5.1.2. Stellar Kinematics and Dynamics of Bars.
I. Introduction to Observing Program

The basic nature of bars is still poorly understood (see Kormendy 1981 and Binney 1982a for reviews). The bar pattern is inferred to rotate rigidly, while the stars rotate differentially and therefore flow through it or circulate around it. We have no self-consistent model of a bar, i.e., one in which the non-axisymmetric potential forces a collection of elongated orbits to align themselves appropriately and then to precess together in such a way that the potential is preserved. We do not know the typical shapes of orbits. For example, we need to know whether orbits are very elongated and always confined to the bar (Lynden-Bell 1979), or less elongated, thereby producing density enhancements where they concentrate or where stars move slowly (Contopoulos 1979, and references therein). In the latter case the bar is a density wave, while in the former it is a material body with internal circulation. Both kinds of orbits may exist, even in the same galaxy. Our theoretical understanding is hampered by a lack of observational constraints. In fact, non-stellar motions have not yet been detected.

A program has therefore been undertaken to measure stellar velocity fields in barred and related galaxies (Kormendy 1982a, b; see Kormendy 1981 for preliminary results). The purpose is to study the dynamical properties of bars and associated components, and to compare these with n-body models. To illustrate the questions investigated, I will here discuss the results for NGC 936, which is the galaxy with the most detailed observations.

Like all galaxies in the program, NGC 936 is chosen because it is a high-luminosity, prototypical barred galaxy (Figure 40). The absolute magnitude is M_B = -20.0, for a distance of 16.6 Mpc on the Mould et al. (1980) and Aaronson et al. (1982) distance scale. This high luminosity is important, because it ensures that velocities and dispersions, and small fractional changes in them associated with the bar, are as large as possible. Highly luminous galaxies also tend to be more regular than fainter galaxies. NGC 936 is an SB0; early-type galaxies are chosen because they are relatively transparent and because I want as much as possible to study a purely stellar-dynamical problem. The axial ratio at large radius implies that the disk is inclined at i = 49° to the line of sight, a favorable value which gives reasonably large measured velocities without being so nearly edge-on that the minor-axis data are dominated by the bulge. For a first study it is useful to choose a bar which is not aligned with either principal axis. It is then relatively easy to distinguish non-circular motions from circular rotation combined with an error in the inclination. The apparent angle between the major axis and the bar is = 59°; the true angle in the disk plane is ' = 66°. A final advantage of NGC 936 is that the disk beyond the end of the bar is unusually bright (Figure 40). Normally, SB disks are too faint to measure outside the B(lens) structure. Here the data reach out to 1.6 bar radii.

Absorption-line velocities and velocity dispersions have been mapped with the High Gain Video Spectrometer (Kormendy and Illingworth 1982a) and the Kitt Peak National Observatory 2.1 m and 4 m telescopes. Spectra were obtained with the slit aligned with the major axis, minor axis, and bar, and at two intermediate orientations. The scale along the slit was 2.7" pixel^-1 on the 2.1 m and 1.4" pixel^-1 on the 4 m. The limiting surface brightness is ~ 24 B mag arcsec^-2 in ~ 3 h of integration, when 10-20 rows of the spectrum are averaged. Velocities V, dispersions and line strengths are calculated using a Fourier quotient program originally written and kindly made available by P. Schechter (section 4.2.1).