|Annu. Rev. Astron. Astrophys. 1989. 27:
Copyright © 1989 by . All rights reserved
Shells or ripples are faint, arc-shaped structures in galaxy halos (Arp 1966, Malin 1979, Schweizer 1980, 1983, Malin & Carter 1980, 1983; Malin et al. 1983, Schweizer & Ford 1985, Schweizer & Seitzer 1988, Prieur 1988). They are reviewed by Schweizer (1983), Quinn (1984), Athanassoula & Bosma (1985), Dupraz & Combes (1986), and Quinn & Hernquist (1987); we discuss only their most important implications.
Schweizer (1980, 1982, 1983) was the first to suggest that shells result from the accretion of small galaxies. Numerical simulations by Quinn (1982, 1984), Toomre (1983), Dupraz & Combes (1985, 1986), Quinn & Hernquist (1987), Hernquist & Quinn (1987, 1988), and others make a convincing case that they form through phase and spatial wrapping of cold material sloshing back and forth in the gravitational potential of an elliptical. Supporting this interpretation are the observations that (a) shells are made of stars similar in color to or slightly bluer than the underlying elliptical (Schweizer 1980, Carter et al. 1982, 1988, Bosma et al. 1985, Fort et al. 1986, Pence 1986, Clark et al. 1987, Schombert & Wallin 1987); (b) shells at successive increasing radii alternate on opposite sides of the center [e.g. NGC 3923; Quinn 1982, 1984, Malin & Carter 1983, Fort et al. 1986, Pence 1986 (but see Prieur 1988)]; and (c) their outer edges are sharp and often edge brightened, like folded sheets (e.g. Malin & Carter 1980, 1983). Such structures form when a small accreted galaxy falls almost radially into a smooth and stationary potential.
Shell detection frequencies show that accretion of small companions is a normal event in the life of a galaxy. Surveys by Malin & Carter (1983) and by Schweizer & Ford (1985) find shells in 17% and 44% of field ellipticals, respectively. Not all encounter geometries and viewing angles produce visible shells, so the real percentage is larger. This suggests that a typical elliptical has experienced one or more accretion events. Disk galaxies also contain shells, albeit less often than ellipticals (Schweizer & Ford 1985, Schweizer & Seitzer 1988). There is no reason to believe that disk galaxies do not accrete companions. However, if the victim is too massive, the disk is destroyed. If the disk is very robust, the resulting flattened potential is not gentle enough to form ordered shells (Quinn 1984, Dupraz & Combes 1986).
At first it was hoped that shells could be used to measure galaxy mass distributions, because the number of shells find their radial distribution depend on the gravitational potential (Quinn 1984, Dupraz & Combes 1986, Quinn & Hernquist 1987, Hernquist & Quinn 1987). The steep potential gradient of an r1/4-law mass distribution predicts a large number (100-200) of shells. The reason is that stars in the innermost shells have much shorter orbital periods than those at large r; one new shell forms every time inner stars complete an extra half-oscillation with respect to outer ones. But galaxies typically contain 20 shells. This can be understood if dark matter is added at large radii to reduce orbital periods there. However, this argument is oversimplified. The predictions were based on the assumption that the accreted galaxy is disrupted instantaneously. If not, it sinks toward the center through dynamical friction. Then inner shell stars have spent less time at small r than we thought, so fewer shells are predicted. Dark matter is no longer required. This also solves the problem that some inner shells are surprisingly close to the center. All of this is discussed by Dupraz & Combes (1987), Hernquist & Quinn (1988), and Prieur (1988), who now conclude that shells cannot be used to measure mass distributions. However, Piran & Villumsen (1987) find that shells are not formed at all if stars are stripped too slowly from the victim. It remains to be demonstrated that we know how to construct regular shell systems over the largest observed radius range.
Shells may tell us something about galaxy shapes. Dupraz & Combes (1985, 1986) suggest that (a) when shells are short arcs bisected by the major axis, the elliptical is likely to be prolate and edge-on; (b) when shells align with the minor axis, the elliptical is oblate and edge-on; and (c) when the shells are randomly distributed in azimuth and the elliptical is nearly round, it is likely to be oblate and face-on (e.g. 0422-476; Wilkinson et al. 1987). As in Section 4.3, these results are statistical, except perhaps in the most regular cases. However, they suggest independently that ellipticals span a wide range of shapes from oblate triaxial to prolate triaxial.