Accurate masses for dispersion-supported galaxies

We derive an accurate mass estimator for dispersion-supported stellar systems and demonstrate its validity by analysing resolved line-of-sight velocity data for globular clusters, dwarf galaxies and elliptical galaxies. Specifically, by manipulating the spherical Jeans equation we show that the mass enclosed within the 3D deprojected half-light radius no can be determined with only mild assumptions about the spatial variation of the stellar velocity dispersion anisotropy as long as the projected velocity dispersion profile is fairly flat near the half-light radius, as is typically observed. We find M-1/2 = 3G(-1) <sigma(2)(los)> r(1/2) similar or equal to 4G(-1) <sigma(2)(los)> R-e, where <sigma(2)(los)> is the luminosity-weighted square of the line-of-sight velocity dispersion and Re is the 2D projected half-light radius. While deceptively familiar in form, this formula is not the virial theorem, which cannot be used to determine accurate masses unless the radial profile of the total mass is known a priori. We utilize this finding to show that all of the Milky Way dwarf spheroidal galaxies (MW dSphs) are consistent with having formed within a halo of a mass of approximately 3 x 10(9) M-circle dot, assuming a A cold dark matter cosmology. The faintest MW dSphs seem to have formed in dark matter haloes that are at least as massive as those of the brightest MW dSphs, despite the almost five orders of magnitude spread in luminosity between them. We expand our analysis to the full range of observed dispersion-supported stellar systems and examine their dynamical I-band mass-to-light ratios gamma(I)(1/2). The gamma(I)(1/2) versus M-1/2 relation for dispersion-supported galaxies follows a U shape, with a broad minimum near gamma(I)(1/2) similar or equal to 3 that spans dwarf elliptical galaxies to normal ellipticals, a steep rise to "Ito 3200 for ultra-faint dSphs and a more shallow rise to gamma(I)(1/2) similar or equal to 800 for galaxy cluster spheroids.