Astron. Astrophys. 321, 84-104 (1997)

8. Discussion and conclusions

Our study of the properties of emission-line galaxies in clusters has yielded several results which we will now try and put together into a more or less coherent picture. Some results are not unexpected and confirm earlier results by other authors. On the other hand, we also obtained some results that are totally new (among which the analysis of the ELG orbits), and which are based on a sample of many tens of rich clusters. Thereby our data provide evidence for the general occurence of dynamical effects that up to now were seen only in one or two individual clusters.

In the following discussion it must always be clearly realized that our ELG simply are galaxies that had detectable emission lines in the ENACS spectra. In several instances we will think of them as (mostly late-type) spirals. This is justified as we found (for a subset) that well over 90% of them were classified as spirals. However, they represent only about 1/3 of the total spiral population, as a result of our observational limit, and the variation in gas content of spirals of different types. When comparing the ELG with the non-ELG, we are thus always comparing a very homogeneous class (ELG; read: spirals) with a heterogeneous class consisting of both non-spirals (ellipticals and SO's) and spirals.

The most striking results that we obtained concern the spatial distribution and the kinematics of the ELG, especially when compared with those of the non-ELG.

First, the ELG very clearly avoid the central regions of clusters, and the difference in central concentration of ELG and non-ELG probably implies an even larger difference in the central concentration of spirals and non-spirals. In the cluster Abell 576, Mohr et al. (1996) recently also found a clear deficit of ELG in the central region. The different spatial distributions are totally consistent with the well-documented dependence of the mix of early- and late-type galaxies on local galaxy density. The clear dependence of the fraction of ELG on the global of a cluster that we found can easily be understood as a manifestation of this dependence. Several physical mechanisms have been proposed for the dependence of galaxy mix on local density. We will argue below that the kinematics of the ELG make it likely that they still contain gas that produces detectable emission lines because they have not yet been inside the high-density central cluster region.

The characteristic of the kinematics of the ELG that supports this explanation most convincingly is their high velocity dispersion. For the reasons explained above, we expect the observed ratio of the 's of ELG and non-ELG of 1.2 to translate into a larger ratio for spirals and non-spirals. In this respect it is noteworthy that Colless & Dunn (1996) find that of the late-type galaxies in the main concentration in the Coma cluster is very close to times that of the early-type galaxies, which they interpret as suggesting that the late-type galaxies are freely falling into the cluster core.

Since we applied an interloper removal criterion irrespective of whether a galaxy was classified as ELG or non-ELG, all our ELG (including those projected onto the cluster core) are cluster members, i.e. are within the present turn-around radius of their cluster. We expect therefore that, had we been able to compare of the spirals with that of the early-type non-ELG (i.e. excluding the non-ELG spirals) we would have obtained the same result as did Colless & Dunn.

Our result of a systematically larger for ELG than for non-ELG (which is based on an ensemble of many clusters) is supported by recent observations of some individual clusters; Mohr et al. (ibid.) find a similar effect in Abell 576, and Carlberg et al. (1996) conclude for a sample of about 15 clusters with redshifts between 0.15 and 0.55 that '... the bluer galaxies, which often contain measurable emission lines, statistically are found to have a higher velocity dispersion than the redder absorption line galaxies, an effect that is particularly prominent near the projected center of the cluster ...'.

We believe that the larger of ELG (and of the spirals) is a generic aspect of the dynamics of galaxy clusters. It probably indicates that the spirals that we see today avoid the central regions because they either have not yet got there (the free-fall time is certainly not much shorter than the Hubble time), or have passed by the core on orbits that did not traverse the very dense central region. In other words: the dynamical state of the ELG reflects the phase of fairly ordered infall (of spirals) rather than the virialized condition in the relaxed core, the size of which is probably only 0.5 h^-1 Mpc (e.g. den Hartog and Katgert 1996).

In this picture, the orbits of the ELG (and therefore of the spirals) are expected to be fairly radial, and their velocity distribution is expected to be quite anisotropic. The statistical weight of our synthetic cluster with 549 ELG has allowed, we think for the first time, a meaningful check of this prediction to be made. The uncertainties of the ratio of the inner and outer 's still prevent the anisotropy parameter to be solved for with high precision. However, the strong rise of of the ELG towards the centre, which was also seen by Mohr et al. in Abell 576 and by Carlberg et al. (see above), quite strongly supports the notion of predominantly radial orbits of at least the ELG that are projected onto the central region.

A moderate to fairly strong anisotropy of the velocity distribution of the ELG can also solve the apparent discrepancy of the mass estimates based on ELG and non-ELG. We do not have a very accurate estimate of the magnitude of the discrepancy because the non-ELG category does contain spirals. Yet, the mass derived from spirals will (in terms of the mass indicated by the non-spirals) be at least as large as that derived from the ELG compared to the mass of the non-ELG, unless the kinematics of the non-ELG spirals is totally different from that of the ELG. From the discussion in Sect.  7, we estimate a discrepancy of at least a factor 1.5. In the projected mass estimator

where N is the number of galaxies in the system, the observed velocity along the line-of-sight, and the projected clustercentric distance of the galaxy (Heisler et al. 1985), the factor is a projection factor that depends on the distribution of orbits. It is equal to and for the cases of radial () and isotropic () orbits, respectively. More generally, one can show that

where the anisotropy parameter is assumed constant throughout the system (see also Perea et al. 1990). If the ELG indeed are not in equilibrium with the relaxed core, the mass estimate based on them could be twice as large as the one based on the other galaxies. This means that f() may be as large as 4/3, which implies that may be as large as 0.7, which is consistent with the best value for derived in Sect.  7.

The assumption that the spirals that we observe today are mostly falling in for the first time is also consistent with the fact that, contrary to earlier claims, we do not see any need for different emission-line properties of ELG in clusters and ELG in the field. In this respect it is noteworthy that only a fairly small fraction of the ELG occur in compact subgroups. Our data are thus consistent with a picture in which the infall of the spirals is rather isotropic. Whether this is indeed so, or an artefact of our analysis, in which we combined many clusters most of which contain only a fairly small number of ELG, will become clear as soon as the results from the more extensive multi-object cluster spectroscopy, that is presently under way, will become available.

© European Southern Observatory (ESO) 1997

Online publication: June 30, 1998
helpdesk.link@springer.de