In the Hubble Deep Field (HDF) we have identified candidate regions wh
ere primordial galaxies might be forming. These regions are identified
from negative or positive peaks in the difference maps obtained from
the HDF maps smoothed over 0.'' 8 and 4 ''. They have apparent V magni
tudes typically between 29 and 31 (missing flux below the local averag
e level for the dark objects) and are much fainter than the nearby L-
galaxies. The identified objects are shown to be real in two ways. Fi
rst, the cross-correlations of these peaks detected in different filte
rs are strong. The bright objects have the cross-correlation lengths o
f about 0.'' 3. Second, their autocorrelation functions indicate that
these faint diffuse objects are self-clustered. Furthermore, the autoc
orrelation function for the high-redshift starburst subset of bright o
bjects selected by color, has an amplitude significantly higher than t
hat of the total sample. The subset of objects, dark in the F450W and
F606W bandpasses, but bright in F814W, also shows stronger correlation
compared to the whole dark sample. This further supports that our sam
ples are indeed physical objects. The amplitude and slope of the angul
ar correlation function of the bright objects indicate that these obje
cts are ancestors of the present nearby bright galaxies. It is shown t
hat the data reduction artifacts cannot be responsible for our sample.
We have inspected individual bright objects and noted that they have
several tiny spots embedded in extended backgrounds. Their radial ligh
t distributions are diverse and quite different from those of nearby b
right galaxies. They are likely to be the primordial galaxies at high
redshifts in the process of active star formation and merging. The dar
k objects in general appear smooth. Our subset of the dark objects is
thought to be the ''intergalactic dark clouds'' blocking the backgroun
d far-UV light (at the rest frame) at high redshifts instead of empty
spaces between the first galaxies at the edge of the universe of galax
ies.