Lyman limit imaging of high-redshift galaxies .3. New observations of four QSO fields

We present the results of deep Lyman limit imaging in four new fields as part of a continuing search for galaxies at 3.0 less than or equal to z less than or equal to 3.5. The technique uses a custom broadband filter set (U(n)GR) designed to isolate objects having Lyman continuum breaks superposed on otherwise flat-spectrum ultraviolet continua. The observations are specifically aimed at detecting known galaxies producing optically thick QSO metal line absorption systems, but are equally sensitive to more generally distributed objects at the redshifts of interest. We have identified a plausible candidate for the QSO absorber in one out of the four new fields surveyed; in the other three the absorbing galaxy must be either fainter than R = 25.5 or closer than similar to 1 arcsec to the QSO sight line (or both). Together with previously published results of the survey, we have now detected in two out of six cases objects with the expected properties of star-forming galaxies at z greater than or similar to 3 within 3 arcsec of the QSO sight lines; the two candidate absorbers have similar luminosities, M(B) similar or equal to 22 (q(0) = 0.5, H-0 = 50 km s(-1) Mpc(-1)), and impact parameters, R similar to 10 h(-1) kpc. We find the average surface density of robust Lyman break objects to be similar to 0.5 galaxies per square arcminute to a magnitude limit R = 25.0. A simple, ''no evolution'' model based on the properties of normal galaxies at z less than or similar to 1 predicts a density of Lyman break objects only a few times larger than observed. We conclude that there is a substantial population of star-forming galaxies, of relatively normal luminosity, already in place at z = 3-3.5. If normal galaxies experienced a period of very high star formation early on in their history, it must have occurred prior to z = 3.5. It is possible that the observed surface density of z > 3 objects is biased by the presence of a luminous QSO in the redshift range of interest; the surveyed fields are too small to examine the possibility of a spatial association of the z > 3 candidates with the QSOs. We suggest a number of future observations that would resolve this question and clarify the nature of the putative ''normal'' galaxy population beyond z similar to 3. (C) 1995 American Astronomical Society.