Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 317, 670-675 (1997)

Previous Section Next Section Title Page Table of Contents

2. The sample

The present work is based on a redshift catalogue involving the central regions of 15 clusters of different richnesses in the redshift range 0.01 [FORMULA] z [FORMULA] 0.06 (Paper I). Note that several of the clusters are well studied, nearby objects like Centaurus and Abell 400 or have been extensively observed for specific purposes (e.g. Abell 3558: Bardelli et al. 1994). Inside these regions nearly complete (80%) redshift information down to [FORMULA] [FORMULA] [FORMULA] could be obtained, as well as raw morphological types for most of the galaxies. The present sample differs from those of previous investigations because it concentrates onto the very cores of the clusters (R [FORMULA] 0.5 h-1 Mpc), and because of the homogeneity of kinematical information.
Cluster centers were chosen in order of preference from (a) published X-ray centers (Lahav et al. 1989, Edge & Stewart 1991), (b) the position of the cd-galaxy, if present, (c) the Abell cluster center or (d) by taking the center of the relevant Optopus field (see Paper I), which roughly coincides with the galaxy density peak. The only exception to this procedure was Abell 1736, whose OPTOPUS field was offset by 0.05 h-1 Mpc with respect to the X-ray center.

A common selection radius of 0.5 h-1 Mpc around each cluster centre was therefore chosen, using approximate mean cluster velocities and assuming all clusters to be located in the Hubble flow. For Hydra and Centaurus (z  [FORMULA]  0.01) two adjacent fields had been taken, and for the other very nearby clusters data from the literature was available to help fill the region.

Table 1 lists the fields and also gives the name of each cluster around whose centre the selection has been done (2), kind of center determination (3), where X: X-rays, cd: cd-galaxy position, A: Abell center, O: center of Optopus field, as well as centre coordinates (4), and selection radius in arcminutes (5).


Table 1. Cluster sample

2.1. Membership selection

As a first step towards meaningful cluster definitions, we then looked at velocity histograms in the direction of all 15 fields, eliminating obvious (5 [FORMULA]) fore- or background galaxies. In one case (field 12) two distinct structures could be identified: the main cluster at v = 4500 km s-1 and a smaller group at v = 11000 km s-1. Moreover, some non-gaussian velocity histograms with indication of bimodality could be recognized. A statistical test was thus applied which returns the likelihood that the biggest gap in the dataset could occur in a normal distribution (adapted from the ROSTAT package, Beers et al. 1990). Only in the case of field 9 a gap was found whose size was inconsistent at the 5% level with an underlying gaussian distribution. Field 9 was thus considered bimodal and subdivided at v = 11500 km s-1, the big gap location. To assess final cluster membership, a 3- [FORMULA] clipping technique (Yahil & Vidal 1977) was then applied to each of the 17 galaxy units recognized so far. Table 2 lists the number of member galaxies after 3- [FORMULA] clipping (column 3), as well as the resulting limits in redshift space (4).


Table 2. Redshift ranges and kinematical parameters

Mean cluster velocities and velocity dispersions were then determined for each cluster in the sample, using biweight estimators (Beers et al. 1990). These estimators have the advantage of being robust against outliers and are particularly useful when working with small datasets. Cosmological effects are taken into consideration following Danese et al. (1980). The resulting values and their uncertainties are listed in Table  2, columns 5 and 6.

Our cluster mean velocities and velocity dispersions do agree quite well with those quoted by Struble & Rood (1991) and Girardi et al. (1993). Nevertheless, there are some cases where a discrepancy is found. The reason herefore must be related to the effect of substructures in the central cluster regions, which tend to inflate velocity dispersions, and to the presence of luminous galaxies of low dispersion in the cluster cores (as shown in the next sections). Nevertheless, given that our results are virtually free from larger scale contamination, we consider them good estimators of central mean velocity and velocity dispersion. With the present data it was not possible to discern the bimodal structure of Centaurus using a gapper test. For this reason we give a global value for the mean cluster velocity and for the velocity dispersion. Nevertheless, substructure is detected in this and in some of the other clusters, as we will see below. It may be, therefore, that the kinematic parameters given in Table 2 do not reflect the true dynamical state in 50% of the cases. They should be treated as first estimations and were calculated because a measure for the cluster dispersion is needed for the Lee-test simulations (see below).

Some of the galaxy samples are clearly too poor to be used for the substructure analysis and have to be discarded. Computation of the Dressler & Shectman (1988) substructure test requires the evaluation of velocity dispersions from samples of [FORMULA] neighbours around each galaxy, where N is the total number of galaxies (Bird 1994). Given that 5 is a lower limit for the determination of standard deviations, we chose a number of [FORMULA] galaxies as the minimum richness.

We are thus left with 12 clusters (flag "Y" in Table 2, column 7) and a total of 576 galaxy redshifts, 2/3 of which are taken from Paper I and have mean errors well below 50 km s-1.

2.2. Photometric data

As we wanted to look at dynamical friction effects, luminosity information was necessary. For three of the clusters detailed photometrical studies could already be found in the literature: A 3526 = Centaurus by Dickens et al. (1986), A S 805 by Millington & Peach (1989) and A 4038 by Green et al. (1990). In addition, [FORMULA] magnitudes from the COSMOS catalogue were kindly provided by H. MacGillivray (1993) for ten clusters of high galactic latitude. Given the selection of high latitude clusters no correction for reddening or extinction was applied.
A comparison of COSMOS magnitudes with those of Green et al. (1990) in A 4038 shows an excellent agreement, with differences of only a few hundredths magnitudes. For some of the brightest spiral galaxies COSMOS magnitudes were not available (due to the problem of resolved HII regions), which meant that magnitudes had to be taken from a bright galaxy catalogue. It should be noted that magnitudes from the COSMOS catalogue are [FORMULA] magnitudes, while those for Centaurus are G26.5 magnitudes, and for most of the other clusters generic optical magnitudes were taken from different sources. Magnitudes were not transformed to a common scale, because for our statistical analysis a spread of a few tenth of magnitudes for individual galaxies could be taken into account.

2.3. Morphological data

Galaxy types were searched for in the literature, mainly resulting in a sample from Dressler's (1980) and UGC (Nilson 1973) catalogues, as well as from Huchra's (1991) collection, as implemented in the DIRA2 database (Astronet Data Base Group Italy, Bologna). All galaxies were then divided into three classes: E, S0 and S. In addition, many of the remaining galaxies were classified by the author into one of the above classes, after visual inspection of ESO-Schmidt plates. Because careful classification is difficult on this kind of plates, a check was made involving 105 galaxies with independently known types. Of these, 15 were classified by the author with an "earlier" type and 13 with a "later" type than literature values, corresponding to an agreement of around 75%.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: July 8, 1998