VOLUME 72, NUMBER 13 P H YSICA L R EV I E% LETTERS 28 MARCH 1994

Observation of a New Charmed Strange MesonY. Kubota, M. Lattery, J.K. Nelson, S. Patton, ~ D. Perticone, ~ R. Poling, V. Savinov, S. Schrenk,R. Wang) M.S. Alarn) I.J. Kim, B. Nemati) J.J. O' Neill) H. Severini, C.R. Sun) M.M. Zoeller,G. Crawford, s C. M. Daubenmier, s R. Fulton, s D. Fujino, s K.K. Gan, a K. Honscheid, a H. Kagan,R. Kass, J. Lee, 3 R. Malchow, F. Morrow, Y. Skovpen, 3 * M. Sung, 3 C. White, s F. Butler, X. Fu, 4

G. Kalb8eisch, 4 W.R. Ross, 4 P. Skubic, 4 J. Snow, 4 P.L. Wang, 4 M. Wood, 4 D.N. Brown, 5 J. Fast, ""

R.L. Mcllwain, s T. Miao, s D.H. Miller, s M. Modesitt, 5 D. Payne, s E.I. Shibata, s I.P.J. Shipsey, 5

P.N. Wang, M. Battle, J. Ernst, Y. Kwon, S. Roberts, 6 E.H. Thorndike, C.H. Wang, 6 J. Dorninick, 7

M. Lambrecht, 7 S. Sanghera, " V. Shelkov, 7 T. Skwarnicki, ~ R. Stroynowski, 7 I. Volobouev, 7 G.Wei, ~ P. Zadorozhny, ~ M. Artuso, s M. Goldberg, s D. He, s N. Horwitz, s R. Kennett, s R. Mountain,G.C. Moneti, s F. Muheim, s Y. Mukhin, s S. Playfer, s Y. Rozen, s S. Stone, s M. Thulasidas, s

G. Vasseur, s G. Zhu, s J. Bartelt, s S.E. Csorna, s Z. Egyed, s V. Jain, s K. Kinoshita, ~o K.W. Edwards, ~~

M. Ogg ~ D.I. Britton E.R.F. Hyatt D.B. MacFarlane z P.M. Patel D.S. Akerib s B. BarishM. Chadha ' S Chan, 's D.F. Cowen, 's G. Eigen 's J.S. Miller 's C. O' Grady' J Urheim, 'A.J. Weinstein, ~s D. Acosta, ~4 M. Athanas, ~4 G. Masek, ~4 H.P. Paar, ~4 J. Gronberg, ~s R. Kutschke, ~s

S. Menary, ~s R.J. Morrison, ~s S. Nakanishi, ~s H.N. Nelson, ~s T.K. Nelson, ~s C. ciao, ~s J.D. Richman, ~5

A. Ryd H. Tajima D. Schmidt D. Sperka M.S. Witherell M. Procario ~ R. Balest 'r K. Cho, '7

M. Daoudi 7 W.T. Ford D.R. Johnson r K. Lingel M Lohner P Rankin, J.G. Smith, ~7

J.P. Alexander C. Bebek K. Berkelman K. Bloom T.E. Browder D.G. Cassel ~ H.A. Cho, ~s

D.M. Coffman, ~s P.S. Drell, ~s R. Ehrlich, ~s M. Garcia-Sciveres, ~s B. Geiser, ~s B. Gittelman, 's

S.W. Gray, ~s D.L. Hartill, s B.K. Heltsley, ' C.D. Jones, ~s S.L. Jones, ' J. Kandaswamy, 's

N. Katayama, ~s P.C. Kim, ~s D.L. Kreinick, 's G.S. Ludwig, ~s J. Masui, ~s J. Mevissen, 's N.B. Mistry, 's

C.R. Ng, ~s E. Nordberg, ~s J.R. Patterson, ~s D. Peterson, ~s D. Riley, ~s S. Salman, ~s M. Sapper, ~s

F. Wurthwein, ~s P. Avery, ~s A. Freyberger, ~s J. Rodriguez, R. Stephens, ~s S. Yang, J. Yelton, ~s

D. Cinabro, zo S. Henderson, za T. Liu, zo M. Saulnier, zo R. Wilson, zo H. Yamamoto, o T. Bergfeld, 2~

B.I. Eisensteinz~ G. Gollinz~ B. Ongz~ M. Palmer z' M. Selenz~ J J Thaler z~ A.J. Sado', zz

R. Ammar zs S. Ball, zs P. Baringer, zs A. Bean, zs D. Besson, zs D. Coppage, zs N. Copty, zs R. Davis, 2s

N. Hancock zs M. Kelly zs N Kwak zs and H. Lamzs

(CLEO Collaboration)' University of Minnesota, Minneapo/is, Minnesota 55/55State University of Nevj York at Albany, Albany, New York 1M99

Ohio State University, Columbus, Ohio, )M10University of Oklahoma, Norman, Oklahoma 7$019Purdue University, West Lafayette, Indiana $7907

University of Rochester, Rochester, New York 1/697Southern Methodist University, Dallas, Texas 7M75

Syracuse University, Syracuse, New York 1M)gVanderbilt University, Nashville, Tennessee 878M

Virginia Polytechnic Institute and State University, B/acksburg, Virginia 9/061"Carleton University, Ottawa, Ontario, Canada K1S 5B6 and the Institute of Particle Physics, Montrda/, Quebec, Canada

McGill University, Montrda/, Quebec, Canada II$A 8T8 and the Institute of Particle Physics, Montrda/, Quebec, CanadaCalifornia Institute of Technology, Pasadena, CaLifornia 91195University of California, San Diego, Ia Jolla, Cahfornia 9M9$

University of California, Santa Barbara, CaLifornia 9$106'' Carnegie-Mellon University, Pittsburgh, Pennsylvania 15218

University of Colorado, Boulder, Colorado 80$09 0$90-Cornell University, Ithaca, New York 1$85$

University of F/orida, Gainesville, Florida $9611Harvard University, Cambridge, Massat."husetts 0815'8

University of I/linois, Champaign Urbana, Illi-nois 61801Ithaca College, Ithaca, Netv York 1/850

University of Kansas, Laamnce, Kansas 660)5(Received 14 January 1994)

Using the CLEO II detector, we have obtained evidence for a new meson decaying to D K+,Its mass is 2573.2+~'e + 0.8 + 0.5 MeV/c and its width is 16+4 + 3 MeV/c . Although we do notestablish its spin and parity, the new meson is consistent with predictions for an L = 1, 8' = 1,J = 2+ charmed strange state.

PACS numbers: 14.40.Lb, 13.25.Ft

003 ] -9007/94/72(1 3)/1972(5)$06.001994 The American Physical Society

VOLUME 72, NUMBER 13 PHYSICAL REVIEW LETTERS 28 MARcH 1994

Mesons with one unit of orbital angular momentum be-tween the quarks exist in four states. If one of the quarksis heavy, then in the limit that its mass approaches in-

finity, its spin decouples from the light degrees of free-dom: the light quark and gluons. The spin of the heavyquark and the total angular momentum of the light de-grees of freedom are separately conserved by the stronginteraction. This is called heavy quark symmetry (HQS);an approximate symmetry remains even for finite massquarks, as long as they are heavy compared to the QCDscale AqcD. Thus models inspired by HQS indicate weshould consider the four states as two doublets [1,2]. Themembers of one doublet, in which the angular momen-tum of the light quark is jg = 3/2, are relatively narrow;they have JP = 1+, 2+. The mesons in the other dou-blet, with jg = 1/2 and J = 0+, 1+, are predicted to bevery broad [2]. When the heavy quark is a charm quark,the light quark can be either an up, a down, or a strangequark. Thus there should be 12 L = 1 charmed mesons:6 relatively narrow states and 6 broad states.

All of these narrow resonances have been observed

[3,4], except for the charmed strange Jp = 2+ meson,designated D,'~+ [5]. The allowed decay modes of theD;2+ are DK and D"K, both proceeding through a Dwave. Because of the limited phase space, the latter is

highly suppressed. Godfrey and Kokoski predict the par-tial width for the decay to DK to be 6 to 10 times largerthan for D'K [2]. The decay to D~+vrs is forbidden byisospin conservation; modes such as D+xm. are suppressedby the Okubo-Zweig-Iizuka rule. Thus we have searchedfor the decays D,'&+ ~ DOK+ and D' K+. Throughoutthis paper, reference to a particular charge state impliesthe inclusion of the charge-conjugate state as well.

The data used in this analysis were collected withthe CLEO II detector at the Cornell Electron StorageRing (CESR). The detector consists of a charged parti-cle tracking system surrounded by time-of-flight (TOF)scintillation counters. These are followed by an electro-magnetic calorimeter, which consists of 7800 thallium-doped CsI crystals. The inner detector is operated in a1.5 T solenoidal magnetic field, generated by a supercon-ducting coil. Finally, the magnet coil is surrounded byiron slabs and muon counters. A detailed description ofthe detector can be found elsewhere [6].

The data were taken at center-of-mass energies equal tothe masses of the T(38) and T(4S), and in the continuumabove and below the T(48). The total integrated lumi-nosity is 2.16 fb i. Events were required to have a mini-mum of five charged tracks and energy in the calorimeterof at least 1570 of the center-of-mass energy.

Specific ionization measurements from the main driftchamber and TOF measurements were used to identifycharged particles. Particles were required to pass a con-sistency cut for the hypothesis in question: kaon or pion.We define y:—( ) + ( ), where Aq is the dif-

ference between the measured and expected specific ion-ization for the hypothesis. Similarly, AT is the differ-

ence between the measured and expected time of flightfor the same hypothesis. Time-of-flight information was

only used when the track's polar angle with respect tothe beain line, 8, met the requirement

~cos8~ & 0.71.

For a track to be considered a pion or kaon candidate,y2 & 6.25 was required for the corresponding hypothesis.

Energy clusters in the calorimeter not matched toa charged track, and which had E & 50 MeV, wereaccepted as photon candidates. To reconstruct mo's,

we used pairs of photons from the good barrel region,

~cos8[ & 0.71, which has the best energy resolution, or

one photon from the good barrel and one photon fromthe "good end eap" region (0.86 &

~

cos 8] ( 0.94), whichhas nearly as good resolution. The invariant mass ofthe two photons was required to be within 2.5 standarddeviations of the zo mass; the mo candidates were thenkinematically fit to the pro mass to improve momentumresolution. The 7rs's used to reconstruct Do's were re-quired to have a minimum energy of 300 MeV; those usedto reconstruct the decay D's ~ Domo were only requiredto have an energy greater than 150 MeV.

We reconstructed De's in the decay modes DK ~+ and Do -+ K ++as. In both cases, the Do can-didates were required to have a measured invariant masswithin 1.65 standard deviations of the observed Do masspeak. The rms mass resolutions are 10 MeV/c for theK n+ mode and 14MeV/c~ for the K n+zo mode. Thedecay angle, o.a-, is defined as the angle between the Kdirection in the De rest frame and the Do direction inthe lab frame. We required Do ~ K z.+ candidatesto satisfy cosua-- & 0.8. This requirement is effectivein reducing the background because the signal is flat in

cosa'-, while the background peaks at 1.In order to reduce the combinatoric background in the

Do -+ K ~+n 0 mode, a parameter is calculated for eachDo candidate, based on its position in the Dalitz plot.The parameter varies from 0 to 1, and is proportionalto the square of the amplitude for decay to the observedlocation in the Dalitz plot. The calculation takes intoaccount the three most important two-body decays andthe nonresonant three-body decay which feed into thisfinal state [7]. The two-body decays included are K p+,K' n+, and K'e~o. We require that the parameter begreater than or equal to 0.1. The efficiency of this cutwas measured to be 64.4 + 1.4% by using the inclusiveD —+ K vr+m sample.

The D candidates were then combined with each pos-itively charged track consistent with being a kaon. Inthe search for D;2+ ~ DOK+, there are several sourcesof background to consider. There is combinatorial back-ground from the various combinations of real and "fake"D 's with real and "fake" K+'s. The fake D 's arecombinations of tracks which accidentally fall withinthe D mass window; the fake K+'s are misidentifiedtracks, mostly pions. The background from real D 's andfake K+'s includes a component from the decay of theD&(2470)+ to Doe+. If these pions are misidentified as

1973

VOLUME 72, NUMBER 13 PHYSICAL REVIEW LETTERS 28 MARCH 1994

kaons, the DOK+ mass reconstructed is "reHected" intothe mass region near our expected signal. This contribu-tion to the background has been measured by recalculat-ing the energy of the K+ candidates using the pion massand the measured momenta. The new momentum-energyfour-vectors were then combined with our D candidates.We observed a peak near 2470 MeV/c2 in the Do7r+ massdistribution. Fitting the peak, we found 27+21 events inthe K 7r+ mode and 23+16 events in the K ~+7ro mode.Using a Monte Carlo simulation, we have parametrizeda shape for this reflected DJ(2470)+ background whichwill be included in our fits to the data. No other res-onance was observed in the Davr+ mass distribution; inparticular there was no peak from partially reconstructedDg(2440)+'s The. widths of the jg = 1/2 mesons arepredicted to be very large [2], and thus should not signif-icantly modify the shape of the background.

To reduce the background from misidentified pions, weimposed an additional ID requirement on the K+ track.We required that the yz for the pion hypothesis for thistrack be at least 2 units larger than that for the kaonhypothesis. The effectiveness of this cut was evaluatedusing the D, i(2536)+ feed-down peak, described below.The cut has an efficiency of (79 6 10)Fo, while reducingthe background by about a factor of 3.

The decay angle, n~+, is defined as the angle betweenthe direction of the K+ in the DaK+ rest frame andthe DOK+ direction in the lab frame. We required thatthe DsK+ combinations have cosn~+ & 0.8. This re-duces the combinatoric background which peaks nearcosa~+ = 1, and also eliminates some of the back-ground from the D&(2470)+ The dis. tribution of the sig-nal events in this angle is unknown, except that it mustbe symmetric about cosn~+ = 0.

Finally, to reduce the background, we take advantageof the hard fragmentation of continuum charm and im-

pose a cut on the x of each D K+ combination. We de-fine x = p/p~ and p~:—(Ebz, —[M(DaK+)]z)i~z For.the K n+ mode, we require x & 0.7; for the K 7r+vrs

mode, which has larger combinatoric background, we re-quire z & 0.8. To improve the mass resolution, we calcu-lated the corrected mass, M' =—M(DOK+) —M(Da) +1864.5 MeV/cz, using the measured invariant masses andthe known Do mass. The corrected mass for D K+ com-binations passing all of the above criteria is shown in

Flg. 1.Two features are prominent in Fig. 1: a feed-down

peak from the D, (i2 53)6+ at about 2392 MeV/c2, and

a broader peak near 2575 MeV/cz, which is a new reso-nance. Each of these is discussed below.

The feed-down peak at 2392 MeV/cz is from theD,i(2536)+, the narrow cs 1+ state, which decays pre-dominantly to D*K. The Q value for the D"a -+ Dana

decay is very small, so even though the 7r is not detected,the peak at 2392 MeV/cz is still very narrow.

Also shown in Fig. 1 is a histogram of M* for

(K sr+)K+ and (K 7r+vro)K+ combinations, where the

[20I

~ t I sI ~ s & s

Is s i s

Is x I &

Is s e

IOO—

CU 80—al

ED II60—C0)

LLI 40 I II I 11ll

20

IIII II

II

IIII s I I

se ~ q s

0 I I I l I I I I I I I I I I

2575 2450 2525I I I I I

2600I i s

2675 2750

M (D K ) (Mevic )

FIG. 1. M', "corrected" invariant mass, of (K sr+[sr ])K+combinations. Data points are for K sr+[sr ] combina-tions in the D signal region; the histogram shows M for

(K m+[rr ])K+ combinations where the K 7r+[m ] combina-tions were chosen in Do sidebands.

K ~+ and K ~+sr combinations were chosen fromthe Do mass sidebands. The K x+ combinationswere chosen from 1800.0 to 1816.5 MeV/cz and 1912.5to 1929.0 MeV/c . The K m+m combinations werechosen from 1777.0 to 1800.0 MeV/cz and 1927.0 to1950.0 MeV/cz. This histogram suggests that about halfof our background cornea from fake Da's; the other halfmust come from real Do's combined with real and fakeK+'s. There appears to be some signal in the sidebandhistogram under both the feed-down peak and the new

resonance. This is due to D —+ K sr+sr events in whicha poorly measured photon, or the wrong photon, wasused to form the vr . We have corrected for this eKect inthe cross-section calculation below. The mass and widthmeasurements are not significantly influenced by this ef-

fect. We have also examined the wrong-sign combina-tions, DOK, and see no enhancements in the invariantmass distribution of such combinations.

To extract the mass and width of the new resonance,we fit the M' distribution for the D K+ combinations,as shown in Fig 2. Th.e fits were done separately forthe two D decay modes. To parametrize the signal we

used a spin-2 relativistic Breit-Wigner convoluted with aGaussian of fixed resolution. The rms resolution, o, was

determined by a Monte Carlo simulation.For the Do ~ K sr+ events the Gaussian had a. =

3.2 MeV/c2. The background, in the mass range from2430 to 2750 MeV/c2, was fit with a first-order poly-nomial, plus the D&(2470)+ background function witha fixed area of 27 events The fit. finds 116+zs signal

events. The mass is 2573.3+z i MeV/cz and the nat-

ural width is 15.6+4 s MeV/cz (statistical errors only).Vfe 6t the D —+ K vr+7r data using the same Gttingfunctions, but with o = 3.7 MeV/cz. The number ofDz(2470)+ background events was fixed at 23. We find

VDLUME 72, NUMsER 13 PH YSICAL R EVI E% LETTERS 28 MAacH 1994

~ I ~ ~ I ~I

~ I ~ II

I ~ I II

I I ~ II

I 1 ~ ~ I I ~ ~

80—

N QQ

4P

(0EA

C:Cl

LLJ

II40 Ill I II '

II

20—

pl s I s s s

2450

I

t&i'

III I I 4 I I I I I I

~ I I~ I ~ ~ - II,„'I~ ~ I I ~ I I / IIt"

II II

,r'/I

I./

I.J.~e" i s s I I s s s I s i i s I i I ~

'I

2500 2550 2600 2650 2TOO 2750

M (0 K ) (MeV/c )

FIG. 2. Histogram of M'(D K+), with fit. The solid lineshows the complete signal and background fitting functions.The sum of the background functions is shown by the dashedline. The dotted line shows just the polynomial used torepresent the combinatoric background. The shape of theDz(2470)+ background function is shown at the bottom bythe dash-dotted line, with the area scaled up by a factor of 5.

101+~s signal events. Using this mode, the mass is mea-

sured to be 2573.1+& s MeV/c~, and the natural width

to be 17.6+s 0 MeV/c~ (statistical errors only), in goodagreement with the first mode.

Although the fits were done separately for the two Dsdecay modes, the data for the two modes are added to-gether in Fig. 2, and the sum of the fitting functions issuperimposed. The complete signal and background fit isshown by the solid line. The total background is shown

by the dashed line. The dotted line shows the polynomialrepresenting the combinatoric background. The shape ofthe Dz(2470)+ background, with the area scaled up bya factor of 5, is shown at the bottom by the dash-dotline. The total signal has a significance of more than sixstandard deviations.

Following the nomenclature of the Particle Data Groupfor a meson of unknown spin [5], we will use the tempo-rary designation "D;&(2573)+" for this new resonance.We estimate the systematic error on the D,'& —D massdifference to be +0.8 MeV/c2 and on the width to be+3 MeV/c2. This includes contributions from varyingthe assumed mass resolution and spin, the mass of theD&(2470)+ and the number of events it contributes tothe background, the order of the polynomial used forthe combinatoric background, the binning, and other de-tails of the fitting. The largest contribution to the sys-tematic error on the width, +1.8 MeV/c2, comes fromvarying the order of the polynomial used to fit the back-ground. The next largest, +1.5 MeV/c~, comes from theuncertainty in the number of Dz(2470)+ events in thebackground. The latter also contributes the majority ofthe systematic error on the mass: +0.6 MeV/c~. Thesesystematic errors are common to both D decay modes.

Averaging the two sets of values, we find that the D;&+

has a natural width of 16+4 + 3 MeV/c and a mass of2573.2+i s + 0.8 + 0.5 MeV/c~, where the third error onthe mass is due to the uncertainty in the D mass.

Our width measurement implies that this new statedecays strongly. Assuming this is so, angular momentumand parity conservation require that its spin-parity be inthe so-called "natural" series: 0+, 1, 2+, 3, etc.

To measure the production cross section for x & 0.7times the branching ratio to DOK+, we have remea-sured the yield in the second decay mode with the 2:cut reduced to 0.7. We have also removed our cuts oncos n~+ in both modes, since the distribution of the sig-nal events in this variable is unknown. Our efficiencyfor reconstructing the DOK+, Do -+ K n.+ combina-tions is (32 + 4)%. The efficiency for reconstructing theDOK+, Do ~ K a+no combinations is (9.2 6 1.4)%.For the Do ~ K a+no mode, the number of eventswas reduced by 22% to account for the peaking of thebackground under the signal. A contribution of 9% wasincluded in the systematic error to account for the un-

certainty in this correction. Using the recent CLEOmeasurement of 8(DD ~ K n+) = (3.91 6 0.19)% [8],the Particle Data Group's value for the ratio 8(DD -+K ++~0)/8(D0 ~ K n+) = 3.10 6 0.26 [5) and ourmeasured luminosity of 2.16 6 0.02 fb i, we find thatfor z ) 0.7, the cross section times branching fractionis o(x ) 0.7)8(D;+ -+ DOK+) = 4.4 + 0.9 6 0.7 pb.The first error refiects the uncertainty in the number ofevents, both statistical and systematic; the two contri-butions are comparable. It was calculated separately forthe two modes and averaged. The second error is the sys-tematic error common to both modes and is dominatedby the uncertainty in the efficiencies.

We have also searched for the decay of this new res-onance to D'OK+, D'0 ~ Done. This mode is allowedfor the D;~+, but expected to be highly suppressed bythe limited phase space. We used the same Do decaymodes and similar cuts as in the previous analysis. Inaddition, we consider the helicity angle of the pro fromthe D*c decay. The helicity angle, 8, is defined as theangle between the D;J+ and the pro, both measured inthe D"0 rest frame. For any meson with spin-parity inthe "natural" series, the helicity angle of the m.o from theD'0 decay must have a sin 8 distribution. We require

] cos 8 ]( 0.75.

We find no signal above background and set the fol-lowing limit on the ratio of branching fractions:

8(D;J(2573)+ ~ D'OK+)

8(D;z(2573)+ ~ DOK+)

at the 90% confidence level. For the D;2+, this ratio ispredicted to be 0.1—0.16 [2].

In summary, we find a signal with a significance ofmore than six standard deviations for a meson decayingto DO%+. The mass and natural width of the new state

1975

VOLUME 72, NUMBER 13 PH YSICAL REV I EW LETTERS 28 MARCH 1994

are measured to be 2573.2+~ s 6 0.8 6 0.5 MeV/c2 and16+4 + 3 MeV/cs, respectively. Though its spin-parityis not established, it must be in the "natural" 0+, 1

2, 3, . . . series. We tentatively identify this state asthe D;2+, the 2+ partner of the D,q(2536)+. The masssplitting between these two states, 38.1+&'s+0.8 MeV/c,is comparable to that seen between the Dq(2420)s andthe Dz(2460)0, the neutral 1+ and 2+ states. The D+~ isestablished as a 1+ meson [4]; thus the D+& and this newresonance appear to form a similar doublet. Its width isinconsistent with that predicted for a 0+ state, while bothits width and decay modes are consistent with predictionsfor the D;z [2].

We gratefully acknowledge the effort of the CESR stafFin providing us with excellent luminosity and runningconditions. This work was supported by the NationalScience Foundation, the U.S. Department of Energy, theHeisenberg Foundation, the SSC Fellowship program ofTNRLC, and the A.P. Sloan Foundation.

Permanent address: INP, Novosibirsk, Russia.[1] N. Isgur and M.B.Wise, Phys. Rev. Lett. 66, 1130 (1991).[2] S. Godfrey and R. Kokoski, Phys. Rev. D 4$, 1679 (1991).

We have recalculated the relative widths for D,'2 ~ DKand D,2 ~ D'K using the measured meson masses.

[3] ARGUS Co//aboration, H. Albrecht et a/. , Phys. Rev. Lett.56, 549 (1986); Phys. Lett. B 2$0, 162 (1989); 2$1, 208(1989); 2$2, 398 (1989); E691 Collaboration, J. C. Anjoset a/. , Phys. Rev. Lett. 62, 1717 (1989); CLEO Collabo-ration, P. Avery e/ a/. , Phys. Rev. D 41, ?74 (1990).

[4] CLEO Collaboration, J. P. Alexander et aL, Phys. Lett.B $0$, 377 (1993).

[5] Particle Data Group, K. Hikasa e/ a/. , Phys. Rev. D 45,Sl (1992).

[6] CLEO Collaboration, Y. Kubota e/ a/. , Nucl. Instrum.Methods Phys. Res. , Sect A $20, 66 (1992).

[7] E691 Collaboration, J. C. Anjos et a/. , Phys. Rev. D 48,56 (1993).

[8] CLEO Collaboration, D. S. Akerib e/ a/. , Phys. Rev. Lett.71, 3070 (1993).

1976