The search began with a low-resolution ( = 10 Å) survey for absorption features in the Ly- forest that were candidates for `Ly- disks.' A total of 68 QSOs with emission redshifts, zem > 2.25 were observed with the Lick 3-m telescope. The results, reported in Wolfe et al. (1986) and Wolfe (1986b), are that 47 out of 476 statistically significant absorption features qualify as candidate `Ly- disks', with W(Ly-) 5 Å. Using accurate follow-up spectroscopy described below we have determined the nature of 31 of the candidates. We find that 14 of the Ly- features are not damped. What appeared as a single strong feature at low resolution, breaks up into a series of multiple narrow velocity components at high resolution. The velocity intervals spanned by the components range from 1000 to 3000 km s-1. In addition the equivalent width distribution of the multi-component systems peaks near the 5 Å threshold. The remaining 17 candidates are in fact damped Ly-. Accurate spectroscopy shows that these Ly- profiles do not break up at higher resolution. Rather, the observed features are well fit with Voigt damping profiles. Moreover in each case, damped Ly- is associated with narrow low-ion absorption lines at the same redshift. In contrast to the multi-component systems, the damped absorbers dominate the equivalent-width distribution above W(Ly-) 10 Å.
The detection of so many damped Ly- systems is surprising. Voigt profile fits to the Ly- features show that 15 out of the 17 damped systems have N(HI) 2 x 1020 cm-2, the HI isophote out to which spiral galaxies have been extensively studied (Bosma 1981). By contrast, no more than 3 disks are predicted for (a) the redshift path (z = 55) of the survey, and (b) the cross-sectional radius corresponding to this N(HI). The implication of this discrepancy is that the HI radius R(HI) 3.5 x RHo (RHo is the Holmberg radius), if there is a one to one correspondence between the damped systems and present-day spiral galaxies. Another possibility is that the high detection rate is caused by an inflated population of small objects, such as dwarf galaxies, at large redshifts.
That a particular class of redshift system occurs along the line-of-sight more frequently than predicted for intervening galaxies is nothing new. Every known class of QSO absorption system shares this property. The CIV systems (Young, Sargent and Boksenberg 1982), the MgII systems (Lanzetta, Turnshek and Wolfe 1987), and the Ly- forest systems (Sargent et al. 1980) all occur more frequently per unit redshift interval than the damped systems. Thus it is entirely possible that the damped systems lie on the tail of the N(HI) distribution characterizing any one of these populations. But the evidence indicates otherwise. For example, the frequency distribution of Ly- equivalent widths above 2 Å has been derived from the Lick data (cf Wolfe 1986b), and indicates a change in slope above W(Ly-) 10 Å. Below this equivalent width, the distribution is indistinguishable from the exponential Ly--metal distribution proposed by Sargent et al. (1980) to describe Ly- lines associated with CIV and perhaps MgII absorption systems. Above 10 Å, where the damped systems dominate, the distribution is noticeably flatter, indicating a different parent population. The evidence for a population shift is further supported by the kinematic structure of the absorbers. The sequence in Ly--metal equivalent widths is actually a sequence in velocity spread, with the weak features consisting of one or two narrow components distributed over a small velocity spread, and the stronger systems made up of many components distributed over much larger velocity spreads. It is also likely that the total N(HI) in these systems is correlated with velocity spread, owing to the correlation of number of components with the total width of the blended feature. On the other hand, the sequence of the damped Ly- equivalent widths is independent of velocity spread, and solely reflects a sequence in N(HI). In this case N(HI) is uncorrelated with velocity spread. Therefore the evidence suggests that the multi-component systems originate by a different mechanism than the damped systems.