THE ASTRONOMICAL JOURNAL, 116 :102È110, 1998 July1998. The American Astronomical Society. All rights reserved. Printed in U.S.A.(

DEEP OPTICAL IMAGING OF THE BRIGHT SEYFERT GALAXY NGC 5548 : A LONG, VERY LOWSURFACE BRIGHTNESS TAIL

J. ANTHONY PHILIPPE PURAGRA PETER RICHARDTYSON,1 FISCHER,2,3 GUHATHAKURTA,4,5 MCILROY,6 WENK,7JOHN LUCAS LYMAN VICKI KARLHUCHRA,8 MACRI,8 NEUSCHAEFER,9 SARAJEDINI,10 GLAZEBROOK,11

KAVAN AND RICHARDRATNATUNGA,12 GRIFFITHS12Received 1997 June 26 ; revised 1998 February 19

ABSTRACTDeep Hubble Space Telescope Wide Field Camera (WFC-1) imaging of a region east of the V \ 133@.7

mag Seyfert galaxy NGC 5548 in the F555W and F785LP bands, and deep ground-based V and Iimaging, show a new extended, blue, low surface brightness structure, apparently a remnant of a tidalinteraction associated with NGC 5548. If this straight, low surface brightness (V D 27È28 mag arcsec~2)tidal tail is associated with NGC 5548, it extends at least 80 kpc from it and has an absolute magnitudeof km s~1 Mpc~1). Previous imaging surveys of Seyfert galaxies would haveM

VB [16.4 (H0\ 65

missed such low surface brightness tails. Morphologically similar to some brighter tails seen in otherinteracting systems and in simulations of merging galaxies, this faint tail is a useful diagnostic of anearlier interaction and of the progenitorsÏ halo-to-disk mass ratios. Luminous ripples and a brighter tailwrapped around the galaxy are seen in the inner 1È10 kpc region. Unresolved blue objects in the longtail have the colors and absolute magnitudes of young globular clusters. Taken together, the two tailsand other features suggest that two spirals merged less than B1 Gyr ago. The inner luminosity proÐle ofNGC 5548 is a good Ðt to a de Vaucouleurs proÐle with kpc. Recent simulations of mergedreff \ 5.8galaxies with high-mass halos fail to form lasting tidal tails, suggesting that the NGC 5548 progenitorspirals had modest halo masses.Key words : galaxies : clusters : general È galaxies : individual (NGC 5548) È galaxies : Seyfert

1. INTRODUCTION

While there have been many observations of the nuclearactivity in Seyfert galaxies (see Osterbrock 1991 ; Petersonet al. comparatively little is known of the formation1994),process for these systems. It has long been thought thatinteractions between galaxies could trigger active galacticnucleus (AGN) activity by redistribution of the angularmomentum of nonnuclear material, and its subsequentinfall toward the bottom of the central potential well (Gunn

As discussed below, remnants of disk material at1979).large radii can be a diagnostic of the extent of dark matterhalos in the progenitors.

Most studies of the disturbed morphology of Seyfert gal-

ÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈ1 Bell Labs, Lucent Technologies, 700 Mountain Avenue, Murray Hill,

NJ 07974.2 Department of Astronomy, University of Michigan, 830 Dennison

Building, Ann Arbor, MI 48109.3 Visiting Astronomer, National Optical Astronomy Observatories,

operated by the Association of Universities for Research in Astronomy,Inc., under cooperative agreement with the National Science Foundation.

4 UCO/Lick Observatory, University of California, Santa Cruz, CA95064.

5 Alfred P. Sloan Fellow.6 Department of Computer Science, Rutgers University, Piscataway,

NJ 08855.7 AT&T, 600 Mountain Avenue, Murray Hill, NJ 07974.8 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,

Cambridge, MA 02138.9 BCSi, Inc., Suite 1430, 2 North Nevada Avenue, Colorado Springs,

CO 80908.10 Steward Observatory, University of Arizona, 933 North Cherry

Avenue, Tucson, AZ 85719.11 Anglo-Australian Observatory, P.O. Box 296, Epping, NSW 2121,

Australia.12 Department of Physics, Carnegie Mellon University, 5000 Forbes

Avenue, Pittsburgh, PA 15213.

axies have concentrated on the inner few kiloparsecs, wherethe surface brightness of ripples is high but where thedynamical timescales are only B107 yr. Given that the ac-cretion or ““ feeding ÏÏ can last for B109 yr, the chance ofseeing evidence for the disruption in the inner region maybe less than 1% et al. Theoretical studies(MacKenty 1994).of gravitationally interacting systems have a long history.Following the early spiral galaxy interaction N-body testsof & Toomre made aPÑeiderer (1963), Toomre (1970)movie showing tidal disruption of interacting spirals andthe creation of a pair of thin tails. foundWright (1972)nonescaping counterarms in his studies ; & ToomreToomre

Ðrst showed that long escaping tails can be remnants(1972)of the collision of two disk systems. According to recentsimulations of interacting disk galaxies, longer-lived fea-tures, such as tidal tails, should exist at radii as large as 100kpc early in the encounter Barnes & Hern-(Mihos 1995 ;quist However, late remnants of such events1992a, 1992b).in simulations seldom resemble Seyfert galaxies, being mor-phologically more similar to elliptical and peculiar galaxies.Long tails are observed in a small fraction of ““ peculiar ÏÏgalaxies and these(Arp 1966 ; MacKenty 1990 ; Malin 1997),features are expected to fade over time to the point wherethey may escape detection in shallow surveys.

In general, there are two scenarios by which long tails canbe created : (1) capture of a massive companion and (2) atidal encounter without capture. In the second case, onemight expect to see the scattered object nearby in the Ðeld,provided it is sufficiently luminous. Tidal interactions(without merging) are sufficient to induce long-term feedingof the nucleus. Given enough mass, dwarf galaxies may playan important role in the assembly of bright galaxies and thefueling of compact objects in their nuclei (see & WyseSilk

Frenk, & White The fraction of gal-1993 ; Navarro, 1994).axies that exhibit AGN activity is ultimately related to the

102

NGC 5548 103

supply of massive candidates for accretion and tidallyinduced feeding of the nucleus.

It is possible that dwarf galaxies dominate the galaxymass function, resulting in more frequent encountersinvolving a bright (L*) galaxy and a dwarf than thoseinvolving a pair of bright galaxies. A large population ofhigh-M/L objects, each having roughly the mass of an L*galaxy, could inÑuence the frequency of AGN activity in anobservable way. & Quinn argued thatHernquist (1989)mergers of spirals with dwarf galaxies cause shells orripples, while & Spergel have shown thatHernquist (1992)collisions of equal-mass progenitors can also create shells.This latter scenario could result in a long tidal tail as well.Recently, Mihos, & Hernquist andDubinski, (1996) Mihos,Dubinski, & Hernquist performed a numerical study(1998)of the e†ects of massive halos on the creation of tails ingalaxy mergers. They Ðnd that if the halo mass is more than10 times the sum of disk and bulge mass, the formation oflong tidal tails is strongly suppressed. For merger remnantsthat exhibited long tidal tails in their simulations, the pro-genitors had halo masses that were signiÐcantly smallerthan expected in standard cold dark matter scenarios. Asdiscussed below, these considerations may be relevant toNGC 5548.

NGC 5548 has been extensively studied as a prototypicalSeyfert galaxy, and in two surveys the possibility of an inter-action was noted. & Seitzer carried outSchweizer (1988)CCD imaging of 145 E and S0 galaxies in the Second Refer-ence Catalogue of Bright Galaxies and found that NGC5548, with three ripples intersecting the spiral arms, was oneof nine disk galaxies in their sample with ripples. In a CCDsurvey of 51 Seyfert galaxies, remarkedMacKenty (1990)that NGC 5548 was among the 25%È35% of the galaxies inhis sample that exhibited some evidence of distorted mor-phology. The distorted morphology he studied was at fairlyhigh surface brightness, in the inner few kiloparsecs. In thispaper, we present evidence for a very low surface brightnesstidal tail or streamer in this system, extending beyond 50h~1 kpc, and discuss its dynamical implications.

2. OPTICAL OBSERVATIONS

In terms of limiting surface brightness, these data areamong the deepest preÈHubble Deep Field imaging by theHubble Space Telescope. During spectroscopic monitoringof the broad-line region of the nucleus of the Seyfert galaxyNGC 5548 with the prerepair Hubble Space TelescopeÏs(HST ) Faint Object Spectrograph et al.(Korista 1995),parallel Wide Field Camera (WFC-1) data in F555W andF785LP were obtained of a region centered from the3@.7galaxy as a part of the HST Medium Deep Survey. TheHST F555W Ðlter is very close to Johnson V , while theF785LP Ðlter is close to the I@ Ðlter used in the ““BjRI@photometric system ÏÏ et al. As discussed(Gullixson 1995).below, we tie our V and I photometry to the Johnson-Cousins system. In the rest of this paper, we use theV [ICterms ““ V ÏÏ and ““ I ÏÏ to refer to these HST bands. A total of20 orbits in V (30,000 s) and 29 orbits in I@ (64,500 s) wereobtained during a 39 day period beginning on 1993 April 19UT. These images were Ñat-Ðelded using the Medium DeepSurvey ““ superÑats ÏÏ et al. et al.(Phillips 1994 ; Ratnatunga

an improvement over the standard Ñat-Ðeld images1994),then used in the Space Telescope Science Institute data-processing pipeline. The HST solar array deployment elec-tronics had malfunctioned immediately prior to these

observations, so the solar arrays were allowed to follow theSun and the WFC-1 swept out an arc around NGC 5548during the 39 day observing period. These WFC-1 expo-sures were registered by translation and rotation. Thesealigned exposures were Ðrst used to carry out internalchecks of photometric consistency and were then combinedto optimize the signal-to-noise ratio. The resulting interme-diate image was used as a support function for cosmic-rayremoval from the individual exposures. These cleaned expo-sures were then reregistered and combined via medianaveraging to make the Ðnal image.

A color image constructed from these WFC-1 V and Idata is shown in One can clearly see the large blueFigure 1.low surface brightness (LSB) nebulosity near the center ofthis image. The LSB feature covers about 0.8 arcmin2 andhas a mean surface brightness of mag arcsec~2k6

V\ 27.7

and a peak surface brightness of mag arcsec~2.kV,peak\ 27

Also visible is an apparent ““ arc ÏÏ of galaxies cutting acrossthe LSB feature in the northeast part of this image. At thenorth end of the arc, the Ðeld includes a group or cluster ofgalaxies, the brightest member of which has a redshift of0.174. To increase the angular resolution of these WFC-1data, we have also processed this image with a modiÐedLucy-Hook algorithm (discussed below).

Follow-up ground-based images were obtained with theKitt Peak National Observatory (KPNO) 4 m telescope on1994 March 13È15 with a 2048 ] 2048 Tektronix CCD

pixel~1). A cumulative exposure time of 4500 s in V(0A.47and 6000 s in I, consisting of slightly disregistered 500 sexposures, were obtained in nonphotometric conditions.These data were combined to make deep 16@] 16@ V - andI-band images. The central 8@] 8@ portion of the combinedV image is shown in and the full image is usedFigure 2,below in The peak of the LSB object identiÐedFigure 5.in the HST image can be seen east and south3@.4 1@.3of the center of NGC 5548 [a(center)\ 14h15m43s.49,d(center)\ 25¡22@01A, B1950.0], a projected length of 52h~1 kpc at the 50 h~1 Mpc distance of NGC 5548. Theseground-based observations further reveal that the LSBobject extends all the way into NGC 5548. (The jetlikenegative feature extending directly east of the nucleus ofNGC 5548 is the residual of an undershoot problem in theKPNO CCD camera preampliÐer electronics, which wehave partially corrected in processing. This is due to ACcoupling to the preampliÐer and the absence of clampingduring reset, which causes an undershoot for many pixelsalong a line extending to the left [downstream] of brightsaturated pixels.)

In order to calibrate these KPNO 4 m data, and to searchfor line emission, several photometric runs on three othertelescopes were obtained. Follow-up CCD photometry wasobtained in the UBV RI and redshifted Ha (j0\ 6650 A� ,*j\ 100 bands, using the F. L. Whipple ObservatoryA� )1.2 m telescope on Mount Hopkins in photometric condi-tions. As an aid to photometric monitoring of this Ðeld, welist in the J2000.0 coordinates, Johnson B magni-Table 1tude, Johnson B[R colors, and associated errors for 99stellar objects in the 10@ Ðeld centered on NGC 5548. Five ofthe brightest stars had saturated R images ; their B[Rcolors are listed as 0.00^ 0.00. The Steward 2.3 m telescopewas used for BV R photometry of the Ðeld, also in photo-metric conditions. Additional narrowband images in theredshifted Ha and redshifted [O III] bands(jrest \ 5007 A� )were obtained with the 0.6 m Burrell Schmidt telescope on

FIG. 1.È““ True color ÏÏ reconstruction from WFC-1 exposures totaling 30,000 s in F555W (V ) and 60,000 s in F785LP (I), obtained as part of the HSTMedium Deep Survey. The Ðeld is from the center of the bright galaxy NGC 5548 (V \ 13 mag) and measures As indicated, the vertical axis of3@.7 2@.4] 2@.3.this image is oriented counterclockwise relative to north. The faintest objects seen are V \ 25È26 mag, some of which are part of the faint background17¡.3blue galaxy population, and some of which may be dwarf systems forming in the shocked gas in a tidal tail resulting from an accretion event in NGC 5548.Evidence for the brighter part of a very low surface brightness tail may be seen in the center of this image in the form of V D 31 mag arcsec~2 blue nebulosity.Individual galaxies of such a low surface brightness would be undetectable in this image if their angular sizes were A concentration of V \ 20È22 mag[1A.red elliptical galaxies (possibly a high-redshift cluster) can also be seen in the western part (right) of the image.

TABLE 1

BR PHOTOMETRY IN THE NGC 5548 FIELD

ID a (J2000.0) d (J2000.0) B B[R

NGC 5548È01 . . . . . . 14 17 48.31 ]25 05 08.0 21.48 ^ 0.08 0.97 ^ 0.15NGC 5548È02 . . . . . . 14 17 48.54 ]25 05 49.3 21.47 ^ 0.08 0.84 ^ 0.14NGC 5548È03 . . . . . . 14 17 49.14 ]25 02 58.4 21.49 ^ 0.07 2.56 ^ 0.12NGC 5548È04 . . . . . . 14 17 49.15 ]25 07 50.4 21.78 ^ 0.10 1.91 ^ 0.16NGC 5548È05 . . . . . . 14 17 49.91 ]25 08 10.9 21.57 ^ 0.09 2.71 ^ 0.13

NOTES.ÈTable 1 is presented in its entirety in the electronic edition of the Astro-nomical Journal. A portion is shown here for guidance regarding its form and content.Units of right ascension are hours, minutes, and seconds, and units of declination aredegrees, arcminutes, and arcseconds.

NGC 5548 105

FIG. 2.ÈDeep V -band image of an 8@] 8@ portion of the NGC 5548 Ðeld, showing the HST Ðeld to the east of the Seyfert galaxy. This image wasreconstructed from a total of 4500 s of V -band KPNO 4 m exposures. North is up and east is left. The faintest features seen are V \ 28 mag arcsec~2. Notethat the blue LSB tail detected in extends all the way into the nucleus of NGC 5548.Fig. 1

Kitt Peak. The central wavelengths of these narrowbandÐlters are and III])\ 5146 andj0(Ha) \ 6658 A� j0([O A� ,their bandwidths are *j(Ha) \ 84 and *j([O III])\ 177A�

The total integration time was 6000 s in each of the HaA� .and [O III] bands. No evidence of emission lines was found,to 1 p of the sky noise in these images. Finally, we calibratedour V and I magnitudes on the standard Johnson V andCousins I system; V [I colors quoted below are all trans-formed onto the system. In one exception, notedV [ICbelow, we compare with models of young globular clustercolors in the Johnson system.

The software package FOCAS was used to generate acatalog of photometry for all objects. Surface photometry ofthe LSB structure was obtained by recursively cleaningcompact objects (stars and distant galaxies) and rebinningthe image.

3. RADIO AND OPTICAL VELOCITY DATA

NGC 5548 is a 21 cm line source with a Ñux of B1.3 Jykm s~1 Clarke, & Fricke & Bier-(Biermann, 1979 ; Biegingmann which corresponds to about 8 ] 108 h~21983), M

_of neutral hydrogen at a distance of 50 h~1 Mpc. Thecentral peak of the radio continuum emission (see &WilsonUlvestad is within 1A of the optical centroid position1982)of the nucleus (note that there may be a systematic o†set ofup to 1A between the optical and radio coordinate systems).There are three sources within 5A of the center. There are

also seven other radio continuum sources in the vicinity ofNGC 5548, and one of them (Wilson & Ulvestad source 5,unresolved) is very near the end of the 52 h~1 kpc LSB tail(o†set east and south from NGC 5548Ïs nucleus) and4@.0 0@.8has a continuum Ñux density of 21 mJy at 1465 MHz (20.5cm). This last source could of course be a chance projection.NGC 5548Ïs H I velocity is 200 km s~1 higher than itsoptical velocity (see et al. Finally, theHeckman 1983).optical absorption velocity of NGC 5548 is 210 km s~1higher than the optical emission velocity. We take this to bean indication of the velocity spread in the interactingsystem. NGC 5548 has a ratio of H I mass to blue lumi-nosity that is lower than any of[MH I/L B

B 0.02 (M/L )_]

the Seyfert galaxies studied by & BiermannBieging (1983),which are in the range 0.05È0.63 (M/L )

_.

4. INNER PROFILE AND RIPPLES

The region within 7 h~1 kpc of NGC 5548Ïs center isdominated by the bright central source and host galaxy disklight. In we plot the surface brightness proÐle ofFigure 3,NGC 5548 out to a projected radius of 100A (24 h~1 kpc)based on the KPNO 4 m deep V -band data. In this region,the brightness proÐle is in good agreement with an r1@4proÐle having (3.8 h~1 kpc) and magreff \ 16A k

V,eff \ 22.0arcsec~2. In order to reveal departures from a smooth r1@4-law proÐle, we subtract from the V - and I-band images of

106 TYSON ET AL. Vol. 116

FIG. 3.ÈV -band surface brightness proÐle of NGC 5548 plotted vs.r1@4. The solid line is the best-Ðt de Vaucoulers proÐle. The pixel scale is

pixel~1, and the e†ective radius of the de Vaucoulers proÐle is0A.47 reff \16A (3.8 h~1 kpc). This is strong evidence for an elliptical-like remnant. Theripple at 6.7 h~1 kpc radius is clearly visible.

NGC 5548 an azimuthally symmetric (circular) model withbased on this proÐle. These V - and I-band di†er-reff \ 16A

ence images are combined into a ““ stretched color ÏÏ image inThe full range in color coding here corresponds toFigure 4.

0.3 mag in V [I. One can see that there remain a number ofbright blue partial ring structures (““ ripples ÏÏ). These ripplesare morphologically indistinguishable from remnants ofinner spiral arms. There is no obvious exponential diskcomponent in the V light. While the hump in the V proÐlefrom r \ 23A to 39A could be interpreted as a signature of aweak disk as seen in some Sa or S0/a galaxies, it is insteaddue to the curved arm & Seitzer as may be(Schweizer 1988),seen in our Figures and The overall proÐle is best Ðtted2 4.by an r1@4 law, suggesting that an elliptical galaxy may be inthe process of emerging from this encounter. In a com-prehensive study, et al. found a similarHibbard (1994)brightness proÐle for the late-stage merger remnant NGC7252.

The ripples have a surface brightness of V B 23È25 magarcsec~2 and are morphologically typical of those seen inother shell galaxies & Carter and in a variety(Malin 1983)of tidal encounter simulations with a satellite whose mass isgreater than about 20% of the mass of the host galaxy

et al. The blue color (V [ID 0.8) of these(Howard 1993).ripples is consistent with shock-induced rapid star forma-tion ; it is also consistent with a scenario in which themerging satellite galaxy has a bluer (younger) stellar popu-lation than the larger host galaxy. These features are remi-niscent of the shell galaxy NGC 5018, which contains ano†-center dust lane and optical and H I plumes, suggestiveof a recent interaction or merger event et al.(Kim 1988 ;

et al. In fact, about 17% of all ellipticalGuhathakurta 1990).galaxies show ripples, shells, or both & Carter(Malin 1983).These shells or ripples are often observed to be bluer thanthe mean color of the host galaxy (Prieur 1990 ; Forbes,Reitzel, & Williger Such ““ Ðne structure ÏÏ is thought1995).

to be caused by recent merger events (cf. &HernquistQuinn & Spergel1989 ; Hernquist 1992).

An independent means of subtracting the bright Ñux nearthe nucleus is to subtract a scaled I image from the V image,yielding a map of the V [I ““ color.ÏÏ Such a subtraction isnot biased by a particular choice of model for the lightdistribution, but it only reveals features whose colors aredi†erent from the average color of the galaxy. A scalingfactor of 0.5 for the I image was found to cancel much of theinterior Ñux in the V image. We made a monochrome imagefrom the logarithm of the image, where andF

V[ 0.5F

IF

Vare the instrumental Ñuxes in the V and I bands. TheFImorphology of the ripples seen in this image is no di†erent

than that seen in the model ÐtÈsubtracted color image inimplying that there are no systematic errorsFigure 4,

resulting from the details of the model used in generatingFigure 4. Note that the V [I color of the galaxy becomessomewhat bluer with decreasing radius ; in addition, thenucleus of NGC 5548 is very blue in color.

5. YOUNG GLOBULAR CLUSTERS IN THE LSB TAIL

Several studies have shown that galaxies that have under-gone recent mergers contain bright blue unresolved, orbarely resolved, sources & Chu(Kennicutt 1988 ; Lutz 1991 ;

et al. et al. & VaccaHoltzman 1992 ; Whitmore 1993 ; ContiGallagher, & Hunter et al.1994 ; OÏConnell, 1994 ; Shaya& Schweizer & Smecker-1994 ; Whitmore 1995 ; Crabtree

Hane It is commonly believed that these objects are1994).young globular clusters, whose formation is thought to betriggered by galaxy-galaxy mergers (Schweizer 1987 ;

& ZepfAshman 1992).The faint blue galaxies in the background naturally con-

tribute to the blue object counts everywhere in the image.Down to a limiting magnitude of V \ 26, there are D30faint objects per square arcminute, mostly background Ðeldgalaxies. However, it is possible that some of the objectsseen in the direction of the faint LSB tail in NGC 5548 arephysically associated with it. shows smoothed iso-Figure 5pleths (contours of constant surface number density) forobjects with 22¹ V ¹ 26 detected by FOCAS in theKPNO 4 m image, overlaid on the image. A Gaussiansmoothing radius of p \ 24A was used. There are two sig-niÐcant peaks in the isopleths, one corresponding to whatappears to be a high-redshift cluster of faint red ellipticalgalaxies and the other a clump of blue objects centered nearthe peak of the LSB object. Many of these blue objects inthe LSB tail appear to have half-light angular sizes ([0A.2)smaller than the average value for the background faintblue galaxy population at comparable magnitudes.

In order to improve the angular resolution of these pre-repair HST WFC-1 images, we developed software that usesa modiÐed Lucy-Hook approach for a maximum residuallikelihood reconstruction based on data from all the HSTand ground-based imaging. The resulting V image covers awide range of surface brightness while having stellar imageswith FWHM less than The V [I colors of the faint0A.1.objects detected on this reconstructed image span the range0 \ V [I\ 4 ; there are many ““ faint blue objects ÏÏ withV [I\ 1. About half of these faint blue objects are unre-solved or barely resolved in this reconstructed HST image,with scale lengths less than (20 h~1 pc at the distance of0A.1NGC 5548). Blue knots corresponding to emission at rest-frame ultraviolet wavelengths have been found in high-redshift galaxies studied with HST et al.(GrifÐths 1994 ;

No. 1, 1998 NGC 5548 107

FIG. 4.ÈColor residual image in the V and I bands of a Ðeld centered on NGC 5548 (45 h~1 kpc on a side at the distance of NGC 5548), in which3@.1 ] 3@.1an r1@4 law has been Ðtted and subtracted from each of the two bands. This image was made from a total of 10,500 s in V and I on the KPNO 4 m telescope.North is up and east is left. The faintest features seen are V D 26 mag arcsec~2. The bright blue ripples (V [ID 0.8), characteristic of tidal encounters, may beseen at projected radii of 6.5 and 3.7 h~1 kpc, and at smaller radii. The full range in V [I color in this ““ stretched color ÏÏ picture is 0.3 mag.

Tyson, & Turner Our observational searchColley, 1996).for related emission lines in NGC 5548Ïs tidal features hasbeen limited by their low surface brightness. We havesearched unsuccessfully for redshifted Ha and [O III] emis-sion from the LSB tail and the compact faint objects in it.However, in a spectroscopic study of the merger remnantNGC 7252 & Seitzer no emission lines(Schweizer 1993),were found in the blue knots.

These unresolved objects may be young globular clusterset al. et al. or dense star-(Whitmore 1993 ; Huchra 1996)

forming regions. A few of these faint blue objects in theHST Ðeld are more di†use and appear to have colors, mor-phologies, and angular sizes consistent with those ofextreme dwarf galaxies around NGC 5548. However, mostof them appear to be associated with the LSB tail region.Most objects in the WFC-1 Ðeld span the magnitude(Fig. 1)range 19 \ V \ 25 and have colors ranging betweenV [IC1 and 3. The compact blue objects in the LSB tail regionhave mean color. Of the 19 objectsV [IC\ 0.42^ 0.04brighter than 25.5 V mag in the whole Ðeld, and bluer than

16 of them are within the area of the LSB tail.V [IC \ 0.5,

Either this is a chance excess of projected objects or thesecompact blue objects are physically associated with the LSBtail itself. Most of these faint blue objects, compact anddi†use, have V \ 23È25 mag, corresponding to absolutemagnitudes of at the distance of NGC[11.4\M

V\ [9.4

5548 (for km s~1 Mpc~1).H0\ 65

6. DISCUSSION AND TESTS OF EVOLUTIONARY SCENARIOS

We Ðnd strong evidence supporting the hypothesis thatNGC 5548 has experienced a recent interaction : the moder-ately low surface brightness arm or curved tail wrappingaround the galaxy at 18 h~1 kpc radius, the long straightLSB tail (50 h~1 kpc), and the ripples in the region interiorto 7 h~1 kpc radius. CCD data for NGCMacKentyÏs (1990)5548 show the outer 18 h~1 kpc curved arm but not theLSB tidal tail (Figs. and or the inner ripples1 2) (Fig. 4).This straight LSB tail is the second tidal tail feature dis-covered in NGC 5548, the Ðrst being the curved tail ordistorted arm at 18 h~1 kpc radius & Seitzer(Schweizer

If this arm were unwrapped, it would be comparable1988).in length to the LSB tail. Both the distorted arm at 18 h~1

108 TYSON ET AL. Vol. 116

FIG. 5.ÈSmoothed number density contour plot for V \ 22È26 mag objects in the 14@] 14@ Ðeld around NGC 5548, superposed on the KPNO 4 mV -band image. North is up and east is left. The lowest level contoured corresponds to 21.9 arcmin~2 (1 p above mean), and subsequent contours are spacedby 3.1 arcmin~2 (0.5 p). There is an excess of faint blue objects in the region of the ““ tail ÏÏ to the southeast, and another associated with faint red galaxies in anapparent high-redshift cluster to the east of NGC 5548Ïs nucleus.5@.5

kpc and the brighter ripple at 7 h~1 kpc are common inter-action features, and appear in a variety of simulations ofretrograde disk-disk encounters p. 489) and(Barnes 1992,diskÈmassive companion encounters. Such features arefairly transient phenomena, typically lasting 1 Gyr (J. C.Mihos 1995, private communication). The ripples near thenucleus have even shorter lifetimes et al.(Howard 1993).

There is evidence that the a priori probability for inter-action was high for NGC 5548 : it appears to lie inside a““ wall ÏÏ structure with high galaxy density. showsFigure 6the large-scale distribution of galaxies in the neighborhoodof NGC 5548. In this declination wedge, it is clear thatNGC 5548 is located in an overdense structure with voidsor low-density regions at both higher and lower redshifts.A similar view is also seen in a right ascension wedge.This overdensity is roughly a factor of 3 larger than themean (depending somewhat on the deÐnition of regionboundaries). This increase in the local density of galaxiesaround NGC 5548 also implies an increase in the probabil-ity of interaction, up to a factor of 10 higher than average.

The straight blue LSB tail seen in both the HST andground-based data is not a common feature of existing

simulations of galaxy interactions : most simulations do notproduce straight tidal tails. Observationally, such a straighttail is not without precedent ; morphologically similar long,but brighter, tails have been seen in a few images of inter-acting galaxies, notably Arp 174, Arp 178, and NGC 7252.Both elliptical and early-type disk galaxies feature such tailsmore often than previously thought It is likely(Malin 1997).that the LSB wings in a galaxyÏs outer proÐle could getstretched out in an encounter. Moreover, tidal tails fadewith time as they get stretched out. The tail seen in NGC5548 has a low surface brightness ; we speculate that suchLSB tails may be fairly common but have been missedbecause of the relatively high surface brightness limits ofmost Seyfert imaging studies.

As an example, assuming a relative velocity of 200 kms~1 between the main body of NGC 5548 and the materialin its tidal tail and a 45¡ inclination of the tail to the line ofsight, the resulting 80 h~1 kpc (true length) tail in NGC5548 would have taken 0.6 Gyr to generate (for kmH0\ 65s~1 Mpc~1). shows that the end of the tail hasFigure 2about twice the surface brightness of the rest : V \ 27 magarcsec~2. This is consistent with the tail falling back on

No. 1, 1998 NGC 5548 109

FIG. 6.ÈDeclination vs. redshift cone diagram showing the distribution of galaxies in a 10¡ box centered on NGC 5548 (large circle) and within 9000 kms~1. The galaxy velocities are from the latest version of the CfA Redshift Catalog (J. Huchra 1997, private communication). NGC 5548 is embedded in anoverdense region, giving rise to an enhanced interaction probability.

itself, a feature that ultimately occurs in all the tidal encoun-ter simulations that produce tails. Of course, we cannot tellthe geometry of the tail ; we may be looking through acurved tail in the plane of the curve, or most of the tail maybe straight. It is instructive to review simulations of binarygalaxy encounters. The simulations show a range of suchtail geometries. If we normalize to the time steps used insimulation 78 of et al. the estimatedHoward (1993),dynamical age of the tail in NGC 5548 would be B1 Gyr.

In order to produce these strong features in a tidalencounter, the interacting companion galaxy must have amass at least 20% that of NGC 5548 et al.(Howard 1993).Most simulated encounters, however, produce tails that donot point straight back toward the nucleus of the hostgalaxy. Nearly polar encounters between galaxies of modesthalo masses can produce a one-sided projected tail pointingback nearly to the nucleus ; this is shown in simulation 78,time steps 400È450, in the video of et al. InHoward (1993).this simulation of a polar encounter, the progenitors hadequal masses and each had unity disk-to-halo mass ratio, sothat the disks were dynamically active. Comparing this withthe case of NGC 5548 seen in our and inferringFigure 2,from the spiral arms that in that Ðgure NGC 5548 is rotat-ing counterclockwise, we see that the LSB tail in NGC 5548points back slightly to the same leading side of the nucleusas the tail in the dynamical simulations of Howard et al.Note that the ripples in this simulation are similar to thoseobserved. While modest companion masses and largeimpact parameters seem to be required to make very longtails, there is no evidence for a companion galaxy of lumi-nosity greater than 10~3L* within 100 h~1 kpc of thenucleus on NGC 5548. Thus, either the LSB tail is theremnant of an interaction with a high-M/L [Z1000

dark companion or the event resulted in a com-(M/L )_

]plete merger. The galaxy just to the west and in the halo ofNGC 5548 has a velocity of 42,000 km s~1 and is thus in thebackground. The V \ 17 blue, di†use clump(M

V\ [16),

of luminosity 2@ north-northeast of the nucleus is at the endof the curved inner tail discovered by & SeitzerSchweizer

It is likely that this inner wrapped tail and the long(1988).LSB tail are remnants of spiral arms of two spiral galaxiesthat merged. The LSB tail extending from NGC 5548 hasan absolute magnitude of log h, comparedM

VB [12 ] 5

with log h for NGC 5548. In summary,MV

\ [21.55] 5for the tidal interaction scenario, there do not appear to beany other candidates for remnants of the interacting galaxyin the 16@ Ðeld. A more plausible scenario is a completemerger.

In a study of young globular clusters in the interactingpair NGC 4038/4039, & Schweizer foundWhitmore (1995)

mag and for the nonreddenedMV

D[11 V [IJ D 0.36globular clusters. These are very similar to the absolutemagnitudes and colors of the compact objects we Ðnd in theLSB tail of NGC 5548. Comparing our V [I colors(converted to the Johnson system) with the models of

& Charlot we can estimate their age.Bruzual (1993),Assuming that these are young globular clusters, their meancolor of then suggests an age of 0.1 Gyr. FromV [IJ \ 0.54the same model, the absolute magnitude of the brightest ofthem, at is more consistent with an age of 0.2M

V\[11.4,

Gyr. This is compatible with the dynamical age limit for theinteraction tail of less than 1 Gyr.

Tidal debris are a diagnostic of dark matter halos (Faber& Gallagher et al. examine simulations1979). Mihos (1998)of binary collisions of galaxies with bulges and extendeddisks, studying the tidal tails made from loosely boundmaterial. Mergers of galaxies with high halo masses lead totidal tails that fall back onto the parent galaxies before theÐnal merger. In their modeling of NGC 7252, Mihos et al.Ðnd a halo mass much less than the sum of the disk andbulge masses. SigniÐcantly, they Ðnd that only low-masshalos produce massive curving tails. At high halo mass,tidal tails become shorter and disappear. On the basis ofthese general results, and from the appearance of NGC

110 TYSON ET AL.

5548, we must conclude that the progenitors in the NGC5548 system had a relatively small halo mass. Given theseobservations, it would be useful to explore detailed binarycollision models for NGC 5548.

There is one other Seyfert galaxy that has apparentlybeen caught shortly after a merger or encounter, Mrk 315.The morphology of the inner ripples seen in NGC 5548 isstrikingly similar to the morphology of the inner severalkiloparsecs of Mrk 315 et al. compare(MacKenty 1994 ;their Fig. 1c with our Since the ripples in Mrk 315Fig. 4).were found to have strong Ha emission, the authorsexplained them as starburst regions, triggered by a recentmerger. The ripples in NGC 5548 extend out to larger radiithan in Mrk 315, which also has a long Ðlamentary struc-ture that may be similar to the LSB tail seen in NGC 5548,but of higher surface brightness.

7. SUMMARY

Like Mrk 315, NGC 5548 may be an example of a mergercaught in the act. Long, low surface brightness tails of thesort found in NGC 5548 may be a common feature inmerging galaxy pairs, as some simulations suggest, but theywould have been missed in previous brighter surveys (see

With no evidence for a companion galaxy ofMalin 1997).

appropriate luminosity in the Ðeld, the two tails and innerripplesÈpossibly remnants of spiral armsÈsuggest themerging of two spiral galaxies. In any case, it appears thatthe progenitors had small halo masses. Ultimately, otherAGNs should be imaged deeply to search for faint blue tails.

One of the purposes of the HST Medium Deep Surveywas serendipity. Given the number of pointed GeneralObserver observations of bright objects, parallel imagingdata will often be centered on Ðelds that are adjacent tointeresting bright objects. It is therefore not surprising thatfaint associated features might be discovered in long paral-lel exposures with the Wide Field Camera in an o†set Ðeld.

We acknowledge helpful discussions with Chris Mihosand Alex Toomre, useful E-mail from Steve Zepf, and veryhelpful comments from the referee. Nigel Sharp obtainedsome of the Ha and [O III] data (Burrel Schmidt), andRogier Windhorst obtained a Ðeld galaxy redshift (Steward2.3 m telescope). This paper is based partially on obser-vations with the NASA/ESA Hubble Space Telescope,obtained at the Space Telescope Science Institute, which isoperated by the Association of Universities for Research inAstronomy, Inc., under NASA contract NAS 5-26555. ThisMedium Deep Survey work was funded by STScI grantsGO-2684.0X-87A.

REFERENCESH. 1966, Atlas of Peculiar Galaxies (Pasadena :Arp, Caltech)

K. M., & Zepf, S. E. 1992, ApJ, 384,Ashman, 50J. E. 1992, ApJ, 393,Barnes, 484J. E., & Hernquist, L. 1992a, ARA&A, 30,Barnes, 7051992b, Nature, 360,ÈÈÈ. 715J. H., & Biermann, P. 1983, AJ, 88,Bieging, 161

P., Clarke, J. N., & Fricke, K. J. 1979, A&A, 75,Biermann, 19A., G., & Charlot, S. 1993, ApJ, 405,Bruzual 538

W., Tyson, J. A., & Turner, E. L. 1996, ApJ, 461,Colley, L83P. S., & Vacca, W. D. 1994, ApJ, 423,Conti, L97

D. R., & Smecker-Hane, T. 1994, BAAS, 26,Crabtree, 1499J., Mihos, J. C., & Hernquist, L. 1996, ApJ, 462,Dubinski, 576

S. M., & Gallagher, J. S. 1979, ARA&A, 17,Faber, 135D. A., Reitzel, D. B., & Williger, G. M. 1995, AJ, 109,Forbes, 1576R. E., et al. 1994, ApJ, 435,GrifÐths, L19

P., Knapp, G. R., van Gorkom, J. H., & Kim, D.-W. 1990,Guhathakurta,in The Interstellar Medium in External Galaxies, ed. D. J. Hollenbach &H. A. Thronson, Jr. (NASA CP-3084) (Washington : NASA), 26

C. A., Boeshaar, P. C., Tyson, J. A., & Seitzer, P. 1995, ApJS, 99,Gullixson,281

J. E. 1979, in Active Galactic Nuclei, ed. C. Hazard & S. MittonGunn,(Cambridge : Cambridge Univ. Press), 213

T. M., Balick, B., van Breugel, W. J. M., & Miley, G. K. 1983,Heckman,AJ, 88, 583

L., & Quinn, P. J. 1989, ApJ, 342,Hernquist, 1L., & Spergel, D. N. 1992, ApJ, 399,Hernquist, L117

J. E., Guhathakurta, P., van Gorkom, J. H., & Schweizer, F.Hibbard,1994, AJ, 104, 67

J. A., et al. 1992, AJ, 103,Holtzman, 691S., Keel, W. C., Byrd, G., & Burkey, J. 1993, ApJ, 417,Howard, 502J. P., Brodie, J. P., Caldwell, N., Christian, C., & Schommer, R.Huchra,

1996, ApJS, 102, 29R. C., Jr., & Chu, Y.-H. 1988, AJ, 95,Kennicutt, 720

D.-W., Guhathakurta, P., van Gorkom, J. H., Jura, M., & Knapp,Kim,G. R. 1988, ApJ, 330, 684

K. T., et al. 1995, ApJS, 97,Korista, 285

D. 1991, A&A, 245,Lutz, L31J. W. 1990, ApJS, 72,MacKenty, 231J. W., Simkin, S. M., Griffiths, R. E., Ulvestad, J. S., & Wilson,MacKenty,

A. S. 1994, ApJ, 435, 71D. F. 1997, in ASP Conf. Ser. 116, The Second Stromlo Sympo-Malin,

sium: The Nature of Elliptical Galaxies, ed. M. Arnaboldi, G. S. DaCosta, & P. Saha (San Francisco : ASP), 460

D. F., & Carter, D. 1983, ApJ, 274,Malin, 534J. C. 1995, ApJ, 438,Mihos, L75J. C., Dubinski, J., & Hernquist, L. 1998, ApJ, 494,Mihos, 183

J., Frenk, C. S., & White, S. D. M. 1994, MNRAS, 267,Navarro, L1R., Gallagher, J. S., & Hunter, D. A. 1994, ApJ, 433,OÏConnell, 65D. 1991, Rep. Prog. Phys., 54,Osterbrock, 579

B. M., et al. 1994, ApJ, 425,Peterson, 622A. C., Bershady, M. A., Forbes, D. A., Koo, D. C., Illingworth,Phillips,

G. D., Reitzel, D. B., Griffiths, R. E., & Windhorst, R. A. 1995, ApJ, 444,21

J. 1963, Z. Astrophys., 58,PÑeiderer, 12J. L. 1990, in Dynamics and Interactions of Galaxies, ed. R. WielenPrieur,

(Berlin : Springer), 72K. U., Griffiths, R. E., Casertano, S., Neuschaefer, L. W., &Ratnatunga,

Wycko†, E. W. 1994, AJ, 108, 2362F. 1987, in Nearly Normal Galaxies, ed. S. M. Faber (NewSchweizer,

York : Springer), 18F., & Seitzer, P. 1988, ApJ, 328,Schweizer, 88

1993, ApJ, 417,ÈÈÈ. L29E. J., Dowling, D. M., Currie, D. G., Faber, S. M., & Groth, E. J.Shaya,

1994, AJ, 107, 1675J., & Wyse, R. F. G. 1993, Phys. Rep., 231,Silk, 293

A., & Toomre, J. 1970, BAAS, 2,Toomre, 501972, ApJ, 178,ÈÈÈ. 623

B. C., & Schweizer, F. 1995, AJ, 109,Whitmore, 960B. C., Schweizer, F., Leitherer, C., Borne, K., & Robert, C. 1993,Whitmore,

AJ, 106, 1354A. S., & Ulvestad, J. S. 1982, ApJ, 260,Wilson, 56A. E. 1972, MNRAS, 157,Wright, 309