2016A&A...586A..90P

Despite the thorough work of van Leeuwen, the parallax of α Centauri is still far from being carved in stone. Any derivation of the individual masses is therefore uncertain, if not questionable. And yet, that does not prevent this system from being used for calibration purpose in several studies. We aim for more accurate model-free parallax and individual masses of this system. With HARPS, the radial velocities are not only precise but also accurate. Ten years of HARPS data are enough to derive the complement of the visual orbit for a full 3D orbit of α Cen. We locate α Cen (743mas) right where Hipparcos had originally put it, i.e. slightly farther away than derived by Soederhjelm. The components are thus a bit more massive than previously thought (1.13 and 0.97M_☉ for A and B, respectively). These values are now in excellent agreement with the latest asteroseismologic results.