1994A&A...286..701K

We present ultra-deep B, R and I photometry in A370 and in a reference blank field, up to a completeness limit B≃26.5. We discuss the photometry at such faint levels, define criteria to select a sample of 33 arclet candidates in the range 24<B<27 and analyze their photometric data. The distribution of B magnitudes and color indices are found to be similar for arclets and for faint galaxies in the blank field. We also give the main geometrical parameters of arclets, such as the position, axis ratio and position angle, useful for any lens modeling or statistical analysis in the framework of gravitational lenses. We propose a method to determine the distance of the background population of the galaxies, sources of gravitational arcs and arclets, as faint as B=27, without any spectroscopy. The method relies on our recent modeling of the gravitational potential in A370 based on deep, high resolution imaging in the cluster center, showing that the giant arc consists of three merging images and that three faint pairs are likely to be multiply imaged galaxies (Kneib et al. 1993). Making use of properties of the ellipticity distribution of the sources and of the redshift conditional probability, we estimate the most probable redshift of each arclet candidate. Major uncertainties are reviewed and a comparison is made with color redshifts computed from Bruzual's evolutionary code. Although there remain uncertainties in the application of the method in the cluster A370, the main conclusions are that 2/3 of the galaxies with intrinsic magnitudes ranging from B=24 to 27 lie at redshift z<1, in excellent agreement with the extrapolation of recent faint spectroscopic surveys, and that a no-evolution model of galaxies is favoured from the resulting redshift-magnitude plot.