1 Introduction

The analysis of the distribution of dark matter in clusters of galaxies provides an important insight into the history of structure formation in the Universe. Their epoch of formation and their evolution with redshift are dependent on cosmological parameters and the nature of (dark) matter. In particular, it is widely believed that the existence of even a few massive clusters at redshift $z \approx 1$ will be a strong indicator of a low mass density universe (e.g. Bahcall et al. 1997).

It is essential that we develop reliable tools to probe the amount and distribution of each matter component and follow their evolution with redshift. Gravitational lensing and bremsstrahlung emission from hot intra-cluster gas are two processes which help us in addressing these issues. Unfortunately, the results from these two approaches have not always been consistent with each other.

Indeed, X-ray mass estimates show discrepancies with weak- and strong-lensing mass estimates of clusters of galaxies. The reasons for the discrepancy are not yet fully understood (see Mellier 1999 for a review). The total mass inferred from lensing exceeds the X-ray mass by a factor of about two for some clusters including well studied ones like A2218 (Miralda-Escudé & Babul 1995). Investigations of a dozen clusters by Smail et al. (1997), Allen (1998) and Lewis et al. (1999) have not provided conclusive answers. Allen (1998) compared cooling flow and non-cooling flow clusters and observed that the former do not show the mass discrepancy. This result suggests that the assumptions regarding the dynamical and thermal state of the hot intra-cluster gas, a key ingredient for the X-ray mass estimate, are not realistic enough for a satisfactory model of non-cooling flow (i.e. presumably non-relaxed) clusters of galaxies. This interpretation was reinforced by Böhringer et al. (2000) and Allen et al. (2001) who found an excellent agreement between the X-ray data from ROSAT and Chandra and the strong- and weak-lensing analyses of the relaxed cluster A2390. However, Lewis et al. (1999) found significant discrepancies even in some cooling flow clusters between the X-ray and lensing mass, particularly with strong-lensing estimates.

It is likely that this mass discrepancy is the result of several less-than-valid assumptions. For example, the comparison between X-ray and weak gravitational lensing is done by extrapolating the best fit of the X-ray profile far beyond the region where data are reliable, where uncertainties are obviously significant and the shape of the (assumed) analytical profile used for extrapolation has a considerable impact on the mass estimate (Lewis et al. 1999; Böhringer et al. 1999).

Lensing mass estimates are not free from bias either. N-body simulations by Cen (1997) and Metzler et al. (1999) show that projection effects of in-falling filaments of matter towards the cluster centre can significantly bias the projected mass density inferred from weak lensing analysis to values higher than those derived from X-ray. Estimates of the bias are typically between 10 and 20 per cent but projection effects due to structures along the line of sight can overestimate the total mass by about 30 per cent (Reblinsky & Bartelmann 1999). It is therefore important to improve the accuracy of weak-lensing mass estimates by minimising systematic errors in particular. This will help us in identifying the assumptions in the X-ray analysis which are responsible for the mass discrepancy as opposed to errors which have their origins in the lensing analysis.

Very deep observations of clusters of galaxies in multiple bands and with sub-arcsecond seeing can considerably improve the reliability of mass estimates from weak lensing; the depth increases the number density of lensed galaxies thereby improving the resolution of the mass reconstruction; multicolour observations allow estimation of photometric redshifts and hence the redshift distribution of background sources; finally, sub-arcsecond seeing makes for a better determination of object shapes and accurate PSF correction. Rarely are all of these stringent requirements satisfied simultaneously in ground based observations. Fortunately, the observations obtained during the Science Verification Programme for the FORS1 (FOcal Reducer/low dispersion Spectrograph; Appenzeller et al. 1998) and the ISAAC (Infrared Spectrometer and Array Camera; Moorwood et al. 1999) instruments mounted on the first VLT unit, UT1/ANTU, at Paranal provide an excellent dataset on the lensing cluster MS 1008-1224. The quality (seeing and depth in 6 bands) of these images are among the best data ever obtained from the ground for weak lensing mass reconstruction of a cluster.

MS 1008-1224 is a galaxy cluster selected from the Einstein Medium Sensitivity Survey (Gioia & Luppino 1994). It is one of the 16 EMSS clusters observed by Le Fèvre et al. (1994) in which they found strong-lensing features. The cluster is at redshift z=0.3062 (Lewis et al. 1999) and is dominated by a cD galaxy. The X-ray luminosity is $L_{\rm X}$(0.3-3.5 keV $)=4.5 \times 10^{44}$ ergs^-1 (From Lewis et al. 1999, with H₀=100 km s^-1 Mpc^-1 and q₀=0.1) and its temperature inferred from ASCA observation is $T_{\rm X}=7.29$ keV (Mushotzky & Scharf 1997). According to Lewis et al. (1999), the X-ray contours are circular and are centered 15 arcsec to the north of the cD and also show an extension towards the north.

The paper is organized as follows: Sect. 2 details the optical properties of the VLT images. Section 3 deals in some detail with the photometric redshift estimation of galaxies in the field of MS 1008-1224. The mass reconstructions, from weak shear analysis as well as from depletion produced by magnification bias, are presented in Sect. 4. The results are discussed in Sect. 5. Finally, a summary of this study is provided in Sect. 6.

In this paper we have used H₀=100 h^-1 km s^-1 Mpc^-1, $\Omega_0=0.3$, $\Lambda = 0$. This corres- ponds to a scale of 1 arcmin = 176 h^-1 kpc at the redshift of the cluster.