5.4. Cooling flows and the evolution of cD galaxies
The phenomenology of cD galaxies was first described in 1964 by Matthews et al. . Eight years later Gunn & Gott  and Gallagher & Ostriker  proposed two alternative mechanisms for the formation and evolution of these cDs. Gunn & Gott  were possibly the first to suggest the existence of a physical link between the IC gas and cD galaxies. They showed that the cooling of IC gas, by thermal bremsstrahlung, would produce a flow of material in the central cluster region, that might accrete onto the cD galaxy. An alternative mechanism for the formation of cD galaxies was proposed by Gallagher & Ostriker  who suggested that the cD might form out of stars stripped from other galaxies. In this case one expects the outer parts of the cD to be in equilibrium with the cluster (rather than the galaxy) gravitational potential. Consistently, Dressler 's observations of the cD in Abell 2029 showed a rapidly growing galaxy velocity dispersion with radius, implying that the mass-to-light ratio of the cD was also rising with distance from the galaxy centre - see Fig. 34. A year later, in 1980, Gallagher et al.  showed the envelopes of cDs to be bluer than the mean galaxy colour, again consistent with the tidal debris scenario.
Figure 34. The rising mass-to-light ratio with radius in the cD galaxy of the cluster Abell 2029. From Dressler (1979).
Another popular scenario for the formation of cD galaxies was proposed in the 70's by Ostriker & Tremaine , and developed by Ostriker & Hausman . The cD galaxy would grow by cannibalism of its neighbours. This scenario was supported by the n-body simulations of White . White showed that as the cluster evolves, the dynamical friction mechanism can drive galaxies to the centre, and thus favour merging phenomena. Carnevali et al.  modelled the evolution of small groups, and showed that the ``merging instability'' leads to the formation of a large central object (they anticipated the discovery of fossil compact groups, see PONMAN, these proceedings). In the 80's the merging scenario for the formation of cD galaxies was re-examined by Roos & Aarseth  who concluded for an early creation of cDs via merging in small groups of galaxies, before the cluster formation.
The merging scenario was supported by several observational evidences. Oemler  determined the luminosities of cD envelopes and showed them to be correlated with the total luminosities of their parent clusters. Dressler  pointed out that the lack of significant luminosity segregation in cD-type clusters was another indication that cD galaxies had cannibalized neighbouring galaxies. Carter & Metcalfe  showed the cD major axis to be aligned with the distribution of surrounding galaxies.
The merging scenario for the formation of cDs was shattered in 1978, when White 's simulations showed that merging can produce giant elliptical galaxies, but not the the cD extended halos. In those years, Lea et al. , Silk , Cowie & Binney  and Fabian & Nulsen  estimated the cooling time of the IC gas in the dense X-ray emitting clusters to be lower than a Hubble time. Fabian & Nulsen noted that ``slow-moving galaxies in core of X-ray emitting clusters can accrete large quantities of cooling gas'', and Quintana & Lawrie  showed cD galaxies to be characterized by small velocities relative to the cluster mean. This gave new strength to the hypothesis of cD growth via accretion of the cooling IC gas.
A first observational evidence for the existence of cool gas in the cluster centres came in 1979 with the detection of soft X-ray components in the spectrum of the Perseus galaxy NGC 1275 (Mushotzky & Smith ). Another observational evidence came with the detection of optical emission-line filaments near the centre of clusters, that were interpreted by Cowie et al.  and Fabian et al.  as arising from the IC gas cooling down to ~ 10000 K.
Gorenstein et al.  remarked upon the different X-ray emissions of the central galaxies in Virgo and Perseus, on one side, and the two dominant galaxies in Coma, on the other side. They correctly pointed out that the difference was related to the lack of cooling flows in the Coma cluster. If NGC 4874 and NGC 4889 were moving through the IC gas, their motion could prevent the formation of a cooling flow (Mathews & Bregman ). A significant motion of the two dominant Coma galaxies with respect to the cluster was later discovered .
In the 80's Lea et al.  and Sarazin & O'Connell  noted that the inferred mass deposition rates in cooling flows were much higher than the inferred star formation rates as derived from UV observations (e.g. Bertola et al.  for M87). The hypothesis was made  that only low-mass stars, characterized by small UV emission, can form in the high-density cooling flow regions.
Cooling flows have since become a major research topic in cluster astrophysics. Two thirds of all clusters contain a cooling flow at their centre. The deposited mass is still unaccounted for, but there exist evidence for X-ray absorption in the centres of cooling-flow clusters, which might be related to the deposited material (see FABIAN, these proceedings). Maybe the active nucleus which is often present in galaxies with cooling flows plays a significant role in re-distributing the accreted material (see MCNAMARA, these proceedings).