The connection between radio-loudness and central surface brightness profiles in optically-selected low-luminosity active galaxies
aa r X i v : . [ a s t r o - ph . GA ] A p r Mon. Not. R. Astron. Soc. , 1–16 (2011) Printed 15 November 2018 (MN L A TEX style file v2.2)
The connection between radio-loudness and central surfacebrightness profiles in optically-selected low-luminosityactive galaxies
A. J. Richings , P. Uttley and E. K¨ording School of Physics and Astronomy, University of Southampton, Southampton, S017 1BJ, UK Radboud Universiteit Nijmegen, Dept. Astronomy, IMAPP, The Netherlands
15 November 2018
ABSTRACT
Recent results indicate a correlation between nuclear radio-loudness of active galaxiesand their central stellar surface-brightness profiles, in that ‘core’ galaxies (with innerlogarithmic slope γ .
3) are significantly more radio loud than ‘power-law’ galaxies( γ > . T HST ).We use these fits to classify the galaxies as ‘Core’, ‘S´ersic’ or ‘Double-S´ersic’. Wecompare the properties of the Active Galactic Nuclei (AGNs) and their host galaxieswith this classification, and we recover the already established trend for Core galaxiesto be more luminous and contain a higher-mass supermassive black hole. Defining theradio-loudness of an AGN as the ratio of the nuclear radio luminosity to [O iii ] lineluminosity, which allows us to include most of the AGN in our sample and prevents abias against dim nuclei that are harder to extract from the brightness profiles, we findthat AGN hosted in Core galaxies are generally more radio-loud than those hostedin S´ersic galaxies, although there is a large overlap between the two subsamples. Thecorrelation between radio-loudness and brightness profile can partly be explained bya correlation between radio-loudness and black hole mass. Additionally, there is asignificant (99 per cent confidence) partial correlation between radio-loudness andthe Core/S´ersic classification of the host galaxy, which lends support to the previousresults based on the radio-selected sample, although it is possible that this partialcorrelation arises because AGN in core galaxies tend to have a lower accretion rate aswell as a higher central black hole mass.
Key words: galaxies: active – galaxies: jets – galaxies: nuclei – galaxies: photometry– radio continuum: galaxies
Before the
Hubble Space Telescope ( HST ) was launched, ob-servations of the central regions of galaxies were limited byatmospheric seeing, which prevented resolutions better than ∼ ′′ from being achieved. This made it difficult to deter-mine how the brightness profiles of these galaxies behaved atsmall radii. However, high resolution images taken with the HST (Crane et al. 1993; Ferrarese et al. 1994; Forbes et al.1995) demonstrated that the surface brightness profiles of el-liptical galaxies and disk galaxy bulges continue to increaseup to the resolution limit, confirming the results of ground-based observations (Kormendy 1985; Lauer 1985) that con-cluded that the profiles of most elliptical galaxies cannot c (cid:13) A. J. Richings, P. Uttley and E. K¨ording be described by models with flat, isothermal cores such asdescribed by King (1966).Early studies using
HST suggested that the centralbrightness profiles can be divided into two distinct classes —in “power-law” galaxies, the logarithmic slope of the bright-ness profile remained steep towards the resolution limit of
HST , while in “core” galaxies the slope flattened signifi-cantly. To describe these brightness profiles in the central re-gions, Lauer et al. (1995) introduced a form of broken powerlaw, the so-called “Nuker-law”, with the logarithmic slopein the inner region, γ , being flatter than in the outer region.These classes then correspond to a dichotomy in the val-ues of γ , with power-law galaxies having γ > . γ .
3. Faber et al. (1997) demonstratedthat there were no galaxies in the range 0 . < γ < .
5, al-though subsequent studies (e.g. Ravindranath et al. 2001)have found a small number of such “intermediate” galaxies.This dichotomy in brightness profiles has been linkedto global galaxy properties, for example core galaxies tendto be very luminous, slowly rotating galaxies with boxyisophotes, while power-law galaxies tend to be less luminous,rapidly rotating galaxies with disky isophotes (Jaffe et al.1994; Faber et al. 1997). For galaxies that contain activegalactic nuclei (AGNs) some of these properties of the hostgalaxy also correlate with properties of the AGN, for ex-ample the mass of the central supermassive black hole(SMBH) correlates strongly with the velocity dispersion ofthe host galaxy (Ferrarese & Merritt 2000; Gebhardt et al.2000), suggesting that the evolution of the host galaxy andAGN are interdependent. Investigating how the brightnessprofiles of the central regions of these host galaxies are re-lated to the properties of the AGN is important in under-standing how they influence one another and how they formand evolve together, for example Faber et al. (1997) proposea model whereby a merger between two galaxies can pro-duce a core galaxy as their black holes sink to the centre,ejecting stars in the central region and scouring out a core,while van der Marel (1999) suggests that the cusp slopes ofcore and power-law galaxies can be reproduced by adiabaticgrowth of the black hole in an isothermal core.More recent studies (e.g. Graham et al. 2003;Ferrarese et al. 2006a) have shown that the Nuker modeldoes not give an accurate parameterization of galaxysurface brightness profiles. It was only intended to modelthe central regions and so cannot describe the brightnessprofiles at larger radii, where the outer power law of theNuker model fails to properly fit the curvature of thebrightness profiles in most galaxies, and Graham et al.(2003) demonstrate that the parameters of the Nuker modelare sensitive to the radial extent of the profile to which itis fitted. However, the brightness profiles of “power-law”galaxies are well fitted by either a S´ersic law over theirentire radial extent, or by a S´ersic law modified with theaddition of a second S´ersic component, the latter meant torepresent an inner compact stellar nucleus (generally belowa few tenths of an arcsec). Meanwhile, “core” galaxiesshow a central deficit with respect to their outer S´ersicprofile (Trujillo et al. 2004; Ferrarese et al. 2006b). Thisprovides an alternative classification to the core/powerlaw dichotomy based on inner logarithmic slope, in whichgalaxies can be separated into those that are well fitted bya S´ersic law at all radii, those that require an additional inner S´ersic component to account for an inner compactstellar nucleus and those that show a central deficit fromtheir outer S´ersic profile. The profiles of galaxies that showa central deficit can be parameterized by a “Core-S´ersic”model (introduced by Graham et al. 2003) that combines aS´ersic profile at large radii with a power law in the centralregions (see e.g. Graham 2004; Cˆot´e et al. 2007).Recently, Capetti & Balmaverde published a seriesof three papers (Capetti & Balmaverde 2005 & 2006 andBalmaverde & Capetti 2006) in which they investigate thecore/power-law dichotomy, based on the Nuker law, in tworadio-selected samples of galaxies that are likely to host anAGN, requiring a minimum radio flux of ∼ HST images to analyse the brightness profiles of the Palomar sur-vey active galaxies, using the S´ersic/Core-S´ersic classifica- c (cid:13) , 1–16 adio-loudness and central surface brightness profiles tion scheme rather than the core/power-law classificationobtained from the Nuker model. We then use these resultsto investigate how the radio properties of these AGNs arerelated to their S´ersic/Core-S´ersic classification, which al-lows us to show whether Capetti & Balmaverde’s resultsare reproduced in a relatively complete sample of optically-selected AGN using this alternative classification scheme. To reliably detect the presence of a galaxy core it mustbe nearby so that a core could easily be resolved by the
HST . The Palomar spectroscopic survey of nearby galaxynuclei (Ho et al. 1995) uses a flux-limited sample of 488bright Northern galaxies, with B T . δ > ◦ , and naturally satisfies our requirement fornearby galaxies. Spectroscopic observations of these galaxiesshow that they include 197 AGNs. Nagar et al. (2005) con-ducted a high resolution radio survey of all of these AGNs,which will enable us to compare their radio properties totheir Core/S´ersic classification. For spiral galaxies it is thebrightness profile of the bulge component that determinesthis classification, however late-type spiral galaxies have rel-atively small bulges compared to the disk component, whichwill make it more difficult to classify them. Therefore we takeas our sample the 150 AGNs in Nagar et al. (2005) for whichthe host galaxy is of Hubble type T iii ] and H β narrow line lumi-nosities for the AGN were obtained using the combinationof unreddened line ratio data and H α fluxes in Table 2 ofHo et al. (1997). We dereddened these fluxes by applyingthe extinction curve of Cardelli, Clayton & Mathis (1989),assuming the Balmer decrement to be 3.1.We also estimated black hole masses for the AGN inour subset of the Palomar sample using the central ve-locity dispersions measured by Ho et al. (2009). We con-verted velocity dispersion to black hole mass using therelation: log( M BH / M ⊙ ) = 8 .
13 + 4 .
02 log( σ/ − )(Tremaine et al. 2002). For consistency this method wasused for all of the galaxies, including those for which directkinematic measurements were also available. HST images of the galaxies in our sample were obtainedfrom the public archive. Previous studies (e.g. Rest et al.2001) have demonstrated that many early-type galaxies con-tain dust structures that may affect the brightness profiles.It is especially important to deal with the effects of dust inthe AGN sample we study here, since dust structures are al-most ubiquitous in LLAGN (Gonz´alez Delgado et al. 2008) and are probably more common in LLAGN than in normalgalaxies (Martini et al. 2003). Therefore to minimise the ef-fects of dust we preferred to use images taken in the near-infrared using NICMOS, with the F160W filter. However, formany of the galaxies in our sample NICMOS images wereunavailable, so for these we used images taken in the opticalusing WFPC2 or ACS. Only broadband filters were used,and preference was given to filters at longer wavelengths —any shorter than F555W were rejected. Using different fil-ters might affect our results if the brightness profile dependson the wavelengths at which the galaxy is observed, howeverRavindranath et al. (2001) showed that the classification ofa galaxy as core or power-law, based on Nuker fits, is in-dependent of the wavelength that is used. We were able toobtain suitable images for 110 galaxies in our sample; thebasic properties of these galaxies are presented in table 1.As described in Section 2.1 the data presented in this ta-ble were obtained from Ho et al. (1997), although data wasonly available for 107 of the 110 Palomar survey galaxies forwhich we were able to obtain
HST images.The images from WFPC2 and ACS were calibrated us-ing the standard On The Fly Reprocessing (OTFR) sys-tem , , however the NICMOS images calibrated in this waycontained a “pedestal”, which is a time-variable bias that isdifferent in each quadrant. It is believed that this is causedby small temperature changes, which are different in eachquadrant because they use separate amplifiers. To correctfor the pedestal the standard CalnicA software that is usedin the NICMOS OTFR system was applied using IRAF tothe raw images, omitting all of the steps after and includingflat field correction. The task BIASEQ in IRAF was thenused to equalise the different biases in the 4 quadrants. Fi-nally, CalnicA was again applied using all of the steps afterand including flat field correction, and omitting the stepsthat had already been performed. The final images still con-tain a bias, but this is now constant over the whole imageand so can be accounted for by including a constant back-ground level when fitting the profile model to the image.These NICMOS images also contain several bad pixels —some of these are cold pixels, which appear dark becausethey have an abnormally low sensitivity; and some of theseare hot pixels, which appear bright because they have anabnormally high dark current. There are also regions, calledgrots, that appear dark because flecks of antireflective paintscraped off of the optical baffles have reduced their sensi-tivity. These pixels are identified in the data quality filesand are masked in the following analysis. The NIC2 cam-era also has a coronographic hole for observing targets thatare close to bright objects, but when it is not used it stillappears in the image as a bright patch, caused by emissionfrom warmer structures that are behind it. This hole movesover time, which results in 2 patches in the image due to us-ing reference files that were taken at a different time. Theseregions have been masked. (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording
Table 1.
Basic properties of the galaxies in our sample for which
HST images were obtained.Galaxy T D M B,total M B,bulge log (cid:18) M BH M ⊙ (cid:19) log L [ OIII ] log L Hβ log L ν,radio AGN ClassIC0356 2 18.1 -21.12 -19.89 7.70 39.27 38.94 < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < iii ] luminosity (erg s − ); log H β luminosity(erg s − ); log 15 GHz radio luminosity (erg s − ); AGN optical spectral classification (S:Seyfert, L:LINER, T:transition, numbersdenote type 1/2/intermediate, multiple classifications denote intermediate types, : and :: denote uncertain and highly uncertainclassifications. See Ho et al. (1997) for details); nuclear surface brightness profile classification, based on S´ersic, Core-S´ersic andDouble-S´ersic fits (see text). For brightness profile classifications with a question mark, the fits were uncertain.c (cid:13) , 1–16 adio-loudness and central surface brightness profiles Table 1 – continued Basic properties of the galaxies in our sample for which
HST images were obtained.Galaxy T D M B,total M B,bulge log (cid:18) M BH M ⊙ (cid:19) log L [ OIII ] log L Hβ log L ν,radio AGN ClassNGC4314 1 9.7 -18.76 -17.74 7.19 37.94 38.09 < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < We fitted the brightness profiles of all galaxies in our samplewith two models. The S´ersic (1968) model is parameterised as follows: I ( r ) = I e exp − b n "(cid:18) rr e (cid:19) /n − (1)This model is described by three parameters: the effectiveradius r e , the intensity at the effective radius I e , and theS´ersic index n , which parameterises the curvature of the c (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording profile. For 1 n
10, the quantity b n can be approximatedby b n ≈ . n − . I ( r ) = I ′ h I + (cid:16) r b r (cid:17) α i γ/α exp " − b n (cid:18) r α + r αb r αe (cid:19) /αn (2)This model uses six parameters, including the S´ersic index n and effective radius r e of the outer S´ersic profile. Theinner power law profile has a logarithmic slope γ , and thebreak radius r b is the radius separating the inner power lawand outer S´ersic profiles. The parameter α determines thesharpness of the transition between the two profiles, with ahigh value of α indicating a sharp transition. The intensity I ′ is related to the intensity at the break radius, I b , as follows: I ′ = I b − γ/α exp " b n (cid:18) /α r b r e (cid:19) /n (3)The quantity b n is calculated as in the S´ersic model.As noted in section 2.1 we excluded late type spiralgalaxies from our sample as their brightness profiles aredominated by the disc, however there are still several earlytype spiral galaxies in our sample. It is possible that thereis still a significant contribution from the disc component inthe central surface brightness profiles of these galaxies, sofor all of the spiral and lenticular galaxies in our sample wealso added an exponential disc component to the model.The profiles of several galaxies in our sample showedan unresolved nuclear source due to the AGN. For thesegalaxies a central point source was added to the S´ersic andCore-S´ersic models.Cˆot´e et al. (2007) demonstrate that low and intermedi-ate luminosity early-type galaxies often contain a resolvednuclear source, for example a compact spheroidal or flat-tened stellar component. They model the brightness profilesof these galaxies using a Double-S´ersic model: I ( r ) = I e, exp − b n "(cid:18) rr e, (cid:19) /n − + I e, exp − b n "(cid:18) rr e, (cid:19) /n − (4)The inner S´ersic model (parameterised by I e, , r e, and n I e, , r e, and n
2) describesthe host galaxy. The brightness profiles of several galaxiesin our sample showed a central excess above a single S´ersicprofile that was too extended to be described by a pointsource, so these galaxies were also fitted with this DoubleS´ersic model.Using the fits with the S´ersic and Core-S´ersic laws, weclassified a galaxy as ‘Core’ if it satisfied the following cri-teria (based on those used by Trujillo et al. 2004):(i) The Core-S´ersic fit must give a lower reduced χ thanthe S´ersic fit. We note that Trujillo et al. (2004) requiredthat the reduced χ of the Core-S´ersic fit must be at leasta factor of 2 less than that of the S´ersic fit, however wefound that very few galaxies in our sample satisfied such astrict criterion. This discrepancy is likely due to differencesin how the data points are weighted in the fits, as they use equal weights whereas we weight the data points using theirPoissonian errors, which are larger at smaller radii.(ii) To ensure that the core is well sampled by the datathe break radius r b must be greater than the radius of thesecond data point.(iii) The slope of the inner power law profile, γ , must beless than the logarithmic slope of the best-fit S´ersic modelwithin the break radius.Galaxies for which the Double-S´ersic model gave thelowest reduced χ were classified as ‘Double’ galaxies (i.e.contain a resolved nuclear source). All other galaxies wereclassified as ‘S´ersic’ galaxies. The 1D semi-major axis brightness profiles of each galaxywere extracted using the ELLIPSE task in IRAF, which fitselliptical isophotes to the galaxy images. Regions of dustthat were visible in the images were masked from these fits.The position angle and ellipticity were free to vary with eachisophote, enabling radial variations in these quantities, suchas isophotal twists, to be accounted for.
The S´ersic, Core-S´ersic and (where necessary) Double-S´ersicmodels were fitted to the 1D brightness profiles by minimis-ing the χ statistic using the Sherpa modelling and fittingsoftware package, which is a part of the Chandra
InteractiveAnalysis of Observations (CIAO ) software. The data pointswere weighted in these fits using their Poissonian errors.To accurately reproduce the brightness profile at smallradii it is necessary to account for the point spread function(psf) of the instrument used. For each image we created a2D model of the psf at the position of the galaxy’s centre onthe detector using the Tiny Tim software . It was necessaryto create a separate psf model for each image because thepsf depends on the position on the detector. The ELLIPSEtask in IRAF was used to extract the 1D profile of the psf,then the galaxy model was convolved with this psf in Sherpa before fitting to the observed brightness profile.In the Core-S´ersic model the parameters α , n and γ tend to be degenerate. Trujillo et al. (2004) recommend that α should be fixed at a large value to force an abrupt transi-tion between the outer S´ersic and inner power-law profiles,however we found that in several galaxies the profile waspoorly fitted by such a fixed- α model, but it was well fittedif α was free to vary. Therefore we fixed α at α = 50 toobtain an initial fit and then we allowed α to vary freely toobtain the final fit. We investigated how robust the parame-ters were for the free- α fits and we found that the parametersof the outer S´ersic component were uncertain in some of thegalaxies, particularly those with a low value of α . This mayalso be caused by the limited radial extent of the fittingregions, which were comparable to the effective radius r e in some galaxies. However, the uncertainties in γ were rela-tively small, and the break radii r b of the Core-S´ersic models http://cxc.harvard.edu/sherpa/ http://cxc.harvard.edu/ciao/ (cid:13) , 1–16 adio-loudness and central surface brightness profiles were well within the fitting region, so the classifications asCore or S´ersic were robust. We were able to obtain fits for all 110 galaxies in our sample,however we are only confident in 62 of these fits, which arereported in table 2. 44 of the remaining galaxies showed largeregions of dust that made the fits uncertain, even after wehad masked most of this dust. The profiles of the remaining4 galaxies could not be well described by a S´ersic, Core-S´ersic or Double-S´ersic model. We report these fits in table 3,however they will be discarded for most of the followinganalysis. We present examples of confident fits using theS´ersic, Core-S´ersic and Double-S´ersic models in fig. 1 andtwo uncertain fits in fig. 2, showing for each the galaxy imageand the radial profiles of the galaxy and the best fittingmodel.We included an exponential disc component in the mod-els for the spiral and lenticular galaxies in our sample, how-ever in many cases this component was negligible. We there-fore only include the disc component parameters in tables 2and 3 for the galaxies where it is significant.To enable us to compare our results with previous stud-ies such as Capetti & Balmaverde (2005), which used theNuker classification scheme, we also fitted the Nuker modelto all galaxies in our sample. If we assume that ‘intermedi-ate’ as well as ‘core’ galaxies from the Nuker scheme cor-respond to ‘Core-S´ersic’ galaxies and ‘Double-S´ersic’ and‘S´ersic’ galaxies both correspond to ‘power-law’ galaxies inthe Nuker scheme then the Nuker classifications of the galax-ies in our sample agree with those that we present in table 2for 82% of the galaxies for which we have a confident fit.This suggests that while the classification of many galaxiesin our sample remain unchanged when we use the Nukermodel, there are still some discrepancies between the twoclassification schemes.
Of the 62 galaxies for which we are confident in the bright-ness profile fits, we have 21 Core galaxies, 26 S´ersic galax-ies and 15 Double-S´ersic galaxies. We only have both lu-minosity data and a confident classification for 60 of the197 galaxies in the Palomar sample that are confirmed asAGN hosts, so we need to verify that we have not intro-duced any observational biases, for example if bright galax-ies are more likely to be imaged with
HST then there wouldbe a disproportionately large number of them in our finalsample. In fig. 3 we compare the distributions of the Hub-ble type, bulge B band magnitude, nuclear radio luminos-ity and [O iii ] line luminosity for the 60 galaxies in our fi-nal sample, the 107 galaxies for which we have a confidentor uncertain fit and luminosity data, and the 197 galax-ies hosting AGN in the Palomar sample, that were stud-ied in Nagar et al. (2005) (the Nagar sample). We used theASURV Rev 1.2 software (LaValley, Isobe & Feigelson 1992)to test the significance of correlations and differences be-tween distributions for censored data, which includes non-detections with limits. This software implements the meth-ods presented in Feigelson & Nelson (1985) for univariate problems and Isobe, Feigelson & Nelson (1986) for bivariateproblems.According to the generalised Wilcoxon tests the distri-butions of Hubble type in our final sample and the Nagarsample are inconsistent with being drawn from the same dis-tribution at the 99.99 per cent confidence level. We can see infig. 3, top left panel, that our final sample has a higher pro-portion of elliptical galaxies, which is to be expected becausewe discarded all late spirals, and a disproportionately highnumber of the fits for early spirals were uncertain becausethey are more likely to contain dust. Using the generalisedWilcoxon tests, we also find that the distributions of bulgemagnitude in our final sample and the Nagar sample are onlymarginally inconsistent with being drawn from the same dis-tribution at the 90 per cent confidence level. There are asmall number of galaxies in the Nagar sample with very dimbulges that we missed because we discarded late type spirals.Also, we were more likely to find
HST images for brightergalaxies. Of the galaxies for which we did obtain
HST data,a high proportion of those with intermediate bulge magni-tudes had uncertain Nuker fits, probably because these tendto correspond to later type galaxies which are more likely tobe dusty. According to the generalised Wilcoxon tests thedistributions of nuclear radio luminosities in our final sam-ple and the Nagar sample are also marginally inconsistentwith being drawn from the same distribution at the 90 percent confidence level. From fig. 3, bottom left panel, we cansee that there is a higher proportion of the small numberof AGNs with higher radio luminosities in our final sample.Finally, using the generalised Wilcoxon tests we find thatthe distributions of [O iii ] line luminosities in our final sam-ple and the Nagar sample are consistent with being drawnfrom the same distribution, although we can see in fig. 3,bottom right panel, that the galaxies with the lowest [O iii ]line luminosities do not appear in our final sample.
We find 15 galaxies that are best fit by a Double-S´ersicmodel. Previous studies (e.g. Cˆot´e et al. 2007) attributedthe inner S´ersic component in such models to a resolvedstellar nucleus, with a radius ≈ . r e , where r e is the ef-fective radius of the outer component. However, several ofthe Double-S´ersic galaxies in our sample have inner compo-nents with a much larger radius. This suggests that theymight not be due to resolved stellar nuclei, for examplesome of them may instead be inner discs or nuclear bars(Erwin & Sparke 2002). We studied the isophotes of thegalaxies and we did not find any with flattened isophotestowards the centre, which suggests that nuclear bars areunlikely. However, searching through the literature we findthat inner discs have been discovered in some of the galax-ies in our sample, for example NGC 3945 (Moiseev et al.2004) and NGC 7217 (Zasov & Sil’chenko 1997). We there-fore cannot be certain whether the Double-S´ersic galaxies inour sample contain resolved stellar nuclei, or if they insteadcontain other nuclear structures such as inner discs. Fig. 4, left panel, shows the distributions of central blackhole mass for Core, S´ersic and Double-S´ersic galaxies. The c (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording
Figure 1.
Galaxy images (top row) and radial brightness profiles (bottom row) for a confident S´ersic fit (NGC7742; left), Core fit (NGC3379; centre) and Double S´ersic fit (NGC7217; right).
Figure 2.
Galaxy images (top row) and radial brightness profiles (bottom row) for a fit that is uncertain due to dust (IC 356; left) anda fit that is uncertain because the galaxy has a complex profile (NGC 4111; right). c (cid:13) , 1–16 adio-loudness and central surface brightness profiles Table 2.
Best fit parameters of the Core, S´ersic and Double S´ersic galaxies from the confident fits.Core GalaxiesGalaxy Instrument Filter µ ′ a r b b γ n r be α m pt c NGC524 NICMOS F160W 13.32 0.18 0.11 2.80 14.88 134.14 -NGC2832 NICMOS F160W 13.51 0.64 0.11 2.90 9.03 4.69 -NGC2841 NICMOS F160W 10.05 0.17 0.35 5.03 77.48 102.27 -NGC3193 WFPC2 F702W 5.39 0.26 0.19 9.36 302.17 1.54 -NGC3379 NICMOS F160W 10.89 1.19 0.06 2.90 6.93 1.3 -NGC3607 NICMOS F160W 13.38 0.22 0.00 1.93 7.63 1.00 -NGC3608 WFPC2 F814W 9.00 0.31 0.29 6.65 71.59 4.01 -NGC3898 NICMOS F160W 9.44 0.22 0.27 4.79 17.77 4.60 -NGC4168 WFPC2 F702W 14.63 0.49 0.00 3.67 41.84 1.30 23.05NGC4261 NICMOS F160W 11.30 1.55 0.00 3.10 7.06 1.87 21.50NGC4278 NICMOS F160W 9.26 0.81 0.13 5.38 48.56 3.22 22.90NGC4350 NICMOS F160W 7.19 0.55 0.25 5.77 18.99 2.22 -NGC4374 NICMOS F160W 11.20 1.86 0.12 3.20 12.36 1.88 20.64NGC4472 NICMOS F160W 10.86 1.76 0.00 4.35 52.52 2.11 24.33NGC4486 WFPC2 F814W 10.89 6.68 0.27 5.32 117.19 3.38 16.90NGC4552 NICMOS F160W 11.67 0.36 0.14 2.71 6.98 7.73 -NGC4589 NICMOS F160W 10.57 0.14 0.00 4.64 21.19 2.12 -NGC5322 ACS F814W 5.73 1.23 0.41 8.14 93.09 2.62 -NGC5485 ACS F814W 16.14 0.32 0.13 2.74 19.79 113.48 -NGC5813 WFPC2 F814W 6.06 0.74 0.16 8.86 215.30 3.88 -NGC7626 NICMOS F160W 9.26 0.47 0.30 5.39 19.73 2.79 -S´ersic GalaxiesGalaxy Instrument Filter µ ae n r be µ e,disc r e,disc m cpt NGC474 NICMOS F160W 18.12 2.09 4.22 - - 20.37NGC488 NICMOS F160W 20.26 3.53 22.83 - - 22.05NGC2685 WFPC2 F814W 20.93 3.91 29.76 - - -NGC2985 NICMOS F160W 19.50 3.45 16.52 - - 20.77NGC3227 NICMOS F160W 17.80 4.07 3.63 - - 17.89NGC3245 NICMOS F160W 17.59 3.03 6.02 - - 19.52NGC3414 WFPC2 F814W 22.38 5.95 48.81 - - -NGC3516 NICMOS F160W 16.64 1.43 2.24 - - 17.17NGC3718 NICMOS F160W 15.95 1.50 1.04 18.05 4.78 19.21NGC3900 NICMOS F160W 22.22 6.55 38.11 - - -NGC3982 NICMOS F160W 19.60 2.47 4.84 - - 20.17NGC4150 NICMOS F160W 18.21 3.10 4.52 - - -NGC4151 NICMOS F160W 18.76 4.30 9.00 - - 15.13NGC4192 NICMOS F160W 16.76 1.93 5.27 20.76 369.78 15.51NGC4203 ACS F814W 21.49 4.30 22.92 - - 21.22NGC4314 NICMOS F160W 22.90 5.41 154.78 - - -NGC4378 WFPC2 F606W 19.75 2.78 5.93 21.51 24.46 -NGC4450 WFPC2 F814W 25.61 8.01 502.69 - - -NGC4459 NICMOS F160W 18.12 3.56 8.62 - - -NGC4636 NICMOS F160W 17.87 1.20 7.21 - - -NGC4698 WFPC2 F814W 16.65 2.03 1.64 18.81 11.65 20.60NGC5273 NICMOS F160W 17.85 1.66 2.08 - - 19.90NGC5838 NICMOS F160W 20.50 5.21 37.30 - - -NGC5982 NICMOS F160W 17.86 2.03 5.54 - - -NGC7742 NICMOS F160W 19.94 4.64 9.55 - - -NGC7743 NICMOS F160W 20.17 6.34 9.66 - - - a Corrected for extinction using values from Ho et al. (1997) and converted to the Johnson V band, assuming that thegalaxy has the spectrum of a K2 giant. b Measured in arcsec. c Point source magnitude, corrected for extinction using values from Ho et al. (1997) and converted to the Johnson V band,assuming a spectral index of 1.c (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording
Table 2 – continued Best fit parameters of the Core, S´ersic and Double S´ersic galaxies from the confident fits.Double S´ersic GalaxiesGalaxy Instrument Filter µ ae n r be µ ae n r be µ e,disc r e,disc m cpt NGC404 NICMOS F160W 19.74 15.72 2.55 24.97 6.35 712.26 - - -NGC3169 NICMOS F160W 14.90 0.81 0.38 17.24 1.60 6.72 - - -NGC3945 WFPC2 F814W 17.12 1.57 1.10 18.67 1.04 9.33 - - 20.60NGC3998 ACS F814W 17.11 3.04 0.35 19.66 2.90 9.75 - - -NGC4125 NICMOS F160W 16.55 1.04 0.84 18.46 2.32 17.56 - - -NGC4143 NICMOS F160W 18.24 0.48 3.25 17.69 2.51 5.59 - - 21.12NGC4435 NICMOS F160W 16.12 0.59 0.39 16.94 1.64 5.52 - - 19.56NGC4550 NICMOS F160W 16.96 0.16 1.92 19.12 3.56 5.10 - - -NGC4725 NICMOS F160W 15.20 1.06 0.13 18.70 2.83 14.10 - - -NGC4736 NICMOS F160W 16.14 0.56 8.25 17.98 3.54 19.42 - - -NGC5377 NICMOS F160W 15.06 0.95 0.26 16.91 0.63 2.81 20.41 440.82 -NGC5678 NICMOS F160W 15.68 0.16 0.02 23.47 6.07 133.73 22.11 367.00 -NGC5746 NICMOS F160W 17.44 0.16 1.41 17.56 2.09 6.74 19.06 623.32 20.61NGC6703 WFPC2 F814W 20.57 0.82 10.87 21.95 6.22 25.78 - - -NGC7217 NICMOS F160W 16.52 0.93 0.79 19.60 2.64 20.05 20.51 167.78 - −5 0 5 10 N Hubble Type −15 −20 N M B, bulge
34 36 38 40 N Log L ν , rad
36 38 40 42 N Log L [OIII]
Figure 3.
Histograms showing the distributions of Hubble type (top left), bulge B band magnitude (top right), nuclear radio luminosity(bottom left) and [O iii ] line luminosity (bottom right) in our final sample (dotted line), the galaxies for which we have a confident oruncertain fit and luminosity data (dashed line) and the Nagar sample (solid line). The arrows denote numbers of galaxies in the histogrambins which are limits only. c (cid:13) , 1–16 adio-loudness and central surface brightness profiles Table 3.
Best fit parameters of the Core, S´ersic and Double S´ersic galaxies from the uncertain fits.Core GalaxiesGalaxy Instrument Filter µ ′ a r b b γ n r be α m pt c UncertaintyNGC315 WFPC2 F814W 12.51 0.89 0.00 5.06 100.29 3.27 21.27 DustyNGC5846 WFPC2 F702W 11.23 1.24 0.00 6.54 419.02 4.40 - DustyS´ersic GalaxiesGalaxy Instrument Filter µ ae n r be µ e,disc r e,disc m cpt UncertaintyIC356 WFPC2 F814W 25.83 9.16 1491.05 21.32 2904.34 - DustyNGC660 ACS F814W 19.25 0.23 8.98 - - - DustyNGC2273 NICMOS F160W 16.17 0.74 1.82 19.55 15.45 18.76 DustyNGC2681 NICMOS F160W 23.12 12.35 103.2 - - - ComplexNGC2683 NICMOS F160W 22.79 6.17 184.16 19.43 369.32 - DustyNGC3190 ACS F814W 18.85 1.97 9.54 - - - DustyNGC3368 NICMOS F160W 17.89 2.42 14.26 - - 16.81 DustyNGC3489 WFPC2 F814W 19.51 4.22 15.24 - - - DustyNGC3507 WFPC2 F606W 21.01 4.8 7.95 21.75 43.06 - DustyNGC3623 WFPC2 F814W 21.16 5.81 63.13 19.68 579.52 - DustyNGC3626 NICMOS F160W 16.67 2.32 3.14 - - 17.87 DustyNGC3627 NICMOS F160W 17.79 2.78 10.46 19.57 344.24 - DustyNGC3628 NICMOS F160W 25.21 5.6 5850.16 - - - DustyNGC4013 WFPC2 F814W 19.37 0.18 7.9 20.55 228.1 - DustyNGC4281 WFPC2 F606W 25.97 8.71 509.97 - - - DustyNGC4388 NICMOS F160W 24.09 7.79 121.02 - - - DustyNGC4429 WFPC2 F606W 22.78 4.52 134.39 - - 17.43 DustyNGC4438 WFPC2 F814W 18.58 3.42 12.99 - - - DustyNGC4477 WFPC2 F606W 21.3 3.58 35.93 - - - DustyNGC4494 WFPC2 F814W 33.22 17.85 46101.91 - - - DustyNGC4565 NICMOS F160W 16 1.44 3.9 18.3 61.7 18.66 DustyNGC4579 WFPC2 F791W 19.24 3.32 9.77 20.22 48.67 19.39 DustyNGC4772 WFPC2 F606W 23.7 6.27 68.5 21.68 36.77 - DustyNGC4826 NICMOS F160W 16.68 2.21 6.87 19.5 80.47 - DustyNGC4866 ACS F814W 33.20 13.65 33227.88 - - - DustyNGC5448 NICMOS F160W 17.36 1.49 2.56 20.71 130.06 - DustyNGC5566 WFPC2 F606W 17.16 0.86 5.23 - - - DustyNGC5866 NICMOS F160W 18.26 1.77 23.21 - - - DustyNGC5985 NICMOS F160W 19.56 2.03 4.91 - - 22.46 ComplexNGC6340 ACS F814W 20.28 3.44 5.07 22.14 27.77 - DustyNGC7177 NICMOS F160W 18.16 1.89 5.97 - - - DustyNGC7331 NICMOS F160W 22.96 8.09 344.32 18 21.64 - DustyNGC7814 NICMOS F160W 17.82 1.91 13.16 - - - DustyDouble S´ersic GalaxiesGalaxy Instrument Filter µ ae n r be µ ae n r be µ e,disc r e,disc m cpt UncertaintyNGC2768 WFPC2 F814W 17.14 1.00 0.44 24.88 5.98 547.44 - - - DustyNGC2787 WFPC2 F814W 16.74 1.54 0.58 19.78 2.20 20.34 - - - DustyNGC2859 ACS F814W 20.00 0.16 0.31 20.20 2.32 9.46 - - - ComplexNGC3675 NICMOS F160W 19.14 0.17 8.96 21.11 5.51 42.71 21.11 1141.66 - DustyNGC3705 WFPC2 F814W 14.32 0.4 0.22 20.58 4.01 28.64 20.82 215.19 - DustyNGC4111 NICMOS F160W 17.58 0.16 1.41 18.75 3.69 19.13 - - - ComplexNGC4220 ACS F814W 20.60 2.25 5.57 20.29 0.14 46.12 - - - DustyNGC4293 NICMOS F160W 16.4 0.43 0.82 29.77 8.65 25512.11 22.23 282.46 17.99 DustyNGC4394 ACS F814W 19.83 0.16 1.02 21.45 3.81 9.06 22.03 24.62 - DustyNGC4501 WFPC2 F606W 15.97 0.73 0.64 19.06 1.11 12.09 22.08 202.65 19.89 DustyNGC4569 NICMOS F160W 15.15 3.72 0.72 15.86 0.59 2.36 18.95 22.24 18 DustyNGC4596 WFPC2 F606W 15.54 0.89 0.07 22.45 4.22 66.19 - - - DustyNGC4750 NICMOS F160W 14.46 0.16 0.07 17.6 1.36 2.39 20.36 157.09 - Dusty a Corrected for extinction using values from Ho et al. (1997) and converted to the Johnson V band, assuming that thegalaxy has the spectrum of a K2 giant. b Measured in arcsec. c Point source magnitude, corrected for extinction using values from Ho et al. (1997) and converted to the Johnson V band,assuming a spectral index of 1.c (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording N N N Log M BH /M sol N N −16 −18 −20 −22 N M B, bulge
Figure 4.
Histograms showing the distributions of central black hole mass (left) and bulge B band magnitude (right) for Core galaxies(top), S´ersic galaxies (middle) and Double-S´ersic galaxies (bottom). black hole masses of the Core galaxies are typically higherthan those of the S´ersic or Double-S´ersic galaxies, for exam-ple the median of log M BH /M ⊙ is 8.39 for the Core galaxies,compared to 7.57 and 7.84 for the S´ersic and Double-S´ersicgalaxies respectively. We also find that the most massiveblack holes are almost exclusively found in Core galaxies —with the exception of one Double-S´ersic galaxy, NGC 3998,all SMBHs with log M BH /M ⊙ & . M BH /M ⊙ . . M B,bulge . − .
6, andfor all S´ersic galaxies M B,bulge & − .
4. Most Double-S´ersicgalaxies are within the same range as the S´ersic galaxies,although there are two examples that are more luminousand one that is much less luminous. Capetti & Balmaverde(2005) found a similar trend in K band magnitudes, althoughthey also found core galaxies covering a much broader rangeof magnitudes. It is also worth noting that, with the ex-ception of one Double-S´ersic galaxy, NGC 4125, all galaxieswith bulge B magnitudes below -21 are Core galaxies.We note that the most massive black holes being foundin Core galaxies is to be expected as they are at the moreluminous end of the luminosity function (e.g. Ferrarese et al.2006b, and also confirmed for our sample by Figure 4, rightpanel), so they also tend to have larger velocity dispersions(Faber & Jackson 1976) and hence black hole masses, whichwe derived from the velocity dispersions.
The radio-loudness of an AGN is typically measured as theratio of the radio to optical luminosity of the nucleus, how-ever we were only able to extract a point source in 23 of thegalaxies for which we are confident in the Core/S´ersic fits,so this would limit the number of nuclei for which we canmeasure the radio-loudness. Furthermore this could bias any N N N −4 −3 −2 −1 0 Log L ν , rad /L [OIII] Figure 5.
Histograms showing the distribution of L ν,rad /L [ O III] for Core galaxies (top), S´ersic galaxies (middle) and Double-S´ersicgalaxies (bottom). The arrows denote numbers of galaxies in thehistogram bins which are limits only. correlations that we observe because we would be preferen-tially selecting bright, type 1 nuclei that are unobscured bynuclear dust. We therefore use the narrow [O iii ] line lumi-nosity instead of the total optical luminosity of the nucleus,because we have this data for most of the galaxies in oursample, and since [O iii ] line emission originates from thenarrow line region, there should not be a bias between type1 and type 2 nuclei.In Fig. 5 we show the distribution of the radio-loudness L ν,radio /L [ O III] for Core, S´ersic and Double-S´ersic galaxies.According to the generalised Wilcoxon tests the distribu-tions of radio-loudness in Core and S´ersic galaxies are in-consistent with being drawn from the same distribution atthe 99.9 per cent confidence level, while the distributions ofradio-loudness in S´ersic and Double-S´ersic galaxies are con-sistent with being drawn from the same distribution. Thisconfirms that the radio-loudness of an AGN is related towhether its host galaxy is classified as a Core galaxy. Wecan see in fig. 5 that the most radio-loud AGNs are hosted c (cid:13) , 1–16 adio-loudness and central surface brightness profiles in Core galaxies, however AGNs with log( L ν,radio /L [ O III] ) . − . iii ] line luminosities of power-law galaxies are aslow as those of core galaxies, although they do extend tohigher [O iii ] line luminosities than core galaxies. This sug-gests that the discrepancy between the correlation that weobserve and that observed by Capetti & Balmaverde (2006)is due to the different definitions used for the radio-loudness.To test this possibility we looked at the galaxies for which wewere able to extract a point source in the profile fitting andwe used this point source to determine the ratio of nuclearradio luminosity to nuclear optical continuum luminosity, L ν,radio /L o . For these galaxies we then compared the dis-tributions of both L ν,radio /L o and L ν,radio /L [ O III] for Core,S´ersic and Double-S´ersic galaxies; these results are presentedin Fig. 6. According to the generalised Wilcoxon tests thedistributions of L ν,radio /L o in Core and S´ersic galaxies areinconsistent with being drawn from the same distribution atthe 99.99 per cent confidence level, while the distributionsof L ν,radio /L [ O III] are inconsistent at the 99.9 per cent level.Furthermore, we can see from Fig. 6 that there is no overlapbetween the Core and S´ersic subsamples when we use the nu-clear optical continuum luminosity. This demonstrates thatthe split between Core and S´ersic galaxies is clearer whenwe use L ν,radio /L o .We also looked at using the narrow H β line luminos-ity instead of [O iii ]. Fig. 7 shows that using this definitionfor the radio-loudness produces the same difference betweenCore and S´ersic galaxy radio-loudness distributions that wasobserved when [O iii ] was used. We saw in § N N −4 −3 −2 −1 0 N Log L ν , rad /L H β Figure 7.
Histograms showing the distribution of L ν,rad /L Hβ forCore (top), S´ersic (middle) and Double-S´ersic (bottom) galaxies.The arrows denote numbers of galaxies in the histogram binswhich are limits only. and Eddington ratio for a sample of 48 LLAGN taken fromthe Palomar spectroscopic survey. In fig. 8, left panel, wecan see that there is still considerable scatter between thesevariables, which suggests that other factors may be involved.We also used Spearman’s rho to test the correlation be-tween radio-loudness and bulge B band magnitude, and wefound that there is a correlation at the 99 per cent con-fidence level, with luminous galaxies generally being moreradio loud. However, fig. 8, right panel shows that there isalso scatter between these two variables, which again sug-gests that there may be other factors involved.It is possible that the scatter in the correlations of radio-loudness with bulge magnitude and black hole mass can beexplained by the Core/S´ersic/Double-S´ersic classification.To test correlations with this classification we parameterisedthe galaxies using the central deficit or excess in the surfacebrightness profile compared to the inward extrapolation ofthe outer S´ersic component, using the ∆ . parameter in-troduced by Cˆot´e et al. (2007). This parameter is defined as∆ . = log L G / L S , where L G is the total luminosity of thebest fit galaxy model within a radius of 0 . r e and L S isthe total luminosity of the outer S´ersic component in thisregion. Core galaxies have ∆ . <
0, S´ersic galaxies have∆ . = 0 and Double-S´ersic galaxies have ∆ . >
0. Wethen tested the partial correlations for both black hole massand bulge magnitude with radio loudness and ∆ . , usingthe cens tau code , which uses the methods presented inAkritas & Siebert (1996) to search for partial correlations incensored data. These partial correlations test whether thereare residual correlations between radio-loudness and ∆ . after the overall correlations have been accounted for.Zero partial correlation between the bulge magnitude,radio-loudness and ∆ . is rejected at the 99 per cent con-fidence level, and zero partial correlation between the blackhole mass, radio-loudness and ∆ . is rejected at the 99.99per cent confidence level. These results suggest that there isa significant residual correlation between the radio-loudness http://astrostatistics.psu.edu/statcodes/cens tauc (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording N N −5 −4 −3 −2 −1 0 N Log L ν , rad /L o N N −4 −3 −2 −1 0 N Log L ν , rad /L [OIII] Figure 6.
Histograms comparing the distributions of L ν,radio /L o (left panel) and L ν,radio /L [ O III] (right panel) for the Core (top),S´ersic (middle) and Double-S´ersic (bottom) galaxies for which we were able to extract a point source. The arrows denote numbers ofgalaxies in the histogram bins which are limits only. − − − − L og L ν , r a d / L [ O III]
Log M BH /M sol −18 −20 − − − − L og L ν , r a d / L [ O III] M B, bulge
Figure 8.
Radio-loudness L ν,rad /L [ O III] plotted against black hole mass (left panel) and against bulge B band magnitude (right panel).Core galaxies are represented by filled squares, S´ersic galaxies by empty circles and Double-S´ersic galaxies by empty triangles. of an AGN and the classification of its host galaxy as Core,S´ersic or Double-S´ersic, after the effects of bulge magni-tude and black hole mass have been accounted for. How-ever, Doi et al. (2006) demonstrated that the radio-loudnesscorrelates with both the black hole mass and the accre-tion rate in a sample of 48 LLAGN, so we also need toconsider whether ∆ . correlates with the accretion rate.We used the ratio of [O iii ] line luminosity to Eddingtonluminosity, log( L [ OIII ] /L Edd ), as a proxy for the accre-tion rate and tested the partial correlation of ∆ . withlog( L [ OIII ] /L Edd ) and black hole mass, following the samemethod as above. We found that ∆ . is partially corre-lated with log( L [ OIII ] /L Edd ) at the 99.99 per cent confi-dence level. Therefore it is possible that the partial correla-tion between radio-loudness and ∆ . arises because Coregalaxies tend to have a higher central black hole mass and alower accretion rate. This possibility needs further inves-tigation. The correlation between accretion rate and theclassification of the host galaxy as Core, S´ersic or Double-S´ersic is illustrated in fig. 9, where we plot log L ν,rad /L [ OIII ] against log L [ OIII ] /L Edd for the Core, S´ersic and Double-S´ersic galaxies.
In this paper we have studied the brightness profiles of el-liptical and early type spiral galaxies in the Palomar spec-troscopic survey that are confirmed as AGN hosts. We fit-ted S´ersic, Core-S´ersic and, where necessary, Double-S´ersicmodels, plus a point source where needed, to the 1D semi-major axis brightness profiles extracted from high resolution
HST images of these galaxies. By comparing these fits wewere able to classify the galaxies as Core, S´ersic or Double-S´ersic galaxies, and then we investigated how other proper-ties of the host galaxy and the AGN relate to this classifi-cation.We found that Core galaxies were generally more lu-minous and hosted higher mass black holes than S´ersicand Double-S´ersic galaxies, although there was considerable c (cid:13) , 1–16 adio-loudness and central surface brightness profiles −9 −8 −7 −6 −5 −4 − − − − L og L ν , r a d / L [ O III]
Log L [OIII] /L Edd
Figure 9.
Radio-loudness L ν,rad /L [ OIII ] plotted against the[O iii ] line luminosity as a fraction of the Eddington luminosity, L [ OIII ] /L Edd , for Core galaxies (filled squares), S´ersic galaxies(empty circles) and Double-S´ersic galaxies (empty triangles). overlap between these subsamples. These results agree withprevious studies (e.g. Capetti & Balmaverde 2006, who usedthe Nuker classification scheme).To measure the radio-loudness of the AGNs we took theratio of nuclear radio luminosity to [O iii ] line luminosity, be-cause we could only estimate the optical continuum luminos-ity of the AGN for approximately one third of the galaxies,for which we could extract a point source in the profile fit-ting. Furthermore, using the [O iii ] line luminosity preventsbias against faint nuclei that are more difficult to extract.Using this definition of the radio-loudness we found thatCore galaxies were generally more radio-loud than S´ersicand Double-S´ersic galaxies, in agreement with the resultsof Capetti & Balmaverde (2006) for a radio-selected sam-ple of AGN based on the Nuker scheme. However, we foundsignificantly more overlap between these subsamples com-pared to Capetti & Balmaverde (2006). This difference ismost likely because we use a different definition of the radioloudness, as we find that there is much less overlap when weuse the nuclear optical continuum luminosity. Therefore weconclude that the radio-loudness/brightness-profile connec-tion uncovered for radio-selected AGN also applies to ouroptically-selected sample.We also looked at how the radio-loudness, defined us-ing the [O iii ] line luminosity, correlated with the black holemass and bulge B band magnitude of the host galaxies. Wefound that, while the radio -loudness does show correlationswith both black hole mass and bulge magnitude, which couldexplain at least some of the correlation between brightness-profile and radio-loudness, we still found a significant partialcorrelation with the classification of the host galaxy bright-ness profile as Core or S´ersic. However, the host galaxy clas-sification is also correlated with the accretion rate, so it ispossible that the observed connection between brightness-profile and radio-loudness arises because Core galaxies tendto have a higher black hole mass and a lower accretion rate.This possibility requires further investigation.
ACKNOWLEDGMENTS
AJR acknowledges the support of a Nuffield FoundationUndergraduate Research Bursary. PU is supported by anSTFC Advanced Fellowship, and funding from the EuropeanCommunity’s Seventh Framework Programme (FP7/2007-2013) under grant agreement number ITN 215212 BlackHole Universe. This work is based on observations madewith the NASA/ESA Hubble Space Telescope, and ob-tained from the Hubble Legacy Archive, which is a col-laboration between the Space Telescope Science Institute(STScI/NASA), the Space Telescope European Coordinat-ing Facility (ST-ECF/ESA) and the Canadian AstronomyData Centre (CADC/NRC/CSA). We also thank the ref-eree, Laura Ferrarese, for useful comments that have im-proved this paper.
REFERENCES
Akritas, M. G., & Siebert, J. 1996, MNRAS, 278, 919Balmaverde, B., & Capetti, A., A&A, 447, 97Best, P. N., Kauffmann, G., Heckman, T. M., Brinchmann,J., Charlot, S., Ivezic, Z., White, S. D. M., 2005, MNRAS,362, 25Caon, N., Capaccioli, M., & D’Onofrio, M. 1993, MNRAS,265, 1013Capetti, A., & Balmaverde, B., 2005, A&A, 440, 73Capetti, A., & Balmaverde, B. 2006, A&A, 453, 27Capetti, A., & Balmaverde, B. 2007, A&A, 469, 75Cardelli, J. A., Clayton, G. C., Mathis, J. S., 1989, ApJ,345, 245Cˆot´e et al., 2007, ApJ, 671, 1456Crane, P., et al. 1993, AJ, 106, 1371Doi, A., Kameno, S., & Inoue, M., 2006, J. Phys.: Conf.Ser. 54 335Erwin, P., & Sparke, L. S. 2002, AJ, 124, 65Faber, S. M., & Jackson, R. E. 1976, ApJ, 204, 668Faber, S. M., et al. 1997, AJ, 114, 1771Feigelson, E. D., & Nelson, P. I. 1985, ApJ, 293, 192Ferrarese, L., van den Bosch, F. C., Ford, H. C., Jaffe, W.,& O’Connell, R. W. 1994, AJ, 108, 1598Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9Ferrarese, L., et al. 2006a, in Proc. IAU Symp. 238, BlackHoles: from Stars to Galaxies — Across the Range ofMasses, ed. V. Karad & G. Matt (Cambridge: CambridgeUniv. Press), in press (astro-ph/0612139)Ferrarese, L., et al. 2006b, ApJS, 164, 334Forbes, D. A., Franx, M., & Illingworth, G. D. 1995, AJ,109, 1988Gebhardt, K. et al. 2000, ApJ, 539, L13Gonz´alez Delgado, R. M., P´erez, E., Cid Fernandes, R., &Schmitt, H., 2008, AJ, 135, 747Graham, A. W., Erwin, P., Trujillo, I., & Asensio Ramos,A. 2003, AJ, 125, 2951Graham, A. W. 2004, ApJ, 613, L33Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1995,ApJS, 98, 477Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997,ApJS, 112, 315Ho, L. C., Greene, J. E., Filippenko, A. V., & Sargent, W.L. W. 2009, ApJS, 183, 1 c (cid:13) , 1–16 A. J. Richings, P. Uttley and E. K¨ording
Isobe, T., Feigelson, E. D., & Nelson, P.I. 1986, ApJ, 306,490Jaffe, W., Ford, H. C., O’Connell, R. W., van den Bosch,F. C., & Ferrarese, L. 1994, AJ, 108, 1567King, I. R. 1966, AJ, 71, 64Kormendy, J. 1985, ApJ, 292, L9Laor, A., 2000, ApJ, 543, L111Lauer, T. R. 1985, ApJ, 292, 104Lauer, T. R., et al. 1995, AJ, 110, 2622LaValley, M., Isobe, T., & Feigelson, E. D. 1992, BAAS,24, 839Martini, P., Regan, M. W., Mulchaey, J. S., & Pogge, R.W., 2003, ApJ, 589, 774McLure, R. J., & Jarvis, M. J., 2004, MNRAS, 353, L45Moiseev, A. V., Vald´ev, J. R., & Chavushyan, V. H. 2004,A&A, 421, 433Nagar, N. M., Falcke, H., & Wilson, A. S. 2005, A&A, 435,521Paturel, G., Teerikorpi, P., Theureau, G., Fouqu´e, P.,Musella, I., Terry, J. N., 2002, A&A, 389, 19Ravindranath, S., Ho, L. C., Peng, C. Y., Filippenko, A.V., & Sargent, W. L. W. 2001, AJ, 122, 653Rest, A., van den Bosch, F. C., Jaffe, W., Tran, H., Tsve-tanov, Z., Ford, H. C., Davies, J., & Schafer, J. 2001, AJ,121, 2431S´ersic, J. L. 1968, Atlas de Galaxias Australes (C´ordoba:Obs. Astron., Univ. Nac. C´ordova)Tonry, J. L., et al.., 2001, ApJ, 546, 681Tremaine, S., et al., 2002, ApJ, 574, 740Trujillo, I., Erwin, P., Asensio Ramos, A., & Graham, A.W. AJ, 127, 1917van der Marel, R. P. 1999, Galaxy Dynamics, ASP Con-ference Series, 182, ed. Merritt, D., Sellwood, J. A., &Valluri, M.Zasov, A. V., & Sil’chenko, O. K. 1997, ARep, 41, 734 c (cid:13)000