The polar ring galaxy AM1934-563 revisited
N. Brosch, A. Kniazev, D. Buckley, D. O'Donoghue, Y. Hashimoto, N. Loaring, E. Romero, M. Still, P. Vaisanen, E.B. Burgh, K. Nordsieck
aa r X i v : . [ a s t r o - ph ] J un Mon. Not. R. Astron. Soc. , 1–19 (2007) Printed 24 October 2018 (MN L A TEX style file v2.2)
The polar ring galaxy AM1934-563 revisited ⋆ Noah Brosch, , † Alexei Y. Kniazev, , David Buckley, Darragh O’Donoghue, Yas Hashimoto, Nicola Loaring, Encarni Romero, Martin Still, Petri Vaisanen, Eric B. Burgh, Kenneth Nordsieck The Wise Observatory and the School of Physics and Astronomy, the Raymond and Beverly Sackler Faculty of Exact Sciences,Tel Aviv University, Tel Aviv 69978, Israel South African Astronomical Observatory, Observatory Road, Cape Town, South Africa Special Astrophysical Observatory, Nizhnij Arkhyz, Karachai-Circassia, 369167, Russia Space Astronomy Laboratory, University of Wisconsin, Madison, WI 53706, USA
Accepted 2007 April ??. Received 2007 March ??; in original form 2007 March ??
ABSTRACT
We report long-slit spectroscopic observations of the dust-lane polar-ring galaxyAM1934-563 obtained with the Southern African Large Telescope (SALT) during itsperformance-verification phase. The observations target the spectral region of the H α ,[N ii ] and [S ii ] emission-lines, but show also deep Na I stellar absorption lines thatwe interpret as produced by stars in the galaxy. We derive rotation curves along themajor axis of the galaxy that extend out to about 8 kpc from the center for boththe gaseous and the stellar components, using the emission and absorption lines. Wederive similar rotation curves along the major axis of the polar ring and point outdifferences between these and the ones of the main galaxy.We identify a small diffuse object visible only in H α emission and with a lowvelocity dispersion as a dwarf H ii galaxy and argue that it is probably metal-poor. Itsvelocity indicates that it is a fourth member of the galaxy group in which AM1934-563belongs.We discuss the observations in the context of the proposal that the object is theresult of a major merger and point out some observational discrepancies from this ex-planation. We argue that an alternative scenario that could better fit the observationsmay be the slow accretion of cold intergalactic gas, focused by a dense filament ofgalaxies in which this object is embedded.Given the pattern of rotation we found, with the asymptotic rotation of the gas inthe ring being slower than that in the disk while both components have approximatelythe same extent, we point out that AM1934-563 may be a galaxy in which a darkmatter halo is flattened along the galactic disk and the first object in which thispredicted behaviour of polar ring galaxies in dark matter haloes is fulfilled. Key words: galaxies: ring galaxies — galaxies: evolution — galaxies: individual:AM1934-563 — galaxies: dark matter — galaxies: galaxy haloes
INTRODUCTION
Ring galaxies posed significant astronomical interest sinceLynds & Toomre (1976) modelled the Cartwheel galaxy asthe result of a small galaxy passing through a larger one. ⋆ Based on observations obtained with the Southern AfricanLarge Telescope (SALT). † E-mail: [email protected] (NB); [email protected](AYK); [email protected] (DB); [email protected] (DOD);[email protected] (YH); [email protected] (NL); [email protected](ER); [email protected] (MS); [email protected] (PV);[email protected] (EBB); [email protected] (KN)
While such events probably happen and produce some ofthe ring galaxies, in other instances different mechanismsmight be at work. A particularly interesting kind of ringgalaxy is the polar ring galaxy (PRG) where a flattened diskgalaxy exhibits an outer ring of stars and interstellar matterthat rotate in a plane approximately perpendicular to thecentral disk. An extensive catalog of PRGs was produced byWhitmore et al. (1990).The issue of PRGs was reviewed by Combes (2006).She reviewed a number of formation mechanisms for PRGs:minor or major mergers, tidal accretion events, or directcold gas accretion from filaments of the cosmic web. Combes c (cid:13) Noah Brosch et al. (2006) proposed that these objects can be used to probe thethree-dimensional shape of dark matter (DM) haloes, pro-vided the PRG is in equilibrium in the gravitational poten-tial. The well-known Spindle Galaxy (NGC 2685), anarchetypal PRG, exhibits two sets of rings: an outer onevisible only on HI maps and which might be in the plane ofthe galaxy, and an inner one that is helix-shaped, is perpen-dicular to the main axis of the galaxy, is optically bright,shows embedded present-day star formation, and is associ-ated with prominent dust lanes. Shane (1980) explained thesystem as consisting of a lenticular galaxy that recently ac-creted an HI gas cloud that formed the inner ring, whilethe outer gas ring might be a remnant of the formation ofthe galaxy. Hagen-Thorn et al. (2005) found that the stellarpopulation of the inner system of dust and gas, arranged ina spiral around the ”spindle” but really in a disk, is 1.4 × years old.In a different ring galaxy, NGC 660, Karataeva et al. (2004) detected red and blue supergiants belonging to thering system. They showed that the age of the youngest starsthere is only ∼ et al. ±
10 km sec − located at l=341.02, b=-28.73, alsoidentified as PRC B-18 in Whitmore et al. (1990). Theobject was recently studied by Reshetnikov et al. (2006),who showed that this is a giant galaxy in a compacttriplet, together with PGC 400092 (classified Sd/Irr:) andPGC 399718 (classified SBc:) at approximately the sameredshift. The authors used the 1.6-meter telescope of thePico dos Dias Observatory in Brazil for imaging in BVRI,the CTIO 1.5-meter telescope to collect spectral observa-tions, and included data from IRAS and 21-cm line observa-tions. However, most of their conclusions about the natureof the object rely on the morphological appearance of thegalaxy.Reshetnikov et al. (2006) modelled AM1934-563 usingan N-body code that includes gas dynamics using stickyparticles and star formation. They concluded that the best-fitting model is of a major merger, whereby a gas-rich galaxytransferred a sizable amount of matter to AM1934-563 dur- Table 1.
Details of the AM1934-563 RSS observationsDate Exp.time Spec. Range Slit PA Disp.(sec) (˚A) ( ′′ ) ( ◦ ) (˚A/pix)16.07.2006 2 ×
600 3650–6740 1.5 140 0.9816.07.2006 1 ×
600 3650–6740 1.5 35 0.9820.09.2006 2 ×
900 6050–7315 1.5 140 0.4020.09.2006 1 ×
750 6050–7315 1.5 27 0.4021.09.2006 3 ×
900 6050–7315 1.5 27 0.40 ing a parabolic encounter. The matter subsequently relaxedand now forms a complete ring of stars, gas, and dust aroundAM1934-563 whereas the donor galaxy is one of the twoother galaxies in the same group.The reason to revisit this object was the availability ofhigh-quality spectra obtained with the effectively 8-meter di-ameter Southern African Large Telescope (SALT) telescope.We derive, for the first time, rotation curves for the ionizedgas and for the stellar components of both the main galaxyand the polar ring. Since PRGs might make good test casesfor the properties of dark matter haloes in and around galax-ies, as argued by Combes (2006), the more observationaldata collected on these objects and with higher quality, thebetter.Very few PRG observations obtained with large tele-scopes have been published. A noticeable one is by Swaters& Rubin (2003), with the Baade 6.5-meter telescope on LasCampanas, tracing the dynamics of the stellar component ofthe prototype PRG NGC 4650A where they showed that thepolar ring is actually a polar disk, an extended feature ratherthan a narrow gas disk. They favour a scenario by whichthe ring/disk was formed by the polar merger to two similardisks, as previously suggested by Iodice et al. (2002). Iodice et al. (2006) observed the gaseous component in the ring ofN4650A with ESO’s FORS2 on UT4 and concluded that ascenario by which it could be formed was through slow gasaccretion from the cosmic web filaments. We propose thatthe same situation could be taking place for AM1934-563.This paper is organized as follows: § § §
3, and present our in-terpretation in §
4. The conclusions drawn from this studyare summarized in § SALT was described by Buckley et al. (2006) and byO’Donoghue et al. (2006), its Robert Stobie Spectrograph(RSS) was described by Burgh et al. (2003) and Kobulnickyet al. (2003), and the first scientific papers based on its ob-servations were published by Woudt et al. (2006) and byO’Donoghue et al. (2006). We used the SALT and RSS toobserve AM1934-563. The observations of AM1934-563 wereobtained during the Performance Verification (PV) phase ofthe SALT telescope with the RSS spectrograph and are de-scribed in Table 1.The July 2006 spectra (see Table 1) were obtained dur-ing unstable weather conditions (high humidity, seeing worsethan 5 ′′ ), without fully stacking the SALT mirrors. Theycover the range 3650–6740 ˚A with a spectral resolution of c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 1.
V-band image of AM1934-563 with SALTICAM. This is a 2 sec exposure, primarily reduced, using binning on-chip of 2 × ∼ − or a FWHM of 6–7 ˚A. These spectra do notshow strong and extended emission lines but were used tomeasure equivalent widths (EWs) of absorption lines in thatspectral range following observations.The spectra obtained on the nights of September 2006were taken during stable weather conditions with seeing ∼ . ′′
5. They cover the range from ∼ ∼ − or 2.4 ˚A FWHM. Alldata were taken with a 1 . ′′ . ′′
258 pixel − (after binning the CCDs by a factorof two). Each exposure was broken up into 2–3 sub-exposuresto allow the removal of cosmic rays. Spectra of a Cu–Ar com-parison lamp were obtained after the science exposures tocalibrate the wavelength scale.The September 2006 data include two spectra obtainedat position angle 140 ◦ centered on AM1934-563 extend-ing about four arcmin along the galaxy’s major axis and at a shallow angle to the dust lane, where the northernpart passes also through the “northwest companion” PGC400092 (Reshetnikov et al. 2006), and three spectra cen-tered on the same position but obtained at position angle27 ◦ , along the major axis of the ”polar ring” described byReshetnikov et al. (2006). We emphasize that the samplingof the major axis spectra was at PA=140 ◦ , not at 130 ◦ asdone by Reshetnikov et al. (2006), since 140 ◦ is closer tothe position angle of the disk as given by Reshetnikov et al. (148 ◦ ) and allows for a moderate degree of disk warping.Although the observations discussed here are mostlyspectroscopic, one image of the galaxy was obtained with atwo-sec exposure in the V filter with the SALTICAM camera(O’Donoghue et al. ∼ . ′′ c (cid:13) , 1–19 Noah Brosch et al.
Figure 2.
A 220 ×
220 arcsec image extracted from the one shown in Fig. 1. The three galaxies of the tight group are indicated, as isthe newly detected H α knot (see text for more details). The slit positions used here are over-plotted and each slit is 1.5 ′′ wide.Note a fewother diffuse images in the neighbourhood. time, which can be evaluated from the stellar images onFigure 13 (see below), caused the images far from the good-quality ∼ ∼
10 arcmin across with 0.28 arc-sec/pixel (after binning on-chip by a factor of two).The data for each RSS chip were bias and overscan sub-tracted, gain corrected, trimmed and cross-talk corrected,sky-subtracted and mosaiced. All the primary reduction wasdone using the IRAF package salt developed for the pri-mary reduction of SALT data. Cosmic ray removal was donewith the FILTER/COSMIC task in MIDAS. We used theIRAF software tasks in the twodspec package to perform thewavelength calibration and to correct each frame for dis-tortion and tilt. One-dimensional (1D) spectra were thenextracted using the IRAF APALL task.Figures 3 and 4 show parts of fully reduced and com-bined spectral images for PA=140 ◦ and PA=27 ◦ , respec- IRAF: the Image Reduction and Analysis Facility is distributedby the National Optical Astronomy Observatory, which is oper-ated by the Association of Universities for Research in Astron-omy, In. (AURA) under cooperative agreement with the NationalScience Foundation (NSF). MIDAS is an acronym for the European Southern Observatorypackage – Munich Image Data Analysis System. tively. Figure 5 shows the spectrum of the central part ofAM1934-563. The ∼ ∼ λλ ∼ ◦ .The derived internal errors for the 2D wavelength cal-ibrations were small and did not exceed 0.04 ˚A for a reso-lution of 0.4 ˚A pixel − , or < − at the wavelength ofredshifted H α line. To exclude systematic shifts originatingfrom known RSS flexure, we calculated line-of-sight velocitydistributions along the slit for both emission and absorbtionlines using a suite of MIDAS programs described in detail inZasov et al. (2000). These programs allow the use of addi-tional correction factors derived from tracing nearby night-sky lines whose accurate wavelengths are very well knownto correct the observed wavelengths of the Na I D, H α [N ii ] λ ii ] λ ∼ − . Most of the calcu-lated velocity distributions are shown in Figures 8–12. Allvelocities derived with this procedure are heliocentric. c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 3.
Part of 2D reduced spectrum for PA = 140 ◦ . NW is up. The slit was positioned along the major axis of AM1934-563 andexhibits the redshifted H α , [N ii ] λλ ii ] λλ λλ ± ′′ along the slit, but the emissionand absorption lines can be reliably traced only up to ± ′′ . The spectrum of the Sd/Irr galaxy PGC 400092 is located approximately ∼ ′′ away from AM1934-563 . The PGC 400092 spectrum shows the same emission lines as AM1934-563, but there is no indication ofNa I absorption. Weak [O i ] λ i λ ′′ = 0.8 kpc and the imageextent is ∼
100 kpc.
All emission lines were measured with the MIDAS pro-grams described in detail in Kniazev et al. (2004, 2005).These programs determine the location of the continuum,perform a robust noise estimation, and fit separate lines withsingle Gaussian components superposed on the continuum-subtracted spectrum. Nearby lines, such as the H α and [N ii ] λλ ii ] λλ λλ A cursory inspection of the spectra obtained at PA=140 ◦ (see Figure 3) shows rotation detectable in the same amountand behaviour exhibited by the H α , [N ii ] λλ ii ] λλ λλ i ] λ λ Table 2.
EWs of absorption lines in spectra of AM1934-563Absorption Line Equivalent Width(˚A) (˚A)CaII H 8.9 ± ± δ ± γ ± β ± ± ± spectrum of AM1934-563 shows [O i ] λ β , H γ and H δ lines in absorption. The CaII H andK doublet is seen in absorption at the blue end of the spec-trum. The spectra also show very weak [O iii ] λλ c (cid:13) , 1–19 Noah Brosch et al.
Figure 4.
Part of 2D reduced spectrum obtained at PA = 27 ◦ that covers the same spectral region as the spectrum for PA = 140 ◦ andshows the same spectral features. NE direction is up. The slit is positioned along the major axis of the polar ring of AM1934-563. Thespectrum of AM1934-563 is visible at position 0 ± ′′ . Note the H α emission line produced by the newly detected group member ∼ ′′ away from AM1934-563 , near the top and close to the right edge of the image. The linear scale and extent of this images are identicalto those of Fig. 3. Table 2. Measurements of lines detected in more than onespectrum were averaged.The rotation curve of AM1934-563 along the majoraxis, derived from the two-spectra combination shown inFigure 3, is shown in Figures 8 and 9. Figure 8 show thevelocity-position plot and Figure 9 shows the galacto-centricvelocity-distance plot. In general, the emission-line rotationcurve derived here corresponds with that shown in Figure 5of Reshetnikov et al. (2006), except that ours is better sam-pled, has a higher signal-to-noise, and the rotation curvesderived from the different emission lines practically coincide,as can be estimated from the formal 1 σ error bars plottedin the figures and from the scatter of the individual points.Figure 9 shows also a comparison of our measurements withthose of Reshetnikov et al. (2006).Deriving the rotation curves shown in Figures 8 and 9we found that the systemic radial velocity of AM1934-563is 11663 ± − , formally higher by some 14 km sec − than the value given by Reshetnikov et al. (2006) in theirTable 3 but consistent with their value within the quoteduncertainties. This offset might be the result of a slightly dif-ferent definition of the systemic velocity; we chose the valuefor which the NW branch of the rotation curve matched bestthat for the SE branch and by this procedure also found therotation center of the galaxy. Independently, we found thatthis location on the velocity curve is also the central pointfor the linear fitting of all the measurements for the Na I Dlines seen in absorption, as shown in Figure 8. We obtaineda best-fit line following the relation:V r = (11663 ±
2) + (15 . ± . × R (1) where R is the distance in arcsec from the point where theradial velocity of AM1934-563, defined using the emissionlines, equals 11663 km sec − and we adopt this locationas the kinematic centre of the galaxy. The different symbolsindicate the H α velocity (black squares), the [N ii ] λ ii ] λ ± − for the NW companion PGC 400092 as well, whereour value is significantly lower than the 11735 ± − given in Reshetnikov et al. (2006). Since the velocity dis-crepancies for AM1934-563 and for PGC 400092 are in op-posite directions, we can probably rule out a systematic shiftbetween our velocity scale and the one of Reshetnikov et al.(2006). This is confirmed also by the plot in Figure 9 wheretheir derived velocity curve points are plotted over our re-sults. The shift between our data for PGC 400092 and thatfrom Reshetnikov et al. (2006) could be the result of the slitposition for PA = 140 ◦ used here that did not cross exactlythe physical center of that galaxy.We could also derive the velocity dispersion of the H α line along the slit for PA = 140 ◦ ; this is shown in the bottompanel of Figure 9. The dispersion is shown as the FWHM ofthe line after correcting for the intrinsic spectrometer linewidth. The corrected H α line FWHM=5–7 ˚A found for thecentral part ( ± − . The corrected FWHM= < c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 5.
Top panel:
The 1D spectrum of the central part of AM1934-563 extracted from the 2D spectrum observed at PA = 140 ◦ with a setup that covers 3650–6740˚A and with a spectral scale of ∼ − . The “reddest” part of the spectrum is not shown. Thespectrum shows some absorption lines and possibly very weak [O iii ] λλ Bottom panel:
The 1D spectrum of thecentral part of AM1934-563 extracted from the 2D spectrum observed at PA = 27 ◦ . All detected lines have been marked.c (cid:13) , 1–19 Noah Brosch et al.
Figure 6.
The 1D spectrum of PGC 400092, extracted from the 2D spectrum observed at PA = 140 ◦ . All the detected emission lineshave been marked. Note that no Na I D λλ measured for the H α line of PGC 400092 indicates internalmotions slower than 45 km s − .The rotation curve along the polar ring axis, at PA =27 ◦ , is shown in Figure 12 as a velocity-position plot. This, asalready mentioned, relies mostly on the emission lines sincethe Na I absorptions are visible only in the central part ofthe spectrum, and is therefore more limited in extent. Thespectra for PA = 27 ◦ show a linearly increasing rotationfor ∼ ′′ SW of the galaxy centre outwards, where the centerposition is that derived for the major axis. Since the NEand SW branches of the ring’s major axis show very differentbehaviour from that observed along the galaxy’s major axis,the method used previously to find the rotation center bymatching the two branches could not be used in this case,thus we do not show a folded and combined velocity curvefor the major axis of the ring.The NE branch shows an approximately flat rotationfrom ∼ ′′ away from the centre, as derived from the emissionlines, with some oscillations from the center to the peripheryat 10 arcsec from the center. These oscillations are evident inboth H α and [N ii ] λ α emission is encountered closeto the location of the most intense continuum contribution(compare the solid and the dashed lines). Our spectra along PA = 27 ◦ show a completelydifferent kinematic behaviour than the one described byReshetnikov et al. (2006). Their Fig. 7 shows a ∼
50 kmsec − difference between the velocity of the [N ii ] λ α at the galaxy centre that increases to ∼
100 km sec − atthe SW end of the ring. We, on the other hand, see no dif-ference between the velocities of these two lines. Moreover,the [S ii ] lines in our observed spectrum also show the samebehavior as the [N ii ] λ α lines. We also note thatthe extent to which the rotation is defined and measurablefor this position angle and using the emission lines is prac-tically the same as for the major axis of AM1934-563 , some8 kpc from the center (at 167 Mpc).Similar to the case of the major axis, PA = 140 ◦ , wesee here also a straight-line behaviour with galacto-centricdistance of the Na I absorption lines. We find a formal linearfit of the formV r = (11662 ±
2) + (14 . ± . × R (2)The Na I rotation curve is linear from 1 . ′′ ∼
5” NE of the kinematic centre. Note that the value foundfor the slope at this position angle is virtually identical withthat for the major axis in equation (1).A comparison of the two panels of Fig. 12, the lowerone which is a velocity-position plot for PA = 27 ◦ and theupper one which is a plot of the line intensity vs. position c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 7.
Line count ratios along the slit for PA = 140 ◦ . All points displayed here have a signal-to-noise ratio of at least four. Top tobottom: a). Profile of the net H α flux in total counts. b). Profile of the [N ii ] λ α ratio. c). Profile of the [S ii ] 6716+6731/H α ratio.d). Profile of the electron-density sensitive ratio R SII =[S ii ]6716/[S ii ]6731. The value R SII =1.4 is plotted with a dotted line. The valuesR
SII =1.35 and 1.0 are plotted with dotted lines; these indicate electron densities n e = 50 and 500 cm − respectively. along the slit, shows that the region where most of the lineemission is produced is about 4 ′′ to the NE of the kinematiccenter of AM1934-563 and that the emission is practicallyonly along the NE part of the ring.As for PA = 140 ◦ , we derive the velocity dispersionfor this position angle as the FWHM of the H α line vs.galacto-centric distance. This is shown in the bottom panelof Fig. 10 after correction for the intrinsic width of the linesusing the night sky spectrum. The corrected FWHM=7 ˚Afor the redshifted H α indicates internal motions of ∼
300 kms − . Although not spectrophotometrically calibrated, ourspectra allow the derivation of a few physical parametersof the gas using line ratios. The good signal-to-noise of thespectra allows the derivation of these ratios along the slit,as shown in Figs. 7 and 11. The ratios plotted in Fig. 7 al-low a derivation along the galaxy major axis and for its NW companion. Since these ratios are based on the very closelylocated emission lines, they practically do not depend onwhether the spectral data were corrected for sensitivity ornot. For the red spectral range, using the sensitivity curvecannot change these ratios by more that a few percent; thisis less than the displayed errors.Creating these ratios we took into account the possi-ble stellar absorption in the H α line. Checking Table 2, andconsidering the Balmer spectra of Gonz´alez-Delgado et al.(1999) we suggest that EW abs (H α )=6 ˚A with a constantvalue along the slit. Since EW(H α ) ≈ ii ]/H α and [S ii ]/H α would in-crease from the AM1934-563 centre to the edges. Thatcould be interpreted as an increase in of metallicity withgalacto-centric distance, which is not correct. With a c (cid:13) , 1–19 Noah Brosch et al.
Figure 8.
Top panel:
The solid line shows the profile of the H α flux along the slit for PA = 140 ◦ after continuum subtraction. Theshort-dashed line shows the continuum intensity distribution along the slit and in the spectral region of the H α line. Middle panel:
Radialvelocity distribution along the major axis of AM1934-563. The black squares, red squares and blue triangles represent measurements ofthe emission lines H α , [N ii ] λ ii ] λ λλ σ error bars have been overplotted for all measurements. The solid blue lineis result of a linear fit to all measurements of the Na I D lines. Bottom panel:
The measured FWHM of the H α line, corrected for theintrinsic line width of the RSS. The FWHM of the reference night-sky line measured in each row is shown with open squares. measured line ratio for the central part of AM1934-563( ± N II ] λ / H α )=0.54 ± ± ± e ≃
50 cm − .The measurements for detected part of PGC 400092 give12+log(O/H) = 8.45 ± e ≃
500 cm − .In a similar way, we derive the gas properties along themajor axis of the ring (see Fig. 11). With the line ratios mea-sured in the central part of AM1934-563 ( ± N II ] λ / H α )=0.51 ± ± ii ] lines ratio we obtainthe same value found previously: n e ≃
50 cm − . α emission knot An isolated H α emission knot was detected at α . = 19 h m s .7; δ . = − ◦ :26’:18”, some78 ′′ away from the main body of the galaxy to the NE and c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 9.
The galacto-centric velocity distributions along the major axis of AM1934-563. The black and red filled circles are for the NWbranch using the emission lines of H α and [N ii ] λ α and [N ii ] λ λλ α and [N ii ] λ on the extension of the ring’s major axis. This knot is realand was detected on all spectra observed at PA = 27 ◦ takenon 2006 September 20 and 21. The velocity distributionwith distance is shown in the top panel of Figure 10. It isevident that the line emitting knot is fairly isolated and isvery distant from the galaxy, yet its radial velocity is closeto that of the AM1934-563 systemic velocity. The measuredvelocity for the knot is 11645 ± − ; this is more thanthree standard deviations away from the systemic velocityof AM1934-563 and very many standard deviations awayfrom the recession velocity measured for H α at the NW tipof the galaxy. It is also very different from the velocity ofPGC 400092, the NW companion of AM1934-563 , or fromthat of PGC 399718, the other companion in the triplet.Our observations do not show a significant velocity dis-persion of the H α line observed from the knot, as shownin the bottom panel of Fig. 10; a formal measurement indi-cates that this H α line has the same FWHM ( ∼ α from the knot indicates internal motionsslower than 40 km s − . The size of the line-emitting region isonly ∼ α ) = 120 ±
15 ˚A. No additional emission linesare visible in the spectrum.
The image of the field displayed in Figure 1 shows not onlyAM1934-563 but also its two companion galaxies. Fig. 1 isa V-band image of the field obtained with SALTICAM inthe same night as the spectroscopic observations on Septem-ber 21. The image of the three galaxies allows one to notethat (a) the region around the target contains many diffuse,low surface brightness (LSB) images that might be partsof galaxies or LSB dwarfs at the same redshift, or distantobjects in the background, and (b) the appearance of thecompanion galaxy PGC 400092 to the NW is that of a Sdgalaxy with a similar overall size to that of AM1934-563 .The LSB objects are also visible on Digitized Sky Surveyimages of the region.We performed unsharp masking of Figure 1 to empha-size the dust lane; this is shown in Figure 13 and, contrary tothe claim of Reshetnikov et al. (2006) that the dust lane issplit and embraces the galaxy nucleus from SE and NW, in-dicates that the lane is fairly straight, passes south and westof the brightest part of the galaxy, and is probably not splitat all. The stars in Fig. 13 have the shapes of crescent moons.This arises from telescope optical problems which are beingironed out during the Performance Verification process, andhave been emphasized by the unsharp masking.The measured ratio of emission lines to corrected H α , c (cid:13) , 1–19 Noah Brosch et al.
Figure 10.
Top panel:
The radial velocity distribution of the H α emission line along the major axis of the ring of AM1934-563. The H α emission line produced by the newly detected group member appears ∼ ′′ away from the center of AM1934-563 . This newly detectedgroup member has a small velocity dispersion and only a ∼
20 km s − difference from the systemic velocity of AM1934-563 , which isplotted with a short-dashed line. Middle panel:
The solid line shows the profile of the H α flux along the slit at PA = 27 ◦ with thecontinuum subtracted. The short-dashed line shows the continuum intensity distribution in the region of the line and along the slit. Bottom panel:
The measured FWHM for H α line corrected for the RSS intrinsic line width. The FWHM of the reference night-sky lineis shown as the solid line. and the possibly very weak [O iii ] λ ∼
240 km sec − andnot at 195 km sec − as given by Reshetnikov et al. (2006),and that this maximum is reached asymptotically for the NEpart of the galaxy. Figire 8 shows that our measurementsare compatible with those of Reshetnikov et al. (2006) forthe regions of overlap. The last points of the rotation curve branch of the SE part of the galaxy, from galacto-centric dis-tance of 6 ′′ to 10 ′′ , drop from 200 ± − to 150 ± − in both H α and [N ii ] λ ′′ to 8 ′′ but shows a step-like drop at this location, fol-lowed by a recovery with a similar distance-velocity gradientas for the central part of the galaxy.A comparison of the major axis rotation curves shownin Fig. 9 shows clearly the difference between the kinematicbehaviour of the two Na I D absorption lines and the H α ,[N ii ] λ ii ] λ c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 11.
Line count ratios along the slit for PA = 27 ◦ .All points plotted here have a signal-to-noise ratio of at leastfour. Top to bottom: a). Profile of the H α flux in total counts.b). Profile of the [N ii ] λ α ratio. c). Profile of the [S ii ]6716+6731/H α ratio. d). Profile of the electron-density sensitive[S ii ]6716/[S ii ]6731 R [ S ii ] ratio. The value R [ S ii ] =1.35 is plot-ted with a dotted line and indicates an electron density n e = 50cm − . used giant and supergiant stars to show that the EW of theMg I triplet near 5180˚A should be twice that of the Na Ilines. If this would be the case for AM1934-563 then ourblue spectrum where the Mg I triplet is barely visible wouldrule out a major Na I absorption contribution from stars.However, in giant galaxies such as AM1934-563 thestellar populations are better represented by main sequencestars. These have stronger photospheric Na I than Mg I (e.g.,a M0V star from the same library as used by Schwartz& Martin (2004) has EW(Mg I)=20˚A and EW(Na I)=12˚A .While it is not possible to separate the stellar Na I fromthe interstellar absorption, we can accept that at the least afraction, and perhaps all of the observed absorption repre-sents the stars in the galaxy. For example, in M82 Saito etal. (1984) detected Na I absorption that they attributed tostars and interpreted as solid-body rotation. Assuming that most of the Na I absorption is photo-spheric, this would indicate that, while the gaseous com-ponent follows a “normal” galactic rotation law, the stellarcomponent rotates almost like a solid body for ∼ ′′ awayfrom the centre. The maximal rotation velocity exhibited bythe stellar component is only ∼
150 km sec − at 10 ′′ fromthe centre for both ends of the major axis.The extent over which the emission is observed for the“polar ring” is almost the same as for the major axis, some18 ′′ overall as shown in Fig. 12, but the derived rotationcurve is completely different. The rotation curve indicatessolid-body like rotation for 1 . ′′ ′′ to the SW. The velocity difference between the outermostpoints on the slit where the absorption lines are measuredis only 90 km sec − . The velocity gradients shown by thestellar components along the major axis of the galaxy andalong the axis of the PR, in regions where a linear rota-tion curve can be defined, are very similar as equations (1)and (2) show. In both cases the gradients are ∼
19 km sec − kpc − , where we converted the observational gradients fromequations (1) and (2) to physical units. At a distance to the object of 167 Mpc (H =70 km sec − Mpc − ) the radius of the galaxy to the outermost pointwhere emission lines are visible is ∼ ∼ ◦ to the major axis of the galaxy. The stellar component ob-served with the spectrometer slit oriented along the majoraxis of the ring is also rotating as a solid body and with asimilar velocity-distance gradient to that observed for themain body of the galaxy.Reshetnikov et al. (2006) concluded from their photom-etry and spectroscopy, coupled with results of N-body mod-elling, that AM1934-563 is a PRG. Their models indicatethat the system might be the result of a major interactionbetween a ”donor” galaxy with a 17 kpc stellar disk anda 42 kpc gaseous disk, with a total mass of 3.6 × M ⊙ ,which encountered a 2 × M ⊙ and 14 kpc wide ”recep-tor” galaxy some 1.6 Gyrs ago with an impact parameter of130 kpc and a relative velocity of 145 km sec − . This en-counter transferred a large quantity of matter (stars, gas,and dust) from the donor to the receptor galaxy resultingin the formation of the polar ring which is inclined with re-spect to the galaxy disk and is warped. Reshetnikov et al.(2006) suggested that the donor galaxy survived and isPGC 399718, the southern companion in the triplet, andargued that their suggestion is supported by the reddish(B-V) colour of the galaxy and by its somewhat disturbedappearance.In selecting this scenario in preference to those of mi-nor mergers calculated by them, or of other possible modelsfor the formation of ring galaxies, Reshetnikov et al. (2006)relied primarily on the morphological appearance of thegalaxy. In particular, the minor merger models rejected by c (cid:13)000
19 km sec − kpc − , where we converted the observational gradients fromequations (1) and (2) to physical units. At a distance to the object of 167 Mpc (H =70 km sec − Mpc − ) the radius of the galaxy to the outermost pointwhere emission lines are visible is ∼ ∼ ◦ to the major axis of the galaxy. The stellar component ob-served with the spectrometer slit oriented along the majoraxis of the ring is also rotating as a solid body and with asimilar velocity-distance gradient to that observed for themain body of the galaxy.Reshetnikov et al. (2006) concluded from their photom-etry and spectroscopy, coupled with results of N-body mod-elling, that AM1934-563 is a PRG. Their models indicatethat the system might be the result of a major interactionbetween a ”donor” galaxy with a 17 kpc stellar disk anda 42 kpc gaseous disk, with a total mass of 3.6 × M ⊙ ,which encountered a 2 × M ⊙ and 14 kpc wide ”recep-tor” galaxy some 1.6 Gyrs ago with an impact parameter of130 kpc and a relative velocity of 145 km sec − . This en-counter transferred a large quantity of matter (stars, gas,and dust) from the donor to the receptor galaxy resultingin the formation of the polar ring which is inclined with re-spect to the galaxy disk and is warped. Reshetnikov et al.(2006) suggested that the donor galaxy survived and isPGC 399718, the southern companion in the triplet, andargued that their suggestion is supported by the reddish(B-V) colour of the galaxy and by its somewhat disturbedappearance.In selecting this scenario in preference to those of mi-nor mergers calculated by them, or of other possible modelsfor the formation of ring galaxies, Reshetnikov et al. (2006)relied primarily on the morphological appearance of thegalaxy. In particular, the minor merger models rejected by c (cid:13)000 , 1–19 Noah Brosch et al.
Figure 12.
Top panel:
The solid line shows the profile of the net H α flux along the slit at PA = 27 ◦ with the continuum subtracted. NEis to the right. The short-dashed line shows the continuum intensity distribution along the slit and near the H α line. Bottom panel:
Theradial velocity distribution along the major axis of the ring of AM1934-563 at PA=27 ◦ . The black squares, red squares and blue trianglesrepresent measurements of the emission lines H α , [N ii ] λ ii ] λ λλ Reshetnikov et al. (2006) produced only partially-open ringsthat were not closed, whereas the preferred major mergermodel produced a “closed and regular ring” a few 10 yearsfollowing the interaction.Since the acceptance of the Reshetnikov et al. (2006)scenario as the explanation for the appearance of this sys-tem relies on their interpretation that the ring is closed andregular, it is worth examining whether the observations pre-sented here support this assertion.The specific items resulting from our observations thatrequire understanding are:(i) Solid-body rotation is observed for stars vs. a ”regu-lar” rotation for the gas at the same (projected) locations.No differential rotation, as expected from a stellar disk, isobserved. This is true for the main body of the galaxy aswell as for the ring, though with the gas showing a differentdistance-velocity gradient than the stars.(ii) The ring is very faint and there is no evidence thatit contains a considerable number of stars, as would be ex-pected from the major merger claimed by Reshetnikov et al.(2006). Our observations of the intensity distribution alongthe slit at PA=27 ◦ show that the stars producing the con- tinuum are located mostly where the HII is, namely some2-5 ′′ NE of the centre.(iii) The ring dynamics are different at its SW end, wherethe line and continuum emissions are very weak and the ringis more extended (Reshetnikov et al. 2006), in comparisonwith the other end of the ring.(iv) The gas dynamics for the ring are very different fromthose of the gas in the galaxy. Specifically, at similar extentsfrom the dynamical centre the gas in the ring spins muchslower than the gas in the galaxy. This, while the stellarcomponents have similar kinematic behaviours as evaluatedfrom the velocity-distance gradients.Apparent solid-body rotation of a galaxy could be pro-duced, for example, by dust extinction. Baes et al. (2003)modelled the light propagation through a dusty galactic diskand showed that, unless the disk is perfectly edge-on, no ef-fects in the kinematics would be observable. The more thedisk is edge on, and the stronger the extinction caused bythe dust in the disk is, the more would the rotation curveresemble that of a solid body. Perusal of the DSS images ofthe object, of the image shown in Fig. 1 of Reshetnikov et al.(2006), and of our Figs. 1 and 13, shows that AM1934-563is not a purely edge-on galaxy and that, since the disk devi- c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Figure 13.
Unsharply masked image in the V-band ofAM1934-563 obtained with SALTICAM. This was cropped fromFigure 1 to show the three galaxies and to emphasize the shapeof the dark lane. ation from edge-on is definitely more than “a few degrees”but rather ∼ ◦ , as explained below, we should not expectto see a solid-body rotation just because of dust obscurationand light scattering. We can, therefore, reject the possibilitythat the solid-body rotation is an effect of dust obscuration. The key observation reported here is the difference in rota-tion curves between the emission lines produced by the gasand the stars as represented by the absorption lines. Suchcases of different kinematic behaviour of the gas and thestars are known in the literature, e.g.,
Bettoni et al. (1990),where NGC 2217 was shown to exhibit “counter-rotation”in that the gas motions in the inner parts of the galaxy indicated motions opposite those of the stars. This was in-terpreted there as a consequence of a warp in the disk cou-pled with the presence of a bar; this situation may exist forAM1934-563 as well.Macci`o et al. (2006) tried to explain the origin of PRGsby accretion of cold intergalactic gas. They provide in theirFig. 4 plots of simulated velocity-position diagrams for gasand stars; the upper one, where the slit is aligned withthe major axis of the galaxy, can be compared with ourFigs. 8 and 9. It seems that the presence of a stellar bar inAM1934-563 could be producing the linearly-rising stellarrotation curve, whereas the rotation curve for the gas fitsthe simulation quite well.Since none of our observations are of photometric-quality, we rely on parameters derived by Reshetnikov et al.(2006) to characterize the galaxy. In particular, weadopt their photometric disk parameters: a disk expo-nential scale length h(B)=5”.1 ± ± ± ± et al. (2004) for the stellar kinematics and fromKregel & van der Kruit (2004) for the gas kinematics. Fig. 6in Kregel et al. shows that the stellar rotation curve can bealmost linear with galacto-centric distance for about 1.5 diskscale lengths and this for galaxies earlier than Sbc. Note thatthis galaxy sample does not include barred galaxies, thoughKregel et al. mention that some do show boxy or peanut-shaped bulges. The gas in none of their galaxies (Kregel &van der Kruit 2004) rotates with as small a gradient withdistance from the center as observed in AM1934-563.It is also possible to compare both the imaged galaxyand its stellar kinematics with the diagnostic plots calcu-lated by Bureau & Athanassoula (2005). Inspection of theirFigs. 1 and 4 indicates that a good fit with AM1934-563could be obtained for an intermediate or strong bar viewedat least at 45 ◦ to the bar or even edge-on, and at a disk in-clination of at least 80 ◦ to the line of sight. The conclusionis that AM1934-563 does probably have a fairly strong barthat is almost side-on to our line of sight, and its disk is seenalmost edge-on.Another comparison for our rotation curve is with thecollection of template rotation curves of Catinella et al. (2006) who, however, studied normal galaxies, not PRGs.They normalize the rotation curves between 2 and 3 diskradii; applying this to AM1934-563, with the peak rotationderived from the curve, indicates that the galaxy should havean absolute I-band magnitude brighter than –23 mag. In-deed, using the photometry from Reshetnikov et al. (2006),with a measured M B ≃ –21 mag and a color index (B-I)=2.06, the absolute I magnitude of AM1934-563 is –23.06mag. This confirms the assumption that, in analyzing thegaseous rotation curve along the major axis, it is a valid as-sumption to adopt the rotation pattern of a regular galaxy,not that of a PRG, since the presence of the polar ring doesnot affect significantly the kinematics of the galaxy. c (cid:13) , 1–19 Noah Brosch et al.
The HI in a number of PRGs, including AM1934-563, wasstudied by van Driel et al. (2002) with the Parkes radiotelescope. This observation produced a puzzling and trou-blesome result for AM1934-563 ; van Driel et al. reportedthe HI line at a heliocentric velocity of 11282 ±
24 km sec − with a full-width at half-maximum of the two-horned profileof 193 km sec − . Note that their data were taken with theParkes multibeam system, which implies a beam width of14 . ′ − bandwidth was centeredat 10000 km sec − and the channel separation was 6.6 kmsec − .If the HI would have been associated with AM1934-563 ,we would expect to find the neutral hydrogen lineat a similar systemic velocity to that measured here,that in Reshetnikov et al. (2001), or that measured byReshetnikov et al. (2006). We would also expect a muchwider HI profile than quoted by van Driel et al. (2002),since the H α kinematics indicate a width of ∼
450 km sec − along the major axis, as befitting a major galaxy givenits bright absolute magnitude of M B =–21.1 measured byReshetnikov et al. (2006). The very wide Parkes beam im-plies that all three objects were included in the measure-ment, and probably many outlying HI clouds that may ex-ist in this neighbourhood as well, but does not explain thevelocity discrepancy since all three galaxies should have ap-peared on the red shoulder of the HI profile shown by vanDriel et al. Another indication that something is wrong with theHI measurement comes from applying the Tully-Fisher re-lation to AM1934-563 . Combes (2006) gives a Tully-Fisherdiagram for PRGs in Fig. 2 of her paper; these galaxiesseem to follow the T-F relation for spirals and S0 galaxiesand it is worthwhile to check where AM1934-563 fits in thisdiagram. Adopting the HI width given in van Driel et al. (2002) indicates that AM1934-563 should have an M B ≃ –18 mag, completely different from the magnitude measuredby Reshetnikov et al. (2006). Adopting a velocity width asmeasured by us albeit from the emission lines and not fromthe HI profile, namely 450 km sec − , yields the proper valueof M B ≃ –21 mag.Irrespective of the explanation regarding the HI redshiftdiscrepancy, it is possible that extended HI is present in thesystem. The possibility that such HI clouds or other gas-richgalaxies might be present is supported by our discovery ofthe H α knot (see below), and by the presence of a few lowsurface brightness (LSB) extended objects in the immediatevicinity. These resemble LSBs the nearby Universe that areoften found to be very gas-rich. In addition, there are a fewvery blue star-like objects that stand out in comparisons ofthe Second Digitized Sky Survey images in different bands.We do not have redshifts for these LSB objects butthe fact that they are of similar sizes to the main galaxiesin the AM1934-563 group hints that they might be groupmembers; such companions are seen in other groups as well( e.g., Grossi et al. α knot discovered by uswill prove to be actually gas-rich members of this group. α knot The H α knot reported above, which is ∼
78 arcsec away tothe NE from the galaxy center but almost at the same ve-locity, is in reality ∼
630 kpc away in projected distance.Its detectable H α emission, combined with a lack of [N ii ],[S ii ] and only weak continuum emissions, argue that thisis probably a metal-poor dwarf galaxy that belongs to thesame group as AM1934-563 . Such objects are known as ”HIIgalaxies” (Sargent & Searle 1970) since they show an HIIregion spectrum with negligible continuum and have con-siderable redshifts.Our fitting procedure to the emission lines, used for thegalaxies and for the ring, allows the derivation of an upperlimit for the [N ii ] λ N ii ] λ / H α ) = − .
46 the metal-licity upper limit is 12+log(O/H) < α -emitting knots in the neighbourhoods of a few star-forming galaxies qualified as “dwarfs” (M B > –18) and lo-cated in some very under-dense regions of the nearby Uni-verse. The study revealed these prospective neighbour galax-ies through the presence of H α emission at or near the cen-tral galaxy redshift. It is possible that the knot found hereis a similar type of object. Spectroscopy of the ring in NGC 4650A has been reported bySwaters & Rubin (2003). They found a ring rotation curvethat seems to flatten out from the center to the North, butwhich is steadily increasing from the center to 20 arcsecon the South side and then flattens out. This is more pro-nounced for the stellar component of the ring than for itsgaseous component. The galaxy itself, an S0 as most PRGsare, shows solid-body-like stellar rotation from the centerto ∼
15 arcsec out while the emission lines show a differ-ent pattern of constant velocity. This they interpret as dueto the galaxy being devoid of gas while the line emissionis produced only in the ring. Comparisons of the rotationalproperties of polar rings and of galaxy disks are valuable inunderstanding PRGs.We return now to the appearance of the stellar rota-tion curves observed at PA = 140 ◦ and at PA = 27 ◦ . Thesecurves, derived from the Na I absorption lines, are very simi-lar. They appear linear for a considerable distance and theirvelocity-distance gradients are ∼
19 km sec − kpc − .We point the reader back to Fig. 4 where the extent ofthe continuum that allows the detection and measurement ofthe Na I D lines is considerably narrower than that for PA =140 ◦ . The discrepancy could be resolved by assuming thatthe absorption lines, and therefore most of the continuum,would not be produced by stars in the ring, as implicitlyassumed in the previous sections, but by stars in the maingalaxy, perhaps in a stellar disk or in a strong bar. TheAM1934-563 inclination can be derived from the axial ratioof the galaxy given in Reshetnikov et al. (2006): i ≃ ◦ . c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited In this case, the angle difference between the two slitpositions, 67 ◦ , would explain the difference in the extent ofthe linear rotation curves at the two position angles as acombination of foreshortening and obscuration by the dustlane. The dust lane produces about one magnitude of ex-tinction, as the intensity profiles along the slit in Fig. 4c ofReshetnikov et al. (2006) show. The sudden disappearanceof the absorption lines only 1 . ′′ α line, plotted with a short-dashedline in the top panel of Fig. 12, supports this interpretation.The ring would, in this case, be composed mostly of gas,would be located between us and the disk with its dark lane,and would necessarily be much less massive than assumedby Reshetnikov et al. (2006). The emission lines measuredwithin ± ′′ of the kinematic centre (see e.g., Fig 12) wouldthen be produced primarily in the disk, while those for PA =27 ◦ but measured at a galacto-centric distance of more than6 . ′′ et al. (2006), specifically those inPanel 6 of their Figure 4, indicates that the location ofAM1934-563, at l ≃ .
02, b ≃ − .
73, corresponds tothe tip of a galaxy filament extending out of the Zone ofAvoidance. This might be a distant structure related to theCentaurus wall and the Norma and Pavo II clusters of galax-ies at lower redshifts, through which intergalactic matter isaccreted by the galaxy and forms the ring.Models of cold gas accretion from cosmic filaments byMacci`o et al. (2006) show how a ring galaxy, such as NGC4650A or for that matter AM1934-563 could be formed bysuch a process. Their simulations show that the accretedgas is not completely cold but rather at 15,000K due to itscollapse within the gravitational potential of the filamen-tary structure. Moreover, they mention that some of the gasmight also be shock-heated by the halo potential.A similar process could take place in AM1934-563 .There is no clear-cut evidence that the ring is closed or re-laxed, or that it has a substantial stellar component. Itsdisturbed appearance at its SW end is more similar to thatof an assemblage of diffuse gas clouds, not of a coherent andrelaxed structure. The NE part is smaller and sharper; it ispossible that accreted gas collides there with itself, becomescompressed and shocked, and reaches higher temperaturesthat produce the enhanced line emission. At this location theaccreted gas could perhaps enter a circular or quasi-circularorbit.An alternative could be that in the AM1934-563 case we are indeed witnessing a merger with a gas-rich galaxy, whichtakes place in a polar configuration. This is, in a way, similarto the major merger scenario of Reshetnikov et al. (2006)with the exception that the ”donor” galaxy would now bethe ring itself. The argument reducing the likelihood of thisexplanation is the lack of a significant stellar continuum fromthe ring, indicating its low mass.
Considering the two gas rotation curves, the one along thegalaxy’s major axis and the other along the ring’s majoraxis, one observation is in order. The two rotation curvesderived from the emission lines extend a similar distancefrom the galaxy’s kinematic centre, are presumably in thesame dark matter potential well if AM1934-563 is indeed aPRG, yet show a completely different full amplitude. Whilethe galaxy major axis rotation curve has a full end-to-endamplitude of ∼
450 km sec − , that for the ring has a fullamplitude of only ∼
240 km sec − . The asymptotic rotationof the ring is slower than the asymptotic rotation of thegalaxy.The formation of PRGs has been studied by Bournaud& Combes (2003) via N-body simulations. They discussed,in particular, cases when both the galaxy disk and the ringcontain gas. Their argument was that in such cases the polarring must, by necessity, be wider than the galaxy. If this isnot the case, the gas in the ring would interact with the gasin the disk and one of the components would join the other.Two orthogonal, or almost orthogonal gas rings, can coexistin the same galaxy only if they have different radii and donot cross each other. Such crossing presumably occurs inNGC 660, where both the disk and the ring contain gas;the N660 system is unstable and according to Bournaud &Combes did not have sufficient time to dissolve the ring sinceits formation.The specific question of the DM halo shape in PRGswas studied by Iodice et al. (2003). They explained thatthe ring material would move slower than the gas in thedisk if the gravitational potential would be oblate, like theflattened disk galaxy. In this case the ring would be ellipti-cal and would show a lower observed velocity than the diskat its outermost locations (see their Fig. 3). In the case ofAM1934-563 , since the ring and the galaxy appear to havethe same size but the ring must be wider in order to avoidcrossing the disk, a possible conclusion would be that thering is elliptical with its major axis close to our line of sightto the object and its minor axis seen almost perpendicularto the disk. This way, the ring could indeed be larger thanthe galaxy, the gas in the ring and that in the galaxy wouldnot cross, and the velocities at the apo-galactic ring loca-tions would be slower than in the galaxy disk. In this casethe outermost visible ring segments would correspond to lo-cations near the ends of the minor axis and ring materialshould show there its highest orbital speed, larger than thatof the galactic disk. As this is not observed, we concludethat the disk and the ring in AM1934-563 are of similarsizes, their contents do cross, and the system is unstable.With the additional kinematic information now avail-able, AM1934-563 could be considered a test case for DMgravitational potential tracing. The discussion of PRGs by c (cid:13) , 1–19 Noah Brosch et al.
Combes (2006) was based on the hope that PRGs wouldprove to be useful probes of the DM potential in which agalaxy and its polar ring find themselves. Combes (2006)found that the rings in observed PRGs show faster rotationthan the maximal velocity observed in the host galaxy. Thetheoretical prediction is in the opposite direction to the ob-servations, namely rings in PRGs devoid of DM halos orwith spherical halos should be rotating slower than theirgalaxies. According to Combes (2006), this effect should beaccentuated for flattened or oblate DM halos.We find that AM1934-563 fulfills the theoretical pre-dictions for non-spherical oblate haloes in that the polarring does rotate slower than its host galaxy. Any DM haloof AM1934-563 , if it exists at all, would have to be flat-tened along the barred disk, but this configuration couldnot be stable on the long run because the ring would crossthe disk. Resolving this possibility and deriving more con-straints on the existence and shape of a possible DM halo forAM1934-563 would require detailed modelling and furtherobservations that are not within the scope of this paper.
We presented observations obtained with SALT and RSSduring their performance verification phase that emphasizethe long-slit capabilities of the RSS for galaxy observations.We traced the stellar and gaseous rotation curves for themajor axis of the galaxy and for the major axis of a po-lar ring-like feature almost perpendicular to the disk of thegalaxy. We showed that, while the gas rotates regularly whensampled along the galaxy major axis, the stellar componentshows rotation like a solid body, supporting an interpreta-tion that this is an object with a strong bar viewed almostside-on.The ionized gas rotation along the major axis of thering was found to be much less regular than along the majoraxis of the galaxy and shows a somewhat shallower gradientwith galacto-centric distance. The Na I stellar rotation fromthe ∼ α knot at a projected distanceof about 700 kpc from AM1934-563 but at a similar velocity,which we interpret as a fourth member of this compact groupof galaxies, presumably a metal-poor dwarf galaxy. The lackof continuum emission for this object while only the H α lineis detected indicates that it might be forming stars for thefirst time. The low velocity dispersion measured from theknot indicates its low mass.We argue that a more plausible explanation to the ma-jor merger scenario proposed by Reshetnikov et al. (2006) toexplain AM1934-563 could be the slow accretion of cold cos-mic gas along a galaxy filament directed to the AM1934-563region. In the cold gas accretion case the flow is probablytowards the galaxy from the South-West and becomes more compressed at the NE end of the polar ring feature. Wepoint out that the kinematic properties we measured followthe theoretical predictions for PRGs in a dark matter halothat is not spherical, but is flattened along the plane of thegalaxy. ACKNOWLEDGMENTS
This paper was written while NB was a sabbatical visitorat the South African Astronomical Observatory in CapeTown; NB is grateful for this opportunity offered by theSAAO management. We are grateful for the generous allo-cation of SALT observing time during the PV phase to com-plete this project. We acknowledge a private communicationfrom Vladimir P. Reshetnikov concerning this galaxy. We ac-knowledge the use of products of the second Digitized SkySurvey produced at the Space Telescope Science Instituteunder U.S. Government grant NAG W-2166. The images arebased on photographic data obtained using the UK SchmidtTelescope. The UK Schmidt Telescope was operated by theRoyal Observatory Edinburgh, with funding from the UKScience and Engineering Research Council (later the UKParticle Physics and Astronomy Research Council), until1988 June, and thereafter by the Anglo-Australian Obser-vatory. The blue plates of the southern Sky Atlas and itsEquatorial Extension (together known as the SERC-J), aswell as the Equatorial Red (ER), and the Second Epoch[red] Survey (SES) were all taken with the UK Schmidt. Ananonymous referee provided some insightful comments thatimproved the clarity of the presentation.
REFERENCES
Allen at al. 1991, MNRAS, 248, 528Baes M., et al., 2003, MNRAS, 343, 1081Bekki K., 1998, ApJ, 499, 635Bettoni D., Fasano G., Galletta G., 1990, AJ, 99, 1789Bournaud F., Combes F., 2003, A&A, 401, 817Brosch, N., Bar-Or, C., and Malka, D. 2006, MNRAS, 368,864Brosch, N. 1985, A&A, 153, 199Brosch, N. 1987, Mercury, 16, 174Buckley D. A. H., Swart G. P., Meiring J. G., 2006, SPIE,6267,Bureau M., Athanassoula E., 2005, ApJ, 626, 159Burgh E. B., Nordsieck K. H., Kobulnicky H. A., WilliamsT. B., O’Donoghue D., Smith M. P., Percival J. W., 2003,SPIE, 4841, 1463Combes, F. 2006, EAS, 20, 97Catinella B., Giovanelli R., Haynes M. P., 2006, ApJ, 640,751Denicol´o, G., Terlevich, R., & Terlevich, E. 2002, MNRAS,330, 69Gonz´alez-Delgado, R.M., Leitherer, C., & Heckman, T.M.1999, ApJS, 125, 489Grossi M., Disney M. J., Pritzl B. J., Knezek P. M., Gal-lagher J. S., Minchin R. F., Freeman K. C., 2007, MNRAS,374, 107Kregel M., van der Kruit P. C., 2004, MNRAS, 352, 787 c (cid:13) , 1–19 he polar ring galaxy AM1934-563 revisited Kregel M., van der Kruit P. C., Freeman K. C., 2004, MN-RAS, 351, 1247Hagen-Thorn, V.A., Shalyapina, L.V., Karataeva, G.M.,Yakovleva, V.A., Moiseev, A.V., and Burenkov, A.N.2005, Astron. Reports, 49, 958Iodice E., Arnaboldi M., Bournaud F., Combes F., SparkeL. S., van Driel W., Capaccioli M., 2003, ApJ, 585, 730Karataeva, G.M., Tikhonov, N.A., Galazutdinova, O.A.,Hagen-Thorn, V.A., and Yakovleva, V.A. 2004, A&A, 421,833Kobulnicky H. A., Nordsieck K. H., Burgh E. B., SmithM. P., Percival J. W., Williams T. B., O’Donoghue D.,2003, SPIE, 4841, 1634Kniazev, A.Y., Pustilnik, S.A., Grebel, E.K., Lee, H., &Pramskij, A.G. 2004, ApJS, 153, 429Kniazev, A.Y., Grebel, E.K., Pustilnik, S.A., Pramskij,A.G., & Zucker, D. 2005, AJ, 130, 1558Iodice E., Arnaboldi M., De Lucia G., Gallagher J. S., III,Sparke L. S., Freeman K. C., 2002, AJ, 123, 195Iodice E., et al., 2006, ApJ, 643, 200Leitherer, C. et al.
The Central kiloparsec of Starbursts and AGN ,J.H. Knapen et al. , eds. San Francisco: ASP, 573Norris R. P., Kesteven M. J., Troup E. R., Allen D. A.,Sramek R. A., 1990, ApJ, 359, 291O’Donoghue D., et al., 2006, MNRAS, 372, 151Persic M., Salucci P., Stel F., 1996, MNRAS, 281, 27 (er-ratum 1996, MNRAS, 283, 1102)Radburn-Smith D. J., Lucey J. R., Woudt P. A., Kraan-Korteweg R. C., Watson F. G., 2006, MNRAS, 369, 1131Reshetnikov V. P., Fa´undez-Abans M., de Oliveira-AbansM., 2001, MNRAS, 322, 689Reshetnikov, V.P. 2004, A&A, 416, 889Reshetnikov, V., Bournaud, F., Combes, F., Fa´undez-Abans, M., and de Oliveira-Abans, M.: 2006,
Astron. As-troph. A TEX file preparedby the author. c (cid:13)000