2006ApJ...646..730J

As large optical quasar surveys for DLAs become a reality and the study of star-forming gas in the early universe achieves statistical robustness, it is now vital to identify and quantify the sources of systematic error. Because the nature of optically selected quasar surveys makes them vulnerable to dust obscuration, we have undertaken a radio-selected quasar survey for DLAs to address this bias. We present the definition and results of this survey. We then combine our sample with the CORALS data set to investigate the H I column density distribution function f_HI(N,X) of DLAs toward radio-selected quasars. We find that f_HI(N,X) is well fitted by a power law f_HI(N,X)=k₁N^α1_^, with logk₁=22.90 and α₁=-2.18^+0.20_–0.26. This power law is in excellent agreement with that of optically selected samples at low N_HI, an important yet expected result given that obscuration should have negligible effect at these gas columns. However, because of the relatively small size of the radio-selected sample, 26 DLAs in 119 quasars, f_HI(N,X) is not well constrained at large N_HI, and the first moment of the H I distribution function, Ω_g, is, strictly speaking, a lower limit. The power law is steep enough, however, that extrapolating it to higher column densities implies only a modest, logarithmic increase in Ω_g. The radio-selected value of Ω_g=1.15^+0.37_–0.38x10^–3 agrees well with the results of optically selected surveys. While our results indicate that dust obscuration is likely not a major issue for surveys of DLAs, we estimate that a radio-selected sample of ~100 DLAs will be required to obtain the precision necessary to absolutely confirm an absence of dust bias.