|Annu. Rev. Astron. Astrophys. 1992. 30:
Copyright © 1992 by Annual Reviews. All rights reserved
The redshift surveys just discussed (i.e. magnitude- limited samples of field galaxies) have so far failed to reveal a significant population of galaxies at z > 0.8. Nevertheless, such galaxies are known to exist based on other kinds of discovery techniques (including serendipitous discoveries). In some cases the high-redshift galaxies would have met the criteria for selection by the field-galaxy surveys and can therefore be regarded as ``normal.'' However, for galaxies that were not found in a magnitude-limited sample, it is not straightforward to assess the constraints on evolution because the selection criteria are usually not simply defined. Still, the growing number of plausibly normal field galaxies at z > 1 show that at least some galaxies were luminous many Gyr ago, consistent with an early phase of enhanced star formation in massive systems.
To give an impression of the number and nature of the known high-redshift galaxies, in the following several paragraphs we relate examples of the discoveries. We also review other kinds of selection criteria, for example searches for line emission, and what the limits from such surveys tell us about the redshift distribution, and therefore evolution, at very faint magnitudes.
The spectrum from a long slit that was positioned over a quasar yielded an apparently normal galaxy at z = 1.018 and B ~ 24 (Thompson & Djorgovski 1991). Other galaxies with redshifts of 0.72 and 0.92 have also been found accidentally with long-slit observations (H. Spinrad, private communication).
High-redshift galaxies may be identified as companions to quasars. In their first attempt with a narrow-band filter tuned to the Lyman line of a quasar, Djorgovski et al. (1985) discovered a companion at z = 3.22. Heckman et al. (1991a, b) and Hu et al. (1991) have reported finding several such candidates among radio-bright quasars. They also detected image extension that is perhaps associated with the host galaxy of the active nucleus. Another case which was originally found by broad-band imaging is a companion to Q1548+0917 with z = 2.76 and R ~ 23 (Steidel, Sargent, and Dickinson 1991).
Still another method is statistical and relies on using the excess number of objects found around distant quasars. In one early study, Tyson (1986) used a wide R band and found an excess of companion galaxies to quasars at z = 0.9 - 1.5 at a relatively bright limit of R = 21, thus requiring a surprisingly large amount (2 to 3 magnitudes) of luminosity brightening at these redshifts. Hintzen, Romanishin, and Valdes (1991) also found an excess of galaxies in the angular vicinity of quasars, but only by going deeper, to R = 23. Even if high-redshift quasars do show nearby companion galaxies, the effect of ionization by the quasars may complicate any interpretation of these galaxies as being normal.
Another powerful technique is to search for galaxies associated with intervening absorption lines seen in the quasar spectra, especially the strong resonance lines of MgII or damped Lyman , resulting from the halos or disks of ordinary galaxies. Searching for galaxies at z > 0.4 via MgII absorbers has great promise, as evidenced by the large number of emission-line candidates in the surveys of Yanny, York, and Williams (1990), Yanny (1990), and Bergeron and Boisse (1991), with one galaxy having been found at z = 1.025 (Bergeron 1991). Damped Lyman searches at significantly higher redshifts, z > 1.8, have been less successful (e.g. Smith et al. 1989). A few recent candidates lend some encouragement however, at z = 2.466 (Wolfe et al. 1992); z = 1.998 (Elston et al. 1991); z = 2.309 (Lowenthal et al. 1991); and the most distant at z = 3.409 (Turnshek et al. 1991). The inferred rates of star formation are quite low (< 10 M/year; Smith et al. 1989), and the continuum of the galaxy underlying the ionized gas may be extremely faint (mAB ~ 28; Wolfe et al. 1992).
Galaxies undergoing their initial collapse and star formation may be recognized from strong Lyman emission or a substantial Lyman-continuum break, and the surveys described above are sensitive to such objects. Pritchet and Hartwick (1987) used narrow-band (100 Å) imaging to detect Lyman emission in the red, and later Pritchet and Hartwick (1990) conducted a similar narrow-band search but at shorter wavelengths for sensitivity at z ~ 1.9. Both surveys yielded only upper limits to the surface density of Lyman emitters. Lowenthal et al. (1990) made a long-slit search of ``blank sky'' for Lyman between z = 2.7 and 4.7, also without a detection.
Surveys of distant clusters may yield sizeable samples of interloper field galaxies. Most noteworthy is the recent investigation of seven clusters with redshifts between 0.35 and 0.55 by Dressler and Gunn (1992), in which 51 field galaxies with good quality redshifts up to 0.85 were observed (Figure 4). This survey contains some of the faintest and highest-redshift field galaxies known. Dressler and Gunn (1992) applied color criteria because of their primary interest in the properties of the members of the high-redshift clusters, and it is therefore difficult to apply their results to field galaxies in general.
Finally, the large masses of distant clusters can be used as gravitational telescopes to magnify background galaxies (Tyson 1990). The distortions can increase the apparent size of such galaxies and thus amplify their total flux (surface brightness is conserved by lensing). Moreover, elongation of the images results in arc-like structures that are relatively easy to recognize. Redshifts for some of the arcs have been measured spectroscopically (Mellier et al. 1991), ranging from z = 0.73 to a possible 2.24. Despite very long exposures on large telescopes (e.g. 6 hours on Cl 0024+16 with the ESO 3.6-m telescope by Mellier et al. 1991), convincing redshifts have not yet been measured for other arcs. Continued surveys for the detection of arcs and subsequent spectroscopy will be completed to fainter surface brightness limits, and the potential of using arcs to obtain a sample of galaxies at z > 1 may yet be fulfilled. Even without spectroscopic redshifts, the spatial distribution of lensed galaxies may place limits on their redshifts statistically (Tyson, Valdes, and Wenk 1990).
The future development of these various techniques will identify larger and better-defined samples of high-redshift galaxies. The spectroscopic study of these galaxies will help constrain models for galaxy evolution, especially if the selection effects can be understood in detail. For the moment, the redshift distribution at, say, B = 26 is still very much unknown and poorly constrained.