We have constructed from the IRAS database,a sample of data reaching
a limiting 60microns flux of about 100mJy in a few regions of the sky,
and covering about 100 square degrees both in the northern and in the
southern hemisphere.This should allow to study the reliability of IRAS
detections,the nature of the faintest Iras sources and the number counts
down to this faint flux with a reasonable sky coverage.While the number-
counts reveal an excess above model-predictions (with and without
evolution)for galaxies,but at a much lower flux then previously found,
thus questioning the effective amplitude of the effect,the colours and
fluxes indicate that the "excess" cannot be due to the already known
populations of galaxies,including the blue dwarfs at z=0.5.
  On the other hand,the follow-up work to find the optical counterparts
shows an increasing number of cases at faint flux level where more then
one galaxy could contribute to the Iras flux.The signification of the
number-counts can therefore not be assessed before one has an estimate
of the fraction of multiple-objects entering the counts.Furthermore,
we need to know whether at these flux levels,we see the same population
of super-starbursters further out,or whether one starts to see a
contribution from closeby but fainter IR-emitters.
 We therefore ask, for a representative subsample of sources,for CAM
imaging  in the 12mu band to separate the various contributors in the
far IR,together with two additional CAM and one PHOT point to establish
the spectral energy distribution of the sources. Complementary
ground-based photometry and spectroscopy is under way for a larger
sample,in order to establish the nature of the extragalactic sources
seen at the limit of sensitivity by Iras.