Paradigms for galaxy clusters are changing. As in all tearing away from secure positions (Kuhn 1970) the process is controversial, yet continuing. Most papers in this volume suggest directions that will probably lead to even stronger new ideas about cluster cosmogony. We are concerned in this review with physical properties that have relevance for the question of whether clusters of galaxies are generally stationary, changing only slowly in a crossing time or if they are dynamically young. We examine if parts of a cluster may still be forming, falling onto an old dense core that would have been the first part of a density fluctuation to collapse even if all galaxies in a cluster are the same age, having formed before the cluster. During the 1930's the stationary nature of clusters seemed beyond doubt. A suggestion that they are dynamically young would have been too radical even for Zwicky who was the model of prophetic radicals. Rather, Zwicky (1937) took the stationary state to be given in making his calculation of a total mass, following an earlier calculation by Sinclair Smith (1936). The justification was that rich clusters such as Coma (1257 +2812; or Abell A1656), Cor Bor (1520 +2754; A2065), Bootis (1431 +3146; A1930), and Ursa Major No.2 (1055 +5702; A1132), known already to Hubble (1936) and to Humason (1936), appear so regular. The projected density distributions of clusters imitate that of an Emden truncated isothermal sphere (Hubble 1930), leading Zwicky (1937, 1957 p. 138 ff) to the large mass found by Smith for Virgo and even earlier to the clear statement that the total mass of the Coma cluster is 400 times its visible mass (Zwicky 1933).
However, signs that clusters may not be stationary were already known 30 years ago. Zwicky (1957, his pp. 78-79) was aware of the different areal distributions of ellipticals and spirals in the Virgo cluster. Later, the increased velocity dispersion of spirals relative to ellipticals was demonstrated (de Vaucouleurs 1961, Sandage and Tammann 1976), violating the known decrease in velocity dispersion with radius for stationary structures if spirals and E galaxies were well mixed. It is also known that the mean velocity dispersion varies by more than a factor of 3 between the north and south parts of Virgo (Figure 26 of Binggeli, Tammann, and Sandage 1987, hereafter BTS 87).
The most telling new datum is Dressler's (1980) demonstration that the morphological-density (hereafter M-D) relation (Hubble and Humason 1931) is satisfied in detail in regions of any given cluster, showing that kinematical mixing of cluster members cannot have occurred in a cluster's lifetime. Otherwise the different Hubble galaxy types would be well mixed in only several cluster crossings. The conclusion of little mixing is firm provided that there are no transformations of one galaxy type into another by mergers for example, or by sweeping in the harsh cluster environment. Because Dressler's M-D relation is so central to ideas of cluster cosmogony, its explanation is required if cluster formation and evolution are to be understood.
We set out evidence concerning these problems in this review. New data obtained in recent surveys of the Virgo cluster (Binggeli, Sandage, and Tammann 1985, hereafter BST 85), the Fornax cluster (Ferguson and Sandage 1988, hereafter FS 88; Ferguson 1989), and of several other groups and in the general field are described in sections 2, 3, and 4 from which we determine how luminosity functions vary with richness. Correlations of diameters and surface brightness with absolute magnitudes are shown in Section 5 for E galaxies in the Virgo, Fornax, and Coma clusters and in the nearby general field. The significance of the new data are discussed in the final Sections for the problems of (a) the dynamical youth of clusters, (b) the origin of the morphology-density relation, and (c) possible transformations of one galaxy morphological type into another in cluster centers.