Redshifted damped hydrogen Ly (DLA) lines are sometimes detected in the spectra of background quasars. They present neutral hydrogen column densities cm-2 similar to the value found at the optical radius of present-day galactic disks (Wolfe et al. 1986), which suggests that they arise in progenitors of present-day spiral galaxies.
Because the impact parameters necessary to produce such absorption lines are likely to be small, the galaxy gravitational potential may lead to significant gravitational lensing effects, such as amplification of the apparent luminosity of the background quasar, or even formation of multiple images.
In turn, these effects can induce severe biases in the samples of quasars and absorbers. First, impact parameters are increased because of lensing (the by-pass effect; see Smette et al. 1997, hereafter, SCS) thus decreasing the cross-section for DLA absorption. Second, background quasars which present DLA absorptions in their spectra also have their apparent luminosity increased because of lensing: hence, intrinsically faint quasars can enter apparent magnitude limited samples. As a consequence, these samples may show an over-representation of quasars that present DLA absorption systems: this amplification bias apparently increases the cross-section for damped Ly- absorption. These aspects can be evaluated in a statistical sense for a well-defined magnitude limited sample (cf. SCS; Bartelmann & Loeb 1996). For bright quasar samples, the amplification bias was found to dominate the by-pass effect, which leads to predictions that the number density of DLA systems and the cosmological density of neutral hydrogen (since DLA absorbers contain most of the H I of the Universe) may be significantly overestimated. Note that extinction by dust introduces other biases competing with the ones just described (Fall & Pei 1993; Perna et al. 1997; Bartelmann & Loeb 1998).
It must be noted however, that existing theoretical estimates are based on a series of simplifying assumptions, one of which being that the DLA absorbers are very similar to present-day disk galaxies. As a matter of fact, Le Brun et al. (1997) presented the first identifications of damped Ly absorbing galaxies at intermediate redshifts: in a sample of 7 DLA absorbers, 3 turn out to be spiral galaxies of various sizes and luminosities, 2 are compact objects and 2 are amorphous, low-surface brightness galaxies (an additional compact object appears to be responsible for a higher-redshift DLA system). This study showed that the population of galaxies giving rise to DLA systems is more heterogeneous than previously thought.
Independently, SCS found that lensing effects may be present in the IUE (Lanzetta et al. 1995) and Rao et al. (1995) Mg II surveys. In addition, in a follow-up of this latter survey, Rao & Turnshek (2000) find a surprisingly high value of at , as expected if lensing effects are important (SCS).
Here we consider the lensing properties of the absorbing galaxies detected by Le Brun et al. (1997). These observations were made using the WFPC2 camera onboard the Hubble Space Telescope, so that the absorbing galaxy candidates are clearly separated from the quasar images, and their characteristics (impact parameter, luminosity, morphology) are fairly well estimated.
The sample of quasars described here is not complete in any sense except that it contains all the DLA systems known at the time of writing the HST proposal (cycle 6). For illustration purpose only , we have evaluated the effects of lensing following SCS (exponential model plus Gaussian hole, or E+GH). These theoretical estimates are based on the observed magnitude and emission redshift of the quasar and on the DLA absorber redshift . Would the DLA sample be representative of a well-defined, complete magnitude limited sample, we find that the number density of DLA systems would have been overestimated by 18%. In other words, statistically speaking, one of the quasars in this sample would not have been observed if gravitational lensing was not taking place. In addition, the value of would have been over-estimated by 28%. We also find that the probability that one quasar in the sample presents multiple-imaging would be close to 50%. The values and are used throughout this paper.
© European Southern Observatory (ESO) 2000
Online publication: December 5, 2000