1 Introduction

Among the unsolved questions of cosmology is the beginning and evolution of the star formation process at early cosmic epochs. From the presence of heavy nuclei in high-z quasars and galaxies it is clear that star formation started rather early (see e.g. Hamann & Ferland 1999; Dietrich et al. 1999). Moreover, galaxy counts and emission line studies of high-z galaxies indicate that the star formation rate (SFR) declined rapidly since at least the epoch corresponding to redshifts of $z \approx 2$ (see e.g. Madau et al. 1996; Madau 2001). On the other hand intrinsic interstellar extinction (which may affect strongly the rest-frame UV of distant galaxies) and uncertainties concerning the relation between line emission and the star formation introduce considerable uncertainties into the derivation of the global star formation rate at early cosmic epochs from visual galaxy counts and emission line studies, while investigations using IR and mm-wave data are hampered by small samples.

An alternative approach to investigate the early star formation history of the universe is the evaluation of the chemical enrichment history of the universe, as star formation and the rapid evolution of massive stars results in a production of heavy nuclei more or less proportional to the SFR. Steidel et al. (1996a, 1996b) and Lowenthal et al. (1997) have demonstrated that galaxies with redshifts up to about $z \approx 5$ can be observed with optical photometry and spectroscopy during periods of high star formation activity (resulting in high rest-frame UV emission which is redshifted into the optical wavelength range in the observers frame). Basic properties of these objects, such as number densities, luminosities, colors, sizes, morphologies, star formation rates, overall chemical abundances, dynamics and clustering have been investigated in various recent papers (cf. e.g. Steidel et al. 1996b; Yee et al. 1996; Lowenthal et al. 1997; Pettini et al. 2000; Leitherer et al. 2001). Steidel et al. (1996a, 1996b) already noted that the high-z galaxies show, on average, relatively weak metallic absorption lines, and they ascribe this finding tentatively to a lower metal content (a suggestion which has later been reiterated in several subsequent papers). On the other hand most of the published spectra of high-z galaxies are not of sufficient S/N to provide qualitative information on the metal content. Therefore, we obtained new high S/N spectra of galaxies with $z \le 3.5$ with the aim of studying the chemical evolution of starburst galaxies at high redshifts.

In the present paper we describe results on the C  IV absorption line strength and their interpretation in terms of chemical evolution with cosmic age at redshifts 0<z<3.5. Most of the new spectra were obtained with the FORS instruments at the ESO VLT in the course of a photometric and spectroscopic study of distant galaxies in the FORS Deep Field (FDF) (Appenzeller et al. 2000; Heidt et al. 2001; Bender et al. 2001). Due to its combination of depth and a (compared to the HDFs) relatively large area the FDF is particularly well suited for statistical studies of high-redshift galaxies. In the present investigation we restrict ourselves to redshifts $z \leq 3.5$ since at higher redshifts the position of the redshifted C  IV resonance lines tends to coincide with strong OH night sky lines. Hence, for an accurate sky subtraction a higher spectral resolution or longer exposure times than we could achieve so far would be needed. In Sect. 2 we describe the sample selection and the observations. In Sect. 3 we present and discuss the measurement of the C  IV (and Si  IV) equivalent widths, from which we estimate the metallicities of the investigated objects in Sect. 4. In Sect. 5 we compare our results with data available in the literature, in Sect. 6 we draw our conclusions.