1r?" ' ‘
Via-:3
say
3?-

s. V.. .
Viv...”
y S

V .V

C
V
”f
q

t.
2
g‘:
i

7. 3,
i \\

' 1%.
i

31
l
1

i

6
i

E
a:

I. (... nu . V . , V V V , _ , , V is“!
V . . . .., . P! .V . t $95.1
unfit. Vain, . V . . , tbuaflhuk‘zfiifimwun .
:51. 2». V . . .. ,. .a~¥.V...fl ..fiumumt .
A V . . . .V V. V V . . ...w .hyfimkuio; flaunt? . .
5V. .V V V V. . it .1 a fit 3 . :21.
“V . . . V . V V Elmo“ Aflflfi 4.» a“ .\ 331‘
V Q V. ank‘ ‘8. 31 3..“h"; .
v 5.1 I fig... . 4v .
Sufitflvkfl‘tr .1?! «I! $kuui «fl V .
. V V, malt v. . ii
Ftumnxfiuéaw v.2 Em: yam 29$
2V «sump-Mn!» nun-i V V . .i . l. .3331!
. V 0‘. “a";

’f

. O
. ....
V VA 0:. a .14 ~34.
{Samumumflfi
V .. .11.}.tk‘lvvfl...
4.. V1.

W

nVV
.119! . 14.. . . V)
xx gunmnumwufiiz
Y"

‘lq
55.0 h .3 .. I...
Q." 31.... a ..tkd...

3V
1
i

‘9
3%

‘ .

Cr: .8.‘
tiff

£1

.
ntiVsL.
c

..
”'1“

. .35).. V .. .
“359x333“... . V . . . . V
. V asafziVViiVVVVVV V .. V V .V . , . . . V. V
...... Qiillti¥1§itd. r. V . V . V V . . . 5:: V9 . V,
rvu%ar...r..l.:§.5.i..3.uz.nv V . . V V . . V . . V
33.12193.‘\.T1a.1frDotti—Vskflivwl V V .. . V . . . V V .a .
177????» ..V‘1.LVflI - £1; . . V . _ V V 4 Eva;
V .v..fl.1!{1$'35!flahwddflu. \t 1‘, . . V V V . . . V . Vvvx‘kt ... 1.8L
.zifit‘hJHleiixf‘IsVLBRQAu-flI . V . . V D V V . . V V . V q a)": I . . V .
.lé. V .1?~.9!..C‘..0. 4 . . .V . .. . . . V, V V fiiiVififiulktit
“why! A‘ih 9". V V V V. V N .. V . .. A xtrwh‘absbtt x Q“... 1.6L.
fivfifluufiltfi . . V. V . .V .. V V ...RuN-An. 1: (EL.
«3%.. gnaw. . , . . . . . V . V m? anti... 4.2.... Vi...
Vern? .. fivovt V . V V . . . A ..w . V ... a? . . V
Eitnhug . . . . V . . V . . 1H1.” gyoflsxl V...- « .
Zhub V.\1 . .V . . . . . . V , ... . V V . V r 1...! "finalftrib aim .. if .91.?!
, . . V .V . . . V V . . . . V#u.vo..bh.§§nlt. . a lir‘ulblh..§nt:21 A h!‘;{
. V .. V . .V . . . V . V . ... _ VV .Vofiixtnfiugibnfl; I I. F! thflfl? . .3. tug: . V
V . — . .V 4 V v V in. I Z l ‘b' . u.
_ V . . V V ,VVLVfiLVIVEVsmfiu aftéumwufimfr ..an «at. V V s. Varmiyw. ti
V V . . an”?! §¥§.V.£fih 554“. V .1 ...».V {1- itfi-hfiil‘
. V . i 5. .. t.» v V .I’V— ‘i, E .
‘1'! l g ,

 

.a 0t
9

Vuhuw}..u.u.7 . . .V. .V . . ... V . V
“.A}>a tr‘V V . V .. V . .. . . V . MOM...“ V
. . . V . . . “no...

. i Ir, 1
V y. 1105..) fi. A
. . . “SAW... 5.1:]:
V . {1. ‘V§.AE~L£ $3.3?! 1.1.1 V...r\ «H 0.1
£E61.r!;fioh uiJMMA.-V.VIi!7 thé.
In . I?! 0%”... ..lttlfirnlkblrhauflu
.255}?! ltl V.

5V V
t il¥§fx n .
. ltzt L11; {‘5 Vii)...“
..Iiszia, “I. (Be... Iii;
‘\ 6||£ .V... bl‘tk V
. . AnthL.ll.|~L
. « twilight, Ink-ft. V

V V .3310. AK . .11t‘6’1'x.

JAN-httfiflv-l9‘h‘r.
3.2113137}: Aztlslttifzg
Ebédfinfluz~zké

         

 

 

. .; . V V.V . . V . V
. . . . V V. . , ..u . . . fi‘ttrt‘x‘
V . titéux ...». 16:32» \..‘~‘ .
V pi. {i3l‘tiifii
V V. I :7- VV. 5.»:tttvoiiovh
. . . . V
v

gig in.»

 

              

 
  

 

    

 

1“: 3......VVVV. V .V ;
"V V. LAWYMVVLfiVhwmfifl 3.3V LEV] . V . . V V EVVVV:
. . . V . V iavgs fixgfififiw

     

 

 

 

 

......m

V .7. .-
. . V. .V..IVVV...VY~VV Vi. ...... V
w , V . . V»... . V V . . V. V V
.V 2; V, .. . ”V V ....n LViVs‘uNd-zUbVH AV. A \ALV.
V x. . ..L:Z.VV.\...
V . V...V :.....
EV. ...... .AV
. . V. k.

er.' ...VVVrv'

 

THESIS

 

This is to certify that the

dissertation entitled

PHI MESON AND ACCOMPANYING STARNGE PARTICLE
PRODUCTION IN 16 GeV/c w+p INTERACTIONS

presented by

JAMES EDWARD HYLEN

has been accepted towards fulfillment
of the requirements for

Ph.D. Physics

degree in

 

 

M .W professor

 

Date September 19, 1981

 

MS U is an Affirmative Action/Equal Opportuniry Institution 0-12771

l_.._.

 

 

MSU

LIBRARIES

 

”-

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

PHI MESON AND ACCOMPANYING STRANGE PARTICLE PRODUCTION
IN 16 GeV/c n*p INTERACTIONS

By

James Edward Hylen

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Physics

1982

J ‘5‘
k;

(3/1/7'i

ABSTRACT

PHI MESON AND ACCOMPANYING+STRANGE PARTICLE PRODUCTION
IN 16 GeV/C n p INTERACTIONS

By

James Edward Hylen

We have studied O production in 16 GeV/c D+p interactions. K"
identified in a Cerenkov counter were used to trigger the Stanford
Linear Accelerator Center Z-meter streamer chamber. A sample of
inclusive ¢ events was thus obtained where O + K'K+. Information
from a second Cerenkov counter was used to identify the K+ off-line.
In addition, the visible decays in the streamer chamber of A0 + pn'
and K: + n+n' were used to obtain events of the type n+p + OAO +
anything and D+p + O K: + anything.-

With 1.26 x 109

n+ onto a 61 cm liquid hydrogen target, the
experiment had a raw sensitivity of 3.1 events/nb. For the film
measured so far, taking into account geometric acceptance, unseen
decay modes, and all efficiency corrections, the sensitivities
are 7.1 events/Db, 22 events/Db, and 27 events/ub respectively
for the ¢ inclusive, O with A0, and O with K: samples. The
acceptance covers the region in Feynman's x, x > 0.5.

¢
The measured cross sections for x > 0.5 are 0(O + X) =

¢
1.3 i 2.0 pb, o(¢Ao + x) = 0.37 i 0.16 Nb, and o(¢K§ + x) =
0.33 i 0.15 ub. When account is taken of unseen associated
channels, it is concluded that ¢ is predominanly produced con-

jointly with pairs of strange particles. This agrees with the

ACKNOWLEDGMENTS

This dissertation is dedicated to my parents, Edward and Doris
Hylen, whose faith, support, and love have meant so much to me.

Although it is impossible to mention by name the very large
number Of people whose efforts have made this work possible, I would
like to give particular thanks to some of them. In particular, I
would like to give my greatest thanks to Horace Z. Ming Ma for his
guidance, support, and friendship during this experiment. It has
been a pleasure working with our collaborators, Richard Milburn,
William A. Mann, Takashi Maruyama, and especially Abolhassan Jawahery,
whose friendship I value highly. Thanks are due to Virginia
Whitelaw, Kenneth Blakken, and Daniel Edmunds for help during the
design stage, and to Eric Werner, Steven Buchbinder, and Greg Smith
for analysis computer programming. The support of SLAC Group D and
the long hours put in by the scanning/measuring crew also deserve
mention, as does the work of Tracy Stone in preparing this manuscript.
I also wish to thank Richard Hartung and Wayne Repko for informative
discussions. Finally, I would like to express my appreciation to
Theodate Lawlor, Karen Habermehl, Mehdi Ghods, and Daniel Bauer,

whose friendship helped me through trying times.

ii

TABLE OF CONTENTS

page
LIST OF TABLES vi
LIST OF FIGURES viii
I. INTRODUCTION 1
II. EXPERIMENTAL APPARATUS
A. Overview 9
B. Beam 14
C. Target 16
D. Trigger Scintillation Counter Hodoscope 16
E. Large Aperture Cerenkov Counters 24
F. Streamer Chamber 26
G. Trigger Logic 28
H. On Line Monitoring 35
I. Data Taking 42
III. DATA REDUCTION
A. Overview ' 44
B. Scanning 47
C. Measurement of Film 50
D. Preprocessing - 52
E. Track Reconstruction: TVGP 53
F. Vertex Reconstruction: APACHE 54
G. Event Fit Selection: FILTER ‘ 60
H. Event Reconstruction Efficiency 64

I. Resolution 67

TABLE OF CONTENTS (continued)

IV. ANALYSIS
A. The K- Trigger and K* Production
8. The ¢ Signal
C. Sensitivity Calculation for ¢
D. Contamination of a Sample
E. Conclusions

V. SUMMARY

APPENDICES

I. BEAM LINE EERENKOV COUNTER
A. Pressure Spectrum Fit
B. Beam K+ Contamination

II. LARGE APERTURE CERENKOV COUNTERS
A. Physical Configuration and Gas System
B. Interferometer
C. Light Collection and Optical System
D. Magnetic Shielding

E. Performance Characteristics and Analysis
of Test Data

III. OPTICAL CORRECTIONS FOR IMAGE PLANE DIGITIZER
MEASUREMENTS

A. Overview

B. Scales, Fiducials, and Grid

iv

72
74
80
89
91
97

98
105

108
109
112
114
118

128
129

 

TABLE OF CONTENTS (continued)

C. Optical Constants Fitting Program
D. Fiducial Comparison

LIST OF REFERENCES

Table
Table
Table
Table
Table

Table

Table
Table
Table
Table

Table

Table
Table
Table

Table

Table

Table

Table

10:
11:

12:
13:
14:
15:

16:

17:

18:

LIST OF TABLES

Hodoscope Counter Sizes and Positions
Hodoscope Coincidence Efficiencies

E 131 Latch Tape Format

Latch Channel Contents

Information Displayed on Visual Scalars

Streamer Chamber High Voltage Supply Hourly
Check-list

Six Hour Check—list and Gas Supply Read-out
Results of Double Scan

Event Measurement Summary

Processing Summary

APACHE Separation of 4-prong Events Combined
into 8-prong Events

TVGP Reconstruction of Beam Tracks
Sensitivity and Cross Section Calculation
Weighted Average Hodoscope Efficiency

Vee Decay Fiducial Volume for Geometric
Acceptance Calculations

Table of Strange Particle Pairs Which Can
Be Produced Conjointly With ¢

x2 Minimization Fit to Beam Cerenkov Spectrum
+
K+ Beam Triggers as a Percentage of n Beam

Triggers From Beam Line Cerenkov Counter Fit
Parameters

vi

page
19
23
36
37
37
40

41
49
52
62
65

69
81
85
9O

92

103
106

Table
Table
Table

Table
Table

19:
20:
21:

22:
23:

LIST OF TABLES (continued)

Pion Rejection Efficiency (9 GeV/c n')
Streamer Chamber Fiducial Survey

Root Mean Square Deviation of IPD Measured
Grid From FPD Measured Grid

Resolution of TVGP Reconstructed Fiducials

Comparison of Fiducial Reconstruction with
Survey

vii

page
119
131

136

138
138

 

Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
Figure 7:
Figure 8:
Figure 9:
Figure 10:

Figure 11:

LIST OF FIGURES

page

Meson_Decay Quark Line Diagrams. a) OZI allowed 3
¢ +>K K b) OZI disallowed ¢ + n+n-n0 c) OZI
disallowed ¢> + on d) 021 allowed (I) -> n+n'n0

e) OZI disallowed w + non-charmed where q , q2

can be u, g, or s quarks f) OZI disalloéed

w' + w + qq where q can be u, d, or 5

Exclusive Production Reactions a) OZI suppressed 5
n p + on b) 021 allowed n p + wn

a) Conjoint ¢ Production Eb + ¢K+K' b) Disconnected, 5
Non-hairpin Diagram for n p + ¢¢n

E 131 Experimental Set-up at the SLAC Streamer 10
Chamber Facility

An Event in the Streamer Chamber With Four Charged 13
Prongs and One Neutral "Vee” Decay

Scintillation Counter Beam Telescope 15
E131 Target: 1) vacuum jacket mylar end cap .005“ 17

thick, 2) target mylar end cap .005" thick,

3) delta ray absorber, 4) vacuum jacket mylar

tube 1.437" I.D. x 0.45” thick, 5) target mylar
tube 1.250“ O.D. x .006” thick, 6) target mylar
tube 1.125" O.D. x .003“ thick, 7) liquid hydrogen,
8) spacer mylar end cap .005“ thick, 9) spacer

G-lO epoxy tube 1.0" O.D. x .063" thick, 10) vacuum
space, 11) streamer chamber wall

Side View of Experimental Set-up 18
Illustration of Hodoscope Corresponding Element 20
Logic Trigger. High momentum particle can hit

the corresponding elements V101, V201, V301 and

V401. Low momentum particle shown his non-

corresponding elements V101, V203, and V304

Large Aperture Cerenkov Counter 25

Streamer Chamber 27

viii

Figure

Figure

Figure
Figure
Figure

Figure

Figure

Figure
Figure

Figure

Figure

Figure

Figure

12:
13:

14:
15:
16:
17:

18:

19:

20:

21:

22:

23:

24:

LIST OF FIGURES (continued)

Logic Diagram for n+ Beam Signal

Hodoscope Common Element Logic Trigger Diagram
for the Channel 1U

Gating Logic Diagram
Example of On-line Event Display and Summary
Data Reduction Flow Diagram

2 .
XVERTEX of Events Measured as 4 Prong Events.

Dashed curve shows an unnormalized X2 distribution
for five degrees of freedom

Transverse Momentum of Decay Particle With Respect
to Decaying Neutral Line of Flight for Visible
Vee Decays

Invarient Masses of Visible Vee

Becays With the
Expected Resolution for KS and A

Superimposed

Quark*Line Diagram of Forward K' Production Through
the K Resonance

a) Geometric Trigger Acceptance for K-
b) Geometric Acceptance for ¢ + K K with

K' Triggering and K Cerenkov Identified

+

)

+

Invarient Mass of Kn System. a) M<Ktriggerfi

o 3)M(KtriggerTT )
for events containing KS c) M(Ksn ) for events
with K' trigger d) M(K'
containing A0

with all positive prongs as n+

. n+) for events
trigger

Invarient Mass of+Trigger K- with All Positive
Prongs Taken as K (Inclusive Scan Sample)

Invarient Mass of K' and "n+”, Where n+

trigger
Are Separated by Cerenkov Signal

ix

page
29
31

33
38
45
56

63

70

72

73

75

76

77

Figure

Figure

Figure

Figure
Figure

Figure

Figure

Figure

Figure

Figure

Figure
Figure

Figure

Figure

25:

26:
27:

28:
29:
30:

31:

32:

33:
34:

35:
36:
37:

38:

LIST OF FIGURES (continued)

Invarient Massvof K+K' System, Both Kaons
Identified in Cerenkov Counters

Distribution of Beam n+ per Event Trigger

Number of Prongs in Final Fit as a Percentage
of Events Measured

Mechanisms for o Production
Differential Beam Line Cerenkov Counter

Shape Used to Fit Beam Line Cerenkov Counter
Pressure Spectrum Peaks

Differential Beam Line Cerenkov Counter Pressure
Spectrum with Fit

x2 of Beam Line Cerenkov Counter Spectrum Fit
Versus Beam Momentum

Temperature Dependence of Beam Momentum Fit

Cerenkov Counter Gas Index of Refraction As
Measured During a Gas Filling Operation. The
open circles show the loss of weight of the
isobutane bottle during the filling operation

Dual Interferometer
Interferometer Fringe Counter Electronics

Pulse Height Spectrum of a Cerenkov Counter Cell
for a 9 GeV/c n' Beam. The Solid Curve is a
Poisson Fit

The Expected Ratio of Standard Deviation to Mean
of the Pulse Height Spectrum as a Function of the
Parameter b for Various Mean Numbers of Photo—
electrons

 

 

 

page
79

83
87

94
99
99

100

104

104
110

111
113
121

124

Figure 39:

Figure 40:
Figure 41:

Figure 42:

LIST OF FIGURES (continued)

The Expected Ratio of Standard Deviation to
Mean of the Pulse Height Spectrum as a Function
of the Mean Number of Photoelectrons for Poisson
Statistics in Electron Multiplication (b=O)

E 131 Coordinate Systems and Fiducial Numbering

Comparison of Measurement of Grid on Image Plane
Digitizer with Measurement of Grid on Film Plane
Digitizer. Lengths of Arrows Give the Differences
Between the Measurements, Illustrating the Image
Plane Digitizer Distortions

Comparison of Measurement of Grid on Image Plane
Digitizer with Measurement of Grid on Film Plane
Digitizer, After Optical Corrections Were Applied
to the Image Plane Digitizer Measurement

xi

Page
126

130
134

135

I. INTRODUCTION

2 3

During the early l960's, 0kubo1, Zweig , and Iizuka introduced
various versions of an ansatz to describe the systematic suppression of
certain strong interaction processes in the framework of unitary
symmetry models. The eight-fold SU(3) scheme of Gell-Mann and Ne'eman4
placed the hadronic resonances and so called stable particles in well
defined patterns. A major triumph for the model was the prediction

of the mass and decay modes, as well as the existance, of the 0'.

On the basis of SU(3), it was expected5 that the decay mode ¢ + n+n'n0
(predominantly through ¢ +-pn, p + n n ) would happen at approximately

the same rate as w'+ 3n and ¢ + K Kl In fact the 3n decay of o is much

 

 

smaller. Specifically6:
P ( ¢ + n+n-n0) I ( ¢.+ n+n-no)
+ _ 2 0.074 _ 2 .l8
P ( w + n n no) P ( ¢ + K K )

This is in spite of the fact that there is significantly more phase
space for the ¢ + 3n decay mode. A further difficulty was that the

quadratic mass formula for the JP = l' vector octet,

m2(K*) = I. [3 #68) + m2(p)]

predicted the mass m( w8 ) = 928 MeV/c2 for the eighth member w8 of
the octet, where as for the I = 0, JP = l’ mesons, m (¢) = l020 MeV/c2
and m (w) = 783 MeV/cz. Sakurai5 in l962 proposed the w-¢ mixing
model to explain this mass difference, speculating that the physical

w and o are linear combinations of ”8 and a vector singlet, wd. With
the mixing angle determined from the mass formula, Okubo1 combined

the octet and singlet into a nonet. He then showed that the additional

assumption that the trace G: of the reducible tensor representation

1

2

of the nonet not enter into any physical expression leads to zero

0 2 3

+ - . .
matrix elements for ¢ + Ian and o +ln n n . Zweig and Iizuka
gave a more pictorial version of this assumption in terms of a

quark model.

In the SU(3) quark model, a meson is formed by a quark anti-quark
pair. The quarks come in three flavors, up, down and strange, which
form the fundamental triplet of SU(3). The octet and singlet come from
the group relation 3603 = 8@l. In this representation, the neutral
members of the nonet are not pure quark states. For example, the
singlet would be mo = l. ( uU'+ dH-+ 53 ). Somewhat suprisingly, when
the mixing angle derived3from the w-¢ mixing model is used, the phy-
sical 6 turns out to be nearly a pure s§' state. This is in contrast
to the m which can be written as %- ( uU' + da'). ,Assuming then that
the quark flavor quantum numbers are conserved in strong interactions,
one can draw quark line diagrams for various possible ¢ decays
(Figures l a-c). The dominant decay mode, ¢ + KK', can be drawn as a
"connected" diagram, where the initial o and the decay particles are
connected by quark lines. The decays ¢ + on and o + n+n'n0, however,
can only be drawn with so called "hairpin" diagrams, where the s and

E' in the same particle annihilate. The Okubo-Zweig-Iizuka (OZI)

Rule is then that hairpin diagrams are disallowed. As seen in Figure

ld, the w + 3n decay is an OZI allowed diagram. That ¢ decays at all
to 3n has been ascribed to small admixtures of non-strange quarks

in the o wave function or to the operation of other, higher order
processes. The decay ¢ + 3n is thus said to be suppressed as com-

pared to the allowed decays, ¢ + KK and w + 3n.

s K- 3' Tr+
u
f 5
¢ 3 :::::> (::::‘U‘ 0
.§_\___ k+ ¢ — U (T
's‘ 5 CU ,-
d
a) b)
3(3)
0 U
5 ”6(3) am")
9 ’ ) w (d) U Tl
S TT U Gig)
u d n
c) d)

ql

c q2

v :> a
C— 2

ql

 

e)

Figure l: Meson Decay Quark Line Diagrams. a) OZI allowed o—>K_K+
b) OZI disallowed ¢ + n+n'n0 c) OZI disallowed ¢ + on
d) OZI allowed w + n+n'n0 e) OZI disallowed w + non-
charmed where q , q can be u, d, or s quarks f) OZI
disallowed w' + w + qQ'where q can be u, d, or s

 

Supporting evidence for the rule was also found7 in the 2+ nonet,
where the decay of the g; state f' to n+n' was seen to be suppressed

as compared to the decays f' + K+K' and f + n+n'.

More recently with the addition of the quark flavor charm, the

8 the extreme narrowness of

OZI rule is used to qualitatively explain
the V and V' resonances. For the cE' state V, the connected diagram
V + DD which is the analog of ¢-+ KK is not allowed because the mass
of two D mesons is greater than the V mass. Thus the only strong
decays possible proceed by OZI suppressed hairpin diagrams (Figures
le—f).

Exclusive channels were looked at to see if the OZI rule also

applied to production reactions. For example, w and ¢ are both

JP = l', iso-singlet mesons. Yet

0 ( T' p + 0 n )

 

_ = 0.0035 i 0.0015
0 (n p + w n )

at laboratory incident momentum 5 to 6 GeV/c.9

As the quark line
diagrams in Figure 2 illustrate, the smallness of this ratio can
again be explained as suppression of the hairpin diagram necessary
to produce the ¢- A list of other decays and exclusive production
reactions, which support the OZI rule, can be found in the survey

by Okubo10.

As pointed out by Siversn

, a non-strange initial state can
produce ¢ via an OZI allowed diagram if other strange particles are
produced along with the ¢- An example is shown in Figure 3a, where
PP. + ¢ K+K-. This type of allowed production, where the s and E

in the ¢ terminate on adjoining particles instead of on each other,

 

"I

n §,:::::::] l-EEEEE ¢

0.

 

 

 

U
p d dn
U U
M TI- 9.10
- d

:3

Figure 2: Exclusive Production Reactions a) OZI suppressed
n p + ¢ n b) OZI allowed n p + w n

‘0
Q
Q.
3

 

 

C
C

a)

 

X

7<

‘0 UI
ca 04cm
culmmlm c|
9

+

 

b) -d

:1

c c|
mun U1 m
e- 9

 

 

Figure 3: a) Conjoint ¢ Production E'p + 4 K+K' b) Disconnected,
Non-hairpin Diagram for n p + ¢ ¢ n

 

6
will be termed conjoint production. The few experimental results on

conjoint ¢ production seem to be contradictory.

Two experiments using the CERN 2 meter bubble chamber examined

12

conjoint ¢ production. Donald et al. using an antiproton beam

at 3.6 GeV/c examined the exclusive channels pp + K+K'K+K’ and

pp + K+K'n+n'. They found approximately ten ¢'+ K+K' above background
in each channel. Since this was a large fraction of the l6 K+K'K+K_
"events, but only a small part of the 8l8 events in the K+K'n+n'

channel, this may indicate that the suppression due to having to

produce two extra kaons with the limited center of mass energy is

of the same order as the OZI suppression of the OZI violating process.
They also observed similar results in the pp + K+K'K+K'nO and K+K'n+n'n0

channels. In sharp contrast, Blobel et al.13

reported that OZI
violating mechanisms strongly_dominated in the inclusive reaction
pp + ¢ + anything at 24 GeV/c. The ¢ signal was observed in an in-
varient mass plot of all pairs of oppositely charged particles in
which one of the particles was identified as a kaon by its decay in
the bubble chamber. Conjoint production was then looked for by
counting the number of extra strange particles, again identified by
visible decay, produced with "K+K'" pairs in each mass range. They
found if anything a dip in the number of extra strange particles with
"K+K'" pairs in the 0 region, where an excess would be expected if the
¢ were conjointly produced. Specifically, the fraction of 0 pro-
duced conjointly was 0.0 i 0.2.

Three counter experiments have also been reported. Woodworth

1.14

. . . - + - + - -
et a examined the exclu51ve reactions n p + K K n n n p and

n-p + K+K_K+K'n'p at l9 GeV/c with the CERN Omega spectrometer.
Their findings are similar to those of Donald et al., in that the

0 signal is approximately equal in the two channels, but this is only

m 2% of the first reaction whereas it is m 60% of the second reaction.
Again in contrast, Akerlof et al.15 using a Fermilab double arm spec-
trometer to look at p + Be + 0 + anything at 400 GeV/c reported no
significant excess of charged kaons produced with 0, indicating
that OZI violating mechanisms dominated over conjoint 0 production.
The ratios of associated kaons were, however, only approximately two
standard deviations from the ratios expected for l00% conjoint pro—
duction.

Recently the ACCMOR collaboration reported16 the results of ex-
periment WA3 using the CERN SPS. Using primarily a 93 GeV/c n-
beam on protons, they found that 32 i 9% of inclusive 0 events had an
extra K0, and 42 i l2% had an extra charged kaon. Their acceptance

for ¢ covered the range in Feynman X17

, 0.05 < XF < 0.25. Their
conclusion was that at least on the order of 40% of centrally pro-
duced inclusive ¢ were produced conjointly.

One other relevant experimental result may be interesting. Using
the Brookhaven National Laboratory multiparticle spectrometer, Etkin

‘8 studied the interaction n'p + K+K7K+K'n at 22.6 GeV/c. They

et al.
found that the quark-line disconnected reaction n-p + ¢ 4 n was approx-
imately l0% of the OZI allowed reaction n'p + o K+K'n . As shown in
figure 3b, the reaction n'p + ¢ o n can be drawn as a disconnected

'quark line diagram without any hairpins. (A hairpin diagram is taken

to mean that both ends of the quark line are in the same hadron).

Etkin et al. state that it is 021 forbidden because it is disconnected.
Thus its large rate with respect to the OZI allowed reaction would
indicate a serious violation of the OZI rule. However, because the
s and E' in a 0 in Figure 3b do not terminate on themselves, it _
seems that the reaction could also be termed conjoint. With respect
to another disconnected, non-hairpin diagram Okubo9 noted that his
original algebraic formulation of the quark line rule forbids only
hairpin diagrams. He further stated that an extended quark line
rule might be formulated to cover all disconnected diagrams. Without
knowing the actual dynamical origin of the OZI rule, it thus seems
difficult to interpret the experimental result as a violation of the
rule. Rather it clarifies the definition of OZI.

The experiment on which this dissertation is based was designed
to study inclusive ¢ production in n+p interactions at 16 GeV/c. In
particular, the analysis is focused on mechanisms for and the role

of the OZI rule in inclusive 0 production.

 

II. EXPERIMENTAL APPARATUS

A. Overview

The experiment was designed to study inclusive 0 production in
fl+p interactions at l6 GeV/c. It was performed at the Stanford
Linear Accelerator Center under the designation experiment El3l.
Because the ¢ lifetime is so short, 0 must be reconstructed from its
decay products rather than being studied directly. The 0 decays
to K+K' 49% of the time, and this mode was selected as the easiest
to use. Pions are much more copiously produced in l6 GeV/c n+p in-
teractions than kaons. One priority therefore became good Cerenkov
counter identification of K' and K+ to suppress the background due
to pions. The SLAC 2-meter streamer chamber was used to actually

view the events. With a sensitive volume of approximately l m3

, it
allowed for the reconstruction of the charged particle tracks from
an event, and also any neutral particles which decayed to charged
states before leaving the chamber. The entire streamer chamber was
in a l3 kG magnetic field, so that particle momentum could be de-
termined by the curvature of charged tracks in the chamber. The
basic design of the experiment also included an event trigger.
An event was recorded only if fast trigger logic indicated that a
K" was identified as coming from the event. Such a trigger was
required so that we would not be swamped by irrelevant events. The
trigger K' were selected using a scintillation counter hodoscope
and a Cerenkov counter.

The layout of the experiment is shown in Figure 4. A seconary

hadron beam from the accelerator was brought through a beam defining

10

kpwpwomu LmnEmsu LmEmmme u<4m msp pm azuumm Fmpcosvgmaxm PMF m ”w mgzmwu

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 Q .1.
mm mmauzs um mm mu man an
6% 8 I. We. 9 b c. mm. mm
> .2.
a = a > _ /\ = =
134.101: 2 _> .2 ..
:
.—
..820T. 9.2 ....
FL _il 3......
”an
3.09
3.58 vacant/\I/ u _ no 0
O .0
6.8 w /_l
+ _$lmmm<
3929.? 24
. 4...... 1.62.4... __T.M.____-...iii___T..
.258 _> x 03¢: 83%
5:530 ~> fl W
I! <

 

 

 

293.1.

11

scintillation counter telescope (S1, 52, AN, AS’ and S3 in the figure)
and a beam line Cerenkov counter (6%2 in the figure). The beam tele-
scope signaled when a beam particle would pass through the target,

and the Cerenkov counter signaled that the particle was a n+, rather
than K+ or proton. The beam then passed through a liquid hydrogen
target, where some beam particles interact with the protons. The
charged particles produced in the n+p interaction then leave ion-
ization trails through the streamer chamber. As stated above, it

was desired to trigger the apparatus when a secondary K' was iden-
tified. The large acceptance threshold Cerenkov counter used for this
purpose had a n- threshold of 2.7 GeV/c and a K' threshold of 9.7 GeV/c.
Thus for particles with momenta between these values, n' would pro-
duce light and give a signal in the counter, whereas K' would not
give a signal. One hundred scintillation counters were used to form
a hodoscope. The rows of counters are labeled Vl, V2, V3 and V4 in
the diagram. The counters were placed and logic set up so that
coincidences were obtained only for particles with momenta greater
than 3 GeV/c. Thus the combination of a coincidence from the hodo-
scope and no signal from the K"Cerenkov counter identified a K',

and triggered the recording of the event.

An event trigger caused a very short high voltage pulse to be
sent to the streamer chamber. This produced avalanching from the
ionization trails. The pulse was cut off before the avalanches could
grow more than a few millimeters. The light from the ”streamers“
was recorded on film, giving a record of the charged particle
tracks. Also, the signals from the segmented cells in the Cerenkov

counters were recorded on magnetic tape, along with counts of the

12

number of beam particles, etc.

The n+p center of mass was moving rapidly forward. Thus part-
icles produced in the n+p interactions tended to be collumated
forward due to the Lorentz boost. The K' then tended to be deflected
by the magnetic field toward the "K' Cerenkov counter". The associated
K+.from a o decay would then tend to be deflected Oppositely, toward
the "|? Cerenkov counter". The momentum of a particle which entered
the K+ Cerenkov counter can be determined by off-line measurement of
the track in the streamer chamber picture. If above 2.7GeV/c, the
(Cerenkov output then determines whether it was a n+ or K+. The
other charged particle tracks,from the event while not Cerenkov
identified, can be reconstructed from the streamer chamber picture.
In particular, the strange particle decays K; + n+n' and A' + pn'
which occured in the streamer chamber can be reconstructed and iden-

tified. (Such a "vee is shown in Figure 5). This allows the search
for the conjoint channels n+p + ¢ K; + anything and n+p + ¢ A° +
anything.

The Cerenkov threshold for antiprotons is higher than that for
kaons. Thus the particle satisfying the l'K' trigger" logic could
also be an antiproton. This was not expected to raise the trigger
rate very much. Although the inclusive cross sections in n+p in-
teractions are not available, it is known19 that antiproton pro—

duction in pp interactions is an order of magnitude less than K'

production in this energy range.

13

 

xmuwo =mm>z Pmcpzmz
mco new mucosa wmmcmgu L30; gpkz Lansmsu cmsmwepm msp cw pcm>m :<

x x x x

”m mczowd

 

14

8. Beam

The #23 SLAC beam line (Figure 4) was at 00 production angle to the
incident electron beam at the primary target, corresponding to maximum n+
production with respect to K+ contaminationzz. The n+ beam trigger
consisted of three "yes" scintillation counters ( S], 52, S3 ), two
scintillation anti-counters ( AN,AS ) to veto halo tracks, a differ-
ential beam line Cerenkov counter (CFZ) to identify n+ and thus
further reduce beam contamination and, finally, a beam spill gate.
The beam counters are sketched in Figure 6. The beam Spill was
approximately l.6 us long, with a repetition rate of from l0 to 240
spills per second. Typical running rates were 60 spills per second,
depending on how many other experiments were sharing the beam. There
were an average of 9.6 beam particles, as indicated by S1 - S2
coincidences, per spill. Of these, an average of 7.0 per spill were
satisfactorily on target, that is, satisfied the logic X+ 3 S1 - $2 .
S3 - (Ag—:TAE). On average, 5.7 per spill were then counted as good
n+ beam, as defined by the further coincidence, n+ E X+ - EFZ'

Beam momentum was determined to be l6.0 i 0.3 GeV/c by use of
the beam line Cerenkov counter (see Appendix I for details). Three
sets of streamer chamber measurements of beam tracks gave l5.5 i
3.3 GeV/c, l6.3 i l.8 GeV/c, and l5.3 i 1.2 GeV/c, all in good agree-
ment with l6.0 GeV/c. The streamer chamber measurements are described
in section 111.1.

The 8.l% K+ contamination of the pion beam was reduced to l.5%

contamination by use of the beam Cerenkov counter (Appendix I).

15

 

 

 

 

-—%-
Beam
Sl
Magnet Q
T 1.59 CIT]
AN ——L A5
L— l.62-u
cm

TVGP Coordinates X Y

Center of veto counter hole 0.3l cm -139.07 cm

Center of target housing -0.04 cm -114.72 cm

Figure 6: Scintillation Counter Beam Telescope

 

2
47.04 cm
46.99 cm

16
C. Target

The target (Figune7) was 6l cm of liquid hydrogen maintained at
a pressure of l8 PSIA. This correspondglto a temperature of 21K and a
density of .0699 g/cm3. Target diameter was 3.l4 cm, with an entrance
hole preceding it of 2.22 cm diameter. This was comforably larger than

the l.6 cm diameter beam spot allowed by the ANAS veto counters.

D. Trigger Scintillation Counter Hodoscope

As stated in the Overview, the hodoscope was designed to select
negative particles coming from the target with momentum greater than
3 GeV/c. To do this, four planes of scintillation counters were built.
They are labeled Vl, V2, V3 and V4 in Figure 8. There were actually
two sets of V3 counters, V3U which covered the upper half of the K'
Cerenkov, and V3D for the lower half. Each V plane contained 20
counters, making a total of lOO counters. Counters in Vl were placed
next to each other, edge to edge, in a sort of picket fence array.

The V2, V3U, and V3D counters were arranged similarly. V4 was in two
planes, with alternating counters placed edge to edge. The individual
counters are numbered from Ol to 20 starting with the one closest to
the beam center line. The design positions for all the‘counters in
the hodoscope are given in Table l.

The principle of the corresponding element logic is illustrated
in Figure 9. The positions and widths of two corresponding counters
in VT and V3 (e.g. V101 and V301) were chosen so that a negative
particle from the target with high momentum ( > 3 GeV/c) would hit
both, but a low momentum particle, being deflected by the magnetic

field to a larger angle, could not. The V2 counters were added to

17

Fsz cmnamzu cmEmmepm APP .mumam Ezzom> Ao— .xomgu :moo. x .o.o zo._ wasp xxoam orig

cmumcm Am .xorcp =moo. ago new cmrxe cmumam Am .cmoocvxz Uwzowfi AN .xuwsu emoo. x

.o.o sz_._ menu Lassa Saute“ Am .xawgp .000. x .o.o .omm._ was“ Lassa Samamb Am

.xuwgp =mv.o x .o.H zmm¢.P wasp Lafixe pmxumn Essom> Aw .Lmneomnm he; much Am .xowgu
:moo. amo vcm LmPXE pmmcmu AN .xumsu :moo. and ocm LmeE pmxomh Eszum> AF ”gmmcmk Pmpu "N mesmru

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“1 Eu m.___
Ea o._© y
we
.\ . I v,
_\ \. \\ \wl . mee

 

 

 

 

 

\
\
\

 

 

 

18

g:-pwm Faucmswcmaxm mo 3mw> mvwm ”m mesmwu

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

s o s m s a ... m e N s _ o
I) . p L b p p .
.505. go: Louczou 34:35 +x J J
2:35...“ uzwcoazl “U1 ”1
E .283 32.. U n”;
nunnnnuj “mmmmmwfl
€3.25.
_ u_um:oc:
cog.
.r W¢Dm36 ... r u
:3: _
Lost: —1 .
_ cm> " “macaw
/ II
~> _> Loosccu a Loammcum
:m> =
LUHCDCU >OJCCLUU IX -
Hill .
u
.
.
.
_\ .
\ \ .
- , F
oxo> \
eczema
“acme: upwou ”memo:

 

 

19

Table l: Hodoscope Counter Sizes and Positions

V4(up V4(doun
Counter Plane V1 V2 V3up V3doun stream) stream)
Offset of counter plane
from beam center line (x) 0cm -3 .5 .5 .13,5 .21,5
Width of counter 01 3cm 4.5 5 5 17.5
"°”'"t (A‘) 02 3 4.5 6 6 17.5
03 3 4.5 6 6 17.5
04 3 4.5 6 6 17.5
05 3 4 5 5 17.5
06 2.5 3.75 S S 17.5
07 2.5 3.75 S 5 15.5
06 2.5 3.75 5 5 15.5
09 2.5 3.75 S 5 15.5
10 2.5 3.75 S 5 15.5
11 2.5 3.75 5 S 15 5
12 2.5 3.75 5 S 15.5
13 2.5 3.75 5 5 15.5
14 2.5 3.75 S 5 15.5
15 2.5 3.25 4 4 13
16 2 3 4 4 13
17 Z 3 4 4 13
18 2 3 4 4 13
19 2 3 4 4 ~13
20 2 3 4 4 13
Height (2) -20cm -30 0 .40 65 65
thickness (Ay) to to to to
20 30 4O 0
Distance from center of
""‘“" ch‘”°" (y) 200 cm 300 400 400 645 645

Scintillator.th1ckness 1/8" 1/8” 1/8" 1/8" 1/4“ 1/4”

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20

V401

 

1

1

i
l

K' CERENKov \

i ‘ K+ CERENKov
-- -V3(T4 V3 01

 

 

 

 

 

l I TARGET

STREAMER CHAMBER

 

 

 

Figure 9: Illustration of Hodoscope Corresponding Element Logic
Trigger. High momentum particle can hit the correspond-
ing elements VlOl, V20l, V30l and V40l. Low momentum
particle shown hits non-corresponding elements VlOl,

V203, and V304

 

21

the coincidence in order to decrease background triggers due to random
coincidences in the counters. The V4 counters were added to the
coincidence requirements to check that a trigger particle made it

through the Cerenkov counter. The hodoscope coincidence logic was

2
v11. . v21. . (V3U1. G) V301.) - V4].

"Mo

i l
where @is exclusive -OR. Hodoscope counter sizes and positions were
optimized by use of a Monte Carlo program which generated "events"
and tracked the resulting particles through the system.

A "K'" could be faked by a low momentum particle which didn't
come from the target. Examples are particles from downstream decays
or interactions with the streamer chamber gas. Such triggers were
rarely seen during scanning. Two or more low momentum particles in
combination could also hit enough counters to give a trigger. Most
such events could be rejected during scanning, and the others elim-
inated by the software after measurement. However, this process
accounted for more than half the event triggers. The dominant con-
tribution appears to have come from particles interacting with the
downstream wall of the streamer chamber, producing a "spray" of
particles into the hodoscope. In anticipation of such problems,
the amount of material in the hodoscope and Cerenkov counter window
had been kept to a minimum.

The counters were wrapped with aluminum foil, packed edge to
edge and sandwiched between two styrofoam sheets, which provided a
low mass supporting structure. The entire package was wrapped with
a layer of 4 mil black polyethylene film to make it light tight. The

entire structure, including ultra-violet transmitting Plexiglas light

22

guides, 2 inch photomultiplier tubes and iron and Mu-metal magnetic

shields, was clamped in a unistrut frame hung on rollers to an over-
head beam. By rolling the frame, each scintillator could be posit-

ioned in the beam for testing. Amperex 2203 and RCA 8575 phototubes
were Used for Vl and V2. Amperex 56 AVP and 56 DVP phototubes were

used for V3 and V4.

The Vl counters were in the fringe of the TB kilogauss magnetic
field. Since even 1 gauss can affect the efficiency of a phototube,
considerable effort was necessary to make these counters work. The
phototubes were placed as far from the magnet as possible, necess-
itating l.5"m to 2 m long light guides. Each phototube for Vl was
inside a T6' thick cyclindrical Mu-metal magnetic shield. This was
wrapped with lOO turns of wire to form a "bucking coil". The current
in the coil was adjusted to cancel out the axial field at the photo-
tube. Currents for the Vl coils ranged from O to 0.5 amperes. The
coil was inside a %r thick cylindrical soft iron magnetic shield.
Also, a %fi thick iron plate was positioned above this phototube array.
The V2 counters had similar long light guides, Mu-metal shields and
iron shields. These counters were far enough away from the magnet,
however, so that it was not necessary to use bucking coils on them.

V3 and V4 needed only short light guides, but also had Mu-metal and
%" thick cylindrical iron shields.

Using a 9 GeV/c pion beam, the scintillator counter efficiencies
and the 4-fold logic (V1 . V2 . V3 - V4) efficiencies were measured
before and after the data run. The 4-fold coincidence efficiencies are

listed in Table 2. They are an important correction to the triggering

cross section calculations.

23

Table 2: Hodoscope Coincidence Efficiencies

Element Start Finish
IU .230 i .004 .383 i .003
10 .620 i .005 .490 i .003
2U .737 i .004 .602 i .003
20 .736 i .003 .578 i .002
3U .625 i .005 .248 i .004
30 .616 i .005 .294 i .005
4U .394 i .005 .319 i .005
4D .407 i .005 .334 i .005
5U .438 i .005 .383 i .005
50 .417 i .005 .396 i .005
6U .508 i .005 .359 i .004
60 .419 t .005 .620 i .005
7U .859 i .003 .514 i .005
70 .867 i .003 .572 i .005
8U .737 i .004 .216 i .004
80 .680 i .005 .235 i .004
9U .494 i .005 .436 ii.005
90 .519 i .005 .422 1:.005
IOU .540 ::.005 .572 i:.005
100 .491 1:.005 .538 11.005
11U .524 1:.005 .346 it.005
110 .562 2: .005 .626 i .005
12U .674 i .005 .653 i .005
120 .654 i .005 .693 i .004
13U .500 1:.005 . .492 i=.005
130 .570 3:.005 .549 12.004
14U .492 i .005 .448 i .005
140 .523 t .005 .494 i .005
15U .392 32.003 .095 i=.003
150 .416 i .005 .229 i .004
16U .501 i .004 .188 i .004
160 .565 :t.003 .372 It.005
17U .173 i .003 .081 ”5.002
170 .190 21.004 .124 it.OO3
18U .309 :t.005 .343 it.OO4
180 .349 ::.005 .385 Ii.005
19U .371 :t.005 .345 i.005
190 .448 :i.005 .387 i.005
20U .196 :.004

 

 

 

24
E. Large Aperture Cerenkov Counters

Having selected high momentum particles with the scintillator
hodoscope, n" are distinguished from K' and B'through the use of the
K' trigger Cerenkov counter (Figure TO). The counter consisted of
10 cells in a 2 x 5 matrix, each cell observing an average of 4
channels of scintillation counter hodoscope. For instance, coinci-
dences in the l up ( = Vl0] - V20] - V3U0] - V40] ), 2 up, or 3 up
hodoscope logic could be vetoed by a signal in the l up Cerenkov
cell. Hodoscope channel 4 up, overlapping Cerenkov cells l up
and 2 up, could be vetoed by either.

The Cerenkov counters were designed to operate with gas at
atmospheric pressure partly because of ease of construction, but more
importantly because this allowed the use of very low mass windows,
greatly reducing false triggers which could have been caused by pions
interacting in a more massive window. Isobutane gas was chosen be-
cause of its high index of refraction ( l.00l37 ), which gives a
reasonably low threshold for pions ( 2.7 GeV/c ).' Isobutane also
scintillates very little and has good transmission characteristics
for ultra-Violet light, where the Cerenkov light spectrum has its
maximum intensity.

22 to be

Inclusive K' production in n+p at l6 GeV/c was estimated
0.57 i O.l9 mb. Since 0 ( n+p + n' + anything ) is23 23.0 i 0.2 mb,
and since kaon triggers were to be produced by null signals in the
trigger Cerenkov counter, it was critical that the counter have ex-
tremely good efficiency to suppress the n' background. When measured

with a 9 GeV/c n' test beam, the pion rejection efficiency, averaged

over the counter window, was 99.848 i .006%. This holds pion

25

 

O

 

 

 
  
 

 

.
.
”M_--_--

 

 

 

 

 

 

 

 

l/J'f/
6% "N N ”N "\\
015101010
77

 

   

 

Figure l0: Large Aperture Cerenkov Counter

 

 

 

 

26

contamination to the trigger to approximately 6%.

A second Cerenkov counter, identical to the first except for
placement and alignment, was used to tag particles associated with
the trigger.

The details of construction, testing, and monitoring of these

Cerenkov counters are given in Appendix II.

F. Streamer Chamber
The primary track recording device in this experiment was the

SLAC 2-meter streamer chamber.24

The chamber was 2 m long, 0.8 m wide
and 0.6 m deep ( see Figure ll ). Three planes of wire mesh formed
transparent electrodes. When the chamber was triggered, a Marx gener-
ator provided a lO nanosecond, 600 kV pulse to the electrodes. The
potential difference of 20 kV/cm caused avalanching along the ion
trails left by particle tracks through the chamber. The pulse was

cut off before the streamers of ions grew beyond a few millimeters.

The ion avalanches gave off light which was recorded on film.

The chamber gas was 90% neon, l0% helium with about 0.l% iso-
butane. The chamber was in a 13 k6 magnetic field to provide momentum
measurement of the tracks. It was viewed through a thin mylar window
by three cameras, with approximately l40 and l8° stereo angles.

The chamber has a memory time of several hundred us. Therefore
all tracks produced during a single l.6 us beam spill were visible on
the film. However, at the highest repetition rate SLAC beam spills
are still 2.8 ms apart, so only one spill is visible.

Figure 5 is a streamer chamber picture showing a 4-prong event

with a vee. The 4-prong vertex is obscured by the target and 6-ray

27

cmnswzo LmEmmLum ”HH mcsmwu

mnocuompw chucmo

 

 

 

 

 

     

..‘~..‘.......~......‘..c _.
-....‘.. ................
..‘.'......".‘...-.~...‘ .

.................... ‘...~ I

 

-gmme mcwz vcm zovcwz Emp»2mv

28

attenuator. Slow spiralling electrons can cause flairs and obscure
parts of the chamber, so acrylic slats were placed above and below the
target to quickly attenuate the 6-ray tracks (see Figure 7). Figure
5 also shows a slow proton nearly parallel to the chamber electric
field, which leaves such a large ionization trail that it is essen-
tially unmeasurable.

Spacial resolution in the chamber varied from 0.2 mm to 0.5 mm
perpendicular to the electric field and from l mm to 3 mm parallel
to the field,depending on the measurer. This is based on point
measurements of fiducials on the film, and is further discussed in

Appendix III.

G. Trigger Logic
Introduction

The basic logic trigger was a coincidence between a beam n+, an
identified secondary K", and the system ready to take data. On this
trigger, the streamer chamber fired, and counter data was recorded.
The trigger system was then disabled to allow the streamer chamber to
recover and the latch data to be written on tape by the on-line POP-9
computer. This ”dead time" was used to count how many times n" would
have triggered the scintillator counter hodoscope logic without the
Cerenkov veto. Hence, during live time K' triggers fire the system.

During dead time so called ”n'" triggers are counted.

Beam Logic
The logic begins with a maChine beam spill gate from SLAC main
control, indicating that a beam spill was on the way. As shown in

Figure l2, this started a 2 us beam gate which covered the l.6 us

29

szmwm Emmm +: Low Emgmmwo demo; ”NF mesmwu

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmwmmm MWW .mcmn53c _mc:m;o LmFmom vchn op comm; mg¢nEzc om
2\oN—k Auv . .Prwam mewsome >2
2\<¢ommh Anv empmm mcm mcoumcwewcomwn Emmnw,vcm .mm .mm .Fm
2\¢ommk Amy “muwc: oemom .mspvwz mmrza oz» op comm» mmewp
mcom Auvpamnu
e as 2 mm
N.NUm :uV mcm An— NW
_.Nom Auv
mpmc
Faucmuwuum
+x Ame?“ mo use «.mom Acv
«.mum mcom ummemuv mcop m
\T - <
N.mom _..m0m m.NUm
22 E .6 E 72.; .3; 25m 2: -22 e 24
Emmmm.+x u+=//Illh ./IlII—I < . mm . Fm Am< + z<v n.K .
m N F u I.
.K. m. m. mn+x

30

SLAC beam spill. This gated the beam scintillation counters S], 82,
S3 and the beam Cerenkov counter EFZ on. An ouptut pulse from this
gate was also used to clear the "ARTURO" computer interface so it
would be ready for an event.

A coincidence between S1, S2, and S3 not vetoed by the halo
scintillation counters AN and AS indicated a beam particle on target.
As shown in Figure 12, this logic is x+ = 31 - $2 - s3 4W).
A coincidence with the beam line Cerenkov counter (used to suppress
beam kaon and antiproton contamination) then gave the beam trigger,

- 52 ° S3 -(A—_:—A—) "EFZ’ Halo was also counted with the

l N S

logic S1 - $2 - (AN + AS) .

K' Identification Logic

The beam trigger indicated a n+ into the target. A corresponding
Element Logic Trigger (CELT) signaled a K7 from the target, and formed
another part of the overall trigger requirement.

Each scintillator hodoscope plane had 20 elements set up in a
horizontal picket fence. The spacing was chosen so that to hit
corresponding elements (for example, the first element of each, as
shown in Figure 9) a particle from the target had to have a mo-
mentum greater than 3 GeV/c. This was above pion threshold in
the Cerenkov counter. Thus the K" signal was a coincidence between
four corresponding hodoscope elements, and a null signal from the
corresponding Cereknov cell (Figure l3). The Cerenkov cells and
the V3 hodoscope plane were split into up and down arrays.

A coincidence could be formed by the up array or by the down array.

31

[IS 12]
"0|

5202

 

 

CEU l-lw

E6 5 6 "20/!

'30] u

 

«5 I
~ _’ I m splitter to add pulses

"01

 

from 21!

 

[RS 62] III—fl.
‘-C '0 AA ‘
Iurst ward of!

 

 

   
 
 
 
  
    

 

to latch
to ADC
Lls 22.9. [G I 6 C124

 

CCU
1-10 0

 

 

 

[RS ‘29

   

[RS 623 l

‘6 ('6 no 0 so similar to 11!

(E w. w. w. 50. 20. so. do
sililer to ID

to All: to latch

Figure l3: Hodoscope Common Element Logic Trigger Diagram for the
Channel lU

 

 

 

32

If both the up and down scintillators were hit in V3, it is
likely that a fake trigger was actually being formed. For instance,
a fast pion may trigger hodoscope elements V101, V201, V3U01, V401
and veto itself by a Cerenkov signal in K'CIU. If a simultaneous

slow pion hits V300I there would be the coincidence V1 V2

01’ 01’
V3D01, V40], with a null signal in K'ClD. Hence the logic was set
up so that coincidences in corresponding up and down scintillators
vetoed each other.

The logic for the other 19 channels was similar to that in
Figure 13. Since there are 40 hodoscope coincidence channels, and
only 10 Cerenkov cells, each Cerenkov cell looked at several scintil-
lator channels. For instance, scintillator channe4$ 1U ( E VlOl-VZOlf

.

v30 V4 , 20, and 30 can all be vetoed by k'E cell 1u. Scintilla-

01' 01 )
tor channel 4U, being on the edge of two cells would be vetoed by a
signal in either K'CIU or K'C2U. This pattern was repeated so that
channels 8, 12, and 16 were also vetoed by either of two Cerenkov cells.

The coincidence channels were then OR-ed together by quadrants
(CELT 1U~10U, CELT 11U-20U, CELT 1D-IOD, CELT 110-200).

The V1 and V2 scintillator counter arrays had long thin light
pipes out from the magnetic field region to the phototubes, and
thus had very small pulse heights. These signals were therefore
put through an LRS 612 amplifier before being sent to the discrim-
inators.

Gating

The gating scheme is shown in Figure 14. When a SLAC machine

beam spill signal started the beam gate, the system was ready to

133

Sesameo UWGOJ mcwpmw “QF weaned

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

puma; noun; vac compo uo< ounliuao umxm—mo g 42\oo~oo
mcouacwspcumpn uix x o».||||||u unwm_y o o~-pp pguu
+ oueu
mcmpaum cwoawsu .xsuwnwccp “sown; wewh
moo “Loam
mcapaum canoes» x “wows: Iiiiill uzo i n -
- P ..- [II No.4 0 o, . esuu
Lomocto case 826 Nmm mm; 6; Nope
one; Hung use 0»
gmaovgu xx»: 0» mem*cp umeJ Emma +.hh
newsmu cu
m. no mm o a o» -
UD< U S H P JZ\OONOO 2 ON P— Hgmu
TIL
. , mumc
oeseeoue, oowue_ steam mwm
cmuaneou o» . swasmgu mm;
zmpmo cosmmgum 2 o_-F HAND
Nopw amam cmaaaeou
62663 tempo .4uawu >o<mm eoztxxxguLL
o
o==e¢< a S euo> zmpm>m,///~w gou_3m .ccoe oeoeweoaxm
«you
ammo? scam
ages 65 use “scum oeem _Pwem econ 045m

 

 

 

w

34

trigger. The coincidence of a n+ beam trigger and a CELT then gave a
logic trigger. The logic trigger starts two timed gates and in-
itiates the electronic recording of the event.

The streamer chamber spark gate shut off the beam gate until
the noise from the streamer chamber firing died down. Otherwise
noise from the chamber could cause another trigger signal.

In approximate time sequence, the electonic recording of the

event was as follows. The latches, ADC's and PWC's were strobed.
Then the Marx generator fired, and the event was recorded on film.
Also the electo-luminescent fiducials flashed. After a 4 ms delay, a
computer interrupt was sent. The computer busy signal now held the

beam gate closed for the approximatley 20 ms needed to do the CAMAC
readout to the POP-9.

At this point, the electronics would be ready to go again. How-
ever, recharging the Marx generator, advancing the film, and writing
the event on tape takes somewhat longer. During this 200 ms, Marx
triggers were suppressed by the dead time gate and the trigger rate
of the hodosc0pe for pions was scaled as a check. This was done by
gating out the Cerenkov veto,st0pping the normal K' trigger scalars,
and turning on scalars to count n- triggers. During this time the
streamer chamber spark gate was off and the beam gate operated
normally. The gate delayed output pulse which occurs at the end of
the dead time was used to clear the ADC's and latches. The system

was then ready for another K' trigger.

35

Latch Tapes

The 100 scintillation counter hodoscope discriminators and the
40 coincidence channels were latched and read out. ADC's recorded
the output pulses from the l0 K+ Cerenkov cells and the lO K' Ceren-
kov cells. The K' Cerenkov discriminators were also latched. The
format of the output buffer as written on the latch tapes by the on-
line POP-9 is shown in Table 3 and Table 4.

H. On Line Monitoring

During the data run, monitoring of the experiment included visual

scalar readings, a continuously updated computer display, an hourly
checklist, and a six-hour checklist. In addition, various alarms

would signal certain problems.

l. Visual scalars displayed the running totals of the logic
signals shown in Table 5. These totals were entered in the log each
time a new role of film was mounted.

2. A background program on the on-line POP-9 gave a contin-
uously updated summary on a Tektronics computer display scope, as
shown in Figure l5. The top half shows the scintillator hodoscope
latches and the Cerenkov counter ADC's for the latest event. For
the event shown, roll 72 frame 2384, the coincidence of (Vl()5) -
(V2()5) . (V30 05) ° (V4(35) - (K'C_2U) - (V3USE—) gives the logic
trigger 05-DOWN. This is the pattern of a high momentum, negative
particle below Cerenkov threshold. The other latches set, VT

17’

.13, V411 and V412 appear to be due to a Single p05itive

particle. The overlapping configuration of the V4 counters is the

V214 , V30

36

Table 3: E 131 Latch Tape Format

word contents
l 338 ( number of words in record)
3 Roll number (in 8CD)
5 Frame number 8CD)
12 ADC of K‘C lu
l3 ADC of K‘C 2u
14 ADC of K’C 3u
15 ADC of K'C 4o
16 ADC of K'C Su
17 ADC of K‘C TD
18 ADC of K'C 20
19 ADC of K‘C 3D
20 ADC of K'C 4D
21 ADC of K'C SD
28 ADC of KIC lu
30 ADC of K+C 3u
31 ADC of K+C 4u
32 ADC of K+C Su
33 ADC of K+C TD
34 ADC of K+C 20
35 ADC of K*C 3D
36 ADC of K+c 4D
37 ADC of K+c SD
38 ADC of K C Zu
44-49 Latch unit l to 6
50 Latch unit 8
51 Scalar 1.1
52 Scalar 1.2
53 Scalar 1.3
54 Scalar 1.4 total number of beam spills
55 Scalar 2.1 S] . 2
56 Scalar 2.2 $1 - S2 - §
57 Scalar 2.3 S1 - S2 -
58 Scalar 2.4 $1 - S2 . A
59 Scalar 3.1 x: a 51 . 52. 53 . A
60 Scalar 3.2 n a X+ . CF2
61 Scalar 3.3 n’
62 Scalar 3.4 AccideZtal
63 Scalar 4.1 K‘ trigger CELT 1-10 0
64 Scalar 4.2 K‘ trigger CELT 11-20 u
65 Scalar 4.3 K- trigger CELT 1-10 0
66 Scalar 4.4 K‘ trigger CELT 11-20 D
67 Scalar 5.1 n‘ trigger CELT 1-10 u
68 Scalar 5.2 n: trigger CELT 11-20 u
69 Scalar 5.3 n trigger CELT l- 10 D
70 Scalar 5.4 n’ trig gger CELT 11-20 D
71 K;C index of refraction (n-l) x 10:; in BCD
72 K c index of refraction (n-l) x 10 in 8CD
270-333 HPHC latches
334 Latch unit 7
336 Latch unit 9
338 Latch unit 10

lane Format: Unlabled 9-track 800 SP!
16 bits/word. 338 words/record

37

Table 4: Latch Channel Contents

Latch Contents (16 bits per unit, lowest order bit at
Unit right)

1 v116,... ,v101
2 V212,...,V201,V120,... ,v117
3 V3u08,...,vsu01,vzzo,... ,vz13
4 V3004,...,V3001,V3U20,... ,V3u09
5 V3020,... ,vao05
6 V416,... ,V401
7 COIN 12o... ,COIN 01U V420, ,V417
8 COIN 080,.. ,COIN OIO, COIN zou, ,COIN 13U
9 (blank 4 bits) COIN 200,... ,COIN 09D

IO (blank 6 bits) EK'SO,...,CK'IO,CK'5U,... ,EK‘IU

Table 5: Information Displayed on Visual Scalars

$1 $2 = beam scintillation counter coincidence, upstream
2

51-So (AN+AS) = beam into halo counters
X+ = 51-52 {AN+ASS°53 = beam into target
+ v . + .
n = X+-Cbeam = beam identified as n by beam line
Cerenkov counter
prs = number of accelerator beam spills

Accidentals = (4 fold v coincidence) - (128 ns delayed n+)

LTlive = logic triggers during live time, normally equal
to the number of frames taken
+ ‘ b + d ' l'v time

LTdead = logic triggers during dead time, with Cerenkov

veto off, to see the w hodOSCOpe trigger rate
+ = b + d r d ad time
“dead eam n u ing e

Sum of coincidences in each quadrant, that is, 1U to 10U,
ID to 100, 11U to 20U, and 110 to 20D

38

O 7 2 0 0 5 O 0 2 3 8 4 239
....* ........... t... V1

a ........ t ...... V2
.................... V3U

* ....... t ....... V30

6 a n ........ V4
.................... UP

* ............... DOWN
.................... K-C
ADC K-

106 95 115 139 97
132 96 111 114 118

113 74 S3 66 S9
481 43 55 74 63

 

a aaaalan I LAAAIIAA
<
H

V2 V3U V30 V4

 

 

l a aaaaua

l aaaum

V1234 COIN UP V1234 COIN DOWN CHER LATCHS CHER GT THRESH ADC OF COIN

 

 

 

 

 

 

 

N

 

will“

:1:

 

Figure 15: Example of On-line Event Display and Summary

39

reaSon two V4 counters are hit by each particle. The Cerenkov
counter ADC readouts shown are all in their normal pedistal ranges
except for K+C 10, which probably indicates the presence of an
associated high momentum pion.

The graphs in Figure 15 show summary information, in this case
for the last 239 events. The first eight graphs show, on a log
scale, how many times each latch channel was set. The downward
slope from channel 0l to channel 20 is normal since the 01
counters are closest to the beam line, and expected to be hit more

often. Although the accumulated statistics on this particular

summary were low, it is seen that V3006 was dead. V3D was in
fact the only major counter failure. Because of the redundency in
the 4-fold coincidence, it was possible to simply remove this counter
from the coincidence requirement and continue data taking. The

ninth graph indicates how many times each Cerenkov ADC was greater
than its normal pedestal value. The last graph whows the pedistal
adjusted ADC for the Cerenkov cell associated with the trigger
channel for each event.

3. Streamer chamber settings which were checked and adjusted
hourly are shown in Table 6.

4. Equipment which was checked and adjusted every six hours is
listed in Table 7.

5. Several alarms were set to warn Of serious problems. Three of
these had to do with the large atmospheric pressure Cerenkov counters.
The counters were filled with isobutane, which is flamable from 1.8%
to 8.4%. To guard against explosive mixtures, leak detectors were

placed around the counters and connected to alarms and automatic

40

Table 6: Streamer Chamber High Voltage Supply

Hourly Check-list

Blumlein (transmission line) pressure
Marx (500kV supply) pressure

Trigger Gap pressure
HV Sensor - positive

negative
HV Helipot - positive

negative

59 psig
8 psig

11 psig
7.56 kV

7.69 kV
6.56 kV
6.56 kV

41

Table 7:

Streamer Chamber
Neohe gas flow-
Ne-He gas supply
Isobutane gas flow
Visual check on streamer quality
Purity monitor

Beam line Cerenkov counter
Pressure

Proportional chambers
Bubblers
Gas supply

K+/K' Cerenkov counters
Isobutane supply
Bubbler
Pressure

Liquid nitrogen supply tank
(for gas recirculation system)
Check supply

Beam line magnet shunt-hold to 0.5! of nominal

1 23 DO
2 23 01
3 23 02
4 23 03
5 23 01
6 23 O4
7 23 05
8 23 02
9 23 D3
10 23 D4
11 23 05
12 23 Q7
13 23 08
14 23 05.2
15 23 Q9
16 23 05.1
17 23 05.6
18 23 05.6

19-3 Streamer chamber magnet

Six Hour Check-list and Gas Supply Read-out

1500 l/hr
record pressure
3 t/hr

check

3.5 x 10‘6

13 psig

check bubbling
record pressure

record weight
check bubbling
1.0 inch water

>1600 1b

9.95 mV
55.52 mV
75.18 mv
55.20 mV
42.40 mV
54.70 mV
51.20 mV
off
61.50 mV
61.50 mV
off
51.70 mV
53.05 mV

.34 mV
12.80 mV

4.00 mV
off
of!
50.33 mV

42

shutoff valves on the isobutane supply. llmecounters were main-
tained at approximately 1.0 inch of water over pressure, which pre-
vented air from seeping in in possible leaks, and which continuously
drove the counter gas through an index of refraction monitoring cell.
Rather than a complicated feedback system, a simple bubbler was used
to keep the pressure constant. During the day, expansion of the gas
in the counters caused excess gas to bubble out. In the evening,
the contraction of the gas and loss through the bubblers was compen-
sated for by cracking open the isobutane supply valves. Each counter
3)

used approximately 3 lb (20 ft of isobutane per day. Approximately

5 ft3/day of this total flowed out through the interferometer mon-
iroring cell, and less than 3 ft3/day was attributed to leaks. The
remainder was lost due to the day/night temperature cycle mentioned
above. As a check, an alarm was set to sound if the pressure in the
counters differed by more than .25 inch from the desired 1.0 inch of
water pressure. Finally, an alarm triggered on tank weight was set
to indicate a low isobutane gas supply.
Other alarm systems, in place frOm previous experiments,

indicated target trouble, Marx generator High Voltage supply trip,

and streamer chamber gas recirculation system fault.

I. Data Taking

Construction of the scintillation counter hodoscope and the
large Cerenkov counters began in early 1977. Set up and testing
at SLAC began in November of that year, and continued through
May of 1978. During the setup the beam was tuned to give the

sharpest possible focus on target. The beam line Cerenkov was

43

slightly modified, carefully tuned, and a pressure spectrum was
run. The hodoscope was assembles and aligned. The "bucking" coils
which were used to cancel the axial magnetic field on the hodoscope
phototubes were tuned. The hodoscope efficiency was measured with
a pion test beam. The large Cerenkov counters were assembled, and
their optics aligned. They were gas and light leak tested, and

had their bucking coils tuned. The Cerenkov counter efficiencies
were mapped using a 9 GeV/c pion test beam. A pulse height spectrum
was also taken for one Cerenkov counter cell. The electronics was
cabled up and the relative timing of the pulses from counters

and coincidences adjusted. Existing data-taking software for the
on-line PDP-9 computer was modified and new display software
written.

The actual data taking ran from May 24 to June 30, 1978. There
were 91 rolls of film used, containing a total of 305 thousand
frames for each of the three views. Of these pictures, approxi-
mately 1% were taken with the trigger Cerenkov counter deliberately
out of the trigger coincidence, and 3% were unusable because of
hardware failures. For the normal trigger events, there were
1.26 x 109 beam n+ on target, giving the experiment a raw sensitivity

of 3.1 events per nanobarn.

III. DATA REDUCTION
.A. Overview

From the pictures of events and latch tapes, one wishes to recon-
struct particle momenta at event vertices and identify particle types.
This information can then be used to study resonances, kinematic
distributions, etc. A simplified schematic of the event reconstruction
process is shown in Figure 16. After a brief general description,
details of each step will be given in later sections.

The pictures were first scanned to pick out ones with events to

be measured. For the first pass on the film, essentially the only
criterion was that the frame have a track which could have caused
the trigger coincidence and which could come from an event in the
target. Once a reasonably large sample of these general events were
done, the rest of the film was scanned selecting only those events
which also contained a visible particle decay. Points along the
tracks in the selected frames were then measured in each of the three
views using X-Y digitizers. These digitized events, on tape, were then
edited and cleaned up. Hardware and measurer errors were corrected
where possible. A bookkeeping file was also created at this point.
It contained only a very short description of each measured event,
and was very useful for sorting, getting summary information, and
keeping track of the processing.

I A version of the Three View Geometry Program (TVGP)25 was used
to take the measured points in the three views and calculate a track
curve in three dimensions. It used a magnetic field map to find
track momentum. It also calculated error matrices for track para-

meters. The target region was not visible in the streamer chamber

44

45

 

Pictures

La tch

Tapes

 

 

V

Sort latch
records into
measurement
order

 

Figure 16:

 

 

 

1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Phasure: digitize
Scan: ch-ec-n Measure points on tracks
for track — list ...: ”'0,
matching {Its a.
trigger ‘ u 'a “w.
' coincidence »‘ ’.
1—1
r
Digitized events
'
bookkeeping
file —- Edit a cleanup
I
Clean events
1
TVGP: fit curves
through digitized
._1 po‘nt‘ ‘ Cllculltl
status to
book file I
_ Tracks
' V
. APACHE: extend
tracks to vertex 6
Add APACHE
status to delete extraneous
000k file I
‘ I
R MERGE latch infor-
mation onto end of
' 4 APACHE records
f
SQJAH: get K: of
vee to vertex
‘ I Kill duplicate events
!
FILTER: select good
APACHE fit

 

r
Final date tape
1

 

 

GeoIIetri c acceptance
6 Analysis

 

Data Reduction Flow Diagram

 

 

46

pictures. Thus the interaction vertex could not be seen. Program
APACHE26 extrapolated the tracks back to a vertex. It also deleted
tracks which were measured, but which apparently did not come from
the event vertex. Because of extraneous tracks and multiple inter-
actions in the target, APACHE frequently found multiple solutions
(called "fits") which contained different sets Of tracks.

The Cerenkov and latch information for the measured events was
'takerl'fronl the latch tapes, sorted into the measurement order,
then added to the APACHE fit records. Program SQUAH’27 was used to
get chi-squares for different types of neutral decay hypotheses for
the vee decays. The chi-squares were used later to label the vees
as A°, Kg, or y. By using the event vertex, the energy-momentum
balance for the decay is overconstrained by three degrees.

Events for which tracks did not reconstruct properly in TVGP
were remeasured. As the editing and TVGP processing were upgraded,
previously failing events could later pass. Also, some events were
deliberately measured twice on different machines for comparison
purposes. A doubly sorted bookkeeping file was used to look for
and delete duplicates so that only one measurement of each event
would be on the final data tape. Special measurements, such as
measurement of non-interacting beam particles, were also deleted.

At the next stage, program FILTER selected the best of the
APACHE fits for an event. While doing the selection, the event
”trigger track" was extrapolated through the hodoscope to make sure
that it agreed with the coincidence counters that triggered the
event. Also particle identification was done where possible.

For the vees, this meant settling on either a A°, a K:, or a v

47
interpretation. For charged prongs, the Cerenkov information was

used for those which extrapolated into the Cerenkov counters to

. . . i i
distingu15h K and n . The reconstructed events were then ready

for analysis.

8. Scanning

The film was scanned at Tufts University to select
events for later measurement. Initially, a portion of
the film was scanned for all legitimate K- trigger events.
In the next stage, the total film sample was scanned with

the further selection criterion of a visible vee or kink.

K' Inclusive Scan

In this pass about 7% of the film was scanned for a
legitimate lC'trigger coming from the target.

Each scanning table was provided with a template which
displayed the positions of the downstream hodoscope elements
and streamer chamber fiducial marks from the viewpoint of
camera #3. With the projected fiducials of view #3 aligned
with the template, fast negative tracks were visually extra-
polated into the hodoscope elements. The scanner compared the
element apparently hit with the element caUsing the event
trigger, taken from a printout of the latch tapes. If the
extrapolation agreed with the latch within : 2 elements, and
the trigger track could have come from an event in the target,
the event was selected for later measurement.

8488 frames were selected using the K' inclusive

criterion from 21,787 frames scanned.

48

Strange Particle Scan

The resources were not available to scan and measure
the entire exposure completely, so a strategy was developed

to select the most interesting events. In addition to the

K- trigger track latch and the target criteria, the events
selected were required to have a visible vee or kink. This
sample thus includes all events with a visible strange particle
decay. It also includes a background of hi decays and I
conversions, which can be separated out by SQUAH fitting
after measurement and processing.

Two views of every frame were examined to find vees
or kinks which came roughly from the target region. When
one was found, the frame was then checked for a trigger
track. Since visible vees and kinks were relatively rare,
and looking for them didn't require extrapolation into
the hodoscope, this scanning proceeded faster than the
k- inclusive scan. 6,657 were selected out of 253,628
frames scanned.
Scanning Efficiency

In order to estimate how many legitimate events were
missed by the scanners, portions of the film were rescanned
under each set of scanning rules.

If one assumes that the loss of events is random
(Geiger-Werner method) , then

N1 = elN N2 = €2N N12 = €1€2N

 

 

49
where
N = total number of events in sample
N1= number of events found by scanner #1
N2= number of events found by scanner #2
N12= number of events found by both, that is, the
events that they agree on

E1 efficiency of scanner #1
82 = efficiency of scanner #2
The double scan gives N1, N2, and N12. Solving the above

equations in terms of these gives N = NlNz/le’ 61 = N12/N2, and

E2 = N12/N1' The results of this calculation are shown in

the following table.

Table 8: Results of Double Scan

 

K" Inclusive Vee/kink
Number Of frames 5,410 29,000
double scanned
N1 1,713 751
N2 2,041 846
N12 1,478 625
N 2,366 1,017
81 .72 i .03 .73 i .05
52 .86 i .04 .83 i .06

One precaution taken in the vee/kink double scan should
be mentioned. Two prongs emerging from the target may be
misidentified by a scanner as a vee (especially during the

early part of the scan when relatively inexperienced). Such

50
events cause no problem in the analysis since the prongs will,

in general, fail SQUAW fits to A0 and kc, and be dropped.

However, their inclusion in the scan by one particularly “liberal'I
scanner, and exclusion by another, perhaps more experienced scanner,
would incorrectly lower the calculated scanning efficiency of the
second scanner. To preclude this, events not agreed upon by both
scans were adjudicated by the supervising physicist and senior

scanner.

C. Measurement of Film

Measurement of the selected events was done at Tufts University
on two image plane digitizers. These were interfaced to a PDP-8
computer which did some on line checks as well as writing the measure-
ments on tape.

The measurer assigned each event a five digit measurement event
type code which described the number of pOsitive prongs, negative
prongs, vees, positive kinks and negative kinks in the event. Four
fiducials were measured in each view. Six points were measured on
each track in each view. To avoid confusing the reconstruction pro-
gram TVGP, tracks which turned through large angles were only measured
through the first 600. Events with a fiducial obscured were rejected
as unmeasurable. Also cut were events with more than 15 prongs.

With an everage of 7 beam particles in each picture, nearly half
of the events have a second target generated interaction. Because
the target region was not visible and trigger tracks were forward
and high momentum, the measurer was normally unable to tell which

event produced the trigger track. In fact, the presence of multiple

51

events, halo tracks, and delta rays produced major difficulties in
this experiment. All tracks which came from the target region and
might have been associated with the trigger track were measured as
one event, and the sorting out of which tracks actually formed the
triggering event was left to the vertex finding program APACHE.

One lesson learned was to leave as little open to human error
as possible. The on-line measurement software system required the
measurer to type in the date and the measuring machine number. Head-
aches could have been saved through the automatic recording of the
information port number to indicate the measuring machine, and the
installation of a computer clock. The on-line system did require that
the fiducials be a set distance apart. This check would immediately
catch measurement of an incorrect fiducial, sloppy measurement, or
major digitizer failures. The system automatically changed the
measuring machine view, so that the three views of each track were
measured in the correct order. The on-line system also required that
the number of tracks measured was consistant with the event code.

A summary of event measurement status is given in Table 9.
The table shows that in the non-vee, non-kink sample, only
229 events still fail after remeasurement, giving a final
measurement efficiency of 96.1%. Calculation of the efficiency -
for the vee sample is complicated by the fact that not all
failed events were remeasured. The original measurement had
a failure rate of 28.8%. The remeasurement had a failure rate

of 25.1% Thus the pass rate for the 62% of the sample for

which remeasurement was done was 92.8%.

events, the pass rate is 73.4% after one measurement, and

95.5% after remeasurement.

52

Table 9: Event Measurement Summary

 

Non Vee Vee

Non Kink Associated
First Pass
Events Measured 5,824 4,799
PANEL failures 475 75
TVGP failures 769 1,306
Failed Events
Remeasured 1,262 853
Remeasurement
PANEL failures 75 7
Remeasurement
TVGP failures 154 207

D. Preprocessing

For kink associated

Kink
Associated

 

1.404
17
357

306

51

After measurement, gross errors on the tapes such as record

fragments due to hardware failure were edited out on the Tufts PDP-l

computer and a copy sent to Michigan State University for processing.

After preprocessing to get a translated tape and event summary in-

formation, the data was interactively edited to correct obvious

errors in event types, measurement dates, measuring machine numbers,

roll numbers, etc.

After editing, program PANEL was run, which made sure each

53

event had the proper heading, proper number of fiducials, and that the
number of tracks measured agreed with the event type. It checked for
bad codes and missing views. It recombined events which had to be
measured in two pieces due to the buffer size limitations of the
measuring program. It assigned each event a sequential ordinal number,
and output a summary record for bookkeeping purposes for each event.
Finally, if the event was acceptable, the data was written out in

TVGP-input compatible format.

E. Track Reconstruction: TVGP

Track reconstruction was performed using a version of the
Three View Geometry Program (TVGP)25. This program takes the
point measurements of a track from the three film views and fits
an analytical three dimensional space curve. It calculates the
track parameters position, direction, and momentum at the
beginning and end of the track. It also produces error matrices
for the parameters, and the correlations between the beginning
and end parameters.

To project the measured points on the film into the streamer
chamber space, one obviously needs to know the camera positions,
film to lens distances, and any lens distortions. Constants
describing these quantities were refined by using the chamber
fiducials. The fiducials were 28 electroluminescent crosses
which were flashed once per frame. Their positions were fixed and

they served as reference points in the chamber space. Comparisons

of measurements of the fiducials on film with a survey of the

54

positions in the chamber were done by a program called NEASEL.
Using a chi-square minimization proceedure it produced the camera
constants. The necessary film fiducial measurements were done
using a Tufts University "film-plane“ (distortion free) X-Y
digitizer.

One modification was made to TVGP. It had used two fiducials
in each view1x>"align the frame", that is, as the reference points
to determine where the measured track points were with respect

to the camera axis and the orientation around the axis. Because

there is measurer error in measuring these points, the program was
modified to use more fiducials. Four fiducials were used for an
initial alignment fit for each frame. Then the fiducial measurement
which was farthest off was deleted, and the alignment redone using
the remaining three fiducials.

Nearly all of the frames were measured on image plane digitizers,
where an image of the film is projected onto a digitizer table.
The projection lens again produces distortions. The correction for
these distortions was also done in TVGP. The details are described

in Appendix III.

F. Vertex Reconstruction: APACHE

The event vertex in the liquid hydrogen target is not visible.
Program APACHE was written to do the vertex finding. It extrapolates
the TVGP reconstructed tracks back to a vertex and eliminates extran-
eous tracks such as halo, delta rays, tracks from a second event,
and decays in the target. The basic routines have been used for

. . 26
several streamer chamber experiments, and are described elsewhere.

55

The algorithm for finding a vertex is based on a X2 minimization
using spacial error matrices of track beginning points. Initially, the
tracks are stepped through the magnetic field, with the proper changes
in the error matrices, to their point of closest approach to the center
of the target. The first iteration vertex is then the error matrix
weighted average of the track end points. After stepping the tracks
to their closest approach to this point, a second iteration weighted
vertex can be calculated. The program iterates until all tracks move
less than 0.1 mm during an iteration. The weighted average of the
beginning points is forced to find the vertex by making the track
spacial error matrices artificially very large in the track direction.
The new weighted vertex thus wants to end up on a tangent to all tracks.

Although the average number of tracks measured for an event was
seven, events with up to fourteen tracks were allowed. Finding the
correct vertex for the event producing the trigger and excluding
extraneous tracks is a difficult problem. The basic scheme Used was
to try iterating a subset of tracks to a vertex, then deciding on
the basis of a x2 cut whether the tracks form an acceptable vertex.

The remainder of this section contains discussions of the distributions
on which X2 cut decisions were made and the algorithm used for
choosing which subsets of tracks to try.

The vertex x2, calculated from the iterated vertex and the
extrapolated TVPG error matrices, is plotted in Figure 17 for all
events for which the scanner measured exactly four prongs. Separate
plots are shown for different ranges Of FRMS, the root mean square
of the measured point scatter around the fitted track curve, which

indicates how well tracks are measured. The film setting error

 

 

Events per unit x2

 

56

 

 

 

 

 

 

  
    

 

 

 

 

50.J l. < FRMS < 2.
40:-
+ cut for FRMS = 1.5
30 _. __
20.—
10 —‘ l
O 5 10 15 20
30
W 2. < FRMS < 3.
20 _
+ cut for FRMS = 2.5
10 -
O 5 10 15 20
30 ‘a
3. < FRMS < 4.
20 —-
10 _. + cut for FRMS = 3.5
\
I I I I
20 0 5 10 15 20
— 4. < FRMS < 5.
10 -
cut for FRMS = 4.5 g
0 I " 1 I '_‘ l
O 5 10 15 20
2
XVERTEX

XSERTEX of Events Measured as 4 Prong

Egents. Dashed curve shows an unnormalized
x distribution for five degrees of freedom

Figure 17 :

57

which TVGP uses in calculating error matrices was fixed at 4.
Hence those events with an FRMS of 4 should have their errors
correctly estimated. In fact, the shape of the theoretical X2
distribution for five degrees of freedom shows good qualitative
agreement with the events with FRMS between 3 and 4. Of course

there are extra events in the long tail of the distribution.

These events we suppose to be those with an extraneous track, and

. 2
thus are events me wish to cut. The cut used was a Xvertex of
3 per degree of freedom, scaled by the FRMS:
2 2
Cutoff Xvertex = (NDOF) x (3) x FRMS

 

film setting error

NDOF’ the number of degrees of freedom, is

NOOF = zntr ' 3

where n is the number of tracks used in finding the vertex. For

tr
the 4-prong events NDOF = 5 and theoretically only 1% of real vertex
solutions would be cut. Average cuts are also shown in Figure 17.

A cut was also made on each track's contribution to x2, so that
several tracks fitting very well could not offset the large con—O
tribution of a really poor track. This cut was made a factor of

two looser than the overall X2 cut. Thus the incremental X2 for

addition of a track was cut off at a X2

of 12, also scaled by
the FRMS as for XSERTEX'

2 criterion is chosen there is still the question of

Once a x
selecting subsets of tracks to try it on. Adding or deleting a track

moves the iterated vertex, and thus Changes all of the track X25-

58

Taking a vertex with all tracks and deleting the one with the largest
X2 contrubtion might work for getting rid of a single halo or delta-
ray track, but certainly fails when one is trying to sort out the
tracks from two separate events in the target. With up to 14
tracks per event, trying all possible combinations of tracks is
unmanagable. The method actually used was to try to first find
three tracks making an acceptable vertex, and then add tracks one
at a time to see if they are consistant with that vertex. For

two tracks there is in general a second, local X2 minimum. Falling
into this minimum can prevent the finding of the actual vertex.
This problem was largely circumvented by never using less than
three tracks, and by starting the iteration vertex in the center

of the target. More specifically, a possible K' trigger track is
identified (normally there is only one candidate per event) and
then all possible pairs of other tracks are tried with it until

a vertex with an acceptable X2 is found. Then, all other tracks
are added one at a time, each being kept in the solution if the

new iterated vertex which includes that track fits the acceptance
criterion. The event solution which has been built up, called an
APACHE fit, is then saved. However this fit is not always the only
one possible. Thus the trigger candidate is then tried with each
track that didn't make the first fit, cycling through all the other
tracks for the necessary three track trial fit. If an acceptable
three track trial fit is found, again as many other tracks as
possible are swept in and the final fit saved. Several APACHE

fits may be produced for one event by this method. For the final

59

data sample, one fit was selected by program FILTER, described later,
as the best fit.

For events containing visible vee decays, the two decay tracks
were fitted to the decay hypotheses Kgi+ n+n- and A0 +-p n'. Since
the outgoing three-momenta were found by TVGP, using the assumed
particle masses for each decay mode leaves only the neutral
three-momentum unknown. Energy-momentum conservation gives four con-
straining equations for the three unknowns, allowing calculation of
a X2 and the best fit momentum of the neutral. The program SWAN27
was used to do this fitting. The neutral particle fit giving the

2 5) was

lowest X (unless it had a confidence level worse than 10'
then used by APACHE the same way as charged prongs, having to
meet the same X2 criterion to be included in the vertex fit.

In an effort to clean up the data, some tracks were deleted
before the APACHE fitting precess. All tracks with momentum less
than 200 MeV/c were deleted to cut down on delta- ray contamination.
Similarly, all tracks with momentum greater than 10 GeV/c were
deleted to hold down contamination by beam halo tracks. Tracks
from o decays will not be cut with these limits. Furthermore,
having an extraneous track in the fit would move the calculated
vertex, thus smearing the resolution.

— As a final step, APACHE does a correction to track parameters
basedon the calculated vertex. This refinement can be especially
important for K' trigger tracks which tend to have a large part of

their length obscured by the target and non-interacting beam

tracks in the same spill.

60

G. Event Fit Selection: FILTER

The APACHE fit was rejected if the vertex did not satisfy the
target cut. Since the target was not aligned with a TVGP axis,
the vertex was first transformed into the "target coordinate system.“
This system has its origin at TVGP coordinates (1.384 cm, 3.123 cm,
48.04 cm ) and, with a rotation defined as positive in the right
hand screw direction with respect to an axis,is rotated - 0.0105383
radians around the TVGP Z-axis and 0.0046900 radians around the
TVGP X-axis. The vertex was required to be within a 1.8 cm radius
of the Y (target) axis, and have a Y (target) coordinate between
-66.5 cm and -4.5 cm.

In a n+p interaction, there should be two more positively charged
secondaries than negative secondaries. The net charge imbalance
(NCI) was defined as the number of positive prongs minus the number
of negative prongs minus two, and would be zero for an event where
all the correct prongs were included in a fit. However, a slow
proton may not make it out of the target, or may be unmeasurable,
giving an apparent net charge imbalance to a fit which is otherwise
correct. Similarly, a secondary may interact before leaving the
target, and thus be lost to the fit. Thus while NCI is a guide to
the quality of an APACHE fit, some leeway should be allowed so
that useful events are not rejected. APACHE fits were rejected if
the absolute NCI was greater than two. In doing this, events with
an extraneous track or two in the fit can be let through. Random
extraneous tracks included in fits, while making charged prong

distributions unreliable, should not affect resonance peaks except

 

61
to smear the resolution somewhat.

An APACHE fit was also rejected if its trigger track candidate,
when extrapolated through the scintillator counter hodoscope, did
not hit the elements for the trigger coincidence latched on tape
for that event.

If there were still multiple APACHE fits for an event after the
above cuts, one fit was selected as the best, and the others rejected,
on the basis of a quality factor. Based on individual examination of
several hundred events, a quality factor which was the product of
three factors was chosen. The first, to indicate how well the tracks
fitted a vertex, was simply XSERTEX / NDOF' Second was a factor to
indicate how well charge balanced the event was. Somewhat arbit-

rarily, this quality factor was chosen as

1.0 if NCI = 0
1.25 if NCI = -l
QFNCI = 1.5 if NCI = +1
1.75.1f NCI = -2
2.0 if NCI = +2

Lastly, a vee quality factor of l was assigned if a vee made a
passing constrained fit with—the event vertex, 2 if it did not.

(With the primary interaction vertex and the decay vertex giving the
direction of the neutral particle, the energy-momentum balance at

the decay vertex is now over constrained by three constraints. If
the SQUAH calculated X2 for the three constaint fit was less than 26.
the fit was considered passing.) The fit with the lowest product of
these three factors was then selected as the best fit.

At this point a decision was also made about vees which passed

the SQUAH X2 criterion for more than one of the interpretations,

 

 

62
K: + n+n: A0 + p n', or the photon conversion Y.+ e+e'

with recoil proton. Although a more.complex scheme was tried

(described in Reference 28), the simple expedient of calling A0
those vees which were ambiguous between K: and A0, and elim-
inating the v fits produced acceptable separation, as illustrated in
the transverse momentum distributions in Figure 18. Because of its
larger decay energy, a K: + n+n' decay has a larger maximum trans-
berse momentum (206 MeV/c) than a A0 +'p n' decay (100 MeV/c).
Superimposed on the plots are the distibutions expected from simple
isotropic decay, and ignoring resolution smearing. The agreement
with the expected distributions is good. Specifically, the K%
tail under the A0 peak agrees with the expected number of events,
whereas a bump would be expected here if A0 were being misidentified
as K2, or a dip sould be seen if K: were being lost into the
A0 sample.

Table 10 gives the number of events reaching each stage in the

reconstruction process.
Table 10: Processing Summary

Vertex reconstruction (APACHE) failures 1996

APACHE passing events 9540
Number of duplicate events removed 312
APACHE fits 13177 for 9228 events
Fits passing FILTER cuts 9031 for 6980 events
FILTER selecteg fits 6980

-with A 199

-with K: 548

63

mamumo mm> wpnwmw>
com ucmwpu mo mew; Facuzmz mcwzmuwo op pumammm cue: mpowpcma xmowo mo saucmeoz mmcw>mcmce “ma wczmwm

 

 

 

 

 

 

U\>mz wow u ”Nana saw: magma uwaogpomw Low compznwcpmwu Fmowpmcowge .......
o\>mz ooH u .m gym: amumu owaocpomw com cowpzmwcumwu Facepmcomsh 11.11
m ox “I “523:8 m ....11
o< mm :mxmp mzozmwasm o¥\o< use o< we» pcwmcpmcou m .IIII
7565 .a
0mm com omp oo_
. . P . .
lea-“0a“ . Ill—ll. _
.P o‘---‘a4.44.|..
ae- . knit. ..... .
.. 1| . .
.. “a... _
.H .... . 3
-. ..nleet — A
a” . . W
.H n.1iu _ “N
“H it a
.H ..n I]. a
. . Id
.. . I.
_H II» 0
.H W \ “w
.n M
u . :2: W
.
—
u
.
.
.
r

 

I-‘cee II... ...

__ [of

j:

64

H. Event Reconstruction Efficiency

Estimation of the efficiency Of the reconstruction software con-
sists of trying to determine the probability that a prong which actually
belonged to an event was not included in the final FILTERed APACHE
fit. If one had a set of frames where one knew a priori which tracks
belonged to the events proper and which were extraneous, then one
could determine the efficiency by simply running these frames through
the software and counting how many good tracks were lost. Such a set
was not available, but the problem was examined with an approximately
clean sample.

Events which were measured as 4—prong events, and which passed
2

VERTEX
had. The method used to examine APACHE's reconstruction was to mix

the APACHE X test as 4-prongs, were as clean a sample as we
these up, and see how confused APACHE became.

There were 626 4-prongs reconsturcted by APACHE, of which 575
passed FILTER. The 525 4-Pr009 ‘TVGPKECOFdS were combined into 313 8-
prong records, which were fed into APACHE-FILTER. 553 fits resulted,
as compared to 575 previously, indicating that 4% were lost in the net
charge imbalance cut or now reconstructed with the vertex outside the
target. Table 11 indicates for each trigger track how many of the
original 3 prongs with it were lost, and how many prongs it picked up

from the other event.

0f the 3 x 553 = 1659 positive prongs which started out with
the 553 passing trigger tracks, only 1308 are still associated with
their respective trigger tracks, a loss of 21 i 1%. It is noted

that the probability of losing 1 or 2 prongs is small. The bulk of

65

Table 11: APACHE Separation of 4-prong Events Combined into 8-prong

Events
Number of extra prongs with Total

Number of trigger track
prongs
lost 0 l 2 3 4

0 252 98 3O 10 4 394
1 15 ll 13 2 3 44
2 0 5 9 3 21 38
3 0 0 l 8 68 77

Total 267 114 53 23 96 553

66

the inefficiency is where 3 prongs are lost, and 4 gained. That
is, a trigger track has actually been swept into the vertex of the
other event.

Another efficiency estimation comes from visible vees. The A05
which were measured are more likely to come from target generated in-
elastic events than charged prongs, which may be beam halo, delta
rays, or the result of elastic scattering. Also the relatively
short lifetime of A0 argues against them having been generated
very far upstream. In any case, if one assumes all A0 are from
target generated events and then finds how many reconstruct with
an event, this should set a lower limit on reconstruction effic-
iency for vees.

For this test, all frames which contained a 1-constraint fit
A0 and from which an APACHE fit with a legitimate trigger track
emerged were examined. Possible v contamination was first removed
from the A0 sample. In 194 of the 325 events with passing fits the
A0 was included in the fit, implying a raw efficiency of 60 i 3%.
This raw number needs to be corrected for those A0 which don't come
from the triggering event, but from a second event in the target.

In order to see how many Aos should have come from a
second, non-trigger event in the target, one needs to assume that
inelastic non—trigger events have approximately equal probability
of producing Aos as trigger events do. This assumption would be
poor for K2, since the associated strange production of K-KO pairs
is large. Thus K: would be biased towards K' trigger events.
(Indeed, when l-constraint K: were examined, 78% reconstructed

with the trigger event, much higher than the 60% for A0).

67

This bias was the reason for deriving the efficiency from the A0
rather than K: events, since AOK' cannot be a strangeness associated
pair.

Using the total n+p cross section of 24.02 i .08 mb at 16 GeV/c?9
the interaction probability for a beam n+ is 0.059. Of this, 17%

is elastic30

and will be ignored. There were an average of 5.7
identified beam n+ per spill, out of an average 7.0 beam particles on
target per pulse. From this, assuming a Poisson distribution for the
number of interactions in the target per beam spill, one can calculate
that 25% of interactions visible in triggered frames would be from
second, non-trigger events. With this correction, the reconstruction
efficiency for A0 with trigger events becomes 80 i 4 %.

Thus the two estimates of reconstruction efficiency are

79% for charged prongs and 80% for vees. The two numbers are

perhaps surprisingly close together.

I. Resolution

2 cuts depend on being able to calculate track

Because various X
resolution accurately, it is important to show that the calculated
resoltuion agrees with the actual resolution. Such an agreement is
also reassuring in demonstrating that one understands what is going on.
The spacial resolution in the streamer chamber ( approximately 0.2 mm
to 0.5 mm in X and Y and 1 mm to 3 mm in Z as determined from fiducial
measurements) depends upon how well a measurer can digitize points
on the streamer chamber film, and is discussed in Appendix III,

The quantities which are known and whose smearing can be checked

against resolution calculated from the measured point scatter are the

68

beam momentum and the invarient mass of various particles and
resonances. As will be seen the resolution correlates very well with
individual measurer's accuracy, as measured by the track root mean
square film fit (FRMS), TVGP's calculation of the measured point
scatter around the curve fitted to the track.
BeamITracks

The beam momentum bite was only approximately 1.5%. Since the
measured width ranged frOm AP/P = 8% to AP/P = 21%, the beam momentum
can be taken as fixed, and the entire spread attributed to measurement
resolution.

Three sets of beam tracks were measured:

i) Roll 793 - measured at SLAC with a film plane digitizer,
and expected to be the best set.

ii) Roll 91 - measured at Tufts with an image plane digitizer.
Each frame had several beam tracks, and ones at the edge of
the beam bundle were preferentially measured since the
central ones were too clustered to separate. They were
done by one Of our better measurers and have a good FRMS.
iii) Roll 46 - measured at Tufts with an image plane digitizer.
Film measured had only one beam track per frame. The FRMS
is a little poorer than average for these measUrements.
These tracks were reconstructed with TVGP's "film setting error" set
at a constant of 4.0, which was approximately equal to the average
FRMS of our early event measurements. As seen in Table 12, the TVGP
calculated error, when properly scaled by the measurer FRMS, agrees
very well with the actual standard deviation of the momentum measure-
ments.

That these resolutions are reasonable given the measured jitter

in reconstructing fiducials can be shown by a quick calculation. The

69
Table 12: TVGP Reconstruction of Beam Tracks

R011 46 R011 91 R011 793

Average measured beam momentum 15.5 16.3 15.3
Standard deviation of momentum

measurements 3.3 1.8 1.2
Median FRMS 3.8 2.1 1.5

TVGP 5P calculated with

 

film setting error = 4 3.3 3.9 3.2
6P x FRMS
film setting error 3.1 2.0 1.2

sagitta s of a segment of length "2” of a circle of radius of
curvature "R” is given by s m 22/(8R) . Using P= 0.3 HR, where P
is the track momentum in MeV/c, H the magnetic field in kilogauss,

and R is in cm, one finds that 8P = §§_ and that for our
P s
16000 MeV/c, l m long beam tracks in a 13 kG field, s=3.0 mm.

Taking OS to be 0.2 mm to 0.5 mm as found for (IX of the fiducial
measurements, one predicts AlVF’ = As/s to be 7% to 17%, as compared

to Alyl>of 8%, 11%, and 21% for the three beam track sets measured.

The momentum resolution thus appears to be well understood, and
is limited by the measurers point scatter. The beam tracks, however,
all travel through one small volume in the chamber, and other signals
need to be looked at to detect systematic problems such as errors in

the magnetic field map, optical distortion problems, etc.
K; and A° Resolution

The invarient mass distributions for visible vee decays are

Shown in Figure 19. The shapes of the K: and A0 peaks are

250 .-
200

“‘5 150

\

>

0)

Z 100

53

u 50

4..)

5

L: 0
100 4
80

NA

3 60

>

(D

Z

5 4O

\

:3 20

C

(U

>

w 0

Figure 19:

 

 

 

7O

 

 

400 450 500 550

M ( n+n' ) from Vee

 

 

100 110 120 130 140

M (p n") from Vee

Invarient Masses of Visible Vee Decays With the Expected
Resolution for K; and A° Superimposed

 

71

not exactly gaussian. This is understandable since vees with large
dip angles would have only short lengths of their decay tracks
visible, where as others could have nearly 1.5 m visible. Thus
what one might expect for the resolution is a sum of gaussians,
with good resolution for long tracks and poor resolution for short
ones. Thus what has been plotted for comparison purposes is a
gaussian ideogram, that is, a sum of gaussians, each gaussian
having as width the resolution calculated for one event. The
standard deviations of the gaussian ideograms of 16 MeV/c2 for
K: and 3.4 MeV/c2 for A0 agree well with the measured standard
deviations of 17 MeV/c2 and 4.2 MeV/cz, respectively.
K30 and 4 Resolution

' The standard deviation of events above background for the
K;°(892) resonance was measured to be 53 MeV/CZ. This is almost
exactly what one obtains from a convolution of a standard deviation
of 44 MeV/c2 (predicted from estimated track errors) with the
F = 50 MeV/c2 natural width of the K‘o. Again from the track error
matrices, one predicts a standard deviation of 17 MeV/c2 for the
invarient mass resolution in the 0 mass region. This again is in
good agreement with the value of 14 MeV/c2 found when fitting the
o in events with associated vees. The details of the fit to o

are in section IV 8.

IV. ANALYSIS
A. The K' trigger and K* production

Our trigger acceptance as a function Of Feynman X and transverse
momentum is shown in Figure 21. Sown are Monte Carlo acceptance
contours for K' which trigger, and Countours for 4 where O + K'K+ with
the K' triggering and the K+ identified in the Cerenkov counters. As

seen in the figure, we are sensitive to the high X region, X > 0.5.

F 4

Because our trigger acceptance is in the very forward X region,
one expects the valance quarks from the incident n+, which carry large
forward momentum, to dominate in particle production. (The u and d
which are the primary constituants of the n+ are called "valence"
quarks. This is to distinguish them from the clouds of uU, dB and 55‘
quark pairs from the vacuum, called ”sea" quarks, which are thought to
surround hadrons.) The K' (which is our trigger) does not have any
of the same valence quarks as the n+, and thus cannot be produced
by simply exchanging one of its quarks. The H'in n+ can, however,

pick up an s quark to form K79, which can then decay giving a fast

forward K', as shown in Figure 20.

{Q1
C11

 

 

 

Figure 20: Quark Line Diagram of Forward K' Production
Through the K?" Resonance

72

 

(GeV/C)

Fl

73

 

 

 
 

 

 

 

11 (GeV/c)

 

 

 

 

1.5 «-
1.0 -1
10% .2 j
20%
0.5 - ~— 30% N‘
@4070 \—
0 ( 80'%7O%6O% 50% \
I I I T
0 0.2 0.4 0.5 0.8 1,0
xK-

 

/

10%
LO“ 20%

40%7’\
0.5-- 60% ‘
70%'\
l
I I
O 2 O 4

 

 

{'50

’I I ”i
0.6 0.8 1.0
Figure 21: a) Geometric Trigger Acceptance for K"

b) Geometric Acceptance for 4 + K'K’with
K' Triggering and K Cerenkov Identified

74

Indeed, K30 production accounts for 30% of all K' triggers as shown in
Figure 22 (a). The magnitude and XFEYNMAN distrubution of the In-

28 to be

clusive K‘b production in this experiment have been shown
in good agreement with a quark fusion mode13]. When one looks at
_events with a K: , (Figures 22 (b) and 22 (c) ), one sees prominent

associated strange particle production in the channels K‘bKO and K- K*+.

as one would expect. The no associated events (Figure 22 (d) ) on
the other hand, show no hint of K50 production. Since the A0 and the
K' trigger both have strangeness -1, two pairs of strange quarks must
have been produced. 50 it is not surprising that these events are
qualitatively different from the K3 associated sample, where only
one pair of strange quarks need be produced. In fact, this striking
difference can be taken as reassurance that the event reconstruction

and vee separation are working well.

B. The 6 signal

The invarient mass of the K' trigger and each positive secondary
taken as K+ is plotted in Figure 23. NO 6 signal is apparent. How-
V
ever, Cerenkov information about the positive secondaries can be

used to remove the pion background. To illustrate that this sep—

trigger
V
Figure 24. The figure shows that when the Cerenkov counter gave a

aration works well, K n+ interpretations are plotted in
signal (indicating that the n+ interpretation was correct) a strong
_. \/

K"O signal was seen, and when a Cerenkov signal was not present
(indicating that the n+ interpretation was wrong) the Nib signal
was indeed suppressed. In order to be identified in the Cerenkov

counter, a positive secondary had to enter the counter and also had

3°C

75

 

3O

(unbuutlom I (20 HeV/Cz)
N
0

 

~

~ 4) awe-vijx—rH-x 7
WW ..
o I . ' .

115° (.8961

(“1.6961

 
 
  
   
  

a) 17°+P"’(K- ir°l+x

  

1

   
      

)1r’+P">K;+(K'rr’1+X

rm -

c) 1r'+P->K‘+(K; ir')+ x

  

[W (.892)

 

 

 

Figure 22:

0.8

1.0 1:2 ' I 4 1.6 16
Z
N" (Cell/c )

Invarient Mass of Kn System. a) M(K trigger n )
with all positive prongs as n+ b)M(K' n+)

trigger
for events containing K; c) M(K: n+) for events
with K' trigger d) M (K
containing A°

- +
trigger n ) for events

76

 

200 ‘- l_J——l_I"

150 '-

100 '-

Combinations / (20 MeV/cz)

50-

 

 

 

I 1 1 f I
1.0 1.1 1.2 1.3 1.4

2
MK-K+ (GeV/c )

Figure 23: Invarient Mass of Trigger K' with All Positive
Prongs Taken as K (Inclusive Scan Sample)

 

77

 

 

 

 

 

 

1 r 1 1 1
1o
K (892)
60 . _
7 11+ wnh Cerenkov
" signal --«
A 40 - ..
N
8 — _
:>
w 20 -- _.
2
8 "' u—I
: Q rrL._.n_
(D
E
(D
>
u.) ,4 ,0 .
60, 11 Without Cerenkov _
_. Signal __
4O — ..
20 — ..
O 1 l 1 1 4 _
700 1100 1500
MK’n"
Figure 24: Invarient Mass ova-trigger and "n+", Where n+

Are Separated by Cerenkov Signal

 

78

to have a momentum above the threshold where pions would give
Cerenkov light with good efficiency, taken to be 3 GeV/c. This cuts
down the acceptance, but Monte Carlo studies indicated that only

25% of 6's would be lost by this cut. The Monte Carlo assumed the

2
e—4.66x -2.5Pl

6 production distribution to vary as

and e The

invarient mass distributions with both kaons Cerenkov identified are

shown in Figure 25. The "inclusive" distribution contains all events
from rolls where the only criterion for measurement was a trigger
track. The plots for A0 and Kg associated events have, in addition,
events from the special vee scans. 0 peaks are apparent in the A0

and K: associated plots.

Because the 4 peak is on the edge of the phase space distribution,
some care must be taken with the background subtraction. A combin—
atorial background was generated by taking the invarient mass of the
K" from one event, and the K+ from another event. The distribution
which results is the dashed curve in Figure 25. Refinements, such
as only mixing events with the same number of prongs and swimming
events vertices through the target so that they coincide, have prac-
tically no effect on the shape of the background curve produced.

Because the number of events is small, the first fit to the 6
peak was done using the combined sample of A° and K: associated events

to improve the statistics. A x2 minimization was done to the form:
f(m) = C(m) x [ A + B x g(m,m¢,o) ]

where C(m) is the combinatorial background curve, 9 is a normal dis-
tribution around m¢ of standard deviation 0, and A and B are the rel-

ative contributions of background and 4 peak. With A, B, m¢ and o

79

 

     
 
 
    

BC) I 1 I I I

- (o) -
50 _ inclusive
40-

20-

 

(b)
with visibleA°

Events/ (20 MeV/CZ)
O

 

 

 

5- _
O 12:11 .-. .-—1

(c) 0
I0 _ Wllh v151ble Ks _
5- _
0 P15 [—1

 

 

l
1000 1200 I400
Mn. (MeV/cz)

Figure 25: Invarient Mass of K+K' System, Both Kaons Identified
in Cerenkov Counters

 

0

80

2 of 9.1 for 12 degrees of freedom was

allowed to vary, a fit with x
obtained over the region 980 MeV/c2 to 1300 MeV/cz. The fit values

were m = 1024 i 6 MeV/c2 and o = 14 i 6 MeV/cz, and the integrated

4
peak over background was 16.8 t 5.3 events. The mass thus obtained
is consistant with the 4 mass of 1019.6, and the width is consistant

with our resolution, calculated to be 17 MeV/c2 in the 0 region.

Taking m¢ and 0 fixed at the above values, the A0 associated sample.
the K: associated sample, and the inclusive sample were then each
fit, giving respectively 8.1 i 3.6 events, 8.9 i 4.1 events, and
9.1 i 14.2 events above background.

It was decided to use a normal distribution since the resolution

is larger than the o width ( F = 4 MeV/c2 ). However a fit using a

Breit-Wigner distribution produced practically identical results.

C. Sensitivity Calculation for 4

Table 13 summarizes the calculations necessary to turn the
raw event counts above into cross sections. Some of these corrections

are discussed elsewhere, and others are obvious. The rest are de-

scribed below.
Beam Scattering

+
Beam 0 which interact in the first part of the target do not see
the rest of the target. The n+p total cross section at 16 GeV/c is

24.05 i 0.28 mb, which implies that the effective target length is

61 cm 61 cm

23 ,g
e-Ci‘gpodg =J e-P. (6.022xlO )(0.0699cm3)(24.05mbc)m = 59.2 cm

0
and a correction factor of 59.2 cm/6lcm = 0.970 is needed.

81

Table 13: Sensitivity and Cross Section Calculation

AH = Atomic weight of hydrogen = 1.008 g/mole

N0 = Avagadros Number = 6.022 x 1023 Nucleons/mole
p = Target density = 0.0699 g/cm3

1 = Target length = 61.0 cm

N = Number of beam n+ = 1 255 538 849 (vee sample)

= 93 124 939 (inclusive sample)

 

 

Vee Inclusive

N t N Sample Sample

Raw Sensitivity = o p 3195 event/Mb 237 event/11b
AH
Correction factors:
Effective target length .970 .970
Beam scattered in S3 .997 .997
Dead time correction .9326 .9326
Hodoscope scintillator efficiency .482 .482
K' decay lost to trigger - .83 .83
n veto of K‘ trigger in Cerenkov .97 .97
Roll 84 Scan latch loss .97
Limit of 15 tracks per frame .922 .877
Obscured frames .982 .982
Scan efficiency .73 .72
Measurement efficiency .846 .961
Apache reconstruction efficiency .79 .79
526 event/Db 40.4 event/vb
4 with O with O inclusive

observed A0 observed K:

4 geometr1c acceptancev

 

( K’ trigger, K+ into Cerenkov) .40 .44 .36

Vee visible after 0 accepted .33 .35

Fraction of 6 into K+K' .486 .486 .486
Fraction of A° into pn' .642

Fraction of K; into n n' .686

Corrected Sensitivity 21.7 event/ub 27.0 event/pb 7.1 event/ub
Events seen 8.1 i 3.6 8.9 t 4.1 9,] i 14. 2

Cross Sections ( X¢ > 0.5) 0.37 i.16pb 0.33 i.15pb 1.3 i 2.0ub

82

Beam 0+ which interact in the S3 beam scintillation counter
would have satisfied all the beam logic, but do not transverse the
target. For 1/8 inch of plastic scintillator, this is a correction

of 0.996.

Dead Time Correction

The trigger dead time gate was on for 200 ms for each event, where-
as the beam gate was held off for only 20 ms to 60 ms. Thus for some-
where between 140 ms and 180 ms beam 0+ would count, but events could
not trigger. At 110 beam spills per second with 5.7 n+ per spill,
between 88 and 113 beam 0+ would count during the dead time. These
counts, amounting to a few percent of the beam n+ per trigger, should

be subtracted out when calculating cross sections.

Actual beam conditions varied widely. Thus an estimate as cal—
culated above, although sufficient to show that the effect is important,
could not be made accurate. A good calculation can be done, however,
by looking at the distribution of n+ beam between events.

The distribution of the number of beam n+ since the previous
event is ploted in Figure 26. One would expect a pure exponential
fall off if every beam particle had the same probability of producing
a trigger event. As seen in the graph, an exponential fits very well
except near zero, where the loss of triggers is attributed to the
dead time problem.

A x2 minimization fit was done using the form

1n y e - X x + b
where y is the number of events and x is the number of beam pions

for the event. With the first three bins excluded, the fit gave

83

20,000'--1

10,000... 'e

1111

a
.e
e
e
5000 — 'e.
... '0.
.e
__ e
e..
a...
— 'e
P. ...

U1 ..
8 a.
g 1000 ...... ‘5
E ; Ia
+.> — 5.
at) -— e
,3, 500 — x.

_ 0 e

1
:,

 

 

.e'
e
.... 'e
0.:
..
. e
100__ ..e
— ...
... O’e.
—- e
50 : I .' ’
"" e
30
I I I I I I I l l 1
O 5000 10,000 15 .000 20 ,000 25,000

Beam 0+ per event trigger

Figure 26: Distribution of Beam n+ per Event Trigger

84

X = 0.00024456 and o = 9.8181 with a x2 of 128 for 95 degrees of
freedom. The plot contained 278,468 events from 1,204,019,250 beam
pions. The dip in the plot near zero would be filled in up to the
extrapolated fit by an extra 21,091 i 144 events from 3,930,375

more beam pions. Thus the overall event rate should be scaled up

by 1.0723 i .0005 to compensate for the dead time.

Hodoscope Scintillation Counter Efficiency

An average hodoscope efficiency was calculated for K' triggers
from O by averaging the measured efficiency for the 40 four-fold
scintillation counter coincidence channels, each weighted by the 6

Monte Carlo predicted probability of hitting that channel (Table 14).

K' decay and w+ veto

If the K" from o decay itself decays before traversing the
hodoscope and Cerenkov counter, a potential trigger is lost. A
Monte Carlo study of this process indicates that 17% of K' were lost
this way, giving a correction factor of 0.83. The same study indicated
a 3% probability that a high momentum charged pion would hit the same
Cerenkov cell as a K" trigger track, producing light which would veto

the trigger.
Measuring Corrections

Due to a computer programming error, a sample of events on roll
84 were incorrectly rejected during scanning. These events correspond
to 3% of the inclusive sample. Unfortunately this roll had already
been sent to Jammu University, India, when the error was discovered,

so the rejected events could not be measured. The 3% estimate is made

85

Table 14: Weighted Average Hodoscope Efficiency

Coincidence Weight(fraction Average efficiency Weight

channel of K- triggers of channel (begin- x

number from 6 into each ning and end, up Efficiency

channel) and down)

1 .098 .431 .042
2 .122 .663 .081
3 .104 .446 .046
4 .094 .364 .034
5 .079 .409 .032
6 .073 .477 .035
7 .048 .703 .034
8 .049 .467 .023
9 .039 .468 .018
10 .046 .535 .025
11 .039 .515 .020
12 .049 .669 .033
13 .033 .528 .018
14 .018 .489 .009
15 .024 .283 .007
16 .021 .407 .009
17 .021 .142 .003
18 .013 .347 .005
19 .022 .388 .008
20

.011 .196 .002

Weighted Average Efficiency .482

86

by comparison to other rolls still at Tufts where the rejected events
were rescanned and measured.

In some frames the number of prongs becomes very large. This
is the result of at least two, and often three or more interactions,
plus halo tracks and delta-rays in the same frame. At some point it
becomes impossible to extract useful events from the jumble. Although
written onto the scan/measure lists, frames with more than 15 prongs
were not measured. For the vee sample, 2014 of 25910 frames initially
selected were rejected for this reason. Events reconstructed from
frames with 12, 13 or 14 measured prongs end up with numbers of
prongs in the final fits ( Figure 27 ) which are more or less char—
acteristic of the entire sample. Assuming frames with more than 15
prongs are not dissimilar to the 13 and 14 prong frames, then weighting
all events up equally by the fraction not measured should not skew the
results. For the inclusive sample, 1044 out of 8488 frames had more
than 15 prongs, giving a correction factor of .877 i .004.

Also not measured were those events where a fiducial was obscured
by a flare. This is a small correction, as only 102 events of the nor-
mal, inclusive sample were rejected for this reason, compared to 5824
measured.

Scanning, measuring, and APACHE reconstruction efficiencies have

already been discussed.

87

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

30% - From frames with 11 30% -1 From frames with 12
measured prongs measured prongs
20% .1 20% _
10% .. 10% .1
0% 0%
5 10 5 10
Prongs in Final Fit Prongs in Final Fit
40% q
30% ' From frames with 13 '1 Frmn all events
measured prongs
30% '-
20% ., "'
20% ‘
10% d
L.l-1- 10% .-
0% 0%
5 10 .5 . .
Prongs in Final Fit Prongs 1n Final Fit

Figure 27: Number of Prongs in Final Fit as a
Percentage of Events Measured

88

Geometric Acceptance

The geometric acceptance for the 4 sample was calculated by
taking each reconstructed event in the 0 region, doing many random
rotations of the event and projecting the results through the apparatus.
The geometric acceptance is the percentage of these random trials
which would be accepted by the apparatus. To be accepted for the o
with vee sample, three conditions had to be met: the K' from the o
decay must have hit corresponding elements of the scintillator hodo-
scope, the K+ from the o decay must have entered a Cerenkov cell
with momentum greater than 3 GeV/c, and the vee decay must have been
visible in the streamer chamber. Because the vee momentum four-vector
and the o momentum four vector may be correlated, the 4 and vee accept—
ances were calculated together rather than being separately Monte
Carlo simulated.

More specifically, events with a trigger K' and a Cerenkov iden-
tified K+ which have invarient mass less than 1080 MeV/c2 were taken
as 4 candidates. The o momentum and the angle between 6 and the beam
axis were considered sacred. If the event contained a vee, then the vee
momentum, the angle between the vee and the beam axis, and the angle
between the vee and the 4 were also taken as fixed. The entire event
was then randomly rotated around the beam axis. The 4 decay was then
generated, with K' produced isotropically in the 0 center of the mass.
The K' was then projected through the scintillator hodoscope. If it
did not hit corresponding elements of the hodoscope, the trial event
was rejected. Otherwise, the program proceeded to see if the K+,

produced opposite to the K' in the 6 center of mass, had momentum

89

greater then 3 GeV/c and hit a Cerenkov cell. If not, the trial event
was rejected. If it passed this test, then it was counted as an
accepted 0 trial, and the program would proceed to check whether the
vee would still be accepted. This was done by generating a vee decay

vertex along the rotated vee direction according to the decay distance

distribution E$11M 0 < random number < 1 ), where

p is the vee momentum, m is the mass and t'the known lifetime of A°
or K: . The generated vee decay vertex was then checked to see if it
was inside a fiducial volume, given in Table 15, which was within
the streamer chamber but not obsured by the target and O-ray shield.
Thus the trial passed if both the O and the vee from the rotated
event were accepted.

2000 trial rotation-decays were done for each 0 with vee candidate
and 500 were done for each 0 candidate that wasn't vee associated,

giving the results in Table 13.

D. 6 Sample Contamination

Because we are looking for final states where strangenss is
created from n+p interactions, contamination of the beam by particles
carrying strangeness could be a serious problem. The beam trigger K+
contamination is estimated in the beam line Cerenkov counter appendix

to be 1.5%. The ABBCCHILVW collaboration give 32

0(K'p)+¢+X

 

= 13.05 i 2.5
olfip1+d+x

90

Table 15: Vee Decay Fiducial Volume for Geometric Acceptance
Calculations

Inside fiducial volume of streamer chamber if:

I X | < 30.0 cm
and | Y | < 75.0 cm
and I Z - 41.2 I < 20.0 cm

Obsured by target O-ray shield if:
I 0.9999 ( X - 0.29 ) - 0.011 (Y + 96.9) | < 2.2 cm
and | 0.011 ( X - 0.29 ) + 0.9999 ( Y + 96.9 ) I < 102.1 cm
and | -0.006 ( Y + 96.9 ) + 0.9999 (Z) | < 52.0 cm

91

for XF > 0.5 with 16 GeV/c beams. Assuming that inclusive 0 pro-
duction from K+ is equal to that from K', then (1.5%) x 13 = 19.5%

of our inclusive 0 signal would come from K+p interactions. K°¢
events could also be produced from K+p with the creation of one pair
of strange quarks. A° on the other hand has the opposite strangeness

of K+, and ¢A° production would have to be through the creation of

two pairs of strange quarks, the same as in the n+p interactions.
This suggeststhat the 6A0 sample is free of any beam contamination

problem.

E. Conclusions
The inclusive 4 cross section in n+p at 16 GeV/c is calculated to

be 1.3 i 2.0 ub for X¢ > 0.5. Ghidini et a1.33 measured inclusive

0 production in n'p interactions at 16 GeV/c. Integrating their X dis-

tribution over 0.5 < X¢ < 1.0 gives a cross section of approximately

4.0 i 0.6 ub. These results are not inconsistant given the large
uncertainty in our measurement. Further, the inclusive 0 cross section

in n+p would be expected to be somewhat less than that in n'p if val-

31

ence quark annihilation is important in 4 production. The ABBCCHILV

collaboration measured32

the inclusive 6 cross section in 16 GeV/c
+ . . . .
n 1nteract1on to be 130 i 30 ub. Compar1son with our measurement of

1.3 ub for X > 0.5 indicates that 0 production is very central in

4
n+p interactions.

In order to compare the conjoint production with non-conjoint, we
need to estimate conjoint production in channels we did not identify,
such as o K+K' . Table 16 lists the channels with the lowest mass

strange mesons and hyperons. The number of each type Of visible decay

 

92

Table 16: Table of Stange Particle Pairs Which Can Be Produced
Conjointly With 6

Strange particle Height of Fraction resulting in visible
pair channel visible vee
o o o o
Ks A Ks Ks
k*k° 1 - - -

o 0
“1‘1
K° k°
KORO Ki; S 1 I, . 1‘ x 29
s "i .
K2 K2
K+A° 1 . 1 .
O O
A
k°1° {K15 o} 1 ‘1 l -
Ks A
18:" {2° - 71°y} 1 - 1 .
O O
2
K020 {Kb } 1 is 1 a
x 2°
s
(2* 1 - - -
o 4'
x°z‘ KL 1:
K0 _+ 1 1, - -
S I.
K 2' 1 . . .
o c
h _ 2
K E {K18 .} 1 H - -
Ks 2
Totals 12 311 4 5
o I. 4
Fraction with K8 T2—

Fraction with 11° - 12‘—

o
" the factor of two is because we may see either of the Ks.

93

are listed for each channel. Note that because 20 decays to A0 y

9 cm, 20 production

nearly 100% of the time with a CT of 1.7 x 10'
is here indistinguishable from direct A0 production, and is included
in the A0 column. Without some theory to indicate the relative
strengths of these channels, we make the simple assumption that all
channels contribute equally. One check on this assumption is that the
table indicates we should see equal numbers of K: and A0, and our
actual measurement of o (O + A0) = 0.37 f 0.16 ub and o (O + K?) =
0.33 0 0.15 pb indicates that this is correct within our statistical
errors. Thus scaling up the conjoint cross section by the unseen
channels, we estimate (for X¢> 0.5):
o(o conjoint) = (0.37 + 0.33) (1%) ab = 1.1 nb

Comparison of this with our measurement of 1.3 1 2.0 pb 0 total
inclusive cross section for X¢>0.5 indicates that OZI allowed conjoint
processes are a large fraction of the total.

Our data indicate that inclusive 4 production at large X is
predominantly by the OZI allowed conjoint mechanism. Although there

have beend attempts34

to justify the OZI rule using quantum
chromodynamics (0CD), the OZI rule remains more or less an
empirical description of the results of processes whose underlying
mechanisms are not known. It is thus interesting to see what
implications this result has for actual dynamical models. In

terms of quark and gluon fusion models, Kinnunen31 lists four

processes (see Figure 28) which would contribute to 6 production.

94

FUSION 0F STRANGE SEA QUARKS:

ANNIHILATION OF LIGHT QUARKS:

 

 

_CE

GLUON FUSION:

 

Figure 28: Mechanisms for 0 Production

95

These are

i) the fusion of a strange quark from the sea of one incident
hadron with a strange antiquark from the sea of the other
hadron. The left over s and 3' will make other strange part-
icles, and this is thus a conjoint process.

ii) annihilation of non-strange (predominatly valence) quark-
antiquark pairs to produce 4 through a three gluon inter-
mediate state. Three gluons are needed since the 4 has
J=l and two massless gauge bosons cannot form a J=l state.

iii) annhihlation of light quarks producting a virtual gluon which
decays into an 55' pair. Subsequent emission of a soft gluon
can then produce a colorless state.

iv) gluon fusion, to form a virtual gluon or exchange a quark.

Again, subsequent soft gluon emmission is required.
Our data would indicate that the first process is dominant, since itis
the only one that would produce the O conjointly with extra strange part-
icles.
In order to calculate the magnitude of the conjoint cross section

in the fusion model, one needs to know the sea quark distribution

functions for the proton and the pion. .The valence quark distribut-

ions are found35

using deep inelastic lepton nucleon data. Because the
nucleon structure function is not very sensitive to the sea contribution,
the sea distributions for the proton are not as well known. The sea
quark distributions for the pion are even less well known. Using the
sea quark distribution functions listed by Kinnunen, one obtains a con-
joint 6 cross section of 0.22 pb for X > 0.5, a factor of five less

4
than our data. However, because of the previous scarcity of data sen-

96

sitive to the pion strange sea quark content, it would perhaps be
better to estimate the strange sea from our results.
Using the shape of the X distribution for 4 from the SE' fusion

mechanism, one finds 1.2% of the production is greater than X¢= 0.5. Our

result of 1.1 ub conjoint 0 production for X > 0.5 would then indicate

4
a total conjoint cross section of m 90 ub. This much of the 130 pb

total cross section being conjoint would be consistant with the results16
of the ACCMOR collaboration, who found conjoint production to be > 40%

of central inclusive 4 production in 93 GeV/c n-p interactions.

V. SUMMARY
We have analyzed 4 production in n+p interactions at 16 GeV/c.
The cross section for inclusive 0 production for X > 0.5 is measured

4
to be 1.3 i 2.0 ub. Comparison of our result with the total cross

32

section of 130 i 30 ub indicates that 6 production is very central.

In addition, we find significant conjoint production. For X¢ > 0.5,
we found 0 ( n+p + ¢A° + anything ) to be 0.37 i 0.16 ub and o ( n+p
+ 0K: + anything ) to be 0.33 i 0.15 ub. With an estimate of unseen
conjoint channels, the cross section for producing pairs of strange
particles conjointly with 4 would be 1.1 pb for X¢ > 0.5. This is a
large fraction of the 1.3 1 2.0 ub inclusive 4 cross section found, in
agreement with expectations from the OZI rule. 0f the models for
inclusive o production (strange sea quark fusion, quark annihilation,
gluon fusion), only strange sea quark fusion produces ¢ conjointly.
Our results therefore indicate that the fusion mechanism dominates,
or is at least as strong as, quark annihilation and gluon fusion in
0 production in n+p interactions.

Finally, one may speculate on the implications of our results
on hadronic 4 production. The 4 is the CE analog of 4. 0n the
basis of the OZI rule, it had been speculated11 that the 4 would be
produced primarily with extra pairs of charmed particles. Such
conjoint production was not observed36. Recent experiments37 now
indicate that the 4 may be produced dominantly by the electromagnetic
decay of x states. Thus conjoint production could still be a large

fraction of the direct production of u, but be obscured by the

competing radiative X decays.

97

APPENDICES

APPENDIX I. BEAM LINE CERENKOV COUNTER

A differential Cerenkov counter (Figure 29) was used to distin-
guish between pions, kaons and protons in the beam. At a given beam
momentum, particles of different mass have different velocities, and
hence radiate Cerenkov light at different angles. Specifically, the
Cerenkov angle 0C is given by cos 0c =§%—-, where B = V/c and n is
the index of refraction. By means of a lens and a mask, the counter

accepts light at one fixed CC = 6 By variying the pressure of

mask ‘
the gas in the counter, the index of refraction can be changed,
allowing selection of the particle type which will radiate light at
emask and be counted.

Freon 13 was the gas used in the beam line Cerenkov counter.
A complete pressure spectrum, shown in Figure 31, was taken to verify
counter operation. During normal data taking, the counter was kept

at 13 PSIG to select beam n+. A fit to the spectrum also allows us

to determine beam momentum.and K+ contamination at 13 PSIG.

A. Pressure Spectrum Fit

The spectrum peak shapes depend on the mask and lens geometry.
As the pressure (and hence the index of refraction) increases the
ring Of light on the mask will edge into the opening, cross the
opening, then edge off the other side. We choose to fit the peak
with a function which has a flat top (corresponding to the ring Of
light being entirely inside the mask opening) of half width HW and
gaussian edges (due to some resolution smearing by spread in beam
collimation, beam momentum spread, etc.) with standard deviation 0.

This shape is sketched in Figure 30.

98

99

 

 

Q)

1 l:

O

U

L

$- '6

Q) 7—

ep z

Ln!- E
NCL

mud) U

COP-13 OJ

:3 N

<5“ ..-

Uo C

014-) "-

O E

..C 3

a. r-

, <:

1:

— ~
\misf
\

 

\3 9

 

 

 

 

Cylindrical Mirrored Walls

Beam
~-0C
\
\

\

 

 

 

Figure 29: Differential Beam Line Cerenkov Counter

 

1..-
.1-..

1
1
I
l
l
1
1

1,, I--———-

Pi

eak Center
Pressure

Figure 30: Shape Used to Fit Beam Line Cerenkov Counter
Pressure Spectrum Peaks

Fraction of Beam Particles Giving Cerenkov Signal

100

to

 

1% d

1 I
I

-

 

 

T l l 1 l

10.00 22.00 34.00 46.00 58.00 70.00
Pressure (PSIA)

Figure 31: Differential Beam Line Cerenkov Counter
Pressure Spectrum with Fit

101

The peak positions depend on the angle selected by the mask

 

(Bmask), the particle momentum (P) and particle mass(mi) according to
-_1__-1
COS (emask) - 8n —n P
P2 + m?

where the index of refraction n contains the pressure dependence.
To obtain this dependence explicitly, an analytic expression was
fitted to the index of refraction versus pressure measurements of

38

Hayes, Schuter and Tamosaitis for Freon 13. The parameterization

p = a ( n-l )2 + b ( n—l ) (PSIA)

gave a = -266370 and b = 20233. (Index of refraction measurements
for the wavelength 3650 A , matching the peak spectal response of
the RCA 8575 photomultiplier tube used in the counter, were used
for the fit). Thus peaks in the pressure spectrum for pions, kaons

and protons are expected at

- b (PSIA), i= n,K,p

 

2
[ b 1 + 1 Pz+mi 2
1

4a 2 2 cos (0 ) P

mask

Taking N“, NK and NP to be the fractions of pions, kaons and

protons in the beam, the fraction of beam particles, N(p), which

should count at a given pressure is then given by

2 2
N(p) = z N. e ‘(ai) (201

where “i = 0 if | p-pi | < HW (Flat top)

102

ai = I P-Pi l - HW if I p-pi | > HW (Gaussian edges)

This shows explicitly the spectrum dependence on the parameters N”,

dependency is in the peak

N N , HW and o. The momentum and e

K’ p mask

pOSItions, pi= pi (e P, mi).

mask’
One final refinement was added. The peak widths are slightly
pressure dependent. If the pressure widths are taken to be caused

by the A0 across the mask opening, then the ratio of the widths

 

 

should be
3P / /
HW] 534p] B1 (2a-b+ b2+4ap])2 b2+4ap1
H 2 421 8 (2a-b+/b2+4ap )2 Vb2+4ap
30 p2 2 2 2

For our final fit, this gave HW1T : HWK : HWp = 1.67: 1.65:

1.60 for the pion, kaon and proton peak pressures Of 12.66, 21.15,

44.69 PSIG respectively.

Table 17 shows the values obtained from the seven parameter
X2 minimization. The fitted curve is plotted in Figure 31. The
x2 of 1145 for 28 included data points and seven parameters seems
to be large mostly because the gaussian edged don't follow peak
shape exactly. The peak positions are well fitted.

The fit error in the momentum determination is estimated from
Figure 32 (x2 vs momentum). If the X2 scale is renormalized to 1
per Degee of Freeedom, an increase of 1 (8x2 = 54 on the graph)

gives AP = i 0.08 GeV/c. ‘The temperature dependence of the fit

momentum is shown in Figure 33. Since the temperature was not

103

Table 17: X2 Minimization Fit to Beam Cerenkov Spectrum

Fitted Estimate for
Parameter Value nggarison
P (momentum) 16.022.13 CeV/c 15.5:.8 from beam track measure-
ments
amask (Cerenkov angle 2.96° 2.7S° from measured mask size
accepted by mask) and distance to lens
HW1r (Flat top half width 1.67 PSIA

+
of a peak) 6.0 PSIA calculated pressure change

for light cone to cross
width of anular ring mask.

a (o of gaussian edges 2.30 PSIA
peak) should be between 2 (NW)

of:

N“. (.‘ beam fraction) .858 ‘"° 21”") * 2°
"KT (K+ beam fraction) .081
Np+ (p+ beam fraction) .037

Other quantities calculated from fit parameters

P («I pressure peak) 12.66 PSIG
K (K+ pressure peak) 21.15 PSIG

P (Proton pressure 44.69 PSIG
peak)

HHK (Half width of k+ 1.65 PSI
peak flat top)

HHP (Half width of proton 1.60 PSI
peak flat top)

°K (a of K+ peak 2.27 PSI
gaussian edges)

o (o of proton peak 1.20 PS1
gaussian edges)

1500

1400

1300

X 1200

1100

Figure 32:

16.4

16.2

16.0

15.8

Plab (GeV/c)

15.6

Figure 33:

104

*' \

 

 

_’ C)
_ O... ---7/
J L l l 1
15.84 15.94 16.04 16.14 16.24

Beam Momentum (GeV/c)

x2 of Beam Line Cerenkov Counter Spectrum
Fit Versus Beam Momentum

 

 

_ o
T e
_ C
C

b e

1 J l l l l

15 20 25 30 35 40

Temperature (0C)

Temperature Dependence of Beam Momentum Fit

105

measured, we add a systematic error Of i .l GeV/c to cover the prob-

able temperature range.

B. Beam K+ Contamination

Since this experiment looks for strange particles produced in
n+p interactions, beam K+ not vetoed by the beam Cerenkov counter
could cause serious background because K+ carry strangeness. In
estimating this contamination, three processes are examined.

In Figure 31, a more or less flat region is seen between the
kaon and proton peaks. No particles should be giving Cerenkov
light through the mask at these pressures. This background of 0.68
i 0.02% may be due to scintillation light in the Cerenkov gas,
photomultiplier tube noise, etc. In any case, if we assume that
these counts happen randomly, then 0.68% of the beam kaons, which
are 8.1% of the beam, have noise counts. Thus .055% of the beam is
K+ contamination.

Off axis kaons could also give (Cerenkov counts. The beam
spread is supposed to be the major source of the gaussian tails of
the spectrum peaks. Extrapolation of the fitted kaon tail for various
conditions is shown in Table 18. During the run, the counter pres-
sure was nominally 13.0 : 0.5 PSIG. Actually, it never got above
13.0 PSIG, although due to a very slow leak it occasionally got as
low as 12.0 PSIG. If an operating temperature of 260C is assumed,
the tail then gives a K+ contamination of 0.17% of beam triggers.

If a beam n+ is in time with a beam K+, then n+ gives the
Cerenkov light necessary so that a K+p event can trigger. There were

an average 9.6 particles (S1 - $2 coincidences) per 1.6 us beam spill.

106

Table 18:,, K+ Beam Triggers as a Percentage of 11+ Beam Triggers From
Beam Line Cerenkov Counter Fit Parameters

Beam Line Cerenkov Counter Pressure

 

 

 

Temperature 12 PSIG 13 PSIG l4 PSIG
16 °C .2211 i .0009% .662 i .002 1.6 i .005
21 °C .1046 i .0004 .351 i .001 .974 i .003
26 °C .0463 i .0002 .1738 i .0007 .540 i .002
31 °C .0191 i .0001 .0804 i .0004 .200 i .001
36 °C .00738 t .00004 .0348 i .0002 .1356 i .0006

107

86% of these were pions. Since the Cerenkov counter pulses were 20
ns wide and had to have a 2 ns overlap with the 8 ns beam X+ coinci-

dence, a kaon could trigger

26 ns x 9.6 x 86%
= 13%

 

1.6 us
of the time. Since kaons were 8.1% of the beam, this gives a kaon
contamination Of 1.1%.
Taking these three sources together, and using the fact that
only 86% of beam X+ gave beam triggers, it is estimated that 1.5%

of beam triggers were due to K+ contamination.

APPENDIX II. LARGE APERTURE CERENKOV COUNTERS

A. Physical Configuration and Gas System

The physical configuration of the Cerenkov counter is illustra-
ted in Figure 10. Each counter was 2.25 m long by 1.75 m wide and
had an entrance aperture 1 m and an exit aperture 1.5 m high. The
windows were made with two layers of 76 pm thick aluminized mylar
and two layers of 102 pm thick black polyethylene film. Since the
counters were designed to contain isobutane gas, it was vital that
the tank be capable of sustaining sufficient over-pressure to prevent
an accidental backfilling with air which could lead to an explosive
mixture. The high tensile strength of mylar plus the opaqueness of
black polyethylene film provides an excellent combination for a
light-weight window. The tank, with the exception of two 6.4 mm
aluminum plates supporting the ten phototube-Winston collector-cone
assemblies, was welded from 3.2 mm aluminum sheets and extruded
aluminum channels. Each of the ten phototubes, with its associated
cone was responsible for monitoring the Cerenkov light produced by
charged particles entering the entrance window as reflected by the
corresponding mirror.

The counter had an interior volume of about 5.7 m3. A gas
manifold and a gas sampling port were provided on the top as well
as on the bottom of the tank. Since isobutane is denser than air,
the gas was introduced at the bottom and exhausted at the top
during filling. A small quantity of the gas was drawn continuously
from a sampling port at the top of the tank and passed through a

laser interferometer of the Rayleigh type. Index-of—refraction

108

109

measurements during one of the filling operations is shown in Figure
34 as solid dots. Since the isobutane gas was introduced from the
bottom of the tank and the interferometer samples were taken from
the top, the curve indicates that a typical fill time was about

four hours. Isobutane is normally supplied in a liquid state in

a cylinder. The depletion of the supply was measured by weighing
the isobutane bottle. A weight measurement is also shown in Figure
34 as open circles and indicates that about 20 kg of isobutane was

required for each filling.

B. Interferometer
The dual interferometer and CAMAC interface used to continuously

monitor the index of refraction of the two Cerenkov counters were
designed and built at M.S.U. As shown in Figure 35, a laser beam
was split,part going through an evacuated reference cell, another
part through the gas sample cell. The beams were merged using the
two lenses, producing a fringe pattern. The fringes were magnified
onto the detector. (The mirrors shown were used only to fold the
optics and thus conserve space).

Initial calibration was done by evacuating the sample cell and
zeroing the counter. Gas was then slowly allowed into the cell.
The cell was 6.35 cm long and a Ne-He laser (X = 6328 A) was used.
Hence the number of fringes counted was approximately (n-l) x105,
where n is the index of refraction of the gas.

In the continuous mode, increases and decreases of the index of

refraction had to be automatically differentiated. This was accom-

plished by monitoring the direction of fringe movement, using two

110

 

 

 

 

15C) 1 I I I 1
E -—60
100- g
:5
.. -40 g
9 3..
x 28
"f 5.2.
5 5
as
5()- .____;> c:
- 20
l I l l I
o 2 4 60

Time (hours)

Figure 34: Cerenkov Counter Gas Index of Refraction As
Measured During a Gas Filling Operation. The
open circles show the loss of weight of the
isobutane bottle during the filling operation

 

111

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P1
0
t
S
t .1I
ml sr
O ne
0. S at
sn min
ntne tu
oeeqllroo
.1111 etc
tt €410
aUXthe
”09.0.1019
C vcn n
asnngo..l
vaooawr
EGCCMTf
7890.12
1].]:
ml
e .l.
5 1|-
a 8
IL C
m e
u C
.1... n
..IISPIe
erer
HOte
rtf
nwl.11e t
0.1]r1lle
emp 1II1II
N smen
e “Cc-I
wnmu
m aaCSS
Ieaaa
SPBVGG

Dual Interferometer

Figure 35:

112

phototransistors in each detector. This logic fed an up~down counter,
which was visually displayed and also recorded along with latch infor-
mation for every streamer chamber picture. A block diagram of the
electronics is shown in Figure 36.

The unit was compact, and relatively easy to align. It was re-
markably immune to vibration and drift, probably because both parts
of the beam travel through the same optics. That is, when a mirror
or lens vibrates, or moves because of temperature expansion, both

beams are affected and the disturbance cancels.

C. Light Collection and Optical System

The Cerenkov radiation produced by charged particles entering
the upstream window was viewed by a system of ten 73.8 cm high
acrylic mirrors. The widths of the mirrors, four each of 37.80
cm and 30.7l cm and two each of 26.59 cm, were tailored specifically
to the geometry of the experiment. The light focussed by each mirror
was further concentrated into a ll.5 cm spot using a 2.5x Winston
cone39. The mirrors as well as the cones were manufactured by the
Plastics Shop at the Argonne National Laboratory.

The mirrors were pressure formed with 6.4 mm acrylic sheets in
an oven to a nominal radius of 2 m. The blanks were then coated
with a thin aluminum film by evaporation in a vacuum chamber. The
bank of ten mirrors was arranged in the form of a 2 x 5 matrix, 2 m
downstream from the entrance window. The mirrors were each respon-
sible for focussing Cerenkov light into one of the ten corresponding
Winston cones.The cones had openings of diameter 25 cm at the en-

trance and ll.5 cm at the exit and were made of successive epoxy re-

sin layers built up on a spinning aluminum mandrel.when the thickness

113

Block Diagram of Fringe Counter:

 

 

 

 

‘Detector Counter f Up Counter i Output

. D‘
Houswng - Control m agaplgymh Buffer

 

 

 

 

D [7 Update
[Update

 

 

 

'Control

 

 

 

 

 

 

 

 

One of Two Phototransistor Curcuits in Each Fringe Detector Housing:

“5'
“AA +I5v
SIM-A

LF355N

7—

x-
#

LED

 

'1
:JW

Counter Control Circuit:

 

 

.'

we».
W4’ :r:.

Figure 36: Interferometer Fringe Counter Electronics

”A

 

 

 

 

 

 

 

I L“.

SIGNAL
Zfi!‘ r
I ZszzZ
‘ -15v
3 18 K-n-
‘ INDIcM‘oR

“Count
Up" to
Counter

"Count
Down“ to
. nter

 

114

reached about 3 mm, the cones were heat-released and coated with a

layer of lacquer and then aluminized in a vacuum chamber.
A flange holding a UVT-lucite plano-concave lens was affixed
to the narrow exit end of each cone. The flat side of the window
was coated with a wavelength shifter, paraterphenyl, which served
to convert the primarily ultraviolet Cerenkov radiation to a visible

light40

, thus providing a better match to the frequency response

of phototubes having glass windows. The concave side of the lens
had a radius of curvature of 12.95 cm congruent to the front window
on a RCA 4522 phototube. The lens was held between two aluminum
rings. The ring closer to the phototube was connected electrically
to the photocathode potential and mechanically fastened, using

6.35 mm diameter threaded G-lO glass epoxy rods, to the other ring
which in turn was glued to the Winston cone and was electrically
grounded. The space between the two rings, (l.59 cm thick) was
filled with black RTV silicone rubber which, in addition to being

a light shield also served to remove surface potential gradients

from the vicinity of the photocathode and forestall noise from dis-

charges.

D. Magnetic Shielding

While magnetic shielding of large phototubes is, in general,
desirable to ensure optimum efficiency of collection of photo-
electrons in the presence of the earth's magnetic field, it was
essential in this experiment where few electron statistical fluc-
tuations were critical. The SLAC streamer chamber magnet had a

stray ambient field of about 50 gauss, and this field was known to

 

115

lie skew to the phototube axis and thus to require compensation of
both parallel and transverse components. The lightweight construc-
tion of the Cerenkov counter tank, with phototubes externally mounted
on relatively brittle epoxy Winston cones, mechanically precluded
conventional massive solid-iron shielding of either the individual
tubes or of the entire detector. These would have been expensive

in any case. In addition, the mounting of massive iron shields
would have interfered with adjacent counter elements. Consequently,
a relatively light and inexpensive multi-layer shield was developed
to cover the individual phototuves and associated Winston cones.

The primary shield element was cylindrical with an outside
diameter of 30.5 cm and a length of 76.2 cm. The inside diameter was
l7.8 cm and was tapered out gradually toward one end over an inside
length of 43.2 cm. to accomodate the Winston cone while providing
maximal shielding coverage of the tube up to its photocathode. The
other end of the shield was flat. The structure was composed of
eleven thin concentric cylinders of successive radii differing by
0.63 cm. The individual cylinders were formed from sheets of trans-
former iron ( a non-oriented electrical steel) of thickness 0.58 mm.
The sheets were cut to size while flat, rolled to an approximate
diameter, and then riveted to more exact diameter while mounted on
a simple jig, with about l.9 cm overlap. The final structure was
then assembled using at each end four radial strips of aluminum in
which appropriate saw cuts had been made to receive the edges of the
respective cylinders. These maintained the appropriate spacings and
overall conformation of the assembly. Longitudinal seams were at

the same azimuth to permit alignment parallel to the field and minimize

 

116

the flux through them. The spacers were then glued in place to the
shields with epoxy resin. The final unit weighed about 24 kg. This
weight, and the simple cylindrical outer shape, made it possible to
dismount the shields from the concentric phototubes and inner shield
assembly simply by resting the former on parallel rails made of an
aluminum channel and sliding it off longitudinally whenever an interior
access was required.

The transformer iron used was a mixture of ARMCO - M-l9 and
M-36 steels, expected to have permeability of the order of 500 at
fields of a few tens of gauss, a coercive force of less than l
Oersted, and to saturate at about 20 kgauss. In the present shield
design it was thus a very convenient compromise between cold-rolled
steel, with a permeability of about l80, and mu-metal and similar
expensive alloys which have much higher permeability but saturate
at about a the field. Indeed, saturation will in any case limit the
utility of thin high-permeability shields in ambient fields above a
few gauss. While some improvement might have been obtained using
a grain-oriented steel instead of non-oriented transformer iron, the
additional expense and complex annealing required to realize the 2-3
fold higher permeabilities after fabrication were not deemed to be
necessary for the present system. An exact calculation based upon
ll infinitely long concentric cylinders of the sizes used and a con-
stant permeability of 500 indicated that a 50-gauss transverse field
would be reduced to 0.8 gauss inside, with a maximum induction of
5460 gauss in the Sides of the outer cylinder. Thus, the design
appeared to be conservative magnetically, as well as in cost and

weight.

 

 

 

117

The phototube itself was mounted inside a separate mu-metal
shield of diameter l4 cm which extended backwards 30.5 cm from the
region of the photocathode, with an additional l4 cm at a reduced
diameter. This shield also formed the mechanical and optical en-
closure of the phototube, and contained the base and resistor string
as well as the phototube itself. With an outside diameter of l6 cm,
this assembly permitted easy removal of the l7.8 cm inside-diameter
outer shield unit described above. An exact calculation for the
magnetic shielding by an idealized outer assembly indicated that
the addition of this inner mu—metal shield, with presumed perme—
ability of 20,000 would further reduce a 50-gauss external transverse
field to about l2 milligauss, with at most 237 gauss induction in the
mu-metal. Although these calculations did not reflect in detail
either the complex dependiencies of permeability upon the field,
or the end-effects of the geometry used, they did provide grounds
for confidence that the overall structure would give a satisfactory
reduction of the transverse component of the external field.

The longitudinal field component, whose approximate effect was
expected to be the longitudinal polarization of the inner mu—metal
shield surrounding the phototube, was compensated by winding a
coaxial coil of 60 turns of insulated wire (AWG #l8 and #20) on
the outside of this inner shield to achieve a bucking field of
about 0.09 gauss/ampere-turn inside the phototube. The actual
bucking current for each tube was established empirically by max-
imizing the Cerenkov pulse height with the external magnetic field
present. The currents required ranged between 0.3 and l.8 amperes

for the 20 phototubes used on the two identical Cerenkov counters

 

118

in this experiment.

E. Performance Characteristics and Analysis of Test Data

During a test run period, a 9 GeV/c n- beam was used to measure
the efficiency of each of the twenty Cerenkov cells with an atmos-
pheric-pressure isobutane filling. At each of the four crossing
points on an imaginary 3 x 3 grid that divided the useful area of
a mirror surface into nine equal sectors, the probability that a
Cerenkov signal from a beam particle traversing the Cerenkov

counter was measured by the ratio

n ' E'° S6 / n- - S6,

where n' was a logic signal for a 9 GeV/c n' entering the streamer
chamber. The counter S6 was a 2.54 cm x 2.54 cm square scintillation
counter of 6.35 mm thickness placed downstream from the Cerenkov
counter. The signal ( n'- S6), therefore, defined a n' particle
transverisng the Cerenkov counter. 'This measurement was taken at

a beam rate of less than one n" per l.5 us beam spill. Even so,
account had to be taken of the long decay time of the RCA 4522
.Cerenkov phototube pulses in order to obtain the efficiency measure-
ment to the high precision desired. Hence a veto circuit was used

so as to ignore beam n" occuring within 300 ns of another beam
particle. The efficiency ranged from 98.8% to 100.0% for the eighty
points measured. These results were combined for each mirror and
are tabulated in Table 19. Averaged over the ten cells in each of the
two Cerenkov counters, the overall efficiencies were 99.848 i .006%
and 99.7l7 i .008% for the K' and K+ Cerenkov counter respectively.

Since no wide variations in efficiencies were observed among the

 

119

Table 19: Pion Rejection Efficiency (9 GeV/c n')

K" Cerenkov Counter

 

Cell No.

Upper l 99.78
2 99.82
3 99.89
4 99.84
5 99.84

Lower l 99.85
2 99.80
3 99.88
4 99.88
5 99.88

Average 99.848

i

:1:

H-

H-

H-

H:

1+

H-

H-

H-

H-

.02%
.02
.02
.02
.02
.02
.02
.Ol
.02
.02

.006%

K? Cerenkov Counter

99.57
99.62
99.75
99.42
99.83
99.74
99.78
99.87
99.89
99.78
99.717

H-

H-

H-

H-

H-

H-

H-

H-

|+

H-

H-

.03%
.03
.03
.04
.02
.02
.02
.02
.Ol
.02
.008%

120

cells, one was chosen at random for the study of the photoelectron
statistics. Figure 37 shows a sample pulse height spectrum from one
of the cells in the K' Cerenkov counter for a 9 GeV/c n' beam. The
channel number has been corrected for the internal pedestal of the
pulse height analyzer.

The spectrum exhibits a sharp rise from zero to a maximum
near channel number 375 corresponding to 100 DC into 50 ohms. From
there, the spectrum declines sharply and is followed by what resem-
bles a shoulder in the region of channel number 750. A l00 nanosecond
gate was used in the pulse height analyzer. The fraction of counts
in, and the position of this shoulder are consistant with its being
caused by a second pulse occuring within the gate time of the previous
pulse.

For the purpose of this analysis, cuts are made on the data.
Only events between channels llO and 630 are used. Fitted to a
Poisson distribution, a mean value of u = l7.2 i 0.1 photoelectrons
is obtained. The fitted distribution is shown as the solid curve
in Figure 37. Since the statistical fluctuations due to the dynode
multiplicaiton add to that due to the photoelectric effect at the
cathode, this mean value obtained using a simple Poisson distribution
should be viewed as a lower limit to the mean photoelectron number.

The analytic form of an experimentlaly observed pulse height
spectrum and its implication on photoelectron statistics have been
studied by a number of authors41. It is generally believed that a
simple Poisson distribution cannot adequately account for the observed
spectra. In particular, the single-photo-electron spectrum does

not always exhibit a clear peak as is expected from Poisson statistics

121

 

1000 1 I 1 I 1 l l I I

500“ —

Number of Events / IO Channels

b -

1 1 12 J l 1 1 1 '!q

o 500 I000
Pulse Height Channel Number

 

 

Figure 37: Pulse Height Spectrum of a Cerenkov Counter Cell for
a 9 GeV/c n" Beam. The Solid Curve is a Poisson Fit

122
for a mean value of one. In a number of cases, the spectrum resem-
bles that of an exponetial function. The actual explanation for
such behavior may be complex. However, it is not difficult to com-
prehend some of the dynamic factors involved in the collection and
multiplication of the few photoelectrons produced by the photocathode.
For example, the collection efficiency of the primary photoelectrons
by the first dynode depends on the local electric field shape, thus
the probability of secondary electron emission may vary across the
surface of a dynode. Similar dependence may also be expected in
electron multiplication in later dynodes. To account for these
non-Poissonian effects, it was suggested42 that instead of a con-
ventional fit to the Poisson distribution, one may adopt the use of

the Polya distribution

for electon multiplication at each dynode. Here P is the probability
of producing m secondaries given that m’is the mean number of second-
ary electrons emitted due to a single electron hitting the dynode.
The parameter b is to be determined using experimental data. The
Polya distribution approaches Poissonian as b tends towards zero
and becomes a monotonically decreasing function of n as b approaches
unity.

The experimentally observed pulse height spectrum from a photo-

multiplier tube of k-stages may be compared with a k-fold convolution

123

of equation (l). The RCA 4522 phototubes used have l4 stages with

an expected overall current amplification factor of 3 x 107 at the
volatage used, corresponding to an average per-stage gain of 3.42.
With the exception of the two tail regions, the pulse height spectrum
shown in Figure 37 can be well approximated by the Poisson distribu-
tions as is evidenced by the fitted curve. Therefore, the shape

of the spectrum would appear to be well represented by a single para-
meter, o/N , where the quantities N' and o are the mean and the
standard deviation of the spectrum peak. In the case of a pure
Poissonian distribution, the ratio o/N' can be represented by

(10';2 where u is fitted Poisson mean. The fact that the average
per-stage gain was only 3.42 suggests that the fluctuations to be
observed at the fourteenth dynode will be the end product of the
photocathode statistics broadened at the earlier dynode stages by

the stochastic nature of secondary electron emission. The ratio
o/N'is expected to approach a constant value at later stages of the
dynode chain.

Using a Monte Carlo technique, a Poisson distribution of photo-
electons was generated at the photocathode, and broadened by a Polya
distribution of secondaries at each dynode. For a, the mean number
of photoelectrons at the photocathode, between 20 and 65 the ratio
o/N is within 0.5% of its asymptotic value by the fifth dynode.
Figure 38 shows the expected ratio o/N' as a function of the para-
meter b at the fifth dynode for 5' between 25 and 65. As may be
expected, a larger value of b effectively broadens the pulse height
spectrum in a manner similar to that caused by lower photoelectron

statistics.

 

124
o 35 _ A 5:25
_. o fi=3o
_ A D a: 35
__ A 5:40
A -
- 0 n= 45
a 6:65
0.30 ...
O
'- A
_. <3 u
A
o a
b A
_ D
0.25
A o A o
A.
A
bIIZ " o o
L.
(J .A I.
0.20 Ill ' '
f“ I
0‘ I
I
045 l

o—NI
I

L-

)—.

..

._

...

...

h-

._

b-

Figure 38: The Expected Ratio of Standard Deviation to
Mean of the Pulse Height Spectrum as a Function
of the Parameter b for Various Mean Numbers
of Photoelectrons

125

The same pulse height cuts were applied to the model that were
used on the data. The ratio o/N' for the observed spectrum was found
to be 0.245. Since the value of b varies between 0 and I, this in-
dicates that the average number of photoelectrons from the photo-
cathode was actually below 45. For acceptable values of b, the Monte
Carlo results suggest an 5' value less than 35. In order to determine
the value of b independent of 3} it would be necessary to deal with
data of known 5' value, for example, using single photoelectron data,
which were not available. For the current analysis, it assumed that
b = 0, or that the Poisson statistics dominates the electron multi-
plication by the dynodes as well as the photoelectron production.
Figure 39 shows the expected ratio of o/N' as a function of 51 The
observed value of 0.245 corresponds to an 5' of 20 photoelectrons.
For Cerenkov light emission at angle a over path length t given by

n' = A2 sinze,

the data suggest a value of 44 for the parameter A. The result is
comparable to that obtained by other larger aperture Cerenkov counters?3
For comparison, integrating the manufacturer's stated relative
spectral response over the Cerenkov spectrum and quantum efficiency
gives an A of 266. Several known factors may be inferred to have
contributed to the rather low collection efficiency. These are:
(a) the residual magnetic field at the phototube may have contributed
to some loss of electrons. They cylindrical mu-metal shields avail-
able were not specifically designed for the set-up, ending about

even with the photocathode, and leaving the task of reducing the

stray field to the multi-layer laminated iron shields and the bucking

126

o.3- ‘ '

t>|lz '

0.2- ’

 

0.! __

.l
i . ... .1

l
0 IO 20 3 4O 5 60 70

fl

Figure 39: The Expected Ratio of Standard Deviation to Mean
of the Pulse Height Spectrum as a Function of
the Mean Number of Photoelectrons for Poisson
Statistics in Electron Multiplication (b=O)

 

 

127

coil. (b) The optical quality of the mirrors was not the best
possible, due to exposure and a semi-transparent layer of the evap-
orated aluminum. However, the counter performance was adequate to
meet the original goals, with a rejection efficiency against fast

pions of m 99.8%

APPENDIX III OPTICAL CORRECTIONS FOR IMAGE PLANE DIGITIZER MEASURE-
MEASUREMENTS

A. Overview of Method

Most of the film measurements were done on Image Plane Digitizers
(hereafter refered to as IPDs). These machines project the image of
the film onto an x-y digitizer. Some distortion in the image is
expected due to spherical aberration of the projection lens. Also,
the film plane may not be parallel to the table surface. This appen-
dix describes the proceedure used to correct the digitized points
for these distortions.

It was assumed that the (presumably small) optical distortions
can be parameterized by a third order polynomial expansion in the
image coordinates. A set of points on film were measured once
using a distortion free Film Plane Digitizer (hereafter FPD), and
again using each of the IPD projection lenses. A least squares
fit was then done comparing the two sets of measured points to

determine the polynomial expansion coefficiencts. This mapping

xfilm ' 3’ )’

1c(Ximage’ yimage)’ yfilm = g(xim699: 1"“999

was then put in subroutine ENGINE in TVGP to correct all IPD measured
points.

Actually, for comparison purposes three spearate sets of
correction Constants were generated by using three different sets of
film points. One set was generated using measurements of the chamber
fiducials on the film. Another set used measurements of a grid.

The third set used an expanded grid with more points.

Testing started with plotting the measurements and mapping

128

129
functions. The three sets of constants were compared, and the final
set selected. The TVGP reconstruction of the chamber fiducials, which
could be compared with survey values, was a major test of the re-

sulting resolution.

8. Scales, Fiducials, and Grid

The Tufts IPD digitizers project an image of the film onto the
digitizing table with a magnification of approximately 14.6. The
image is digitized with a smallest unit of measurement (called a
"least count") of 0.001 inch. The Tufts FPD digitizer, which digit-
izes directly on the film has a least count of 0.0001 inch. The
SLAC FPD, on which some beam tracks and test events were digitized,
has a least count of l um. One least count of the Tufts IPD on
our film corresponds to approximately 70 pm to 80 pm in the streamer
chamber.

The ficucials are electroluminescent sheets which were flashed
once per event. They appear as x's on the film, and provided the
reference frame for measuring tracks. The fiducial pattern and
numbering, with coordinate systems, are shown in Figure 40. In a
normal event, four fiducials were measured. In measured order,they
are fiducials 4, 2, l, 27. The surveyed positions of the fiducials
are given in Table 20.

The original grid was a piece of film with horizontal and ver-
tical lines drawn on it with 5 mm spacing. The measured points were
the points where vertical and horizontal lines crossed. There were
52 measured points, in a 4 by 13 array. Later, extra horizontal
lines were inserted, so that the spacing in the IPD x direction was

only m 2.5 mm. An array of 10 by 13 points was measured on this so

130

x
mmpmcwucoou
and muwah

mcwcmnssz meozcwm
use msmumxm mpocmucoou _m_ u now mesmwd

mm 0 am
e a 2 S «x A u
x x x x Nx
me om XwFXmFX
¢F NP m Mk 0 Nb . PH
X x x + V x + + flammm
FNX 2.x wimp”
mx. :x mu » mx 8w i
Rx mm
m
X X
m 3
x x N z
mmumcwvcooo o—P mcquwsm u<4m

oaH mumsh Emume mumcwngoou aw>H cw :owumucmwgo

Table 20:

Fiducial

LDmNONU'I-bOON—I

Streamer

131

Chamber Fiducial Survey

Position (October 31, 1977)

X Y
-29.87 cm 6.07 cm
30.18 6.17
-30.19 6.10
29.84 6.16
-29.96 6.14

0.11 6.18
30.15 6.20
-30.03 6.12
- 0.06 6.13
29.99 6.17
-30.12 6.19
— 0.14 6.20
31.92 6.17
- 0.21 6.17
-14.81 6.12
15.19 6.16
-14.89 6.12
15.19 6.16
-14.88 6.15
15.15 6.19
-14.88 6.17
15.12 6.19
-28.43 78.80
32.08 77 82
-28 46 78 22
31 95 78 06
-30 14 78 07

132

called I'big grid".

C. Optical Constants Fitting Program
The program OPTIPD was written to produce the optical constants

necessary to correct the IPD distortion. The correction is parameter-

ized as
2 2
xcorr = a1 + a2x + a3y + a4x + asxy + a6y
+ a7x2y + 68xy2 + 69y3
ycorr = bi + bzx + b3y + b4x2 T b5xy + bsyz

+

b7X2y + b8xy2 + b9y3

where (x,y) is an IPD measured point, and (x y ) is the

corr’ corr

corrected point. The eighteen constants a1, bi are produced by

a least squares fit using

 

 

no. of
points . _ . 2 . _ . 2
X2 = Z xcorr(1) XFPD(1) + ycorr(1) yFPD(1)
1“ o (1') a (i)
xFPD yFPD

and taking x,y,x2, xy, etc. as the "independent” variables. (Each
measured point used in the above fit was actually the average of
at least 10 measurements, in order to lessen the effect of measurer
jitter on the result).

In order to see the distortion pattern, and how well it is
corrected, a program was written to rotate, translate and magnify
an IPD measurement of the grid onto the FPD measurement by a least

squares fit. (IPD and FPO can't be directly compared since they

133

have different zeroing and magnification, and the coordinate axes
’may be slightly different). Figure 41 shows the distortion of

the 52 IPD measured grid points with respect to the FPD measured grid
points for the view 1 projection lens of IPD machine number 4.

The arrows showing the differences between the IPD and FPO measure-
ments of the grid points have been mangified 50 times with respect
to the position scale to make them easily visible. Figure 42

shows the same comparison after the IPD measurements have been
corrected using the above distortion parameterization. Table 21
lists the root mean square distance between corresponding points

of theIPD and FPO grid measurements for the uncorrected IPD grid
points, and for the grid points using optical correction constants
derived from the "big grid" measurements, and from fiducial measure-
ments. 0n the average, both sets of optical constants do about
equally well, and both give an improvement over the uncorrected
measurements. The fiducial produced optical constants may have an
advantage. TVGP will first apply the IPD optical corrections to
measured points. Rays from this point are then traced back into the
chamber using camera constants. The camera constants were produced
by a x2 minimization in program WEASEL comparing FPD measured fid-
ucials to the fiducial survey. If we use the same FPD measured fid-
ucials for both the IPD least squares and the WEASEL minimization,
that intermediate mapping between them will match exactly, and

one systematic error is removed. .The fiducial produced constants

were therefore used for the entire data set.

134

35000 ' /
/ ’ 1
\ a f
30000" 1 \ f
I
I ’ ' ‘f
u f
25000 - ‘ ’
’ v- \ f
y ' \ I
20000“ ‘ ‘ 4 I
R \. O .‘

 

Film Plane Digitizer Y Coordinate (Least Counts)

 

 

15000-
\\,\
\ 4
t
10000- \ \ \\‘
‘\
0 7'00
5000' Arrow Length (FPD
Least Counts)
o ,. , fl
0 5000 10000 15000

Film Plane Digitizer X Coordinate (Least Counts)

Figure 41: Comparison of Measurement of Grid on Image Plane
Digitizer with Measurement of Grid on Film Plane
Digitizer. Lengths of Arrows Give the Differences
Between the Measurements, Illustrating the Image
Plane Digitizer Distortions

Film Plane Digitizer Y Coordinate (Least Counts)

Figure 42:

135

 

 

 

35000 '1
V. l ‘ O
\ \ ‘ '
130000 "‘ § \ 1 .
Q Q ‘ \
’ p o 6
25000 7
c v Q P
' ‘ Q ‘
20000 7 a . . v
s a ‘ '
v ‘ s ‘
15000 '
0 V ' V
~ ‘ ‘ ‘
10000 " , 0 ‘ v
d 100
5000 5 Arrow Length (FPD
Least Counts)
0 1 1 n
0 5000 10000 15000

Film Plane Digitizer X Coordinate (Least Counts)

Comparison of Measurement of Grid on Image Plane
Digitizer with Measurement of Grid on Film Plane
Digitizer, After Optical Corrections Were Applied
to the Image Plane Digitizer Measurement

136

 

o.m n.o
N.n m.m
m.o 0.x
N.¢ m.o
¢.¢ o.n
N.m P.m
m.o N.o
~.m m.m
m.e w.n
o.m m.o
> x

.umeou _au.pao
Fawusewa s5
555502.50 oaH

 

m.o m.m
m.o m.m
«.0 N.m
5.0 m.o
o.m N.o
o.m m.¢
o.m m.w
—.n m.o
¢.m o.o
o.o_ m.o
> x

.pmeoo _mo.5ao
ewcw m.m >5
505002250 oaH

 

N.¢F v.5
N.FF N.w
¢.P~ m.o
~.NP N.m
m.—p o.o
¢.NP o.m
m.NF v.m
o.PN m.m
F.m~ ¢.m
m.o~ w.m
> x

cwuumgcoucs oaH

Ampczou umwmg onuv mucmoa vwgw mm mg» we cowumw>mo mzz

mammcw><

 

m
N

F Amkm_v N e
m
N

_ Aw~m_v F . e
m
N

P F m

mcwzumz

Longsz mcwgzmmmz

zmw> pcmsmgsmmwz 95H

vmcw vmgammmz can sock urea 5023mmmz oaw mo cowum_>mo menacm cam: uoom "Pm mpnmh

137

D. Fiducial Comparison

28 chamber fiducials and 3 target fiducials are visible in the
chamber. The 28 chamber fiducials were measured on the Tufts
IPDs as 14 two point tracks. The fiducials have also been measured
on a SLAC FPD. They have also been surveyed in real space. The
fiducials have been processed through TVGP with and without the
IPD optical corrections, and the results compared with the survey
values. Due to some fiducials not being visible in all views, and
SLAC having measured a slightly different set of fiducials, three
tracks are dropped, leaving 22 fiducials in the comparison.

Table 22 gives the average standard deviation of fiducial point
measurements. This indicates the limit on chamber resolution due
to how well a measureer can set the digitizer on a point. Table 23
shows the RMS deviations of the fiducial positions reconstructed by
TVGP from the chamber survey positions. For this test, the IPD
optical corrections produced by using the grid were used. The camera
optical constants used here were produced using the SLAC FPD fiducial
measurements. The SLAC deviations in the table are therefore
artificially low, since the same values used in the fit are also
being used in the test. The IPD optical constants produce an
obvious improvement. The results in the table are also consistant
with several known facts. The chamber is more than three times as
long as it is wide. Distortions would be expected to be more apparent
over the greater length, and indeed the deviations for Y are worse
than those for X for the uncorrected IPD measurements. For the

corrected set, deviations in X and Y are of the same order, as

Table 22:

Measuring
Machine

SLAC FPD

Tufts IPD #3

Tufts IPD #4 Set 1
Tufts IPD #4 Set 2

Table 23:

Measuring
Machine

SLAC FPD

Tufts IPD #3

Tufts IPD #4 Set 1
Tufts IPD #4 Set 2

138

Resolution of TVGP Reconstructed Fiducials

Standard deviation of 10 measurements of
each point (averaged for the 22 fiducials,
in the streamer chamber coordinate system)

0x

0.17 mm
1.24 mm
0.44 mm
0.30 mm

0y

0.33 mm
0.30 mm
0.56 mm
0.25 mm

02

1.48 mm
1.29 mm
2.76 mm
1.01 mm

Comparison of Fiducial Reconstruction with Survey

Root Mean Square Deviation of TVGP recon-

structed fiducials from survey values
(millimeters)

Without IPD optical
corrections in TVGP

O
X

0.42
1.13
0.71
0.72

0y

.0.30

2.15
1.75
1.64

0'
Z

0.83
4.33
1.71
1.42

With IPD optical
corrections in TVGP

0x

0.75
0.45
0.90

0y

0.62
0.57
0.52

G2

2.51
2.26
1.38

139

expected since both directions are perpendicular to the camera
axis. The Z coordinate must be determined by small angle stereo

from the cameras, and its deviation is correspondingly worse.

LIST OF REFERENCES

10.
11.
12.
13.
14.
15.
16.
17.

LIST OF REFERENCES

S. Okubo, Phys. Lett. 5, 165 (1963)
G. Zweig, CERN Report Nos. TH-401 and TH-412, 1964 (unpublished)

LC;

. Iizuka, Prog. Theor. Phys. Suppl. 37-38, 21 (1966),
. Iizuka, K. Okada, and 0. Shito, Prog. Theor. Phys. 55, 1061
(1966)

.g. M. Gell-Mann and Y. Ne'eman, The Eightfold Way (Benjamin,
New York, 1964)

m

0.0. Sakurai, Phys. Rev. Lett. 5, 472 (1962)

Particle Data Group, Rev. Mod. Phys. 48 (1976), N0. 2, Part II

S.L. Glashow and R.H. Sokolow, Phys. Rev. Lett. 15, 329 (1965)

See, for example, H. Harari, in Proceedings of the 1975 Interna-
tional Symposium on Lepton and Photon Interactions at High
Energy, Stanford, California, edited by W.T. Kirk (SLAC,
Stanford University Stanford, 1976) p.317

D.S. Ayres et al., Phys. Rev Lett. §g_, 1463 (1974)
0. Cohen et al., Phys. Rev. Lett. 55, 269 (1977)

S. Okubo, Prog. Theor. Phys. Suppl. 55, l (1978)
Dennis Sivers, Phys. Rev. 511, 353 (1975)

R.A. Donald et al., Phys. Lett. 515, 210 (1976)
V. Blobel et al., Phys. Lett. 555, 88 (1975)
P.L. Woodworth et al., Phys. Lett. 555, 89 (1976)

C.W. Akerlof et al., Phys. Rev. Lett. 55, 861 (1977)

C. Daum et al., Phys. Lett. 555,313 (1981)

In the inclusive reaction A + B + C + anything, Feynman's X for
m

max'
. cm ' .
momentum of particle C, and pmax 15 the max1mum momentum

particle C is defined as XF = pcm / p pa? is the c.m.
possible consistant with quantum conservation laws. (Particle
Data Group, Rev. Mod. Phys. 52 (1980) No. 2. Originally pro-
posed by R.P. Feyman, Phys. Rev. Lett. 25, 1415 (1969) ). The
definition XF = ZpfiP /‘V§-, whereiFE is the total c.m. energy,
is also in use. The two definitions become the same at high

140

18.
19.
20.
21.
22.
23.
24.

25.

26.

27.

28.
29.
30.
31.
32.
33.
34.
35.
36.

37.

141

energy.

A. Etkin et al., Phys. Rev. Lett. 35, 784 (1978)

J.V. Allaby et al., CERN report 70-12 (1970)

SLAC Users Handbook, p. C 2-25 (1968)

URCL - 3421

F.T. Dao et al., SLAC Proposal E-131 (1976)

K. Bockmann et al., Nucl. Phys. 5_115, 253 (1977)

F. Bulos, A. 0dian, F. Villa and D Yount, SLAC Report No. 74
Clemehlgg?)Heusch et al., Nucl. Instrum. Methods 155, 165 (1975)

F.T Solmitz, A.D. Johnson, and 1.8. Day,Alvarez Group Programming
Note-P-ll7, Lawrence Radiation Laboratory

M. Alston Garnjost, Group A Programming Note P-239, Lawrence
Berkeley Laboratory

O.I. Dahl, T.B. Day, and F.T. Solmitz, Alvarez Group Programmer's
Note p-126, Lawrence Radiation Laboratory

Abolhassan Jawahery, Ph.D. dissertation, Tufts University, 1981
Foley, Phys. Rev. Lett. 15, 330 (1967)

B. Svensson, CERN 67-24

R. Kinnunen, Report HU-P-l78 (1980), University of Helsinki
Douglas R.0. Morrison, CERN/EP 79-102 (1979)

B. Ghidini et al., Phys. Lett. 555, 186 (1977)

I.J. Muzinich and Frank E. Paige, Phys. Rev 551, 1151 (1980)
R.D. Field and R Feynman, Phys. Rev. 515, 2590 (1977)

M.Bink1ey et al., Phys. Rev. Lett. 51, 578 (1976)
JJG. Branson et al., Phys. Rev. Lett. 55, 580 (1977)

J.H. Cobb et al., Phys. Lett 255, 497 (1978)
T.B.W. Kirk et al., Phys. Rev. Lett. 55, 619 (1979)

38.

39.

40.

41.

42.
43.

"'1

C:

can

142

.R. Hayes, R.A. Schluter and A. Tamosaitis, Index 555_
Dispersion 5f_Some Cerenkov Counter Gases, ANL-6916,
(1964)

. Winston, J. Opt. Soc. Am. 55, 245 (1970)

 

 

 

.L. Garwin, Y. Tomkiewicz and D. Trines, Nucl. Instrum. Methods
151, 365 (1973)

. R Prescott, Nucl. Instrum. Methods 55, 173 (1966), also see
references to previous works cited therein

.R. Prescott, ibid.

.0. Auber et al., flucl. Instrum. Methods 51, 79 (1970)

. Trines, The Gas Cerenkov Counter for E-89 (1.6 GeV Sggctro-
meter), SLAC Group A Note N0. 11, (March 1, 1973), un-
published

 

.B Bowden et al., A Large Pressurized Cerenkov Counter, SLAC
Group BC Internal Memo (1977), unpublished

 

. Hinterberger et al., Rev. Sci Instrum. 11, 413 (1970)

.H. Williams, A. Kilert and D.W.G.S. Leith, Nucl. Instrum.
Methods 155, 483 (1972)

 

“1111111 111711111111111111111110
3 1293 03083 0628