A common feature in most galaxy clusters is subclustering (many papers in this volume). This fact together with the morphological and kinematical differences between parts of clusters (such as in Virgo as in section 2.1) makes it clear that many clusters are not in a stationary state. The strongest evidence comes from the morphological-density relation (Dressler 1980) for the reasons discussed in the first section.
The ideas by Tully and Shaya (1984), Shaya and Tully (1984), and by Tammann and Binggeli (1988) for an explanation appear highly satisfactory. We suppose that E and S0 galaxies form in the densest perturbations of the initial fluctuation spectrum (Sandage, Freedman, and Stokes 1970). Hence, in the shells of the small Friedmann universes that have the highest over-critical density, and therefore are the first to begin collapse (after the initial expansion as part of the general expanding manifold of the large scale universe), E galaxies predominate. Because these densest shells collapse first, they will form the oldest core of the developing cluster composed at first entirely of E and S0 galaxies. Less dense parts of the fluctuation expand further before turn around and take longer to collapse onto the already formed core. Galaxies formed in these intermediate density shells are not mainly E and S0 types but are spirals of Hubble class Sa and early Sb. Such galaxies will, then, arrive onto the core after the E types but before the late type Sc, Sd, Sm, and Im types that form in the smallest density regions of the perturbation that will arrive to the core even later.
Tully and Shaya's picture is, then, that late type spirals in the Virgo
region are only
now arriving near the cluster proper, having been in the outer shells of
the hierarchy
of Friedmann low density shells. Nevertheless, these shells have
densities that are still
above the critical value so that collapse toward the core will occur
eventually. Because
of their low 
(effective)
values, which, however are still slightly greater than 1, these
outer shells take, of course, a very long time to turnaround. In this
picture, the Virgo cluster is still in the process of buildup;
successive layers still being placed upon the old
E and S0 rich core, and each layer with a different morphological mix
from those that have gone before.
Said differently, the morphological mix of a cluster depends on time. The newly arriving shells are composed of successively later Hubble galaxy types because the type of galaxy that is made from the density fluctuations depends on the density out of which it is formed (Sandage, Freeman, and Stokes 1970).
This model, explains naturally the Dressler M-D details. It requires the clusters that show the morphology-density relation to still be in the process of formation. At earlier times the percentage of E to spiral members near the core should be larger than at later times (at the earliest times all galaxies in the cluster proper were E and S0 types), because the outer shells of mainly spirals had not yet arrived. Hence the form of the M-D relation should be time dependent. This prediction provides a test. The amplitude of the morphological- density relation should depend on redshift. Clusters at large redshift should be spiral-deficient compared with nearby clusters.
The paradigm change concerning cluster youth therefore seems necessary. The many new data now available on cluster properties set out in previous sections can apparently be understood only if clusters are not stationary, but are still in the process of forming.
Much of this review has centered about data obtained in the Las Campanas surveys of the Virgo and Fornax clusters and selected loose groups. The Virgo survey was made in a team effort by Binggeli, Tammann, Tarenghi, and the author. The Fornax cluster plus loose groups survey was completed by Ferguson as part of his PhD thesis research. The field survey was a joint effort with Binggeli and Tarenghi, as was the work with graduate student J.-M. Perelmuter of Johns Hopkins on the surface brightness problems. I am grateful to these astronomers for the enjoyable times we have had in these collaborations.
It is a special pleasure to acknowledge the conversations with Gustav Tammann on the problem of the youth of clusters and on his model discussed in the text for the explanation of the Dressler morphological-density relation that is time dependent. His insight was crucial to the completion of this review.
It is also a pleasure to thank the staffs at Space Telescope Science Institute and at the Physics and Astronomy Department of The Johns Hopkins University for their hospitality during the organization of this review in March and April 1989.