Annu. Rev. Astron. Astrophys. 1997. 35: 389-443 Copyright © 1997 by Annual Reviews. All rights reserved

1. INTRODUCTION

The nature of the faintest galaxies detectable with our large telescopes has a long and intriguing history and remains one of astronomy's grand questions. The quest to measure galaxy counts faint enough to verify the cosmological principle and to constrain world models motivated the construction of a series of ever larger telescopes, and this, in turn, led to a renaissance of observational cosmology that has progressed apace during the twentieth century (Hubble 1926, 1936). The interest in the deep universe as probed by faint galaxy statistics and the puzzling results obtained with modern instrumentation have been major driving forces in extragalactic astronomy through the 1980s and 1990s. The most recent chapter in this observational story is the Hubble Deep Field (HDF) project (Williams et al 1996) - an unprecedented long exposure of an area of undistinguished sky observed with the Hubble Space Telescope (HST) undertaken with the purpose of extending the known limits of the faint universe as delineated by field galaxies.

Although the original motivation for what has always been difficult observational work was an attempt to quantify the cosmological world model, the bulk of this review is concerned with the study of faint galaxies as a way of probing their evolutionary history. The evolutionary possibilities were, of course, recognized by the early pioneers (Hubble 1936, Humason et al 1956). However, galaxy evolution was surprisingly slow to emerge as the dominant observational motivation (see Sandage 1995 for a historical account). Even in the 1970s much of the relevant time on the Palomar 200-inch telescope was aimed at estimating cosmological parameters by using galaxies as tracers of the cosmic deceleration (Tammann 1984). Evolution was considered primarily as a necessary "correction" to apply in the grander quest for the nature of the world model; the term evolutionary correction remains (cf Poggianti 1997). Galaxy evolution only rose to prominence when quantitative predictions for the star-formation histories of normal galaxies became available (Sandage 1961, Tinsley 1972). An equally important theoretical motivation for studying evolution was the development of the statistical machinery to understand how physical structures form in the expanding universe (see e.g. Peebles 1971). Both propelled observers to construct catalogs of faint field galaxies and to consider comparing their statistical properties with theoretical predictions.

An appropriate birthplace for the modern era of faint field galaxy studies was the symposium Evolution of Stellar Populations, held at Yale University Larson & Tinsley 1977. The theoretical ingredients were largely in place in time for the first deep optical surveys made possible by a new generation of wide-field 4-m telescope prime focus cameras. In an era in which the photographic plate is so often disregarded, it is salutary to realize that many of the basic observational results that generated the discussion that follows were first determined from photographic surveys exploited by the new generation of computer-controlled measuring machines (Kron 1978, Peterson et al 1979, Tyson & Jarvis 1979, Koo 1981).

The question of interpreting faint galaxy data in the context of the evolutionary history of field galaxies was comprehensively reviewed by Koo & Kron (1992). Their article summarized the developments from the early photographic results through to deeper charge-coupled device (CCD) galaxy counts and, most importantly, provided a critical account of the interpretation of the first deep redshift surveys. As they stressed, the addition of spectroscopic data provided a valuable first glimpse at the distributions in redshift, luminosities, and star-formation characteristics of representative populations of field galaxies a few billion years ago. Several puzzling features emerged that have collectively been referred to as the faint blue galaxy problem (Kron 1978). In its simplest manifestation, an apparent excess of faint blue galaxies is seen in the source counts over the number expected on the basis of local galaxy properties. A more specific version of the problem that attracted much attention followed the results of the first faint redshift surveys (Broadhurst et al 1988, Colless et al 1990). The count-redshift data from these surveys did not solve the number problem by revealing a redshift range (at either low or high redshift) where this additional population could be logically placed. Relatively complex evolutionary hypotheses were then proposed to reconcile these results, including luminosity-dependent evolution, galaxy merging, and the existence of a new population of source present at modest redshift but, mysteriously, absent locally. Koo & Kron (1992) reviewed the material at hand but considered many of these hypotheses to be premature. They stressed the relatively poor knowledge of the local galaxy population and urged a more cautious approach. Meanwhile, however, the fundamental question remains as to how to account for the high surface density of faint blue sources seen to limits well beyond those of the spectroscopic surveys.

Although it is only five years since Koo & Kron's review, it is timely to address the question anew for several reasons. First, there has been an explosion of interest in the subject as evidenced by a number of diagnostics including large numbers of articles and the statistics of telescope time applications. Clearly, in many minds, the controversy remains. Second, considerable observational progress has been made through larger, more comprehensive, ground-based redshift surveys including those from the first generation of 8- to 10-m telescopes. The redshift boundary for statistically complete field surveys has receded from z = 0.7 (Colless et al 1990, 1993) to 1.6 (Cowie et al 1996). Third, significant new data bearing on the question has arrived from the refurbished HST including the Medium Deep Survey (MDS) (Griffiths et al 1994, Windhorst et al 1996) and the HDF (Williams et al 1996). Reliable morphologies and sizes of faint field galaxies have become available for the first time, providing a surge of new data similar to that provided by the first redshift surveys discussed by Koo & Kron.

The major questions addressed in this review are as follows: What are the faint blue galaxies seen in the deep optical galaxy counts and what role do they play in the evolution and formation of normal field galaxies; is there convincing evidence for recent evolution of the forms proposed; and, if so, what are the physical processes involved? "Faint" in this context is defined, for convenience, to be at or beyond the limiting magnitude of the Schmidt sky surveys (i.e. V > 21). "Faint" need not necessarily imply "distant," although a major motivation is the hope of learning something fundamental about distant young galaxies. The adjective "blue" is not applied rigorously in most of the articles in this area, and the term primarily reflects the significance of the excess population when observed at optical wavebands sensitive to changes in the short-term star-formation rate. "Field" implies selection without regard to the local environment (Koo & Kron 1992). Normally this is in the context of systematic surveys of randomly chosen areas that span large cosmic volumes. The distinction between field and cluster observations is straightforward to apply, but the physical significance of the different results remains unclear. The 1977 Yale symposium also saw the first quantitative evidence for an increase with redshift in the blue galaxy population in rich clusters (Butcher & Oemler 1978). Recent work (Couch et al 1994, Dressler et al 1994) interprets this evolution as produced via a changing star-formation rate in certain types of cluster galaxies, presumably as a result of environmental processes. However, until the physical cause of these activities is better understood, a connection between field and cluster evolutionary processes should not be ruled out.

Two points should be made about the scope of this review and its intended audience. First, a quantitative review of the evolution of galaxies would require a critical discussion of many related issues such as the stellar and dynamical history of nearby galaxies, indirect probes of the high redshift universe such as QSO absorption line statistics, primeval galaxy searches, the growth of large-scale structure, uncertainties in stellar evolution, and the reliability of evolutionary modeling used to interpret the wealth of data now available. All of these are active areas that impact heavily on galaxy evolution and merit reviews of their own. The strategy here is to focus solely on the faint blue galaxy question, drawing on additional evidence from these fields where appropriate. Second, the subject is developing rapidly with many new observational claims, some of which appear to contradict one another. A detailed resolution of these issues is usually highly technical and in many cases not yet possible. The review is therefore primarily aimed at a fresh graduate student entering the field rather than at the expert who is active in the subject. The hope is to bring out the key issues without getting overly bogged down in the numerical detail.