We present the results of our analyses of the X-ray emission and of the strong lensing systems in the galaxy cluster Abell 611 [z = 0.288]. This cluster is an optimal candidate for a comparison of the mass reconstructions obtained through X-ray and lensing techniques, because of its very relaxed dynamical appearance and its exceptional strong lensing system. We infer the X-ray mass estimate deriving the density and temperature profile of the intra-cluster medium within the radius r similar or equal to 700 kpc through a non-parametric approach, taking advantage of the high spatial resolution of a Chandra observation. Assuming that the cluster is in hydrostatic equilibrium and adopting a matter density profile, we can recover the total mass distribution of Abell 611 via the X-ray data. Moreover, we derive the total projected mass in the central regions of Abell 611 performing a parametric analysis of its strong lensing features through the publicly available analysis software Lenstool. As a final step we compare the results obtained with both methods. We derive a good agreement between the X-ray and strong lensing total mass estimates in the central regions where the strong lensing constraints are present (i.e. within the radius r similar or equal to 100 kpc), while a marginal disagreement is found between the two mass estimates when extrapolating the strong lensing results in the outer spatial range. We suggest that in this case the X-ray/strong lensing mass disagreement can be explained by an incorrect estimate of the relative contributions of the baryonic component and of the dark matter, caused by the intrinsic degeneracy between the different mass components in the strong lensing analysis. We discuss the effect of some possible systematic errors that influence both mass estimates. We find a slight dependence of the measurements of the X-ray temperatures (and therefore of the X-ray total masses) on the background adopted in the spectral analysis, with total deviations on the value of M-200 of the order of the 1 sigma statistical error. The strong lensing mass results are instead sensitive to the parameterisation of the galactic halo mass in the central regions, in particular to the modelling of the brightest cluster galaxy (BCG) baryonic component, which induces a significant scatter in the strong lensing mass results.

SL2SJ02176-0513 is a remarkable lens for the presence of two multiply-imaged systems at different redshifts lensed by a foreground massive galaxy at z(lens) = 0.656: a bright cusp arc at z(arc) = 1.847 and an additional double-image system at an estimated redshift of z(dbl) similar to 2.9 based on photometry and lensing geometry. The system is located about 400 kpc away from the centre of a massive group of galaxies. Mass estimates for the group are available from X-ray observations and satellite kinematics. Multicolour photometry provides an estimate of the stellar mass of the main lens galaxy. The lensing galaxy is modelled with two components (stars and dark matter), and we include the perturbing effect of the group environment and all available constraints. We find that classic lensing degeneracies, are significantly reduced with respect to standard systems and infer tight constraints on the mass density profile: (i) the dark matter content of the main lens galaxy is in line with that of typical galaxies f(dm)(< R-e) = 0.41(-0.06)(+0.09). (ii) the required mass associated with the dark matter halo of the nearby group is consistent with X-ray (sigma(grn) = 550(-240)(+130)); (iii) accounting for the group contribution in the form of an external convergence, the slope of the mass density profile of the main lens galaxy alone is found to be alpha = -1.03(-0.16)(+0.22), consistent with the isothermal (alpha = -1) slope.

Context. The new generation of wide-field optical imaging as the Canada France Hawaii Telescope Legacy Survey (CFHTLS) enables discoveries of all types of gravitational lenses present in the sky. The Strong Lensing Legacy Survey (SL2S) project has started an inventory of clusters or groups of galaxies lenses and of Einstein rings around distant massive ellipticals. Aims. Here we attempt to extend this inventory by finding lensing events produced by massive edge-on disk galaxies that remain a poorly documented class of lenses. Methods. We implemented and tested an automated search procedure of edge-on galaxy lenses in the CFHTLS Wide fields with magnitude 18 < i < 21, inclination angle lower than 25 ◦ , and a photometric redshift determination. This procedure estimated the lensing convergence of each galaxy from the Tully-Fisher law and selected only those few candidates that exhibit a possibly nearby arc configuration at a radius compatible with this convergence (rarc 2rE). The efficiency of the procedure was tested after a visual examination of the whole initial sample of 30 444 individual edge-on disks. Results. After calculating the surface density of edge-on lenses possibly detected in a survey for a given seeing, we deduce that this theoretical number is about 10 for the CFHTLS Wide, a number in broad agreement with the 2 good candidates detected here. We show that the Tully-Fisher selection method is very efficient at finding valuable candidates, though its accuracy depends on the quality of the photometric redshift of the lenses. Finally, we argue that future surveys will detect at least a hundred of such lens candidates.

Context. This paper reports the results obtained on the photometric redshifts measurement and accuracy, and cluster tomography in the ESO Distant Cluster Survey (EDisCS) fields.Aims. We present the methods used to determine photometric redshifts to discriminate between member and non-member galaxies and reduce the contamination by faint stars in subsequent spectroscopic studies.Methods. Photometric redshifts were computed using two independent codes both based on standard spectral energy distribution (SED) fitting methods (Hyperz and Rudnick's code). Simulations were used to determine the redshift regions for which a reliable determination of photometric redshifts was expected. The accuracy of the photometric redshifts was assessed by comparing our estimates with the spectroscopic redshifts of similar to 1400 galaxies in the 0.3 <= z <= 1.0 domain. The accuracy expected for galaxies fainter than the spectroscopic control sample was estimated using a degraded version of the photometric catalog for the spectroscopic sample.Results. The accuracy of photometric redshifts is typically sigma(Delta z/(1 + z)) similar to 0.05 +/- 0.01, depending on the field, the filter set, and the spectral type of the galaxies. The quality of the photometric redshifts degrades by a factor of two in sigma(Delta z/(1 + z)) between the brightest (I less than or similar to 22) and the faintest (I similar to 24-24.5) galaxies in the EDisCS sample. The photometric determination of cluster redshifts in the EDisCS fields using a simple algorithm based on z(phot) is in excellent agreement with the spectroscopic values, such that delta z similar to 0.03-0.04 in the high-z sample and delta z similar to 0.05 in the low-z sample, i.e. the z(phot) cluster redshifts are at least a factor similar to(1 + z) more accurate than the measurements of z(phot) for individual galaxies. We also developed a method that uses both photometric redshift codes jointly to reject interlopers at magnitudes fainter than the spectroscopic limit. When applied to the spectroscopic sample, this method rejects similar to 50-90% of all spectroscopically confirmed non-members, while retaining greater than or similar to 90% of all confirmed members.Conclusions. Photometric redshifts are found to be particularly useful for the identification and study of clusters of galaxies in large surveys. They enable efficient and complete pre-selection of cluster members for spectroscopy, allow accurate determinations of the cluster redshifts based on photometry alone, and provide a means of determining cluster membership, especially for bright sources.

Context. The Strong Lensing Legacy Survey (SL2S) provides a sample of the strong lensing events associated with massive distant galaxies, some of which lie in the outskirts of galaxy groups and clusters.Aims. We investigate the internal structure of early-type galaxies in overdense environments, where tidal forces are expected to alter dark matter haloes of infalling galaxies.Methods. SL2SJ02176-0513 is a remarkable lens for the presence of two multiply-imaged systems at different redshifts lensed by a foreground massive galaxy at z(lens) = 0.656: a bright cusp arc at z(arc) = 1.847 and an additional double-image system at an estimated redshift of z(dbl) similar to 2.9 based on photometry and lensing geometry. The system is located about 400 kpc away from the centre of a massive group of galaxies. Mass estimates for the group are available from X-ray observations and satellite kinematics. Multicolour photometry provides an estimate of the stellar mass of the main lens galaxy. The lensing galaxy is modelled with two components (stars and dark matter), and we include the perturbing effect of the group environment and all available constraints.Results. We find that classic lensing degeneracies, e. g. between external convergence and mass density slope, are significantly reduced with respect to standard systems and infer tight constraints on the mass density profile: (i) the dark matter content of the main lens galaxy is in line with that of typical galaxies f(dm)(<R-e) = 0.41(-0.06)(+0.09); (ii) the required mass associated with the dark matter halo of the nearby group is consistent with X-ray (sigma(grp) = 550(-240)(+130) ); (iii) accounting for the group contribution in the form of an external convergence, the slope of the mass density profile of the main lens galaxy alone is found to be alpha = -1.03(-0.16)(+0.22) , consistent with the isothermal (alpha = -1) slope.Conclusions. We demonstrate that multiple source plane systems together with good ancillary dataset can be used to disentangle local and environmental effects.

The existence of strong lensing systems with Einstein radii covering the full mass spectrum, from ˜ 1-2 arcsec (produced by galaxy scale dark matter haloes) to >10 arcsec (produced by galaxy cluster scale haloes) have long been predicted. Many lenses with Einstein radii around 1-2 arcsec and above 10 arcsec have been reported but very few in between. In this article, we present a sample of 13 strong lensing systems with Einstein radii in the range 3 arcsec-8 arcsec (or image separations in the range 6 arcsec-16 arcsec), i.e. systems produced by galaxy group scale dark matter haloes. This group sample spans a redshift range from 0.3 to 0.8. This opens a new window of exploration in the mass spectrum, around 1013-1014 Ms, a crucial range for understanding the transition between galaxies and galaxy clusters, and a range that have not been extensively probed with lensing techniques. These systems constitute a subsample of the Strong Lensing Legacy Survey (SL2S), which aims to discover strong lensing systems in the Canada France Hawaii Telescope Legacy Survey (CFHTLS). The sample is based on a search over 100 square degrees, implying a number density of ~0.13 groups per square degree. Our analysis is based on multi-colour CFHTLS images complemented with Hubble Space Telescope imaging and ground based spectroscopy. Large scale properties are derived from both the light distribution of elliptical galaxies group members and weak lensing of the faint background galaxy population. On small scales, the strong lensing analysis yields Einstein radii between 2.5 arcsec and 8 arcsec. On larger scales, strong lens centres coincide with peaks of light distribution, suggesting that light traces mass. Most of the luminosity maps have complicated shapes, implying that these intermediate mass structures may be dynamically young. A weak lensing signal is detected for 6 groups and upper limits are provided for 6 others. Fitting the reduced shear with a Singular Isothermal Sphere, we find σ_SIS ˜ 500 km s-1 with large error bars and an upper limit of ~900 km s-1 for the whole sample (except for the highest redshift structure whose velocity dispersion is consistent with that of a galaxy cluster). The mass-to-light ratio for the sample is found to be M/Li ~ 250 (solar units, corrected for evolution), with an upper limit of 500. This compares with mass-to-light ratios of small groups (with σ_SIS ˜ 300 km s-1) and galaxy clusters (with σ_SIS > 1000 km s-1), thus bridging the gap between these mass scales. The group sample released in this paper will be complemented with other observations, providing a unique sample to study this important intermediate mass range in further detail. Appendix A is only available in electronic form via http://www.aanda.org Based on observations obtained with MegaPrime/MegaCam, a joint project of CFHT and CEA/DAPNIA, at the Canada-France-Hawaii Telescope (CFHT) which is operated by the National Research Council (NRC) of Canada, the Institut National des Sciences de l'Univers of the Centre National de la Recherche Scientifique (CNRS) of France, and the University of Hawaii. This work is based in part on data products produced at TERAPIX and the Canadian Astronomy Data Centre as part of the Canada-France-Hawaii Telescope Legacy Survey, a collaborative project of NRC and CNRS. Based on observations made with the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. These observations are associated with programs 10876 and 11289.

We present a reconstruction of the mass distribution of galaxy cluster Abell 1689 at z = 0.18 using detected strong lensing features from deep acs observations and extensive ground based spectroscopy. Earlier analyses have reported up to 32 multiply imaged systems in this cluster, of which only 3 were spectroscopically confirmed. In this work, we present a parametric strong lensing mass reconstruction using 24 multiply imaged systems with newly determined spectroscopic redshifts, which is a major step forward in building a robust mass model. In turn, the new spectroscopic data allows a more secure identification of multiply imaged systems. The resultant mass model enables us to reliably predict the redshifts of additional multiply imaged systems for which no spectra are currently available, and to use the location of these systems to further constrain the mass model. In particular, we have detected 5 strong galaxy-galaxy lensing systems just outside the Einstein ring region, further constraining the mass profile. Our strong lensing mass model is consistent with that inferred from our large scale weak lensing analysis derived using cfh12k wide field images. Thanks to a new method for reliably selecting a well defined background lensed galaxy population, we resolve the discrepancy found between the strong and weak lensing mass models reported in earlier work. The derived parameters for the best fit nfw model profile is found to be c200 = 7.6 ± 1.6 and r200 = 2.16 ± 0.10 h −1 70 Mpc (corresponding to a 3D mass equal to M200 = 1.32 ± 0.2 × 10 15 h70 M⊙). We find that the projected mass enclosed within the Einstein radius for Abell 1689 is M (45) = 1.91±0.27×10 Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, 2100 Copenhagen, Denmark; marceau@dark-cosmology.dk Department of Astronomy, California Institute of Technology, 105-24, Pasadena, CA91125, USA OAMP, Laboratoire d’Astrophysique de Marseille UMR 6110 Traverse du siphon, 13012 Marseille, France European Southern Observatory, Alonso de Cordova, Santiago, Chile Institut d’Astrophysique de Paris, Bvd Arago, 75014, Paris, France Laboratoire d’Astrophysique de Toulouse-Tarbes, CNRS-UMR 5572 & Université Paul Sabatier Toulouse III, 14 Avenue Edouard Belin, 31400 Toulouse, France Astronomy Department, Yale University, P.O. Box 208101, New Haven, CT 06520-8101, USA Department of Physics, Yale University, P.O. Box 208101, New Haven, CT 06520-8101, USA Institute for Computational Cosmology, Durham University, South Road, Durham DH1 3LE, UK Kapteyn Astronomical Institute, P.O.Box 800, 9700 AV Groningen, Netherlands Argelander-Institut für Astronomie, Universität Bonn, Auf dem Hügel 71, 53121 Bonn, Germany L3AB UMR 5804 2, rue de l’Observatoire, BP 89, 33270 Floirac, France Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822, USA

AIMS: We present data from the CFHTLS Strong Lensing Legacy Survey (SL2S). Due to the unsurpassed combined depth, area and image quality of the Canada-France-Hawaii Legacy Survey it is becoming possible to uncover a large, statistically well-defined sample of strong gravitational lenses which spans the dark halo mass spectrum predicted by the concordance model from galaxy to cluster haloes. METHODS: We describe the development of several automated procedures to find strong lenses of various mass regimes in CFHTLS images. RESULTS: The preliminary sample of about 40 strong lensing candidates discovered in the CFHTLS T0002 release, covering an effective field of view of 28 deg$^2$ is presented. These strong lensing systems were discovered using an automated search and consist mainly of gravitational arc systems with splitting angles between 2 and 15 arcsec. This sample shows for the first time that it is possible to uncover a large population of strong lenses from galaxy groups with typical halo masses of about $10^{13}h^{-1}M_\odot$. We discuss the future evolution of the SL2S project and its main scientific aims for the next 3 years, in particular our observational strategy to extract the hundreds of gravitational rings also present in these fields.

Aims.We present data from the CFHTLS Strong Lensing Legacy Survey (SL2S). Due to the unsurpassed combined depth, area and image quality of the Canada-France-Hawaii Legacy Survey it is becoming possible to uncover a large, statistically well-defined sample of strong gravitational lenses which spans the dark halo mass spectrum predicted by the concordance model from galaxy to cluster haloes.

Caustics are high-density structures that form in collisionless media. Under self-gravity, cold dark matter flows focus onto caustics which are yet to be resolved in numerical simulations and observed in the real world. If detected, caustics would provide strong evidence for dark matter and would rule out alternative models such as those with modified dynamics. Here, we demonstrate how they might be observed in weak lensing data. We evaluate the shear distortion and show that its radial profile is marked by a characteristic sawtooth pattern due to the caustics in dark matter haloes that form by selfsimilar accretion. We discuss the observational complications, mainly due to the poor knowledge of the virial radii of the haloes and demonstrate that a superposition of about 600 cluster-size haloes would give a signal-to-noise ratio which is sufficiently large for the detection of caustics with ground-based observations. This number is reduced to 200 for space-based observations. These bounds can be easily achieved by the ongoing wide field optical surveys such as CFHTLS and the future space-based projects SNAP and DUNE which have to be accompanied by an X-ray follow-up of the selected clusters for a precise determination of their virial radius.

Caustics are formally singular structures, with infinite density, that form in collisionless media. The non-negligible velocity dispersion of dark matter particles renders their density finite. We evaluate the maximum density of the caustics within the framework of secondary infall model of formation of dark matter haloes. The result is then used to demonstrate that caustics can be probed by properly stacking the weak-lensing signal of about 600 haloes. CFHTLS accompanied by X-ray observations and the space-based experiments like SNAP or DUNE can provide us with the required statistics and hence a way of distinguishing between the viable dark matter particle candidates. The extension of our results to more realistic models including the effects of mergers of haloes is briefly outlined.

Observations of the core of the massive cluster Cl 0024+1654, at a redshift z=0.39, were obtained with the Infrared Space Observatory using ISOCAM at 6.7 mum and 14.3 mum (hereafter 15 mum). Thirty five sources were detected at 15 mum and thirteen of them are spectroscopically identified with cluster galaxies. The remaining sources consist of four stars, one quasar, one foreground galaxy, three background galaxies and thirteen sources with unknown redshift. The ISOCAM sources have best-fit SEDs typical of spiral or starburst models observed 1 Gyr after the main starburst event. The median infrared luminosity of the twelve cluster galaxies is 1.0x10^11 Lsun, with 10 having infrared luminosity above 9x10^10 Lsun, and so lying near or above the 1x10^11 Lsun threshold for identification as a luminous infrared galaxy (LIRG). The [OII] star formation rates obtained for 3 cluster galaxies are one to two orders of magnitude lower than the infrared values, implying that most of the star formation is missed in the optical because it is enshrouded by dust in the starburst galaxy. The counterparts of about half of the 15 mum cluster sources are blue, luminous, star-forming systems and the type of galaxy that is usually associated with the Butcher-Oemler effect. Dust obscuration may be a major cause of the 15 mum sources appearing on the cluster main sequence. The majority of the ISOCAM sources in the Butcher-Oemler region of the colour-magnitude diagram are best fit by spiral-type SEDs whereas post-starburst models are preferred on the main sequence, with the starburst event probably triggered by interaction with one or more galaxies. Finally, the mid-infrared results on Cl 0024+1654 are compared with four other clusters observed with ISOCAM.

We present the ESO Distant Cluster Survey (EDisCS), a survey of 20 fields containing distant galaxy clusters with redshifts ranging from 0.4 to almost 1.0. Candidate clusters were chosen from among the brightest objects identified in the Las Campanas Distant Cluster Survey, half with estimated redshift z(est) similar to 0.5 and half with z(est) similar to 0.8. They were confirmed by identifying red sequences in moderately deep two colour data from VLT/FORS2. For confirmed candidates we have assembled deep three-band optical photometry using VLT/FORS2, deep near-infrared photometry in one or two bands using NTT/SOFI, deep optical spectroscopy using VLT/FORS2, wide field imaging in two or three bands using the ESO Wide Field Imager, and HST/ACS mosaic images for 10 of the most distant clusters. This first paper presents our sample and the VLT photometry we have obtained. We present images, colour-magnitude diagrams and richness estimates for our clusters, as well as giving redshifts and positions for the brightest cluster members. Subsequent papers will present our infrared photometry, spectroscopy, HST and wide-field imaging, as well as a wealth of further analysis and science results. Our reduced data become publicly available as these papers are accepted.

We investigate the supercluster MS0302+17 (zapproximate to0.42) using weak lensing analysis and deep wide field BVR photometry with the CFH12K camera. Using (B-V) vs. (V-R) evolution tracks we identify early-type members of the supercluster, and foreground ellipticals. We derive a R band catalogue of background galaxies for weak lensing analysis. We compute the correlation functions of light and mass and their cross-correlation and test if light traces mass on supercluster, cluster and galaxy scales.We show that the data are consistent with this assertion. The zeta-statistic applied in regions close to cluster centers and global correlation analyses over the whole field converge toward the simple relation M/L=300+/-30 h(70) (M/L)(circle dot) in the B band. This independently confirms the earlier results obtained by Kaiser et al. (1998).If we model dark matter halos around each early-type galaxy by a truncated isothermal sphere, we find that a linear relation M proportional to L still holds. In this case, the average halo truncation radius is s(*) less than or similar to200 h(70)(-1) kpc close to clusters cores whereas it reaches a lower limit of similar to300 h(70)(-1) kpc at the periphery. This change of s(*) as a function of radial distance may be interpreted as a result of tidal stripping of early type galaxies. Nevertheless the lack of information on the spatial distribution of late-type galaxies affects such conclusions concerning variations of s(*).Though all the data at hand are clearly consistent with the assumption that mass is faithfully traced by light from early-type galaxies, we are not able to describe in detail the contribution of late type galaxies. We however found it to be small. Forthcoming wide surveys in UV, visible, and near infrared wavelengths will provide large enough samples to extend this analysis to late-type galaxies using photometric redshifts.

We reanalyze the strong lens modeling of the cluster of galaxies MS 2137.3-2353 using a new UBVRIJK data set obtained with the ESO Very Large Telescope. We infer the photometric redshifts of the two main arc systems which are both found to be at z=1.6+/- 0.1. After subtraction of the central cD star light in the previous F702/HST imaging we found only one object lying underneath. This object has the expected properties of the fifth image associated to the tangential arc. It lies at the right location, shows the right orientation and has the expected signal-to-noise ratio. We improve the previous lens modelings of the central dark matter distribution of the cluster, using two density profiles: an isothermal model with a core, and the NFW-like model with a cusp. Without the fifth image, the arc properties together with the shear map profile are equally well fit by the isothermal model and by a sub-class of generalized-NFW mass profiles having inner slope power index in the range 0.7〈= alpha 〈= 1.2. Adding new constrains on the center lens position provided by the fifth image favors isothermal profiles that better predict the fifth image properties. A detailed model including nearby cluster galaxy perturbations or the effect of the stellar mass distribution to the total mass inward does not change our conclusions but imposes the M/LI of the cD stellar component is below 10 at a 99% confidence level. Using our new detailed strong+weak lensing model together with Chandra X-ray data and the cD stellar component we finally discuss intrinsic properties of the gravitational potential. Whereas X-ray and dark matter have a similar orientation and ellipticity at various radius, the cD stellar isophotes are twisted by 13o +/- 3o. The sub-arc-second azimuthal shift we observe between the radial arc position and the predictions of elliptical models correspond to what is expected from a mass distribution twist. This shift may result from a projection effect of the cD and the cluster halos, thus revealing the triaxiality of the mass components. Based on observations obtained at the Very Large Telescope (VLT) at Cerro Paranal operated by European Southern Observatory.

Back on several on-sky tests using the Multi-Object Spectrography at the Canada-France-Hawaii Telescope, we present refinements of the faint objects optical spectroscopy technique we proposed in 1994 (initially names "Va-et-Vient" = "Back-and-Forth"), a nod-and-shuttle technique that preserves a sky dominated photon noise regime. We validated schemes that allow 100% of the total telescope time used to integrate the light from the scientific objects. We also investigated how to optimize the configuration of the charge sifting in order to use minimal space on the detector. For typical distant galaxy fields (profile dominated by seeing), we demonstrated an increase by a factor of approximately 2 of the multiplex gain and even higher for more compact sources. The technique proves perfectly apt at eliminating all the systemic errors that may cause the saturation of the signal-to-noise ratio: slit defects, CCD fringing, and flat-field residuals. This technique is also the most efficient and unique approach to use curved slits.

We reanalyze the strong lens modeling of the cluster of galaxies MS2137-23 using a new data set obtained with the ESO VLT. We found the photometric redshifts of the two main arc systems are both at z=1.6. After subtraction of the central cD star light of the HST image we found that only one object lying underneath has the expected properties of the fifth image associated to the tangential arc. We improve the previous lens modelings of the central dark matter distribution of the cluster, using an isothermal model with a core (IS) and the NFW-like model with a cusp. Without the fifth image, the arc properties together with the shear map profile are equally well fit by the and by an IS and a sub-class of generalized-NFW mass profiles having inner slope power index in the range 0.7<alpha<1.2. Adding new constrains provided by the fifth image favors IS profiles that better predict the fifth image properties. A model including cluster galaxy perturbations or the the stellar mass distribution does not change our conclusions but imposes the M/L_I of the cD stellar component is below 10 at a 99 Using our new detailed lensing model together with Chandra X-ray data and the cD stellar component we finally discuss intrinsic properties of the gravitational potential. Whereas X-ray and dark matter have a similar shape at various radius, the cD stellar isophotes are twisted by 13 deg. The sub- arc-second azimuthal shift we observe between the radial arc position and the predictions of elliptical models correspond to what is expected from a mass distribution twist. This shift may result from a projection effect of the cD and the cluster halos, thus revealing the triaxiality of the system.