Carrying out galaxy redshift surveys using HI observations introduces an obvious bias into the sample: a galaxy must have significant amounts of neutral hydrogen gas in order to be detected. Because of the roughly anti-correlated variations in HI content and optical surface brightness with Hubble type, HI redshift surveys have contributed most to studies of the distribution of spiral and irregular galaxies, and are thus complementary to optical redshift surveys. While the latter are very efficient at obtaining redshifts for higher surface brightness early-type galaxies (E and S0), they often completely miss the LSB galaxian population. Because the redshifts of late-type spirals and irregulars in particular are more easily measured in the HI line than optically, the 21-cm line surveys play a critical role in tracing the large-scale structures defined by such systems.
A particular importance of the HI line for the study of large-scale structure arises because of the segregation of galaxies in different regimes of local galaxy density according to their morphology, surface brightness and, perhaps, luminosity. Numerous observational and theoretical studies have investigated the morphology-density relation (Dressler 1980; Postman & Geller 1984) and the universality of the luminosity function (Binggeli et al. 1988). Clear evidence of morphological segregation is visible in Figure 6, which shows the sky distribution of galaxies in the PPS region with redshifts in the range 4000 km s-1 < cz < 7500 km s-1. The upper panel shows all galaxies in the redshift window, while the middle and lower panels show, respectively, subsamples of early-type galaxies (E, S0, S0/a) and late-types (Sc, Irr). Giovanelli et al. (1986) have shown that morphological segregation can be seen even among the spiral classes, with corresponding increases in the relative proportion of Sa, Sab spirals and decreases in Sbc, Sc galaxies as the density increases. Indeed, the gradient in population fraction is evident in nearly all density regimes. Using the redshift and morphological information in the PPS database, Iovino et al. (1993) have concluded that morphology and luminosity are two independent parameters that correlate with both the clustering properties of galaxies and their distribution as a function of local density. Bright galaxies are relatively scarce in low density regions, while faint spirals and irregulars are rare in high density ones. Within each morphological class, the more luminous galaxies are preferentially found in regions of higher density than are the low luminosity ones. It is not yet known whether the process(es) responsible for such segregation reflects the local conditions at the epoch of galaxy formation or result from secular interactions with neighbors or the intergalactic environment.
Figure 6. The same region as shown in Figure 4, but restricted to galaxies in the velocity range 4000 to 7500 km s-1. The upper panel shows all galaxies, the middle plots only those with morphological types of E or S0, while the lower panel plots the late-type galaxies (Sc and Irr). Note the difference in the degree of clustering between the middle and lower panels.
Dwarf galaxies, especially those that are not actively forming stars, are notoriously difficult to observe optically and are correspondingly absent in optical redshift surveys. Due to their low surface brightnesses, these galaxies are often purposefully excluded from "magnitude-limited" redshift surveys even though their integrated magnitudes might be substantially brighter than the quoted limit. An added complication arises because the total magnitudes for LSB objects are systematically underestimated in most existing galaxy catalogs (often by 1-2 magnitudes), so they often fail to be included in survey lists in the first place. These biases apply equally to nearby LSB dwarf galaxies and more distant LSB disk galaxies. The impact of these deficiencies in the optical redshift surveys on such things as the slope of the faint end of the galaxian luminosity function is hard to assess, but could be severe (McGaugh 1994; Ferguson & McGaugh 1995).
HI observations offer at least partial rectification of this problem. As long as a galaxy has substantial amounts of HI gas, it is relatively easy to observe at 21-cm. It is therefore no accident that some of the largest HI redshift surveys have targeted dwarf and/or LSB galaxies (Fisher & Tully 1975; Thuan & Seitzer 1979; Schneider et al. 1990; 1992; Bothun et al. 1986). Studies of the spatial distributions of low luminosity galaxies have relied heavily on the use of HI redshifts (Thuan et al. 1991; Eder et al. 1989; Salzer et al. 1990; Schombert et al. 1994). Such studies are important for answering questions about the details of the formation of the large-scale structures, since the relative clustering properties of high- and low-luminosity galaxies is an important discriminant between competing theories (e.g., White et al. 1987). Of course, constructing complete samples of dwarf and LSB galaxies remains a significant challenge, but HI observations hold out the promise of obtaining fairly complete redshift information once the appropriate survey lists are available.
A final advantage of 21-cm observations is that they provide fairly complete information about the global gas content in the galaxies observed. This is due, in part, to the fact that the beam size at 21-cm is typically larger than the size of the optical disk. Therefore, HI emission profiles contribute not only accurate redshifts, but also measures of total HI content and velocity widths. This latter quantity is useful for applications of the Tully-Fisher relation (see below).