The knowledge of the masses of small groups of galaxies is important, because the timescales for collapse and virialization of groups depend on their mass-density. While the large inefficiency of the virial mass statistic is well known (Bahcall and Tremaine 1981, hereafter BT, and references therein), biases in the virial mass may produce wrong evolution timescales for small groups. These biases originate from group contamination by interlopers, and from the different galaxy and dark matter distributions inside the groups (caused by mass segregation). This second bias was first studied by Smith (1980, 1984) although already implicit in the work of Limber (1959). It formally arises because the ratio 2T/C where C = ΣFα·Rα (the Clausius virial) is not the same for the luminous and global matter distributions. We illustrate here some quantitative aspects of this Limber bias from the output of N-body simulations of groups of 8 galaxies described elsewhere (Mamon 1986).