We present the results of spectroscopic studies with the LRIS spectrograph on Keck of two of the Hawaii deep survey fields. The 393 objects observed cover a 26.2 arcmin 2 area and constitute a nearly complete sample down to K = 20, I = 23, and B = 24.5. The rest-frame K -band luminosity function and its evolution with redshift are described. Comparisons are made with other optically selected (B and I ) samples in the literature, and the corresponding rest-frame Bband luminosity function evolution is presented. The B -band counts at B ∼ 24 are shown to be a mixture of normal galaxies at modest redshifts and galaxies undergoing rapid star formation, which have a wide range of masses and which are spread over the redshift interval from z = 0.2 to beyond z = 1.7.The luminosity functions, number counts, and color distributions at optical and IR wavelengths are discussed in terms of a consistent picture of the starforming history of the galaxy sample. [O II] emission-line diagnostics or restframe ultra-violet-infrared color information are used in combination with restframe absolute K magnitudes to construct a "fundamental plane" in which the evolution of the global star-formation rate with redshift can be shown, and we find that the maximum rest-frame K luminosity of galaxies undergoing rapid star formation has been declining smoothly with decreasing redshift from a value near L * at z > 1. This smooth decrease in the characteristic luminosity of galaxies dominated by star formation can simultaneously account for the high B -band galaxy counts at faint magnitudes and the redshift distribution at z < 1 in both the B -and K -selected samples. Finally, the overall K -band light density evolution is discussed as a tracer of the baryonic mass in stars and compared with the overall rates of star formation inferred from the rest-frame ultraviolet light density as a function of redshift.

The text of this paper has today been replaced with a version which contains all the figures and tables, including the galaxy redshifts.Unfortunately, I and my group, being small (because I have received no federal funding for this research for several years) have not finished two final pieces of analysis: (1) a paper on the evolution in the galaxy luminosity function and (2) a paper on the relationships between broadband colors, morphologies, and narrow spectral features (lines and breaks). I therefore request that the redshifts not be used for projects which are substantially similar to either of these two projects until I have submitted such papers to astro-ph. I expect this to happen by 2000 October. Thank you very much.

ApJ, 561, L153 (2001); revised 3/2/02 ABSTRACT It appears that the Lyman α forest is becoming thick at a redshift of about 5.5, cutting off the higher redshift intergalactic medium from view in neutral hydrogen. However, the effects of star formation at higher redshift are still readable in the intergalactic metal lines. In this paper I use observations of 32 quasars with emission redshifts in the range 2.31 to 5.86 to study the evolution of the intergalactic metal density from z = 1.5 to z = 5.5. The C IV column density distribution function is consistent with being invariant throughout this redshift range. From direct integration, I determine Ω CIV to be in the range (2.5 − 7) × 10 −8 and Ω SiIV in the range (0.9 − 3) × 10 −8 between z = 1.5 and z = 5. The metallicity at z = 5 exceeds 3.5 × 10 −4 , which in turn implies that this fraction of the universal massive star formation took place beyond this redshift. This is sufficient to have ionized the intergalactic medium. Subject headings: early universe -intergalactic medium -quasars: absorption lines -galaxies: formation 1 Visiting astronomer, W. M. Keck Observatory, jointly operated by the California Institute of Technology and the University of California.