1 Introduction

Clusters of galaxies are increasingly viewed not as simple isolated and relaxed systems but as embedded in and connected to the general large-scale structure in the Universe. This view of clusters in a larger context has consequences for the interpretation of cluster galaxy populations, cluster dynamics and mass estimates.

Clusters of galaxies grow by continuously accreting galaxies and groups of galaxies from the surrounding field, mostly along filamentary structures. In the process, galaxies are transformed from the predominantly blue, actively star-forming, spiral population characteristic of the field to the red, passive and elliptical population characteristic of the inner and denser regions of clusters (Dressler 1980; Abraham et al. 1996; Balogh et al. 1998). Cluster galaxy populations evolve with redshift: rich clusters at high redshift contain a larger fraction of blue galaxies than local ones (Butcher & Oemler 1978, 1984; Cl0024+1654 is an example of a "Butcher-Oemler'' cluster). The exact nature of the interaction of infalling galaxies with the cluster environment (hot intra-cluster medium, tidal gravitational field) and its influence on the morphology of galaxies, their gas content and star-formation rates (as measured by galaxy colours and spectral type) are as yet ill-understood; hence the interest in investigating the "infall region'' beyond $\sim1\,h^{-1}\,{\rm Mpc}$ distance from the cluster centre, where the transition from field to cluster galaxies is taking place. The advent of new wide-field CCD mosaic cameras available on a number of large telescopes (e.g. CFHT, CTIO, Subaru, ESO 2.2 m) makes it possible to obtain photometric and morphological information on 1-10 Mpc scales around the cluster centres. However, wide-field investigation of clusters demands both imaging and spectroscopic observations. Individual spectra of galaxies describe their spectral energy distribution and provide their redshift, which is indispensable to produce a catalogue of cluster members with radial velocity and information regarding their stellar content and star formation history. At present there is only a limited number of clusters with more than $\sim$200 spectroscopically identified member galaxies (e.g. Abraham et al. 1996), and especially at high redshift ( ) spectroscopically well-studied clusters become very rare, mostly due to the fact that contamination by field galaxies increases rapidly with redshift.

The fact that clusters are not isolated systems also raises questions concerning the traditional ways of estimating masses of clusters of galaxies through different mass estimators: gravitational lensing analyses, kinematical analyses from redshifts of cluster member galaxies and X-ray observations.

Gravitational lensing is sensitive to the total integrated mass along the line-of-sight from the observer to the lensed sources, weighted by the appropriate combination of angular size distances between observer, lens and source (e.g. Schneider et al. 1992; Bartelmann & Schneider 2001). In the presence of massive structures other than the cluster along the line of sight, the mass derived from gravitational lensing overestimates the mass of the cluster proper. Large spectroscopic surveys provide additional information needed to correctly interpret the lensing analysis in this case. Metzler et al. (2000) investigate the influence of the presence of filaments and groups of galaxies in the vicinity of a cluster on weak lensing estimates of the cluster mass and find that significant overestimates (up to a factor 1.5 to 2) are possible and even likely. Similar investigations with comparable results were conducted by Cen et al. (1997) and Reblinsky et al. (1999).

A similar bias should be expected to affect measurements of velocity dispersions: if foreground or background groups of galaxies in the immediate neighbourhood of the cluster are added into the redshift histogram, but are not resolved and recognized as separate entities, the velocity dispersion of the cluster itself will be overestimated. Generally only 30 to 50 member galaxies are used to estimate the line-of-sight velocity dispersion (and virial cluster mass; see e.g. the large compilations of Girardi et al. 1998; Girardi & Mezzetti 2000). Furthermore, the measured redshifts are generally concentrated within a relatively small region within a projected radius of $\sim$ $500\,h^{-1}\,{\rm kpc}$ of the cluster centre. Whereas one can argue that these numbers might be sufficient for relaxed clusters with regular spatial and velocity distributions, the various derived estimates will contain large systematic errors if unresolved substructures are present. What can be obtained from redshift surveys is a galaxy number density weighted line-of-sight velocity dispersion averaged along the line-of-sight. With a sufficiently large number of cluster member redshifts it is possible to measure the variation of the line-of-sight velocity dispersion with projected distance from the cluster centre (Carlberg et al. 1997), but determination of even more detailed information on the dynamical status (e.g. velocity anisotropy profile) of a cluster requires a forbiddingly large number of redshifts (e.g. Merritt 1987).

Clearly, combining information coming from gravitational lensing (weak and strong), the X-ray emission from the hot intra-cluster gas, the Sunyaev-Zeldovich effect and the galaxy velocity distribution is the best way to arrive at secure mass estimates for clusters of galaxies (see e.g. Castander et al. 2000). This is now possible for several cases. Of particular interest are clusters for which a significant discrepancy between different mass estimates is found. One such cluster is the well-known lensing cluster Cl0024+1654. About 100 redshifts of galaxies in this cluster were obtained by Dressler & Gunn (1992) and Dressler et al. (1999), resulting in a velocity dispersion of $\sigma\simeq1300\,{\rm km\,s}^{-1}$, which is consistent with mass estimates derived from the spectacular arc system in the cluster centre (Kassiola et al. 1992; Smail et al. 1996; Tyson et al. 1998; Broadhurst et al. 2000). Cl0024+1654 was among the first clusters in which a coherent shear signal due to weak lensing was found (Bonnet et al. 1994). A crude mass estimate from this analysis was consistent with the strong lensing and kinematical estimates. In addition to the signal due to Cl0024+1654 itself, Bonnet et al. (1994) also found a coherent signal to the north-east of the cluster centre in an area where no obvious galaxy overdensity could be seen. The X-ray luminosity of Cl0024+1654 on the other hand is unusually low for a cluster of this velocity dispersion, and mass estimates from the X-ray observations are a factor of two to three lower than the lensing and kinematical estimates (Soucail et al. 2000).

In order to better understand the dynamics of Cl0024+1654 and how it is embedded in the surrounding large-scale structure, we have conducted a wide-field spectroscopic survey at the Canada-France-Hawaii Telescope (CFHT) and the William Herschel Telescope (WHT) from 1992 to 1996. In this paper we present the catalogue of the galaxies observed for this survey. Section 2 summarizes the photometric and spectroscopic observations. The data reduction and analysis are presented in Sect. 3. Section 4 discusses some interesting global results of the survey and describes some structures found in the redshift distribution that are not directly related to the cluster Cl0024+1654. A summary is given in Sect. 5. A detailed analysis of the dynamics of the cluster itself and its environment as well as the spectral properties of its member galaxies will be the subject of a forthcoming paper (Czoske et al. 2001).

Throughout this paper we use a Hubble constant $H_0 = 100\,h^{-1}\,{\rm km}\,{\rm s}^{-1}\,{\rm Mpc}^{-1}$, $\Omega_{\rm M} = 1$ and ${\Omega_\Lambda} = 0$, which gives a physical scale of $3.195\, h^{-1}\,{\rm kpc}\,{\rm arcsec}^{-1}$at the cluster redshift.