Present address: Princeton University Observatory, Peyton Hall, Princeton, NJ 08544, USA.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

One should be cautious, as the use of OLS can be very misleading: linear fits not accounting for errors are known to underestimated the true slope of the relation. Merritt & Ferrarese (2001a), for example, have argued that the shallower slope found for the M- $\sigma$ relation by Gebhardt et al. (2000) is due mostly to a bias introduced by neglecting the observational errors. In our sample, however, it is hard to assign errorbars to the BH masses and velocity dispersions for galaxies.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.