Annu. Rev. Astron. Astrophys. 1991. 29: 239-274 Copyright © 1991 by Annual Reviews. All rights reserved

3.3 Detailed Kinematics

Detector improvements have made it possible to determine accurate profiles of the line-of-sight radial velocity and velocity dispersion, with good spatial resolution. It was found that elliptical galaxies have complicated rotation curves, as exemplified by the discovery of kinematically distinct cores.

3.3.1 DISTINCT CORES The velocity fields of about 25% of all ellipticals show remarkable differences between the central regions and the outer parts (25, 170). Typical examples are NGC 5813, which rotates rapidly in the center, and shows almost no rotation in its outer parts (104); IC 1459 and NGC 5322, which rotate rapidly in the center and show opposite rotation along the major axis in the outer parts (24, 122); NGC 4365 and NGC 4406, which show rapid central rotation, and slow rotation in the outer parts, but along the minor axis (24, 125, 182, 360). Whereas in these and other cases the core kinematics appears to be decoupled from that in the outer regions, the central rotation axis generally does not point in a random direction, but is usually aligned with the minor axis. Only one example has been reported where the rotation axis of the central parts lies along the major axis [NGC 5982 (360)]. Current observations can only detect velocity reversals in the inner parts of galaxies, as most data do not extend beyond one r_e.

The velocity dispersion curves of the galaxies with distinct cores do not appear unusual, although in some cases decreases towards the center. The photometry generally does not show marked peculiarities, and no strange color gradients have been reported (e.g., 126, 270).

The line profiles of the kinematically distinct cores show a strong asymmetry that reverses sign across the nucleus and is not due to stellar mismatch or instrumental effects [e.g., IC 1459 (122), NGC 5322 (25)]. Such line profiles arise naturally in a two-component galaxy, consisting of a slowly rotating main body with a large velocity dispersion, and a small central component which is counter-rotating rapidly, and has a small velocity dispersion. This second component may therefore well be a disk. Such a cold dynamical component can influence strongly the observed rotation curve, even if its contribution to the observed light is as low as 20% (e.g., 122, 231, 370), but its effect on the velocity dispersion profiles is small.

Kormendy was the first to stress that the formation of these systems may be due to mergers (190). Specifically, he explained the observations of NGC 5813 by the hypothesis that a small galaxy had fallen into the center of a large galaxy, and that the central rotation reflected the rotation of the small galaxy. In this interpretation the light from the center comes mainly from the small galaxy. Thus the center is expected to be blue and to have a low velocity dispersion. Numerical simulations of this type of merger (12) have shown that the dynamical structure of the large galaxy is in fact changed significantly by the process of merging. The central rotation of the merger remnant corresponds to the orbital angular momentum, rather than the internal angular momentum, of the small infalling galaxy. The remnant of the small galaxy dominates the light in the central parts.

The disklike structure of the subsystems suggests that they have formed from gas that has settled to the center. This may have involved a starburst, or a full-fledged merger of two spirals (122, 310). We can only put lower limits on the fractional mass needed to form these subsystems. It is impossible at present to decide if their formation is part of the formation of the whole galaxy, or is a secondary event. Observations of nearby mergers may help to elucidate the nature of the kinematic subsystems (13, 319).

Some elliptical galaxies show irregular rotation curves (181, 359), which may indicate that these systems are not yet fully relaxed. This result is unexpected, as the dynamical time in the central parts is quite short. Possibly, these systems can survive for much longer than a central dynamical time, or they form repeatedly, e.g., through regular infall of material, or by condensations and star formation from cooling flows.

3.3.2 VELOCITY FIELDS Detailed kinematical ``maps'' have been obtained for some ellipticals and bulges by combination of data taken at many slit positions (44, 171, 175, 194, 231, 373). Two-integral axisymmetric models can fit bulge velocity fields satisfactorily (178). The Jeans equations have been used to predict the apparent velocity dispersions and radial velocities under certain assumptions concerning the anisotropy of the velocity dispersions and the mass distribution (e.g., 44, 344). For some ellipticals constant M / L models with f = f (E, L_z), i.e., _z = _R (see Section 2.1), give satisfactory fits. The badly fitting galaxies may have distribution functions depending on three integrals, and/or changing M / L as a function of radius (but see Section 3.5.2).

The purpose of these detailed studies is to obtain a better description of the distribution function. This might give useful information about the formation mechanism. The problems with this type of approach can be illustrated by considering the interpretation of the cylindrical rotation observed in box-shaped bulges and ellipticals (175, 194). Binney and Petrou (47) modeled this with three-integral distribution functions which were sharply peaked, as expected from mergers or infall of cold systems (369). Rowley (296) constructed two-integral models, and concluded that the boxiness is caused by dissipative processes. Box-shaped bulges can be produced also by spin-up of spheroids (230). Yet another explanation is that boxy bulges are part of a bar (62). Observations of line profiles may help to distinguish between some of these suggestions. Detailed simulations will be extremely useful, even if they do not include gaseous processes.