3. The observed sample and its characteristics
The observed QSOs were essentially chosen from the WMFH sample, which is a set of BAL and non-BAL QSOs from the Large Bright Quasar Survey (LBQS, cf. Hewett et al. 1995), augmented by several BAL QSOs from other sources. The selection was achieved during the observations depending on the QSO observability (position on the sky) and magnitude (priority to the brighter objects). A priority was also given to the BAL QSOs with low-ionization features. Five objects observable in the southern sky were added: 3 BAL QSOs from the Hartig & Baldwin (1986, hereafter HB) sample (0254-3327, 0333-3801, 2240-3702), and 2 non-BAL QSOs from the LBQS (2114-4346, 2122-4231). Finally, an additional 7 true or possible gravitationally lensed optically selected QSOs (cf. the compilation by Refsdal & Surdej 1994) were included in the sample.
The final sample then consists of 42 moderate to high redshift optically selected QSOs (cf. Tables 2 & 3). It contains 29 BAL QSOs, 12 non-BAL QSOs, and 1 "intermediate" object (2211-1915, cf. WMFH). 8 of them are true or possible gravitationally lensed QSOs, including 2 BAL QSOs: 1413+1143 and 1120+0154 2.
Among the 42 optically selected QSOs, 36 are definitely radio-quiet while only 1 is radio-loud (2211-1915, the "intermediate" object) (Stocke et al. 1992, Hooper et al. 1995, Véron & Véron 1996, Djorgovski & Meylan 1989, Bechtold et al. 1994, Reimers et al. 1995). The 5 remaining objects (3 BAL QSOs and 2 non-BAL QSOs: 0333-3801, 0335-3339, 2154-2005, 2114-4346, 2122-4231) have apparently not been measured at radio-wavelengths. However, they are most probably radio-quiet too (Stocke et al. 1992, Hooper et al. 1995).
3.1. The low-ionization BAL QSOs
Approximately 15% of BAL QSOs have deep low-ionization BALs (Mg ii 2800 and/or Al iii 1860) in addition to the usual high-ionization BAL troughs (WMFH, Voit et al. 1993). These objects might be significantly reddened by dust (Sprayberry & Foltz 1992). They also possibly constitute a physically different class of BAL QSOs (Boroson & Meyers 1992).
While objects with strong low-ionization (LI) features are recognized as LIBAL QSOs by most authors, the classification of objects with weaker features is controversial. We therefore define three categories of LIBAL QSOs: strong (S), weak (W), and marginal (M) LIBAL QSOs. The strong and weak LIBAL QSOs in our sample were all considered and first classified as such by WMFH. The strong ones are 0059-2735, 1011+0906, 1232+1325 and 1331-0108; the weak ones are 0335-3339, 1231+1320, 2225-0534 and 2350-0045 (WMFH "a" parameter ). But the classification by WMFH is rather conservative and includes only clear LIBAL QSOs, while several authors have reported faint LIBAL features in a number of other objects. We classify the latter objects as marginal LIBAL QSOs. These are 0043+0048, 1246-0542 and 2240-3702 (HB), 1413+1143 (Hazard et al. 1984, Angonin et al. 1990), 1120+0154 (Meylan & Djorgovski 1989), and 1212+1445 (this work). The marginal LIBAL QSOs are characterized by very weak Mg ii and/or Al iii BALs. The asymmetry of the Mg ii or C iii] emission lines, when cut on the blue side, is also considered as evidence for marginal LIBALs. Note finally that line strengths may be variable in some objects and that weak LIBALs could have been observed only once (namely due to possible microlensing effects as suspected in e.g. 1413+1143; Angonin et al. 1990, Hutsemékers 1993).
The remaining BAL QSOs are classified as high ionization (HI) only, except 0903+1734 and 1235+0857 which are unclassified, the Mg ii line being outside the observed spectral range and no Al iii BAL being detected. These classifications are summarized in Tables 2 and 3. Note that most spectra available in the literature were carefully re-inspected to check for the consistency of the classification. Altogether, the strong, weak and marginal LIBAL QSOs constitute approximately 50% of our BAL QSO sample (but this is not representative of the actual proportion of LIBAL QSOs among BAL QSOs since priority was given to these objects).
3.2. The BAL QSO spectral characteristics
WMFH provide a series of spectral indices characterizing the absorption and emission features of BAL QSOs. For the absorption lines, they define the balnicity index (BI, in km s-1) which is a modified velocity equivalent width of the C iv BAL, and the detachment index (DI, unitless) which measures the onset velocity of the strongest C iv BAL trough in units of the adjacent emission line half-width, that is, the degree of detachment of the absorption line relative to the emission one (see also HB who first distinguish between detached and P Cygni-type BAL profiles). Estimates of BI are also given by Korista et al. (1993) for most objects of our sample, such that we adopt for BI an average of these values and those from WMFH. WMFH also provide "clever" half-widths at half-maximum (HW, in km s-1) and equivalent widths (EW, in Å) for the C iv, C iii] and Fe ii emission lines. For a more detailed definition of these indices, see WMFH.
For a few objects (0254-3327, 0333-3801, 2240-3702, and 1120+0154), some spectral indices were not provided. We therefore computed them using C iv spectra published by Korista et al. (1993) and Steidel & Sargent (1992). The spectra were digitally scanned, and the measurements done following the prescriptions given by WMFH. The measurements were also done for spectra of objects with published indices: a good agreement was found, giving confidence in our new values. For the C iii] and Fe ii emission lines, half-widths and equivalent widths were simply rescaled from those measured by HB. All these quantities are reported in Table 3.
Table 3. BAL QSO spectral characteristics
In addition, we have evaluated the slope of the continuum using BAL QSO spectra digitally scanned from the papers by WMFH, HB, and Steidel & Sargent (1992). After some trials, we realized that some spectra cannot be easily fitted with a single power-law continuum: the slope often breaks roughly near C iii], probably due to reddening and/or extended Fe ii emission (compare for example the spectra of 1246-0542 and 1442-0011 in WMFH). We therefore decided to fit the continuum blueward and redward of C iii], independently. The derived slopes and are given in Table 3, assuming a power-law continuum . The values of and are affected by large uncertainties (not smaller than 0.3), mainly due to the difficulty to accurately identify the continuum when the BALs are very large, when the Fe ii emission/absorption is strong, or when the Mg ii absorption is wide.
© European Southern Observatory (ESO) 1998
Online publication: November 9, 1998