaf H-pztdiele goes through the cycle of transforming into a real
%M Fields Chap, 3

R-particle and phulon, then back into a (no6 necessttrily the same) vi&uf
H-particle that is detect& by the pro& source used to define the field. The
exciting fidd that appear8 in fhe inhgral expression surxznrg~zesthe @fleetof
%heinitial source excitation and of the u n l i ~ t e drepetiLions of Lhe~erevemible
conversions and is, therefore, considering all pDfa'together, the very field that
is being construeled. This point of view is similar to a xnuXLipfe setzt%!ring
analysis in terms of the last eoXXision. Xf this integral equation were t a be solvd
By iterafion, we would indeed be conside~ngmcceg~ivelymore; elabora;ts repeti-
fions of the same bmis grocesa, The compa~sonwith fhe h a m descriptian aE
ane such action then identifies the matrix n. Tkis coqarison is ffl.cilitaM
by wfiting.

w h m the modified pmp6t.gation funetion obeys the inbgral equettion

X &p:az ,,a ..a. (20,-- zO'). (8-16.24)

The identification of (3-16.14) with %hefimt two t e r m of %heihrative solution
of (3- 16.%), @ves, using transfom propagation functions,

The corfegpanding farm of the intttgral equation (3-18.24) can be presented as

Al%bou& these are ra%hergeneral equ&ions we shall prsduee only an ap-

psoxinrrak solution that is aipplieabfe ordinary circumstances, as indicactrtd
by the specialization,
rp~tofiapo..~t(~') 6.~.~rpa~(Po). (3- 16.27)
Such staftjmenfs exprw the. rotational i n v ~ ~ a o cofe isolackd systerna, when
the a' are identified ss angular momentum quantum numbers. Only equal
mer@e~are consider& in (3-16.27) since attention is abo resdll.ic-(lddo the
d o d n a n t elements of the propagation matrix:
3-1 6 Instability and multipartick exchange 367

The resulting simplified equation is

which is consistent with the symmetry

-
This symmetry is maintained when c(p0')e(P0)is replaced by unity, as is justified
by the predominance of the contributions for PO p''. The inference that the
integral is only of interest for p0 p' would seem to be contradicted by the
factor (p0' - which vanishes strongly under just these circumstances.
To see which tendency prevails, we approximate l',o0 (Po) by

and consider the integral

which uses a complex equivalent of the principal value of integrals, according

to [Eq.(2-1-62)]

The integral is evaluated by closing the contour a t infinity in either half-plane,

as is convenient, with the result

The imaginary term in (3-16.34) is more directly inferred by writing (3-16.32) ss

Thus, the structure that appears in (3-16.29) is

368 Firstds Chap. 3

the faetor - doe8 illdeed s u p p u s the reill part of the integrsl, but
not its ima@naw part.
i f ?,et Z 0, the finite imagina~yterm maintains the sign of the infinitesimal
imaginary quantity, -ipO'c, and the latter is superfluous in the resulting
approximate equation:

The implied time behavior is

eoxlsistent ~viththe symnnet~y

The timedependent connglex phase factor continues to identify the enerw

p@' > 0, but the variable smplitude 1 - fr,o.l, 1 = z0 -- zO' > 0, haa been
replaced by exp(--- +"u,o#t). This is quite satishtory sinw the implid prabeEbility,
exp( -?,at t ) , never exceeds unity and decreases monotonicslly to zero with
inorewing time. The expnenLiaX, function generalizes the linear deerewe of
probability over short time intern&, extending it from the initial instant to
arbitrary later times, according h

where r At << 1. Xonrmlativistic .~lpproximationscan be introduced in (3-1 6.92,

and tvc arrive at a simptt?explicit formufa for r,ol [earnpare (3-15.6ti)j,

This statement is xlat restriekd ta the nonrelativistic limit, of courrj-e, for it

c?quates the initial rate a-t kvbieh the probability of persistence deereases to the
cometsponding r a k a t ~vhichtransitions are made ta H-padicles of Iobver enerm.
It is inkresting ta verify ChsG this bdanee of probability persi~ts%tarbitr&ry
timtits. We first consider the most elementary siluation, an un~tatsleH-pa&icle,
designated X I , trl-zich e m only radiate down La the stable H-particle of lotvest
energy, 1,zbele-d1, indicated by
3-1 6 tnstsbitity and multfparZlelaaxehanga M9

The probability t l ~ a tH-particle 11 still exists at at time t after its creation is

exp(-rnl) < 1. Is t h h 10s~of prabfLbili@ cornpensaw by fhe probabilify %h&
H-paAic1e I exists, accompanied by ra, photon? 9"s evduah the latker prob8bility
we must extend the formula for Wzl, Eq. (3-Is.@), .tto the new esilu&ian of
urntable particles, From the \riekvp.point of H-partcicles thirs formula de~erib8
t ~ v astages of emissions and absorptions. After the creation of the initial E-
particle it propagates until the moment that the phobn is e m i t t d when iG ee
to exist. At that imtant the final H-padicle is created and is eventudfy de&l;ed,
Evidently, we must now US@ the modified propega%ionfunction to ~ p r w n t
these voyage8 bet1t;een emission and absavtiom acts. Sourm are i n t m d u d do
dmeribre what is common to aIE emimion and abmvtion mwha~smrsof a gw-
ticular type, and the corresponding propagation function is of general applieabil-
iky, f ndividud realistic mechanisms will a180 have specific features fha%n d
additional characterization. This we s b l l d i ~ e u later.
a In the present essentially
no~lrelativisticsituation, Itowever, the description of the mmhani~mfor H-
particle transmutation. and photon emimisn requires no gisignifielznt eomwtion
and the introduction of modified propagation. funetions is quite su&cient for our
purpose@-= we shall we.
What has just been said is that (%fS.fiO) eontinucjtss apply with the
changed meaning of field given by (3-16.23) and the simplificaticion (3-16.28).
In the situation being considered, the; camally I%beledH-parficle fidd +2(z)
that appears in
(3-1 6.43)

is given by [we use the nonrelativistic approximation (3-16.67) but retain the
relativistic origin of energy]

The time t2 is R fiducial poillt \vithin the source K 2II(z"), and correspondingly
tlxe definition of H-particle emission source appeam as

The use! of a reference point that is interior to the source rather than arbitrarilJ;
&oen is always posgible and can be useful in identifying the xnwhanical propep
ties of states. It hcorrseg mnndataw in demribing unstable pa&icles.
376 Fields Ctrrrp, 3

The coupti~rgof emissioxl and afssorptiorr sources for tz ga~rticularunstable

H-particlc is given by

tvfrere, illustrated by particle 11,

and t l is a. reference time that locates the source K l ( z O ) . The fa~toriastionof

(3-X(i.46) clearly separabs the three st;ages of emission, propagation for the
interval t = t x - t2, and absorption. The possibility sf describing unstable
particles as stable over shod time intervals must certainly apply to $he large&%
values assumed by z0 - t I and t z - zO', which are limited to displacement8
within the sourem, Thus the decay factors in the H-particle sour= definitions
(3-26.45) and (3-16.47) can be omitted, and these sources play the same role
as with stable particles. In l;his way, %hen,the weakening af the coupling
(s16.46) tvith incremirrg time int;ervaE between the sources, owing to the
instability of the particle, is specifically associated with the process of prop%-
gation only.
The same kind of description is used for particle I, even thougl~?I = 0,
and the probability amplitude deduced from (3-16.43) is

The initial and fixlai times are now explicit in the specification of stateg, although
only t ==: t - t is significant?,as we have emphasized by using E l as the reference
time for the photon fielid. The time integration is evaluated as

tltld the traxtsition probabilitjr, summed aver photon polarization8 and emission
directions, but still differential in the photon energy, becomes
k0 dko 1 -- 2e-"'2""L cos (kO -- k f I r ) t + em'"' , (3-16,50)
711 - -
%2 (kO - h.! l d Z f ( * 7 1 1 ) ~
3-1 6 lnstobillty and multlpartlcta sx~hange 371

where
k f 11 = E11 - El
and, of eoursc?,
~ K=I $akf rrl (IlplmlII)I 2.
The total probability is produced by carrying out the k' irltegration. That is
approximated under the assumption of \~\.eakinstability, r 1 1 << k f 11, by re-
placing k v l f 11 w~ithunity and evaluating the integrals as

together with the speeializwtion to t -= 0. In this j v v \ye obtailt tllc probnbifit~

of finding H-particle I afkr time t as

5vhicE-i is the required value. The spectral distributiox~of ttre emitkd photon is
also exhibited on evaluating (3-16.50) at a time rrlt >> 1, such that the radiativ~
transition has certainly occurred, The result,

is the familiar Lorentzisn shape th%tidentifies the decay constant 711,the recip-
rocd of the mean lifetime, tvith. the tvidth of the spectral line at half-maimurn.
This is the shape of a spectral line emitted in a transition to the stable
W-particle. But what if the final H-particle is also unstable? Now eonsider n
third H-psrtiele 111, ~vhiehcan only decay into X I , with the subsequent trans-
mutation of the latter into the stable variety 1. In this situation two photons
are emitted and we must use W z zto describe the process. There are tutoanalo-
gous terms in the relevant probability amplitude which are related by the
B. E. symmetry of the phobns. But, apart from the special circumstance
kf"lr 2rl h:& 1x1, only one of them terms is appreciable depending upon tvhich of
the photons has its frequency near kp 11 while the other frequency is close to
kfI 111. Thus, it suffices to regard the photons as distinguishable through fheir
frequerlcies and use only one of these terms. The probability amplitude for the
rvhole process is
tf"l1ere the time propagation functions detail the sueeessivc enusul nets of tllc
drama, In writing this expression, we have proeeedd ss thaugh the H-padicles
of types II and III were unique, although additional indices a11, a111are neces-
saw. T h a e detdls can be inse*d and do not d e e t the resulb, under the
physical circumstances indicated in (3-16.27). The z0 time integretioll is thc
one already performed, with zO' supplying the latver limit instead of tn. The
Lime inkgrd faetor of (S16.26) is, therefore,

The next integration poses no dificultiies, but we shall be confed to ev81uak

it only for such large 1, r l ~ >t> 1 , r111t >> 1 , that it represents the completed
process of cascading decay. Only the exponential containing Er contributes, in
contrast ~rittrtthe one containing Ell, to $ive the value

The implied transition prsbability Chat refcrs only Lo the spectrail distribution
of the photons is

The successive emissions are not independent. I t is the energy k' +

El of
the p h o h snd pa&icie i n b which I1 dec%ys,rather than the energy E ~ Ithat ,
determines the spectral distribution of kO'. When one integrates over kO', the
result is just (3-16,55), h i e h means that W-partide 11 is certainly produced
at wme time by decay from Iff, after whi& the previous disemion applies.
The sns\\*erto the question concerning the spe~traldistribution of the ptrobn
radiakd in a Cransitian bett-veen unstable H-partidea is obtained by. intiegrating
over k'. It is instructive to write this integral as (E = ka f E I )

which describes an energy-conmwing radiative %r&nsitian.b t w e n two enerw

p have the widths 711 and rgrr, respectively*
didcibutions of Xlorent~lian~ h a that
Aocarding to an elemenl%vcontour integ~alwaluatioa, Lhe ~ s u l t i n gspectral
3-1 6 Instability and multiparticIsexchsnge 373

wiLh a width given by the sum af the individual H-parLicle widths. This conclu-
sion is parfieufarly transparent if one recognise8 that the double enerw iatepal
of (3-16.W) is equivslenl to s single time irtkgral:

It ~-outdbe hard aot to suspect the existence of another approach that is capable
of producing this formula directly. We shdl find it, not surprisindy, in the
time cycle description.
But, first, 1eL us give an. analogous discussion of photon scattering, in order
to verify that the unphysicat infinite ttross section a t exact resonance has been.
removed by the explicit recognition of H-parLicle instability, Elmtic wscattering
by the stable E-pa&icle L will, be comidered, Then it ~u&ee1;3tro introduce
modified E-particlis propagation functions in (3-15.88), which will be used only
in the aonrelativistie limit and in the gauge of (3-15.lM). The sigrmifiesnf
change is the ineroduction in (3-15.101) (\v@ ignore the l / n l term) of the
substitution

while E -- (E1 - k') remains unaltered. To understand this it is necessary to

be somewhat more general than (3-16.37), where r,ot (P') is considered only for
PO = We return to (3-16.29) and proceed as in (3-16.351, but with
rPo.(P') retained, and get

near resonance, p' -- p'', or EX k' -

showing the general form of the imaginary term. This distinction is unnecessary
+ E, but it is needed far from resonance
conditions. Otherwise we should have, incorrectly, added an im~tginarykrxn
to E -- (E1 -- k'), where%?
rE(E1 - k') 0, (3- 16.615)
sinee no photon emission can occur if the Lots1 energy is less than E x .
H-particle If becomes strongly excited -when
Under the% circumstances the dominant contribution ta the differentid cross
sectian of (3-15.101) is

This differential cross section far specified polarizations is replactea by the b t a l

G ~ S mction
S on summing over final poltzrizations and directions and averaging
aver the initial palwization (and direction). Recalling Ghat

together with the orthsgonality stated in (3-16.27), we find that

whem g11 is the mulLiplieity of parti~leXI, the number of different values

m8urrtd by axr. The fwm of the cross section, a t exaet resonance,

is typicd of any rwonant scatbring process. The bmic resananf crass seetion
is 4.1rg2, h m 4rr/(kf x1)2, which is multiplied by the number of resonant states,
grx, and divided by the multiplicity of the initid particles. Tfist is jusf the
fmtor of 2, referdng to the two phobn pofsri~stions,sinee H-particle I has bmn
amurn& to be unique.
The promise to exhibik another and, more direct derivation of (3-16.62) will
be fulfilled, even to the point of generalizing this formulst so that it, refers to any
pair of urntable H-padicles, which are capable of dwaying in other sequences
khan XI1 + 11 X. Here is the statement of the mare general problem. The
arbidra~ unstable H-particle f I1 is creELCed near time zero. I t can decay to a par-
ticulrzr unstable H-particle I f as we11 W in other ways, and these secondary un-
stable particles continue the eabsedc?until the stable particle I is reach&. Wh&
is the differential probability for finding B photon of frequency ko 11 k f r 111,
wiLhout reference to the ather photons of different frequency &hat are also
emitbd? For s spwified polarization, that probability is expressed by

which msumes 8 time in,terval long exlough to have the probability attain. its
final value. Let us supply two additional hetom, Gtkpthe pmbability amplitude
for detecting the photon kX, and -i& ,complex conjugate. This produces
it8
 tnstebiiity and mulripsrti~leexchange 376

a quantiLy th%tcan be presented as

Apparing here &resuceesaive s t s s s of a time cycle, in which two anafogom

photon sourcw act, ooe on the fortvard time path and the 0 t h on LXle refum
path. Thus, we are now inkrested in fhe time cycle gener~liaationaf Wzz.
T o use a, cornistent nonrelativistic dmeription, one should subtrmt m from
the f rtltquencies in the propagation function (3- 16.38). This reduce8 the positive
frequencies to nonrelativistic energies, for z0 > xof, but converts the negative
frequencies to values 3 -- 2m, for '2 < g@'. The latter produce negligible
contributions to time inhgraXs, and the nonrelativistic version of (3-18.38) is,
accordingly,

When the g a u p (3-15.106,107) is used, the stmcture of z"W22,w~t'(cenin a

simplified matrix notafion, becomes

The transition to the time cycle is made after time t2. Time t1 is now emounhred
on the return path, which is certainly 'later' than tz, and q(tl -- i2) is replac4
by unity. Also, time t is reached 'efLer9irne tr and q(t - t l ) mu& be replaced
by q(tl - t). Since both t and tI refer to the return time path, there is no sign
change in the integral. The proper treatment of the r Lerrns is fixed by %he
physical necessity of maintaining the damping, the weakening of the ~oupling
~ C inereasing
h dime interval, All this gives the subsfitution:

With the H-particle saurew operating in the vicinity of 1 = 0,

3 Fields Chap. 3

and the coefficient of ( - i ~1:)($Kr

~ supplies the required time eyele quantity,
in the f o m
HI-) J~-~s"(+~ = e2 dt dCa exp[2'EIIx~ - pzIrttl

We must still extraet the ooeEcient of snd of --gkk

from E(t2) and
E(tZ),re~pmtively. According to (3-15.1 07), Lhct: fir~kof thme is

and i b coxnplex conjugate, evaluaLed a t 11, applies on the rever8e Gime path.
Since the restriction k0 czs k f i 111 picks out the contribution from the specific
Ef-pafiieXe XI, the de~iredpdability, as it is dedueed from (%16.77), is

With dk0/2?r removed, the factor in front of the double time integral, summed
over polarizatiarrs, is the A-eoeseient for the dmay XXI -+XI, Ia the h k r m k
of a more uniform no%%tion we: shailt now denote it by 711 ux. T o aimplify W
..time irrbgrals we introduce new v t t ~ a b l e ~ :

which mnges from -m t~ QD, and t<, the snadier of the two %inn@#
whiah vahw
from O to m, Then, %hetransformations

When XII can only radiab to If, 7x1 1x1 = "Frrr, a ~ we d have ~ e p r d ~ ~ c x f
(3-16.62). Mart3 generally, the probability of emitting any hquency in the
~ is (TI1 I I ~ / Y I I ~ <
neighborhood of k f 111 ) 1, ~ecarding
tO
3-1 5 Instability end multfparti~leexchange 377

and this expresses the competition between the specified transition and all1
othem that 311 can undergo, The sum af these fractions over a11 decay mode8
of If1 is equal to unity.
The time cycle extension of W z z also gives a direct derivation of the rem-
nance scattering emss sction (3-llieCi9), or, rather, ifs generaliszttion in which
I1 beeomes an W-particle that can decay in ways other than down to the gtable
particle 1, and rr becomes the corresponding total. cross swltion. A photon is
incident on X, and eventuizlly one again finds I, aecompzznic;d by one or more
photons. The total probability for these phenomena, \vith s given interaction
time, is

When the initial particXes are introduced by appropriate sources, two of each
kind, this beeomes a Lime cycle vacuum amplitude, deseribd by iW2z. The
result is obtained from (3-16.77) by replacing XI1 with the stable I, and using
the field of an incoming photon insted of (&X6,78) :

The integrand depends only upon the time variabb 1 =. l z -- t l , and the integra-
tion over t< is identified with the duration of the inter~etion. The total erom
section is found by dividing the photon flux 2k0 dwr into the transition proba-
bility per unit; time, On recognizing thaf
e2(k!11)' /(IIajx * e 1)i2 = T ~ I ~ YXI,
I (3-16*88)
a

we get the dominant contribution to the cross section, for k' kf 11, as
,P

When 71 11 == we regain (S16.69). The additional factor, Tr II/rIx C X,

evidently represents the diminished ability to excitc! I1 directly from I, w h i ~ h
is the reciprocal aspect of the fractional probabiIiLy for reaching I directly from
fE in decay. From this point of view, the elmtic scattering cross section should
be obtained from the total cross section by multipfyirrg the lathr tYith an addi-
tional rx II/rrrfachr. That is indeed the result dducrsd from (2-16.67) whm
the quantity mntaind in (3-16.68) is $;iven. its pneral interpretation W the
partial width r 11 :

The above discussion is ixreompfek since no mention has been made of the
ddi.tionsl t e r n in IFzz that is dewnded by the cmsaing symmetry of tbe pho-
tons. It is produced by reversing the sign of k'. This tRrm is certainly non-
resonant. But, more important is the appeesance of the initial energy as EI - k a ;
the value that should be a~signedto the damping constant of H-pa&icle I1 is
not rI1 but %em, as in (3-16.65). Then the resuXting dime inlegrd gives
+
6(kf 1- k 4 0.
All the developmends of this aeetiotoxl have used the exampie of spinless
particle8 that are bound to form El[-particlw. A similar treatment for spin +
p a ~ i c l e swould run in exact paraIXet, with occasisn~iinserLians ar deletion8 of
c(p') factors, for example, to represent the changed statistics. The nonrelativ-
istic results are identical.
The natural instability of H-particles has direckd attention to the necessity
of considering multigartiele exchanges, in addition to single-particlepropaetion.
It is a complementary aspect of $he principle of spaee-time uiformity that
couplings identified through the examination, of red proeesms eontinue to be
meaningful when. appfid Lo virGual processes, This says thad multiptzrtiele
exchanges are significant, although the energy CO produce sevemX red par-tietes
m6y not be available. Thw, &henext stage of d y x l a ~ c a levolution. is %be
~ystematicgeneraliz;ation of all single-partick exchanges b e t w ~ nsources to
those involving two particllcls, including their unlimited repetition. Before em-
barking on. this masive progfam, however, wre shall give a relatively b ~ e dis- f
cug~ionof the ~avitationalversion of ~ u e hconeepk as primitive interactions
and gauge invariance.

3-17 THE GRAVITATIONAL FIELD

The field wsociahd ~ t massless
h particles of ihelicity &2 has not yet been given
an independent discwsion, We refer back to (2-4.241, but use the mechanical
mea8uxl.e of Tp,,according to (2-4.33) :

The symmetrical tensor field h,,(%)is defined by

3-$7 Tha gravitationat field 379

subject to the source restriction

a, 6 ~ ~ ~ (=x0.)
The corresponding arbitrariness in the identifieatiion of h,,(z) is exhibited in

Contraction of indices in the tensor gives

(3-1 7.5)
and therefore

The introduction of the source restriction, through the divergence of this equa-
tion isslate?a the a~pectof the field h,,(z) that is governed by the arbitrary
(z) vector,
a,(hpv(x) - ~ ~ " h ( z= ) )a2ty(z). (3- 17.7)
Returning to (3-1 7,4), we deduce

which is the second-order difirential field equation,

It is also the form of the equation obtained by placing m = Q in Eq. (3-3.19).

GaxrLractixlg the indices in (3-t7.9), or converting (3-17.5) to a diEerentiaX
equation, implies
-a2h(z) -+ ~ )-*KT(z),
a p a v h p Y (= (3-X7.9a)
from which we derive another version of the differential field equations,

it is (3-3,17), with m = 0. The structure of the left-band side of t h i ~

equation.
is such thst its divergence is identically zero. The vanishing divergence of the
source tensor now appeam a8 an algebraic consequence of the field ~ U ~ Z ~ ~ Q X ~ F
380 Flalds Chap, 3

Since the arbitrariness of the vector E,(z) is still maintained in these field equa-
tions, they are unaffected by a redefinition of the field h,,(%) having the form

which is a gravidationd gauge transformation.

The following definition, analopus to (3-3.23),

with its eonsequenee

F,($) = r,kh(x) = a,h(~),
pravides a fimt-ar8er form of the field eqiuations [(3--3.2-2), wiLh m = Oj

The gauge i n v ~ i a n c eof the Iefbhsnd side of (3-17.14) is not realized through
the invariance of r,,~,but rather

and

Note, however, that thege gwgr?, transformation rmpsnses do not invdve first
derivatives of Lfre Ex fz).
Another ~ y a k mof first-order differential equations fit is (3-3.20, 211, with
m = Oj is praducd by the definitions
@~XE(%) ~ ) aph~W(z)- dlh~@(je)
= - w V X ~ (= (3- 17.17)
and
= aph(z) - dhhyh(s),
oy(z) = wPhh(~) (3-17.18)
name1y,
ah@,.k(z) - a.w,(z) = K(T~.(S)- + ~ , , T ( Z ) ) . (3-17.19)
The reBponse of thme fields to gabuge transformations i s gven by

and
@ph~(~)@~xP(z)ak(ap~r(~) +
- ~P&(x)) (3-17.2131)

@p (4 + @p (z)4- a p a k t k ( ~
- )d2& ( X ) . (3-17.21)
It is observed that Lbe divergence af the vector field m,($) is gauge invariant, A
comparison of the form of the divergenee inferred from (3-17.18) with (3-17.9a)
shows that
d,wg(x) = +KT(%), (3-17.22)
which is dso the eantraction of (3-17.19), sinee
3-1 7 The gravitational field 389

An arlalwus w e of (3-17.14), however, introduces ta. new vector field,

&I"(z)= I"",x(x)= 2dVhv~(~)axh(x), (3-17.24)
such that
rx(z) - XT'(X)
= ~wx(z). (3-17.25)
This is s contraction of the tensor relations

Another connwtion between the two third-rank tensors, which implies this one, is

The sueeessive stages involved in producing an action expression Lo represent

the first-order field equations (3-17.12, 24) are

where the last version introduces the relation

+
dhhpr(x)= d ( ~ ~ * ~ ( x~ ) ~ ~ ~ ( z ) ) , (3- 17.21))
i t s contraction
dyhpv(z) E i(rP(z)i- 'r(z)), (3-17.30)
and (3-117.13). The action is
(3-17.31)
with

Apart from a, divergence krm, this Lapange funetion is the analogue of (3-5.411,
with m -. 0. For simplicity, we cfo not include a source for the third-rank
tensor field, in contrast with (3-5.40). The stationary requirement for va~ations
of hp', or h""" ---- +@@"h,recavers (3-17.14)) and variaLions of F,,x, &er re-
arrangement~indieakd by the structure of (3-3.45), reproduce (3-17.12). The
Laf~;r&nge funetion is not gaum invariant,
382 Fields Chap. 3

but the aetion is invariant. If we relinquish the use of pas an hdepndent

variable and Xet it be befind by (3-17.12), an appropriab Lapange function is
~ . e ( h )= f(r"'Yr,.h - Arrh)e ( ~ 1 7 . ~ )
When writfen out as a qudrtttic funetion of the fint defivativm of h,,(z), this
fi-ange funetion differ@from that of Eq. (3-5.99) only in the 1-6 germ:

which itlustrat@ the freedom to add divergence terms, This athmative has
d r e d y been n o M in Eqs. (3-5.31,32).
The similar development that i s bmed on the fimborder diEerentiaX equs-
Gions (3-17.17, 19) stafts with

~vherethe second version u& the substitutions

The Lagange furretian is obtained rzs

Apart from di-vergenee terms, it is (3-5.34) wifh m = O. Again, the muree

coupled to the third-rank tensor will not be u s d . The vafiatiaion of hp" - 3 f " h
reproduces (3-1 7.19) and thlikt of oh,, yields (3--11.17) reanangements that;
are indieat4 by %hestructure of (3-5.38). The response of this Lapange furze-
tian to gauge transformations is

which msures the inva~anceof the tbction. Whm wh,, Io~wi k independenf
sfatus and is defined by (3-17.f7), the Lavange function clan be chomn M

which is the quadratic function of the fir& defivakives of h,,(tt) that is pmdued
by averadng the two alternatives of (3-17.35) :
3-1 7 Ths gravitational field 383

The stress tensor Fy(s) has been given a kinematical definition, m~hichis
not unique, through the response to infinihsimal coordinate deformations,

It aequires a dynamical defiaition by imitating the role of the paviton source

Tpp(2),This is inacated, in the re~ponsc:of an action expression Lo infiniksintal
gauge transformations, by adding t'" to !Py,

The two concqts are identified by requiring invariance of the action under the
unified gauge-coordinate transformation,

Thus, infinitesimal coordinah transformations induce the infinitesimal gauge

transformations
6h,,(z) = i (a, 6zv(x) t a* sz,(z)),
6I",,x(z)= d,3, Qixx(xZ

+
The use of the total stmss tensor !Py P",as the fwtor of h,, in the mtim,
is the introduction of a primitive interaction. Same modification of tB"is needed
since it is not conserved inside pa&iek, aaurces, and a gravitational model of
particle mmes must; be introduced. But let us dekr the biscugsion of that ques-
tion and proceed with the development, whieh is modeled so ctosdy on the
eleclromagnetic one, in order to reach Lbe point of divergence between the two
very different physical sy~tems. Consider %heexample of spinless pa&icles,
using the simplesl stress Lensor form, Eq. (3-7.81,

As in the electromagnetic anetlogue, the eoupling term &,P"will be combined

with the particle Lagrange function, -to form

At this stage in the elwlromagnetic discussion, faeilidahd by the usc? of a sfightly

different Lagrange function, the gauge eovsrisnt derivative a, - ieqA, ap-
pear&, and one vefified invariance under the whole Abelisn uaup of gauge
transformations. The related gravilalional situation is only partly produced
by the suh~litutionkqA, -+ h,,ay; there is also a pavitaLiond eoupling that
does not refer to derivatives, But, much more significant is what underlies the
replacement of the single matrix czq by the four diaerential operator8 (I/.k)d,.
384 Fields Chap. 3

The general coordinate transformation group is non-Abelian, as indicated by

and the extension of invariance under infinitesimal transformations to cover the

whole group is not trivial.
It is instructive to examine just how infinitesimal coordinate transformation
invariance comes about. The transformations associated with

describe the scalar nature of the particle field,

and give the induced gauge transformation of the graviton field,

including
= hpV(x) + *(a, ~ x , ( x )+ a, ~ x , c , ( x ) ) , (3-17.52)

E(T) =h(~) + a, 6 # ( ~ ) . (3-17.53)

The invariance of the mass term in the action is stated by

= /(d5) ( l + ~ ( z ) ) + ( z ) ~ ,(3-17.54)
which is satisfied if
(dz)( l + h(x)) = ( d z ) ( l + E(@). (3-17.55)
For the infinitesimal transformation (3-17.50)) the transformation law of
volume elements becomes
(dz) = (dx) det (aF/azV)
= (dx)(i - a, ~ x ' ( x ) ) , (3-17.56)
and it is required that

This can only mean that h(x) is restricted to be a very small quantity, per-
mitting ha, 6 9 to be neglected as a second-order object.
The situation is similar for the quadratic derivative term of the action.
We first notice that

isolates the factor 1 + h that compensates the transformation behavior of (h).

Then,
g"'(%)= Q"' - 2hpv(x) (3-17.60)
3-1 7 The gravitrrtionslfZeld 388

must transform appropriately to produce a watar combination:

& 6 ; ( ~ ~ 8 " " ( ~ ) ~ vzzz
~ ga,@(.z)8"""(z)av&(g)t
Sz) (s17.61)
or
~ " ( 2=
) s""~)a.z@ahz@. (3.- r 7.62)
This property eharacbrizes g p v ( z ) as s contravarjlant bnsor of the second rank
under general coordinate transfornnationa For infinik~imaltransfomation~,
that transformation.taw becomes

which doe8 reduce to the first statement of (3-17-45), if one neglects seeond-order
quantities by replzccixrg gg"(x) with g@', on the right-hand side,
The ten~org,,(%), in~verseto gpv(z),

bfts the transformation law of a c o v a ~ a ntensor

t of the swond rmk,

The implied behavior of the dekrminant

S(%) zzz det gpv(z) (3-17.66)
is
g(z) = g(z)[det ( a x @ / d ~ ' ) ) ~ , (3-X 7-67)
and therefore
(-@(z)) " 2 ( d ~ =
) ( - g ( ~ ) )l l Z ( d ~ ) . (3- 17.68)
E t is consistent do regard this as the generalization of (&17,55) for, under the
weak field conditions ia whieh the latter refers,

and

Aeeordingty, to enBure invariance of the action under arbitrary coordin&k

transformations, the Lagrange ftrnckion of (3-17,48), which is appropriate to
weak gravitationd fields, should be replaced by

We must find a ximilar generalisation of the weak field fmm of the p a v i h -

tionat Lagrange funetion,

which i s (3-1x32) with dl reference to third-rank tensors s%a%d

in kms of
386 Fields

One rec~gnigesin hp" - ggp"hsrt of the weak field evaluation

The missing constant term can be added in (3-17.72) since it changes the
Lagrange function by a divergence. Then the strong field generalization is
clearly indica;ted:
Z K (g(%),
~ ~ ( 4 ) (--II(x)) li2gpv(2)Rpu(2),
E (3-17.75)
with
Rhv 8hrtv- aPr;h ~ ; ~ -+ r ~k~r:h,
$~ (3-1 7.76)
This will indeed contribute an invariant action if yMvRp.is a scalar with respect
to arbitrary coordinate transformations, The required covsriant knsor behavior
of R,,(z) must emerge from the transformation law of the three-index symbol
) The latter should resemble a third-rank tensor but cannot be entirely
of this nature, according to the ~veilk.field transformation of (3-17.45) which
~ontainssecond derivatives with respwt to coordinates. A suitable generaliza-
tioxr is stated by
r;,(i-);1.~" rig(~);i,~Pa,~g+ a,a,zk. (3-17.77)
This transformation Xatv is such that a coordinate covariant derivative of first-
rank contrsvariant vectors can be defined:
= aa.VP(z)+ r:&(z)vk(z). (3-17.78)
= (a. + r,(z))~v"(z
v,vP(~)
The matfix notation facilitates the consideration, af
[v,, v,lw" =$X V&, (3-17.79)
where
R,,"~ =. a,r:~- aVr:& 3- T:,T";~ - r:,r,Ph (3- 17 .so)
is indeed a fauxrth-rank tensor, which is antisymmet~ealin p and, v. We can now
recognize the tensor character of

The nation of covariant derivative, identified with ordinary differentiation

for sc~lars,is extended to firrsbrank covariant vectors by the requirement

and to arbitrary bnsors by generalizing the difierentiation rule for prodtteb.

As an application, we note that 6~3,(3), which is any infinitesimal change of
the r,",(z),
does transform as tensor, and
&R,, = [a, ar;, + r", - ar",ri. - arigrk.]- [a, art, - 8r",r:,]
= V& 8 ~ -
; ~V@ (3-1 7.84)
The eovariant derivative of g(s) is defined by the deterrninantal diflerentiatian
fomu18,
Vxs(z1 = g(z)gHV(s)Vxs&y(2)
= g(x)fvIz)P~~~~(2> - %x(z)f"xv(z)l (3-17.85)
= axa(x) - zg(z)f""x(x),
01"
vk(-S(~)) H2 = ak( - @ ( X ) ) - (-g(%)) 112riv(~).(3-17.86)
A. simple consequence is the divergence formdla,

These results are used in applying the stationary action principle to varis-
tions of P:, as it appears in g(g, F). The vanishing coefficient of 6Pi, in 6W
states that [gppis gH"(z)j
v,[(-o) 1j2s"l - a i ~ , [ ( - ~ ) =
lizy'Y] 0, (3- 17.88)
which imp1ies
v~[(-~)'~=
~ ~0." ] (3-57.89)
From the latter pmperty one d e ~ v e s ,successively, the vanishing of the co-
variant derivalives for g ( ~ ) g'"(~),
, and g , , ( ~ ) . The last statement,

leads to the explicit construction

mfhieh is the strong field generalization of (3-17-12). The weak field vemion af
(3-f7*90) appears in (3-17.29). As one can verify directly, the vani~hingco-
variant derivative of g ( ~ )implies, according to (3-17,86),%h&

which generalizes (3-1 7.13). This form ensures that Bp,, as defined in (3-17.76),
is rz symmetrical tensor.
The variation of gp"(z)in the pul;.eXy gravitational contribution to the action
induee~

where
4, R,,
X $gPvEp
m R = FvR,v, (3- 1%94)
and we have used the deterninandal property
The Zlensor G,, obeys a differential identityt which is a consequence of the
coardinatti?invariance of the gravitatisnd action term. We firrat note %heinfini-
tesimal response of ( X ) , analogous to (3- 17.63),

which generalizes the weak field e u g e transformation of (3-17.45). On ~ t i n g

where

we eonelude from the invariance of the action Ghat

The variation of ggvfz) in the matter part, of the action, defines a hnsor
t,,(z) that generaIizes the stress tensor,

f nspectiorr of the Lagrange! function (3-17.71) show8 that

We slm note the generalization. of (3-7.91, in source-free regions.

When, the atalionary properky with mspect La 6 variations is invoked, the

coordinate invaeance of this action k r m leads, as before, LE) a differentid
slabemend,
V,t"""(gz) = 0. (3-l"i"lQ3)
Another form of this generalized local conmrvation law is

The field equation dedued by va;ti).ing g"" in the complete action

3-1 7 The gravitational Qield 389

it lis Eimtein" pgra;vitationat field equation The sfress tensor divergence eondi-
tion (3-17.103) appesm again, now as an identity demande-d by the ~tmeture
of the p a ~ t a t i s n a field
l equation.
The replacement of spin O particles as the model of mavitating matbr by
okher inkger spin, parYtielm is relatively straighithmard, A rather special but
interesting example is provided by photons. The Lagranp function

is innmediahly getneraliaed to realize invrariance under arbitrary caordinab

tramsformatiom, while maintaining electromagneticgtauge inva~ance,by writing

Af points not occupied by electrorngra;gneficssurees, the implid field equations are

(3-17.1 IQ)

The stress demor deived by varying gPY(z)in. (3-17.1W) is

where aX1 eontrava~antand covsrianL indiees are reIated by mean8 of the

t~3n~0rgP,(2), Another instructive derivstioa can bet dven, By redefining FP'
to absorb (-g)Ii2, which praduees the Lagrange function

aU reference to g,, is ctoncentrsled in the Imt term. The stress bnsor (3-17.1 12)
i s regained, divided by (-g) to conform -4th the d t e r d meaning af F""". BuL
now we can see something very clearly: (--g)-"2~r.g.h i s homogenwu~of degree
gero in the campanexrts of g,,(zj, which is to gay that the Lapange fmetion
(3-27.113) is i n v a ~ a nunder
t the transformation

for arbitray X(%), The impEiesLlion, for an infiniksimal deviation of X(z) fmm
unity, i~
112 v
(&)(-g) P fi, SX = 0 (3-1 7,115)
or
l(%)= gb,(z)t""(x)= 0, (3-17.116)
which, is tme.. Id is e ~ c l e nthat
t we me now considering a generalization of $he
cadormaf transformE1Ciona that were o~@nallyinfroducd through the can-
3 Fields Chap. 3

sider~tionof isotropic dilakions ((3-7.153)l. Incidentally, while the alternative

Lagrange function. af (3-1 7.113) was helpful in recognizing eonformaf inv~rialrtce,
one can also use (3-17.109), combining the eonfarmal transfomation (3-17.1 M)
with the field transformations

ta attsin the invariance of C.

As ure have noted eariier in, connection with (3-7.168), the kinematical:
arbitrariness in stress tensors must be considered in tes"cing for conformal
invariance. The arbitrariness can be placed in a dynamical, conbxt, akin to
$he electromagnetic procedure illustrated in (3-10.63). Returning to the weak
gr8vitational field situation, we examine the possibility of replacing any given
stress tensor P" by [Eqs. (3-7.83, 84,8511

where mpYsKk
is symmetrical in p and v , in K and X, and obeys

The addition, to the Lagrange function term tP" is, effectively,

mr**"bKa&h,y
= -$mp'e"QR,A (3-17.120)
where
-+
R , , , ~= aJlaxhxv ava,hxp-- a,axb, -- +a,h,~ (3--17.~2~
is produced by using the eyelic property (3-1 "l.X 192,

Also contained here, in

is the requirement that m&**"'be symmetrical in the two pairs of indices, as

already found in, pa~ieutarsi-t;uations[f 3-7.88), (3-7.11 l), (3-7.137)f.
The four index object EpKVk Inas many symmetries. 1%is antisymm&rica;l
in p and K, in v and X, and symmetrical in the two pairs, prc and v&. The sum
of the three krms obtained by cyclic permutation, with one index held ked,
equals zero. As the notation betrays, R,,,x is the weak fieXd version of the hasor
derived from (3- 17.80):
 We conclude from these resulk $hat posrsible dditiontzl brms in %heLa-
giznge funetion of matter have fhc? form

"'
where mNF9is a tensor, refening to the matter field and the gravitational fieid,
that h@ the symrnetries previousl-y noted. An illustration for s p h OIgenerfziized
from (3-7.88), is

For definibeness, the coacient is ehosen so that the new ~trmstensor, in the
absenee of the gravihtiomtal field,
.,8 = a,$aR - f g,v(a" + m2+2)- it(a,a"*2-- g,. (3-17.127)
has the prope&y
1 -"c.

and vanishe8 for m = 0. When (3-17.126) is uged in the pme-ding equation

we encounter
R,K.ksps'h= R,.'hgK" E (3-17.129)
and the modified spin O Lagrange function is

1%would be indere~tingto ve-fify that this system, with m = 0, is con-

formally iwari%nk,in the sense of (3-17.114) supplemented by an itppropria6e
response for +(X). One sws $hat

is suitable, ifX(z) is eonstanl, To complede the test If is 8uBFieienl .t;o cornider

an, infinitesimd variation of X(%) from uniiGy, 6X(z), Then,

where

is to be computed from
&l,v = &kg,,*
1rrvolved here are
= 6(3, log (-g) 'l2] = 2gP bX (3- 17,135)
and
5/""vr;, .=; -gwap &X,
thus assuring the invariance of the action, for m = 0, under the ~ o u pof
conformal transformations.
Harold interjwts a question,
H. Your preoccupation with conformal trangformrationa in the context of
what is eustonna~lycdted general relativity makes me suspc?cd that you, inkad
to give a aource theory setting for some of the more recent atbnrpts t a enlarge
the fmmework aE general relativity. I am thinking particulshriy of the ideas of
P, Jordan and Brans-Dickc: (B-D), and of Diekek related eRort8 to e~labfish.s
discrepancy between the residual perihelion, precession of Mercuv and the
Einstein prediction, The B-D progosat is based on Mach'@ppriciple bvhich,
while a very intriguing notion, is devoid of immediate obsewational. content.
caxt one sugge~tp o ~ ~ i b i l i t of
i e ~modifying the Einskin theory on somewhat
more ph ysiealt gounds?
S. That is indeed my intent;ion.
h t uss begin, by asking wbedher, through some exhnsion of the theory,
conformal ixlva~aneecould be made an. exact symmetry property. Celrtainly
the msss term of (3-17.130) can be multiplied by cr(z)', where a(%)is a new
scalar field that responds to eonformal transformations as

Furthermore, the conforrnd response of the grravitadional Lwrange function

(3-17.75) could also be compensated by multiplication with ~ ( z ) at ~ ,less* for
constant X(%) which leaves R,, unchanged. And, when one recognizes %had
f -g) 'Rc2 is part of the conformally invariant Lsgrange function (3- 17.130)
[with m = Q and 9 --P a]the generatigafion h arbitrary. X(%) is clear, leading to
the complete?earzformally invariant

where R = g@"R,,retains its meaning in terms of the g,, and their derivatives.
It would seem thsf we have acquired rt new nnasslws particle of spin 0, repre
~ n h byd the scalar field ~ ( 2 ) .But something is amiss, In a weak field situs-
Lion, with
g(%) 1 -4- ~ ( z ) , (3-17. 140)
the dominant f/;, Ldtrms in this Lagraxlgr: funetion are
The p derivative term has the wrong sign. And the source of the p field, pro-
+-
portional h R ~ i ? , vanishes according to (3-17.107). All this indicaks &at
the p field does not descrirbe a physicd excitation. It can be transformed away,
by introducing the conformal transformation \%rith(X(%)) I t 2 equal to to(z), which
reduces the latkr to unity.
Nevertheless, the eonfsrmal invariant version is valuable in pointing out a
new direction, As we have b e n learning in high energy particle physics, nature
does not always seleet what we, in our ignorance, lvould judge to be the most
symmetrical and harmonious possibility. Perhaps the formal inva~anceunder
conformal transformations is broken in suck ai. way that tz, ma8sless, zero spin
particle does exist, IDespik the principle of noriloeality for mmsle-ss padicles,
one cannot object to such a psrlticle on experimental grounds if it interacts with
matter sufficiently more weakly than a ~ a v i t o n ,In order to realize this sugges-
tion, we must add an additional contribution tro the Lagrange function that
eaeictively reverses the sign of the ~r derivative t e r n and msigns it an arbitrary
coeacient. That is not enough, however, for the tr field would still have no
source; it is necessary to destroy the combination R $- ~ t This . can be done
arbitrtzriily, but $he possibilities are illustrated by two elementaq stfkrnative
procedures: remove the g2 faetor that multiplies m 2 ; remove the g2 factor that
multiplies R. The firs%procedure gives a version of %hie?" B-D theory. The
second one has the same practical csnsequenees, and seems somewhat simpler.
Xt is described below.
The modified Lagange function is

+-
where a > O is a new empirical constant. The factor I a is introduced in order
to retain the o ~ G n aphysical
l significance of K * This Lagrange function leads to
the f ~lowing
f field equations :

where t,, is the total stress tensor, adding to the matter contribution that af
the er field,
394 Fields Chap. 3

The equatianv (3-11.143-145) are independent of the made1 used for matter,
provided the matter part of the L%grarrgefunction has been made conformally
invafisxrt by the local introduction of the a field, implying

The field equation (3-17.144) first appear8 as

which is reivritden in the sfstGed form by eliminating 1,. It is also obtained direetly
by applying the sction principle to the conformal response of the noninvariank
f i a ~ a n g efunction,

giving
= --@R -1.
CUK
-(-s)-"2a,[(-g) l~ZB@uau@l] 14-a!
(3-17. 1.50)

The quickest to draw the practical consequences of the modified theory

is by returning to the souree procedures of' Section 2-4, now supglemenkb by a
k r m referring do spin O particles. The respective spin (helicity) 2 and spin O
+ +
sources are appropriately normalized ss [ ~ / ( 1 a ) ]'j2Tp'and [ja~/(1 C X ) ] ' ~ ~ T
where the latter is bmed on the field 4(02 - 1). This produces
1
wz=-- EL:
2 14-ar
fdz)(drt.')[PP(z)D,(z- z")T,,(z" - *T(x)l>+(z - zt)T(zt)
-+ +ffT(s)l>+(a: - zt)T(zf)f, (3-17.151)
and the inkractian enerm with a fixed body of mass M, replacing (2-4.36,37), is

Under the eireumstanees trr << to@,the Newtonisn potential energy is retained,
along with the gravitational red shift. For light, with t k k = to', the deflection
and the slowing of the spwd of light are reduced by the factor 1/(1 $ a). In
+
diseus~ingperihefion precession, the klnetie e n e w correction factor L (22"/m)
+ +
isr hanged it~to1 [(l --- a)/(b a)](2T/w), which gives
3-3 7 The gravitational f Sald 395

and the perihelion precession is reduced by the factor (1 - ia)/(l + a). A

convenient presentation of the correction factors is
light phenomena: 1 --- fa/(1 -$- a)];
perihelion precession: +
l - * [ a / ( l a)].
At the time of writing, memurernents on the time delays in radar echoes
from Venus have produced a result that is 0"8f 0.2 of that expected from the
fensor, or Einstein, theory. T h i ~limits the parameter a to 0.1 & 0.2. Z t hss
also been claimed that 8 percent of the Mercury perihelion precession cart be
assiwed to a solar mass qundrupole moment, leaving 92 percent to be accounted
for by the scalar-tensor ~odificationof the Eingtein theow, This @wes

Until independent evidenee for the solar quadmpole moment is fartheaming,

perhaps from continued obsewation af the asteroid Zcarus, the question whether
a weaHy coupled, massles~, spin O particle exists must still be considered
sub judice.
Now that a scalar-tensor t;beory of gravitation has been devised without
reference to Mach's princi;ple, perhaps a word about this cosmic speculaLian is
in order, I t is a natural hypothesis from the soume theory viewpoint, for it
asserts that the weak field decompositions,

ideniify thc! Gdds of nearby sources----2h,,(z), p(x)-.-and Lhe fields of very

distant sources-g,,, 1. To draw a qualitative inference fmm &is idea withoul
having to use weak field approximrttians throughout the cosmos, we eonsider
an averaged situation in whieh the 'mass of the universe,' M, and the 'radius
of the universe,", provide the only scales for fields and sources. This is ex-
pressed by

where the funetions r,,, s, s,,, IP are all of order unity which is to be judged here
on a logarithmic scale. Then, if we exhibit; only the scale f a t o r s on opposite
sides of the two field equations (3-17.1.1-3, 1441, they read

The implications are

a- X,
logarithmically eonsistenl with a == 0.6 X IQ-', and
398 Fields Chap, 3

tionsl constant K l ~ cm2

- - -
The latkr is a well-known empirical eonnection, between the eonvent;ianal
orders of magnitude, R 1 0 em, ~ ~M 10" g 10" cm-', and the grsvits-
- [Eq.
~ (2-4.40)].
~ If this relation is viewed as char-
aetessdi~of Mach%principle, it cannot be said of source t h w y that the situation.
is qualitatively altered by the infroduction of the scalar field, for (3-17.lfi0) aho
%ppliesto the pure Lensor theor-y.
The suggestion tha%the value of a would be significantly restricted by a
common solution of the two field equations is not qttib b r n e out on examining
a mast e f e m e n t a ~model. To describ-r?it we use geometrical tanwage (for the
first time) and char~eterizethe space as w homogeneous, isotsopie, three-
dimensionally flat space. This is one of the Ffiedmann models;

To be consistent with the purely time-dependent tensor field, the scalar field is
also of that character, a(t), Xf it is assumed Lhgt the matder stress tensor has
only the energy component, too = Pm, the field equations imply

where the dot cjtesignates time derivative. We shall be csnknt ta pick out a
particular solution:
= (t/T)"3, ) (a/3)'j2 log (l/$@),
~ ( i= p, = 0, (3-17.163)
where T indicabs the present era, and

This solution describes the matter density as negXidbXe compared go the energy
density contributed by the a field, whieh, evafuakd at the present era, is

where H is the Hubbfe expansion paramekr

- --
Here is an illustration of (3-17.160), with R T,M pT3. The currently
aeceptcd value of H -. 2.4 X 10-18 sec-' implies p 1.0 X 1 0 g/em3p ~ 8s-
~ ~
s u n n i ~th& a is fairly small compared to unity, Pre~umablythe simplifying
feature of the model, p,, = 0, means that the nnalkr density ip, at least an order
3-"t The grevitationaf field 397

of mszgnitude less than this value of p, bvhich is not inconsistent with the observ*
tional data. The only sensifive dependence on a oecurs in to, the time a t which
Lhe laws of physics bmonne quafitatively similar to those now prevailing, in that
c($)> 0, t >z to, To the e-xtent that there is evidence for the winbnanee of
these laws over a significant fraction of the age of the univerm, a is correspond-
ingly bounded from above. The nominal value a = 0.06 gives to ~ o - ~ T .
The ease with which integer spin L s ~ a n g efunetions acquire general co-
-
ordinette invltriance by suitably introducing doe8 not extend to psrtieles
of integer ++ spin, T o appreciate the difference let us folloui. the earlier wwk
fidd procedure, now using the apin 4 Lagrange function

and the stress tensor [(3-7,1=)iO)f

to form
+
t,. = ~ + ~ O b [ ~ , ( l l i ) ar.(l/i)a,lJ.
v + (3- 17.1438)

g($,h ) = g($)f t,,hp' -- --(l $ h)f#r0[r,(grP -- hrp)(l/i)a, + m]+.

(3-17.169)
Appearing here is, not
Q ~ ~ (CZ ). - 2hFy(x),
A%g&Y (3-17. 17a)
but something resembling the square root of this combination,

The necessary generali~atioscan, be car~r;dout, hawever, if we distinwish

the vector index in gc"" - hP"(z)that is asmciszted with the coordinab derivative
d, from the vecbr indw &at is tied to the matrices 7,. To emphasize this we
henceforth use fatin fetters to indicate a focal Minkourski coardindtte system,
which will be ret;ained for the description of spin. Thrzt; notational diskinctian is
used in writing the generalization. of (3-71.171) m
g" (z) = efia(z)g.aePb( X ) = eB'(z) etfz), (3- 17.172)
which, is maintained under general coordinate transformations if

There is an. independent invariance under Xocal Lorentz tnznsformations:

aPa(~)= lab(%)@p6(z) ,
where
l", (z)gab16d(z) = grd.
With the definition
(4 = @PP ($14 (2)
we deducd from (3- 17.172) that
 Chap. 3

and
g,.(z) = e,"(s)g.be!(z) = e;(z)@~.fr) (3-17,178)
Regarding the fist form of the l a ~ equation
% aa a, matfix prduet, vc-e infer the
dekrminantal relation
-B($) = (det (3-17.179)
or
(-g(z)) ' l 2 = det @,.(X) = &(g). (347.18C1;)
The indicatd provisional generalization of (3-17.16I)) is

The eammponding aetion is certainly inmriant under arbitrav c o d i n % &

dmnsfclrmaptions i f the field $(z) is transformed W a, scdar,

But we shdX also require inva~anceunder arbitrav Xocd Lorentg dr&mfornn&

%ions,for wbick (3-17.181) ia irradequak.
To ztpprwbte the physical significance of the lmt requirement, Iet US can-
sider the response of the matbr action Co va~atioxrsof e : ( ~:)

which defines tE(z). An infinihsimai local Loreats transformation is

with

Invariance wiLh rapeet to arbitraw transformalions of this tyw 'thus deman&s

that
tb"(z)= e@"(ztC(x) = tU"(z), (3-17.186)
which i s dso the symmetw p r a p e ~ y

The t e m r that is required to be symmetrical is indeed the stress knsclr of

matter. This fotlaw~on. ~ f i n g

which repmduces the 8Gress knsor definition (3-7.1002,

The response of +(S) to the local LorerrLz Lransformation (3-17.174) is

3-1 7 The gravitational field 399

where
L~TOL = ?02 = l"flb. (3-17.191)
It is the action of the coordinate derivatives on &(l($)) that disturbs the
foeal invafiance of the Lagrange function. (93-17.181). A coordinab displace
ment induces an infinitesimat. Lorentz transformation and an associated field
tansformation :
Eag(z 4- dz) = la,(x>[8i4- hcb(~)]p
(3-17.192)
L(l(z f dz)) = L ( l ( z ) ) [ lt &'d~.b(z)+@"I,
in which
dwab ==t -doba E lCa dlcb, (3-17-193)
and therefore
L-'d,L = +iF,a,lebgab. (3-17.194)
I n order do compnsate this eRect the coordin&e derivative in the tagrange
function is replaced by
a, - t i ~ . , ~ @ " ~ , (3--~7.195)
where ua,b(;c) behwes like a covariant vector with respect t o general coordinate
fransfarmatians, and responds do Xoeal Lorentz transformations in such a
manner that
L-'(a, - ai@.,boa6)L = dg - aiwapboab. (3- 17.196)
The required transformation law is

A fundamental mixed t e n ~ a is
r defined by the cornrnutrtLor

which has the character of an antisymmetricat Lensor of the seeond rank for
general coordinate transformations, and of an antisymmetl.iea1 tensor of the
second rank with respect to focal Lorenta transformations. A scalar in both
senses is construct4 by
= R(z),
e'"(z)ePb(z)Rpv.b(x) (&17.200)
and provides the basis for a gravitational Lagrange function:

I n the weak field limit, where the linguistic distinction of %heindices is re-
moved and
@"(X)c?i g@@- h p a ( % ) , (3-17.262)
 Flslda Chap, 3

this Lagange function rduces to (3-1?.38), ixpa& from s divergence km.

The ~ p l i e a t i o nof the stationary action prinoiple to variations of U,,&
in, a(e, CJ) E;ives
8.[e(e'@ePb- - oo,,[e(&'ePb - @'epC)]- w$,[e(&@epe- dcePo)]= 0.
(3-1 ?.m31
The following definition of ii quantity $h& is not a tensor,

enablm one to prrtsent this equation W

where
aabc + @aeb "- @cab "“k gbcha - @&BAG Q:, (3-17.m)

A. = e-'a.(eeL) - W.* = -(nut -+ (3- 17.m7)

meordintg t-o the dehrminsntal fornnuta

On eantrm%ingb and c in (3-17.209 we get various muItipIm of ,X, implying that

from which X, = O ia %%ilk

recovered. The East ref%tionis solved by

it is the ~ t r o field
~ g generalizeztion of (3-1 7.17).
Amth:er w a k Md propem, Eq. (3-17.n), is generalized by defining

vvhich syrnrnetry p r o ~ & y=presses the rc3latian of (3-17.210)* The additional

definition
rip= e z e ; ~ . ~ . e ~=' F,X (3-l"1.213)
enables on@.to w ~ t (3-17.212)
e EH

Mul%iplicatiom;af this equation by ePa, foflowed by symmetrisatioa in y and v ,

removes the ua&b krm and givm
3-1 7 The gravitational field 401

This is recognized aa the statement that the covariant derivative of gf"" vanish=,
, : ' l with the quantities of (3-1 7.91), known usually a s Christoffel
and identifies
symbols. After this, it is abundantly clear %hat the two objects defined in.
(3- 17.124) and (3- 17..f99) are connected by

where the correctness of the algebraic sign can be verified in the weak field
limit* Thus, the two gravitational Lagange functions, $(g, l") and c(@,O)
are idemticd.
Returning do the spin $ Lagrange function (3-X7.181), we insert the eo-
ordinate d e ~ v a t i v egeneralization stated in (3-17. 195) and obtain

where, i d should be noted, $he total antieommutativity of the field extracts the
antisymmetrical part of the matrices u0r*obC.This removes the terms with
a = b or a = c, trhieh are proportional to the symmetrical matrices roro.
Then, since

we can write the Lagrange function as

+ + m]$,
c(+,e) = - e i $ ~ ~ [ ~ ' e i ( l / i ) a , *oGiv,r, (3-17.219)
where
*& = aC*bcdWabe. (3- 17.220)
The notation g($,e) might have been elaboraled as C(#, e, w). When this
structure i8 added to the gravitational Lagrange function (3-17,201), in whieh
eg and are used as independent variables, the W,,& dependence of g(+,e, W )
produces an additional term in (3- 17.203), ~vhich removes the symmetry
property noted in (3-l"7.212). This is a natural form of the theory, But we shdl
prefer, far simplicity only, to regard a a b e as defined by (3-17.21 1) and therefore
not subject to redefirzition through the appearance of @,be in the matter Lagrange
function. I n order to identify the stress tensor t,, directly, we consider the
special variation
6eg = -$-ligPve,,, (3-17.221)
die21 is consistent with the construction of gP"ronrr the e,: and gives

A little cafeulalian sho\vc.s that

 Chap, 3

and thus,

where the first two teirns can be united through the reintroduction of the spinar
eovariant derivative (3-17.f 95).
The sedar defived from this k n m r is

A. aimplifieation can be made through the we of the field equation implid by

the Lagmnp funetion. But what is requird here can km obtain& marts directly
by 8ppfying &heaction principle to Lke gadicular field vadation

The rmpontse of the mat;ter action term is

since no contributit;ion involving 8,6h appears, owing Lo Lhs symmetry of the

matrices rave. The conclusion is that the Lagrsnge function vanishes (at points
not occupid by pa&iele sources) and

The lmt wnsiderakion is intimately relabd to the p o ~ b i l i l yof exhibiting s

@onformallyinvafiant matter Lagrange funetion khrough the introduclion of
the scdar field cr(s).
Conformal tmnsfarmLions on the t e t r d of veehr fields G(z) appear as

or, using infinitesimal &mnsformations,

The approprlak conformal bhavior of 9(2) is alwady @tat&in (3-17.22fi).

It is such that %hereplacement of m by mrr in ($17.219) maces to produce a
3-1 7 The gravitational Pfafd 403

conformally invariant Lagrange funetion. The deriv~tionof the ~cralart thmugkr

the tr-dependence of g($,e, a) according to (3-17.147) then @ves (S17.228)
directly, with mcr substituted for m,
The temptation to extend this diseusgion to arbitrary multispinor fields will
be resisted. Instead, wc %urnto the long deferred topic of the pavita&ional
model. of particle and gra.viton sources, I t is wodh appreciating why id ia thaf
we have managed thus far u~ithoutexamining this question, The arena of gravi-
tational phenomena is confined essentially ta ~tronomicalbodtim, which are
beyond our exger"rxnentft1 control. Ftceordingly, urehave no overt use for ~ o u r e e ~ ,
lvhich give idealized expression to the experimenter's ability to manipulate
the physical situation being studied, The graviton source concept has already
fulfilled its primary mission by serving as the modet upon which the eoordinale
invariant dynamical theory has been erected. NeverLheless, some remark8 are
ewled for, &though, as the pmce$ing comment; indicales, they ean be limiM
to the use of particle and gravihn sources under the weak field graviLaLiond
conditions prevailing in terrestrial experiments. The following brief analysis
wed not be applicable to experiments conducted on a. spaceship in d m orbit
about a, rapidiy spinning neutron star.
The first point a t issue has previously been raised far charged parliiicle
sources. The generalized stress tensor consewation law, (3-17.103, 1M), ~411
fail inside particle sourees unless one recognizes the pre-exisknee of the enerw
and momentum that is trzlnsferrd to the emitted particle, There is, however,
no eIectromag;netic andogue to the graviton murce problem. Photons are
electrical1y neutral, rvhereas gravitons carry enera-momentum which must dso
be transferred rather &an created within the source. Just as an explicit A,-
dependence was introduced to provide the correet gauge transformation bhavior
of charged pal-l;iele sources, these source problems can be viewed as the search
for the explicit gPY-dependence$fiat will give the varisus BOUI"~C?Sthe eomeef
response to general coordinate transformations.
The simplest; example is a scalar source K(s), appea,~ngin the action
through the term
(dz)(-g(z)) L'Z~(z)@(z). (3-1 7.234)

We must replace x"" by a functional of the g,,,zpfx, g), such thstt under a generd
coordinate transformation
xp(z,8) z== ~ ' C x g),
t (3-17.235)
for then
shows the requird dynsannkal equivafeme of the field8 y,,(a), #(s) a d 51,,(2),
+(X), Under weak field conditions we write

where the inva~ancepropedy (3-1 7.235), staged for infixritmimal f retnsforma-

%ions(2""= f l - &Y),requires that

This must hold as a, consequence of the gawgcj translcrrmation (3-17.45). A

solution is

(dz" )(dx")fIr (X:- xylf" (X' - S'') 4, (S"),

(3-17.239)
where f"(z - x') )is one of the familiar elws of functions obeying
aufc1(z - X') = &(X - a'). (3-17.XO)
Tbe matter strms kngor that is rtow dc?rived, slakd in. the absence of %he
pavitation field for simplicity, is the consewed objeek

(3-17.241)
where tC""is fhe tensor given in (3-7.8), which. is sueh Lhst

The possibility of ma,vil;a>rremission otscu~ngdirectly from the rrtafkr

source is exbibiM, for single graviton rta&a%ion,by writing the coordinate
i n v a ~ a nft o m of the murce term (3-17.234) m

Asmrning that the graviton detection sources do not overlap fhe K support re-
@on,one can use Lhe sourwfree, weak gr~viLtls1ionalfield equations (3-17.13,14)
La derive
 The gravitational f Deld 40Ei

This enabf es one do present (3-1x243) as

Alternativety, one might have b w n with the la& form, where the additional
k r m mrvw tO remove the responw of + ( X ) to infinitesimal coordinate trans-
formation~.Similar discussions can be given for any other type of matter source
and fidd, wifh appmpriate &Lention to their transfornnation properties.
The weak field form of the gravitorr source term in the action is
TRdnI"pCo> 4 (dx)TY(z)~lr.P(z)
(d.) --) P (3-1 7.246)
where
d J I T ~ ( x=
) 0,
and a eonstanL is added to &rrive a t the second vemion of (3-17.246). The
physical property to be represented is thaL the radiation of an ztdditional grsviton
can accompany the working of a graviton bouree as well as a ma;tder source. The
mathematical problem is the removal of the responsle of g,(z) to infinitegimal
coodinale transfarmatians, apad fmm gradient hrrns-gsuge transfarmationls
-which do not contribute in (3-17.246). If we use the symbol to indicak
identiw apart from gadient term^, Che response of gr,, to infinitesimal coordinate
tr~nsformations, Eq. (3- 17..96), is expregsed by

Then, appf ying the weak field stakement (3- 17.45),

which gives the required generalization of (3-17.246) :

The source Pp(%)

that is notv derived through variation of g,,(z) is

Through its dependence on the gravitatiortd field to the required accuracy,

P" doe8 respend appropriately to infinitesimd coordinate transformalions, and
o b y s the divergence equation (3-17.104). Finally, it should be said that, tzs
 Fields Chap, 3

in the plnoton situation, the consider4ttion of additional radiation h r n the

Bources ean be avoided by &opting s n equivalent g&uge, The pavitational
gauge eon&tion is
(dz')fP(z - z')hiP(xp)= 0, (3-1 7.253)

feding to %bevanishing of the X h terms.

This volume closm with s short exchange bet~ieenEarold arid the author.
H. How Can it be the end of the bmkl You have haray hewn. There are
any numbw of addifionail topics X should like to 8% developed from the view-
point of source theory. And think of the field day you wiIl the revimers,
who usually prefer to list all the mbjeets not included in a volume r ~ t h e rthan
discuss what it dws contain.
S. Quite tme. But we have now reslelned the point of tr&nsifionLo the next
dynamical fevd. And, since this volume is already- of ft re-mnable ~ize,and
many of the ideas of source thmfy are in it, if" hardly fully developed and applied,
it m m s better to put it before the public as the first valume of a s e ~ e s .Hope-
fully, the next volume will be prepsped in time to meet the pawing demand
for more Souree Thwry.
 WOW T 0 READ VOLUME I
The first volunle was described as a resewch document, and a textboak. Un-
fortunately, the beginning student was given no guidelines to tell him into whicfi
category a particular section fell. hecordingly, here are some suggestions for a
first encounter with source the or!^, and relativistic quantum mechanics,
a) In Chapter I , omit Section 1-4.
b) In Section 2-1, the derivation of the Lorentz transformation behavior of the
source function from that of states can be omitted, I t is sufficiently evident;
from the form of Eq, (2-1.38), for example, that K ( x ) is a scalar function.
c) Omit the muiti-particle generalizations of the vacuum amplitude in Section 2-2,
They are of interest primarily in many-particle applications, which are not
yet a t the center of attention,
df Sec3tion 2-5 need be read only to appreciate the general linear transformation
of sources and its relation to spin, tagether with the .possibility of campsing
arbitrary spins from more elementary ones.
e) The discussion in Section 2-6 that begins with Eq. (2-6.24) can be omitted
by recognizing directly that (2-6.26) is the covariant generalization at the
+
projection matrix 4 (1 ps), which selects a definite parity in the rest frame
and, tbereby, the two components appropriate to spin 4,
f) Omit the multi-particle generalizations in Section 2-7,
g) I t is sufficient, In Section 2-8, t o read the discussion of spin $.
11) In Sections 3-1 and 3-2, omit multi-particle generalizations.
i)Omit the discussion of spins 3 and in Section 3-25.
j) Xn Section 3-4, restrict attention to multispinors of ranks 2 and 3.
k) The spin liznitations already noted should be continued in Section 3-8,
E) The rambling discussion about the arbitrariness of stress tensors that appears
in Section 3-7 should only be skimmed,
m) The lengthy account of magnetic charge and its conceivable relevance to
hadronic behavior [Sections 3-8, 3-93 is optional, However, don't miss the
debut of Harold on p. 2401 nor the remarks on mass normalization [p. 2471.
n) Most of Section 3-17 is optional reading, particularly the discussions of broken
conformal invariance, cosmology, and spin gravitational coupling.
 Appendix

Finally, we add two minor comments about specific topics in Volume I.

X . The discussion of Eqt . (1-1.44) does not make clear that cornmutativity of the two
displacement operators remains an alternative possibility (the numerical
coefficient zero cannot be changed to unity by redefining the operators),
2. The f o m s of Lagrange functions that yield first o d e r differential equations
were merely stated in the text. The genesis of thew expressions might be
clarified by this illustration far spin Q. Begir*ningwith the second order fom
[Eq. (&5.12)]

we introduce the independent veetor\field 4, by adding to 9 the term

The nature of the system is not changed thereby since, on extending the action
grinGiple to $, we learn that QI, - vanishes, apart fram a pssible source
term. But, on adding (A-1) and (A-21, the squares of the first derivativs
cancel, producing the Lagrange function (3-15.161, from which the fimt-order
field equations fallow, This procedure is the analope of one for ordinary
mechanics that begins wit h the quadratic Lagangian

where m is a mn-singular symmetrical matrix, and inlrducw the indegn$ent

variables $ by adding

The sum of (A-3) and (8-41,

yields the equivalent first-order Hamiltonian description,

 Index

AeLiont 186, See aka Lerange function for spin 1, 188

additivity of, for noninteracting for spin g, 191
particles, 256 for spin 2, 189
for arbitrary spin, 191 for spin 8, 191
coardixl~teinvariamt gravihn sourw for spin 3, X96
term in, 405 stationary, 187
discontinuous ehange of, and eharge of time cycle deaription, 197
quantigation, 242 for two kinds of photon sources,
electrom~gnetie,displacement charge 239
of, 248 Advanced Green" function, 148-149
with field strengths as variables, 340 for Dirac equation, l62
modificsticln for point charge@, AmpBr-Ian inkraetion, 77
%3-244 Analytic contiauation, 36
for point charges, 247 Analytieity, 35
of ebctromqnetic field, 229, 235 Angular momentum : commutation
for gravitoa, 198 relations in three and four
of graviton field, 381 dimensions, 88
pavifon source brxn in, 485 exgeetation value, relation to flux.
helieity decomposition of in nr-tas~lesa vector, 224
limit, f 97 and souree rot&ions, 223
for interacting par-ticies, 282 sgecification of stab@,53
ixzva~anceof, 283 Angular momentum aux vector, local
under gauge and coordinab conservation of, outside murces,
transformadionru, 383 223
under TCP, 277 Angular momentum operskr, 8
partial, 278, 279 Antiparticles and t3pace-time uliformity,
for particle with grewribt3-d motion, 24'7 47
for photon, 198 Antiperiodie Green" function, 163
quantum unit of, 8 Area, f WO-dimensional,247
mduced form, for second-rank qinor, Astronarnicd appfieadions of source:
192 theory, 82
mcond-order formulation for third-rank Auxiliary fields: for third-rank
spinor, 194 multispinnors, 179
for spin 0, X 87 for fourth-rank mul6iirrpimrs, 181
with primitive electromagnetic for fif th-rank muftiqinor~,183-184
interaction, 256 for @pin$ parli&s, 117
for 8pln B, 190 for @pin3 particles, 171, 172
bawd on srjcond4rder quafion, 193 hid degcription, 84
409
410 index

Bargmann-Wigner equations, 264 conservation of, 72

Barryolxs, 251 twa-dimensional lattice far, 251-252
magnetic model of, 253 universality of unit of, 250
Basis, 2 electric land magnetic, coexistence of,
Beam of particles, 41, 58,57 242
Bhabha sclcbttering, See Scattering of exmctation V ~ U Z Irelation
, ta flux.
spin -& particles with opposite?! vector, 200
charges and fermions, f 13
Booster, 8 leptonic, 125
Bose-Einstein statistics, 55 Iocalisation of, 243
Boson, 113 magnetic, 231
Boundary condition: for Gfe(?n88 unit of, 250
function, 146 of multiparticle states, 53
outgoing W&*, 153, 157 purely magnetic, unit of, 252
of periodicity, 150 Charge fluctuatim ftux vector, 207
retarded, 148 marge Buetuation~and p h m
of vacuum time cycle, 148, 162 transformations, 207
Bound slates, 346 Charge flux vector: local conservation
Brrtns-Dieke, 392 of, outside sources, 200
Brernssdrahlung. See Photon emission for spin O particles, 200
in Coulomb scatteGng for spin particles, 204
Broken conformal inva~aneetheory, for spin 1 particles, 203
383 for spin 2 particles, 204
prscticsl consrtquences of, 394 for third-rgnk spinors, 206
Broken symmetry, 393 Cbsrge matrix, 48
Charge quantization, 239
condition for, 249
C, 49 Charge reflection mratrix, 49
Cauchy principal value, 46 Charge symmetry, 49
Causal analysis of vacuum amplitude, Christoffel symbols, 401
51-52, 109, 119 Clifford-Dirac algebra-, 105
Causal control of H-particle transitions, Coherence in scattering, 358
362 Collisions, 37
Causality, 313, 37 caussll control of, 36
and unitarity, 61, 122 Commutator, 2
Causal structure of propagation functions, Complernentarity, 38
42, 5& 559, l20 of source descriptions, 39
Center of mass, 35 Complettlness, 41, 53, 58, 59, 121, 158
G.G.8, syatem, iv of H-partieb states, 348
Chalcidian slababet, 234 for particles of arbitrary spin, 137
Charge, 47-48 of spin particle states, 110
accelerated, and rdistian, 265 of spinors, 319
conservation of, 255 Complex conjugation :of Fermi-Dirac
in interaction, 285 sources, t 10
distribution of, 77 of sources, 131
dynamical and kinernatical arspeet;gs, 2 s Complex fields, 153
eigenveetors of, 284 charge interpretation of, 1%
electric: %xisof, 250 of time cycle description, 154
Components of a vector, 2 of covariant veetor, 386
Composite systems, 344 of g, P",g,", 337
consistency of phenomenologieal of bnsors, 386
description, 357 Critique of particle theories, 24
Compton scattering. See Scattering, of Crossed reactions and anslytieity, 36
photoas by charged particles Crossing relations, 33, 290
Conformal group, 225 and A- and B-cwffieients, 353
and conmrvittion laws, 2226 and electron-photon scattering, 318
for electromagnetic field, 230 and electron-positron scattering, 308
represented by %otatians,2225 and pair creation by a, photon in a
and stress scalar, 226 Coulomb field, 343
Conformal invariance of action for spurious minus sign in, 318-31 9
massless spin 0 particles, 39% Cross section : definition of, 286
Conformal transformations, 388-3W differential, 287
Gonmrved current, construction of, 255 elastic, near resonance of photon
Constructive principles: of S-matrix scattering by H-particle, 377-378
theory, 35 total: for arbitrary energy electron-
of source theory, 31 photon scattering, 320
Contact terms, 144 for electron-positron creation in
and field digerential equations, If38 two-photon collision, 31S
introduced by source redefinitions, 173 for high-energy electron-photon
in multispinor description, f 77 scattering, 319
for spin g particles, 176-177 near resonance of photon scattering
for spin 3 particles, 171 by H-particle, 374, 377
Coordinrcte displacement, local Lorentl; for slow electron-positron
transformation induced by, 399 annihilation into two photons,
Coordinak invariance, general, and 312-313
+
particles of integer 4 spin, 397 for spin O particfe pair annihilation
Coordinate invariance, infinitesimal, of into two photons, 299
spin O Lagrange function, 384 for two-photon annihiiation of high-
Coordinate systems: in @ace-time, 7 energy electror, and positron, 315
transformations between, 332 for two-photon annihilation of
Correction factors for light phenomena, unpolarized electron and positron,
and perihelion precession, 395 317
Coulomb gauge. See Rdiation gauge Current algebra, iii
Coulomb inleraction, 77 Current vector, 201. See also Charge flux
Coulomb pobntial, 320-321 vector
Coulomb scattering : connection with ambiguity in, 203
nonrelativistic limit, 323 associated with particle stste, 201
and dynamical levels, 322-323 electric : for arbitrary multispinor, 2434
of spin O particles, 321 asmeisted with charged particle
of spin particles, 322 murces, 258
Coulomb correction to, 324 identity of kinematics1 and
phaw shift between helicity dynamical definitions, 262
transitions in, 327 for spin 4, 260
spin dependence of, 328 for spin I particles, 263
Govariant derivative: of contravariant for spinor-tensor field, 265
vector, 386 of nonconsrved point charge, 258
412 Index

Decay constant and width of spectral for high-energy urrpalarised electron-

line, 371 positron scatbring, 308
rlelta function, four-dimensional, 146 for low-energy spin 4 scattering of
Determinant: e, relation to g, 398 unpolrzrized like chetrges, 304
and Fermi-Dirac statistics, If 8 for low-energy unpolarised ebclron-
g, transformation bhavior of, 385 positron scathring, 308
Differential cross wetion: for arbitrary near resonance, for photczn scattering
energy electran-photon scattering, by H-particb, 373-3cr4
31+320 for photon emission in Coubmb
for arbitr%ryenergy spin 4 scatkring scattering of nonrelativistic spin 0
of unpdt-srized like ehtzrges, 305 particle, 330
for Compton scattering by electron for photon scatbring, by H-particies,
a t rest, 341 358
for Cornpton scattering by spin O by realistie EX-partichs, 3%
particle at rest, 335 Rutherford, 289
for Coulomb scatte~ngof &pin0 for scatbring, of circularly polarizled
particle, 321-322 photons by spin O charged
Coulomb correction to, 324 particles, 286
for Coulomb scattering of spin -& of linearly polarized photons by
parlicle, 322 spin 0 charged particles, 295
Coufonnb correction to, 328 of spin 0 rtnd unpolarised spin 3
definition of, 287 pmticles, 310
for forward and backward emission in of win. Q particle by massive spin
high-energy electron-positron 9 psrticle, 31 t
annihilation, 316 of anpolarized spin 4 particle by
for high-energy electron-photon massive spin O particle, 31 X
scattering, 318 far &pinO particle pair creation, by
far high-energy electron-positron cireularly polari~edphotons, 298
wattering, with equd helicities, by Iineetriy polarized photons, 297
307 for spin O particle pair annihiltiltian
with opposib bhecities, 308 into circularly polari~edphotons,
high-energy, for two-photon 299
annihilation of urrpolariged for spin 0 particles, of like charge, 288
electron and positron, 314 relation btween like and unlike
for high-energy pair creation by a charges, 8 0
photon in a Coulomb field, 343 for f wo-photon annihilation of
for highenergy photon emission in unpalarized electron and positron,
Coulomb scattering, of spin O 317
particle, 338 unit of, 296
of spin 4 particle, 343 far unpolarizred etectron-positron
for high-energy spin $ scattering of scattering at arbitrary energies,
like charges, with equal helicities, 309
302 for unpolariaed photon %tatbring by
with opposite helieities, 303 f pin O efiarged psrticfes, B 4
for high-energy spin $. scattering Dif3Ferential equations, firstorder :for
of unpolarized like ehares, electromagnetic field, 228
303 for srbitrary-rank multi~pinors,f 85
 Index 413

for fourth-rank multispinors, 182 Dynamical evolutian: first level of, Z8

for fifth-rank multispinors, 184 wcond stage of, 378
for graviton field, 380 Dynsmics: in operator field &gory, 33
for heliciw 4 fields, 176 in S-matrix theory, 36
for spin O field, 187 Dyon, 263
for q i n field, 160
for spin l fields, f M,178, 188
for spin # fields, 173 Effective local interaction. betwwn
for spin 2 fields, 166, 189 Maxwefl and Diratc Gelds, 313
for sipin 8 fields, 177 EEeedive tweparticlc? source: for photon
for third-rank muldispinors, 179 and spin 0 charged particle, 266
DiBerential equations, fourth-order, for for photon and spin 5 charged particle,
fourth-rank muldisginorfi, 182, 276275
DiBerential equations, %can$-order : for Eigenfunctions: related do nonrelativistie
electromagnetic field, 228 wavefunctions, 352
gauge covftrianl for spin O field, spin 0,orthonormality of, 345
for gravilon field, 379 spin 4, eomplehness of, 35S360
for helicity 3 fields, momentum space orthonormality of, 348
transcription of, I69 Einstein A-coefficient. See Transition
for second- and first-rank spinors, probability, per unit time, for
180 R-particle radiative transition
for spin 1 fields, 164 Einskixx B-coescient, 353
for spin 2 fields, f 66 Einsbin grsvitational field equation,
momentum space transcription of, 388-389
167 Einsteinian rebtivity group, composition
for spin 3 fields, 172 properties of, 15
for third-rank multi~inors,180 generator eommutrttian relations, 16
Differential equations, transformed by Einstein theory, obmrvationat tests of, 82
source redefinitions, f 74 EEectrie charge, Xoeal conservation law, 28
Dilations, 215 EIectrie dipole moment, exltt-rnal, of
isstropic, 224 H-pal-ticle, 359
Dipole xraoment :electric, 13 inkrnal, of H-pafiicle, 358
of spin g particle, 261 Electromagnetic fieM, 227
magnetic, 13 commutation relations for, 26
Birac equation, wage of fields in, 1611 Electramagnetic model of sources, 255
Displscernent : of coordinate frame, 39 Eleetromagnetie source models, 257
of muree, 39 EXeetron, 34
Dispbcements: arbitrariness in, field Electron+lectron scsltering. Bee Mplller
responm do, 214 scattering
in causal situation^, 210 Electron-positron scattering. See
variable, 209 Bhabha soatbring
Double scattering and polarization, Energy :internal, 11
328 of quasi-static source distribution, 77,
E)wl charged particbs, 251 82
charge assignments of, 253 Energy-monnentum : con~rvationof, 8I
Dug1 field strength tensor, 228 expectation value, relation to flux
Dual knsar, definition of, 17, 74 vector, 220
 of multiparticle states, 53 Field, electromagnetic, 26
prwxistence Bvikhin sources, 403 Field equrttiona. See alr~oBiSerenthl
Energy-momentum eon~rvationand equeztions
kinemstical integrals, 288 of broken eonformal invariance theary,
Eneru-momentum flux vectar, 209 393
lac& conmrvation of, out~idesources, Dirw, with electromagnetic
210 inbraetion, 262
Enmgy operator, 8 nonlinear, 279
q u a t i o n of motion, proper Lime, 23 gravitational, 388-381)
Euclidesn Green's function, 146 MaxweU, with chsrged particle current,
Euclidean postulate, 44 257
and arbitrary spin, 189 in grltvitational field, 388
and *in 3 particles, 111 nonlinear, and physical proeessa,
Euclidean propagation function, 44 B7-278
inequalities for, 45 for spin 0 padiebs with prinaitivcs
Euclidean space, attached Ito Minkowgki ellectromagnetie interaclion, 256
space, 43 for spin I. particles with interactions,
Exclusion prineipb, 109, 119 263
E w c t a t i o n values, fr13 Fields, 145
Exbnded sources, 265 msoeiabd wit h individual emission,
Exbriar algebra, 106 absorption act, 154
auxiliary, 171, 177, 180, 182, 183
causal evaluation of, 153
f@: cfsss of, 233 complex, X53
eovariant choiee of, 260 far spin. 3 particles, 15%160
covariftnt, physical interpretation of, of Lime cycle; description, 154
269 correlation funetions of, 30,32
and electromagnetic source models, 257 electronnagnetic, 227
quation for, 233 elimination of, 278
quivalenee of electromagnetic warm equd time commutation relations of, 32
model. and gauge intnz-rpretations, I[",,x, of gravitons, 380
269 grsvitationd, 378
equivalence of timelike and spacelike graviton, as wesk gravitational, 385
vectors, 259 helieity decomposition of in massleas
snd gravitation81 gauge condition, 406 limit, 197
and gravilcttional source models, 404 of helicity 8 prtrtieles, differentid
and magnetic charge, 241 equations for, 176
W noneonserved etrrrent veetar, 258 of helicity 3 psrticles, dif"ferclntia1
and scatbring experiments, 279 equations far, 169
upp part of, 248 H-particle, 351-352, 354
asymmetry restrickion on, 234 integral equation for, 365
uwd in conmrved curreat consItruelion, Maxwel.1, conformal Lr%nsformationof,
255 390
m d in gauge condition, 238 multispinor, 1'7'7
to eharacbriz;e gawe, 255 WBhr, of ~ ~ B V ~ L O R S380
,
1Fermi-Dirac statistics, X03 operator, 24
Fermion, 113 r e d and imaginary, 2CZS
 retarded, 147 Grassmann algebra, 106
of Dirac equation, 162 Gravitational constant, 82
spin 0, for causal arrangement, 284 Gravitational deflection of light, 83
of spin 0 particles, 145 Gravitationd field, 378
spin 9, causal expressions far, 159 conformal Lrsnsformation of, 389-390
response to local Lorentz e;, 397
transf ornnations, 398-399 e;, 307
of time eycle description, f 61 g@",384385.
of spin particles, 157 g,,, 385
differential equations for, 160 rak, f l i p , 400
of spin 1 particles, 164 F;,, 385
differential equations for, 164 l'$, construction of, 387
of spin 8 psrticles, E72 X",: transformation law of, 386
difffsrential equations for, 173 400
of q i n 2 particles, 165 %&b, 399
digerential equations for, l66 Gravitational red shift, 82
of spin g particles, 175 Gravitational slowing of light, 83
diEerentia1 equations for, 177 Graviton, 80
of ;spin 3 particles, 167 wtian for, 198
differential equations for, X72 etdditional, ernitkd by graviton
tensor, of gravitons, 378-379 source, 405
of time cyele description, 147, 152 emission from nnatbr source, 404
of unstable H-particles, 369 helicity states of, 81
Field strength Censor: d u d to, 228 a parable, 81
electromagnetiic, 228 vacuum probability amplitude for, 80
Fine stxrrcture constant, 250 Grsviton source concept, 403
Fluctuations, 64 Grftviton source problem, 403
Friedmann models, 396 Green" function, 146. See also
Pragw8ticzn funet ion
g, for arbitrltry multispixlor description, advanced, 148-149
265 of Dirae equa;tion, 162
dependence an description, 262 antiperiodic, 165
for spin 3 particle, 280 associated Euelidean, f 46
for spin l particles, 264 boundary condition for, 146
Galiban relativity group, composition for charged spin Q particle, 271
properties of, 8 for charged spin g particle, 218
generator commutation rehtions, 9 Dirac, causal expressions for, 158
Gawe conditions, 238, 258 of Dirae equation, 157
on MftxweXl Green" ffunction, 279 Birac, momentum integrals for, 158
Gauge covariant derivative, 262 of Laplaeek equation, 260
Oauw invariance, B8 Maxwell, olzeying gauge condition,
grwitalional, 380 278-279
Gsuge transformations, 228 modified periodic, 155
Abelian group of, 262 periodic, 149
grtrvitstional, 380 Fourier series construction, af, 151
Cwa kinds of, 2363 retarded, 148
Generators, 3 of Dirac equation, 162
 spin 0,eigenfunction construction of, of neutfinos, 125
344 spin I states of, 69
spin 0, expanded in powers of vector spin 2 staks of, 79
potential, 281 Hermitisn adjoin%,1
spin 0, iterative constmction of, 351 Hermitisn g matrix reafigation, crihrian,
spin -$,eigenfunction construction of, for, 6
348 Hermitisn operator@,1
spin $, exlpanded in powers of veetor infinibsirnetl, 2
potential, 280 Homogeneous electromagnetic field, and
spin 3, high energy limit of, 35s3fi0 spin O Crwn" function, 272
spin ikratim construction of, H-partides, 344
383-354 fields of, 351-352, 354
Group :Abdisn, of gBuge transformations, I' matrix, 3fi3.
262 ixthgrlat equation for fields of, 365
AbeIian, of translations, 5 modifid propag~tionfunction, a(i3
cornmutation relation8 of gener~brs,4 persistence probedbility, 3Ci8
eommutation relations for three photon scattering by, 355
paramekra, 7 Il matrix, 365366
composition properlies, 5 propagation function, time limitstion
conformal, 225 on, 365
Euclidesn, connected pieees of, 49 rdiative transitions btween, 350
finib dimension&lreali~ationof, 7 skeletlllt irtterwtions with photons,
of Galilean transformations, 8 351,354
generators of, 3 sources of, M?,349
Lorents, connected pieees of, 49 unstable, fields of, 369
hrentz, finite repremntcation~of, 86 photon emission by, 368
non-Abelian, of general coordinate time kbavior of propagation.
transformations, 384 function, 368
parameter space, 7 s s unstable particles, 360-361
structure constantsp4 virtud, 362
of unitary transformations, 3 as effective wurees, 365
Gyromagnetic ratio, See ako g Hubble e w s n i o n parameter, 396
of electron, 13
of muon, 13 Ido, Zchir6, 81
Infinitesimal rotiztions, response to, 9
Hadrons, 251 1n6niLesimaI Lrsnslations, response to, 10
m magnetically neutral composites, 251 Infra-red catastrophe, 273
Harold: on history, 338 Interaction skeleton, 277
on magnetic chsrge, 24%241 Interaction volume, 283
on modifying Einstein" theory, 392 f nvariance transformations, 199, 209
on source theory and reviewers, 406 Invariant flux, 287
on speculation, 2% InvarianL momentum apace measure, 31
Heaviside step function, 75, 234 f rreducible processes, 283
Helicity, 20 Isotropic dilations, 224
eigenvectors of, for arbitrary spin, 136
for particles of inkger 4 spin, 132 Jacobi identity, 3
+
fsr spin partiehs, 129 Jordan, P,, 392
Klein-Nisfiina formula. DiBerential Lafiace" equation, Green's function of,
cross mction, for Compton 260
scatbring by electron a t rest and trttceless tensum, 93, 98
bgendtre" ppolynonnirtl, 93
hptonic charge, 125
Lsgrange function, See also Action hptoxls, 125,2-53
arbitrariness of, 188 Light : gravit &lion&deflection of, 83
far srbitrary-rank spinar, 195 graviktional sfawing of, 83
with broken confomal inva~ance,393 XIight phenomena, in scalar-hnsor theory,
conformally invariant, of nnwive 395
pin O partides and gravilatiana1 Linear momentum operabr, 8
field, 392 healised exeitslions, propagatian
of electromagnetic field, 229 characteristics of, 30
gauge invariancto of, 256 1;o~aEh r e n t ~transformations:
gravitational, 385, 399 requirement of invariance for, 398
of grsviton field, 381 response of spin field ta, 398-8W
of gravitons, response to gauge h a 1 Ninkowski coordinate system, 397
t ransformations, 381-382 h n g wave length gauge, and photon,
of interacting Dirac and Msxwell scattering by realistie H-pa&iefes,
fields, 261 358
for massless spin O particles, eonformal b r e n t z gauge, 281
transformation khabviar of, 391 Lorentzian shape of speckrat line, 871
of matter, arbitrariness in grsvitational Lorentz Lransfomations: composifion of,
field, 391 843, 10&101
modified, af spin O particles in linear respone to, 85
gravitational field, 391 represctntatisn by matrix similarity
of photons in gravitfttionai field, 389 transformations, 103
for second-rank spinor, 191 response ctf -spin nratriees to, 101
for spin 0, 187 Lorentz transformations, infinitelml:
of spin O particles, in gravitational ogertztor field theory construction for
fieid, 385 generators of, 24
with primitive electromagnetic redization of generators of, 19
interaction, 2565 response of electromagnetic field to, 27
with primitive gravitational response of single particle states to, 39
interaction, 383 responm af saurce to, 89
for spin *, 191 responrxt of stale h, 89
conformaXly invariant form, 402 responm of &tressdenwr to, 25
in gravitatianaI field, 401 responm of vector ta, 89,85
in weak gravitational field, 397 respoxlst: to, I43
for spin l , 188 and the space reffeetiun matrix, 1OO
for spin 1 particles with primitive and spin matrices, 86
interaction, 263
for spin g, 191.
for spin 2, 189 Mach" principle, 395
for spin g, X91 hfagnetie charge, 15, 231
for spin 3, 190 apparent srbitrariness in description,
for third-rank spinor, 194 241
Magnetic moment, See also Dipole of rank 2, 177
moment, magnetic relation to knsors, 195
relation to angular moments for pin 5, of rsnk 3, 178
261 of rsnk 4, 180
Magnetic quantum number of of rank 5, l82
multiparticle state, 54 symmetry and spin, 185
Mass, 11, 17 Muon, 34
Masshss particles: of arbitrary helicity,
141 Neutrinos, 125
of arbitrary integer helicity, 9G97 helieity of, 20
of fnelicity O and 1, 144 Neutron star, 403
of helicity *, 125 Newtonian interaction, 82
of helicity l , 72 Nonunitary transformations, 87
of heticity %, 129
of heXieity 2, 80 Operator field theory,iii, 24
of helicity 8, 175 consistency of, 28
of helicity 3, 99, 168 dynamics in, 33
infinite spin limit, 21 particles in, 30
+
af integer $ helieity, X33 speeulstion in, 34
nonlocality of, 20 and strong interactions, 34
nonlocalizlability principle, 22 Orbit, equation of, 85
position vector of, 20 Orthonormal spin-angle funelions, 115
Mass normali~atian,247 Ortfxonormal vector functions, 70
Ntaxwell" equations, 28, 228
with charged particle eument, 257 P, 50
with electric and magnetic currents, 237 Pair creation by a photon in a Coulomb
Mesons, 251 field, 343
magnetic model of, 253 Pair creation by two photons for spin O
Metric tensor, 16 particles, 286-297
Mode, 119 Parameter : Hubble, 396
M#Iler scsttering. See Scattering of of scalar-tensor theory, observations
spin g particles with equal charges concerning, 395
Momentum cells, 201 Parametric device for combining
Multiparticle states, 52, 68 denominators, 336
parztnletrized mixture of, 149, 155 Parity, 50
for particles of arbitrary spin, 13&137 for arbitrary spin, 141
of spit1 particles, 110, l18 intrinsic, 1 17
Multiphoton emission, soft photons, orbital, 117
270 spaee-reflection, 20, 95
Multipole moments: a
for spin particle angule~rmomentum
electric and magnetic, X17 states, 117
radiation by, 354 for spin % particles, 109
and spin, 280 Particle fiux vector, 208
Multispinor, 134 Particle occupation number, 52
relation to tensors, 142 average, 149, 155
Multispinor fields, 177 of Fermi-Dirac system, 163
of arbitrary rank, 184 in Fermi-Dirsc statistics, 101)
Particles, 1 Yermanexzt, and Dose-Einstein statistics,
with arbitrary integer spin, 85 55
sources for, 92 hrsistence probability of H-particles,
causal source description of s beam of, 368
41-42 Phase shift between helieity transitions
charged, two types of, 252: in Coulomb scatbring of spin g
composite, 36 particle, 327
statistics of, 252 Phase transformations, 48
creation of, 37 in causal situations, 200
critique of theories of, 24 and charge Auctuations, 207
detection of, 38 variable, 199
w i t h dual charges, 252 Photon, 34, '72
elementary, 11, 22 action far, 198
executing prescribed motions, 245 emission and &mrplion, 266
of integer -/- spin, X27 helicity of, 20
and general coordinate invariance, heXieity states of, 93
397 minimum detectable energy, 274
inberetcling, 12, 23 polarigation vectors of, 73
in a macroscopic environment, f 2 single-particle states of, 72
massless, 19 soft, 268
of arbitrary integer spin, 96-97 average number emitted, 273
of helieity g, 129 vacuum probability amplitude for, 12
af helicity 3, 99 Photon emission : in Coulomb scaltering,
+-
of integer $ helicity, 133 related t-a Cornptaiz scattering,
in operator field theory, 30 331
phexlomenolagical theory of, 37 in Coulomb scattering of spin O
reactions of, 32 psrticles, 329
real, 267 in Coulomb; scattering of spin 4
source of, 37 psrtieles, 341
space-time description of exchange of, Photon propagation function, 77
54-55, 118 Physical system, 28
of spin 0, 38 Paint charges, electric and mwnetic,
elzarged, fields associated with, 249
285 Poisson distribution, 65
of spin *, 99 Polarization : and double scattering, 328
of spin 1, 67 of spin particles in C?ouEomb
of spin 2, 78 scattering, 328
stable, 31 Polarisation vectors: arbitrsrine~sin,
unstable, 12, 32, 360-361 and gauge invariance, 294
virtual, 2437 for massless particIes, 98
Pauli matrices, 99 for particles of arbitrary integer spin,
Perihelion precession, 83 95
in. scalar-tensor theory, 395 for photons, 73
Periodic Green" function, 1149 rotation of, 76, 232
causal forms of, 150 far spin 1 particles, 68, 69, 91
Fourier series construction of, 151 for spin 2 particles, 79
modified, 155 Position vector operator, 10, 17
Primitive electromagnetic interactions, of spin O particles, 42-43
25-4 of spin partiefes, X06
physieal context of, 266 for angular momentum staks, 316
P ~ m i t i v einteraction, gravitiationd, 383 far arbitr~rymdes, 120
Principal value integrds in modifid invaiviag charge matrix, 123
propasgatio~function, 3 M generalisation of, 120
multip-~le"lrticIe
Probability :for e m c d e decrty of unstable timedependent, for spin 0, 346
H-particle, 372 far win *, 348-841)
for emitting prtrticbs, 65, 66 of unstable H-particlers, time behswisr
of H-particle deeay, 371 of, 388
for persisten~eof the vacuum, 43,52-53 Propagator. See Propaie;atia:onfunction
for ~ p c i f i cdecay of unstable P r o p r lifetime, 32
H-particle, 374 Proper orthoehranous Lorent~poup,
Probability amplitude :for radiative 16-17
decsy of unstable H-par$icla, 370
far two-pbobn deeay of unstable QuadrupoEe moment, ebetric, for spin f
H-particle, 371 particles, 264
Probability amplitudes, generating Quantu:n degree of frwdom, p h w spwe
funetion for, 282 f ranslation group, 5
Probe source, 55, 118 &uesntum ehectrodynamies, 34
Fro~eetionmatrix: for helieity, 141 Quantum mechanics, 1, 7
for helicity in relation to efrarge, 142 dynamical variables of, 33
for inGeger -f- spin, 131
Projmtion, operal;ors, 30 Rdiation gauge, 350
for angular momentum, l14 vector potential in, 350-351
Proieetion. tenssr :dy d i c eonstruetian of, Reactions, 33
94 twwpartiele, 35
for inkger q i n , 94, 130 3Reduced mass, 357
for massless particles, 97 Relativity :Einsteinial~,13
dyadic eonstruetion of, 98 Galileart, 7
relation to kgendre's polyn~mia1,94 &normdisstion, 32
for spin 2 particles, 79 Retarded fields, 147-348
for spin 3 particles, 170 of Dirsle equation, 162
lh.)ropwstionfunetion. See Green'8 Retarded Green's function, 1.48
funetian for Dirac equation, 162
asymptotic forms for, 46 btations: and angular momentum, 22%
causal structure of, 42-43, 57,51), 120 in causal: situstions, 223
complex conjugate of, 59 in charge space, 232
ditrerential equation for, 145 invarisnts of, 24%
fourdimmsional repremndation. of, 46 in Eueiidean charge space, 48
of H-partielea, spproximate of sources, 54
conadruction of, 36G36.7 Rutherford c r o s =etion, 28%
integral equation for, 366
nnsdified, 363 Satl;redo, 241
nonrelativistic form of, 375 Yalviati, 241
mdf iparLic1e generalisation of, 56, 65 Scalar eonfarmal field, 392
Scalar-bne;or mobification of Kingbin dynamics in, 343
theory. See Broken conformal q@cufstionsin, 36
invariance theory Soft photons, 268
Scsle transformations and eonformsl Sources, 37
group, 226 algebraic properties of, 135, 137
Scathring: elastic, of phohn by algebraic redefinitions of, 173
H-particb, 373 for arbitrary spin, response to TCP, 139
electron-photon, 318 charged particle, containing vector
multiple, lmf eallision snlsfysis of, 366 potential, 257
of photons by H-particles, 355 complex, 46
of photons by spin Q charged partiebs, for lspin particles, 159
29&291 complex conjugation of, 137
resonant, sf H-particle and time cycle contmt (overlap) h r n s , 144
description, 377 definition of, for masslew particles of
of spin O and spin 3 particles, 309 arbitrary helieity, 142
of spin 0 particles, 284 for massless particles; of helicity 8,
with equal charges, 281i-286 X 30
with opposide charges, 289 for ma~slessparticks of inkger 7t. 4
of spin particles, with eqml charges, heliciw, 134
300 for neutrinos, f 26,
with opposik charges, 306 for psrtieles of arbitrary i n e e r
Thornson, 34,3fM1), 360 spin, 96
and H-particles, 356 for particles of arbitrary spin, 136
Scattering c r o s seretions, 286 +
for particles of integer 4 spin, 133
Seatbring procesms: csusal analysis of, for spin 0 particles, 41
282 far spin O particles and antipecrtiebs,
irreducible, 283 47
S(catte~ng)matrix, 35 for spin particbs, f 08, l f 3
Selection rubs, 283 far spin 1 particles, 68
SeE4onsistency of muree coneept, 268 for spin 3 partiele~,129
Shell game, 50 for spin 2 particles, 79
Simple source, 267 directiondity of, 50
Simultaneity, 15 dispfacement of, 39
Single parlicfe state, 38 egective, woeiated with virtual
Singlet spin funetion, W H-particles, 365
Skebtsl interaction :incompleteness of, for photon and H-particle emimion
360 and absowtion, 361
for particlepartiele scatbring, a;79 for two-particle procesws, 266,
for particlephoton procems, 280 276275
for spin O particle-photon procews, electromagnetic model of, 255, 267
28 1 extended, 265
for apin -3; parliel*photon proceBm8, for ernision and absorption, refakd
280-281 by TCP, 2716
S matrix, 35 Fermi-Dirac, eampbx conjugation of,
S-mat rix theory, iii 110
constructive principle8 of, 35 gravitational m d e l of, 403
 Index

of gravitons, and energy-momentum, of unstable partiefes, use of reference

81 pain%in, $69
enechanicd measure of, 81 weak, efFect on multipartide states,
of H-particks, 346-347,349 65, 119
extended, 362 of spin O particles, 38
multiparticle exchange htween, 5656, Source restriction: for heIicity m
118, 361 parLieles, 126
non-eonserva3;Cion of current in, 202 ++
for helicitly .n mwsless partieie~,
non-consftrvation of stress tensor in, 133
21 l for helieity 1 nnsless pa;r.ticles, 72,2B
for partieles of arbitrsry inkger spin, 92 for hdicity 1 and 2 m ~ s l ~ psrticbs,
8s
+
far particles of integer 4 spin, 132 I68
photon, coupling of different kinds, 232 for heficity If mltsslftss pafticles, 129,
two t y p s of, 231 173
for photon helicity staks, 73 for helieity 2 massless particbs, 81.
of photons, and electric charge, 72 for helicity 4 mw1ess partieles, 176
reality of, 40 for helicity 3 mwless partieles, -99,f 68
relation between emission and for massless partiefes of arbitrary
abmwtian, 41 integer helieity, 97
restricted, 72, 81, 97, 9.9, 125, 129, 133, Souree theory, phenomenatogicsl
168, 173, 176, 233 orientation of, 2497
and general physieal laws, 2.43 phenomenologieakl viewpaint of, 365
rotation of, 54, 223 Space reflection, 50
as scatterers, 320 for arbitrary spin, 141
mlf+onsistency of concept of, 268 lineetr responm of Bource to, Bfi.
simple, 2437 matrix, for spin 4 particles, 100
spin O particle, containing grmitationd Space-time, 7
field, 403 Spsce-time refieetion as a Eucrideri~m
a
of spin particles, 99 transformation, 49
for angular momentum states, 116 Space-time uniformity, 37, 42, 43, 51
coupled by single-particlc! excfianw, and antiparticles, 47
1011, 115 and energy, 77
definition of for arbitrary modes, 120 momentum complement of, 269
response to TGP, 111.1, 117 and photon helicity, 75
of spin X particles, 67 Spctral distribution :of phohn emithet
spin g esnstructisn of, $30 by unstable H-particle, 371.
of spin # psrticfes, 127 of photon in transition btween.
eoupled by ~ingle-parti~lft exchange, unstable B-padjcles, 372-373
127 of urtstable H-partielt? nlxdiation and
of spin 2 partieles, 78 time cych description, 374
pino or, of spin O and @in f particles, Speculations: in operator field theory,
142 34
of spin 3 particles, 106 in S-matrix theory, 36
strong, of q i n Q p~rticles,50 Spherical harmonies: ddition theorem
tensor, 92 of, 94
and transition prabtbbiliLies, 282 generating function of, 94
of unstabb L3[-particlt3s, 369. Spin, 10
Spin and statistics, 134 scattering of apposite chargw, 3W
unified proof of connection btovwn, 138 source of, 99
Spin matrices: commutstion relations of, space refleetion matrix for, 100
86 symmetry properties of source
for spin 4 particles, 99 coupling, 102
symmetry and reality of, 86 Spin 1 particles, fi7
Spinor, 106 definition of source, li8
antisymmetrical, 136 fields af, X64
Spin-parity, 141 helicity states for, 69
Spin 0 particles, 38 polarization vectors for, 68, 6Q
behavior of sources under Lorentz seeand-rank spinor des~riptionof,
transformations, 40 177
charged, description by real sources, source of, 67
284 Spin 8 particles, 127
deffnition of source, 41 definition of murce, 126)
fields of, 145 fields of, 172
primitive electromagnetic interaction helieity eigenvectors for, 129
for, 256 source of, 227
propagation funetian of, 42-43 Spin 2 particles, 78
scattering, 284 definition of source, '79
of like chargesj 285-.286 fields of, L65
of opposite cbsrges, 289 heftieity states for, 78
of photons by, 290-291 polarization vectors for, 79
strsng sources for, 50 source of, 78
Spin matrices: as, 127 Spin particles, fields of, 176
a",U .
, = 0, , . ,3, 101 Spin 3 particles, fields of, 16'2
a,,@ = 1 ,..., 4,112 State vectors, 1
eigenvectors of: u, 107 symmetry of, 33
eigenvectors of: v, 106 Stationary tbction, principle of, 187
T 5 , 104 Statistics: Bose-Einstein, 33, 55
Y", X03 of composite particles, 2-52
F,, 103 Fermi-Dirae, 33, 99
rat, 133 and win, 134
a,,, 104 Step function, Heaviside, 234
@&*E, 112 Stimulated emission, 56
trace of, 107, 143 Sdone, Max, 83
Spin particles, 99 Stress scalar: and confomal
and charge, 113, 123 transformations, 226
definition of source, l OS of electromagnetic field, 230
fields of, 157 of photons, and conformal invcarlanee,
parity of angular momentum states, 390
117 for q i n 0 particles, 209, 216
parity of single-particle states, 109 far spin particles, 220
propagation functions for, 106,136, f 20, far spin 1 psrticle~,213
122, 162, X63 for spin 2 particles, 219
response of sources ta TCP, X 14, 117 Stress tensor, 24, 210. See alga Energy-
scattering of like ch&rges,300 momentum Aux vector
 arbitrariness of, 217, 390 @p,, 387
and angular momentum, 1224 diftFerential identity for, 388
for spin O particles, 2x6 B,,k, 390
~rbitrarinesswithin sowees, 216 R,,, 388
for arbitrary multispinor, 222 a,&, 399
wwciated with particle state, 210, 221 R & ~ 386
*A~
commutation relations with hrenrttl ~trese,t,,, 388
generators, 25 symmetrical: t,,, 3E)8
conserved, of spin Q field, 404 t;, 398
of ebctronnagnetic field, 26, 230 tracelessr, 83
equal time eommutaficrn rel%tionsfor, Tensor-spinor, 132
25 Tetrad vector field. See also Gn-tvitatio~tt~X
generalized conservation law in field, et:
gravitational field, 388 conformal behavior of, 402
identity of kinemzttieal m d dynamical Thornson erorss wetian, %4, 320, 356
definitions, 383 Tints cycle description, 62
implied by gravitont source, 405 fieids of, 147, 161
of massive particles, 245 for parsmetrised multiparticle stabs,
of photons, in gravitatianal field, S 9 1511, 1516
mssponse of to infinitesimal h r e n f z and photon spectrum of unrptabb
transformations, 25 ~ - ~ a r t i d374
e,
and scalar, for spin 0 particles in. and resonant seatwring, 8 7
gravitational field, 388 Time eyele vacuum amplitude, 62,154,
for spin -$ pmtiele in graviteLtiona1 161
field, 401-402 for charged particles, 156
far seeand-rank spinor, 221 and expectation values, 6.8
for spin 0 particles, 212 generdisstion to multigar&ie!e sgate, 165
for spin particles, m for spin 4 particles, 161
using third-r~nkmultispinor, 222 Time refleetion, 49
for spin. 1 particles, 213 Time scale and unstable particbs, 381
for spin 2 particles, 218 T matrix element: for Coulomb
symmetry of, 214 seal"dring, of spin O psrticb, 821
S%ronginteractions and operatar field of spin 4 particle, 322
theory, 34 for electron-positron mnhilation into
Sum rules, 32, 356 two photons, 31 l
m pherrornenological consiskncy for photon absorption by H-parfiele,
eonditione, 357 353
Super rsource, 282 for photon ennisgisn in Gulamb
Swfaee, directpea element of, 247 scattering of spin 0 partieb, 329
for photon scattering, by H-particies,
I", f;O 355
Tchebichef % polynomial, 98 by spin 3 pa&icle, 340
T W ,50,54, $9, 113, 117,126,129,276, for radiative H-particfe transitian, M2,
2w 3M
defined for arbitrary spin, 139 for scsttering af photons by pin Q
Tertmr: eo~ltravariant,385 charged partielw, B2
eovariant, 385 e r o ~ i n gsymmelries of, 292
 for scattering of spin Q and spin for H-pnrtielcs, 347
particles, 369 for spilt 2 particles, 78
far spin 0 particle pair creation by two Traeuum, probabilrity =trnplitude : causal
photons, 297 analysis of, 51-52, 55, 109, l f 9
for spin O scattering, of like charges, complex conjugate of, 59
286 decomposition in zero mass kinnit, 80
of opposite charges, 289-290 with differetlt kinds of particles, 256
for spin 3 scattering, of like charges, Eucfideai~tra~lscriptionof, 45, 60, l 13
306 for gravitons, 80
of opposite charges, 306 for integer spin in EucXidean space, 140
+-
Total a~rgularmomentum: classific%tio~~ for i n t ~ g e r 4 s p i l l in Euclidean space,
of sources and states, 70, t 14 140
quantum numbers, 71 for xnasslcss particles, of helicity 8, 129
Total symmetry of zt multispinor, 134 of kelieity 3, 99
Trace, 107 of integer+ helicity, 133
T(ransition) matrix, 35 multiparticle gelleralisatiot~of, 56, 58,
definitioll of, 283, 321 1X9
physical interpretatiol~of, 284, 321 far neutrinos, Euclidettn tra~~scriptisn
Transition probabilities in source of, 126
description, 282 for particles of arbitrary integer spin,
Transition probability amplitudes, 56 95
for spin 3 particfes, I21 for particles of arbitrary spin, X35
Transition probability per unit time for for particles of integer --/- -& spin, 132
H-particle radiative transition, 352 for photons, 72
Triplet spin functions, 90 in refsttion to energy, 77
Two-photon ttnllihiliation : of high-energy far spin O partieIes, 42, 47, 51
electron and positron, 323 for spill $ particles, 106
of slow electron and positron, 312 Euclidercn trtlnscriptio~~ of, 113
for spin 1 particles, 67
Unitarity, 35, 59 for spin 8 psrlicles, X28
and causality, 61, 122 for spin 2 particles, 79-80
Unitary transformations, I for two kinds of photon sources, 235
group of, 3 Vacuum state, 16, 29, 38
invariance of, 41
infinitesimal, 2
sueeesslve, 2 Vector, response of to Lorentz
Universe :age of, 396-397 transformations, 69
Vector potential: electromagnetic, 227
geometricai description of, 396
mass and radius related by Mach's two kinds of, 236
exhibited in terms of field strengths, 238
principle, 896
Vector-spinor, 127
mass of, 395
matter density in, 396-397 Virtual particle, 267
model of, in scalar-tensor theory, 396
Wavefunction for spilt 3 particIes, 159
radius of, 395
Weir;s&cker-Wi11iamsmethad. See
Photon emission in Coulomb
Vacuum persistence probability, 43, scattering, related to Compton
52-53, 68 scattering