Compact Stellar Systems in the Fornax Cluster: Super-massive Star Clusters or Extremely Compact Dwarf Galaxies?
M. J. Drinkwater, J. B. Jones, M. D. Gregg, S. Phillipps, PASA, 17 (3), 227.

Next Section: Discussion
Title/Abstract Page: Compact Stellar Systems in
Previous Section: Discovery Observations: The Fornax
Contents Page: Volume 17, Number 3

Properties of the compact objects

Sizes

These object images are unresolved and classified ``stellar'' in our UKST plate data, although imaging with the CTIO Curtis Schmidt shows that the brightest two objects have marginal signs of extended structure. In Fig. 1 we present R-band (Tech Pan emulsion + OG 590 filter) photographic images of these compact objects from the UKST. These were taken in seeing of about 2.2 arcseconds FWHM and the third object (FCSS J033854.1-353333) is resolved with a 3.2 arcsecond FWHM. Applying a very simple deconvolution of the seeing this corresponds to a scale size (HWHM) of about 80 pc. This is much larger than any known globular cluster, so this object, at least, is not a globular cluster. The other objects are all unresolved, so must have scale sizes smaller than this.

**Figure 1:** R-band photographic images of the new compact objects. The images are all from a single UKST exposure on Tech-Pan emulsion, digitised by SuperCOSMOS (Miller et al. 1992). Each image is 2.5 arcminutes across with North at the top and East to the left.
$\begin{figure} \hfil \psfig{file=fig_im.ps,angle=0,width=18cm}\end{figure}$

Luminosity and Colours

These new objects have absolute magnitudes -13<M_B<-11, based on a distance modulus of 30.9 mag to the Fornax Cluster (Bureau et al. 1996). These values are at the lower limit of dwarf galaxy luminosities (Mateo 1998), but are much more luminous than any known Galactic globular clusters (Harris, 1996) and the most luminous of the NGC1399 globulars (Forbes et al. 1998) which have $M_B\approx-10$ . The luminosities of the compact objects are compared to several other populations of dwarf galaxy and star cluster in Fig. 2. We note that the magnitude limit of the 2dF data corresponds to an absolute magnitude of $M_B\approx-11$ here. In order of decreasing luminosity the first comparison is with the dwarf ellipticals listed in the FCC as members of the Fornax Cluster. The possible M32-type galaxies in the FCC are not included as none of them have yet been shown to be cluster members (Drinkwater, Gregg & Holman 1997). The Figure shows that the Fornax dEs have considerable overlap in luminosity with the compact objects, but morphologically they are very different, being fully resolved low surface brightness galaxies. Recently, several new compact dwarf galaxies have been discovered in the Fornax Cluster (Drinkwater & Gregg 1998) but these are all brighter than M_B=-14 and do not match any of the objects we discuss here. Binggeli & Cameron (1991) measured the luminosity function of the nuclei of nucleated dwarf elliptical galaxies in the Virgo Cluster. The Figure shows that this also overlaps the distribution of the new compact objects. In this case the morphology is the same, so the compact objects could originate from the dwarf nuclei. The Figure also shows the luminosity functions of both the NGC 1399 globular clusters (Bridges, Hanes & Harris, 1991) and Galactic globular clusters (Harris 1996). These are quite similar and have no overlap with the compact objects.

For completeness we note that the luminosities of the compact objects have considerable overlap with the luminosities of dwarf galaxies in the Local Group (Mateo 1998), but even the most compact of the Local Group dwarfs, Leo I (M_B=-11.1) would be resolved (

$r_e\approx 3''$ ) in our images at the distance of Fornax. The only population they match in both luminosity and morphology is the bright end of the nuclei of nucleated dwarf ellipticals.

**Figure 2:** Distribution of absolute magnitude of the compact objects (filled histogram) compared to dEs in the Fornax Cluster (Ferguson 1989; solid histogram), the nuclei of dE,Ns in the Virgo Cluster (Binggeli & Cameron 1991; short dashes), a model fit to the globular clusters around NGC 1399 (Bridges, Hanes & Harris, 1991; long dashes) and Galactic globular clusters (Harris 1996; dotted). Note: the magnitude limit of our survey that found the compact objects corresponds to M_B=-11.
$\begin{figure} \hfil \psfig{file=fig_lf.ps,angle=0,width=9cm}\end{figure}$

Spectral Properties

The 2dF discovery spectra of these compact objects are shown in Figure 3. They have spectra similar to those of early type dwarf galaxies in the sample (two are shown for comparison in the Figure) with no detectable emission lines. As part of the spectral identification process in the Fornax Spectroscopic Survey, we cross-correlate all spectra with a sample of stellar templates from the Jacoby et al. (1984) library. The spectra of the new compact objects were best fit by K-type stellar templates, consistent with an old (metal-rich) stellar population. The dE galaxies observed with the same system by contrast are best fit by younger F and early G-type templates. This gives some indication in favour of the compact objects being related to globular clusters, although we note that two of them were analysed by Hilker et al. (1999) in more detail without any conclusive results. We do not have the spectrum of a dE nucleus available for direct comparison, but since our 2dF spectra are taken through a 2 arcsec diameter fibre aperture, the spectrum of FCC 211, a nucleated dE, is dominated by the nucleus. This spectrum was fitted by a younger F-type stellar template, again suggestive of a younger population than the compact objects. We cannot draw any strong conclusions from these low-resolution, low signal-to-noise spectra.

**Figure 3:** 2dF discovery spectra of the five compact objects as well as two cluster dwarf galaxies for comparison. Note: the large scale ripple in the spectrum of the fifth object (FCSS J033952.5-350424) is an instrumental effect caused by deterioration in the optical fibre used.
$\begin{figure} \hfil \psfig{file=plot_spec.ps,angle=0,width=9cm}\end{figure}$

Radial Distribution

The main advantage of our survey over the previous studies of the NGC 1399 globular cluster system (e.g. Grillmair et al. 1994, Hilker et al. 1999) is that we have complete spectroscopic data over a much larger field, extending to a radius of 1 degree (projected distance of 270 kpc) from the cluster centre. This means that we can determine the spatial distribution of the new compact objects. In Figure 4 we plot the normalised, cumulative radial distribution of the new compact objects compared to that of foreground stars and cluster galaxies. This plot is the one used to calculate Kolmogorov-Smirnov statistics and allows us to compare the distributions of objects independent of their mean surface densities. It is clear from the Figure that the new compact objects are very concentrated towards the centre of the cluster, at radii between 5 and 30 arcminutes (20-130 kpc). Their distribution is more centrally concentrated than the King profile fitted to cluster members by Ferguson (1989) with a core radius of 0.7 degrees (190 kpc). The Kolmogorov-Smirnov (KS) test gives a probability of 0.01 that the compact objects have the same distributions as the FCC galaxies: they are clearly not formed (or acreted) the same way as average cluster galaxies. To test the hypothesis that the compact objects are formed from nucleated dwarfs, we also plot the distribution of all the FCC nucleated dwarfs, as these are more clustered than other dwarfs (Ferguson & Binggeli 1994). However in the central region of interest here the nucleated dwarf profile lies very close to the King profile of all the FCC galaxies, so this does not provide any evidence for a direct link with the new compact objects.

West et al. (1995) suggest that a smaller core radius should be used for intra-cluster globular clusters (GCs). This profile, also shown in the Figure, is more consistent with the distribution of the new objects: the KS probability of the compact objects being drawn from this distribution is 0.39.

**Figure 4:** Cumulative radial distribution of the new compact objects compared to the predicted distribution for intra-cluster globular clusters (West et al 1995) and the profile fit to the distribution of all FCC. Also shown is the distribution of all nucleated dwarfs in the FCC and all the unresolved objects (stars) observed in our 2dF survey.
$\begin{figure} \hfil \psfig{file=fig_radial.ps,angle=0,width=9cm}\end{figure}$

We also note that the radial distribution of the compact objects is much more extended than the NGC 1399 globular cluster system as discussed by Grillmair et al. and extends to three times the projected radius of that sample. It unlikely that all the compact objects are associated with NGC 1399. This is emphasised by a finding chart for the central 55 arcminutes of the cluster in Fig. 5 which indicates the location of the compact objects. They are widely distributed over this field and the third object (FCSS J033854.1-353333) in particular is much closer to NGC 1404, although we note that its velocity is not close to that of NGC 1404 (see below).

**Figure 5:** The central region of the Fornax Cluster with the positions of the new compact objects indicated by squares. This R-band photographic image is from a single UKST exposure on Tech-Pan emulsion, digitised by SuperCOSMOS (Miller et al. 1992).
$\begin{figure} \hfil \psfig{file=figure5_10n.ps,width=15cm}\end{figure}$

Velocity Distribution

We have some limited information from the radial velocities of the compact objects. The mean velocity of all 5 (

$1530\pm110 \hbox{${\rm\thinspace km}{\rm\thinspace s}^{-1}\,$}$ ) is consistent with that of the whole cluster (

$1540\pm50 \hbox{${\rm\thinspace km}{\rm\thinspace s}^{-1}\,$}$ ) (Jones & Jones (1980). However, given the small sample, it is also consistent with the velocity of NGC 1399 (

$1425\pm 4 \hbox{${\rm\thinspace km}{\rm\thinspace s}^{-1}\,$}$ ) as might be expected for a system of globular clusters. Interestingly, the analysis of the dynamics of 74 globular clusters associated with NGC 1399 by Kissler-Patig et al. (1999) notes that their radial velocity distribution has two peaks, at about 1300 and 1800