In this section, we briefly discuss recent results from kinematical studies of extragalactic globular clusters. The required measurements were discussed in Sect. 3. Kinematics can be used both to understand the formation of the globular cluster systems, as well as to derive dynamics of galaxies at large radii.
6.1. Globular cluster system formation
Globular cluster system kinematics are used since a long time to constrain their formation. In the Milky Way, kinematics support the association of the various clusters with the halo and the bulge (see Harris 2000 and references therein). In M31, similar results were derived (Huchra et al. 1991, Barmby et al. 1999). In M81, the situation appears very similar again (Schroder et al. 2000). Beyond the Local group, radial velocities for globular clusters are somewhat harder to obtain. Nevertheless, studies of globular cluster kinematics in elliptical galaxies started over a decade ago (Mould et al. 1987, 1990, Huchra & Brodie 1987, Harris 1988, Grillmair et al. 1994).
Figure 6 illustrates one example where the kinematics of globular clusters allowed to gain some insight into the globular cluster system formation (from Kissler-Patig et al. 1999b). The figure shows the velocity dispersion as a function of radius around NGC 1399, the central giant elliptical in Fornax. The velocity dispersion of the globular clusters increases with radius, rising from a value not unlike that for the outermost stellar measurements at 100", to values almost twice as high at ~ 300". The outer velocity dispersion measurements are in good agreement with the temperature of the X-ray gas and the velocity dispersion of galaxies in the Fornax cluster. Thus, a large fraction of the globular clusters which we associate with NGC 1399 could rather be attributed to the whole of the Fornax cluster. By association, this would be true for the stars in the cD envelope too. This picture strongly favors the accretion or pre-galactic scenarios for the formation of the metal-poor clusters in this galaxy.
Figure 6. Velocity dispersion as a function of radius for various components around NGC 1399, see Kissler-Patig et al. (1998a) for details. The two solid lines are fits to the velocity data of the globular clusters and of the Fornax galaxies. The dashed line shows the X-ray temperature converted to a velocity dispersion, the triangles are stellar measurements.
As another example, Fig. 7 shows the velocity dispersion and rotational velocity for the metal-poor and metal-rich globular clusters around M87, the central giant elliptical in Virgo. There is some evidence that the rotation is confined to the metal-poor globular clusters. If, as assumed, the last merger was mainly dissipationless (and did not form a significant amount of metal-rich clusters), this kinematic difference between the two sub-populations could reflect the situation in the progenitor galaxies of M87. These would then be compatible (see Hernquist & Bolte 1992) with a formation in a gas-rich merger event (see Ashman & Zepf 1992).
Figure 7. Kinematics of red (thin line) and blue (thick line) globular clusters in M87. Projected velocity dispersion, and projected rotational velocity as functions of radius for a fixed position angle of 120°. Dotted lines mark the 68% confidence bands. Taken from Kissler-Patig & Gebhardt 1998.
Generally, the data seem to support the view that the metal-poor globular clusters form a hot system with some rotation, or tangentially biased orbits. The metal-rich globular clusters have a lower velocity dispersion in comparison, and exhibit only weak rotation, if at all (Cohen & Ryzhov 1997, Kissler-Patig et al. 1999b, Sharples et al. 1999, Kissler-Patig & Gebhardt 1998, Cohen 2000). The interpretation of these results in the frame of the different formation scenarios presented in Sect. 5 is unclear, since no scenario makes clear and unique predictions for the kinematics of the clusters. Furthermore, some events unrelated to the formation of the globular clusters can alter the dynamics: e.g. a late dissipationless mergers of two ellipticals could convey angular momentum to both metal-rich and metal-poor clusters, bluring kinematical signatures present in the past. Detailed dynamical simulations of globular cluster accretion and galaxy mergers are necessary in order to compare the data with scenario predictions. But clearly, kinematics can help understanding differences in the metal-poor and metal-rich components, exploring intra-cluster globular clusters, and studying the formation of globular cluster systems as a whole.
6.2. Galaxy dynamics
Kinematical studies of globular clusters can also be used to study galaxy dynamics. The globular clusters do only represent discrete probes in the gravitational potential of the galaxy, as opposed to the diffuse stellar light that can be used as a continuous probe with radius, but globular clusters have the advantage (such as planetary nebulae) to extend further out. Globular clusters can be measured out to several effective radii, probing the dark halo and dynamics at large radii.
The velocity dispersion around NGC 1399, presented above, is one example. Another example was presented by Cohen & Ryzhov (1997) who derived from the velocity dispersion of the globular clusters in M87 a mass of 3 × 1012 M at 44 kpc and a mass-to-light ratio > 30, strongly supporting the presence of a massive dark halo around this galaxy. With the same data, Kissler-Patig & Gebhardt (1998) derived a spin for M87 of ~ 0.2, at the very high end of what is predicted by cosmological N-body simulations. The authors suggested as most likely explanation for the data a major (dissipationless) merger as the last major event in the building of M87.
These examples illustrate what can be learned about the galaxy formation history from kinematical studies of globular cluster system. In the future, instruments such as VIMOS and DEIMOS will allow to get many hundreds velocities in a single night for a given galaxy. These data will allow to constrain even more strongly galaxy dynamics at large radii.