Can we choose now between oblate axisymmetric main body and tumbling prolate one in the case of NGC 2685? Yes; as there is no stellar velocity gradient along the minor axis, we can definitely state that it is not a tumbling prolate ellipsoid. The velocity field of stars agrees well with a picture of an ordinary edge-on disk. The cosine-like azimuthal dependence of the central line-of-sight velocity gradient for the gaseous component (Fig. 3) with the maximum at the photometric minor axis can be obtained in two configurations: circular rotation in the plane of the polar ring or pure radial motions in the plane of the galactic disk. Our spectral resolution for the red spectra of NGC 2685, , allows to estimate a velocity dispersion of the nuclear ionized gas; it appears to be near 200 km/s - much higher than that of stars in the center of NGC 2685. The estimates of the stellar velocity dispersion range from 60 km/s to 114 km/s (Whitmore et al. 1990, Di Nella et al. 1995, McElroy 1995). So, the high velocity dispersion of the gas keeps a possibility of radial motion dominance over the rotation. However several other reasons force us to prefer a hypothesis of circular rotation in the plane of polar ring: neutral hydrogen in the polar ring demonstrates the same sense and amplitude of velocity variations (Shane 1980), and the molecular gas does so (Watson et al. 1994, km/s), so it seems improbable to suggest an existence of supersonically expanding or compressing gaseous disk with a radius of 5 kpc taking into account its regular appearance.
It is a case of IC 1689 where a hypothesis of a prolate tumbling potential may justify: we observe a small velocity gradient along the minor axis for the stars, and an overall shape of the polar ring is roundish as if it has a central orthogonal "spindle" directed almost towards us. There are also rather strong evidences for radial gas motions towards the nucleus: the emission surface brightness distribution has a shape of bar elongated in while a dynamical major axis of the gas demonstrates an orientation intermediate between that of the gaseous bar and continuum isophotes; such a turn together with a ratio of to close to implies a superposition of tangential and radial motions of roughly equal amplitudes. A difference of systemic velocities determined by using the velocity fields of gas and stars can also be explained in the frame of strong radial motion hypothesis: we observe mostly the southern half of the nuclear gaseous disk which, according to Reshetnikov et al. (1995), is seen fore the main body of the galaxy and which has positive velocity excess (motions towards the nucleus), so the average (systemic velocity) must be overestimated. For the gas distributed in the polar plane a projection of radial velocity component onto the line of sight is zero at and ; and indeed a marginal deviation of points down from the cosine curve is seen at these PA s in Fig. 3. That means that if we calculate a gas systemic velocity by averaging only values at and , we would obtain a value much closer to the systemic velocity for stars.
Another interesting perspective is opened by discovering a possible age difference between the decoupled nucleus and the surrounding bulge in IC 1689. An origin of chemically decoupled nuclei may be related to some gas accretion event (Sil'chenko et al. 1997); if it is the same event that produced also a appearance of the polar ring, there must be age similarity between the nuclei and the polar rings. Some qualitative effect may be traced in the case of NGC 2685 and IC 1689: the colour of the ring in NGC 2685 (0.7, Peletier & Christodoulou 1993) is redder than that in IC 1689 (0.3, Reshetnikov et al. 1995), and the mean age of the stellar population in the nucleus of NGC 2685 is significantly larger than that of IC 1689 (see Fig. 5). But a quantitative agreement is absent: Peletier & Christodoulou estimate the age of the ring in NGC 2685 as 2-5 billion years, we give a age of the nucleus equal to 10 billion years or more. The situation in IC 1689 gives a hint for solving this problem: the age of the stellar population in the nucleus is 5 billion years (Fig. 5), but the emission spectrum in the middle of the ring provides evidences for intense present star formation, so the epoch of the star formation in the ring may be long enough. It is not predicted by any models, and if this finding is confirmed views on polar ring formation would be changed.
© European Southern Observatory (ESO) 1998
Online publication: January 16, 1998