Heating Cluster Gas

It is now generally agreed that some process prevents the diffuse gas in galaxy clusters from cooling significantly, although there is less agreement about the nature of this process. I suggest that cluster gas may be heated by a natural extension of the mechanism establishing the $\mbh-\sigma$ and $\mbh - M_{\rm bulge}$ relations in galaxies, namely outflows resulting from super--Eddington accretion on to the galaxy's central black hole. The black holes in cD galaxies are sporadically fed at unusually high Eddington ratios. These are triggered as the cluster gas tries to cool, but rapidly quenched by the resulting shock heating. This mechanism is close to the optimum efficiency for using accretion energy to reheat cluster gas, and probably more effective than `radio mode' heating by jets for example. The excess energy is radiated in active phases of the cD galaxy nucleus, probably highly anisotropically.