Our inability to detect primeval galaxies (PGs) has been one of the
 major problems in the current observational cosmology.  Here, PGs are
 defined as progenitors of the present-day normal, luminous galaxies
 undergoing their first major bursts of star formation at large
 redshifts.  Despite the great sensitivities of current detectors in
 both optical and near-IR, we have still not seen PGs.  In fact, the
 null results from recent deep imaging and spectroscopic surveys are so
 puzzling that we are completely at a loss with respect to our
 understanding of galaxy formation processes.

   One way to explain the current situation is that PGs are heavily
 obscured by dusts.  For example, if there is a PG at z = 4, even
 its K-band light, being B-band in the restframe, could suffer
 significant dust extinction.

   In order to examine this possibility, we propose to perform a serach
 for PGs using ISOCAM at 6.75 micron (LW2).  At this wavelength, we can
 considerably reduce the effect of dust extinction.  Currently, only ISO
 has a capability of doing deep imaging at such a long wavelength.

   We will specifically look at fields around already known z>4 quasars.
 Since galaxies tend to cluster, we could enhance our chance of finding
 high-z galaxies by looking at the vicinity of known high-z galaxies.
 We have already obtained deep images of these fields in the B, I, and
 K band in our related PG survey from the ground.  By looking for objects
 which are detected by ISO but not in the ground data, we will examine
 the possibility that PGs are heavily obscured and therefore cannot be
 detected at shorter (< 2.2 micron) wavelengths with the sensitivities
 currently available on the ground.