2. POSSIBLE MECHANISMS

A wide variety of theoretical mechanisms have been proposed for affecting the structure and dynamics of galaxies in clusters. It is uncertain which of these are actually operating, and to what degree they are important. For example, even after more than a decade of concerted effort we are not certain that ram-pressure sweeping, one of the most "accepted" mechanisms, is the cause of HI deficiencies in cluster galaxies. The main difficulty in distinguishing which of the different mechanisms are occurring is that quite often the resulting observational signatures are nearly identical. In this case, gas evaporation may also be responsible for the HI deficiencies.

This review will often focus on one specific test. Is the effect caused by a local mechanism, such as the tidal interactions between a few nearby galaxies, or a global mechanism, such as the mean tidal shear from the potential well of the cluster? More specifically, we will examine whether correlations are better versus the local projected galaxy density, or a global property such as the distance from the center of the cluster, R_cluster. This rough breakdown is only a first step in unraveling the puzzle. Several other tests and observations will be required to be certain of specific mechanisms. For example, HI deficiencies appear to correlate quite well with R_cluster (although it is not clear whether they have been tested versus local galaxy density). This would suggest that something about the global condition of the cluster, such as the presence of intergalactic gas near the center, is responsible for the effect. In order to pin it down further we need to test whether HI deficient galaxies have a larger velocity dispersion than normal galaxies, since the effectiveness of ram-pressure sweeping should vary as the square of the velocity with respect to the cluster (see Haynes 1990).

2.1 Local Mechanisms

The strong correlation between morphology and local projected galaxy density (Dressler 1980), and the recent surge in interest concerning the possibility that subclustering exists (Geller and Beers 1982, Dressler and Shectman 1988, Fitchett 1988) has led to the general opinion that the subclusters are physical entities and play a dominant role in determining several galactic properties (see West, Oemler, and Dekel 1988, for a dissenting opinion). These subclusters have smaller internal velocity dispersions than the cluster as a whole, making galaxy-galaxy encounters stronger, and removing one of the early criticisms that interactions between galaxies traveling at large relative velocities ( 1000 km s^-1) would have little effect on the galaxies.

It is important to carefully define what we mean by a subcluster. In some studies it has been used to mean a small density enhancement with only a few members. In other studies it may contain nearly as many members as the main cluster. Our goal is to test whether the galaxies in an isolated subcluster have the same properties as the galaxies in an equally dense region in the cluster core, so we will assume a subcluster contains relatively few galaxies. The difference between clusters and subclusters becomes semantic if a subcluster is nearly as rich as the main cluster.

Some of the major local mechanisms are outlined below.

• Initial Conditions (isolated protocloud) - A decade ago, the standard picture for galaxy formation was that initial conditions in an isolated protogalactic cloud determined the final morphology of a galaxy. For example, Sandage, Freeman, and Stokes (1970) suggested that the fundamental parameter was the amount of angular momentum in the protocloud, with the higher momentum material having a slower rate of star formation and therefore settling into a disk after the initial collapse that formed the spheroidal component. Gott and Thuan (1976) suggested that the the strength of the original density enhancement was the prime determinant. The ratio of the rate of star formation (faster in dense protoclouds) to the collapse time determined whether the galaxy would be an elliptical or spiral galaxy.

• Tidal Shaking (galaxy-galaxy) - Miller (1988) has argued that a severe "shaking" of a galaxy as it passes by galaxies in the high density core of a cluster may rearrange the distribution of mass into a more elliptical-like distribution without actually adding or removing mass.

• Tidal stripping (galaxy-galaxy) - Gallagher and Ostriker (1972) originally suggested that interpenetrating collisions between galaxies in clusters may liberate material from the interstellar medium and deposit it on the cD galaxies found near the centers of most dense clusters. This material would have the velocity dispersion of the cluster ( 1000 km s^-1), rather than the velocity dispersion of the galaxy ( 300 km s^-1). It now appears that direct collisions are relatively rare, but lower speed encounters between galaxies that pass near each other may cause even more damage (Toomre and Toomre 1972), especially before the cluster has collapsed and the relative velocities between galaxies have increased. One of the reassuring aspects of galaxy-galaxy collisions is that we know they are happening at some level, unlike many of the theoretical mechanisms we are discussing in this section, as becomes evident by a quick look through the Arp Atlas (1966) or Arp-Madore Atlas (1987). The main questions are how frequently do these interactions occur, how much mass is really lost (especially how much dark mass), and will the remnant be distinguishable from the progenitor. Reviews by White (1982), Dressler (1984a), and Kormendy and Djorgovski (1989) discuss tidal interactions in more detail, but answers to these three questions are largely unknown.

• Merging - In its simplest form, this theory begins with an initial population of disk galaxies which subsequently merge to form ellipticals. Toomre (1977) noted that based on the frequency of severely interacting galaxies (e.g., the "Antennae galaxy"), and ages estimated from simulations, the number of elliptical galaxies was approximately equal to the number of remnants from these violent interactions. Roos (1981) has extended the merger hypothesis by suggesting that merging can also explain the presence of spheroidal bulges in disk galaxies. Several elliptical and S0 galaxies have now been observed with gas that counterrotates with respect to the stars (Galletta 1987, Bertola and Bettoni 1989, Rubin, Ford, and Hunter 1988), providing some support for this idea. However, no counterrotating bulges have been found in spirals (Kormendy and Illingworth 1982, Whitmore, Rubin, and Ford 1984, Fillmore, Boroson, and Dressler 1986). Peanut shaped bulges may also result from the recent accretion of material into a disk galaxy (Binney and Petrou 1985, Whitmore and Bell 1988). Several cosmological N-body simulations (Aarseth and Fall 1980, Roos 1981, Frenk et al. 1985, Zurek, Quinn, and Salmon 1988, and Carlberg and Couchman 1989) provide more support for the potential importance of merging.

• Galactic Cannibalism - Ostriker and Tremaine (1975) suggested that the central galaxy in a cluster may grow into a cD galaxy by the accretion of several nearby galaxies. As dynamical friction slows the victims they slowly spiral in and are devoured. This is the merger theory at its extreme, but is defined as a separate item since the remnant in the two cases is different (i.e., a cD galaxy for galactic cannibalism and a normal elliptical for the merger theory), and the existence of one mechanism does not necessarily indicate that both mechanisms are operating.