5 Results

94 radial velocities were measured (not including the velocities of the giant galaxies of Table 1). The data are listed in Table 2 in order of increasing right ascension. The first column is the catalog name from Paper I. Column 2 gives the galaxy type as determined by morphological appearance, surface brightness profile (see Paper I), and spectral type. For the spectral classification the spectrophotometric atlas of galaxies by Kennicutt (1992) was used for comparison. Columns 3 - 5 are the right ascension, declination (epoch 2000.0) and total V magnitude from Paper I. Columns 6 - 9 are the heliocentric velocity of the cross correlation technique $v_{\rm cross}$, the R value, the velocity of the second determination method $v_{\rm rvid}$, and the emission line velocity $v_{\rm em}$, if available. In Cols. 10 and 11 the final adopted velocity and a mean error of the different determinations are given. In most of the cases, the final velocity is the mean of $v_{\rm cross}$ and $v_{\rm rvid}$.In the cases where the emission line velocity is based on at least 2 clearly identified lines it was also included. Column 12 indicates the quality parameter Q of the velocity determination. Table 2 is also available in electronic form at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/Abstract.html.

We defined 4 classes of quality due to the clearness and evidence of the velocity identification: (1) a clear and evident identification, both methods agree very well, (2) very probable identification, almost clear, (3) doubtful, but most probable value, and (4) very doubtful, only a try.

  

Table 3: Statistics of the quality of velocity determinations

Table 3 summarises the statistics of quality classes divided into the following sub-samples: velocities between 700 and 2500 km s^-1 (Fornax cluster, not including the galaxies of Table 1), velocities between 33000 and 34500 km s^-1 (background cluster, see Sect. 5.3), and all other background galaxies.

  

Figure 3: In this magnitude - surface brightness diagram all galaxies of our photometric sample brighter than V = 21.5 mag are plotted. Galaxies with velocity determinations are encircled with small circles. The background galaxies that show emission lines are indicated with triangles. Only the galaxies encircled with large circles are Fornax members. The spectra of most dwarf galaxies have a too low signal-to-noise for line identifications. Note that the dE, Ns FCC 188, FCC 222, and FCC 274 are not plotted, since they do not belong to our photometric sample

  

Figure 4: Velocity histogram for all galaxies with velocity determinations. The bin size is 2000 km s-1 starting at 500 km s-1. The concentration of galaxies at about 33700 km s-1 belongs to a background cluster just behind the center of the Fornax cluster

In Fig. 3 we show the selection of our sample in a $\mu_{\rm eff}-V_{\rm tot}$ diagram (see also Paper I, Figs. 3 and 7). All velocity determinations of the quality classes 1 to 3 are indicated. The limits are $V_{\rm limit} \simeq 20$ mag and $\mu_{\rm eff_{\rm lim}} \simeq 23.5$ mag arcsec^-2. The velocities of the fainter galaxies are mainly based on emission line identifications. Due to the low surface brightness of most dwarf galaxies (FCC numbers) the signal-to-noise of their spectra is too low for line identifications. Among the galaxies with higher surface brightness two objects with velocities consistent with Fornax members were found.

The distribution of velocities in our sample is shown in Fig. 4. The first bin represents the Fornax cluster velocities. The highest velocity (cz = 97000 km s^-1) corresponds to a redshift of z = 0.3. A striking feature is the concentration of galaxies around 33700 km s^-1 (z = 0.11). They belong to a background cluster behind the center of the Fornax cluster (see Sect. 5.3). Another group of galaxies with velocities around 26000 km s^-1 is concentrated in a region about $12\hbox{$^\prime$}$ east of NGC 1399 (field B1; see Paper I). The galaxies around 55000 km s^-1 are not spatially correlated.

5.1 Previously measured galaxies

Our velocity sample includes 12 galaxies studied before. Table 4 gives the different names, membership classifications, and previous radial velocity measurements.

Irwin et al. (1990) had no morphological classification for their targets, but noted that their compilation of low surface brightness galaxies, which they classified as members (Davies et al. 1988), might be contaminated by background galaxies in the magnitude range 17 < B < 19 mag and of central surface brightnesses of $22.0 < \mu_B <$ 22.5 mag/arcsec². Indeed, all galaxies in this range turn out to be background galaxies (CGF 9-1, CGF 3-6, and CGF 1-10). Concerning the FCC, all "likely background'' galaxies are proved to be true background galaxies (CGF 9-1, CGF 3-7, and CGF 10-2). In addition, the likely Fornax member FCC 141 (CGF 9-5) has a velocity of $v_{\rm helio} = 16845$ km s^-1. Due to its blue color, (V - I) = 0.8 mag, it was classified as irregular. The definite FCC galaxies are confirmed to be members. Their velocities agree well with previous results.

  

Table 4: Cross references and comparison with velocities to previous identifications

Recently, Minniti et al. (1998) have measured radial velocities of globular clusters around NGC 1399. Their brightest object, whose nature as GC was questionned, is the same as the nucleus-like object CGF 1-4 (see next section). Their velocity of $v_{\rm helio} = 1459~\pm~52$ km s^-1 agrees very well with our value: $v_{\rm helio} = 1485~\pm~38$ km s^-1.

  

Table 5: Number counts of GCs in the bright end of the GCLF from MC simulations. The total number of GCs is 5800. The bin centers are given in absolute V magnitudes

  

Table 6: Line indices for Fornax members, given in magnitudes and measured on the unfluxed spectra

5.2 New Fornax members

Three objects in our sample have velocities of the Fornax cluster, but have not been identified as members yet. Two of them, CGF 1-4 and CGF 5-4, are located at distances of $8\hbox{$.\mkern-4mu^\prime$}3$ and $13\hbox{$.\mkern-4mu^\prime$}6$ from NGC 1399, CGF 1-4 in the direction of NGC 1404 and CGF 5-4 in east direction. They are hardly resolved, have a circular shape, (V - I) colors of 1.1 and 1.0 mag, respectively, and a very high central surface brightness. These properties are typical for globular clusters as well as for nuclei of nucleated dwarf ellipticals, or even for compact ellipticals. Adopting a distance modulus to the Fornax cluster of (m-M)₀ = 31.3 mag, implying a distance of 18.2 Mpc, (Kohle et al. 1996, recalibrated with new distances of Galactic GCs, Gratton et al. 1997, following Della Valle et al. 1998), their absolute magnitudes are M_V = -13.4 mag and M_V = -12.2 mag, which is about 3 magnitudes fainter than M 32 (M_V = -16.4 mag) and 1 - 2 magnitudes brighter than the brightest GC of the investigated luminosity function (Kohle et al. 1996). For comparison, the brightest GCs in the central Virgo galaxy M 87 have absolute V magnitude in the order of -11.5 mag (Elson & Santiago 1996; Whitmore et al. 1995).

In order to investigate the possibility, whether these two objects might be "normal'' GCs in a very rich GCS, the number of GCs that populate the bright end of the luminosity function (LF) of the GCS in NGC 1399 was estimated in Monte Carlo simulations. As representation of the LF both a Gaussian and a t5-function (see e.g. Kohle et al. 1996) with different dispersions $\sigma$ were adopted. In 100 runs 5800 GCs were randomly distributed. In Table 5 the number counts in 6 bright bins (bin width 0.5 mag) are given for different functions and dispersions. Very bright GCs with M_V = -13.3 mag can statistically exist in a rich GCS, if the t5-function is representative for the bright wing of the LF.

According to numerical simulations by Bassino et al. (1994) nuclei of nucleated dwarf galaxies can survive the dissolution in the gravitational field of a giant elliptical over the lifetime of the universe. They expect globular-cluster-like remnants and less concentrated remnants with masses in the range between 2.8 to $7.4\ 10^6\ M_{\odot}$ and tidal radii from 170 to 400 pc.

Assuming that the two nucleus-like objects have mass-to-light ratios resembling GCs, their masses can be estimated. Adopting the relation of Mandushev et al. (1991), log$(M/M_\odot) = -0.431~M_V +2.01$, we derive masses of $6.1\ 10^7\ M_\odot$ and $1.9\ 10^7\ M_\odot$respectively. This is one order of magnitude more massive than $\omega$ Centauri ($2.5\ 10^6\ M_\odot$), the most massive cluster in the Milky Way. Alternatively, the mass of both objects can be compared with that of M 32. Nolthenius & Ford (1986) derived a mass of about $8\ 10^8\ M_\odot$and a M/L_B = 3-4 from velocity dispersion measurements. If we adopt a mass-to-light ratio of 3.5 for the two compact objects, we get masses of $6.3\ 10^7\ M_\odot$ and $2.1\ 10^7\ M_\odot$, more than one magnitude less massive than M 32.

The third object that has the velocity of the Fornax cluster, CGF 7-9, is located about $6\hbox{$^\prime$}$ north-east of NGC 1427A. Its velocity is very uncertain (Q = 4). Nevertheless, in case of membership, this galaxy belongs to the late-type dwarf population due to its blue color ((V - I) = 0.9) and irregular shape.

  

Figure 5: Mg2 is plotted versus the equivalent width of <Fe> = (Fe5270+Fe5335)/2. The filled circles are the nucleus-like objects CGF 1-4 and CGF 5-4. Open circles are the dE, Ns FCC 122 and NGC 1396. The dashed lines are tracks from population synthesis models for single stellar populations (long dashed: Fritze-v. Alvensleben & Burkert 1995, short dashed: Worthey 1994). The age range is 1.5 to 17 Gyr, the metallicity range $-2.0 < [{\rm Fe/H}] < 0.5$dex.

5.2.1 Line indices of the Fornax members

Metal abundances of the three dE, Ns NGC 1396, FCC 188, and FCC 222 and the two nucleus-like objects were measured in order to compare them with the properties of Milky Way, LMC and NGC 1399 GCs. The procedure for the measurement of the line indices is described in Brodie & Huchra (1990). The Lick/IDS bandpasses were used, as defined by Burstein et al. (1984) and updated by Trager (1997). In Table 6 the indices, given in magnitudes and measured on the unfluxed spectra, are summarised. In the Figs. 5, 6, and 7 the line indices of H$\beta$,<Fe>, MgFe, and Mgb$\ast$Fe52 are converted into equivalent widths by the relation $W_{\lambda}(I) = (\lambda_2 - \lambda_1)(1 - 10^{-I/2.5})$,where $\lambda_2$ and $\lambda_1$ are the maximum and minimum wavelengths of the bandpass. The errors were estimated from the photon noise in the bandpasses, $\sigma_{\rm P} = (\sigma^2_{\rm C1} + \sigma^2_{\rm L} + \sigma^2_{\rm C2})^{1/2}$, where $\sigma_{\rm L}$, $\sigma_{\rm C1}$, and $\sigma_{\rm C2}$ are the statistical errors in the line and adjacent continuum bandpasses (see Brodie & Huchra 1990). The statistical error in the flux in a bandpass is $\sigma = (O + S)^{1/2}/O$,where O and S are the total accumulated counts in the bandpass in the object and sky, respectively.

  

Figure 6: H$\beta$ versus Mg2 with tracks from Fritze-v. Alvensleben & Burkert (burk) (long dashed lines, ages on the left, metallicity between Z = 0.001 and Z = 0.04) and from Worthey (1994) (short dashed lines, ages on the right, metallicity between $[{\rm Fe/H}] = -2.0$ to 0.5 dex)

In Fig. 5 and Fig. 6 we plotted Mg2 versus the equivalent widths of <Fe> (= (Fe5270 + Fe5335)/2) and H$\beta$,respectively. The dE, N FCC 188 was omitted in the Mg2-<Fe> plot, since the iron lines in this spectra are too weak. Also plotted are the relations for the indices as derived from population synthesis models for single stellar populations. The long dashed lines are the models from Fritze-v. Alvensleben & Burkert (1995) for 1.5 to 16 Gyr old populations, with metallicities between Z = 0.001 and Z = 0.04. Short dashed lines are Worthey's (1994) models for 1.5 to 17 Gyr and $[{\rm Fe/H}] = -2.0$ to 0.5 dex.

The larger errors (due to the low signal-to-noise of the spectra) do not allow an age separation of the objects. Notice that the nuclei of the dE, Ns and CGF 5-4 fall in the range of metal-poor old single stellar populations. Similarly, metal-poor GCs in the Milky Way and M 31 (Burstein et al. 1984; Brodie & Huchra 1990) as well as in NGC 1399 (Kissler-Patig et al. 1998) are located in this region of the plot. On the other hand, in this diagram CGF 1-4 is clearly separated from these objects and shows a more metal-rich stellar population. Its line indices are comparable to those of the metal-rich GCs in the MW, M 31, and NGC 1399 as well as the line indices of the center of M 32 (Burstein et al. 1984), which shows a slight enhancement of H$\beta$ compared to GCs.

  

Figure 7: [MgFe] versus H$\beta$ for the dwarf galaxies (filled circles) and Fornax ellipticals (open circles) (upper panel), and [Mgb$\ast$Fe52] (see text) versus H$\beta$ for Virgo dwarf galaxies (open triangles) and NGC 1399 globular clusters (solid triangles) (lower panel). For reference, we show two models from Worthey (1994) for 17 and 8 Gyr, spanning a metallicity range [Fe/H] from -2.0 to 0.5 dex (dashed lines)

Adopting the Mg2$-[{\rm Fe/H}]$ relation from Brodie & Huchra (1990), the 4 more metal-poor objects have metallicities that range between $[{\rm Fe/H}] \simeq -1.6$ and -0.9 dex. This agrees well with the spectroscopic measurements of 10 dE, Ns in Fornax by Held & Mould (1994). The metallicity of CGF 1-4 is $[{\rm Fe/H}] \simeq 0.4$ dex, when adopting the Mg2 - [Fe/H] relation by Brodie & Huchra (1990), or $[{\rm Fe/H}] \simeq -0.1$ dex considering the non-linear behaviour of Mg at higher metallicities (Worthey 1994; Kissler-Patig et al. 1998).

Finally, we show in Fig. 7 a comparison between the abundances of the dwarf galaxies and elliptical galaxies in Fornax measured by Kuntschner & Davies (1998) (upper panel), as well as dwarf galaxies in Virgo (taken from Huchra et al. 1996) and globular clusters in NGC 1399 (taken from Kissler-Patig et al. 1998) (lower panel). The [MgFe] index ($\sqrt{{\rm Mgb} \times <{\rm Fe}\gt}$) was plotted versus H$\beta$, following Kuntschner & Davies. The Virgo dwarfs have no Fe5335 index measured. For them and the NGC 1399 GCs $\sqrt{{\rm Mgb} \times {\rm Fe5270}}$ (labeled [Mgb$\ast$Fe52]) was plotted versus H$\beta$. In both panels models from Worthey (1994) for 17 and 8 Gyr, spanning a metallicity range [Fe/H] from -2.0 to 0.5 dex, are shown for reference. The large errors in our measurement prevent a detailed comparison; age differences cannot be discriminated within a factor of 2 or 3. However, we note that, as expected, the dwarf galaxies (except CGF 1-4) are less metal rich than the giant ellipticals and consistent with having similar ages. The comparison with the Virgo dwarf galaxies and the NGC 1399 GCs shows that all the Fornax dwarfs (including CGF 1-4) fall in ranges span by them. However, CGF 1-4 is more metal rich than the bulge-like GCs in NGC 1399, and could belong to the "very metal rich'' group of GCs found by Kissler-Patig et al. (1998). In this case, it might also be somewhat younger, which would increase its mass-to-light ratio and reduce the estimated mass. But CGF 1-4 and CGF 5-4 remain puzzling objects; it can not yet be decided whether they are nuclei of disrupted dwarf galaxies, cEs, or true extremely massive GCs. Further spectroscopic observations with a higher signal-to-noise are needed to uncover the nature of these objects.

  

Figure 8: Velocity histogram for background cluster galaxies. The velocity dispersion is typical for a relatively poor galaxy cluster

5.3 Background cluster at z = 0.11

19 galaxies with velocities around 33700 km s^-1 (Fig. 8) were found. The velocity dispersion is about 360 km s^-1 which is typical for poor clusters (e.g. den Hartog & Katgert 1996). The ratio of early type (E+S0) to late type (S+Irr) giant galaxies is about 1.1. It is slightly lower than in the Fornax cluster. The spatial distribution of the 19 galaxies is shown in Fig. 9 (bold hexagons). It matches well the density distribution of the fainter galaxies down to V = 21.5 mag; see Fig. 9 in Paper I. Since no countable amount of new Fornax dwarf galaxies was found, we conclude that the excess population of galaxies near NGC 1399 mainly belongs to the background galaxy cluster.

Assuming a Hubble constant of H₀ = 75 km s^-1 Mpc^-1 the distance to the cluster is 450 Mpc, or (m - M)_V = 38.3 mag. The brightest cluster galaxy (CGF 1-1), located $1\hbox{$.\mkern-4mu^\prime$}1$ south of NGC 1399, would then have an absolute luminosity comparable to NGC 1399 (M_V = -22.1 mag). Note that the K-corrections at this redshift are in the order of 0.15 mag in V for old stellar populations and 0.05 mag for late-type spirals (Coleman et al. 1980). The radial surface brightness profile of this galaxy (see Paper I, Fig. 4) follows a de Vaucouleur profile in the inner part and becomes significantly flatter outside a radius of $7\hbox{$.\!\!^{\prime\prime}$}0$ ($\simeq 16$ kpc), which is typical for a cD galaxy (e.g. Schombert 1986).

  

Figure 9: Position of the background cluster galaxies in the central Fornax CCD fields. The dashed circle is the central galaxy NGC 1399. The points are galaxies brighter than V = 21 mag, the larger the point the brighter the galaxy. The hexagons indicate the velocities of the galaxies, the bold ones being velocities around 33700 km s-1. The larger the hexagons around a galaxy the smaller is its velocity. The dashed trapezium-like region was used for the analysis of the luminosity and color distribution (see text)

The luminosity and color distribution of the galaxies were analyzed inside a trapezium-like region that encloses the member galaxies of the background cluster (see Fig. 9). The distribution of an arbitrary sample of field galaxies around the other Fornax giant galaxies was subtracted from the distribution of the cluster galaxies.

A dip in the luminosity distribution at V = 19.5 was found that corresponds to an absolute magnitude of about M_V = -18.8 mag at a redshift of z=0.11. This is the luminosity range where the Gaussian shaped giant galaxy luminosity function decreases and the counts of dwarf galaxies start to rise (see e.g. Jerjen & Tammann 1997). Of course, the dwarf galaxy counts are severely uncomplete at the cluster distance due to their very small angular diameters below our resolution limit ($1\hbox{$.\!\!^{\prime\prime}$}5 \simeq 3.3$ kpc). However, at the fainter counts, an excess population of blue, (V - I) < 0.8 mag, dwarf galaxies was found, which probably represents a population of star-forming dwarf irregulars in the cluster. Most of these galaxies are clustered around the brightest cluster galaxy (CGF 1-1) and around a bright elliptical in the north-east of the cluster (CGF 2-2).

The galaxy CGF 2-2 is already known as a narrow-line radio galaxy (Carter & Malin 1983). It is correlated with the radio source PKS 0336-35. Its spectrum shows strong OII, NeIII, and NeV emission lines that Carter & Malin explained by a very bright high-excitation emission-line nucleus. Some of the blue faint galaxies in the direct environment of this galaxy have knots and tails. We propose that CGF 2-2 is the center of an interacting subgroup of galaxies within the background cluster.

In further investigations, this cluster might be used to constrain the amount of extinction within the central part of the Fornax cluster.