Determining an accurate position for a submm galaxy (SMG) is the crucial step that enables us to move from the basic properties of an SMG sample - source counts and 2-D clustering - to an assessment of their detailed, multi-wavelength properties, their con

Identi cationofsubmillimetregalaxiesintheSHADESSourceCatalogue

13

z

HG4.1S/mµ42S z

Figure7.RatioofS24µm/S1.4GHzasafunctionofredshift,z,forSHADESsourceswithrobustcounterparts( lledcircles:LH;emptycir-cles:SXDF).Thosewithoutspectroscopicredshifts–themajority–areplottedarbitrarilyatz=2.3.ThetracksofArp220,Mrk231andNGC6240areshowntogetherwithasampleoffaintradiosourcesinSXDF(smalldots–Ibaretal.,in

preparation).

Figure8.Log10S24µm/S850µmversuslog10S1.4GHz/S850µmforSHADESSMGswithbothmid-IRandradioidenti cations( lledcircles),withonlyradioidenti cations(squares)andwithonlymid-IRidenti ca-tions(opencircles).Arepresentativeerrorbarisshown,lowerleft.TheredshiftparameterisationofChapmanetal.(2005)isshownasahorizontalbaratlog10S24µm/S850µm=0(see§7.2).

Summarisingtheseplots,thebestavailablecomplementarydataintheLH–equivalenttothoseavailableintheGreatObserva-toriesOriginsDeepSurvey(GOODS)northern eld–allowsustoidentifyrobustlyovertwothirdsofSMGstocurrentsubmmdetec-tionlimits.Theobservedtrendsinidenti cationrategivenostrongrationaleforrejectinganysourcesfromtheparentSHADESSourceCatalogue,althoughaslightquestionmarkisthrownoversomeofthelowestSNRsources.

c0000RAS,MNRAS000,000–0007CONSTRAINTSFROMSPECTRALINDICES

7.1S24µm/S1.4GHz

Sincethespectralslopesat24µmand1.4GHzaresimilar,itmayproveinstructivetoexaminethebehaviourofS24µm/S1.4GHzasafunctionofredshift,asshowninFig.7.Weexpectthisplottobe-trayAGNcontributionstotheradio uxdensityinso-called‘radio-excessAGN’(Drakeetal.2003;Donleyetal.2005)or,conversely,‘mid-IR-excessAGN’whichhaveQSO-heateddustbutlittleornoAGN-relatedemissionintheradio.Forstar-forminggalaxiesthisratioistightlyconstrainedouttoz=1(Appletonetal.2004).GalaxieswithlowvaluesofS24µm/S1.4GHz,i.e.thosewithstrongradiowithrespectto24-µmemission,areunlikelytobedominatedbystarformation.

TheSHADESSMGsshareapproximatelythesamedistribu-tionofS24µm/S1.4GHzvaluesastheotherradiosourcesinSXDF(Ibaretal.,inpreparation).Fig.7showstheredshifttracksofArp220,NGC6240andMrk231–archetypalultraluminousIRgalaxieswithincreasingdegreesofAGNcontribution.MeasuredvaluesofS24µm/S1.4GHzfortheSHADESSMGsareconsistentwithanyoftheseSEDsbutMrk231isthepreferredtemplate,implyinganAGNcontributiontothemid-IRluminosity.Onlyatz<1couldthemostextremeSMGbeclassi edcon dentlyashavingaradioexcess.7.2S850µm/S1.4GHz

Hughesetal.(1998)andCarilli&Yun(1999)pointedoutthevalueofS850µm/S1.4GHzasanindicatorofredshiftforSMGs,atleastforz<3.Smailetal.(2000)andIvisonetal.(2002)werethe rsttoemploythetechniqueforsigni cantsamplesofSMGs, ndingmedianredshifts,z>～2.

Chapmanetal.(2005)foundthattherelationshowedalargedispersionfortheirsampleofradio-identi edSMGswithspectro-scopicredshifts,indicativeofarangeofSEDs.Theynotedthatapurelysubmm-selectedsampleshouldshowanevenwiderrangeofS850µm/S1.4GHzthantheirradio-identi edSMGs,sincetheneedforanaccurateradiopositionbiasesthesampleinredshiftandtem-perature.

Thesurprisingly attrendidenti edbyChapmanetal.,un-correctedforaprobableredshift-dependent～0.3dexshiftat-tributabletotheirradioselectioncriteria,wasparameterisedasS850µm/S1.4GHz=11.1+35.2z.Thisparameterisationwasnotintendedasacarefulphotometricredshifttechnique–ther.m.s.scatterinredshiftis～1,afterall–butlikelyremainsthebestwaytoestimatethemedianredshiftofradio-identi edSMGsamples.Applyingthistooursampleof65SMGswithrobustradiocoun-terpartsyieldsamedianredshiftof2.8,withaninterquartilerangeof1.3–3.8,somewhathigherandbroaderthanthespectroscopicredshiftdistributionreportedbyChapmanetal.(medianz=2.2,interquartile1.7–2.8,beforetheirsmallcorrectionfortheradiose-lectionfunction).TheChapmanetal.parameterisationisnotap-propriateforSMGswithoutradioidenti cations,butfortheentireSHADESsample(adoptingthelimitsinTables1–2forthoselack-ingformaldetections)itindicatesamedianredshiftof3.3.

ThedifferencebetweenthedistributionreportedhereandthatofChapmanetal.(2005)isquitemarked,butcanbeexplainedbyavarietyofeffects:spectroscopicbias; eld-to- eldvariations;strongclusteringoftheSMGpopulation(Blainetal.2004);ouradoptionofdeboosted uxdensitiesforallSHADESsources(alargeproportionoftheChapmanetal.sampleislikelytohavesuf-feredasubmm uxdensityboostofoneformoranother);and,not