EPJmanuscriptNo. (willbeinsertedbytheeditor) Transversity Distribution and Polarized Fragmentation Function from Semi-inclusive Pion Electroproduction V.A.Korotkov1,2,W.–D.Nowak2,andK.A.Oganessyan2,3,4 1 0 1 IHEP,142284Protvino,Russia 0 2 DESYZeuthen,15738Zeuthen,Germany 2 3 INFN,LaboratoriNazionalidiFrascati,00044Frascati,Italy n 4 YerevanPhysicsInstitute,375036Yerevan,Armenia a J Received:date/Revisedversion:date 2 2 Abstract. A method is discussed to determine the hitherto unknown u-quark transversity distribution δu(x) from 2 a planned HERMES measurement of a single target-spin asymmetry in semi-inclusive pion electroproduction off a v transverselypolarizedtarget.Assumingu-quarkdominance,themeasurementyieldstheshapesofthetransversitydis- 8 tribution δu(x) and of the ratio H1⊥(1)u(z)/D1u(z), of polarized and unpolarized u-quark fragmentation functions. 6 Theunknownrelativenormalizationcanbeobtainedbyidentifyingthetransversitydistributionwiththewell-known 2 helicitydistributionatlargexandsmallQ2.Thesystematicuncertaintyofthemethodisdominatedbytheassumption 2 ofu-quarkdominance. 0 0 0 1 Introduction at some scale, they will be differentat Q2 values higher than / h that. p Deepinelasticchargedleptonscatteringoffatransverselypo- The chiral-odd nature of transversity distributions makes - larizednucleontargetisanimportanttooltofurtherstudythe p theirexperimentaldeterminationdifficult;uptonownoexper- internal spin structure of the nucleon. While a lot of experi- e imentalinformationonδq(x,Q2)isavailable.Itcannotbeac- mental data on the longitudinal spin structure of the nucleon h cessed in inclusive deep inelastic scattering (DIS) due to chi- : hasbeencollectedoverthelast10years,thestudyofitstrans- v rality conservation; it decouples from all hard processes that versespinstructureisjustabouttobegin.Onlyaverylimited i involveonlyonequarkdistribution(orfragmentation)function X number of preliminary experimentalresults is available up to (seee.g.[8]).Thisisin contrastto thecase ofthe chiral-even now: r a number density and helicity distribution functions, which are (1) measurements of the nucleon structure function g2(x) at directlyaccessibleininclusiveleptonDIS. CERN[1]andSLAC[2]-[4], (2) a first measurementof a single target-spin asymmetry for Inprinciple,transversitydistributionscanbeextractedfrom pions produced in lepton scattering off longitudinally po- cross section asymmetries in polarized processes involving a larizedprotonsatHERMES[5], transverselypolarizednucleon.Thecorrespondingasymmetry (3) afirststudyofhadronazimuthaldistributionsinDISoflep- can be expressed through a flavor sum involving products of tonsoffatransverselypolarizedtargetatSMC[6]. twochiral-oddtransversitydistributionsinthecaseofhadron- hadronscattering,whileinthecaseofsemi-inclusiveDIS(SIDIS) A quarkofa givenflavouris characterizedbythreetwist- achiral-oddquarkdistributionfunctionalwaysappearsincom- 2 parton distributions. The quark number density distribution binationwithachiral-oddquarkfragmentationfunction.These q(x,Q2)hasbeenstudiednowfordecadesandiswellknown fragmentationfunctionscaninprinciplebemeasuredine+e− for all flavors. The helicity distribution ∆q(x,Q2) was only annihilation. recently measured more accurately for u and d quarks [7] − − andisstillessentiallyunknownfors quarks.Thethirdparton ThetransversitydistributionwasfirstdiscussedbyRalston − distribution,knowngenerallyas‘transversitydistribution’and andSoper[9]indoublytransversepolarizedDrell-Yanscatter- denotedδq(x,Q2)characterizesthedistributionofthequark’s ing.Itsmeasurementisoneofthemaingoalsofthespinpro- transversespininatransverselypolarizednucleon. gramatRHIC[10].Anevaluationofthecorrespondingasym- Fornon-relativisticquarks,whereboostsandrotationscom- metry A was carried out [11] by assuming the saturation TT mute,δq(x) = ∆q(x).Sincequarksinthenucleonareknown ofSoffer’sinequality[12]forthetransversitydistribution.The toberelativistic,thedifferencebetweenbothdistributionswill maximumpossibleasymmetryatRHICenergieswasestimated providefurtherinformationontheirrelativisticnature.Thetrans- to be A = 1 2%. At smaller energies, e.g. for a possi- TT versitydistributiondoesnotmixwithgluonsunderQCDevolu- blefixed-targetha÷dron-hadronspinexperimentHERA-N[13] tion,i.e.eveniftransversityandhelicitydistributionscoincide (√s 40GeV),theasymmetryisexpectedtobehigher. ≃ 2 V.A.Korotkovetal.:TransversityDistributionandPolarizedFragmentationFunction Insemi-inclusivedeepinelasticleptonscatteringofftrans- wherethesuperscript(1)indicatesthatanoriginallyk depen- T versely polarized nucleons there exist several methods to ac- dent function was integrated over k with the weight kT2 . cess transversity distributions; all of them can in principle be T 2Mh2 Here,M isthemassoftheproducedhadronh. realized at HERMES. One of them, namely twist-3 pion pro- h To facilitate access to transversityand polarizedfragmen- duction[14], useslongitudinallypolarizedleptonsand a dou- tation functionsfrom SIDIS, single-spin asymmetriesmay be blespinasymmetryismeasured.Theothermethodsdonotre- formedthroughintegrationofthepolarizedcrosssectionover quire a polarized beam; they rely on polarimetry of the scat- P , the transverse momentum of the final hadron, with ap- teredtransverselypolarizedquark: h⊥ propriateweights.Intheparticularcaseofanunpolarizedbeam (1) measurement of the transverse polarization of Λ’s in the andatransverselypolarizedtargetthefollowingweightedasym- currentfragmentationregion[15]-[17], metryprovidesaccesstothequarktransversitydistributionvia (2) observation of a correlation between the transverse spin theCollinseffect[25]: vectorofthetargetnucleonandthenormaltothetwome- A (x,y,z) sonplane[8,18], T ≡ (3) observation of the Collins effect in quark fragmentation dφℓ d2P |Ph⊥|sin(φℓ +φℓ) dσ↑ dσ↓ throughthe measurementof pionsingle target-spinasym- h⊥ zMh s h − . (3) R R dφℓ d2P (dσ↑+dσ(cid:0)↓) (cid:1) metries[19]-[23]. h⊥ R R Here ( ) denotes target up (down) transverse polarization. TheHERMESexperiment[24]hasexcellentcapabilitiestoin- ↑ ↓ Theazimuthalanglesaredefinedinthetransversespacegiving vestigate semi-inclusive particle production. Taking the mea- the orientation of the lepton plane (φℓ) and the orientation of surementoftheCollinseffectasanexample,itwillbeshown the hadron plane (φℓ = φ φℓ) or spin vector (φℓ = φ inthefollowingthatHERMESwillbecapabletoextractboth h h − s s − transversityandchiral-oddfragmentationfunctionatthesame φℓ) withrespecttothe leptonplane.Theanglesare measured timeandwithgoodstatisticalprecision. aroundthez-axiswhichisdefinedbythemomentaqandP of thevirtualphotonandthetargetnucleon,respectively.Theraw asymmetry(3)canbeestimated[25]using 2 Single Target-Spin Asymmetry in Pion e2δq(x)H⊥(1)q(z) Electroproduction A (x,y,z)=P D q q 1 , (4) T T · nn· P e2q(x)Dq(z) q q 1 P A complete analysis of polarized SIDIS with non-zero trans- whereP isthetargetpolarizationandD =(1 y)/(1 y+ T nn versemomentumeffectsinboththequarkdistributionandfrag- y2/2)isthetransversespintransfercoefficient.Th−emagn−itude mentation functions was performed in the framework of the oftheasymmetrydependsontheunknownfunctionsδq(x)and quark-partonmodelin [21] and in the field theoreticalframe- H⊥(1)q(z). workofQCDin[22].Animportantingredientofthisanalysis 1 isthefactorizationpropertythatwasprovenfork integrated T functionsandthatcanreasonablybeassumedfork depending T 3 Transversity Distribution and Polarized functions[22]. In the situation thatthe final state polarization Fragmentation Function is not considered, two quark fragmentation functions are in- volved:Dq(z,z2k2)andH⊥q(z,z2k2).Herek istheintrin- 1 T 1 T T No experimentaldata are available on any of the transversity sic quarktransversemomentumandz isthefractionofquark distributions δq(x), while their behaviour under QCD-evolu- momentumtransferedto the hadronin the fragmentationpro- tionis theoreticallywellestablished [16]. An exampleforthe cess. The ‘polarized’ fragmentation function H1⊥q allows for leadingorderevolutionoftheprotonstructurefunctionsg (x,Q2) a correlation between the transverse polarization of the frag- = 1 e2∆q (x,Q2) and h (x,Q2) = 1 e2δq (x,Q12) is mentingquarkandthe transversemomentumofthe produced sho2wPniiniFigi. 1. It was ass1umed that h2p(Px)i cioinicides with hadron. It may be non-zero because time reversal invariance gp(x) at the scale Q2 = 0.4 GeV2 and1both functions were is notapplicablein a decayprocess,as was first discussed by 1 0 evolvedto the scale Q2 = 10 GeV2. The evolutionwas per- Collins[19]. formedusingtheprogramsfrom[27,28]forg andh ,respec- Since quark transverse momenta cannot be measured di- 1 1 tively.Theimportantconclusion,whichwasalreadydiscussed rectly,integralsoverk (withsuitable weights)aredefinedto T earlier(seee.g.[29]),followsthatwithincreasingQ2 thetwo arriveatexperimentallyaccessiblefragmentationfunctions: functionsarebecomingmoreandmoredifferentfordecreasing xwhileatlargexthedifferenceremainsquitesmall. z2 d2k Dq(z,z2k2) Dq(z) (1) Results fromtwo independentmeasurementsindicate that Z T 1 T ≡ 1 the polarized fragmentation function H⊥(1)q(z) may be non- 1 isthefamiliarunpolarizedfragmentationfunction,normalized zero:i)azimuthalcorrelationsmeasuredbetweenparticlespro- by the momentumsum rule dzzDq→h(z) = 1. Corre- ducedfromoppositejetsinZdecayatDELPHI[30]andii)the h 1 spondingly,thepolarizedfragPmenRtationfunctionisobtainedas single target-spinasymmetry measured for pions producedin SIDISofleptonsoffalongitudinallypolarizedtargetatHER- k2 MES[5].Theapproachof[25]isadoptedtoestimatethepos- z2Z d2kT (cid:18)2MTh2(cid:19) H1⊥q(z,z2kT2)≡ H1⊥(1)q(z), (2) siblevalueofH1⊥(1)q(z).Collins[19]suggestedthefollowing V.A.Korotkovetal.:TransversityDistributionandPolarizedFragmentationFunction 3 4 Projected Statistical Accuracy and 22px,Q), g(x,Q)10.715 ghgppp111,,, hQQp122,== Q11002= GG0.ee4VV G22eV2 SAyfusllteanmalaystiisctsoextracttransversityandpolarizedfragmenta- ph(1 tionfunctionsthrough(4)requiresonetotakeintoaccountall 0.5 quark flavours contributing to the measured asymmetry. Ac- cording to calculations with the HERMES Monte Carlo pro- gramHMC,thefractionofpositivepionsoriginatingfromthe 0.25 fragmentation of a struck u-quark ranges, depending on the valueofx,between70and90%foraprotontargetandisonly slightlysmallerforadeuterontarget.Therefore,inafirstanal- 0 -2 -1 ysis,theassumptionofu-quarkdominanceintheπ+ produc- 10 10 1 x tion cross-section appears to be reasonable.This is supported Fig. 1. The transversity distribution hp(x,Q2) (continuous line) bythesumruleforT-oddfragmentationfunctionsrecentlyde- 1 0 whichcoincideswiththehelicitydistributiongp(x,Q2)at thescale rivedin[33].Theseauthorsconcludedthatcontributionsfrom 1 0 Q2 = 0.4 GeV2 (asgiven by theGRSV LOparameterization [26] non-leading parton fragmentation, like d π+, is severely in0the’standard’scenario).TheirevolvedLOdistributionshp(x,Q2) suppressedforallT-oddfragmentationfunc→tions. 1 (dotted)andg1p(x,Q2)(dot-dashed)areshownatQ2 =10GeV2. Consequently, the assumption of u-quark dominance was usedtocalculateprojectionsforthestatisticalaccuracyinmea- suringtheasymmetryAπ+(x).Theexpectedstatisticsforscat- parameterizationfortheanalyzingpowerintransverselypolar- T teringatHERMESunpolarizedleptonsoffatransverselypolar- izedquarkfragmentation: izedtarget(protonordeuteronoptionsareunderconsideration) k H⊥q(z,z2k2) M k willconsistofaboutsevenmillionsreconstructedDISevents. AC(z,kT)≡ |MT|D1q(z,z2k2T) = M2C+| Tk2| , (5) ThestandarddefinitionofaDISeventatHERMESisgivenby h 1 T C | T| thefollowingsetofkinematiccuts1: with M 0.3 1.0 GeV being a typical hadronic mass. C Choosinga≃Gaussi÷anparameterizationforthequarktransverse Q2 >1GeV2,W >2GeV,0.02 < x < 0.7,y < 0.85. momentumdependenceintheunpolarizedfragmentationfunc- tion AnadditionalcutW2 >10GeV2wasintroducedintheanaly- R2 sistoimprovetheseparationofthestruckquarkfragmentation D1q(z,z2kT2)=D1q(z)πz2 exp(−R2kT2), (6) region.AnaveragetargetpolarizationofPT =75%isusedfor theanalysis. leadsto Considering only u-quarks the expression for the asymmetry H⊥(1)q(z)= (4)reducestothesimpleform 1 M ∞ exp( x) D1q(z)2MCh (cid:18)1−MC2R2Z0 dxx+M−C2R2(cid:19). (7) AT(x,y,z)=PT ·Dnn· δuu((xx)) · H1D⊥(u1()uz()z) (10) 1 Here R2 = z2/b2, and b2 is the mean-squaremomentumthe hadronacquiresinthequarkfragmentationprocess.Inthefol- foraprotontarget,and lowingtheparametersettingsM = 0.7GeV andb2 = 0.25 C GeV2 are used because they are consistent [31] with the sin- A (x,y,z)= T gletarget-spinasymmetrymeasuredatHERMES[5].Theyare also compatible with the analysis of [30], as can be seen by (1 3ω ) P D δu(x)+δd(x) H1⊥(1)u(z) (11) evaluatingtheratio − 2 D · T · nn· u(x)+d(x) · Du(z) 1 1 dzH⊥(z) foradeuterontarget.Hereω =0.05 0.01istheprobability R(z )= zmin 1 , (8) D ± min R 1 ofthedeuterontobeintheD-state. dzD (z) zmin 1 TosimulateameasurementofAT theapproximationδq(x) = R ∆q(x) could be used in view of the relatively low Q2-values where H⊥(z), in contrast to Eq.(2), is the unweighted polar- 1 at HERMES, in accordance with the above discussion. The izedfragmentationfunctionusedin[30] Gehrmann-Stirlingparameterizationinleadingorder[34]was takenfor∆q(x) and the GRV94LO parameterization[35] for z2Z d2kT H1⊥q(z,z2kT2) ≡ H1⊥q(z). (9) q(x).TheQ2evolutionofthequarkdistributionswasneglected and Q2 = 2.5 GeV2 was taken as an average value for the The BKK parameterization [32] was used to estimate the in- tegralovertheunpolarizedfragmentationfunctionD1(z).The 1 Q2 andν arethephoton’svirtualityandlaboratoryenergy,x = values obtained for the ratio, R(0.1) = 0.048 and R(0.2) = Q2/2Mν istheBjorkenscalingvariable,y = ν/E isthefractional 0.070,aretobecomparedtotheexperimentalresult[30]:0.063 photonenergyandW isthec.m.energyofthephoton-nucleon sys- ± 0.017. tem;E =27.5GeVatHERMES. 4 V.A.Korotkovetal.:TransversityDistributionandPolarizedFragmentationFunction HERMESkinematicalregion.TheHERMESMonteCarlopro- thefunctionK(x ,z ),asopposedto9unknownfunctionval- i j gram HMC was used to account for the spectrometer accep- ues:4valuesforδu(x )and5valuesforH⊥(1)u(z )/Du(z ), tance. The following cuts were applied to the kinematic vari- wheretheindicesiandi j enumeratetheexp1erimentjalint1ervajls ablesofthepion2: inxandinz,respectively.Thestandardprocedureofχ2 min- x >0,z >0.1,P >0.05GeV. imizationwasappliedtoreconstructthevaluesforbothδu(x) F h⊥ andH⊥(1)u(z)/Du(z)andtoevaluatetheirprojectedstatisti- Thesimulateddataweredividedinto5x5binsin(x,z).The 1 1 expectationsfortheasymmetryAπ+(x)aswouldbemeasured cal accuracies expected for a real measurementat HERMES. T TheresultsareshowninFigs.3a,b,respectively. byHERMESusingaprotontarget,arepresentedinFig.2ain Inananalogouswaytheconsiderationofthedeuteronasym- differentintervalsofthepionvariablez.Theprojectedaccura- metry(11)allowstheevaluationoftheprojectedstatisticalac- ciesfortheasymmetrywereestimatedaccordingto curaciesforameasurementofthefunctionsδu(x)+δd(x)and H⊥(1)u(z)/Du(z)(seeFigs.4a,b,respectively).Theprojected 1 1 statisticalaccuracyisconsiderablyworsethanthatforthepro- π+AT0.18 00..12e<<pzz<< →00..23 eπ+X K(x, z) 56 ep → eπ+X toofnthtaeragseyt;mthmisetirsyc(a1u1s)e,dwmhiacihnliynbtuyrtnheisedxupeectotetdhesmloawlleerrvvaalluuee 0.3<z<0.5 4 of(δu(x)+δd(x))/(u(x)+d(x))comparedtoδu(x)/u(x). 0.6 0.5<z<0.7 0.7<z<1.0 3 0.4 2 0.2 x) 3 z)2.5 0 -1 a) 01 b) -1 δu(2.4 ep → eπ+Xa) uz)/D(1 2 ep → eπ+Xb) 10 x 10 x 1.8 u(1.5 1) Fig.2.Protontarget.a)TheweightedasymmetryATπ+(x)indifferent 1.2 ⊥(H1 1 intervalsofz;b)thefunctionK(x,z). 0.6 0.5 0 0 1 10-1 x 0 0.2 0.4 0.6 0.8 z1 δA = Ph⊥ sin(φℓ +φℓ) 2 2 1 , (12) T (cid:28)(cid:16)zmπ s h (cid:17) (cid:29) · √Nπ Fig.3.a)Thetransversitydistributionδu(x),andb)theratioofthe fragmentationfunctionsH⊥(1)u(z)andDu(z)aswouldbemeasured whereN isthetotalnumberofmeasuredpositivepionsafter 1 1 π byHERMESwithaprotontarget.Theasteriskina)showsthenormal- kinematiccuts,and< >meansaveragingoverallaccepted izationpoint. ··· events.In thisway the productofthe transversitydistribution andtheratioofthefragmentationfunctions, x) 3 z)2.5 K(x,z)=δu(x)· H1D⊥(1u1()uz()z), (13) δx) + d(2.25 ed → eπ+Xa) uu(z)/D(11.25 ed → eπ+Xb) aswellastheprojectedstatisticalaccuracyforameasurement δu(1.5 ⊥(1)H1 1 ofthisfunctionwerecalculatedandareshowninFig.2b,again 1 forthecaseofaprotontarget. 0.5 0.5 Thefactorizedformofexpression(10)with respectto the svhaaripaebfleosrtxheantwdozuanllkonwoswthnefusnimctuioltnasnδeouu(xs)reacnodnHstr⊥u(c1t)iuo(nzo)f/tDheu(z), 0 10-1 x 00 0.2 0.4 0.6 0.8 z1 1 1 whiletherelativenormalizationcannotbefixedwithoutafur- Fig. 4. a) The transversity distribution δu(x) + δd(x), and b) the therassumption.Aswasdiscussedabove,thetransversitydis- ratioofthefragmentationfunctionsH⊥(1)u(z)andDu(z)aswould tributionδq(x)conceivablycoincideswiththehelicitydistribu- bemeasured by HERMESwithadeut1eron target. The1asteriskina) tion∆q(x)atsmallvaluesofQ2 wheretherelativisticeffects showsthenormalizationpoint. are expected to be small. According to Fig. 1 the differences aresmallestintheregionofintermediateandlargevaluesofx. Hencetheassumption Two sources of systematic uncertainties arising from ap- proximationsusedintheanalysiswereinvestigated.Toevalu- δq(x0)=∆q(x0) (14) ate the contributionof the normalizationassumption(14),the relativedifferencebetweentransversitydistributionδu(x,Q2) atx0 =0.25wasmadetoresolvethenormalizationambiguity. andhelicity distribution∆u(x,Q2) was studied as a function The experimental data then consist of 25 measured values of ofxandQ2intheHERMESkinematics.Startingfromδu(x)= 2 xF = 2pL/W, where pL is thelongitudinal momentum of the ∆u(x)atthescaleQ20 = 0.4GeV2bothfunctionswereevolved hadronwithrespecttothevirtualphotoninthephoton-nucleonc.m.s., to higher values of Q2. The results are shown in Fig. 5 and andz=Eh/ν,whereEhistheenergyoftheproducedhadron. allow the conclusion that the relative difference is small for V.A.Korotkovetal.:TransversityDistributionandPolarizedFragmentationFunction 5 x above 0.2 0.3; the corresponding systematic uncertainty Acknowledgements ÷ is on the level of 2 5%. The same conclusion is valid for ÷ WethankKlausRithfordiscussionswhichledtothedevelop- the evolution of δu(x) + δd(x). A larger contribution to the mentofthemethoddescribedinthispaperandtheHERMES CollaborationforthekindpermissiontousetheirMonteCarlo Program.WeareindebtedtoRalfKaiserforthecarefulreading 2Q)0.1 ofthemanuscript. x, u( ∆ 2Q))/ 0 References x, 2∆Q) - u(-0.1 Q21, .G0eV2 12.. SEM15C4CCoollllaabboorraattiioonn,,DK..AAdbaemestaelt.a,lP.,hPyhs.yLs.eRtte.vB.4D0546,,3573730(1(919979)7.). x, 3. E143Collaboration,K.Abeetal.,Phys.Rev.D58,112003(1998). δ(u( 2.0 4. P.Bosted,VeryPreliminaryResultsfortheSpinStructureFunction 3.0 -0.2 5.0 g2 fromSLACE155x,Proc.of15thInt.Conf.onPart.andNuclei (PANIC99),Uppsala,Sweden,June1999. 10. 5. HERMESCollaboration,A.Airapetianetal.,Phys.Rev.Lett.84, -0.3 4047(2000). -2 -1 10 10 x 1 6. A.Bravar,Nucl.Phys.Proc.Suppl.79,520(1999). 7. HERMES Collaboration, K. Ackerstaff et al., Phys. Lett. B464, Fig.5.Relativedifferencebetweentransversitydistributionδu(x,Q2) 123(1999). andhelicitydistribution∆u(x,Q2)asafunctionofxandQ2inthe 8. R.L. Jaffe, Proc. of the 2nd Topical Workshop Deep Inelastic kinematicalregionaccessibletotheHERMESexperiment(<Q2 >≃ ScatteringoffPolarizedTargets:TheoryMeetsExperiment,DESY 2.5GeV2). 97-200, Zeuthen, 1997, p.167, ed. by J. Blu¨mlein, A. de Roeck, T.Gehrmann,andW.-D.Nowak;hep-ph/9710465. 9. J.Ralston,D.E.Soper,Nucl.Phys.B152,109(1979). 10. D.Hilletal.,RHICSpinCollaborationboration:LetterofIntent, systematic uncertainty originates from the above mentioned BNL,May1991. ‘contamination’ of other quark flavors than u to π+ produc- 11. O.Martinetal.,Phys.Rev.D60,117502(1999). tion, when assuming u-quark dominance in the analysis. The 12. J.Soffer,Phys.Rev.Lett.74,1292(1995). xandz dependenceofthiscontaminationwasevaluatedwith 13. V.A. Korotkov and W.-D. Nowak, Proc. of the Workshop Po- HMC.Bothcontributionswereaddedlinearly;theresultingto- larized Protons at High Energies - Accelerator Challenges and tal projected systematic uncertainties on the extraction of the Physics Opportunities, DESY-PROC-1999-03, 1999, p.19, ed. by transversitydistributionδu(x) andthefragmentationfunction A.de Roeck, D. Barber, and G. Ra¨del; DESY 99-122; hep- ratioH⊥(1)u(z)/Du(z),aswouldbemeasuredusingaproton ph/9908490. 1 1 14. R.L.Jaffe,X.J.Ji,Phys.Rev.Lett.71,2547(1993). target,areshownashatchedbandsinFigs.3a,b,asafunctionof 15. F.Baldracchinietal.,FortschrittederPhys.29,505(1981). xandz,respectively.Thesameprocedureforadeuterontarget 16. X.Artru,M.Mekhfi,Zeit.Phys.C45,669(1990). yieldsprojectedsystematicuncertaintiesforδu(x)+δd(x)and 17. R.L.Jaffe,Phys.Rev.D54,6581(1996). H⊥(1)u(z)/Du(z), as shown as hatched bands in Figs. 4a,b, 18. R.L.Jaffe,X.Jin,J.Tang,Phys.Rev.Lett.80,1166(1998). 1 1 respectively. 19. J.C.Collins,Nucl.Phys.B396,161(1993). 20. J.C.Collins,S.F.Heppelmann,G.A.Ladinsky,Nucl.Phys.B420, 565(1994). 21. A.Kotzinian,Nucl.Phys.B441,234(1995). 22. P.J.Mulders,R.D.Tangerman,Nucl.Phys.B461,197(1996); 5 Conclusions Erratum-ibid.B484,538(1997). 23. M. Anselmino, E. Leader, F. Murgia, Phys. Rev. D56, 6021 (1997). 24. K.Ackerstaffetal.,Nucl.Instr.andMethodsA417,230(1998). In conclusion, the HERMES experiment using a transversely 25. A.M.Kotzinian,P.J.Mulders,Phys.Lett.B406,373(1997). polarizedproton targetwill be capable to measure simultane- 26. M.Glu¨cketal.,Phys.Rev.D53,4775(1996). ously and with good statistical precision the shapes of the u- 27. M.Hirai,S.Kumano,M.Miyama,Comp.Phys.Comm.108,38 (1998). quark transversity distribution δu(x) and of the ratio of the 28. M.Hirai,S.Kumano,M.Miyama,Comp.Phys.Comm.111,150 fragmentationfunctionsH⊥(1)u(z)/Du(z).Thenormalization 1 1 (1998). can be fixed under the assumption that in the HERMES Q2 29. S.Scopetta,V.Vento,Phys.Lett.B424,25(1998). rangethetransversitydistributioniswelldescribedbythehe- 30. A.V.Efremov,O.G.Smirnova,L.G.Tkachev, Nucl.Phys.Proc. licitydistributionatlargex.Usingadeuterontarget,informa- Suppl.74,49(1999). tiononδu(x)+δd(x)willbeavailable,butwithconsiderably 31. A.M.Kotzinianetal.,hep-ph/9908466. less statistical accuracycomparedto a measurementofδu(x) 32. J. Binnewies, B.A. Kniehl, G. Kramer, Zeit. Phys. C65, 471 (1995). fromaprotontarget.Thesystematicuncertaintyofthemethod 33. A.Scha¨fer,O.Teryaev,Phys.Rev.D61,077903(2000). proposed in this paper is dominated by the assumption of u- 34. T.Gehrmann,W.J.Stirling,Phys.Rev.D53,6100(1996). quarkdominance. 35. M.Glu¨ck,E.Reya,A.Vogt,Zeit.Phys.C67,433(1995).