Three-dimensional redshift surveys, which densely sample the local galaxy distribution, are essential to characterize the properties of galaxies and the nature of the large-scale structures at the present epoch. Dense sampling is critical for studying the morphology of large scale structures, while the number of galaxies and surveyed volume are necessary for detailed statistical analyses. It is important to emphasize that the nearby Universe is in some respects unique. For instance, only nearby can one expect to use the peculiar velocity field of galaxies to map the mass distribution in the framework of the gravitational instability paradigm. Comparison between the galaxy and the reconstructed mass distributions provides a valuable probe of the relation between galaxies and the underlying dark matter, at least on large scales (Dekel 1994, da Costa et al. 1996). Furthermore, linear theory gives a relation between galaxy density and peculiar velocity, which can be used to derive a velocity field from all-sky redshift surveys. The derived velocity field can be compared to that observed to measure = 0.6 / b, where is the cosmological density parameter and b is the linear biasing factor (e.g. Davis et al. 1996, da Costa et al. 1998).
Until the mid-70s all the information available to cosmologists was the projected galaxy distribution. Although it was apparent that the distribution was far from homogeneous nothing was known about the reality of the observed structures in three-dimensions. The pioneering pencil-beam surveys of nearby clusters and surroundings (e.g. Gregory & Thompson 1978) provided the first hint that the galaxy distribution was irregular, motivating wider-angle surveys. Such an attempt was first carried out by Sandage (1978). However, the sample, which had a median radial velocity of 1500 km s-1, was too shallow to probe the nature of the large-scale structures. The CfA Redshift Survey (Davis et al. 1982) was the first wide-angle survey to reach beyond the Local Supercluster. It provided strong evidence that the galaxy distribution was far from homogeneous, showing instead a complex topology made up of large empty regions and filaments. However, the structures were poorly defined because of the sparseness of the sample. The picture that emerged was considerably different from that envisioned just a few years earlier, in which clusters were believed to be rare, isolated regions of high density in an otherwise uniform background. In parallel, the KOSS survey (Kirshner et al. 1981), using a series of pencil-beam surveys, identified a large empty region with an estimated size of 6000 km s-1. Pressing the observations to fainter magnitudes, the survey of a thin 6° wide slice was completed (de Lapparent et al. 1986) showing that these empty regions were bound by remarkably sharp and coherent structures with scales comparable to the linear size of the survey (~ 100 h-1 Mpc). However,the slice-like geometry of the survey did not allow one to differentiate between two-dimensional structures or one-dimensional filaments. Further evidence of large coherent structures came from the HI Arecibo Survey of the Perseus-Pisces region (e.g. Giovanelli & Haynes 1991). More convincing proof that these coherent structures were not filaments but two-dimensional sheets came from the Southern Sky Redshift Survey (SSRS, da Costa et al. 1988). The SSRS was designed to extend the CfA survey to the southern hemisphere in order to: 1) obtain a unobstructed view of the LSS by avoiding the Virgo cluster; 2) to produce the first all-sky sample to moderate depth; 3) to test the reproducibility of different statistics employed in the analysis of the CfA data. By construction the southern sample was required to have the same surface density as in the north. Since there were no prominent nearby clusters this requirement led to a slightly deeper survey which allowed for the detection of the Southern Wall, a coherent, thin structure seen over the entire declination range probed by the SSRS.
The need for better sampling of the structures motivated the extension of the CfA and SSRS surveys to fainter magnitude limits. The CfA2 Redshift Survey (Geller & Huchra 1989) soon produced the striking map of the full extent of the structure seen in their earlier slice-survey - the Great Wall - a spectacular example of a thin two-dimensional structure containing several rich clusters. Following suit, the SSRS2 (da Costa et al. 1994a) confirmed that the Great Wall was not a rare event, even though unique is some aspects, but that large voids and walls are in fact common features of the galaxy distribution.
Combined, CfA2 and SSRS2 now cover over 30% of the sky to the same depth. The sample consists of over 20,000 galaxies, a remarkable progress relative to the first generation of wide-angle surveys. Together, the CfA2 and SSRS2 surveys provide a unique and, so far, unmatched database which combines dense sampling with an almost complete three-dimensional view of the present-day galaxy distribution. They extend out to a moderate depth (cz 15,000 km s-1), which allows one to probe a linear scale of the order of 300 h-1 Mpc.
Figure 1 shows a cross-section of the local galaxy distribution obtained by combining the CfA2 and SSRS2. From the redshift maps alone one finds that large coherent structures appear to be a common feature of the galaxy distribution. Walls and voids, 5000 km s-1 in diameter, are seen in every region large enough to contain them. The qualitative picture that emerges is one in which the galaxy distribution consists of a volume-filling network of voids.
Figure 1. Redshift versus right ascension diagram for galaxies brighter mB 15.5 and within a 10° wedge taken from the combined CfA2-SSRS2 sample.
Unfortunately, the scope of optical surveys is limited to relative high galactic latitudes because of the zone of avoidance. Therefore, to extend the sky coverage one must resort to infrared-selected samples. Examples of redshift surveys based on IRAS galaxies include the 1.2 Jy IRAS Survey (Fisher et al. 1995), QDOT (e.g. Kaiser et al. 1991) and more recently PSCz (Saunders et al. 1998). Although considerably more sparse than their optical counterparts, the main advantage of IRAS galaxy redshift surveys is the uniform and unmatched all-sky coverage. Full sky-coverage greatly simplifies statistical analyses (e.g. power-spectrum analysis, counts-in-cells), bypassing some of the problems associated with edge effects and survey geometry. Equally important is the uniformity of the parent sample, which eliminates some of the uncertainties that have plagued the nearby optical surveys. But, above all, the main contribution of redshift surveys of IRAS galaxies is the fact that only from them can one compute a reliable peculiar velocity field as predicted from the galaxy density field, a key element in understanding the dynamics of the local Universe.
Because of the unexpected large size of the structures observed nearby, it became essential to extend the surveys to greater depths. To achieve this goal in a reasonable amount of time, the Stromlo-APM survey (Loveday et al. 1992) used a sparse-sampling technique advocated by Kaiser (1986), ideal for low order statistics, measuring redshifts for about 1800 galaxies randomly drawn at a rate of 1 in 20 from a complete magnitude-limited catalog selected from the APM Galaxy Survey (Maddox et al. 1990a). The survey probes a depth of ~ 200 h-1 Mpc, sampling a volume about five times that of the CfA2-SSRS2, at the expense of small-scale information. The data were used to measure the luminosity function of galaxies and their clustering properties on large scale,with the large volume allowing for the sampling a large number of different structures. Analysis of the radial density variation also showed no evidence of a large local void, one of the proposed explanations for the strong variation of the APM galaxy number counts at the bright end (Maddox et al. 1990b).
All the previous surveys were carried out observing a galaxy at a time. A major progress in redshift surveys came about with the multiplexing capability of multi-object spectrographs. An outstanding example of the benefits of the combination of fiber-fed spectrographs and wide-field telescopes is the recently completed Las Campanas Redshift Survey (LCRS, Schectman et al. 1996) carried out on the 2.5m du Pont telescope at Las Campanas. The LCRS contains redshifts for over 25,000 R-selected galaxies covering 0.2 steradians in six strips, each 1.5° x 80°, in the south and north Galactic caps. The median redshift of the survey is z ~ 0.1. Although probing a much larger volume, about five times larger than the combined CfA2-SSRS2 survey, inspection of the redshift maps supports the picture that the galaxy distribution consists of a closely-packed network of voids ~ 5000 km s-1 in diameter bounded by thin, large walls, with no strong evidence of inhomogeneities on larger scales.
More recently, other surveys to comparable depth to the LCRS, but adopting different selection criteria and observing strategies, have been completed: the Century Survey (CS, Geller et al. 1997) with about 1800 galaxies covering 0.03 steradians and the ESO Project Slice (EPS, Vettolani et al. 1997) with about 3,300 galaxies covering 0.008 steradians. Again the large-scale features are qualitatively similar to those seen in earlier surveys. However, both surveys claim to find evidence of inhomogeneities, such as the Corona Borealis supercluster, on a scale of ~ 100 h-1 Mpc.
Further progress in this range of redshifts will have to await the completion of 2dF and SDSS which will measure of the order of a million redshifts, providing a complete and unprecedented wide-angle coverage of the galaxy distribution to a depth of about 300 h-1 Mpc. The impact that these surveys will have can already be appreciated from the preliminary results of the 2dF survey (Maddox, this proceedings).