Introduction

Polar ring (PR) galaxies have been the targets for several attempts to constrain the shape of dark matter halos: e.g. Schweizer et al. 1983, Whitmore et al. 1987 (WMS), Sackett & Sparke 1990 (SS), Sackett et al. 1994 (S94). They are early type galaxies with a ring or annulus of gas, stars, and dust orbiting in a plane nearly perpendicular to the equatorial plane of the host galaxy. This peculiarity allows the sampling of the velocity field in two nearly perpendicular planes, which may give constraints on the shape of the dark halo mass distribution. This is a difficult measurement because it is based on the extrapolation of the inner mass model, determined mainly by the visible mass, into the outer polar region, where the information is given only by the PR HI gas.
So far the results on the dark halo shape have been contradictory. WMS concluded that the dark halo associated with the S0 was nearly spherical; the same optical data were analysed by SS in a more detailed study, and they showed that a range of dark halo flattening, from E0 to E8, were compatible with the data. Their best fit was E5, and the spherical case gave a relatively poor fit. Recently S94 proposed a dynamical model for NGC 4650A\ using new optical observations of the radial velocities and velocity dispersion for the host S0 galaxy which rules out the spherical halo option for the dark halo aligned with the S0. The S94 best fit model includes a dark halo with an E6 to E7 flattening.
A recent dynamical model proposed by Combes & Arnaboldi (1996) for the kinematics of the prototype PR galaxy NGC 4650A (Whitmore et al. 1990) shows that the dark halo is only needed to account for the kinematics of the PR HI gas. This result is very similar to the situation observed for late type spirals: here the luminous component completely accounts for the observed kinematics in the inner regions, while the dark halo is needed to explain the kinematics in the far outer regions. The similarity between spirals and the wide PR of S0 systems is strengthened by a recent study of the BRK broad band photometry for a sample of PR galaxies (Arnaboldi et al. 1995). This survey shows (i) a colour gradient in the polar annuli towards bluer colours at larger radii, very similar to the observed colour gradients of the late-type spiral disks (de Jong & van der Kruit 1994); and (ii) integrated B-R, R-K colours of PRs are very similar to those of spirals. In spiral galaxies, there is some evidence that the dark matter surface density distribution is proportional to the HI distribution (see Freeman 1993). In the PR galaxy NGC 4650A, the HI lies entirely in the PR (van Gorkom et al.\ 1987, vGSK87). This led Combes and Arnaboldi to try a dynamical model for NGC 4650A, in which the dark component is flattened with the same axes as the PR itself. In this model, the dark matter distribution is then simply proportional to the HI gas distribution of the NGC 4650A system, just as it appears to be in spiral galaxies. The model was able to explain very nicely the published optical rotation curves and velocity dispersion profiles along the major axis of the S0 and the PR, and the low angular resolution HI velocity field by vGSK87. This question of the dark halo shape is very important because cosmological simulations of Cold Dark Matter universes predict the 3-D halo shapes to be triaxial spheroids with a maximum flattening c/a = 0.4 (Warren et al. 1992), so there is an urgent need to test the Combes-Arnaboldi hypothesis more thoroughly. To this aim, we obtained high resolution 21 cm data with the ATCA for the PR galaxy NGC 4650A, and the results are discussed in Section 2.

© Copyright Astronomical Society of Australia 1997