Annu. Rev. Astron. Astrophys. 1991. 29: 499-541
Copyright © 1991 by Annual Reviews Inc. All rights reserved

Next Contents Previous

3.1 Pencil-Beam Surveys

In trying to balance the requirements of depth, completeness, and limited telescope time, pencil-beam surveys are the most economical. The selection of the parameters associated with a pencil-beam survey, such as the beam cross section, is often tied to instrumental constraints, e.g. the size of a spectrograph's aperture plate. In order to overcome the effect of small-scale fluctuations in the galaxian distribution, one would thus ideally choose the cross section of a pencil-beam survey (at its nominal redshift depth) to be comparable to or larger than r0. The sampling density can then be set by matching the associated sampling error with that arising from the clustering properties and the chosen geometry of the sampled region.

A survey spread over many separate fields can afford to sample deeply in the redshift direction while also covering a large solid angle. Thus a combination of pencil-beam surveys can economically yield information on very large-scale features in xi (r). On the other hand, topological details that could be investigated with a wide-angle survey tend to be lost. In addition, the approach may complicate the definition of the survey's selection function; for example, a space dependence is introduced by coupling the number of redshifts obtainable per field to the multiplexing factor of a multiobject spectrograph when the survey field size is comparable to that of the spectrograph. The special niche for discrete field surveys of the pencil-beam variety is in the trailblazing determination of the rough, large-scale properties of the galaxian distribution, ideally followed by more detailed studies over wider solid angles.

KOSS In a series of joint papers, Kirshner, Oemler, and Schechter - later joined by Schechtman (hence KOSS) - , approached the observational study of large-scale structure by sampling almost completely to a fairly deep magnitude limit several widely spaced fields, each about one square degree in size. At the mean depth of their survey, one pencil-beam field is somewhat narrower than r0, so the structure seen in each tends to be dominated by the small-scale fluctuations in the galaxian distribution. Nonetheless their survey showed considerable deviations from homogeneity on very large scales. In 1981, on the basis of 133 redshifts in three fields ~ 35° apart, KOSS reported the discovery of a large void in Boötes, centered at cz ~ 15,000 km s-1 and extending over 6000 km s-1 in the radial direction (see also KOSS 1983). In 1987, a more elaborate analysis confirmed the existence and size of the void (although not as extreme a density contrast as initially reported); they sampled 283 small fields - each 15' on a side - spread between 13h30m and 16h30m, +30° to +70°, to mlim ~ 17. KOSS (1990) are currently extending their discrete sampling approach in a survey that covers the two galactic caps to a redshift of 0.2, to include eventually a total of 10,000 to 20,000 redshifts. The strategy combines the advantages of a multifiber spectrograph with the large field of view of the Las Campanas 2.5-m duPont telescope. They are surveying 200 fields, each 1.5° x 3° in size, covering about 25% of the sky between 10h15m and 15h15m, 0° to - 18°, for the northern galactic hemisphere, and between 21h00m and 04h30m, -30° to -48°, for the southern one. They are obtaining their own photometric and astrometric catalog by means of a CCD drift survey. A 60-fiber spectrograph can yield 40 to 45 redshifts of galaxies with magnitudes 16.0 < R < 17.3, per 2h exposure, and about 4 fields per night can be observed. By summer 1990, they obtained about 2000 redshifts.

DURHAM-AAT-SAAO Peterson et al (1986) surveyed five high-galactic-latitude UK Schmidt Telescope (UKST) fields, each with an average area of 14 deg2, widely separated between the equator and - 50°. To a mean limiting magnitude of about bJ ~ 17.0, each field contains about 70 galaxies. This first sample, observed with the Anglo-Australian Telescope (AAT), is commonly referred to as the Durham-AAT redshift survey (DARS). Comprising redshifts of 329 galaxies, it also benefits from high-quality photometric data, obtained digitally on UKST IIIa-J plates. This effort was recently complemented by that of Metcalfe et al (1989), who selected a sample of about 750 galaxies in nine UKST fields, to a depth comparable to that of the earlier DARS, and measured the redshift of every third galaxy in the sample for a total of 264 redshifts. These observations, carried out at the 1.9-m telescope of the South African Astronomical Observatory (SAAO), are referred to as the Durham-SAAO sample. A notable parallel effort, also at the SAAO, concentrates on the southern sky survey field 349, in which Parker et al (1986) reported 107 redshifts to bJ ~ 16.5. These studies have concentrated on the determination of the statistical properties of the galaxian distribution, namely the two-point correlation function (Shanks et al 1983, 1989), the galaxy luminosity function (Efstathiou et al 1988), and applications of the cosmic virial theorem (Hale-Sutton et al 1989). The Durham effort has been particularly effective in the study of high-redshift objects, as we discuss below.

GALAXIES AT HIGH REDSHIFT (z > 0.1) The study of high-redshift galaxies aims at several goals, mainly: (a) the approximation of an elusive fair sample, whereby reliable mean properties of the universe can be obtained; (b) the understanding of the evolution of galaxies and of the field luminosity function; and (c) a knowledge of the evolution of clusters and their galaxian content. Recent reviews that emphasize particularly aspect b can be found in Ellis (1987) and Koo (1988). Because of the low fluxes, redshift surveys employ different strategies for distant galaxies than for nearby ones. At high z, spectroscopic observations can be time consuming (at z ~ 0.5, a 200 km s-1 quality redshift requires the better part of a night's observing on a 4-m class telescope); the surface density of targets, however, is high, and the use of multiobject spectrographs is well-suited for the task. Alternatively, large numbers of galaxies can be surveyed photometrically, and the broad band spectral information can be used as an indicator of distance, a photometric red shift.

Photometric redshifts The technique, first applied by Baum (1962), is well reviewed by Koo (1985). Observations in many-color passbands yield an estimate of redshift that relies critically on assumptions of each galaxy's stellar population and evolutionary state. Evolutionary models, most notably those of Bruzual (1983 and refs. therein), need to be adopted, and redshifts are derived by best fits of the model spectral energy distributions to the data. Of particular importance to these studies is the identification of the 4000-Å break, a blend of the Ca II H and K and other metal lines whose cosmological potential is discussed by Dressler & Schectman (1987). Koo (1985) finds that two thirds of his photometric redshifts agree with spectroscopic values with an accuracy of ±0.04 in z for z leq 0.35 and ±0.06 to z leq 0.6.

Loh & Spillar (1986 and refs. therein) have attempted to apply the results of their photometric survey to place constraints on cosmological parameters. They obtained six-band photometry for a sample of about 1000 galaxies with a median z appeq 0.5 within a pencil beam of 24 µsr, and derived photometric redshifts by best-fitting the location of the 4000-Å break. The number of galaxies observed is essentially the product of the galaxy luminosity function and the volume sampled, and it can be expressed in terms of the mean luminosity density (phi*) and the density parameter Omega. Somewhat controversially, Loh & Spillar (1986) derived a relatively high value of Omega = 0.9 ± 0.3 for a cosmological constant Lambda = 0.

While the photometric technique of estimating redshifts is an economic one, it remains uncertain, mainly because of evolutionary effects that are themselves the object of study. Photometric techniques allow the probing of faint magnitude regimes (mB ~ 23), and as evolutionary effects become better understood, they might provide a very fruitful tool for the study of large-scale structure.

Spectroscopic redshifts Using a multifiber spectrograph at the AAT, Broadhurst et al (1988) extended an approach similar to that used in DARS to survey 200 galaxies of 20.0 < bj < 21.5, in five fields of 200 to 400 arcmin2 each, at a mean redshift of 0.22. Also at the AAT, Colless et al (1990) used a multislit spectrograph to probe even deeper (21 < bj < 22.5), obtaining 149 redshifts in three fields. With a multislit arrangement at the Kitt Peak 4-m telescope, Koo & Kron (1988) surveyed three fields near the North Galactic Pole; their survey currently includes 401 redshifts (Kron 1990, personal communication). These two complementary efforts have been combined to produce a remarkable view out to z ~ 0.5 in the direction of the two galactic poles (Broadhurst et al 1990). The most striking result is the appearance of a strong feature in the clustering at scales of 128h-1 Mpc. Ongoing efforts will extend the scope of the Durham-AAT and Kitt Peak probes to encompass as many as 30 or 40 pencil beams in both hemispheres, distributed over two patches (each several degrees wide, with one in each hemisphere) for a total of some 1000 redshifts. This combination approach of both pencil beam and sparse sampling emphasizes sensitivity to large transverse structures, since at z ~ 0.15, the pencil beam separation is ~ 10-60 Mpc. The other important thrust of these surveys is the understanding of the evolution of the field luminosity function, as briefly discussed in the last section.

Next Contents Previous