We have come a long way from Shapley's "rare freaks" to the recognition that collisions are a primary force driving the continuing evolution of galaxies. Most of this progress is the result of the work of the last 25-30 years, beginning with the seminal work of Toomre and Toomre. There are two basic reasons that collisions are such an important process in galaxy evolution. The first is that collisions can strongly affect the structure, dynamics, and SFRs of the galaxies involved. This point was vigorously argued by the Toomres and Eneev, Kozlov, and Sunyaev, and their models triggered an ever-growing interest in the field which continued up to the present. Evidence for the strong effects of collisions can be found in almost every section of this paper. In particular, the discovery of massive halos around galaxies, and the realization that these halos would provide very strong frictional braking, virtually guaranteed that most galaxy collisions (outside of clusters) would result in mergers, with global consequences, as discussed above.
The second basic reason for the importance of collisional processes is that collisions are not the result of rare, chance encounters. Rather they are the inevitable result of the continuing, (hierarchical) growth of large-scale structure in the universe. That is, collisions are written into the initial conditions.
So why did Shapley and other galaxy pioneers mistake the importance and frequency of galaxy collisions? The vast distances between galaxies, and the consequent view of them as isolated, unevolving island universes certainly were important factors. With no knowledge of the dark halos it would have been difficult to imagine that collisions were anything but rare, though Baade's recognition that galaxies rarely traveled alone was an early hint to the contrary. The early results of Hubble's classification project, which found that most galaxies could be fit into a simple scheme with only a few basic categories provided further evidence that the freaks were rare. The relatively short time needed to complete a merger, and the rapid relaxation of the remnant into a surprisingly normal-looking state, were also not realized.
With extensive help from numerical simulations we now understand not only these basic points, but also many details of the processes that create the enormous variety of collisional forms, and drive the evolution of the colliding galaxies. Following Shapley, Zwicky, Schweizer and others, I have presented descriptions and simple classifications of many of these collisional forms in hopes of highlighting the connections between different forms.
Despite the vast morphological variety, collisional phenomenology drives largely from a few basic dynamical processes. The most extreme morphologies are the result of tidal torqueing, centrifugal bounce, and hydrodynamic splash. The first two can be reasonably approximated as the combination of an impulsive disturbance, and subsequent ballistic kinematics, which provides a simple conceptual picture. Yet, this approximation is only applicable to transient disturbances. On longer timescales, globally fluctuating gravitational fields and resonant couplings are responsible for dynamical friction and violent relaxation, which drive dynamical heating and the merger process. (Though we recall that dynamical friction also has an impulsive aspect.) These effects have been demonstrated in detail by self-consistent numerical models, and analytic theory provides a firm foundation for understanding them, as described above.
Our understanding of the process of mass transfer in nonmerging collisions is not so well developed. However, in the case of tidal torqueing in flyby collisions there are at least some good rules of thumb. Unfortunately, in the case of hydrodynamic "splashes" it appears that the amount of mass transfer is a very sensitive function of the relative orientation of the disks, and the orbital inclination. Thus, this case is more complex.
Our understanding of interaction induced SF is even less complete. We have learned a great deal in the last decade about the processes that organize and compress the gas on large scales. The most dramatic of these is the funneling of gas into the central regions of merging galaxies as a result of the redistribution of angular momentum and the increased gravity. Funneling provides the fuel and processes like gravitational instability and cloud crushing provide the match to light the largest bonfires in the universe, the super-starbursts. A significant fraction of the stars in the universe, especially in early-type galaxies may be formed in such events. Interaction induced bars also funnel gas and foster wave and resonant ring star formation. In this case, the fireworks are less spectacular, but also longer-lived, and so, probably contribute significantly to the net SF of the universe as well. The intensity of SF induced by transient spiral and ring waves is comparable to the bar case, but more short-lived. However, waves effect the outer disk gas reservoirs of late-type galaxies, and so, have an effect by stimulating star formation and metal production in new regions.
There is still a great deal to learn about the processes by which vigorous star formation is actually initiated in large gas clouds. Waves and other transient structures can serve as useful tools in advancing our understanding in this area. The fact that for some time after an impulsive collision, these structures largely follow ballistic kinematics (except for shocks in the gas), provides us with a good deal of understanding about the environment in which SF occurs. Probably the most universal SF process is the local, time-dependent, gravitational instability. The range of unstable wavelengths and the duration of the instability are predictable in transient compression waves. Observations are also more easily interpreted in these environments, which have not experienced multiple cycles of disturbance. Nonetheless, clear confrontations between theory and observation still require the highest available resolution in both multiwaveband observations and numerical simulations.
As discussed above, a number of other processes may also play important roles in interaction induced SF. For example, pressure-induced SF (cloud crushing) may be important in very dense environments. Processes that limit or inhibit SF may be equally important. Young star heat and momentum inputs may heat and disperse cool, dense star-forming environments. The starburst phenomenon may be the result of such self-limiting behavior, together with the generation of galactic winds and fountains by the heating processes. Isolated galaxy disks appear to be strongly self-regulated. If so, collisional systems are one of the few places to study the nonlinear, nonequilibrium expression of these processes.
The phenomena of gas disk disruption and reformation in the smaller partners in disk-disk collisions, and of accretional inhibition of SF in the central regions of the larger partner, provide special examples of collisions as laboratories for the study of nonlinear gas dynamical and SF processes. The study of such cases has only begun, but they may help explain how starbursts can be delayed in interacting systems.
All of these dynamical and SF processes are important aspects of galaxy evolution, and thus, astrophysical cosmology. For example, an understanding of galaxy-wide SF processes at low-redshift would be of great help in answering questions about the efficiency and duration of SF in the epoch of galaxy formation. Analogous statements apply to AGN activity. These questions bear not only on the observability of young galaxies, but also on the hard photon flux available for heating and reionizing gas on the largest scales, and also on the question of the degree to which intergalactic gas is enriched with heavy elements at early times.
At present, only quite tentative answers are available for questions about how galaxy collisions depend on environment and redshift. This includes questions of the relative frequency of galaxy collisions of different types as a function of redshift. However, the prospects for progress on these questions in the coming decade look very good. A vigorous, multi-faceted array of investigations is already underway. The most sensitive and highest resolution instruments in space and on the ground are being used to study galaxy collisions in unprecented detail, and at unprecented distances. Better instruments and more powerful computers are on the horizon. It seems likely that the overall picture of galaxy formation and evolution will be worked out (observed?) in the next few decades. It's an exciting time in this field!