THE BALDWIN EFFECT IN COMPLETE OPTICALLY SELECTED SAMPLES OF QUASARS

From an analysis of the equivalent widths of the C IV and Mg II lines in the largest available data set of unbiased, complete samples of optically selected quasars, we have found that the least-squares fit of the equivalent width versus absolute magnitude gives the following relations: EW(C IV) is-proportional-to L1550(-0.13 +/- 0.03) and EW(Mg II) is-proportional-to L2798(-0.08 +/- 0.03) or, equivalently, L(C IV) is-proportional-to L1550(0.87 +/- 0.03) and L(Mg II) is-proportional-to L2798(0.92 +/- 0.03). For both lines the slope of the correlation between the equivalent widths and the luminosities is very shallow. In particular, the slope for the C IV line is significantly different from the previous `canonical' value for the slope of the `Baldwin effect' in quasars: EW(C IV) is-proportional-to L1550(-(0.3-0.4)). While our data set is composed of optically selected quasars, the latter result was mainly based on the analysis of flat-spectrum, radio-loud quasars. It is likely that the enhanced optical variability of radio-loud quasars may be the fundamental factor in producing the observed difference between the two correlations. Our slope for the C IV line is in agreement with that obtained by Kinney et al., in their study of the Baldwin effect which was principally based on IUE data. Our normalization, however, is about a factor of 2 smaller than theirs. Possible selection effects which may have biased the normalization of the Kinney et al. result are suggested. In the past, various models (with different degrees of success) have been suggested to explain the 'old' Baldwin relation. We feel that the data presented and discussed here should be taken as the basis for new theoretical attempts to determine the best physical parameters to reproduce the `new' slope of the Baldwin effect for optically selected quasars.