1996Aj 111 . 1551M the Astronomical Journal

1996AJ 111 . 1551M THE ASTRONOMICALJOURNAL gests thattheywillnotbewellresolvedatÁ.21cm.Also, largely ignored,andisunderrepresentedinHisynthesis compact dwarf(BCD;Thuan&Martin1981)classhasbeen wards spiralanddwarfirregulargalaxies.TheHl-richblue galaxies usingHIasaprobeispredominantlyweightedto- double hornprofilesindicative ofextendedrotatingdisks, only 8%oftheBCDsinThuanandMartin’ssamplehave sizes, oftensmallerthantypicalsynthesizedbeamsug- studies. Thismaybebecausetheirsmallopticalangular similar galaxies,havebeenmade (e.g.,Tayloretal.1993, Some importantHisynthesis observations ofBCDs,and suggesting thattheyarenotideal candidatesforDMstudies. 1551 Astron.J.Ill (4),April1996 The questforunderstandingdarkmatter(DM)halosof © American Astronomical Society • Provided by theNASA Astrophysics Data System c3an 21-2 _1 -1 9 =1 Mount StromloandSidingSpringObservatories,TheAustralianNationalUniversity,PrivateBag,P.O.WestonCreek,ACT2611, Département dePhysiqueandObservatoireduMontMégantic,UniversitéMontréal,C.P.6128,Suce.“A”,Québec, Nevertheless itisclearthatthecoreofDMhalounusuallydense(po^O.lP~)dcompact <ÆIL=16 withinthelastmeasuredpoint.ThusNGC2915isoneofdarkestdiskgalaxiesknown.The rotation curveisderivedandfittedwithamassmodelconsistingofstellardisk,neutralgas NGC 2915doesnotobeytheTully-Fisher[A&A,54,661(1977)]relation,beingunderluminousforits Pressure supportisprobablysignificant,anditnotclearwhetherthecoreinequilibrium.Beyond complex HIdynamicsofthecentralregionresultsinahighuncertaintymanyfittedparameters. counterpart correspondstoacentralHIbar.Thedistributionandkinematicsarediscussedindetail.A over fivetimestheHolmbergradius,and22exponentialscalelengthinBband.Theoptical compact dwarf(BCD)galaxyNGC2915.Itisshownthat2915hastheHIpropertiesofalate-type This paperpresentsAustraliaTelescopeCompactArrayHIsynthesisobservationsoftheweakblue beyond theopticaldisk.©1996AmericanAstronomicalSociety. 2 i,crit^10cmadequatelydescribesthelocationofcurrentstarformation.AlthoughHIproperties optical disktheaverageHIlineofsightvelocitydispersionis8kms,whichnormalforgalaxies. dark matter(DM)halo.TheDMhalodominatesatnearlyallradii.totalmasstobluelightratio, stellar disk.SplitandbroadHilines(velocitydispersion^35kms)areseeninthecentralregion. spiral galaxy(Sd-Sm),includingadoublehornglobalprofile,andHIarms.Theextendsoutto BCDs, whichhavesimilarradialprofilesofHIdensityandvelocitydispersion,extendingwell of NGC2915areextremerelativetonormalgalaxiestheyappearlessincomparisonother [ApJ, 344,685(1989)],whenonlytheneutralgasisconsidered.AsimpleHIsurfacedensitythresholdof (R^l kpc).Theneutralgascomponent,withmass=1.27X10isprobablymoremassivethanthe Vrot8 kms“byafactorofnine.Italsodoesnotobeythestar-formationthresholdmodelKennicutt tb h c NGC 2915.II.ADARKSPIRALGALAXYWITHBLUECOMPACTDWARFCORE Department ofPhysicsandAstronomy,TheJohnsHopkinsUniversity,Baltimore,Maryland21218-2695 1. INTRODUCTION Electronic mail:[email protected],[email protected] Sylvie F.BeaulieuandKennethC.Freeman Received 1995December4;revised1996January16 Electronic mail:[email protected] Electronic mail:[email protected] 0004-6256/96/111 (4)/l551/15/$6.00 VOLUME 111,NUMBER4 Gerhardt R.Meurer Claude Carignan Canada H3C3J7 ABSTRACT Australia r_1 per I).ThereitwasshowntobeaweakBCD,thatisits this starformationisneartheverycenterofgalaxywith NGC 2915’sHirotationcurve, andfromit,constrainthe Previous singledishHIobservations indicatethatithasa these yieldsthedistanceD=5.3±1.6Mpc.Atthis of thecenter.NGC2915resolvesintostars;brightest integrated star-formationrateisonly0.05y-Mostof optical propertieswerepresentedbyMeureretal.(1994;Pa- vations presentedhereweremade primarilytodetermine very extendedHIdisk(Becker etal.1988).TheHIobser- some enhancedstarformationalongthemajoraxistoSE other thantosearchforDM. 1" correspondsto26pc,and1'1.54kpc. 1995, 1996;Hunteretal.1994)buttheirprimaryaimswere Here wepresentHiobservationsofNGC2915whose © 1996Am.Astron. Soc.1551 APRIL 1996 1996AJ 111 . 1551M 1552 MEURERETAL:NGC2915.IL how theHIpropertiesofthisBCDdifferfromothertypes lates starformationinBCDs. galaxies, andwhethertheygiveanyindicationofwhatregu- pressure supportcorrectionsapplied,andmassmodelsfitted. ISM areanalyzedinSec.4.Therotationcurveismeasured, structure ofitsDMhalo.Inadditionwewishtoexamine sents anoverviewoftheHIproperties.Thedynamics Fisher (1977)relationandthestar-formationthresholdlaw. comparison withothergalaxies,adiscussionoftheTully- Section 5presentsadiscussionoftheresultsincluding we discussapossibleinteractionpartnerofNGC2915. Our conclusionsaresummarizedinSec.6.IntheAppendix baselines—up to6km)werediscardedsincetheyproduced However, baselineswithantenna6(whichgivesthelongest Australia TelescopeCompactArrayusingallsixantennas. receiver wasemployedwiththecorrelatorsetto256chan- insignificant correlationamplitudes.ThedualpolarizationAT nels perpolarization,andwitheachchannelseparatedby figurations aregiveninTableI.Theintegrationtimewas20 to equipmentfailureandweather. AIPS package(ATNFversion).Therelevanttasksusedare s pervisibilitymeasurement.Somedataineachrunwerelost noted hereinparentheses.Badorsuspectdatawereedited were calibrated(CALIB)usingthesecondarycalibrator, out (TYFLAG,SPFLAG).Temporalgainandphasedrifts from theobservationsof1934—638and0407—658 min. Spectralvariationsinthecalibrationweredetermined 0906-682 whichwasobservedat50minintervalsfor3-6 15.7 KHz(3.31km/s).Theobservinglogsforthethreecon- calibrator 1934-638(SETJY,GETJY;fluxreference:Walsh UV plane(UYLSF)andtheresultant continuumandline two polarizationchannelscombined toformStokes“I”data set using0407-658whichisalsoafluxstandard(Walsh (BPASS). Theabsolutefluxlevelwassetbytheprimary ibility. Theline data setswereshiftedtotheheliocentric rest data setswereaveragedintime (UVAYG)to1minpervis- sets (SPLIT).Thecontinuumwas fittedandsubtractedinthe 1992). Thecalibrationwascheckedforthe1.5kmarraydata 1934—638 calibration.Thecalibrations wereappliedandthe 1992). Thecalibrationagreedtowithin2%withtheadopted The observationsarepresentedinSec.2.Section3pre- NGC 2915wasobservedwiththreeconfigurationsofthe The datawerereducedusingstandardsoftwareinthe © American Astronomical Society • Provided by theNASA Astrophysics Data System antenna 6. 11 config. UTdatebaselineonsource a 0.375 3May,199231-45911.34 0.75 C22Jan,199346-75010.52 1.5 D12Mar,1993107-143910.80 name range(hrs) In meters,excludingbaselineswith 2. OBSERVATIONSANDDATAREDUCTION Table 1.Hiobservinglog. 17 bined (DBCON).Thedatawereimagedand“CLEANed” frame (CVEL),andthedatafromthreerunswerecom- respectively). FortheNAcube,twospectralchannelsata form andnaturalweighteddatacubes(hereafterUNNA, about thenoiselevelperresultantchanneltomakebothuni- time wereaveragedattheimagingstage.TheUNdata level areWo=45"(circular)fortheNAdatacubeand imaged atfullspectralresolution.Thebeamsizesthe50% (MX; Schwab1984;seealsoClark1980;Högbom1974)to with two-dimensionalGaussianbeamshavingtheabove were madebyreconvolvingthecleancomponentimages rected forthefall-offinprimarybeamresponsewithdistance MNT). Themomentmapsanddatacubeswerethencor- and linebroadeningwereconstructedfromthezeroth,first, sizes. PlanesoftheresultantNAdatacubeareshowninFig. from thepointingcenter(PBCOR).Table2summarizes and secondmomentsofthedatacubes(w.r.t.velocity;MO- W=27"X23" (PA=0°)fortheUNcube.Thefinalcubes the beamsize,pixelandnoiselevelinresultantdata cubes. some ofthepropertiesNAandUNdatasetsincluding Gaussian fitstotheUNdatacubeweremade(XGAUS).As 1. UN datacubewasblockaveragedinto30"X30"pixels,and a furthercheck,andtoexaminethelineprofilesindetail 5 tiple GaussianfittingalgorithmusingtheIRAFprogram all theresultingprofileswereexaminedandfitwithamul- SPLOT. plotted asgreyscales.Panel(a)showstheNAmoment0map Figure 2(Plate52)showsthreeHisurfacedensitymaps, 50 beam), andpanel(c),theXGAUSpeakamplitudemap. NGC 2915clearlyhasaspiralmorphology,andthisismost shows thattheyareverydifferentfromitsopticalproperties. readily apparentintheNAmap.Thereappearstobetwo (45" beam);panel(b),theUNmoment0map(27"X23" map, whiletheUNmoment0contoursareshowninFig.4 arms whichcanbetracedbacktotheendsofcentralbar Anglo-Australian Telescope.ThisclearlyshowsthattheHI, extent ofNGC2915. including thespiralarms,extendswellbeyondoptical structure. Figure3showsacontourplotoftheNAmoment0 tion demonstratingthatthisisarealbar,andnothighly is mostclearlyseeninpanel(a)ofFig.2.Thevelocityfield, (Plate 53)overlaidonanIbandimageobtainedwiththe Each ofthesecloudshasaneutral gasmassof—2.5X10 inclined disk.ThebarisresolvedintotwocloudsintheUN analyzed inSec.4.1,showsthesignatureofanovaldistor- data ascanbeseeninthebottom twopanelsofFigs.2and4. neutral Hecontent. ^he neutralgasmassistheHi multiplied by1.33tocorrectforthe From eachcube,mapsoftotalintensity,meanvelocity, In ordertocheckthemomentanalysisresults,single A cursoryexaminationoftheHIpropertiesNGC2915 The prominentcentralbarisabout2.5'(3.9kpc)long.It 7aca (1.4X10 ifl°lbackground issubtracted) 3. OVERVIEWOFHIPROPERTIES 1552 1553 MEURER ETAL: NGC 2915. IL 1553

09 28 26 24 09 28 26 24 09 28 26 24 09 28 26 24 RIGHT ASCENSION (J2000) Fig. 1. Contour plots of the channels with signal in the CLEAN NA data cube. The contour levels are at 10, 25, 50, 75, and 90 percent of the peak surface -1 20 -2 -1 brightness of 4.7 Jy beam (AHi = 1.69 X 10 cm ). The Vr in km s of each channel is shown in the upper right of each panel. The beam size (1^50=45") is shown on the lower left of the upper left panel, and again four panels beneath it.

Table 2. Dataset properties. h) Quantity NA value UN value units LO beam W arcsec x arcsec CO 50 45x45 27x23 -1 CO pixel size 14 x 14 x 6.62 10 x 10 x 3.31 arcsec3 x arcsec-1 x km s noise/channel 2.9 3.0 10“ Jy -1Beam fSdV 145 139 Jy8 km s Mm 9.58 9.21 1O M-10 ^(global) 151 146 km s-1 ^„(global) 170 163 km s 1 V, V4 (global) 471 468 km s“-1 V,w4 (dynamical) 467 km s ^HI 14.9 kpc21 2 ArHi(max) 0.88 1.26 10 cm" Mm/Ls 2.7 2.6 (M/Lb)® Rmt Rho 5.2 Rm/otB 22.6

which is the majority of the mass in the bar. They are located 17" NW and 36" SE of cluster 1 (Paper I). They encompass many of the embedded objects discussed in Paper I and bracket the brighter central clusters as well as some of the peculiar structure on the SE side of the galaxy. The faint “jets,” first mentioned by Sérsic et al (1977) (cf. Fig. 2 1 Paper I) project outwards from the core of these clouds. The Fig. 5. Velocity field from the NA first moment map. The Vr in km s of XGAUS map in panel (c) of Fig. 2 is very noisy, and clearly the contours are indicated. shows only the highest S/N features. In this map the morphology is reminiscent of a spiral with a bar interior to a detached ring. galaxy, but also along the optical major axis, or the bar ridge The optically visible portion of NGC 2915 coincides in line. Such a distribution of star formation is common in position, orientation, and size to the central bar. Thus while barred galaxies (Friedli & Benz 1995, and references the exponential optical surface brightness profile (for therein). R>35") implies a disk structure, the Hi data suggests that There is a large “hole” ~130"X180" (3.4X4.6 kpc) lo- the stars are in a bar configuration. This interpretation is cated 310" (8.9 kpc) to the east of the galaxy’s center. Ex- consistent with the optical morphology (Paper I) which amination of a position velocity cut through this structure shows enhanced star formation primarily in the center of the does not reveal split line profiles, indicating that it is probably not a kinematic shell, but an interarm clearing. The HI in NGC 2915 does not appear to be very porous, as is found to be the case in many nearby disk galaxies, such as HoII (Puche et al. 1992). This may be largely a result of resolution; the HoII data have a resolution of 66 pc, and typical bubble sizes are a few hundred pc, while the linear resolution of our data is ten times worse (640 pc). The velocity field from the NA and UN moment 1 maps are shown in Fig. 5 and Fig. 6 (Plate 54), respectively. They clearly show the expected pattern of a rotating disk. The disk must be warped as indicated by the twisting of the kinematic major and minor axes. Thus although there are closed velocity contours, it is not readily apparent whether these are due to a declining rotation curve, or the warped disk. The rotation curve is derived in detail in Sec. 4.1. The NA and UN global velocity profiles are shown in Fig. 7. These were made from the data cubes by summing the flux in each channel within a “blanking aperture.” The aperture was defined by eye using the total intensity map as a guide. The twin homed nature of the profile is the classic profile of a rotating disk. The bump near Vsys is due to the large central concentration of H I. Fig. 3. Contour plot of total H i intensity from the NA zeroth moment map. Figure 8 shows the line of sight velocity dispersion, b, Contour levels are at 2.5, 7.5, 15, 25, 50, 75, and 90 percent of the peak H i -1 -1 21 -2 map from the XGAUS profile fitting. It shows that b in- surface brightness of 3.1 Jy beam km s (NHi = 1.71 X 10 cm ). In this creases by several tens of km s-1 towards the center of NGC and the remaining maps the crosses mark the positions of fiducial stars and the star marks the optical center from Paper I, and the beam size is shown in 2915. This is a much larger gradient than typically observed the lower left comer. in spiral and irregular galaxies (Kamphuis 1992; Lo et al.

noise ratio of their observations and perhaps inadequacies in their beam profile correction. The systemic velocities measured from the global profile, Vsys(global), and the tilted ring analysis, V (dynamical) (see Sec. 4.1), agree well with each h) sys other, and also with the optical velocity Vsys(optical) -1 3 =468 ±5 km s reported in Paper I. E From the total intensity maps we measure RHl, the radius where the face-on HI column density N(R i)=5X 1019 cm-2, and AHl(max), the maximum Hi face-on column density. The correction to face-on is for the mean inclination i=59°, as derived below. The remaining properties in Table 2 com- com Fig. 7. Global H i Vr profiles derived from the NA (solid line) and UN pare the Hi and optical properties (Paper I). ^Hi is - (dotted line) data cubes. The systemic velocity (first moment of the profiles) pared to the B band luminosity, LB, and RHl is compared to are indicated with arrows of the corresponding line type. the Holmberg radius RUo and B band exponential scale l length aB . The resultant ratios quantify what is already 1993). These maps indicate that pressure support may be clear: NGC 2915 is gas rich and very extended in H I. important to the dynamics of NGC 2915. The line profiles are discussed in detail in Sec. 4.2, while the pressure support 4. ISM DYNAMICS or asymmetric drift, corrections to the rotation curve are discussed in Sec. 43. 4.1 Tilted Ring Analysis Table 2 presents some properties measured from our data. The HI rotation curve was derived using the now stan- From the global HI profiles we measured the total line flux dard tilted ring algorithm, ROTCUR (Begeman 1989), in a fS dV, the corresponding Hi mass, ^hi> the velocity manner similar to that outlined by Martimbeau et al. (1994). widths at 50% and 20% of the maximum intensity, and The UN and NA first moment maps, and the UN-XGAUS W o, respectively, and the systemic velocity V . That 2 sys velocity field were analyzed separately, using ring widths of fS dV from the UN and NA global profiles agree so well, 25", and 45" for the UN and NA data, respectively. There are indicates that the UN data is not missing much short spacing up to six parameters for each ring; the central coordinates flux compared to the NA data. The Hi mass, = 9.6 X ,Y , the systemic velocity V , the inclination i, the po- ± 0.2 X 108 fr°m the NA data, is significantly higher C C sys sition angle (j) of the kinematic major axis (receding side), than that derived by Becker et al. (1988), = 7.0 X 108 and the rotation velocity, V . The parameters were derived (after adjusting to our adopted distance) using a low resolu- rot using five iterations of the algorithm. In the first iteration all tion Hi map derived from Parkes 64 m observations. We parameters are allowed to be free. This iteration gives a feel suspect that the flux difference is due to the lower signal-to- for how the parameters vary with radius R. In the second iteration, i and 0 are held fixed while the other parameters are allowed to vary. The aim of this iteration is to find the best average value of Xc, Yc, and Vsys, which are then held fixed at a constant value for all subsequent iterations. The resultant mean Vsys values are reported in Table 2 as Vsys (dynamical). In the third iteration, new radial profiles of i, 4>, and vro, are derived. Smooth curves are drawn through these profiles and the interpolated values of these parameters are used as the initial guess for these parameters in the fourth iteration, which is meant to test how stable the solutions are to small changes in the initial guesses. It was found that the NA-MOMNT and UN-XGAUS solutions are stable to small perturbations. However the inner three rings of the UN-MOMNT field (Æ<90") are not. No unique solutions could be found. This is due to the kinks in the central isovelocity contours seen in Fig. 6. Therefore we adopt the UN-XGAUS results for R^90", the UN-MOMNT results for 90"<Æ^390" and the NA results for Æ>390". Finally each half (approaching and receding) of the galaxy was analyzed separately. This final step allows the uncer- 09 27 30 26 30 00 25 30 RIGHT ASCENSION (J2000) tainty of the parameters to be estimated from the level of asymmetry in the velocity field; the errors in the parameters Fig. 8. Line broadening, b, map from automatic single Gaussian fitting are taken to be the absolute difference between the value for using XGAUS. The results are shown as both a greyscale and contours at ¿7 = 10, 20, 30, 40, and 50 kms-1. Blank areas are regions of either low the full ring and either the receding or approaching half, signal to noise or complex profiles which could not be fit with a single whichever is larger. The Vrot errors estimated by this method Gaussian. are always larger than or equal to the formal errors from the American Astronomical Society • Provided by the NASA Astrophysics Data System 1996AJ 111 . 1551M -1 1556 MEURERETAL:NGC2915.IL jor axispositionangle,(f)\inclination/,androtationvelocityV. MOMNT, UN-MOMNT,andUN-XGAUS.Theparametersshownarema- Fig. 9.Resultsofthetiltedringanalysisthreevelocityfields:NA- least-squares fitting.Theformalerrorsiniandareocca- cases theformerisadopted. Table 3.TheHIdiskshowsastrongwarpinboth(/>(& sionally largerthantheasymmetryoffielderrors.Inthese proximately constantuntilrisingagainforR>290”.Figure beyond thisradius.Theadoptedrotationcurveshowsarapid for 150",withrelativelymildwarping(A=15°,A/=4°) rot rise toVt^80kmsatÆ=202.5".Itthenremainsap- shown inFig.9.Theadoptedfitparametersaretabulated =40°) andi(A¿=25°).Thewarpisstrongest,especiallyin/, 10 showsapositionvelocitycutthroughtheNAdatacube ro The resultsintermsof/,0,andVforeachdatasetare rot © American Astronomical Society • Provided by theNASA Astrophysics Data System 252.5 227.5 202.5 277.5 412.5 352.5 327.5 302.5 457.5 377.5 177.5 152.5 127.5 102.5 592.5 547.5 502.5 (") 52.5 27.5 77.5 R NA -MOMNTUNXGAUS -48.1 ±2.1 -61.4 ±1.9 -58.6 ±1.7 -56.3 ±0.5 -53.0 ±0.6 -53.7 ±0.6 -55.7 ±4.2 -56.8 ±3.9 -56.3 ±0.8 -65.0 ±1.1 -64.0 ±1.0 -62.8 ±2.1 -65.1 ±1.0 -66.3 ±0.4 -67.5 ±1.6 -68.8 ±0.9 -68.5 ±2.2 -68.8 ±0.5 -28 dh12 -28 ±11 Table 3.Resultsoftiltedringanalysis. (°) 62.2 ±2.3 78.0 ±7.1 55.2 ±1.4 53.8 ±1.4 56.0 ±2.7 70.3 ±8.4 73.1 ±9.7 74.8 ±5.4 52.2 ±1.1 53.0 ±1.2 53.8 ±3.1 53.1 ±1.5 58.4 ±0.7 56.2 ±2.0 54.6 ±2.3 58.7 ±1.9 58.9 ±2.3 58.2 ±1.1 63 ±11 73 ±24 (°) 1 65.5 ±2.0 82.1 ±4.6 81.5 ±1.0 71.9 ±9.0 80.2 ±1.6 79.7 ±4.6 79.9 ±1.3 76.8 ±6.7 81.0 ±0.7 79.9 ±1.5 79.9 ±0.7 79.7 ±3.7 81.2 ±3.2 84.2 ±1.5 82.6 ±2.1 87.0 ±0.5 92.3 ±1.3 (kms 0) 29 ±17 37 ±22 53 ±27 Kot VDV c 38.1 26.5 24.6 31.5 37.7 11.2 12.6 15.2 19.0 10.8 10.5 10.5 11.9 11.3 12.5 13.3 14.3 15.2 16.9 16.1 82.5 72.7 82.8 81.4 79.2 76.0 80.4 80.4 80.9 80.8 80.7 82.3 82.0 83.8 88.5 85.5 93.9 39 65 53 -1 2 -1 -1 2 Fig. 10.Contoursthroughaposition-velocitycutoftheNAdatacube.The corresponding tothekinematicmajoraxis.TheXaxisindicateslocation cut isthroughthecenterofNGC2915andatapositionangle-70°, along thispositionangle;thetickmarksareseparatedby208".Thecontours are at-3.5,3.5,7,14,21,28,and35mly/Beam.Notethe“secondrise”in real. directly onbothsidesofthegalaxy,demonstratingthatitis along themajoraxis(0=—70°).Thesecondrisecanbeseen cal deviationsfromthefithave|AV|^15kms.There- V seenonbothsidesofthegalaxy. tively. ThisindicatesthatVvarieswithRsuggesting the innerportionsofgalaxyarenotinequilibrium. and mostlypositiveresidualsatR^65"150",respec- siduals intheinnerportionshowringsofmostlynegative The residualsalongtheminoraxishaveopposingsignsupto twists andtheproblemsinUN-MOMNTfitnotedabove. deviations arelargeenoughtocausetheisovelocitycontour axies, butatlargeradii.InthecenterwhereLislowthese Kamphuis (1992)findsVvariationsinnormalspiralgal- matic majorandminoraxesarenonorthogonal(Bosma the SW).Thisisasignatureofanovaldistortion:kine- bits arenotstrictlycircularfortheserings,althoughone distortion probablyarisesfromthecentralbar,andisstron- the amplitudeofkmsattheseradii. gest fortheringswith200"^/?^450".Itshowsthator- ^ ±10kmsoneithersideofthecenter(+toNE,— r rot Fig. 8whichgivesamapofbvaluesfromXGAUS.The should bearinmindthattheresidualsaresmallcomparedto broadening iscenteredontheSE HIcloud(seeFig.4). map isnotsymmetricaboutthe centerofNGC2915.There sysy is strongerlinebroadeningonthe SEsidethantheNW.The Since thelinewidthsalongminor axisarenotelevatedrelativetothe 1981). TheeffectcanalsobediscernedinFigs.5and6. rot sys major axis,itisnotlikelythatresiduals areduetoagalacticwind. Figure 11showstheresidualmapsfromringfits.Lo- The resultsoftheautomaticGaussianfittingareshownin 4.2 LineProfiles 1556 1557 MEURER ET AL. : NGC 2915. II. 1557

Residual VT [Km/s] Residual Vr [Km/s] -20 0 2o O 20

Residual Vr [Km/s] -20 o 20

Fig. 11. Greyscale plus contour plots of the residual maps from the tilted ring fits. The contours shown are for AVr=-20, -15, -10, -5, 5, 10, 15, and 20 kms“1. Panels (a), (b), and (c) show the NA-MOMNT, UN- MOMNT, and UN-XGAUS residuals, respectively. Note that the panels have different scales.

■] I I I I I 09 29 28 27 26 25 24 (b) RIGHT ASCENSION (J2000)

XGAUS was set to make only single Gaussian fits. The Could the broad profiles be due to beam smearing of the SPLOT measurements were designed to test the XGAUS re- observed steep rotation curve? This is not likely, because the sults and measure complex line profiles. The position of the knee in the rotation curve is four to six beam widths from the splot fits are shown in Fig. 12, with the positions of profiles center, and thus the central velocity gradient is not that steep. best fit by multiple Gaussians highlighted. The HI profiles Beam smearing only shows its effects over one to two beam are predominantly single peaked. The main region of split widths. We verified this by simulations of the data including lines is confined to within 2' of the center, which is also the effects of beam smearing on the UN data cube. A change -1 where the lines become the broadest. Profiles at the position in Vrot of 80 km s in one beam width can result in ¿?=27 of both HI clouds are split. Thus the total HI width distri- kms-1. But we can rule out such a steep rotation curve, bution (including splitting) is a bit more symmetric than the because it would require the knee in the rotation curve to be b map shown in Fig. 8. The average line split is A Vr = 41 at Æ =50". Only higher resolution observations can determine km s“1. A few representative profiles and their SPLOT fits if there are large velocity gradients within any given beam. are shown in Fig. 13. However any such gradients are likely to be turbulent in © American Astronomical Society • Provided by the NASA Astrophysics Data System 1558 MEURER ETAL: NGC 2915. IL 1558

(< East) ARA [arcsec] (West >) R [arcsec] Fig. 12. Position with respect to the center of NGC 2915 (shown as a star) of SPLOT profile measurements. Positions marked with small dots are well Fig. 14. Comparison of line-of-sight velocity dispersion, b, measurements. represented by single Gaussian fits, while double Gaussians are required for Azimuthal averages of the UN-MOMNT, and UN-XGAUS measurements the positions marked with large dots. The crosses mark the position of the are shown as filled squares and triangles, respectively. Individual SPLOT fiducial stars shown in Figs. 4 and 6. measurements with signal/noise >3 are shown as X’s, lower signal/noise splot measurements are indicated with dots. nature since the broad (b^lO kms“1) line region is eight good agreement, indicating that the additional beam smear- beam diameters wide. ing in the 30" pixel data has not substantially increased b Figure 14 compares the SPLOT b measurements with the with respect to the full resolution UN data. The MOMNT azimuthally averaged b from the MOMNT and XGAUS re- analysis, on the other hand, yields b substantially lower than sults. It shows that the SPLOT and XGAUS results are in the Gaussian fits. This is due to an artificial bias in the MO- MNT analysis technique; MOMNT only uses the data where the signal is above a certain threshold in a smoothed version of the data cube. This systematically excludes the profile wings, and thus the results are biased towards lower b. The dotted line in Fig. 14 shows the two pixel resolution of the UN data cube (¿7=2.8 km s-1). Most profiles are well sampled by our data, those having ¿?^2.8 kms-1 mostly have low signal to noise ratios. A notable exception is shown in Fig. 13(f). At large R, the mean b is around 8 kms-1, which is typical for disk galaxies (Kamphuis 1992). Figure 13 shows that at this width the profiles remain well sampled.

43 Asymmetric Drift Correction The broadening of the HI profiles in the central region is significant compared to the measured rotation, as can be seen by comparing Fig. 14 with Fig. 9. The random motions of the gas may then provide significant dynamical support in the central region. It is not clear that the gas there is in dynamical equilibrium. Indeed the asymmetry of the b map suggests 400 450 500 550 400 450 500 550 that it is not. Nevertheless, here we will assume that gas is in equilibrium and account for the pressure support by estimat- Vr [km/s] ing an “asymmetric drift” correction following Oort (1965: see his Eq. 10). We assume that the gas velocity ellipsoid Fig. 13. Examples of SPLOT Gaussian fits (continuous lines) to Hi line (i.e., pressure) is isotropic. This greatly simplifies the profiles (histogram style lines). The profiles were chosen to illustrate (a) a 3 typical high signal/noise outer profile; (b) a typical low signal/noise outer correction. In a cylindrical coordinate system, the radial (R) profile; (c) a broad single profile near the center; (d) a split profile near the acceleration KR about the rotation axis z is given by center; (e) a split profile in the outer disk; (f) an atypical profile: a narrow core on top of a normal single profile. The positions of the profiles in ±E, ±N offsets (in arcsec) from the center of NGC 2915 are: (a) -295,101; (b) 3Binney & Tremaine (1978) treat the case of the asymmetric drift in colli- -25,311; (c) 5,11; (d) 35,-19; (e) -265,-79; (f) -325,41. sionless systems (stars only) which have nonisotropic velocity ellipsoids.

© American Astronomical Society • Provided by the NASA Astrophysics Data System 1996AJ 111 . 1551M Fig. 15.Thetoppanelshowstheline-of-sightlinebroadening,asa the solidline. tom panelwiththecorrespondinglinestyle.ThetotalXcurveisshownas fits tothesedataasgiveninthetext.ThecontributionofbandX using concentricringswith0=—60.8,i=59.6.Thebrokenlinesshowthe from theUN-XGAUSbmap(top)andUNzerothmomentmaps(center) Xg .Thefilledcirclesshowthedatawhichareazimuthalaveragesextracted function ofradius,whilethemiddlepanelshowsgassurfacedensity, 1559 MEURERETAL:NGC2915.IL gradients tothetotalasymmetricdriftcorrection,(Jareshowninbot- where pisthedensity.Iftherewerenopressuresupport The densitycanbewrittenasp=£/(2/*),wherehisthe orbits wouldbecircularwithvelocitiesVgivenby have acore-halostructureandarewellfitbyfunctioncon- The othertwotermscanbeestimatedfromtheradialprofiles negative, makingthecorrectionpositive.Wedonothavea Usually thecontributionsofeachgradientsarezeroor taken asHIplusneutralHe).Thecircularvelocityisthen vertical scaleheightandthegassurfacedensity(here of ln(Z?)andln(X)whichareshowninFig.15.Bothprofiles direct measurementofhbutwillassumedlnih^ldR=0. where (tistheasymmetricdriftcorrectiongivenby sisting ofaGaussiancenteredatÆ=0ontoppolynomial (in R)background: ß D gz c g z d -4 2 3 62 2 Krb ln(¿?) =2.40+6.57X10/?+1.23 exp(-{Æ/85.1}/2), ln(2) =0.812+1.057X10“/?-7.13X 10~/? a Vc=vL+°-d (1) ~ —rM g 2 © American Astronomical Society • Provided by theNASA Astrophysics Data System D~ 2 _ _v? àiHpb]) ot O 200400600 -Rb 2 + 1.51exp(-{Æ/56.2} /2), (4) 2 d ln(2,) dR + 2 R [arcsec] d \n(b)\n(h) z dR Jr (2) (3) _1 -2 2 -1 2 rms of0.05and0.07in\n(b)ln(2),respectively.These where Risinarcsec,bkms,andthegassurface density inpc.Thefitswereunweightedandhavean equilibrium attheseradii. the massdistribution,withcaveatthatHImaynotbe quite significantforR^152".Wecannowproceedtomodel circular velocities,V.Asexpectedthecrcorrectionsare last twocolumnsofTable3tabulateaandthecorrected fits (convertedtolinearunits)areshowninFig.15,aswell curve. Thecomponentsarethestellardisk,aneutralISM as thecrcurveresultingfromdifferentiatingfits.The free parameters.TheDMhaloisassumedtohaveaspheri- mined fromtheobservationspresentedhere,andthereareno distribution isdeterminedfromthesurfacebrightnesspro- bution therotationalvelocityatlargeÆ,Voc,andhaloveloc- cally symmetricdensitydistributiongivenby distribution oftheHIdisk(includingneutralHe)isdeter- files inPaperI,andhasonefreeparameter^M!L.Themass ity dispersioncraregivenby(Lakeetal.1990): core radiusR.Hereweadopty=2.Forthisdensitydistri- where thefreeparametersarecentraldensitypand Fitting wasdonewitha^minimizationtechnique.Thusthe (H l)disk,andadarkmatter(DM)halo.Thestellarmass the associatederrors. points thantotheouterpoints.Theaminimizationthen rings. IntheendweadoptauniformerrorinVof3kms. errors totheVvaluesinTable3.Theseincludedadopting fitted parametersdependnotonlyontheinputdata,butalso components. the fittedrotationcurveandcontributionofeach eters aretabulatedinTable4.EachpanelofFig.16shows discussed. TheyareillustratedinFig.16,andtheirparam- we arenolongerdoingtrueaminimization,buteffectively ing producesareasonablefitatallÆ.Sofortheadoptedfits make amarginalimprovementatlargeR.Onlyequalweight- effectively sacrificesthefitininnerregionsorderto other methodsinvariablyassignlessweighttotheinner This correspondstotheaverageerrorinVforR>90".The rors, andtakinglocalaveragesoftheerrorsoverseveral the Verrors,combiningerrorswithlikelycrer- a nonweightedleast-squaresminimization.Fourmodelsare the pressuresupport. ThedifferencebetweenModels Aand models generatedhere. needed tofittherotationcurve. Thehaloparametersarefor are allowedtobefree,withtheonlyconstraintthatthey g a verycompactanddensecore, themostextremeof cD D D ^0. Theresultingfithas^Æ/L =0., i.e.,thediskisnot B 0 c 0 c c rot rotD b PyW Three componentmassmodelswerefittotherotation We experimentedwithnumerousschemesforassigning \+(RIR)' Model AisthebestfittoVdata.Allthreeparameters V„ =477Gp/?c4.9o-g. Model BisafittoVinstead of V;ineffectitignores c c 0 rot c Po 4.4 MassModels 1559 1996AJ 111 . 1551M from theUN-XGAUSdata,circlesUN-MOMNTandsquares Fig. 16.Therotationcurvewithmassmodelfits.Trianglesrepresentresults The Holmbergradiusisindicatedbythearrow.thicksolidlineineach from theNA-MOMNTdata.ClosedsymbolsareVandopen. 1560 MEURERETAL:NGC2915.IL panel showsthemodelfit,andbrokenlinesshowcontributionof individual components:stellardisk,dottedline;neutralgas,longdashed vidual models. line; DMhalo,shortdashedline.Seethetextforadescriptionofindi- B isthusindicativeoftheuncertaindynamicalstate central region.Neglectingtheasymmetricdriftcorrectionre- core parametersisquitedramatic,afactorofabout1.5inR yields ^/L=0.0. and morethan2inp.Sincewemayhaveunderestimated sults inalarger,lessdensecore.Theeffectontheinferred isothermal sphere(e.g.,Binney&Tremaine1987).Forthis and compactthanthatofModelA.LikeA,B density distribution cr, itispossiblethatNGC2915’shaloevenmoredense crot c ß 0 D -1 Model CshowsafitwheretheDMhaloisnonsingular limits. TheyweredeterminedfromMonte-Carlosimulations assuminguniformVand Kot errorsof±3kms. -1 9 2 -1 3 -1 1 1 c Fit parameters: Within lastpoint(15.2kpc): At Ego(2.9kpc): © American Astronomical Society • Provided by theNASA Astrophysics Data System rms (kms) MjLb (solarunits) M dark/AlLuminous ■Mt/Lb (solarunits) Al (10AÍ©) AÍ dark/AiLuminous Mt/Lb (solarunits) M (10®Ai©) Reduced x ob (kms) Po (AÍ0PC“) V (kms) V (kms") Rc (kpc) Note. —Errorsare90%confidencelimitsand upper limitsare95%confidence Voo (kins") r t c c Model: R [Kpc] Table 4.Massmodels. 0.24 ±0.06 39.9 ±0.8 0.78*$ 88 ±2 <0.4 1.01 2.8 0.10 ±0.020.07 41.2 ±0.951.9 1.24i$ 2.5±0.3 < 0.140.75±0.40 91 ±2 3.8 73 ±1 2.14 10.2 4.1 3.6 8.3 23 0.10 ±0.02 1.23 ±0.15 41.0 ±1.1 91 ±2 1.80 3.8 9.6 3.4 1.2 5.2 71 27 76 -1 This istheformofDMdistributionusuallyfavoredby .Æ/Lg asthebestfit. for R^6kpc.ThedifferencebetweenModelsAandCillus- Carignan andcollaborators(e.g.,Puche&1991, represents approximatelythemaximumdiskcomponent halo modelsgivenbyEq.(5)andfix^S/L=1.2,which lation inNGC2915.ForModelDwereturntotheanalytic expect fortheredcolorofoutlyingdiffusestellarpopu- free. Theresultinglimitsonhi)35kmsisthemeanvelocitydispersionin north oftheopticalcenterclearlycorrespondstoHa The protrusionintheb=30kmscontourabout25"to edges ofthebubbles(neargalacticcenter),and ter morphology.Thatistheirionizingclustersareatthe energized neutralISM.TherearetwoHabubbles,ofablis- energy of the core.Thus£(turb)=3.8X10erg.Theenergyinturbu- ergetic HIisassociatedwiththeSEcloud.Itat gun” thatimplicateshotyoungstellarpopulationswiththe image ofMarloweetal.(1995).Thisshowsthe“smoking bubbles. Theultimateenergysourceisthestellarwindsand it. ThusthebroadHIprofilesareassociatedwithturbu- edge oftheSEHabubblewhichappearstobeexpandinto sion, althoughbeamsmearingmaycausesomeconfusion lent ionizedISMandinparticulartheexpandingHa L/L=l.S (Leitherer&Heckman1995),whereis IMF andaconstantstarformationrate,theequilibriumratio (Plate 55)showsthebcontoursofFig.8overlaidonHa 1562 MEURERETAL:NGC2915.IL the ratemechanicalenergyisproducedbysupemovaeexplo- SNe producedbytheyoungcentralstellarpopulations. ergs. Thecrossingtimeofthecoreistcross^3.1kpc/35 population inNGC2915issufficienttoexplaintheobserved typical velocity).Duringthistimethemechanicalenergyre- km s=87Myr(i.e.,thelengthofregiondividedby end upaskineticenergy,therestbeingthermalized(Ostriker erg, orseventimesE.Only10%-20%ofLislikelyto kinetic energyintheneutralISM. approximation, theenergyreleasedbyyoungcentral sions andstellarwinds.FromtheL=6.4X10ergs (Paper I;Marloweetal1995)wededuceL^^LIXIO magnitude versusHIlinewidthdiagram.Thetightcorrela- & McKee1988;MacLowMcCray1988).Sotoafirst lease ofthecentralpopulationis£'ech=¿mechfcros3Xl0 hold evenformoderatemasses. the lowmassend.NGC2915shows thatitdoesnotalways of theTFR(e.g.,Pierce&Tully1988)whendataare This discrepancyisalsoapparentinotherrecentcalibrations and X’s(Sandage1988),istheTully-Fisher(1977)relation- tion shownbythefilledcircles(Sandage&Tamman1976) Â: adjusted tothesamevalueofH.Carignan&Beaulieu ship (TFR)fornearbygalaxies.NGC2915,likeDD0154 defined asstars andallphasesoftheISM—and linewidth, (1989) notethatDD0154shows theTFRmaybreakdownat (Carignan &Beaulieu1989),fallswayoffthiscorrelation. ably beconsideredacorrelation betweenluminousmatter— mechna kmech Ha mS 0 54 E=(Qxp) +£¿(turb)=4.2X10erg. What isthesourceforthiskineticenergy?Figure17 For asolarmetallicitypopulationwithSalpeter(1955) Figure 18showsNGC2915andDD0154intheabsolute k Milgrom &Braun(1988)note that theTFRshouldprob- © American Astronomical Society • Provided by theNASA Astrophysics Data System 5.3 Tully-FisherRelationship The solidcirclesaredatafromSandage&Tamman(1976);theX’s with DDO154andNGC2915.AV21istheinclinationcorrectedHiline Fig. 18.TheTully-Fisherrelationshipfornearbydiskgalaxiescompared both samplesandwehavetakenM=+0.1wherenecessary. Table 2ofSandage(1988).Galaxieswithi<40°havebeenexcludedfrom width at20%ofthepeakvalue,andM\sabsoluteBbandmagnitude. not justopticalluminosityandlinewidth.TheTFRholdsfor ^ÆIL—0.3-A{MIL)q (higherifthereisasignificant ISM. However,forNGC2915andDDO154theneutralgas most galaxiesbecausethestellarmassdominatesover under thisscenario,thediscrepancywithTFRmayindi- mass inmoleculargas)theywouldfallontheTFR.Thus their ISMwereconvertedtostarswitha component ismoresignificantthanthestellarcomponent.If minous matterandtheTFR.InanormalTFRgalaxy, unevolved comparedtonormalgalaxies. cate thatthedisksofDDO154andNGC2915arerelatively much moreconcentratedthantheDM.Thecontributionsof rotation curveisnearlyflatatV=fromaboutadisk mum valuesV,butatdifferentradii,suchthattheoverall each componenttotherotationcurveproducessimilarmaxi- become centrallyconcentratedbeforestarformation.Like- called “disk-haloconspiracy.”IftheneutralISMofdis- stellar distribution,whichdominatestheluminousmass,is DDO 154wouldnotlooklikenormalTFRgalaxiesevenif well belowthatoftheDM.Inotherwordstherewouldbeno will becoextensivewiththeDM(seebelow)andhaveV the resultantluminosityisonTFR,stellardistribution crepant galaxiesweretransformedtostarsinsitusuchthat Paper I)suchthatNGC2915 falls ontheTFR,butthere wise, onecouldadjustthedistanceordustextinctionA(see their ISMwereconvertedintostars,unlessto disk-halo conspiracyinsuchcases.ThusNGC2915and like anynormalTFRgalaxy. would benegligible.Soeventhen NGC2915wouldbeun- would stillbenodisk-haloconspiracy andthestellar^Æ!L scale lengthouttothelastmeasuredpoint.Thisisso- ßpg DM haloisperhaps anewtwistonthedisk-halo conspiracy B b rotmax max max B B -1 There stillremainstheproblemofdistributionlu- The coincidenceofthecentral b=A0kms=crofthe 6 Sf "18 0 -16 -20 - -22 -14 • % □DD0154 2.2 2.42.6 _L logCAVj,, [Km/s]) ÛNGC2915 x x • x >S< 2.8 1562 1996AJ 111 . 1551M -1 1563 MEURERETAL.:NGC2915.IL Fig. 19.Upperpanel:radialvariationoflogarithmicsurfacedensities.The the ,S/LratioofModelD.Thecriticaldensityrequiredforefficientstar DM density(dashedline)isthatderivedfromModelD.Thestellarprofile neutral gasmeasurements(solidline)areazimuthalaveragesusingtheUN normalized byX.Linestylesarethesameasforupperpanel. curve andthebfitofEq.(3).Lowerpanel:logarithmsurfacedensities formation (Kennicutt1989;dotdashedline)isderivedfromtherotation data forRÎOkpc(390")andtheNAR>10kpc.Theprojected late onthemeaningofthiscoincidenceinSec.6. (dotted line)istheBbandsurfacebrightnessprofile(PaperI)normalizedto to themassofNGC2915fromModelD:neutralgas,, ratio ofthefirsttwothesewithX.Itisimmediately the stars,andDMhalo,S.Figure19(b)shows (although onecasedoesnotmakeaconspiracy).Wespecu- monly observedindiskgalaxies,asfirstpointedoutby neutral gasdoes.Arelativelyflat2Æprofileiscom- the centerofNGC2915wherehighmassstarformationis less thanXbyafactorofabout3(range2to9),evenin Beaulieu 1989;Broeils1992). constant factortoobtaintheDMcontribution(Carignan& Beaulieu 1989;Carignanetal.1990).Rotationcurvesof Bosma (1981;morerecentexamplesincludeCarignan& apparent thatthestarsdonottraceDMverywell,but does notobeyKennicutt’sstarformationlaw.Thediscrep- star formation,2(Kennicutt1989)and/2-is some galaxiesarewellfitbyscalingtheXprofilea not bringX<.Thisiscontrary towhatisfoundfora However, evenadoptingaconstantb=5kmsprofiledoes 2915 because2andthisiswhereT?thehighest. ancy betweenXandiisworstinthecenterofNGC seen. WhenonlytheneutralISMisconsidered,NGC2915 b the neutralISMisconsidered.Of coursetheremaybenoreal ness galaxies(vanderHulstet al.1993)andotherBCDs wide varietyofgalaxiesincluding bothlowsurfacebright- DM discrepancy, since muchoftheISMmaybe in molecular (Taylor etal.1994)whichobey theKennicuttlawwhenonly DM DM gDM cút critDM g cútg crit gcrt Figure 19(a)showsthesurfacedensityofcontributors Figure 19alsoshowsthecriticaldensityforhighmass © American Astronomical Society • Provided by theNASA Astrophysics Data System 5.4 StarFormationThreshold R [Kpc] 21-2 21-2 8 7 :=20-2 21-2 -1 _1 form. StarformationinNGC2915occursatR^30"(Paper molecular gasfortheKennicuttlawtohold. formation inNGC2915isconsistentwith“simple”Hi rection Si=1.2X10cm.Recenthighmassstarforma- clouds andisX!=2.410cmorafterinclinationcor- to satisfyKennicutt’slawthereshouldbeM5=3X10 I), whereandM(/?=s30")=3.1X10Thus threshold schemesforhighmassstarformation;example tion isseennearthepositionsofboththeseclouds.Star tion lawofKennicutt(1989)doesnotholdforNGC2915,a while Skillman(1987)similarlyproposesthatSn^crit Gallagher &Hunter(1984)proposethatefficienthighmass formation viaclouddestruction.InNGC2915andtheBCDs star formationoccurswhereSn>i,rit5X10cm, rotation isalmostsolidbody,andthustidalshearmini- the risingportionofrotationcurve.Thisiswhere in thesampleofTayloretal.(1994),starformingregion simple thresholdlawdoes. mized. SimilarlythegalaxiesinSkillman(1987)study and indeedmostoftheopticallyvisiblegalaxyislocatedin cated. all havesolidbodyrotationwheretheirHIIregionsarelo- appearance. TheHimorphologyandglobalproperties 2915 showsittohaveanatureunexpectedfromitsoptical = 10cm.Thus,althoughthemorecomplexstarforma- The HIline-of-sightvelocitydispersioninthiscentralregion the barisresolvedintotwoHIcloudsseparatedby1.5kpc. oval distortioninthevelocityfield.Atourhighestresolution axy. Thepresenceofabarisconfirmedbythesignaturean (mass, linewidth)arethoseofalate-typebarredspiralgal- H H molecular i the opticalgalaxy,outtofivetimesHolmbergradius.At young stellarpopulationsthere.TheHIextendswellbeyond ics ofthecentralregioncanbeaccountedforbyknown is veryhigh,¿^40kms,andtheHIlinesplitbyAV measured points(RÎOkpc).For/?<3kpc,randommotions there areindicationsofanovaldistortioninthevelocity these radii,thedynamicsaredominatedbyrotation,although blue-light ratioisÎL=16, making NGC2915oneofthe nous componentsalone.Largeamountsofdarkmatter(DM) is inequilibriumattheseradii. velocity varieswithradius.Itisnotclearwhetherthesystem are significantandthereindicationsthatthesystemic approximately flatforR>4kpc,butslightlyrisingatthelast field, andsomewarpingofthedisk.Therotationcurveis HC rotation curvedifficult.Despite thisuncertainty,itisclear that NGC2915’sDMdistribution hasanunusuallydense central region,makeinterpretation ofmassmodelfitstothe of thelastmeasuredpoint(R= 15.2kpc)thetotalmass-to- are required,dominatingatnearlyallradii.Withintheradius «=40 kmsatthepositionofbothHIclouds.Theenerget- darkest diskgalaxiesknown.The complexdynamicsofthe TB The peakHIsurfacedensityoccursinthetwocentral Kenney eial.(1993)showhowtidalshearcanlimitstar Our detailedHistudyoftheweakBCDgalaxyNGC The rotationcurvecannotbeaccountedforbythelumi- Only about1%oftheso-calleddarkmatterneedbe 6. CONCLUSIONS/SUMMARY 1563 1996AJ 111 . 1551M c3 R=0.$ to2.5kpc).Thediskratioisnotwellcon- regulating starformation:abar,anddeepcentralpotential. particular theyrevealtwokeyingredientsinitsrecipefor Friedli &Benz(1995)showedthroughnumericalsimula- NGC 2915,atopicourobservationsshednewlighton.In Gallagher &Hunter1984;Skillman1987). have formedrecently.However,itdoesobeysimplecritical nicutt (1989)starformationthresholdlaw.Unlessthereis redistributes angularmomentumresultinginaninflowalong tion, theISM,andbarinisolatedbarredgalaxies.The tions thatthereisacomplexinteractionbetweenstarforma- density thresholdmodelsforhighmassstarformation(cf. tion, eveninthecenterwherehighmassstarsareobservedto line width(orrotationalvelocity).NordoesitobeytheKen- tionship, beingunderluminousbyafactoroftenforits the bar.SomestarformationalongbarenergizesISM considerable amountofmolecularmaterial,itsgasdensityis stellar componentislessmassivethantheneutralISMdisk. strained byourmodels,althoughitseemslikelythatthe reaches thecenter.Thecoincidenceinvelocitydispersions, always belowthecriticaldensityforhighmassstarforma- (central densitypoÔ.lP~)andcompactcore(radius enough toinhibitadditionalstarformationuntilthematerial inferred fortheDMhalo,suggeststhatintensityof that ofHImeasuredinthecentergalaxy,and 1564 MEURERETAL\NGC2915.IL fountain.” ThestellarcomponentofNGC2915andother that suchanoutflowcancontributetothespindownof mation ishalted,atleasttemporarily.Somemodelsindicate too strongtheISMisthrownclearofcoreandstarfor- tial well(Bothunetal1986);ifthestarformationbecomes central starformationisregulatedbythedepthofpoten- (via theresultantsupemovaeexplosionsandstellarwinds) radius indicatingthattidalshearprobablystronglyinhibits BCD/amorphous galaxiesislimitedtowithintheturnover Charlton &Salpeter1989)thusestablishinga“galactic disk resultinginfurtherinflow(Corbelli&Salpeter1988; NGC 2915andSGO0938.1—7623.Stimulatingdiscussions man whoassistedinobtainingandreducingIbandimagesof by Noguchi(1988)orTayloretal(1993)cannotberuled ner (Appendix)soaninteractionmodelsuchasthatsketched tion, whichisconvenientsinceNGC2915inalowdensity lated starformationforNGC2915doesnotrequireinterac- tally theopticalmajoraxis.AscenarioofbarandDMregu- counterpart isabar,oratleastthatthebaraxiscoinciden- and orientationofthecentralbar,suggestingthatstellar c proved itscontentandpresentation. Wearealsogratefulto Michael Dahlem,RonAllen, and StefanoCasertanoim- early draftofthispaperfrom Rosie Wyse,TimHeckman, Renzo Sancisigreatlyinfluenced thispaper.Critiquesofan with AlbertBosma,AnnetteFerguson, ChrisMihos,and ing HaimagesandEchellespectraavailable,ChrisLid- out. environment. However,thereisonepossibleinteractionpart- star formation.TheopticalimagesofNGC2915hasthesize the anonymousrefereeforcomments thathelpedimprove Finally wespeculateonthenatureofstarformationin NGC 2915doesnotobeytheTully-Fisher(1977)rela- We thankAmandaMarloweandTimHeckmanformak- © American Astronomical Society • Provided by theNASA Astrophysics Data System 1 1 4 4 -1 _1 knowledges fundingthroughNASAGrantNo.NAGW-3138, nautics andSpaceAdministration.G.R.M.gratefullyac- Laboratory, Caltech,undercontractwiththeNationalAero- the earlyphasesofthisinvestigation.Wearegratefulto Montréal, andTheSpaceTelescopeScienceInstituteduring this paper.LucTurbideandHarryPayneprovidedessential BCDs presentburstofstarformation.Suchcompanionswere tion withanunseencompanionmayberesponsibleforthe lated BCDsforcompanionswiththerationalethatinterac- maintaining theirsuperbfacility,andefficientlybringingthe lent facilities.WethanktheNarrabristaffofATNFfor directors oftheAustraliaTelescopeNationalFacilityand administered byTheJohnsHopkinsUniversity. tragalactic Database,afacilityoperatedbytheJetPropulsion Compact Arraybackonlineafterthunderstorms.Literature software support.G.R.M.wasemployedatUniversitéde searches wereperformedusingNED,theNASA/IPACEx- Space TelescopeScienceInstitutefortheuseoftheirexcel- which correspondstoaradiusof25kpc.Attypicalrelative However theprimarybeamofATdishesisW=3' rated fromNGC2915by42',aprojecteddistanceof64kpc, face brightnessobject,SGC0938.1-7623,whichissepa- this distanceinlessthan100Myr. velocity of300kms“aninteractingpartnercantransverse successfully searchedourHIdatacubeforfaintcompanions. successfully foundin12/21galaxiestheirsample.Weun- band CCDimageofSGC0938.1-7623obtainedusingthe teracting partnerofNGC2915? 2915. Therearenogalaxieswithcatalogedvelocitieswithin FI imagingsystemontheAAT.Itwasmadefromfour200s is alowsurfacebrightnessgalaxyordiffusenebulosityas- and hasnomeasuredvelocity.IsSGC0938.1-7623thein- 500 kms“ofNGC2915’s.Howeverthereisonelowsur- body hasadiameterof—130".Therearefaintextensionsto There arenumerousforegroundstarsoverthefaceofSGC 0938.1—7623, butnoobviousoverabundanceofsourcesthat exposures obtainedduringnon-photometricconditions. (NED) forpossibleinteractionpartnerswithin5°ofNGC The NASA/IPACExtragalacticDatabase (NED)isoperatedbytheJetPro- was obtainedwiththeParkes64mradiotelescope.Theob- overall symmetryofthesystemsuggeststhatitisagalaxy. the SEandNWyieldingatotaldiameterof—270".The would suggestSGC0938.1—7623isresolved.Themain sociated withourgalaxy.Figure20(Plate56)showsanI (1985) notethatitisnotcertainwhetherSGC0938.1-7623 galactic Hiandanarrowspike havingV=30kms, over 1024channelsandreachanoiselevelof50mJy.There the NationalAeronautics andSpaceAdministration. pulsion Laboratory,CaliforniaInstitute ofTechnology,undercontractwith are noobviousHIfeaturesinthe spectrumexceptforstrong servations covertherangeV=-1000to+4000kmssplit 50 r T Taylor etal(1993,1995,1996)surveyedapparentlyiso- We searchedtheNASA/IPACExtragalacticDatabase Unfortunately theanswerisnotyetclear.Corwinetal An HispectruminthedirectionofSGC0938.1-7623 APPENDIX: ANINTERACTIONPARTNER? 1564 1996AJ 111 . 1551M l 1 probably correspondstotheedgeofhighvelocitycloud HVC 300-24+274,whichMorras(1982)showstohave narrow profilesandvelocitiesvaryingovertherange W=30 kms\and/Sdv=0.72±0.06Jykms.This 240^^330 kms“. 1565 MEURERETAL.:NGG2915.II. Begeman, K.G.1989,A&A,223,47 Becker, R.,Mebold,U.,Reif,K.,&vanWoerden,H.1988,A&A,203,21 Bajaja, E.,Huchtmeier,W.K.,&Klein,U.1994,A&A,285,385 Binney, J.,&Tremaine,S.1987,GalacticDynamics(PrincetonUniversity Begeman, K.G.,Broeils,A.H.,&Sanders,R.H.1991,MNRAS,249,523 Bosma, A.1991,inWarpedDisksandInclinedRingsAroundGalaxies, Bosma, A.1981,AJ,86,1825 Bothun, G.D.,Mould,J.R.,Caldwell,N.,&MacGillivray,H.T.1986,AJ, Brinks, E.,&Klein,U.1988,MNRAS,231,63P Bruzual, A.G.,&Chariot,S.1993,ApJ,405,538 Broeils, A.H.1992,A&A,256,19 Carignan, C.,Charbonneau,P,Boulanger,F.,&Viallefond,F.1990,A&A, Carignan, C,&Beaulieu,S.1989,ApJ,347,760 50 Clark, B.G.1980,A&A,89,377 Charlton, J.C,&Salpeter,E.1989,ApJ,346,101 Corwin, Jr.,H.G.,deVaucouleurs,A.,&G.1985,Southern Corbelli, E.,&Salpeter,E.1988,ApJ,326,551 Dickey, J.M.,Hanson,M.&Helou,G.1990,ApJ,352,522 Côté, S.,Carignan,C.,&Sancisi,R.1991,AJ,102,904 Friedli, D.,&Benz,W.1995,A&A,301,649 Gallagher, J.S.,&Hunter,D.A.1984,ARA&A,22,37 Gerola, H.,&Seiden,P.E.1978,ApJ,223,129 Grillmair, C.J.,Freeman,K.C.,Bicknell,G.V.,Carter,D.,Couch,W., Hoffman, G.L.,Lu,L.U.,Salpeter,E.E.,Farhat,B.,Lamphier,C.,&Roos, Högbom, J.1974,ApJS,15,417 Hunter, D.A.,vanWoerden,H.,&Gallagher,J.1994,ApJS,91,79 Hutchmeier, W.K.,&Seiradakis,J.K.1985,A&A,143,216 Holmberg, E.1958,Medd.Lund.Ast.Obs.,Ser.II,No.136 Marlowe, A.T,Heckman,T.M.,Wyse,R.F.G.,&Schommer,1995, Mac Low,M.-M.,&McCray,R.1988,ApJ,324,776 Lo, K.Y,Sargent,W.L.W,&Young1993,AJ,106,507 Leitherer, C.,&Heckman,T.M.1995,ApJS,96,9 Lake, G.,Schommer,R.A.,&vanGorkom,J.H.1990,ApJ,320,493 Kennicutt, R.C.1989,ApJ,344,685 Kenney, J.D.R,Carlstrom,E.,&Young,S.1993,ApJ,418,687 Kamphuis, J.1992,Ph.D.thesis,Rijksuniversiteit,Groningen Mateo, M.,Olszewski,E.W.,Pryor,C.,Welch,D.L.,&Fischer,P.1993, Martimbeau, M.,Carignan,C.,&Roy,J.-R.1994,AJ,107,543 Press, Princeton) edited byS.Casertano,P.Sackett,andF.Briggs(CambridgeUniversity Press, Cambridge),p.181 92, 1007 234, 43 Galaxy Catalogue(UniversityofTexasPress,Austin) T. 1993,AJ,106,39 Sommer-Larsen, J.,&Taylor,K.1994,ApJ,422,L9 ApJ, 438,563 AJ, 105,510 We concludethatthereisnoevidenceSGC0938.1 © American Astronomical Society • Provided by theNASA Astrophysics Data System REFERENCES 6 1 yet beruledoutasaninteractionpartner.Forassumed50 ^ >2.6X10withS/N>5atNGC2915’sdistance. km s“linewidththeHIspectrumshouldbeabletodetect poor. If SGC0938.1-7623isatthedistanceofNGC2915itgas Meurer, G.R.1994,inESO/OHPWorkshoponDwarfGalaxies,ESOCon- Milgrom, M.,&Braun,E.1988,ApJ,334,130 Meurer, G.R.,Mackie,G.,&Carignan,C.1994,AJ,107,2021(PaperI) —7623 isinteractingwithNGC2915.However,itcannot Noguchi, M.1988,A&A,201,37 Morras, R.1982,A&A,115,249 Oort, J.H.1965,inStarsandStellarSystemsV:GalacticStructure,edited Pierce, M.J.&Tully,R.B.1988,ApJ,330,579 Puche, D.,Carignan,C.,&Wainscoat,R.J.1991,101,447 Puche, D.,&Carignan,C.1991,ApJ,378,487 Ostriker, M.P,&McKee,C.F.1988,Rev.Mod.Phys.,60,1 Hl Puche, D.,Westpfahl,Brinks,E.,&Roy,J.-R.1992,AJ,103,1841 Roberts, M.S.,&Haynes,P.1994,ARA&A,32,115 Sackett, P.D.,Morrison,H.L.,Harding,R,&Boroson,T.A.1994,Nature, Sandage, A.,&Tamman,G.A.1976,ApJ,210,7 Sandage, A.,Saha,Tamman,G.Labhardt,L.,Schwengeler,H., Sandage, A.1988,ApJ,331,605 Salpeter, E.1955,ApJ,121,161 Sérsic, J.L.,Bajaja,E.,&Colomb,R.1977,A&A,59,19 Schwab, F.R.1984,AJ,89,1076 Skillman, E.D.1987,inStarFormationGalaxies,editedbyC.J. Shostak, G.S.1978,A&A,68,321 Taylor, C.L.,Brinks,E.,&Skillman,E.D.1993,AJ,105,128 Taylor, C.L.,Brinks,E.,Pogge,R.W.,&Skillman,E.D.1994,AJ,107, Skillman, E.D.,Bothun,G.Murray,M.A.,&Warmels,R.H.1987, Taylor, C.L.,Brinks,E.,Grashuis,&Skillman,E.D.1995,ApJS,99, Taylor, C.L.,Brinks,E.,Grashuis,&Skillman,E.D.1996,ApJS,102, Tully, R.B.,&Fisher,J.1977,A&A,54,661 Toomre, A.1964,ApJ,139,1217 Thuan, T.X.,&Martin,G.E.1981,ApJ,247,823 Walsh, W,editor1992,AustraliaTelescopeCompactArrayUsersGuide Warmels, R.H.1988,A&AS,72,427 Viallefond, F.,&Thuan,T.X.1983,ApJ,269,444 van Zee,L.,Haynes,M.R,&Giovanelli,R.1995,AJ,109,990 van derHulst,J.M.,Skillman,E.D.,Bothun,G.McGaugh,S.S.,&de Wevers, B.M.H.R.,vanderKruit,P.C.,&Allen,R.J.1986,A&AS,66, Prugniel (ESO,Garching),p.351 ference &WorkshopProceedingsNo.49,editedbyG.MeylanandP. by A.BlaauwandM.Schmidt(UniversityofChicagoPress,Chicago), p. 455 Panagia, N.,&Macchetto,F.P.1994,ApJ,423,L13 370, 441 Lonsdale Persson(NASAConf.Pub.CP-2466),p.263 A&A, 185,61 971 427 Blök, W.J.G.1993,AJ,106,548 505 (Version 1.0),(AustraliaTelescopeNationalFacility,Sydney) 189 1565 1996AJ 111 . 1551M up andeastistotheleftinallpanels.Thescalebarbottompanel is2'longand20"wide. UN moment0map(25"resolution).Thebottompanelshowsthepeak intensity (amplitude)fromtheXGAUSsingleGaussianfitstoUNdata.Northis Fig. 2.GreyscalerepresentationsofHIintensitymaps,alltothesame scale. ToppanelistheNAmoment0map(45"resolution).Themiddle 1766 PLATE 52 © American Astronomical Society •Provided bythe NASAAstrophysics Data System Meurer etal.(seepage1552) PLATE 53

-76 30

O H < 38 i—] wO Q 40

09 28 27 26 25 24 RIGHT ASCENSION (J2000)

Fig. 4. 800 s I band CCD image of NGC 2915, obtained with the AAT FI system overlain with HI total intensity contours from the UN moment 0 map. 1 1 21 Contours are at 5, 13, 25, 50, 75, and 90 percent of the peak surface brightness of 1.39 Jy beam“ km s“ (AHi = 2.46 X 10 cm—2). The strong glow at lower right in this figure and Fig. 6 is due to a bright star just off the edge of the frame. Meurer et al. (see page 1552)

Fig. 6. Velocity field from the UN first moment map overlain on the / band image. The contour interval is 10 km s l. The contour passing through the center -1 _1 has V, =470 km s ; the prominent closed contours to the SE- and NW of the center have Vr=400 and 540 km s , respectively. Meurer et al. (see page 1554)

1768 © American Astronomical Society • Provided by the NASA Astrophysics Data System 1996AJ 111 . 1551M Fig. 8.Thescalebaratlowerrightis30"long. Fig. 17.Contoursofline-of-sightvelocitydispersion,b,overlainontheHo?imageMarloweetal.(1995).Thebcontourscorrespondtothose shownin © American Astronomical Society •Provided bythe NASAAstrophysics Data System Meurer etal.(seepage1562) PLATE 55 1769 1996AJ 111 . 1551M dimension. Fig. 20.800sIbandCCDimageofSGC0938.1—7623,obtainedwiththe AATFIsystem.NorthisupEasttotheleftandimage500"acrossineach 1770 PLATE 56 © American Astronomical Society •Provided bythe NASAAstrophysics Data System Meurer etal(seepage1564)