DEEP IMAGING OF HIGH-REDSHIFT QSO FIELDS BELOW THE LYMAN LIMIT .2. NUMBER COUNTS AND COLORS OF FIELD GALAXIES

We present an analysis of the number counts and colors of faint galaxies to approximately 26.5 mag in the fields of two high Galactic latitude, very-high-redshift QSOs. The images, obtained in three optical passbands (U(n), G, and R), are comparable in depth to the deepest imaging surveys published to date, and were obtained as part of a program aimed at combining ultradeep imaging surveys with the information available from the absorption line spectra of the high-redshift QSOs. In this paper, the second in a series, we concentrate on the general properties of the field galaxies at faint magnitudes. The study was motivated by the significant disagreement between the two previously published studies using similar data. In particular, we readdress the faint galaxy number counts and colors as a function of apparent magnitude, and we reexamine the possible contribution of very-high-redshift (z greater than or similar to 3) galaxies to the faint samples. We have reached the following principal conclusions: (1) After extensive modeling of the incompleteness in the detection of faint galaxies in our images, we find that the number counts to R=26 are well fitted by the relation log N(m)=0.31R+C, which is consistent with the slope of the number-count/magnitude relation obtained recently in the K band. The G-band (very close to B(J)) counts for the same galaxies are consistent with the same slope (with a factor of approximately 3 smaller normalization at a given magnitude) fainter than G approximately 23.5, but exhibit a much steeper slope at brighter magnitudes. At R=25.5, the differential number counts have reached approximately 1.2X10(5)/deg2; the same surface density of galaxies is reached at G=26.5. (2) We confirm the existence of a gradual ''blueing'' trend of the field galaxies toward fainter apparent magnitude; however, the blueing trend appears to extend only as faint as G approximately 24 (or R approximately 23), fainter than which both the (G-R) and (U(n)-G) colors appear to level off. The mean colors of faint galaxies are considerably redder than flat spectrum, in agreement with some previous work, but in substantial disagreement with other work. There are essentially no objects to R=26 which have spectral energy distributions which are bluer than fiat spectrum. (3) The potential contribution of very-high-redshift (z>3) galaxies may have been underestimated in previous analyses; the current data, based on the number of objects which have faint apparent R magnitudes, relatively fiat (G-R) colors, and very red (U(n)-G) colors, are consistent with the same population of relatively luminous (approximately L*) galaxies at z approximately 3 as exist at z approximately 0.7. Two objects in the 0000-263 field have been identified previously as z approximately 3.4 galaxies, and hence serve to represent the expected optical colors of the general galaxy population at very-high redshift. Neither of these objects would have been found in the deep spectroscopic surveys which have been performed so far, if the two identified galaxies are typical (or, more likely, if they represent the bright end of the luminosity function) then one must go at least 1.5 mag deeper in the spectroscopic surveys (to B approximately 25.5) to routinely detect such objects.