198 4ApJ. . .276. .491H AST 78-27879. is operatedbytheUSNational ScienceFoundationundercontract elliptical ,NGC5128=CenA.Thisaclassical, accepted asyetanotherpeculiarityofthenearestgiant unusual, oftenstriking,characteristics.Amongtheseare:(1) a strong, double-lobedradiosource,isalsonotedforahost of by Graham(1979)ledhimtoreviveBaadeandMinkowski’s wavelengths, andwhichprovidesevidenceforveryrecent thick dustbandandassociatedevidenceforvigorous star formation toatleast20(D/5)kpcfromitsnucleus; and formation; (2)a“jet”thatappearsatradio,optical,andX-ray The AstrophysicalJournal,276:491-508,1984January15 @ 1984.TheAmericanAstronomicalSociety.Allrightsreserved.PrintedinU.S.A. (1954) suggestionthatNGC5128representstwogalaxies in (3) detectabilityaty-rayenergies.Radialvelocityobservations 2 1 3 PresentlyaVisitorattheDominion AstrophysicalObservatory. For manyyearsthelackofaglobularstarclustersystemwas VisitingAstronomer,CerroTololo Inter-AmericanObservatory,which ContributionsoftheUniversityWaterloo Observatory,No.90. 12 -1 mass of^1.6x10(D/5)Mwithin^36(Z)/5)kpc.Theimpliedvaluethemass-to-lightratio,M/L, is ^16(5/D)insolarunits.Analysisofvelocitiescompaniongalaxieswithin350(D/5)kpcsuggeststhat luminosity andsizeto,orbrighterthan,coCen,themostmassiveluminousglobularclusterinGalaxy. (3) Inprojecteddistancetheconfirmedclustersliebetween(2.9and35.6)(D/5)kpcfromnucleus.Thatis,they to +860kms.In1982threeadditionalclusterswerefoundbyspectroscopicobservationsofasample at theCTIO4mtelescopein1981,12areclusterswithheliocentricradialvelocitiesrange+340 velocities, andtheluminosityofSN1972einNGC5253. orientation oftheclustersystemwithitsmajoraxisalignedalongisophotes visually selectedcandidatesandofthespectroscopicallyconfirmedclustershintsatapossiblepreferential , scaledtoatotalpopulationof^600,suggestseitherthatthebrightestNGC5128clustersare effects. (4)ApplicationofaprojectedmassestimatortoourvelocitydatasuggeststhatNGC5128has globular clustersassociatedwithNGC5128.Fromthemwefindthefollowing:(1)Thehave these resultswithobservationsobtainedin1980bringsto20thenumberofspectroscopicallyconfirmed observed NGC5128globularclusterluminosityfunctionwiththatoftheclustersLocalGroup of threenewclustersinacompletesamplephotometricallyselectedcandidates.(6)Comparisonthe counts supporttheexistenceofasubstantialclustersystemæ600members,asdoesspectroscopicdiscovery of themostdistantstarclustersthathavesofarbeenidentifiedaroundNGC5128,itself.(5)NewStar there ismassintheNGC5128halo(orclusterofgalaxies)atorbeyondradius lie primarilyoutsidethemainbodyofopticalgalaxy,atendencylikelytohaveresultedfromselection smaller galactocentricdistancesthandothebluerclusters.(2)Thebrightestclustersareprobablysimilarin Subject headings:clusters:globular—galaxies:individualstellarcontent outer spheroidofthegalaxy.Othertopicsdiscussedincludeselectioneffects,propertiesradial 0v 17.0 K } CLUSTERS x X CANDIDATES (?) Xx x % XXxXx V x YX xxX • ^X Xo* V Xx A x Æ x X x v XX V MAGNITUDE *7 OL n 1r NGC 5128 •o ••X A XX V-15-5 Vol. 276 _L 198 4ApJ. . .276. .491H -1 , o _ 1 No. Z1984NGC5128GLOBULARCLUSTERSYSTEM493 0.67 Q a-1 23 20 22 26 27 14 17 18 24 12 11 10 16 7 6 4 5 Notes.—(1)GP’sanalysisoftheirimage-tubespectrumgaveV=568kms.OurPaperI 2 r>0 Object ID 207 501 Internal 123 82-107 gg 82-95 21 15 82-105 78 o 7 5 6 4 8 2 1981-1982 SpectroscopicObservationsofNGC5128Clusters 1981 Apr5 1981 Apr3 1981 Apr3 1981 Apr3 1982 May21 1981 Apr4 1981 Apr3 1981 Apr4 1981 Apr5 1981 Apr4 1981 Apr4 1982 May21 1981 Apr5 1981 Apr4 1981 Apr4 1981 Apr5 1981 Apr3 1982 May22 1981 Apr4 Date (UT) (min.) Exp. 20 45 60 25 60 60 60 63 60 60 55 90 30 90 90 90 30 90 38 TABLE 1 6 -1 1 cannot bedistinguishedfromstarsbyvisualselectionwith predicted bythestarcounts(PaperIand§VI),mostofwhich experiment afterinspectionofour clusters andcandidatesononeofJohn presently availableplatematerial.Theradialvelocitiesofthe barely nonstellarobjectswouldbe easier ontheHaemulsions.Thereason stars andgalaxies,werescanned(asdescribedin§II)with the remainder ofthe1982candidatesrangedfrom—110to+200 Graham’s excellentseeingIIIa-Jplates suggestedthatvisualidentificationof to befieldstarsandexcludethemfromFigure3. radial velocityandspectralclassificationcriteriawejudgethem for thismaybethatthehighercontrast oftheIliaemulsionsmaskssubtle images ofGraham’s(1981)photoelectricsequencestars.Since DAO PDSmicrodensitometerinlate1981April.Threeplates NE regioninbothBandVwiththePickering(1891)-Racine NGC 5128.Unfortunatelywewereabletocompleteonly the intensity gradients. (P5236, P5237,andP5254)werecalibratedusingthesecondary Table 2),whichwasusedwheneveritavailable.Visual km s[(Vo}=16±32(m.e.)].Onthebasisof include firstplatesfromanewsurveyoftheouterregions inspection oftheplateslistedinTable2iscontinuing, and a numberofpromisingnewcandidateshavebeenidentified. (1969)prism (indicatedasP-Rinthecolumnheaded“Plate” in rf (km s)^Lines 6 683 The visuallyselectedclustercandidates,aswellpresumed 859 688 521 782 WechoseIla-OandIla-DratherthanIIIa-JIIIa-Fplatesforthis 801 638 710 445 796 523 The available4mprime-focusplatesdescribedinTable2 641 495 345 315 644 492 771 309:: K,o_ a_No. m 49 26 21 41 54 43 24 31 51 14 18 32 19 46 45 25 55 50 18 b) PhotometricObservations 10 11 12 10 9 9 7 7 8 7 5 9 8 7 6 9 8 3 Notes 198 4ApJ. . .276. .491H _1 No. 2,1984NGC5128GLOBULARCLUSTERSYSTEM495 we obtainedmagnitudesbyintegratingtheintensityimages, we eliminated(tofirstorder,atleast)theeffectsofdiffering structure, ifany,betweentheprimaryandsecondaryimages made brighterby0.05mag.ForplateslackingP-Rimages, that mayaffectirisphotometry(Racine1969;Blanco1982). was adoptedfortheNGC5128plates.Subsequentanalysis A Amof7.00magbetweenprimaryandsecondaryimages electrically measuredstarsonœCenplates(whichweretaken expressly tocalibratetheprismoneachofsametwonights this casethemagnitudesadoptedhereinwouldneedtobe perhaps Am=6.95magismarginallymoreappropriate.In of CannonandStewart’s(1981)largersamplephoto- column thenumberofplatepairsusedis2,thenmagnitudes column headingsarelargelyself-explanatory.(Ifintheseventh magnitudes weredeterminedfromplatesP1015,P5236,and both thephotoelectricsequencestarsandP-Rimages.Visual we usedsecondarystandardscalibratedonplatescontaining used forprimefocusobservingofNGC5128),suggeststhat from platesP5237andP5254werenotavailableforinclusion 0.06 mag.Wehavenotmadeanexhaustivestudyofpossible in themeans.Notealsothatvelocitiesareroundedto the are <0.1mag. candidates hasameaninternalerrorforoneobservationof P5237, andbluemagnitudesweredeterminedfromP1013, nearest 10kms).Thelastsevenobjectsinthetable are P1014, andP5254.Thephotometryforthemultiplyobserved candidates webelievetohaveahighprobabilityofbeing clusters, butforwhichnoradialvelocitiesareyetavailable. systematic errorsinthephotometry,butweexpectthatthey properties oftheNGC5128globularclustersystem.Thecon- clusions belowaresomewhattentative,however,for two globular clusterswecannowattempttoinfersomeglobal reasons. First,small-number statisticalfluctuationsare processes arebiasedagainst certain kindsofclusters.These selection effectsareprobablyquite importantinsomeareasand important inasampleofonly 27objects.Second,theselection The resultsoftheseeffortsaregiveninTable3,wherethe With oursampleof20confirmedandsevensuspected © American Astronomical Society • Provided by the NASA Astrophysics Data System IV. PROPERTIESOFTHECLUSTERS P1013 P1015 P1014 P1031 P1016 P2110 P2109 P5238 +P-R P5237 +P-R P5236 +P-R P5253 +P-R P5251 +P-R P5256 P5255 P5254 +P-R 3 vdB:vandenBergh;RJD:Dufour;JEH:Hesser. Plate Journal ofAvailable4MeterTelescopePrimeFocusPlates 1975 Apr12 1975 Apr12 1975 Apr12 1975 Apr14 1975 Apr12 1981 Apr1 1981 Apr1 1981 Apr1 1981 Apr2 1981 Apr2 1981 Apr2 1976 Jun21 1976 Jun20 1981 Apr2 1981 Apr2 Date 3 Observer RJD RJD JEH JEH JEH JEH JEH JEH JEH JEH vdB vdB vdB vdB vdB TABLE 2 necessary tokeepinmindhoweachsampleofclusterswas complicate interpretationofourresults.Consequently,itis chosen. Inthefollowingparagraphswethereforesometimes clusters selectedfromphotometryinthesmallregionnear their extendedimages(sampleA)fromthesubsampleofthree separate thesubsampleof17clustersidentifiedonbasis center ofNGC5128(sampleB).Werefertothe20confirmed rather thanthe5Mpcusuallyquoted,althoughwealso and sevenprobableclustersasthe“combinedsample.” comparison withtheNGC5128literature. parameterize ourresultsintermsofD/5Mpctofacilitate the analysesof§VI,wefrequentlyadoptavalue3Mpc require avalueofthedistancetoNGC5128.Anticipating Furthermore, thediscussionsofthissectionwilloccasionally (min.) Fig. 4),arangethatextendsbeyondtheredlimitobservedfor sample. Thesymbolsusedare:X’s indicate sampleAclusters;-l-’sarethe sample Bclusters;andtheopencircles representvisuallyselectedcluster candidates. Exp. 45 45 25 15 12 12 15 15 15 15 10 15 10 10 10 A widerangeofcolorsisobservedfortheclusters(see Fig. 4.—Acolor-magnitudediagram forthecombinedNGC5128cluster Emulsion V 17 I 9 103a-O 18 103a-D 103a-O 103a-D 103a-O 103a-O IIIa-J Ila-D Ila-D Ila-D Ila-O Ila-D a) ColorsandMagnitudesoftheClusters Ila-O Ila-O Ila-O 0 GG 495 GG 385 GG 385 GG 495 GG 385 UG 2 UG 2 GG 495 GG 495 GG 495 GG 385 GG 495 GG 385 GG 385 GG 385 Filter B-V , Comments 20'N 20'W 20'N 20'E 20'S 20'W 20S 20'E 20'S 20'E 20'N 20'W 20'N 20'E 198 4ApJ. . .276. .491H 496 HESSERETAL.Vol.276 mag forNGC5128(vandenBergh1976),clusters11and26 have creptintothephotographicphotometryforthese and severalcandidateshave(B—V)>0.9mag.Ifnoerrors Galacticglobulars [(B—V)^0.9mag].WithE(B—V)=0.10 objects, theymusteitherbereddenedbylocalabsorptionor intrinsically veryred.Cluster11mayliebehindpartofadust lane, butcluster26certainlydoesnot. The symbolsareasinFig.4.This figuresuggeststhatclusterslyingat metal-rich thanthoseatlargedistances. smaller projecteddistancesareonaverage redderand,hence,presumablymore 0 0 Figure 5revealsatendencyforthebluest(andpresumably Fig. 5.—(B-F)pvs.projecteddistance (arcmin)forthecombinedsample. g > CD t i © American Astronomical Society • Provided by the NASA Astrophysics Data System 0-6 0-8 1-2 0 81624 23 . 20 . 17 . 18 . 22 26 14 . 27 24 21 11 25 12 10 16 15 13 19 7 . 6 . 5 . 4 . 3 . 1 . 2 d(arcmin) -•» Object Values AdoptedforPosition,Photometry,andVelocity arcmin kpc(D/5) 21.5 15.9 24.5 10.1 19.6 14.9 12.8 10.0 11.9 10.6 14.4 4.9 2.0 8.1 4.7 8.2 9.5 7.2 5.3 7.2 4.8 5.8 4.9 6.5 7.2 7.3 6.9 Distance 23.1 31.3 11.9 11.8 10.5 28.5 21.7 10.5 13.8 14.7 35.6 20.9 14.5 18.6 10.5 15.4 10.5 10.0 17.2 2.9 7.1 7.7 6.8 7.1 8.4 7.0 9.5 TABLE 3 (mag.) 17.03 17.09 17.60 17.25 17.60 17.70 17.70 17.47 18.09 17.96 18.06 17.90 18.31 17.98 <^> 18.12 18.56 18.65 18.50 18.73 18.59 17.72 18.68 18.62 18.81 18.56 18.11 19.16 -1 -1 the dustlane,andiftheyareomitted,tendencyremains. most metal-poor)clusterstolieatgreaterradialdistances clusters withB—V>0.93is11.5±1.5(D/5)kpc.Thecorrela- The meanprojecteddistanceforthe14clusterswith The meanB—Vcolorforall27probableclustersis0.93mag. two likelycandidates(clusters9and21)areseenprojectedon from NGC5128.Onlyoneconfirmedcluster(cluster11)and tion ofcolorwithpositionisonlymarginallysignificant(atthe B—V <0.93is17.0±2.5(D/5)kpc,whilethatforthe13 tion (—0.007magkpcforadistanceof3Mpc)agreeswith 70 %confidencelevel),butthesenseandslopeofcorrela- Accounting forprojectioneffectsinthepresentNGC5128 the slopeofcolorversusdistancerelationfoundinour sample willchangetheobservedgradient,butitstill will Galaxy (—0.008magkpc;HarrisandRacine1979). agree withthatfoundinourGalaxytowellwithintheerrors. Nevertheless theabsenceofpopulousblueclusters the sequent spectroscopicobservationsofsampleAclusters. introduced inthevisualselectionprocedureor sub- Finally, wenotethatnocolorbiaswas,toourknowledge, candidates tendtolieawayfromthedustbandwheresuch to ascertainspectroscopicallyifanyofthedozenobjects objects presumablywouldbefound.Itmightinteresting Magellanic Cloudtypeinoursampleisnotsurprising,as our mentioned byvandenBergh(1979)tolienearthedustband are populousblueclusters. This effectisillustratedinFigure 6,whichshowsthevisual fainter clusterstolieatgreater radialdistancesfromthegalaxy. Inspection ofTable3revealsa tendencyinoursampleforthe (mag.) 17.97 18.40 18.37 18.34 18.07 18.72 18.50 18.40 18.82 18.81 18.84 19.02 19.04 19.28 19.10 19.36 19.40 19.59 19.50 19.48 18.81 19.80 19.77 19.76 19.72 19.72 19.07 - (mag.) 0.74 0.88 0.93 0.80 0.85 0.80 0.73 0.78 0.79 1.12 1.04 0.84 0.71 0.98 1.02 0.86 0.91 1.12 0.91 0.96 0.64 1.06 1.16 1.09 1.16 1.09 1.14 N 1 (kms") 600 780 690 690 860 510 710 590 520 660 800 440 800 640 490 640 340 770 500 198 4ApJ. . .276. .491H -2 7 No. 2,1984NGC5128GLOBULARCLUSTERSYSTEM497 of projectedradialdistance(inarcmin)fromthecentergalaxy.Thesymbolsusedasfollows:X’sindicatesampleAclusters; +’sarethe sample Bclusters;opencirclesrepresentvisuallyselectedclustercandidates;andthedotsgivevandenBergh’s(1976)surfacebrightness measuresfor distance inarcminutesfromNGC5128.The13brightest plates andavaluearbitrarilyreducedbyfactorof10torepresenttheouterpartsimages.Atendencyformoredistantsample Aclustersand NGC 5128,itself,fromhisTable1.Finally,twosquares,arbitrarilyplacedat6',representthetypicalcentralsurfacebrightnessofcluster imagesonour beyond aradiusof3'-4'thebackgroundlightisnolonger could accountforadeficiencyoffaintclusterswithdistances extended clustersagainstthebackgroundlightofgalaxy clusters haveameanprojecteddistanceof12.3±2.1(D/5)kpc magnitudes forthecombinedsampleplottedagainstprojected suspects tobefainterisapparent. the photographicphotometryaremuchtoosmalltoaccount compared to16.9±2.3(D/5)kpcforthe13faintest.Errorsin very deeplyexposed,havingbackgroundspeculardensitiesof, less thanæ3'to4'or5(D/5)kpc.However,webelievethat for thedifference.Abiasagainstvisuallyselectingfaint typically, only0.7aboveclearplateatthatradius;and(2) a seriousproblemfortworeasons:(1)ourplatesarenot the clusterimageshaveacentralsurfacebrightness,typically, bias whichmaybeimportanttoexplainwhywefindfainter brightness ofNGC5128atthatradius(cf.Fig.6).Another factor of10abovethe21.4magarcsecbackgroundsurface extended clustersarerareintheinnerregionsofgalaxy clusters atlargerdistancesarisesfromthedifficultyof plate, thisdifferenceisinsufficientto affectourconclusions. identifying faint,compactclusters.Becausewefind that clusters inoursamplenearerthecenterofNGC5128. radii andanassumeddistance, convolvedthemwiththePSF, clusters usingKing(1962)profiles withvariouscoreandtidal plates, weareabletoestimatetheirsizes.Thiswasdone by (see §IVh),weareleftwithprimarilybright(andcompact) converting thephotographicdensityintointensityand then stars ontheplatewithhighestresolution(P5236). The measuring thepointspreadfunction(PSF)foranumber of FWHM onthisplateisT.'O.Wethenconstructedmodels of 7 Fig. 6.—TheVmagnitudeforclustersandthesurfacebrightness,oy,underlyingNGC5128light(vandenBergh1976)areplotted asafunction AlthoughthemeasuredPSFisslightly broaderneartheedgesof Since theclustersinsampleAareresolvedonourbest © American Astronomical Society • Provided by the NASA Astrophysics Data System b) SizesoftheClusters 8 plotted inFigure7for16ofthesampleAclusters.Cluster6 procedure isstilladequatetodetermineapproximatesizes.Wenoteherethat, cluster, andcluster1isnoticeablyelongated(seeFig.2).However,our the mostflattenedglobular.Perhapsnoticeableflatteningofcluster1 in theGalaxy,œCentauriisbothmostluminous(M=—10.2)and also pointstohighintrinsicluminosity.(WithV=17.6,A0.3[vanden core andtidalradius,wewereunabletodetermineeach acceptable fitcouldbeobtainedforseveralcombinationsof and fittedthemtotheobservedclusterprofiles.Sincean Bergh 1976]andD=5Mpccluster1wouldhaveM—11.2.) by Galacticstandardsusuallyproducedanadequatefitwithin individually. Almostanyratiooftidal/coreradiusreasonable a ratioofr/r=33,typicalforGalacticglobulars. the uncertaintyinobservedprofile.Therefore,weassumed to aKingmodelwithr=5(D/5)pc andr=170(D/5)pc.Thedashedline v represents theadoptedreferencemodel. v v tc c t 8 Strictlyspeaking,aKingmodelisnotappropriatefornonspherical The sizesderivedrelativetotheabovereferencemodelare Fig. 7.—Thesizeof16sampleAclusters vs.projecteddistancerelative 0 1020 d (kpcifD=3Mpc)-► 198 4ApJ. . .276. .491H was notincludedbecausethehighbackgroundfromNGC5128 makes itssizeveryuncertain.Thethreeconfirmedclustersfrom 498 MESSER number ofœCen-sizedclusterscanbemoredirectly sufficiently forustoobtainameaningfulsize.Mostofthe œ Cen(oneofthelargestinGalaxy,withr=4pcand clusters plottedinFigure7rangefromaboutaslarge sample Barealsoexcludedsincetheirimagesnotresolved proportion totheassumeddistanceNGC5128.Thelarger r =90pc)toconsiderablylarger.Thesesizeswillincreasein by TysonandJarvis(1979);thesemomentsaredominated appreciated byexaminingFigure8,whichisaplotofthe the centralanddiffusecomponentsofimage,respectively. second andfourthmomentsoftheintensitydistributiondefined As inFigure7thecomparisonisforKingmodelsof16 an assumeddistanceof3Mpcandtheindicatedcoreradii. sample AclustersconvolvedwiththeobservedstellarPSFfor Two remarksshouldbemadehere.First,eventhough § IVb). NGC 5128isrelativelynearby,theresolutionofbestphoto- tion for16sampleAclustersarecompared totheoreticalmodelsforcore radii of1and5pc,atidalradius 50 pc,andadistanceof3Mpc(seetext, c graphic materialavailabletousdoesnotallowstraightforward t separation ofstarsfromclusters(seealsoFigs.1and2). Even withPDSanalysis,itwouldbeverydifficulttoidentify barely resolvedatall.Onepossiblecausefortheradialtrendis compact thanthosefartherout,andthattheinnerclustersare close tothecenterofNGC5128aresystematicallymore defined bystarsinFigure8.Thiswouldmeanthatthe objects similartoamoretypicalGalacticglobularcluster that thetidalforcesofNGC5128havelimitedsizes of 5Mpc,themodelcurveswouldmoveclosertoline as theinverseofgalactocentricdistance(Innanen,Harris, may beanotherargumentagainstdistances>5Mpc. well asthetidalradiidependonGalactocentricdistance and Webbink1983).InourGalaxy,however,thecoreradiias the innerclusters.Tidalradiiareexpectedtovaryroughly sample AclustersareallsignificantlylargerthancdCen,which (with ræ1pc).Second,atorbeyondthecanonicaldistance c Fig. 8.—Thesecond(C2)andfourth (C4)momentsoftheintensitydistribu- One obviousfeatureofFigure7isthattheclustersprojected © American Astronomical Society • Provided by the NASA Astrophysics Data System ET ÁL.Vol.276 9 2 complicated. Regardlessofitscause,thesystematicdependence that thecauseofsizetrendinNGC5128maybemore introduced biasesinourselectionofcandidatessampleA. of clustersizeonpositioninNGC5128hasundoubtedly in §IVaandthelargeextentofclustersystemcompared to thehalolightfoundin§IVccanbothbeexplained,atleast In particular,thedependenceofmagnitudeonpositionfound to beidentifiednearthecenterofNGC5128. clusters insampleBandtheresultsofstarcounts(§VI), result, combinedwiththefindingoftwobright,compact in part,byourpreferentialselectionoflargerclusters.This reinforces theconclusionofPaperIthatmanyclustersremain (Harris andRacine1979;vandenBergh1983),suggesting distances greaterthan36kpcandonly9%liebeyond21 radial distancesofatleast36(D/5)kpc,whiletheknown globular clustersinourGalaxy,only4%havegalactocentric projected distanceswhichcluster16wouldhaveifNGC5128 probable clustersinthecombinedsample)liewithin5!5 were at5and3Mpc,respectively.Theaverageprojected see Fig.6).Adifferentrepresentationofthistrendcanbeseen radius thatcontainshalfofthelight(vandenBergh1976; in NGC5128andifDæ5Mpc,itissurprisinglylarge. radial distancefromNGC5128ofour17sampleAclustersis particularly iftheNGC5128distanceis>5Mpc.However, bright clustersatlargeradiiinNGC5128thantheGalaxy, the Galaxy.Thiscomparisonsuggeststhattherearemore tions foroursampleAclustersandthebrightestglobularsin in Figure9,whichshowsacomparisonoftheradialdistribu- Only threeofthe17sampleAclusters(andfive27 projection effects,thesemaybecomparedto=11.1kpc clusters is14.4(D/5)kpc(8.6atD=3Mpc).Ignoring (cf. HarrisandRacine1979).Thesevaluescorrespondto pact clusters,absorptionbythedustband,and/orcompetition distribution inFigure9andmightexplainthelargeapparent the selectioneffectsdiscussedin§IVfr(abiasagainstcom- for 112GalacticglobularslistedinHarrisandRacine. tidal frictionmayhavedestroyedthemostmassiveclusters scale sizeoftheNGC5128clustersystem.Inaddition, from backgroundlight)arelikelytobeseverelyaffectingthe much closerthantabulatedbyHarris andRacine(1979). bright, sotidalfrictionisprobablynotdominant. the clusterswhichwehavefoundnearcentertendto be originally lyingnearthecenterofNGC5128.Nevertheless, the spheroidofNGC5128.Axtestsupportsanexcess of and clustercandidates(seeFig.3)mayhintatapreferential 16.0(D/5) kpc(9.6atD=3Mpc),andofall27probable axis isalignedwiththemajorofouterisophotes of orientation oftheclustersystem,insensethatitsmajor visually selectedclustersinthepolarregionsofgalaxy at spatial distributionwasintroducedinselecting(fromthelist of the 5%confidencelevel.Webelieveitunlikelythatabias in candidates) specificobjectsforspectroscopicobservation. selection techniquesaresosubjective, thatthisapparentalign- However, thesampleofconfirmed clustersissosmall,andthe 9 The NGC5128clustersystemextends,inprojection,to If thisextensionisrepresentativeoftheentireclustersystem WeexcludePal1sinceDaCostaand Mould(1982)havefoundittolie The distributionontheskyofvisuallyselectedclusters c) DistributioninNGC5128 198 4ApJ. . .276. .491H 1 No. 2,1984 Nonetheless wementionitbecauseatendencyhasbeennoted ment shouldonlyberegardedasatantalizingpossibility. in thepolarregions.Also,NGC5128’souterisophotesshow when visuallysurveyingourplatestofindmoreclustersuspects a markedincreaseinellipticityovertheinnerones(Cannon isophotes arepresumablyrepresentativeoftheoldestpopula- share commonspatialdistributions.Indeed,themuchmore tion inNGC5128,itwouldnotbesurprisingweretheyto populous clustersystemofM87showsjustsuchanalignment with themajoraxisdefinedbyitsouterisophotes(Harris and R =100kpc.TheGalacticglobularsabovethedashedlineexhibitsame velocity dataforanasymmetry(rotation)bydividing the 1981). Sinceboththeglobularclustersandfaintouter sample Aclustersintonorthernandsouthernhemisphere Smith 1976).Finally,wenotethathavesearched our compared withMvs.Galactocentricdistance,R,fortheGalacticglobular NGC 5128globulars(triangles)scaledtodistancesof3and5Mpcare centric velocities,600+55and660±45kms~(s.d.m.), components (withrespecttothedustlane).Themeanhelio- range inmagnitudesasdoesoursampleofNGC5128clusters. studies substantiateanelongatedclustersystem,further respectively, arestatisticallyindistinguishable.Shouldfuture clusters (circles).ThearrowindicatestheMofNGC2419,whichhas warranted. attempts toelucidateitsrotationpropertieswillclearly be clusters havestrongermetallic linesthandothebluerones. Due totherelativelyshortexposures (forsuchfaintobjects) v v Fig. 9.—Vmagnitudesvs.projecteddistance(inkpc)forthesampleA Inspection ofthevidiconspectra showsthattheredder © American Astronomical Society • Provided by the NASA Astrophysics Data System d) StrengthsofSpectralLines NGC 5128GLOBULARCLUSTERSYSTEM R Kpc -1 -1 -1 time. However,wenotethatFrogel’s(1980)(V—K)colorof attempt morequantitativeanalysisoflinestrengthsatthis and theattendantnoiseinspectra,wearereluctantto overall metallicitysimilarto47Tucanae.Fromourdatacluster GP’s cluster(our7)ledhimtosuggestthatithadan may beconsiderablymoremetal-richthan47Tue.Weplanto its colorandlinestrengths;thus,otherclustersinNGC5128 7 appearstobeatypicalNGC5128clusterwithrespect firm thissuggestion.Suchspectramightalsobeusedtotest obtain spectrawithhighersignal-to-noiseratiostotrycon- Cohen’s (1982)intriguing,butstilltentative,identificationof seen infrontoftwoglobularclusters. high-velocity interstellarlinesduetotheNGC5128halogas where Nisthenumberoftestparticles(globularclusters), We adoptthehighervalueforderivationofNGC5128 for theithparticle,andisprojectedradialdistanceof 0 estimate themassofamassivecentralobjectembeddedin better thanthevirialtheoremwhichistraditionallyusedto that particle.SubsequentstudybyTremaine(1981)oftheform v isthedifferenceinsystemicandobservedradialvelocities “projected massestimator”whichtheyargueconvincinglyis mass fromtheglobularclustervelocitiesofTable3. a multiplicativefactorof64forequation(1)ratherthan24. in whichthemassdensityisassumedtofollowlightyields of theprojectedmassestimatorrequiredforasphericalsystem equation 23,namely spherical systemoflow-masstestparticles.Theysuggestthat, spanning arangeofæ190kms,areavailableinthe appropriate valueforthesystemicvelocity.Manyvalues, orbital eccentricities,thebestestimatoristhatgivenbytheir in theabsenceofspecificinformationondistribution literature (seeTable4).Graham(1979)tentativelysuggeststhat sky linesintheearlieropticalwork.Anotherfactormay, this lackofagreementmayarisefromcontaminationbynight however, alsobeinvolved.Forthevelocitiesdeterminedsince papers themeanis489+22kms.SérsicandCarranza zi velocity field,anditslargeassociatedgradientsinthevicinity others, havegraphicallyillustratedNGC5128’scomplex of thedustband.Perhapsvariationsinexactslitplacement (1969), Graham(1979),andMarcelinetal(1982),among with thevelocitygradientstoproduceapartofspread 1970, themeanis549+6kms;forsevenearlier and largerslitsizesinsomeoftheearlierworkcombined the NGC5128velocitybyadoptingmeanradial of seen inTable4.Wemaysidesteptheissueofuncertainties in Table 5,lieabovethemost recentvaluesgiveninTable4 calculated forvarioussubsetsofourdataandgiven in our globularclustersasthesystemicvelocity.Suchvalues, for thegalaxy,itself.Asdiscussed in§IIIandtheAppendix, we believeourSITvelocities tobeonthesystemdefinedby the Galacticglobularclusters. Unlesssomesystematicerror has beeninadvertentlyintroduced (andobservationsofcom- Bahcall andTremaine(1981)haverecentlyformulateda To undertakethecalculationofv'swerequirean zi a) DeterminationfromGlobularClusterVelocities V. THEMASSOFNGC5128 499 198 4ApJ. . .276. .491H -1 _ 1 -1 1 500 12 parably faintstarsinœCenargueagainstthispossibility), systemic velocityofNGC5128andthemeanour an intriguingdifferenceofæ90±30kmsbetweenthe each sampleofclustersand(2)Graham’sv=548km sample ofglobularclustersissuggested. systemic velocity(1)themeanclusterdeterminedfor be a=140kmsrelativetotheirmeanvelocityandcr 165 km srelativetoGraham’svelocity.Regardlessoftheexact s". Thevelocitydispersionsofthe19clustersarefoundto in thepresentprogramliewithin24'.5ofcenter of is foundfromequation(1)(modified)tobe^ 1.6 sample orsystemicvelocityadopted,themassofNGC5128 systemic x 10(D/5)M.Sincealloftheglobularclustersobserved 0 We thereforecarriedoutourmasscalculationsusingforthe \ 484: \ 420: { { { 1 470±50 f 630 f 434±75 f 462±25 Optical 468 ±40 450 465 ±50 605 475 ±47 508 ±11 602 ±38 607 ±18 548 ±5 559 ±24 551.4 536 ±30 545 ±5 541 ±8 557.6 ‘1981 © American Astronomical Society • Provided by the NASA Astrophysics Data System 1980 1980 +19811982. 1980 +1981 Mean RadialVelocitiesfromtheNGC5128 “ r,© {551.1 : Radio Sample N(kms) 563 551 546 ±3 530 550 551.8 551 553.2 Abs.:21-cm(VLA) Globular Clusters Abs. +[0II]3727em. Em.: Ha,[NII]6584 Em.: [0II]3727,Hß,Hy Em.: [0II]3727\ Abs.: CaII+otherAbsorptionLines Em. +Abs.Lines Em.: Ha(Fabry-Perot) Abs.: CaIIH,H£,FeI4383J Abs.: CaIIK,Hô,G-bandJ Em.: [0II]3727,Ha,[N6548'! Abs.: 21-cm,profilemidpointT Abs.: 21-cm,strongestcomponentJ Em.: Ha\ Abs.: NaDJ Abs.: H£CO,avg.of3components Abs.: 21-cm,strongestcomponentJ Em.: 21-cm1 Abs.: 21-cm,strongestcomponent Abs.: OH,strongestcomponent Abs.: H2CO,strongestcomponent Abs.: CaIIH,K,I4226,Fe4271,4383 Abs.: NaDJ Em.: Ha1 Abs.: FeI4383,Hß,Mg5170,5183 Em.: Hß,[0III]4959,5007,[NII]6548,6584 Em.: Ha,[NII]6584 Em.: [0II]3727,Hß,III]4959,5007 Em.: Ha,[NII]6548,6584,[0I]6300,6364,[S6716, TABLE 5 Em.: Ha(Fabry-Perot) Ha, [SII]6717,6731 15 63340 17 63036 19 63732 5 60837 ^ Estimates oftheSystemicVelocityforNGC5128 Features HESSER ETAL.Vol.276 j } TABLE 4 10 kpc. TheresultspresentedinTable6separatelyforthe1980, NGC 5128,thisvaluereferstothemassinterior36(D/5) the exactsampleandsystemicvelocitychosen.[For 1981, andthecombined1980-1982sampleareinsensitiveto equation (1)was,infact,employed,therebyaccountingforthedifference that Tremaine’s(1981)formulationwasusedforthemasscalculations, earlier reference. of afactor2.7betweenthemassesinTable6andthosegiven the 10 Althoughanearlierreportonthiswork(Hesseretal.1981)claimed All 191415 d> 10'825 20 dclO' 115.6 1980 53.49 particular sampleofglobularclusters. 1981 151716 NGC 5128MassEstimatesfromtheGlobular a b Systemictakentobethemeanvelocity ofthe ^systemicistakenfromGraham1979. SAMPLE N^tciobulars^^^Graham^ 67311 Cluster RadialVelocities Baade andMinkowski(1954) Burbidge and(1962) Burbidge and(1959) Humason, MayallandSandage(1956) Sérsic (1969) Sersic andCarranza(1969) Evans andHarding(1961) Kunkel andBradt(1971) Roberts (1970) Gardner andWhiteoak(1976a) Whiteoak andGardner(1971) Gardner andWhiteoak(1976b) Graham (1979) Marcelin etal.(1982) Appenzeller andMöllenhoff(1980) Möllenhoff (1981) Rodgers andHarding(1980) Whiteoak andGardner(1979) van derHuist,Golischand Haschick (1983) TABLE 6 1 Reference Mass [x10(D/5)M] 0 198 4ApJ. . .276. .491H 1 1 12 -1 -1 No. 2.1984NGC5128GLOBULARCLUSTERSYSTEM501 comparison withtheresultsfromprojectedmassestimator, differences inmassestimatesforothersystemsanalyzedby method (cf.PaperI)isæ3x10(D/5)M.Similar we notethatthemassfoundbystandardvirialtheorem the twotechniqueswerereportedbyBahcallandTremaine Tremaine’s (1981)remarkthattheirprojectedmassestimator produced onlyminordifferencesfromthevaluesgivenin Table 6.OurcalculationsarethusinaccordwithBahcalland is relativelystablewhenappliedtosmallnumbersoftest (1981).] Eliminationofspecificobjects,suchascluster26, This lieswithintherange5.4,below.Adoption (2) thegalaxiesinNGC5128subgrouphaveisotropic of thesecondassumptionyieldsmassesthatrepresentlower from theglobularclusters.Ifgalaxyorbitsarelinear, limits, butwhicharealreadylargerthanthosederived (1979) andbyWebsteretal(1979),whilethedistance to the quotedmasseshavetobemultipliedbyafactor ziz Fig. 10.—Distributionof17probablememberstheNGC5128/5253 Although Websteretal(1979)concludedthatthe We haveadoptedthefollowingthreeassumptionsfor o 40° - 30° - b 10- © American Astronomical Society • Provided by the NASA Astrophysics Data System 330° 320°310°^300° 12 12 Mass including1332-453 7.6x10(D/5)M5.0 Mass excluding1332—453 2.4x10(D/5)M1.7(D/5)M 0 0 a b CorrespondstolinearradiusR= 350 (D/5)kpc. CorrespondstolinearradiusR= 700 (D/5)kpc. ab Sample R<4?08?0 Mass EstimatesforNGC5128Subcluster TABLE 8 12 2 made abouttheirorbitsandTremaine’s(1981)modification of 2.If(aswasdonefortheglobulars)noassumptionsare In viewofthelackastrongcentralmassconcentration is employed,themassesinTable8mustbemultipliedby4. NGC 5128/5236Grouplieintherange8-25x10{D/5)M. in theCentaurusGroup,lattervaluemaybemore M ä1.6x1012(D/5),suggeststhatNGC5128hasa appropriate. Ourestimatesofthetotalmass of themassinNGC5128/5236clustermightconsist massive halothatextendsbeyondthedistancessampledby Comparison withthevaluederivedfromglobulars, more smoothlydistributedintraclustermaterial. the globularclustersidentifiedinFig.3.Alternatively,much galaxy appearsnearthecornerofJ270.Thusitisimpossible vignetting fromthecalibrationwedgepresumablyadversely starlike objectsinthevicinityofNGC5128.However,this Sky SurveyfilmJ270toestimatetheexcesspopulationof 0 to makecircularlysymmetriccountsbeyondæ28',and 0 be surveyed.Second,ifthedistanceofNGC5128is<5Mpc, with newplatematerialwasdesirableforatleasttworeasons. surprisingly (cf.§I)largenumberofexcessstarlikeobjects influences portionsofthecountsatsmallerradii.Dueto background measurementfreeofstarclusters(Hanes1980). associated withtheclusteringofgalaxiesinareasampled To thatendwemadetwoindependentsetsofcounts it isimportanttosampleatlargeradiiinorderhavea for determiningthebackgroundlevel,averylargeareashould First, inordertosmoothoutthepossiblestatisticalfluctuations (æ 600)found,thosecountssuggestedthataredetermination centered onNGC5128(kindlyprovidedbytheRoyal starlike objectsonafilmcopyofUKSchmidtplateJ3986 parison withthePickering-Racinesecondaryimagesonplate Cannon). Observatory, EdinburghPhotolabsthroughDr.R.D. were madeofJ3986byvandenBerghandHesser.For the formerset,countsweremadethroughalow-power P5254 (cf.Table2).Independentcountsusingdifferentreseaux each of606arcmin,centeredæ77'north,south,east, and west ofthegalaxycenter.Ringsectors30°widthwere counted alternatelyinwardandoutward.Nonstellarimages eyepiece in25ringsofwidthbb'.'Ocenteredonthisgalaxy (cf. Table9).Backgroundcountswereobtainedinfourareas, (which amountedtoperhaps20%ofthetotal)were not VI. NEWRESULTSFROMSTARCOUNTSCONCERNINGTHESIZE In PaperIweusedstarcountsmadeontheSRCSouthern J3986 reachesJ>22.0magaccordingtoavisualcom- OF THENGC5128GLOBULARCLUSTERSYSTEMAND a) CountsonUKSchmidtPlateJ3986 DISTANCE TONGC5128 198 4ApJ. . .276. .491H 2 12 while therestmaybefaintextensions oftheopticallightNGC5128, faint foreground“highlatitudecirrus” illuminatedbytheGalacticplane, to 5>19ina24'x36'areacentered onNGC5128. itself. EvansandHarding(1961)also foundpatchinessintheirstarcounts (Cannon 1981).Someofthematerial visibleonsuchplatesisundoubtedly NGC 5128toderivethenumberofglobularclusters No. 2,1984 over background(whichispresumedtobedueglobular surrounding thisgalaxy.FromthedatainTable9,excess quadrant, itseemssafesttousecountsinringssurrounding clusters) isfoundtobe325±79s.e.Extrapolatingtheobserved counted. Thesurfacedensitiesandtheirformalmeanerrors 475 ±120(estimatederror),avaluewhichisingood surface densitiestor=0yieldsatotalclusterpopulationof is significantlyhigherthanthe=5.68±0.05, counts inSRCfieldJ270.Abreakdownofthecluster agreement withthatofPaperIfromtheslightlydeeper errors. Anadditional(butprobablysmall)uncertaintyis net countsmaybesignificantlylargerthantheirquotedmean particularly highconcentrationofdistantgalaxiesinthe surrounding NGC5128.Themostlikelyexplanationforthe for anannuluswith17.05 a 9051 5.41±0.06 240 6.32±0.41 423 6.18±0.30 281 6.16±0.37 488 5.84±0.26 468 6.16±0.28 367 6.04±0.32 321 6.03±0.34 500 5.48±0.25 587 5.42±0.23 538 5.45±0.23 186: 6.12:±0.45 NGC 5128GLOBULARCLUSTERSYSTEM 13 14 15 by Hesserusinga10xbinocularmicroscope,andthesame region thatwasusedtodeterminethebackgroundinTable9, at =22',ratherthanthesignificantdecrease(compared computed forring20isverysimilartothatfoundbyvanden reseaux andpreceptsdescribedinPaperI.Thebackground ground densityofgalaxiesseverelylimitstheaccuracywith entries ofTable9,weconcludethatvariationsintheback- reasons, andthestatisticaluncertaintiesassociatedwith showing thatsomedistantclustersdo,indeed,exist.Forthese more, threeclusters,clusters1,16,and27(cf.Table3),lieinthe globulars haveR>35kpc,andmanyofthemoredistantones find ahighpercentageofremote,ordinaryNGC5128globularswhenwe rather luminousandlargeclusters,itseemsimprobablethatwewould counts inTable11alsoconfirmthehighbackgroundtoward with theskyareas)seeninTable9forradii16!5-27'5.The Bergh. However,thecountsofTable11dosuggesta NGC 5128. which thetotalclusterpopulationcanbedeterminedfor are sparselypopulated,low-luminosityclusters.Further- the æ100kpcradiusofGalacticglobularclustersystem, population inthatdirection. must discriminateagainstasignificantforeground-starandbackground-galaxy for thebackgroundwemayestimate,bylinearextrapolation in thebackgroundcounts.Byadoptingo=5.78±0.05(m.e.) the southofNGC5128,aswellimportancevariations it isnotparticularlybothersome.Lessthan4%oftheGalactic that derivedearlierfromthedataofTable9. of starlikeobjectsassociatedwithNGC5128æ440±110 of thedataforrings2-4toR=0,atotalexcesspopulation (statistically insignificant)increaseoverbackgroundinring4 examining thelimitingmagnitude.Predictionsofstellarsurfacedensities for surface densitiesofdistantgalaxiesprovideanindependentmethod for in thelimitingmagnitudeadopted.ModelsofGalactichaloand the (estimated error).Thisresultisinexcellentagreementwith Jarvis andTyson(1981)havederivedarelationforthedependenceof the described elsewhere(Pritchet1983u).Fromastudyof12Galactichalofields, were kindlycalculatedforusbyPritchet(1983b)basedonGalaxymodels 9 18.1maginTable13 If NGC5128liesatadistanceof^3Mpc,ourstarcounts © American Astronomical Society • Provided by the NASA Astrophysics Data System V trast insurfacedensityofglobularsversusotherforeground examined thisquestioninasomewhatgeneralfashionwhile and backgroundobjectsmightbeexpectedtooccur.We was chosentoyieldtheobservednumberofæ600objects namely: (1)atotalpopulationof700clusters;thisnumber brighter thanJæ22foranassumeddistanceof3Mpc; adopting severalassumptionsappropriatetoNGC5128, described andgiveninTable12.FromFigure12wecansee luminosity functionadoptedearlier(i.e.,=—6.8, within rasafunctionofassumed galaxydistance.Thefigureshowsgood results areshowninFigure12alongwithoalready than J=22forvariousassumedgalaxydistances.These the averagesurfacedensityofclusterswithinrandbrighter o =1.2).InTable14wegivethetotalnumberofclustersand contrast betweenclustersandbackground plusforegroundfor322isdueto therapidincreaseincontribution luminosity functionaswelltheM/sandinferredsizes system, aspresentlyexplored(namelythecumulative of thebrightestclusters),seemmostconsistentwithadistance depends uponanassumptionwhichisdifficulttotest,e.g.,that usually quoted.Wecaution,however,thatourreasoning globular clustersystemsassociatedwithLocalGroupgalaxies. luminosity functionsimilartotheof the NGC5128globularclustersystemhasaGaussian sg s +g Hm @s+g g The factthatthereisanoptimumdistancefordetectionofa Using anassumeddistanceof3MpcforNGC5128we Fig. 13.—Comparison(foranassumeddistanceof3Mpc)betweenthe In summary,thepropertiesofNGC5128globularcluster © American Astronomical Society • Provided by the NASA Astrophysics Data System ET AL.Vol.276 16 12 distance estimatetoNGC5128concernsthedeterminationof NGC 5253.Since5128and5253arebothmembers the MvalueforSN1972e,aTypeIsupernovain electric BmagnitudesmeasuredbyArdeberganddeGroot NGC 5128shouldalsoapplyto5253.Fittingthephoto- of thesamegroup,newlydetermineddistanceto Type IsupernovabyBarbon,Ciatti,andRosino(1973)yields A =0.40±0.03(basedonthereddeningdiscussionofvanden B(max) =8.35±0.10(estimatedfittingerror).Combining this valueofM(max)(whichis>1magfainterthanwould Bergh 1976)givesM(max)=—19.45±0.4.Comparisonof this with(m—M)=27.39±0.36(basedonD3Mpc)and be estimatedusingD=5Mpc)withB(max)ofaccurately (1973) forthissupernovatothemeanlightcurveoffast observed TypeIsupernovaeinellipticalsbeyondtheLocal efficacious, anditseemslikelythatmanymoreclusterswill global valueoftheHubbleparameter. Supercluster shouldbeabletoprovideadeterminationofthe metal-rich than47Tue.Furthermore,theredderclusterslie be foundbythisapproach.Thespectroscopicallyconfirmed nearer (inprojection)toNGC5128thanthebluerones. NGC 5128globularclustersmaybesignificantlymore clusters showawiderangeinB—Vcolorandimplythatsome b value ß(max)=7.75±0.1quotedby ArdeberganddeGroot(1974)results estimate themassofNGC5128tobeæ1.6x10{D/5)M© cluster systemwiththatdefinedbytheouterisophotesof to accountforapossiblealignmentofthemajoraxis confirmed clusters.However,wearguethattheydonotseem significantly totheobservedspatialdistributionof from theirattempttofittwophotographic prediscoveryobservationsby interior to36(D/5)kpc.TheimpliedM/Lyratiois^16(5/D). NGC 5128,itself.Fromtheglobularclustervelocities,we Austin (1972)tothelightcurve. From analysesofthevelocitiescompaniongalaxiesweinfer Selection effects(discussedin§IV)probablycontribute b that thereismassintheNGC5128halo(or ß ß 0 star clustersthathavesofarbeenidentified.Finally,we Group ofgalaxies)atorbeyondtheradiusmostdistant a totalclustersystemnumberingæ600members,asdeduced used thecumulativeclusterluminosityfunction(normalizedto cited. Adoptionofsuchadistancehasramificationsformodels NGC 5128is^3Mpc,ratherthanthelargervaluesfrequently from newstarcounts)toarguethatthedistance realms thatliebeyondthescopeofthispaper. of NGC5128anditsenergyfluxthroughouttheradio-to-y-ray peculiar galaxyNGC5128havetooffer.Werematerial at mation thatobservationsoftheoldestcomponents the craft available,anenormousincreaseinthenumberofreliable cluster candidateswouldalmostcertainlyemerge.Assuch we higher platescaleordiffraction-limitedimageryfromspace- offer yetanotherjustificationforimageryofNGC5128from Space Telescope(or,evenbetter,theproposed,widerfield c) ApplicationoftheNewCentaurusGroupDistanceto 16 One ofthenumerouspotentialimplicationsnew Our visualselectiontechniqueshaveprovenremarkably WeareindebtedtoGustavTammann forpointingouttousthatthe Our workhasonlybeguntotapthewealthofnewinforma- Luminosity ofSN1972einNGC5253 VII. SUMMARY 198 4ApJ. . .276. .491H subgiant starsineachcluster;and(3)spectraofROA40 third categorieswereobtained,andthosefewerthanin monitor spectrograph+SITbehavior.(Alternatedesignations 6352 =C1721—484.)In1982,onlyspectrainthesecondand for theglobularclustersare:NGC1851=C0512-400 and ROA65inœCenobservedseveraltimespernightto reduced asdescribedin§IliaandHesserHarris(1981) observed duringthe1981Apriland1982MaySITruns also kindlysharedplateswithus.B.andR.Agguirre 5129 =C1323-470,NGC6752C1906-600,and (Hesser, Bell,Cannon,andHarris1982),œCen=NGC velocities derivedfromobservationsofanumbergiantand taken withthetelescopeinzenith;(2)clustermean tions for:(1)thesolar(twilight)absorption-linespectrum tions byothers.InTable15wesummarizeour1981observa- yield averagevelocitiesthatmaybecomparedwithdetermina- provided continualencouragementandconstructivecriticism, Telescope AllocationCommittee,havegenerouslysupported work. CTIODirectorsV.BlancoandP.Osmer,aswellthe deserve carefulconsiderationasameansofeliminatingsome bility ofdiffractionlimitedimageryfromspace,multiplex No. 2,1984NGC5128GLOBULARCLUSTERSYSTEM507 and allowedustoexaminehisplatecollection.R.Dufour spectroscopic observationsandreductions.J.Graham the observationalprogram.AtCTIOeffortsofB.Atwood, separating globularsfromstarsandgalaxiesoverawidefield conditions achievablefromtheground.Priortoavaila- of theselectioneffectsinherentinourwork. spectroscopic and/orphotometrictechniquescapableof 1981. Inthecolumnsoftablewepresent:(1)object; J. Baldwin,A.Gómez,G.Martin,M.Navarrette,C.Poblete, Star labinstrument),aswellunderthebestpossibleseeing O. Sáa,S.Schaller,andR.Venegaswereessentialtothe Velocity reductionscarriedoutfortheGalacticobjects Numerous colleaguesandfriendshavecontributedtothis © American Astronomical Society • Provided by the NASA Astrophysics Data System Twilight Sky-30171024 0 10 NGC 5139-ROA65-471424818 245 3 NGC 1851-401835638 319 +37 NGC 6752-6013-5230 -32 -20 NGC 6352..-4810-12828 -121 -7 NGC 5139-472022051 228 -8 NGC 5139-ROA40-471324014 223 17 NGC 6352-48 10-11313-1218 NGC 5139-ROA40...-47 92452122317 NGC 5139-ROA65-47 7251362456 -1 -1 Object (°)N(kms) ) (kms (1) (2)(3)(4)(5)(6) (7) Summary ofSITRadialVelocityDeterminationsforGalacticObjects THE SITVELOCITYSYSTEM APPENDIX TABLE 15 1981 Data 1982 Data y \ 'rQ7Vicat'r,G^cat y -1 -1 _1 which issufficientlysmallthatwefeeljustifiedinneglectingit. data inTable15.First,theweightedmeandifferencetreating deleted.) Second,thereisnoconvincingevidencefora which weregenerallysecuredatverylargehourangles, all determinationsasbeingofequalweightis+5.8kms, Webbink (1981)exceptforthe0kmsadopted velocity determinedfromtheNdatapoints;(5)standard times anobjectwasobservedor,inthecaseofclusters, the differencebetweenobservedandcatalogvelocities. twilight skyandthe(lowerweight)NGC6352valuefrom velocity fortheobject,whichisgenerallytakenfrom deviation ofonethosedatapoints,<7;(6)the“catalog” number ofindividualstarsobserved;(4)themeanradial were veryhelpful.K.Freeman,W.Harris,A.Sandage,F. copies essentialtothestarcountingefforts.G.,Burbidge computer terminals.R.CannonandthePhotolabsatRoyal relaxation at(very)oddhoursawayfromtheLaSerena (It wouldbe—0.7kmsiftheNGC1851observations, (2) itsdeclination(inwholedegrees);(3)thenumber,A,of We alsothankD.Crowe,Duncan,andR.Haapalafor much appreciated,constructivecommentsonourefforts.To valuable commentsonthestarcountanalysisbyC.Pritchet at KittPeak.CalculationsofGalactichalomodelsand accorded ustheopportunitytousemeasuringequipment Observatory, Edinburghprovidedverypromptlythefilm and R.Gayosoprovidedmuchappreciatedmomentsof Shawl, Hesser,andMeyer’s(1981)image-tubestudy;(7) colloquia haveguidedus,weexpressourdeepestappreciation. Engineering ResearchCouncilgrant. appreciates thefinancialsupportofaNaturalSciencesand their helpwiththemanuscriptpreparation.G.L.H. all theabove,andtocolleagueswhosequestionsraisedat Schweizer, G.Tammann,andS.Tremainealsocommunicated ¿ Several remarksmaybemadeonthebasisof1981 198 4ApJ. . .276. .491H - 508 dependence ofthederivedvelocitiesupondeclination.Third, candidates) nights.Themeanvelocitydifference(f^—F) the observationsofcoCen,whichislocatedneartoNGC values forœCenitselfreflectthefactthatmostofstars system issatisfactoryforNGC5128.Fourth,thehigher<7; 5128 onthesky,suggestthatzero-pointofourvelocity that theywereusuallyobservedforonly30minutes. observed werefaintsubgiantsinthevicinityofturnoffand Ardeberg, A.,anddeGroot,M.1973,Astr.Ap.,28,295. nights (May19-21UT),althoughtheNGC5128datawere .1974,inSupernovaeandSupernovaRemnants,ed.C.B.Cosmovici Allen, D.A.1974,M.N.R.A.S.,168,27P. secured onlyonthesecond(10candidates)andthird(three of allsuitablecomparisonstarsfromtherunis8kms\which Appenzeller, I.,andMöllenhoff,C.1980,Astr.Ap.,81,54. Atwood, B.,Ingerson,T.,Lasker,B.M.,andOsmer,P.S.1979,Pub.A.S.P., bs?at .1980,Ap.J.(Letters),240,L93. .1979,A.J.,84,1270. Austin, R.D.1972,IAUCire.,No.2421. Baade, W.J.,andMinkowski,R.1954,Ap.119,215. Freeman, K.C,Karlsson,B.,Lyngâ,G.,Burrell,J.F.,vanWoerden,H., —.1962,Nature,194,367. Evans, D.S.,andHarding,G.A.1961,M.N.A.S.SouthAfrica,20,65. Dottöri, H.A.,andFourcade,C.R.1973,Astr.Ap.,23,405. Barbon, R.,Ciatti,F.,andRosino,L.1973,Astr.Ap.,25,241. Bahcall, J.N.,andTremaine,S.1981,Ap.J.,244,805. Blanco, V.M.,Graham,J.A.,Lasker,B.andOsmer,P.S.1975, Blanco, V.M.1982,Pub.A.S.P.,94,201. Faber, S.M.,andGallagher,J.1979,Ann.Rev.Astr.Ap.,17,135. Caldwell, C.N.1983,Ap.J.,inpress. Fairall, A.P.1981,M.N.R.A.S.,196,IIP. de Vaucouleurs,G.1975,inStarsandStellarSystems,Vol.9,Galaxiesthe Cohen, J.G.1982,Ap.(Letters),260,L45. Cannon, R.D.,andStewart,N.J.1981,M.N.R.A.S.,195,15. Cannon, R.D.1981,inProc.SecondESO/ESAWorkshop(ESASP-162),p.45. Burbidge, E.M.,andG.1959,Ap.J.,129,271. —.19766,Proc.Astr.Soc.Australia,3,63. .1981,Pub.A.S.P.,93,291. .1979,Ap.J.,232,60. Gardner, F.F.,andWhiteoak,J.B.1976a,M.N.R.A.S.,175,9P. Frogel, J.A.1980,Ap.(Letters),241,L41. Da Costa,G.,andMould,J.R.1982,privatecommunication, Graham, J.A.1978,Pub.A.S.P.,90,237. Graham, J.A.,andPhillips,M.1980,Ap.(Letters),239,L97(GP). Hanes, D.A.1980,inIAUSymposium85,StarClusters,ed.J.E.Hesser Harris, W.E.,andRacine,R.1979,Ann.Rev.Astr.Ap.,17,241. Harris, W.E.,andSmith,M.G.1976,Ap.J.,207,1036. Hesser, J.E.,Harris,H.C,vandenBergh,S.,andG.L.1981, in Hesser, J.E.,Bell,R.A.,Cannon,D.,andHarris,G.L.H.1982, A.J., Hesser, J.E.,andHarris,G.L.H.1981,Pub.A.S.P.,93,139. J. E.HesserandS.vanden Bergh:DominionAstrophysicalObservatory,5071West SaaichRoad,Victoria,B.C.V8X4M6, H. C.Harris:PhysicsDepartment, McMasterUniversity,Hamilton,Ontario,L854M 1,Canada G. L.H.Harris:PhysicsDepartment, UniversityofWaterloo,Ontario,N2L 3G1,Canada Canada 91, 120. (Dordrecht :Reidel),p.103. Ap. J.(Letters),198,L63. Goss, W.M.,andMebold,U.1977,Astr.Ap.,55,445. For the1982runspectrawereobtainedonpartsofthree of ChicagoPress),p.557. Universe, ed.A.Sandage,M.andJ.Kristian(Chicago:University 87, 1470. (Dordrecht: D.Reidel),p.10. A. G.D.PhilipandS.Hayes(Schenectady,N.Y.:L.DavisPress), I AUColloquium68,AstrophysicalParametersforGlobularClusters, ed. p. 467. © American Astronomical Society • Provided by the NASA Astrophysics Data System HESSER ETAL. REFERENCES -1 again issufficientlysmallthatwehavechosennottoapplyit the velocitiesinTable1.However,errorsarelargerfor clusters 22and23,wereobserved)seemstobelargerthanfor due toazero-pointshift,inthesensethatmean difference basedonœCenobservationsaloneforthesecond night (theoneonwhichmostNGC5128candidates,including the thirdnight.Duetoweatherproblemsandsmallnumber convincingly mapthisbehavior,andweignoreit.Therefore, statistics, wedonotfeelthathaveenoughdatato Hesser, J.E.,Harris,H.C,vandenBergh,S.,andG.L.1982, it ispossiblethatthevelocitieslistedinTable1forclusters22 and 23shouldbedecreasedbyæ30kms. 1982 observationsofROA40and65.Thisappearstobe Jarvis, J.F.,andTyson,T.A.1981,A.J.,86,476. Innanen, K.A.,Harris,W.E.,andWebbink,R.F.1983,A.J.,inpress. Humason, M.L.,Mayall,N.U.,andSandage,A.R.1956,A.J.,61,97. King, I.1962,A.J.,67,471. Kunkel, W.E.,andBradt,H.V.1971,Ap.J.(Letters),170,L7. Marcelin, M.,Boulesteix,J.,Courtes,G.,andMilliard,B.1982,Nature, ———. 19836,privatecommunication. Peterson, B.A.,Dickens,R.J.,andCannon,D.1975,Proc.Astr.Soc. Möllenhoff, C.1981,Astr.Ap.,93,248. Materne, J.,andTammann,G.A.1974,Astr.Ap.,37,383. Racine, R.1969,A.J.,74,1013. Pritchet, C.J.1983a,A.J.,inpress. .1969,Nature,224,253. Rood, H.J.,Rothman,V.C.A.,andTurnrose,B.E.1970,Ap.162,411. Rodgers, A.W.,andHarding,W.1980,Ap.J.(Letters),236,L17. Roberts, M.S.1970,Ap.J.(Letters),161,L9. Schaller, S.,Osmer,P.Albrecht,R.,andLasker,B.M.1978,inCTIO Schaller, S.1982,privatecommunication. Sandage, A.R.,andTammann,G.1974,Ap.J.,194,559. Pickering, E.1891,HarvardAnnals,26,14. -.1979,Pub.A.S.P.,91,639. Sérsic, J.L.1960,Zs.Ap.,51,64. Tyson, J.A.,andJarvis,F.1979,Ap.(Letters),230,L153. Tremaine, S.1981,privatecommunication. Tammann, G.A.1983,inHighlightsAstr.,88,1476. Shawl, S.J.,Hesser,J.E.,andMeyer,E.1981,inIAUColloquium68, Sérsic, J.L.,andCarranza,G.1969,Inf.Bull.SouthernHemisphere,14,32. van denBergh,S.1976,Ap.J.,208,673. van denBergh,S.,Hesser,J.E.,andHarris,G.L.H.1981,A.J.,86,24(PaperI), — .1983,Mem.Soc.Astr.Italiana,54,199. van derHulst,J.M.,Golisch,W.F.,andHaschick,A.D.1983,Ap.(Letters), .1979,Proc.Astr.Soc.Australia,3,319. Webster, B.L.,Goss,W.M.,Hawarden,T.G.,Longmore,A.J.,andMebold, Webbink, R.F.1981,Ap.J.Suppl,45,259. Whiteoak, J.B.,andGardner,F.1971,Ap.Letters,8,57. Pub. A.S.P.,94,754. 297, 38. Australia, 2,366. p. [9J.2.20.5. Facilities Manual,ed.J.E.Hesser(2ded.;Tucson:AURA,Inc.,), Astrophysical ParametersforGlobularClusters,ed.A.G.D.Philipand D. S.Hayes(Schenectady,N.Y.:L.DavisPress),p.193. 264, L37. U. 1979,M.N.R.A.S.,186,31.