Feynman, Richard - Photon-Hadron Interactions (Westview, 1998)

PHOTON" HADRON
INTERACTIONS
PttOTON--HADRON
INTERACTIONS
RICHARD P. FEYNMAN
late, California Institute of Technology
�e�'o/.i�
.. . � ...... · ·
AdvAm:ed Book Program

.
\
\
·- -.._.., A Member of the Perseus Books Group

Library of Congress Cataloging-in-Publication Data
Feynman, Richard Phillips.

Photon-hadron interactions/ Richard Feynman.
p. cm. - (Advanced book classics series)
Originally published : Reading, Mass. : W. A. Benjamin, Advanced
Book Program, 197 2. (Frontiers in physics)
Includes index.
l. Photon-hadron interactions. 2. Hadron interactions.
l. Title. 11. Series.
QC793.5.P428F49 1989 539. 75
' 4-dcl 9 89-30697
ISBN 0 -201-360 74-8 (paperback)
Copyright© 1972, 1998
All rights reserved. No part of this publication may be reproduced, stored in a

retrieval system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior written permission of the
publisher. Printed in the United States of America.
Wcstvicw Press is a Member of the Perseus Books Group.
Cover design by Suzanne Heiser
2 3 456 78 910
First printing, February 1998
Find us on the World Wide Web at

http://www.westviewpress.com
Editor's Foreword
Addison-Wesley'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
monograph. Indeed, throughout its nearly forty-year existence, the series has
emphasized informality in both style and content, as well as pedagogical clari
ty. Over time, it was expected that these informal accounts would be replaced
by more formal counterparts-textbooks or monographs-as the cutting-edge
topics they treated gradually became integrated into the body of physics knowl
edge and reader interest dwindled. However, this has not proven to be the case
for a number 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
community 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 notes on Richard Feynman's lectures at Caltech on the topic of pho
ton-hadron interactions, in which he developed his theory of partons, were first
published some twenty-five years ago. As is the case with all of Feynman's lec
tures, the presentation in this work reflects his deep physical insight, the fresh
ness and originality of his approach to understanding physics, and his overall
pedagogical wizardry. As a result, this volume will always be of fundamental
v
vi Editor's Foreword
importance to anyone interested in understanding the development of quantum

chromodynamics (QCD)-the theory of quarks and gluons-which explains
hadron-hadron interactions at high energies.
David Pines
Urbana, Illinois
December 1997
Special Preface
Many of us were first introduced to the concepts of the parton model from this book. At
that time the competing view was one in which there were no elementary particles. Every
particle was supposed to be a composite of every other particle. The ideas and concepts
in this book have helped pave the way for our understanding of the constituent nature of
hadrons which eventually led to the Quantum Chromodynamic (QCD) theory of quarks
and gluons. As is true of most of Feynman's books, the maximum benefit is obtained if
one has previously studied the subject in some detail. Feynman's unique perspective can
best be appreciated by readers with a solid background in the subject.
Although this book is almost 18 years old, it still is an excellent reference. It
appears on the recommended reading list of all the current QCD books. The boo k provides
a good understanding of the parton model frt>m the man who invented it. In the "pre-QCD"
or "naive" parton model the constituents within hadrons were assumed to be bounded in
the transverse direction. The probability of finding a parton within a high momentwn
hadron with a large transverse momentum was assumed to fall like a Gaussian or an
exponential. QCD tells us that this is not exactly true and gives a power law fall-off in the
transverse momentum. Because of this, many "naive" parton model expectations are
modified (in an important way) by logarithmic factors. Feynman used to laugh when his
parton model was referred to as "naive," and he would say, "At least I got it right up to
logarithms." We all miss Feynman very much and it is through books like this that his
ideas live on.
R.D. Field
March, 1989
vii
Vita
Richard P. Feynman
Born in 1918inBrooklyn,Richard P.Feynmanreceivedhis Ph.D. from Princeton in 1942.
Despite his youth, he played an important part in the Manhattan Project at Los Alamos
during World War II. Subsequently, he taught at Cornell and at the California Institute
of Technology. In 1965 he received the Nobel Prize in Physics, along with Sin-Itero
Tomanaga and Julian Schwinger, for his work in quantum electrodynamics.
Dr. Feynman won his Nobel Prize for successfully resolving problems with the
theory of quantum electrodynamics. He also created a mathematical theory which
accounts for the phenomenon of superfluidity in liquid helium. Thereafter, with Murray
Gell-Mann, he did fundamental work in the area of weak interactions such as beta decay.
In later years Feynman played a key role in the development of quark theory by putting
foiward his parton model of high energy proton collision processes.
Beyond these achievements, Dr. Feynman introduced basic new computational
techniques and notations into physics, above all, the ubiquitous Feynman diagrams which,
perhaps more than any other formalism in recent scientific history, have changed the way
in which basic physical processes are conceptualized and calculated.
Feynman was a remarkably effective educator. Of all his numerous awards, he
was especially proud of the Oersted Medal for Teaching which he won in 1972. The
Feynman Lectures on Physics, originally published in 1963, were described by a reviewer
in Scientific American as "tough, but nourishing and full of flavor. After 25 years it is the
guide for teachers and for the best of beginning students." In order to increase the
understanding of physics among the lay public, Dr. Feynman wrote The Character of
Physical Law and Q.ED.: The Strange Theory ofLight and Matter. He also authored a
number of advanced publications that have become classic references and textbooks for
researchers and students.
Richard Feynman died on February 15, 1988.
viii
Contents
Editor's Foreword v
Special Preface vii
Vita viii
Preface xiv
1-5 GENERAL THEORETICAL BACKGROUND 3
First Order Coupling 5

Conservation of Current 6
2nd Order Coupling 7
Unitarity 2nd Order 8
Proof 10
End of Proof 12
Research Problem 13
Conservation of Current 14
Remark 15
Isotopic Spin, Strangeness, Generalized Currents 16
Conservation of Generalized Currents 18
Singularities on the Light Cone 21
Vacuum Expectation of Vµv (1, 2) 22
e• + e· .. Any Hadrons 23
Note: Annoying Point 28
Commutator 29
Problems 30
6-8 LOW ENERGY PHOTON REACTIONS 31
Pion Photoproduction Low Energy (0 to 2GeV) 31

Problem 35
ix
Note 38
8-10 QUARK MODEL OF RESONANCES 42
The Quark Model 42

Note 44
Problem 49
Calculation of Matrix Elements so
Feynman, Kislinger and Ravndal, Phys. Rev. (1971) SI
11-12 PSEUDOSCALAR MESON

PHOTOPRODUCTION, HIGH ENERGY 60
Pseudoscalar Meson Photoproduction-Higher Energies 60

References 60
13-14 t-CHANNEL EXCHANGE PHENOMENA 71
t-Channel Exchange Phenomena 71

Comments 7S
s-Channel Resonances 76
Veneziano Fonnula 78
Estimates of Coupling Constants 79
14-21 VECTOR MESONS AND VECTOR MESON

DOMINANCE HYPOTHESIS 82
Properties of Vector Mesons 82

Electron Production of Vector Mesons 82
Vector Meson Dominance Model 89
0 as ss 94
VDM and Photon Hadron Interactions 96
Diffractive Production of P , w , • 102
Other Tests of VDM 108
Shadowing in Nuclei 109
To Summarize the Position of VDM 113
22-24 ELECTROMAGNETIC FORM FACTORS 114
Electromagnetic Form Factors 114

Nucleon 114
Contents xi
In Lab 115
Electromagnetic Form Factors (continued) 116
Pion Fonn Factor 118
Proton Fonn Factor for Positive q2 121
Note 122
25-26 ELECTRON-PROTON SCATIERING.

DEEP INELASTIC REGION 125
Other Photon Processes for q2< 0 125

Inelastic Electron Nucleon Scattering 126
Inelastic Electron Nucleon Scattering (continued) 128
Theory of the Inelastic Electron Proton Scattering 130
26-33 PARTON MODEL 132
Parton Model 132

Parton Model (continued) 134
The Wee Region 136
The Region Near x = 1 140
The Region -q2 Large M2 Finite Resonances
x
143
Argument that Y ' = Y 145
General Remark about the Power Law (q2) -y 147
Partons as Quarks 147
Momentum Carried by the Quarks 151
Models 152
Future Tests of Charged Partons = Quarks 153
Deep Inelastic Scattering with Spin 155
34-35 TESTS OF THE PARTON MODEL 160
Angular Momentum in Parton Wave Functions 160

Other Experiments Testing Parton Idea (Drell) 160
p + P ... /11 - + Anything (continued) 162
Electron Pair Production of Hadrons · 163
36-37 INELASTIC SCATTERING AS

PROPERTIES OF OPERATORS 167
Inelastic e p Scattering as Properties of Operators 167

Properties of Operators (continued) 171
xii Contents
38 LIGHT CONE ALGEBRA 178
Light Cone Algebra 178
39-41 PROPERTIES OF COMMUTATORS IN

MOMENTUM SPACE 182
Propenies of Commutators in Momentum Space 182

Region I 185
Bose or Fermi Quarks 187
Region II 188
Region III 188
Scattering in the Deser, Gilbert, Sudarshan Representation 190
42-47 ELECTROMAGNETIC SELF ENERGY 199
Electromagnetic Self Energy 199

Cottingham Formula 202
Expression for Self Energy in Terms of W Only 205
Other Electromagnetic Energies, Quark Model 206
Electromagnetic Self Mass, Quark Model (continued) 208
I = 2 Mass Differences 211
Further Comments on Electromagnetic Mass Differences 212
Compton Effect y p ... yp or y n ... y n 215
Compton Effect for Very Small Q, v 217
Forward Compton Scattering from Non-Relativistic
Schroedinger Equation 219
48-49 OTHER TWO-CURRENT EFFECTS 221
Other Quantities Involving T 11v 221

Other Two-Current Effects 225
50-51 HYPOTHESES IN THE PARTON MODEL 229
Hypotheses in the Parton Model 229

General Framework 230
Hypotheses in the Parton Model (continued) 232
xiii
Contents
52-54 HADRON-HADRON COLLISIONS AT

EXTREME ENERGIES 237
Hadron-Hadron Collisions at Extreme Energies 237

Hadron-Hadron Collisions at Extreme Energies (continued) 241
High Energy Hadron-Hadron Collisions (continued) 245
55 FINAL HADRONIC STATES IN DEEP

INELASTIC SCATIERING 250
Interaction of Partons with the Electtomagnetic Field 250

Region of Finite q2, v + • 256
Continuity of Large q2 and Small x Region 258
56-57 PARTONS AS QUARKS 259

Partons as Quarks (continued) 263
Product Predictions 264
APPENDIX A 271
The Isospin of Quark Fragmentation Products 271
APPENDIX B 276
A Test of Partons as Quarles 276
INDEX 279
Preface
The most advanced course in graduate theoretical physics at Caltech is "Special Topics
in Theoretical Physics." Each year the professor chooses the topic with which he will deal.
This year (1971-72), having just come back from the 1971 International Symposium on
Electron and Photon Interactions at High Energies, held at Cornell University. my own
interest in the subject was aroused, and I chose to analyze the various theoretical questions
related to that conference. The lectures themselves became so extensive that the decision
was made to put them into book form, with the thought that other people might also be
interested. Thus, the report of the Cornell conference should be considered as a
companion volume to these lecture notes. The references given here are far from
complete, but a full list of references is given in the Proceedings of the Symposium,
published by the Laboratory of Nuclear Studies, Cornell University, January 1972.
The material is dealt with on an advanced level; for instance, knowledge of the
theory of hadron-hadron interactions is assumed. I have tried to analyze in detail where
we stand theoretically today. The treatment is somewhat uneven; for example, I should
have liked to study the theory of the decay of the in more detail than I was able to do.
On the other hand, there are long discussions of vector meson dominance and of deep
inelastic scattering. The possible consequences of the parton model are fully discussed.
Time did not permit me to complete the original plan which was to include the
theory of weak interaction currents which are so closely related to electromagnetic
currents.
Many thanks must go to Arturo Cisneros who edited, corrected, and extended the
lectures from my class notes. Without his effort, this book would not have been possible.
I also wish to thank Mrs. Helen Tuck for typing the lecture notes.
Richard P. Feynman
Pasadena, California
Summer 1972
xiv
Photon-Hadron
Interactions
General Theoretical Background
Lecture 1
One very powerful way of experimentally investigating the strongly

interacting particles (hadrons) is to look at them, to probe them with a
known particle; in particular the photon (no other is known as well). This
permits a much finer control of variables , and probably decreases the

theoretical complexity of the interactions. For example in an ordinary
hadron-hadron collision like wp + wp we are hitting two unknowns together,
and further, we can only vary the energy, we cannot vary the q2 of the pion
which must be mw 2• In fact a "pion far off its mass shell" may be a
meaningless - or at least highly complicated idea, On the other hand in
y + p � p + w we know the y is single and definite, and we can vary the q2
of the y by using virtual y's via, for example, electron scattering
e+ p + e + p + w.
We are assuming that we do know the photon. QED has been checked so
closely that we know that if the photon propagator were off by a factor of
the form ( 1 - q2 /A2)-l then A exceeds 4 or 5 GeV. The amplitudes are known
to about for q2 as high as (1 GeV)2• For the rest of this course we
5%
shall assume QED is exact. There is already evidence, as we shall see, that
3
4 Photo11-Hadro11 lntc'ractions
in virtual photon-hadron collisions the photon acts normally (i.e . , obeys

QED expectations) up to II of & to 8 GeV .
At any rate we shall suppose QED exact - where we mean by QED the standard
interaction theory for electrons , muons and photons . Exact , but incomplete -
for hadrons are charged and interact also with the QED system . We discuss
first how we shall assume we can describe this interaction .
S ince e2 is small it is natural to describe their interaction in a series
of orders in e. One photon exchange , two pho ton exchange , etc . ( I t might
be thought that to describe this coupling we shall have to have some detailed
dynamical theory of the hadrons - ultimately , of course , yes - but some
things can be said in general restricting the matrix elements whatever the
underlying hadron dynamics - and it is these restrictions we seek in this
lecture . )
The no coupling case presents no problem . The factor giving the amplitude
that a hadron system goes from an incoming state I n , in> , to an outgoing
state <m , out ! is:
smn - <m , out I n, in> (1 . 1)
The S matrix is the transformation matrix from the "in" representation to the
"out" representation
E S <n , in l = <m, out ! (1 . 2 )

n mn
The state I n , in> means a state which far i n the past i s asymp totically
free stab le hadrons (stab le in strong interactions only , e . g . 11° is "stab le")
described by momenta, and helicities all contained via the index n . The state
m, out has the set of indices m with the same space of indices , but represents
a state which in the future is asymptotically in si tuation m.
Thus the S matrix is really the uni t matrix but in a mixed represen t at i on,
a different labeling for incoming and outgoing states . Supposing these states
are all there are , conservation of probability requires
(i.e. , ( 1 . 3)
(In the special case that the state n represents a single stab le particle the
in and out states are the same).
General Theoretical Background
First Order Coupling.

The general coupling of electrons and hadrons is represented by the
m
diagram: M
N n
The electron-photon system goes from state N to M, the hadrons from n in to
m out . We suppose the only interaction possible is by the exchange of a
photon - and this pho ton is characterized by a polarizationµ , momentum q:
<M/j µ {q) / N> �2� <m , out ! Jµ {q) l n , in> .

2
4ne i
amp = (1.4)
q
That is to say (supposing we could measure the amplitude) we define in a given

experiment the quantity
'1 (q) mn = <m , out l Jµ {q) / n , in> (1 . 5)

-::f µ
This is done by removing from the measured amplitude the known (by QED theory)
factors
I t is then our first supposition that this quantity d1 µ (q) mn depends

only on the states m , n of the hadron system and only the virtual momentum
q and polarization µ of the virtual photon . �
,l
µ depends in no way on how
the pho ton was made (eg . whether by µ's or electrons or on the angles and
energies of the electron for fixed q and pho ton polarization) .
This is a strong assumption . I t has been verified most completely for

the case of proton form factor measurements, but is o ften assumed in checking
equipment , comparing results from one lab to another etc . We assume it .
We emphasize then that J µ (q) is an experimentally defined quantity -
definab le in principle for all q .
We find i t convenient to define a new matrix Jµ (q) kn defined in a
non-mixed representation as
(1 .6)
so we now write
6 Photon-Hadron Interactions
(1. 7)
4ne- 2
To deal with this a li t tle more abs tractly , the quantity j • - can , in
u q 2
QED be described as the matrix element (between the leptons ) o f an operator ,
the vector potential operator for the leptons a (q) , all cons idered known
ll
and compatible by QED in any specific case, Thus the firs t order interaction
is described by the matrix
1 + i f a (q) J (q) d\ + higher order .

µ II
The 1 comes from the zero order (where we saw the S matrix is really the
unit matrix in an unmixed representation) .

*
Unitarity requires to first order (a+•a (-q) transpose)
f
I au (q) Ju (q) d q +)
4 4
(1 - i a+ (q) J+ (q) d q +·� (1 + i - 1 (1.8)
u u
or , s ince a (q) is arbitrary, J •J .

ll u
: J
ll
is hermitian . Since all q are
available we can define Fourier Transforms
(and for a /space ) .

u
Thus the coupling is
Lecture 2
Conservation of Current .
You may , if you wish simply assume V J a

O, conservation of current for
ll u
hadrons , or else note the following discussion .
It is not , s trictly , true that J from E q (1.5) can be comple tely

.
u
obtained from experiment . That is because a (q) is not completely arbitrary.
ll
When it comes from the usual diagram rules it always satis fies q a (q) •O.
u ll
ll II
Thus one component of a ( the one in direction q ) is always missing (unless
General Theoretical Background 7
q2 • 0) and thus one component of J , the component in the direction qll is

ll
mis s ing . We f inish the definition of J by choos ing qllJ\J • O.
ll
2
This we do in the following way . Firs t , for q • 0 , a free proton of
polarizat ion e the coup ling is e / (q) - but for a free photon e is undefined -
ll ll ll
to it could be added aq\J (e � = e
ll
+ aqll) any a. This can make no e f fe ct s o
our cons istency with Q E D demands q\JJ ll • 0 at leas t when q2 = 0, This is a
physical property Jll must satisfy . If it is not true for general q2 redefine
to replace the old via J ' • J

a new J\J' ll ll - qll (qvJv )/q2• Evident ly
q\J J \J O, and no new pole at q • 0 is introduced by the new l/q

2 2
' = term for
its numerator qvJv vanishes at q

2
= O,
What of other restrictions on J ( l ) ? These we wish to find . For example

ll
G �
s ince it is a local operator field theory sugges ts that i f hadrons are governed
( l ) , J ,,( 2 = 0 if 1
�
by an underlying field theory then and 2 are space
\J
like separated (symboli zed by l • 2 ). I f you wish you may assume this -
but it is very in teresting that we can prove it from our as s umption that the
s t rong system interacts with QED (subject to sys tematic errors in proof due
to taci t as sumptions . This may not be an important poin t but it is interes t ing
so I will waste your time by proving i t ) .
2nd Order Coupling .
This is via diagrams of the type whose ampli tude depends on a computable
factor from the leptons* times a ma trix element depending on the two momen ta
and polari zat ions of the two virtual pho tons -1 / 2 Vil" (q1, q ). As defined
2
q1 ++ v,
<-+
this i s symm e trical in q , ll for Bose s tatistics among pho tons
2
does not permit us to distinguish the photons so no other function can b e
experimen tally defined . Us ing an unmixed representation , and coordinate
space (via doub le Fourier trans form) we can represent th is amplitude as
( 2 .1)
\J ll+
*
in part I of th is course "lep tons " will mean e , e
+, , only .
8 Photo11-Hadro11 l11ternctii111s
The contribution from the QED lepton side can be figured for each diagram ,
and can , as shown in QED theory , be generally wri t ten as the matrix e lement
(be tween in and out photon-lept on states) of the time ordered product of the
operator a ( 1 ) and a ( 2 ) (symbolized by

µ µ
{ } T> . S ince arb i t rary a ( l ) a (2) can
be made V is experimen tally defined .

µv
I wish now to prove a number of things , so it will be mo re convenient
to restrict the integral to t > t and write this and the first two orde rs as
1 2
T • 1 - i J J (l) a / 1 ) dt -
µ 1
t
J>� v v
µ
(1, 2) a (l ) a) 2 ) dt dt
µ 1 2
(2 . 2 )
1 2
Evident ly one can write an entire series of functions to ever increasing
order .
Uni tari ty 2nd Order .

+
This gives the restrict ion (using a • a)
- f> J.
t
1
t
2
v \JV \I
( 1 , 2) a ( 1 ) a ( 2 ) dt
V 1
dt
2
• o (2 .3)
+
ob tained by writing T T • 1 and expanding to 2 n d order . Th e 2nd integral
is over all t , t •
1 2
It can be spl i t into a part t
1
> t
2
and a part t
2
< t
1
in the lat ter relab le variab les 1 , 2 (and µ , v ) to get
Now , i f we could assume that wi thin the range of possib le QED states arb i trary
values for a ( l ) av ( 2 ) can be generated (whi ch is t rue) and also independently

\J
for av ( 2 ) a ( l ) (wh i ch is false) we could conclude that
\J
v ( 1 , 2) - J ( l ) J v ( 2 ) for t l > t
2
(2 . 5 )
µv \J
But outside the light cone for example � (2) , a(l � • 0 so the two products
are not independent , they are equal . To proceed therefore more carefully
write
av(2) a (1) •a (1) aV (2)

11 11
+ r:
[v(2), a lJ (1)1
J
to get
Now the first factor must vanish, for we could take the case that a(2), a(l)
coDDDUte first (for example one photon from an electron, another from a muon in
lowest order) hence we surely always must have
J II (1) Jv(2) + Jv(2) J,,(l) - v IJV(1 2) + v+IJV (1, 2)

• '
(2. 7)
determining the real part of v (l, 2). In addition in general we must have
11v
(taking adjoint of last term)
The commutator is zero outside the light cone. Inside I believe we can make
it arbitrary (although some little further study of special cases is necessary
to verify this) hence we deduce
if (2 .9)
1 is in forward light cone of 2.

We have almost proved Eq. (2.5) but not for any t1 > t , only for t1
2
inside the light cone of 2. The difference is very important because
Eq. to be relativistically invariant requires �J (1), Jv(2)� 0
(2 . 5 ) , =
11
outside the light cone. Eq. (2.5) also is natural if hadrons come from any
underlying field theory, for then our picture of coupling if t1 > t2 can
Leptons
be cut at a t between t and t1; the first coupling is J11(1) and the second
2
is J {2}, so we get the product . But one may be averse to assuming that
v
s trong interactions can be described by a complete set of states (and that the
complete set can be taken as In, in> } at any arbitrary time . Nevertheless
if we continue our s tudy of the raquirements for consis tency with QED to
4 th order we can do it .
'nlis point is not important , but we do include the proof for completeness .
If ( 2 . 5 } were right then T would be {exp - i JJ11 (1} a\J (l} dt1 } T . In general
it is some polynomial in a/1}, {a (2) a

ll
) l >} T , {all (1) a) 2 } a
0
C3>} T etc .
(From now on we omit the polarization indices - they always go in an obvious
way with the position indices}. 'nle first order is made to agree with J (l}
\J
so in general we can write (U(l2 } represents the deviation of V ( 12 ) from
J(l} J(2)} for t > t }

1 2
JJ
J �t 2 U ( l , 2 ) a (l} a (2 )
-i (1) a (1) d t 1 [';
T • e 11 [ + i
�
t
l
+ J t f t JU (l 2 3) a (l} a (2 } a ( 3 }
�
t
1 z � 3
+ f t J J t JU(l 2 3 4) a (l) a (2 } a (3) a (4 ) + " • � (2 . 10}

t >, � t >,
l 2 3 4
+
Now in forming T T to check unitarity the exp -i /J a factors go out. To 2nd
and 4th order in a the U (1 2 3) term does not enter. To second order we have
- � +
u (l 2) a(2)•a(l} + J uc1 2) a(l} a(2} • o
so we conclude U is hermitian u+ • U everywhere , and U (l 2) • 0 if
as before .
Now in 4th order we have
f�+> (lt 2 3 4) a(4} a(3} a(2} a(l} + {t

u c 1 2 3 4) a ( l} a (2 } a ( 3} a ( 4 }
> t > t > t
t > t3 > t4 l 2 3 4
l 2
+ �U ( 5 6)
t > t
U (7 8) a (5 } a (6 } a (7} a (S) • 0 (2 .11}
s 6
t > ts
7
In the last integral we have t5 > t6 and t7 > t8 but no particular relation
of t5 and t7 there are 6 relative orders. We replace in each the variables
via t1, t2, t3, t in that order
4
1 5 5 5 7 7 7 call a(l) •A
2 6 7 7 5 5 8 a(2) • B
3 7 6 8 6 8 5 a(J) •c
4 8 8 6 8 6 6 a(4) • D.
The last term gives
U(l2) U(34) a(l) a(2) a(J) a(4) ABCD ABCD

=
U(l3) U(24) a(l) a(3) a(2) a(4) ACBD A�, �D +ABCD

•
U(l4) U(23) a(l) a(4) a(2) a(3) ADBC. A[p, �c +AB[!>, c] +ABCD
U(23) U(l4) a(2) a(3) a(l) a(4) BCAD s[C, .go + (!, �CD
• +ABCD
U(24) U(l3) a(2) a(4) a(l) a(3) BDAC. s[!>, .gc + �. �DC + AB[n, cJ +ABCD
U(34) U(l2) a(3) a(4) a(l) a(2) CDAB. c[n, A]B + [E. �DB +Ac(E, s] +
+A@, �D +ABCD (2 .12)
and the first term is
u+(l 2 3 4) DCBA. [p, s"]BA +c[p, �A +CB@, A] + @, �AD + sf1=, �D +

+ (!!, �CD +ABCD (2.13)
Now we get many obvious relations. For example the coefficient of ABCD must
be zero (take case all 4 potentials commute). We believe that the vector
potential a is an arbitrary function of space and time. We can therefore
choose it to be different from zero only in four small regions of space time
around the points 1, 2,3 and 4; call these regions o1, o2, o3, o4• For
our special interest here take the case that the variables have the following
light cone properties
[A, c] • o • [A, s]
[s, ] n • o • (n, cJ
Only [}J, c] and [A, � </- O
1 is outside the light cone of 2 , 3 is outside the light cone of 4 . Omi tting
the ABCD term which we noted was independently zero and collecting what is
left in this case we get
+ U(l3) U(24) @, �AD

+ u (24) u (l3) [!>, '9sc
+ U (34) u c 12 >{1P . �CB+@. !JAD} ) dT l dT
2
dT 3 dT
4 (2 . 14)
Only the third term needs to be turned around to [!>, ajcB+ [!>, £1 [!, SJ then
all the tel"lll8 are coefficients of @, !JAD or CB(!?, A] and must all vanish.
But ultimately we are left with U (24) U(l3) [!>, A'] Cs, � or finally
U (24) U (l3) G � �
<2> , a ( 3 . (4) , a ( l � dT l dT2 dT dT4. 0
3
(2.15)
Since the commutators are sufficiently general , I think we can conclude the
integral will be zero only if the integrand is and U(l3) • 0 even if 1 is
outside the light cone of 3 .
End of Proof
We can therefore conclude
(2 . 16)
In the proofs of (2 . 9 ) and (2,16) we have assumed that a(i 1 , t 1 ) and
� Cx1 , ]
t 1 ) , a (i 2 , t 2 > are sufficiently general functions of 1 and 2 (when
2 is in the light cone of 1 ) . We believe this to be true . It is left to
those interes ted in more rigorous proofs to verify this , for example by
trying to cons truct a basis .
Lecture 3
We found in the previous lecture
(3 . 1 )
furthermore
for (3. 2)
This means the original symmetric V v(l, 2) can be written

µ
seagulls 04(1-2) (3.3)
µv
The last term comes because there could be a 6(t1 - t2) term, or by relativity
4
a 0 (1-2) term or gradients thereof--leading to just a constant or polynomial
under Fourier transform.
{ -}
These seagull terms would mean that the abstract form for T would be like
e
-i [fJµ(l)aµ(l)dT + Jsµv(l)aµ(l) av(l) dT1 + ··
T
thus adding a local term at one space time point, but second order in a(l)
as appears in QED for the interaction with a scalar particle, for example.
Of course, instead it could contain gradients, as F v(l) F (l) for example.
µ µv
{ lf f J}
There can be higher terms for higher orders. In short what we have found is
that T must be expressible in the form Tex i L(l)dT1 • where L(l) is
T
an operator depending on a(l) only, for example of form J (l) a (l) +
µ µ
+s v (1) a (1) a (1) + where J , S are operators in the hadron
µ µ v
variables. Such a form is of course, also, the inunediate result of
supposing a local field theory for hadrons.
Research Problem
What experiments could best establish existence or non-existence
of seagulls?
In QED for spin 1/2 there is no seagull, for spin 0 there is (but
with Kemmer Duffin matrices there is not - resolve this!) Since all
quantities are defined by experiment the reality of such seagulls for
hadrons is an experimentally determinable fact.
Thus we see a knowledge of matrix elements of Jµ alone should determine
all scattering amplitudes V.
The restriction on the matrix elements of J are very important.

( 3 . 2)
µ
They lead to many relations--dispersion relations for example. We return to
this subject later when we discuss deep inelastic ep scattering. There are
some special technical difficultiea coming from the highly divergent nature
of some of these expressions so mathematical rigor requires a little more
attention. In practice they give trouble only in the vacu11111 expectation
(of, for example v v) and in no other problem and so they can be avoided beet
11
by disregarding them. ('Ibey have been analyzed by Schwinger, and are called
trivial Schwinger terms).
Conservation of Current
We suppose now that current ia conserved in the more conventional sense
that we will take it to be true that quantum electrodynamic• cannot determine
II II II
a (1) completely, but a gradient a (1) +v x(l) (x(l) is an arbitrary function,
aot an operator) can be added to it (i.e. if a different gauge were uaed 1a
the lepton theory) without altering the phyaice. Thus T �] · T � +Vx ] 10
1 - i J J11(l) a11(1) dtl - f J v11 v

(l2) f 11(1) •v(2 1 dt dt 2 +
{(
l
J (ai1m
T
1- i J11C1> +911 xm ( )) dt - i J v11vu2> a11c1> +
)}
1
+911x<1> )( av(2) +Vvx<2> dt dt +...

1 2
(3.4)
or to firat order ia x
0 (3.5)
' l 11(l) •
v < v o (3.6)
1>
II )IV
Cl 2) • etc.
I
Equation (3.5) we have already discussed; (3.6) give• aomethiag new. Uaiag (3.3)
�Lii �J
4
9 (1, 2). v
II )IV 11 V )IV
v (1) J (2 +' 11 •••gulls a c1, 2)
T
'11t11
� �J IvLii �J
but
0
II
(1) J (2 • J (1) J (2 •
T T
v v
(3.5;
II
by equation except that' dou not comute with the time ordering
operation in
lLii (1) J c2
v
�J T
• eCt
1 - t
2
> J Cl> J (2) + eCt -t > Jv (2) J (1).
II v 2 1 11
We have also to differentiate the 8 with raapect to t. 'Ihua whea 11 • O,

we ge t from this an extra te rm
the equal time commutator of J0(1) and J (2). Thus in general

v
Vµ Vµ/l, 2) • f µ 3µ (1 , Jµ (2
:
} + O(t1-t ) �0(1), J/28 +
2
+ vµ seagulls µv 6 (1-2) •
In our application this lea ds to
6 3 Cx-1-x-
or
[J (1), J �
0 v
(2)
t •t
1 2
• -Vµ seagulls µ v
2
) (3 . 7)
Actually I have simplified the discussion of the seagulls, for gradients also
appear on the 6 4(1, 2), but the point we want to make is that the equal t ime
commutator of charge density and current is determined by seagulls. In
particular, if as many people ( e.g. Gell-Mann ) have suggested seagulls vanish
(by analogy to QED where the coupling is purelyJ�A µ y µ �dT to a spin 1/2 field
there are no seagull diagrams) we would have
[J (1), J (2�
0 v
�t •t
• 0 ( 3.8)
1
2
we have no direct test of this yet (,although we do have tests of 'ii'µ Vµv(l, 2) • O ) .
Remark
Schwinger has pointed out that ( 3 . 8) is impossible. Because if J 0 • p
and J • 1, 2, is written on the vector J it says

v
3
�<i1>.J<i2� -o
Therefore taking divergence � <i1>. v

2
• J( i�
2
. 0 or � <i1>. * <i �
2
• 0
because V
-
J le. But le. is t he operator Hp-pH where H is the
• •
a t
• at
Hamiltonian of the system (assuming there is one�generally the energy of
s t a te opera t or) or
pHp - ppH • 0
Now take the vacuum expectation value. Let the energy of the inte rme d i ate
state n be En, we have
but p0 •
n p!0 and if the vacuum is the lowest state we must have E
n
- E0 > 0 so
t(En - E o >21P no 12• 0

n
which is impossible.
But this argU111n

11 t applies as well to QED itself where we know we have no
seagulls in the original field operator for the lagrangian. It comes becaUBe
we do not really compare thia formal field theory directly to experiment but
remove some divergent vacuum diagr81DB at the beginning. This problem appears
entirely aaaociated with the vacuum problem and could be removed. We can indeed
have the analogy of QED; 'no seagulls' and have equation (3,8) satisfied for
every problem except the vacuum. The precise statement is that ( 3 . 8) holds
if from the co1111u

1 tator you subtract its vacuum expectation value times a unit
matrix.
Lecture 4
Isotopic Spin, Strangeness, Generalized Currents
The hadronic states n, m of a matrix element of J such as <mlJ ln>

µ µ
can be classified into definite nondynamical quantum numbers of isospin and
strangeness (both of which, we assume, are perfectly conaerved by the strongly
interacting system above). The J matrix may have elements between different
t0
values of the quantum numbers, but it must of course conserve charge.
From the very low rate of K0 � w0 + y or A � y + n (although � n + y is
fast enough) we conclude that weak interactions are involved here. Thus we
think J has zero matrix elements between states of different strangeness.
Among sets of different isospin we can describe the result by saying J has
parts of I spin• 0 (isoscalar), I• l (isovector), I• 2 (isotensor) etc.
and use appropriate Clebsch Gordan coefficients to relate amplitudes among
different multiplets. Since J does not change charge it must only involve
the 3rd component, if it is isovector for example. The fact that proton and
neutron have charges + 1, 0 already shows that J is not pure isoscalar
independent of isospin, nor pure isovector (where the c harges would have to be
opposite) but contains a linear combination of these two. No experiment seems
to require I • 2, but I do not know how precisely or exten s i vely this has been
tested. Recently some evidence was claimed for the need of an I • 2 compone nt
in comparing y p -+ 11 +N and yN .,. 11 - p at energies near the A reson an c e --b ut it
appears that corrections for deuteron structure ( for the y neu tron rate is
inferred from yD data) were incorrectly analyzed.
Most theorists today assume Al or 61 1 only for J .

µ
• 0 • (This is
evidently a fundamental question because it tells something of how J is
"ultimately" coupled; for further s trong interactions conserving I -spin

cannot alter this rule--we see "in" through the strong dynamic coupling in
this respect at least because the strong coupling conserves this I spin
character .)
Having available matrix elements <m j J J n> for a variety of s t ates n (and m)
µ
all of the s ame I-spin multiplet permits one by appropriate linear combination
always to isolate the pieces due to the isos calar and isovecto r part separately.
Thus we can de f i ne matrix eleme nts and therefore op erato r s for J5(q) and JVJ(q).
µ µ
But for the vector we could alao ca lcu l a t e (via Clebscb Gordan coefficients}
matrix elements between specific states of other components of the ve cto r

v
cur rent J ( q} or J - (q } (with In this way new kinds
v+
isospin +l or -1). of
µ µ
currents are de f inab le .
This would j ust be an exercise in Clebscb Gordan coeffi cients , but we
think some of these currents are physically important also. We think the
v+
current J (q) is the nonstrangeness changing nonparity violating part of
µ
weak interaction (an assumption known as CVC). This leads to a suggestion
by which these extensions of current are useful in a powe r ful theoretical way
(Gell-Mann).
That QED is coupled to J ' or weak interaction to JV+ is (in our point
µ µ
of view in these lectures )akind of accident irrelevant to s t rong interactions.
They just lead us to a tool to study hadrons, but hadron in terac t ions can be
analyzed alone. Nevertheless we deduced a number of re lations from assuming

either tha t these currents come from some operator in a field theory under ly ing
hadrons, or that hadrons are such that weak perturbation fields could be coupled
(we will use the latter hypothesis), We can expect, for any two points and
components of current J:(l) and J� (2) that they co111Dute if 1, 2 are space
like separated
(4. 1)
This is a We are trying to induce new laws and restrictions
2X
new assumption,
on J and the hadron system. We know �11(1), J}2� • 0 for ol where J,

our electromagnetic current, is an isovector and isoscalar. Using isospin only
bow far is it possible to go to prove say that the isovector par t, or the
�: (1)' J: (2B etc. commute? In the realm of isospin what we assume here is that
the space-like commutation law is true not only for the total current
J S + JV 3 but also for the isoscalar part alone with itself, the isovector
part alone with itself, the isoscalar part with the isovector part and
generalizations for the isovector part with different I spin component
directions.
We can also have the scattering of an imaginary a-type vector "photon"
to a b-type photon governed by
That is, extend the concept of vector potential a (1) to contain another
11
index a, the type, carrying for example isospin or strangeness. Then the coupling
to external potentials could be
T • 1 - i J J= (l) a:(l) dT -
1 t JJv:�(l 2) { a:(l) a�(2 � T
dT1 dT
2 +
Thus Vab (l ,
\IV
2) • { a
\I
b
J (l)• J ( 2 )
V
l +
1r seagulls
ab
\IV
4
S Cl-2)
(4.2)
(4. 3)
Conservation of Generalized Currents
We want the conservation laws analogous to V (1 2) • O . Take first

\I \IV
V ·
the case if I spin which we know is exactly conserved. Consider the scatteri ng
of an I • 1 particle via J� +
. The charge of the final state is one higher
than the initial, so
(4.4)
x'>
(4 . 5 )
but the las t , equal time commutator is by hypothesis ( 4 . 1) zero outside the
3
light cone�it mus t be a multiple of 6 (x-x ' ) , say s ( x) 6 3 ( x-x ' ) . Equation (4 . 1)
v+ -
shows s mus t be J (x) . Hence the equal time commutation relation results ,
µ
(4. 6 )
(this assumes there are no special terms in � (x1 , x2 ) which would integrate
out , a question we shall see related to seagulls again) .
Equation (4. 6) and its generalization to the much wider group su 3 x su 3
are Gell-Mann ' s equal time commutator relations . They represent the first
guessed dynamical property of hadrons that is not simply a consequence of
relativistic quantum mechanics general principles .
We can also describe this from the point of view of a property of the
scattering function y&bv , Take the case one potential is + isospin , the
µ
other 1 3 • Since current is conserved you might at firs t think V µ l Vµ3+ (1 , 2) • O ,
v
but the error is that the potential v, 2 carries in a charge + . So only if the
electromagnetic potential (coupling 3) also couples to the + meson is charge
conserved . E . g . diagrams are like
L
�
a•
// I y
Their sum conserves charge , and if the photon had polarization e • aq

//\,
µ µ
proportional to its own momentum the sum would give zero . V J+v (l ' 2) is j us t
µ
the sum of the first two , the last one is easily computab le and is clearly a
first order hadronic matrix element of a current J ' in this case J+ itself .
µ
Thus one can easily show that
(j us t by I-spin from V µ J µ (1) • 0

3
(4 . 7)
Aa we shall show in a moment (4 . 7) is equivalent to (4. 6) if no seagulls exis t ;

i f they do exist (4 . 7) i s true but (4 . 6) has t o b e modified . (4 . 7) is the
20 Photon-Hadro11 lnteractio11s
more fundamental relation .

To show the relation of (4.7) to ( 4 . 6 ) we substitute (4 . 3) into (4 . 7) .
t'
J
> (l) . J+(2 � -
v 1T
{ 1
µL µ
v/�(l) , J:(2
T
+ 6(tl -t2 ) �o(tl , xl) , <(2� t •t (4 . 8)
l 2
so writing 6 4(1-2) on the right side of (4.7) or 6(t1-t ) 6 3(x 1-x2 ) we obtain
2
the result (4 . 6 ) since 'ii J3 O . In general seagull terms must be added , but
=
µ µ
we do not know if they exist.
The generalization to a general Lie group with generators Ga with
commutation relation ( Ga , Gb ) • • fcab
c Gc ( • Ga•b by definition) (a•b)
f ab c
is
(4 . 9 )
(4 . 10 )
These may b e obtained from noting that the generalization of a gauge transformation
a� a + VµX is , in the group a: -> a: + V x a + (x l< a ) a. Supposing T [a) is
->
µ µ µ
unchanged by such a transformation we find
T �+ Vx + x )( � - T [a]
as a functional relation�or calling 6T/ 6a(l) the functional derivative one

easily deduces
f � x(l) + X(l) )( a(l� 6T/6a(l) dT l ,. 0
for all x(l) , so integrating by parts we have
Vµ6T/6aµa(l) • a (l) X 6 T/6a (1) (4 .11)

µ µ
When T is written in a power series in a (4 . 2) and is substituted into (4.11) ,

zero and first order terms give (4.9 ) and (4.10 ) .
Since isospin is exactly conserved (4.9 ) , (4.10) must be exactly satisfied
when restricted to the three spin components of vector currents. What of
su3 which is only "almost" satisfied? Gell-Mann has proposed that su 3 , although
not exactly satisfied for the entire hadronic system may be more and more
accurately satisfied as shorter and shorter apace-time intervals are involved.

That is how it would behave if an underlying field theory had propagator
gradient terms satisfying su3 , but mass-like terms violating su3 • (E.g. q 'J q +
+ qmq where m is a nonsu3 invariant matrix, q are quark operators) . If Ja

is a strangeness changing current, having for example a matrix element between
A, N then µJaµ (l) 0 because A, N have different amasses . (If A, N at reat
V +
<A I Jxy z l N> • O,

u
<A I J: I N> • a say, therefore <Al a ; I N> - <A IV·J I N> (w A-wN )a
•
cannot be zero. ) That is J8(1) O

µ is equivalent to another operator, say
II
ns(l) . Then
f
VµV:� (l, 2) • 6(t l-t 2 ) �:. J�J t 1• t2 + a (l) J� (2 T } (4.12)
Now the latter presumably does not contain a singularity as strong as 6 4 (1-2)
but if su3 is valid at small enough distances�let us say the 6 4(1-2) singularity
ofVµv8µbv is correctly given by (4. 10) , Thus we say
V •
µv8b (l 2) 6 4 (1-2) Jaxb (2) + "smooth"
11\1
•
\)
(4.1 3 )
where "smooth" is less singular than 6 4 (1-Z) . Then we can still deduce the
equal time co1111u11 tator relations under the above assumptions of smoothness of
the su3 violating term. Equating the singular terms in (4.12) and (4.12)
we find
(4.14)
(seagull terms have been ignored) .
These relations are of very great interest because they are nonlinear
requiring absolute scales, Thus (if valid) they can serve as supplying
absolute scale definitions to the currents so that the rule that weak
interaction of hadrons is + A (rather than .7A) is definable and
V V -
therefore testable. This particular teat has been made by Adler and
Weisberger using PCAC to take the pion coupling as a measure of the divergence
of the axial current. We discuss how somewhat more direct tests can be made
by neutrino scattering later on in the course (in Part II) .
Singularities on the Light Cone
The commutator �µ (1) , Jv (Z� is zero if 1 is outside the light cone of
2 , nonzero inside , What kind of singularity does it have j us t pas s ing
through , or near the ligh t cone , For free fie lds o f any mass in field
2
theory the commutator has a 6 (s ) type s ingulari ty across the cone . There
12
� ]
is experimental evidence from inelas tic s cat tering experimen ts o f electrons
on ( l ) , J ( 2 ) is of the same type

pro tons that the s ingulari ty of
µ v
2
(gradients o f ) 6 (s > where s is the interval from 1 to 2 . We shall discuss
12 12
this mat ter in cons iderab le det ail where we discuss these experiments . At this
time we shall also develop further other formal properties of the commutator
or time ordered products of J operators . These mat ters wi ll therefore b e

µ
deferred until later , f o r I find them easier to discuss when cer tain
experiments are in mind .
Vacuum Expe c tat ion of Vµ �

In orde r , howeve r , not to leave this theore tical discuss ion en tire ly
in the air , I shall illus trate one applicat ion of i t --the s imp les t , the vacuum
expectation of Vµv (l , 2 ) . This is a func tion only of the dif ference 1 - 2 .
Its fourier trans form, into variab le q --which we will call Vµv (q) is needed ,
2
for example , to calculate vacuum po lari zation correct ions ( t o order e ) due
to hadrons . It represen ts the diagram

Hadrons
If we write in momen tum s pace <O l v)JV (-q , q) I O> = -q q v (q 2 ) + 6 µV b (q 2 )

\l V
(by
2
rela tivis tic invariance , gauge invariance implies b � q v or we have
<O j vµv (-q , q) j O> • ( 6 µvq 2 - qµqv)

2
v (q ) . (4 . 15 )
q \J qV
If we wri te ( 6 µ b we see b mus t go to zero as q + 0 in order to avoid
v - -2 -)
q
a new pole at q2 • O. Ac ting on conserved currents the las t term vanishes .
The series of b ubbles propagating be tween two currents is
+ . . .
+
2 2 2 2 2 2
• s
4ne i
av + s
4ne i ,
q 2v (q 2 ) 4ne i
av + s
\l
4ne i
q 2v 4ne i
q 2v 411e i
-2- s v . - .
\l
-- -2- -2- -2-
\J
--2 - 2-
q q q q q q
• 4ne
2 1
s �--=----- s v ( 4 . 16 )
\J
q 2 ( l-411e 2 iv (q 2 ) )
We note there is no mass renormalization of2 the photon, the pole is still at
q2 • 0 , but the residue is altered to 4l-4we •
"e 2 where a v(O) (which may
ia
we
turn out infinite) is lost in charge renormalization. If start expanding
near q2 • 0 to get v(q2 ) a + q� , we can write the renormalized propagator
•
to first order in e 2 as
1
so 411e2 ib measures the vacuum polarization correction due to hadrons in such

predominantly low energy problems as the Shift etc.
QED Lamb
The imaginary part of v(q2 ) for q2 > 0 is the "virtual photon lifetime"
and gives the rate of production of hadrons in (say) an electron-position
collision. Because the imaginary part of the amplitude represents a loss in
s
probability that a photon remains photon
*
".JVV' + • 1 • 1
..
"-"0"""' + 411e2 iv(q2 ) i.e. Prob . + 411e 2 i(v-v ) •
It is therefore directly accessible to experiment. The real part is related

to the imaginary part by a dispersion relation. Therefore hadronic vacuum
polarization effects (to ordere2 ) could be completely determined after suitable
experiments are done. We will discuss this matter in detail in the next lecture.
Lecture 5
e+ + e -+ Any Hadrons
Consider the process e+ + e - + hadrons in state m out. It is governed
by an amplitude
Hadrons
. x.
( S . 1)
Hence the probability is proportional to
2
l <m I J (q) I O> l • <O I J (-q) i m > <m I J (q) I O >
out \J \J out out \J
-
lbus if we could measure the total cross section for e+ + e + hadrons in any
state as a function of the energy E of the electron or position in the case
(q
0
• 2E , Q- • 0 , q2 • 4E ) we can directly measure Em
2
O Ut
<O I J (-q) I m
)J OUt
> <m
OUt
l J (q) I O >
\J
r <O I J (-q> l m > <m I J (q) I O > -

\J out out '.)
m out
<O I J C-q)J (q) I O > P (q) (5. 2)

\J V \JV
•
Relativity and gauge invariance permits us to write this in the form
(qµ q - q2 o µ > e.(q0)p(q2) because we know that if q < 0 (the vacuum state
v v 0
being lowest) no hadron state m could be excited. p (q) • 0 for q < O.

out \JV 0
(In fact if Q , the space like momentum of q is nonzero , the lowest possible
state is a pair of pions with momentum Q hence P µ (q) • 0 unless

v
q > 2 1.2 + m Q2 /4 or q2 > (2 m )2 and q > o. ) Thus p(q2) exists only for
0 11 11 0
i > (2 m )2 • For example, p(q2) • 0 for q2 O, i. e . , space like q2 • Further

11
<
p (q 2 ) must be pos itive , for example if q • (q , O , 0 , O) and the polarization
0
of the virtual photon were x , the sum of absolute squares must be positive on
2 2
the right hand side of (5. 2). lberefore p(q2 ) is positive for q > (2 m11) ,
zero if q2 < ( 2 m )2 •
11
Therefore
F. T .
<O I J Cl)J c2 > i o> •
\J
(5 . 3)
v
F.T.
2 2
<O I J (2 ) J (l) j o> • <O j J (q)J (-q) I O> • (q q - o q ) e(-q0)p(q )
\J v \J V \J V \JV
( 5 . 4)
(F. T. s tands for Fourier transform) .
Now we can work out the commutator and time ordered product. Firs t the commutator
Note this can be written as
r· 2 2
o (q2 -m2 )p(m )dm • (5 . 6 )
(2m )2
11
There fore the vacuum expectation of the commutator in space is
�
jJ
1V 1 - v.12 ) !"' 2dm2 p (m2 )
V
6
I.JV
· ..Ill
i; (1,2) , (5 . 7 )
(2111,,)
which is zero outside the light cone , because C111 ( 1,2) is . It is interes ting
that the proof that the coDDDu tator vanishes outside the light cone , assuming
relativity is so simple for the vacuum expectation . (At the end of these notes
the coDDDu tator , p ropagator and their Fourier t rans forms are listed) . To get
the time ordered product we need
<O ! { Jv ( l ) J
ll
(2) }T I O>�<o� ( t 1 -t 2 ) J v
(l)J
µ
(2) + l (t 2 -t 1 ) J µ (2 ) Jv (l) i o>
we 8(t1-t2 )
f e (t ) e+i Cq0 t-Q0x> d 3xdt (2n) 3 3 (Q)
To get the F. T. need a convolution with the F . T . of
o 0!ic •
q
f8 (-t ) e+i(q0t-Q0x> d 3xdt (2n) 3o 3 (Q) •
-i
q0+it
(2) }T i O> �n ff�q -qlo' +i 8 (qo' )

There fore
F . T. <O i {J ( l ) J •
v ll t
o
1
8 ( -q0' )) (q 'q' q' 2 )p(q' 2 ) dq0'
IJ V
- 6
I.JV
- v
I.JV
(q) ( 5 . 8)
(q�, Q) . rs t do the case t , t for then q 'q ' - 6 q ' 2 µ ,v •

jJ V I.JV
where q' means Fi
• Q2 • q t qt - o t q 2 and comes out of the integral . Then change the sign o f

t
q� in the second term integral to get
The integral is j us t
2n f p(q , 2 >e<qo')2q 'dq'

L
o o
1
or changing variable q� q�2 - Q2 m2
from to • get
)dm
2 2
v
tt (q ) (q q t - 0 t tq2 ) J p(q2m 2
• t
-m +i t
-
iS
tt
(5 .9)
where S t t is the F. T. of a possible seagull expectation in the vacuum. It
mus t b e a constant, o r a finite polynomial in q, and mus t be real. We can

guess (generali zing µv from tt--however see no te below) that wri ting
vµv (q) • (qµqv - 6µv q2 ) v (q 2 ) we have ,
...
J(2m
2 2
p (m ) dm
-i s
)2 q -m
2 2 (5 . 10 )
+i£
11
where S , at wors t , is a finite polynomial i n q ; at b es t i t i s zero . N o t e the
imaginary part of iv ( q 2 ) • t p (q 2 ) as previously remarked phys ically we can also
wri te
iv (q )
2
• - -
11
l �v(q2 ))
Im
J q2-m2+i£ at m2
dm2 + S (5 . 11)
This is a dispersion relat ion for v(q2) , express ing v for all q2 in terms of
i ts imaginary part (wh ich in th is · •se is (a) nonzero on l y for q2 >

2
11
(2m ) , (b)
measurable by accessib le experiments .)
I f a (s ) is th e cross sect i on for annihilation of electron-pos i tron into
a is the to tal C . m . energy squared , we a (s ) • (411e ) p ( s ) /2

2 2
any hadrons where have
so the vacuum po lari zation is given directly in terms of experiment by
(5 . 13)
The Lamb shi f t corre ction (or correction to the magnet i c moment o f the e lect ron ,
e t c . ) due to hadron vacuum polari zat ion depends on 4ne2iv ' (O) which is
l
J
a (s ) ds
w -s--
I t is expected today for reasons to be dis cussed later that a (s ) may , for
large s vary as cons tant/a so the integral would converge and be determined
experimentally .
In general the pos s ib ili ty of an S mus t be resolved to ge t the greates t
use of s uch a r e lat i on . Otherwise i f v (q2) i s known somewhere at some special
value o f q2 (say O or "' ) we can convert to a sub tracted dispe rs ion rela t i on ,
I.e. s uppose i t is known S is a cons tan t , but not a polynomial in q2 , and
v ( q ) is known . i ts value when q q1 to get

2 2 2 2
Sub tract from (5. 10) for q •
1
2 2}
2 } dm2 (ql-q
i p (m
v (q }
2
-
v (q )
l
2
• 211 J (q2 -m2+i£ ) (q 2 -m +iE ) 2
(5 . 12 )
e en
1
same trick helps i f the i nt e gral on l o ok s div
2
S does no t appear . This m rg t.
Tii e in tegral o n m2 ma y now converge bet ter for large m2
In our application any const an t S is uninteresting, as we have seen the
value of v (O ) is of no interest today because it is los t in charge renormalization.
Hence supposing S does not have a q2 term (assumption of no bad seagulls) we
would write a dispersion relation for the quan tity v (q2 ) - v (O ) , the only quan tity
of physical significance
i 2
v(q 2 ) - v (O) • �
211 J�
(2m > 2
(5 . 14)
1T
n.us we may look forward in the near future (when experimen tal results for
p(m2 ) will be more complete) to being able to make a (first order in e2 )
correction in QED calculations of the effects of loops involving hadrons.

+ - -
In a recent measurement of cr(e e ..,. µ+µ ) an effect of the contribution
of the <P resonance to p(m2 ) has been observed . 'This comes from the interference
of the graphs
, '
+ µ+ · y µ
2 2
X
e+ e e
+
A
Hadrons
at a q near m . In the first order in which the effect contributes only the
,
real part of i (v (q2 ) -v(o)) is seen, this is observed as a slight oscillation
2
in a near q2 • m,.
- -
As an example we calculate the modification to cr(e+e ..,. µ+ µ ) when "Hadrons"
in the above graphs is the cp resonance only. In this case
Substituting in to (5.12 )
c dm2
( 5 . 16 )
2
� m2 -m;) 2+r 2/4 ) (m2 -q -i E ) m2
Im (
i (v(q2 ) - v(o)) ) ( 5 .17)
Re (
i (v (q2 ) - v(o)) ) --!;;-
2
p r
J_
�
c
(5 .18)
where P means principal par t .

1lie lower limi t in the integral ( 5 . 18) can be set to -"' with negli gib le
error because the resonance is very narrow . Carrying ou t the integral we find :
(5 . 19 )
(5 . 20 )
where o0 i s the cross section for the process i n the ab sence of � -pho ton
interactions .
iv(q2) .
We need not have evaluated the in tegral in (5 . 1 8) to get the real part
2
rm (iv(q2))
of What the relation (5 . 15 ) says is tha t iv ( q ) i s an analy t i c
• } p (m2 ) .
( )
function wi th n o poles i n the lower half plane and such that
Re
2
2
To ob tain iv (q ) all we need to do is guess an an alytic func t i on whose
imaginary part is 2 p (m ) and whi ch has the correc t poles in the uppe r hal f
1
plane . In this case , the function is easy to guess
(5 .2 1)
the refore
(5 . 22 )
Making the sub t raction at q2 • 0 we ge t (5 . 19 ) .
( ) ' ' "µ' v q ' 2

Note : Annoyin g Point
Let us calculate other µ v componen ts in 5.8 • Firs t tx so q q v -

µ
• q� QX .
1lierefore get
q
� Qx 6 ( q � )
o �
q -q +i £
s o this is OK be cause the factor q 0 � mul tip lies the s ame in tegral a s i n the
xx
- (q0• 2 -q2) q2
tt cas e . Troub le comes however in the case for then the coe ff icient is
0
2 2
q�2 - q0 2
Q for the q • does no t change dire c t ly to as required , but
x
we have ins tead an extra correct ion of which cance ls the denominator
General Tht'Ol't'firnl Backgro1111d 2')
so we ge t
( Qx� - �xxq 2 )
J I
2 2
Vxx ( q ) •
i p (m ) dm i
2 11
+
211 p (m2 ) dm2
q -Ill +i£
2 2
µvq ) J
2 2
v
i
( q ) s 2ii ( q µqv - 2 p (m ) elm .L C + iS
or 6
2 2 + 2 11 (6 -6 6 ) ( 5 . 2 3)
µv µv µt vt µv
� p (m2 ) dm2 •
q -m +i£
where C is a constant (infinite no doubt) •
We could get rid of C by a seagull type term but we are confused because
the time ordered product alone seems not to be relativistically invariant .
Apparently <O l [J µ ( 1) , J v ( 2 � 1 0> • 0 outs ide light cone isn ' t enough . Perhaps
some limit on the singular behavior when t • t near i1 • i2 is also required

1 2
( for relativistic invariance) and is not satisfied here . It is satisfied
J
for every other real problem. Hereafter we drop it supposing some term like
A • A d
3i
is added to the Hamiltonian to straighten it out . It is called
the trivial Schwinger term . In quantum electrodynamics the trivial Schwinger
term is controlled by calculating by regulators , i . e . , propagating electrons
m2 m
2 A2 2
J
of mass minus a term w ith mass + so p (m ) need not be positive and
2 2
p (m ) dm is taken zero .
Certain functions and their transfol'lllll are importan t . Propagator
�•
J
- ip · (t i)
m - e , 1 2) m (2 )
I (x,t) •
+ 2 2
p -m +i£ (211)
4 -
� (s +
811S
H
l (ms ) (5 . 24)
H / • Jl - iN
1
where s • + lt 2 -x2 in time like regions and s • -i � in space like regions .
�
-ims
For large s , H is asymptotically H e . For small s
+
,
I (i t) •
m - .L411 2
6 (a ) +
-n
i
4 11 s
+ i_
1611 -
m
2
i8112 En (�
2 ) + Y -
1
_
2] + (5 . 25)
y • • 5 77 2 . . . .
For free fields of mass m it is j us t <O l < + C 2 ) , + Cl) } l o>

T
Comnutator .
The c0111t
111U ator function
C (i , t) • sgn t • Re ( I (i , t )
+
) (5 . 26)
F . T . C (x , t) • sgn
2 2
q 0 & C q -m ) (5 . 27)
For free fields it is <O I E ( 2 ) , + (l� l o>

-
C (x , t ) • s gn t
r: �
6 ' s2 ' m
J1 (ms��
L
+
Btr s
for real s only . Outs ide the ligh t cone it is zero . For small s
- [ §Jii 2
J
m2 m4 s
C(x,t) • sgn t - 4"' + I6ii - Tiiii + ( 5 . 2 8)
··
Prob lem 5 . 1 .
Show that the total cross section to produce any hadrons from e
+ -
+ e
at energy E + E is
(4tr 2> 2
a • ; p (q2 )
q
2
Problem 5 . .
2
Find p (q2 ) for muon pair production .
Problem 5 . 3 .
Use p (q2 ) of prob lem 5 .

2
to find v ( q 2 ) - v (O ) for the vacuum polarization
of muons .
Low Energy Photon Reactions
Lec ture 6
Pion Photoproduction Low Energy (O to 2 GeV) .
Reference : R. L . Walker , Phys . Rev . �. 17 2 9 (1969 ) .
As our firs t experimental s ub j e c t we take up low energy photoproduct ion
of pious from nucleons . Al1long the individual reactions are
/
y + p + 1 + n
y + p + '11 0 + p
y + n + 11° + n
The first two are the mos t extensively s tudied . Today data on various
polarization and spin cases are also avai l ab le , Evidently we are measuring
2 •
things which depend on <N 11 I Jµ (q) j N > for q O. The ( total energy in the
2 2 2
C . M. ) is of ten called s . s • (p+q) • m + 2mvLAB where vLAB is the energy
of the photon in th e lab sys tem ( for us v goes from 0 to 2 s ay ) . Thes e are
not th e only mat rix elements from N s t ates at thes e energies
+ o +
or y + p + n + A , o
Y + P + n + 'II + 'll y + p + p + n
etc . are also poss ib le but we dis cus s these la ter .
31
The mos t characteris tic f eature of the behav i o r is rapid ch ange s in angular
dis tributions with energy . This feature is characteristic of hadron collisions
such aa 11 - + p + 11- + p also and is interpreted as due to resonances . The
same resonances appear here , of course , as in 11 scat tering , The photon data
has b een explained and analyzed in considerab le detail on this b as is . Firs t
we dis cuas the general " theory" of resonances .
A general element of the T ma trix mus t have a special property of
separab ility defined by the following circums tances , Suppos e we imagine a
collision like A+ B + C + D + E + F. On e po ss ib ility i s that A+ B hit first

to make D + X in one part of space (via element <DX ! s ! AB> and then X as a
real particle goes off across the room to hit C to make E + F via C + X + E + F.
That is <EFD ! T ! ABC> must have an infinity--a singularity when the momenta
satisfy PA + Pc • Px Px
2 2
p8 - and has the value
Mx. for a real particle x .
r es idue
The of the singularity is <EF ! r l cx> • <XD I T I AB> .
2
When Px is very near
2
Mx. ( and Ex is positive) the b ehavior is ve ry much as if an ideal free particle
-1
mass Mx. p rop agated so Mx.2 +
of
2 the amplitude varies as (p
2 - iE) •
X
<DEF j T ! ABC> :\:<EF j T j CX> 1 <DX ! T I AB> (6 . l)

2 2
(pA+pB-p C) -Mx. +iE
4 4
More specifically if s is written s fi • o fi - ( 2 11 ) i o (p f p i) Tfi near
-
a resonance (don ' t confuse T with r (A] us ed previously ) .
<CD I T F I AB> • < CD I TF I Rea> -2,,---1-.,2,---- <Res l t !AB>

p - (� -HR�)
2
wh ere (pA + pB - P c> 2 is sufficiently close to Mx. ( and the s ign of the energy
is righ t ) . There must be a corresponding singular b eh avi or for every way
that the momenta can be combined into the momentum of some s t ab le p art i cl e ,
The exact form of the singularity comes from an an aly t ic extrapolation from
the imaginary part
which unitarity demands because processes A + B + C + X and D + X + E + F

are possib le .
If a small perturb ation (such as weak interaction) ac ts to make s tate X
slight ly uns table and of decay rate r1 it can be s hown that the modification
Low Energy Photon Reactions
near the resonance is to replace
1
by
Mx.2+i £
2
Px -
(There a r e als o s light modifi cat ions of � an d t h e residues , but the f o rm
is the s ame) . The non-relativistic f o rm i s
I n non-relativis t ic theory th is is the f o rm n o mat ter what r is due to .
It has b een found that such res onant behavior of the T matrix is exhib i ted
by s trong interact ions . (Here the width r is not produced by weak external
perturbation , but is a cons equence of the s t rong interac t ion itself . ) They have
( )
the following characteris tics .
-1
T 2 2
varies in Bre i t Wigner fashion p - (ll\t -ir ll\t ) .
R
This only means something preci s e if r is small enough that the variat ions
in the res idue factors < cd l T I R> <R I T l ab> are either negligib le or well known
(as near threshold b ehavior-- there a proper variation of r with Q may have
to be included} ,
The same resonances appear in many reac tions at the same mass and width
(given I-spin , angular momentum , s trangenes s , parity } , The Table of Particle
Properties is a tab le of such resonances . What is the meaning o f the exis tence
of resonances in strong interactions ? There are examples of narrow res onances
in other fields .
a) Atomic systems , The excited levels of atoms are stable except for
coupling with ligh t . The narrowness of the ob s erved resonances in the emiss ion
spectrum of an atom , for example , is known to be due to the smal l value of the
2
coupling e •
b) Nuclei , The coupling is known to be large b ut narrow resonances are
ob served . In thi s cas e the origin of the narrowness of the resonances can
still be unders tood . In the decay of an excited nucleus of high A by the
emission of a neutron ; for example , it is hard for the wave function to
concentrate enough energy on a s ingle neutron to escape . Other effects such
as centrifugal b arriers account for the narrowness of res onances in nuclear
physics .
None of the above effects seem to appear in particle physics . But also
resonances are generally not so narrow , the width is about a fourth of the
spacing to the next resonance in the same channe l . The meaning of the appearanc
of resonances in strong interactions is not unders tood . At firs t theoretical
physicists did not expect resonances from strong coupling field theory . Then
they realized if it were strong enou&h there were "isobaric states" . But the
exact meaning of the existence of these resonances for underlying theory is
not clear .
Lecture 7
We discuss a number of prob le1111 in general in using resonances , in the

specific form used by Walker .
a) How much is resonances and how much is background? Can the b ackground
below a resonance be simply tails of other resonances ? If we write a resonance
off resonance it is
If a varies like a (E) � a( Ea_) + m(E- Ea_) , then our resonance i•

a (Ea_)
� + m .
R
Thus a general background m canno t be defined IDlleas one is specific ab out

the variation of <T> off resonance . Some people like to define <T> as pure
constant and aak then if background is zero--but it may depend on how we
write <T> , e . g . aa \, or other explicit forms and seems a bit arb itrary
althou&h it haa merit of b eing precise.
In practice sometimes we know r mus t vary because we are close to a
threshold where r varies aa Q Zt+l for small Q where t is the orb i tal angular
11101Dentum. Thia form canno t be correct for large Q because it b lawa up too
fas t . T o deal with that empirically Walker chose
Low Energy Photon Reactions 35
for the rate a meson couples to a proton With angular momentum 1. For a
photon a similar
term was used . x was arb i trary (at ab out , JSO GeV for a l l res onances except the
1236 where . 160 was us ed) . But changing x is like changing a (E) ,--it distorts
the form and leaves the ques tion--How b ig is the tail of a resonance?
Impoasible to answer except adl itrarily .
�
The use of r • r {Q/Q ) 112 has implications for the express ion below
0 0
threshold . These are not used by Walker (violating dispersion relations ) .
What are they?
Theoris ts like to choose the infinitely narrow resonance approximation ;

a is cons tant , the background is
a (at resonance)
E-Ea
in defining resonance backgrounda--or rather "effects of resonances far from
their resonant energy" . This last concept in quotes is very sub tle and hard
to define but is used glibly for all kinds of things today in theory , as we
shall see .
b ) Can T b e expressed entirely as a s um o f resonant terms ? Consider
the case a + b + c + d
a
x d
b
s
t
•
•
(p a + ) 2
�
(p c - P a ) 2
u . (p a - p d ) 2
These are resonances in the s channel of the kind
1
and have a s um
(7 . 1)
Then there ar e resonances in the t channel from di agrams o f the form
c>-<d (7 . 2)
a b
For a real experiment none of thes e are res onant s ince t < O , a l l are far llMay .
A theoret ical ques tion is : Should we add ( 7 . 1) to ( 7 . 2 ) (and u resonances ) ?
This only means some thing i f ( 7 . 1) is precisely defined .
It is a tenet of theory (called Veneziano or extended duality) today
that we should no t add ( 7 . 1 ) and ( 7 . 2 ) but that in the sum of s channel
resonances alone is the t channe l b ehavior completely defined ; this is supposed
to apply to the imaginary part only .
An express ion has b een found for ( 7 . 1) which when s ummed gives something
that can b e wri tten as a s um of t channel res onances (Veneziano ) . These are
difficulties of definition in practice . What Walker did was t o take a term
for each resonance th at h e knew about , p lus a b ackground . In fact he wrote
the amplitude as the sum o f three parts :
(1) s- channel resonances
( 2 ) the pion exchange
( 3 ) background .
(l) T h e resonant masses and widths were taken from the w N scattering
data which had b een analyzed by ampli tudes for each channel (of angular
I spin , parity ) . i.e.

µ
momentum , A constant , the matrix <R\ J \ N> <R \ T \ yN>
for each resonance had to be determined empiracally , by adj us t ing to fit dat a .
(The other factor <wN \ T \ R> w a s availab le from w s cat tering) .
(3) The ''b ackground" was a s lowly varying amplitude in each channel .
Goodness of f i t o f parameters in (1) is j udged i f the b ackground c an b e
made t o vary s lowly , o r a t leas t smoothly . I t was hoped that all these
background terms would b e real b ut some small imaginary was ultimat ely
needed in some channels , pos sib ly b e cause resonances are le f t out , or have
inaccurate parameters. (Next time Walker does this he will relax the cond i tion
that the background amplitude be real . )
Lecture 8
We continue with the discus sion of the terms used by Walker .
(2) t channel 1T pole . Thie is from diagram A
· �- r
r 'Ip
A
k
0
(a diagram which does not exist for 1T --but we will say more about this later) .
It leads to a term like
(2 q-k) - e
where g is an empirical determination of how 1T 1 a couple at their

2 5
2 14 . 8 . 2
pole , empirically g � This factor t-m1T vaTies rapidly with t for
small t, leading to a rapid angular dependence . Small t means an effect far
away from the nucleon. The effect is that a nucleon has a chance to be seen
+
as a nucleon with a 1T around it spread relatively far out , as
-m r
1T
e /r
in amplitude. If this virtual 1T is hit with a photon it can be sent out
forwards and appear real ("unbound " ) having received its necessary energy and
momentum from the photon . It goes fairly sharply forward--at higher energies
at least and affects the amplitudes in all valuea of a-channel angular momentum
including high values , where for low energies and momenta no resonances are
large . So it gives the maj or contr:lhution at the higher angular momenta .
Since there the rapid variation with t near t • 0 is essential we are near the
2 2
pole m1T • . 0 2 GeV and we can consider it trus tworthy and accurate . To be
sure , at the higher energies lower angular momenta involve it for t large
2
and so the form l / t-m etc . may b e wrong or amb iguous (we mean we want to
1T
follow the principle� the contribution of poles is well defined only for
parameters near their resonance) . But here the amplitude is slowly energy
dependent and its exact value for the lowes t one or two angular momenta
(s , p waves) is los t in the background amplitudes we are adding anyW'ay .
(No background was added for very high angular momentum) .
The expression we have written for the amplitude is not gauge invariant ,
however . The ff'+ springs from a source and doesn ' t conserve charge . It must
be combined to other terms to make it completely gauge invariant . One obvious
contender is the s channel resonance corresponding to the unexcited nucleon
at 9 38 . We can put these together aa follows (let charges on initial nucleon
these give
• 4w ( 2q-k) · e 1
C u2y 5u 1 ) ew/2 g ( 8 . 1)
- 2k • q+k2
M
I
•
- � +l(-+M e µl
M
II
4 wu 2Y5 2l
k +2p 1 · k
( 1' +
4M ('l( - U > ) ul l2 g (8.2)
•
µ2 } � 2-l(+m
Mxn 4wU2 ( e2 ' + iiM <''It - l(') Y u
k 2-2p 2 · k S l /2
g ( 8 . 3)
With ' replaced by l( the three expressions give
The inclusion of diagram III for the nucleon resonance reminds us of

the question of whether we should not also have added to our law energy s channel
resonances , such as the A, such diagrams as
The answer is certainly yes . They were not included by Walker but again they
are relatively slowly varying for a positive energy w. They are far off from
their resonance (which appears below threshold at negative w .)
The gauge invariance of Hz + M11 + Kz II is seen to hold even if the

anomalous moment terms are omitted . These terms contribute only low angular
momentum at low energy as does the entire II and III , and does not vary
particularly rapidly with k . For small k all three terms have the expected
l/k singular behavior . The anomalous magnetic terms in I I and III go to a
cons tant ae k + 0 because there is a k in the numerator als o . Since the
background term was arbitrary except it was assumed that it was a) small in
higher angular momentum states , b) slowly varying in energy ; we could omit the
anomalous moment terms and include them in the b ackground . Walker did this
and found , perhaps surprisingly , that if he left these anomalous moment terms
out, the background was smaller than if he left them in. Thus his explicit
definition of the "pion exchange term" is I + II + III with µ l = µ 2 • 0
(an expression he calls the "electric Born term" ) .
We see that he could j ust have well left out the nonanomalous moment terms
too and used a simpler expression--readily generalized to make gauge invariant
in a rational way .!!!I. pion exchange term. For
the las t term vanishes because � l • M on the initial s tate . The second term
is like a magnetic moment term in that it has no pole as k + O; it is of lc.w
angular momentum and slow energy behavior ; it can be taken into the b ackground .
Leaving out all such moment terms , as they only contribute to undefined
background we find the pion exchange term can be written in a gauge invariant
form:
"pion exchange term" •
( 8.4 )
(for w 0 , e w • 0 and .-lg is replaced by g, e 1 • e 2 so we s till have a gauge

invariant express ion) .
It is evident that the "pion exchange term" has no precise definition
(some part of the nucleon resonance term is added) . The amb iguity of exactly
what part is academic because the amb iguity is s lowly varying and easily
absorbed in the definition of ''background" . We would sugges t for a general
cas e , at any k (not only k • O) and any initial and final s tate that we take
2 2
j ust the term
( 8 . 5)
tt '\_J pf
y..f � pi
<f l w l i> is the amplitude for the diagram
· -{
Since s trictly this is off shell some definition by theory may be used , although
all we can be sure of is that the res idue on shell is correct . The "w 0 exchange
term" does not have 1/ (t-mw2 ) • The angular dependence is slow and shows no
forward peak .
Any form like
(2q-k) • e - a • e
( 2q-k) • k a• k
for any a would be all right , but ( 8 . 5 ) has the virtue that the singular
behavior as k + O is phys ically correct .
I have gone through this long discussion to show that Walker commi t ted
no theoretical error of principle in his method of fitting . His adj ustable
parameters of fit were the desired quantities <Res l J I N> and the b ackground
µ
amplitudes , adj us ted to be as small and as slowly varying as possible. In th is
way all the data is distilled to produce theoretically s ignificant numbers
<Res l J I N> . Walker gives the value of the matrix element as A:
µ
A • Q (2M2 � 2Q) -l / 2 <f l J · e p01 l i>

rather than <f j J • e J i> directly . Suppose a resonance of mass m1 decays into a
nucleon energy E 2 , and a photon each of momentum Q . (Q (m1 2-m2 ) / 2m1 )
2
• •
If the photon goes off in the z direction with positive helicity +l , there
are two helicities possib le for the proton - 1 /2 , +1 / 2 , in which case the
helici ty of the decaying resonance is +3/2 or +1 / 2 . A is given for each case ,
and for whether it is proton or neutron ( see Table 8 . 1) .
Quark Model of Resonances
The Quark Model

The various baryon resonances have been successfully classified by the
quark model ; they are as if they were made of three quarks , Each quark has
spin 1/2 ; three possib le unitary spin s tates u, d, s ; u, d are I spin +1 / 2 ,
-1/2 strangeness zero ; s is I spin O , s • -1 ; charges are +2 / 3 for u , -1/3 for

s , d . I t i s assumed thst the s tate i s separable in its dependence on unitary
spin , regular spin , and some additional internal coordinate or coordinates
having characteris tics of internal orbital motion.
The nucleon octet and the lowes t decimet require no "orbit" contribution ,
but the next set of levels of odd parity are understood as getting their parity
and some of their angular momentum from a first excitation of an orb ital angular
momentum. These "quarks " can be looked upon as abs tract indices to describe
s wsve function (in which case we need three indices each of which csn be a
unitary and a spin contsining index , taking therefore 3 x 2 (or 3 x 4 using
Dirac spinors ? ) values and another 4-space vector index for the angular
momentum. Alternatively if the picture is taken more literally as particles
going around each other we would be lead to expect two internal degrees of
freedom and two angular momenta which could combine . This leads to the
expectation of a greater variety of states than does the other view , but this
42
Quark Model of Resonances 43
exces s only arises at second internal excitation , or s tates of positive parity

of higher masses . Today there is no sure evidence that these two angular momenta
ara nacessary (a s tate needing them on the 19 70 table is now considered uncertain) .
However , we will suppose they exist , and will s tudy a rather "real" physical
view of the model . The abs tract theory with one angular momentum has not been
formulated clearly enough to do calculations with it--it may not exist as a
true alternative .
I will j ua t briefly remind you of the ideas and elementary consequences

of the quark model . A quantum s tate of three obj ects can have one of 4 possible
properties under permutations of these obj ects . It may be symme tric s , or
change s ign , antisymme tric A, or in one of two states of mixed syuaetry a, S
which upon permutations become linear comb inations of the•elves . Thua

if X represents the two mixed s tates , combine S with X say makes a s tate of
mixed symme t ry X, A with A is syD111S t ric S, X with X makes with different linear
combinations of the four s tates either s , X or A, thua
product s x A
symme tries s s x A
x x s , x, A x
A A X S
With unitary spin there are three choices. for each quark or 27 states . There
are 10 synaetric s tatea l uuu> , l uud> etc . and only 1 antisymme tric l uds>
s s A
(aince each mus t be different) . The remaining 16 are Ba and BS . Thua the
unitary symmetry is
S • lQ decimet
X • ! octet
A • ,!-. s inglet
Only such su3 multiplets are found among the b aryons . For spin with two
poss ib ilities there is no antisy1111111 t ric s tate , The sy1111111 t ric is spin 3/2
quartet , the X (two) is spin 1/2 doublet , syllbolized by superscript 4 or 2.
Therefore collbining spin and unitar,r spin we can make the following
multiplets (SU
6
multiplets ) . S can come from S• S or X · X via ch�rt above so
56 states
� s tates
2 0 s tates .
Finally we combine this with internal orbital motion . We can th ink of a

harmonic oscillator force for definiteness . 'l1le answer depends on the degree
of excitation of the orbits N ,

We are t o make an overall completely symuetrical state .
'l1lerefore pure I S > 56 • !Q3/ 2 , 2 !!1 1 2
4
�· No internal motion . •
the well known lowest decimet and octe t .

N • 1. One unit o f excitation h as the X symme t ry . Because the CM canno t
move (relative coordinates only) x1 + x2 + x3 wh ere X is a pos it ion of a quark

is impos s ible. Only differences can be involved , the l a > , l a > states :
l a> _!
16 ( 2X1 - X -
2 X
3)
I a> _!
./!
(X - X ) 3
2
To get overall I S > , this orbital I X> must be combined with unitary spin and
spin I x> or 70 . Finally when the unit of angular momentum is combined with the
spin in various ways to make different total angular momentum j , all of odd
parity , we find
2!1 1 2.!3/
2 2
2111 2
2 J3/ 2
4 4 4
.Ql/ 2 J.31 2 .!5 1 2
2 l:.Q.1/ 2 3/
2 l2. 2
4
All these mult iplets (except the .§.31 2 > have been found .
If an obj ect can be in one of a number of conditions x, y , z • • • we can
with three such obj ects form s tates like I zzy> meaning the firs t obj ect is in
condition z, the second x , third y . From th is and its permutations l xzy> etc .
Quark Model of R <•so11a11c<'S
we can form the states of definite symmetry
I s > • l xyz> 5 • ....! C l xyz> + l xzy> + l yxz> + l yzx> + l zxy> + j zyx> )

16
l a> • j xyz> • _l_ ( j xyz> + l xzy> + l yxz> + l yzx> -2 I zxy> -2 I zyx> )

2 13
I B> • l xyz> .
1
2 C l xyz> - l xzy> - l y zx> + l yxz> )
I A> . l xyz>A • ....! c- l xyz> + l xzy> - l y zx> + l yxz> - l zxy> + l zyx> )

16
If two say xy are the same s tate y • x replace l xyz> + l yxz> by 12 1 xxz> .
If three are equal only the S survives l xxx> • l xxx> .
5
Comb ining two such s tates in a product we find
S•S • S A• S • A S • .L (aa + 88)

12
S•a • a A• a • B a • .L (-aa + BB )
12
S•B • B A• B • -a 8 • .L (al! + Ba)

12
S ·A • A A•A • S A • .L (- al! + Ba) ,
12
With these rules and the rules for combining angular momenta any specific
state in the quark model can be cons tructed .
"""'
°'
TAEm 8 . 1 Paotoelectric 1!13trix elc=nts .
St!l.te 1'bltiplet
J.,_ Iz
( f' ! Jµe µ I 1) /F A(aev} -1/2 A�(�V)-1/2 1-.e:ICP(c-ev)-1/2
p33 (!.2?.5) 0
4�3/�· +1 +3/2 p - ./6 p -0 . 190 -O . J.78 -o .2g
+l/2 p - ./2 p -0 . 110 -0 . 103 -0 . 136
D13(1520 ) + ff. +c . 115

2!!31J�. 1-1 +3/2 p +O . ll:! -;.0 .151
"tl
�
+l/2 + -0 . 0 36 <:)
p ),p JI ,/Ci -0 .029 -0 .026
- ./3 Ci
+3/2 n - lfi -0 . 115 -O . l.l2 -O . l.!52
;::
+l/2 n + � ),p - ff JO -0 . 0 33 -0 .030

<:)
�.,
;::
;::
....
s11(1535) 2!1/l!2·l-1 +l/2 p + # N> + � Jn +o . 165 +o. 160 -;.O .OS6 �
i:l
+l/2 •
�
n - - ),p - -0 . 115 -o . ::.og .r: . ue
6 s 6"
� n:. JO ;::
�/2
....,
Di,s(l570) p 0 0 0 o .°'o t
4!s;J�.i-]i
+l/2 p 0 0 0 - 0
+3/2 n - ff ),p -0 . 0 57 -0 .053
+l/2 n -0 . 041 -0 .038

- � >.p 10
2 +o . 050
331( 1650 ) �l?2·1-1 +l/2 p - � ),p + � rn 40 .04.7
TA13!,;l 8 . l( con·l; . }
,. e I 1)/r! A<'� ('!:-:!V) ·l/2

S"Ate !>"11t iplet
Jz -:o:
( 1' J J A (C ".JV) -l/2 Ai::t (c-av) •1/2
\l l!
2
n35 ( l670 ) !.�3/-ci.r}3.• 1 -1 +s/2 p + /n +o . 09 1 +') .OSl
+l/2. :P + /$ i\p + ri Iii +o . 09 5 ·!..� .0�2

� .;:, � "
+ f3 K::>
pll ( 14 70 ) +J../2 p (+ Jt 7-p) A +o . 0 2 8 .)0.0:32 s::
I:'>
...,
2!:t/J.�· 0 1 ..,..
.J;-l/2 n -0 . 019 -c .oro
3 ;s:
Cl
(- � >.p) J. I:>.
+-0 . 064
�
:::-15 (1G80) 2s Jss,2+1
-5/:-_.,.,..... ,J2 +::i/2 :P � .070 ..0 . 139 �
(+ � /if) i\
:::i.:i
+!./2 :P ffi -0 . 0 11 -0 . 015 - o
(- A:> + 5 /n} �· �
� i\ ::s
I:'>
+5/2 n 0 0 0 - 0
::s
a
+l/2 n +o . 038 � .OU
(+ .fl >.p) i\
w ( 7 84) 0 !( 0 . 29 7 0 . 22 ± 0 . 0 2
38i �o
A(15 20 ) +3/2 A
+ � J?i +o . 10 9
0 . 09 7 + 0 . 0 10
2131I?E· 1-J
-
+1/2 A Ni -
+o . 012
j
- � [J3 � .m] 1
""
'I
Resonance mass m1 goes to photon momentum Q and final hadrons (see I z column)
mass m . J z is the resonance spin proj ection along the direction of motion
2
of the decay photon of helicity +l .
2 2
�-2
Q • -2-
ml • >.
• n" Q ,
"
p -
-m2 Q 2
• exp ( 1 "'
F "' 1
n (mi'+111 )
2 2
1
n .. 1 . 05 (GeV) 2
Lecture 9
The Quark Model (cont . )
!!....:..1.· Two orbital excitations each of type X can make s , X, A and the total
orbital angular mowi. can b e 2 , 1 , 0 as it turns out the s tate of symmetry A
has total L • 1 , s tates S or X can have 2 or O. Hence w e c an make
2 2
A11 2 !31 2 ·
The 20 has not been seen . It has no matrix element of any operator (like
J 11 ) operating on one quark at a time to · the proton . Two quarks at leas t mus t
have their motion changed to get to the 20 from the fundamental �6 +J N
,O • 0•
We have definite need for the �6 , 2+] b ecause of a d of spin 7/2+ which
can be gotten in no other way . Two other d ' s of this set of decimets at
j • 1/2+ , 3/2+ have als o been seen at roughly the same energy . This whole
multiplet is expected as a Regge recurrence of the fundamental • �6,0�N 0
and is 2 . 1 (GeV) 2 in mass 2 ab ove . This also accounts well for the octet
at 5/2+ (m2 • 2 . 85) .
A puzzling s tate is the 1/2+ Roper resonance at 2 . 16 . I t i s beat fitted
as evidence of the js6,0�N • 2•
What is puzzling about thia state is its law
ma s s a s compared to the
21
512 at 2 . 85 which is b e s t f i t ted in �6 , 2� N •2 •
These two s t at es are identi cal in their SU {6) properties . The ctuark model
with harmonic dynamics predicts them to be of the s ame mass s ince they b o th
corres pond to two internal exci tat ions . The l arge mass difference shows that
harmonic dynamics may be too s imp le . The Roper res onance is a b reathing mode
os c i l lat ion of the nu l eon ; it

c is like ly that all b re athing modes o f o s cillation
lead to a resonance of lower mass than those with a
� 6 , 0�N 2 has
net orb i t al angular momentum.
These �6,oJN ,. �6 , 2 l a the states that a

•
The corresponding decimet in th e no t b een seen .
2, 2
+
re re expected i f
•
w ere only one internal deg ree o f f re edom . The 70 ' s wou ld �
o
th ere b e expected .
Row g od is the evidence that they are there . There are three s t ates that may
require them :
+
l/ 2 N (l780) poss ibly
2
r7o,o+J
m • 3 . 16
3/2 N (l860 ) pos s ib ly [10 , 2+] m2 • 3 . 46

+
7 / 2 N {l990 ) [ 70 , 2+:i m • J . 9 6 .
+ 2
Th e l a t t e r s t ate would have t o be �O ] ,2

+
; it would b e impos s ib le with one
de gr e e of not {It was th e 19 70
� J
freed01D only--but it is well es t ab lis h ed . on
P ar t i c le tab les , but no t on the 1971 tab les ! ) Th e f i rs t could also be 6,o .
N 4
a hi gh er "b reathing" mode excitation of the nucleon , the s e t being nucleon
m • . 881 , Roper m 2 . 16 , N ( l 7 80 ) m •
2 2
•
2 3 . 16 .
i f only in t e rna l motion is allowed , would have to b e
E6 , 2 ]
The second , one the
at
+ 2
miss ing o ct t
e from the J • 3/2 . If so its 1118S s • 3 . 46 is very
s t a te so
2
dif feren t from the mas s = 2 . 85 of the J • S /2 the usual rule o f
small spin orb it coupling would h ave to be abandoned .
Thus the evidence is in favor of rectuiring two internal degrees of freedom
but it is not conclus ive .
�·
to c a re ful ly
o
A good res earch p roblem is s tudy the propert ies of t he 20
via the quark m d el and propos e the wises t way to look for it exp er imental ly .
The di f i cu l ty
f in ob serving the 20 is that if we as s ume we prob e the b ary ons
th rou gh an o p e ra t o r which acts on one quark at a time (r e g rd less of the nature

a
of the operator) then the 20 cannot be reached from the nucleon . The r e as on for
this is that the is antisymme tric in its su 6 indices while the nucleon is
20
totally symmetric . Operating on one of the quarks in the nucleon still leaves
the other two in a symmetric s tate .
There are many s tates possib le at N a Th e mos t important i s surely the
�0,1-J at �0 , 3J
3.
Regge recurrence of . A few states of this are apparently
N �6,4J N
known .
• 4, A t; o f spin 1/2 probably from 4 is known .

•
Calculation of Matrix Elements .

The simples t application of the quark model is to calculate matrix elements
assuming a nonrelativis tic SchrBdinger equation for the quarks . There are
amb iguities of factors like m2 /m1 or E 2 /m2 (energies and masses of final and
initial s tate) but for our photoelectric matrix elements they are small .
See R. Walker "Single pion photoproduction i n the resonance region" International
Symposium on Electron and Photon Interactions at High Energies , Daresbury ,
England , 1969 . See also references therein.
One s imply works out the states of a system with Hamiltonian of three
harmonic oscillators :
H •
l + 2 + 2 +P 2 mw2
2m ( p l + P2 + 3 ) + -2- (9 .1)
One takes the spin and unitary spin dependences to factor out from the internal
motion .
The interaction of one quark , say 1 , with an electromagnetic field is taken
to be
where e 1 is the charge on the quark . Matrix elements of this operator (with A
the appropriate plane wave for a photon) between states n of the Hamiltonian.
(9 , 1 ) gives us
our nuni>ers for <Res l J I P > to compare to experiment .
µ
The quantities m , w , g , are parameters ; g is the gyro-moment of the quark .
Th e Dirac value g • 1 works very well an d it w as chosen to be 1. Th e value o f
w i s guessed a s 400 MeV so the level spacing will be roughly right--the value
of m is chosen (as 340 GeV) to fit the proton magnetic moment . The values of
the current elements that result are given in table 8 . 1 next to last column
Anon-rel . • ANR . It is seen that there is remarkably good agreement with the
known values . All the s igns are right and small values are predicted to be
small , sometimes by selection rules generated by the model--sometimes by cancel-
lation of tel'111S for orbit and spin coupling . (The terms containing p in the
column for <f l J e I i> are the ones contrib uted by spin-- the formulas are from
µ µ
a relativis tic modification . They are virtually the same as the nonrelativis tic
cas e ) . Finally the orders of magnitude are generally very close� the wors t
is the F15 (1688) wi th +3/2 he l i ci ty on proton which is off by a factor 2 . It

is as possib le that this is an error of the theory , as that it is a sys tematic
error of Walker ' s fit where he omitted , for simplicity , a number of resonances
which the quark model says should not be small .
Several questions arise i f this i s real . Firs t a non-relativis tic theory
is surely not valid containing particles of mass 340 MeV in which the firs t
excitation is 400 MeV . Second , with such low quark mass there should be
QQQQQ s tates .
We have us ed Bose s tatis tics . If the quarks can be free it can be proven
that spin 1/2 implies Fermi statis tics . The only way out is to assume three
kinda or "colors " of quarks for each su3 type and say that the bound s tates are
pure singlets in the new index . For example , baryons mus t be in the 3Q s tate
containing one of each color thus antisymme tric in the color , to say now that
quarks obey Fermi s tatistics require� symme t ry in the other indices . (A formal
way of dealing with this called parastatis tics can be shown to be exactly
equivalent to the three colors of Fermi quark theory . )
Lecture 10
Feynman , Kis linger and Ravndal , Phys . Rev . (1971) , tried to remedy the point
that a nonrelativistic theory was ambiguous in haw you applied it to a prob lem
of relativis tic kinematics by making a theory where the rules at leas t were
relativis tically s tated . Such a theory to be complete would have to be a
relativistic field theory which is too complicated to calculate . They made a
s imple theory of states , but at the expense of unitarity--which meant cons iderable
trouble later on--but at least relativis tically unambiguous rules could be
formulated . The net result for our photon matrix elements is to give almost
exactly the asme nume r ical result as the non-relativis tic caae--so it is not
of great interest h er e . (It do es so with one less arbitrary cons tant , however . )
Nevertheless we outline the meth od .

They noted that the
harmonic os cillator Hamiltonian could be written
2
2mH p2 + m2111 x2 and that 2mH was nearly the difference of the s quares of the
•
energy m. + W (or m.aaa ) of the s t ates . Thus to get {m.aaa ) 2 to be equally spaced
2
they could use an eigen-vector operator for maa s which was a harmonic os cillato r .
The only cona tant is m.2 1112 wh i ch they call ri 2 , so the quark mass disappears on
the righ t hand side. Thus let
K • J (p a2 + pb 2
2
+ p
c
2) + ;-
n (u -
a �
)
2 + (� - u ) 2
c + {u c - u a )
2 + C
( 10 . 1)
cons t ant , P ua • 4-dimensional p o s i t i on o f

a
C is a is 4-dimensional lllOln
lle tum ,
quark a . This in r e c i p r o cal is a propaga tor l/K. Its poles (K • O) give aiass 2
as seen as follows :
Take out the cm. motion
( 10 . 2)
to get relative 1DD11&n tum ope rat or s � . n (with coordinates x , y) find
K • p2 - N
( 10 . 3)
N is the Hamiltonian for two oscillators . T hu s s tates of eigen-value

2 -1 as particles o f mass 2 • N0
N • N0 have poles in K as (p - N) , or propagate
fl . indicate mass inc r eas e s by
2
Evidently these are separated by Regge slope&
l.05 (Gev ) 2 per uni t of angular momen t um , hence take fl • 1 . 05 GeV ,
{For the b aryon photoelectric ma trix e lements the re are no fur the r (e f f e c tive )
adj us table cons tants . )

Quark Model of R t•soncmct•s 53
If K is perturbed by 6K the propagator becomes
.!K + .!K 6 K .!K + .!K 6K .!

K
6K .!
K
ao in first order the disturbance 6K becomes
� <f l 6K l i> -1---2

p -m i p - mi
or the appropriate perturbations of m2--whic:b. is the relativis tic T matrix

elements defined by the usual relativis tic rules-- is j us t the matrix elements
of 6K b e tween eigen s tates of N .
What about spin? Th e lack of much spin orbit coupling s ugges ts that spin
factors from the rest of the wave function--thus the operator K is the s ame
expression times a wave function with three dirac 4 component indices although
it does not affect those indices in no electromagnetic field . However p a2 is
interpreted as its equal c; a >2 so when an electromagnetic field is operating
the operator is
KinA • 31}a - e .' C ua
8 �( ) (
; a - e 8Aa < ua> + � - �Ai, <� >
2
)
(
+ ;c - e cA( uc > )� + �: Gua - �) 2 + (� - uc ) 2 + ( uc - ua> 2 ]
(10 . 4)
e8 , � · e are the charges on the three quarks .

c
The f i r s t order perturb at ion thus gives the current operator for momentum q :
(10 . 5 )
Matrix elements of this are taken between states of the system .

But the system has too many states . (1) The 4-dimensional oscillator has
time- component states which have negative norm (or negative energy - FKR prefer
the former but see below) . To avoid this it is assumed they are not excited -
so the experimental states satisfy the subsidiary condition of being in the
res t s tate of the oscillator in the direction of the 4-1110ment11111 P � of the state
*
( if aµ and b µ * are creat ion operatr�s f or the two oscillators ) ,
*
P µ a µ I +> • o
( 10 . 6 )
( 2) By taking + to involve Dirac spinors twice as many s tates as des ired are
included (Q and Q) . To avoid the parts of the s pinor appropriate to ant i quark
s tates other sub s id iary conditions are assumed .
Y I +> m l +>
JI )I C
P - c10 . n
2
where m is the mas s of the s t a t e (m • P JI P JI ) . ( 3 ) The curren t is not gauge
d i f ferences used are exactly Nn ( i . e . C is a t rue

2
invarian t unless the mass
2
cons tant ) . Thes e mat rix element s are computed with the true experimental mas s
s separa tion Nn .
2
which may not have exactly Transi tions from 8p to quar t e t
2
s tates a r e unaffected . Also t h e F ( 1688) is unaf fected for its mass is almo s t
15
exac tly 2 n above t h e proton . The uncertainty involved for a l l e lements of the
table ( except w+wy ) is probab ly very smal l .
Om i t t ing such s t a tes [10 . 6) and ( 10 . 7 ) ) means a complete set i s not used
and , s ince the omit ted s tates have nega tive norms , means that mat rix elemen ts
will be generally too large b y ever increas ing factors as the mas s discrepancy
of initial and f inal s tate rises . FKR compu te all kinds of meson decay widths
(via PCAC us ing the divergence of the axial current for the pion coupling
operator) and f ind this to be the case . To compensate they divide by a factor
- -3
)
(u2 u 1 ) for the spins , and include a cutoff factor
{_ a
� ml
2
( 1 0 . 8)
\
exp
2)
n cm + m2
1
(with a determined empirically to f i t , a '"" l /n .
Thes e do not seriously affect our matrix elements , of curren t , for lower
states ( the theoret ical 1688 would be 40% higher without them) they have been
included in the numbers given in table ( 8 . 1 ) . I t is seen that no essential
modificat ion of the non-relativi s t ic model is implied for these ma t rix elemen ts
of J11 •
The table also contains one meson ent ry , that for w � w + Y i here
relativity is very important . The agreement is good but recent experimen ts
give still lower resul t s (more like . 15 ) . The value comes from ob servation
of the b ranching rat i o rate of the w into w + y.
Bes ide these non- diagonal mat rix e lements there is , of course , the diagonal
elemen t <Proton IJ µ I Proton> . This is known for all moment um Q as the form
factor of the proton - we discuss here only the low q resul t s , the charge
(which of course comes out righ t ) and the magne tic momen t . The quark model
( a s used here) i s quite unsuc ces sful there . The rela t ivistic theo ry gives
µ • 3 . 00 Nuclear magne tons for the p ro t on and - 2 . 00 for the neutron ( this
compares to 2 . 7 9 and -1 . 91 experimentally) . I t ( like SU3) gives µ A � 1 / 2 ll �

N
-1 . 00 , b u t in the magneton of the A hence -0 . 84 Nuclear magnetons .
Exper imen t is -0 . 70 ± 0 . 0 7 . (The reas on it is magne t ons of the mas s of the s tate
•
2
i s ins tead o f energy perturbat ion • µ (a • B ) we have perturbat ion o f mas s
mµ (a • B ) so the values of mµ are the s imple numbers ) .
Suppos ing that all this serves as evidence for the quark model , where do
we go f rom here ? We have these pos s ib i l i t ies :
I Extensions and improvemen ts in principle of theory :
a) Extension to o ther c urren t ma t r ix elemen t s . Naturally axial current
elements can be calculated dire c t ly from the s ame model (pu t .I( � y 5'0 and a
few are known by a -decay . They come out wrong by the same factor , if theory is
mul t iplied by . 7 1 it agrees wi th experiment ( e . g . , for g we get 5/3 from the

A
model and 1 . 23 experimentally) . Us ing PCAC amplitudes for resonan ce pion
emission to nucleon can be calculated with general succes s , except for a
numb e r of very sens i tive matrix elemen t s whic h come out as the d i f ference o f
two large terms . For details see FKR - we will discuss the photoelec tric
2
mat rix element s for q • 0 later .
b) States o f negat ive norm . These c ause a l l kinds o f amb iguit ies and
uncer t aint ies . Firs t , nobody has any idea for the factors int roduced by the
spin (except t o cons ider the wave equat ion t o be that for a two-component
spinor wave funct ion - this has no t been worked out - on the face of it , i t
appears it would violate parity .) Second , w e have negat ive n o rm t ime-like
s t a te s . In this regard , a sugge s t ion was made by Fuj imara , Kobayashi , Narniki ,
Progr . Theor . Phys . (Kyo t o ) .!2_ 1 9 3 ( 19 7 0 ) . FKR use the "wave fun c tion "
z ] for the gr ound s tate of the osc illators (all o thers are
2 2 2 2
exp [ t - x - y -
2
t
s imply polynomials t imes this factor . ) The explod ing e can only be c on t rolled
2 2 2 2
by negative norm . FKN use instead exp [ -t - x - y - z ) which they po int
out is al so a solut ion of
- (t
2
- x
2
- y
2 2
- z )
) ' • E� ( 10 . 9 )
which is after all the s um o f four independent oscillators for t , x , y , z
each of wh ich can be in its ground s tate . The time oscillator now contributes
have s tates below the ground s tate

2
negat ively to mas s , and we can and even
2
s tates of negat ive m Again we shall have to say something like "t ime s tates
are not excited " or something - but at leas t i t is an alternat ive view .
Thus the wave functions in momentum variab les have the form
exp -
1 ( p
2
-
(P . p) (P • p)
p2
) ( 10 . 10 )
20
2
where P is the four-vector momentum of the overall s tate P •
�s 2 • (The
ground s tate wave funct ion is simply this factor - others mult iply by polynomials . )
i • x
between two states P 1 to P ; P - P •
1
Ma trix elements o f currents e �
2 2
q
involve overlap integrals like
( 10 . 11)
which is easily integrated to give a form factor
exp ( 10 . 12 )
Thus , for examp le , the form fac tor for the pro ton (assuming no spin factors )
2 2
where P 2 P �. ( P -P )
2
• • •
1 2 1 1
q comes out
1 1
( 10 . 1 3 )
2 4
it is seen tha t for large (negative q ) this behaves as l/q as does the real
form facto r , and it generally fits rather we ll !

2 •
For the photoelectric matrix elemen ts (q 0) assuming the s tates observed
are those whose time s tate is not exci ted , al l the matrix elements are nearly
2M M
l 2�
exac t ly the same as FKR (expect for an oc cas ional factor of M 1� / P 1• P •
2 �2� 2
M -tff
l 2
which is close to 1 ) excep t they all carry the form factor F , it is theoreti cally
much more sat i s fying than FKR be cause it is no longer chosen arb i t rarily) wh i ch
in this case works out to be
(10 . 14)
which is much l ike FKR empir ical F exce p t f o r the 4 i n the exponent ( s o i t i s
very close to FKR ' s F to the fourth powe r) . Thi s cuts o f f much too fas t and
makes the poor f i t o f the 1 6 8 8 even much more unsatis factory and a l s o des t roys other
cas es that f i t te d we ll .
c) Extens ion to higher order perturba t ion , in particular ma trix e lements o f
the s uccess ion o f two current ope rato rs . Sca t t e r in g of photon , elect romagne t i c
ma s s di fferences or non-lep tonic weak decays require the matrix e lemen t of the
product of two curren ts . In the lat te r case the two currents are at the same
space time point and we can prove that i f the quarks are Bose the famous 6I • 1 /2
rule for these decays is explained . A natural s ugge s t ion is the s e cond order
pe r t urba tion 6K i 6K . However , in s umming over all s tates in the p ropagator l/K
what should we do?
a) Sum over all s t a tes including time s ta tes wi th negative norn , and include
no arb i t rary form fac tors .
a> Sum ove r only physical s tates (no time s ta tes exci ted} using some
assumed form facto r , - if so , what factor?
y) Use the sys tem o f Fuj imara e t al. s ummin g over all s t a tes including those
of negat ive energy .
There are awkward prob lems associated with each way (wi ll the commut ators
vanish outside the ligh t cone ? } and no thing precise or defiaite has yet been
proposed to compare to experimen t . We shall d i s cuss this prob lem again when we
have developed further theore t i cal too ls , e . g . dispers ion relat ions .
II Modo fications of the Speci f i c Mode l of the Theory .
Can the theory be modi f ied to be in grea te r degree in agreement with
experimen t ? The bes t clue to s ta r t with is probably the real value of the
mass , wh ich really do depend on the su 3 ;i.�d s pin character .

2
Perhaps the
'
mos t impor tant clue i s the large separation of the n, n, n sys tem contrasted
2
with the small differen ces (all accounted for by the extra mass for st rangenes s )
of p , w, �. I t s eems to indicate the interaction fo rce is not a vec tor , but
an axial vector . An interes ting prob lem is the split o f E and A in the baryon
octets . The s imple quark model predicts them to be degenerate .
Next comes the correct ions that the quark model implies to itself for
phys ical cons is tency . Fo r examp le s uppose , us in g PCAC or a modified quark model
tha t has some kind o f direct coupling , that a pion , 6, and P are coup led - we
can calculate an amp li tude for
v
6
I p
This imp lies , of course , that the 6 can decay into n, P and thus the 6 has
a width . But in the original theory the 6 was an exac t energy leve l . We can get
the width by imagining the delta propagates also via vertical diagrams like
p p
the quark model gives the coupling to be used later as the beginning of some
per turb ation expans ion . This succes s fully induces an imaginary part in the
propagator irm - but at the same time corrects the real part so the expe ri-
mental mass is not the uncorrected mas s from the direct us e of the quark model .
@. (D E6, �
How much is the correction and can we use it to understan d some part of say
the I: ,A
4 2
mass dif ference , the o split in the etc1 The
perturbation theory is hard to use , it has a tendency to diverge , it seems to
involve couplings far away f rom their mas s shel l , there are many more s tates
than j us t the Pn available to the 6, etc . Finally and mos t impor tan t - maybe
we are partly counting the same thing twice - perhaps in the equation for the
quarks we s tarted with already contains some of the physics of interaction
that we have later put into the perturbation correction .
Due to these uncertainties nobody has worked these things out . (For
another examp le , h ow· much is the magnetic momen t of the proton corrected ? It
has some ampli tude to be a delta and a p i on , or even a pro ton and a pion) .
Th i s que s t i on is very serious . What is the quark model a model for? For
the calcula t i on of firs t order coup lings among a s e t of ideal s tates wh i ch is
later to be used in some defin i te prescrip t ion ( f o r example , a se ries of
diagrams like a f ield theory pe r turbat ion expans ion ) to compute the f inal
hadron world?
The quark theory is an incomp lete theory . No definite prescrip tion is
given as to what to do next to corre c t i t , or even to use i t completely in a
corre c t and cons is ten t (wi th uni tarity , for example) manne r . As such i t
reaches a dead end . It shows tha t ce r t ain regulari ties exi s t among the hadron
s tates , but a precise s ta tement o f what this means or what this really
imp lies one should include in des igning a more adequate theory is s ti l l
lacking . It is a mo s t impor t an t prob lem in the theory of the s trong in ter
actions .
This "dead end" is a) a result of our lack of imagination of how to go
further and b) a prob lem that every "narrow resonance approximation" has to
face .
Pseudoscalar Meson Photoproduction, High Energy
Lecture 11
Pseudoscalar Meson Photoproduction - Higher Energies
We now cons ider what happens at higher energy - in parti cular from
5 to 18 GeV (as full data in the t rans i tion region 2 to 4 GeV is lacking) . Here
the analysis directly by sums of resonances is hopeless as too many are involved .
A full analysis of amplitudes does not exis t , mainly cross sec tions and a few
asymme t ry measurements with polari zed pho tons , or polarized targe ts does exis t
but our theo retical s k i l l at analyzing the ampli tude for each helici ty e t c . is
no t yet good . Therefore this will be more of an empirical survey po inting out
problems , rather than a complete theore tical analysis - in sho r t , we do not
understand this region .
References
Diebold "High Energy Photoproduction" , High energy physics confe rence ,
Boulde r , Colorado (1969 ) .
Wiik "Pho toproduction of pseudos calar mesons " , International symposium
on electron and pho ton interactions at high energies , Cornell ( 19 71) .
We shall study in th i s part only a ve ry small f raction of the to tal photon
60
Pseudoscalar Meson Pl10toproc/11ctio11, Higlr £11c•rgy
nuc leon cross section at high ene rgy . The cros s sec tion fo r yp + n +n is , for
example , 0 . 6% o f the t o t a l at 5 GeV .
The to tal cros s section for y on p shows bumps due to the 1 2 36 , 1560 and
1 700 res onances at low energies . It is mos t ly f lat above 3 GeV , it falls
from 12.� �b to 113 �b , from 5 to 15 GeV ( this is about 1 / 2 2 0 of np cross section) .
The reac tions go increaa ingly in the forward direction so that it is
convenient to plot them as functions of t for given s (or k , the pho ton
momentum in the lab . ) . The s lope of the t dependence varies relatively s lowly
with k, and cer tain charac teris tic regions for t are no ticed . There is also
a peak in the backward direc tion , small u, where i t is best to plot u for
given s .
An empirical rule , no t precise , but very good for comparing data at various
-2
energies is that all the cros s sections vary nearly as k in the forward
-J
di re ction , and as k in the b ackward direct ion . The reason for this is no t well
2
unders tood . That is , k da/dt is a nearly f ixed function o f t . Below t about
2
1 (GeV) the curves for different react ions are each dif ferent . They are given
on a chart from Dieb old� repo r t (reproduced here in fig . lZ-1 ) . ( This is no t
precise data , varia tions of f ( t ) with s are smoo thed over - to get exact numbers
see detailed experimen tal repo rts - but they are exce llen t representat ions of
things for a survey such as this . )

3t
Above t • 1 all reac tions agree in falling very much like e The cause
of th is remains unknown . In comparing to hadron collis ions this rule is seen
to be mos t accurate for pho ton react ions but is seen also in many s trong in ter
- 0
action cases , fo r examp le in forward ( n p+ir n or n+p+k+I +) as we ll aa in back-
ward (n+p+pn+) (G . Fox) . It may be related to ano ther s t rong interaction rule ;
that in high energy inclus ive reactions making mesons or b aryons these particles
JP 2
are distributed in transverse momentum P.l. as e .L This is also t rue of
l.
-IP 2
photon induced inclus ive reactions . This is only true in a rough sens e , e
-2 . 2P 2
for small p J. to e J. for large PJ. •
2 2
The theory of transverse momentum distributions at high energy and high t
is wide open . It is a very good p rob lem to work on .
Very supris ingly b ackward photopro duction peaks are the only ones that
seemingly do not obey the rule . They are very flat - flatter than anything else ,
1•2 u• �-A
++ -
(This is not true for y,... w wh ich has a rapid fall
varyi ng as abou t e
3u
of f with u in the backward direct ion , more like e .)
The behaviour a t smaller t ( t < l ) can quali tatively be explained b y the
idea of the exchanges of Regge poles of the following traj ec tories :
w , and s trangenes s analogue K
p and nearly degene rate Az
K* and nearly degenerate K**
-2
The k behaviour would not be expected , for the varioua traj ectories have
-2
di f f erent intercep ts and s lopes . But the k canno t be a true asymptotic rule
about ampli tudes - it mus t be a sort of accidental res ult of several terms
varying nearly that way - although quanti tatively it is always a bit of troub le
-1
to ge t i t out . This can be seen becauas if the ampli tude varied truly as s ,
or rather , in correct normalization , were constant , the pho ton coupling via
isoscalar and isovector would differ in phase by 90 ° and would not in terfere .
+ -
Therefore we would expect da /dt (yp+ w n) • da /dt (yn+w p ) and they certainly
are not equal .
The reas on isos calar and isovector differ in phase by 90

°
is this . For
Thus if f (s ) • 8 s
iw * m ( t)
any ampli tude f ( s ) the antiparticle amplitude is f (e •> .
-iw a a
for antiparticles it is S * e s . But for an isos calar coupling the amp litudes
for particle and an tiparticle are oppos ite , this requires 8 8 • 1c exp ( -iw a / 2 ) ,
c is real . For isovector coup ling the amp li tudes are the same because the
antipart icle has oppos ite I-spin thus S • c ' exp ( -i w a / 2 ) , c ' is real . This
v
shows that a
.
and a di f fer in phase by 90 ° .
v
In Kegge theory the exchange of a traj ectory can be in terpre ted as a sum
over t channel resonances . In each case the polarization of the exchanged
resonance is uaed up in get t ing as high a power of s as poss ib le , by being in
the direct ion of the two longitudinal momenta . Take , for example , the case
where a vector particle is exchanged and there is no spin flip at ei ther vertex
Pseudoscalar Meson Photoproduction, High Energy 63
The ampli tude goes like � (P i' e) {P i e) • P ].' P • s. The spin is therefore
P l 2
not available to carry longitudinal angular momen tum. If there is a helici ty
f lip 6A at one of the ve rtices , the spin of the exchange is again used up in the
6 I
term with the highes t power of s ; hence there is a fac tor Q A or lt'"l A .
6
If
+ 1 6 A2 1
b oth ver ti ces have a f l i p the amplitude goes as ,tt"" l A l l
6
ideally . This
later is true i f the 6A are of oppos ite s i gn , it is required by the overall
conservation o f angular momentum . However i f the 6A are equal , say to 1, we
expe ct t but th i s is not required by angular momen tum conservation . In fact ,
in such cases the e f fect is lost by absorp tion . Absorpt ion corre c t ions for small
impact parameters reduce the t to its lowes t phys ical pos sibility ( in this
example to a constan t ) . Only if the coup lin11 real ly factorize mus t i t be
t l Al l
6 + 1 6 A 2 1 otherwise absorp tion can reduce the powe r .
Lec ture 12
Pseudoscalar Mes on Pho toproduction - Higher Energies (continued)
We now discuss each case in closer detail (following G . Fox) yp+w 6

- ++
6 ** p
Thi s case can be explained s imply by w exchange . At the pho ton to w
vertex we mus t have a flip of helicity, at the p6 ver tex none is necessary
2
(because of the p - 6 mas s d i fference ) ; hence the ampli tude is ,/t'"/ ( t-m ) showing
w
2 -2
a dip at t • 0, a maxim1111 at t • m w exchange expects k cross sections
w
( s 2 a -2 , a Z 0) . Gauge invariant one pion exchange wi th ab s o rpt ion corre c t ions
works fine in this case .

+ -
yp+w n , yn+w p
We have a helicity flip at the yw vertex and one (from parity) at the
we
2) ,
w
nucleon vertex . Hence ideally expect t/ (t-m there should be a drop
at t • 0 but if we look at the data in figure 12-1 we see there is a spike in
the forward direction , The answer to this is : ''because of absorption " ,
To unde rstand how absorption affects the ampli tude, pe rform a Fourier trans form
to the impact parameter rep resentation . The amo li tude t o find a
pion now looks like V (e · µ b ) which is an odd function of b . The photon couples
VV
b
to this pion with amplitude that goes like (e -µ ) , the firs t gradient is
because the pion is in a P wave with res pect to the nucleon , the second gradient
VV
-
is because the photon coup les to the pion in a P wave , (e µ b ) is a function
that looks like
The forward ampli tude is proportional to the integral of this f1mc tion which
is zero so we still havedt explained the observed peak . Absorp tion is here
introduced , pions which are s cattered by photons with small impact parameter
do not always get out because there is s good chance for them to interact
VV
b
strongly wi th the nucleon . The function (e -µ ) mus t be multiplied by a
function a (b ) like :
The integral of the product a (b ) VV (e -µb ) gives the forward amplitude whi ch
is now non zero . The result is that the amplitude is
(U ·l)
G . Fox has noticed the empirical rule that set ting C • 1 works very wel l in
2 2
many cases . The amp li tude becomes m / ( t-i11 ). ( Such peaks are common in
rr 11
s trong interact ions , e . g . , p n change . exchange pn�np ) .
+ -
The difference between do/dt (yp+rr n) and do / d t (yn�rr p ) shows the presence
of p -like exchange . The ampli tudes in the two cases go as
rr n • rr + A2
+
yp � p +
yn � 11 p • rr - p + A
2
The rr and p interfere coherent ly because the p coup ling is mainly due to s pin
flip (see b elow) . The sign of coupling agrees also with what is e xpected b y
naive quark mode l . ( I t also agrees with the s ign i n a similar s t rong in terac tion
case np�n • rr+p and plf+'tll!' • rr-p . ) The expected energy dependence for various
t for the p is no t clearly seen . Perhaps for small t pions domina te . For
b y the fact that a ( p ) • 0 there (roughly

0p
t near --0 . 5 a dip is expected in yp+rr
a (p ) • 0 . 5+ t ) , In Regge theo ry the amplitude for a p traj ec tory exchange is
proportional to
-i rr a
( 12 . 2 )
( 1 -e ) a ( t ) -1
s 1n11a r (a)
s
-l
The po le i n (sinrra ) a t a•O is "killed twice" by the phase in the numerator
, so the contribution of the p t raj e ctory vanishes at t •

-1
and by ( r ( a ) ) --0 . 5 .
+ 0 0
The dip is seen more clearly in 11 p�rr n ; in y p+rr p the zero being in a double
flip amp litude may be obscured by absorp tion . In a s in gle f lip ampli tude i t
should show u p •
Asymme t ry for yp+rr n +

o.L - 011
0..1. 0(1
A • --- ( 12 . 3 )
+
o and 0 11 are the cross sec tion for photons polarized perpendicular and
.L
parallel to the plane of s cat tering respectively . Experimentally A looks like
1 ,I.-- . - --- -
1
(
i
'
.1
a '=" a , for l=t > . 2 a

11 ..L 11
for small t is smal l . This can be explained by con
a a 11 , a
s idering the exchanges which con tribute to .._ and ..L is contributed only
• e.g. p ,w ) a11
J
by natural parity exchange {P {-l ) is contributed only be
e .g . w) .
J
unnatural pari ty exchange {P • - { -l )
To see this cons ider the following picture :
pseudos calar
scalar
The photon comes in with polarization e in plane of shee t , if this is
reflected in a mirror in the p lane o f scattering everything looks the same
except that s ince there is a pseudoacalar particle the amplitude changes s i gn ,
i t i s therefore zero .
Cons ider now a ph o ton coming in with perpendicular polarization :
pseudoscalar
y
"'V\I V\J'\..f\
scalar
Again reflect in a mirror in the plane of s cattering , e changes sign , but
there is another change in s ign b ecause o f the pseudos calar parti cle so a
natural pari ty exchange can contribute to a.L . Similar arguments can be used
in the case o f pion exchange , generalization to traj e c to ries of higher angular
momentum gives our rule ,
The reason for the asymme try curve can now be seen as follows . At t • 0
a11 • a.L by geome t ry , as there is no s pecial axis . For ideal w exchange we
have the amplitudea

--
t (1 2 .4)
Au •
-
t-m11 2
� 0
for p we have A11 • 0 A _/ + O. From absorption ( t•O correction) we must have

A 11 • -1/ 2 A.J.. -1/ 2 to ge t zero cross section using G . Fox ' s rule for size .
( :)
=
The total is
t-+11111
An • � - .!_ • !. (1 2 . 5 )
t-m 2
11
2 t-m11
+ A1..P
1
2
A l..P should g o t o zero a t small t . Neglecting it w e expect the asymme try

I AJ. 1 2 - I A ,,1 2
A• (12 . 6 )
I A..J...1 2 + I A 1 1 1 2
t o start at zero a t t • O , rise rapidly t o 1 a t t • -m11 2 (which it does

experimentally) and to fall off toward zero for higher t . It does fall off
slightly , to perhaps 0 . 8, but not as much as expected ; presumably other terms
like A.J. P are beginning to contribute strongly .
yp+110 p
do /dt for this reaction shows a very sharp forward peak , it is due to
photon exchange (called Primakoff effect) , the diagram is
p p
we know that 11° couples to two photons from its decay . In fact, this effect
provides the most accurate way to measure the lifetime for 11°...-y y ; the exchanged
photon in the Primakoff effect is almost on its mass shell .
For t outside the forward peak , the reaction can come from some exchange -
we expect the w to predominate (see below) and to have a strong single flip
(flip at y11 vertex , non - flip at nucleon ver tex) . Thus it vanishes at t • O , and
does show the dip when a ( t) • 0 at t • 0.5, The energy dependence is entirely
wrong for small t , A plot of experimental a ( t) (from do/dt • s 2 a ( t) -2) shows a

near 0. 2 for all t out to 1 2 where it may be -1 . If w we expect yn+�0n to
t - .
0
be equal . The ratio yn+� 0n is 1 for small t , falls to 0 . 6 near t 0 , 5 and •
yp+� p
rises again to 1 at large t .
The asymmetry for polarized photon parallel a11 and perpendicular a.L to the
production p lane has also been measured . A • (a_._- a11 ) / ( °.i_ +o11 ). The cross
section is predominantly a.L. except near the dip when a11 may be as much as 20% ,
yp+K+A yp+KH
These reactions are due to K* and K* * exchange . A bit of K exchange at

lower t makes A greater than the E because the coupling KPA is b igger than
KPE (see below) .
yp+np
This is dominated by p exchange instead of w exchange , so no dip is seen

st -0 . 5 because of the double flip .
Quantitative fits have been attempted along these lines , but each needs
several complicating features like absorp tion , Regge cuts etc . It is not worth
our while to follow any of them in s pecific detail for no one model automatically
easily gives detailed fits to all the reactions .
What we can say however , is that in this large s region (at leas t for
small t ) the concept of t channel exchanges seems to be a guiding principle
that is fairly success ful . This same effect is seen in hadron collisions in
even more detai l ; i t is not a particular aspect of photon collis ions . There -
fore the discussion o f the next lecture will b e more general .
We derive now the relative strengthsof various couplings from the quark
mode l . The arguments above use these results .
To get these relative s t rengths we assume mesons couple through currents .
The currents are ob tained by taking the meson quark wavefunctions and in ter-
preting the quark states as creation and annihilation operators .
We have p • u-;; -dd, w • u-;j° + dd; hence p couples to a nucleon like a
curren t with u charge +l , d charge -1 . w
couples like u charge +l , d charge +l .
A photon couples like u charge 2/ 3 , d charge -1/3 or like 61 w + 2
1 p . To
a neutron the coupling of the p is reversed in s ign , the w is not . Therefore

yn 21 p + 61 = w which gives p • yp-y n , w • 3 (yn+y p) ,
Pseudoscalar Meson Photoprod11ctio11, High Energy 69
The couplings go like

to charge to magnetic moment
yp 1 3
yn 0 -2
p 1 s
w
3 3
Therefore p couples mainly to spin , the flip amplitude predominates but

for the w no such predominance occurs .
0
From w a uii - dd we get the couplings of w, p to y and w0 • Hence between
w0 and p0 (which have the same uu and dd signs) coupling the photon to u plus
d • •
2
photon to 3 + (- 3)
1
3•
1
to w it is coupling to u minus coupling to
2 1
d - 3 - (- 3> • 1, therefore
ir o wy •
1
ir0 PY 1
For the n • L (2 ss - uii - dd) the ratio is reversed

16
Pseudoscalar mesons couple to nucleon octet via su 3 wi th F • 2, D • 3 , thus

1
/2 (F-D )
( F�D )
Kl:N
iViN .
fl
n-T1-r---i mrn "'�

lffilT
0. <
0
<
<
!:::
t �'°/
{\j
< -
0.
>--
I
I <<
«\ +
�
q
I
I -- 0 >
0
C) '"""!
N
..;
p� f
10 <..O ...... �
0. �
<i'•l� >-- ci '-'
H
+ <l-i 0 "'
� -.
� --
/
/.
/
� 0 '
0
0 0 0
0 0
0
t - Channel Exchange Phenornena
Lecture 13
t-Channel E�change Phenomena

There have been a number of theoretical views aa to how to unders tand
these t channel exchange ef fecta and I will diacuas a number of them . They
are closely interrelated and are undoubtedly to a large extent , different waya
of describing the same thing , I will give a very brief discussion of each ,
indicating their relationships to each other and their various degrees of
incompleteness . Some of the waya of understanding t exchanges at large s are :
a) The geometrical or impact parameter viewpoint
b) The Regge pole formula
c) t-channel resonances
d) a -channel resonances ,
The firs t view (has been discussed mos t completely by Harari) notes that
as s..... , the angular momenta can get very high so that if the tth orb i tal
momentum is selected where t•kb is kept fixed as s increases we can define b
ever more accurately . Thia b ultimately becomes the 2 -dimensional apace vec tor
perpendicular to the plane o f collis ion and is the Fourier transform variable
corresponding to � the transverse momentum t ransfer vector .
71
i�·
Amplitudes A(.Q.L , s) can therefore be written as
J
b -
A ( Ck , s) • e a (b , ( 13 . 1)
and the behaviour of A is studied by describing a (b, s) using geometrical

intui tion to assist in unders tanding a (b, s) rather than trying to gather
int uition of A ( <l.1_ , s) directly .
Diffraction is readily unders tood by supposing there is an amplitude for
a particle to pass through a target without interacting which is very small for
small b , but rises to one outside the target .
A(b, s ) would look like
The energy dependence of this amplitude (nearly independent of energy) is taken

from experiment . (It is vaguely understood as a strong interaction of an obj ect
of fixed size . )
A reaction like charge exchange , say n -p+n°n is described by supposing ,
at fixed b that the n going through has some amplitude to change to a n° . (The
energy dependence of this amplitude may again be guessed from experiment , or
from another theory like Regge pole theory . ) This amplitude is larges t in a
ring
:�
--"-.&------ A
because for large b no interaction occurs while for small b there is lit tle
chance of getting through . In crudes t approximation we can say all comes from
a ring at radius R, leading in Fourier trans form to an amplitude like J0 (/:t R) ,
for no helicity flip . This expects a dip at t • --0 . 15 which is sometimes found.
t -Channel Exchange Phenomena 73
For a s ingle he li c i ty flip where the amplit ude mus t vary acro s s the impact
the result is J 1 ( .r.:t"R} . For R • 1 fermi this has a

i�
parameter plane as e ,
zero at t • -0 . S and is , in thi s model the origin of the dip at th is t in
pion exchange s c a t tering , phot on produc tion of ir0 e t c . (The dip that in Regge
theory we as s o c iated with the zero o f a ( t ) } .
Tb.is model does give , with some exceptions , a rough descript ion of the
general behaviour , but it is not a comp lete theory because the energy dependence
of amplitudes is not unders tood , nor are the coupling s t rengths .
The view that a particle is a point traveling at a fixed impact parameter
rather than a s tructure with f inite dimension in the transverse plane
( i . e . two pie plates co lliding) is an oversimpli ficat ion . At tempts t o deal
with this have s o far , lead to complications and amb igui ties and no clear
definite picture has emerged .
b} The Regge po le formulas result from a sugges tion (mo tivated by an analysis
of non-relativis tic s cattering by Regge) that the amplitude function for two
body react ion A ( s , t ) be s tudied as an analytic funct ion of the energy s and
an angular momentum variable in the t channe l , a . For large s the analy t i c
formula was sugges ted

a1 ( t )
A(s , t) • (E + /) Bi (t) s (13 . 2 )
i i
where a s um over special values o f i ( called poles) an d a continuum of i values
( called cut s ) is in general expected . Nearly any funct ion can be s o expanded
but the usefulne s s of the idea depends on making further assumptions . Eviden tly
as s..., the large s t a ' s would determine A, and i t was hoped that the poles would
have these , definitely above the a valuea where the cuts start , so A would
(t)
behave as S ( t ) a a or a sum of a very few terms o f this fo rm . If so, S ( t)
would be factorizable S (t) s ( t ) in a rea c t ion A + B + C + D, and a ( t ) if

AC 1D
extended to pos i tive t would be an integer for those t equal to the mas s of a
real par t i c le of appropriate angular moment um . I t is only a t s uch a t t h a t A
could have a pole like singular ity in t .
For certain react ions (e . g . w p+w 0n) these expectations have been
brillian t ly confirmad , but generally complications set in . For example , for
p ion exchange we saw how ab s o rp t ion , a concept from viewpoin t (a) of ten
modifie� the simple pole pic ture . In the "pure" Regge view any modi fication
of the simple pole formula for A(s , t) can be described, of course , by adding
cuts - but what cuts to add , and how strongly requires outs ide ideas , like
borrowings from the absorption model . The cuts star t , for large s , righ t at the
position of the poles and cannot be eas ily separated from them .
It is found empirically that a (t) is nearly a straight line . The reason
for this is unknown .
Also to describe the nearly constant total cross sections a special traj ectory
was invente d , the pomeron , with a special value one for a at t = O, without any
dynamic explanation of this part .
This theory is again incomp lete as no information is supplied fo r get ting
B ( t) , and , in practice more important , for determining the rise and behavior of
cuts . Attemp ts to avoid using the absorption model (conspiracies , " etc . ) have
failed .
Obvious ly some kind of integration of the ideas of a , b will be the mos t
profitab le .
c) The relation of the Regge form to the particles on the traj ectory can be
looked a t ano the r way (van Hove) . Suppose we have a theory which would give a
sequence of particles of ever increasing angular momentum . Then the Born terms
representing first order exchange of all these t-
channel parti cles will give
2 2
a Regge - like sum. I give an example . Le t the mass go as mn : mo 2 + nSl , where
m0 2 is the mas s 2 of a scalar , m0 2 + ".. of a vector , m0 2 + 2'1 of a tensor (spin 2 )
particle , etc . The exchange of these for a reaction A + B + C + D where for
s implicity A, B, C, D are s calar , goes as
c D
s2 + • • • ( 1 3 . 3)
-- 2
t-m3
e at
The s for vector comes from (P A + P c ) e at the upper coupling and (P B + Po) •
the lower summe d over e . Likewise s 2 for the tensor comes if we use the full
two tensor indices to carry informat ion about P8 + P D from vertex to couple to
PA + PC to make ( ( P
2
A + P cH P8 + P D )) 2 or s in the limit . e n are squares o f
coupling cons tants . Le t m 2 • m 2 + nn , and let e equal l/n ! say , our sum i s
n o n
1
n!
set ting u • se and (t-m ) /n • a(t) the integral can be trans formed to
-nx 2
0
1 ( t ) s -a (t) - 1 -u
A • s a l0 u e du •
- 0
For large s the upper limit on the integral can be put m, to ge t the asympto tic
form
- fr s a ( t ) r (-a ( t ) ) (13 . 4)
res idues of this function r ( -nt-c ) • 1::!L. reproduce

n!E
The poles and
the original series . In general any Re.gge term can be expanded by its poles
in t (coming from S {t ) and the series is then a van Hove series determining
particles and couplings . Or some guess can be made of states and residues and
a Re.gge term derived . These two forms are mathematical express ions of the same
theory , and permi t di f ferent points of view of the same th ing ; sometimes one ,
sometimes ano ther is more sugges t ive of a new idea .
Comments
s
For negative s the series ( 13 . 3) gives an amplitude rising as e . It is
sugges ted that the coe f ficients canno t fall as quickly as l/n ! Al ternatively
one can say that the values for negative s are taken from the asymp t o tic form
for pos it ive s by analytic con tinuation .
Clearly in t h is sum, for negative t , we are far away from the true resonances
and are again in the position of not knowing what the real sum should be unless
we have a comp lete theory . Such analytic extension formulas are hard to use
practically - for example what does observa tion of the firs t two resonances of
the serial tell us about the asympt otic form?
- ---i + ? • ?
l
-z s
t -m t m
0 - 1
If there were no other terms the asymptotic s behavior would go as s . Yet

2 2
from m , � we can get a lit t le information on the behavior o f a ( t ) . How does
0
it work ? I don ' t know - the analy t i c extens ion o f approximate functions , o r
empirically partially determined funct ions is a mathemat ical p rob lem I d o not
unders tand nor trus t , nor can I guess errors . To show the type of di f f i culty ,
a ( t)
suppose a s um o f t channel resonances gives Bs ; now change the coe f f icient
i • l term by 6C . +
o f the Sure ly the s um is now Bs
a (t) ( d C ); wh ich for
t -m
1
large s goes as s . The point is a modi fication o f the second term only is not
"na tural" or "smo o th " and changes the result dras tically . Whatever phys ically
causes the change in the f irs t term changes all the other terms a b i t in s uch
a way that the asymp t o t i c f o rm is only mildly altered .
To s umma r ize the van Hove sum is not really us eful except theore t ically .
If you have s ome theory of all the t -channel resonances and their coup lings then
you can b e assured the theory will lead to Regge asymptotic behaviour - or i f
y o u have s ome theo ry leading t o Regge asymptotic behaviour you can , b y looking
at the po les ( in t) in i ts exp ression deduce where t -channe l resonances are
imp lied ( a ( t ) • intege r ) and something ab out their coup lings .
Summing the t-channel resonances sugges ted by the harmonic os cillator
quark mode l for meson resonances (QQ) does give an expec ted Regge behaviour in
th is manner . De tailed numerical comparisons have not yet been made .
Lec ture 14
d) s -Channel Resonances
Anothe r theoret ical idea is that all o f the behaviour at high energy also
can b e unde rs tood by j us t cont inuing our s channel res onance analys is (e . g . by
Walker ) to h ighe r energy . A behaviour like a dip f ixed in t , as s in creases means
a feature that appears at gradual ly decreas in g angles e • t /s so that a sharper
and sharper angular dis t r ibution results . Th is me ans that higher and higher
an gular momentum a-channel resonances must be con trib u t ing as s increas es . But
this is , of course , all right for we know higher angular moment \DD resonances are
available a t h ighe r s . The approximate s t abi lity o f f { t ) curves as s varies
does imply a great deal ab out h ow the resonances and their coup lings mus t behave
at large s .
t -Clra1111el Exe/range Phe110111e11a 77
That the entire curve comes f rom s umm ing s chann e l ( and , some wo u l d say ,
u channel) resonances only is a tremendous res t ri c t ion wh ich , i f t rue , mi gh t
b e s o s t rong a s t o guide theory t o the one an d only exact expres s i on that can
do this - thus , much a t tention has been paid to this idea . Mos t s t atemen ts
are , a t f i r s t , no t that ambit ious , because comparison t o expe riment again is
meaningless i f we have to decide on the h yp o the t i c al value of a s um of
resonances far f rom their resonan t ene rgies . Ins t ead it is noted that t h e
+ r / 4) is on ly large near the res onance

2 2 -I
imaginary p a r t of a re s onance r(E-E )
r
so the p rincip le is s t udied in the form
I� A(s , t) • E rm ( s - channel resonan ce s ) ( 14 . 1)
excep t perhaps for the pomeron .
Such ques t i ons as the con tribut ion of u channel res onances (which a re
always o f f res onance so the imaginary p a r t is very smal l , zero for infini tely
narrow resonances ) and whe ther the pion exchange term (which is real) should
be inc luded are s i de s tepped in this way .
I t mus t be emphas ized that a -channel poles are exp l i ci t ly mean t , For i f
an in tegral ove r a con t inuum i s also included nea rly all f un c t ions A ( s , t ) can be
s o rep re s en te d .
I f we add the p r in c i p le (duali ty ) that the t -channel behavior (resul t ing
from the s -channel poles ) mus t i f analytically extended , have poles a t the s ame
2
mass for the same quantum numbers as the s -channel resonances we have s u ch a
seve re res t r i c t ion that p e rhaps only a nearly unique solution exis ts , and hence
a solu tion of the s t ron g intera c t ion phys ics could res ul t . Much work has been
done in this d i re c t ion - success h as not been a t t ained . Naturally the p rinciple
might be wrong -- pe rhaps s -channe l and t-channel resonan ces mus t be separately
added - or what amoun t s to the same thing s channel cuts are necessary if the
amp li t ud e is to corre c t ly represent the h i gh energy behavior in s and t . There
is no phys i cal p rinciple unde rlying the duali ty hypotheses - it is simply a
guess at a p o s s ib le mathematical prope r ty of an amp li tude , It is an ob servation
at low ene-rgy that amp litudes are dominated b y resonances , s o that some so-rt o f
infinitely narrow res onance approximation may b e a good s t a r t ing poin t . But
is this also valid a t high energy where in most collis ions many parti cles are
p roduce d ? Can they be unders t ood as the prod ucts o f cas cade dis integration of
resonances or is the re an important contribution o f direct coupling to the

continuum?
We remember that at high s the two body final s tate we have s tudied seems
to be only a very small part of the cross section . The multibody states which
cons titute the maj or part of the collision cross section will be discussed later .
Veneziano formula
Such ideas are encouraged by the exis tence of a function (Veneziano) which
can be expanded entire ly in terms of s-channel poles - or t-channel poles , and
which has Regge behavior at high energies .
A(s , t) f (-a (s ) ) f ( -a ( t) )
=
(14 . 2 )
r (-a ( s ) -a ( t ) )
where a (s ) i s a straight line in s .
The res idues at the poles in some generalizations are not necessarily
always positive in cases where they should b e .
This i s only a pos sible form for A(s , t ) i f the s cat tering particles are
spinless - much work without clear conclus ion has been done in trying to
generalize this . Also to generalize from 4-point to n-point functions .
A fundamental problem is this : This is a narrow resonance approximation .
How precisely are we to correct i t ? (The same question w e discus sed for the
quark mode l , ) Again perhaps the same ques tion in another language - this is
pres umab ly not directly the formula for A(s , t ) but for a deeper amp litude
that mus t s till be , as we might say , "corrected for absorption . " Do we have
the Born term of some expans ion ?
Attempts to extend duality or Veneziano ideas to photoelectric couplings
have been singularly unsuccessful . However , the problem of the mathematical
behavior of photon couplings should not be disregarded . They are especially
important because they contain beside s , t a new variable q 2 the mass 2 of
the pho ton (which may for example be virtual via electron scattering) . The
amplitude A (s , t , q 2 ) has analytic properties in all 3 variables s , t and also
q 2 (we shall study them later) and this increase in res trictions may be very
helpful in sugges ting forms or theories .
Finally we can ask whe ther a dynamical model (such as the 3 quark model)
of hadrons could be expected to lead to j us t the sum o f resonances and couplings
needed to generate mainly forward scattering . I think it is very likely to have
no dif ficulty . We may draw the following diagram :
If we suppose the photon and pion each couple to one quark - but each to a
different one , then quarks a and b have large momenta and the amplitude that the
final configurat ion a, b , c is in a s imple pro ton s tate , where they presumably
have small relative momentum is very small . Thus at high energy the most
important case is that a single quark a receives the photon and emits the pion :
a b
The net momentum c gets is only the momentum trans fer k-q and the ampli
tude a b c is a simple pro ton is large if , and only i f , this trans fer is small .
Therefore the pion goes forward in a way characterized by a function of the
momentum trans fer .
At any rate a dynamical theory of s channel resonances alone might not
find it too difficul t to ob tain apparent t-channel exchange effects .
Estimates of Coupling Constants
In a number of cases we es timated s izes of coupling constants . For
example , we said p -exchange was mainly flip from a nucleon . Such things come
from theory outs ide the impact parame ter , or Regge theory which left such
cons tants undetermined . They could come from o ther experiments (e . g . flip is
large can be go t from � nucleon scattering) but nevertheless theories are
availab le for what to expect . We used mainly su 3 , but even here F/D ratios
must be known - and these are given well by su6 , or what I have called a
quark model estimate .
The "quark model" used here is really not the same as the harmonic quark
model used for the s resonances applied to t resonances . We are not dealing
with t resonances but rather with t traj ectories . The quantitative theory o f
trajectories to be expected from any (in parti�ular from the harmoni c ) quark
model has not been worked out , Rather we are us ing the "quark model" j us t
t o count - t o make up su 3 and F/D rules ,
Even then we implicitly supposed some thing that many people have used
{but to our knowledge has only las t week been explicitly formulated by
Kis l in g er ) . We assumed the p (or w) traj e ctory was coup led in the s ame p roport ion
as currents are coupled (o f the appropriate quantum numbers I • 1 , 0) . This idea

usually comes in two s teps . (1) the vector mesons on their mas s she ll couple
like currents and (2) the p ,w,+ t raj e c t ories couple like the p ,w,+ :.iesons do .
Stated more precisely consider the amp l i t udes for the fo ll owi n g diagrams .
A x , ·B
,/"
//
x y y
The ratio of these amp l i t udes is l l I < D l j l C> wherc j is the current with
p quantum numbers . We have assumed the amp li tudes have no t been correc ted for
ab sorp t ion .
Ano ther formul at ion of this "quark counting" game is to draw diagrams for
t exchanges be tween hadrons like
implying that baryons are QQQ and mesons are QQ in the usual way but that the
in teraction can be largely under s t oo d as a sum of terms each rep resenting the
scat tering of one quark in one hadron wi th one in another . (Lipkin) .
It is clear tha t our interpre t ations of highe r energy s cat tering comb ines
several ideas . What is evidently lacking today is some overall view that com
bines these nicely togethe r . I n the past too much at tention has been paid t o
forcing the ampli tude into one mold or another rather than combining the more
well es tablished and cent ral ideas o f each of them into some synthesis .
Vector Mesons and Vector Meson
Dominance Hypothesis
Properties of Vector Mesons
Be fore we discuss production of vector mesons by pho tons we first discus s

the s impler prob lem o f photon-vector meson coupling.
Electron production of vector mesons

-
In s colliding beam experiment we can measure e++e + hadrons , This is
interpreted of course as a virtual photon of positive q 2 • ( 2 E cm) 2 and gives
us a direct measure of
<hadrons ! J (q) ! O>

)l
for positive q 2 This matrix element is dominated by three resonances at

suitable values of q2 corresponding to the p ,w,,
-
neutral mesons of spin 1 ,
j us t like the photon .
+ -
For example if the hadrons is a pair of K ' s , K K the cross section is
very small unless q2 is close to (10 2 0) 2 (ie m , 2 ) and then rises way above
background in a beauti ful Breit-Wigner curve of width 4 . 7 t . 7 MeV . The
interpre tation , of course , is that we have gone through a resonance , the ' via
82
Vector Mesons and Vector Meson Dominance Hypothesis 83
a diagram like
To analyze this have two ampli tudes , one purely hadronic <K+K- 1 4>> and
we
2
r-?"
4 2 e «jl j J j O> at the mass of the
a pho ton amplitude which we write as lf4'1fe- µ µ
4> , e is the photon po larization . The second factor may be writ ten
I'
<fj> j J j O> "' F9 elJ where e µ :!.:i •he pu.i.d& � ot .. �ion vector of the vector particle .
JJ
(There is no generally accepted convention for how to write

F.,. . One way is to write m /g4> • F4> for what we have called r, , others write
2
4>
2
m / 2y so y is r 1 g , but still others use the same letter y"' for another
.,. ' ' .,.
way of expressing the coupling . )
The experimenters avoid all the ambiguity by noting that the d iagram above
implies that a free 4> meson would have a certain rate to disintegrate into
e+e - . They give this rate (which comes more directly from experiment anyway)
r fj>+ e+e
- 1 . 36 t . l x 10 -6 GeV
=
either directly or as a branching ratio . A s imp le calculation shows the
connection ( for any vector meson V of mass mv ' neglecting electron mass )
(a •l /137) i s
-3
2
a2
rv+ e+e - • - 3- Fv mv
2
mv (14 . 3)
4na s -
3
For the 4> , F4> • .080 (GeV) or g4> 2 /4 n • 13 . 3 ,

2
One can look in this energy region for other products (such as 3n , or n°y )
and find the resonance again - thus determining the hadronic amplitudes for
fj>+3n or 4>+n ° y , usually given in the form o f branching ratios .
In the case the final state is 3n there i s another resonance at 785 MeV ,
the w meson - it is studied in the same way .
Again studying Zn final s tates a large resonance near 765 of width around
125 Me V . This large width makes considerable ambiguity ( in assumptions of how
r q
z
varies with ) in determining the cons tants . mass and width . Further the
curve is asymme t ric , sharper on the high energy side . A small shoulder is
nearly apparent - this effect is interpreted as interference with the w
resonance assuming there is a finite amplitude for ..,.+Z11 . (This violates isospin ,
and is an electromagnetic effect which we will discus s in the next lecture . )
Values for the various constants for the vector meson may be found in
the Particle Properties Tab le , There is now some additional data from the
Orsay Storage Rings (J , Lefrancois , 19 71 International symposium on electron
and photon interactions at high energy) . Some of the differences from Particle
Table results are new data but some are due to an altered way of reducing data ,
especially for the p. We shall have to wait for new particle tab les to thrash
these differences out .
B • Branching ratio
New Particle Tab les
4> B { cf>+n°y) • (Z . l :!: . 7) x 10 -Z B ( tj>+i(IC-) • 46 . 4 :!: Z . 8%
B (tj>+11°y ) • ( . Z5 :!: . 0 9 ) x 10 -z B ( 4>+ 1S_ - 35 . 4 :!: 4%
�)
• :!: Z . Z) x 10 :!:
B { tj>+311) ( 14 . 7
-Z B ( cj>+311) - 18 . Z 5%
r tj> total = 4 . 7 :!: . 7 MeV 4 .0 :!: .3 Me V
w B (w+11°y) a .0 7 :!: .OZ B (..,.+11°y) • 9.3 :!: 1 . Z:C

B (w+n°y) Not seen < .oz B (..,.+311) • 90 :!: 4%
rw total =
9 . Z :!: 1.0 MeV .. 11 . 4 :!: 0.9
B (..,.+Z11 ) • . 04 :!: .OZ (Phase 87° :!: 15 ° ) B (..,.+Z11 ) • . 00 9 :!: . OOZ
r (w+e e ) . 76 :!:
+ - . 0 8 keV
• :!:
Z
:!:
z
p mp • 780 6 mp 76 5 10
r • 15 3 :!: r • :!: zo
p
total 13 MeV total 1Z5
p
r +e e •
+ -
r +e e :!:
+ - . 5 keV 7.5
p
• 6.1 :!: . 9 keV
p
Calculating from the latest Orsay data . we get

z
g /411 .. Z . Z 7
p
gw /411 • 1 8 . 3
z
z
g /411 13 3 • .
4>
but Orsay makin g "corrections for finite width" reduces the data in some other
way :
g p 2 /411 2 . 56 • :!: . 22
2
g w /4 11 - 19 . 2 :t 2
g ' 2 /411 11 . 3 � :t 0.8
Note . Now that we have data for the rates t+ny and .,..11 y we can compare the
predictions of the quark model with harmonic dynamics of Feynman , Kislinger
and Ravndal .
Quark Model Orsay
r � ny = 1 . 79 x 10 -4 GeV 1 .0 :t .3 + 10 -4 GeV .
r w11y • 1 . 92 x 10 - 3 GeV .6 :t .2 x 10 -3 GeV .
Lecture 15
We will make a di g res s ion from our main discussion of vector mesons to
consider the interesting feature of w

- p interference .
The reason w can go into 211 is that the s tate l w> is not pure l w0> •
_!. (uu + dd) isospin zero but has a small admixture of I p 0> • ...!. (uu - dd)
rz rz
isospin one in i t . This mixing is due to electrodynamics .
We consider the w - p system as a two -s t a te sys tem where Cp0 is the ampli
tude to be in a p0 state and Cw0 is the amplitude to be in an w0 s tate .
( 15 . 1)
the mass matrix is
H •
( m
w
- lli
2 6 )(
- 784-i6
-2-
irp
6 m - 6
p
therefore the true w is
l w> • l w 0 > I P O>

6
mw -mp
+
i r w-irp
2
_
l w > • l w0 >
+
6
19+56i l p o > ( 15 . 2)
o can be given in terms of the branching ratio for w+2n
I
o p +2n
I • 2 .97 x 10 -3 o 2 MeV -2
Prob . w+2n
Prob . w+all • 19+5 6 i w+all ( 15 . 3)
From Orsay experimental data the above branching ratio is 4 t 2% we therefore

obtain l o l • 3 . 7 t . 9 MeV .
The phase of the w-p interference is determined
by fitting the "shoulder" in o (e+e- + n+n -) near the p resonance ; the value
is 87 ° t 15° . If o is negative the phase of o / ( 19 + 5 6 i) is 109 ° , which is
in fair agreement with the experiment .
We can understand the mixing as due to two effects : ( 1) The electromagnetic
energy of uu is not equal to the energy of dd because of the difference in self
energy of the objects and the difference of energy of interaction . (2) There
is a contribution due to annihilation p + y + p , w + w • Y + w0 P + Y +
0 0* 0 0 0
We can get estimates for the two above effects from the knowledge of K + -
K *0 and p+ - p0 mass differences and from Fp • Let the self energy of d be a
and the mutual energy of dd be -b ; then the self energy of u is 4a and the
mutual energy of uu is - 4b because the photons act twice on doub le the charge .
+ ud-, K *+ • us- and K *o ds-
From w __! (uu + dd) , p o • __! (uu - dd) , p
• =
0 12 n.
we ob tain the following electromagnetic self energies
p 0 : • (4a + a) - 21 (4b + b) • 5a - 5b /2
w0 : • ( 4a + a) - } ( 4b + b) • 5a - 5b /2
p+ : • (4a + a) + 2b • 5a + 2b ( 15 . 4)
• 5a + 2b
*
K o • (a + a) - b • 2a - b
But we also have a matrix element between 

 • (4a - a) - } (4b - b) • 3a - 3b /2
from this alone the mass matrix is
w
0
Sa - 5b /2 3a - 3b /2
w0 3a - 3b/2 Sa - Sb /2
S ince K *+ - K *0 • 3a + 3b
+ 0
p - p - 9b / 2
*+ - K *0 ) - (p + - po ' )
we have 3a - 3b/ 2 • (K (15 . 5)
in terms o f measurab le quantities .
o' o
No te that in eq . (15 . 5) we have writ ten p · instead of p , this is only
to indicate that p o ' does not contain the contribution from the annihilation
2
term which we calculate next . To first order in e the change in mass is given
by
2
llm 2 2- <o j J j p >
-
4TTe v v
p
- (15 .6)
q
or
4TTe 2 F 2
llm 2 . - z- (15 . 7)
p m p
p
= 1 . 53
p
llm MeV
Let x be the amplitude for _!. (uu - dd) .,. , the amplitude for _!. (uu + dd)
./2
y .,. y
./2
is x/3 ( see lecture 16) . The mass matrix due to the annihilation term is there fore
proportional to
Po w
0
2 2 ( 15 . 8)
Po x x /3
;
2
w
0 x /3 x2 9
We already have one entry of this matrix , lim

p
z 1 .53 MeV we therefore have
for the full matrix
w0
1 .53 . 51
MeV (15 .9)
.51 . 17
Adding the contributions from the electromagnetic self mass and from the
annihilation term to the non-diagonal matrix element we find
01
5 = .5 MeV + (K*+ - K*o ) - (p+ - p ) ( 15 . 10 )
and s ince
( 15 . 11)
5 • -1 . 0 2 MeV + (K*+ - K* o ) - ( p+ - p o )
0
from data in the particle tables , (K*+ - K*0 ) • - 8 ±3 MeV and ( p + - p ) =
-
2 4 ± 2 . 1 MeV .
. But we do not trust these results , especially the p mass
difference , we only say that they indicate that 5 is likely to be negative .
'
. r;;:i
The yp coupling of the form y+ Tre- F e lJY e ' is no t gauge invariant and
lJ
2
would give a finite mas s for the photon via
y p y p y
�
The coupling is A YB P • However we can couple the fields Fµ v of the photon and
lJ µ
p , for examp le . It leads to
( 15 . 12 )
which is evidently gauge invariant . S ince eY . q is always zero we see our
pole behaves like

2
(15 13) .
The latter term has no pole singularity and is "los t " in the background of
other than pole term and effectively we can use a residue j us t proportional
to ey • e p ( i . e . 2 C p mp 2 Fp ) •
=
Lecture 16
Vector mesons (continued)
We now return to our discussion of vector mesons . We can see what su 3
2
gives for the ratios of the couplings 8v given at the end of lecture 14 . We
use the simple quark model ( for cowiting only) and assuming the + is purely
st range quarks , J couples to QQ pair proportional to its charge
p • ..1. (uil - dif) + - ..1.

12 12
(IJ - ..1. (uu + dd) + ..1. (- - ( 16 . l}

12 12 (� + t» _
3 1_12
• • SS + - 1/ 3 • - 1/3
Noting that these are values for the coppling F , and that g is the reciprocal
we f ind
: g (IJ
-2 -2 9 : 1 2
: g' :
V arious people have tried to correct this for su 3 breaking , but nobody
really knows how, There are two questions . The first is , to what extent is
the ' pure ss? I see no way to determine that . The low value of the ' + 2w
b ranching ratio is interpreted in the quark model by saying that the ' being
made of purely s trange quarks finds it hard to go into non-strange obj ects (w) ,
If so , all the amplitude to go to wy comes from an admixture of the w s tate ,

(uu + dd) / 12 in the + wave funct ion . In this way the amplitude to find w in
the ' comes out to about . 10 , although there is some wicertainty t . 0 3 as to
how mass factors enter . This is only 1% probability , the s tate is 99% pure ss ,
k
P roblem. Ma e a theory to es timate how much <j> + 3w would be expected if
it is due to the fact that <j> has a small admixture of w state ; assuming (for
no excellent reason} that ss cannot go to 3w .
More important is the question o f su 3 break ing because the masses are not
equal , For one example of ambiguities, should we compare F ' s directly with
v
the su 3 predicted ratios , or should it be Fv /m , or F /m or what , which
2
v v v
should bear the s imple ratios? su 3 cannot tell us and nobody really knows how
to calculate i t , although various guesses have been made . (I prefer F,, / ll\,
for a theoretical reason , but one which is really not profowid or necessary} .
These ques tions probably do not strongly affect the p ,w ratio , as the masses are
c lose together . As you see the ratio 9 : 1 is not very far of f , (and , in spite
of the wicertainties , the ' is also about right ) .
Vector Meson Dominance Model

To summarize we exhibit the photon to w+ w - amplitude near the pole of the
p (the only one of the three ' ' w, p which couples to the pion) as
1
""'" fp 11 11 (p l11+p211 > q2 2+ir m
-m
F
p
p pp (16 .2)
�
m2
{f p 1111 / gp ) 2 � {p
-mpq +Lr P mP l11+p2 11 >
when q2 ::::. mp 2 • One would ordinarily expect other terms {such as a direct
coupling of photon to pions or various other "intermediate states") to be added
to this , lbis , of course, would not change the behaviour near the pole (which,
unfortunately for the p is not so definite as the width rp is rather large, which
makes many difficulties in practice in fitting it, nevertheless in this theo -
retical discussion we shall neglect it) . Another way to say it is to suppose
the numerator has a factor t{q2 ) which varies with q2 such that t{mp 2 ) · � or
again to say the "constants" fP 1111 or gp vary with q2 only their value at q2 •
mp 2 being defined. All these ways say the same thing so we shall not argue
about them.
Nevertheless a bold hypothes�.s has been suggested (vector meson dominance)
that this expression with fP 11 11 /gp constant is all there is to it - there are n�
other terms. I see no good physical reason for this. It is made in analogy to
another mysterious hopothesis of the same kind, PCAC , that works . lbere we do
an analogous thing with a pion pole - but only extend it from q2 m112 .02 • •
at the pole, to q2 0, Here we suggest eq. (16 . 2 ) is valid not only near q 2
• •
mp 2 but also for all q2 or at least to q2 O . •
We know for long wave lengths the pion looks like a simple point charge
so that as q2 0 (use crossing to convert <11+11- I J l o> to <11+I J 1 11+> we have
+
II II
2
< 11+ I J11 1 11+> • (p1 + p 2 11 > for q + O ,
11
(16 .3)
hence for (16 .2) to be true we must have {neglect problems generated by rp ,
or suppose rp depends on q 2 and vanishes at the (2m ) 2 threshold and below) .
11
• 1
Experimentally fp21111 / 4 11 • 2. 43
Vector Mesons and Vector Meson Dom inance Hypothesis 91
= 2 . 56
well wi hi the 10%
t n \lll ce r t a i n t i es t q u t ti s .
in d e t e nnining h ese an i e
2
It would also imply that the form factor F l q 1 for the pion, written Tr
< n+ I J I n+ > • F (q 2 ) (p 1 + p 2 ) would, for negative q 2 , have to be exactly

Tr
µ µ µ
(l+(-q 2 )/mp 2 ) -1 . We shall lo k o at what evidence - there is l a te r so far i t
looks very good,

Th e i dea is extended t o the case o f interact ions wi th an y s t a t e - f o r
example in interactions with the nucleon the part due to isovector would be
<N I J I N> isovector
2
..£....
l m
( 16 . 4)
µ gp • -
p with polarizat i on µ . <N j p l N> , the c up l i g

The t e rm ul o n of a p to a n c eon
is defined only at the p pole, and there has an electric and magnetic part so
we can write as <N j c1y µ + c2 f (y �-(y µ ) IN> Thus , expressing the lsovector
•
part of the current coupling in terms of the usually defined fonn we f a c t ors
1 )
( 6 .5
Thus we predict that F 1 /F2 • c1Jc2 is independent of q2 (fair) and F 1 varies

as l / (l -q 2 /mp 2 ) . This is o
� t ter l/(l-q2 /mp 2 ) 2 is
n t good , b e t ter . Here we
are testing (1) for large negative q 2 - far from the p pole and it fails .
Obviously VMD canno t be an exact principle for all q2 But maybe it is nearly
valid from q 2 mp 2 to O . We must discuss extensions of the idea and
• f u r th e r
evidence.
A further conclusion from the nucleon case above is that c 1 /g • 1. There
p
is no way to get he p t up ling he nucle n , c1 or c2 , without ambiguity
co to t o at
present (or ever? is it a matter of definition <N j p l N> cannot have all the
particles physically mass h l l ! ) . Some kno ledg ,
on the s e nN scattering w e from
for example, exists for the ccupling of the p-like trajectory - or for what we
call p-exchange - but the extrapolation from p-trajectory to p-pole exchange
is ambiguous . The latter (p-pole exchange) has an amplitude varying s but as
the terms associated with p trajectory exchange do not , and they are not supposed
to in Regge t or ts l f .
he y i e
A similar expression using the w trajectory is to give the isoscalar

coupling, Since the mass of the w is nearly equal to that of the p , the
isoscalar and isovector form factors should have the same dependence as q2
they do even though this dependence is not the (l-q2m 2) -l expected .
p
Lecture 17
Vector Meson Dominance Model (continued)

In greater generality all three neutral vector mesons may be involved.
For example, in coupling to an arbitrary hadron state we write
gv ( g2 -m ;+if m ) <Hadrons / V>

<Hadrons /J / O> -m 2
)J
l (17 . 1)
v vv
V with polarization µ
(or of course by crossing for any transition hadrons X hadrons Y)

+
(l7 .2)
That the left side of (17 . 2 ) gives the total of all the ways that J µ can
couple is the general statement of the hypothesis of vector dominance. We have
already seen how it predicted the p n n coupling constant. Can we do a similar
thing for the $KK coupling, say $K+K- ? We know that the current operator at
q 2 � 0 might be written (p1µ+ Pz µ > with coefficient 1 because the charge is 1 ,
so we have the sum o f lg (K+K- / V) muat equal 1 . Thus i f we write
v
(l 7. 3)
we would have
(17 . 4 )
The problem now is that although we do know the three gv (from experiment) we
have too many unknowns in ( 1 7 . 4 ) to relate it to experiment . (The rate that
$ +K+K- implies fv + - / 4 w = 1 . 4 7 ) , We need some rule to partition the bits of
K K
current between p , w , and � .

A very suggestive thing to try is to express the quark model idea that
the $ is pure ss, and thus couples only with the s quarks in system. This
a
suggests we invent three currents Ju , Jd , Js which couple only to the u quarks ,

Vector Mesons and Vector Meson Dominance Hypothe.fiJ 93
the d quarks , or the s quarks (as though each had unit charge) , Then , for
example , the electric current (charges on u, d , s , are + 2 / 3 , -1/ 3 , -1/ 3

respectively) is :
J el • .J. Ju _ .!. J d _ .!. s

3 3 3J (l7 .S)
1 1
p jj (uu - dd) , l2 ( J - J ) ,
u d
Then , the being we define a 11 p - type current " aa iP •
.!.
likewise an w-type Jw • 12 (Ju + Jd ) and a ' type , which since ' • ss is jus t J' •
Js hence for example
(17 . 6 )
el
Now obviously w e suppose the three p ieces into which J i s split i n (1 7 . 6)
are the corresponding vector meson resonant pieces of (17 . S) . This may be writ ten
in a very s imple way , if the gv really have the ratios of the quark model (Lecture
( -m;2
14) we are saying, of course
<Y j Jv j X>
µ µ
• cans t , <YV I X> ( 17 . 7 )
q 2 -m +ir m
v v
where the constant , the s ame for all quarks is /2/g •

p
we
u d s
However s till mus t define i n an experimentally definite way what J , J , J ,
or JP , J w , J f are . We can use isospin , hypercharge and quark number (or
rather , baryon numbers ) as three conserved quanti ties (each definite for a
particular transition X � Y) to serve ins tead of quark numbers . Thus let J Z
represen t the curren t of Z-component isospin ; JY current of hypercharge , JB c urrent
of baryon number (equals 1/3 times the current of quark number) to write
Ju . i + .!.2 Jy + JB
z
Jd • -J + .!.2 Jy + J B (17 , 8)
JB - J
y
+ JB
(because if you substitute for example the quantum numbers of the u quark , Iz •
+ 1/2 , Y • + 1/3, B • 1/3 for the currents you get one for J , zero for J
u d
and
s
J , etc , )
Hence
(17 .9)
lb.is then defines precisely what we mean by the currents in (14.5) and
(14 .6) .
For example, let us apply this to the K+K- de ay of the + . We need the
c
couplings of J+ to K+ (we also calculate JP , Jw for completeness) . For the K+ ,

Iz + 1 /2 , Y • + 1, B
• o , so at zero q 2 ;
•
(17 .10)
Thus combining this with (14.9) and (14 . 7) we see we mus t have
fwK+K- /gw !.6 '

- .
defining how the total of (1) i n {17.4) is partitioned. A t any rate we predict
(f + 2K+K- /4n} /(g+ 2 / 4n) 1 / 9
•
2
which agrees perfectly with the f
+ KK/4n • 1.47 from + + K K and s, 2 / 4w 1 3.3
+ -
a
from + -+ e+e-.
How s low is + + J n compared to w + 3n7 ' goes 18% t o 3n, b u t r is only 4 MeV
so the partial width is only . 7 MeV while i t is 10 MeV for the w . Phase s pa ce
is lar ger for the + but more detailed analys is would require knowledge of how
the matrix e lement varies , If e is the polarization of the vector meson
and P1 , p 2 , p 3 the four vectors of the three pious the amplitude must be cons tant
e e p p . The " cons tan t" for the ; has to be about 0 ,l of i ts value
µ vap µ P lv 20 3P
for the w .
Lecture 18
The + appears to act as pure as . What is the significance of this - or
rather how can we de fine this in terms of quantum numbers or rules without
referring to the quark model? We don't know how. We try to say that it couples
weakly to states which have no strange particles in them, so not to 3n but yes
to KK. Yet such an idea "a state that has no strange particles in it" is not
readily definable . Due to strong interactions a state like KK has no overall
quantum numbers to distinguish it from 3n ; in fact via virtual interactions
the KK should couple strongly to 3n (for example through a virtual w , KK ++ w +-+ 3n . )
There fore the ques tion remains , what keeps the � from coup ling s t rongly with 3n 7
From point
the of view of a to that s state
quark model we might t ry s ay the s
is selected from the !..._

12
uu+dd) state of the w simply by the fact that s quarks
carry a larger mass (or a different interaction energy) than non-strange quarks .
So in lowest order of perturbation theory the eigenstates of energy are ss
and !..._ (uu+dd) . But then we find these states are unstable, and as a kind of
12
perturbation they decay, it tums out with small widths . But this "perturbation"
is not small - there is a coupling of $0 and w0 to every meson state (with the
w coupling equal to - !..._ of the $ coupling) . Thus for these virtual meson
12
states for example $ + KK + w there are diagrams
which mix the $, state. These diagrams cannot be calculated, (the above
w
diverges quadratically) and no calculation of such virtual strong interactions

in any problem has ever been successful. But usually the quantitative idea
that if the states can be connected via strongly interacting virtual states they
can go into each other is valid. The coupling constants f $KK and fwKK are so
large they would seem to have effects of order 1 , that is, to strongly mix the
original pure ss and !..._ (uu+dd) •
12
Of course it is always possible that the world is simply complicated,
that the ss
comb ina tion but the
is not i s o lated by principle in the beginning ;
very virtual interactions select some linear combination of ss and !..._ (uu+dd) to
12
be the $ (and the orthogonal combination for the w) and it just comes out that
the combination is ss .
Then perhaps it is really possible to understand why $ + ny is so small,
but is the smallness of $ + 3n really obvious in this model of the state?
But still more striking is the fact that this "accident" is repeated again!
For the mesons of spin 2+ where the f (1260) goes nicely into 2n or 4n and into
KK. The f ( l514) goes predominate ly into Kif and even into KK*+ KK* and no t into
nn in spite of the even larger phase s pace , (The quark model rates (Feynman ,
Ki slinge r , Ravndal) us ing pure !._ (u'U+dd) for the f ( 1260) and ss for f (1514) are
ri
all high , b ut the relative proportions are right . )
The experimental and quark model (FKR) rates for 2 + mesons are :
Experimental FKR
f ( l260 ) + nn 120 MeV 244 MeV
Kif "' 8 13
2 n +2 n - '0
f ( 15 14) + Kif 52 10 3
Kk• + I<K• 7 15
nn < 10 0
nnn 13 ± 7
nn < 30
NOTE : There are in addition two hadronic phenomena associated with the �
which are unexpected un les s the � is , like as, weakly coupled to hadronic s t a tes
of zero s t rangenes s , The first is backward production of � in n + N + N + �
: �· :
I t is very weak , as though the � were not coupled to the N , N* (a nucleon
traj ectory) j unc tion . Again p + p + ' + nn is very s trongly suppressed
( relative to p + nn for example) as though the ' ' s s and s cannot be made from
the non -s trange quarks in the o ther parti cles .
VDM and Photon Hadron Interactions
Comple te vector dominance implies that the photon cannot interact with a
hadron except as it firs t becomes a p, w or � . and then this p , w or � (say V
in general) interacts with the hadrons . Thus we would expec t to find some kind
of relation between ampli tudes like
Amp (y + A + B) • J4we 2 2 .!_8v (V + A + B) .

v
( 18 . 1)
There are several ques tions involved in us ing the formula . Firs t , there
2
are theore tical questions of mean,ing . S ince y has ; • O, the vector meson
2
on the right s ide should be a vir tual vector meson of q • O , for which s t rictly
no de finition can be given . It might be argued that s ince mass on the initial
s tate incoming particle might have the leas t effect at h igh s , (where E ';:: p
anyway for finite mas s ) this relation may be mos t nearly correc t for real vector
mes ons on the right -hand s ide at very high s . Another related ambiguity is
that the photon has only two polarizations , helicity ± 1, so the equation has
only meaning for the corresponding helicity ± 1 s tates of V. But helicity is
not a relativistically invariant concept and depends on the frame . Mos t
theorists have come to the conclusion it is in the a -channel frame that the
p helicity should be ± 1. At any rate this uncertainty is reduced by going to
large s als o . Thus we shal l , to avoid much comp licated dis cus sion , limit our-
selves to comparisons to experiment at the highes t energies now availab le .
The next ques tion is how we are to ob tain the vector meson cross sections
or ampli tudes V + A + B, after all V meson beams are not available . Some t imes
theo ret i cal arguments are available , but i f they are too complicated they are
not useful to tes t VDM . The most useful s imple cases are :
a) Pseudoscalar meson production , in particular p + N + n + N.
b ) Diffraction (elas tic) scattering from nucleons p + N + p + N or nucleus
p + Nucleus + p + Nucleus .
We dis cuss each in tum .
We can s t udy the reaction p + N + n + N by experimentally s tudying the
reversed reac tion n + N + p + N , which should have the same amplitude . This
has been s tudied at 15 GeV and reported by D . W . G . S . Leith , Phenomenology
Conference , 19 71 (Caltech) p . 555 . Naturally the p is not ob served dire c t ly

o -
n -p + p n p + n n - n looking in
+
but n is inferred from a complete s t udy of
the appropriate mass region of the two outgoing pions . Some correc tions mus t
be made for pairs of pions , say in mutual s waves , which are not due to
virtual p decay . However the data at low t has been nicely analyzed in detail
theoretically and we can describe the results , and compare to VDM expectations .
We evident ly wish to compare these results to reactions like y + p + n+ + n
0 0
and y N + n N is related direct ly by VDM to n -p + v n where v is a vector
meson linear combination o f P , w, f· Since the f coupling to mesons is smal l
we drop i t , and the w f interference . VDM then predic ts for example that (by
squaring appropriate amplitude s )
+
(yp+w n) +
da -
2 dt
1 [ da (yn+ 11 p) ]
dt
4we 2 [ l l p dT '° P n) + 2 [Pnl

] (18.2)
gp2
1 da - o 1 da
• + at (11 - p + wn)
.,
[p ] p
g
"'
These [P 11 J are density matrix elements to project out the helicity � 1 reactions .
-2 -2
p "'
Data in the "' reaction are lacking , but g / g is about 9 / 1 so the second term is
(w
da - o
pro b ab ly never large . In addition , at 8 GeV p + p n) is itself about
dt
10 t imes
da
dt
(w -
p + "' n) in the forward dire c tion , so the "' contribution migh t
be as small as 1 % to the VDM result . Th e same could n o t have been said of a
P -lil interference term ( for an amplitude of order .1 interfering with 1 can make
a 20% effect , but its s quare is only 1% ) . Tha t is the reason that the comparison
is made to the sum of yp w+ n and yn + 11 p cros s sec tions - for in that sum
-
+
the p -w interference term ( f rom isospin cons iderations) cancels out ( as does the
p -+ interference) ,
The cros s sections for all the helicity combinations of the p have been
measured ( and reported in the form of dens ity matrices) by measuring the angular
+ -
distributions of the pair of pions lf lf resulting from the dis integration of
the polarized p2 • We also have data with polarized photons ( perpendicular and
parallel to the plane of produc tion and so can make two comparisons . The vector
dominance model predicts then ,
41fe 2
da
(yN wN) • -z ( p l l + p l-1 ) dt c w -p + p n)
.J. da o
dt + ( 18 . 3 )
g
p
da1 2
( yN ( w p + P m)
"'"'d't
1 4we [ p - p da - o
l-1 J dt
+
11N) · -- ( 18 . 4)
g 2
ll
p
where da / dt (yN + wN) means the average of da / d t () n

+
+ w -p) and da /dt (yp + 11 n)
and the p ' s are density mat rix elemen ts in the helicity frame , The results of
such a comparison appear as follows (polarization data for photons at 15 GeV

are not yet availab le , but a guessed extrapolation from 8 GeV was made - this
does no t affect the comparison of the s um o f � and a11 which is of course
meas ured as the unpolarized cross sec tion at 15 GeV) ,

Vector Mesons and Vector Meson Dominance Hvpothesis 99
�r- - ttf t
4 1/2 [crh + CTy11] + 112 ["";, · err,!
.. <:!• 300
�
t !V!
Y,2 (P,II-P,H J (g_di<r) + ,(P,, •P,.,)(�)
I'
,u
� p � p I'
lj
200 E
!:'!
D
:I.
-t D
:I.
200
lll;; lll;;
<:!.. <:!.. I
---'r�ww-------------------------
100 -
l1i 1/IfI + 11
.. .. .. ..
:I! :I!
'
100 • 0
!!! !!!
I
•
0
I + .. . . . . . . . .
----1..:._"_ _J _ l__ _ _J___J_ __J____L
0 o. 4 o.e o. 4 o. e
k GeV / c k GeV/c
The results sre seen to agree in exactly the forward direction and generally
for the unnatural parity exchange a11 , but clearly disagree away from t • 0 for
".i · Thus VDM as a general theorem cannot be exactly correct (barring the poss i
bility of an eDGrmous w contribut ion in this region) .
Thus photons do not couple exactly as off mass sbe.11 vector mesons . But
j us t how do they couple? To make progress we should s t udy the nature of the
deviations f rom a s t rict VDM and find their characteris tics . For example had
we a good theory , which at presen t we do not have - but should be ab le to
develop - we could say why these various cross sections agree for a" and for
the forward direction , and why not for � ? Perhaps some clues can b e ob tained
from this example so we s t udy it further ,
The values of t we are s tudying here are so smal l that single pion
exchange ought to dominate both cross section• ( i , e . y as well as p ) , If this
were exactly true without absorp tion corrections we would expec t a re lation
between the cross sections s imilar to ( 18 , 2 ) , ( 1 8 . 3) , or (18 . 4) simply because
diagrams correspond:
11 \_J N 11 LJN
Y /\ N p nN
di ffe'rence of p , y and selection
2
The only difference ( as ide from the mass
of the correct helici ty ampli tudes to compare) would be that the y is coupled
to the p
lµ
+ p
2µ
of the pions via a facto � and the p via f
P1111• Hence
we would expect a relation like
Amp (yN 11N) .,p tp1111 (pN

+ Amp + 11N) (18 . 5 )
J OO Photon-Hadron lnteractions
to replace (18. 1) ,
and corresponding relations for the cross sections (i .e . (18. 3) ,
(18.4) 2
with g p replaced by f 2 ) . Therefore insofar as these relations agree
p 1111
for small t we have no t made
any new tes t of the vector dominance relations other
than that f
P 11W
2
� g
p
2
, a test that we have al ready seen works very well . If
that equality of coupling cons tants was an
acciden t then the low t agreement
"'" 'uld
only mean that s ingle pion exchange dominates both reactions at small t -
someth ing we expect to be true on other grounds .
On the other hand even at low t , .r.:f • m11 for example, t he two theories
deviate - how could that come about ? The one pion exchange term mus t , of course ,
be corrected for absorption . The differences between a yN an d a p N at low t most
likely can be attributed to differences in the degree to which absorption
u.odifies the one pion exchange expectation .
In the first place (see Leith ' s report) a fair fit to the p data is given
by a one pion exchange model co rre c ted fo r absorption (due to P . K . Williams ,
Phys . Rev . Dl 1812 (1970)) for the amplitudes that concern this gives us
(18.6)
and
(18. 7)
We have already seen that for

the unabsorbed one pion exchange gives
these amplitudes ( in connection with the discussion of photoproduction)
A.J. � 0
and that ideal ab sorption would subtract 1/ 2

from each amplitude . This leads
to zero Aff -t m11 2 if • and
for large t equal amplitudes for AJ_ , A11 and hence
to an asymme try that rises from 0 at t 0 to l at t m11 and falls beyond tha t
• •
2
to O . This agrees with the asymme t ry observed for he t p, but not for the
photon . In the photon case the fall off for large t is only moderate , to say
0 .5 or so , Thus for the photon it looks like a more appropriate correction
for absorption might be to sub tract more like 2 /3

from the two amplitudes .
(Thus the asymme t ry is 1 t • 2m11 2 for
falls to and
for large t not in 0.6 -
Vector Mesons and Vector Meso n Dominance Hypothesis 101
disagreement with experiment) . We really do not exactly know why the sub-
tracted term for absorption should always be the full 1/2 . Possibly in
this case extra contributions (isoscalar or isovector?) are contributing to
the photon case to increase the 1/2 appropriate to the p case. The point
here is an effect different for p and y and hence different from expectations
of VDM (unless it could possibly arise from the w term) . However the correct
cons tant term to use is determined experimen tally by the cross section for
t 0 which is just proportional to the square of this constant with known

•
coefficients. This determines that the constant is indeed 1 / 2 to 10% .

But a much more striking difference is the difference of oy� and o PJ.
This difference is already large at small t - the o Pl drops off much faster
than does oYl Why? I don't know. I think the mystery here is �y the o pj
falls so fast . In the fit to OPE plus absorption it was found empirically
that a factor e lO(t-m ) needed to be included in these cross sections .
n
2
P. K . Williams had suggested such a factor for absorption effects but expected
a much slower fall off (like e) (t -mn > ) . To what is such a rapid fall off due?
2
Such a rapid variation is entirely unexpected . It is very possibly due to the

method of analyzing the data. The effect of an amplitude to produce a pair of
n's in a mutual s-wave (not a p ) has to be subtracted away. The value of p l l
near • mn 2 is especially sensitive to what is done here (at t 0 or -t � lOmn 2
-t •
it is not sensitive.) Nevertheless there is a contribution to natural parity

exchange which is different for y and p (violating VDM) . How can we make a
theory of where and when such deviations should arise? We leave it as a
problem to analyze this in more detail . One obvious possibility seeing the
large number in the exponent is to be reminded that total absorption effects
(as seen in elastic scattering) do fall off as expbt with b of order 8 or 9
for nucleon scattering. So perhaps a careful analysis of absorption, or
n
also of the possibility that the source of the pion is indefinite by the size
of the nucleon or the will explain it. The point of explaining it, however,
p,
is that the drop off is much slower for the photon case . Therefore whatever
the cause it works differently in p and y pseudoscalar production. In its study
lies the possibility of understanding physically where the ideas of VDM go
wrong. The problem cannot be difficult - the effect shows up for small t and
1 02 Photon-Hadron Interactions
hence for large impact parameter and therefore in a realm where physical
phenomena are usually understandab le ,
Lecture 19
Diffractive P roduction of p , w, t
The next topic we take up is the di f fractive p roduction of vector mesons
by photons , At first we s t udy the p meson for mo re detailed data is available
here . I will not go into as much detail in the results as we are accus tomed
to - for a full recent report see :
Wolf : 1971 International Conference on Electron an d Photon Interac tions a t
High Energy , Cornell , I thaca , N . Y . , (19 71) .

Our ma:ln concern will be a comp arison to VDM .
The p0 p roduction i n yN + p 0N looks very much like diffraction s cat tering ,
the cross section approaches a cons tant at high energy and has the typical
dependence of such scat tering . But how does the photon di ffract into a p ? One
answer is provided by VDM , from the expected ampli tude relation (18 . 1) we expect
2
do o 411e
( N + P N) (19 . 1)
dt y " -- 2
g
p
(because of I spin change it is expected that w0N + p 0N or t0N + p 0N will fall
rapidly with energy so only the p term remains ) . We do not know directly the
cross section of p0 on nucleons but we can expe c t it to be a typical e lastic
scat tering .
A crude use of the quark model supposing the quarks to scat ter independently
of their spin direction and to be similarly dis t ributed inside the p 0 and 11°
suggests a (p 0N) • 0 (11°N) . This latter is not known by direct experiment but
1 N � 11 N)
- - +
0 (11 N + N) ] ,
+
isospin gives it as It does turn out that the
Z ( 0 ( 11 + 11
total cross sec tion for yN + p 0N varies with energy in j us t the way 2 [ 0 ( 11 N)+ 0 ( 11 N) ]
1 + -
behaves , In fact the total cros s sect ion is given correctly by this rule and
e.quation (19 . l) with g 2 • 2 . 8.

p
1mcertainty because the observed p rocess is yN + N + 11 11
+
There is a 15% +
and there are uncertainties in interpretation due to so-called Deck-type diagrams
in which the y becomes two pions not at the p resonance and one pion s cat ters like
y
�
Vector Mesons and Vector Meson Dominance Hypothesis
11
11
103
11
N N
ins tead of
11
y
11
A strict adherence to VDM would , as far as l can see expect the analogue
for p scattering to 211 via
but the point is in es timating j us t the effect of the s imple p elastic scattering
so we can use our 11 ° scattering analogy .
1bis idea that yN + pN tells us about pN + pN permits us to summarize the
behavior in another way . 1be polari z ation of the p coming out can be measured .
l t is very nearly purely transverse and polarized as the incident y ray . 1bis
is excellent evidence that the process of elas tic s cattering (also called
pomeron exchange) does not change the helicity in the center of mass sys tem .
w production . Here the diffraction term is smaller so other processes are also
effective , in particular one pion exchange at the lower energies . 1be natural
parity exchange has besides the pomeron exchange also poss ib le � exchange so
the purely diffractive part has not been separated out yet for a clear tes t .
+ production . 1be experiments show some inconsis tency but data is availab le .
One way to test the relations of VDM i& to compare the forward differential
cross section for yN + VN with the forward differential cross section for
VN + VN . 1bis latter can probably b e es timated fairly well from the optical
theorcom using quark model estimates for the total cross sections for VN .
(1be quark mode l est imate for the p, w and t are :
cr T (K+p) + cr T (K-p) - 0 T ( 11+p) • 15 mb at 5 GeV . )

crT ( . p) •
This gives a fit with gp 2 /411 • 2 . 6 t . 3 , gw 2 2 4 t 5 and g � 2 • 22 t 6. Only

..
•
I 04 Photon-Hadron Interactions
2
the g
$
seems too b i g . It could mean the quark es t imate of a 4N i s too b ig -
10 mb would f i t b e t te r .
Lec ture 20
Di f f raction production of ve ctor mes ons can also be s een , and more
copious ly and clearly from y on nucle i , A . (Reference : K. Go t t f ried "Nuclear
Pho toprocesses and ve c t o r dominance " , Cornell Confe ren ce , 19 71 . ) There are
difficulties for the w since the final 3n s tate is hard t o measure , and the p
is so wide that theore ti cal que s t ions of in terp re t ing data are involved . The
cleare s t case expe r imen tally is $ product i on , however much more data exis t s
for the p , In these exper iments on p p roduc tion obse rvations are made o f 2n
production by photons on nuclei , o f course . A b eauti ful res onance in mas s o f
the 2 n is seen at the p mas s . I t i s asyane trical showing the in terference
effect with w + -
+ Z n j us t as in e e production .
The 4> data is now goo d . The dependence o n A , the ma s s number o f the
a
4>P
nucleus , gives us some i dea of , and the ab s o lute cross s e c tion pe rmits a
determination of g . However an unce r tainty arises for the results depend

4>
sens i t ively on the cho i ce of an unknown quantity , the real part of the
forward $N s cat tering amp litude f ' Then the dat a

4> 4>
is ins u fficien t to dete rmine all three and If
a
2
g / 4 n is its Orsay value 1 3 . 3 then we can only conclude that may be
4>
perhaps from - . 3 to - . 5 and a4>N in the range 8 t o 14 mb (at about 7 GeV) ,
pos s ib ly a b i t lower than the quark model rules would sugges t , but lit t le
can be said .
The p data is more extens ive . Here the large width causes some confus ion
as to what part of the data is to be at tributed to p production . Again we

2
have three parame ters g p /4n , a pN and a
pN
' One ge ts agreemen t with expe ctat ions
if we choose ap N • -0 . 2 4 (but it is no t well dete rmined) . The data gives
/4n • 2 . 6 and a p N • 27 ap N
2
the "good value s " (a t 8 . 8 GeV) g mb . ( But is not
p
well de termined and res ults depend on i t . ) There fore the phenomenon o f ve ctor
meson produc t ion exi s t s and behaves closely like a diffraction proces s , and
coul d wel l be in quan t i t ative agreement with the VDM . Does this tes t the
Vector Mesons and Vector Meson Dominance Hypothesis 1 05
mo de l ? I think not because formula ( 19 . 1) applied to dif frac tion s cat tering
can be derived ( at very high energy ) from ano ther hypothesis - namely that
the elas tic scattering of the p is much larger than diffractive dissociation
scat tering o f the p ( the process by which p N + p *N where p* is some other
s tate of the same isospin as the p , and which survives as s + � ) . Such
diffraction dissociat ion ( as NN + N*N or nN + n *N ) are in other reactions
pe rhaps only 30% of the e las t i c s c a t t e ring f rom the nucleon s o there is
no reason not to assume it for the p , Fur thermore for the p we shall only
have to assume that the part of diffraction dissociation to those p * having
spin 1- ( like the photon ) is small compared to the elas tic . There is also
a direct experimental confirmation of the fact that yN + p* N where p* is
any isospin one , 1- s tate , produced in a diffrac t ion dissociative way does
not fall with energy .
For production on nuclei our assump tion ( of the dominance of elas tic
diffraction dissociation ) is ever more true as A rises - for the elas tic comes
from diffraction from the shadow of the entire nucleus - whereas the
particles produced by dissociation can come only from its edge .
The point is it is no t dominance of the y to p that we require for
the validity of equation (19 . 1) , it would also follow from dominance of
P + p in the products o f the diffraction of p on nucleon and nucleus .
We can see how this works most clearly in the case of scattering from
a large nucleus where our assumption is most nearly valid . Ins tead of
considering yA + pA analyze the reverse reaction p A + yA and compare it
to pA + pA. The high energy pA scattering appears , in say the c .m . as follows :
The waves of the incoming p come in from the left and fall upon the
nucleus where they are ab sorbed . A short dis tance (which can be many
wave lengths if s + � beyond the targe t say at the dot ted line the wave
function is nearly the original p wave funct ion with a hole as a function o f
b , the impact paramete r . Beyond that the waves slowly diffract into the
shadow in a way determined by Huygens principle . If 1 - a(b) is the ampli-
t ude to find a p at b (so a is near 1 for b smaller than the nuclear radius ,
and near 0 beyond) the amplitude for an outgoing wave of transverse momentum
Q is ( l-a (b ) ) e iQ • b d2b • f (Q) . The amplit ude at P is
f
ikr 2
1/1 (P) • f (l-a (b) ) __ d b (20 . 1)
e approximately
r
Of course the 1 gives only the forward beam and doesn ' t interes t us - i t
is j us t a s i f the p came out only across the face of the nucleus . To be sure ,
if the p were not a poin t particle but say made of parts ( indicated by the
braided line) we could no t precisely define every thing in terms of an amplitude
to give a p , but would need a function of all the parts as variab les . But
clearly for b outside the nucleus the parts are in the same relative amp litude
as in a p - only near the edge is there some distortion of the re lative behavior
of the parts - and so a proj ection possible into some other than the p . But
the amplitude a (b ) is clearly that for a p except near the edge ; so that elas tic
scattering is much larger ( for nuclei , at least ) than diffraction dissociation .
N ow we have no ted from experiment by e+e - that whenever we have a p
presen t there is an amplitude that this p will disintegrate to e+e (via a

-
/
pho ton) and therefore there is a source of the electromagne tic field - a current
given by j µ Y 4'fle F •amp. to find p polarized in direction µ , (F • - M /g ) .
2
=
P P p
Thus in our prob lem at each point P there is a current source of electro-
.p
magnetic field of size l/l(P) - thus the total amplitude for finding a photon
going out with wave vector kout is f l/l (P ) e out d P . If we subs titute our
-ik 3
-ik r/ r and e -ikout. p j us t
express ion for ljJ (P) in this the convolution o f e
-ik ·b 2 m 2 and k 2 • 0) so we even tually ge t the
1
gives e out 2 • (But k
•
k -k t
2 p out
outgoing amplitude f8� photon is
�F p f (1-a (b ) ) e
-i Q out .p 2
d b (20 . 2)
---
mp 2
where Qout is the transverse part of kout " The integral is the same as for the
p elastic scat tering so the amplitudes are proportion. al - the cross sections
2 2
mus t bear the ratio 4 'fle /gp •
One can of course carry the mathematics out in more detail separating
the z and transve rse momentum care fully to see j ua t how the as s ump t ion o f
high energy is valid . I t is eas y to unde rstand i f t h e diffraction o f the
wave in to the shadow is omi t ted . Then the wave going forward behind the
ikP z
z
shadow is j ua t e ( 1-a (b ) ) where the in cident p is of energy E and
p
k ::
z
j�p
• E "'111
p
2
/ 2E . The amp li tude to make a photon of ene rgy E , transverse
momentum Q , hence longitudinal momentum k y •

2
E - Q /2E as
z
iQ · b
2
( 1-a (b ) ) e d b
2E
411e F
J'X
f (Q) -
p
f (Q) (20 . 3)
2 2
m _Q
p
A l/ 2E factor has been included to t ake account of the relat ivis tic l / lfi
2 2
normalizations for the incoming and outgoing particles , The m - Q is wrong
p
2
p
here , the dif frac tion into the shadow makes it m (as though the
p
k were E for p ' s moving in th is dire c t ion ) .
z
It is evident tha t the polarizat ion of the 'I is the same as that of the p
for this proces s . Th i s feature has been checked expe rimen tally very care fully
and is ve ry st riking , the p polarizat ion p roduced by polarized pho tons is nearly
pure ly that of the pho ton .
Of ten it has been s ugges ted that VDM predicts its formula (19 . 1 ) via a
�
diagram like
1'1ucleus
s ugges ting the pho ton becomes a vi rtual p o I ze ro

qz , and hence th at the
coup ling F (or - m / g ) should be that appropriate for q •

2 2
P
0 ins tead of
p p
m
2 2 + -
q • where i t is measured in e e . It is a tes t of VDM to find tha t this
p
constant is unchanged in this range . But we· have seen , by cons iderat ion o f
the reverse reac tion , neglecting dif fraction dissociation that the appropriate
mp
2
cons tant should be that of and no extrapolat ion is involve d .
J. Mandula has con firmed my argumen t by an argument from dispers ion theory
and the reduction formula .
The assump t ion of negle ct o f di f f raction dissociation comes in here . Near

the edge , the parts of the p may not be in precisely the same relative mo tion
to produce a p,
but they might s till produce a photon , Clearly this is small ,
2
It means an in terfering term o f order FP * /mP * times the amplitude to make p *
where (FP * i s the coupling of p * to photons ) summed on p * . For nuclei at leas t
it must be small coming only near the edge . The point is not so much that F *
P
is small (which it may be , as so little p* seems to be produced by photons on
nuclei (which measures F p *2 a p *N ) but that the number of p * produced by p via
diffraction dissociation is small .
Other Tes ta o f VDM
If VDM were correc t we expect the pho tons to have amplitudes l/ "v to be
various vector mesons so that the total cross section of y ' s on p is l / g,, 2
times the total cross section of each of the vector mesons . We expect
2.
2
° t 0 t ( Vp)
0 tot (yp) • 411e (20 . 4)
p , 111 , 41
-- 2
"v
(neglecting possib le interference e ffects of w and 41) , This is dominated by
the p (because of the 9:1:2 ratio of s.y2
is small) . We es timate and a ( 41P )
a ( p p ) • o (wp) by the quark model as equal o (11 p) (a s cheme which we che cked
°
t t t
in our a (yp -+ Vp) con s i de rat ions above ) . We find experimen tally that o tot (yp)
is too large by 40% , as though o ther processes were available to the photon to
interact with the proton besides going through virtual vector mesons .
A very s imilar check giving precisely the s ame result is to use eq . ( 18 . 1)

to say
A (y p -+ yp ) • � "'\' l A (yp -+ Vp ) (20 . 5 )
L.v g v
Again the dominant term is the p . Independently o f the relative phases we
IL
get the inequality
2
1
1
do/ dt (yp �
2
2
do
-+ yp) 4 11 e - (yp -+ Vp) (20 .6)
Bv
p ,w, dt
The quantities on the right hand B ide are available directly by experiment .
The dependence of do /dt (yp
0
-+ Vp) on s ( from 2 . 7 to 5 . 2 GeV) and on t ( from
ia
2
O to . 4 Ge V ) are well represented by the right hand s ide , but it always only
about 1/ 2 of the experimental cross section ! Thus VDM fails here .
This ia in accord with the 40% error in the total cro s s s e c t i on text , for
the optical theoreom relates do/dt (yp -+ yp) in the forward direction to the
s quare o f o tot (yp) (assuming something about phases ) .
Vec tor Mesons and Vector Meson Dominance Hypothesis 109
Lecture 2 1
Shadowing in Nuclei
How should the total cross section for y+ Nucleus vary with the mass
nuni>er of the nucleus A ? We know for collisions with hadrons nuclear matter
is nearly opaque (because a hadron-nucleon is comparable to the spacing of the
nucleons) and therefore the cross section for A nucleons is not the sum of
a contribution from each (hence a tot A a nucleon) because nucleons in the
�
front shadow those in back so ultimately for large A it goes as the area of
the nucleus or A2 /3 .
On the other hand for photon-nucleus collisions� first glance would
suggest that the nucleus now being transparent (ay nucleon is mu ch smaller than
nuclear spacing in a nucleus) each nucleon would see the full beam and hence
the cross section would go as A (times the simple nucleon photon cross section) ,
On the other hand the VDM shows that this latter conclusion cannot be
right in general for it is incorrect if the VDM is correct. A photon amplitude

should be proportional to the p amplitude - the latter is a hadronic amplitude -
hence the photon cross section proportional to the p cross section and therefore
varying as A2/3 (I neglect the contributions of and + in this qualitative
w
discussion, their contributions are easily reinstated) .

The reason is that in the simple view we imagined the interaction
y + nucleon • X to be a local process of y on nucleon, the interacting near
y
where the nucleon is located - but VDM reminds us that this ia not so. A y
can become virtual hadrons (e.g. 2 w , but most importantly a p ) far away in
front of the target, and the virtual hadrons propagate a long way to the nucleon
to interact, A real photon is a pure ideal photon plus a virtual hadron with
amplitude l/6E according to perturbation theory where 6 E is the energy difference
of the state of given momentum as photon and as hadron. Thus if it is a p
the energy ia � while that of the photon is v so 6E · VR - v "' mp 2 / 2 v ,
very small for large v . This is also the distance ahead of the nucleon where
the photon-hadron conversion occurs . If this is large compared to the mean free
path of p in the nucleus then shadowing occurs (A2 1 3 ) , if it is small, no
shadowing (A)•
The physical idea can be seen best first for a simple model consisting of
JJO Photon-Hadron Interactions
two thin s l abs a, b of nuclear matter one in front of the o ther by d . We shall
calculate according to VDM.

d
If a p were impinging s uppose the small prob ab ility of being ab sorbed
in the first layer is f ; in b alone i f p impinged f , But the probabili ty

a b
that the p ge ts to b is (1-f ) so the product made in b is (1-f ) f . The

a a b
total cross sect ion is thus f + ( l- f ) f • f +f -f f the las t term represen ting
a a b a b a b
shadowing .
Le t us calculate the p rob ab ility for any products made in b i f a is present
by pho tons via VDM. I f a i s not p resent the amp li t ude to find a p at b
making product is proportional to the amp litude the y converts to p at x

2
< 'V411e FY ) and then the p arrives at d . This is
i (k k ) (x-d)
( 21 . 1 )
Y P
-
_ r;:::r
4Tre F e
'V P
( the phase has been taken relative to the pho ton phase at x • d i . e . the
above express ion contains additional factor exp (-ik d) ) where the k , kp
y y
an
2
are the k-ve c tor for y or p at the s ame frequency . Hence - k • m / 2v for
y p p
k
large v (•k ) .
y
When a is present (2 1 . 1) is valid only for d > x > O. For x < 0 we have
the additional factor fo r p to ge t through a • y l-f a ="" 1-f /2 . Thus for

x < 0 we have
(2 1 . 2 )
;
The total ampli tude (in tegrate x) is there fore proportional to
P Fe k -k
y p
(1 _
f
2
a
e
- i C k -k ) d
Y P (2 1 . 3)
The main term has an amp litude proportional to

� FP
4we
2
2
as we
m gp
expe c t , the term shadowing f comes in various phaies - and in a more complete
a
in
Vector Mesons and Vector Meson Dominance Hypothesis Ill
problem with a continuum of layers would cancel out for (ky -kp )d > 1.
Thus the criterion for scattering involves shadowing should be complete
v,
for v + "'
One extend this easily to a thick layer in one dimension. We seek the
�an
amplitude to find a at x (for that will later be used by squaring and summing x
p
to get the total cross section.
We have two cases y 0 and y > 0

<
Y < 0 : y converts at Y , propagates as p to O ; propagates as p in medium to x
Y > 0 : y converts at y, propagates in medium to x.

The amplitude is proportional to
i(k -k ) y -i(k -k ')x y < 0
e Y P e Y P
i(kY -kP ' ) y -i(kY -kP ' )x (21.4)
e e y > 0
(the phases have been taken relative to the photon phase at x, i.e. the above
expressions contain the additional factot exp ( -iky x) ) .
Integrating over Y the amplitude at x is
1 -i(ky -kp ' )x { -i(k -k ' )x )
ky -kp e + ky -k1p ' \1-e Y P (21 .5)
Here kp ' is kp in the medium. At very high v it has an imaginary part (representing
absorption) which is finite and fixed, the real part goes , of course with v , but
may differ from it by a finite amount. Th.us as v + ky -kp k a fixed
"'
•
' °""
number (whose imaginary part gives the absorption cross section and whose
p
phase is the phase of forward p nucleon scattering) .

2
Thus if ky -kp m2 v k, the first term in (2 1.5) dominates and we get
a «
exact proportion to p absorption, hence absorption.

A full calculation including the spherical geometry of the nucleus has
b een made . They show a v dependence of a to t (A) /Ao (nucleon) for vario us nuclei
( see Go t tf re id ' s report ).
1
-- c . 83
a t o t ( A)
- Cu
-
Ao t (N) , __
--
. 65
.s .53
to
. . ___ Pb
. 25
5 10 15 2 0 v (GeV)
The data does not show much v dependence - (but it is not far off ,
expe rimen t al errors are large enough , and the expe rimen t is very h a rd at
lower energies ) but no t outs ide expec t a tions - howeve r the asymp totic values
of a t ot (A) /Ao tot (N) are c lo s e r to . 87 for C, . 75 for Cu and . 60 (wi th large
error) for Ph .
These a re all unders tandable i f the total cro s s sec tion is considered
70% VDM and 30% something els e ; (as we have s e a n is t rue) and i f the something
e lse goes jus t as A p u re ly , showing no sh a d owin g . Then the asymptotic

( fo r v + m ) values o f a t (A) /Aa (N) should di f fer from 1 by on ly 70% of the VDM
to
the o re t i cal value for this difference .

Fo r them q 2 • -Q and
2
Measuremen ts also exi � t fo! virtual pho t ons ,
2 + Q
•
m
v - g:_ k -k • e we agai n ge t
y 2v y p v
k so Thus we exp e c t from VDM that
a v depen den t shadowing , the parame t e r determining the extent o f shad owing
being
2
(m + Q ) /2 v
2
instead of m2 /2v . No v dependence o f shadowing is seen
Au
2
( in for Q • 0 . 5) in s t rong disagreement with VDM predi c t ions . If we corre c t
VDM by al lowing a certain percentage of "other pro ces ses " which do n o t show
shadowing , of cours e , less shadowing is expe c ted . But we s ho u l d also no t e the
with Q • The p amp l i t ude falls as l /(pi. + Q )

fraction of something e ls e mus t i
r se
2 2 2
(m2 + Q 2 -2 The actual cross s e c t ion
so VDM cro s s sections sho uld fall as ) •
probably falls only as Q -2 for large Q (see later ) hen ce the fraction of photon
2
cro s s s e c tions wh i ch can be ass ociated wi th virtual vector meson production
(VDM cont ribution) mus t fall as ph o t ons b e come more vi r tual . Thus th e results
for Au are not s urpris ing.

To Summarize the Pos ition of VDM

Interact ion of photons o f small q
2
(or q
2
• O ) is very s t rongly influenced
by the nearby po les at the mass o f the vector mesons . Sub s t an tial parts of the
interaction can be unde rstood as being due to intermediate virtual vec tor
mes on coupling - but , o f course , not all the coup ling can b e so des cribed
( for example only 70% of the total yp cross section at high ene rgy is so
des cribed ) •
In the special case of the coupling to pion the dominance is mo re complete ,
evidenced by direct experiment e e - � (showing mainly the p , w

+
as n+ n - peak ;
• 4Gev
2 2
not a large background and not much else res onant has b een found to q
and by the fit of cons tants f ::> g . (This will be dis cus sed further in
p nn p
lec ture 2 4 , in reference to the dispers ion theory o f the pion form facto r . )
Finally measurements exis t on nuclei for photon production of incoherent
p (- t > O . l GeV) as we ll as for production of other particles such as n+, n

0
and even n° ( also K+) . Here the theory is comp licated by a number of feature s -
such as the ab sorp tion of the particle leaving the nucleus . The VDM theory
predicts rapid variations in A with energy , whereas experiments indicate

eff
only a small vari a tion , if any . (See Go t t fried , I thaca Con ference (19 71) and
Diebold , High Energy Physics Con ference at Boulder , April 1969 ) . There is at
present no explanation known for this discrepancy - its s tudy would make a
nice prob lem . I doub t tha t the fact that the y couples only partially to VDM
is sufficien t to account for the l arge dis c repancy . Some i tem in the theory o f
thes e incoheren t final s tates is prob ab ly invo lved .

Electromagnetic Form Factors
Lec ture 2 2
Elect romagne tic Form Factors
Nucleon : The matrix e lement for the intera c t ion of an e le c t ron , s ay , and a
proton involves the diagonal e lement of the curren t operator .
lJ
Us ing
relativis t i c invariance and gauge invariance this can be writ ten in the form
where u u
are 4-spinors des c rib ing the in and out proton of momen ta p , p
1 2 1 2
2
momentum t rans fer q • p - p • F and F are functions of q only . Other
2 1 1 2
linear comb inat ions o f F and F have also b een de fine d , mos t used are ( the
1 2
so-called e lectric and magne tic form factors )
G • F + 2MF
M 1 2
clearly as q
2
� O F
1
i s the charge : F
1 pro ton
(O) • 1, F
1 neut ron
(O ) • 0
and F
2
are the anomalous magne tic momen t a . Hence G ( O )
E
• 1 for p , 0 for n
G (O) • \J P is the total magne t i c moment .
M
These have been meas ured by s ca ttering elect rons from pro tons e las t i cally .
The firs t order ampl i t ude is (u4, u

3
are elect ron spinors )
1 14
(22 . 1)
ep 3
�''r(" Pl
The cross section aris ing from this is :
[ GE t � +
J
2 2
a (9 ) • ()
CINS 9 ------- +
1 + t
2 2
2t GM tan ( 0 / 2 ) (22 . 2)
( Rosenb luth)
2
t -
e 4 cos 0/2 (22 . 3)
2Eo
+
2
M
0/2 ( 1 sin 6 / 2) 2 2
-l
4E s in 0 /2
_o _____
• ;;_..,
1 + T s in 0/2
2E 2
By varying both E 0 and 0 we can vary q and 0 independently , and thus

2
2
verify whether the b racket in (22 . 2) does vary linearly wi th tan (e/2 ) for
It works wel l , so both GE and � can be extracted for the pro ton .
2
fixed q
Data for � is more accurate than GE . For the neutron uncertainties of the
deuteron wave functions make the determinations , especially o f G much
En
less cer tain .
Higher e lectromagnetic corrections do come in , especially effects from the
brems trahlung diagram
e \f . - / p
e /v\ p
but they are largely unders tood and allowed for . This i s an important but
tedious correction to the data to get the ideal pho ton exchange cross section .
Experimental res ults are usually described by giving deviations from a set
( i
o f completely empirical approximations , namely that
2 -2
1 + Ti
The latter function is called the dipole function GD .
Most accurately known is �p /�p ' I t vaguely follows the function GD .
JJ6 Photon-Hadron Interactions
For small q2 it varies as about 1 + (2 .6 ± . l)(+q2), very close to the dipole.

GMp jµp G D � 1 . 0 8 for 1 < -q2 < 6 . Above -q2 8 GeV2 the ratio falls below 1.
•
Thus GM falls somewhat faster than l/q 4 for large -q2 •
The equality �/µn GMp /µp

• as within the accuracy of the neutron
data.
For the neutron GEM2 at q2 • 1 is . 004 5 ± . 004 3 .
On e can fit with GEn : GMn(T /(l + 10 T)).

Neutron -electron scattering at low energy (using electrons in atoms ) gives
0.9 2
n / 4M
= µ
For the p roton GE/ (G / µ ) falls for larger q l near 2 or 3 so if put as

Kp p
1 + A q21 A is perhaps .8 (large error). It is not possible that GEp GMp / µ •
p
for all q2, for at q2 4M2 the definition of these two G 's in terms of F1 and
•
F become identical, so \Ulless GEp and G vanish here (which is very \Ullikely)
2 Mp
we must have GEp GMp at q2 • 4� . Thus GEp /(� /µ ) 2. 79 at q2 • + 4M 2
• •
P p
and •l at q2 O . •
The simple ratios among GMp ' GMn, GEp ' GE n which are nearly right would be
expected by pure vector meson dominance (assuming the mass of p and w equal , and
that � couples only weakly to the proton) . However, all the form factors would
be multiples not of GD but of the meson propagator � 2 /(� 2 -q2) (1 + Q2 /.5 8) -1 • •
This supposes that the photon couples to a point charge nucleon via the p . Thus
it has the wrong large Q2 behavior Q-2• Its small Q2 behavior is 1 - 1.7 Q2
instead of 1 2.6 Q2 • It is, of course, not expected that VDM continue to
-
work as we go to negative Q further and further from the pole at Q2 -mp

2 2
• •
But there is little doub t that the large size of the charge square radius
(i. e . the coefficient -2.6 of Q2 ) is contributed to a large extent by virtual
p ' s. It is not really clear why these ratios do maintain their values at
Q2 • 0 out to Q2 • 2 or 3 GeV2 as they do to 10 or 20%.
Lecture 23
Electromagnetic Form Factors (continued)
We can gain some insight into the meaning of form factors by looking at
Electromag11etic Form Factors 117
the non-relativistic case . For a s p in les s sys tem the form fa c t o r is j us t the
momentum representation of the charge de nsi ty .
f (Q ) • J p(R) e
iQ •R d 3if •I p (r ) 4 11r2 dr { si �Qr
r} ( 3 . 1)
2
Expanding in powers of Q2 for small Q2 , with p (r) normalized so that
Jp (r) 411r2dr • 1 we ge t (sinQr/Qr • 1 - Q2r 2 /6)

1 2 + ••• (23 2 )
f (Q} • 1 - 6 r w
2 .
p
where r 2 is the mean square charge r adius
J p
p
rp 2 (r ) 411r 2 dr
2
• r
An e xp onen t i a l e-ar for the charge density gives the d p o le dis t ribution
i
f (Q) • ( 1 + Q2/ a2 ) -2 • The Q-4 dependence says the slope at o rigin must be
fini te not z e ro . It is , howeve r , no t a va l i d conclusion to use non-relativis tic
views to s tudy large Q2 • Nevertheless the con clusion is e s s en t i a l ly corre c t .
Q + m
2
Relativis tically the fact that as the form factors go t o ze ro
me ans that there is z ero p rob ab i l i ty to find the proton alone as s point charge .
This co uld result in field theory ei the r from ther e being no ideal field entity
(parton) with quan tum numbers o f the proton , or e ls e , if there were such the real
pro ton is always dressed , it h as zero qmpli tude of being an i de a l p ro ton .
To make a relativistic analogue of a thing like the hydrogen atom (whose

charge dis t ribution is e-a r ) we imagine two s p in le s s particles held toge ther with
a spinless Yukawa p otent ia l generated by the weak exchange of some s c alar meson ,
shows s behavior , (2) the

-ar
then ( 1) the non-relativis tic wave function e
form factor varies as l/Q4 for large Q2 •
For example , let one p ar t ic le A be charged of mass m1 , the other 8 ne u t ral

of mas s m2 held weakly non relativis tically . The mass o f the pro ton is near
m + � and neglecting in z eroth o rder the relat ive mo tion the incoming p rot on
A
of momentum p l ' means A has momentum ap 1 and B has momen tum
MA
bp 1 (a • MA+� , a+b•l) .
Finally we wan t a bound p ro ton of momentum P z hence parts of momen tum ap , bp •

2 2
We do this by h i t ting A with the very large Q p -p , whereas for the p ro ton •
1 2
ultimately to hold toge ther p
A
mus t change only by aQ , hence an amoun t of
momentum bQ mus t be passed o ve r t o b (by an exchange o f t h e meson res ponsib le for
interaction of A and B . Th e diagrams are

The two propaga tors each give l/Q2 for l a r ge Q2 hence l/Q4 for the form
facto r . Thus the first diagram is
1 1 . _a_
2
( ap +Q )
2
(bQ) b 3Q4
l
the second
1 1 a
(ap 2 -Q) 2 (bQ) 2

4
• bQ
at large Q2 •
If ve c to r mesons are exchanged between a par t i c l e of spin 1/2 and on e of

2
s p in zero , a Q appears in the nume r a t o r due to magne tic coupling .
N on re la tivis t i cally if a sys tem is made of 3 p a r t i c les , the large q2
asymptotic behavior depends not on the s ingularity when j us t two come toge ther ,
but rather when all three are on top o f one anothe r .
As we sh a ll see , s uch pic tures are too s imple and inadequate . We are on ly
making ve ry naive models to compare to the non-relat ivis ti c ideas . They wil l
not agree with the ine las t i c scattering.
Lecture 24
Pion Form Factor
ex i s t s also for the pion fo rm factor , fo r q2 < O (spacelike)
:"' '� ·�··�.

Da ta
<1.+ 1 J µ I 1T+ > • (p l

2
+ Pz>u F'll (q ) F'll (O) • 1 ( 24 . 1)
It is b es t ob tained from the reaction of yp + n 11+ produced by a vir tual
...... , .... , .... by
By going to s u f f icient energy o ne can ob tain a region where virtual pion
2
exchange is mos t important - and tha t this virtual 11 has as ne a r ly zero q
(and is therefore as close to i t s mas s shel l as pos s ib le ) . The coupling of the

virtual to the nucleon is known in this region {gN 1111y5 , g2 15) so the only
11 •
unknown is the photon-pion coupling between a real and one that is nearly real,
11
but a large momentum transfer Q .

Data is available out to q 2 -1. 2 where F has fallen to 0.28. This
•
11
is just about exactly {for no known very good reason) the isovector part of
the nucleon form factor F 1P -F 1n which also fits all lower Q2 • It will also fit
2 l 2
fairly well to the pure VDM or p meson pole {l + g:_ • 56 ) , which is . 32 at Q
- -1.2, •
2
but not at all well to the form factor GD of q which is too low { .15 at Q -1 .2) . •
We have already discussed this form factor for positive q 2 where it has been
measured in the reaction e+e - + 11+11-. We have an experimental curve which is
dominated by the p pole expression {with small understood corrections for the
w-p interference, etc. ) . It is interesting that an extrapolation so far into
negative Q2 still works reasonably well. The rate e+e -+ 11+11 - measures , of course,
only the absolute square J F 11 (q2 > J 2 not the real and imaginary part separately.
However the shape of the curve (p resonance) is so simple that our physical
understanding may make a separation into real and imaginary parts not hopeless
(e.g. we wrote previously, for the part near the resonance (-mp 2 / (q2 -'mp 2 + ir p p )) m
and had some guesses as to how r varied with q2 • If this can indeed be done we
would have information now known or soon available on F11(q2 ) over a large range
of q 2 both positive and negative. We can then check the hypothesis that every
one believes - that F 11 (q2 ) must be an analytic function of its variable q 2 with
a branch cut along the positive real q 2 - i.e. that F11 satisfies a dispersion
relation.
(24 .2)
(µ •m11)
(211) 2 where at least j F 11 (q2 > J 2 is
measured. A check on a guess for the imaginary part would be to compute the real
part from (24.2) for q2 > 0 and check J F , J 2 (alternatively a much easier equivalent
method is to fit I F J 2 by choosing physically appropriate analytic expressions
11
for the complex function F11 (q2 ) ) . Then, having established a reasonable
Im F11 {q2 ) one calculates F11 (q2 ) for negative q2 (where it is real) to compare to
experiment. We already see that it will work well - in the positive q2 region
the main feature is the p resonance and up to rather large q2 no other resonance,
in fact not a large production, has been found.

Below the inelastic threshold q 2 (4µ) 2 (and in practice probably some
•
distance above it also) the phase of F n (q 2 ) must, by first order unitarity, be

the same the n-n scattering phase shift . Information on this is recently
as
becoming available from pion-virtual pion scattering analyses . This may help
in analyzing the dispersion relation.
How should we expect Fn (q2 ) to go for large q 2 ? It must fall off because
the probability of producing just two n's where so many other states are
available must be very low. The isospin current generated by the pair of pions
must be coupled to other hadrons and we require this coupling not to radiate .
As we shall understand better later, this leads one to expect F to fall as some
n
power of q2 for large q2 , but we do not know the power. However, these con-
siderations make it not so suprising (granting the experimental behavior of
Fn (q 2 ) for positive q2) that for negative q2 even out to -1.2 the expression
is close to the pole formula (which arises from (24.2) if most of the imaginary
p
part is near mp 2 ) .
Sum rules result from the following considerations. If the unsubtracted
dispersion relation (24. 2 ) is valid, since we know Fn (O) 1 we must have

•
"'
-� I m Fn (q' 2 ) dq' 2 1
" J
• (24 . 3)
(2µ) 2 q' 2
a "sum rule".
If we put q2 � "' in (24. 2 ) , we would expect Fn (q2 ) to go as l/q2 times
- -1 1 Im F(q' 2 ) dq' 2 (if F (q2 ) went more slowly than l/q2 we would have to
n n
say Jim F(q' 2 )dq' 2 diverges , as indeed we would expect for Im F(q' 2 ) would
probably also fall slower than l/q' 2 ) . If we expect Fn (q2 ) to fall faster
than l/q2 (we know no sure argument for this) we would have another relation
called a superconvergence relation.

We expect to have an unsubtrscted dispersion relation valid because if
(24.2) had an unknown constant added to the right-hand side then Fn (q 2 ) could
not approach zero as q2 + � . In any event we have the subtracted relation

(24.2) - (24.3) :
(24.4)
In practice (24.4) is more useful than (24.2) to compare F11 (q2 ) for
negative q 2 to its experimental value since uncertainties from the contributions
for large q1 2 where data is unavailable are much smaller,
We can now see using (24.4) why from what we already know experimentally
for F 11 (q 2 ) for positive q2 the value for negative � 2 should be close to the
2
p pole value even rather far out. In (24.4) we know llDF(q ) is dominated by
the pole - there is not much I F11 (q2 ) 1 2 for larger q2 (at least no resonance to
p
4Gev 2 > and (24. 4) does not contribute integrand for still larger q 2 , for q 2
even as negative as -1. 2 . Thus F (0) - F (q 2 ) is given by the analytic extension
11 11
of the pole formula nearly. But F11 (0) is (the accident of the VDM constraints
p
f /g 1, see lecture 16) given by the p pole formula nearly, as we have seen
p ll ll p
•
earlier; thus F11 (q 2 ) should also b e . An interesting study would be to try to

make this discussion quantitative. One can also ask if the data on (e+e - + 11+11 -)
is accurate enough to also dominate (24.3) and thus explain the accident that
fp 1111 /g p 1 . •
Proton Form Factor for Positive q

2
The proton form factor for positive q 2 could be ob tained from the reaction
+
e e -+ p+p. lb.ere are not yet any quantitative experiments but they are expected
momentarily. One experiment near threshold saw no pairs , indicating �· GE are
small there - about small (or smaller) than the dipole formula calculated
as
there q 2 4. We would expect again that F1 Cq2 ) and F2 (q2 ) are analytic
�
functions satisfying a dispersion relation like

1 r
2 ImF(q' 2 )dq' 2
F(q )
J q2 -q ,2+ie
• -
11
- (24.S)
(When amplitudes csn be expressed in more than one way, such as helicity
flip amplitudes , or coeficients of Dirac matrices in spinor express ions - it is
for the latter that the simplest dispersion relations are supposed to hold .
Relations for other combinations s t be deduced from these . )mu
This time we have a difficulty. F(q2 ) is measurable experimentally for

q2 0 by electron scattering and for q2 > 4� by e+e - annihilation into
<
proton-antiproton pair. But how can it be defined experimentally in the region

q2 • 0 to 4� ? I do not know of a physical definition. But it is expected
(fron field theory examples etc. ) that ImF(q·· 2 ) is not zero in this region - in
fact that the integral in ( 2 4.5) has a threshold at ( 2 µ ) 2 • Thus no amount of
experimental data will permit a very detailed test of ( 2 4.5) for a piece of the
integrand is completely unavailable. Of course, if a theory is available for
F(q 2 ) it can be defined in no-man's land (O < q2 4M2 ) by analytical continua-
<
tion - but if we had a theory it would presumably already satisfy ( 2 4 .5)

and the proper thing to do would be to compare it directly to experiment in
the physically available region.
Thus the dispersion relations for nucleon form factor are not of much
direct use for comparing to experiment.
!iQ:!! : We append here a theoretical note on the expected position of the
threshold ( 2 4.5) below which Im F(q2 ) can be expected to be zero.
For a charge density e 1r classical physics gives F l/(Q2 + a2 ) which has
•
a first singularity at Q 2 -a2 • A superposition of exponentials ��f(A)e -ArdA

•
gives J� ...i 2
Ai. l dA giving a continuum of singularities the lowest A€owever
Ao Q + A
being Q2 -A 2 the slowest exponential tail. This idea is valid relativistically .
•
0
•
(It has been shown that the position of the singularity is the same as the position
of the singularities in Feynman diagramsof a process as if real particles were
virtual) .
For the proton the virtual state N + n+ exists with a extending out as
n
(a gradient of) e -µ r/r. Its square, the charge density involves polynomials in
l/r times e -2 µr , the longest exponential is therefore 2 µ, and the threshold at
2 µ.
A more interesting case is one say like the E, of mass M1 which virtually
can emit a K (mass m) to be a nucleon mass M2 , with M2 M1 • We obtain the
<
singularity by studying the perturbation diagram but can see its position
physically this way :
We ask : how sharp a rising (time independent } exponential electric
potential can we tolerate before we get a divergence beca\18e the fie l d rises
faster than the charge densi ty falls ?
Thua we have a field of momentum Q ( i t is pure imaginary • ia} and are j ll8 t
able t o make the K meson from the t. Th e virtual state i s a particle of mass M2
momentum O , and one of mass m momentum + Q 2 where the initial s tate is one of
mass M1 of momentum G/q (Brei t frame , 111 • 0) .

The condition of j us t divergence is that this s tate j us t has zero energy
denominator
(this equation is relativistic because it conserves both energy and momentum} .
This has the solution
2 2 2
2 2 2 2 2
Q •-4m + (M1 - M2 - m ) /M2
-a •
if Hz < M1 - m (o therwise the square roots are not pos itive . ) The threshold
2 2 2
2 2 2
is at q +a •
That i t is below the energy 4m needed to make a pair , which
was the first guess , we call this an " anomalous " threshold .
2 2 2
For M2 > M - m the equation cannot be s olve d , the s ingularity now
1
2 2
really comes (normal threshold} when the pho ton can firs t make a pair q 4m ••
Note on the quark model of Feynman, Kislinger and Ravndal.

The form factor for the proton by FKR is
GM
µ
• (l
exp (1- 2
20 4M
L)) n •(1 GeV)
2
M Mass of the proton
•
which is very poor - it cuts off too fast at high -q 2 (there is no reason to
expect the model valid for large Q2 ) . For small q2 it goes as about 1 - .7 q2
instead of 1 + 2.6 q2 In the model the photon coupling is direct (processes
like VDM are not considered if they were we would add a factor (l-q2 /mp 2 ) -l or
1 -1 . 7 q 2 giving better understanding ) . The ratio GE µ /GM comes out 1 + q2 /2'1'1
2
or 1 + , 5 q2 which falls too fast with q2 (experiment is around 1 + .06 q 2 ) .

(Another way of dealing with the quark equations giving a much better form
factor, but much worse fits to photon electric matrix elements ls given in
Fujimara et al. Prog. Theor. Phys. 44 193 (1970) .)
For the pion FKR would get (m 2 11 2 • .02) .
IT
•
which is patently absurd (for example, having a zero at q2 -0 .08!) It is by

•
all odds the worst disaster of their model -(it comes from the ad hoc way they
dealt with the spin of the quark unaffected by the photon) that one would think in
honesty they would include it in their paper. They didn't because, surprisingly,
they didn't think to calculate it when they wrote their paper!
Electron - Proton Scattering. Deep Inelastic Region
Lecture 25
Other Photon Processes for q2 � 0
Electron beams permit us to study reactions such as e p � e x where x is

some hadron final state .
2
It is controlled by the amplitude <'U2 Y µ u1 ) -
411e (x l
µ
2- J I P > in which only
q
the factor (x IJ IP > is unknown , and so we can measure this matrix element (or
II
rather its square) . The mos t novel feature is the ability to vary the q 2 o f the
virtual photon away from q 2 • 0 ( real photon experiments , which we already
2
discussed ) to q negative , and in fact to the far negative region . In addition ,
very energetic (high mass ) s tates x can be excited in the high energy machines .
A great deal o f information is becoming available , and there are many theoretical
2
questions especially for the new large q and high mass s tates x .
Th e simples t experiments to do are those which d o not require studying the
x s tates at all (sum over possible states x ) but j us t s tudy the ele ctron beam -
125
its def lection and energy determines the momen t11111 transfer and its energy loaa
determines (in the lab system} the energy of x above the proton mass . these
experiments were done first and are moat complete - we discuss first therefore their
results and theoretical interpretation (which will occupy us for many lectures } .
Then we shall return and discuss what we can expect the final s tates x to look
like and how they do behave experimentally as far as ia known .
Inelastic Electron Nucleon Scattering

'
E
E
The energy loss v •

E -E ' in the lab, of the electron is measured as well
as its 4-1110 me ntum trans fer q 2 . The invariants o f the problem are thus q2 and
p • q • Mv where p ia the 4 -1110mentum of the pro ton . All quantities measured proper
to the hadrons are expressed as functions of these two variables . 'Ihe prob -
ability we measure is then proportional to
In the case of unpolarized electrons ( there are no experiments with polarized

electrons and pro tons yet , but we shall discuss the theory later} , the sum and
average over electron s tates contributes the spur (u y v u1 ) Cu2 Y µ u1 ) •
2
+ S \1111
kl vk 2 µ - o µ v k i k2 > · While the hadron s tate
1 / 2 ap 0! Jt ) 2 (
2 yv l yµ . kl µk2 v
gives an expression we shall call Kµv
2 2
(25 . 2 )
L <x j J µ (q) j p> 2116 (Mx - (p+q) )
"\'
K µv •
x
(spin of p averaged , is also µv symme t ric part of expression for any spin) .
(Note , the mas a 2 of the final states which we are generating is expressed in terms
2
of our variables q , v as
2
Mx2 • (p+q) 2 • ; + 2 Mv + q 2 q is negative .
By varying v for given q2 we can vary Mx2 and look , for example , for final state
resonances , )
Electron-Proton Scattering. Deep Inelastic Region 127
From argUJDents o f g auge invariance and re lativity the synunetrical part

o f the tenso r � v (which is all that counts for unpolarized electron , and
is all that survives if the proton is unpolarized) mus t have the form
D •n. n •n
q qv
....).. -aII - 4
2 2 2 µ
M K
\I V
w 1 (q , v ) � ( 6 ,.:;- -
11 II
- • 4 w2 (q , v) (p (p . , )
-q µ v 2. )
...,. � · 2-
q
2 q q
(25 . 3)
Several properties o f W1 , W2 :
• 0
2
The apparent , but unwanted poles at q come from the way we have
wri tten our expression and of course cannot be real . To remove the highest pole
2 2 4 2
we mus t have ( p • q ) w2 • -M q w1 as q 2 + 0 o r w2 • -q /v w 1 order q as q + 0 , +
2 2 2
b u t w1 can approach a cons tant w1 (0 , v ) a s q + O . Hence for small q , w 1 • w1 (0 , q ) ;

2 2 2
w2 Cq , v ) + -q /v w 1 CO , v) .
2 2 2
The direct total absorp tion cross section for an imaginary pho ton of mass q 2
to hit a p ro ton would be
2
a • 4 11e "\"
� l <x l e J j p> j 2 2 116 (Mx 2 - (p+q) 2 )
2k2M µ µ
• 411e 2
e JI e V K \I V (25 . 4 )
2k2M
k • momentUJD o f photon in p res t system ,

We have two polarization cases :
P
Transverse , e 11 is perpendicular to q11 and to µ
2
a 4 11 e 2
c • - k- W1 (25 . 5 )
Langi tudinal . Some t imes called scalar
e t • q /v -q -
z /:f ' ez •
(25 . 6 )
These two cross sections can be used as parameters ins tead of w 1 , w2 •

From the above formulas
wl
•
f1
+ v2 2) (�)
a +a (25 . 7 )
-q
w2
t s
\
It is convenient to de fine
Combining the electron spur to the expression for Kµv we ob tain the total
cross section in the lab .
-1L - _,..._
2
__
.. a_..,.... 2
8cos 8/2) w2 + 2 (sin2 8/2 ) w� (25 . 8)
dndE ' 2
4E sin4 8/2
In principle it is possible to get both w1 , w2 , this is done in several cases -

but mos t accurate is w2 and o ften what is given is w under some assumption
2
about w1 • For example we have
(25 .9 )
£ • � + 2 tan2 0/2 (1 + � J
-l
-q
and often data is given for a t + co s without resolving them , £ is often small .
P roblem: Show that for proton e las tic scattering
2
o (v + q /2M)
Lecture 26
Jnelas tic E lectron Nucleon Scattering (continued)

We shall now give a kind of preliminary descrip tion of the experimental
results - some details will be dealt with further later . First a use ful variable
to discuss the lower energy resul ts in is the mass 2 of the final s tate Mx2 •
Mx 2 M2 + 2M v + q 2 We shall , at first , use M 2 and q 2 in our discussion . As
•
x
a function of Mx2 we , of course , get nothing below �
then a large elastic peak
2 2
corresponding to Mx M when the final s tate is j us t a proton - proportional
•
2
to o (q + 2Mv) - and depending on <\t• GE as we have previously studied . This
2 2
contribution falls as q2 rises as 1 / (1 - q / . 71) •
Next , for fixed q2 , as M 2 varies , we see a "firs t " resonance at Mx 12 36 , •
x
a "second" one near Mx • 1520 and finally a " third" around 1700 and some indication
of a fourth at 1900 . Theo retically we should see more resonances , e . g . 1535 but
undoubtedly the "second" peak is an unresolved mixture of these two while the o e n
near 1700 sometimes called the 1688 resonance is probab ly that with four others
expected near that energy , The resonance at 140 7 (the Roper resonance) has not been
seen in these experiments .
These resonances can be fit ted as Bre i t -Wigner p eaks on a background which
2
gradually dominates into a smooth curve as Mx increases .
How do the resonance strengths vary with q 2 ? At very low q 2 the beh avior
depends theoretically on the angul ar momentum of the state and star ts as an
appropriate power of Q (Q 2 -q 2 ) . For higher q2 , however , the s t rength of the
•
resonances all fall more or less as does the elastic peak . In fact if one plots
the ratio (da /dn) with -q
2 one finds curves which ri s e rap i d ly
res / (do /dQ) elas tic
from threshold ( pho toelectric) values to saturation (in the vicinity of one or
2
just below) for q 2 > ab out 1 GeV •
For larger q 2 and large v

there is nothing left of the resonances . There
2 2
i t was suggested by Bj erken that the func t ion vw (q ,v) (and also w (q , v ) ) should
2 1
2
be a function o f the variab le x • -q /2 Mv only . (Data is often presented also
as a function of w 2Mv / -q 2 • l/x) . This has turned out to be rather accurately
•
t rue , This feature , that as -q2 + m , v + � such that -q 2 /2Mv • x is kept fixed ,
vW2 Cq 2 , v ) approaches a func tion of x, F (x) is known as Bj orken scaling . We shall
have a gre a t deal t o s ay late r of i ts theore tical s i gn i fi cance ,
Present (January 19 72) data is available in the following regions ,
20 / w-2
I Resonance Region
q2 II S caling Region
III W1 , w2 S e p ara t e d
10 �3 4
---
w c
2 2 In the
In the re gion 4 , s caling is observed above Mx • (2 . 6 ) •
2
region 12 w > 4 for Mx > 2 -q > l; vw is a constant within errors and s cales
>
2
in any variable .
Tii e data for vW2 looks as follows
l
f0 vW2p . 18 l for the proton
dx • t
J vw2ndx "' .12

1
for the neutron
0
Data fo'l" the neutron (by subtraction from the deuteron) is also available.
Tiie ratio to w2p looks like
l
l
It is consistent with a straight line (l - x) from x • 0 to . 75 . lbere are
indications that it curves away above the line for larger x and could reach a
value as high as . 4 ( ? ) for x + l, but higher values are probably ruled out.
In the region where w1 and w2 can be separated, values of R can be found -
they vary from 0 to O . S with large errors so no definite trend can be seen.
If R is assumed constant it is . 18 . 10 , the data is also compatible with
t
R • . 0 3 (-l) /r(- or with R � q 2 / v 2 . In the latter case we would expect with

scaling that R + 0 for fixed x.
lbeory of the Inelastic Electron Proton Scattering
We now begin rather long theoretical discussion of the inelastic e - p
a
scattering. We shall first discuss the "deep" inelastic region -q 2 + v +

"'
•
00 ,
-q2 /2M x, or region of Bjorken scaling. First we shall discuss what is

•
called the parton model briefly; then we shall discuss general properties of w2 ,
w1 and see that they are connected to the commutator of two currents . Finally
we come back to the parton model in more detail, discussing suggestions that
the partons be identified with quarks , etc . - and then discuss the relation of
this to a more abstract representation called Gell-Mann's light cone algebra.
lbere then follows miscellaneous discussion, ending with attempts to understand
things in other regions of the q 2 , v plane (i . e . for smaller q 2 , and , or at

resonances) etc . , (actually our di scuss ions will not be so neatly organize d , we
shall wander among these subj ects ) .
Parton Model
Parton Model
The parton mo de l is the conceptually eas ies t to understand , al though i t
appears a b i t special , a s special assump tions seem t o b e made . Mo re general
abs t ract cons iderations are therefore more s a t i s fying , b ut fi rs t we dis cus s
the more e lement ary viewpoin t , I n dis cus s ing these mat ters it is bes t to
keep in mind all the princi p les of relativity , quantum mechanics , uni tarity ,
analy ti city , e t c . One way to do this is to work in a conceptual model whi ch
satis fies all these principles s imul taneously . There is no known s imp le model
wh ich does this except field theo ry (and that may not do it - all examp les
diverge ! ) , and that i s a very complicated mode l indeed . Nevertheless we shall
try to see wha t field theory migh t s uggest .
In field theory the wave fun ction for a s tate , such as a pro ton , could be
given by giving the amplitudes to find various kinds of b are field parti cles
moving wi th various momen ta . These b are field parti cles we call "par tons " .
Some times some phenomena can be unders tood dire c t ly in terms of this wave
function but us ually ma trix e lemen ts to wave functions of o ther s t ates mus t be
analyzed . The wave funct ion , however , is not eas ily trans formed from one
re lativi s t i c sys tem to ano ther ( the Hamiltoni an mus t be known ) be cause it rep re
sents a slice at a given t ime . Ce rtain properties are there fore more eas i ly
1 32
Parton Model 1 33
seen from the wave funct ion in one sys tem than in anothe r . The deep ine las t i c
s c a t te ring behavio r i s bes t un de r s t ood from the wave fun c t ion for the p ro ton
wi th extremely high momen tum P ( in +z direction) . In fact we s tudy the limiting

form o f the wave fun c t ion as P+ � .
In what variables will the wave function have a limiting form? From a study
o f the character o f typical field hamiltonian in a few examples , and from a study
of very high energy inclusive hadron collisions we conclude that i f the momentum
o f the parton is measured proportional to P as �p and the transverse momentum
in abs o lute uni t s k then the ampli tudes depend only on �. k as P rises - (except
for � so small that �P i s a few GeV , � of order 1 GeV/P are called ''wee" -
the dis t ribution of"wee" partons is probably bes t described in absolute p z
k variables . As we shal l see the "wee" region is hard to analyze - b u t the
main features o f the deep ine las t i c scattering does no t involve them) .
The fac t that k� i s fini te is !!.2!_, at leas t in any obvious way , a direct
consequence o f field theory - ( in fact perturbation theory does not give this
result , and therefore must no t be reliable here) it is simply guessed at from the
ubiqui tous result o f high energy colliss ions that the transverse momenta
available to the products averages about p� (.4
-2 • GeV)
2 •
But granting this , the s caling is expected �p from field theo ry (although
today we coul d base thi s also directly on experiment , it is the s caling l aw
for the longitudinal momentum of the products in very high energy collisions ) ,
Some s ugge s t i on o f how i t works is this . The ampli tude that a s tate of energy E
is seen to be made up o f two parts of energy En • Er + E2 is dominated ( in
per turb a t ion th e o ry) by a f acto r
(26 .1)
z p1 + p .
But the component of momentum of the parts is that of the whole P •
2
Set p l • f; l P2 • P, � 2 p' � 1 + � 2 l, E - --.//Pl 2+M2 - P �2 p
� + = M
I f 1 i s th e mass of
the firs t part and k1 i t s transverse momentum, we have

k 2 2
+M
E 1 - Jr. 12 P2+k/+M/ � �l + -1--1- approximately . (26 . 2 )
2 f;l
k 2+M2_2
Likewise
� + _2__
p
2
2 f; 2 p
Therefore A/ (E - E1 - E 2 ) •
(26 .3)
(The P in the numerator is usually absorbed in normalization or in the form

for A , the essen tially complicated behavior of this is a function of t1 , t2 ,
k12 , k 22 • Arguments like these were originally used to predict the inclusive
scaling, and they have been confirmed by experiment ,
Furthermore we see t1 (outsi de the wee region) must b e positive - no
partons are going vigorously backward - because (26 .2) is not valid if c1 is
negative , it should be l t 1 1 on the righ t hand side . Then (if t 1 is negative)
E - E 1 - E 2 � (1- I t 1 i -t 2 )P • 2 E; 2 P so the denominator is no t small but large
2
and the amplitude is l/P smaller than the preceeding case .
Lecture 27
Parton Model ( continued)
We envisage the proton of momentum P as being made of partons of momenta
E; i P all sharing ill various proportions ti the momentum o f the proton , all E; lie
between 0 and 1 (else some other would have to be negative since I ti • l) .
We shall therefore think of the incoming proton as a box o f partons sharing
the 1110mentum and practically free , Another way to look at this is to take a
dynamic view of the parts in the res t sys tem and assume finite energy of inter -
action a1110n g parts so as time goes on they change their momenta , are created or
annihilate d , etc . , in finite times . But moving at large 1110mentum P these times
are dilated by the relativistic trans formation so as P rises things change more
and mo re s lowly , until ultimately they appear not to be interacting at all .
When the proton is hit by the photon the interac tion operator Jµ couples to one
parton or another and knocks it to a new state of momentum p 2 • p 1 + q (p • nearly)
l
E;P
2
If the parton had mass m we would expect a rate proportional to
(2 7 . 1)
-
Pi
Parton Model 135
(the factors l /2E 1 and l / 2E are included because of the normalizations of

I M l 2 and K)JV ) , In fact, due to the interaction energies assumed finite , the
energy loss is not just that for a free particle but differs from it by an
unknown (but finite which may vary with x, etc.)
For spin zero partons
I M l 2 • (p l µ + P 2 µ > (p lv + P 2v > · (27.2)
For spin 1 / 2 partons
=(2 P 1 )J p 'V, + 2 p1V p2 )J - 2 6 )JV (p 1 · q >J (27.3)

(We shall calculate for spin 1/2 , just state results for spin 0 , ) To make
things easy we will omit all terms in q ll in K)Jv because we know how to get them
by gauge invariance . Thus in j M j 2 put p lv in place of p2v . In our limiting
circumstances we can write p l µ tP ll to an excellent approximation where P ll is
=
the four-momentum of the proton (strictly it is valid for the z component only,
but we have seen that implies it for the t component E 1 • f;E; and the transverse
components being only finite are relatively small) .
Therefore if
f(x)dx • number of partons with momentum between x and x+dx each weighed
by the charge squared (in units of electron charge) we have
K)JV • !.M ( 4 P )J Pv W2 - 4 6 )JV M2w1 )
· J f(C) (4 C2 PI' Pv _
(27 . 4 )
c
(the first l / C comes from the normalizing factors 2E/2E 1 ) . Inside the 6 function
we have (CP+q) 2 -DI2 - fl • 2E; (P • q) + q2 + f; 2 !r2 - m2 - but as -q 2 .,. .. and P ·q
fl ;
• Mv goes to with -q2 • 2 Mvx this becomes 2 Mv (f;-x) + (finite) or nearly

00
6 ( 2 Mv (C-x)) _.!._ 6 (C-x) , Therefore we have, integrating the 6 function,

=
2Mv
for large v
4 P P W - 4 6
2
)JV 2 )JV M W1 =
Thus
• f ( x)
2
2MW ( q v )
1 1 ( 2 7 .5)
and we have functions of x • -q

2 / 2Mv only .
If we used s calar partons , the formula for v W2 is unch ange d b ut w1 • o .
If the fraction of the partons which a spin 1 / 2 (weighed by their charge squared)
(l+ \J
at momentum x is y (x) we ge t w unchange d , 2MW • y (x) f (x) . The ratio
v 2 1
w -w
- -"----
__-q
cr 2 1
R m
s a
in the scaling limit would be
(2 7 . 6 )
In the region s t udied R is o f o rder .18 t 10 so y is less than .2, not many
scalar partons . A much 100 re likely hypothesi s is

The curren t carryin g partons are all sp in 1/2
Th e value of R remaining today experimentally being due t o our having n o t enough

2 2 2
energy and q to be fully in the scaling limi t . E . g . , R "" q I v would also fit
the data and give zero in the s caling limi t .
This i s a very profound conclus ion about the s t ruct ure o f the underlying
theory for hadrons . We mus t wa tch to see if R really does approach zero as the
scaling limit is reached .
If charged partons all carried the fundamental charge te then charge
� xf (x)
squared is 1 in our units and thus we could s ay only 18% of the moment um is
carried by charged partons in the proton (be cause • , 18 t . 0 1) - the

remainder , 82% would be carried by neutrals . This 1 8% is surprisingly s mall .
If the partons are quarks and carry charges l ike t 2/3 or t l/ 3 the percent age o f
momentum carried by the quarks could be higher. We shall discuss such a model in
more de tail , b ut it turns out even then it is neces sary to assume something e l s e
neutral carries part of the momentum.
Lecture 28
The Wee Region

The ideas leading to the s caling formulas (2 7 . 5) are very reliable . By
Parton Model 137
making further assumptions about the wee region we can unders tand other aspects
o f w1 and w2 • It mus t be realized we are now elaborating on our original parton
ideas developing them further to understand more features of the f (x) curve . The
data for vW2 seems to approach a cons tant ( . 32 ) as x + 0 which means that f (x)
would go as . 32/x showing s mean nuui>er of partons ris ing as x falls into the
wee region (such that the number of wee partons is finite and independent of P ,
and the mean total number of partons in a state of momentum P rises logarithmically
wi th P ) . lbis is not entirely unexpected , it is the same as the dis tribution of
p roducts in hadronic collis ions . The way this appears to happen can be gathered
by studying the field theory equations at high energy , and also the perturbation
theory of bremsstrahlung . In the latter case neutral particles are generated
J
in a dk/E • "' dPz / pz 2+� -tta
2
dx/x distribution . These neutrals can generate
+
pairs so the small x region contains large and nearly equal numbers of particles
and antiparticles . The field equation approach sugges ts the same thing; and
further that the low region is generated in higher order perturbation from the
higher momenta by a series of cascades x + x' + x" going down in x . In either
case we conclude the character of the pai J:B will be as a whole neut ral and
therefore the same for proton and neutron . lbese expectations have been made
by other means also , for example , the dx/x leads to a constant cross section
for virtual photons (of fixed but large , negative mass squared) , That these
cross sections should be a cons tant is expected from cons iderations that these
pho tons have an amplitude to be virtual hadrons ( like the p) and hadrons give
cona tant cross sections (the magic word "Pomeron" is used to "exp lain" this ) ,
the same for proton and neutron . Thus we expec t vW2 n to equal vW2 as x as
P
+ 0
indeed they turn out to do ; both f (x) for p and n app �oach the same . 3 2/x for
small x. We also know experimentally that at q 2 • 0 a • a to 3% at 16 GeV .
YP yn
lbe fact that f (x) seems to go as . 32/x as x + 0 implies that the total
cross section for virtual photons of energy v on protons is s cons tant as energy
goes to m for fixed and large negative q 2 , j us t as the real photon cross section
does . We can expect , as long as we are on the high side of the wee region , that
x • b/P with b large enough that , as our meas urements tell us , f (x) ""'" . 3 2 / x .
Hence the transverse virtual photon cross section is
1
w f (x ) • !!...!._ f (x)
2 2 2 2 2
0t !!!_ �
•
k 1
•
/22· 2M v 2M (28 . 1)
,._j v · -q
in the s caling limi t . And for x on the high s ide of the wee region (-q 2 large
enough , )
- 115 µb c · 32 �v2) (28. 2)

-q
At this point we may well guess ( these considerations are independent of
the parton model) what happens for v large b ut -q 2 is not large enough that
-q2 /2Mv is far enough out of the wee region - 1 GeV/P . We know for q 2 • O
2
the total pho to-cross section is independent of v . For o ther -q , the unreliable
2 2
VDM would give a factor -mp f { l -m )
p
in the amplitude or (1/ ( 1-q /mp 2 ) ) 2 in the
cross section times a v independent term ( large v ) . 'nlis is clearly wrong for
2 2 2
large -q ( far from the pole) for it falls as (-q ) -2 , ins tead of ( -q ) -l as
p
we have just seen is experimentally found, above . But we can cer tainly guess
that , for each -q 2 the cross section is constant but dependent on -q 2 for large v ,
a
2 2 2 2 2
t 4 w e C (-q ) . We do not at presen t know C (-q ) excep t fo r q
• 0 and large •
2
-q • We need a good theory for this function .
Although the number of partons is infinite , the momentum contained in
them is fini te , of course , because the total momentum of all partons neutral
and charged is 1 (units of P ) . We have for the momentum carried by all charged
�
partons weighed by e 2 xf (x) dx • . 18 t . 0 1 .
0
Formula for R
We may get an idea of how R behaves as we approach the scaling limi t by
calculating a s directly in this region . We work in the system in which q is
pure space-like
p (E , 00) p - l..:.i. • (E , O , O , O )
µ II
P,
q 2 qll
q q
q (0 , -2Px , 0 , 0 ) 6 ll - - ll V .. diag . ( 1 , o,
-
q2
-1 , -1)
V
2
Use Equation (2 7 . 4) with all q put back into I M l
11
-
11
M (4P11 P V w2 - 46 llV ; W1 ) •
Parton Model 139
(the expression in the 6 function becomes (4 P 2 x(f;-x) -ti)) .

Now to get CJ s
we wan t the t component of both sides ; however , if the RHS were exact the
z component would exactly vanish (from gauge invariance) . To avoid erro rs of
di fferences of large quan tities we take µ • t-z , » • t-z so oµv vanish.es ,
p lµ p 2 v (t 1 - p 1 z ) (t 2 - p 2 z ) (this is equivalent to replacing the polarization
+
of the photon e µ • (1 , O , O , O) by e ' : ( 1 , 1 , O , O) ; we should ge t the same

µ
2z runs backwards so e: 2 - P2z
result by gauge invariance) . Now p ""' 2 j p 2z l •
' /. 2 2 2' +m2

2 (2x - f; ) p . But p lz runs forward so t 1 - P l z ,.,,,..j P lz +k, +m -P 1z .,,,;o kl2 p •
Here ki2+m2 is the mean perpendicular momentum plus the maSB of the parton (at f;) ,
2
whatever that means , but at least i t is the square of some fini te energy . o s
and R are given by
4 (k�+m2 )
( 4E 2w2 -4 m2w1 ) • 2Mf (x) 2
-q at x
R •
2
-q
Thia explicit dependence on m2 is erroneous . The deviations in the parton

model of calculations using free partons are probably to set :!:ti (a binding energy
correction) uncertainties on all effective parton masses . In this special case
the error comes in assuming in e: 1 - p lz that e: 1 ,is j ust the kinetic energy of the
free parton and not more complicated . When operators involve d/dt in S chroedinger
perturbation theory their exact expression is usually more complicated than
j us t the kine tic energy operator . The formula for R should be
R • (28 . 3)
at x
The expected falling trend with l/ (-q2 ) is not evident in the data , but
the errors are large . If we estimate k2 � . 25 (GeV) 2 independent of x as in
2
hadronic collis ions , and m of the same order (else we could not explain why
kJ. is so small) , we find R � (2:!:4ll) I (-q 2 ) ; for -q 2 • 7 (near center of data)
2
R :11> • 3 :!: 4ll/7 not unreasonable compared to the average . 18 .

Lecture 29
'Ole region near x • 1
We can ge t some qualitative ideas of the region near x • 1 by firs t s t udying the
extreme case x • 1 , the elastic form factors , at large -q 2 We take the
coordinate sys tem wi th q • (0 , -Q , 0 , 0 ) a pure space-like vector . P the
µ
momentum o f the proton • (E , P , O, 0) mus t have P • Q/2 , after the collis ion
the proton has moment um P � • P µ" 'lµ • (E, -P , 0 , 0) 1 with z component in the
opposite direction . For large Q, hence large P, we can des cribe the ini tial
s tate o f the proton as an ampli tude for various con figurations (in x, kl ) . Le t
us suppose for example , the configurat ion contains two non-wee partons and two
wee , and draw a pic ture :
x2
!
2
�
Be fore
Xl
After x2
! Figure 29 . l
Now we ask that the final state ( the "after" picture) b e only a proton -
that is, a factor < Proton at -P I "after s t ate ">, the (comple x conj ugate of)
amplitude tha t the after s tate appe�rs in the pro ton wave function . But a le ft
moving proton can never ( for large P) look like the after pic ture because it
contains a backward moving parton (x2 ) , (and any estimate of i ts mass square is
2
order P 2 , requi ring order P 2 b inding energy contributions to ge t the mass down
2
to Mp ) .
'nlus insofar as the proton looks like "before " it contributes very little
to elastic s cat tering. To scatter elas tical ly we mus t s tart wi th a configuration
where one par ton has nearly all the momentum , x • 1 and all the res t are· wee ,
say below l/P .
Before
li---1---· �
1
After Figure 29 . 2
Parton Model 141
Now the s ingle fas t one is reversed by the incoming pho ton , the energy
is not changed much , and the final configuration is pos sible fo r a pro ton
because wee partons ( those wi th absolute p finite say below order 1 GeV) can
z
go backwards as well as fo rwards . The refore the s cattering ampli tude ( form
factor) is proportional to the square o f the ampli tude that a proton looks
like "be fore " , hence proportional to the probabili ty that the pro ton has all
partons but one in the wee region .
We now see quali tatively why this should be some inverse power of Q in
the following manner . If the b reml5trshlung analysis of the "sea" (dx/x)
region has any qualitative reality we can judge the probability that no parti cles
have their x above x of order l/Q to be e -cn whe re ii is the mean number expec ted
0
above this ir,, and c is some cons tant to allow for correlations (unlike first order
bre-trahlung which is purely Poisson , perhaps each "photon" could make two of
some o ther object , etc . , so that the mean number of particles would be higher but
proportional to some vaguely defined "mean number o f s t atis tically independent
event s ") .
Now we have seen , for small x at leas t , n goes as xa/x where a is
J
l
another cons tan t - hence ii, the mean number above x0 is a x �
• a ln x -
1
o
a ln x where x1 is some unknown upper limit where the dx/x ideas b reak down .
0
Thus for small enough x , e -en varies as exp ( ca ( ln x - ln x 1 ) ) as a cons tant
0 0
times a power of x • That is , the probabili ty no "sea" particles are outside
0
the wee region falls as a powe r of Q .
There may b e some special partons to make up total quantum numbers which
are not parts of the sea which mus t also be in the wee region . But s caling
studies of the field equations as well as o ther theore tical arguments (related
to Regge-type behavior) sugges t very s trongly that they would show behavior o f
the form xa dx/ x where a ! O, and thus thei r probability to be found in the
region below x would also fal l as a power of x0 •
0
We conclude that the probability the proton looks like our parton near
I
x 1 with all the others be low some low limit x falls as a power of x , s ay
•
0 0
as xY .
0
In particular then for x wee of order l/Q , we find the p robability
0
varying as Q4 •
Thus the form factor GM of a proton (and GE als o , see below) should fal l
as a power of Q , Q-Y . Experimentally y is probably a lit tle larger than 4 .
The fact that the form factor falls toward zero as Q + m means that a
proton (as P + �) has a zero amplitude to appear as a single parton . This
may be because no sin g le parton has the precise quantum numbers of the p roton ,
but even i f i t did , a more cogent reason is that the probability of finding a
parton without o ther fields generated by interaction from it (represented by the
sea dx/x) is zero , (j us t as is expected from the theory of b re1111Str ahlung) .
Our analysis of the ratio R • 0 8 /o t des cribed in the previous lecture can
be as well applied to the elastic piece x • 1 . We therefore find
2
as 4 (k� + m :!: ti)
0t I elastic
•
-q I x - 1
+ m2 :!:
2
where k.L. ti is the value for the s ingle parton with nearly all the momentum .
But this ratio exp ressed in terms of G ' G is simply
E M
Thus we expect G /G to approach a constant (k� + m :!:ti 1/M- -'-

2 2
( M • proton mass ) .
2 E M
for large q A great prob lem is found when we es timate the cons tant from
If G :"I µ GE µ • 2 . 79 we find (k� + m :!: ti) 1 • . 11 (GeV)
2 2 2
experiment .
M
disturbingly small ! But data at q • 2 . 6 even shows G /G is les s than 1 ,
2
E M
about . 6 (and falling ? ) so (k.i... + m :!: ti) 1 < . 04 ?
2 2
Now we can see how f (x) should vary for x near 1 , say x • 1 -y with y small .
Let us find the total probab ility that x exceeds 1 i . e . 1 f (x) dx . We mus t -y {�
again find one parton with nearly all the momentum, the total momentum of all
the o thers cannot exceed y , That i s not the same as the probability that none
can ex ceed x0 • y which we worked out as y -Y , but a little cons ideration of the
problem soon shows that it does vary wi th the same powe r , but with a di fferent
cons tant in fron t . (This i s mo s t easily seen b y a kind o f dimensional analysi s ;
o r alternative ly one can work out the case o f a dx/ x Poisson dis tribution in
detail to check i t . ) Hence
J1-y f (x) dx .. yy
1
Parton Model 143
for small y, or
f(x) ( 1 - x) y -l
�
for near l . That is , we expect the power law of the elastic scattering to be
x
related to the power law of the behavior or f(x) near x 1 . In particular we

•
expect f(x) (1 x) 3 or a little higher. Data is not in disagreement with

� -
this . (This relation was first published by Drell and Yan.)
The region -q2 large, � finite

Resonances . Can we extend these ideas to get some understanding of why the
resonances fall with about the same power of Q for large Q2 as does the elastic
peak? (Here we are studying the region large i 2Mv -q2 , more precisely
- , �
Mx2 �2 + 2Mv - (-q 2 ) finite.) Perhaps it is better to frankly tum the

•
question about the other way and ask what the fact that the resonances fall
off as the elastic peak with -q2 tells us about the wave function of these
resonances .
If we excite say the 6+ our picture again cannot look like figure 29 .1
for the "after" picture could not be a 6+ moving rapidly to the left because
it contains some partons going in the wrong direction. We must instead have
as in figure 29 . 2, but this time instead of asking whether the last picture
looks like a proton1 we ask: with what amplitude does it look like a 6+? We
again expect a to have a sea region, etc. , and to have a difficulty measured
6+
by a factor Q-y'/ 2 (in amplitude, for a proton Q -y / 2 )to appear with a single
non-wee parton. We conclude from experiment the y ' probably equals and
y -
can probably invent a posteriori argument to justify this . (For example , the
sea may be very similar in the two cases because they are generated by the same
parton - that is , the chance this parton is bare of non-wee particles is what
the factor measures - and it is the same parton in either case.) The wee partons
left over from the original right moving protons have only a certain amplitude
to be in the correct proportions (of numbers non-Q dependent momenta, etc.) to be
appropriate to a left-moving proton - and a different amplitude to be appropriate
to a 6+ . It is the ratio of these non-Q dependent amplitudes (squared) which
determines the ultimate ratio of the probability of producing 6+ to proton at
large Q2 • (It is always possible in principle of course that for some particular
resonance this amplitude is zero and such a resonance would fall fas ter with Q
than the elastic peak . )
These relations o f power laws for elas tic peak , resonances , and asymptotic
f (x) are so interes ting that it is important to no tice that they can be expressed
in a general principle ( I am st rongly indebted to J . D . Bj orken for discussions
on this idea) quite apart from any parton theory interpretation . S tated in
this way they are seen to be rather profound and represent a fundamental new
property of high energy collisions ( for corresponding relations are expected
relating exclus ive and inclusive hadron reactions at high energy) .
2 2 2
Suppose we plot vW2 for a given large q against Mx - M • 2Mv (l-x)
2 2 2
'"" -q (1 -x) (x is near 1 and 2Mv ""' -q ) . We note the scale of Mx - ; and
2
1-x are j us t proportional . We find a set of resonances at app ropriate Mx
Elas tic Peak
Mx 2 -M·��--t-�
2 2 2 2
and the tail of the "background"curve ( 1-x) y-l dx or (-q ) -y (Mx -M ) y -l d (Mx ) .
2 • 2
Now when we increase q al l the resonances fall (but stay in the same place Mx ) ,
2
as (q ) -y as we have seen , but the si ze of the "background" curve (at a given Mx )
2
2
also falls as (q ) Y . That is the "background" (by background I mean the high x
tail of the scaling region function f (x) ) bears a fixed ratio to the resonances .
p
It is not possible by going to higher q 2 (or in• hadronic collisions to higher s )
to separate even more clearly the resonances from the background , nor do the
resonances fall into the background and get lost . Put another way , the back-
ground can be consis tently viewed as due to overlapping resonances or tails of
resonances itself and all resonances ultimately fall off the same way (as a
2
power of q as q2 increases ) . Bohr would like this , he would say there are two
complementary ways o f viewing the background , as continuum single particle
states leading to scaling expectations , or as large numbers of resonances - and
nature conspires to make i� impossible to decide in the intermediate region
2
(large Mx - ; ) which is corre c t .
Parton Model 145
Lecture 30
Argumen t that y ' m y
We will the assumption that for any state (e . g . p or A)
make as momen tum
Q increases the distribution wee x (• finite among ap proaches a definite momentum)
distribution of Q .
in dependen t mo re explicitly for our case ,
Rathe r , ass ume we
that the a p l i tud for the proton to be

m e single parton a given distribution
a and
amon g wee momen tum ( e . g . below 2 GeV ) and nothing in between falls somehow with
Q b ut the among the wees approaches a de finite asymp tote , (e . g .

dist rib ution
amplitude for no for

wee /ampli tude 300 MeV / ampli tude for two a t
one o f momen t um
100 MeV nd 400 MeV e t c .

a constant ratios) ,
be come
<one p ar on at x ""' 1 ,
t es we
k'sJ Q>• Pp (Q) fp , i (k ' s )
at momenta p ro ton at
(the index i indi cates parton type) an d s imi larly for a resonance . Hence the
a
sc a ttering of photon, wh i ch turns arolmd fast parton from Q Q ' (e . g .
the to
in a coordinate q • (Q-Q' , O, 0, Q+Q' ) ) .

sys tem where
�
l., ,.,
Before p Q
. �
After A
•
Q'
t . . ->J
The ampli tude is p roport iona l to
We explicitly suppose the las t factor is not (For certain resonances , zero .
or quantum number I spin) reasons i t might be

(e . g . hos may fall fas ter ze ro , t e
wi th Q, but study the ones which are not spe a in this way , ) The answer
we ci l
muat be relativistically invariant, however, and be function thus a only o f

2
q • 4QQ ' . Hence FP(Q) FA (Q ' } ia flmction of QQ' only, from which we can
a
deduce that both Pp (Q) and FA {Q) be of the form 1/ (2Q)y/Z (a con s t ant
mus t
can be abs o rbed in the s e cond factor) .
Thus amp . p + A • (q2 ) -Y l 2a (p , A}

whe re
a (p , A ) • \
L .
ff' i (k} f Ai (-k) d3k
p,
i 2
depends on the s t ates p , A, but not on q •
If two states have di fferent powers of Q in their F (Q) • Q-r /2 then they
.J"
3
mus t be orthogonal in the sense 7r, Ck) fb (-k) d k 0 - usual ly because of •
di fferent quantum numbers , like charge s trangeness or isospin (or z angular

momentum) in the wee system .
The principle dis cussed at the end of the previous lecture very likely applies
to high energy hadron co llisions also ; although not our direct interes t in this
course , I will briefly outline how it might work there .
Consider an exclusive reaction (at fixed or zero momentum trans fer t) A+B +
C + Resonance X, where A , B , c, are fixed . Let the incoming momenta in the

c .m. be p .and the outgoing momentum of C •
2
Pc px , s 4p goes to infinity ;
• •
2
and imagine we meas ure the missing mass of the resonance by � • • P X2
2 2 2 2
( PA + PB • P c > """ s ( 1-x) + MA +Ha -M Thus plot resonance data as M"" •
2
C •
2 2 2 2
+MC -MA -� � s (l -x) . As s varies resonances s tay at fixed M"" , of course , but
2
�
tleir size goes down very likely as a power of s, say as s ar cil- > dtl- ( a is negative) ;
as expected for exclusive reactions , where a is the leas t negative possib ility for
an exchange of something that acco1D1ts for the quantum number changes from A to C .
But since we are no t really looking at the products � this can be thought o f
a a an inclusive reaction where , i t has been argued , ( R . P . Feynman , Phya . Rev .
Lett . !! 14 15 (1969 )) the probability should be a flDlction of x only . Hence
2
this s caling region near x • 1 will go into the resonances for large M only if
a aF (rl- ) � • s a F (s (l-x) ) ds (l-x) i• a f1D1ction of x only , not of a. Hence this

-a-1
flDlction F (a (l-x) ) mus t go as (a ( l-x) ) -a -1 and the s caled result goes as ( 1 -x) dx
for x near 1. (There are situations , for example , where C and A are both pro tons
that it is technically hard to get to high enough x to avoid contamination of p ' s
disintegrated from other hadron resonances C', etc . , but these matters canno t
concern us in detail here ; the result baa not been readily checked , there fore . )
Lecture 3 1
We a ummar ize what we or surmise about vW and w1 in the high v region .
2
know
2 but 2
Region Large v, large -q - q / 2Mv • x finite . 2MW ia a function of
I)
1
2 2
x only , called f (x) . w2 • (-q /v ) W1 •
Parton Model 147
2
Region II) Large v , fixed -q as v + • ,
2 • 2
2
Mv g ( -q Z
function of q )
-q
2 2 2
Region I I I )Large v , large -q 2 but finite M -M • 2 Mv - ( -q ) ,
x
2 2 -y 2 z
2MW • (2Mv) -y+l function of M only • (-q ) h (Mx -M ) ZMv
1 x
2
(The las t factor 2Mv is the normalization from dMx to dx} .
In order that the three regions fit together we have ,
(I ; II) for small x , f (x) goes as a/x (a • . 3 2)
2
for large -q , g ( -q 2 ) goes as a .
( I I ; III) for x near f (x) goes as A(l-x) y-l
1, y 4 or 5 ""
2 � 2- 2 2 �
for large Mx - , h ( Mx M ) goes as A ( Mx - ) y
-l
In region I we expect o/ o t to fall with increas ing v as l/v or 1/ (-q )

2
•
In region II we expect o s / o t to approach some finite limit depending on -q 2

2 2
as v + (This means vW is finite , w i s of the same order as (-q /v ) W 1
m,
2 2
but is not equal to it . ) In region III we expect o s /o t to fall with increasing
2 2
v as l/v . In the special region of small q : w • -q w /v 2 + order q /v ;
4
2 1
2
as� q •
2 -y
General remark about the power law (q ) , namely it may have a logarithmically
-
falling coefficient (q2 ) y / (atnq 2+b ). In the discussion of the previous lecture
we saw powers where y could depend on the quantum numbers of the wee group , like
total isospin , angular momentum etc , These are discrete so the y i ' s are discrete
and there is a lowes t one . However , there is transverse momentum which is not
discre te ; ycould depend on the t rans ve rse momentum of the fas t p a rt on say as
• 2
Y Yo + Y 1k� so the result is
� -
(q 2 , ( y o+y 1k� ) dk; .
Y
( qz , - o
( 31 . 1)
Y1 ln q
2
Partons as Quarks
Naturally we should like to go further and find out mo re about what quantum
numb e rs the p ar tons carry . As it turns out from present experiments only a few
things can b e learned , but future exper imen ts can add informa t i on wh i ch greatly
res t ri c ts the pos s ib il i ties . We dis cuss this b y choo s in g an example , tha t
ch arged par tons carry t h e quan t um numbers o f quarks . ( In s uch a mode l Ge l l -Mann ' s
equal time curren t commutat ion relat ions are automati cally s a t i s fied . ) We are
not making the three q uark mo de l , the number o f quarks may be , indeed mus t be ,
infini te .
One is immedia tely s t ruck with the que s t ion as to whe the r our explana tion
of s caling is pos s ib le at all i f the partons do not have in te ger char ges . The
idea was that the ou tgoing par ton (now quark) could es cape wi thout further la rge
( as P ) intera c t ion to ul t ima te ly reso lve i ts e l f in to an outgoing b widle o f
hadrons .
-------.. a
p
But it is a q uark ( kn owing no low energy hadrons are of charge 2 / 3 say) it canno t
turn int o hadrons un les s it p i cks up the ext ra 1 / 3 charge by dragging along
ano the r parton quark . I f th is e x t ra par ton is from the ongoing ones o f momentum
o f order P the momen tum changes ( d i re c tion B to A) would b e o f o rder P and the
as sump t ion that b inding forces are s ma l l , apparen t ly required in our derivation ,
would f ai l . On the o ther h and , i t i s pos sible perhaps that the extra quarks
are found among the wee par tons , whi ch a re natural to any had ron wave fwic tion and
s ince hadrons going in direc tion A o r B share the s ame wee region they could
exch ange quarks there , making up the neces sary integers . Thi s is not clearly
satis factory but i t is an exciting adventure to t ry the idea that ch arged par tons
are s imply quarks , see consequences , devise experimen ts t o tes t them; and if they
are found to succeed , to re turn with more zest to the ques t ion of how Nature
mus t then b e reso lving the apparen t paradox among :
a. Partons carry quark quan tum numbers
b. Hadrons do no t
c. S caling works .
Parton Model 149
I n this spirit we look t o discovering b y what means we c an che ck (a) . At
the very leas t we are giving an example of how experiment could lead to fur the r
identifi cation of the character of partons .
To des cribe the parton dis tribution in a pro ton we would have s ix functions .
u (x) • No . of up quarks with momentum x to ir+-dx in the pro ton

II 11 " ti
d (x) down
II II II
s (x) s t range
Similarly u(x) , d(x) and s(x) are the numbers o f anti-up , anti-down and
an ti-s trange quarks with momen tum between x and ir+-dx in the proton .
The total charge on a proton is +l so
1 • 3
2
! - 1
[ u (x) - u ( x) ] dx - 3
1
j Cd (x) - -
1
d (x) ] dx - 3 1 1
[ s (x) - s(x) ] dx .
0 0
The z isospin is +1/ 2 so
1
j
f • } [ u (x) - -u (x) ] dx - 21 [ d (x) - d{x) ] dx .J
0 0
The s t rangeness is zero so
[ 1
[ s (x) - s(x) ] dx - o .
These equations have the solution
]cu<x>
0
- u(x) ] dx . 2
t [ d (x)
0
- d(x) ] dx - 1
J [ s (x)
1
- s(x) ] dx • 0 ( 31 . 2 )
0
That is , the ne t number of each kind o f quark is j us t the number of that kind in
The observed function f p (x) from 2 MW1
the s imple non-re lativis tic 3-quark model .
(or vW /x) is the number of each kind weighed by the square of the charge :
2
hence
- 1 1
f p (x) = 9
4
(u (x) + u (x) ) + 9 (d (x)+d-(x) ) + 9 (s (x)+s-(x) ) ( 31 . 3)
Measurements on the ep s cat tering o f course give us only this sum and we cannot
ext ract the separate terms from this alone . On the o ther hand we have data on
neutron s cat tering too . The neutron is given by the same formula excep t u (x)
would now be the number of up-quarks in the �· However , by isospin
reflection the neutron funct ions are got from those of the proton by rep lacing
u by d, d by u. Thus , if u(x) continues to be as defined , the number of
"ups " in the proton is also the number of downs in the neutron , we have
• 9
1 (u (x) + -u(x) 4 (d (x)+d- -
fn ( x) ) +9 1
(x) ) + 9 (s (x)+s (x) ) (31 . 4)
P
We have data on both these functions . The ratio fn / f behaves like
1
fn
r
p
-�
It has been s ugges ted that possibly fn / fp 1-x falling to zero as x O . We +
see immediately that thi s is impossible if charged partons are quarks ; s ince
u {x ) , 'il{x) e tc . , are all positive fn / f cannot fall below . 25 . The original data
P
only went up to about x -. . 8 and had fallen to perhaps . 3 so . 25 -was a possible
limi t as x + 1 but we have s ince learned that there was a mistake in the computer
program and i t is closer to .4 for x near 1 .
However , j us t for an exercise , what could we deduce i f fn / f was shown to be
P
j us t 1/4 as x l? That means d (x) falls to zero fas ter than u (x) as x + 1 . Thi s
+
2
would result i f w e s upposed that the y (in (q ) -y or ( 1-x) y-l ) f o r wee partons o f
total isospin zero is smaller than the y f o r isospin one (only one non-wee
parton) . Then the leading term (lowest y) would require tha t the fast parton
have the same isospin as the pro ton , up-quark only (since the wees are isospin
zero) . Howeve r , in that case the probability of exciting the 6 resonance (of
isospin 3/2 ) would fall fas ter than the elas tic peak for one can only make a
6 from wees with isospin one (as the leading quark can only con tribute 1/2) .
Thus we would expec t the 6 resonance to fall off with q 2 fas ter than others
( data is insufficient to indicate whether this is true or false in fact ) .
We can argue (if partons are quarks ) that if fn / f p falls below 1 i t "almost
surely" falls to 1/ 4 as follows : We first assume that almos t sure ly ( ? ) the
values for y for different quantum numbers among the wee are not equal and that
one therefore dominates . First among the s impler ones o f zero strangeness quarks
we have y0 for I • 0 and y 1 for I • 1 . Hence depending on whi ch i s smaller we
have even tually (as x + 1) a pure I • 0 or pure I • 1 among the wees . For I • 0
the non-wee quark mus t be (1/2 , 1/2) or u, this gives immediately fn / fp •
Parton Model 151
2 2 •
( charge of d) / (charge of u) • 1/ 4 . If I 1 the non-wee quark c an b e u or d
with weight given by the appropriate Clebsch-Gordan coefficient p • lfT3 (+l) d -
lf13 (O ) u , n • lf13 (O) d - lfl3 (-l ) u there fore
fn ( 2 / 3} (4 / 9 } + ( 1 / 3} ( 1 / 9 ) 3 if I a 1.
fp - (2/ 3) ( 1/ 9) + ( 1/3) (4/9 ) 2
Hence on the assump tion o f unequal y ' s e i ther 1/4 or 3/ 2 is the limi t ( there fore
if below 1 , it is 1/4) . (There is the possibility that the non-wee quark is an

antiquark , in that case if I dominates fn / fp • 0 • 2/ 3 but this possibili ty is
considered very unlikely for p and n} .
Another s tate may be wees of s t rangeness +l and the non-wee pure s . It is
unlikely this is eas ier than the non-s trange case , but even so the ratio would be
one , hence excluded .

Several models have been made which predict ratios below 1 b ut above 1/4 ,
All of these assume the lack of corre lation of fas t quark quantum numbers and wee
state character which we explicitly argued for , that is , they imply a degeneracy
r1 • y0 in which case , of course , other values are possible .
The reader can show that the ratio o f G for proton and neutron comes out
M
2 2
-2 for isospin z ero wees , for large -q ( i t s tarts at µ p /µ n -1 . 4 at q O} . = •
The case that the total angular momentum among the wees is ze ro should be tried
2
firs t . Show also that GM for the proton is posi tive for large -q , so no sign
2
change is necessary as q rises , which is sa tis factory because experimentally
GM for the pro ton never seems to fall to zero . Isospin one for the wees gives
GMp / GM + 0 and GM is positive at large -q
n n 2
•
Lecture 32
Momentum Carried by the Quarks
Another interes ting s tudy is the total momentum carried by each kind of
J [u (x) + u(x) ] x dx: J [ d (x) + d(x} ] x dx; J [ s (x) + s(x) ] x dx

quar� .
Let U • D • S •
J x fp (x) dx J x fn ( x) dx
be the total momentum carried by up (plus anti-up} quarks , etc . Then we know
• . 18 , • . 12 so
. 18
. 12 (32 . 1)
Now if we asswne that there are no neutral partons beside quarks (e . g . no

"gluons ") all the momentum of the proton is quarks , we would have
U + D + S 1
Solving these equations we have U • . 2 1 , D • . OJ , S • . 76 . This result is clearly
unreasonable - that 3/4 of the momentum of the non-s trange obj e c t proton is to
be found in the s trange quarks . This indicates most s t rongly that all partons
cannot be quarks - there mus t be some neutral ones carrying momentum , say N . This
momentum can be thought of as in the field , i f any , by which the quarks in teract ;
(possib ly for example a neutral pseudovector) which if represented in fie ld theory
by inte rmediate field partons are usually called gluons .
In the region near x �
0 the functions fp and fn diverge as . 32/ x , corres
ponding to infinite numbers of quarks and antiquarks . (The net number of quarks
is finite , see eq . 31 . 2 and discussion . ) Further fp an d fn must have the same
behavior so if u(x) • a/x for small x then -;;(x) • u(x) • d (x) d{x) . We do
•
not think we could p rove s (x) and its equal s(x) mus t equal that value also
because S U ( J) is not perfect , but it may not be far away ( if indeed it is not equal) .
An interes ting quantity is the measure of how u and d approach each othe r
�
J
as x O ; although each is inf ini te , the difference is finite and integrab le .
Experiment gives about . 17 for the integral
1
[ f (x) - fn (x) ) dx but the erro r
0 P
•
J1cvwP2 - v�2 ) dx/x . )

is large for it depends precisely on the small di fference near x 0 (as it is
Some people (ss for exmaple Pas chos , Weisskopf and Kuti) have tried to guess
the functions u, d, e tc . , by making s impli fying assump t ions whi ch are chosen for
their s implicity rather than their phys ical ne cessity . Such predictions may or may
not s ucceed , i f they fai l all we say is nature is not so s imple . Unlike the ideas
of the parton model we have used so far , the additional assump tions made in these
modelP have in general no physical backing . In case any o f you are interes ted
we give an exmaple o f s uch a theory here : I t is that the wees or dx/x region
Parton Model 153
is made en tirely of uncorrelated pairs , and tha t on top of thes e there are three
"valence quarks " of the kind of the three quark model and the wave function o f
these is a simple product of independen t pieces . (We do not expect such
simplicity . ) That is , i f a (x) is the typical sea dis t ribution and v(x ) that of
a valence quark we would write
u • a + 2v u • a
d • a + v d • a
s a s • a
fo r the p ro ton . j 1
v (x) dx • l ( to get 31 , 2 ) . Near x + O i t is a (x) that goes
to in fini ty (as . 9 4/x) but x v ( x) + 0 ss x + O. This predicts for the total
momen ta carried by quarks S • 2D -u so the data ( 32 , l) gives U • . 36 , D • .18
•
J x fn dx to exceed t J x fp dx .
and S 0 (surprise) hence we are i n s ome kind of trouble (a cannot b e zero)
unless the experimental erro rs allow
l
At any ra te some momentum (46%) must be in gluons . ( fp (x) -f (x) ) dx
nIt predicts
which is hard but not imposs ib le to reconcile with the experimental . 1 7 . I t also
f • 3l
predi cts vw2n /vw2p + 2 / 3 (hence probab ly wrong) by suppos ing that as x + l , a falls
fas ter than v .
Lec ture 33
Future tests of charged partons • quarks
So far no direct data is accurate enough to test or eliminate the idea that
charged partons are quarks . However further experiments can . We give two examples ,
neutrino s cat tering and polarized electron and pro ton s cattering . •
We shall discuss the inelas tic s cat te ring o f neutrinos from pro ton \-P + µ
plus anything in detail in the las t part of this course . We do not know weak
interaction theory as well as quantum electrodynamics but can use these experi -
ments themse lves to test i t (e , g , if interaction is not a point-like four fermion
in teraction s caling will fail in the form derived below) - but s upposing it is
all righ t for now, we suppose the coupling wi th quarks is (ii y µ (l+iy 5 ) v ) (Q'y ( l+iy 5 ) Q)
u
wi th the quark quan tum numbers from u quark to cos e c d + sin e c 8 where e c
2
is Cabibbo ' s angle , s in e c ""' . 06 , For simp licity o f our rough discuss ion he re
we take e c - o .
Then if we s catter � to II + it can affect only up quarks (sending u to d)
or d quarks (d to ii) . Thus we would expect to measure j us t u + d if the curren t
were pure vecto r , but it is V-A . Th e A i s j us t like V fo r relativistic
parti cles (excep t the sign is not reversed for an tiparticles as i t is fo r V . )
l f for the V -A current matrix e lement we wri te
in analogy to electrodynamics we find that the general form by convention is
•
2
M11 v i [ 4 P 11P vw2 - 4it o 11 vw 1-2i & µ va A P 0 q Aw3 J ( 33 . 2 )
( leaving out all terms proportional to q because they give small effects at
11
2
the µ , v end propo rtional to (mass µ ) ) . w1 , w2 are as in ele ctricity except
they con tain VV and AA (nearly equal) ; the new term w3 comes from interference VA .
ln the scaling region vW , 2MW 1 , vw3 scale to three functions f (x ) , £ 1 (x) , t 3 (x)
2 2
(x • -q /2 Mv ) . lf partons have spin 1/2 , t • 2 x t • For the quark model we ge t
2
2 1
f 1 v p - 2 (u + d cos e + s s in e )
2 2
f 1 vn
2 2
• 2 (d + u cos e + s sin B )
f 1vp • 2 (u + d cos e + ii s in e )
2 2
f l vn
-
• 2 ( d + u cos e + s s in B)
2 - 2
2 2
f 3v p � 2 (u - d cos e - s s in e)
f 3 vn
2 2
= 2 (d - u cos e - s sin e)
f 3 vp -2 (u - d cos e - 8 s in e )
2 2
•
f
vn •
2 2
-2 (d - u cos e - ii s in e) ( 33 . 3 )
3
Thus we have many re lat ions , for example
f vn + s sin e)
2 2
_
f vn 4 (u cos e
•
l 3
- 2
f 1vp + +
vp - 2
£3 •4 (d cos e s sin e)
+ s sin B) ( 33 . 4)
- 2 - 2
f 3vn • 4 (u cos e
* See also Appendix B

Parton Model 1 55
2
Negle cting sin 8
p p p p
4 u · f 1 v + f 3v 4 u - f 1v f v
3
(33 .5)
n n n n
4 d • f 1 v + f 3v 4 d - f 1v f v
3
Sum rules
(33 . 6)
the firs t is Adler ' s sum rule (not quark model dependent) , the thi rd was first
derived by Llewellyn Smith and Gross .
Dis tribution predictions
+ cos e (d-d) + s in e (a-ii) )

vp p 2 2
f
3 f v - 2 (- (u-u)
3
( 3 3 . 7)
the las t term can probably be neglected , it gives a very close check on the model .
Also
fv
n
3 f 3v
n •
2
2 ( - (d-d) + cos a (u-u) + 2
rtn e C s-8> >
- 6 COB
2 ep e 2
ec (f - f n ) + sin e ( (s-8> - (d-d) ) ( 33 . 8)
by sub traction with the previous relation can be used to eliminate the las t
term if there was any doub t .
allows us to ob tain u + u, d + d , ii + a
p p ep en
Knowledge o f f 1v , f 1v , f , f
permits further separation into u, u, d , d, a + ii negle cting

v p
separately and f 3
2
sin e
Deep inelastic scattering wi th apin
If the electron beam and the pro ton is polarized in the deep inelas ti c
s cat tering we can obtain s till more information , this time about how the
partons are polarized in a polarized pro ton . For defini teness we shall assume
all the charged par tons have spin 1/2 .

Fo r a particle at res t the s tate may be des cribed by saying the spin is in some
particular s pace direction 8. Relativis tically trans formed spin s tates can be
described by a uni t four-vector s µ ( the t ran s formed s is as though i t were a
polar vector) satis fying s µs µ • l and s p 0 (p momen t um of s tate) . When
µ µ µ
• •
taking matrix elements we can use s purs and sum over s tates if we replace the
usual p roj ection operator ' + m by the mo re comple te {l+m) (i-iy s � ) /2 to proj ect
into the spin s tate s . Thi s s is a pseudovector . The s pin s tate o f p in the
µ µ
J factor o f
v
2 2
wµ v - '\� < X I J v l p ' >*<X I J I P> 2 110 (Mx - (P + q) )
L µ
need not be the san� as the spin of p in the J factor ; but the quantity we want
µ
depends on s pin as a 2 x 2 matrix (dens ity matrix) one of two states in , one o f
two out . However a l l the information is con tained i f the diagonal is known fo r
all s tates , not only for p • p ' • +z or -z but also for p • p' • x • ( (+ z) + (-z) '/12
for example ( in fact , these three and p • p' • -x are enough in our case) . Thus
we consider only diagonal cases p • spin p' • s but for various s .

µ µ
W v can then
µ
be shown to be o f the form
(33.9)
( disregarding te rms proportional t o q or q v ) where w1 , w are a s before and

µ 2
2
G 1 , G 2 are two new f unc t i one O f q . v .
What do we expect for these new functions G 1 , G 2 in the deep inelas tic
region acco rding to the parton model? If a parton has momen tum p 1 • xP and
spin described �y w µ the s cat tering to p 2 • p 1 +q is described by
( 33 . 10 )
Let u s use f o r example the coordinate system where the pho ton has· pure z
component
q • (0 , -2 P x, 0 , 0) 2Mv • 4p 2 x
2
P • (£ , P, O , 0 ) -q
µ
Now f i rs t s uppose the helici ty of the pro ton is +1 thus s • (P , £, O , O ) /M .
µ
Parton Model 157
Fur ther le t
2
h (x) • No . of partons ( charge ) wi th helici ty + in a pro ton of hel . +
+
2
h (x) • No . of par tons
_
( charge) with he licity - in a p ro ton of he l . -
µ µ µ µ
we have w • (P , e , O , O) /m • xMS /m for hel . + and w • xMS /m for hel . -
1 1
We there fo re h ave
( 3 3 . 11)
The function gives 6 ( �-x) / 2Mv in the s caling limit . The firs t integral gives
the res ults we already know 2 MW • f (x) • h -+h and v W • xf (x) . The se cond
1 + - 2
in tegral gives
( 3 3 . 12 )
the re fore the funct ion G - 2 x G s cales .

1 2
Suppose now that the p ro ton is polarized in + x direc tion , we have
s
0 0
• w • (O , O , 1 1 0) for par ton with pol . + x and w
0
• -s
0
for pol . -x .
2
k (x) • No . of par tons ( charge) with pol . + x in a pro ton of po l . + x
+
with po l . - x in
2
k (x) • No , of partons ( charge ) a pro ton of po l . - x .
_
We have an expres sion o f the type ( 3 3 . 1 1 ) with h+ replaced b y k and h rep laced
+ _
by k from whi ch we derive
_
3 2 2
2M vG + 2!r v G • m(k (x) - k (x)H' x ( 3 3 . 13)
1 2 + _
3 2 2
therefore the function 2M vG + 2M�v G s cales .
1 2
2 2 3
Hence from ( 33 . 12 ) and ( 3 3 . 13) !r v G s cales and Mv G s cales and we wr i te
2 1
2 2
M vG 1 • g Cx) , Mv G • g (x) . These are the s caling p redic t ions for spin cases .
1 2 2
The term in G in ( 33 . 12 ) is relatively negligib le (order l / v times term in G ) .

2 1
We have
( 33 . 14)
( 33 . 15 )
h
+
+ h
_
• k +
+
k_ • f (x)
158 Photon-Hadron lnteractions
Of course , these new functions g 1 , g 2 cannot be obtained from f (x) because new
dis tributions are involved , the number of partons with various spins is s tates o f
various helicity , so that w e derive nothing b u t the s caling rules even if w e use
a special model like quarks , etc. However , in that model we can ob tain a
remarkable sum rule of Bjorken .
We can use our proton wave function (described by parton) not only to ob tain
predi ctions for deep inelastic scaling but like any o ther wave function we can
write the expectation value of various operators for i t . We have already done
that trivially when we wrote j ll as the operator L ks e iy and found the
to tal charge on the p ro ton to be expressed as /l� �::; �
- t d-d) - t (s -s dx .3
Now we will discuss the expectation of a less trivial operator , the axial S
decay operato r for p + n transitions . It is wri t ten empirically as

 i (u y y 5 u ) (GA /GV ) and experimentally GA /GV • 1 . 2 3 t .02 . Theoretically
•
ll P ll n
+
in the quark model the operator J+All is the sum over quarks of T +y lly 5 (T I spin
raising op,) . If we change the I spin to third component we can use the relation
We calculate both sides for a proton moving fast to the right wi th momentum P
and spin s 0 and sum on proton s tates via a spur operator correctly projecting. On
the left s ide we have the corresponding sum on quarks , thus
Total of sp l <j+m) (l-i y wf) y 5y µ ] ! (GA /GV) sp 1 ¢+M) (l -ir 5f> r 5r µ l L mwµ
-
•
•
summe d with weight + 1/2 for u, u; 1/2 for d, d • (GA /GV ) s0M/ 2 . Now the sum
on the left is exactly the integral over x of 2 g1 , (or 2 (g 1+g2 ) depending on
the s pin case) that we needed for deep inelastic ep s cat tering excep t the
weights in the latter case were 4/9 for u, u, 1/9 for d, d and 1/9 for s , s for
-
the proton (or 1/9 for u, u, 4/9 for d, d, 1/9 for s , 8 for the neutron) while
-
here we want 1/2 for u, u, 1/2 for d , d. But the di f fe rence of s1 for p roton
g
lp and neutron gln has weights 1 / 3 for u , u, 1 / 3 for d , d or 2 / 3 of what
we want . Hence we have for the two polarization cases
2 J
1
(glp - g ln ) dx •
1
-
3
() GA
-
Gv
( 33 .1 6)
2 J 1(glp - gln + g2p -

(:�)
( 33 . 17)
0
Parton Model 159
( ( 3 3 . 16 ) is Bj o rken ' s relation ) . From these we conc lude J� g

2p
- g
2n
) dx • O.
0
This comes me rely from the equal i ty of the coe f f i cien t for the polarization z
[i t
and x cases , an equali ty whi ch would have to hold no matte r wh a t the wei gh ts
we re ; so we can conclude individually dx • 0 and g dx • O, i .e.

2p 2n
Ji 2
dx • 0 ( 3 3 . 18)
0
a res ult of ro tat ional s ymme t ry (angular momen t um conservation) .
The fi rs t relat ion is more remarkable and would tes t s i multaneously the
quark view , and the in terp re t ation of the weak current as y ( l+iv ) no t only for
µ 5
the elemen tary parti cles e , v , µ, v but also for the elemen tary cons ti tuent
e µ
quarks (or quark quantum numbers on spin 1/ 2 partons ) . Its ve rification , or
failure , would have a mo s t decis ive e f fect on the dire c tion o f future high
energy theore tical phys ics .

Tests of the Parton Model
Lecture 34
Angular momentlllD in par ton wave functions
The e11ua tion

t
0
g dx
2
• 0 comes from angular momentlllD considerations ( that
the pro ton is spin 1/2 and the parton is spin 1/2 ) . This b rings up a general
research prob lem about angular momentum and part on wave funct ions ( func tions
of x and transverse momentum k.J. ) . The dependence o f these functions on the
transverse momenta kl. is no t known , and are very in teres ting theoretically . Can
any information on this , or other 11ues tions , be go t by s tudying the angular
momentum proper ties of parton wave functions ? For examp le , what res t rictions
if any on the wave function come from the fac t that the to tal angular momentum
of the proton is j us t 1/2? Or , s uppose the parton probab ility dis tribution f (x)
were known for a 6 o f helicity + 3/2 . What could you say about the dis trib ution
for a 6 of helicity + 1/2 ?
Othe r expe riments tes ting parton idea (Drell)
Dre ll has suggested another expe riment which can be predi cted by knowledge
of the parton dis tribution . Again we use the parton as quark model to des cribe
the idea, but it can of course be analyzed the same way in o ther representa tions
+ µ µ-
for par tons . The reaction is p + p + + any thin g . P ro ton p lus p roton (or
+ -
p + p) at very high s (e . g . P in CM is large ) makes a µ µ pair whose energy is
160
Tests of the Parton Model 161
propo r tional to P , and whose z componen t of momen tum is of order P , as P + m .
The picture is
the µ pair is made from the annihi lation of a p arton in pro ton A and an ti -parton
in p ro ton B and the o ther partons j us t keep going on to make final hadrons , the
"any th ing" . By sui table cho ice o f the energy and momentum ( in z dire c t ion) o f
the µ ' s w e can de termine x and x o f t h e annihilating partons . (Thus i f ,

1 2
i n the CM, the momen t um o f the colli ding pro tons is P , an d the ene rgy an d to tal
z component momen t um of the µ ' s is £, P ' we have by conservation of momen tum

z
and energy
� " ( c -p ) / 2P
z
It will be eas ies t to meas ure the pair at large angles ( the angular dis tri
b ution is 1 + cos e ) so no great loss o f intens ity o ccurs here , and b ackgro und
2
hadrons which should move p rimarily in z di re ction are avo i ded . If 6 " 0 is
chosen howeve r , the momenta of the two µ 1 s give direct ly P x , P � be cause in s uch
1
a one-dimentional collis ion of re lativi s t i c particles the momen tum of the final
parti c les is the s ame as tha t of the inciden t pa rticles .
The cross section for th is is the c ross sect ion fo r relativis tic p air
4 2 2
annihilation ( e ( 1 + coa e ) / 16 c ) t imes the chan ce C o f finding the app rop riate
partons to annih i late ; e . g . for p + p, if partons are quarks :
4 - 1 -
C • 9 1 2
-
2 1
-
(u (x ) u (x ) +u (x ) u (x ) ) + 9 ( d (x ) d (x )+d (x ) d (x ) ) +
2 2
1 1
1
+ 9 1
-
2
-
( s (x ) s (x )+s (x ) s (x2 ) )
1 ( 34 . 1)
To be valid , o f course , the wee region mus t be avoided ( that is , P mus t be
high enough that Px 1 , Px2 are large compared to 1 GeV) . One might at f i rs t be
concerned tha t the j P> j P> s t a te we s tart wi th is no t a s t a tionary s ta te , fo r

pro tons scatter via s trong interact ions so the pos s ib i li ty exi s t s tha t the parton
distribut ions are firs t dis turbed by " s t rong" interact ions be fore they annihilate
to make 11 ' s . But we a re coming to unde rs tand tha t the "s t rong in teractions "
are not so s t rong in th is sense - that in hadron-hadron col lis ions it is not
the f as t par tons that interact , b ut the wees . This accounts for low momen tum
trans fe r and s caling in ve ry high energy hadron collis ions - s o as long as we
avoid this wee interacting reg ion (interact ions via fini te energies , not o f
o rder P) the partons act a s free particles nearly ( as P + w ) .
Obviously wi th enough data we could s ee i f a form like ( 34 . 1 ) could work -
or if we know u , u e t c . from neut rino s ca tte ring we could tes t i t dire c t ly .
Lec ture 35
p p + u u- +
+ +
any thing ( con tinued)
Gronau has pointed out that with no further data than we already have we
could calculate this factor C in at leas t one region . If x

1 is small , we
know all the f un c tions u , u , d , d go as a / x wi th the same a , and s , s go as
1
a /x , where a may be close to a (but not neces sari ly exa c t ly equal , say
s 1 s
a • a -8 with 8 sma ll ; su says a • a but it is incorre c t ) . We shall firs t
s 3 s
cons ide r 8 • O , and then say what the e f fect o f i t is . Hence in this region
( a • . 2 4) we have
ep
But the las t fac tor is j u s t f (x) , the 2MW1 for the pro ton and is known ,
so C is predictable in thi s region . I t shows that Dre ll ' s experiment could
easily se rve as a test of any parton mo del - and will be an interes ting experiment
to watch .
If B is not zero we ge t
so it seems that there is l i t t le e ffect f rom any reasonab le B . For example ,

Tests of the Parton Model 1 63
a very weak dependence on 8.
In a s imilar way we could describe the pro ton antipro ton case Jl+P+µ+µ - +
anything . I t would be determined by a fac tor
( 35 . 2)
If finally , the functions u , ii etc . are determined for the pro ton , the
corresponding functions for the pion u ( x) , ii ( x) etc . could be ob tained through
x x
+
( difficult to get enough energy) w + p + µ µ - + x for now
Cw • . 94 ( up ( xl ) -u w ( x 2 > +-uJ x 1 ) u w ( x2 ) ) + 91 ( dp ( x l) -d w ( x 2 ) +dp ( xl ) d w ( x2 ) ) +

-
1 s x )-s x ) - x ) s x ) )
+ 9 ( p ( l w ( 2 +s p ( l w ( 2 ( 35 . 3 )
Electron pair production of hadrons

Dre l l has sugges ted that another fundamental experiment may yield information
about the character of partons . I t is e++e- + any hadrons at high energy . We
would expect to produce some spin 1/2 parton pair of charge e i ,
e4
do
cID • e·i 2 2 2
( l+cos e ) ( 35 . 4)
4q
• ( e i /e) 2 o 0 where o 0 is the cross section fo r epin 1/ 2 pair production ( e . g .

e+e - +µ+µ-) . Now this pair would turn into hadrons - and if the energy is high
enough, of nearly the same energy as the virtual s tate of a parton pair . Thus
if there were a number of parton pair types of charges e 1 , e 2 etc . , the total
cross section oe++e-+ hadrons is just the total to make each pair :
o ( e+e - + hadrons)
( 35 . 5 )
o ( e+e - + µ+µ - )
That is , the ratio approaches a constant , and the constant is the sum of
squares of parton charge summe d on each type that exists . If there are any
1 2
1 as strongly e .g . 4
spin 0 partons they would contribute 4 ( e for each spin zero
par ton) • We call D • 4 e / "Drell ' s constan t �·
"' 2 4 1 1 • 2/ 3
For example if partons are quarks we expect L... e i • 9 + 9+ 9 for D .
I f partons must carry integral charges , and be spin 1/2 , the sum mus t be a t least
l. In the case o f quarks we would have the dif ficulties we have dis regarded
as to how a pair of opposi tely 1110 ving charges of ± 2 / 3 say can turn into only
hadrons of integral charge . There may be s ome way to get the required extra
quarks from a soft sea o f pairs , but one can perhaps legi timately question
whether the above arguments are entirely clear , involving as they do an assumption
that the parton s tates go entirely into hadron s tates . I s i t really consistent
to do this and omit s tates of quark quantum numbers ? I t would be mos t exc iting
if the ratio does come out 2 / 3 as well it migh t , for the theore tical ques tions
produced by such a simple answer would be very interes ting . I think it would
be worthwhile as an exercise beforehand to assume the result was really 2 / 3 and
see what paradoxes , if any , would then have to be resolved .
Aswe discussed p reviously theoretically ( lecture 5 ) , the total cross section
+ - 2 2
for o (e e � any hadrons) as a func tion of q ( q • 4El, E • CM energy) de te rmines
the vacuum expectation value of the product of currents
(q µ q v - o µv q ) a
2 2
<O / J µ (-q ) Jv (q) / O> • ( q0 ) p (q ) ( 35 .6)
with
2 2
CJ -
(4 TTe ) 2
--- p (q )
2q 2
The vacuum sca t tering o f virtual photons by virtual hadrons is determined by
2
( q µ q - q o \lV ) v (q )
2
F . T . <O j {J (l) J (2) } T j O>
v v
= ( 35 . 7)
µ
l 2
where dispersion theory told us Im( iv) - 2 p (q ) '
00
J
_.i.._
2
4TTe i (v(q ) -v(O) J •
2 2 cr ( s ) ds
-- - ( 35 . 8)
2 2
rr (4rre ) 4m 2 s -q
"
(l + above quantity) is the factor by which a photon propagator l/q 2 mus t be
multiplied for first order effect of virtual hadrons i .e .
2
CA _
_ _
,.� _
•
1
2 4TTe 2 i [ v ( q ) -v(O) ]
q
The low energy tes ts , like e f fect on Lamb shift depend j us t on the lowest
2
q or only on iv ' (O) that is os -1 ds . J
Today we can
begin to say something quite detailed about this , for in the
2 2
region ( 2m " ) to jus t above 1 GeV the cross section is dominated by the P , w , �
2 2
production . Perhaps p (q ) set tles down to D/6 TT for large q fairly soon .
Al though one migh t s uppose we migh t have to go decidedly above the nucleon
Tests of the Parton Model 1 65
pair creation , this may not be t rue ; soft parti cle produc tion may domina te at
all ene rgies and reach i ts asymptote soone r .
A . Cisneros has calculated the various hadronic con tributions t o iv ' (O) .
The con tributions from the p , w and + turn out to be ten times larger than
2
the large q contribution wider reasonable assump tions about the likely value of
2 2
Drell ' s cons tant and the value o f q at which p (q ) attains i ts asymptote .
+
Data from e e - intersecting rings can be used directly to ob tain the
con tribution from the p. The data is simply inserted in the formula
..
f
2
2
iv ' (O) .tl!L.l
1
• 2 2 2 dq (35 . 9 )
2
11 ( 411e ) 4m11 q
to ob tain
( 35 . 10 )
(What was used here was the data on the pion form factor which is almos t
everything there is in y (virtual ) + p + hadrons . The 411 inelas ticity is very
2 2
smal l even at q ( . 8 GeV) ) .
•
In the cas e of the w and + we assume VDM, which works wel l in this type
of process . In this case the contribution from the vector meson v • w o r + is
2
p (q ) ( 35 . 1 1 )
2
From the relation Im ( iv(q ) ) 2 1 p (q 2 ) we deduce
•
2
2 1 mv
i vv ( q ) - 2 --..,2.---,2.----
•
( 35 . 1 2 )
gv (q -mv ) -ir m
v v
For a very narrow resonance , which is a good approximation for w and + ,
2 2 2 2 2
we have p (q ) • mv g,,- 2 116 (q -mv ) the value o f iv ' (O) is simply
( 35 . 13)
-2 2
(This formula gives i v ' (0) 5 . 3 x 10 GeV- for the p, in good agreement with
p
•
what was ob tained above using the data directly . ) The contributions o f the w
and 4> are the refore
( 35 . 14)
( 35 . 15 )
2
We now evaluate the con tribution from the " tail" of p (q ) assuming it has
2 2
at tained i ts asymptotic value D/6n at q • q0
2
S.... . _D _ _L (35 . 16)
l2 i q 0
4 2
q
•
-� 2 2 2
This gives iv ' (O) . 84 x 10 Dq0 - If q 0 is as low as l GeV and D
t
-2 -2
is the quark value 2 / 3 we have . 56 x 10 GeV for iv' (O) , We believe the
t
actual value of the non -vector meson contribution not to be much larger than this ,
we ass ign to it 100% uncert ainty . Adding the various con tributions we get
finally
( 7 . 3 t 1 . 1) x 10
-2 2
iv' (0) • Gev -
hadrons
This corresponds to a correction to the magne tic moment of the muons o f
8
(5 , 5 ;t . 7 ) x 10 - in (g -2 ) /2 .
2 2 l •
2 2
For comparison , iv' (O) for muons is [ 4 n 15 M 1 - 15 , 3 x 10 - Ge v - ,
µ
so the hadronic contribution is about one half as large .
2 2 2
Away from q ""' O , the contributions to i (v( q ) - (0) ) from the tail of p ( q )
grow as the vector meson con tributions fal l , so the uncertainties are greater .
Ano ther expec t ation of intere s t from the parton model is the angular dis tri
+ -
bution o f hadronic products in high energy e e collisions .
,f;' hadrons
hadrons �
We expect that partons of spin 1/2 are produced with an ( l+cos 0) angular
2
dis tribution and that the final observed hadrons will have small transverse momenta
relative to the direction 6 in which the partons were pro duced at high enough
energy . Two bursts of oppositely moving hadrons are expected which will determine
the angle e . If there are also charged partons of spin 0 the angular dis tribution
will be
D 1 (l + cos e) + 2 D0 (1 - cos 0 )
2 l 2
1 2
where D 1 and D0 are Dre ll ' s cons tants for partons of spin 1 / 2 and spin 0
1 2
respectively . For quarks we expect of course only a (1 + cos 0 ) dis tribution .
2
Inelastic Scattering as Properties of Operators
Lecture 36
Inelastic e p scat tering as properties of operators

We should now like to turn back to a point of view , described more par-
ticularly in the firs t lectures of this course , of the electro dynamic properties
of hadrons as being measurements of properties of the curren t operato rs J ( l) .
µ
[J
For example there we said we expected that the co11111u tator v (2) , J ( l) ) would
µ
vanish if 1, 2 were outs ide each o thers light cone . Second order interactions
were described by matrix elements of a second operator V v ( 2 , 1) which we showed
µ
would be expressible as a time ordered p roduct of the firs t ones J (written
µ
{ J v (2 ) J ( l) } ) .
µ T
As we have seen the e p s catterin� measures a function which we have written as
We now cons ider it in a more abstract way . Firs t we note that if the s tates
p and x are imagined to con tain their center of mass momentum factors , the
2 2
2 w6 (M - (p+q) ) would be unnecessary , for the class of s tates x ' now mean one
x
of mass Mx moving with momentum P . Thus
. allI x ' <x ' I J (q) j p>

µ
167
or by completeness ,
Thus K µ is the expectation on the proton of the product o f two operators .

v
2 2
We have measured it only for q < 0 but it exis ts also for q > O and of
2
course the entire func tion for all q both posi tive and negative belongs toge ther
theoretically . For pos i tive q 2 , however , there is a small technical poin t . We
wish to alter the definition because we do not wish diagrams which do no t affect
the proton at al l , thus proton
O
J
hadrons
J
Such terms do no t ex:lst for the non-di.igonal matrix elemen ts of the product of
operators term and we should like to define K as the limit
v µ
of this as p + p to ob tain a sui tab ly use ful defini tion o f matrix elements
of the pro duct operator. We can get the same res ult if we exp lici t ly s ub t ract
the contributions of the disconnected diagrams o f the type above . Hence we
write
K
µv
(q ) • < pj J v ( -q ) J µ (q ) ip > - <O / Jv (-q) J µ (q) j O> ( 36 . 1)
Then if the final p is changed to p ' slightly di fferent , the express ion
de f ined by ( 36 . 1) is cont inuous .
We would like to consider , for theoretical purposes , various o ther currents
beside electromagnetic - we can have vario...s su3 generalizat ions in an octet (or
nonet) (i . e . currents wi th different charge numbers ) and axial curren ts as well .
Thus we le t these indices (su 3 and axial) as well as the µ , v indices o f space
be contained in a s ingle letter A ( calling J µ (l) A ( l) e t c . ) to save wri ting•
in general arguments , and can always go back and insert indi ces at the end .
Thus i f A is any operator o f our allowed set ( for our examples , a vecto r or
axial vector local current in su 3 octet or poss ib ly s inglet) we wri te (obvious ly
the diagonal pro ton s tate can be generalized immediately too , but we leave chat
for the s tudent) .
� (q ) • - <O j B ( -q ) A (q) j O > ( 36 . 2)

Inelastic Scattering as Properties of Operators 169
And las t ly we omi t writing the l as t term for s impli city , b u t you mus t
rememb e r that , fo r d iagon al ma t r i x elements it is the re .
�A is the diagonal ma t r ix e lemen t o f the p ro d u c t of two ope rators .
E idv ent ly we can de fine a cor re s p ondin g thing in s p ace :
Since the diagonal eleDslt K depends on ly on the d i f f e rence x2 \J - x1\J or
2-1 , t
i s Fourier trans form s KBA (q) . We exp li ci t ly
i call the compon e n ts of
q • (v , Q) in the sys tem where the p roton i s a t res t (o the rs via re la tivis t i c
trans formation) . Thus I P > is at res t
•
�l x
2 z!
x
<x ! A (Q) j p >5 ( Ex -E -v)
P
( 36 . 3)
Since K gives the product ope rator we can find from i t the commuta to r as
we ll as th e time o rdered ope rator so us e ful in s ca t tering .
We note , in our case (in the res t sys tem of the proton) , since the p r ot on
is the lowe s t mass of all s tates o f baryon number one (and A does not change
baryon number) that Ex > E fo r all x. Hence if v < 0 th e 6 fwiction is zero -
p
no s t a te can be reached by A (v , Q) l p> if v < 0 - and C l p> at re st)
�A (v , Q) • 0 if v < 0 (36 . 4)
Now the commut a t o r mat rix e lement is defined by
so its Fourier trans form satisfies
We no te that K can b e ob t ained from C and vic e versa b e cause o f ( 36 . 4) . Thus

�A (v , Q) • C BA (v , Q) for v > 0 (36 . 6 )
and C BA ( -v , ...Q ) • -CAB ( v , Q) .

Thus in measuring �A ( v , Q) we are measuring the Fourier trans form of the
co111111U t ator of two currents .
We shall discuss the consequences of this interes ting result in a moment ,
b u t wh i le we have these equations before us we wi sh to de rive a few fo rmulas for
the s cattering amplitude whi ch we will need later in the course . As we have
discus sed the s cat tering ampli tude for an incoming photon (virtual or real)
coupled to J (l) say A (l) to an outgoing one coupled to B is determined by the
operator
VBA (2 , 1) • { B (2) A (l) } T + seagulls ( 36 . 7)
we discuss the effect of seagulls (if there are any , 6 (2 -l) type terms ) later
and omit them for a while .
I f in particular we are interes ted in the forward s cattering ampli tude

T BA (q) on a pro ton with a photon of momentum q we need the Fourier trans form
( 36 8 )
.
(The superscript F indicates that the choice o f s ign of the imaginary part
for negative frequencies is taken according to the convention of Feynman in his
c
QED papers ; there is a different choice called causal amplitude , T BA ' which is
of ten very useful . )
To take the Fourier transform o f the firs t term in ( 36 . 8) we have a product

of e ( t 2 -t 1 ) whose F . T . is i/ ( v+i£ ) and B ( 2 ) A ( l) whose F . T . is 'SiA (v , Q) and hence
-+
we have the convolution of these . In the same way the second term is the
convolution of -i/ (v-i£) ( the F . T . of 9 ( t -t ) and K (-v ,
1 2 AB ..Q) .) Hence
( 36 . 9 )
t
v ' -v
l
Now use
+ i£
. p ,.!-
v -v
ho (v ' -v)
( 36 . lO)
sgn (v) • +l v > 0

-1 \) < 0
where we have explicitly added possib le seagull terms oBA which is simply an
unknown ( finite order) polynomial in v , Q . Thus the scattering ampli tude is ,
except for a polynomial , given in terms of the commutator .
Negative �. is of course , not defined by experiment . However it can be
ob tained from measurements of the reaction with antiparticles (using A for A)
via the connection implied by ( 36 . 10) and the relation for commutators ( resulting
Inelastic Scattering as Properties of Operators 1 71
from the fact that A is the adj oint operator to A)
so that ( 36 . 10) implies
Lecture 37
Properties of operators , continued

Another more usual convention (causal amplitude) to define the negative frequency
extension changes the s ign of the imaginary part for negative frequencies
�
T A( v , Q) • :
T A (v , Q) + i8 ( -v) CBA ( v , Q} ( 3 7 . 1)
c c * ;t
Thus (TBA ( v , Q} } • TBA (-v , -1.1)
and also ( 3 7 . 1) becomes now
�
T A (v , Q) • J CBA (v 1 ' Q) v ��: + 0BA ( 3 7 2)
.
c
The significance of T BA (which is j ust as good a way of des cribing scattering
as T ) in coordinate space can now be seen . (TC is given by an expression like
F
( 36 . 9) except that the sign of the +iE in the las t term is reversed , } In ( 3 7 . 1)
we have expressed things in terms of the commutator but the las t term can be
more s imply expressed as the product operator , from ( 36 . 5) and ( 36 . 6) we have
8 ( -v ) CBA( v , Q) • -KAB (-v , -Q) , the •
F . T . of A( l ) B ( 2 ) . Hence ( 37 . 1) says
- iT C • F . T . { B (2)A (l) } T - F .T . A ( l) B (2 )
for t 2 > t 1 w e have B (2)A(l) - A(l) B (2) • [B (2) , A(l) ) and for t 2 < t 1 , w e have
A(l) B (2) - A(l) B (2) •. 0
Hence the scatte ring amplitude ( causal)is the Fourier trans form of the
re tarded commutator + seagulls . By retarded commutator we mean the commutator
( 3 7 . 3)
from this , ( 3 7 . 2) is directly obvious .

(Remark . The 8 ( t 2 -t 1 ) in 37 . 3 at first sight makes the result not rela
tivistically invariant , until it is realized that the commutator factor is zero
for apacelike regions so the plane t 2 • t 1 can be tilted arbitrarily as required
by Lorentz transformation . This is true at leas t if the commutator is not too

singular at equal times which is often true , Difficulties sometimes arise making
c
TBA defined here not relativis tically invariant unless corresponding non-invariant
terms ( called Schwinger terms ) are added into the "seagull" part of ( 3 7 , 3) . )
Expression (37 . 3) serves a s a general defini tion o f the Chew amplitude ,
i . e . in space-time it is the retarded commutator , even when we do not have the
diagonal elements , or lowest s t ates (so the produc t operator is 0 for v < 0) ,
The general defini tion of the Feynman amplitude is the time o rdered opera tor ( 36 , 8) ,
They differ by the product operator ,

Note on various relations
Reality conditions
(37.5)
{ C BA (v , Q) } * • CAB (v , Q) • -ciiA (-v , -Q) (37 . 6 )
Cross ing
{T � (v , Q) } * • T� (-v , -Q)
(37 . 7 )
Imaginary Part
{T� (v , Q) } * - T� (v , Q) • -iCAB (v , Q) (37 . 8)
The o BA satis fy reality and crossing relations required to keep them
valid for T . Namely
• O BA ( -v ,
/
oB (v , Q) -Q) • oAB (v , Q)
The commutator at equal times t 2 • t 1 can be obtained by integrating CBA (v , Q)

with respect to v for all v , since e �
iv (t z -t 1) dv /2w � C t -t ) . Hence Gell Mann ' s
• 2 1
equal time commutation relation
(3 7 . 9 )
Inelastic Scattering as Properties of Operators 1 7)
becomes upon Fourier t rans form for our diagonal element on the pro ton ,
..
1 C a b ( v , Q ) dv / 2 w ( 3 7 . 10)
a cons t an t , independent of Q. This is called a sum rule .
We now re turn to our study of the fact that in measuring �A ( v , Q) we are

measuring the F . T . of the commutator o f two currents . As examples o f questions
we shall ask are ( 1) What limitations on the F . T . result from the fact that the
coamutator vanishes outside the light cone? ( 2 ) From experimental facts about
the behavior of K (e . g . Bj erken scaling) what do we learn about the character

of the coamutator?
I t behooves us to study the general behavior of coumutstors , and we begin
our s tudy with the commutator o f two s calar fields of mass m in a sys tem without
interac t ion :
( 3 7 . 11 )
We can express • C l) in terms of creation and annihilation operators in the
( )
usual way
" ' , i) • 2: . " "' ' -1 /2

...
.r. ...... , + .: -·· .......
• ,
( 37 . 12)
Here "'k -� is the correct frequency to describe the operato r ' s

development in time for there i s n o interaction so the energy is that of a free
par ticle . The 8k ' s commute with each o the r , and the 8k * ' s , only a and a * do not
coDDDU te if they belong to the same k:
*
C8£• 8£• ! • olt'K' ( 3 7 . 1 3)
We work out ( 37 . 1 1 ) immediately by forming the commutato r , us ing ( 3 7 . 1 3) , of

an express ion like ( 3 7 . 1 2) . Dropping terms which obviously comnute we are lef t
with :
<; ( 2-1 ) • C ( t ,x) • 2

'K
1 (
2 "'k e
-i "'k t ik· x
e
( 37 . 1 4)
The actual integral can now be done - i t involves Bessel functions .
Here we will just do the special case m • O , and s ay wha t the case m2 ; 0
gives , leaving details to the student .
In ( 3 7 . 14 ) put k· x • kRcos e , "'k • k for m • 0 to get
-) •
C ( t ,x j 2 irk2 dk dcos e
3 e
-ikt ikRcos e
e + C,C,
(21') 2k
s ingular 6 function on the light cone ,
For f inite mass we obtain the same s ingulari ty on the light cone , zero
outs ide it of course , and a Bessel function inside :
Thus we see that the commu tator for free particles is zero outs ide the ligh t
cone and s ingular ( like 6 function) on the light cone .
We have dis cussed the s ingularities of two fields . Also inst ructive is to
dis cuss the s ingularities from two currents . One such current might be in a free
field theory J (1) - 'i'C l) (a

µ µ
+ aµ ) , ( 1 ) or � ( l) y \I ; C l) for spinor fields . As
an example we will leave out these gradients , e t c . and find the commutator for
" currents" that are s imply squares of a s calar field
where C ( 2 , l) is the free field coDDDu tator we worked out before . Clearly K
m
2
has the same s ingulari ty sgn ( t -t ) 6 ( s ) as noted before for C , but this time
2 1 12
multiplied by an operator function of two positions F ( 2 , l) cal led of ten s
b ilocal operator . This operator is needed , of course , only on the light cone .
Its mat rix e lements give functions of x -x ( for example of t -t for a

2 11 1 11 2 1
diagonal element in a system of a particle at res t) , So in general the
s ingularities along the cone are modulated by a function of the dis t ance from
the origin of the cone .
For interacting hadrons the commu tator of currents is expected to be zero
outside the light cone also , and non-zero only inside . An interes ting question is
how i t makes the t ransi tion as we cross the light cone , Some sort of j ump is
expected presumably . Is it in value or only in s lope ? Or possibly it has 6

function as the free particle case does . Such a ques tion is a fundamental one
only answerable by experiment .
We have the Fourier transform experimentally in K v (v , Q} . A singular
µ
behavior corresponds to some sort of high v , high Q limit ; so it is the behavior
at large v, Q that gives us the answer to the question . But in this region
experiments indicate K(v , Q} satis fies B j o rken scaling - that is, it is only a
fwiction of E; • -q / 2Mv .
2
The mos t straightforward way to analyze this is to take the inverse trans form
o f K, using the Bj o rken limit and seeing what s ingularities it gives - then
concluding the character of the s ingularities , because they depend only on the
high v , Q limit are the same for the complete function as they are for the
Bj o rken limit . We will not try to be rigorous , because for one thing we really
do not know , experimentally , K for q 2 > 0 , nevertheless we can see what happens .
2 2 2
Let Q be along the z axis . -q • Q -v • (Q z+v} (Q -v} • 2Mv ( , hence for
z z
•
large v , Q z is nearly v in fact Q z v+M( . The Fourier inverse of a function
of ( only would look like
Je Je
1 i ( vt-Q z} iQ z ( t-z} -iM(t
-- z f ( E;} dvdQ 1
• -- e f ( E;} d(dQ z
(211 )
2 3 2
(211)
• s ( t} O (t-z)
where s ( t} is the F . T . of the structure function f (E;} .

To put it in another slightly more rigorous way we note
J o (q2+2MvB} f ( B } dB
2
f (-q /2Mv} • 2Mv
Now omit for a moment the 2Mv (suppose we took the trans form of w2 } and
2
note for large v , q 2 that to a good enough approximation the 6 (q +2Mv8} csn be
replaced by 6 ( q +2MvB+� s 2 } . So we get for large enough v , ( q +2MvB+� s • (q+$P} 2 )
2 2 2
Now we can inser t , if we wish , a sgn(v} to keep the asymme try as is required
for a F . T . of a commutator, and for large v this is just sgn (v+BM} ( if the
proton is at res t} , Hence asymptotically
Now the F . T . is easy for we know that
2 2
F . T . sgn ( t ) o ( s ) • sgn (v) o ( q )
so that for a four vector a
hence
Jf ( S ) dS e
F . T .W2 -
-iSP · xsgn ( t ) o ( t 2 -R2 )
no ting P •x • Mt we have our previous result - that the significance of Sj orken

scaling is that the s ingularity of the current operators has a 6 function-like
singularity on the light cone . To be more precise we will have to include all
the P µ Pv factors , etc . , and define everything precisely - there will then be
gradients of & -functions involved . The bee t way to say it in a general way is
that s caling shows that the s ingularities on the light cone of the current
cOllDDu tators are j us t of the same severity as they would be for free particle
fields .
This is , of course , what we expect from the parton interpretation for there
we assumed in the final statl!s x (which become the intermediate states of the
commutator) the partons can be considered as free . Thus there is no surprise
in the conclus ion , it is only that we wish to state what we have discussed
(scaling) in an abs tract and general way (as current commutator s ingularities)
with minimum references to a mode l . Although the s tatement s till seems to refer
to free particle fields , that is only a shorthand to wri ting 6 and o ' functions
on the light cone .
Each general property of partons we assumed said something more explicit
about the character of the singularity . For example , to say charged partons
are spin 1/2 is to say �he singularities are like those o or 6 ' functions charsc-
teris tic of free Dirac field commutators . In addition , vector and axial vector
results are related . I f , for example , one adds that partons are quarks certain
numerical relations are implied , as we have seen , between the singular parts of
the commutators o f various kinds of currents . (That is to say for example , our
� p
�
relations among f ep and f P , f etc . implied by expressing everything in terms
of the six functions u, u, d , d, s , s. )
Quarks , as free particles , have not been found . There are many questions
as to whether the detailed views of the parton model are correct for quarks . In
particular , the question whether a single par ton quark moving out in recoil need
or need no t take its non-integral quantum number with it .
In addition it is often ve ry use ful to see , when results have been obtained
from a mode l , j ust how much depends on the model and whether in fact the results
canno t be stated as a general mathematical principle without recourse to a

specific model . In that way , if it proves later that too many details of the
model are faulty, we can s t ill begin again having learned some general properties
of hadrons without being committed to all the others .
Therefore it is interesting to note that the parton model is equivalent
to the general statement (and one much closer to direct experimental results)
that commutators have 6 -like light-cone singularities .
Light Cone Algebra
Lecture 38
Li gh t Cone Algebra
An interes ting ques tion answered by Fri tzsch snd Gell-Mann ( 19 71 Coral
Gables conference) is how the s caling results of the "partons as quarks " csn
be stated in a way which - at the end - doesn ' t involve quark wave functions
or ope rators at all .
We wish to s tate that the ligh t cone s ingulari ties are like those for free
quark (spin l/2 , su 3 triplet) c0Dm1utators . We mesn , of course , the largest
&' or & type of s ingularity , as we say the "leading" (in high frequency)
singularity . Firs t we see what the singulari ties are like if currents were
represented by quark fields •<x) , a Dirac spinor carrying su 3 indices on which

3 x 3 matri ces A can act .
The c0Dm1utator o f two spinor fields o f mass m is easily worked out (as we
did for Bose fields) it is
f
( Jus t as the propagator is < + m) / (p
2 2 2 2
- m ) instead of l/ (p - m ) so this
j + m comes likewise into the cODm1utator ( the real part of the propagator) .
Since we are looking for the leading singularity near the light cone Cm is
1 78
Light Cone A lgebra 1 79
2
� ( • ) and the mass term is smaller than the large gradient , thus
[ • C 2 ) , i'C l ) ] .;. !11 f/2 s gn ( t

2
- t 1 ) 6 (s2
.
/> ( 38 .1)
2 ia the s quare interval t:rom 1 to 2 and .; means " the leading

where s
21
s ingularities near the light cone are equal . "
I t i s eas i ly ve rified that the results we ob tain he re f o r the commu tators
o f two curren ts each a b i l inear form in �. are exactly the same whe ther we
s uppose a) the fields obey the us ual anticommu ts tion relations at equal times
3
• <i2 ) �( i1 ) + �(x > • <i2 ) • 6 (i - i2 ) and � (i1 ) , � (i ) an ti commute as is
1 1 2
app ropriate to Fe rmi parti cles ; o r we suppose b) the spin 1/2 f i elds obey
3 -
commutat ion relations � ci2 > �Ci1 > - �cx1 H <x 2 > • 6 Cx 2 - x > and � <x H <i2 >
1 1
co1111u
1 te as is appro p ri ate to Bose part icles ,
This is very interes ting b e cause it says the "Bose quark" model which
is appropriate at low ene rgies is in no fundamen tal way in con tradiction
to 11 partons as quarks " at higher energy . We can c a l l this parton model "partons
as Bose quarks . "
a
Next a current o f su 3 type a (described � ). ) is
•
a a
\I II
J (l) �(l) :>. y • ( l) (38 .2)
where f or example f o r e le c t r i c current :>. a i s diagonal s a y :>. Y • dia g . ( 2 / 3 , - 1 / 3 ,

y
- 1 / 3) . For axial currents change y 11 to y 11 5 ,
Now when we commute two currents , we find a s imple but tediously complicated
resu l t (see re ference for details) , To illustrate the idea , which is all we
intend to do here , we take the case of two electric currents , and also drop
several terms which would vanish i f we took spin averaged mat rix e lements , then
all that survives is
( p lus other terms d ropped)
where
(38 . 4)
(2 , 1 on mutual light cone} an d :>. 6 is diag , ( 4/ 9 , 1 /9 , 1/9} • ;1. Y ;1. Y in our case .
Now we s t ill s ee the quark fields in ( 38 . 4) , but the idea now is to dis -
regard equation ( 38 . 4) and to s ugge s t that equation ( 38 . 3) ( it s gene ralization
to arb i trary currents and including omi t ted terms of course} are gene rally valid .
They give the ligh t cone s ingularities and de f ine , on the ligh t cone at leas t
a set of new operators :� ( 2 -1) . It is the matrix e lements of these ope rato rs
wh ich give us our s t ruc t ure functions .
No dire c t implication is made that they can be expressed as ( 38 . 4 ) .
Equat ion ( 3 8 . 4) as we ll as ( 38 . 1) , ( 38 . 2 ) are j us t s caffolding
to arrive a t ( 38 . 3 ) and are hence fo rth to be forgo t ten . In ( 38 . 3) no explicit
q uark ope rato rs are seen . Ins tead only some new bilocal operators are defined
(meaning depending on two poin ts ) . They are de fined only on the light cone by
the very equa tions ( 38 . 3 )
[ Th is sys tem woul d be a true " algeb ra " if the p ropert ies o f the '.t' could
now be de f ined indepen den tly . Fo r example , an i de al s i tuation ( in fact a
comple te theory of the hadrons ) wo uld res u l t i f equations giving the commutators
o f s uch ·:P a could be given in terms o f J ' s and '.f' s themse lves . This does no t
seem pos s ib le . A li ttle can be done , howeve r . Wi th the form ( 38 . 4 ) the
commutators of two :ft s which have their variab les on the s ame ligh t ray
( e . g . [:;'{ 3 , 4) , 'Ji'( l , 2) ] where 3 , 4 , 1 , 2 all lie on a s ingle generator of a ligh t
cone ) again can be expres sed a s a n '.f'. This re lation has been a l s o hypo thes ized ,
but i t is not much . I t gives predict ions for certain two-current inclus ive
+ -
reactions l ike e + p � e + µ + µ + any hadrons . They are , of course , j us t
those expe cted f rom the par ton model in the same condi tions . We discus s them
later . ]
In order to use equation ( 3 8 . 3) we mus t take the Fourier trans fo rm , and
for that we can use the result (ve ry s imilar to those we have already der ived ,
we leave de tails to you) if q • ( v , O, O, v + MO for large '!.•
J d
4
x e
iq • X
6 ( x ) F ( x , x1 , x2 , x3 )
1 2
0
11
J - e
d t - iMf; t
F(t , O, O, t)
Iti
that is , i t involves on ly the integral along the ray t • z common to the plane
i v ( t -z)
t • z ( from e ) and the l ight cone .
Naturally every result o f this theory is also a result of the " par ton as
quark " theory for the latter is a model of the former ; but not eve ry parton
res ult can be der ived from ( 38 . 3) . It i s pos sib le , there fore , that many of
them are wrong and only ( 3 8 . 3 ) and no t the comp lete field model may survive .
Light Cone A lgebra 181
Therefore it is interes ting to compare the theories for various types o f

predictions made b y the parton a s quarks theory . They seem t o b e of three
classes .
A. Scaling , relations among scaling functions , sum rules .
B. Special arguments about the s tructure functions derived by arguments
about hadron collisions and other arguments ; like dx/x behavior ,
relation of form factor power l/QY t o x near 1 , structure func tion
(1-x) Y etc .
C . Applications to experiments o f Drell type e + e + hadrons or
+
p + p + µ+ µ - + any hadrons , etc .

Class A are exac tly those derived from light-cone algebra . Class B are not
obtained from light cone algebra by default . That is , they simply amount to a
discussion of how the matrix element of 'J(2 , l) might behave . It is not a
specific assump tion of the model but an at tempt to go further that leads to
Class B results - a serious attempt to discuss the matrix elements of the light
cone algebra would lead to results in this class .
Class C are very interes ting, as they seem beyond ( 38 . 3) and require some
extens ion , even though they appear evident from the parton view . It is therefore
here , testing these that a real choice can be made as to whether the extensions
of the parton model beyond the expectations of ( 38 . 3) are really sound .
The reason the current commutator on the light cone is not sufficient for
reactions like p + p + µ + µ + X is that we do need matrix elements like
+
< pp l JJ l pp >
but this time as we take the limit as the momentum of the current operator
increases we are also changing the state pp ; the relative momentum of the two
protons mus t also increase , This is an awkward limit for a theory of operators .
It is a very interesting ques tion , Is there some general abs tract way
without quark fields (partons) to describe all these Class C parton predictions
also ? Or are they perhaps wrong , unreliable extensions of the idea?
And what becomes of the question o f how the partons come apart in the
proton without exhibiting quark quantum numbers among the final s tates ? Does
current algebra help us to solve i t ? Perhaps , it seems to be translated into
"are the re any rep resentations pos s ib le of the algebra (38 . 3) which do not imply
quark quantum numbers among the localized states ? It may be more tractable in
this mathema tical form than in physical arguments .
Properties of Commutators in Momentum Space
Lecture 39
Properties of Commutators in Momentum Space
We now come to discuss what general properties a commutator has to have in

momentum space , such that its Fourier trans form will be zero outs ide the light
cone of configuration space ,
This is clearly a promis ing line to follow to supplement phys ical intuition
on the properties of Wµv ' At present not a great deal can be said , but for your
interest in your future research what has been done here may prove fruit ful ,
First we can make some obvious remarks . If we multiply a commutator ( in
space-time) by a function G which is 1 inside and zero outside a light cone we
recover the commutator (except , as is the case , that it is singular right on
the light cone where our function G ie poorly defined) , S ince the F . T . of G
is 1611P . V . ( l/q 2 ) 2 (P . V . • principal value) we have the convolution theorem
C ( q) •
J
C ( u) P , V .
l
16 11
c -u >
2 2 --
4
d u
(211)
4
( 39 . 1)
;& o
(plus pieces from the light cone) , (C (q) is the F . T . of a commutator, )
In a s imilar vein , since the F . T . of
�
we see that C must have the form
1 82
Properties of Commutators in Momentum Space 183
C (q) •
J 2 d4
X( u) sgn ( qo -uo ) o ' ( (q-u) )
( 2 w)
� (16 wi ) ( 39 . 2)
plus light cone pieces , Of course we can say also in (39 . 1) that the C (u) of
the integrand is any function F(u) at all , and the C that comes out on the le ft
will be the F . T . of a function zero outside the light cone . I have not been
able to make much use of these observations .
In us ing the simple views above we mus t be careful of one point . The
func tion we 111U l tiplied C (x , t) b y , G ( x , t) , was defined as 1 ins ide the light cone
and 0 outs ide ; what is it exactly on the light cone ? It is indefinite there ,
ordinarily that is such a small region that it makes no difference in the integral
ove r space t ime of C (x , t) G(x, t) but in fact it makes a great uncertainty because
C ( x , t) has a 6 function singularity j us t where G(x , t) is poorly defined . We can
correctly straighten out our formulas for this effect in the following way , We
call Ca (q) the asymptotic C ( q) giving j us t the light-cone 6 s ingularities . Than
C (q) -Ca (q) has no light cone s ingularity and so equat ion ( 39 . 1) for example , holds
if C is replaced by c-c a on both s ides . One can then simplify or rearrange the
equation to obtain one like (39 . 1) b ut with some addi tional terms related to
the light cone behavior .
A far more s ub tle and useful observat ion was made by Dyson . He noticed ,
in many problems we also know C(q) • 0 fo r certain regions of q a

s p ce (where
no in termediate s tates may be ava ilable) . For examp le , for q • v > 0 the
0
lowest s tate available for the final s tate x is the proton itself , moving with
momentum Q ( space part of q )

)I
hence with energy JN hence
v > - M + J�� 2
and likewise
v < + M - � ( 3 9 . 3)
Dyson has proved that the ne cessary and suf ficient condi tion fo r C ( q) to
vanish in region S • s ( q) < q0 < s (q) and to have a Fourier trans form vanish
1 2
outs ide the light cone is that C ( q) can be wri tten as
C ( q) • J J: 4
d u s
2 2 2 2
s gn ( q0 -u ) 6 ( (q-u) -s ) t ( u , s )
0
( 39 . 4)
o ( Dyson Rep resentation)
where � vanishes outs ide a region R , but is otherwiae arbi trary . The region R
is such that the (q-u) 2 -s 2 hyperboloid does no t pene trate region S .

That i t is a suffi cien t condit ion is easy to see ( the great dif ficulty
o f the proo f is to show it is necessary) . We have already seen that the free
particle commutator C (x , t ) is zero outs ide the light cone , so the F . T . of
m
(q 2 -m2 ) sgn ( q0 ) is zero outside the light cone . But if C (x, t) is multiplied
m
iu
by anything , in particular by P · x ( or any superposition over u) it is still
zero outside the light cone , hence F . T . e iu · xcm ( x , t)
sgn ( q o -uo ) 6 ( (q-u) 2 -m2 ) . •
2
Hence superposing with weight � (u , s 2 ) various cases of u and m s 2 we get •
( 39 . 4)
It i s easy to see , by drawing hyperbolas that for our case the region R
of integrat ion whe re � (u , s 2 ) does no t vanish is as follows
inside two cones

2
For s > ; l u0 l < l u l
ins ide region

For s 2 < ; bounded by
M - l u0 I • �u2
+ (M-s) 2
Incidentally , the s cattering amplitude that corresponds to this commutator
(using 37 . 2 ) becomes
T c (q) �
1
2 11
J J
R
d 4u ds 2 � (u s 2 )
---'---=--�...._ ___
2 2
(q-u) -s +i£ (qo -uo )
+ a (q ) ( 39 . 5 )
where a ( from seagulls) is a polynomial in q ( for the ampli tude

µ
� replace
One o f the di f fi cul ties i n using the Dyson representation i s that the func tion
� is no t unique , many can give the same C (q) . Another very s imilar repres en tation
has been derived (I sus p e c t not as rigorous ly as Dyson ' s ) by Deser , Gilbert and
Sudarshan especially for a problem like ours whe re C is a function only of the
two invariants q2 , v, it is
C (q 2 , v ) s � �
2 Mv
...
da
+l
dB H (a , S ) s gn (q +SM) 6 (q+Sp ) 2 -o )
0
( 39 . 6 )
0 - 1
This is , of course , j us t Dyson ' s representation i f the four vector uµ can be
assumed to have only a time component , that is , only a component in the P µ

direction .
There is again the expectation tha t H (o , S ) is zero outs ide a mo re limited
range if o <
2
M • In fact if o < �B
M on ly runs from -/a/M to + /a/M . but we
leave our expressions intact , j us t remembering H (o , S ) • 0 for I s l > /a/M in
this region of o.
I t is clear tha t i f C is zero on the light cone various gradients are

also , so several poss ib ilities exis t wi th di fferent integral powers of v ( f rom
ze ro up to any finite value) with co rresponding H functions ; but we have taken
the one appropriate to 2MW •

1
Equivalently we can write our fo rm with a new
de finition for o .
2
2MW (q , v )
1
• 2wv � �
m
do
+l
2
dBh (o , S ) s gn (v+MB ) 6 (q +2SMv -a ) ( 39 . 7)
0 -1
In addit ion , for our case , the asymme try in time for the coumutator becomes
the p roperty that
h (o , -B ) • -h (o , B ) ( 39 . 8)
This representation is very nice , and h i s probably uniquely de termined

by w1 • The weakness , however, is that no physical interpretation or expression
in terms of matrix elements is given for h (o , S ) . Therefore we cannot use any
physical intui tion in guessing how h should behave (other than using our knowledge
of how W behaves and working backward) . Tha t is to say we cannot at any stage
say - such a function for h (o , a ) is too "crazy" physically - or it ought to
behave so and so in this region , et c. As we shall see , this is a serious
weakness in this case .
We now turn to see how h (o , l!) must behave in order to produce a function
2MW 1 behaving as we expe ct (see Lo! cture 3l) for large v (we take v > 0 throughout) ,
Region I
2 2
First of all we have the s caling limit v + • , -q + • , -q / 2Mv • x finite .
Here 2MW 1 • f (x) a function of x only . Eq . (39 . 7) gives (sgn ( v+SM) • 1 for
large v , s ince v > M) .

+l
2MW 1 2Mv •
JJ
dB h (o , B ) 6 (-2Mvx + 2MvB-o ) do
0 -1
Now if we assume h (o , B) falls away rapidly enough with a only fini te a are
involved here , so as v + "' the a can be dropped in the 6 function and we have
h (o , x) do a function o f x as required
f (x) • f ..
h ( o , x) do ( 39 . 9 )
0
I believe i t was this argument with the Dyson representation which either
led Bjorken to his scaling hypothesis , or helped to confirm his suspicions of
i ts truth .
q 2 > 0 s caling
We can now determine how the function 2MW 1 (q , v )
2
We have here a bonus .
q + 00 such that
2 2 00 ,
+
f
should go in the posi tive q s caling region . Letting v
l 00
+ /2Mv • x ' is finite we have 2MW 1 • h (o , -x ' ) do which again scales ; but
even more , from the symmetry of h it sc�es to the same function fep (x ' )
-f ep (x ' )
2
2MW • q > 0
1
How can we see this remarkable result from the parton model? Can we
2
equally well derive the function for q > 0 by using physical arguments ? We
will explain qualitatively how it comes about .
For positive q we mus t be concerned in the commutator ( 36 . 1 ) with the
2
subtracted vacuum piece . In the vacuum we can make pairs , and at high l j us t
a pair o f partons
We are concerned with how this pair production probability is modified by the
presence of a proton
t? ·
At firs t one might think the modification mus t come through interaction of
the newly made partons with those in the proton - in interaction about which
we know lit tle except that it is fini te and is no t involved in our derivation of
2
f (x) for negative q But at high energy a mo re importan t effect is the
exclus ion principle - partons cannot be made if they are already present in
the pro ton (if Fermi s tatistics is assumed) . Thus in our diagram the proba
bility to produce a u to the left and u to the right at x' is 4/9 units (for
the charge is 2 / 3) . However , it is altered at those x by the chance , u (x) , that
a u parton is present in the proton . Thus we have a contrib ution proportional
to (4/9u (x' ) ; if the u p arton is moving forward it is (4/9 ) u (x ' ) . If a d par
ton pair is produced the probab ility is 1/9 , modi fied by the chance of finding
a d parton already in the proton , the contribution is (l/9 ) d (x ' ) , e tc . ) Thus
the entire contribution is 4/9 [u(x' )+ii (x ' ) ]+ l/9 [ d ( x ' )+d (x ' ) ] + l/9 [ s (x ' )+s(x ' ) ]
or f ep (x ' ) as require d .
Bose or Fermi quarks

In making the analysis for q 2 positive we assumed the quarks obeyed Fermi
statistics . If Bose s tatistics are used in the formal expression of the com-
pleted sum over s tates ( 36 . l) we have some changes . Firs t the re is a change in
s ign of the Fermi exclusion effect turning into a Bose tendency to increase
emission because a particle is already present . 'l'ben there is another change in
s ign in the closed-loop diagram <O j JJ I O> when changing from Fermi to Bose
s tatis tics [see , for example , R . P . Feynman , Phys . Rev .z! 756 ( 1949) ] .
•
On the other hand with the Bose statistics s ign the imaginary part o f
<O j JJ : O> i s negative and i t cannot be written a s a sum of positive probabilities
E j <O j J j x> I 2 but is minus that sum. This circumstance is the b as is of the
,
x·
usual proofs o f the relation of spin an d statistics ; spin one-half particles
canno t be consistently interpreted as Bose particles . The moat straightforward
interpretation of Bose quarks would mean the "Drell. Constant" defined for the
e+e - + any hadrons crose section would have the imposeible value -2/ 3 . Of
course a completely naive interp retation of quarks as Bose particles would also
lead to the wrong result that the wave function would be synmetrical under
the interchange of two protons . If quarks obey para-s tatistics both of these
problems would be removed .
Region II
We continue our discussion of equa tion ( 39 . 7) now turning to the region

v + m , -q fini te . Here we expect 2MW1 to go as v times a func tion of q
2 2
2 2
which we have writ ten as 2MW 1 2Mv g (-q ) / ( -q ) In this region our equation
• .
( 39 . 7) reads
2MW 1 • 2Mv �� m
da
+l
2
dB h (a , S) o (q +2Mv S -a )
0 -1
at first sigh t , for v + m , we could forge t -q 2 -a in the 6 and ob tain 2MW 1 •
�h ( a , o) da , a cons tant independent of q 2 and evidently incorre c t .However ,

0
the "cons tant" would , by ( 39 . 9 ) be f (o) which we know is infini te , as for small x ,
f (x) � a/x . This sugges ts that we assume for small B that h (a , S ) is singular .
h (a , S) • k(a) / S ( 39 . 1 0)
for small S . Substituting in the integral above gives
2!ti 1 • 2Mv f m
k (a
a-q
� da
0
Thus
�
-q
i�0
m
a-q
da
�
gives the -q dependence of the totalmvirtual photon cross section . The limit of
g ( -q 2 ) as -q 2 gets large approaches k (a ) da .
0
Lec ture 40
Region III
2 2 2
v +
m , large -q but Mx � • 2Mv - (-q )
2
Finally we turn to region III :
2 2 2
kept finite . There we expect 2MW 1 (-q ) -y h (Mx -M ) 2Mv where h is a function
•
2 -Y
of M z _; only . The (-q ) 2 2
is typi fied by the elastic s cattering (Mx -M • 0)
x
2 2
and y is perhaps 4 or 5 . For x near 1 , f (x) ..... A ( l -x) y-1 and for large Mx -M •
2 2 .2 1
h (Mx -� ) goes as A (Mx2 -� ) y- •
Properties of Commutators in Mo111e11t11111 Space 189
In th is region our equation ( 39 . 7) reads
J f
"' +l
2 2 (40 . 1)
2MW1 • 2Mv da dB O ( Mx -M +2Mv (B-l) -a )h (o , fl )
0 -1
Again a "natural" guess (neglect M 2 -� -a relative to 2 Mv (B-1) in the 6 function)
leads to the wrong answe r : 2MW •

1
x ..
1
do h (o , 1) , a constant independent of
�
Mx -� . But again we can fix it up by a mor! comp li cated guess for the behavior
of h (o , B) near a 1 . Because , from ( 39 . 9 )
• do h (a , x) i s zero ss x + 1 (like
0
A (l-x) y - l ) , Therefore one way to get the answer would be to assume that h (a , B )
goes to zero near a • 1. For example s uppose
h (a , B) • D (o ) (1-B) y - l as fl + 1
l
"'
then D (a) do • A. (If we assume y depends on a we can be more general and

ge t nearly the same final result except with some slow logarithmi c dependences ,
which might in fact be the re in this region . This is not known very well
experimentally) . Substituting into ( 40 . 1) and integrating ( the 6 function says
2 2
1 -B • (Mx -M -a ) /2Mv ) we get
2 2 1
2MW1 • (2Mv) -y+ l (Mx -M -a) y - D (a ) da ( 40 .2)
2
hence of the correct form if ( call Mx -� • A)
J
A
h (A) • ( A -o ) y -l D ( a ) d11 (40 . 3)
0
This appears a t firs t satis factory ; h will have the right asymptotic behavior
AA y-l for large A if D(a) falls off fast enough with o, and j D (o ) da • A .
0
However , we will again be fooled in our qualitative e xpectations for h (A) from
this equation if we suppose that D(a) is a s imp le function . The operation of
integration in ( 40 . 3 ) is a very powerful smooth ing operation ; for no s imple
D(o) would we be lead to guess what we know is true - that h ( A ) shows a series
of peaks and valleys for small A , the resonance peaks , It is no t that (40 . 3)
says resonance peak.a are impossib le - it is that (40 . 3) plus a guess that D (o )
i s not too complicated (i . e . does not vary i n a rapid and peculiar way ) fails
to do so . To see this take the case y • 4 and note that ( 40 . 3) can be solved as
1 d4
DO.) • 6
-:-:ii h ( >. ) (40 . 4 )
d >.
The fourth derivative of a resonance oscillates in a very special manner . (For
a much more intelligent way to deal with the resonance region see Cornwall ,
Corrigan and Norton , Phys . Rev . DJ 5 36 , ( 19 71) and the next section . )
To summar ize this inves tigation , the Dese r , Gilbert , Sudarshsn representation
is use ful as a formal tool to get dispersion relations , etc . It is , as of course
it ought to be , capable of representing what we know of the experimental data
with some function h (o , S ) . But it is disappointing in that its predictive
utility - by guessing "reasonable" behavior for h (o , S ) - is not very great first
because we have no phys ical picture o f what the function is , and second i t seems
like a function that behaves in a way as complicated as (if not more complicated
than) the data i tself .
It seems that h (o , S ) is not , in fact , a function that has much direct
physical s ignificance and is purely an artificial expression in a certain
mathematical form of the fact that the commutator vanishes outside the light
cone .
It might pay to try to find ano ther representation o f this fact in which
the kernel functions may have some direct physical representation .
Scattering in the Deser, Gilber t , Sudarshan representation

Given the commutator in any form we can find the scattering amplitude
2
T (q , v ) corresponding to it by us ing the dispersion relation (3 7 . 2 ) . It is
easy to see by direct substitution of ( 39 . 7) that the s cattering which has
the right spin dependence (i . e . o v q -q qv part) to cor�espond to 2MW1 (call it
2
µ µ
_,,.,1 ., r
2MT 1 ) is
2 -l ( 40 . 5 )
2KT 1 • dS h (a , O) (q +2K,""'1< op (_) ) + , ...... ..
0 - 1
ii
... +l
2MW1 2 ( 40 . 6 )
• 2Mv d dB h (o , S ) sgn (v+Ml! ) O (q +2Mv-o)
0 -1
Properties of Commutators in Momentum Space 1 91
This is the causal amp l i t ude convention , in the Feynman conven t ion the
it sgn (v+MI!) is replaced s imply by i t . The "seagulls" are a polynomial in q 2 ,v .
This can be readi ly verified ( s imp ly n o t i ce tha t sgn (v+MS ) 6 (q2+2MvS-o) • s gn (v+MS )
a«v+MS ) 2 -E2) • ( 6 (v+MS -E) -6 (v+MS+E»l2E where w e have writ ten E • +(o+M2 S 2 -+Q2 ) 1 12) .
�: Be cause we are expressing T1 as an int e gra l wi th a v facto r times the

in tegral we have really been w o rk in g with W/v an d T/v ; but there is a pole in
v (W1 _,. 0 as v + O b u t T1 does no t ) . To avoi d this use in our dispe rs ion relation
W/ v to ge t
2 2
(T 1 (q , v ) -T1 (q , O ) ) /v .
�: I t is evident from ( 40 . 5 ) and ( 40 . 6 ) that a l th o ugh derived from the

dispersion relation in v for fixed Q ( s p a ce part of q) they also s a tis fy a
dispers ion re l ati on in v f o r fixed q2 ( for q2 < 0) (a once s ub t racted dispersion
re lation)
2 ...
d (v ') 2
•
2'r
.Y...... J 0
(40 . 6a)
(Assuming no seagulls b eyon d a con s t an t term ab s o rbe d in T1 (q 2 ,O) . Thia can be
s een by dire c t s ub s t i tution o f ( 40 . 6 ) into (40 . 6a) , remembering that W(q 2 , -v ' ) •
-W (q 2 , v 1 ) , i . e . th a t h (o , I! ) •- h ( o , -S ) .
C ons i de r
J dv '
v ' - (v+i t )
o ci+2Ml!v ' -o) sgn (v '+MS) • -l
q 2 -o+2MSµ+iugnS
eve rythin g is clear exc ep t the i ts gn S shoul d be i e s gn ( �B ) . Let us ch o o s e to
work with v > 0 (T is symme tric in v ) , Then for q2 < O there is no po le in
(q2+2MSv-o+it ) - l i f B is negative , the sign of ic is a rb i t rary ; we may replace

sgnS by s g n ( v MS ) .
+ I f I! > O , s gnl! • + • sgn (v+MS) so it is s till correc t .
( For v ne ga ti ve the s ign reverses . ) Hence we have
J dv '
v ' - (v+i c ) v
'
W/ v ' is symme tric so we can set the range from 0 to wi th l / (v ' - (v+iE)) +
00
2
1/(-v ' - (v+iE ) ) -2v/ (-v ' 2+ (v+iE ) ) . From th is equation (40 . 6a) follows .
•
Fo r w2 which converges fas ter we expect a corresponding unsub tracted

dispersion relat ion
f ---
2
2
d (v 1---
) 2
W (q v ' )
(v ' ) - (v+iE) 2 2 •
for l < 0 •
2
For q • 0 these are of course the Kronig-Kramers formulas relating real and
imaginary parts o f the index of re fraction ( forward light scat tering) which
is necessary if signals are not to come out before they go into a block of
s cat tering material .
We try now once again (unsuccess fully as it wil l turn out ) to try to
ob tain some expected limi tations on 2MW 1 because we know it is a causal ( ze ro
outs i de the light cone in space) commutator . This time we shall use behavior
valid not only for large v but for any region . For example we know for the
elas tic s cattering if the p ro ton were a point charge we would have for v > O ,
2MW1 2Mv6 ( (p+q) 2 -M""2 )
• =
2
2Mv6 (q +2Mv ) . (To get the correct symme try for v
posi tive and negative this can be writ ten more correctly as 2MW 1 2 Mv (sgn(v+M) •
6 (q +2Mv) - sgn (v-M) 6 ( l -2M v) ) . This is obviously causal - as we have seen -

2
for i t comes from perturbation theory o f fields .

Now in the real world this is mul tiplied by a factor (the square of the
2 2
elas tic form factor) , a function of q , say f (q ) which falls off gradually ,
2 2
for negative q 2 , from q 2 O , to behave as ( -q ) -y for large -q • One would
•
expect such modulation of the function expected for point-like particles to

represent some kind of smearing of the point and to perhaps imply a lack of
causali ty - a lack which mus t be balanced by correct contributions from terms
off the elastic mass shell ( corresponding to o ther resonances , etc) . Thus , can
we not make s ome requiremen ts o f behavior in other regions , especially non
2
elastic regions by our knowledge of f (q ) for negative q 2 ? Unfor tunately not
( aaCornwall , Corrigan and Norton , (Phys . Rev . 03 537 (1971) ) show) it is
possib le to s tay on the elastic mass shell for negative q 2 and only alter the
behavior for posi tive q 2 and s till arrange virtually any f (q 2 ) which falls off
fai rly smoothly . In parti cular if f ( q2 ) can be written in the form
=f0
�
q 2 -µ
(40 . 7 )
for nega tive q 2 , i t can be done . We can see why this is , and also get a
better physical fee l for const ructing causal functions by the fo llowing con-
siderat ions . I t is eas iest to deal wi th the s cat tering functions T , fo r these
to be causal they must be the Fourier trans form of a retarded coumutator and
thus zero outs ide a forward light cone , Let a ( x , t) be such a function and A(q)
its F . T . and let b (x , t) be another such function , B (q) its F , T , Thus the con-
vo lution of a (x , t) and b (x , t) is obviously , by geometry , such a function (zero
outside forward ligh t cone ) and hence i ts F . T . or s imply A ( q ) B (q ) is again
satis factory (a causal scat tering function) .
We see that comb inations by multiplication (and addition of causal scatte ring
functions are causal scat tering funct ions . The s implest causal scattering
function is
[q2 - m2 + i£ sgn q 4 ] -l (40 . 8)
i
We can generalize this (multiply by e u · x in space time) to find for any four-
vector u
lJ
that
(40 .9)
is a causal function . Thus the e lastic scat tering function for p oin t p a rti cles
[ (q+p) 2 -!r2 + i£ sgn ( +M) )
-1
(40 . 10)
evidently is causal (as well as what you ge t by putting -q for q) , We can
multiply this by an express ion like ( 40 . 8) , we see
-l
2 2
[q -t11 +i£ sgn v ) [ (q+p) 2 -i/- + i£ sgn(v+M) ] -l
is causal . This is true for any m2 • µ and hence any superposition with weight
p (µ) dµ is , so a scattering amplitude like (note (q+p) 2 � • q2+2 Mw
J p ( µ) dµ
Tl
•
1
(40 . ll)
q - µ+- i£ sgn v
2
v q 2+2Mv+ i£ sgn ( v+M)
is , by itself causal . To get the correct symme try for v one need only add the
corresponding expression with v replaced by -v ; we will suppose it is always
done and not write i t explicitly .
(You might find it physically more satis factory and easier to in terpret
if the factor is considered as a form factor due to virtual particles like p ' s
at each vertex . Those at one vertex would contribute a factor
-1
w(µ) d µ
2 (40 . 12)
q -µ+ i£ sgnv
0
where µ is the mass squared of the virtual particle and w measures the weigh t
o f its contribution ; g i s causal . W e would then expect t o multiply ( 40 . 10) by
2 2
( g ( q ) ) , one factor for each coupling ; the result would still be causal ,
and possibly physically easier to unders t and . )
To get w 1 / v (commutator) from (40 . 11) w e need only take its imaginary par t
v
•f Pv + q -µ
2
p ( µ ) d µ " sgn(v-Hl) 6 (q -2Mv) +
0
f
2 1 ( 40 . 13)
+ ngnv6 (q -µ) p (µ ) dµ PV -2--
q +2Mv
0
In the region q 2 <

0 the las t term disappears and we are j us t lef t with
2
the elas tic point charge scattering multiplied by a factor f(q ) given by (40 . 7) .
It is disappointing that the res trictions of causality do no t affect our
region of experimental observation (q < 0) . Furthermo re , it will be hard in
2
practice to ge t p ( µ) exactly from knowledge of the integral (-q

2
• Q2 )
( 40 . 14)
even if fairly exact . It is not easy to reverse the integral into accurate
knowledge of p (m) unless physical arguments ( P dominance etc . ) are availab le .
Thus again we are thrown back in this problem to unders tand the process
physically ; the mathematical properties do not help as much as we had hoped .

2 2
(We know that if f (Q ) falls faster than l /Q , say as (l/Q 2 ) 4 ; then we
J0
1
can conclude , by (40 . 14) , that p (µ) dµ • O. Such a relation is called a
2 2
superconvergence relation -f (Q ) converges for large Q more rapidly than
4
the form woultl suggest . Again s ince f falls as ( l / Q2 ) we can conclude that
the moments �
µ np (µ) d11 vanish for n 0 , 1, .
2
• Alternatively , f can be
expressed as ( g ( q2 ) ) 2 with g ( q2 ) • J y ( 11 ) ( q2 -11 )

-1
d11 J
and y ( 11) d11 s 0.)
Lecture 41
The idea discussed at the end of the previous lecture is iDD11e diately
generalizable to scattering through an intermediate resonance , say of mas s Mx2 '
2
M -i/- • >. . (q2+ Mv ->. + i£ sgn (v +M) ) -1 .
2
call A point coupling would give scattering as
x
We can mul.tiply by any form factor, say with a p ( >. , 11) • We are thus representing
things by a s um of s channel resonances each of which has a square forir factor
l
f { >. , - > • J p ( i , \I ) dµ { 41 . 1)
-q + 11
The total scattering from all this {s channel resonance representation)

would then be
+
p ( >. , µ ) d >.dp
q 2 -l!+i £ s gnv q 2+2 Mv-Hi£ sgn ( v+M)
+ s ame term (v + -v ) + seagulls (41 . 2 )
The w 1 which goes with this { the imaginary part of T 1 / v ) is
"' "'
w1
-
v
•
jj p { >. 1 11) PV 22
q
11
-11
[ sgn {v+M) H q 2+2Mv->.) -s gn ( v-M) 6 ( q2 -2Mv ->.) ] +
0 0
+ 11S gnv JJ "' "'

p ( >. 1 11 ) 6 { q 2 -11 2 ) PV ( z
l
q +2Mv ->.
( 4 1 . 3)
For q2 < 0 the �a s� term vanishes and we have for v > 0 simply
w /v
1
•
j
11
..
f ( >. , -q
2 2
) 6 (q +2Mv->. ) d>. • f (q2+2 Mv , -q
2
) ( 41 . 4 )
0
a superposition of contributions for each e f fective mas s with a fo rm factor
2
f ( A , -q ) of form given in ( 41 . 1) .
Have we not gone in a complete circle ? Our original express ion giving
1 96 Photon-Hadron lnteractions
(except for pho ton polariza tion factors ) w 1 was of the fo rm
2
( for q > O , v > 0)
2
This looks j us t like (41 . 4) , The 6 is of course 6 (q +2Mv-A) so we interp re t
2
f ( A , -q ) as L
l l 2 summed over s t ates of a given mass 2 • �+ A ,
2
(At first this would seem to make a function of Q , the space part of the
2 2 2
momentum trans fer s quare d , ins tead of q v -Q ; but it is the same because
=
2
the 6 function relates v to Q and it can be expressed either way . ) We see
2 2
f ( A , -q ) mus t be posi tive for -q > O ; of course , s ince the lowes t elas t i c
2
M
x M ( A • O) is separa ted from the con tinuum at Mx M+mn ( A th • 2 mn M+mn ) .
• a
2
The function f ( A , q ) and hence p (A , µ ) will have a 6 ( A) cont ribution and after
that the integral in (41 . 3) will s t art at an inelas tic thresho ld A to m
,
th
We may not have gone in a comp lete circle . Firs t we now know (a) that the
weigh t factor f( >., -q ) mus t be expressib le in the form (41 . 1 ) and (b ) we know
2
2
what the func tion looks like in the experimentall y unavailable region +q > 0
(see 41 . 3) and , of course , the scat tering funct ion (41 . 2 ) that goes with i t .
But do we know this ?

We only proved that the form (41 . 2 ) was causal ; we have no t proven tha t
every causal funct ion must be exp ress ible as (41 . 2) and at the moment we do
no t think we can .
Since (41 . 2 ) is causal it mus t be express ible in the DGS form ( 39 . 6 ) . One
way to do this (suggested by Comwall , Corrigan and Nor ton , Phys . Rev . D3 5 36 ( 19 71) )
is to comb ine denominators (i t is eas ie r to use the Feynman amplitudes replacing
iE sgn ( • • ) by iE) via

1
dB 1
l
'"'l'---
2 • 2 -
2
2 q _,, q +2Mv-,
[ q +2Mva- AS- µ ( 1-B ) l � A
to get a s ingle denominator. Thus we call a •AS+µ (l- B ) and integrate on a by
parts to prove
h (a , B) • ��
m m
p ( A , µ ) 6 1 ( A B+ µ ( l-B) -o) dAd µ (4 1 . 5 )
0 0
Of course this can now be s impli fied . I f from h (a , B ) we could always find a
Properties of Commutators in Moment11111 Space 1 97
p ( >. , µ ) which would give this h we would have a proof (assuming the DGS rep re-
sentation is proved) that (41 . 2 ) is also a necessary form . We suspect it
cannot be done and i t is possible that (41 . 2 ) , although very physical , is not
the complete exp ression ; but o ther forms (o ther types of diagrams - or "channels "
o ther than s channels) might have to be added to it to ge t a complete repre-
sentation . This is a good problem .
<
2 2
W for all q in terms o f W for q O?
We have a form that suggests the quest ion a s to whether knowledge o f the
2
fact tha t W is causal , and knowledge of i ts value for negative q only, enables
us to find it for all q , v .
2
We are not now concerned with the practical fact
2
that knowledge of f ( >. , -q ) to a given limited experimen tal accuracy does no t
permi t discovering p ( .l. , µ ) very well and does not mathematically alone define
f ( >. , -q ) for q > O . Rather we suppose w1 perfectly known for q < 0 and ask
2 2 2
to what exten t T is defined everywhere .

This has great interes t for there are quan tities , such as electromagnetic
f
self energies , which can be defined in terms o f T as integrals perhaps (e . g .
2 2
T ( q ) d4 q/q ) . If T were un i que 1y de termine d b y W in the experimental 1 y
accessible region we might search for express ions for these integrals directly
in terms of this W at q < O . (Cottingham formula for self-energy . )
2
2 2
Given W(q , v ) in the availab le region (momentum like q ) how unique is
W(q , v) in the energy-like region ? Let W (q 2 , v ) and W (q 2 , v ) be two causal
2
a b
functions identical in the q < 0 region . Le t us study what is possible for
2
their di fference w - w - w . W (q 2 , v ) • 0 for q2 < 0 and hence is causal .

d a b d
What form must it have ? We see immediately it need not be ze ro because
2 2
sgnv6 (q -m ) is such a function . To get the mos t general form we use Dyson ' s
theorem which says (s ince i t is causal) that i t mus t b e expressib le in the form
wd -
J J..
4
d u
0
ds
2
sgn (q o -uo ) o ( (q-u) -a
2 2 2
) t( u , s ) (41 . 6 )
4>
2 2
where i s non-zero ins ide the region where the hyperboloid (q-u) ..s does not
penetrate the region S of q space where we know Wd vanishes . Thus S is the
< 5 u •
2
entire q 0 region . It is seen that every hyperboloid cuts unless O.
Thus the mos t general form fo r W is
d
m
1 •
2 2 2 2 2
d
- sgnv6 (q -s ) ' (s ) ds sgnv � (q ) ( 41 . 7)
0
where by de finition Hx) • 0 for x < O.
2
Thus complete knowledge o f w 1 (q , v ) in the experimentally available
q < 0 region permits definition o f w1 everywhere within an arbitrary constant ,
2
2 2
( independent o f v ) function of q for positive q • T is de termined also up
to the function
(41 .8)
Strictly this argument is not valid , gradients o f 6 func tions in space time
2
could be used in Dyson ' s theorem. More correctly , at each posi tive q the
function W is de termined by the behavior of W(q , v ) for q < 0 up to an unknown
2 2
2
polynomial in v , the coe fficients of which are arbitrary functions of q •
Physical arguments about large v asymptotic behavior would have to be used to
limit the degree of these polynomials . The s caling limit for
0 and the x <
fact that W is odd tells us that (41 . 7) must be of the form v' (q ) ( for v > 0)
2
and that T 1 is determined up to a function of q 2 given by (41 . 8) . Note this

function is not zero for i < 0 so T1 is not completely de termined for q 2 < 0
by W . This agrees with the dispe rs ion result (40 . 6a) where a subtraction had
2
to be made and an arbitrary function (T 1 (q ,0) ) lef t undetermined .
Electromagnetic Self Energy
Lecture 42
Elect romagnetic Self Energy
We now discuss s few places where knowledge of T would help us calculate the
electrodynamic p roperties of pro tons and neutrons . Since we have measured , in W ,
the electromagnetic coupling o f p ro tons w e migh t hope t o use the experimental
knowledge to determine the electromagnetic energy of pro ton and neutron and
their measured difference , to compare to experiment . As we shall see the hope
2
is , at presen t , frus trated be cause knowledge of W for q < 0 where it is avail-
able is not quite enough to determine the electromagnetic coupling everywhere
(T) - the arbi trary func tion of �(s) men tioned in the p revious lecture is not
de termine d . Since the answer for the p-n mass diffe rence is only one number we
are frustrated until we can find a theoretical or experimen tal way to determine
2
T unique ly - for example to determine t 1 (q ,0) of the dispersion relation (40 . 6a)
2
for q < O .
Be fore we dis cuss this by formal mathema tics let us see what we can expect .
As is well known , the self mass of a point spin 1 /2 particle diverges logarith-
mi cally . For the electron t. (m2 ) m z-;;-
2 Je 2 A2
• tn 2 where A is some upper cut off
m
for electrodynami cs if we replace the photon propagator
:..;, �q 1 99
This Am2 is experimentally undefinab le . A s imilar infinity for the AM2 electro-
magnetic for the proton uould also - by itself not be observab le , b ut
• - 1 . 2 9 34 MeV
is , in fact , observab le and is measured to 5 sign i ficant figures . Can we

calcula te it - or even see what order of magnitude it is? - for example does
our present theory say i t must be infinite ?
As long a s the electromagne tic interaction of the nucleons were mysterious
one could always say anything could happen - but now that we have some knowledge
of them we mus t answer more speci fi cally .
The divergence o ccurs from high frequencies and at first it was thought the
hadrons mi ght be soft at high frequency and the electromagne tic self energy con-
vergent . But now we know for inelas tic s cattering a t leas t they look like they
are made of point-like cona ti tuents . Does this point-like behavior mean the
energy mus t diverge ? Of course , the pro ton could divex ge and the neutron also-
only the di f ference need converge - but the difference in point-like s truc ture is
w 1p -Wln is also finite and point-like in the s caling limi t . lt is the coincidence
of the 6 -func tion of the photon prepagator and of the electron propagator which
makes the divergent self energy - and now we see the pro ton , the neutron , and
proton minus neutron all have s ingular behavior on the light cone so i t at
firs t looks like divergence is inevi tab le ,
Le t us es t imate how much . Since we are dis cuss ing the high energy behavior
we can use the i dea that proton•partons . Clearly the self energy diagram of
mos t importance at high energy in the s caling limit are when a photon is emitted
and absorbed by the same parton - therefore , as far as the divergent (tn A 2 )
part is con ce rned it is as i f each parton gets a shift in mass proportional to
Am 2 • m 2 e 2 tnA 2 /m 2 where e , mi are the charge and mass o f a parton . How
i i i i i
much does this change the mass of the p ro ton? We cannot honestly say . One
might try to calculate the change in E - P z � / 2 P by calculating the sum o f
•
2 2
the ch anges in £ -P • 9 � of each parton . We would find
2
( AM2 ) nucleon • L Am2

partons x � x f
m2 fep
(x) tnA 2 dx
There are obj e ctions to this (by the way it is also even mo re divergen t
since f "' l/ x and the x integral cannot be carried to 0) . The energy is not
ep
j us t the sum of the kine tic ene rgies of the partons , interaction energies among
the partons are also invo l ved . Thi s is reflected in the dangerous formula con-
taining the mas s squared of a parton - a thing we said was meaningles s to
t.Aup to now . The dis t rib ut ion of partons is changed in f i rs t order also so
2
the total m change is not co rrec tly evaluated . ( I t might be though t we are
right to take < ljJ I AV l 11» for the per t urba t i on on the Hamil tonian us ing the old
AV • Am a•a is
wave funct ion - b u t
2 not the change in the Hamiltonian , only
of the Lagrangian - in the Hamiltonian there are many o ther e f f e c t s on the
interac t ion terms through 1 / /2; factors e t c . , so we have not computed the e f fect
o f the perturbation co rre ctly . )
It may well be that m 2 O , or is effectively z ero (a suggestion to me
•
i
due to Zachariasen) due to in terac tions (or as a matter o f principle) so the
2
logarithmically divergent perturbations A i depending on inA neve r arise . This
may be so , but we do not know. We mus t turn to more detailed q uan tit at i ve
analyses if we are to try to study this further .
Electromagnetic mass shifts come from the emission an d reabsorption o f
a virtual pho ton .
J�
of the photon . Hence by taking Fourier trans form
2 2 F 2
AM • 4ire
(211)
2 _
_l
q +i£ \,µ (q , v ) (42 . l )
We have written
so that
(42 . 2)
The expression in square brackets has the imaginary part ( 1 -v 2 /q 2 )W -w which

2 1
is the contribution from longitudinal photons . If partons are spin 1/ 2 it
falls faster with v
than does T whose imaginary part w 1 simply scales to f (x)
1
in the Bj e rken limit . We will hereafter just write T for T + [ ( l v 2 / q 2 ) T2 T ) /4
l
- -
l
and suppose the imag inary part W • w1+[l-v 2 /q 2 )w2 -w1J/4 scales to f (x) in the
Bj o rken limi t .
Lecture 43
C o t t in gham formula
As we have seen we can write for the electromagne t ic mass e f fe c t
(43 . l)
where 4T • T
\l \l
W e mus t integrate this over a l l q2 bu t w e have seen T is
de te rmined by i ts behavior f o r q2 < 0 (where experiment can s a y some thing
ab out i t) so it is pos s ib le that maybe (43 . 1 ) can be wri t ten in terms of T for
q2 < 0 only . How this c a n in f a c t be done w a s s h own by Cot t ingham . H e showed
that in the four-dimens ional integral the conto ur on v could be changed (withou t
pas s ing any s ingulari ties)
iw ,
from the real line v • ......, to
m to the imaginary
n
axis v • 111 • ....... to � . (We show how i t i s done later . ) Suppos e i t i s t rue .
We can then wr ite d 4 q • 2 QdQ2

d 111 q 2 • v 2 -Q2 • -w 2 -Q2
Now
(43 . 2 )
( Co t tingham formul a)
Now the quan t i t y is all righ t for -q 2 b u t is comp le tely unphys ical for v is
imaginary . We can de f ine it howeve r by analytic continuation by our dispersion
relation ( 40 . 6a) , set ting v • iw
T(qz , i111) • t(i , 0) - :z 1 ( 4 3 . 3)

0
m 2j+ P m dv ' 2W{ q2 , v '>}

Hence
�-q2 -wi {�T(qz ,0) - 1112

I d,_ .
I , 2 ( , 2.L, ,2)
dw
n
(43 .4)
V V
� TW
2
.n
o
-q
o - v-q ·
We can carry out the integrals on w to get
· J"
0
dC -\ {!.
-q
>
4
C -<i2>rc -l ,o> -!.]"(�=\+� - )
2
0
v'
1 1
� W(q2 , v ' ) dv ' ;t
2 v '.2
(43 . 5) J
This then succeeds in get ting an explicit formula for the self energy in
terms of in a region access ible to experiment . However , we are frus trated
2
W ( q , v)
2
by the unknown term T ( q , 0) .
It is essential to know something about T(q if we are to be able even

2
, 0)
to determine whether the self energy diverges . Le t us look first at the contri -
bution from the second term in
could come . Put q - 2
•
for large v converges to the limi t f (x) .

(43 .5) in the scaling region from which divergences
2 Mvx and consider W as a function of x and v , W(x , v) which

The term becomes
l�
..
vdv �� 2
+ 1 -1 -} (� W(x, v)
0
or for large v , where the square bracke t is we get

2
-( 2M:x/v) / 4
J i
.. 1
�v ; xf (x) dx
f
1
The v integral diverges logarithmically wi th coe fficient xf (x) dx , the fraction

0
of momentum carried by the charged partons each weighed by the charge squared ,
Of course a cut-off of electromagne tism is used in d ( -q ) / ( -q2 ) is
2
(43 .5) ,
A
d ( ) -:-2"2
replaced by �
2 2
•
-q A ..q
This provides a cut-off for our v integral (of order t / 2 Mx) so the part
2
diverging as ln t. has a coefficient xf (x) dx . �
the ln A 2 term i f it
J f
However , other term in could also produce
2
T ( q , 0) a
2
only falls as C / ( -q 2 ) as -q 2 "• (If · + ( s ) ds/ (q 2 -. ) this C is • < s) ds . )
2
+ T(q ,0)
It is therefore possible that these divergences from T and W cancel and that the
self energy (or at least the proton -neutron difference) is finite and calculable .
There are two views we could take at presen t . We know of course the p -n
mass difference converges so let us talk about T and W for the p-n difference at
least . In principle T could be determined by experiment and is therefore defined
physi cally , ei the r of the following could happen .
(a) Equation (43 .5) with this T s t il l gives a logari thmi c divergence .
The reason is that our theories are wrong for high energy ; this and the
elect romagne t i c s e l f energy calculation of QED are b o th wrong and will
b o th be f ixed b y the same modi f i ca t ion of re lativi s t ic quan tum mechanics
a t high energy some day to b e found .
(b) The T is s uch tha t the integrals converge and agree with the expe ri-
mental mas s d i fferences .

2 2
(c) The T ( q , 0 ) ( for q < 0) is in fact not precis ely definable expe rimental ly ,
that it is to some exten t arb i t rary - hence that our theory is not ab le
to calculate this mass difference precisely and mus t be " renormalized . "
I believe in th is case , if partons are quarks only one renormali z a t ion
cons tan t , corresponding to the elec tromagne tic mass dif ference of u
and d quarks , would suf fice to make all the hadron self energy dif ferences
converge s imul taneous ly .
As Z achari asen has sugges ted to me , the b e s t (mos t limit ing) thing to
do t od ay is to assume (b) is t rue . This puts restric tions on possible theories
and may have predictive value . I f it leads to a paradox or incons is tency we
le arn that ( a) mus t be so . Zachariasen has shown tha t all will be convergent
if the equal t ime commutator o f J and j vanishes [ J

µ jµ ]
, = O. In the quark
model it corresponds to quarks having zero res t mass .
I believe thi s is a very good p roblem to work on . I myse l f haven ' t found
enough time wh ile preparing these notes to analyze it in a more e lemen tary or
clear fashion for you .

2
How can we ever hope to ge t at T ( q ,O) expe rimen t ally or theore t i cally ?
It is the forward scat te ring amp li tude on a proton of a virtual photon of

2
mas s -q • I t woul d be invo lved in a two -electron forward scatte r ing e+e+p -+
e+e+p via the diag ram
This is no t an experiment that can be done .

2
But knowledge of T ( q ,v) anywhere
would be of assistance be cause the dispersion re lations can be used to conve r t

2
i t t o knowledge o f T ( q , 0 ) ; s t i l l no expe rimen t suggests i tse l f .
It is in teres t ing tha t T (O , O) can be obt ained theore tically - the fo rward
Compton s catt2ring from a p ro ton off a real (on shell) pho ton . For Q + O , v + 0
we have very long wave length slow f ie lds to which , of course , the proton looks
to be simply a mass ive point charge of mass M . I t s cat te rs then a s i t would

e2 + +
A •A
2M
clas s i cally ( o r non�re lativis tically via S chr5dinger equa tion from the
term) thus ( called Rayleigh scattering) i t gives
1f •
e2
T ( O ,O) • -
NOTE : How to rotate contour to ge t Co ttingham formula ;
F
Use DGS representation for T
n( a , B )
I
v 2 --0_2+2Mv B -a+ i£
Call E • ../ cr+i-f s+Q2 and no te the s ingularities are at v • Q -i£ , -(J.+i£
v+BM • E -i£ , -E+ ie
S ince E > SM, the poles below the axis are all for v
y----- - > O. The contour goes like
the dot ted line , it can evidently be ro tated to the imaginary axis .
Le cture 44
Express ion for self ene rgy in terms of W only
We didn ' t s ucceed in representing 6i(- in terms of W (q2 , 0 ) fo r negat ive
q
2
only , wi thout at the same time involving ourse lves wi th ano the r unknown
2
function T ( q , O ) ; and furthe r each of two parts is infinite and it is hard to
guess at the di fference . Perhaps we should abandon th is and take a las t look
at j us t the 6M2 exp ressed in terms of W { q v ) for pos i tive and ne gative q2 •
2
, It
is given by
W ( v , Q ) dQdv (44 . 1)
F
T µv
. (J
(This is obtained by expressing TF in the form
K
IJV
v-v ' +iE

(Q , v ' )
_
K
HV
)E
(Q , v ' )
v+v ' - i
dv '
_
2w
+
and ins erting into (42 , 1) . W ( v , Q ) depends only on Q , the magni tude of Q; and
we have d 3 Q • 4wQ 2 dQ) .
To study the possible divergence of the integral at leas t , we go to large
2
v and in fact to the s caling region -q /2Mv x, hence Q v+Mx . We can write • s
(cons ider 2MW a function of x and v , 2MW(x , v ) )
t.M2 •
2
dv::
1
JJ
m
dx 2 MW ( v ,x) ;:: (44 , 2 )
0 -v /M
For large v the uppe r limit for x is kinematically 1 . 2MW approaches the
2
function f (x) for positive x in the s caling limit . Fo r negative x (posi tive q )
it approaches as we have seen -f (-x) ; note , however, that -x is not kinematically
limited to 1 .
Dynamical ly -x i s limi ted , W ( x , v) exists for x < -1 but for large v falls
J
rapidly until there is nothing left of o rder one .
m +l
• o.
I
The contribution of the scaling region gives dv f (x) dx
-1
But this only says that a divergence higher than logarithmic vanishes , a thing
Expanding the factor (v+tlx) I (2v+Mx) seems to give a term
� xf (x) ( dv /v) dx ;
we expect anyway .
Mx/2v leading to this is not all , we shall have to know to
2
order l /v how W differs from i ts Bjorken limi t for both positive and negative q •
Th is sums up the p roblem ; experiment can in p rinciple help us with the
2
approach to the limi t for q negative . But we shall have to rely on theory to
2
obtain the con trib utions to order l/v for positive q before we can decide
whether !!.� diverges according to present theories .
Other electromagnetic energies , Quark model

Having failed with fundamental theory to ge t information on electromagnetic
mas s di fferences , we now turn to much cruder pictures to discuss the poss ib le
relations o f !!.�
in di fferent terms of an SU or SU mul tiplet . We do it in
3 6
the language of the quark model although many of the resul ts come from weaker
assump tions , like s imple su 3 etc .
The proton , for example , in the low energy quark model is made of three
quarks , two u quarks and a d quark wi th total spin 1/2 wi th wave function
I P> • __l (2 uud - udu - duu) ( t H ) symme tri z ed •
16
(44 . 3)
The electromagnetic self energy can be thought of as being made of two

parts :
a) The self energy of the individual quarks . We suppose this is proportional
to the change squared of each , thus 4a : a: a for u: d: s respe ctively . To a
pro ton this contributes AM /2Mp • 9a (we shall normali z e all mass squared changes
2
2
to 2Mp of the proton as a s cale for measuring a , the t rue change in M is then
2Mp a . )
b) An interaction ene rgy be tween pairs whi ch we take as p roportional
to the produc t of ch arges . The interaction must depend on the mutual spin
relation of the p air . Thus write S ( l+y ) i f spin is parallel and S ( l-y) if
spin is antiparallel ; this is S ( l+yP) where P i s the spin exchange operator .
We mul tiply by -2 for ud or us , +4 for uu and +l for dd , ds , s d , ss pairs ,
call this factor xi "
j
The electromagne tic sel f mass operator can there fore be wri tten
2
AM • 2 Mp { 4 a ( No . of u ' s ) + a (No . of d ' s ) + a (No . of s ' s ) +
+ L pairs xij S (l+yP) } (44 . 4)
I t is easy t o get the expe ctation of this operator for every s tate . Thus
on the proton the operator Bx alone does no t see the spins and gives (2 (4-2 -2 ) uud -
( 4 -2 -2 ) udu - (4 -2 -2 ) duu ) t t + / /6 which is zero . For the neutron , change a u to d ,
the coe fficient i s - 3 s o there i s a contribution - 3S . Next we s t udy SyxP
xP (uudtH ) (4uud - 2 udu - 2duu) t t +

xP (udut H ) (-2 duu - 2uud + 4udu) t t +
xP (duut H ) (-2udu + 4duu - 2udu) t t +
there fore
xP (2uud - udu - duu) t t+ / /6 •
• ( (8+2 ) uud + (-4-4+2 ) udu + (-4+2-4) duu) t t + / /6

and
 • c 20 + & + & > = 6
Adding up the various contributions we have for the pro ton
AM2 /2Mp • 9a + 6 Sy
Lecture 45
Electromagne tic self mass , quark mo del (continued)
+
We can calculate the EM self masses for every particle of the 1/2 octet
as wss done in the previous lecture for the proton , the result is :
p • 9a + 6By
n • 6a - JS + JBy
+I: • 9a + 6 By
i:0 • 6a - JS - J By
I: • Ja + JS
:;: • Ja + JB
:; 0 • 6a - JS + JBy
A0 • 6a - JS + } By
Hence
tiM ti.il- / 2Mp
exp
.
p -n 3a + 3 5 + 3 By - 1 . 29 MeV - 1 . 21 MeV
-= :!: :!:
�o - .
3a - 6 B + lBy - 6.6 .7 - 6.7 .9
i: +- i: o 3a + 38 + ·i; By
=
- 3 . 06 - 3.64
oI: -I: - 3a - 6s - 23 �r
.
- 4 . 86 :!: .07 - 5 . 78
ci: +-i: - )-(p-n)-e 0.. :-) o • 0 :!: .7 + .s :!: .9
We have three constants for four maa s di fferences so we have the relation
(an su3 relation)
{I:+-I: -) - (p-n) - e 0- = ->· 0
which fits well . (We choose to use ti.il- for no extremely good reason , the relation
fits better with l!.M but there is li ttle to choose , i t is inside the experimental
error for ti.il- also . )
The values of the constants that we get are

3a • -2 .0 Me V
B { l+y ) .24 MeV parallel spins

B (l-y) • 1 . 32 MeV antiparallel spins
l! • . 78 MeV y • - . 69 MeV
The sign of a is the opposite of what you would expect but then you expect
+ � and it must be renormalized , possibly to a negative value . It means generally

parti cles with fewer u quarks are heavier , i . e . more pos i tively charged b aryons
are lighter . The sign of the 8 term from electrostatic repulsion is pos itive
as expected . We find at traction of paral lel magne ts in an s s tate but repuls ion
of antiparalle l ( the magnets are on top of each other) which is correc t ; the
net repuls ion in the parallel case is less .
Con tinuing in the 5 6 su6 multiple t to the decime t and suppos ing the cons tan ts
are the same we can predict eve rything:
8i /2Mp (predicted)
•
++
8 12a + 12 8 (1-y) - 5 . 12 MeV
8+ = I: + . 9a - 6 . 00
• ;
80 - E o o 6 a - 38 (l+y) - 4 . 72
8- - I: n
- -
. .: 3a + 38 (l+y ) - 1.28
2
8M /2Mp (pred . ) 8M(pred . ) 8M(exp . )
:!:
° ++
8 -8 + 0.4 + 0.3 2 .9 .9
:!:
++
8 --8 + 3.8 + 2.9 7 .9 6.8
- 0.9 - o.7
++
8 +- 8
i: - - E + + 4.7 + 3.2 3.3 :!: 1 .5
- - - _::: o + 3. 4 + 2.2 4.9 :!: 2. . 0
The expe rimental data is not good b ut there is a serious discrepancy

with a very recent experiment on 8 ° - 8++ ; except for this the signs and
gene ral order of sizes is good .
For pseudoscalar mesons , no ting that the sign o f the charges i s reversed
for antiparticles , and that only antisymmetric spin contributes we get
( call B' • B ( l-y ) )
11 t:..m2 /2Mp (exp . )

+
• Sa + 28 '
11 - 5a - 5 B I 11 -11
2 8' .64
0 + 0
2 • !
• 3a + 3S ' - 1 . 95
o
K+ • 5a + 2 B ' K+-K
Ko • 2a - S '
We have two constants to fit with two parameters and have no prediction .
•
•
We ge t S ' . 14 , 3a -2 . 38 again confirming that a is negative . In fact a is
close to i ts value for the baryons wh ich is what FKR'e relativistic quark model
would expect . Furthermore the 8 ' • 8 ( 1-y ) is nearly of the same order of
magni tude .
Finally we do the vec tor mesons . Here we need 8 ( 1 -y ) • b for paralle l spins
p •
+ S a + 2b
0
p Sa - 1_
2 b
s
(j)0 Sa - I b
*
K +• Sa + 2b
K*o• 2a - b
* *
The only data on 6M is that for K + - K o whi ch is -S . 7 ± 1 . 7 MeV o r
•
2
6m /2Mp -S . l ± l . S • 3a + 3b . I f 3a z -2 . 38 this gives 8 m - . 9 ± S , which is
very bad , it is o f the wrong s i gn . If the values for the octet baryons are
used we predict 6m2 /2M • ± 1 . 9 !
p
In this system we also have two o ther e f fects (discussed in Lecture lS)
a) the electromagne tic mixing matrix be tween p0 and w0 , wi th off diagonal term
p0 w0
)
3a - 3b /2 . b) The annihilation term between and which we found to be
0
p
0 �Sl
•
(j)0
6m MeV .
.17
Th e o f f diagonal element 6 i n the mass matrix
J
6m - if /2
p
.. _'., .
is determined by p ,w in terference to be -3 . 7 ± . 9 MeV . Subtracting the

annihilation term + . S l gives -4 . 2 ± .9 MeV . , for the contribution of the self
2
energy . 6m /2M then corresponds to 3a - 3b /2 • 3 . 7 ± . 7 . This s ugges ts if
p
3a • -2 MeV that b is in fact pos itive and near + 1 . 1 ± . 4 MeV , not too con-
sisten t wi th the baryon value of . 2 4 for parallel spins .
To summa r i ze :
SU6 does not , predi cting 6°
the mesons things are very poo r .
-
This i s generally 1D1success ful .
6
++
•
For the baryons su 3 works but
+ 0 . 3 for the experimental 2 . 9 ± . 9 . For
For pseudos calars w e ge t 3 a • -2 . 4 , 8' • .14
( compared to 1 . 3 for b aryons ) . For the vector mesons the si tuation is confused .
I f we use mass di fferences instead of 6m2 /2Mp fo r the rules , the baryon
constants come out 3a • -1 . 9 , S (l+y) 0 . 20 , 11 (1-y) • 0 . 9 8 . The predi cted
•
o ++
l - A is 0.8 so only l i t t le is gaine d here . But the me s on s i tua tion i s
a l t e re d (be caus e o f the small w mas s ) . The cons tants come o ut fo r t h e pseudo-
s c alars 3a • 7.0 , B' • 1.0. For the ve c t o r mesons we have 3a • -4 . 8 , b • -0 . 3 .

The B ( l+y) fo r ba ryons may d i f f e r from b f o r ve c t o r mes ons and B ( l -y ) for
ba ryons may also differ from 8' for pseudo s c alar me s ons be cause the s i ze of
the wave funct ion is so d i f fe rent s o the mean l/r d i f f e rs . The e l e c t r i c and
magne t i c in terac t ion need no t ch an ge in the same ra ti o so it is possible tha t
b is negat ive fo r me s ons and pos i tive fo r b a ryons , b u t it is di f f i cu l t to s ee
2
why S ' is so di f fe rent as the Am gives .
Why the value of Ja should d i f fe r in one case f rom the o ther is not
2
clear . Were it no t for expe riment l would dec ide 3a calcul a te d for Am / 2M is
p
the same f o r pseudos calar and pseudovec t o r mesons and is 0 . 6 o f that for
baryons . Th is is because we migh t gue s s that mass squared shifts due to

2
s t rangenes s come from a mass change in the s quark . The d i f f e rence in m
fo r the baryons is ab o u t . 40 , for the mesons abo u t . 2 4 or 0.6 as much . Thus
the s e l f energy correction e f f e c t - if it is to be associated s imply w i th a

p roper mas s change of the u quark will also be 0 . 6 as e f f e c tive in mes ons as
in ba ryons .
Eviden tly the naive theory does not work well , we do no t unders t an d
things s o we ll . mo re detailed dynami cal theory is ne ces s ary .

A But in any
case , mos t part i cu l arly the large A

0 A++ is mos t disquie t in g . -
Al • 2 ma s s d i f f e rences
There is one observation that can b e made here whi ch s ugges ts tha t dynamic
calculations might be possible fo r some comb inat ions . No tice that a is the
self energy term, po s s ib ly involving high frequency behavior , b u t S (lty) is due
to mutual interac t i on and ought to show no divergence . Cert ain comb inations
o f mass s q uared d i f ferences do not involve a . They a re
*
i (A + A- - A
o
-
+
A )
2 1: 0
+
i: + i:- -
All the s e involve the Al • 2 isospin part of the e l e c t romagne t i c self
energy . Thi s ene r gy depends on th e p ro duc ts o f two curren t ope ra tors JJ each
o f wh i ch con tains Al • 0 or 1, and can make terms whose parts are Al • O , 1 or 2 .

2 1
In terms of lz these go as con s t an t , Iz , and Iz - 3 l (I+l ) respec tive ly .
Th e cons tan t (61 •

0) part i s expe rimen tal ly los t in the s t rong interac tions ,
t h e 61 • l 1 to
z
or terms are measured by differences p ropo r t i ona l I like
z'
2
l: - E the Lil 2 term 1 and is
-
But to z
+
or p-n . R e f fect is p ro po r tion al
meas ured by the d i f fe ren ce s ( like !+ + E- - 2 E 0 ) men tioned above .
We are calculating the ll l z • 0 componen t o f t he Li l • 2 e f fe c t , o f cours e ,
b u t t o i s o la t e i t think o f cal culating an a r t i fi ci al ll lz • +2 th a t would ar i s e
electromagnetic J we re lll +1 Li l •
if cur ren t z • (ins tead of 6 I
z
• 0) , 1. We
mus t then th ink of calculating an e f f e c t of two c ur ren ts like J+ J+ .

Now we see why a does not aris e , and why in ge n e r al , poss ibly the integrals
in calcula ting this might converge rapidly at high v, provided partons are
quarks . For if all the current carrying fundamental operators (psrtons ) have
is o s p in 1 / 2 it is impossib le for two J+ to operate on th e s ame p a rt o n in
succession at h igh energy . Thus the virtual photon exchange diagrams like A
con tribute only to Al • 1 and are imposs ible for Lil • 2 , only B are allowed for
1\1 - 2
s t rong
interac tion
""""""' pho ton
- parton
Baryon
Large vir tual momen ta o f the photon is possible in A no ma t ter how so ft
the s t rong interactions are - i f the ini tial momen tum dis t r ibu tion of the
partons involves only s low ones so can the final s tate . But i n B, i f the
vir tual pho ton momen t um is ve ry h i gh it is un l ike ly that the s o f t s trong
in te ract ion can pull the partons back together to give much o f a di ago nal
ampli tude to b e in the undi s t urbed b aryon s t a te agai n .
Lecture 46
Further comments on elect romagnetic mas s differences

calculat ions of 61
As we foimd in the pre �ious lec ture , dynamic
• 2 mas s
di fferen ces should b e feas ib le by s ummi ng over

n ot many s tates and using al l
we know th e o re t i ca l ly and experime ntally &bou t the expected

behavio r of the
ne cessary matrix elemen ts , Of particular in teres t would be a study of the

+
� - 11° mass dif fe rence .
The elas tic term con t r ibu te s almo s t all o f the 11+ - 11° mass di f fe ren c e
as a s imple calculation will s how . The diagrams are
7·
��
11
The pion form factor is dominated by the p resonance , we th e re fo re include a
2
factor m / (q2 -m 2) at each pho t on co up ling so the mass di fference is
p p
m
4
(46 .1)
in m112 / mp 2 th e firs t t e rm i n the square b racke ts is 1 , this gives
j
To o rde r zero
d q 1
4
2
4
p
(Am ) 0 • 4 2
m
2
-411e 3 --
(211 ) q
The integral is re adily performed remembering th at in four dimensions d 4q • 112q2 dq 2

after integrating over angles . The result of the zero th order calcula tion is
2 2
(Am2 ) • 3e m /411 , do i ng the next order gives
0 p
2
m 2 m
+ --\- .l.n �
m
(46 .2)
m
p 1T
2
In terms o f Am • Am /2m this i s 11 + - 11 ° • 4 . 1 MeV as comp a re d t o the experi-
mental 4 . 6 MeV . Estima t es o f con trib utions of h i ghe r intermediate s t a tes
(which could be done using FKR ' s model) should be small .

+
A similar result is found for E + E- - 2E0 • It seems as though the
elas tic te rm al ready gives almos t the comple te resul t . E s t imat e s of hi gher
resonances give les s than 2 0% , and the result is in good agreement with the
dat a .
The contributions of t he e l as tic term to self energy differences of

baryons were calculated by Gross and Pagels , Phys . Rev . fil 1381 (1968) us ing
su 3for the magnetic momen ts , and GE and � varying like (l + q2 / . 71) -2 •
They ge t
t.M (elast ic) t.M(exp . )
p -n + 0 . 79 MeV - 1 .29 MeV
E+ - t 0 + 0 . 16 - 3 . 06
Eo - E - - 0 . 88 - 4 . 86
::: 0 - :: - 1 . 10 - 6 . 5 :t . 7
(Most o f this i s due t o the charge , since the re are factors o f q from Yµ4-ilYµ
in the magne tic part . ) Note that E + + E - - 2E 0 • +1 .54 MeV (elas tic) compared
to +1 . 8 experimental , as we expec ted this s hould be dominated by the elas t i c
term . (Decimet intermediate states give less than .1 and presumably higher
ones even less . )
Note that o the r combinations (o ther than t.I • 2) do not involve "a"
( see lectures 44 and 45 , "a" i s the EM self energy o f a quark) such as (p-n)
- ( !! 0 - :=: - ) • In this case one canno t show that these differences also involve
a product of currents s uch that each current does no t act on the same quark .
But the fact that "a" is no t involved suggests that we may be able to ge t at
this difference also by estimating ma trix elements to various known s tates
us ing , for example , the quark model . Howeve r , the e lastic term fails utterly
this time t.m(elastic) • +1 . 9 , t.Mexp . • 4 . 2 :t .7. Why? To explain this we
look at the contributions from the s caling region . Let u (x) , u (x) , d (x) , d (x) ,
-
vW Ix f (x) for p , n , :=: o and
2
s (x) , s (x) be as des c ribed in lecture 31 , then •
is
-
•
4
9 (u + u) + 91 (d + -d) + 1
9 (s + s-)
'+-
•
1
9 ( u +
-
u) + 94 (d + -d) 1 (s
9 + s- )
..
_o
- 9 u) + 94 a)
l l (s s- )
fe :: (u + (d + + 9 +
-
•
l
9 (u + u) + 91 (d
- + 94
+ d) (s
-
+ s) (46 . 3)
(The neutron is ob tained from the pro ton by replacing u by d and vice v.ersa .
The :=: 0 is like the neutron but s replaces u and vi ce vers a . The :=: i s like the
:0 but u replaces d and vice versa . ) Therefore the s caling function for (p-n) -
0 - l - - -
( : - : ) is 3 ( u + u + s + s - 2d - 2d) which is no t necessa rily zero so
high frequencies can come in . In the model of valence quarks plus sea it is
zero b u t we do no t believe th is to be like ly .
Fo r further de t ails on all these ma t ters of EM self energy see an article
by W . N , Co t t ingham in "Hadronic In teractions o f Electrons and Pro tons " , Cummings
and Osborn Ed . Academi c Pres s , N . Y . 19 71 .
Le cture 4 7
Comp ton e ffe c t y p + yp or yn + yn
We now go on to dis cuss o ther e f fects involving two pho ton couplings .
The Comp t on e f fe c t is the mos t closely related to wh at we have done . If the
s cat te ring is exac tly in the forward direct ion the s ca t tering ampl i tude is given
o.
2 2
)J V
by T ( q , v ) fo r q • We p revio usly mean t the average ove r pro ton s p ins ,
thus T is the spin averaged forward s c a t terin g , we could also meas ure for
special spin direct ions of the p ro ton . The imaginary part of the forward
scatte ring is , o f course , the to tal cross sect ion o or o wh i ch we have

yp yn
discussed b e fo re . (E . g . a showed resonances at low v , a fall o f f pe rhaps
YP
like (9 7 + 6 7 / 1\i) )Jb , and ( 9 7 + 43 / 1\i/IJ'O) for neu t rons ,
The di ffe ren tial cross s e c t ion can be fitted w i th
!!2. • e
At
dt
-2
For energies from 2 to 7 GeV we ge t around 6 Gev • For energies from 8 t o
-2 2
16 GeV , A • 8 Gev is clos e r ; the re i s some s ign o f a quadratic te rm At + B t
wi th A • 7 . 4 , B • 2 . 0 . (This i s much like hadron di ffract ion s ca t te ring , e . g .
wp at 9 GeV has A • 9, B • 2.5.) Thus pho ton di f fraction looks very much the
s ame as wo ul d be expected for hadrons except for the very much smaller cross
sect ion , of course .
We now dis cuss the forward s c a t tering in mo re detail including spin e f fe cts .
The forward ampl i t ude may be wri t ten
(47 . 1)
as a spin ma t rix operating be tween spin s t ates of the p ro ton in the lab , system ,
The var io us meas ured quan t i t ies are expressed in terms of f and f as fol lows :
1 2
The total cros s section is the imaginary part of the diagonal (in spin)
scattering
Im f 1 (v) v to t
'\N ( 47 . 2)
4 TT
This is known . The forward differential cross sect ion for unpolarized forward
scattering is
1
( 4 7 . 3)
16TT
The real part of f 1 can be ob tained from the imaginary part by a dispersion
relation , (Eq . (40 . 6a ) fo r q2 • O) where we use the fact that f 1 (o) -e 2 /M . �
e2
Re f 1 (v) ( 47 . 4 )
M
Th is has been evaluated (see Damashek and Gilman, Phy s . Rev . Dl 131 9 ( 19 70 )
or Buschhorn e t al . Phys . Le tt . 33B 241 ( 19 70) and the sum of the firs t two
terms of ( 4 7 . 3) is compared to the experimental (do /dt) to see how b ig the
0
las t term is . They agree wi thin errors for v from 2 . 5 to 17 GeV so the n l f j 2 /v
2
contribution is less than 10% over the entire range . (The con tribution is
greates t from the first term in (47 . 3) above about 5 GeV . The se cond is 15%
at 2 GeV and falls away at higher v) .
e2 �
O , f 2 (v) • � µ 2 v where µ is the
We also know , for small v , as v -
2M A A
anomalous part of the magnetic moment of the nucleon in nuclear magnetons .
At finite angles the asymme try parameter has been meas ured
E • --- (47 . 5 )
OJ_ Oii
+
where o1 and 011 are differential cross sections at fixed t for incoming
photons polari zed perpendicular or parallel to the plane of collis ion , For
t • 0 E mus t be zero , of course ; but wi thin the limits of experimental error
( :!:10% for -t < . 2 , ±20% up to -t • . 6) it is zero up to t • -0 . 6 . (The

average of E for t = ,l to . 7 is . 02 � . 06 , )
We have discussed the size o f do/dt for yp in relation to (VDM) p p cross
section (see lecture 20) , it is twice larger than VDM expects . The asymme try
produces no problem for the corresponding asymme try in the p case , i t is also
ve ry small . This is no t unexpected , s channel helicity conservation also
expects the ssme result . The q uestion is : With incident light in the z direction
with x polarization do more pho tons scatter at a small angle e � Q (transverse) /v
in the direction x or in the direction y ? From the poin t of view o f diffraction ,
currents generated by the incident wave must be adequate to produce the correct
forward scat tered wave to interfere with the incident wave to accoun t for the
loss of intens ity of this wave represented by the total cross section . These
currents are obviously limited to the spatial region of the pro ton , and so they
produce scattered waves in other directions , the usual eAt of diffraction from
the pro ton , j us t as in hadronic collisions . B ut these currents mus t make pure
x polarization at leas t in the forward direction , they are x directed currents .
In o ther direct ions at small angles we have the same intensi ty for x and y
deflection excep t for a COB 0 l ab "" 1 - 0.2 lab projection fo r X de flection ,
.
therefore i: ( 1 - cos > I 8iab . > e � ab _ 1 2 :::o -t/ 2 } "" + .03 for
= ( 1 + cos
l\ab . =
-t • 0 .6 , v � 3 . 5 where the data is tslten . Tilus we expect small i: , if any ,

close enough to zero to not be in disagreement with experimen t wi thin i ts errors .
To summa rize , the Compton s cat terin g as a function of t above 2 GeV shows
no surprises other than what we can expect from diffrac tion from the known total
pho ton absorption cross section .
Below 2 GeV , therefore in the resonance region , there is no data , But it
should be possible to make a pret ty good theory of angular distrib ution
and energy variation by conside ring a success ion of a -channel resonances
(many of the matrix e lements of which are known from the study of yp + wp
in the same energy region , unknown ones may be guessed from the quark model) ,
There is also a computable neutral pion exchange term
p x
The coupling o f two photons to a neutral pion is
p
known from the w0 + 2y decay .
All these calculations can be checked and controlled by fit ting the calculated
imaginary part o f the scattering to the nicely meas ured to tal cross section
o yN which shows the expected resonance bumps in this region o f energy .
tot
Compton effect for very small Q, v
S cattering of very low Q, v is like the scattering of radio waves , or

(if q 2 � 0) it depends on the reaction of a particle to nearly cons tant electric
and magnetic fields , This is , of course , given by two experimental constants ,

(obtained by measuremen ts in such fields ) the charge and the magne tic moment
(res trict to spin 1 /2 case) . Therefore we expect the Compton e f fe c t for low
enough v, Q to be given en tirely in terms o f these cons tants . The particles
should act exactly as if they were point p articles . One can compute the effect
semiclassically o r from the non-relativis tic approximation to the Schroedinger
equation with spin (Pauli equa tion) o r again by diagrams ass uming a pure
particle with no internal excited s ta tes . Such a term is cal led a Born term.
Thus we wri te T v (not averaged over spin directions o f the p ro ton ) as a S IDll
µ
from the Born term and the res t f rom o ther diagrams
T (47 . 6 )
µv
The reason T B dominates at low v is that i t has an energy denominator

due t o the intermediate state A of s i zev -E ) s o i f A i s also a proton
(M +
Ap
E is Mp (as Q O) + and we have a leading l/v factor which doesn ' t appear in
A
the remaining terms TR .
As you expect the matrix e lement of the charge densi ty J • P at Q • 0
t
< x i p (Q) I p> •
2
is· the total charge and is diagonal thus orde r Q if x is not
the proton s tate . To show that the off diagonal ma trix e lemen ts of o ther
components of also go to zero we look at q < x i J I p > 0 ( charge conser-

J •
µ µ µ
vation) ; so v < x l p l p> • Q .< x l l Co) l p> , hence if v , Q go to zero together we
see that j matrix elements go to zero . A mo re rigorous (but harder to interpre t)
argument is given below .
We now compute the limit of TB as v , Q + O. If
A
µ is the anomalous
µA
moment in nuclear magnetons , the coupling of a photon is Y µ + 'iii (y 4 - 4Y µ ) so
µ
T v
B • (y +
µA
(y 4 - id.y ) ) ; it
II
+ +
� (y
v
+
µA
(y 11 4 - id.Y11 ) +
µ µ 2M µ 2p • q + q 2M
(47 . 7)
For small q and q 2 • 0 this is eas ily worked out to be

2
_ L
M
R
while the contribution from T s tarts as v 2 and thus we have as v + 0
£ 1 (o) 2
• - e /M
2 2 2
f 2 (v ) /v • f i (o) - e µ /211
• (47 . 8)
A
R
To show how T is small more formally (at least for the f 1 term) note
B
tha t T µ v total as well as T each separately sati s fy the gauge condi tion q µ T µ v O , a
R
h ence we mus t h ave q µ T µ R • O Now we can wri te T µv ( fo r the symmetric spin
•
v
averaged case at least) as a power series Tµ v ap µ p v + b (p µ q v + p v q µ ) +
a
2
c 8 IJV + .order q • There mus t be no po les like l/p • q in a , b , c (wilike T )JV
total
)J
B R
which has such poles coming from T ) . Now qµ T 0 requires •
µv
2 3
(p • q) P µ a + (p . q ) q µ b + q P µ b + qµ c + order q 0 •
We cannot solve this by a • -bq 2 /p • q because no l/p . q terms are allowed .

C learly c • - p · qb , b • - p - qa/q2 with a • aq 2 are the only possibilities ,
therefore a � a q2 , c � a (p • q ) 2 and the term s tarts out as second
b � - a (p • q) ,
R
order in q . The wisymme tric non-spin s umme d T can also be shown to be of
)JV
the same order and the re fo re T Co

mp ton • T B t o order 1 and v . (For a comp le te
)J V IJV
discussion see Low , Phys . Rev . 96 1428 ( 1 954) and Gell-Mann , Phys . Rev. 2! 14 33
(1954) . )
Forward Compton scattering from non-relativis t4.c Schroedinger equation
The equation is (with first order relativistic corrections )
Hijl •
[ l {p - eA) • {p - eA)
-
+ -t + -t
- --..,.
1
.. (p+ • + e {l + µ + • ±
p)2 - - A) 0 11 +
2M BM"' 2M
+ �2 {l + 2 µA ) (V ·E + zt. (p - eA) xE � 1j1 • E ip (4 7 . 9 )
+ +
The incident amplitude is -tA • e i , ±� • ive+ i and ±11 • i +k x e i . In the laboratory
p is zero , hence the leading term comes from the A·A term and is - (e 2 / 2M):i . :f
contributing to £ 1 • Next we have the term t . 8 operating in second order , two
diagrama with energy denominators -v and +v
Finally the last term in ( 4 7 . 9 ) with the term � · A x E gives
2 2
which combines with the previous term to change the (1 + µ )
A
to µ
A
we
therefore have
2
•
e .,.
Amp . ei
2M
Other Two - Current Effects
Lec ture 4 8
O ther quan tities involving T

µv
Another exper imental quan t i ty tha t involves our func tion T • 
µv µ
(no t ave raged over the spin o f the prot on , i t involves the ant i symme tric part of T v)
µ
is the hyper f ine splitting ene rgy in hydrogen respons ib le for the 1420-megacycle
l ine . I t is the d i f fe rence in energy in the ground s s tate o f a tomic hydrogen
depending on whe ther the spins of the elec tron and p ro ton are paral lel o r anti-
paralle l . I n non-relativistic approximation i t depends on the prob ability that

2
the electron is on top of the pro ton j � (o ) j in the ground s tate wave function .
Re lat ivis tically we can wri te ( Rm • Rydberg , µ ' µ magne tic momen t s of p and
p e
e, µ • Bohr magne ton)
0
2
+m
M')
-3 3
U + y a > t ""R -P ( 48 . 1)
-3
The factor ( 1 + m/M) comes f rom reduced mass co rrec tions to the Schroedinger
2 2
equation in ge tting j � (o) j ; 1 + 3a /2 is a modi f icat ion due to the Dirac
equation . The other factors t, -� 1' are all near one and are due to higher
o rder quantum electrodynamic corrections . They have been separated into three
factors for convenience of discuas ion . f collll!S from QED mo difica tions of the
221
mo tion of the electron , diagrams like A below .
A B
� comes from the proton recoil diagrams of type B in which the fact that the
pro ton is not a point charge is included by using measured form fac tors . 'f' comes
from two pho ton exchange terms of the form C , D .
All these factors except 'P which depends on as yet unknown properties
of the proton have been calculated to very high accuracy - further 6 E is
measured to absurdly high accuracy . The cons tants like a, µp etc , are now
known well enough to determine 'f' to about four parts per million . Theo
retically , the deviation due to 1" is about this same order of magnitude .
As long as ';" remains uncertain we cannot use these accurate measurements and
calculations to improve our knowledge of the cons tants - or to put the prob lem
the other way if more accurate values of the other cons tants become available
that will tell us"P, an electromagnetic property of the pro ton - and we would
be challenged to calculate it , write -P • 1 + 6 . Let us see what i s involve d .
Obviously the proton coupling is to two currents . We separate the cases
that the photons are of low energy and momentum from the cases where they are
high . Where they are low , b inding of the electron in initial and intermediate
s tates etc . mus t be considered - but here for low v, Q, the proton acts , as we
have seen , like a particle of charge and magnetic moment and thus we can do
this part of the calculation . For definiteness we do it putting in the experi -
mental form factors once for each photon and integrate over all moments , call
this 61 • Write 6 • 6 1 + 6 2 • Now for high virtual momentum electron binding ,
even the electron mass can be neglected and we can imagine the electron and
p roton to be free and at rest before and after the scattering - thus our two
current operator for proton in and out of same momentum (res t ) is
all that is involve d . Naturally s tates x other than pure proton Born states
are involved . Of course we have already counted some high momentum contri-
butions in 6 1 and we shall have to sub tract them, thus we shall have not j us t
elas tic is calculated from the Born
T but rather T - Tµvelastic where T
w w w
term alone with form factors . Call this T µv ' • T µv - Tµvelas tic . Then 6 2 is
proportional· to diagrams c , D with electron at res t initially and finally .
If Tµv ' is a y matrix on the proton spin it depends on the spin flip amplitude
for the proton - thus on the antisymme trical part of T v ' (proton at res t ) .
µ
"-f �)
62 2 T r (C µ v
:h=t:
f (l+y t ) y zy 5 ) T r C\ v ' f (l+y t ) y z y 5 ) (48 . 2 )
)
2 2
The factor l / (q ) is for the two pho ton propagators . The first trace
is the Compton scattering by the electron
c
• y \)
l l
µv yµ + yµ
� + 4- m � - 4
neglecting the momentum p and mass o f the electron . The second factor is the
coupling of two photons to the pro ton , which we do not know b ut which we are
discuss ing ; we need its antisymmetric part . In Lecture 33 we wro te its
imaginary part wµv in the form
2
where G 1 and c 2 are functions of q and v defining the imaginary part . Let
the complete scattering functions of which c and c 2 are the imaginary part
1
be called s 1 and s 2 • That is , we write a form for T j us t like the above
µv
for W excep t s , s 2 replace G , G 2; and c , G are Im s , Im s • Then by
µv 1 1 1 2 1 2
subs titution we can express 6 2 directly in terms of s , s • One gets ( see
1 2
Drell and Sullivan , Phys . Rev . 154 1477 (1 9 6 7 ) and C .N . Iddings , Phys . Rev .
1388 446 (1 9 65 ) ) •
... 2 .. J d4
� )
� 3 lc 2 q2
�
- }) sl cl , v > 3
+ �
2
q } s2 cl , vj�
Now we can do many things with this to try to estimate it or compare

it to things measurable in principle in ways entirely analogous to our dis-
cuss ion of electromagnetic self energy . For an example , we can use Cottingham ' s
idea of rotating the contour on the dq 0 integral from the real to the imaginary
axes , thus the integrals depend only on s 1 , s 2 in the negative q 2 region .
Finally we can express these complete s 1 , s 2 in terms of G 1 , c 2 ( their imaginary
parts) by a dispersion relation , and thus express A 2 in terms of G 1 , G2 •
Two questions come up :
First , are these unknown functions , like T 1 ( o , q2 ) necessary in the dis-
persion relation or are they as for r2 without cons tants ? ( The answer to this
ques tion is almost certainly known . ) We can guess because we know the asymr totic
scaling behavior of c 1 , G 2 • If there are sub tractions , this me thod is frus trated .
Second , supposing there are no undetermined functions in the dispersion
relation and 6 2 can be entirely expressed in terms of G 1 , G 2 in the experimental
region ; what can we do until c 1 , G2 is directly measured? It becomes a research
problem to guess as completely as possible to see in what ranges of -q 2 , v the
6 2 is most sensi tive and use whatever models or ideas are mo st reliable there .
We could try to incorporate all that is known of low energy theorems and integrals
(like ! g2 dx 0 ) to limit the pos sibilities . At worst , certain limi ts of un
•
certainty can be estab lished since , for example , 2g , which is the dif ference
of up and down spin partons h+ (x) - h_ (x) (see Lec ture 33) cannot exceed the
sum h+ (x) + h (x) which is f (x) and is measured . Generally posi tivity of the
_
trace of Im Tµ v on any diagonal s tate limi t the size of G 1 , c2 in terms of w1 ,
w2 in some way . This problem should be pursued
And if the answer to the first ques tion is yes and there are unknown
functions brought in in the fixed q2 dispersion relations ? Then Cot tingham 1 s
scheme does not work and we shall have to use other methods o f analysis such
as the fixed Q dispersion relations to ob tain expressions on which we can apply
our partial physical understanding of pho ton hadron interactions - partons ,
scaling, quark mode l , etc . , - in conj unction with as many sum rules e tc . , that
we know (as wel l as possibly even the numerical value of the n-p mass differences)
to guide us as much as possible to calculate this quantity 6 2 as accurately as
possible , and with an honest estimate of the possible theoretical limits of
uncertainty .
Lecture 4 9
Other two-current effects
I should j us t like to add a few miscellaneous remarks o f situations in
which the double operation of two currents is involved, namely , in the dis-
integration of pseudoscalar mesons . They all present interesting questions
for study ; this is mean t merely as an introduction . Of course matrix elements
of one J are involved in s ingle photon decays like "' ... 11y and we have already
µ
discussed them .
Two J ' s are obviously involved in two pho ton decays like 11° -> 2 y or
n° ... 2y . An honest calculation of either one o f these would be very interes ting
(one in which the validity o f the assump tions is backed up by considerably more
than the mere fact that the answer agrees with this one experiment) . What does
su 3 say? •
...!. (uii - dd ) , n° • ...!. (uii + d d - 2 sii) to get
Use the quark model 11°
12 16
the ratio o f amplitudes amp 11°/amp n° - ...!. ci - !-) /...!. ci + !. - 2!.) - 13. I t says
12 9 9 16 9 9 9
that the amplitude for 11 is /3 that for n , so the rate should be three times as
high . Experimen tally the widths are r ( 11° ... 2 y) • 7.2 t 1 . 0 eV , and the partial
width n ... 2y is r (n ... 2y) • 1.0 t .3 KeV . The ratio is 1/140 ins tead of l!
O f course , the reason for the abject failure is the very large mass
difference of w and n (mII /m � • 1/15 ) and we mus t be much more careful of
2
n
it. This is where su 3 is indefinite - and no universal way of doing this is

known . First there is phase space : the general formula for dis integration o f
a n object of mass m at res t into two particles each of whose mo111en tum is Q i s
where M i s the relativistic ma t r ix element . In our case ( 2 y ) Q • m/ 2 s o phase

space (i . e . if we assume M is given by su 3 ) works against the n and the dis-
2 /M 2 • 3/ 1 ) .
n 11
crepancy is another factor o f m /m worse (i . e .
11
1/ 540 instead of
n
M
More sens ibly we should write M in its simplest relativis tic form which
for a pseudoscalar meson di sintegrating into 2y ' s of polarization e 1 , e ,
2
momenta k 1 , k2 is M a E vap e1µ e v k
•
µ 2 10 �9 •
Now the guess would be that a is determined by su 3 • This means M goes
mll 3
as Q2 or m2 Rate / Rate • 2 , a 2 /a 2 . _l_ This at leas t moves in
II n
.•.
n m 140
n
the righ t dire ction , it gives
2 2
11 11
a /a - 0.4 ins tead of 3 0 . •
This is not bad but (a) what of the remaining discrepancy and (b) why
2
should a be given by su
3 and not say a/m or a/m ? Eviden tly we have b een doing
too much comparison to experiment and too lit tle thinking . Can we reason out
from other things we know j ust how the 11 rate and 11 rate should be compared , or
how either migh t be calculated or estimated absolutely ?
With regard t o the mechanism o f the decay ever since it was s ugges ted
by Gell-Mann , Sharp and Wagner , Phys . Rev . Le tters ! 261 (1962) it is supposed
to be dominated by the d iagram going through an in termediate p or w meson like
y 11
Thi s connects it to the 11PY coupling cons tant , or through su

y 3 to the pseudoscalar
vector photon coup ling cons tants in general ; determined directly for example
by the w + 11y rate .
Again 11 + 11 11y is interpre ted in the same way as
The p 11 11 coupling being known one can compare this wi th 11 + yy and p redict
the ratio r (11 + 1111y ) / r (11 + yy) with good s ucces s . Ge ll-Mann et al . go t . 25 ,
experiment is . 12 (Gormley , Phys . Rev . 2D 501 (1970) ) . (For two calculatiOns
corresponding to di fferen t choices of how the coupling cons tants symmetry rules
depend on the masses of the states see Brown , Muncek and Singler , Phys . Rev .
Le tters 21 707 (1968) and Chan , Clavelli and Torgerson , Phys . Rev . 185 1 754
(1969 ) . The choice made in the latter paper fi ts very nice . ) This is inter
esting for at first sight the numerator is order o and denominator is order o2
so the o rder should be 137 but here many numerical cons tants accumulate to
overwhelm this factor and make the ratio nearly 1000 times smaller . For this
picture there is a factor in the matrix element for 11 + 1111y like l/ (m

2 2
- m1111 )
p
2 2
11 11
where m is the invariant mass squared (p + p - ) of the four-vecto r sum
+
of the two pions . This distorts the s pe c trum away from the s imples t form toward
2
larger probab ili ties for larger m The experimen ts (Gormley) are so
1111
accurate to see this e ffect , even quantitatively , so there can be na doub t of
the mechanism in this case . It is likely therefore that no deep mys teries lie
in the 2y disintegrations eithe r .
However the n + 311 does present a challenge . The G pari ty of the 11 ' s
i s - , of the n is + s o the disin tegrat ion i s not allowed , s t rongly . Cons ider
+
n + 11°11 11 - , charge conj ugation (which we think is surely satisfied for decays
at this rate) (n and 11° are charge conj ugation + because they can go into 2y )
+
requires that the 11 11 mus t be symmetri c , they mus t be in an I • 0 or 2 s tate .
This added to the I • 1 o f the third meson yields only total I • 1 , 1 , 2 , 3 as
pos s ibilities , there is no I = O. Therefore the decay cannot o ccur excep t by
violation of i so topic spin . Isospin is , of course , violated by e lectrodynamics
so an electrodynamic virtual effect (orde r a in matrix e lemen t ) is involved and

2
the rate is o f order a so n 2y can compete wi th it rather than being completely
swamped by what it would be if n + 311 were a s t rong interaction . (The data on
n decays is
Branching ratio %
- 0
n 11 11
+
11 23.1 ± 1.1
...
°
311 30 . 3 ± 1 . 1
... 11 11 -y
+
4.7 ± 0.2
... yy 38 . 6 ± 1
0
... 11 yy 3.3 :!: 1
r total (n ) - 2 . 70 :!: . 6 7 KeV)
The change in I spin can be AI • 0 , 1 , 2 but only the AI • 1 part violates
G parity . Hence the final s t ate of three pions mus t have I • 1 . From this we
+ +
+ - °
can ge t an es timate of the ratio r (n 311° ) / r (n 11 11 11 ) . Combining three
states of isospin one we can ge t for the to tal I spin :
symme tric 3, 1
skew symme tric 2, 2, 1, 1
antisymme tric 0
If we suppose the 311 are in their lowes t space wave , s wave and symmetri c ,
s ince they are Bose w e will have I spin 3 o r l ; EM permits only the 1 (but
by proper use of space s tates the skew-symme tric I • 1 s tates could come in) .
Fo r t h i s s t ate r (n + Jn° ) / f (n + n+n-n° ) z 3/2 . If th e re is s ome skew symme tric
(momen t um dependen t ) space s tate the o ther I = 1 can come in reducing the 3 / 2 ;
s ince it comes in wi th h i gher angular momen tum it is prob ab ly smaller so the
ratio may b e fairly close to 3 / 2 . I t is 1 . 3 experimen t a l ly .
How do we cal culate the rate n + Jn ? Where does the in termediate photon
ac t ; can we gue ss wh i ch int e rme diate s t ates are mos t l ikely to be impo r t an t ?
A quan t i t ave e s t imate fo r this rate i s a p rob lem that n o dis cus s ion o f the
e f fe c t s o f virtual s e l f ene r gy photon a c t ion can fail to men t i on .

Hypotheses in the Parton Model
Lecture 50
HYfo theses in the Parton Model
We should now like to discuss what we can say about what the products
X would look like in deep inelas tic scattering e + p + e + X. lb.ere are some
measurements for certain definite final s tates X for small energies and low
q 2 of the virtual photon (see Berkelman , 1 9 71 Cornell conference ) . Mos t
of these can be unde rstood from direct extensions o f our theories for photon
(q2 • 0 ) reactions . We have already dis cussed pion production from virtual
pion exchange as yie lding information on the pion form factor . In addi tion
p production has been s t udied from virtual photons , with no s urprises - VDM
gives a fair account - see our theory discus sed in Lectures 16 to 2 1 where we
merely have to replace k2out by q 2 in the equations there yielding a fac tor
m 2 / (m ·2 -q 2 ) (q2 is negative ) relative to the q 2 • 0 case .
p p
It is necessary also to make an assumption of how the longitudinally
po larized photon (possible when q 2 � O) couples in relation to the longi tudinally
polarized p - it is assumed that these ampli tudes are re lated by a factor
m 2
e the extra factor q2 /m 2 is an ansatz made because gauge
m 2 q
2 p
-
p
invariance requires that J longitudinal vanishes as q 2 + O. lb.is may be valid
for no t too large q 2 , but of course if q2 becomes really very large the
229
•
asamp tions about small p + p product ion may begin to fail .
2 2
What happens at large q , and how it ties on to small q , in every case
for very large v will be our present concern . We shall have to be guided by
theory and l will take this opportllll i ty to review the parton model and some
assumptions that can be made about i t . We shall lis t assmptions that we can
make - without today being sure of wh i ch are righ t and which wrong - jus t to
see what their consequences are wi th the hope tha t experiment may la ter make
the selection {e . g . are charged partons quarks ? } , The re fore in our list
some ass umptions will (perhaps } be incons is tent with o thers . The assumptions
will come from a mixture of theoretical guesses and known experimental
facts - so one migh t be warned that if a particular assumpt ion neatly exp lains
some experimental fac t it may not really be a significan t confirmation for the
assumption migh t have been made with that fact in mind . Finally , little
effort will be made to derive one assmption from another - they are ce rtainly
not independent . This will there fore unfortunately not be s mathematical and
sound sys tem, but rather a lengthy "intui tive" or physical dis cussion .
Fo r a good discuss ion along the same lines see J . Bj o rken ' s pape r in
the 1971 Cornell Conference .
General Framework
We s uppose as in field theory , a wave function for a state can be given
by giving the amplitude for finding various numbers of field quan ta , or partons
of various momenta. In particular we dis cuss the wave functions o f s ingle
particles ( sometimes two parti cles in collision) wi th extremely large momentum
P in the z direction (P + �> . The wave func tion is being des cribed in Fock
space giving for a state the amplitude
etc.
where � i s the ampli tude t o find no par tons (usually zero } ; � (p } is the
o 1 1
ampli tude there is one parton (of s uch and such a type , an index we are omit ting)
p2
which has moment um p 1 ; � is the amplitude there are two partons of momenta
2
pl ' etc . This can be written in other ways . For examp le let j VAC>
*
represent the vacuum s tate and ap the operator creating a parton of momentum
*
I p1 I P1P2
p . Then we can write the wave function s tate I �> F j VAC> where =
p* • �o · l +
;
�l (p ) a +
l l
f
is s ome function of the creation operators .
Then we make the fol lowing assumptions :
Al . The amp li tude to find a large P� on any parton falls rapidly with lj_
such that e ffectively we can in first app roximation cons ider all P1 are finite
(as P + "' ).
A2 . The "wave func tion" for longitudinal momentum o f order P , i . e . ft�
depends only on x .
This requires some complicating remarks to make its definition clear
+ "'
•
for there is , as P an ever increas ing contribution to � for small x .
Pn ·
More precisely consider the dens ity matrix. • • •
Let �m (p 1 p 2 pn ) be the amp litude
to find n partons of momenta p 1 to Then for example the dens i ty for one at
L.f1 �n(P 1• P2 • · .. ·· Pn)j 2 2

k is
6 (k-p ) � d 3p • p (k)
i i
n i
That this depends only on k1 and x • k /p when x is finite as P + is the "'
L
L J �: (p 1 , p 2 , . . k ·P
assumption we wan t to make - with all i ts generalizations . E . g . the one
particle densi ty mat rix (like • •
pn ) �n (p 1p 2 . . k ' • ·
n)
p (k , k ' ) depends only on k1 ,kL , k /P;k � /P etc . , when x is fini te .
n
• • • •
I 6 (p i -k1 ) 6 (pj -k2 ) )

dp 1 dpn L
Again the two particle dens ity ( the expectation of
ij
behaves likewise e tc . I t is almo st but not quite the same as saying the
wave function in p i ' x i P Li /P , has a definite limi t . The wave func tion
•
.L
is a function of all of the momenta including those of finite momentum
(which we call wees ) · The s caling doesn ' t work for those - in fact the
mean number of particles rises with P so the wave function �n for any fixed
n falls with P ( like a power o f P) but the "relative wave function" e . g . , a
ratio in which on ly one part icle (or a finite number) is moved depends for
fini te x , p only on x, p •
� 1
A3 . In the wave function the ampli tude to find fini te longitudinal
momentum particles remains finite as P + m. That is to say again that the

dens ity matrix , e . g . the density for finding particles with finite (e . g . < +4 GeV)
values of P L have a de fini te limiting behavio r as P + m, and the expected
numbe r of such wee partons is fini te .
Lecture 5 1
Hypotheses in the Parton Model (continued)
A4 . To have continui ty between Al and A2 the mean number of partons o f
a given type f o r small x goes as dx/x and the wees go as dp � as P + m.
z
The number for p negative falls off rapidly so that for p • xP , finite x
z 2
negative there are no partona ; although for P z negative and finite there are
some (a fixed amount falling rapidly with negative p z ) .
(As a trivial example of the kind of behavior envisaged in the wee region
* *
consider a wave function like exp (E Ck 8k > l o> where ak creates a particle of
longitudinal momentum k and � varies as Ck • a/ (w-k ) w
3/2 with w • lfk 2+ 1 say ,
and a • constant . )
Ai. The behavior o f the wees i s nearly (as P + m completely) independent
of the distribution of the fast ( finite x) partona . Thia again is complicate d .
If we s t retch out the variable p z by , f or examp 1 e , de fi ni ng Y f.n ( • 'vP/ zZ + 1 GeV
2
+ p z ) ao for finite P z • y is f ini te ; for fini te x , y is 1n2P + £nx .
We have particles at every y from finite to f.n2P .
0 1n2P
We have drawn a graph of the mean number of particles in dy . I f we
look near finite y , P + m we see the behavior of the wees ; i f we look at
finite 1n2P-y (finite x) we see the behavior of the s caling fas t ones ; in
between is a plateau with a finite densi ty of partons so the mean number of
partons rises as 1nP .
It is easy to understand the densi ty but how do we understand the wave
function ? This gives the amplitude for every configuration , which is a set
of values o f y for partons present
0 tn2P
How does this ampli tude vary with the posi tion of the dots ? I t behaves
like a wave function for a finite one dimensional drop of liquid of thickness
tn2P , The ampli tude is large only i f particles are mom o r less everywhere ,
with a uniform dens ity except near the surfaces at 1n2P and O . The behavior
at one surface is no t strongly inf luenced by what the configuration is at the
o ther surface - they are insulated from each other by the long ( ? , 1n2P is
never really very large } intermediary plateau ,
Mathemat ically we are dis cussing the solution of H� • E� for a state of
fixed P z (but we do no t know H, o f course } . I f � is the momentum operator
(e . g . like t P z a:8it etc . ) we wan t � � • P� . Cons ider then that � is an
e igenvector of the operator W • H - -r:',

z for a state of mass M, Now as P + "'
(E VP �:= P+M /2P) we can consider the limi t

•
2 2
( 5 1 . 1)
so we are looking at eigenvectors with fixed eigenvalues of the operator PW as

P + •. We are assuming that it looks a s i f , a s P + • the operator PW has a
distinct limi t , expressible in x • p L/P and P� · This would be nice , but there
is t rouble on the small momentum end , The equation is like a cas cade , large x
generates smaller x through interaction terms (like turbulence equa tions , or
cosmic ray showers ) . Smaller momenta pile up until new phenomena s e ts in to
change the equations (like viscosity in turbulence , or ionization loss in
cosmic rays ) to finally determine the wee x ( finite p z ) behavio r . ( Th e app roxi-
mation in showing that PW depends only on x is wrong , for example we can no
I 2
+p -1111 ""' P z . ) But by that time the "shower is fully developed"
2 2•
�
longer write VP z
and the behavior of the wees (except for normalization - total s trength of
wees ) is independent of the way it s tarted at fini te x . (In the wee region
interaction energies are comparable to kinetic energies . )
The behavior at this end is a solution of the equation
w� - o cs1 . 2 }
(Note the omission of P . ) In general the operator W does not have a zer.o
eigenvalue if all boundary conditions are used - but here we relax the boundary
m, �
z +
condition of finiteness as p (It is like solving the Schroedinger
• •
equation Hw Ew for E 0 when it does not have this eigenvalue by relaxing
say the condition at r � m and thus studying open scat tering s tates approxi-
mately to which the real large r behavior will have to be attached - here we
mus t ultimately really at tack the finite x solution o f 2PWw m2 w . ) •
(a Lorentz transformation in the z direction by velocity v , call f ·' (l+v) / (1-v) )

I t can b e shown that s ince equation (51 . 2 ) is invariant under a boos t
in which all large P z are multiplied by f , that w can be of the form f 8 w

(so boos ting does not change w , only the normalization) . This means that fo r
finite but small x the probability of one parton at x varies as x28 dx/x . The
lowest B solution we assume in A4 corresponds to S • 0 ( from experiment , not
theory) . Other solutions exist fo r higher B and the general solution is a

linear combination of these whose coe fficien ts are de termined by how they
fit on to (51 . 1) . I had hoped to get a field theory interpre tation of Regge
theory this way but I have not completed the analysis .
A6 . The distribution o f the wees is the same for all hadrons . This is
a bold assumption partly guided by experimen t . Among o ther things it means
the wees are neutral to isotopic spin ; the wees for the proton look j us t as
they do for the neutron . The difference can only come from a higher B and
hence falls in amplitude relatively as P -B (B > 0 ) . The assump tion that the
wees are su 3 symmetric will not be made , (I think it leads to disagreement
with experiment in the expec ted ratios of � · s to K ' s in ce rtain experiments )
for we imagine interac tion forces are e f fective in de termining the wee dis-
t ribution and such forces are not su 3 invariant .
(It strikes me at thi s momen t , that since the wees are determined by
Ww • O , i . e . , a state o f zero mass squared , and that since pions have a small
mass , the s tate of the wees may be approximately only pions (with kaons , of
larger mass much reduced , hence large su 3 breaking . ) Known pion inte ractions
(perhaps des cribed by intermediary p mesons ) might permit a solution of Ww • 0
in terms appropriate pion base s t ates . If you do this I would sugges t it

might be easier if you work at first with the symme tric (intp z ) wee dis tri-
bution corresponding to two fast hadrons colliding , rather than this one-
s ided , one-particle dis tribution . )
The assump tion A6 is not comple tely obvioua from field theory - for
there might be some long range direct e f fect of the fast partons on the s low
ones in principle . The choice is guided by experiment (which shows that the
righ t-moving products of hadronic collis ions depend only on the right-moving
ini tial colliding parti cle and not on what it collided with . To see how this
assump tion is uaed see Bl, and J . Benecke et al . Phys . Rev . ,!!! 2 159 (1969) .
The phys ical assumption Bl that we make later says speci fically that there is
no such long range e ffect .
!!:]_. Con tinui ty demands , since the wees are adj acen t to the sea (plateau
region ) that we also have that � (e . g . mean numbers and correlations o f
partons) is the same for all hadrons .
AS . The
probability there are no partons in a sufficiently large gap of
where y depends on the quantum numbers

- Ay
rapid!!L..AY-1!!_�-fil..�s as e y
(angular momen t um , isospin , s t rangenes s ) carried by the gap . E . g . suppose we
have a proton s tate and we ask for certain partons a , b , c , for y > y 1 and others
below y
2
s t <E-- Ay � a b c
• •I I• • •
y 1·
Perhaps their s trangenes s is + 1 , the entire proton has S • 0 so there is a
con tribution S 1 across the gap . It is evident that this "quantum numbers
carried across the gap " de fined as the quan tud numbers of the s tate minus
those to the righ t of the gap ( a , b , c) is j us t the sum of those to the left
(s , t ) . This more complicated method of express ion is an anticipation of the
same idea for distributions when two hadrons are colliding . Then it is the
quantum numbers of the hadron moving to the r ight minus the to tal quantum
numbers o f the partons to the right of the gap .
Assumption AS is not s tated clearly . We have to say how the gap Ay is
changed . It is used in two cases : 1) Region a , b , c and s , t are both s tated
comp letely and the gap widens s imply becauae P increaaes ; hence t,y .. .tn2P and
the ampli tude falls as p-Y . This was used in analyzing the p ro ton form factor
in lecture 2 9 , for example . 2 ) The gap is in a plateau . On one s ide or both
there is a large s t re t ch o f plateau. Here a , b , c is fixed and . s , t e t c . , is
anything at all over a wide range of y up to the o ther b oundary (order .e.n2P
away ) ay is kep t fixed as tn2P r ises . Probab ility goes as (x 2 tx1 ) -Y where
x 1 , x2 are the x values at each end of the gap ( used for asymptotic behavior
of deep inelastic scat tering near x l ) .
•
Hadron - Hadron Collisions at Extreme Energies
Lecture 52
Hadron-hadron collisions at ext reme energies

Although our subject is pho ton-hadron interac tions we shall review the
assumptions made in describing hadron-hadron collisions A + B at extreme
energies . We first leave out e las tic scattering and diffraction dissociation
and aim toward the large part of the cross sec tion where several particles are
emi tted A + B + C + D + E + �- .
(
For a hard collis ion suppose the momen ta of A , B are PA ' P B respec tively
)
in the z direction - for example take center of mass PA • P B • P . For a
,...____-__
finite z-velocity v trans format ion from this PA • fP , PB • t P with f • �( l _v ) / (l+v ) .
We only work with PA , P B or P + m .
The asymptotically incoming wave function will be , o f course , (i . e .
before "in teraction") , some kind o f product wave func tion o f A o f momen tum PA
to right and B of momentum P 8 to left , Technical problems arise here . In field
theory this cannot simply be a product o f the wave functions o f each particle
that we have been des cribing for that is not unique ( for example suppose A
contains a Fermion parton at momentum p , and B also contains one of the same
kind at the same momentum,but there cannot be two in the field as they are
*
Fermions ) . Thua if A is represented as a creation operator PA on the vacuum ,
237
* *
�A • FA I VAC> which creates all our partons , and B by FB we can define the
incoming asymptotic wave function as
* *
There is some trouble in the wee region whe re F B FA do not co1111u1 te (note
* *
that no creation operators for fast partons p xP appear both in F and F8
....
A A
because A and B are moving in opposite directions) , Actually this is only
technical because we only want the state after the interaction . The prob lem
would arise only if we were quantitatively calculating the interactions ; but
now we wish to talk about how the wave f\Dlction looks after interaction hence
say after "interaction plus correction for overlap in defining the initial
s tate " , The overlap affects only the wees , but we shall assume the interaction
affec ts only the wees also . (By interaction we mean the e f fects o f the fact
* *
that although FA I VAC> and F8 1 VAC> are both eigenfunctions of H J �> • E l �> ,
* *
FAF8 1 VAC> is not . )
Following we s tate the assump tions we shall make regarding the interacting
wave function ,
.!!!_. Partons interact only if their relative four-momentum is finite ,
J
assuming they have some finite maaa o f order 1 GeV . Thia ia equivalen t to
the statemen t , i f y • i.n ( p z +1 + p z ) (in GeV) ( so y for negative P z ia j us t
2
-y f o r positive p z ) that partona 1 , 2 interact only if their relative y value

.I 1 ::Y 2 is o f order one or smaller .
(I use 1 GeV for the general energy values at which interactions fall
off e t c . suspect i n several applications even a smaller value (e . g . p�
I
average) may be correct , although possibly larger in some circums tances -

it of course cannot be defined precisely without a quantitative theory . )
We use this ass umption to get a t the wave function (in terms o f parton
distrib utions ) for the outgoing final s tate after interac tion . The dia tri-
bution in y of partons for the in s tates A and B have ranges of y small to
i.n2PA and - i.n2P8 to small respective ly . We put them toge ther smearing things
(effect of interaction) over a range &y • 1 . This smearing near y • 0 j oins
the posi tive and negative y regions (from A and B respectively) . Since these
regions were the same for both A , B (see A7) this can be done mos t simply by
j us t extending the common p lateau region from one to the other . The exact
pos i t ion of the CM le aves no t race then . We as s ume th is as a general principle .
B2 . In a hard hadron-hadron collis ion there are no special e f fects tha t
dis tinguish parti cles having finite momenta in the exac t cen ter of mass .
Longi tudinal trans forma tions with a velocity v not too close to c leave the
dist ributions o f such part icles essen tially unchanged . (Ass ump t ion due to
C.N. Yang . )
In our applicat ion the particles are par tons , the t rans formation alters
2 .tn
1 l+v
the pos i tion of the ori gin of y by .tnf • (a fini te amo un t ) and the
l-v
assump tion s ays the dist ribut ion should look the s ame . Hence the smearing
j us t has the e f fect o f extending the p lateau region o f A smoo thly back into B .
plateau
A in
p lateau
of B B in
-4,._��-+-��- y
.tn2P 2
F INAL STATE
p la eau
We mi ght ask i f the smearing of AY, • 1 changes the d is t ribution of part ons
for y near .tn2P where they were de termine d entirely by A . (This region is
called the A fragmen tat ion re g ion , near -.tn2P the B fragmen tation region . )
B
But near A the dis trib ut ion already s a tis f ies ( in the sense that a li quid
surface is nearly independen t of wha t goes on deep be low i f forces extend
over fini te distances) the wave e q uat ion , so we assume it is no t change d .
Therefore we imagine that the final real hadron parti cles come from th e
disintegration o f an "original" parton s t ate wh i ch has the following prope rties .
B J . P.i.!...P. a r ton s ta te appropriate to a fas t hadron collis ion o f A and B

is one whose dens i ty matrix looks like that for the A in the A single particle
fragmentation region near in2 P , or bette r if x • p / P , f o r x > 0) , like
(y
A 1
B alone in the B fragmentation region (y near -in2 / P / , i . e . for x < 0 ) and like
B
the universal plateau region in regions in be tween < l x l small and wee ) .
Thus thi s is completely described in terms of the wave functions for

single particles described in assump tions Al to AB . We emphasi ze again tha t
our assump tions are not independent ,for example , for B3 t o work the pla teaus from
each particle mus t be the same as A7 says . We are not trying to develop a
logical system of assumptions , but j llEl t state a number of mutually cons istent
(or poss ibly inconsistent - see quark assumptions later on) ideas .
The picture we are developing in B3 is a wave function like a liquid in
the variable y with surfaces A, B at which the dis tributions are unique but
an in terior plateau or sea region which separates them . They can be well
separated by taking P large enough , for they are separated by in 2 P - which
is universal . Further the relations or correlations from one place in y
to another have a finite y range o f order one - so the general behavior in
this sea is like a Markov chain - wi th enough separation in y things become
uncorrelated . Many obvious properties expected for such a chain can be
expected here but they will not be all explicitly s tated (for example , the
probability that there will be no parton at all of a given type in a range
6y goes as exp (-C6y) for some C for large enough 6y , etc . )
The word "appropriate" in defining the wave function is purposely vague

for I am not sure whe ther I am des cribing the final outgoing wave func tion
after interaction when all the particles are separating or one in between the
ini tial and final . I have not clearly resolved my confus ion on this matter -
but as I only use the function qualitatively in a manner described in the next
lecture I have not had to clari fy i t .
Technical Footnote on the small momentum region in Assumptions B 2 and B3
y near z
On drawing the y plots to describe the parton wave func tion of assumptions
B 2 and B3 we assumed the wee region near y • 0 where A , B interact/ as being
completely healed over and j us t a smooth continuation of a p lateau through
y • 0. This is physically what I want and leads to C2 (next lecture) with
uniformi ty in the y de fined there for physical particles and in accordance
(/ )
with invariance under fini te velocity trans formations which applies to particles .
2 1
But technically the curve in the y space y • tn 1 P z +1 + P z fo r partons
migh t show a bump near y • 0 , a bump which moves when we make a finite
Lorentz trans formation , which is fine s ince wave functions need not be relativis tic
invariants (I thank F. Merritt for pointing this out) . But it mus t be a bump
so constructed to have no physical effect as a bump in the final real hadron
distribution C2 . I t is a "theoretical artifact" due to carelessness in finding
( J +pJ
the right normalization and def inition of var iables for the wave function . The
2 1
resul t cannot really be smooth in the s cale of y • in p z +1 for if so it
(J +p2)
would not be smooth (near y • O ) i f I had arbitrarily chosen to use
y ' • tn p z +1/ �
2
for dy/dy ' is not cons tant .
Lecture S J
Hadron-hadron col lisions at extreme energies (continued)
We now go on to describe what the products might look like in hadronic
col lisions (still leaving out diffraction dissociation ) . We have , of course ,
no quantitative way to ge t from the wave function des cribed in partons to the
wave function des cribed in outgoing real hadrons . But we shall simply assume
that in y space the relation of parton + hadrons is much like the relation
hadrons + partons described in A. Later on we shall have t o describe the
products expec ted from s t ates which unlike that in B J , have gaps in them , for
example , a s tate with j us t two partons at opposite ends of the y s cale ,
separated by 2P . We assume ( the "complement' of BJ) :
Cl . The hadrons which would result from the disintegration of a state
whose initial wavefunc tion consists of one parton a (or a few) go ing to the right
� a and one b parton going to the lef t at yb separated by a large gap
..Ya ::Yb • tn2P consists o f hadrons o f finite transverse momenta in three regions .
lbere are hadrons going to the r ight with y
near Ya depending on the character
of a; there are hadrons going to the left with y near yb characterized by
parton b and those in between distributed in a universal uniform plateau -
like sea which is the "universal sea of hadrons corresponding to a parton

gap " .
It has been proposed that this "sea corresponding to a parton gap " be
i tself a gap . This will not be inconsis tent with what we will say next (C2 ) ,
and which we use in hadron collisions . But i t does no t seem reasonab le by
itself to me if hadrons make a universal sea as B2 supposes - for I think that
means that if there were any dis tribution of two lumps o f hadrons wi th a gap
between , they would make a sea of par tons , so if the re is no sea of partons
there is no separated lumps of hadrons , but there must be a hadron sea . At
any rate it is my s trong belief that there is in fact such a sea in this case
and not a gap in hadron momenta corresponding to the gap in parton momenta.
As A. Cisneros points out the two outgoing lumps carrying opposite
hadronic quantum numbers ( in the case e+e - + hadrons ) would genera te a dipole
s t rong current which would tend to radiate softer hadrons into low x . To pre
vent this radia tion becomes increasingly mo re di fficult as the energy increases ,
(as for two-body exchange reac tions ) . Any exclusive two-lump probability will
fall as a power of energy relative to the total inclus ive reaction in whi ch the
radiation , generating an intermediate plateau , is permit ted .
Now for a wave function as in B 3 we can imagine the various partons
disin tegrating more or less like in Cl but not really independently , those at
the ends of the y range determine the hadrons there , and tho se in the center
affecting the hadrons in the center , but in a universal manner independent of
y in thi s region . Thus we again ge t a hadron dis tribution like that in Cl
with three regions , but the plateau may be a new and different dis t ribution
but a sea nevertheless . (Whether the two plateau regions , the one in Cl
corresponding to an initial parton gap , and this one for wave function B J ,
are the same or no t i s a di fficult problem I have not yet been able t o decide .
We shall call the assump tion that they are equal C6 , see lecture 55 . ) Thia
assumption can be made by j us t repeating the wording o f Cl j us t changing the
name of the sea , o r it can be put in another totally equivalen t way .
Remark : There is a possible confusion here be tween the "initial" wave

funct ion of Cl and the "appropriate " wave function of BJ . Cl in e +e
-
collisions is j us t after the interaction wi th the pho tons - the parton

pair is j us t c reated . There s till remains time for interactions (via terms
* *
like a a a in the Hamil tonian) to act before we reach the parton representat ion
of the "final" initial s tate - i . e . before we reach the "appropriate " wave
function in the sense of B J . This interaction converts the ini tial fas t
parton into two or mo re , and these again are broken up e tc . , in a cas cade
fashion making profound changes , for example by filling in the gap in the
low P z region and creating some sort of parton plateau in the final outgoing
s tate "appropriate" to the initial s tate Cl .
The reason no such extensive modification is made in going from the
ini tial s tate in a hadron collision to the final appropriate state BJ is this .
The distribution o f the fas t (non-wee) partons in the init ial s tate already
s atis fies H� • E� so little dis turbance is worked there by the Hamiltonian .
Only in the mutually overlapping wee regions does the further action of H
modify things ( to smooth out the plateau) .

ft_. The distribution of final s tate hadrons at each y resulting from a
parton distribution like BJ depends on the nature o f the parton dis tribution
within a finite range in y of y,
This i s true no t because dis tan t partons have n o effect , b u t rather because
they have a universal e ffect , Here (and in Cl) y can be mo re p recisely de fined
( )
for the hadrons are on their mass shell and have a definite mass . We take
1.n (E+p z ) • +m +p 2
2 2
y • 1.n V/p1 z
'
�
+p z say in GeV ( change o f scale means j us t a
change in origin of y) :
CJ. Putting th i s a l l toge ther i t means that i n a hadron collision A + B +
anything in the CM sys tem plotting x •

P /P for x negative the distribution of products
z
depends only on x and B, for x pos itive only on x and A as P + • ; and in
the small tx region the distribution becomes universal varying like dx/ l x l
and mo re generally continuing across the wee region like dp z /!.,
The idea that the right movers depend only on A and the left movers
only like B is called limiting (i . e . as P + •) fragmen tation . It was
suggested by C . N . Yang et al , Phys . Rev . 188 2159 (19 6 9 ) but at the time it
was supposed these regions separated and had no communication , but in fac t
there is a sea between . This sea , howeve r , we suppose is universal and

(although it logically could) we assume it carries no information from the
r ight to left region .
We have omi t ted the diffrac tion dissociation but it is evident that if
it is added in a de finite pe rcentage to the inelas tic will not change our
conclus ion . However , the fraction that elas tic s cattering is of the total
cross sec tion does not seem to be universal ( for example , for pp at large P
+
a e1 / a to t � . 25 whereas for w- p it is closer to . 17 , see G . Giacomelli in

Proceedings Ams terdam Conference on Elementary Particles , North Holland
Press (19 71) ) . Thus limiting fragmentation cannot be absolutely exact . It
is probably generally nearly corre c t ; perhaps in a future mo re exac t under-
standing it will be true for parts of the collision characterized by some
o ther parame ter (e . g . impact parame ter) but when integrated ove r this parameter
i t is no longer exact for differen t cases give various di fferent re lative
weights to the various values of the paramete r . Nevertheless even with this
evidence agains t its perfect universal validi ty we continue to analyze in a
naive and s imple way leaving refinements to some future date .
There are a number of addi tional conclus ions made by assuming the
Markovian idea and extending our ideas such as about gaps A8 from parton
to hadron wave functions . We shall not dis cuss them in de tail for hadron
collis ions is no t our main subject but give some examples . We assume ,
analogously to AB that
C4 . The probability there are no hadrons in a sufficiently large gap o ;

rapidity &y goes as e -a&y where a depends on the quantum numbers carried by
� (right go ing particle quantum numbers o f A minus quantum numbers of
all hadrons to the right o f the gap ) . For example , for the exclusive collision
A+B + C+D so the outgoing state is pure C to righ t , D to lef � the gap is
1 2 I PA l +tn2 I P8 1 • tn4 I PAPB I tns and e -a&y
• (4PAP8 ) -a • s -a where a depends
•
on the quantum numbers of A-C . Looked at from a Regge point of view this
whose a depends on the quantum numbers exchanged in the
-2 (l-a )
should go as s
t channel , which is the same as A-C . Thus a is identi fied wi th 2 ( 1-a ) (or
whatever the correct power law o f energy fall-o f f turns out to be) and we make
a contact with the theory of exclusive reactions . (The same goes if C is
n + p
2
two particles like of fixed total mas s not necessarily at resonance .
I do not know o f examples o f f resonance where this powe r law has been checked -
again we see the uni ve rsal p rin ciple that going to higher ene rgy does not l i f t
a resonance ever higher against "non-resonan t " background - tile latter can
always also be though t of as tails of o ther resonances . )
Again applying this to the case that C only is near the end o f its range ,
so x is nearly 1 bu t D is any thing , even many particles , we see we are

c
-a1o ( l-x ) a
generating a gap of 1n (l-x ) and the amp litude goes as e c (1-x )
c
• •
c
The distribution o f x is then d (l-x )
c
a
a ( l -x c ) a-l dxc with a 2 -2a .
• •
c
Tnis resul t , though " legally " t rue as P + .. , xc + 1 is in prac ti cal cases
nearly unobservable . For example suppose C is a p ro ton produced in a p+p
collision . Protons also come f rom di f fraction dissociation o f a res onance o f
o f mass � say going t o proton W and pion . Th i s spills protons over a range
f rom x • (E -p ) / � to x • (E +e ) / � (as P + m ) , where E ' p are the energy
p p p p p p
and momentum of the proton in the res t frame o f the resonance . Although the
latter is less than one , there is a very small gap (of range . 9 8 to 1 for
Ha2 • 2 . 16) f ree of dif fract ion gene rated pro tons it is too small to isolate
experimentally . I f a gross plot is made for x not sufficiently near 1
c
various numbe rs of dissociat ion pro t ons are included and th e variation o f
•
numbers appears far from (1-x )
c
l-Za
dx
c This di f ficul ty does n o t arise for
pions when the incident par t i c le i s a pro ton .
Lecture 54
High Energy hadron-hadron c ol l i s ions (con tinued )
I should like to make a few coaments about our "conclus ion" from C2 .
First s ince the mean number of particles goes as dx/x in the small region and
continues acro ss x • 0 as dp/e it is evident that the tot al number of particles
of a given kind ( the multiplicity for tha t kind) rises logarithmically with
1nP or with lns . This is al so obvious in the area of the y plot whe re the
plateau expands logarithmi cally with s . But the plateau region is (statistically )
neutral , i ts average for any addi tive quantum number such a s charge , third
component o f isospin , baryon number , hype rcharge , z componen t o f angular

momentum e t c . , mus t be zero (becaus e if not dx/ x would give a ins dependent
value for one o f these fixed cons erved q uan tum numbers ) . We expect this from
t� e cas cade idea of how the plateau is fo rmed .
J
Thus s uch integrals as
dx [Number of w+ at x - Number o f w at x] converge to numbers which are

0
characteris tic of the particle A ( initially moving to the righ t ) independen t
of P as P + .. . Independen tly corresponding " le f t numbers " like the s ame
integra l for x • -1 to 0 can be de f ined , which should depend on B .
In particular then we can define definite quantum numbers ( fo r the
additive quan tum numbers) for the right -moving particles , by s imp ly adding
the to tal number for all for which x > 0 ( i . e . p > 0) in the CM system .
z
This number will vary from event to event , of course , but we want the s tatistical
expected mean ove r many events . Thi s "mean righ t q uan tum number " will approach
a cons tant as s + .. . Thus we can talk of the " righ t mean 3 -isospin " or the
" righ t mean s t rangenes s " . I t is evident that thes e mean right quantum numbers
mus t , under the ideas o f C2 , be the same as those o f the incoming righ t -moving
particle . The plateau region does not le t any quantum numbe rs s l ip through i t .
( I f for examp le we take a symme tric colli sion A +A then by overall quantum
number conservation and symme try the right quantum numbers ( and the left)
mus t be tha t o f A. But by limi ting f ragmentat ion replacing the le f t A by B
to make A + B does no t change the distribution of right move rs hence they
s till carry the quantum numbers of A.)

Thus in te re s t ingly a s P + .. the right-moving particles in the mean carry
the energy , the momen tum (minus a cons tan t ) , the 3 isospin , s t rangenes s , baryon
number , z angular momentum , e t c . , of the incoming righ t move r .
NOTE : We show that , dis regarding quan ti ties of order l / P , the difference of
total energy E and to tal momentum P o f the par tic les moving to the right is a
cons tan t D • E (& - p ) ( independen t of P as P + .. and in fact if the p la teau

i zi
is universal , the same cons tant D for every particle A) . For fini te x ( say
posi tive) the difference from one hadronic particle is & - p • V2 2 2 2

P x +p +m - Px
&
�
2 2
+m ) /2Px (m is the mas s of the hadron) which is of order l/P and therefore
J.
(p
negligib le . The main cont ribution comes from x near zero where the dis tribution
of a particular type is cdp/ & , hence the con tribution to & - p of these is
c
f The integral g ives c .m
V
P.J. TIU , so D
�
p •O
z
is the sum of c ove r the transverse momenta and the types of hadrons
in the plateau . If the pla teaus are universa l , the cons tan t D is unive rsal
and may be eas ily calculated in te rms of already measured quan t i ties .
Thus in a collision of A and B each particle is conve r ted into a train o f
part i c les moving in i t s own direct ion . The " t rain A" has the quantum numbers
of the particle A and its ene rgy (by the conse rva tion of energy) but has l os t
a ce rta in momentum D in the inte rac t ion , it is held back a b i t by the in ter -
action , A and B each lose D to the othe r . ( Such a finite momen tum trans fer
is, of course , consis ten t and un ders tandable i f only wees inte rac t in the
collision . )
(For the wave func t ion of a s ingle hadron described in assumptions
A2 to A6 , the to tal momentum o f the par tons is , o f course P , the total momentum
-Mil +p
2 2
r
Jip2
..L
of the s tate , but the total ene rgy E E • i s no t the to tal
i i
i
energy E "' P because of inte rac tion ene rgy wh i�h compensa tes the expe c ted
f in i te excess of E E above Ep )
i zi'
As a first s tep to describe these things formally , we are trying to
des cribe the s tate
(where " in righ t " means having very large positive longi t udinal momentum P ,
and " in left" means -P) in terms o f outgoing hadron s tates - an element of
the s ma trix . O f course the mos t likely thing is tha t the two parti cles
do not coll ide , making s imply / Aout r ight '

B
out le f t
>. We wish to deal with
the wave funct ions if they collide , so we wri te as usual S • l + iT and we
are s peaking of the T matrix . We shall no t no rmalize it co rrec t ly , but j us t
outline our ideas . Formally this wave func tion can b e given i n terms o f the
*
ampli tude to f ind various outgoing hadrons . If c is the ( formal) operator
to represent the creation o f some kind o f a had ron (kin d , transve r se momentum
"'
p , and longi tudinal momentum p are indices of c ) we can represent s uch
1
*
s t a te s by X / VAC> where X is some operator fun c tion of the c . We have been
discussing how X looks . Let M be an operator to create a p lateau - say a
typical universal pla teau for some range of x around 0 - say x = - . 2 to + . 2

[ the exac t way the plateau of M cuts off for finite x is arb i trary ; i ts
choice affects the de fini tion of GL , GR defined late r , but the final operator
X is not dependent on this } . Next we write X as G G� where G is to modify
L R
*
the sea on the right ( for x > O. It involves creation operators c to add
particles to (and beyond) the plateau operator M , and annihillation operators
c to take particles out (which were put in by our arbitrary cho ice of how the
p lateau M is defined) but all these for x > 0 ( that is the meaning of the R) .
*
Likewise G L is an operator func tion of c , c only for x < O . The operators
L R
G , G commute s ince they contain operators o f different particles (some signs
mus t be adjus ted for Fermi particles ) . Thus we write
I A in R ' B in L> • G!G� I VAC> (54 . 1)
where we have writ ten I M-plateau> 0 M I VAC> . R

The operator GZ depends only
on the particle A, etc . If you want things to look even nicer write the le f t
*
s ide in terms of operators too , say d which create incoming particles , and
then have
* *
TdAR dBL I VAC> a GARGLB /M-plateau>
*
Thus the operator d! is equivalent ( in this two-body equation at leas t) to
*
GAR ' but in an odd representation in which dAR acts on the vacuum and GAR on
the M-plateau s tate .
A research problem which is very important , and virtually unknown
theoretically , is ( the very rare) collis ions at extreme energy in which the
particles come out at large relative momenta to the original direction . For
example , pro ton-p roton elastic scattering at finite angle , e . g . 9 0° , where t
is the same order as s as s + m , What kind o f phys ical view accounts for these
collisions , I shall not discus s ideas which have been tried here , for our
subject is pho tons , bu t shall only commen t that nothing is clearly unders tood
and you can s tart from scratch on your own . ( For example will assump tion B l
have t o be abandoned or quantified? ) (You s tart b y looking first roughly at
the experimental resul ts to remember qualitative salient features that might
need explanation . )
Comments : By assuming that the wee region is the same for each hadron and that
only wees in terac t have we not assumed that all to tal cross sect ions a or a
pp rrp
etc . are equal , clearly contrary to fac t ? I have no t though t this out c learly
but have always suppo sed that the part of the wave func tion which does inter-
act (which is always infinites imal compared to the part whe re they go pa s t
each o ther wi thout inte rac t ing) could s t il l have some normal ization re lated
to the total cross sect ion for that parti cular collision without being in-
cons is tent with o ther ideas . In the formal express ion above , for example ,
each G could carry a numerical coe f f icient g proper to A . This would make
A A
2 2
total cross sec tions proportional to g g , or as is said , factorizable . I t
A B
ma y b e that the previous assump tions d o no t imply that the t otal cross s e c tions
are necessarily equal , but rather perhaps that they are factorizab le . It would
a •
imply for examp l e , that
rr p VfO r;;-
pp \f rr rr
etc . We do not have any evidence on
whe ther thi s is true .
In these s t ud ies we have made no remarks which permit us to understand
transve rse momentum behavior (except to say that transverse momenta in hadronic
collis ions are limi ted , a res ul t taken direc tly from experiment ) . Obviously
lots o f in teres t ing theore tical ques t ions remain , s uch as what func tion is the
transve r se momentum d i s t r ibut ion , how does it differ for various values o f x ,
o r for rr ' s and K ' s ? How should exclus ive cross sections vary with t , e t c . ?
This entire realm o f phenomena has been le ft out o f our analys is , an excel lent
future oppo rtunity for advance lies here .

Final Hadronic States in Deep Inelastic Scattering
Lec ture 5 5
Interact ion of partons with the electromagne ti c field
We assume that in the or iginal f ield Hami l tonian describing hadrons in
�e rms o f partons the re are terms giving the coupling o f partons with the
vector po tent ial of the quantum ele c t romagne tic field . We shall assume in
the spirit of minimum e lec t romagne tic coup ling that they couple in the simplest
way expec ted from the propagat ion operators via gauge invariance . That is we
assume :
Dl . The coupling of partons to the elec t romagnetic field is the ideal
minimum coupling operator . That is also the coupling that would be valid i f
they were ideal free particles .
This coupling is not unique if the partons are s p in 1 or highe r , but
for the present this will not concern us for we shall suppose par tons are
ei the r spin 0 or spin 1/2 . Al though we are in danger of no t having the mos t
gene ral case w e shall never theless explici tly next take the working hypo thes is
( sugges ted , o f course , as we have discussed by experiment on vw2 and w1 and
not a priori by theo ry) that
D2 . All the charged partons are of spin 1 /2 and hence couple through
the current operator e tly where _ e tl is the parton charge (ti is an index for
ll
the kind of parton) .
250
Final Hadronic States in Deep Inelastic Scattering 251
We have seen how the assumptions Al-A8 p lus these two Dl and D2 lead to
the s c a ling expectations for the deep inelas tic s ca t tering ( lecture 2 7 ) and
there is no reason to repeat all that here again . Howeve r , he re we shall
discuss what we can s ay about the p roduc ts in photon collis ions , in part i c u la r
we 2Mvx , Mv •
2
begin wi th the deep ine lastic ep scattering region q • - P •q
(P is the momentum of the proton , q tha t o f the virtual p h o ton) so v • virtual
phot on energy in labora tory (pro t on a t res t ) system . Le t us us e the coordinate
sys tem with the virtual pho ton purely spacelike q • (O , -2Px , 0 , 0 ) P • (P , P , 0 , 0 )
µ µ
q • -4P x , 2 Mv • 4P x . Then as a result o f our awsump tions the parton wave
2 2 2 2
function before and a f te r the collis ions looks like :
�
-2Px
Before
Immediately
After Px
Coup ling
That is , one parton ( say type a) moving to the le f t , the remaining par tons
moving to right j us t as in original proton , less the individual par ton a o f
The relat ive prob ab il i t y o f th i s pi cture is e

2
momentum x .
a na (x)
is the number o f partons of type a with p /P • x in the original proton s tate .
z
Then: the hadron prod u c t s in deep inelas t i c region are those expected from
the partpp dist ribut ion 1 u s t des c r ibed , The total cross sect ion in this s cale
is proport ional to � e
a
2
n (x) so the dens i ty ma t rix per collision is a super-
0
pos i t ion o f cases of d i f fe rent types a of partons with weights , wa depending
es n (x) ;
2 2
on the charac ter o f the collis ion t h r ough x ;
a
w (x) • e
a na (x) /ta a
the sum of the weigh ts be ing 1 .
One obvious consequence of this and our othe r assumpt ions is that in this
sys tem the transve r se momenta will be limi ted an� ( fo r fixed x as we vary P ,
o r i f you like v ) .
The longi tudinal momenta will s cal e as P , i . e . if they are s t a ted in
units of P as say "p the d is t r ibu tions will be independen t of P as P + w
( depend only on " ) .
We expect also near n equal zero to find a d n / n behavior . For p o s i t ive
n (night movers) we expe c t it behaves like the unive rsal M-plateau charac teris t i c
o f a wave funct ion like B 3 , which w e know f r o m hadron collis ions . For nega tive
n we are in the "plateau region of init ial par ton gap " (de fined in connection
wi th Cl) , We have no t assumed these two p lateaus are the same so the coe fficient
of d11 / 11 need not be the same , If they are not we shal l have trouble defining
what happens in the trans ition region - it cannot go s imply as dp z / E for that
leads to the same coe f ficient for plus and minus 11 . We see however that this
ques tion is to some extent an arti fac t of our particular cho ice of coordinate
system . Note that the s ta te described he re as " immediately after coupling" is
an ini tial parton s tate ( in the sense discussed in the remark fol lowing the
discuss ion on Cl) - there s till mus t be interac tions from the Hamiltonian
be fore it becomes the "appropriate" outgoing wave function . This will produce
cascading of the le ft-moving parton into the gap smearing the wee region into
negative 11 and making large readj us tments for the righ t-moving sys tem also
(because , for f inite x; they are no longe r the correct solution of Hw • Ew
since one parton is missing) ,

Thi s all appears qui te complica ted and it is difficul t to make firm pre
dictions . However , we might cont inue t o assume interactions are each limited
in range on a rapidi ty plot - although there are many of them poss ible fil ling
in gaps , etc . But we shall t ry to adhere to the principle at leas t that the
parton moving to the lef t determines the final hadrons to the le f t and like
wise for the righ t . We put this idea formally into the following assump t ion ,
a generalization of Cl , C2 (we wri te i t independently , for it may no t be true
while the s pecial ca se Cl or C2 may be) .
C5 . In the center of mass system (or one moving longitudinally at any

velocity not near c) an ini t ial s tate cons isting of any dis tribution o f
partons to the right and other partons t o the left yields in the final s tate
a distribution of hadrons such that those hadrons moving to the right are
entirely characterized by the ini tial partons moving to the right , and does not
depend on those initial partons moving to the left . (Likewise , exchanging
le f t and righ t . )
Assumption C5 , if it were right and this continui ty in dP z / E , would
seem to sugge s t that both plateaus fit togethe r . I am not sure of myself
here but shall put i t down as an explicit assumption which would , if i t is
true , remove all our difficulties - the dn / n region i s always universal ,
that of hadron-hadron collisions .
Final Hadronic States in Deep Inelastic Scattering 253
C6 . The dis t r ibution of hadrons in the plateau region 1s the same fo r
every ini tial pa rton distribution . This as sump t ion is at p resent , ve ry weakly
based and may eas ily be wrong - it is an in teres t ing conj e c ture .
We now make a more detailed discuss ion of our expe c tations for the le f t -
moving part icles . (I have profi ted grea tly from conversat ions wi th A . Cisneros
on these mat ters . ) Fo r these parti cle s a variable more convenien t than n (which
goes down to -x) is z • - n/x • -p 2 / Px = P • p/P • q the fraction that the left
moving part icle ' s momentum i s o f the to tal le f t -moving momen t um . S ince this
is z • P . p / P . q i t is the ene rgy o f the particle in terms of the energy v o f
the pho ton i n the laboratory sys tem . I t is the p roper variab le for seeing
how the vir tual pho ton fragment s . Of course as v � m , x fixe� the d i s t ributions
in z s cale as we have said .
If we could be sure that only a parton of type a came out (which by the
way , can be much more nearly done for neutrino scat tering - in the quark model ,
neutrino scat ter ing can lead to unique quarks to the le f t ) the dis tribu tion
to the le ft would be unique - say D ( z ) . The number o f � · s wi th a given z is

a
:
a fun ction of z only D ( z ) , the p robab i l i ty o f �. K at z , z depends on
1 2
1iC
z , z as D ( z , z ) - and these funct ions do in no way depend on x . They do no t
1 2 a 1 2
depend on x because the hadrons to the le ft deoend only on the parton to the
left ( a ) and the adj acent wees from the hadron ( i f at all) - and these
latter are unive r sal and unaffected by the removal o f the parton a t x from
the proton . The la t ter does no t affect left-moving hadrons for its rela tive
momen t um to left-moving hadrons is no t finite but grows as P � m.
Th e ac tual d i s t r ib ution seen a t a given x will depend on x be cause the
relat ive probabili ties of producin g di f ferent kinds of partons a will depend
on x . The actual dis t r ibu tions D (x , z ) wil l b e the weigh te d average fo r each
part on
D (x , z ) • "\" wa (x) D (z) (55 . 1 )

L a
2
where the weigh ts w (x) proport ional here to e n (x) are de fined by
�
� a a
2
w (x) - e n ( x)
a a a / � e s 2n s (x)
s
These funct ions D ( z ) , or equivalen tly their creation operator

a
o! on
M-p lateau ( if C6 ) isola tes something characteri st i c o f partons and , if our
assump t ions are all corre c t , the re fo re ve ry fundamental indeed . We shall
dis cuss later a specific par ton mo del (quarks ) as wel l as some practi cal
que s t ions about the possible ext rac tion of the isolated D ( z ) from experimen t ,
a
as well as the possib i l i t ies o f finding the w (x) by special guesses ab out how
a
D ( z ) may behave . To me the possib ility of special funct ions charac teris t i c
a
o f each kind o f parton i s a very interes t ing possib i l i ty , an d one tha t could
be an entrance to a path in to the heart of the me chanisms of strong inter -
actions .
These same funct ions D ( z ) wil l appear in certain other expe riments - for
a
examp le , of course , in deep neut rino p ro ton � µ + p roduc ts experiments . The
analys is is nearly the same as here excep t that the fundamen tal coup ling may
be differen t so al though n (x) are the same the weigh ts w (x) come out dif ferently .
a a
+ -
Again in the e e collis ion the assumption D2 says our ini t ial s tate
- 2
is j us t a pair , parton a and antiparton a with weight e • Thus , ass uming
a
2 R L R L
C6 the final s ta te would be
a a a a a a
E
e (D D- + D- D ) I M -p lateau> again producing
hadrons in any one direction characterized by the di s t r ibution
Ia
e
a
2
D (z)
a
where we sum an a over partons and an tipartons .
I f , for examp le , in some expe riment we could be sure that a cert ain
parton a came out to the left say , then as we have seen we would expect tha t
the t o t a l "le ft-moving quantum number" ( the s um , f o r s o me addi t ive quantum
number o f tha t number fo r all hadrons llK)Ving to the le f t ave raged over al l
even ts ) would be that of the parton a . Thus in p rincip le we could de fine
or dete rmine in terms of experiment the quan t um numbers of the par tons . If
the s tate is n o t pure w e shall have to know some thing of the weigh ts w (x) t o
a
mak e this useful - but there are so many di f ferent kinds o f expe riments
possible that in principle the w (x) can be de termined as we ll as the ove rall
a
quan tum numbers of the partons .
The particles to the righ t under de ep ine las tic e-p s c a t tering come from
fragmen ta tion of a pro ton with one a parton of llK)mentum fract ion x remove d ,
say E (z) . They are evi den tly not ve ry fundamen tal . But it is clear
(p -a , x )
that the same kind of f inal s tate resul ts (on both s i des , le f t and righ t ) in
+
Dre ll ' s experiment p + p � µ + µ + any hadrons so the p ro ducts in this
Final Hadronic States in Deev Inelastic Scattering 255
experimen t can be en t i re ly expres sed in terms o f these E ' and hence

(p-a , x)
( suppo s ing n (x) has been worked ou t) in terms of the p roducts fo r deep
a
ep sc a t te r ing . We leave it for you to wri te the exp licit relat ions and to
sugges t p ractical expe rimen ts to test your ideas .
NOTE : Our f inal hadron s t ate is acco rding to our ass ump tions
w ( ) DL E IM
R
a X
p lateau>
a (p-a , x)
a
where D � i s the operator for a left par ton a, and E

R
(p -a , x )
for the righ t
moving fragments . B u t this can be conside red a s a mnemonic only f o r the
exp ress ion is p robab ly impo s s ib le for one ope rator M always . Be caus e nothing
would seem to p reven t us from wri ting D 0 I M p lateau > whi ch woul d have to tal
R L
a $
quan tum numbers of two quarks (a + $ ) whi ch is impossible to write in terms
o f the hadron ope rato r s having only integral quan t um numbers . (I am indeb ted
to J. Mandul a for po inting this out . ) A va lid mathema t i cal repres en tation
for these i deas is an excellent p roblem.
The reade r should be warned that a number o f these s caling p redic tions
for special p roduc ts of reactions may only hold at much higher energies than
that at whi ch s caling for the total cross sec tion (vW and W ) sets in .
2 1
This warning res ults from theore t i cal experience wi th a numb e r o f . examp les
o f analogo us theo rems in non-relativi s t i c quan tum me chanis whe re the sum
works well before the individual terms do . This is because if cer t ain inter-
ac tions are dis regarded in wo rking the total p robab i l i ty by assuming ce rtain
s tates only are "en tered" subsequent interactions may not change the total
p robab i l i ty the s t ate was "en tered" b u t may redis t ribute that p rob ability
ove r di f fe ren t final s tates than we re expe c ted .
Spe cial case o f small x
In the special case o f small x the predic tions are espe cially simple .
First cons ider the righ t s i de (original p ro t on ) . He re we have a parton dis -
tribu tion j us t l ike that of a p ro ton with only a ve ry low x parton removed
and the wees dis turbed (by interact ion with the p lateau deve loping from the
left ) , Thus all the partons ot any s ub s t antial x are exac tly like that of
a proton - and we can expe c t the s ame dis t ribution of hadrons to come out
(at leas t for z >> x ) as do come out fo r a hadron colli s ion of a p rot on , s ay
,,,;i F
R R
( p-a , x )
Hence for x small E
p
Next for small x all n (x) dx go as
a
Ca dx/ x where Ca is a cons tant , so
C / t C • y approaches a cons t an t ya independent o f x for

•
a
that w (x)
a
a a a
Next call D the mixture D • E y DL
L
small x .
r
L
r a
a of the distribu tions for each
part on , each weigtced with weigh t Y · Our hadron di s t r ibut ion be comes thus ,
a
for small x , nearly
L RI
D
r p
F M-plateau>
That is , for smal l x the p ro t on fragmen ts into a form independent of x and
the s ame as it does for a hadron collis ion . And the virtual pho t on also
fragmen ts in a un iversal way independent o f x . Since w e have ass ume d the low
Ca and ya ' and hence D

r
x region the same for all hadrons , the do not depend
on the part icle s t ruck by the pho ton (normalized to the total cross -sec tion for
collision , of course ) . A smal l x pho ton and a hadron behave j us t like the
collis ion of two hadrons , each fragments in its own charac teris tic way . That
o f the pho ton is independent o f x .
Region of finite q
2 , v + •
For finite q, negative q2 we can s till use our sys tem o f coordinates in
which q has only a s pace component Q .
except if Q • O. I t is clear here however that only x near zero can be
affected by the pho ton Q ; that is only the wees are e f fe c ted . They are ,
howeve r , affected in a very compli cated way for in teraction is important in
the wee region . we cannot there fore predi c t what will happen there , but we
can note (a ) that it is the s ame for every hadron A , A + y + produc ts for
we have the s ame wees for eve ry hadron according to A6 , and (b ) the frag-
men tat ion o f the fini te x in the above sys tem is charac teris tic of partons
of sys tem A only , for only the wees are effec ted by the pho ton .
In consequence of (a) the produc ts on the left which can be desc ribed
z •
!'..:2.
in terms of p . q where P is the proton four-momen tum, p is tha t of a
product and q that of the pho ton , for finite z as v + • , is some kind of a
dis t r ibution D
2
y , qz ( z ) . The dist ribut ion clearly depends on q , the vir tual
Final H adronic States in Deep Inelastic Scattering 257
mass of the pho t on , because the comp licated interact ions of the wees depend
on this momen t um . I n the o t h e r dire c t ion ( the variab le ( q • p / q • P ) the proton
fragmen ts in the same way as it does for hadron col l i s ions . The se cons idera
•
2
tions hold for q 0 also , of course , but our coordinate sys tem is inconvenien t
fo r such a case .
We could also use the cen te r of mass sys tem for any finite q
2
��(P ' , P ' ) ( P ' , -P ' )

2
Conse rvation o f energy and momentum means the virtual pho ton (P ' + m,q fini te)
in teracts only with the wee p artons o f the targe t p ro ton (or h adron A) . Th is
interact ion is complicated but p roduces the same dist rib ution for any hadron
2
fo r given q The hadron b ehaves as it always does where its wees are dis-
turbed whe ther by another hadron or by a photon . Formally our final hadron
s t ate is
L RI (5 5 . 2 )
D 2 F M-plateau>
y,q A
Thus as far as high energy inelas tic collis ions are concerned the
(virtual or real) pho ton acts j us t like a hadron inasmuch as i t appears to

2 2
have i ts own ( q dependent , o r q -0 ) fragmentation products , in its
direction , the hadron fragment ing also in i ts charac te r is tic way .
This of course makes a nice union with the idea o f ve c to r meson dominance ,
2
that a free pho ton ( q -0 ) h as a certain reas onab le prob ab i lity to be a vir tual
vector meson and as such woul d behave in hadron collis ions like a hadron . We
note now we shall � have to de termine with what p rob ability it looks like
2
a hadron and how this var ies with q , for in any event i t , as a whole , should
act j us t like a hadron does in v + m col lis ions .
In the center o f mas s p ic ture (and also in the spacelike q figure ) there
are terms o f coup l ing in whi ch the pho ton f i rs t divides into par tons on the
way in , for example one fast one slow , and these s l ow partons interact or
annihilate with the wee partons o f the hadron . Thus the p i c ture that the
incomin g photon looks with s ome amplitude like partons itsel f is reins t ated .
x • -q /2Mv is finite) the cont ribut ion o f

2 2
As q rises (and cert ainly whe re
such diagrams falls away and only the dire c t coup ling t e rm o f pho ton s c a t tering
a parton of the hadron remain impo r tan t .
x
2
Con t inuity o f large q and small region
x As
Finally we shall match our finite
2
q region to our small region .
2
we have done before we shall suppose when v is very large and -q large but
2
-q /2Mv small the limi t may be taken in either order - i . e . , we can ge t the
+ m x
2
res ult e i ther from our fini te q , v fo rmula or from f ini te , but small ,
v + m formula .
Thus ( S S . 1 ) mus t agree with (5S . 2 ) for large q2 • This is easily done ,
2
the results agree if only we add the res ult : •
D , q 2 Dr for large q That is :
y
2 of
The fragmen tat ion p roduc ts o f a pho ton of large -q become independent
(We are in all casea normali zing to the total cross section
2 2
-q as -q rises .
which is varying , as l /q , of course . )

2
Partons as Quarks
Lecture 56
Partons as Qu&rks
We could now go on to discuss various models of what quantum numbers
partons carry , but we shall limit ourselves to one example , the one that is
most interes ting . The s tudent should try other examples , such as the Sakata
mode l , to see whether we can e lilDinate them by experiments now done , or proposed .
We shall suppose the charged partons come in six varieties , three plus
their antiparticles . The three called u , d , s carry the quantum numbers of the
three quark s tates (of the low energy quark jllO d el) . This we swmnarize by :
El . The charged partons are guarks . Mos t of our previous assumptions
were guided , or we thought they were guided by field theory or considerations
based on high energy experiments . This , of course , is not , it is an inspired
guess . But i t is also contrary to what can be true if the field theory is too
ordinary . For in such a theory there would be a base s tate o f quark number
one (and non-integral charge in a localized wave packet) and , in view of the
conservation of quark numbe r , some eigens tate of the sys tem of quark nuinber
one is expected . In other words we expect to see real particles with quark
quantum number s . They have not been seen . I t is possib le to imagine that
they have very large mass - but this makes it very hard to take all the
259
previous assump t ions about all parton interact ions limi ted to the GeV region ,
etc . The re may be some way to reconc ile all this - i t is one o f the mo s t
intriguing of theoret ical p rob lems . In o rder to emphasize it I will make
anothe r unnecessary assumption that I will not use bu t I add to remind you
o f the p rob lem .
E2 . Phys ical quarks do not exis t . If you prefer to replace it by
"physical quarks have high mas s " , go ahead - you s t i l l h ave theore ti cal work
to do reconciling it with El and the res t of our assumpt ions . Of c ourse E l
may be wrong - one o f the mo s t impo rtant expe rimen tal j obs o f the future i s
to find ou t whe ther El is indeed corre c t or impos s ib le and so w e should work
out as many tes t able consequences of it as possible . So far we have , as
discussed in lec ture 32 , come to the conclusion from experiment that in any
case :
E 3 . Neut ral partons als o exis t . What they may b e like we do not now
know , exce p t perhap s they may not be vecto r (b ecause the �.P degene racy is
not l i f t ed ) .
Although the problem of reconcil ing E l , E2 and f ield theory might be
very difficult , it appears (at least at first s igh t ) to be no t at all d i f ficul t
to re concile E l , E2 and the other assumption s we have expl icitly made about
parton and hadron dis tribut ions :
A careful review o f our assump tions shows this . The re is poss ible doub t ,
as al ready ment ioned , i n B l ( interact ion only between partons o f small
re lat ive y) but this is only used to make the later assump t ions mo re plaus ible
and maybe replaced by these lat ter assumptions .
A place whe re there is an especially in tere s t ing , but no t incons istent ,
conclus ion is in connec t i on with " righ t -moving quantum numbers " (discussed
in lec ture 54) for the parton function D(lR and whi ch should now be non-integral .
These numbers are defined s tatis t ically as the ave rage over all events -
al though each event mus t give inte gral values the average , of course , need no t
be . Fo r example i f we know one quark ( and no an tiquark) is sent to the righ t ,
the mean numb e r o f baryon less antib aryons found on the righ t should ( at
ext remely high ene rgy , a t leas t ) be + 1 / 3 .
The arguments lead in g to this conclus ion are no t obviated by the fac t
that the partons do not have integral quantum numbe rs . Imagine for a very
Partons as Quarks 26 1
large tnP tha t the quarks are dis t ributed in the ( f inal s t ate) wave funct ion
in a long plateau: ( the left end o f the plateau is gene ra ted by a long chain
* *
of cascades via terms l ike a a a f rom the ini tial s ingle quark)
v v
q Q 9
4J_ v
.o.\
11 K 11 11 p K 11
Then in turning into final hadrons various bundles of quarks go together t o
make legitimate hadron quan tum numbers . In do ing so they take combinations
of quarks over a finite range of y as il lus t rated above . (The overall total
triality mus t be zero , o f course , fo r the ini tial s t ate has zero triality ; in
our illus trat ion the initial s t ate has baryon numb e r one . ) We assume tha t
there is some non-zero probab ility pe r dy to p ick of 3 quarks (or 3 ant iquarks)
to make a baryon ( o r antibaryon ) . Likewise there should be a f inite probab i l i ty
to pick up s t range quarks . It is then seen that as a s t a t i s t i cal mat te r with
a sufficien t ly long p lateau in y ( sufficiently long Markovian chain) the quark
number ( and s t rangene s s ) becomes randomized and the central region of the
p lateau i s neu t ral on the ave rage in these variab les . Thi s means tha t the
r i gh t and le ft mean quantum numbers of the f inal hadrons approach a con s t ant
that depends as we have supposed on the ini tial righ t or left quark charac ter .
It doesn ' t make any d i f fe rence exac tly where you cut the plateau in deciding
wh i ch is right o r which le f t , as long as you cut somewhere near the middle .
Thi s result is so interes t ing and i t s experimental verif ication would
rep resen t s u ch a . d i r e c t measure o f the suppo sed non-in tegral quan t um numb e r s
o f the parton quarks that we should s ay some words ab out i t s pos s ib le ve ri-
f ication . First in e le c t ron produc tion in general we do n o t have a single
type of quark thrown to the left so the beauty o f the res ul t would be con fused
by h aving to first know the w (x) by one method or ano ther (see me tho ds below) .
a
On the other hand for x near 1 we have been le d by experiment (ratio o f
v W2n to v w2p ' lec ture 31) to suppose that only u quarks s urvive - hence i f
that i s t rue , near x • 1 our l e f t -mo ving quark ma y be a pure u quark .
Another way t o insure pure quarks is by neut rino s cat t e ring which we dis cuss
be low .
Secondly , although it may be easy to form pions in the plateau region

it may be harder to form K ' s and still more di fficult to form baryons (becauae
of their masses ? ) . If this is the case we should need a very long plateau
indeed to ge t equilibrium in baryon number, although hyperon charge might be
easier and isospin eaaies t of all . In the energy ranges availab le to experi -
men t therefore I would expect the quantum number rules to work best for
isospin , next for hypercharge , and b aryon number las t (i . e . requiring the
largest energy) .
Tiie eas ies t would be isospin . For example we would expect that the
left-moving z isospin : t I ziNi ' the sum of each product hadron moving to the
left ( for ex�le in the system with qµ pure spacelike , or maybe the center of
mass) , will per collision , for a given x be given by
I IziNi • O· (t (u (x) - ii (x) � - t (� (d (x) - d (x» )] / fep (x) , (5 6 . 1 )
that is , i t will rise from zero for small x (where u (x) • u (x) etc . ) , never
rise above + 1/2 (or fall below - 1/2) and approach + 1/2 for x near 1 (where ,
we think u (x) dominates ) .
'nlis relation can serve as can a number of others of this type in one o f
two ways , In one case we migh t suppose u (x) , u (x) e tc . , already known from
some o ther method - such as neutrino scat tering explained in lecture 33 - or
analysis o f p + p + µ+ µ - + anything , or by perhaps one of the equations developed
further on . In that case (56 . 1 ) is a quantitative prediction to tes t the
consis tency o f all the ideas that partons are quarks . Alternatively it can
i tself be used to give further information on the six separate functions
u {x) , ii (x) e tc . , so that they may be separately de termined . niese could then
be compared to the results of other methods of getting them , but do not serve
directly as a check of the model .
However , even without complete separation of all the functions a relation
such as may be used to check the quark mode l becauae of the exis tence
�
(55 . 1)
1 l
of sum rules (Equation 3 1 . 2 ) such as
J
(u (x) - u (x) ) dx . 2 ,
0
(d (x) - d (x) ) dx . 1
so the integral over x of the nume rator of (56 . 1 ) should be 7/18 .
I do not know at this time which kind of experimental information will

become available first and so my general discussion suffers from a confusion
in analyzing these theoretical expectations as to which comes first the
horse or cart . There are a ve ry large numb er of such rel ations whose
o rganization is thereby made difficul t . I will there fore me rely indicate
s ome of the gene ral rela tions expe c ted from our theo ry and leave the choice of
the best way to use them o r comb ine them to compare to expe riment up to you .
Le c ture 57
Parton& as Quarks ( con tinued)
There is ano the r way to ins ure that one ge ts a pure quark o f one type
recoiling in the final s tate , and that is wi th deep inelas tic neu trino or
ant ineutrino scat tering , as we discussed in lec ture 3 3 . We assume of course ,

*
the usual GJ J weak coupling with the hadron part of the weak current given
µ µ
in terms of quarks as Cabibbo sugges ted . That is , we exp licitly assume :
E4 . The weak interaction is via the current Q ' yµ (!±l- 5 ) Q where Q is the
Dirac operator for a u quark and Q' that for a "Cab ibbo quark " - a qu!-��
which is d with amplitude cos e and a with amplitude a ine ( aine � . 2 4) .

c c c
We shall assume for our dis cus s ion that the current in te raction ia
point-like , but that is a mat ter fo r experimen t to de cide . This ve ry
in teresting que s tion is , howeve r , bes ide the s cope o f our discus s ion ; after
i t is determined the same kinds o f ques tions and remarks will apply to
products gene rated by the aforemen tioned current operator . In all cases
we can analyze things as i f they were the e f fe c t of a vi rtual W mes on field
W of momentum q (generated by the lep ton) coupled with the current in E4 .

µ µ
Just as be fore (Equation 3 3 . 2 ) the to tal s quare mat r ix element of J can
µ
be expressed in terms of w , w , w so the p roducts , su111111e d ove r spins and
2 1 3
angles , at leas t , can also be so analyzed . That is the to tal w , w , w can
1 2 3
each be split into part ial w , w , w for special types of products . We
1 2 3
discuss only the s caling region . We have s een vw i s rel ated t o w and the
2 1
relation should hold fo r products also . But more in te res ting we no tice d
f � -f;
p p
(j us t the s caled func tion 2Mw -2MW f o r an tineutrino s cattering on
1 3
protons ) was j us t pure ly u ( x) . Hence the same holds for the products , ie . ,
the probab ili ty of a product in w (determined as the app ropriate coe ffi cient
1
in the variation of cross section wi th labo ratory neut rino angle keeping q , v
f ixed) minus tha t in w for vp s ca t tering are purely the products for a u quark
3
being kno cked backwards (forwards is a pro t on les s a d q uark with p robability
cos ec - the latter can in f irs t approximation be negle cted as s in 2 ec is only

2
.06) . Thus by s t udy ing the products to the le f t in this comb ination we are
s t udying the fragmen tation p roducts , D ( z ) , expected of one s ingle quark , a u

u
quark in fact .
By ch oosing o ther comb inations we can select reco i l quarks of different
types . For example , f �P+f�p for neutrinos on p ro tons gives pure u quarks
and should give var ious p ro ducts with probab i l i ty D u ( z ) u (x) ; a fixe d dis
trib ution independen t o f x, the total cross section being u (x) . Again
f � -f �
p p
for neut rinos on pro tons gives nearly a pure d quark (cos ec
2
• . 94
d quark, s in ec •
2
. 0 6 s quark ) .
From these dis tributions the total quan tl.DD nl.DDb ers of the quarks can be
de te rmined . In this way alone cert ain models can b e e liminated , but o thers
can not . For example we cannot distinguish the quark model from the th ree -
triplet model where the re are 9 partons (and 9 ant ipar tons ) in s e ts of three .
A, B , and C each set having three s t ates like quarks wi th various integer
quant \DD numb ers . Thus what would come out in the expe riment on £ -f for
1 3
v p scat tering , where we expe ct a pure u quark in the quark mode l , woul d ,
i n the three-triplet model , be a + 1 / 2 isospin parton b u t e i ther o f type A o r
B o r C wi th equal p robability s o the mean charge o n o ther nl.DDb ers can b e
1 / 3 integral and j us t equal t o the u value , f o r that is how the integer charges
A, B, C were chosen .
+ -
for (O the r experiments , s uch as e e + any thing or
corre lations of left and right s i de dis integra tions migh t dis t inguish these
models . )
Product predict ions
There are a large number of predi ctions implic i t in the re la tion that
the dis tribut ion o f a given final hadron in the left (photon) direct ion ( for
convenience no t no rmal ized) is in gene ral given for deep inel as t i c ep s cat tering
by : (Equation 5 5 . 1 )
(5 7 . 1 )
where D ( z ) e t c . , are the distributions of the produc t in ques tion for pure
u
up quarks , e t c . There are s ix func tions in general so it is di fficult to
analyze un less u (x ) et c . , were all already available . Howeve r , by taking
cert ain comb inations of measu rements fewer functions are involved . We
illus t ra te this wi th an example .

+ +
+ w w
Suppose we ask to produce a w , call the dis tribution D (x , z ) and D ( z )
u
+
e t c . By iso spin re flect ion the p robab i l i ty a u yields a w is the s ame as
-
that a d p roduces a w ; and by charge conj ugat ion again the s ame as that a u
In this way we see for w product ion there are real ly only
-
produces a w .
three independen t funct ions

+ +
w
o!! • o!!
w
d
•
D • D
u d u
+
o!! o !!
+
w � -
w
D D
d u d u
w w
o!! os!!
+ +
D - D -
(5 7 . 2 )
s s s
+
In fact i f we mea sure the number of w minus the number of w at a
given z i t (Equation 5 7 . 1 ) al l reduces to one func tion :
D
w
\x , z ) - D
w
-
(x , z ) • A(z) [} (u(x) - u (x) ) - } ( d (x) -d (x) � (5 7 . 3)
( for virtual y on p ro ton )
+ -
where A ( z ) • D
w w
(x) - D ( z ) . Thus we expec t the dis tribution (probab i l i ty
u u
as a func tion o f z) to be the same for all x . As we vary x we can determine
4
9 -
(u (x) -u (x) ) -
1
� -
d (x) -d (x) ) , within a cons t an t . This is j us t as in 5 6 . 1 , but
we do no t have to measure ove r all z to in tegrate , and meas ure other parti cles
+
as we ll . A me re measuremen t of w and w- at some convenient z would be enough .
The abso lute coe f f icient can be de termine d in two ways , ei ther from the s um
rules (Equations 31 . 2 ) , or by the hypothes is that as x + 1 only u (x) survives
a +
4
as 9 u (x ) + f
ep
(x) known function as x 1. Additional in forma tion would
come from the same experiment on the neutron , o f course , (we ge t 9 (d-d ) +
4 -
9 (u-u) ) .
1 -
+
The sum o f the number of w and w does not give us much that is new
about the dis t ributions , but we can roughly p redict its x dependence
w+ w- 11'+ w- � -
D (x, z) + D (x , z ) • (D ( z ) + D ( z ) )
u u
\? (u (x ) + u (x) ) +
+9
1 -
(d (x) + d (x ) ) + 9
1
- o"
s
-+-""-- (s (x)
11 11
+
J
s (x) ) (5 7 . 4 )
Du + Du
The express ion i n curly b racke ts i s the same a s fep (x) except for the
coefficient of the las t term (which should be simply 9 1)
. Howeve r , that term
is probably small ( for not only should s , s be less than say u , u in a proton
but also u is enhanced by 94
relative to 9 1)
so the distribution of 11 + plus 11
is p robably nearly independent o f x , and if normalized to f ep (x) depends on
z only .
Arturo Cisneros has suggested an hypothesis which we explain in mo re

detai l below , which amounts to assuming that near z • the functions D ( z)
1
a
fall off with various powers of (1-z) and in particular that as z + 1 the
+ + "+
function Du11 is much larger than o!u or Ds •
This makes the coe fficients
of (5 7 . 3 ) , (5 7 . 4 ) e qual as z + 1 . Thus it means that in this region the
+
probability of finding a 1! as we vary x is a direct measure of u (x) + d (x) ,
-
and the probability of finding a 11 measures u (x) + d (x) in the same s cale .
This is still ano ther sugges tion o f how the individual de termination o f the
functions n (x) may be facili tated . ln fact , this hypothes is , if true ,
a
permits a determination of the six functions u (x) , u (x) etc . (up to an
overall numerical cons tant) , by measur ing the dis tribution functions for
charged mesons only near z • 1, for both proton and neutron . If only the
proton is used as a target the n (x ) cannot be de termine d without making
a
measuremen ts of neutral mesons which is difficult experimentally .
We can do similar things for the p roduc tion o f other particles , for
+
K
example K-mesons . Here there are six independent functions Da (z) , the
others are obtained by isospin reflection or charge conj ugation . (For example ,
+ -
Dl<u • D-
K
• D
Ko DK° The s tudent can verify that for v on pro tons
d d •
u
+
if we measure particles at a given z to the lef t , if N ( z ) is the number o f
K+ e t c . , w e find the following resul ts : The isotopic spin di fference depends
on one function , with the same x dependent coefficient as before :
0 {, 0 ) (
N+-N 0+N -N - • \D+u-Du0+Du-Du- ( z ) 9
4 1 (d (x) -d- (x
(u(x)-u- (x) ) - 9 ) >) ,
0
another comb ination (which does not require K and K° to be distinguished)
als o factorizes into one fun ction o f x, one o f z :
ep
but the las t factor in curly b racke ts is like ly to be close to f (x) for any z .
F inally the fourth re lation , involving two functions is
(a meas ure of hypercharge , b ut unequally sens i t ive to s t r ange and non-s trange
quarks s ince we do not assume su invariance ) .

+ +
3
Cisneros ' assump tion E6 (below ) here means that , as z + 1 only Du and Ds
survive , call them � . s respe ctively . Then a measure of K mes ons to the le f t
as z + 1 is a dire c t measure o f various comb inations o f th e u , d functions .
ttU + S e ad + s ii
+ 0
Number o f K Number of K
au + S s ttd + S s
- 0
Number o f K Number o f K
I n this , a s i n a l l cases , data o n vn gives add i tional info rmat ion, change
u (x) ++ d (x) and u (x) ..... d (x) in the formulas .
We have not di s cus sed the right dis t r ibutions , but the re are re lat ions
here too for various experimen ts . We mention only one as an examp le .
Llewellyn Smi th ' s sum rule (Equation 33 . 6 ) negle c t ing s in ec f (x) -f

2 vp vp
=
3 3
·-6 ( f (x) - f (x) ) works for eve ry x as a to tal cross section . We now
ep en
see that the obj e c t s p roduced at the right ( in the hadron fragmentat ion
region) are the s ame on bo th s ides for the pro ton but no t for the neutron -
for the en experiment we mus t observe the isospin re fle c ted p roducts . If
this is done , the relat ion holds f o r the part i al cro ss sections f o r any
products to the right i f the le f t p ro ducts are no t observe d .
Finally to b ring a l l the hypotheses w e have made ab out par tons together
en ep
into one list we note finally our sugge s t ion Chat the f /f rates s t rongly
suggests that when a pro ton has a quark near x • 1 and a rema inder of small
momen t um , that quark is a u quark . We t ry to generalize this to any b a ryon
of the 56 mul tiplet in su language meaning our su only q ual i t atively , not
6 6
exactly quantitatively .
!! · The ampli tude that a b aryon of the fundamen tal 56 is a parton of
x near 1 and a remainder o f small x varies as (1-x) Y , the lowes t y o ccurs in
the case that the part on is a quark , hence � in su 3 (no t an antiquark) and
the remainde r i s CJL........Sh!._Q) _ �c;! (]

__ making the octe t .
A member o f the decimet A, has a smaller probab i l i ty o f having a quark
near x • 1 than does the octet . We have already discussed the imp lications
for t o tal cross sections . It has imp licat ions for produc ts also , of course .
We also assume that if a s tate is a pure quark to the left i t has an
amplitude to be a baryon to the left with z nearly 1 wh i ch is proportional
the the chance the baryon cont ains a fast quark near x • 1 of the s ame kind .
There are then many imp l ications for produc ts - and we have ment ioned some
that come from u (x) being larger than all the o thers as x � 1 . Th ere are
o the r s , of ano ther type , for example (Cisneros , private communication ) in the
+ -
e e � hadrons s ince uu has four t imes the probab ility of d d the chance of
producing a proton with x near l( in the cen ter o f mass ) and any thing else is
four t imes the p robab i l i ty o f p roducing a neut ron .
If E5 is correct we should like to assume s ome thing analogous for the
mesons . The analogous assumption is tha t when one quark takes mo s t of the
momentum it is of a type that the low energy quark model suppos es the meson
to be made o f . We therefore ( in agreement with Cisneros) assume:
�· Fo r mesons wi th one par ton near x • 1 it is a q uark � and the
remainde r is CJ) wi th s pin 1 / 2 , p lus the charge conj ugate with the s ame
p robability ( for the charge conj ugate i t is Cl} near x • 1 and the remainder
is Q) .
We have built a very tall house of cards making so many weakly-b ased
conj e c tures one upon the other and a great deal may be wrong . (P robab ly
the weake s t is C6 - same p lateau for gap and hadron - but i f it we re wrong i t
does not alte r t h e thrus t o f any o f the others - j us t i n t h e ope rator expres sions
we shall have to be care ful to use the righ t plateaus . ) Neve rtheles s this is
the bes t guess I can make now - and we can try to us e them has working hypo theses .
Probably the greates t challenge to experiment and theo ry is to ge t some
evidence of quark quan tum numbers in h i gh-energy collisions . The low-energy
quark model , good as it is , is not enough , there is always lingering doub t
that the regularities observed h ave some ent i rely di f feren t b as is or are ,
in part , accidental . The es tabl ishment of evidence for the quark mode l
(and we have indicated very many ways - both in the last few lec tures as well
as earlier - (Llewellyn Smith ' s sum rule (Equa tion 3 3 . 6 ) , the spin sum rule
for g 1 p - gln e tc . ) by high energy experimen ts would confirm at once the reality
o f the regular ities interpre ted by the low energy quark model . This would
make firm a conj ecture of deepes t significance to understanding high energy
physics - the importance of quark quantum numbers .
Supposing for a moment this is done , then the next serious ques t ion will
become theoretical - what exactly is the relation of the quark qualities at
high energy and those at low energy . The "partons as quarks " model does no t
imp ly the low energy model ( i . e . why are the wave functions not more complicated
involving quark antiquark pairs ) nor vice vers a . At present their relation
would not be understood . To s tart working on this now will take a little
courage - you might was te your t ime - maybe partons as quarks will not be
confirmed . I f you do start , poss ibly one place to start might be to think
about low energy matrix elements like � � p + y in a fas t moving sys tem in
which all (or some) momen ta are of the order P so parton wave functions can
be used . (We have one relation of this kind in Bj orken ' s sum rule for g ,
Equation 33 . 16 . )
Finally i t should be noted that even i f our house o f cards survives and
proves to be right , we have not thereby p roved the exis tence of partons . The
final resul t of our considerations has been to describe the result of the
J j p> (for large v , -q 2
1J
ope ration o f a curren t on a proton s tate • 2Mvx) as
a linear combination of operators like OaL E R ( M j VAC , creating final out-

p-a , x )
going hadron s tates only . It might be wise to follow this out formally
wi thout mentioning partons (analogous to the way Gell-Mann and Fritzs ch
describe parton results for total inclusive cross sections in terms of
conmutation rules for quanti ties , curren ts , defined in general whe ther par tons
"exist " or no t) .
From this point of view the partons would appear as an unnecessary
scaffolding that was used in building our hous e of carda .
On the other hand, the partons would have been a use ful psychological
guide as to what relations to expe ct - and i f they continued to serve this
way to produce o ther valid expe ctat ions they would of course be gin to become
" re al " , possib ly as real as any other theore tical s t ructure invented to
des cr ibe Nature .
At any rate we shall see . It is good to have s ome thing to look forward
to.
Appendix A. The lsospin of Quark Fragmentatior
Products
APPENDIX A
The Isospin of Quark Fragment ation Products
The discuss ion (Lecture 56 ) lead ing to the i dea tha t addi tive
quark quantum numbe rs could appear as mean to tal quant um numb e rs o f
products moving i n one direction i s s urprising - especially when it is
no ted that what ho lds fo r 3-isospin holds also for any o ther component
s uch as 1-isos pin or 2-iso spin (al though of course in practice they are
nearly imposs ible to measure) . It looks like an i sospin 1 / 2 object
co uld be represen ted by a group o f isospin 1 ob j e c ts (e . g . pions ) -
wh ich at firs t seems impos s ible - except that we have an inde f inite
number of s uch obj ects .
It is there fore of inte res t to make a very s imple s pecial ma the-
matical mo de l , to show that indeed s uch things can be done in p r inciple .
Thi s is especially important when it is reali zed tha t our previous
a t tempts a t mathematical formulation cannot be comp lete and mus t be
looked a t as mere mnemonics (see note in Le ct ure 55 on the D and E
operators) . This l i t tle mo de l may help by i ts example to lead to co rrect
pos sib le fo rmal exp ressions o f our ideas .
ln this mo de l s uppose quarks carry only iso spin 1/2 and hadrons are
271
only pions of isospin l - made of quark antiquark pairs .

Imagine that we s tart with some current annihilation (analogous to
+ -
e e but in more general isospin direction) ini tially disintegrating
into a pair o f quarks Q , QB (a , S are su2 indices fixed in this problem.

0
Immediately after interac tion :
-P p
after Hamiltonian operates : N quark pairs in singlet s tate
makes hadrons
(pions )
of type
Next the action of the Hamiltonian is to p roduce pairs of quarks

in a singlet s tate uniformly spaced in y space - a typical one is Q Qi
i
summe d on all i equally . The number of such pairs N is then proportional
to tn2P which we take to be very large . (One could also ass ume the
number dist ributed via Poisson with a mean N e tc . , b ut we avoid comp li-
cations which only serve to confuse our point - choose N fixed . )
Next this row o f quarks is assumed to conve rt t o pions by a simple
rule , each pion is formed by a pair adj acent in the y space . Thus (in
figure) if the firs t new s ingle t pair had index i , the next j etc . the
firs t n is formed from an antiqua rk index S and a quark , index i ; - the
next by an antiquark index i and a quark index j ; - e tc .
To des cribe the isospin s tate o f a n we us e an isospin 3-vector V .
Thus if the n is a neutral pion n° we have V with only a z component ,
v - (0 , 0 , 1 ) . For a n+ we have V � ( l , i , O ) e t c . The ampli tude that an
•
12
antiquark of index y and a quark o f index 6 form a n characterized by
vector V is then p roportional to the yo matrix element of the two-
by-two matrix o • V where o are the Pauli mat rices . Write this as < y l o V l o >
· .
(We work in relative amplitudes and probabili ties leaving overall

normalization to the end . )
Thus the total amplitude t o find the n ' s in di rections v1v2 • • . V is
N
Amp • � <j j o · V3 • • • • • • • • o • VN l a> , the s ums on
i,j . • •
Appendix A . The lsospin of Quark Fragmentation Products 2 73
i, j e t c . , being because the newly fowid QQ pairs are in s ingle t s tates .
This is o f course
(Al)
Having the amp litude ( in an su 2 invarian t fonn, o f course ) we can
ask many quest ions . The re lat ive probab ility o f finding any con figuration
Ea (a . v1 ) (a . v2 > . . . (a . vN >Pa (a . v/> . . . (a . v2 *> <a ·v/3

is the square
P (V1 . . . v > • rr
N
(A2)
where pa ' P a are 2x2 dens i ty ma tr ices co rresponding to the s t ates ,
pa • a+a · A , Pa • b+a · B . l a > , p a • l a > <a l ;

(�)
say For s tate for example i f
a is 1/2 in z , l a> • then Pa • ( � g) • f + f az .

Thus in that case a • 1 /2 , A • 1/2 , A • 0, A • 0. I� a is a s t a te
z x y
spinning in the direc tion o f a wi i t ve c tor then A/a is that wii t ve c to r .
The expe cted value- o f z isospin i s eviden tly A / 2 a .

z
Now suppose w e obse rve the isospin charac ter o f only a limited
number o f pions , s ummi n g over the charac ter o f the res t . In fac t we
shal l do two cases ; s ummi ng ove r all to ge t the normaliza tion o f our
prob ab il i t y , and s ummi n g over all but one , as i f we s t udied p ro duc ts
w number k + anyth ing . w
��
In any case a sum over an unobserve d means a
)
sum on V over 3 pe rpendi cular values , symbolized by ·· · (A · V ) ( B . V)
A•B . We need the formula
"""
i:...., (o ·V+1'r ) (a+a · +B) (o . +v) • Ja -o • +B (AJ)
v
wh ich is eas ily ve r i f ie d .
Now f ind the no rmal iza t ion "t •L P (V 1 . . . VN ) . On summing

v v . . .v
12 N
over v 1 , Pa is conve rted from b+a • B to Jb-o · B (AJ) . Next s um v 2 and

by
terms to 3Nb+(-l) Na . B whose

2
i t is conve rted to 3 b+a - B . Con tinue for N
trace w i th pa gives ( suppos e t r ( l ) • l)
'71 •
N
3 ab + (-l)
N
(A .B) (A4 )
N
For large N this is almos t exac tly J ab and so we shall divi de by
this to get normalized p robab i l i ties . Thus the normalized prob ab i l i ty

th
that the k p ion is o f type V is
k
Pk (Vk) •+
3 ab I v 1 • . .
Tr E8 (a 0V1 } . . • (a ·Vk ) • • • (a · V } p (a ·V/ )
N a • • •
VN
except Vk
(cr · Vk
*
> . . �
. (cr · V 1
Now again we can sum over ... v 1 , :t-

v 2 up to :t-vk-l converting p B to 3k-1 b+(-1) k-1
(a • B) ; and also sum in an analogous way first on V ' then V
N N-l etc . to
(cr · ) . Thus the net is (scale t r ( l) • 1 )
v N -k N -k A
k+l to convert p a to 3 a+(-l)
from which any question about the one particle dis tribution func tion can
now be answered. For example , sum over all possibili ties for Vk and check
normalization ( for large N) .
+
The mean pro b ab ili ty th e k t h meson i s 11 + is Pk Tl ob taine d b y se t t i ng
11
... 1
a ·V • � (a x+iay ) in
put cr V
(A5 ) . For • = l (a x -icry ) •
12
Thus the
12
mean isospin of the k th hadron is PkTl - Pk!!-
+
-icrx cry
• a term like (A5 ) with
... :t- *
a · Vk • • •
a · v k replaced by · · ·
(- r
o lus icry · · · cr x . No te first us ing
this is 2 a z and hence gives � t B/b ,
-l
only the one coming from p
a
likew ise the term in cr • A only contributes if cr is in the z direction .
We find
(A6 )
This resul t confirms al l our expectations . Firs t if k is in the middle

of the plateau, o f order N/2 , not near either end , then P 11 - P11 � j-
+ - N/2
which is very smal l . Thus the plateau region has become neutral .
+
Pk 11 - P k11 can only avoid being small if either k is small (i . e . near
the S end) or near N (i . e . near the a end) . In the former case neglecting
terms of order 3 -N we have
(A7 )
if k is fini te , near the B end , a result that depends only on B, the

quark at that end, and practically (as N -+- �> not at all on the quark
at the other end. (Evi dently for k near N we have the exact opposite
resul t . )
A ppendix A . The Isospin of Quark Fragmentation Products 275
F inally the total z-iso spin quantum number of all the pions to the
+
11
{P
k
the z-component quan tum number of the le f t -moving quark! The sum on k
is to be taken from k • 1 to near N / 2 to hold only the l e f t move rs , the

-N/ 2
cont ribution from the a-dependent term as only of o rde r 3 , and the
sum is of the s ame orde r ; the same as if we s umme d the left term above
to in fini ty . Obvious ly the result is insens itive to exactly whe re in
the pla teau we s top , it is only necessary that we s top the sum on k at
some point far from ei the r end .

+ 0
{One might further notice that P 11 + P 11 - 2P 11 is zero ,
k k k
-N
negle ct ing 3 , for every k . Th is is a consequen ce of the fact that
the quark h as total isos pin 1 / 2 , and not higher . We leave to the reade r
to show tha t this generalizes to the re al case and coul d in principle
become a tes t of the to tal isos pin character o f the partons . Th a t is

+ 0
11 11 11
N + N - 2N is zero for any z to the le f t i f the left movers
{z) {x) {z)
come f rom a single parton {of any kind or superpo s i t ion ) and if the
isospin of partons i s ei the r zero or one-hal f . )

Appendix B. A Test of Parto ns as Quarks
APPENDIX B
A Tes t o f Partons as Quarks
Fo llowing J . D . Bj o rken we note tha t the sum o f neu t rino and
an ti -neutrino cros s -s e c t ions can be p redicted fai rly closely in terms
o f quan t i ties already known . Since measuremen ts o f thes e t o tal cross
sect ions are the eas ies t tes t o f quark quan t um numb e rs , we give an
analysis here .
Le t us me as ure all cro s s s e c t ions in uni t s o f G s/ 2 � where G is
the Fermi cons tant and s the s quare of the center of mass energy . For
nucleons , then , o ur unit i s GEM/ � where E is the laboratory energy .
The t o t a l c ross s e c tion of a neutrino with a sp in 1 / 2 parti cle is
2. Wi th an an tiparticle it is 2 / 3 . Hence on a p ro t on the cross s e c tion
f
is
l
1 -
2 x (d + 3 u) dx
0
the fac t o r x coming b e cause the cross s e c tion varies w i th s . Fo r
neutrons w e rep lace d b y u , e t c . , s o t h e mean neu trino cross se ct ion o f
a nucleon i s
2 76
Appendix B. A Test of Partons as Quarks 277
J
1
-
- x (d + u + 3
1 -
(d + u) ) dx .
0
the anti-neut rino cross sec tion is
a • f (a vp + avn ) - f x (d + u + t (d +
u} } dx
0
S ince d, u are pos itive , but undoub tedly less than d , u, we see tha t
o/a mus t be s ub s t antial ly les s than 1 b u t gre ater than 1 / 3 .
Th e sum i s
J
1
• x (u + u + d + d) dx
a +a
t
0
However , in tegrat ing th e s um o f (31 . 3) and ( 31 . 4) we have
J � J } J
1 1 1
x ( u + u + d + d ) dx + x (s + s) dx
ep en
x(f + f ) dx -
0 0 0
J
Experimen tally thi s in tegral is 0 . 31 so if we could forge t the
• • 0 . 74 .
- 4 9 -
integral x ( s+S' ) dx we wo uld have a+a 3 · 5 ( . 31) Rut s+s
mus t surely be le as t han u+ii and d+d and , when weighed by x, s urely
much less . It would be hard to manage to make inclus ion of the las t
term p ro duce mo re than a 10% e f fe c t . Thus we have a ve ry s t ringent
tes t of our parton quark mode l : a+a canno t exceed O . 74 and ye t almo s t
s ure ly cannot fall be low 0 . 74 b y mo re than 10% .

p p n + n
One can also cal culate upper limits fo r a v + av and a v av
ep en
separa tely (us ing o ther proport ions of f and f ) ; they are 0 . 64
and 0 . 84 res pe ctively .
These numerical e s t ima tes mus t be revis ed by a few percen t , for we

2
have neglected sin e . They are val id only at asymp t o t i c ene rgy , o f
c
course , b u t T . D . Lee has pointed out t h a t e le c t ron d a t a indicates tha t
this shoul d only requi re a few GeV .

INDEX
A Ad l e r ' s sum rule , 155
Anoma l o us threshold , 122-123
65-66
B B j o rken ' s s um rule , 158
c Commu t a t o r func tion , 2 9- 30
Commut a t o r s in momentum space , 1 82-190
Comp ton s c at tering yN � 'yN , 2 15 - 2 2 0
Conservat i on o f current , 5 , 1 3- 1 5
Conserva t ion o f general i zed curren t s , 18
Co t t ingham fo rmula , 201-203
D Deep inelas t i c s ca t t ering w i th sp in , 155-159
Deser , Gilb e r t , Sudar shan represent a t ion , 1 84-185
Di f fract ive produc t ion o f p, W, ;. 102-108
Dispers ion relations for p ion form

f a c to r , 119- 1 2 1
Dyson rep resen t a tion , 183-184
E Elas t i c term in sel f en e rgy , 2 1 3-215
E l e c t romagne�ic current (defined for

hadrons ) , 5
Electromagne t i c s e l f energy , 199-215
quark mode l , 206-2 1 1
F Final ha d ro ns in high energy pho ton

co l l i s ions , 2 50- 2 5 8
Form factors by VDM , 90-91
279
280
Form facto rs , h i gh energy , 140- 1 4 1
G Gell-Mann ' s ea ual t ime commutat ion

reJ.at1onS' 19
General i zed cur ren t s 16-21
c onserva t ion , 18
H High energy hadron-hadron c o l l i s ions , 237-249
Hyp e r f ine s p l i t t ing in hydrogen , 221-225
I Imp a c t p arame t e r viewp o in t , 71-72
Inelas t i c e l e c t ron nucleon s ca t te ring , 126-131
Inelas t i c neutr ino-nucleon s c a t t ering , 1 5 3-155
I s o s p in of quark f ragmen t s , 2 7 1-2 7 5
L L ight cone , s ingular i t i e s on , 21-22
L i gh t cone algeb ra , 1 7 8- 1 8 1
Light cone b ehavior o f commutat o r s , 1 7 3-1 7 7
L imi t ing f ragmentat ion 243-244
Llewellyn Smi th and G ro s s sum rule , 155
M Moment um carried by q uarks , 152-153
N Neut r ino-nucleon t o t a l c ro s s s e c tions 2 76-2 7 7
Nucleon e l e c t romagne t i c form f a c t o r s 114-116
0 O p e r a t o r s in ine la s t i c s c a t te ring , 167-173
p P a r t i c l e fragmen t a t ion , 239
Par ton fragmenta t ion , 252-253
Parton mod e l 132-140

Index
P arton mode l hyp o theses 2 29 - 2 36
Partons as quarks , 1 4 7- 15 1 , 259-270
Low ene rgy p ion photop roduc tion 31-41
Par ton wave fun c t ion 2 30 - 2 3 1
P ion ab s o rp tion 6 3- 6 4
P ion f o rm fact or , 118-122
dispers ion relations , lH-121
P i on pho toproduc t ion , 31-32 , 37-41
Propagator func tion , 29
Pseudo s calar me son pho toproduc t ion ,

hiv,h energy , 60-70
Q Quark fragmen t a t ion 2 6 0- 2 6 2
Quark f ragmen t a t ion products , 2 7 1-2 7 5
Quark f ragment s , isosp in o f , 2 7 1- 2 7 5
Quark mode l , 42-50
dynami cal relat ivi s t i c , 5 1- 5 9
dynami cal non-relativi s t i c , 5 0- 5 1
R Regge p o l e formula , 73
Resonances , 32-34
Resonances in s caling funct ion , 142-147
s a- channel res onan ce s , 7 6- 7 8
S ma t r ix , 4
S ca t t e r ing in the DGS rep resen t a t i on , 190-19 8

2 82
S eagull s , 13
S e cond- order e l e c t romagne t i c coup l ing , 7
S e cond order uni t a r i ty , 8-10
Shadowing in nuc le i , 109- 1 1 2
S ingul a r i t i e s o n the l i gh t cone , 2 1- 2 2
S p in s caling fun c t ions , 156-1 5 7
Sp in s t ru c t ure fun c t ions , 155
T t- channel exchange phenomena , 7 1- 7 6
Time ordered p roduc t , 9-10
v Vacuum exp e c t a t ion o f conmut a t o r , 2 4- 2 5
Va cuum exp e c tat ion o f t ime-o rdere d product

p roduct 25
Vacuum p o l a r i z a t ion due to hadrons 22 , 164-166
Valence quarks 152-15 3
VDM in yN _., 'llN , 96-102
V e c t o r meson dominance mode l , 89- 1 1 3
Ve c t o r mesons , 8 2 - 8 5 , 88-89
Vene ziano formul a , 78-79
w W e e p a r t on s , 232 , 234

Feynman, Richard - Photon-Hadron Interactions (Westview, 1998)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Feynman, Richard - Photon-Hadron Interactions (Westview, 1998)

Uploaded by

Copyright:

Available Formats

PHOTON" HADRON

AdvAm:ed Book Program

·- -.._.., A Member of the Perseus Books Group

Feynman, Richard Phillips.

ISBN 0 -201-360 74-8 (paperback)

Copyright© 1972, 1998

All rights reserved. No part of this publication may be reproduced, stored in a

Wcstvicw Press is a Member of the Perseus Books Group.

Cover design by Suzanne Heiser

Find us on the World Wide Web at

importance to anyone interested in understanding the development of quantum

1-5 GENERAL THEORETICAL BACKGROUND 3

First Order Coupling 5

6-8 LOW ENERGY PHOTON REACTIONS 31

Pion Photoproduction Low Energy (0 to 2GeV) 31

8-10 QUARK MODEL OF RESONANCES 42

The Quark Model 42

11-12 PSEUDOSCALAR MESON

Pseudoscalar Meson Photoproduction-Higher Energies 60

13-14 t-CHANNEL EXCHANGE PHENOMENA 71

t-Channel Exchange Phenomena 71

14-21 VECTOR MESONS AND VECTOR MESON

Properties of Vector Mesons 82

22-24 ELECTROMAGNETIC FORM FACTORS 114

Electromagnetic Form Factors 114

25-26 ELECTRON-PROTON SCATIERING.

Other Photon Processes for q2< 0 125

26-33 PARTON MODEL 132

Parton Model 132

34-35 TESTS OF THE PARTON MODEL 160

Angular Momentum in Parton Wave Functions 160

36-37 INELASTIC SCATTERING AS

Inelastic e p Scattering as Properties of Operators 167

38 LIGHT CONE ALGEBRA 178

Light Cone Algebra 178

39-41 PROPERTIES OF COMMUTATORS IN

Propenies of Commutators in Momentum Space 182

42-47 ELECTROMAGNETIC SELF ENERGY 199

Electromagnetic Self Energy 199

48-49 OTHER TWO-CURRENT EFFECTS 221

Other Quantities Involving T 11v 221

50-51 HYPOTHESES IN THE PARTON MODEL 229

Hypotheses in the Parton Model 229

52-54 HADRON-HADRON COLLISIONS AT

Hadron-Hadron Collisions at Extreme Energies 237

55 FINAL HADRONIC STATES IN DEEP

Interaction of Partons with the Electtomagnetic Field 250

56-57 PARTONS AS QUARKS 259

Partons as Quarks 259

The Isospin of Quark Fragmentation Products 271

A Test of Partons as Quarles 276

One very powerful way of experimentally investigating the strongly

permits a much finer control of variables , and probably decreases the

in virtual photon-hadron collisions the photon acts normally (i.e . , obeys

smn - <m , out I n, in> (1 . 1)

E S <n , in l = <m, out ! (1 . 2 )

First Order Coupling.

<M/j µ {q) / N> �2� <m , out ! Jµ {q) l n , in> .

That is to say (supposing we could measure the amplitude) we define in a given

'1 (q) mn = <m , out l Jµ {q) / n , in> (1 . 5)

I t is then our first supposition that this quantity d1 µ (q) mn depends

This is a strong assumption . I t has been verified most completely for

is described by the matrix

1 + i f a (q) J (q) d\ + higher order .