4. Morphological and physical properties of Abell 496
4.1. Morphology of the cluster at various wavelengths
We display in Fig. 4 the superposition of the optical image of the cluster with ROSAT PSPC X-ray and radio isocontours. The X-ray contours are quite smooth, with no obvious substructures. However, there appears to be an excess of X-ray emission towards the north west, in the direction where there is also an excess of emission line galaxies (see below). A radio source is visible south east of the cluster, probably associated with a galaxy.
4.2. The galaxy velocity distribution in Abell 496
The cluster Abell 496 corresponds to structure 3 in Table 1 ; it has a BWT mean velocity of 9885 km s-1 and a global velocity dispersion of 715 km s-1. The corresponding velocity interval is [7813,11860 km s-1] and includes 274 galaxies. Note that the cD galaxy has a velocity of 9831 km s-1, close to the mean cluster velocity, and is located very close the X-ray emission center, suggesting that the cD is at the bottom of the cluster gravitational potential well. This is an indication of a quiescent history of the cluster (see e.g. Zabludoff et al. 1993, Oegerle & Hill 1994).
The wavelet reconstruction of the velocity distribution of Abell 496 shown in Fig. 5 (274 galaxies) suggests the presence of a certain amount of substructure. The sample was analyzed with 256 points, and the two smallest scales were excluded. The corresponding velocity distribution is non-gaussian; it shows: a tiny feature at 8300 km s-1; a main asymmetric peak in the [8500, 10700 km s-1] range containing 232 galaxies, with a BWT mean velocity of 9769 km s-1, and a BWT velocity dispersion of 518 km s-1; note that this velocity structure is not quite centered on the velocity of the cD galaxy; a smaller peak at 10940 km s-1 with 36 galaxies in the [10700, 11860 km s-1] range. If we only keep the largest scales, we are left with a rather symmetric velocity distribution showing an excess at high velocities. This excess corresponds to the peak at 10940 km s-1, which contains a small number of galaxies (see Sect. 4.4).
These structures are also found by applying a rank-velocity classification, which gives two breaks globally consistent with those found by analyzing the cluster velocity distribution. Such breaks probably indicate substructures with velocities coherent with the finer analysis based on the wavelet technique. However, the number of galaxies involved in these structures is small, and the velocity distribution in the main cluster therefore appears to be quite smooth, suggesting that Abell 496 is quite well relaxed.
In order to confirm the state of relaxation of Abell 496, we have applied a Serna & Gerbal (1996) analysis to the subsample of 96 galaxies located within a radius of 1800 arcsec around the cD and with magnitudes R17.0; within this limited sample, the completeness of the redshift catalogue is 82% and this type of analysis is expected to give robust results. Note that galaxies in this region with measured velocities but without published magnitudes were discarded. Results are displayed in Fig. 6. At the extreme lower right of the figure, we can see the very tight pair made by the cD (#280, R=12.2) and a satellite galaxy (#292, R=15.6, in the Durret et al. 1999b catalogue): this confirms that the cD is at the bottom of the cluster potential well. We also observe a structure of 11 bright galaxies (10 galaxies with R15.8, plus one with R=16.8) highly concentrated in space around the cD (mean distance to the cD: 216 arcsec, with a dispersion of 48 arcsec) but not in velocity (BWT mean velocity and velocity dispersion: 9745 and 375 km s-1). This result is comparable to what is found in other clusters, where the central core is more or less well discriminated. The main body of the cluster center appears quite relaxed, with no strong subclustering within a radius of 1800 arcsec (1.65 Mpc). This picture agrees with the general shape of Abell 496 seen in X-rays (see Fig. 4).
4.3. Luminosity segregation in Abell 496
After violent relaxation, two-body gravitational interactions lead to a certain level of energy equipartition between galaxies of various masses, and consequently to a certain segregation in velocity dispersion with luminosity (mass). This process concerns essentially massive galaxies, and added to dynamical friction, it creates segregation with distance to the cluster center. The stage of post-violent relaxation therefore leads to a segregation in the [L,] space larger in the central regions than in the overall cluster.
In order to search for such effects in Abell 496, we have derived the velocity dispersion and the average distance of galaxies to the cluster center (defined by the position of the cD) in several magnitude bins. We restrict our sample to galaxies belonging to the cluster, i.e. in the velocity range [7813,11860 km s-1], and within 1000 and 1800 arcsec from the cluster center, in order to have reasonably complete samples:100% and 79% complete for R respectively. The completeness is estimated by comparing the number of galaxies with measured redshifts to the number of galaxies in our photographic plate catalogue, for the same R magnitude limit. Since there are 2 galaxies with R and 6 with 14R15, we chose to fit the data with two different "brightest" bins: one including the 8 galaxies with 12R15, and the other including only the 6 galaxies with 14R15.
The velocity dispersions estimated in several magnitude bins are different for the two samples, as shown in Fig. 7. In a 1000 arcsec radius, the velocity dispersion increases more steeply with magnitude: the corresponding slopes are and km s- 1 mag-1 when the brightest bin is included or not respectively. In a 1800 arcsec radius, the velocity dispersion increases less with magnitude, the corresponding slopes being and km s-1 mag- 1.
As seen in Fig. 8, the average distance to the cluster center is somewhat smaller for the galaxies located in the brightest bin (R15), then remains roughly constant with a possible decrease with increasing magnitude, specially in the broadest sample (1800 arcsec radius).
Interestingly, we can notice that both the distance to the cluster center and the overall velocity dispersion range are reduced when emission line galaxies (hereafter ELGs) are excluded. This agrees with the general scheme that ELGs are often found in the outskirts of clusters of clusters and are not as strongly tied to the cluster gravitationally (e.g. Biviano et al. 1997). We now discuss in more detail the properties of ELGs in Abell 496.
4.4. The emission line galaxy distribution
We now compare the distribution of emission line (ELGs) versus non-emission line (NoELGs) galaxies. There are 85 ELGs and 381 NoELGs in our velocity catalogue, among which 34 ELGs and 241 NoELGs in the velocity range of Abell 496. The global percentage of ELGs in the cluster is therefore %. Note that this percentage is perfectly coherent with the proportions observed by Biviano et al. (1997) in the ENACS survey.
The spatial distribution of the 211 NoELGs and 21 ELGs belonging to the main [8500, 10700 km s-1] velocity peak is displayed in Fig. 9. The fraction of ELGs in this velocity range is 93%, consistent with the global cluster value within the error bars. NoELGs appear rather homogeneously distributed, except for a sort of linear north-south concentration towards the center. On the other hand, a large majority of the ELGs in this velocity range is located west of a north to south line crossing the center, and at least half of these ELGs even seem to be close to the west cluster edge. This agrees with the fact that ELGs tend to concentrate in the outer regions of clusters (see e.g. results based on ENACS data by Biviano et al. 1997). The presence of an excess of ELGs can at least in some cases be interpreted as due to merging events producing shocks which trigger star formation. This was shown to be the case in the zone of Abell 85 where the X-ray filament merges into the main body of the cluster (Durret et al. 1998): an excess of ELGs was observed in that region, together with a temperature increase of the X-ray gas. The ASCA X-ray gas temperature map presently available for Abell 496 (Markevitch et al. 1999, Donnelly, private communication), does not show any temperature increase for the X-ray gas in that area, so we cannot correlate the excess of ELGs towards the west cluster edge with a higher gas temperature zone. However, we can note that this excess is located roughly in the same region as the X-ray excess emission in the north west region of the cluster (Fig. 4). Such an X-ray enhancement could be due to a merging event originating from the north west, but our data cannot show this with certainty.
On the other hand, the spatial distributions of the 23 NoELGs and 13 ELGs in the [10700, 11860 km s-1] velocity interval are comparable (Fig. 9), while the ELG fraction seems much higher: 3618%. Though the small number of objects may introduce errors, there definitely seems to be an excess of ELGs with somewhat higher velocities than the bulk of the cluster; these ELGs account at least partly for the peak at 10940 km s-1 in the wavelet reconstruction of the velocity distribution.
A general picture for the ELG distribution in Abell 496 is that of two samples of galaxies falling on to the main cluster, one from the back (the ELGs concentrated towards the west) and one from the front (the high velocity ELGs).
4.5. The galaxy luminosity function
We have seen in the previous section that Abell 496 appears to have properties common to many clusters, with a relaxed main body and ELGs probably falling on to the cluster. We therefore expect its galaxy luminosity function not to be strongly modified by environmental effects, as observed in some clusters showing more prominent substructures. We discuss below its main features.
4.5.1. The bright end of the galaxy luminosity function
We have first derived the galaxy luminosity function (GLF) of Abell 496 in the R band from the redshift catalogue, within a radius of 1800 arcsec around the center, and for a limiting magnitude R=18.5. There are 196 galaxies in this sample. The completeness of the redshift catalogue within these limits is 79%, and it is 100% in that region for R16.0. The obvious interest of such a GLF is that no background contribution needs to be subtracted, therefore making the results very robust. We have limited the magnitude interval to the [13,18.5] range, because for R13 there is only one galaxy (the cD), which introduces edge effects in the wavelet reconstruction of the GLF, and for R18.5 the completeness sharply decreases. This corresponds to the  interval in R absolute magnitude.
The GLF obtained after a wavelet reconstruction is shown in Fig. 10. The sample was analyzed with 128 points, excluding the two smallest scales. The significance level of the detected features is at least 3. A flattening is observed at R16, corresponding to an absolute magnitude M. This shape is comparable to that found in Virgo (Binggeli et al. 1988) and in Abell 963 (Driver et al. 1994), where a comparable flattening was observed at a common absolute magnitude of -19.8. The GLFs of Coma and Abell 85 are more complex, with a "bump" corresponding to the brightest galaxies, followed by a "dip" at M (see Fig. 9 in Durret et al. 1999a). Note that the flattening of the GLF in Abell 496 occurs exactly at the same absolute magnitude as the dip in Abell 85 and Coma.
The GLF in Abell 496 suggests at least a bimodal galaxy distribution, with bright (mostly elliptical) galaxies in the bright part and dwarf galaxies in the fainter part. We therefore performed a fit of the wavelet reconstructed GLF of Abell 496 by summing two functions: a gaussian, to account for bright galaxies, and a power law (case 1) or a Schechter function (case 2) to represent faint and/or dwarf galaxies. In case 1, we fit the data as a function of apparent R magnitude; the gaussian is then found to be centered on R=15.190.01, with , and the power law varies as R. In case 2, we fit the data as a function of absolute R magnitude, to allow a direct comparison with other authors; the gaussian is then found to be a little broader, centered on R=15.480.04, with ; the Schechter function, defined as in Rauzy et al. (1988, Sect. 3.2.3), has and M (in the  absolute magnitude range).
The GLF resulting from these various fits is shown in Fig. 10; it obviously reproduces the data very well. Except at the faint end where the sample incompleteness most probably modifies the GLF shape, both fits 1 and 2 are good but we cannot distinguish between them. In view of the obvious quality of the fit, we did not attempt to estimate error bars with Monte-Carlo simulations, as done previously e.g. for Abell 85 (see Durret et al. 1999a, Fig. 12).
The various values obtained from these fits of the GLF can be compared to those found in other clusters. The gaussian used to fit the bright part of the GLF in Abell 85 has , comparable to the value we find in case 1. The Schechter function for Abell 496 has a slope comparable to that found by Lumsden et al. (1997), but notably flatter than the values found in other surveys (e.g. Valotto et al. 1997, Rauzy et al. 1998 and references therein). Rauzy et al. (1998) argued that the flatter slope found by Lumsden was due to incompleteness at faint magnitudes; this may also be true in our case, since our sample is 100% complete only to R=16.0 (M), and we also find a brighter value of M* than the above surveys, suggesting that we are missing faint galaxies.
We can note that the GLFs of Coma and Abell 85 were interpreted in a similar way, with the bright part mainly due to ellipticals (with a small contribution of spirals) and the faint part due to dwarfs (Durret et al. 1999a). Comparable shapes were found in several other clusters. The fact that the GLF of Abell 496 shows a flattening at the same value M indicates that the galaxy population in Abell 496 is comparable to those of the above mentioned clusters.
Note that Molinari et al. (1998) have analyzed the GLF of Abell 496 from a photometric catalogue in three colors, reaching magnitudes much fainter than those of our spectroscopic catalogue. We will therefore compare our results to theirs in the next section.
4.5.2. The faint end of the galaxy luminosity function
Our intent was also to derive the luminosity function from the CCD catalogue, which corresponds to a small region of 246 arcmin2 in the cluster center. For this we first performed a wavelet reconstruction of the R magnitude distribution in the R magnitude range [17,23]. Since we have no background exposure, we estimated the background contribution by connecting the counts from the Las Campanas Redshift Survey (LCRS, Lin et al. 1996) and from the ESO-Sculptor Survey (ESS, Arnouts et al. 1997), as described in our study of Abell 85 (Durret et al. 1999a, Fig. 10 and text), and we subtracted this background to the observed number of galaxies. The result is shown in Fig. 11. We have checked that the consistency of the background number counts estimated by Tyson (1988) with those of the LCRS and ESS combined as described above is good.
The difference between the observed number of galaxies and the background (Fig. 11) becomes negative for magnitudes R, while the CCD catalogue is complete at least up to R=21. Therefore, this background cannot be considered as representative of the local background in our CCD field of view. Note that this was already the case for the CCD photometric data of Abell 85.
One notable feature is the dip in the galaxy magnitude distribution at R (M), which is detected at a high confidence level. This dip corresponds to that observed by Molinari et al. (1998), who found a dip at R19 (M). Note that they also find a similar dip in the g band, and possibly in the i band. Molinari et al. (1998) made a second determination of the GLF by selecting cluster members in a colour-magnitude diagram. In this case, they find a small dip, or at least a flattening, for R (M). This value does not agree either with the bright nor with the faint GLF that we derived. It is difficult to understand why, since their colour-magnitude relation appears quite well defined.
In order to investigate the origin of the dip seen in our data, we propose a toy model, which is not a fit but only illustrates how the dip could be accounted for. Let us first note that the contribution of the other structures detected along the line of sight is negligible. Assuming a Gaussian + a Schechter function to model the GLF (see Sect. 4.5.1), we rescaled the number of galaxies produced by this composite function to fit the dimension of the CCD field. We then applied a magnitude cut-off to this GLF, as suggested by Adami et al. (2000), for galaxies fainter than M in the inner core of the Coma cluster. This effect becomes very strong for galaxies fainter than M. The exact shape of such a cut-off is unknown, so we applied a convenient apodization function (the choice of this function influences the shape and smoothness of the dip). The background counts were modeled as the background contribution from the LCRS and ESS described above. We then summed the cluster and background contributions, and the result is shown in Fig. 12.
Such a toy model can reproduce the global GLF shape, with counts similar to the observed data and a dip comparable to the observed one. A fine-tuning of the various parameters involved could make Figs. 11 and 12 more similar, but this would push the model too far. However, we can state that a cut-off in the GLF of Abell 496 similar to that observed in Coma is a solution to account for the observed dip.
© European Southern Observatory (ESO) 2000
Online publication: April 17, 2000