Clusters of galaxies are the largest virialized structures observed in the Universe. Since they arise from exceptionally high peaks of the primordial fluctuation density field, their properties are highly sensitive to the nature of such cosmic fluctuations. Therefore, the mass function of both local (e.g. White et al. 1993; Girardi et al. 1998) and distant clusters (e.g. Oukbir & Blanchard 1992; Carlberg et al. 1997; Eke et al. 1998; Borgani et al. 1999) is a powerful tool to constrain cosmological models for the formation and evolution of cosmic structures. Moreover, clusters are useful laboratories for testing models of galaxy evolution. While early-type galaxies only show evidence for passive evolution (e.g. Stanford et al. 1998), the fraction of blue galaxies increases significantly with redshift (Butcher & Oemler 1978), at least up to , and the fraction of S0's decreases (Dressler et al. 1999). It is therefore essential to have reliable cluster catalogues over the largest possible redshift range.
Most distant clusters, at , have so far been identified through optical follow-ups of X-ray selected clusters (see, e.g. Gioia et al. 1990and Rosati et al. 2000for a recent review), or by looking at the environment of high-redshift radio galaxies (e.g. Smail & Dickinson 1995; Deltorn et al. 1997).
In the optical, clusters at and beyond started to be classified in the 80's (Gunn et al. 1986). In the 90's a large catalogue of objectively selected distant clusters, identified in the optical, became available (Postman et al. 1996). These last clusters are identified using a matched-filter algorithm using both positional and photometric data. In brief, this algorithm filters a galaxy catalogue to remove fluctuations in the projected distribution of galaxies that are not likely to be galaxy clusters. For this purpose, the filter is built around parametrizations of the spatial distribution and luminosity function of cluster galaxies. This algorithm also provides an estimate of the redshift for each candidate cluster (hereafter we refer to the matched-filter estimated redshift as ). Currently, PDCS clusters have been confirmed spectroscopically, most of them at (Holden et al. 1999a, 1999b; Oke et al. 1998).
Recently, Olsen et al. (1999a, 1999b) and Scodeggio et al. (1999) have presented a catalogue of 302 cluster candidates from the I-band images of the ESO Imaging Survey (EIS, see Renzini & da Costa 1997). Clusters are identified in two dimensions (hereafter, 2-d) using the matched filter algorithm of Postman et al. (1996; see Olsen et al. 1999a). The estimated redshifts for EIS clusters span the range , with a median redshift .
Several EIS cluster candidates have been confirmed so far, most at , either from the existence of the red sequence of cluster ellipticals/S0's in colour-magnitude diagrams (Olsen et al. 1999b), or from a combination of photometric and spectroscopic data (da Costa et al. 1999).
The EIS cluster catalogue is the largest optically selected cluster sample currently available in the Southern Hemisphere to this depth. This catalogue constitutes an obvious reference for follow-up observations at the ESO VLT aimed at determining the structure and dynamics of distant clusters, as well as the spectroscopic properties of their member galaxies. Unfortunately, little is currently known on the performance of the matched filter algorithm in detecting real clusters at . As we already pointed out, most confirmed PDCS and EIS clusters have redshifts . Therefore, to point blindly at EIS cluster candidates would make for an inefficient use of VLT time, because we expect several of these candidate clusters not to be real, in particular at .
The aim of our investigation is twofold: we want to confirm as many EIS clusters as possible, in order to build a reliable sample of distant clusters with well determined redshift, and, at the same time, evaluate the performance of the matched filter algorithm in the detection of high-redshift clusters. In order to achieve this purpose, we use two independent methods: (1) multi-object spectroscopic observations of EIS cluster candidates in the redshift range , and (2) the detection of the colour-magnitude sequences traced by early-type galaxies through multi-colour optical and near-IR photometry of the most distant EIS cluster candidates (Scodeggio et al., in preparation).
In this paper we report the first results of the spectroscopic investigations of 6 EIS clusters. We are able to confirm the existence of significant concentrations in redshift space in correspondence of four of the six EIS fields targeted. For two of these confirmed clusters, the spectroscopic mean redshift agrees with the matched-filter estimate to within .
In Sect. 2 we describe our spectroscopic observations, data reduction, and give the new galaxy redshifts. In Sect. 3 we analyse the data, and define sets of galaxies in redshift space. We also discuss the concordance of the mean redshifts of these sets with the matched-filter estimates of the cluster mean redshifts. We then make a likelihood analysis of the reality of the galaxy sets, and flag four of them as reliable at % confidence level (Sect. 4). Finally, we discuss our results and give our conclusions in Sect. 5.
We use H h75 75 km s- 1 Mpc-1, and throughout this paper, unless otherwise stated.
© European Southern Observatory (ESO) 2000
Online publication: August 23, 2000