Fermi National Accelerator Laboratory

FERMILAEi-Pub-94/301-E

A Direct Measurement of the W Boson Width I? (W)

F.Abe et. al The CDF Collaboration

Fermi National Accelerator Laboratory P.O. Box 500, Batavia, Illinois 60510

September 1994

Submitted to Physical Review Letters September 6,1994

a Operated by Universities Research Association Inc. under Contract No. DE-AC02-76CH03000 with the United States Department of Energy

Disclaimer

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

CDF/ANAL’ELECTROWEAWCDFR/2642

FERMILAB-PUB-94/301-E Version 4.0

A Direct Measurement of the WBoson Width r(W)

F. Abe,13 M. G. Albrow, D. Amide&l6 J. Antos,*s C. Anway-Wiese,4 G. Apollinari,*s H. Ared,’ M. Ata~,~ P. Auchincloss,*5 F. Azfar,*r P. Azzi,*o N. Bacchetta,ls W. Badgett,lG

M. W. Bailey,18 J. B~o,~~ P. de Barbaro,*s A. Barbaro-Galtieri,14V. E. Barnes,*4 B. A. Barnett,l* P. Bartalini,23 G. Bauer,ls T. Baumann,s F. Bedeschi,*s S. Behrends,s

S. Belforte,*” G. Bellettini,*s J. Bellinger,3s D. Benjamin,32 J. Benlloch,ls J. Bensinger,s D. Benton,*’ A. Berehas,7 J. P. Berge,’ S. Bertolucci,* A. Bhatti,zs K. Biery,tr M. BinkIey,r F. Blrd,2g D. Bisello,*o R. E. Blair ,r C. Blocker,*9 A. Bodek,zs W. Bokharl,15 V. Bolognesi,*s D. Bortoletto,24 C. Boswell,l* T. Bou10s,~~ G. Brandenburg,g E. Buckley-Geer,7 H. S. Budd,*s K. Burke&l6 G. Busetto,20 A Byon-Wagner,7 K. L Byrt~m,~ J. Cammerata,12 C. Campagnari,7

M. Campbell,16 A. Caner,7 W. Carithers,t4 D. Carlsmith J. Chapman,16 M.-T.

,33 A Ca.~tro,~~ Y. Cen,21 F. Cervelli,*3 Cheng, 28 G. Chiarelli,* T. Chikamatsq31 S. Cihangir,’ A. G. Cla~-k,*~

M. Cobal, M. Contreraq5 J. Conway,27 J. Cooper,7 M. Cordelli,* D. Crane,1 J. D. Cunningham,3 T. Danielsls F. DeJongh? S. Delchamps,’ S. Dell’Agnello,23 M. Dell’Orso,*3

L Demorder,26 B. Denby S. Donati,23 R B.

,23 M. Deninno,* P. F. Denver&l6 T. Devlin,*’ M. Dickson,*5 Drucker, l4 A. Dunn,16 K. Einsweller,14 J. E Elhq7 R Ely,14

E Engels, Jr.,** S. I3t0,~ D. Brrede,lo S. Errede,to Q, Fan,25 B. Farhat,ls I. Fiori,* B. Flaugher,’ G. W. Foster ,7 M. Franklin,g M. Frautschi,l* J. Freeman,’ J. Frk~Iman,~~

H. Frisch,5 A. Fry,zg T. A. Fuess,l Y. Fukui,ls S. Funaki,sr G. Gagliardi,*s S. Galeotd,zs M. Gallinaro,20 A. F. Garfinkel,24 S. Geer,7 D. W. Gerdes,r6 P. Giannetd,23 N. Giokaris,*b

P. Giromini,a L Gladney,*r D. Glenzinski,12 M. Gold,‘* J. Gonzalez,” A. Gordon,9 A. T. Goshaw,c K. Goulianos,26 H. Grassmann,6A. Grewal ,*l G. Griec~,*~ L Groer,*’ C. Grosso-

Pilcher,s C. Haber,l* S. R. Hahn,7 R. Hamilton,9 R Handler,33 R M. Hans,34 K. Hat-a,31 B. Hart~I,~l R. M. Harris,’ S. A. H auger,6 J. Hauser,’ C. Hawk,27 J. Heinrkh,21 D. Cronin-

Hennessy,6 R Hollebeek,*l L Holloway,10 A. Holscher,lt S. Hong16 G. Houk,*l P. Hu,** B. T. H~ffman,~~ R H~ghes,~~ P. Hurqg J. Huston,17 J. H~th,~ J. Hyleq7 M. Incagli,23

J. Incandela,’ H. 1~0,s~ H. Jensen ,I C P. Jessop,g U. Joshi, R W. Kadel,” E Kajfasz,7a

R. D. T. Kamoq30 T. Kaneko,31 D. A. Kardelis,lo H. Kasha,s4 Y. Kato,ls L Keeble,so Kennedy, 27 R Kephart,7 P. Kesten ,14 D. Kestenbaum,g R M. Keup,l” H. Keutelkq7

F. Keyvan,’ D. H. I&q7 H. S. Kim,” S. B. Kim,16 S. H. Klm,31 Y. K. Kim,14 L Kirsch,s P. Koehn,2s K. Kondo,31 J. Konigsberg,s S. Kopp,s K. Kordas,ll W. Koska,7 E Kovacs,ra

W. Kowald,6 M. Krasberg,‘” J. Kroll ,’ M. Kruse,*‘S. E Kuhlmann,’ E Kuns,*7 A. T. Iaasanen,24 N. Labanca,23 S. Lamn~I,~ J. I. Lamoureu~,~ T. LeCompte,lo S. Leone,23 J. D. Lewi~,~ P. Limon? M. Lindgren,’ T. M. Liss,lO N. Lockyer,*l C. Loomis,*7 0. Long,*’ M. Loreti,*O E H. Low,*l J. Lu,~ D. Lucchesi,*s C. B. Luchini,lo P. Lukens,’

K. Maeshima,7 A P. Maas,

Maghakian, 26 P. Maksimovi~,~~ M. Mangan~,*~ J. Mansour,17 M. Mariotti,*s J. P. Marriner,’ A Martin,r” J. A J. Matthews,” R Matdngly,rs P. McIntyqso P. MeIese,26 A Menzione,23 E Meschi,zs C. Mkhail,g S. Mikamo,~3 M. MilIer,s R Miller,17 T. Mhnashi,sr

S. Miscetti,8 M. Mishina,lj H. Mitsushlo,31 S. Miyashita,st Y. Mot-&a,13 S. MouIding,zs J. Mueller,*’ A Mukherjee,’ T. Muller? P. Musgrave,ll L F. Nakae,2g I. Nakano,31 C Nelson,7

D. Neuberger,* C. Newman-H~lmes,~ L Noduhnar~,~ S. Oga~a,~l S. H. Oh,6K. E Oh1,34 R Oishi,31 T. Okusawa,19 C. Pagliarone,zs R Paoletti,*J V. Papadimitrioy7 S. Park,7

J. Patrick,’ G. Pauletta,23 M. PauIini,14 L Pescar~~~ M. D. Peterqr4 T. J. Phillips,6 G. Piacendno,* M. Pillai,25 R Plunkett,’ L Pondrom, 33 N. Prod~it,‘~ J. Proudfoot, F. Ptohos,g

G. Punzi,23 K. Ragan,ll F. Rimondi,2 L Bistorl,23 M. Roach-BeIlino,32 W. J. Robertson,6 T. Rodrigo,’ J. Romano,s L Rosenson,ls W. K. Sak~moto,*~ D. Salt~berg,~ A Sansoni,s

V. Scarpine,so A Schindler,l’ P. Schlabach ,g G. F. Sciacca,zs A !kribano,~3 S.

E E Schmidt,7 M. P. Schmidt,” 0. Schneider,14 Segler ,’ S. Seidei18 Y. Seiya,3t G. Sganos,lt A Sgolacchia,z

M. Shapiro,‘” N. M. Shaw,24 Q, Shen,*4 P. F. Shepard,** M. Shimojima,31 M. Shochet,5 J. Siegrist,2g A SiR7a P. Sinervo ,11 P. Singh,22 J. Skarha,l* K. !%~a,~~ D. A Smith,*3 F. D. Snider,12 L Song7 T. Son&l6 J. Spalding,7 L Spiegel,7 P. Sphkas,ls A Spies,12

Submitted to Physical Review Letters September 6, 1994

L Stanc~,~~ J. Steele,33 A. Stefanini,23 K. Strahl,” J. Strait,7 K. Sumorok,ls R. L Swam, Jr.,lO T

D. Stuart,’ C. Sullivan,s . Takahashi ,lg K. Takikawa,si F. Tartarelli,23 W. Taylor,”

Y. Teramoto,lg S. Tether,t5 D. Theriot,‘J. Thomas,2g T. L T~oIMs,~~ R. Thun,l6 M. Tlmko 32 P. Tipton, A. Titov,26 S. Tkaczyk,’ K. Tollefson,25 A. Tollestrup,7 J. Tonnison 24 ’

J. F. de Troconiqa J. Tseng,l2 M. Turcotte,2g N. Turini, N. Uemurq3r F. Ukegawa,zr G. Unal,21 S. van den Brink,22 S. Vejcik, IILl R. Vidal,’ M. Vondracek,lo R. G. Wagner,1

R. L Wagner,’ N. Wainer ,’ R. C. Walker,25 G. Wang,23 J. Wang,s M. J. Wang,28 Q F. Wang,26 A. Warburton, G. Watts,2s T. Watts, 27 R. Webb,30 C. Wendt,33 H. Wenzel,14

W. C. Wester III,l4 T. Westhusing,lo A, B. Wicklund,t E. Wicklund, R. Wilkinson,2i H. H. Williams,21 P. Wilson,5 B. L Winer,25 J. Wolinski,30 D. Y. Wu,16 X. WU,~~ J. WYSS,~~

A. Yagii,’ W. Yao,14 K. Yasuoka, 31 Y. Ye,lr G. P. Yeh,’ P. Yeh,28 M. Yin,‘j J. Yoh,’ T. Yoshida,lg D. Yovanovitch,71. Yu,~~ J. C Yun,’ A. Zanetti,23 F. Zetti,23 L Zhang,33 S. Zhang,16

W. Zhang,21 and S. Zucchelli2

(CDF Collaboration)

lArgonne National Laboratory, Argonne, Illimis 60439 2IstUu~ Naa-di Plska Nuleare, University of Bobgua, 140126 Bologna, ItAy

3Brandeis University, Walthm. Massachusetts 02254 4univ~ly of chmmia at Los Angeles, Los Angeles, California 90024

5university ofchlcyo, chkag* Iullois 60637 ‘Duke Unlverslty, Durham, North Guolina 27708

‘Fenni National AcceIeratm Labwatory, Batavia, IlUds 60510 8hbontod Nazhali di FrascaU, Istituto Nazicmale dI FIska Nudeare, I-OoWd Frascati, Italy

91ianwdunlverslty, Cambrklge, Massachusetts 02138 lo Unh’ersity of Ullnois, Urbana, Ill&h 61801

’ ’ hstit~te Qf lbtkle F%~~cs, McGffl University, MonUeaI H3A 2T6, and U-ty of Toronto, Toronto MSS 1 A’, Canada 12Tbe Johna Hopkins University, Baltlnmre, Maryland 21218

13Nathul Islb0ntm-y for High Energy Physics (KEK), Tsukuba. Ibarakl305, Japan %awre!ue Berkeley Iabuaw, Berkdey, Califcxnia 94720

%iamchusetts Institute of Technology, CiuhMge, -usetts 02139 16UnhmrsltyofMkh&an,AnnArbor,Mkh@n48109

“b-&an state University, hst Lamin& Idida@n 48824 %nhrsItyofNaHexko,AIbuqucrpuqNew-87131

“Osaka City University, Osaka 588, Japan 29mhrenleadiPadovl,~UtutoNsrzbllsrlediFlslcPNudePre,~~dtPdov;SI-3513lPadwa,I~

21university of Fennsyhfan& Pbhb@h@ Femsylvarda 1910)

22univarslty of Pkbur& Pktsburg& Penusyfvcmlp 15260 23Istltllt0N~IllledlFUolNudeP1P,UniversityludScudpNormalcS~~~ofPfs,E56100PLsa.Italy

2+IuThle Unlvdty, west Iahyeno Indiana 47907 2bdversityof Rochter, Rubeater, New York 14627 26B* Univdty, New York New York 10021 27RuQess University, Plscabway, New Jm 08854

‘hcadeda SJnka, Taiwan 11529, Rep&l&z of U&m 2~~~ndu~ Super cdllder Lab~ratoty. Dallas, Texas 75237

Texas A&M University, Coik+ge Station, T~JGU 77843 31UnhmrsityofTsukuhe,Tsukubn,Ibarakt305, Japan

j2Tuftl uldversily, Medford -uaetb 02 155 =I.hhmity Qfwiscm h4adisQa wkonsin 53706

3%leIJllhmity,NewHaveqcoMecucut06511 %dlKw

2 September 1994

2

Abstract

This letter describes a direct measurement of the W boson decay width, r(W),

using the high-mass tail of the transverse mass spectrum of W+ev decays recorded

by the CDF experiment. We find r(W) = 2.11 f 0.28(s&t)f O.l6(sys.) GeV, and

compare this direct measurement with indirect means of obtaining the width.

PACS numbers: 13.38.+c, 12.15.Ff, 11.80.Dq, 14..50.h-

The W boson width, T(W), is a fundamental parameter that is well-predicted in

the Standard Model. The W decays with approximately equal probability to each of *

three lepton and two quark families kinematically available (with a color factor of

three on the number of quark families). Hence, its branching ratio into (C,Q) is

approximately i. The quark decays are enhanced by as(MW)/lr to first order in QCDso

that the leptonic branching ratio isI 0.1084 f 0.0002, where the 0.0002 reflects the

uncertainty in the value of as at @=I$,$ . The predicted W width may then be

predicted by dividing the leptonic partial width r(W+!vJ - (G&)(&f& /6z) by the

branching ratio. Using MW - 80.14 f 0.27 GeV/cZ,E*] Rosner et al.111 find

Zi%!I = 2.067 f 0.021 GeV where the uncertainty in the prediction is dominated by

the uncertainty in M++c

The W width has been obtained experimentally[31 by an indirect method using

the ratio R I ~B(p~~W~~/~~~p~j~~~~~. A measurement of R, together with

a calculatior#] of the ratio of production cross sections o(p~+W)/c~{p~+zo) and the

LEP measurement&] of r(zo) and Q@-&+t), is used to obtain a measurement of the

W leptonic branching ratio r(W+&j/r(W). A calculation of r(W+&j is then used to

obtain the W width. With this method a precision of 85 MeV has been obtained. As

Rosner et aZ.[l] have noted, however, loops at the W-fermion vertex involving a @ or

3

scalar Higgs @,I61 could alter the effective W-fermion coupling, and hence T(W), but

not affect the ratio r(W+t’v)/r(W). Thus, a measurement of r(W) from

T(W+t’v)/T(W) assumes that the W coupling to leptons is given by the Standard

Model. While in principle such non-standard couplings would also alter the W

production cross section and thus affect the value of I’(W-+k’v)/r(W) extracted from

R, a direct measurement of the full width r(W) is desirable so that these radiative

corrections to r(W) can be observed.

Direct measurements of T(W) have been reported by the UA1[71 and I~~2181

Collaborations. Including systematic uncertainties, they obtain the values

r(w) = 2.8 _‘;1:: f 1.3 GeVand r(W) < 7.0 GeV (9096 CL) respectively. These direct

measurements result from fits of the W transverse mass distribution for the best

values of Mw and r(W). The transverse mass is defined as MT P ?/ 2Pj++( 1-cos(A#)),

where P’T and q are the transverse momenta191 of the electron and neutrino and A#

is the azimuthal angle between them. The fits were performed over a limited range

in MTnear the Mwpeak.

The tail of the transverse mass distribution of the W contains information on

T(W). Events with MT > MW can arise due to the non-zero W width or due to the

calorimeter resolution. The resolution, furthermore, degrades for large values of the

W transverse momentum, Py However, a precise measurement of T(W) from the

high-mass tail is possibleI be cause the q distribution is sufficiently well-known

and because far above MW the Breit-Wigner tail dominates over the gaussian

resolution of the detector. In this analysis the W width is determined from a binned

log-likelihood fit to the transverse mass distribution in the region MT > 110 GeV/c2,

where a Monte Carlo study indicates there is good sensitivity to the width and the

4

systematic uncertainties are small. Possible non-gaussian tails to the resolution are

discussed below.

This measurement is made with 19.7 f 0.7 pb-1 of data collected by the

Collider Detector at Fermilab (CDF) during the 1992-1993 run of the Fermilab Tevatron

Collider. The Tevatron produces pF collisions at 4 = 1.8 TeV. Detailed descriptions of

the detector can be found elsewhere.[lll Th e portions of the detector relevant to this

measurement are (i) electromagnetic and hadronic calorimeters covering the

pseudorapidity[l*] range 1~1 < 4.2 and arranged in a projective tower geometry; (ii) a

drift chamber (CTC) immersed in a 1.4 T solenoidal magnetic field for tracking

charged particles in the range 1~1 < 1.4; (iii) a time-projection chamber (VTX) for

vertex finding; and (iv) two arrays of scintillator hodoscopes located on either side of

the detector for triggering.

To select candidate events we require an electron in the central barrel region

of the detector (I@ < 1.05) with calorimeter transverse energy ET > 30 ~&lo] and

CTC transverse momentum PT > 13 GeV/c. We require the electron track to be

isolated in the CTC, requiring Iso < 5 GeV/c, where Iso is defmed as the

scalar sum of the PT of all tracks except the electron track within a region in q--#

space of -\, (A#)* + (A@* = 0.25 centered on the electron. The ratio of energy in the

hadron (Had) and electromagnetic (EM) calorimeter towers of the electron cluster is

required to satisfy Had/EM < 0.055 + 0.045 x A transverse momentum

imbalance is required to signal the presence of the neutrino. We require

ET s 30 GeV, where the missing transverse momentum, or ET, is defined as the

vector sum of the ET in all calorimeter towers with lql < 3.6. Finally, the pjF

interaction point, which is distributed by an approximate gaussian of width

az = 26 cm along the beam direction, is required to satisfy lZind c 60 cm, and the

5

total accidental calorimeter energy not in time with the pj7 collision is required to be

< 100 GeV. There are 10845 events passing these cuts. Of these, 93 events with second

isolated, high-@ tracks pointing to electromagnetic clusters are removed from the W

sample as @ candidates, and 226 events with clusters of tracks in the CTC which point

at the ET vector in # and at calorimeter cracks in Q are removed as mismeasured QCD

jet events (“dijets”), leaving 10526 events.

Several processes can also mimic the W signal. The process W+z~+evvv has a

signature similar to W+ev decays, but at lower MT The process @+e+e-, where one

electron is detected and the other is lost because it falls into an uninstrumented

region of the detector, can produce the signal of an electron and ET, as can QCD dijet

events where one jet is lost and the other passes our electron selection criteria, We

use the ENJET Monte Carlo program[l3] and a detector simulation to estimate the

number of @+e+e’ and W+w decays contaminating the W sample. We find that the

number of Zo events remaining in the W sample is 50 f 15 events and the

background from W-av is 150 f 45 events. The WD dijet background is estimated by

studying a data sample of events with an “electron” + ET in which the “electron” has

Iso(trk) > 6 GeV/c. These events are presumably mismeasured dijet events. We

study the efficiency of our dijet removal cuts on this sample and normalize to the

number (226) of events in the W sample removed using these cuts. We estimate that

the number of dijets left in the sample is 241 f 40 events. We observe 124 events

with MT > 110 GeV/c*. We expect that 3.5 f 0.5 events are due to Zo and W-av

events and that 53 f 8 events are from QCD dijets .

The W transverse mass spectrum is modeled using a Monte Carlo program that

generates zeroth order diagrams of W production, qq+W, according to an energy-

dependent Breit-Wigner distribution:[141

6

a(i) - 6 Hi - M&J2 + iQ’~M$ ]

where 6 s IS the (possibly off-shell) eV mass. This form of the cross section includes

vacuum polarization contributions to the W propagator. We use the MRSIT’ structure

functions.[lfi] The effects of higher order diagrams for W production are mimicked

by giving the bosons PT according to a previous measurementfl61 of the Py

spectrum. The lepton momenta are passed through a simulation of the detector

response. The same kinematic and geometric cuts as in the data are applied

A cut of Py < 20 GeV/c is imposed in order to suppress many of the expected

backgrounds in the fit region (MT > 110 GeV/d) and in order to suppress the

broader calorimeter resolution which arises at large Pr With this cut, the number

of W+ev candidates is reduced from 10526 to 9701 and the total background is reduced

from 441 f 62 events to 224 f 44 events. The background in the fit region is

reduced to w 1096 of the 58 events observed. Furthermore, Monte Carlo studies

indicate that the uncertainty in r(W) due to the #shape is reduced from 700 MeV to

120 MeV. Figure 1 shows the transverse mass distribution of the 9701 candidate

events after the flT < 20 GeV/c cut, along with the expected background.

To obtain F(W) a binned log-likelihood fit to the W MT distribution, binned in

1 Ge V/c2 bins, is performed for MT > 110 GeV/cZ. Monte Carlo templates are

generated with values of r(W) between 0.667 and 3.667 GeV at 200 MeVintervals. In

the templates the MT shapes are the sum of W Monte Carlo and of backgrounds, where

the background is normalized to 224 events and the W Monte Carlo is normalized to

9701- 224 = 9477 events. The data are fit to each template and a likelihood curve vs.

r(W) is made. Figure 2 shows the likelihood curve for the data. The most likely value

7

is r(W) = 2.04 f 0.28 GeK The uncertainty is purely statistical and is determined by

the point where the likelihood decreases by 0.5.

As a check of the fitting technique, we generate an ensemble of Monte Carlo

“experiments” of 9477 W events each, add 224 background events, and fit to the Monte

Carlo templates above. The mean returned value of QJm> = 210 f 0.02 GeVis

obtained from the ensemble of “experiments,” to be compared to the input value of

2.07 GeK Furthermore, the r.m.s. = 250 f 17 MeV of the fitted values is consistent

with the 280 MeVstatistical uncertainty estimated from the data.

The systematic uncertainties in this determination of the W width are due to

effects that alter the shape of the transverse mass distribution. They are: the Py

distribution, the electron resolution, the neutrino resolution, the backgrounds, and

the electron energy scale. To estimate the uncertainties due to these effects, we allow

these parameters to vary in our Monte Carlo program and then refit the Monte Carlo

transverse mass spectrum with the varied input parameters to the original Monte

Carlo templates.

The Monte Carlo model of the ET resolution is compared to the ET observed in

@+e+e- decays, where any ET in the event is due to the energy response of the

calorimeter and not due to neutrinos. Varying the gaussian width of the ET

resolution by the amount allowed by the Zo data leads to a 4.9% variation in T(W).

Possible non-gaussian tails in the calorimeter ET resolution, which have been

checked using minimum bias data, lead to negligible variations in r(W). The

calorimeter electron energy scale and resolution are determined using the mass peak

position and width of i?+e+e’ decays. Allowing the scale and resolution to vary in

the Monte Carlo leads to 2.0% and 0.6% variations in T(W), respectively. Allowing the

background to vary within its estimated uncertainty causes 0.8% variations in r(W).

Distorting the input $ distribution in the Monte Carlo within its uncertainty

and fitting the Monte Carlo with the distorted q to the nominal templates, we find a

variation of 696, or 120 MeV, in r(W). A different kind of Py uncertainty arises

because of the theoretical expectationfl71 that the Py distribution varies

logarithmically, - aw; ‘+ log( 5 ), with the mass M of the tV pair. We have checked T T

this possible source of bias in two ways: first, we fit a Monte Carlo sample with this

new PW T distribution to the Monte Carlo templates generated with the nominal

distribution and observe a shift of 23 MeVin r(W). Second, we have performed the

entire analysis with a cut of q < 10 GeV/c, observing a shift of 0 MeV. The

contribution of these shifts, when combined in quadrature with the 120 MeV

uncertainty from the input W-spectrum, is negligible.

Finally, we apply a shift to r(W) due to the effect of radiative decays of the W,

Wevx not accounted for in our Monte Carlo. While most of the radiated 7 ‘s are

collinear with the electron and, hence, are clustered in with the electron energy,

some of the photons are radiated at wide angles. These wide angle photons carry

away some of the W mass, and, hence, shift the MT distribution downward. We have

used a Monte Carlo programfl81 to estimate that we must shift our value for r(W) up

by 70 f 28 MeV to correct for radiative decays.

Assembling the results, the final result for the Wwidth is:

r(W) 3: 2.11 f0.28(stat.) f O.lG(sys.) GeK

As a check, we instead fit over the region MT > 120 GeV/cZ, and obtain a value of

r(W) = 2.15 f 0.34(srat.) f O.O9(sys.) GeV. Here the systematic uncertainty is

smaller because the cut-off for the fit is farther from the falling edge of the

resolution. Note that the MT > 110 GeV/cz sample has 58 events and the

MT > 120 GeV/d sample has 35 events.

In conclusion, we have reported on a direct measurement of the Wboson decay

width, r(W) = 2.11 f 0.28( staf.) f O.lG(sys.) GeV, using the tail of the transverse

mass distribution of W+ev decays in 19.7 ~6-1 of m collisions at & = 1800 Get!

With the combined data set of 200 pb-1 from both the CDF and Dj$ detectors

anticipated in the next year, a 100 MeV measurement may be anticipated. This may

be compared with the expected uncertainty1191 from the LEP-200 experiments of

200 MeV, which will be obtained after five years of data-taking. With future runs of

the Fermilab collider, a 30 MeV measurement is possible, which approaches the level

of the radiative corrections to the width.

We thank the Fermilab staff and the technical staffs of the participating

institutions for their vital contributions. We also thank James Cronin and Jonathan

Rosner for helpful discussions. This work was supported by the U.S. Department of

Energy and National Science Foundation; the Italian Istituto Nazionale di Fisica

Nucleare; the Ministry of Education, Science and Culture of Japan; the Natural

Sciences and Engineering Research Council of Canada; the National Science Council

of the Republic of China; the A. P. Sloan Foundation; and the Alexander von

Humboldt-Stiftung.

10

r11

PI

01

r41

PI

PI

[‘I

PI @I

J. Rosner, M. Worah, and T. Takeuchi, Phys. Rev. D 49,1363 ( 1994).

F. Abe et. al., Phys. Rev. D 43, 2070 (1991); J. Alitti et al., Phys. Lett. B 276,354

(1992).

F. Abe et. al., Phys. Rev. Lett. 73, 220 (1994); Phys. Rev. Lett. 64, 152 (1990);

Phys. Rev. Lett. 69, 28 (1992); C. Albajar ef al., Phys. L&t. 253B, 503 (1991); J.

Alitti et al., Phys. Iett. 276B, 365 (1991).

A.D. Martin, RG. Roberts, and W.J. Stirling, Phys. Lett. B 306, 145 (1992).

LEP Electroweak Working Group, CERN Preprint CERN/PPE/93-157, 26 August

1993.

A. Denner and T. Sack, Zeit. Phys. C46, 653 ( 1990); A. Denner, Forts&r. Phys. 41,

307 (1993).

C. Albajar et al., Z. Phys. C 44, 15 (1989).

R Ansari et al., Phys. Lett. B 186,440 (1987).

The transverse momentum is defined as PT- Px sine, where 8 is the angle with

respect to the proton beam direction. Similarly, ET = Ex sine, where E is

measured in the calorimeter.

[lOI S. Kopp, Ph.D. Thesis, The University of Chicago (1994), unpublished.

U1l F. Abe et al., Nucl. Instr. and Meth. A 271, 387 (1988).

fl21 At CDF cylindrical coordinates r, 9, and z, are used, where Q is the azimuthal

angle and z points in the proton beam direction and is zero at the center of the

detector. The pseudorapidity, q, is defined as q = ln(cot(tV2)~.

[131 F. Paige and S. D. Protopopescu, ISAJET Monte Carlo program, BNL Report No.

BNL38034,1986 (unpubJished).

11

1141 F. A. Rerends, “Z tine Shape,” in “Z Physics at LEP I,” CERN-89-08.

1151 A.D. Martin, R.G. Roberts, and W.J. Stirling, Phys. Lett. 306 B, 145 (1993);

Erratum Phys. Lett. 309 B, 492 (1993).

1161 F. Abe et. al, Phys. Rev. Lett. 66,295l (1991)

[l’l P. Arnold and R Kauffman, Nucl. Phys. B349, 381 ( 1991).

1181 R. G. Wagner (unpublished), based on calculations by F. Berends et al., Z. Phys.

C27,155 (1985) and F. Berends and R Kleiss, Z. Phys. C27,365 (1985).

[191 P. Roudeau et. al., in the “EPCA Workshop on LFP200,” CERN-87-08, edited by A.

B&m and W. Hoogland, p. 62.

12

Figure Captions

Figure 1: Transverse mass distribution of the 9701 W candidates surviving a cut of

Py < 20 GeV/c, along with the expectation for the background and the expected W

shape from the Monte Carlo program. In the Monte Carlo, r(W) = 2.067 GeVwas

used.

Figure 2: Results of the log likelihood fit of the data to Monte Carlo templates of

different W widths. Each point represents a log-likelihood fit performed over the

range MT > 110 GeV/cZ. The curve is the best fit of the likelihood points to a cubic

polynomial. The most likely value is at r(W) = 2.04 GeK

13

10:

“0 \ > Q) lo2

0 *

\" m c 10

F W

1

1

L

-

m

m

I

I

I

m

I

.

.

.

r

ta (19.7 pb-‘)

80

+ev Monte Carlo

Bockgrounds

160 280

Transverse Mass (GeV/c2)

-90

-92

-94

0 L -96

I cT, -98

0

-100

-102

-104 111, ,111 ,, l,,,,l,,,,(,,,*~*,,,~, 0,s 1.2 1.6 2 2.4 2.8 3.2 3.6 4

we (W