How do we go about locating the faint and distant galaxies at the heart of our exploration and this review?
We found several successful (or partly successful) methods to locate the faint targets at high redshift: none are without "flaws". For example, some methods are weakened by "contaminants", be they intrinsically faint M, L, or T dwarf stars in the galactic disk, or a mis-identified (longer wavelength, smaller redshift) emission line.
Following the theme in the Stern & Spinrad (1999) review, we shall discuss several of the more successful search techniques; initially we'll review the finding of distant galaxies utilizing non-optical wavelengths. Often these techniques turn out to be "safe" and productive.
4.1. Radio-Loud Galaxies
Radio galaxies at high redshift are rare but interesting guides to the location of large, mature galaxies and correlated structures - sometimes actual (rich) clusters (van Breugel et al. 1999; Lilly & Longair 1984). For some specific cases, like 4C 41.17 (z = 3.798), and also radio sources resembling it, we note that steep radio spectral indices and moderate flux densities correlate with high redshift and great luminosity. Such objects are visible across much of the presently observable Universe.
The stronger radio galaxies, those with fluxes S408 100 mJy, tend to follow a good Hubble relationship in the observer's near-IR bands; that is, their (K,z) magnitude-redshift correlation is linear with only a moderate scatter.
This result shows that the powerful radio galaxies, E systems in morphological appearance, have a fairly strong resemblance to a luminous "standard candle" (van Breugel et al. 1999; Best et al. 1999). The history of the steep radio spectral index "angle" is reviewed by de Breuck et al. (2000). Going for the steeep radio spectral counterparts also tends to minimize the "contamination" by Quasars (radio spectral indices < - 1.3).
We then may inquire: are all steep radio sources luminous galaxies and Quasar candidates? The answer here is mainly negative; it is the medium strength (so as not to exceed some limiting intrinsic luminosity) steep spectrum sources, identified at long wavelengths in the optical and IR that have the greatest promise in pointing out very distant spectrographic targets. These may be radio-loud stellar systems at a large redshift, say z 4.
Somewhat tangential to our central motivation, we note that at both small and large distances, radio galaxies possess some/many of the characteristics of giant E galaxies (or luminous cluster Es). Since these E galaxies here and now have a strong correlation amplitude at small separations, we can anticipate many of the distant radio Es to also have smaller companions - perhaps in a group population. These earmarks of early structure are going to be valuable; the recent paper of Venemans et al. (2002) illustrates a large (2 Mpc) overdense region at a redshift z = 4.1 located "around" the radio galaxy TNJ1338-1942. So the radio galaxy becomes a valuable marker in such a case. We note another, less well-documented case in the HDF(N) is currently being explored by Stern, Dey, Dawson, and Spinrad. Here the redshift is even greater; the first observed galaxies have z 5.2. No radio source takes part in that overdensity region, however. Stern et al. (2003) show a group surrounding the radio galaxy MG0442+0202 at z = 1.11.
The record redshift for a radio galaxy is still z = 5.19 (van Breugel et al. 1999), with TNJ0924-2201. Several observing groups are concentrating on the identification of deep samples showing a steep spectrum, with the expectation that some are ultra-luminous and located at z > 5. These are rare systems; one problem in interpretation is that it should be a fairly slow process to "build" a large and luminous galaxy. Perhaps it requires a cosmic interval in excess of a billion years to do so, either in the model described as a "monolithic collapse" (Eggen, Lynden-Bell & Sandage 1962), or by the accumulation of smaller structures (Searle & Zinn 1978) - a hierarchical model. With the currently popular cosmology [H0 = 65, = 0.7, m = 0.3], the look-back interval between z = 4 and (an arbitrary) z = 20 is only ~ 1.2 Gyr (see Fig. 2 again). That might be sufficient time to build a large galaxy; the implication is then a SFR of ~ 80 M yr-1. That is a rarely observed and atypically high SFR. So it is a clue that massive radio galaxies are unlikely to be found at z > 5. But the near-IR Hubble Diagram of the highest-z radio galaxies plotted by van Breugel et al. (1999) continues to suggest a continuity in galaxy luminosity which we may still extrapolate to stellar (and gaseous) mass similarities.
Under standard CDM-based models of galaxy evolution, we expect the giant elliptical galaxies, which are the hosts of today's radio galaxies, to form late (at z ~ 1) through a process of merging of smaller sub-units. Although these models seem to be consistent with what is known so far about field galaxy evolution (e.g. Barger et al. 1999), and indeed with observations of the hosts of the radio-quiet quasar population (Ridgway 2000), it is clear that radio galaxies are an exception. They seem to only show significant evolution at z > 2, and still appear to be luminous galaxies at z ~ 3 and perhaps beyond. One possible solution is that the most massive galaxies formed first in so-called anti-hierarchical baryonic collapse. In this model (Granato et al. 2001) the high baryon densities in the centers of the most massive dark matter halos cause them to start forming stars early. Thus, the fate of simplistic theoretical analyses suggest the need for a sharper observational analysis.