Probing the High Redshift Universe with the faintest dwarfs

Ultra faint dwarf spheroidal galaxies (UFs) are the least luminous and the least metal-rich stellar systems ever known (L <10(3-5) Ls un, [F e/H] <-2). The Fe-Luminosity relation derived for UFs constitutes a natural extension towards lower metallicity of that of "classical" (L > 105Ls un) dwarf spheroidal galaxies (dSphs), suggesting that the formation of all these Milky Way (MW) satellites has been governed by the same physical processes. However while classical dSphs and UFs together span more than four orders of magnitude in luminosity, their total mass is roughly the same within the innermost 300 pc. Why is then the star formation so inefficient in UFs? Moreover recent observations of metal-poor stars in dSphs have revealed that [F e/H] <-3 stars represent the 25% of the total stellar population in UFs, while they are extremely rare in classical dSphs. What does this observation imply? Are these galaxies among the first star-forming objects in the MW system? We investigate the nature of UFs in a general cosmological context, simultaneously accounting for various classical dSphs and MW properties, including their Metallicity Distribution Function (MDF). The model successfully reproduces the observed Fe-Luminosity relation and the mean MDF of UFs. We found that UFs are left-overs of H2-cooling minihaloes; they are the oldest dSphs in the MW systems formed at z > 8.5, i.e. before the reionization epoch