Lyman-break galaxies at z ~ 3 represent both a large jump in redshift and a substantially different detection/selection technique than has been used previously in galaxy studies. While we do not yet have enough empirical evidence to allow us delineate an evolutionary scenario to link this population to galaxies observed at lower redshifts or in the present-day universe, several important traits of this population can already be traced:
The Lyman-break galaxies form a substantial population of star formers already at 15% of the cosmic age. They have assembled at least ~ 10-15% of all the stars observed in the local universe, but the true fraction is very likely higher by a factor of several due to the presence of dust obscuration in the observed rest-frame far-UV light of these galaxies. These stars seem unavoidably linked to the oldest stellar systems observed in the present-day universe, i.e., the elliptical galaxies and the bulges of spiral galaxies.
The rest-frame UV spectra of these galaxies bear close resemblance to those of local starburst galaxies. While the ISM of these galaxies has certainly undergone some degree of metal enrichment, as indicated by the numerous metal interstellar absorption lines, it is hard to make quantitative statements at this time, because the features are heavily saturated. Despite the high star-formation rates and the abundance of ionizing photons, the Ly line is often weak or absent, very likely due to the presence of dust in the ISM coupled with resonant scattering. The kinematics of the absorption and emission lines shows the existence of large outflows of gas, with velocity in excess of ~ 400-500 km s-1, very likely due to stellar winds and supernova explosions. This stresses the importance of understanding the effects of feedback in the process of star (and galaxy) formation.
Both the UV colors and optical nebular emission lines directly point to the presence of dust obscuration. Correcting factors very likely range between 3x and 7x, but unfortunately the exact numbers depend quite strongly on the shape of the extinction law, which is poorly constrained in these galaxies. The unobscured star formation rates of the individual galaxies, if sustained for ~ 109 yr, have produced stellar masses that span the range observed in the local universe, with a typical value around M*, or 1011 M.
HST high-resolution imaging of the UV morphology of the Lyman-break galaxies shows a dispersion of properties. The brighter systems have compact morphologies with light profiles characteristic of collapsed systems. They appear characterized by a relatively high degree of spherical symmetry, and have half-light radii that compare with those of present-day bulges and elliptical galaxies of intermediate luminosity. At the fainter end of the luminosity distribution the morphology seems to become more irregular and fragmented. A preliminary inspection of rest-frame optical images obtained with HST + NICMOS seems to show morphologies that are rather similar to the UV ones.
The Lyman-break galaxies have clustering properties expected from massive dark halos. As a consequence, they are also heavily biased tracers of the mass. Values of the mass of the dark halos that host Lyman-break galaxies at the bright end of the luminosity distribution are M ~ 1012 M. Large concentrations of these systems with total mass typical of rich clusters (i.e., ~ M ~ 1015 M) are also observed already at redshift z ~ 3. The Lyman-break galaxies are characterized by strong spatial clustering, with comoving correlation length r0 = 2.1 (3.3) h-1 Mpc, only a factor of 3 (2) of the present-day correlation length. The slope of the correlation function is = -1.9, very close to the one observed in the local universe. If the biasing parameter is only mildly varying as a function of the spatial scale, this disfavors standard CDM models and points towards an open universe. We are currently studying the presence of clustering segregation with the UV luminosity (i.e., the dependence of the star-formation activity on the mass of the hosting halos) using the HDF sample of Lyman-break galaxies (Giavalisco et al. 1988, in preparation).
We would like to thank Charles Steidel, Mark Dickinson, Max Pettini, Kurt Adelberger and Mindy Kellogg for many useful discussions and for allowing us to use unpublished material in this paper. We would also like to thank the organizers of this ST ScI Symposium for the excellent meeting. We gratefully acknowledge support from the Hubble Fellowship program through grant HF-01071.O1-94A awarded by the Space Telescope Science Institution, which is operated by the Association of Universities for Research in Astronomy, Inc. under NASA contract NAS 5-26555.