Involved people at OAB: Mignoli, Gruppioni, Zamorani.
The Marano Field is a southern sky area extensively surveyed in the
optical (by means of multicolor imaging, slitless and slit spectroscopy,
variability), in the X-rays (with a ROSAT 
ksec integration)
and in the radio band (with the ATCA radio telescope (
=0.2 mJy)
at 1.4 and 2.4 GHz). Including the recently performed ISO observations and
the future XMM deep pointing, it is one of the best studied regions of the sky
at all wavelengths.
The existing observations have already provided:
70 spectroscopically confirmed AGNs with 
(Zitelli, Mignoli, Zamorani, Marano & Boyle 1992, MNRAS, 256, 349).
A complete sample of 50 X-ray sources with
erg cm
s
,
one of the deepest existing surveys at these
wavelengths. Of these, 84% have been optically identified
(Zamorani, Mignoli, et al., 1999, A&A, 346, 731).
AGNs are by far the dominant class of counterparts of these X-ray sources,
representing 71% of the optical identifications obtained.
This is consistent with the ROSAT data in the Lockman field, which have
shown that at the faintest flux level reached by these observations
about 70-80% of the soft X-ray background is resolved into discrete sources
(Hasinger, ..., Zamorani 1998, A&A, 329, 482) and
about 3/4 of the identifications obtained so far are with classical broad-line
AGNs (Schmidt, ..., Zamorani 1998, A&A, 329, 495; Lehmann, ...,
Zamorani 1999, A&A, 354, 35). It is interesting to note that, while most
of these X-ray selected AGNs would have been selected as AGN candidates
also on the basis of their optical colors and morphology, about (10-15)% of
them would have been missed by a pure optical selection, either because
classified as extended or because their colors are not different from stellar
colors. Both these incompletenesses tend to become more serious at the
faintest sampled magnitudes. This is confirmed also by the spectroscopic
identifications of the X-ray sources in the Lockman field, where a number
of the spectra of the faintest AGNs show a substantial contribution
from continuum of the underlying galaxy. These data suggest that only
a combined multiwavelength approach can provide a complete census of
all AGNs at faint magnitudes.
A deep radio sample for which 
% of optical photometric
identifications and 50% of spectroscopic identifications, at typically
, were obtained (Gruppioni, Mignoli, Zamorani 1999, MNRAS, 304, 1999);
these are the highest identification fractions available so far in
literature for sub-mJy radio samples. This work has suggested
that the identification content of the sub-mJy radio sources may be strongly
dependent on the magnitude limit of the spectroscopic follow up.
While at bright magnitude (B<22.5) most of the optical counterparts are
star-forming galaxies, at fainter magnitudes most of
the optical counterparts appear to be early-type galaxies, probably containing
low luminosity AGNs. As a consequence, any conclusion on the content of the
sub-mJy population based on samples with a large spectroscopic incompleteness
does necessarily require significant and uncertain extrapolations of
evolutionary models for the different classes of optical counterparts.
A deep optical multicolor catalogue of an area of about 0.15 sq.deg.
in the same sky region, has been obtained through CCD photometry at the ESO NTT
telescope in the past years. From this catalog faint quasar candidates with
magnitudes up to 
were selected. A significant fraction
of these candidates has been observed spectroscopically with FORS1 at the VLT.
Preliminary analysis of these suggests that the efficiency of AGN selection
based on the standard criteria (colors + morphology) decreases significantly
at B > 23.0. At these magnitude most of the UV selected, point--like
objects turn out to be extremely compact narrow emission line galaxies
at z 
0.6 - 1.2, with the classical broad line AGNs being only about
20% of the total number of candidates.
These data, when fully reduced, will allow to firmly estimate the surface
density of AGNs at B
23.5, where very few data exist, and to test
at fainter magnitudes the existing models of luminosity function and evolution,
which have now been firmly established on the basis of large samples
(2dF survey) limited at B 
21.