SIMBAD references

Observations have demonstrated the presence of substantial amounts of interstellar dust in elliptical galaxies, most of which is believed to be distributed diffusely over the galaxy. Absorption by dust grains has a major impact on the transfer of stellar radiation, and affects the projection of each physical (and kinematic) quantity. In a previous paper, we have investigated the effects of a diffusely distributed dust component on the observed kinematics of spherical galaxies. In this paper we investigate the effect of not taking dust into account in dynamical modelling procedures. We use a set of semi-analytical dusty galaxy models to create synthetic line-of-sight velocity distribution (LOSVD) data sets, which we then model as if no dust were present.

We find some considerable differences between the best-fitting models and the input models, and we find that these differences are dependent on the orbital structure of the input galaxy. For radial and isotropic models on the one hand, we find that the dynamical mass of the models decreases nearly linearly with optical depth, with an amount of 5per cent per optical depth unit, whereas their orbital structure is hardly affected. For tangential models on the other hand, the dynamical mass decrease is smaller, but their orbital structure is affected: their distribution functions appear less tangentially anisotropic. For all models the mass-to-light ratio will be underestimated, by a factor of around 20per cent per optical depth unit.

We discuss these results in the light of the limited effects of dust extinction on the LOSVDs, as obtained in Paper I, and conclude that the determination of the dynamical mass and the kinematic structure of galaxies is not only determined by the observed kinematic quantities, but is also critically dependent on the potential and hence the observed light profile. We argue that dust, even in rather modest amounts, should therefore be taken into account in kinematic modelling procedures.