THE VELOCITY DISPERSION ANISOTROPY AND MASS-TO-LIGHT RATIO OF ELLIPTIC GALAXIES

Axisymmetric dynamical models are constructed for 37 bright elliptical galaxies for which high-quality photometrical, and both major and minor axis kinematical data are available in the literature. The models are of the type used previously by Binney, Davies & Illingworth and van der Marel, Binney & Davies. The projected kinematics are predicted from the observed surface photometry, assuming a constant mass-to-light ratio, and a velocity ellipsoid with sigma-r = sigma-theta [i.e. f = f(E,L(z)]. For the sample as a whole it is found that these models tend to predict too much motion on the major axis. This implies that elliptical galaxies as a class must have sigma-r > sigma-theta, which has not been demonstrated before. This result is consistent with general expectation as based on N-body simulations of dissipationless collapse. From the models accurate mass-to-light ratios are derived that are corrected for the effects of rotation and radial anisotropy. The average mass-to-light ratio for the galaxies in the sample is (M/L)B = (5.93 +/- 0.25)h50. The mass-to-light ratios correlate with total luminosity according to (M/L) is-proportional-to L0.35 +/- 0.05. The quantity (upsilon/sigma)* is identified as a second parameter in this relation. Galaxies with high (upsilon/sigma)* tend to have a low mass-to-light ratio (for their luminosity). The observed velocities in the outer parts of the galaxies in the sample systematically exceed the velocities predicted by our (constant mass-to-light ratio) models. From this it is argued that no axisymmetric constant mass-to-light ratio models can fit the kinematical data in the outer parts of elliptical galaxies. Either dark haloes must be present, or more detailed triaxial models must be studied.