

 21 Mar 1994

A MEASUREMENT OF B(D+s ! OEl+*)=B(D+s ! OEss+) F. Butler,1 X. Fu,1 G. Kalbfleisch,1 W.R. Ross,1 P. Skubic,1 J. Snow,1 P.L. Wang,1 M. Wood,1 D.N. Brown,2 J.Fast ,2 R.L. McIlwain,2 T. Miao,2 D.H. Miller,2 M. Modesitt,2

D. Payne,2 E.I. Shibata,2 I.P.J. Shipsey,2 P.N. Wang,2 M. Battle,3 J. Ernst,3 Y. Kwon,3 S. Roberts,3 E.H. Thorndike,3 C.H. Wang,3 J. Dominick,4 M. Lambrecht,4 S. Sanghera,4

V. Shelkov,4 T. Skwarnicki,4 R. Stroynowski,4 I. Volobouev,4 G. Wei,4 P. Zadorozhny,4 M. Artuso,5 M. Goldberg,5 D. He,5 N. Horwitz,5 R. Kennett,5 R. Mountain,5 G.C. Moneti,5

F. Muheim,5 Y. Mukhin,5 S. Playfer,5 Y. Rozen,5 S. Stone,5 M. Thulasidas,5 G. Vasseur,5

G. Zhu,5 J. Bartelt,6 S.E. Csorna,6 Z. Egyed,6 V. Jain,6 K. Kinoshita,7 K.W. Edwards,8

M. Ogg,8 D.I. Britton,9 E.R.F. Hyatt,9 D.B. MacFarlane,9 P.M. Patel,9 D.S. Akerib,10

B. Barish,10 M. Chadha,10 S. Chan,10 D.F. Cowen,10 G. Eigen,10 J.S. Miller,10 C. O'Grady,10 J. Urheim,10 A.J. Weinstein,10 D. Acosta,11 M. Athanas,11 G. Masek,11 H.P. Paar,11 J. Gronberg,12 R. Kutschke,12 S. Menary,12 R.J. Morrison,12 S. Nakanishi,12

H.N. Nelson,12 T.K. Nelson,12 C. Qiao,12 J.D. Richman,12 A. Ryd,12 H. Tajima,12 D. Sperka,12 M.S. Witherell,12 M. Procario,13 R. Balest,14 K. Cho,14 M. Daoudi,14 W.T. Ford,14 D.R. Johnson,14 K. Lingel,14 M. Lohner,14 P. Rankin,14 J.G. Smith,14

J.P. Alexander,15 C. Bebek,15 K. Berkelman,15 K. Bloom,15 T.E. Browder,15\Lambda D.G. Cassel,15 H.A. Cho,15 D.M. Coffman,15 P.S. Drell,15 R. Ehrlich,15 P. Gaiderev,15

M. Garcia-Sciveres,15 B. Geiser,15 B. Gittelman,15 S.W. Gray,15 D.L. Hartill,15 B.K. Heltsley,15 C.D. Jones,15 S.L. Jones,15 J. Kandaswamy,15 N. Katayama,15 P.C. Kim,15

D.L. Kreinick,15 G.S. Ludwig,15 J. Masui,15 J. Mevissen,15 N.B. Mistry,15 C.R. Ng,15 E. Nordberg,15 J.R. Patterson,15 D. Peterson,15 D. Riley,15 S. Salman,15 M. Sapper,15 F. W"urthwein,15 P. Avery,16 A. Freyberger,16 J. Rodriguez,16 R. Stephens,16 S. Yang,16

J. Yelton,16 D. Cinabro,17 S. Henderson,17 T. Liu,17 M. Saulnier,17 R. Wilson,17 H. Yamamoto,17 T. Bergfeld,18 B.I. Eisenstein,18 G. Gollin,18 B. Ong,18 M. Palmer,18 M. Selen,18 J. J. Thaler,18 A.J. Sadoff,19 R. Ammar,20 S. Ball,20 P. Baringer,20 A. Bean,20 D. Besson,20 D. Coppage,20 N. Copty,20 R. Davis,20 N. Hancock,20 M. Kelly,20 N. Kwak,20

H. Lam,20 Y. Kubota,21 M. Lattery,21 J.K. Nelson,21 S. Patton,21 D. Perticone,21 R. Poling,21 V. Savinov,21 S. Schrenk,21 R. Wang,21 M.S. Alam,22 I.J. Kim,22 B. Nemati,22

J.J. O'Neill,22 H. Severini,22 C.R. Sun,22 M.M. Zoeller,22 G. Crawford,23 C. M. Daubenmier,23 R. Fulton,23 D. Fujino,23 K.K. Gan,23 K. Honscheid,23 H. Kagan,23

R. Kass,23 J. Lee,23 R. Malchow,23 Y. Skovpen,23y M. Sung,23 and C. White23

(CLEO Collaboration)

1

1University of Oklahoma, Norman, Oklahoma 73019

2Purdue University, West Lafayette, Indiana 47907 3University of Rochester, Rochester, New York 14627

4Southern Methodist University, Dallas, Texas 75275

5Syracuse University, Syracuse, New York 13244 6Vanderbilt University, Nashville, Tennessee 37235 7Virginia Polytechnic Institute and State University, Blacksburg, Virginia, 24061 8Carleton University, Ottawa, Ontario K1S 5B6 and the Institute of Particle Physics, Canada

9McGill University, Montr'eal, Qu'ebec H3A 2T8 and the Institute of Particle Physics, Canada

10California Institute of Technology, Pasadena, California 91125 11University of California, San Diego, La Jolla, California 92093

12University of California, Santa Barbara, California 93106 13Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213

14University of Colorado, Boulder, Colorado 80309-0390

15Cornell University, Ithaca, New York 14853 16University of Florida, Gainesville, Florida 32611 17Harvard University, Cambridge, Massachusetts 02138 18University of Illinois, Champaign-Urbana, Illinois, 61801

19Ithaca College, Ithaca, New York 14850 20University of Kansas, Lawrence, Kansas 66045 21University of Minnesota, Minneapolis, Minnesota 55455 22State University of New York at Albany, Albany, New York 12222

23Ohio State University, Columbus, Ohio, 43210

(January 25, 1994)

Abstract Using the CLEO II detector at CESR, we have measured the ratio of branching fractions B(D+s ! OEe+*)=B(D+s ! OEss+) = 0:54 \Sigma 0:05 \Sigma 0:04. We use this measurement to obtain a model dependent estimate of B(D+s ! OEss+).

\Lambda Permanent address: University of Hawaii at Manoa yPermanent address: INP, Novosibirsk, Russia

2

Most measurements of the D+s meson branching fractions are normalized to the clean D+s ! OEss+ channel. [1] However, the absolute D+s ! OEss+ branching fraction is not well known, and this limits the precision of these measurements. Here we present a measurement of Rs = B(D+s ! OEl+*)=B(D+s ! OEss+), which can be used to extract the D+s ! OEss+ branching fraction by using the known values for the D ! _K\Lambda l* branching fractions, the D and D+s meson lifetimes and theoretical predictions for the ratio of widths: \Gamma (D+s ! OEl+*)=\Gamma (D ! _K\Lambda l+*).

The data consist of an integrated luminosity of 1.71 fb\Gamma 1 of e+e\Gamma collisions recorded with the CLEO II detector at the Cornell Electron Storage Ring (CESR). A detailed description of the CLEO II detector can be found in reference [2]. The data sample contains over two million e+e\Gamma ! c_c events taken at center-of-mass energies on the \Upsilon (4S) resonance and in the nearby continuum (ps , 10:6 GeV).

Due to the undetected neutrino, we cannot fully reconstruct D+s ! OEl+* decays. However, there are very few processes which produce both a OE meson and a lepton contained in the same jet. Consequently, this correlation can be used to extract a clean D+s ! OEl+* signal. The backgrounds due to misidentified leptons and from random OE-lepton combinations can be reliably estimated, and the possible contamination from other decay modes is shown to be negligible.

We identify OE candidates by using the decay mode OE ! K+K\Gamma . In order to suppress combinatoric background, OE candidates are required to have momenta above 1.1 GeV/c. In addition, the kaon candidates must have ionization energy loss and time-of-flight consistent with that expected for a kaon with the measured momentum.

The search for leptons is restricted to the kinematic regions in which the lepton identification efficiencies and hadron misidentification rates are well understood. Hence, electron and muon candidates are required to be in the fiducial regions j cos `j ! 0:91 and j cos `j ! 0:81, respectively, where ` is the polar angle of the track with respect to the beam-axis. In addition, electron candidates must have momenta above 0.9 GeV/c and muon candidates above

3

1.4 GeV/c. The only exception is for muons in the region j cos `j ? 0:61 which are required to have momenta above 1.9 GeV/c. Electrons are identified by comparing their ionization energy loss, time-of-flight, and energy deposited in the electromagnetic calorimeter with that expected for an electron with the measured momentum. Electrons from photon conversions and Dalitz decays of ss0's are rejected by pairing electron candidates with all other oppositely charged tracks in the event and rejecting those which have both small separation and parallel trajectories at their point of closest approach. Muons are identified by matching charged tracks to hits in the muon detectors which lie outside the electromagnetic calorimeter. In order to be identified as a muon, a track must penetrate at least 5 interaction lengths of steel. For leptons in the momentum ranges and fiducial regions considered, the identification efficiencies are approximately 92% for electrons and 90% for muons.

To reduce further the combinatoric background, we require that the OEl+ momentum be greater than 2.4 GeV/c. In order to be consistent with having originated from a D+s decay, the OEl+ candidates must have an invariant mass less than 1.9 GeV/c2. In order to suppress the combinatoric background from \Upsilon (4S) events which tend to be more spherical, we require that the ratio of Fox-Wolfram moments [3], R2 = H2=H0, is greater than 0.30. This eliminates 77% of the \Upsilon (4S) background whilst retaining 92% of the signal.

The efficiencies for reconstructing D+s ! OEl+* decays are obtained from a Monte Carlo simulation which takes the predictions of the ISGW model [4] as input. These events are then passed through a full simulation of the CLEO II detector and the same event reconstruction and analysis chain as the real data. Because of the small q2 value associated with the decay OE ! K+K\Gamma , the kaons tend to overlap in the drift-chamber. This makes it difficult to simulate accurately the ionization energy loss measurement. In order to avoid this problem, the momentum dependent efficiencies for identifying OE mesons are obtained from the data by comparing the inclusive yield of all OE's before and after particle identification. These efficiencies are then combined with the predicted OE momentum spectrum from D+s ! OEl+* decays to give the total OE identification efficiency. Following the above selection criteria, and

4

after correcting for the effects of final-state radiation from the leptons [5], the efficiencies for identifying D+s ! OEe+* decay is 10.5% and for D+s ! OE_+* the efficiency is 2.9%.

Figs 1(a) and 1(b) show the invariant mass distributions of all K+K\Gamma combinations which are accompanied by an electron or muon respectively and which pass the above selection criteria. We fit these distributions with a signal and background function. The signal function is a Gaussian function convoluted with a Breit-Wigner function. The background function is a phase-space background function [6] which accounts for random K+K\Gamma combinations. The width of the Breit-Wigner function is fixed to the natural width of the OE state [8], and the mean and sigma of the Gaussian function are fixed to the values extracted from a fit to all OE candidates with momenta above 1.1 GeV/c. Only the overall normalization of the signal function is allowed to vary in the fits. The fits yield 359 \Sigma 22 D+s ! OEe+* and 123 \Sigma 15 D+s ! OE_+* candidates. There are two main sources of background: OE's accompanied by fake leptons [7], and random OEl+ combinations.

The background due to fake leptons is estimated by first using the real data to measure the momentum dependent probabilities that a hadron will be misidentified as a lepton. These probabilities are typically 0.3% for electrons and 1.2% for muons. These results are then used to randomly label tracks (which do not pass the lepton identification criteria described above) as leptons in the data. With this procedure, we extract the number of OEl+ combinations due to misidentified hadrons. For electrons this estimate is 46 \Sigma 14 events, while for muons it is 27 \Sigma 8. The quoted errors include the contributions from the uncertainties in the misidentification probabilities. In order to check these estimates, we examine the invariant mass distribution of all OEl+ candidates. A peak at the D+s mass is seen which is due to D+s ! OEss+ decays in which the pion is misidentified as either an electron or muon. The technique of randomly labeled hadronic tracks as leptons yields 7:4 \Sigma 2:2 OEss events where the ss track is misidentified as a lepton. This is in good agreement with the 4:5 \Sigma 3:6 events found in the lepton sample.

For the range of lepton momenta considered, random OEl+ combinations come from two

5

sources: from e+e\Gamma ! c_c events in which a OE is produced in the fragmentation process and is combined with a lepton from the semileptonic decay of the charmed hadron in the same jet, and from \Upsilon (4S) decays in which a OE is produced in the decay chain of one of the B mesons and is combined with a lepton from the semileptonic decay of the other B meson. A OE and lepton which originate from the decay of the same B meson do not contribute since they tend to be emitted back-to-back and thus have too large an invariant mass. The background from random OEl+ combinations is estimated using the Monte Carlo simulation. However, this is complicated by the fact that the OE production rate from both fragmentation and B meson decays is not well known. For this reason, an attempt is made to scale the Monte Carlo prediction to account for the OE production rate observed in the data.

In the continuum, not only the rate of OE production, but also the correlation between the OE and the charmed hadron direction is important. The agreement between the data and Monte Carlo is investigated by considering how often a OE is produced in the same hemisphere as a fully reconstructed D meson. Both D0 and D\Lambda + mesons are considered, and are reconstructed using the following decay chains: D0 ! K\Gamma ss+ and D\Lambda + ! D0ss+; D0 ! K\Gamma ss+. The reconstructed D mesons are required to have momenta above 2.5 GeV/c in order to account approximately for the range of D momenta which are expected to contribute leptons in the momentum range of interest. For this particular study, the OE momentum criterion is relaxed to 0.8 GeV/c in order to provide sufficient statistics. In doing this we have assumed that the OE momentum distribution is well reproduced by the Monte Carlo, and that it is the rate of OE production in the fragmentation process which contributes the greatest uncertainty. Both the number of D mesons and the number OE's are obtained by fitting their invariant mass distributions. False combinations due to the D meson combinatoric backgrounds are accounted for by subtracting the number of OE's found when using the D mesons invariant mass sidebands. In the real data 0:17 \Sigma 0:11 OE's are found for every 1000 reconstructed D mesons. This is to be compared with 0:16 \Sigma 0:02 for the e+e\Gamma ! c_c Monte Carlo. The ratio of these two numbers is 1:0 \Sigma 0:7, so no correction is applied in this case. The simulation

6

predicts a background of 12 \Sigma 8 and 1:8 \Sigma 1:2 events for electrons and muons respectively, where the errors include the uncertainty in the above ratio.

The background from random OEl+ combinations in \Upsilon (4S) decays is estimated in a similar manner. In this case the directions of the OE and lepton are uncorrelated. For this reason it is sufficient to compare the number of OE's with momentum above 1.1 GeV/c in the continuum subtracted \Upsilon (4S) data with that observed in the \Upsilon (4S) B _B Monte Carlo. In the real data 5:0 \Sigma 0:5 OE's are found per 1000 B _B events to be compared with 5:3 \Sigma 0:1 in the Monte Carlo. This gives a correction factor of 0:95 \Sigma 0:08. After applying this correction, the predicted background is 19 \Sigma 2 events for electrons and 9 \Sigma 1 events for muons.

Figs 2(a) and 2(b) show the number of OE's which fall in each OEl+ invariant mass bin for electrons and muons respectively; where the number of OE's has been extracted from fits to the K+K\Gamma invariant mass distributions. The combined background estimates are also shown, as well as the simulated predictions for the signal shapes which have been normalized to the number of candidates extracted from the fits to the K+K\Gamma invariant mass spectra. It can be seen that the predicted signal shapes are in good agreement with the data. The background estimates can be checked by comparing the predicted number of candidates which fall outside of the signal region with the number actually observed. For electrons we predict 8 \Sigma 1 events in the region 2:0 ! MOEl+ ! 3:5 GeV/c2 and observe 12 \Sigma 7, and for muons we predict 7 \Sigma 1 and observe 8 \Sigma 5. Both predictions are in good agreement with the data.

We have also estimated the possible contamination from the decays D+ ! OE _K0l+*, D+s ! OEjl+* and D+s ! OEssssl+*. The decay D+s ! OEss0l+* is forbidden from conservation of isospin. The Feynman diagrams for the first two processes are shown in Figs. 3(a) and 3(b), respectively. For the first decay, we first estimate an upper limit on the number of D+ ! OE _K0l+* decays in the data by making the conservative assumption: B(D+ ! OE _K0l+*) =B(D+ ! _K0l+*) ' ff \Delta B(D0 ! (K\Lambda ss)\Gamma _+*)=B(D0 ! K\Gamma _+*), where ff is a suppression factor because in the first numerator it is an s_s pair which must be popped from the vacuum as opposed to a light-quark pair [9]. This assumption is motivated by the similarity of

7

the decay diagrams. The latter ratio has been measured by the E653 collaboration [10], B(D0 ! (K\Lambda ss)\Gamma _+*)=B(D0 ! K\Gamma _+*) ! 0:04 at the 90% confidence level. Taking the number of D+ ! _K0l+* events in our data sample to be 60,000 [11], assuming ff = 1=3, and including the simulated acceptance for D+ ! OE _K0l+* decays, we estimate less than one background event from this source. The contribution from D+s ! OEjl+* decays is estimated in a similar manner. Here we make use of the same E653 result and assume: B(D+s ! OEjl+*)=B(D+s ! OEl+*) ' ff \Delta fi \Delta B(D0 ! (K\Lambda ss)\Gamma _+*)=B(D0 ! K\Gamma _+*), where fi accounts for OZI suppression [12] of the first numerator. Making the assumption fi = 1=10, and including the simulated acceptance for D+s ! OEjl+* decays, we again estimate much less than one background event from this source. The contribution from D+s ! OEssssl+* decays should also be very small for similar reasons. Therefore, it is assumed that the background from these decay modes is negligible.

After subtracting all backgrounds, we find 282 \Sigma 22 D+s ! OEe+* and 85 \Sigma 15 D+s ! OE_+* candidates which fall in the D+s signal region MOEl+ ! 1:9 GeV/c2. After correcting for the detection efficiencies in each channel and for the OE ! K+K\Gamma branching fraction [8], the efficiency corrected yields are 5460 \Sigma 430 for electrons and 6000 \Sigma 1000 for muons. A breakdown of the yields in each channel is given in Table I. To combine these two numbers, we take a weighted average, after first correcting for the fact that the muon rate is predicted to be 5% lower than that for electrons because of the reduced phase-space [13]. Therefore, our result is given in terms of the effective yield in the electron channel which is 5580 \Sigma 400 events.

In order to limit the systematic effects which stem from the selection criteria, the number of D+s ! OEss+ decays is measured in a similar manner. Again, the OE candidates are required to have momenta above 1.1 GeV/c . To account approximately for the fact that no neutrino is produced in this decay, we require the OEss+ momentum to be greater than 2.7 GeV/c [14]. We then require the K+K\Gamma ss+ invariant mass to be within \Sigma 25 MeV/c2 of the known D+s mass [8]. The efficiency for detecting D+s ! OEss+ decays following these selection criteria is

8

17.4%.

The result of the fit to the K+K\Gamma invariant mass distribution is shown in Fig. 4. We find 1049 \Sigma 35 candidates. Also shown is the result of the fit to K +K\Gamma combinations from the D+s mass sidebands, which is used to estimate the contribution from random OEss+ combinations. We find 163 \Sigma 18 candidates due to these random combinations. After subtracting this background, the efficiency corrected yield is 10; 370 \Sigma 460 events.

Finally, since we have already corrected for efficiencies, the ratio of branching fractions is,

Rs = B(D

+ s ! OEe+*)B

(D+s ! OEss+) =

5580 \Sigma 400 10; 370 \Sigma 460 = 0:54 \Sigma 0:05 \Sigma 0:04; (1)

where the first error is statistical, and the second is an estimate of possible systematic effects. This systematic error includes: the uncertainty in the number of fake leptons (6.3%), the uncertainty in the level of continuum charm background (2.7%), the uncertainty in the level of B _B background (0.9%), the uncertainty in the lepton identification efficiency (2.5%), the uncertainty in the OE identification efficiency (1.0%) and that due to the limited number of Monte Carlo events which were used for the efficiency estimates (2.7%). We have also considered our sensitivity to the D+s production mechanism by using the predicted efficiencies for D+s mesons produced in D\Lambda +s decays. However, since both the D+s ! OEl+* and the D+s ! OEss+ efficiencies are affected in the same manner, the effect is small (0.5%). For this analysis the ISGW model was used to generate semileptonic decays in the Monte Carlo simulation. The uncertainty associated with this choice of model was investigated by adopting the (V \Gamma A) prediction for the lepton momentum spectra. From the resulting change in the predicted D+s ! OEl+* acceptance, we assign a systematic error of 1.8%. Various background shapes have also been used to extract the number of OE mesons. In all cases Rs changes by less than 1:1%, which is taken to be the systematic error. After adding these estimates in quadrature, the total systematic error is 7:6%. Table II compares this result with those of previous measurements [15-17].

Having measured Rs, we extract the D+s ! OEss+ branching fraction by using the theo9

retical value for Fs = \Gamma (D+s ! OEl+*)=\Gamma (D ! _K\Lambda l+*). We can write

B(D+s ! OEe+*) = Fs \Theta B(D0 ! K\Lambda \Gamma e+*) \Theta o/D

+ s

o/D0

= Fs \Theta B(D

0 ! K\Lambda \Gamma e+*)

B(D0 ! K\Gamma ss+) \Theta B(D

0 ! K\Gamma ss+) \Theta o/D+s

o/D0 : (2)

where o/D+

s and o/D0 are the D

+ s and D0 lifetimes, respectively. Fs is predicted by various

quark models [18,19]. Here we choose to adopt the result of the modified ISGW model

Fs = 1:00, since to date this is the only model which can account for the measured value for B(D ! _K\Lambda l+*)=B(D ! Kl+*) [19,20]. Past experiments have used Fs = 0:9 for this ratio. We use CLEO II measurements for all quantities except the D0 and D+s lifetimes. In this way some of the systematic errors cancel and the problems associated with averaging the results of many different experiments are avoided. We found B(D0 ! K\Lambda \Gamma e+*)=B(D0 ! K\Gamma ss+) = 0:61 \Sigma 0:07 [11], and when this is combined with our measurement, B(D0 ! K\Gamma ss+) = (3:91 \Sigma 0:19)% [21], and with the E687 measurements, o/D0 = (4:13 \Sigma 0:05) \Theta 10\Gamma 13 s [22] and o/D+

s = (4:75 \Sigma 0:21) \Theta 10

\Gamma 13 s [23], we obtain B(D+

s ! OEe+*) = (2:74 \Sigma 0:36)%. Using our

measurement of Rs, we obtain: B(D+s ! OEss+) = (5:1 \Sigma 0:4 \Sigma 0:4 \Sigma 0:7)%, where the first

error is from the statistical error on Rs, the second from the systematic error on Rs, and the third from the uncertainty in the D+s ! OEe+* branching fraction. This measurement is greater than but consistent with previous estimates and upper limits on B(D+s ! OEss+) [8,24].

In conclusion, we have measured B(D+s ! OEe+*)=B(D+s ! OEss+) = 0:54 \Sigma 0:05 \Sigma 0:04. By using the theoretical prediction Fs = 1:00, we find B(D+s ! OEss+)= (5:1 \Sigma 0:4 \Sigma 0:4 \Sigma 0:7)%.

We gratefully acknowledge the effort of the CESR staff in providing us with excellent luminosity and running conditions. J.P.A. and P.S.D. thank the PYI program of the NSF, I.P.J.S. thanks the YI program of the NSF, G.E. thanks the Heisenberg Foundation, K.K.G., I.P.J.S., and T.S. thank the TNRLC, K.K.G., H.N.N., J.D.R., T.S. and H.Y. thank the OJI program of DOE and P.R. thanks the A.P. Sloan Foundation for support. This work was supported by the National Science Foundation and the U.S. Dept. of Energy.

10

REFERENCES [1] For all states described, the charge conjugate state is also implied. [2] CLEO Collaboration, Y. Kubota et al., Nucl. Inst. and Meth. A320, 66(1992). [3] G.C. Fox and S. Wolfram, Phys. Rev. Lett. 41, 1581(1978). [4] N. Isgur et al., Phys. Rev. D 39, 799(1989). [5] D. Atwood and W.J. Marciano, Phys. Rev. D 41, 1736(1990). Final-state radiation

reduces the D+s ! OEe+* acceptance by 1:1%, and has a negligible effect in the muon channel.

[6] The phase-space background function has the form: a \Delta (m \Gamma m0)ff \Delta e\Gamma fi(m\Gamma m0), where m

is the invariant mass of the K+K\Gamma pair. The parameter m0 is first obtained from a fit to the inclusive K+K\Gamma invariant mass distribution, and ff and fi are free parameters.

[7] Fake leptons are either misidentified hadrons or muons from ss+ and K+ decays in flight. [8] Particle Data Group, K. Hikasa et al., Review of Particle Properties, Phys. Rev. D 45,

1(1992).

[9] The notation (K\Lambda ss)\Gamma is used to represent the sum of all possibilities: _K\Lambda 0ss\Gamma , K\Lambda \Gamma ss0,

and higher order resonances which decay via these states.

[10] E653 Collaboration, K. Kodama et al., Phys. Lett. B 313, 260(1993). [11] CLEO Collaboration, J. Alexander et al., Phys. Lett. B 317, 647(1993). [12] J.L. Rosner, Phys. Rev. Lett. 22, 689(1969). [13] J.G. K"orner and G.A. Schuler, Z. Phys. C 46, 93(1990). [14] This should be compared with the requirement pOEl+ ? 2:4 GeV/c which was used to

select D+s ! OEl+* candidates.

11

[15] CLEO Collaboration, J. Alexander et al., Phys. Rev. Lett. 65, 1531(1990). [16] ARGUS Collaboration, H. Albrecht et al., Phys. Lett. B 255, 634(1991). [17] E687 Collaboration, P.L. Frabetti et al., Phys. Lett. B 313, 253(1993). [18] M. Wirbel et al., Z. Phys. C 29, 269(1985). This model predicts 0.83 for the ratio of

\Gamma (D+s ! OEl+*)=\Gamma (D ! _K\Lambda l+*).

[19] D. Scora, Nucl. Phys. A 527, 743c(1991); D. Scora, Ph.D. thesis, University of Toronto,

1993.

[20] D. Scora and N. Isgur, Phys. Rev. D 40, 1491(1989). [21] CLEO Collaboration, D.S. Akerib et al., Phys. Rev. Lett. 71, 3070(1993). [22] E687 Collaboration, P.L. Frabetti et al., preprint FERMILAB-Pub-93/332-E, submitted

to Phys. Lett. B.

[23] E687 Collaboration, P.L. Frabetti et al., Phys. Rev. Lett. 71, 827(1993). [24] F. Muheim and S. Stone, preprint HEPSY 93-3, to appear in Phys. Rev. D. They

estimate B(D+s ! OEss+) = (3:5 \Sigma 0:6)%.

12

FIGURES (a)

FIG. 1. Fits to the K+K\Gamma invariant mass distributions for (a) K+K\Gamma e+ and (b) K+K\Gamma _+ combinations which lie in the D+s signal region MK+K\Gamma l+ ! 1:9 GeV/c2.

13

(b)

14

(a) FIG. 2. Invariant mass of (a) OEe+ and (b) OE_+ combinations. The data points are obtained by fitting the K+K\Gamma invariant mass distributions for each OEl+ invariant mass bin. The solid histograms show the sums of the predicted backgrounds and the simulated signal shapes. The dashed histograms show the background contributions.

15

(b)

16 c

s s \Phi

(b)

s c

d d

s n

\Phi K0

(a) D+

D+s s

h

n s

s

s

FIG. 3. Feynman diagrams for the possible background modes. (a) D+ ! OE _K0l+*, in which an s_s pair must be "popped" from the vacuum, and (b) D+s ! OEjl+*, which is OZI suppressed.

17

FIG. 4. Fit to the K+K\Gamma invariant mass distribution for K+K\Gamma ss+ combinations which lie within \Sigma 25 MeV/c2 of the D+s mass. The dashed histogram shows the contribution from the D+s mass sidebands.

18

TABLES TABLE I. Summary of D+s ! OEl+* yields. The errors quoted in this table are statistical only.

Decay mode D+s ! OEe+* D+s ! OE_+* Total candidates 359 \Sigma 22 123 \Sigma 15 Fake lepton background 46 \Sigma 0:3 27 \Sigma 0:8 Continuum c_c background 12 \Sigma 0:4 1:8 \Sigma 0:1

B _B background 19 \Sigma 0:8 9 \Sigma 0:5 Background subtracted 282 \Sigma 22 85 \Sigma 15

Efficiency, ffl \Delta B (%) 5.16 1.42 Efficiency corrected yield 5460 \Sigma 430 6000 \Sigma 1000

TABLE II. Comparison of this result with those of previous experiments. We have increased the E687 result by 5% since only muons were used in their analysis.

Experiment Events RS CLEO 1.5 [15] 54 0:49 \Sigma 0:10+0:10\Gamma 0:14

ARGUS [16] 104 0:57 \Sigma 0:15 \Sigma 0:15

E687 [17] 97 0:61 \Sigma 0:18 \Sigma 0:07 This result 367 0:54 \Sigma 0:05 \Sigma 0:04

19

