4.1 The Size of D and cD Galaxies
One of the best indications that the cluster environment can affect the structure of galaxies is the existence of D and cD galaxies near the centers of many clusters. Morgan (1958) and Matthews, Morgan, and Schmidt (1964) originally identified these galaxies as a special class. The "c" stands for supergiant and the "D" means the galaxy is dustless with a large diffuse envelope around it. Kormendy and Djorgovski (1989) have argued that only the cD designation is meaningful, and the defining characteristic is an inflection in the outer brightness profile. The D galaxies do not have this added envelope, but instead have profiles similar to normal elliptical galaxies. However, the D galaxies are too bright to be a statistical fluctuation of the elliptical population (Bhavsar 1989), and hence will be retained in this review.
A typical giant elliptical galaxy may have a luminosity of about 1.5 L*, where L* is the characteristic luminosity from the Schechter luminosity function. An average D galaxy is 9 L* (Tonry 1987), and a cD galaxy is 12 L* (Kormendy and Djorgovski 1989). Similarly, the effective radius of a typical giant elliptical is 10 kpc, while for a D galaxy it is 40 kpc. The outer halo may extend to several hundred kpc. The large size is caused by a very flat luminosity gradient rather than a difference in the surface brightness of D and cD galaxies with respect to giant ellipticals.
An important clue to the origin of D and cD galaxies is that they are almost always found at the dynamical center of the cluster, and their velocity is very near the systemic velocity of the cluster (Quintana and Lawrie 1982). However, Quintana and Lawrie only had nine cD galaxies in their sample, two of which had residual velocities of more than 300 km s-1. Recent observations by Hill et al. (1988) and Sharples, Ellis, and Gray (1988) confirm that the velocities of these two cD galaxies are anomalous. The velocities of D galaxies also tend to be the same as the systemic velocity of the cluster, although there are six of 11 cases with residuals greater than 300 km s-1 in this case. This fundamental point needs to be checked using a much larger sample. D and cD galaxies are never found in the field (Schombert 1988). D galaxies are found in both rich and poor clusters, but cD galaxies are only found in rich clusters (Thuan and Romanishin 1981, Schombert 1988). Thuan and Romanishin suggests that D galaxies may form from local density enhancements (e.g., by cannibalism) while cD galaxies form by the accumulation of tidal debris. This hypothesis would also explain why only cD galaxies have extended outer halos. The identification of D and cD galaxies as the dynamical center is strengthened by the fact that their positions generally coincide with the X-ray centers of the clusters. In addition, the projected cluster profile shows a clear peak which follows a r-1 power law when the D galaxy is used as the center (Beers and Tonry 1986).
D and cD galaxies often have multiple nuclei
(Hoessel and Schneider
1985).
This has usually been taken as direct evidence that these galaxies have grown
to their prodigious size by cannibalizing nearby companion galaxies.
Merrifield and Kent
(1989)
have obtained CCD images for the inner 250 kpc of 29 clusters. They find that
the distribution of galaxies is characteristic of the effects of dynamical
friction. However, recent observations by
Tonry (1985)
have shown that most of the "nuclei" have
velocities with respect to the cluster that are much larger
(
800 km s-1) than
the internal velocity dispersion of the D and cD galaxies
( 300 km
s-1). Tonry interprets this as evidence
that only about 25% of the multiple nuclei are bound. Most of the nuclei
are simply
sharing the bottom of the cluster potential well with the D and cD
galaxies, and are not actually multiple nuclei.
Cowie and Hu (1986),
using a larger sample, argue for a bimodal distribution with
's of 250 km s-1 and
1400 km s-1. They argue that
about 60% of the multiple nuclei are bound. However, they did not
normalize by the
velocity dispersion of each cluster before adding the data together so
this might simply
reflect the spread in the types of clusters they studied (Lauer, private
communication).
Bothun and Schombert
(1988)
suggest an intermediate value for the fraction of bound
companions, since only one of the three clusters they studied
(A 2589,
the cluster with
the most dominant cD galaxy) has a large population of bound nuclei.
Lauer (1988)
has examined the shape of the multiple nuclei and finds that about 50%
show evidence
of an interaction with the central galaxy. Curiously, the distorted
nuclei do not tend
to have lower velocities with respect to the central galaxy as would be
expected if this
50% represented the bound nuclei. This may indicate that essentially all
nuclei passing
near the central galaxy are affected, and the half not showing
distortion yet are still on
their way in toward the central galaxy. However, Bothun and Schombert
find the most
obvious signatures of tidal stripping occur in the low-velocity nuclei
of A 2589.
Lauer (1988) estimates from the fraction of distorted galaxies that about 4 L* of material would be accreted by a central galaxy in 10 Gyr. This implies that either the accretion rate was slightly higher in the past, since a typical cD galaxy has 12 L*, or that another mechanism, such as the accumulation of tidal debris, may be needed to augment the process. In any case, this provides clear evidence for fairly substantial amounts of accretion through galactic cannibalism.
Several explanations have been proposed for the existence of D and cD galaxies, including: statistical fluctuation in the population of elliptical galaxies (very unlikely, see Bhavsar 1989), preferred location in regions without tidal shear (Merritt 1984, unlikely since field ellipticals with little tidal shear are never D or cD galaxies), cooling flows (Fabian, Nulsen, and Canizares 1984; unlikely, see section 2), galactic cannibalism of nearby companion galaxies (Ostriker and Tremaine 1975, Hausman and Ostriker 1978), and accumulation of tidal debris (Richstone 1976, Malumuth and Richstone 1984). We have already listed several supporting observations for the cannibalism and tidal debris hypotheses. Other evidence includes the fact that D and cD galaxies tend to be flatter than normal ellipticals, tend to be aligned with the axis of the cluster (Binggeli 1982), never show much rotation, and have velocity dispersion profiles that rise rapidly in their outer regions (Dressler 1979, and Carter et al. 1985). Observations of intracluster light in some clusters provide more support (Thuan and Kormendy 1977, Malumuth and Richstone 1984, and Struble 1988; see Gudehus 1989 for a dissenting opinion). Schombert (1986, 1987, 1988) has recently completed an extensive photographic study of a large sample of D and cD galaxies which also supports the dynamical formation of envelopes. More specifically, he finds the luminosity gradient in the outer envelope of a cD is about the same as the gradient of the galaxies in the cluster. He also finds correlations between the envelope luminosity and various cluster properties, including richness, X-ray luminosity, Bautz-Morgan type, and Rood-Sastry type.