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

3.2 Rotation

Kinematic observations of elliptical galaxies have provided several surprises. First, it was found that luminous ellipticals rotate slowly (32, 169). A modern observation of NGC 1600 gives a rotational velocity of 1.9 ± 2.3 km s^-1 along the major axis, resulting in a v / < 0.013 (182). Although in one particular case projection effects might play a role, the luminous ellipticals as a class have a mean v / 0.2. Binney showed that this is significantly lower than expected for oblate isotropic rotators (35). Hence the shapes of bright elliptical galaxies are supported by anisotropies in the velocity distribution (34, 36). A useful diagnostic indicator is (v / )^*, which is the measured v / divided by (v / )_iso, the value expected for oblate rotationally flattened galaxies. A good approximation is (v / )_iso = [ / (1 - )]^1/2, where is the ellipticity of the galaxy (189). The parameter (v / )^* is almost independent of inclination for oblate models. Bright ellipticals have a mean (v / )^* 0.4.

The second surprise was the rapid rotation of spiral bulges, with v / > 0.5 (117, 171, 194, 231, 370). Most observations have been done for (nearly) edge-on bulges in early-type spirals. Contamination of bulge light by light from the disk is a major problem, and many studies have tried to avoid this problem by observing at several position angles and/or offset slit positions. The total number of well-observed bulges is still relatively small. Detailed modeling, taking into account the influence of the disk, has shown that bulges rotate about as fast as expected for oblate isotropic rotators (117, 177, 178, 185).

The third surprise was the observation that ellipticals with -18 > M_B > -20.^m5 have (v / )^* 0.9, showing that intrinsically faint ellipticals rotate about as rapidly as bulges, most of which have comparable luminosities (68). There is a relation between luminosity and v / , in the sense that more luminous ellipticals rotate slower. However, the scatter about the relation is large, and its origin is not well-understood. Wyse and Jones noted that v / correlates with surface brightness, in the sense that galaxies with low surface brightness have lower v / (376). They speculated that the high surface brightness galaxies had dissipated more energy during their formation, and have thus a higher rotation. This is not a complete explanation, as the specific angular momentum J / M is also lower for the bright galaxies (27).

Subsequently, it was found that galaxies with ``disky'' isophotes have systematically high v / (23, 25, 58). Galaxies with ``boxy'' isophotes have a large spread in v / . A possible explanation is that the galaxies with disky isophotes contain relatively luminous disks, as indicated by a recent statistical analysis of the available data (295). The disks may dominate the observed kinematics, because the measurement techniques are more sensitive to the low velocity dispersion system in a multi-component galaxy (e.g., 122, 231, 370). Other parameters, like X-ray and radio-emission are also correlated with the isophotal shape (28, 193). This suggests that boxy ellipticals and disky ellipticals are intrinsically different, and may have had a different formation history. Detailed analysis of isophote shapes and kinematic properties may help estimate the light contribution of the disks (e.g., 26, 295, 312).

Very recently, several groups have found slow rotation in dwarf ellipticals (27, 60, 155). Bender and Nieto (27) observed five faint ellipticals with M_B > -18^m. All of these ellipticals have v / < 0.3, and four have (v / )^* < 0.5. The lower surface brightness galaxies generally are rotating slower. This result has added to the confusion in explaining the rotation of ellipticals. It is possible that the low mass and low surface brightness systems have formed in a different way. Supernova driven winds may have been more important during their formation (e.g., 77). These winds may help to decrease the v / of a galaxy. Other types of explanations have been considered also (27).

Figure 2. The rotational parameters of ellipticals correlated with dynamical and structural parameters. In (a), the rotation velocity v is plotted versus the mean velocity dispersion < >. Note the positive correlation for low < > and the negative correlation for high < >, and the large scatter. The filled symbols are upper limits. In (b) and (c), v / and (v / )* are plotted against absolute magnitude. The most luminous galaxies rotate slowly, while intermediate galaxies rotate fast, and the low luminosity galaxies (-MB < 18) are intermediate. In (d), v / is plotted versus the surface brightness at an effective radius. Lower surface brightness galaxies rotate somewhat slower. In (e), the specific angular momentum J / M is plotted against absolute magnitude. J / M is approximated by c re v, with the constant c arbitrarily set at 1. The line is a fit to the distribution of points when v is set equal to < >. Note the large scatter below this line for the luminous galaxies (27). In (f), v / is plotted against the isophotal parameter a4, which measures the deviations of the isophotes from ellipses (23). Disky isophotes give positive a4, while boxy isophotes have negative a4.

Recent results on the rotation of globular cluster systems and planetary nebulae in the outer regions of elliptical galaxies indicate that their kinematics differ from that of the luminous central regions. Data on globular cluster systems in the two Virgo ellipticals M87 and NGC 4472 suggest that their v / 0.3 (256), comparable to the v / of the globular cluster system in our galaxy and Andromeda, and consistent with the flattening of the luminous part of the galaxies. The number of observed globulars in these two cases is still small, and hence these results are very preliminary. The planetary nebula system in NGC 5128 (Cen A) is reported to be rotating with a velocity of 100 km s^-1 at 4r_e, as compared to 40 km s^-1 for the luminous, inner, part of the galaxy (119, 372). This may be interpreted as a gradient in the rotational velocity, but it is also possible that the formation process has caused a systematic difference between the rotational properties of the different components.

We have collected the available high quality data from references (23, 27, 67, 68, 125, 182). Photometric parameters and distances were taken from (114), or, if not available, from the references mentioned. The data are shown in Figure 2. All of these plots show a correlation and large scatter. In Figure 2e, the specific angular momentum is plotted against absolute magnitude. The solid line is the relation expected for galaxies with v / = 1. There is a significant spread below this line for luminous galaxies.

At present, no satisfactory explanation can be given for these results. The systematic differences between low-mass and high-mass galaxies cannot be explained in a scenario of purely dissipationless formation (68). Simulations of dissipationless hierarchical formation predict a large scatter for v / , but no large systematic trend with mass (18, 378). The lower specific angular momentum of bright galaxies cannot be explained completely by differences in the dissipated energy during their formation.

The planetary nebula and globular cluster systems provide very interesting independent results. The fact that they appear to rotate faster than the luminous part of the galaxy they are associated with argues strongly against formation in a monolithic dissipative collapse. In such a picture the central parts are predicted to rotate faster than the outer parts. Merger simulations have shown that the inner parts of galaxies can lose their angular momentum quickly to the outer parts (15, 17). It is possible that the planetary nebula and/or globular cluster systems have the same kinematics as the halo; this exciting possibility deserves great attention.