Emission-Line and Continuum Properties of 92 Bright QSOs: Luminosity Dependence and Differences between Radio-selected and Optically Selected Samples

We have analyzed the emission lines and (rest frame) UV continua in 92 QSO spectra (resolution 4-6 A which had been obtained for a recent study of Mg II absorption in the spectra of 103 low and intermediate redshift QSOs. The sample contains 42 radio-selected QSOs (of which 14 have "steep" radio spectra) and 50 which were discovered either by UV excess or by slitless spectroscopy. The radio-loud and radio-quiet subsamples have roughly the same distribution in both redshift and absolute luminosity. In general, the spectra cover the rest wavelength interval from below the C III] λ1909 emission line to just above the Mg II λ2800 emission line. For many of the objects information is also available for the C IV λ1549 line. The main conclusions are as follows: 1. An examination of their mean emission-line equivalent widths and continuum properties shows that there is no significant difference between the QSOs discovered by the three different methods. Thus, the optically selected sample is not biased by the prominence of emission lines or optical continuum slope. In addition, with the possible exception of the equivalent width of the C III λ1909 emission line, there is no significant difference in the optical/UV spectra of radio-flat versus radio-steep QSOs. 2. Deconvolution of the blended C III] λ1909 feature indicates that the weaker component is almost certainly A1 III λ1859. In addition Si III] λ1892 occasionally makes a small contribution. The mean contribution of A1 III to the blend is 12% +/- 1%. 3. The broad shoulders of the Mg II emission line which have been sometimes attributed to Fe II are probably real features in the Mg II profile. 4. The distributions of the following quantities differ significantly (>2 σ) between the radio-loud and radio-quiet subsamples: W(Al III), I(C III])/I(Mg II), I(C III])/I(C IV), and I(Al III)/I(C III]). All other measured equivalent widths and intensity ratios are statistically indistinguishable between the two subsamples. 5. Significant correlations with continuum luminosity, M_lambda2200_ are found for the continuum ratio f_v_(λ1909)/f_v_(λ2800) and for the quantities W(Mg II), W(C III]), and I(C III])/I(Mg II). All of these correlations, in particular the Baldwin effect, are significantly stronger for the radio-selected sample than for the optically selected QSOs. 6. The C IV and C III] emission lines each have a mean blueshift relative to Mg II of 450 km s^-1^ while C IV] has a mean blueshift relative to C III] of 200 km s^-1^, The actual distribution of velocity differences is very broad and highly asymmetrical; the mode of the distribution is consistent with zero velocity difference and the net offsets are produced by high-velocity tails in the distributions. The velocity shifts are not correlated with QSO luminosity, but radio-quiet QSOs exhibit significantly larger values of z(Mg II) - z(C III]) than radio-loud QSOs. The differences summarized above are relatively subtle; otherwise, the continuum and emission-line properties of optically selected and radio- selected QSOs are remarkably similar. We note that the most significant correlations involve the C III] line which is density sensitive. However, we are unable to relate the observed regularities to any simple physical picture of the structure of QSOs.