The Strong lensing legacy Survey Bernard Fort (IAP, Paris) Thursday February 15th, 11am It has been known for many years that the sky shall display a huge number of gravitational mirages, which occur when the center of a very compact mass condensation (the lense) is almost perfectly aligned along the line of sight of a more distant galaxy (the source). This spectacular effect produces multiple images of a same source when the matter distribution reaches a critical projected mass density. It corresponds to a light deflection regime called strong lensing, first recognized through bright multiple QSOs/radiosources because, even if QSOs are rare background sources, they are also relatively easy to identify on the sky. Although giant arcs located around the center of several X-ray clusters were rapidly recognized as strong cluster-galaxy lens, the most frequent galaxy-galaxy strong lenses are mostly undiscovered because apart from a few (often serendipity) HST survey, we have almost no direct way to find them systematically. In this presentation, I will show that the combined depth, area and image quality of the CFHT-Legacy Survey (170 sq. degree) allow us to constitute a database of about 200 lenses, from gravitational rings around massive galaxies to giant arcs. This Strong Lensing Legacy Survey (SL2S) project needs to use automated search procedures to search for very distant lenses with an average redshift around z=0.6 and will uncover a new population of lens masses bridging the gap between the galaxy and cluster mass. As the geometrical characteristics of the lensed images configuration only depend on the local potential gradient and angular distances between the observer, the lens and the source, the SL2S database can be used to study the evolution of dark matter halo profile with cosmic time, to study highly-magnified distant galaxies and possibly with future large survey, to implement new cosmological tests on the evolution of the geometry of the universe. After a rapid presentation of the strong lensing phenomena and an estimation of the number of expected lenses in the CFHTLS fields, I will show how we detect the lenses, as well as our future plans to improve the success rate of the detection procedure and to contribute to the development of this topic in the perspective of future projects like SNAP or DUNE.