Annu. Rev. Astron. Astrophys. 1992. 30: 705-742 Copyright © 1992 by Annual Reviews. All rights reserved

3. TIDAL TRAUMAS

Although the peculiar appearance of galaxies in a few double and multiple systems had been noticed much earlier, it was Zwicky (1956, 1959) who first called attention to the enormous variety of extended structures seen in such objects. Zwicky described these features as ``clouds, filaments and jets of stars which are ejected massively from galaxies in collision'' by ``large scale tidal effects''. Many astronomers, however, were skeptical that certain remarkably narrow appendages could be blamed on tidal forces (e.g. Gold & Hoyle 1959; Vorontsov-Vel'yaminov 1961), and Zwicky himself suggested that ``electromagnetic actions . . . contribute to the internal viscosity of stellar systems''.

Bridges & Tails

Bridges and tails are common in interacting disk galaxies. The former often appear to link large galaxies to small companions, while the latter stretch far away from the galaxy causing the perturbation. That ``old-fashioned gravity'' could extract these narrow structures was illustrated by Pfleiderer & Seidentopf (1961), Pfleiderer (1963), Yabushita (1971), Clutton-Brock (1972a, b), Toomre & Toomre (1972; hereafter TT), Wright (1972), and Eneev et al. (1973). These calculations generally treated each galaxy as a point mass surrounded by a disk of test particles, a simple numerical method, but adequate for the job. Only close, relatively slow passages - parabolic or even sub-parabolic - readily produce convincing bridges and tails; faster passages produce smaller disturbances. As TT noted, such slow passages would naturally result between pairs of galaxies which, bound gravitationally since formation, have fallen together for the first time along extremely eccentric orbits.

With varying degrees of success, the calculations reproduced bridges and tails as thin, curving ribbons yanked from disks by violent tidal forces. Such features are clearly relics of recent collisions rather than signs of ongoing interactions. The most impressive examples arise in direct passages where the orbital angular speed of the companion temporarily matches that of the stars within the disks, creating a ``quite broad'' resonance.

The tidal hypothesis gained additional credibility from TT's deliberately simple models of four well-known interacting systems. Further studies of these and other galaxies continue to support the tidal picture. For example, Schweizer (1978) and Schombert et al. (1990) have reiterated Zwicky's claim that bridges and tails are composed largely of stars drawn from a population very similar to that of the disks from which they extend. Kinematic studies of interacting galaxies (e.g. Tully 1974; van der Hulst 1979a, b; Combes et al. 1980; Marcelin et al. 1987) have revealed velocity fields quite consistent with tidal models (Toomre 1978).

Early test particle models raised many questions which could only be addressed with self-consistent calculations. One such question concerns effects of self-gravity in structuring tidal bridges and tails. Once launched, such features develop in an essentially kinematic manner (TT), and deep exposures of NGC 4038 / 9 by Schweizer (1978) reveal thick tails consistent with free expansion both parallel and perpendicular to the direction of extension. On the other hand, self-gravity may create small-scale structure within tails. Zwicky (1956) noticed a concentration of luminosity near the end of the southern tail of NGC 4038 / 9 and speculated that such objects might evolve into dwarf galaxies. Schweizer (1978) reported clumps of gas and young stars in the tidal tails of other interacting systems, and Mirabel et al. (1991) describe a beautiful system with at least nine distinct knots spread along a total extent of 350 kpc. Models indicate that dwarf galaxies formed in this manner can incorporate considerable gas from the disks but very little material from the halos of their parent galaxies, and so should have unusually low M/L ratios (Barnes & Hernquist 1992).

Another such question concerned the conjecture that tails extracted from disks might be unable to climb out of deep halo potential wells (e.g. White 1983b). This raised the possibility that the long observed tails could contradict the large masses proposed for invisible halos. However, encounters between self-consistent disk/halo models indicate that galaxies with halos of four times the mass of their luminous components can nonetheless produce tails as long as those of NGC 4038 / 9 (Barnes 1988). Since the energy required to climb out is provided by falling in, mere length is probably not an effective constraint on halo masses.

Wavelike Tidal Spirals

Self-gravity plays an important role in structuring the inner disks of some interacting galaxies. TT noted that galaxies with bisymmetric spiral patterns often have close companions, and suggested that these spirals have a tidal origin. N-body models of galactic disks abundantly illustrate the development of tidally excited spirals (Toomre 1981; Noguchi 1987; Barnes 1990). In a shearing, self-gravitating disk of stars, perturbations can grow by factors >> 10 while swinging around to become trailing spiral patterns (Goldreich & Lynden-Bell 1965; Julian & Toomre 1966). Such ``swing amplification'' occurs because the shearing of the spiral pattern temporarily matches the epicyclic motions of individual stars, permitting a modified form of Jeans instability to develop (e.g. Toomre 1981). This mechanism can amplify tidal perturbations in situ, rapidly bringing forth a trailing spiral over much of the disk. The calculations show that such spirals, although manifested as density waves as well as material arms, do not survive long but wind tightly up over a few rotation periods.

Several groups have attempted to reproduce the inner spiral structure of M 51 with tidal interaction models (Hernquist 1990; Howard & Byrd 1990; Sundelius 1990). Perturbations strong enough to generate the large-scale tidal features originally modeled by TT and Toomre (1978) also give rise to a ``grand-design'' spiral density wave in the inner disk. At present, however, none of the calculations offer a really convincing reconstruction of M 51's spiral structure; better models of the pre-encounter disk of NGC 5194 are probably required.

Cartwheel Galaxies

Some, although by no means all, galaxies with pronounced rings (e.g. Few & Madore 1986) appear to be the result of collisions. As shown in simple N-body simulations (Lynds & Toomre 1976; Theys & Spiegel 1977), rings develop when a companion galaxy makes a close and nearly normal-incidence passage through the plane of the victim, exciting large epicyclic oscillations in the target disk. Initially, all these oscillators are in phase, but since their periods increase with mean radius, the oscillations drift out of phase and orbits crowd together radially to produce an expanding ring. Indeed, kinematic studies of several ring galaxies find velocities consistent with the collision model (e.g. Fosbury & Hawarden 1977; Few et al. 1982). Even the ``folded ring'' galaxy Arp 144, once interpreted as a collision between a galaxy and an intergalactic H I cloud (Freeman & de Vaucouleurs 1974), is now known to include two bodies with infrared colors typical of evolved stellar populations (Joy et al. 1988).

To be sure, not all ring-making collisions are described by the simplest models. The system studied by Taylor & Atherton (1984) exhibits a rather complex velocity field. More remarkable yet is the extreme range of velocities, exceeding 1000 km/s, observed in the messy H II ring of Arp 118 (Hippelen 1989); many features of this system are nicely reproduced by a test particle calculation, but the M/L ratios implied seem uncomfortably high. The ``Cartwheel galaxy'' itself presents a modest puzzle or two; the putative companion does not seem massive enough to produce the very strong ring observed (Davies & Morton 1982), and the spoke-like features giving this galaxy its name are not well understood. Self-consistent three-dimensional models might help account for all these systems (e.g. Appleton & James 1990).

Only head-on collisions leave the bulge of the victim at the center of the ring; if the perturber is somewhat off-center, the bulge can be yanked to one side, producing an empty ring (Lynds & Toomre 1976). More off-center passages, although presumably more common, may not retain their ring-like appearance for as long as a direct hit; the resulting morphology changes smoothly from a ring to an open tidally-induced spiral as the pericentric separation is increased (Toomre 1978). Ring-like shapes resulting from off-center passages are more common than generally thought. One example is the very extended H I tail of M 51 (Appleton et al. 1986), which presents an almost circular outline beautifully illustrated in VLA maps (Rots et al. 1990).

Damaged Ellipticals

Tidal interactions involving elliptical galaxies are more subtle than those involving spirals, since encounters in rich clusters tend to be fast and disturbed ellipticals produce diffuse sprays of stars instead of narrow filaments. Nonetheless, a number of interacting ellipticals have been identified by their luminosity profiles (Kormendy 1977), distorted and off-center isophotes (e.g. Lauer 1986), and peculiar kinematics (e.g. Borne & Hoessel 1988).

The effects of hyperbolic encounters on the luminosity profiles of spherical galaxies have been studied both analytically (e.g. Knobloch 1978) and numerically (e.g. Dekel et al. 1980; Aguilar & White 1985, 1986; McGlynn 1990). Such encounters do not tidally truncate the target; on the contrary, they promote stars to loosely-bound orbits, creating extended halos with r^-4 density profiles, closely following a de Vaucouleurs law (e.g. Jaffe 1987). Aguilar & White (1986) found that stars which have not yet phase-mixed produce transient, outward-moving luminosity excesses. These results provide a natural explanation for the distended profiles of galaxies with nearby companions (Kormendy 1977).

Photometric decomposition techniques provide further evidence of interactions among elliptical galaxies (e.g. Hoessel et al. 1985; Lauer 1986, 1988). In these reductions a smooth, concentric luminosity model is simultaneously fit to each galaxy in the field; the residuals reveal non-concentric isophotes and elongated features referred to as ``dynamical friction wakes''. These studies find little evidence for strong interactions among most multiple-nucleus cluster galaxies, in general accord with the view that such systems are either optical doubles or extremely fast, plunging encounters (Merritt 1984; Tonry 1984). Some points of confusion remain; Hoessel et al. (1985) report that only those pairs with velocity differences smaller than their internal dispersions show signs of interaction, while Lauer (1988) sees evidence for interactions in pairs with velocity differences in excess of 1000 km/s.

Surface photometry and long-slit velocity data of several interacting elliptical galaxies have been used to construct semi-restricted 3-body models (Borne 1988; Borne et al. 1988; Balcells et al. 1989). These models do a fairly convincing job of reproducing the tidally disturbed morphology and ``U-shaped'' velocity profiles (e.g. Borne & Hoessel 1988) of their subjects, but it is difficult to assess their claimed uniqueness since they do not seem to be overconstrained by the data. Fully self-consistent models including dark halos are probably needed to reliably predict the future evolution of these systems.

An interesting exception to most of the rules for interacting ellipticals are ``dumbbell'' systems (e.g. Valentijn & Casertano 1988). These are comparable pairs of giant elliptical galaxies with projected separations of ~ 10 h^-1 kpc, found at the centers of some rich clusters. The distorted morphologies and relatively small pairwise velocity differences observed in these objects imply that many are bona fide interacting systems. Tremaine (1990) suggested that dumbbell galaxies are the last stage in the merger of rich clusters, each containing a D or cD galaxy. This proposal provides a reasonable account of many features of dumbbell systems, including their morphology, separations, velocity differences, and overall frequency. Rix & White (1989) have constructed self-consistent equilibria for dumbbell galaxies and used N-body calculations to show that at least some of these models are free from violent dynamical instabilities; it is not clear if systems with extensive common envelopes can also turn out to be stable or if real dumbbell systems can get themselves into such slowly-evolving states.