Metal abundances have been estimated for less than a dozen absorption line systems. The main impediment is the scarcity of systems which show enough lines from enough ions to allow simultaneous estimates of both ionization and abundances. Difficulties are also encountered with inadequate spectral resolution, line blending, the existence of multiple subcomponents, and slight differences in the velocity distributions of the different ions in a given system.
Published abundance estimates range from near solar, to about 0.01 of solar abundances. These estimates are sufficiently accurate to show that abundances do differ, but they do not allow a determination of the full range or distribution. Moreover they are unlikely to be fully representative of all metal line systems because estimates are usually made for only those systems which show the largest number of metal lines, a bias which will lead to above average abundances.
Metal line systems are sufficiently common that only a few percent can occur within the Holmberg radii of galaxies. We should expect that abundances will usually be at levels anticipated for the outer regions of galactic haloes at generally high redshifts. Values around 0.01 solar might be appropriate, depending on the history of enrichment. An evaluation of the distribution of abundances as a function of epoch can be expected in the coming decade. These results will be of extreme importance in the determination of the sequence of metal enrichment during and prior to galaxy formation. For example, if it is found that metal systems with high abundances are common at the largest redshifts (z 4), then enrichment by Population III stars would be indicated for theories which have galaxies collapsing at later epochs.
Attempts to set upper limits on the abundances of the Ly- systems are thwarted by our lack of precise knowledge of their level of ionization, and hence their total column densities. Two approaches have been followed.
Ly- systems typically have neutral hydrogen column densities of N(HI) = 1014 cm-2 (Sargent et al. 1980; Carswell et al. 1987). Even if these systems have very high ionizations corresponding to total column densities of logN = 19, it is observationally difficult to set abundance limits which are substantially below those found in the metal line systems. Norris, Hartwick and Peterson (1983) attempted to overcome this problem by summing portions of the spectra of two QSOs, shifted to align the expected positions of metal lines at the redshifts of some 65 Ly- systems. They did not detect CIV or NV, but OVI was marginally present. Searches in higher quality spectra of other QSOs by Sargent and Boksenberg (1983) and Norris and Peterson (1986) have failed to confirm the OVI detection, leading to typical upper limits on the abundances of metals of about 0.03 solar, a limit which is sensitive to the assumed level of ionization.
Impressive upper limits on the metal abundances have also been determined for two Ly- systems which have logN(HI) = 17, amongst the largest N(HI) known for Ly- systems. In both cases the limits are about 0.001 solar for an assumed level of ionization (Sargent and Boksenberg 1983; Chaffee et al. 1986). The basic assumption is that these rare systems with logN(HI) = 17 should have about the same level of ionization as the typical systems with logN(HI) = 14. The large N(HI) then corresponds to large total column densities needed to justify the low abundances. Chaffee et al. quantify this argument by assuming that the Ly- systems are in pressure equilibrium with a hot low density intergalactic medium. All clouds at a given redshift should then have about the same density and ionization. It follows that the logN(HI) = 17 systems will be about 1000 times the size of the typical Ly- systems, even though they are only about 1% as common. Since detection probability is proportional to cross-sectional area, the logN(HI) = 17 systems must then represent only 1 in 108 of Ly- absorbing clouds, and they would have 109 times the mass of a typical cloud. It is then far from obvious that these should be considered typical clouds.
Nevertheless these results are extremely important because they show that some systems probably do have low abundances. This follows from the observation that some Ly- and some metal line systems do have sizes in excess of 1 kpc, and it is reasonable to assume that the same applies to those systems with logN(HI) = 17, which are intermediate in N(HI).
In summary, there is no direct observational evidence showing that the typical Ly- system has a metal abundance which is well below the range shown by metal line systems, but there is at least a 3 order of magnitude range in metal abundances for the narrow line systems.