1 Introduction

Deep field studies have become one of the most powerful tools to explore galaxy evolution over a wide redshift range. One of the main aims of this kind of study is to constrain current evolutionary scenarios for galaxies, such as the hierarchical structure formation typical of Cold Dark Matter universes.

Undoubtedly, the Hubble Deep Field North (HDF-N, Williams et al. 1996) and follow-up observations with Keck were of particular importance to improve our knowledge of galaxy evolution in the redshift range z = 1 - 4(see e.g. the contributions to the HDF symposium, 1998, ed. Livio et al.). The HDF-N is the deepest multi-colour view of the sky made so far, with excellent resolution. A disadvantage of the HDF-N (and its southern counterpart, the Hubble Deep Field South (HDF-S, Williams et al. 2000)) is a relatively small field of view ($\sim$5.6 sq.arcmin). Therefore, its statistical results may be affected by the large-scale structure (Kajisawa & Yamada 2001; see also Cohen 1998) and by limitations due to small samples.

Figure 1: DSS plots of the FDF and of a field of the same size surrounding the HDF-S. Also indicated are the field boundaries of the HDF-S. Note the much lower surface density of bright foreground objects and the absence of bright stars in the FDF region.

Following the pioneering work of Tyson (1988) several ground-based deep fields with a wide range of scientific drivers, sizes, limiting magnitudes and resolutions have been initiated. Examples are the NTT SUSI Deep Field (NTTDF, Arnouts et al. 1996), which has a size similar to the HDFs and sub-arcsecond resolution, but is a few magnitudes less deep than the HDFs, or the William Herschel Deep Field (WHTDF, Metcalfe et al. 2001 and references therein), which has a much larger field of view, a depth comparable to the HDFs, but lacks sub-arcsecond resolution. Other surveys, such as the Calar Alto Deep Imaging Survey (CADIS, Meisenheimer et al. 1998), are much shallower, but cover much larger areas (several 100 sq. arcmin in the case of CADIS ) and are specifically designed to search for primeval galaxies in the redshift range z = 4.6 - 6.7.

The aim of the FORS Deep Field (FDF) is to merge some of the strengths of the deep field studies cited above. The FDF programme has been carried out with the ESO VLT and the FORS instruments (Appenzeller et al. 1998) at a site, that offers excellent seeing conditions and allows imaging to almost the depths of the HDFs. The larger field of view compared to the HDFs (about 4 times the combined HDFs) alleviates the problem of the large-scale structure and results in larger samples of interesting objects. Moreover, spectroscopic follow-up studies with FORS can make full use of the entire field. Using the FORS2 MXU-facility, up to $\sim$60 spectra of galaxies (within 40 slitlets) in the FDF can be taken simultaneously.

In the present paper, the field selection of the FDF, the photometric observations and the data reduction are described. The first results have been described in Jäger et al. (1999). A source catalog (available electronically) based on objects detected in the B and I bands and containing 8753 objects in the FDF is described and its properties are discussed. This catalog supersedes a preliminary I-band selected catalog, which had been discussed by Heidt et al. (2000). Photometric redshifts obtained from the FDF will be discussed by Gabasch et al. (in prep.; see Bender et al. 2001 for preliminary results). Spectroscopic follow-up observations of a subsample of the FDF galaxies have been started. Up to now, spectra of about 500 galaxies with redshifts up to $z \sim 5$ have been analyzed. Initial results have been described in Appenzeller et al. (2002), Mehlert et al. (2001, 2002), Noll et al. (2001) and Ziegler et al. (2002).