1. INTRODUCTION

At redshift z 1000, the distribution of matter in the universe was remarkably smooth: density fluctuations in the cosmic microwave background were of order one part in 10⁵ on the degree scale (e.g., Bennett et al. 1996). Locally, 13 h₅₀^-1 Gyr later at z = 0, we observe that the distribution of baryonic matter on the Mpc scale is far from smooth, with baryons largely consigned to luminous, bound structures, such as galaxies and clusters of galaxies. These present-day structures can be explained by the gravitational collapse and coalescence of the overdense regions of the early universe. A detailed understanding of this collapse, identified as galaxy and large-scale structure formation, is uncertain currently and stands as one of the primary challenges to astrophysicists today.

The earliest epoch of galaxy formation lies beyond a redshift of 5. Recent observations have, for the first time, directly measured systems at the large look-back times implied by z > 5 (e.g., Dey et al. 1998; Weymann et al. 1998; Spinrad et al. 1998; Chen, Lanzetta, & Pascarelle 1999; van Breugel et al. 1999; Hu, McMahon, & Cowie 1999). Several lines of evidence support a substantial epoch of galaxy formation prior to z = 5: the presence of metals (in excess of the primordial abundances) in high-z damped Ly systems (e.g., Lu et al. 1996), quasars (Hamann & Ferland 1999), and star-forming galaxies at z ~ 2.5-3.5 (Steidel et al. 1996a; Lowenthal et al. 1997) requires metal creation and dispersal at higher redshifts. The tight photometric sequences in both low-z and intermediate-z clusters also attests to high formation redshift z_f at least for the elliptical galaxy formation in dense environments (e.g., Stanford, Eisenhardt, & Dickinson 1995). Indeed, some elliptical galaxies at z ~ 1.5 are observed to contain evolved stellar populations with ages in excess of 3.5 Gyr (e.g., Dunlop et al. 1996; Spinrad et al. 1997; Dey et al. 1999b), again implying high formation redshifts.

Theoretical paradigms of galaxy formation are vastly different: do large galactic spheroids form primarily via the monolithic collapse of a protogalactic cloud (e.g., Eggen, Lynden-Bell, & Sandage 1962) or are they built up through the hierarchical accretion of a multitude of subgalactic clumps (e.g., Baron & White 1987; Baugh et al. 1998)? Both faint number counts and the apparent lack of massive red systems at z 1 in (K-selected) redshift surveys would seem to favor the latter model (Kauffmann & Charlot 1998). However, the most direct answer will come with detailed studies of protogalaxies in the early universe.

Considerable astronomical expertise and experience have been aimed at identifying protogalaxies in the early universe over the past 40 years (for a recent review, see Pritchet 1994). Table 1 lists the most distant galaxy confirmed as a function of time. There are several established and innovative methods to locate the minority population of distant systems from the confusion of faint, intermediate-luminosity systems that dominate faint galaxy counts (at optical/near-infrared wavelengths). This paper presents a review of these techniques with some attention applied to the implications of the current studies and expectations for this line of research in the near future.

Progress in this field has accelerated with the advent of new facilities, notably, the Keck telescopes. In Section 2 we present a brief historical review of distant galaxy studies followed by a discussion of protogalaxy searches at nonoptical wavelengths in Section 3. In Section 4 we discuss several optical/near-infrared selection techniques for the "normal" population of distant galaxies. The cosmological redshifting of the light from these distant systems implies that our ground-based optical/near-infrared window samples the rest-frame ultraviolet (UV) spectrum; in Section 5 we therefore discuss the results of recent space-based observations of the UV properties of the youngest galaxies locally, as detailed studies of these relatively bright systems can yield considerable insight into observations of the most distant systems. In Section 6 we discuss the biases in the protogalaxy search techniques. In Section 7 we detail some of the highlights of these studies. Finally, Section 8 summarizes the discussion and suggests the primary questions which may occupy workers in this field at the start of the new millennium.

Throughout this paper, unless otherwise explicitly stated, we adopt an Einstein-de Sitter cosmology with a Hubble constant H₀ = 50 h₅₀ km s^-1 Mpc^-1 and no cosmological constant, = 0. For this cosmology, the age of the universe at redshift z is 2/3 H₀^-1(1 + z)^-3/2 = 13.2 h₅₀^-1(1 + z)^-3/2 Gyr. Magnitudes are quoted in the Vega-based system unless otherwise explicitly stated.