Astronomy&Astrophysicsmanuscriptno.8327 (cid:13)c ESO2008 February2,2008 XMM-Newtonobservations of the Galactic globular clusters NGC 2808 and NGC 4372 M.Servillat1,N.A.Webb1,andD.Barret1 CESR,Universite´PaulSabatier,CNRS–9avenueduColonelRoche,31400Toulouse,France e-mail:[email protected] 8 0 Received20July2007/Accepted28November2007 0 2 ABSTRACT n Aims.GalacticglobularclustersharbourbinarysystemsthataredetectedasfaintX-raysources.Theseclosebinariesarethoughtto a playanimportantroleinthestabilityoftheclustersbyliberatingenergyanddelayingtheinevitablecorecollapseofglobularclusters. J Theinventoryofclosebinariesandtheiridentificationisthereforeessential. 7 Methods.WepresentXMM-NewtonobservationsoftwoGalacticglobularclusters:NGC2808andNGC4372.WeuseX-rayspectral andvariabilityanalysiscombinedwithultra-violetobservationsmadewiththeXMM-Newtonopticalmonitorandpublisheddatafrom ] theHubbleSpaceTelescopetoidentifysourcesassociatedwiththeclusters.Wecomparetheresultsofourobservationswithestimates h frompopulationsynthesismodels. p Results.Fivesources out of 96 are likely to be related toNGC 2808. Nine sources are found in the fieldof view of NGC 4372, - o nonebeing locatedinsideitshalf-massradius. Wefindonequiescent neutronstarlow massX-raybinarycandidateinthecoreof r NGC2808,andproposethatthemajorityofthecentralsourcesinNGC2808arecataclysmicvariables.Anestimationleadsto20±10 t cataclysmicvariableswithluminosityabove4.25×1031ergs−1.MillisecondpulsarscouldalsobepresentinthecoreofNGC2808, s a andsomesourcesoutsidethehalf-massradiuscouldpossiblybelinkedtothecluster. [ Keywords.Galaxy:globularclusters:individual:NGC2808,NGC4372–X-rays:general–Stars:binaries:close 2 v 0 1. Introduction served(Hurleyetal.2007)whichcouldexplainthepresenceof 9 close binaries located outside of the half-mass radius, as a CV 6 XMM-Newton and Chandra X-ray observatories are currently in M 22 (Pietrukowiczetal. 2005). D’Amicoetal. (2002) and 0 revealingmoreandmorefaintX-raysourcesinglobularclusters Colpietal. (2003) also discussed the case of two MSPs out- . 2 (GCs)thankstotheirhighsensitivityandhighangularresolution side the half-massradiusofNGC 6752which couldhave been 1 respectively (e.g. Webbetal. 2004, 2006; Heinkeetal. 2003b, ejectedfromtheGCthroughinteractionswithacentralmassive 7 2006). Thirteen bright X-ray sources with L >1036ergs−1 X object. 0 are found in the 152 known Galactic GCs. These are neu- : From early X-ray observations of GCs, we know that they v tron star low-mass X-ray binaries (LMXBs) showing type I areefficientatproducingX-raybinariesintheircorecompared i X-ray bursts (e.g.Lewin&Joss1983). Numerous faint X-ray X sources with L <1034.5ergs−1 have been shown to be a va- to the field. Using the eleven known bright LMXBs in GCs X r riety of objects, mainly binaries such as LMXBs in quies- (now thirteen), Verbunt&Hut (1987) showed that LMXBs in a GCsareproduceddynamicallythroughexchangeencountersof cence (qLMXBs), cataclysmic variables (CVs), active binaries isolated neutron stars with primordial binaries as opposed to (ABs), or millisecond pulsars (MSPs), the probable progenity the much less probable evolution of a primordial binary into of LMXBs. These objects have been classified through multi- an LMXB. Observations also support the fact that qLMXBs wavelength analysis: qLMXBs are usually identified by their in GCs scale with the cluster encounter rate (Gendreetal. soft blackbody-like X-ray spectra (e.g.Gendreetal.2003a,b), 2003a; Pooleyetal. 2003), implying that qLMXBs are formed CVs can be confirmed by their blue, variable optical counter- through dynamical processes which occur in dense stellar sys- part (e.g.Webbetal.2004), ABs by their main-sequence,vari- tems. ConcerningCVs, following their discoveryin significant able optical counterparts(e.g.Edmondsetal.2003), and MSPs numbers in 47 Tuc (Grindlayetal. 2001), it has been pointed bytheirradiocounterpart(e.g.Grindlayetal.2001). outfollowingpopulationsynthesisstudies(Ivanovaetal.2006; It is clear that through mass segregation, heavy objects Trentietal.2007)andfromobservations(e.g.Webbetal.2004; such as binaries are concentrated towards the core of GCs Pooley&Hut2006)thattheymaybeformedinGCseitherdy- (Lightman&Grindlay1982).Therefore,weexpectthemajority namicallyor,foralowerfraction,fromprimordialbinaries. of X-ray binaries, which are more massive than the mean stel- larmass,tobelocatedinsidethehalf-massradius.Hurleyetal. Neglectingthe roleofbinaries,we knowthatGCevolution (2007) showed from simulations that the fraction of primor- leads to a core collapse followed by the GC disruption on a dial binaries destroyed in the core by a variety of processes is timescale shorter than the mean age of GCs, estimated to be balanced by the combination of mass-segregation and creation 11.5 ± 2.6 Gyr (Carrettaetal. 2000). This core collapse must of new binaries in exchange interactions, leading to a marked have been delayed by an internal energy source to explain the increase of the binary fraction in the central regions. Outside GC longevity, and binaries could play this role (see Hutetal. thehalf-massradius,theprimordialbinaryfractioniswellpre- 1992,2003, forareview).Inthecore,binariesaresubjecttoen- 2 Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 Table 1. Globular cluster parameters from Harris (1996, up- 2. X-rayobservationsanddataprocessing datedFeb.2003) NGC 2808 was observed on February 1st 2005, for 41.8 kilo- seconds(ks)withthethreeEuropeanPhotonImagingCameras Parameters NGC2808 NGC4372 (EPIC MOS1, MOS2 and pn) on board the XMM-Newton ob- RightAscension(J2000) 9h12m02.6s 12h25m45.4s Declination(J2000) −64◦51′47′′ −72◦39′33′′ servatory, in imaging mode, using a full frame window and a Distance[kpc] 9.6 5.8 mediumfilter. Coreradius,r [′] 0.26 1.75 The observation of NGC 4372 was performed under the c Halfmassradius,r [′] 0.76 3.90 same conditions on March 23–24th 2005, for 29.7 ks, but the h Tidalradius,r [′] 15.55 34.82 MOS1CCD6wasinoperative(micro-meteoriteeventonMarch t rc/rh 0.34 0.45 9th2005),andthepndatawaslostduetotechnicalissues. Half-massrelaxationtime[yr] 1.35×109 3.89×109 Massa[M⊙] 1.46×106 3.08×105 Metallicity,[Fe/H] −1.15 −2.09 2.1.Datareductionandfiltering ⊙ WeprocessedthedatausingtheXMM-NewtonScienceAnalysis a calculated from the relation M/M⊙ = 3×100.4(Mv −Mv) using the Systemv7.0(SAS).Weusedtheemprocandepprocscriptswith absolutevisualmagnitude. themostrecentcalibrationdatafilestoreducetheEPICobserva- tiondatafiles(ODFs).Duringthisstep,badeventsmostlydueto badrows,edgeeffects,andcosmicrayeventswereflagged.We countersandhardbinariesbecomeharderwhiletransferingtheir filteredthethreeresultingeventlistsforeventpatternsinorderto energyto passing stars. This scenario leads to a globalheating maximisethesignal-to-noiseratioagainstnonX-rayevents.We of the core, and even a small population of close binaries can selected only calibrated patterns, i.e. simple and double events drivethe evolutionofthe entirecluster(Hutetal. 1992).Some for pn data and single to quadruple events for MOS data. For GCscouldalsocontainanintermediatemassblackhole(IMBH) pneventsbelow500eV,weselectedonlysingleeventsbecause of ∼ 103M⊙ or more in their core to explain the distribution in this energy band non X-ray events also affect double events of stars in some clusters, and the stability onlarge time scales. (Ehleetal.2006, §3.3.7). ThepresenceofanIMBHwasclaimedforM15(Gerssenetal. Basedonthelightcurveofsingleeventsexceeding10keV 2002, see also Ho et al. 2003), for G1 in the galaxy M 31 we identified periods of high background, due to soft proton (Gebhardtetal.2002, 2005),andforextragalacticGCshosting flares,andselectedgoodtimeintervalsfortheobservation.For anultraluminousX-raysource(Maccaroneetal.2007). NGC 2808, this operation leads to 38.0, 37.3 and 30.2 ks of ThetargetsstudiedinthispaperaretwoverydifferentGCs. clean observation for MOS1, MOS2 and pn respectively. For NGC2808isamassiveandconcentratedcluster.NGC4372on NGC 4372, the observation was highly affected by flares and thecontraryislessdensewithfewerstars,beingalsoaverylow only15.7ksforMOS1 and17.2ksforMOS2 remainafter fil- metallicitycluster.TheirparametersarelistedinTable1. tering.Wenotethatflaringactivityiscontinuousduringtheex- NGC2808hasalreadybeenstudiedinX-rayswiththeGIS posure and events with energy above 2 keV are affected by a instrument on board ASCA observatory. Only one source was highnoise,evenafterfiltering. reported in the GIS catalogue (Uedaetal. 2001) with a 1.35′ error circle located at 7.5′ from the cluster center. No sources 2.2.Sourcedetection were reported during the ROSAT All Sky Survey observations (Vogesetal. 1999). The core of NGC 2808 was also observed Thelistofeventswasdividedintothreeenergybands(0.5–1.5, withtheSpaceTelescopeImagingSpectrograph(STIS)onboard 1.5–3,and3–10keV)toallowustoderivespectralcolours.The the Hubble Space Telescope (HST) in 2000 with ultra-violet sourcedetectionwasdoneforallavailabledatasimultaneously (UV) filters F25QTZ (far-UV band centered at 159 nm) and (MOS1,MOS2andpnwhereavailable). F25CN270 (near-UV band centered at 270 nm). Dieballetal. We performed the source detection using the script ede- (2005) looked for white dwarfs (WDs) and CVs in this data. tect chainwhichfirstcalculatesthelivetime,thevignetting,the They found ∼40 WD and ∼60 CV candidates in the field of sensitivity map and the backgroundmap for each detector and view. Two of the CV candidates are variable (UV 222 and eachenergyband,andthencallsaslidingboxalgorithm.Finally UV 397), and another has an optical counterpart (UV 170). weranthetask emldetect.Foreachsource,thetaskperformsa NGC 2808 has also been observed in detail in the optical, pointspreadfunction(PSF)fitting,foralltheavailabledetectors and since Harris (1974), it has been known that the horizon- andforthe threebandssimultaneously,refinesthe coordinates, tal branchin NGC 2808is unusual(see also Bedinetal. 2000; andgivesthecountrates,thehardnessratios,andthemaximum Carrettaetal.2006).Themainsequenceisseparatedintothree likelihood(ML)ofeachsourcecandidate. branches, possibly due to successive rounds of star formation, Thefluxeswereobtainedbyprovidingtheenergyconversion withdifferentheliumabundances(Piottoetal.2007). factors(ECFs,inunitsof1011countcm2erg−1)whichallowthe In the optical, the colour-magnitudediagram of NGC 4372 directconversionofcountrates intofluxes(flux= rate / ECF). indicates an old cluster (15±4 Gyr), with high reddening,but Thesefactorswerecalculatedineachenergybandandforeach nospecialfeatures(Alcainoetal.1991).Kaluzny&Krzeminski detector by extracting an on-axis source and generating detec- (1993) found 19 variable stars, of which one has a light curve tor responsefiles for the source(using rmfgen andarfgen SAS consistent with an eclipsing CV of period 0.4 days. In the X- tasks). These response files were used to create a fake spec- ray,theROSATObservatory,withtheHRIinstrument,detected trumcorrespondingtoacommonmodel:apowerlawspectrum 19 sources (Johnstonetal. 1996), of which 9 fall in our field with Γ = 1.7 (mean spectrum of detected sources) and the ab- of view. All these sources are located outside of the half-mass sorption (N ) of the cluster. For NGC 2808, the absorption of H radius of the cluster, and none are consistent with the variable 1.2 × 1021cm−2 was calculated from the reddening of optical starsdetectedintheoptical. observations(Bedinetal. 2000) with the relation computedby Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 3 Burstein&Heiles (1978). For NGC 4372, the reddening esti- matedbyAlcainoetal.(1991)allowsustodetermineanabsorp- tionof2.8×1021cm−2.Finally,ECFswerecalculatedbydivid- ingthecountratesofthefakespectrumbythemodelfluxes.The obtainedfluxeswerethen convertedto unabsorbedfluxes.This methodgivesareliableestimationoftheECFsweneedforour energybands. To give an idea of the errorson these values,by changingthe spectralindexofthe modelto Γ = 2.0,thefluxes changedby1.5%,2%,and8%,intheenergybands0.5–1.5,1.5– 3and3–10keVrespectively. We processedtheentirefieldofviewwithaspatialbinning factor of 80, giving imageswith square pixels of side 4′′. This givespixelsof a similar size to the pn pixelsize (4.1′′×4.1′′), but larger than the MOS pixels (1.1′′×1.1′′). The binning is sufficiently small to sample the PSF of the two detectors cor- rectly, where the pn and MOS PSF FWHM (Full Width Half Maximum)are6′′ and5′′ respectively(Ehleetal.2006, §3.1). Weusedaslidingboxof5×5pixelstodetectthesources,and selectedsourceswithaMLgreaterthan10(4σdetection). ForNGC2808,thisleadstothedetectionof92sources,all of which were visually verified for each detector. Five sources falloutsidethepndetector,fourforMOS1andthreeothersfor MOS2. In the central region of the detectors, along the line of sightofthecenteroftheGC,wenotedthatasourcehasacom- plexform.Inthisregion,thePSFisnarrowerandbetterdefined Fig.1. Combined image of the XMM-Newton observation of and the vignetting is the lowest. We therefore reprocessed the NGC2808.Thethreecenteredcirclesshownrepresentthecore, datawithaspatialbinningof40togetsquaredpixelsofside2′′ half-massandtidalradii.Thedetectedsourcesareplottedwith and thus better sample the PSF for MOS detectors. We listed their 90% error circles. We used a spatial binning factor of 80 fromvisualinspectioneightpossiblesources,andusedthetask andtheimagewassmoothedwithaGaussianfilter.Forthisrea- emldetecttosimultaneouslyfitthosesourcecandidates.Wede- sonthecoreappearsblurry,azoomisshowninFig.2. tectfiveadditionalsourcesataminimumof4σ(C1toC5)inthe halfmassradiusofthecluster,ofwhich3fallinsidethecorera- dius.Threeothersaredetectedat2.5σ.Wehaveatotalnumber of 96 sources detected above 4σ. Their properties are listed in Table2,thecombinedimageispresentedinFig.1,andacolour image,withazoomofthecenter,inFig.2. For NGC 4372, 10 sources were detected. Three sources were detected with the MOS2 only, in the region where the MOS1 CCD6 fell. None of these sources are located inside the half mass radius. The X-ray source properties are listed in Table3,andthecontourmapispresentedinFig.3. 3. Ultra-violetobservationsanddatareduction AlongwiththeEPICinstruments,theOpticalMonitor(OM)on boardtheXMM-Newtonobservatoryperformedthreeexposures of 4000 seconds for both NGC 2808 and NGC 4372 with the UVM2filter,centeredat231nmintheUVband.Forthisfilter, thePSFFWHMis1.8′′andthefieldofviewisapproximatively 16′×16′. The UV data was processed with the SAS task omichain. Thisscriptremovesbadpixels,performsspatialcalibration,and source detection for each image. For each source, the instru- mental magnitude is evaluated from the count rate. Finally the resulting images and source lists are merged. We considered a significancethresholdof3σfordetectedsources. Fig.3. Contour map of the XMM-Newton MOS2 observation ForNGC2808,wedetected598sourcesatalimitingUVM2 ofNGC4372.Coreandhalf-massradiiareshown.Thedetected magnitude of 19.3. The region inside the half-mass radius is sourcesareplottedwiththeir90%errorcirclesandwithcontours overcrowded and poorly resolved. Approximatively 45 X-ray at 3, 5 and 10σ. Small circles are XMM-Newton sources and sources fall inside the OM field of view, for which we found biggerdashedcirclesareROSATsourceswiththeirerrorcircle nine matching UV sources. The properties of the UV counter- asreportedbyJohnstonetal.(1996). partsarelistedinTable4.WetookintoaccounttheX-rayposi- tionaccuracyofthesources(∼ 3.6′′,90%errorcircles)andthe UVpositionaccuracy(∼ 2.0′′,90%errorcircles).Wethuskept 4 Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 Fig.2. Top:XMM-NewtonobservationofNGC2808.Colourscorrespondtodifferentenergybands,red:0.5–1.5keV,green:1.5– 3keV,blue:3–10keV.Thefieldofviewis30′across.Thebinningfactoris40andtheimagewassmoothedwithaGaussianfilter. Bottom:Azoomedcombinedimageofthethreedetectorsisshownontheleftandtheimageontherightshowsareconstructionof thecorewitheightsources(onlyfivearedetectedat4σ,theotherthreearedetectedat2.5σ).Thecoreandthehalf-massradiiare shown. Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 5 Table4.ListofUVcounterpartsintheNGC2808fieldofview. TheX-raysourceIDisgivenwiththepositionoftheUVsource (90% error is 2.0′′), the offset position between UV and X-ray source,theoffsetfromtheclustercenter,andtheUVM2magni- tude. ID RA Dec Offset Offset UVM2 2000 2000 X-ray center 1 09h10m49.9s −64◦48′15.48′′ 1.18′′ 8.51′ 18.40±0.01 2 09h11m28.3s −64◦50′36.60′′ 1.61′′ 3.85′ 18.32±0.01 17 09h11m33.6s −64◦51′04.32′′ 0.54′′ 3.17′ 10.95±0.01 22 09h12m35.3s −64◦53′19.32′′ 2.06′′ 3.77′ 18.01±0.01 24 09h12m01.7s −64◦56′12.84′′ 1.69′′ 4.46′ 14.80±0.01 29 09h12m02.2s −64◦55′10.56′′ 2.54′′ 3.35′ 17.04±0.01 50 09h12m02.2s −64◦55′10.56′′ 2.52′′ 4.27′ 18.97±0.01 77 09h12m23.5s −64◦57′11.88′′ 1.63′′ 5.84′ 18.05±0.01 85 09h12m13.7s −64◦47′52.80′′ 1.24′′ 4.07′ 17.24±0.01 Fig.4. Log(N)−log(S)diagramintheband0.5–2keV.Sisthe limitingfluxandNthenumberofsources.Theempiricalrelation matchingUVsourcesthathaveanoffsetwiththeX-raysource ofHasingeretal.(2005)isreportedforcomparison. lowerthan5.6′′.TheX-raysourceswithUVcounterpartsareall locatedoutsidethehalf-massradius,andarethereforeprobably backgroundorforegroundsources. tool WebPIMMS1 v3.9b (Mukai 1993) to determine the mini- We performed a Monte-Carlo simulation to evaluate the mum detectable unabsorbed flux of each annulus. In the cen- probabilitythatthe superpositionofan X-raysourceanda UV ter, this limiting flux is F =4.1×10−15ergcm−2s−1, source occured by chance. We excluded the central region in- andgoesupto6.1×10−15e0.r5g−1c0mke−V2s−1inthelastannulus.This side a radiusof2.4′ wheretheUV sourcesare notresolved.In lead to a limiting luminosity of L =4.5×1031ergs−1 the remaining area the X-ray and UV sources appeared to be 0.5−10keV forsourcesinthecoreofNGC2808. uniformlydistributed.Thesimulationleadsto1.1±1.1sources We used the log(N) − log(S) relation derived by alignedfortuitously.Weconcludethat8±1oftheUVcounter- Hasingeretal.(2005)fromadeepobservationoftheLHinthe partscouldbeassociatedwiththeircorrespondingX-raysource energy band 0.5–2.0 keV. Our estimated minimum detectable withaprobabilityof99.9927%. fluxallowsustodeterminethenumberofsourcesexpectedper For NGC 4372, 272 sources were detected at a limiting square degree. We took into account the quadratic sum of two UVM2magnitudeof19.6.AbrightA0starinthefieldcausesan errors:anerrorof10%ontheflux,andanimprecisionof10% out-of-focusghostimage(smokering)andleadto∼10spurious onthelog(N)−log(S)relation.ResultsarereportedinTable5. detections.OnlythreeX-raysourcesfallintheregionobserved Asaconsistencycheck,weprocessedalongobservationof bytheOM,andnonehasaUVcounterpart. 90ksoftheLHthattookplaceonNovember27th2002.Thisob- servationusedafullframewindowandamediumfilterasforour observations. We performed exactly the same data processing 4. TheX-raysourcesinNGC2808 thatweusedforourobservations,andplottedthelog(N)−log(S) relationfortheenergyband0.5–2keV(seeFig.4).Wealsoplot- 4.1.MembersofNGC2808 tedthelog(N)−log(S)relationusingthesourcesdetectedinthe Most of the sources detected are background or foreground samebandwithacorrectionforthehigherabsorptioninthefield sources, but some of them are members of the cluster. If we ofviewofNGC2808.Theshapeofthecurvesaresimilar,and assume no cosmic variance in the distribution of sources in we cansee theflux detectionlimitin the differentfieldsdueto thesky(seeYangetal.2003),wecancompareourobservation exposuretime.Thelog(N)−log(S)relationisthereforeapplica- with onewithouta GC todeterminestatistically thenumberof bletoourdata. sources belonging to the cluster. We evaluated the number of TheresultsreportedinTable5indicatethatthefivesources backgroundsourcesbasedonobservationsoftheLockmanHole locatedinthecenterofthefieldofviewarelikelytoberelated (LH)with XMM-Newton.Inthis field, centeredonthe sky po- to the cluster. The probability of membership if we assume a sitionRA 10h52m43s,Dec -+57◦28′48′′,theabsorptionis Poisson distribution is 99.9985%. Moreover a possible excess 2000 2000 verylow,N =5.7×1019cm−2(Lockmanetal.1986). of sources appears in the annuli between radii 3.6′ and 6.5′ H We dividedtheXMM-Newtonfieldofviewintoseveralan- (4.5r to 8r ). The probabilityof membershipto the cluster is h h nulitotakeintoaccountthevignetting,whichbecomesmoreim- 96.75%. As we are dealing with low numbers, statistical fluc- portanttowardstheedgeofthefieldofview.Theannuliarecen- tuationsmightexplainthis excess, but we have nonethelessin- teredonthecenteroftheGC,whichisalsothecenterofthesen- cludedthesesourcesinouranalysis. sitivitymapofthedetectors,andtheirsizewaschosentoencir- cleatleast5detectedsources,andwhenpossible20sources.For 4.2.Spectralanalysis eachannulus,weevaluatedthebackgroundcountrateinseveral 15′′ radius regions without sources assuming a correction for We first plotted the hardness ratios obtained during the source thevignetting.Aminimumdetectablecountratewasestimated detection for sources with more than 50 counts and more than tobeequivalenttoaMLof10abovethisbackgroundcountrate. 3countsineachenergyband(Figs.5and6).Weshowthetracks We assumed a power law spectrum with Γ = 1.7 with the ab- sorptionofthefieldofview,andenteredthecountratesintothe 1 http://heasarc.gsfc.nasa.gov/Tools/w3pimms.html 6 Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 Table 5. Expectedand detectedsourcesforNGC 2808field of view. Annulus Expected Detected (′) 0.5–2keV 0.5–2keV 0–0.76 0.30±0.06 5 0.76–3.6 6.27±1.24 5 3.6–6.5 14.31±2.94 20 6.5–9 17.58±3.79 20 9–12 21.99±5.88 20 Fig.6. Colour-colour diagram of NGC 2808 sources. Same commentsasforFig.5. PO:powerlawwithphotonindices3,2.5,2,1.5,1,0.5. BR:thermalbremsstrahlungwithtemperatures1,5,10,15,20, 50keV. BB: blackbody spectrum with temperatures 0.1, 0.5, 1, 1.5, 2keV. NSA:neutronstarwithhydrogenatmosphere,mass1.4M⊙,ra- dius12km,distanceoftheclusterandlog(T )=5,6,6.5,6.8, eff 7. Fig.5. Flux-colourdiagramofNGC2808sources.Forclarity, onlythe30brightestsourcesareenumerated,thefluxvaluesfor source in a region withoutsources on the same CCD, with the all the sources are listed in Table 2. R is the count rate for the same vignetting, and when possible at the same distance from givenenergybandinkeV.Atypicalerrorbarisshownatthebot- CCDreadoutnodeforpnCCDs. tomright.Blacklinesrepresentthefollowingmodels,assuming Weusedthetaskevselectwithabinningof15eVforMOS anabsorptionof1.2×1021cm−2: data and 5 eV for pn data as recommended (Loiseau 2006, NSA:neutronstarwithhydrogenatmosphere,mass1.4M⊙,ra- §4.9.1).Then foreach spectrum we used rmfgen and arfgen to dius 12 km, distance of the cluster and log(T ) = 5.9, 6, 6.1, eff generate the instrumental response files for a point source, i.e. 6.2. the redistribution matrix file (RMF) and the ancillary response NSA PO: NSA and a power law with photon index of 1 and file(ARF).WefittedthedatawithXspecv11.3.2(Arnaud1996). contribuing50%oftheflux. Abinninggreaterthan20countsallowedtheuseoftheχ2 min- imizationcriterion,butwhenthere were insufficientcountsper bin (under 20 per bin), we used the Cash statistic (Cash 1979) ofsomespectralmodels.InFig.5,possibleqLMXBsarelocated whichprovidesagoodness-of-fitcriterionsimilartothatofχ2. ontheleftsideofthediagram,whileveryabsorbedsourcesare We triedsimple modelsincludedin Xspec such as a power seen on the right side of the diagram. The most luminous and law, a bremsstrahlung, a black body, a Raymond Smith or a absorbedsourceshavecoloursthataresimilartotheseofextra- mekalfit.Forverysoftsourceswetriedahydrogenatmosphere galactic objects as discussed in Sect. 6. In Fig. 6, the qLMXB model (Zavlinetal. 1996), assuming the distance of the GC, a candidatesshouldbelocatedatthe bottomleft,andCVs inthe massof1.4M⊙ andaradiusof12kmforaneutronstar.These middleofthediagramaroundthepowerlawbranchwithphoton parameterscorrespondtootherneutronstarsdetectedpreviously indices1to1.5andthebremsstrahlungbranchwithtemperatures inGCs(Webb&Barret2007;Heinkeetal.2003a).Whenitwas 10to50keVmbox(e.g.Richman1996;Baskilletal.2005). clearthatasimplemodelwasinsufficienttofitthedatawetried For the brightest sources, we extracted and fitted the spec- compositemodels.Theresultsofthespectralfittingareprovided tra. We used an extractionradiusof 30′′ wheneverpossible, so inTable6andarediscussedinSect.6.Wealsoprovidethespec- 80% of the encircled energy was included. In the crowded re- traofC1andC2(Figs.7and8). gion the extractionradius was reducedto 8′′, correspondingto AglobalbackgroundspectrumwasextractedforeachEPIC 50% encircled energy (Ehleetal. 2006, §3.2.1). A correction camera in order to locate possible features in the background is taken into accountin the instrumentalresponsefiles. We ex- that could be present in the source spectra. We selected all re- tracted a background with the same extraction radius for each gions without sources by excluding areas of 1′ radius around Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 7 Table 6. Best fitting modelsto the spectra of sources in the NGC 2808field of view. The unabsorbedflux is in the 0.5–10keV range[×10−14ergcm−2s−1]andforpossibleclustersourcestheluminosityisgiveninthe0.5–10keVrange[×1032ergs−1].The modelstriedareapowerlaw(PO),anabsorbedpowerlaw(APL),abremsstrahlung(BR),ablackbody(BB),aRaymond-Smith (RS),amekal(MK)andaneutronstarwithhydrogenatmosphere(NSA).Weusedmodelswithoneortwocomponents(Comp.1 andComp.2).TheabsorptionN gal[×1021cm−2]isfrozentothevalueofthecluster.Parametervaluesarephoton-indexΓofPO H model,temperaturekT [keV]ofBB, RSandMKmodels,andlog(T )[K]ofNSAmodel.χ2 andC givethegoodnessofthefit, eff reportedwiththenumberofdegreesoffreedom(dof). Src Flux N Model N Γ kT log(T ) Model Γ kT χ2 C dof Hgal H eff ID (Lum) Comp.1 Comp.2 C1 8.6±1.5 1.2 PO – 1.56±0.14 – – – – – – 37.54 43 (9.5±1.7) 1.2 BR – – 12.28+19.42 – – – – – 37.35 43 −5.03 C2 2.4±0.3 1.2 PO – 2.8±0.2 – – – – – – 51.37 43 (2.6±0.4) 1.2 NSA 0.98±0.04 – – 6.016±0.017 – – – – 56.45 43 1.2 NSA 0.82±0.40 – – 5.975±0.027 PO 1.56 – – 38.88 38 1 52.8±8.6 1.2 PO – 1.31±0.05 – – – – – 143.38 – 94 1.2 APL 4.3±1.8 1.7±0.2 – – PO 4.7±1.4 – 98.07 – 93 2 15.4±1.2 1.2 PO – 1.9±0.1 – – – – – 48.28 – 36 3 9.1±1.1 1.2 PO – 2.6±0.1 – – – – – 46.36 – 36 5 17.3±2.7 1.2 APL 35.03+18.63 1.65±0.60 – – – – – 27.17 – 20 −13.71 1.2 APL 49.46+23.53 1.918+0.89 – – PO 2.0 – 15.14 – 19 −10.57 −0.53 13 6.4±0.7 1.2 PO – 0.32±0.25 – – – – – – 53.95 38 1.2 APL 5.64+8.9 0.75+0.64 – – PO 9.5+0.5 – – 41.20 35 −4.93 −0.42 −0.6 17 2.1±0.6 1.2 PO – 2.1±0.3 – – – – – – 28.95 27 1.2 MK – – 2.7±1.0 – – – – – 30.63 27 22 1.1±0.6 1.2 BB – – 0.16±0.02 – MK – 76+4 – 49.54 40 −59 (1.2±0.6) 24 0.9±0.2 1.2 RS – – 0.40±0.09 – – – – – 29.62 27 (1.0±0.2) 25 0.9±0.2 1.2 BB – – 0.17±0.09 – MK – 6.19±6.1 – 18.71 25 (1.0±0.2) Fig.7. Spectrum of C1 (NGC 2808) fitted with a power law Fig.8. Spectrum of C2 (NGC 2808) fitted with a NSA model modelandtheabsorptionofthecluster. and the absorption of the cluster. The contribution of C1 also appearsasapowerlaw. detected sources. We produced the instrumental response files 4.3.Variabilityanalysis for a flat field. Based on this spectrum, we fixed the limits of the energybandswe used: 0.4–15keV for pn events, and 0.2– We performed variability analysis based on the pn data for 10 keV for MOS events. For all sources we could evaluate the sources with more than 300 pn counts, and for sources with fluxinthe0.5–10keVenergyband.ForMOS,twostronglines a fitted spectrum and more than 100 pn counts. We extracted areobservedat1.5and1.75keV(AlKαandSiKαlines).For lightcurves for sources and backgrounds (same regions as for pn,thebackgroundspectrumshowsastronglinefeaturearound spectral analysis), adjusting the binning for each source to ob- 1.5keV(AlKαline),fainterlinesat8 and8.6keV (Culines), tain a mean of 20 countsper bin after correction.We removed andsomefeaturesat9.7and11keV(asseeninEhleetal.2006, thefirst12ksoftheobservationasthispartisaffectedbyflares. §3.3.7). We corrected the pn events for losses due to e.g. vignetting or 8 Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 Fig.9. Lightcurveof source1 (NGC 2808)andcorresponding goodtimeintervals(GTI)asdefinedinSect.2.1.Starttimet is 0 February1st 20054:57:28(MJD2453402.70657). Table7. VariabilityanalysisofNGC2808sources.Wegivefor eachsourcetheBinsize[s],theKolmogorov-Smirnovprobabil- ityofthesourcelightcurvebeingdifferentlydistributedthanthe backgroundlightcurve(K-S),andtheχ2withthenumberofde- greesoffreedom(dof)forthebackgroundsubtractedlightcurve fittedwithaconstant. Fig.10. Flux-colour diagram of NGC 4372 sources detected withXMM-Newton(crosses,unitsontheleftaxis)andROSAT ID Binsize K-S χ2 dof (diamonds, units on the right axis). R is the count rate for the C1 2400 0.15 6.08 10 given energy band in keV. Typical error bars are shown at the C2 3000 0.31 20.05 9 bottomleft. 1 300 1.05×10−4 120.90 82 2 600 0.23 48.71 39 Table 8. Best fitting models to spectra of sources in the 3 600 0.04 67.87 39 NGC4372fieldofview.SameasTable6. 5 1200 0.33 31.90 21 13 2000 0.86 23.65 13 Src Flux N Model Γ kT χ2 dof 17 2400 0.15 31.08 10 Hgal 1 148±10 2.8 PO 4.1±0.2 – 155.94 113 2.8 BB – 0.14±0.01 153.70 113 2 147±10 2.8 PO 4.1±0.2 – 100.40 79 filterswiththeSAStasklccorr.Thetimevariabilityisexamined 2.8 BB – 0.14±0.01 99.02 79 withaKolmogorov-SmirnovtestusingIDL/Astrolib2procedure kstwo by comparing the source lightcurve to the background lightcurve.We also fitted the backgroundsubtracted lightcurve WebPIMMS) is F ∼5±2×10−14ergcm−2s−1, so we withaconstantvalue,andcalculatedtheχ2ofthefit.Theresults 0.5−10keV expecttodetectallROSATsourcesinthisregion.Wenotethat arereportedinTable7.Onlysource1seemstobevariablewitha ourfluxdetectionlimitiscomparabletotheROSATlimit,dueto significantprobability.ThelightcurveispresentedinFig.9and largecutsafterflarefilteringandtheabsenceofpndata,butwe discussedinSect.6.6. increasetheangularresolutionandclearlyresolvethesources. WeperformedthesameanalysisasinSect.4.2.We carried 5. TheX-raysourcesinNGC4372 outspectralanalysisfortwosourceswithsufficientcounts,inthe samemannerasdescribedinSect.4.2.Theresultsarereported Johnstonetal. (1996) found 9 sources in NGC 4372 with the inTable8. ROSAT X-ray observatory in an equivalent region of the sky WecomparedhardnessratiosgivenbyJohnstonetal.(1996) coveredbyourXMM-Newtonobservation.Twoofthesesources totheonesweobtainedusingthesameenergybands.Weplotted werenotdetectedinourobservations(R8andR10),andwede- forbothROSATandXMM-Newtondataaflux-colourdiagram tectthreeadditionalsources(7,8and9).Threemergedsources (Fig. 10). The two distributions are well correlated within the intheROSAT image(R5,R7 andR8) areclearlyresolvedinto errorbars,especiallyforbrightsources,butwenotethatsource two bright sources (1 and 2) and one faint and more diffuse 4seemsharderinourobservation. source(9),seeFig.3. We could not use a log(N)−log(S) relation to discuss the distribution of sources, as in Sect. 4.1 for NGC 2808, because 6. Discussion ofthelownumberofsourcesdetectedwhichleadstolargeerror We discuss here the presence of objects expected in globular bars. Moreover the high background noise due to high flaring clusterssuchasqLMXBs,CVs,MSPsandABs,addingnoteson activitydoesnotallowusto estimateaccuratelythefluxdetec- individualsources.We alsodiscusssomebackgroundandfore- tionlimit.Fromthefaintestsourcesdetected,thislimitisaround groundsourceswhichpresentunusualfeatures. F ∼3±1× 10−14ergcm−2s−1, which leads to a lim- 0.5−10keV itingluminosityof L ∼1032ergs−1 forasourcein the 0.5−10keV coreofthecluster.Fromthecountrateofthefaintestsourcesde- 6.1.LowmassX-raybinariesinquiescence tectedbyROSATinthisregion,theirlimitingflux(obtainedwith For 18 GCs that have been observed deep enough to detect 2 http://idlastro.gsfc.nasa.gov/ all qLMXBs, we plotted the number of qLMXBs against the Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 9 approximate encounter rate for a virialized system (ρ1.5r2), as to the fact that CV1 emission may be complex emission from 0 c done in Gendreetal. (2003a). According to this correlation, severalsources,asitsluminosityisquitehighforasingleCV. one would expect 3 ± 1 qLMXBs in NGC 2808, and none in NGC4372. NGC2808–C3,C4,C5. Thesesourcesarelocatedinthehalf- For NGC 2808 we detect one source consistent with a mass radius and the colours indicate that they are likely to be qLMXB (C2, described below). For NGC 4372, a qLMXB CVs (Fig. 6 and 5, and Table 2). Their luminosity, if they are having the minimum luminosity of the 21 qLMXBs reported membersoftheclusterisconsistentwiththishypothesis. by Heinkeetal. (2003b) would have been detected. The lack of detection is therefore consistent with the prediction of zero qLMXBinthiscluster. 6.3.MillisecondpulsarsinNGC2808? Some GCs are known to harbour a large population of MSPs. NGC 2808 – C2. This source is located in the core radius of In 47 Tuc, which is similar in mass to NGC 2808, a popula- NGC2808.Itisclosetoabrightsource(C1,15′′)andthePSF tionof∼25MSPsisestimated(Heinkeetal.2005).Thebright- wingsaremerged.FromEhleetal.(2006, §3.2.1),weestimated est MSPs have luminosities around our flux detection limit for that10%oftheemissionofC1waspresentintheC2spectrum. NGC2808.However,wenotethatMSPs,whicharehardX-ray We fitted a modelcomposedof 10%of the C1 spectrum anda sources,aredifficulttodistinguishfromCVsatthisluminosity hydrogenatmospheremodelfora neutronstar. Theparameters (Bogdanovetal.2005).Therefore,ourobservationsarenotcon- aregiveninTable6andthespectruminFig.8.Thehardemis- strainingforthepresenceofMSPsinNGC2808.ForNGC4372, sioninthisspectrumappearstobeduetoC1.We notethatthe theirfluxistoofainttobedetected. unabsorbedluminosity(2.6±0.4×1032ergs−1 ifitbelongsto the cluster), and the X-ray spectrum of C2 are consistent with 6.4.OthersourcespossiblylinkedtoNGC2808 theqLMXBhypothesis. We detect 5 sources with a hardness ratio below −0.7 in the flux-colour diagram presented in Fig. 5, and located 6.2.Cataclysmicvariables ∼5−6×r away.Suchsourcescouldexplainthepos- h[NGC2808] sibleexcessofsourcesfoundinSect.4.1ifsomearelinkedtothe Ivanovaetal. (2006), using numerical simulations and taking into account different CV formation channels, predict about cluster. There is a highprobabilitythat these sourcesare back- groundsources,butif theyare linked to the cluster, theycould 200CVs fora GCsimilar toNGC2808.Of these,onlya frac- have been formedfrom primordialbinariesand remain outside tioncanbedetectedinourdata.Fromtheempiricalfunctionof ofthehalf-massradius(Hurleyetal.2007). Pooley&Hut (2006) derivedfrom fitting the numberof bright CVs (at luminosities above L =4.25×1031ergs−1) Dempseyetal.(1993)presentedastudyof44RSCVnsys- 0.5−10keV tems(ABs)observedwiththeROSATobservatory,andshowed against the specific encounter frequency, we estimate a pop- ulation of 20+20 bright CVs in NGC 2808. Given the X- that their spectra are very soft with most of the emission be- −10 low2keV.Fromtheirspectra(Table6),coloursandluminosity, ray unabsorbed luminosity of the core (excluding C2) of L =1.4×1033ergs−1, we cannotexpectmorethan30 sources24and41havesimilaritieswiththebrightestABs. 0.5−10keV brightCVsinthisregion. Evans&Hellier (2007) used a black body model to fit the softexcessbelow2keVfoundinthespectraofsomeintermedi- We detect4 CV candidatesinNGC 2808withluminosities down to L ∼3.0×1032ergs−1. A hidden unresolved atepolars,thehardemissionofintermediatepolarsbeingfitted 0.5−10keV with a mekal model. Sources 22, 25 and 50 have soft spectra population of fainter CVs may exist in the core of the clus- with a hard tail (Table 6), and could therefore be intermediate ter, but only the brightest have been detected in our observa- polars. tion. In particular, we note that the source C1 has a luminos- ity (L =9.5×1032ergs−1 if it belongs to the cluster) Some of the soft sources could also be active stars in the 0.5−10keV foreground.TheX-rayfluxhereisconsistentwithanestimated higherthantheknownluminositiesofCVs(Verbuntetal.1994), distanceof∼100pc. and could be composedof severalsources. After removingthe contribution of C2 from the core global spectrum we found a spectral photon-indexof Γ = 1.44±0.09, consistent with CV 6.5.LookingforanIMBHinNGC2808 emission(e.g.Baskilletal.2005), andonlythebrightestMSPs emission(Bogdanovetal.2005).Thissupportstheideathatthe FollowingTrenti(2006),NGC2808isagoodcandidateforhost- coreemissionisduetoamajorityofCVs. inganIMBHastheratiorc/rhislargerthanthecriticalthreshold of 0.3. This cluster may howevernot be sufficiently relaxed as InNGC4372whichisalessmassivecluster,theestimation the age is only7.4 times the half-massrelaxationtime (Hurley leadstolessthanonebrightCV.IfCVsarepresentinthecluster, 2007). theymustbefainterthanourlimitingluminosity. IfsuchanIMBHexistsinNGC2808,itshouldbelocatedat the center of mass of the cluster due to mass segregation. We NGC 2808 – C1. Source C1 is located in the core ra- found no evidence for an X-ray source at this position but it dius of NGC 2808. This source is hard, and is well fit- maybebelowourdetectionlimit.WeassumethattheBHisfed ted with a power law model or a bremsstrahlung model (see by intracluster gas with a density of ∼ 0.5cm−3 derived from Table 6). These values are consistent with CV X-ray emis- Pfahl&Rappaport(2001)withNGC2808parameters.Itshould sion(e.g.Richman1996;Baskilletal.2005).C1alsohasaUV benoted,however,thatapossibledetectionof200M⊙ ofneu- counterpart (UV 222) proposed to be a CV by Dieballetal. tralhydrogeninthecoreofNGC2808hasbeenreportedinthe (2005),strengtheningthishypothesis.Thespectrumisplottedin literature (Faulkneretal. 1991). If this is confirmed, it implies Fig.7.At2.0keV,afeaturethatatfirstsightcouldbeanemis- thatthe gasdensityis underestimatedhere.Inthesame wayas sion line is not well fitted with a Gaussian. It is possibly due Hoetal. (2003), if we assume a BH of 1000M⊙ accreting at 10 Servillatetal.:XMM-NewtonobservationsofNGC2808andNGC4372 thefullBondirate(Bondi1952)andanopticallythick,geomet- spectrashowasoftexcessbetween0.5and0.9keV,possiblydue ricallythindisk(Shakura&Syunyaev1973),wefoundanX-ray toreprocessedemissionofanabsorbedAGN. accretionluminosityfiveordersofmagnitudeaboveourlimiting luminosity. However, the BH may be radiatively inefficient as NGC 4372 – 4. This source appears to be harder than the foroptically thinadvection-dominatedaccretionflow (ADAFs, source R13 detected previouslywith ROSAT at the same loca- see e.g. Narayanetal. 1998). Following Grindlayetal. (2001), tion (within the error circles). With ROSAT, the photon-index ourlimitingluminosityimpliesanupperlimitof∼ 290M⊙ for correspondingtothehardnessratiois∼ 2,andforourobserva- acentralIMBHinNGC2808. tion,itgoesdownto∼0.5,ifweassumethesameabsorption.If weassumethesamephoton-indexforthetwoobservations,then 6.6.OtherNGC2808sources theabsorptionhasdoubled. NGC 2808 – 1. This source is the brightest source in the field of view, located 8.5′ away from the center of NGC 4372 – 7, 8 and R10. In the contourmap publishedby NGC 2808. A source was previously listed in the ASCA Johnstonetal. (1996), we can see some unresolved features at source catalogue (Uedaetal. 2001) with a compatible po- thepositionsofsources7and8.Wecanestimatethattheirfluxes sition. The unabsorbed flux is lower but consistent with (Table3)haveincreasedbyafactor12and4forsources7and8 our detection (FASCA =3.0±0.7×10−13ergcm−2s−1 and respectively.ROSATsourceR10isnotdetectedinourdata.As 0.7−7keV FXMM =4.0±0.5×10−13ergcm−2s−1). weshouldhavedetectedallROSATsources,itmayhavevaried 0.7−7keV Thissourceiswellfittedwiththeabsorbedpowerlawmodel between the two observations by a factor 2 to become fainter (APL+PO) used by Mainierietal. (2007) for Seyfert 2 active thanourlimitingflux. galacticnuclei(AGN)spectra.ItiscomposedoftheGalacticab- sorption,anabsorbedpowerlaw,andanunabsorbedpowerlaw tomodelthesoftexcessbelow0.9keV.Thelightcurveplottedin 7. Conclusions Fig.9istypicalofanAGN,showingshort-termlowamplitude We have presented XMM-Newton observationsof the globular variability(seeforinstance Gliozzietal.2004).Wenotethata clustersNGC2808andNGC4372. UV counterpartisfoundintheOM data.Thesefeaturesareall ForNGC2808,wehaveshownthatthefivecentralsources consistentwithanAGN,butanopticalorinfraredcounterpartis arelikelytobelinkedtothecluster.Oneoftheseisverylikelyto neededtoconfirmthenatureofthissource. beaqLMXB,andtheemissionoftheremainingcentralsources is consistent with 20 ± 10 bright CVs (at luminosities above NGC 2808 – 5 and 13. These sources show very absorbed L0.5−10keV =4.25×1031ergs−1),ofwhich4aredetectedinour spectra, well fitted with the APL+PO model (Mainierietal. data. A summary of these sources is presented in Table 9. We 2007). These spectra could indicate Seyfert 2 AGN, as for expectto resolvemoreobjectsin the coreof NGC2808in our source1. Chandraobservation(Servillatetal.2008). For NGC 4372, we detect no sources in the half-mass ra- dius,butthelimitingluminosityofourobservationsisnotcon- NGC 2808 – 17. This source is located 3.2′ away from the straining,inparticularforapossiblepopulationoffaintCVs.We center of NGC 2808. We found a possible optical counterpart, comparedoursourcesoutsidethehalf-massradiustopreviously HD 79548 at RA 9h11m33.293s, Dec −64◦51′03.28′′. 2000 2000 detectedsourceswith ROSAT in the field of view, andfounda This star is an A0V star with magnitudes B=10.42 and verygoodcorrelationformostofthesources. V =10.15.AUVcounterpartisobservedintheOMdatawitha magnitudeof10.95,consistentwiththeemissionofHD79548. Acknowledgements. Thisworkisbasedonobservations obtainedwithXMM- Using the distance and the visual magnitude of Vega, a well Newton, an ESA science mission with instruments and contributions directly known A0V star (7.76 pc, V =0.03), we estimated a distance fundedbyESAMemberStatesandNASA.WethanktheCNESforsupportof of∼ 800pcforHD79548,andderivedanX-rayluminosityof theoperationalphaseofthemission.ThisresearchhasmadeuseoftheSIMBAD L =4.9±0.9×1029ergs−1. database,operatedattheCDS,Strasbourg,France.WearegratefultoJ.Grindlay 0.5−10keV whosecommentshavehelpedtoimprovethispaper. The relation between the X-ray source and HD 79548 is unclear and it is possible that we observe a background sourcealignedwithHD79548.However,youngA0Vstarswith References an age less than 107 yr can be such a bright X-ray source (e.g.Peaseetal.2006;Tout&Pringle1995).Anactivestarasa Alcaino,G.,Liller,W.,Alvarado,F.,&Wenderoth,E.1991,AJ,102,159 companionis also consistentwith the X-ray emission obtained Arnaud, K. A. 1996, in ASP Conf. Ser. 101: Astronomical Data Analysis SoftwareandSystemsV,ed.G.H.Jacoby&J.Barnes,17 (Briggs&Pye 2003; Golubetal. 1983). Optical spectroscopic Baskill,D.S.,Wheatley,P.J.,&Osborne,J.P.2005,MNRAS,357,626 observationsofthissourcecouldhelptodeterminethenatureof Bedin,L.R.,Piotto,G.,Zoccali,M.,etal.2000,A&A,363,159 thisobject. Bogdanov,S.,Grindlay,J.E.,&vandenBerg,M.2005,ApJ,630,1029 Bondi,H.1952,MNRAS,112,195 Briggs,K.R.&Pye,J.P.2003,MNRAS,345,714 6.7.PeculiarNGC4372sources Burstein,D.&Heiles,C.1978,Astrophys.Lett.,19,69 Carretta,E.,Bragaglia,A.,Gratton,R.G.,etal.2006,A&A,450,523 AsthesourcesaredistantfromthecoreoftheGC,theyaremore Carretta,E.,Gratton,R.G.,Clementini,G.,&FusiPecci,F.2000,ApJ,533,215 likelytobebackgroundsources. Cash,W.1979,ApJ,228,939 Colpi,M.,Mapelli,M.,&Possenti,A.2003,ApJ,599,1260 D’Amico,N.,Possenti,A.,Fici,L.,etal.2002,ApJ,570,L89 Dempsey,R.C.,Linsky,J.L.,Schmitt,J.H.M.M.,&Fleming,T.A.1993,ApJ, NGC 4372 – 1 and 2. These two soft sources appear to have 413,333 similar parameters (Table 8). They are around 12′ away from Dieball,A.,Knigge,C.,Zurek,D.R.,Shara,M.M.,&Long,K.S.2005,ApJ, thecenterofNGC4372,andare2′awayfromeachother.Their 625,156