Annu. Rev. Astron. Astrophys. 1991. 29: 499-541 Copyright © 1991 by Annual Reviews. All rights reserved

5.4 Structure and Dynamics of Clusters

Cluster-dynamical properties are tightly linked to initial conditions. In particular the presence of substructure, and its evolution with redshift, may be an important gauge of cosmological models. Cavaliere & Colafrancesco (1990 and refs. therein) suggest that cluster substructure evolves slowly, and that after a Hubble time only a relatively small fraction of clusters will appear relaxed. On the other hand, West (1990 and refs. therein) stresses that a lack of significant substructure in the inner regions of most Abell clusters would imply that they are dynamically relaxed systems at present (see the recent review by Geller 1990). Observations, however, yield growing evidence for substructure. A well studied example of substructure in a cluster is that of Centaurus. On the basis of 180 identified cluster members, Lucey et al (1986) point out that what appears as a single structure projected on the sky, actually does split clearly into two separate components: Cen 30 at <cz> = 3041 km s^-1 and Cen 45 at <cz> = 4570 km s^-1, with velocity dispersions of 586 and 262 km s^-1, respectively. In each case, a large elliptical galaxy lies near the dynamical center. A model in which the two clumps represent substructures merging together is favored. Bothun et al (1983) convincingly showed that the Cancer cluster is made up of five discrete groups; in this case however, the groups-all characterized by relatively small velocity dispersions - are possibly unbound. Even in the case of Coma (the stereotypical relaxed cluster), close analysis of the galaxian distribution shows central substructure (Fitchett 1990). The importance of abundant radial velocity information in the study of cluster structure and dynamics cannot be overemphasized. In the robust sample of Dressler & Schectman (1988a, b) evidence for substructure is found to be significant in 11 of their 15 clusters. They also suggest that the process of cluster formation may occur over several discrete accretion stages, protracted over the Hubble time. The evidence that clusters are still forming at the present epoch has been reviewed by Sandage (1990).

Some kinematical information suggests that dynamical decoupling of the inner parts occurs in some clusters, i.e. a distinct population of galaxies, slowed down by dynamic friction, becomes actually bound to a cD, which eventually grows by cannibalism (Cowie & Hu 1986). This effect has been studied in A2029 (Bower et al 1988) and in several clusters including A2589 (Bothun & Schombert 1988). Discordant velocities between cDs and the cluster as a whole have been investigated in A2670 (Sharples et al 1988) and A1795 (Hill 1988). Zabludoff et al (1990) find that 8 of 9 cD galaxies have velocities significantly different from the mean cluster value. Some of these results had already been anticipated by Quintana & Lawrie (1982). While they are still somewhat ambiguous in providing hard quantitative information, they do nonetheless underscore (a) the condition of yet ``unfinished business'' associated with the process of cluster collapse at the present stage of evolution of the universe, and (b) the necessity of large redshift samples that allow proper interpretation of the dynamic circumstances of any given cluster.

Cluster evolution can also be gauged through changes in their galaxian population. After the early work of Butcher & Oemler (1978), it now appears fairly well established that such evolution has been observed. Three kinds of cluster galaxies are seen with a frequency that increases with redshift (Gunn 1990a): (a) blue, narrow emission-line objects possibly quite active in the formation of massive stars, (b) active galactic nuclei, such as Seyferts, and (c) a population of red galaxies that show Balmer absorption, indicative of the presence of A stars, in addition to an older red population. In the absence of emission lines, the latter galaxies show indications of recent but not ongoing star formation and are dubbed E + A objects (Elliptical + Absorption). While the spectra of most red cluster galaxies are indistinguishable from those of nearby ellipticals, the active galaxies are different from typical nearby spirals, implying that evolution has indeed taken place. The report of a correlation between the active population fraction and the cluster velocity dispersion (Newberry et al 1988), possibly a key to what causes the observed differences between high and low redshift cluster populations, redoubles the need for high-quality dynamic information on distant clusters.

A recent review of the space distribution of clusters of galaxies and its implication on the large-scale structure of the universe has been given by Bahcall (1988a), followed by discussions by Bahcall (1988b), Geller & Huchra (1988), and Sutherland (1988). The cluster-cluster correlation function, _cc (r) exhibits amplitudes substantially larger than those of the galaxy-galaxy correlation function, (r), at any separation. Firming our knowledge on the difference between the amplitude of and that of _cc - both on the scales over which significant correlation is revealed by the cluster population and on those in which peculiar motions might be implied by redshift elongation of _cc - should force important constraints on cosmological scenarios. Cold dark matter schemes, which have dominated the interpretational scene throughout the later part of the decade, may meet difficulties in producing coherent structures on scales exceeding 50 h^-1 Mpc and peculiar motions as large as 1000-2000 km s^-1, as might be indicated by some cluster samples. Sutherland (1988) has illustrated the misleading clues introduced by selection effects, incompletely sampled clusters that might break into several components, and limited redshift information. When these effects are compensated for, the evidence for larger peculiar motions and significant clustering on the largest scales is significantly weakened. The ongoing cluster redshift surveys, together with X-ray selected samples to be derived from orbiting observatories scheduled to be launched in the next decade, should resolve this important issue.