7. RELATION OF QUASARS AND BL LAC OBJECTS

The high-redshift members of any flux-limited sample tend to be systematically more luminous than the low-redshift members because of the induced correlation of luminosity and redshift. Among BL Lac objects, such differences have been interpreted as evidence for a ``true'' (Type 0) BL Lac class at low redshift and a ``quasar-like'' (Type 1) BL Lac class at high redshift (Burbidge and Hewitt 1989). A particular motivation for this division (Antonucci 1993) is the fact that broad emission lines, occasionally with large equivalent widths, have been seen in some high-redshift but not low-redshift BL Lac objects. (Here ``high-redshift'' essentially means the distant half of the 1 Jy RBLs, which extend to z ~ 1; few members of any complete XBL sample have z 0.5. In this section we review whether existing data support and/or require a bifurcation via redshift for BL Lacs.

One of the defining features of BL Lac objects is their weak or absent emission lines. For the complete 1 Jy radio-selected sample (Stickel et al. 1991), BL Lacs are defined in part by rest-frame equivalent widths of optical emission lines less than 5 Å. The EMSS and Slew Surveys are defined in a similar way, although using the observed equivalent widths (Stocke et al. 1991; Perlman et al. 1995). With this equivalent width limit, there can in principle be cross-over objects - for example, for fixed emission-line luminosities, the equivalent widths can change when the continuum varies.

Emission line types and strengths are central to the question of whether nearby and distant BL Lacs constitute distinct populations. The nearby BL Lacs, which are associated with nearby ellipticals (typically at z 0.2) with typical galaxy absorption spectra, sometimes have weak narrow emission lines (typically [O III]), while the more distant BL Lacs, around which host galaxies have never been detected, sometimes have weak but broad emission lines (typically Mg II; Stickel et al. 1991, 1993). Observational limitations (and sometimes habits), however, militate against uniform information on emission line properties across the full redshift range. In particular, to detect broad Mg II in the nearby BL Lacs requires high signal-to-noise ultraviolet spectra, and to detect [O III] in the distant BL Lacs requires fairly red spectra.

With available data we first ask whether BL Lac objects have line luminosities comparable to or clearly distinct from FSRQ, for samples selected in the same way (therefore, the 1 Jy BL Lacs and the 2 Jy FSRQ, both selected at high radio frequencies). Figure 11 shows [O III] line luminosities as a function of redshift, which removes observational selection effects due to spectral bandpass and at the same time effectively matches objects for luminosity in the selection band (which should be unbiased with respect to the optical line emission). We found [O III] data in the literature for all the 2 Jy FSRQ with z < 0.7 and for eight of the twelve 1 Jy BL Lacs sources with certain z < 0.5; the other four BL Lac objects have much weaker [O III] lines (Stickel et al. 1993).

Figure 11Figure 11 shows that the [O III] luminosities of BL Lac objects are systematically lower than those of FSRQ. Inclusion of upper limits for BL Lacs, were they available, would exacerbate the difference. There are not enough data points, however, to determine whether the two classes actually have distinct narrow-emission-line properties.

The one BL Lac with quasar-like [O III] luminosity is PKS 0521-365, a nearby low-luminosity object which had been previously classified as a BL Lac (Véron-Cetty and Véron 1993) but which was not included in the 1 Jy BL Lac sample (Stickel et al. 1991) because the equivalent width of two lines was larger than the 5 Å limit (specifically, W_H 9 Å and W_{[O II]} 7 Å; this object also has broad H; Ulrich 1981; Stickel and Kühr 1993). Figure 11Figure 11 suggests that PKS 0521-365 has BL Lac-like line luminosity but qualifies as a quasar because its continuum emission was unusually faint (exactly as found earlier by Ulrich 1981) when Stickel et al. classified it.

Broad emission lines are perhaps more crucial to the comparison of high-redshift BL Lac objects and quasars. We collected from the literature all available Mg II 2798 fluxes for the 2 Jy FSRQ and the 1 Jy BL Lacs, plotted in Fig. 19 versus redshift to facilitate comparisons at similar luminosities and observed wavelengths. Roughly 35% of the 2 Jy FSRQ and 30% of the 1 Jy BL Lacs are represented in the Figure, including six of the seven BL Lac objects with z > 0.7. Additional data for the FSRQ would be relatively easy to obtain, but for the BL Lacs, which by definition have very weak spectral features (in fact ~ 1/3 do not have a firm redshift determination), spectra with higher signal-to-noise ratios are needed. Because of the low mean redshift of BL Lac samples, the Mg II line usually lies in the ultraviolet, accessible at the required signal-to-noise only with HST.

Figure 19. The available Mg II emission line luminosities for 1 Jy BL Lac objects (filled circles) and 2 Jy flat-spectrum radio quasars (open squares), plotted versus redshift so that comparisons at similar luminosity (which is well correlated with redshift in flux-limited samples) and/or observed wavelength are possible. BL Lacs and FSRQ line luminosities differ by more than one order of magnitude up to z ~ 0.8, above which some of the high-redshift BL Lacs have line luminosities comparable to the lower-luminosity quasars.

In contrast to the [O III] plot, Fig. 19 shows that a few 1 Jy BL Lacs have Mg II luminosities as high as those of the weaker-lined FSRQ. In particular, two BL Lacs have L_{Mg II} > 3 x 10⁴³ erg s^-1 (the luminosity for the faintest FSRQ), namely PKS 0537-441 and B2 1308+326, both at z 0.7. The equivalent width of Mg II in PKS 0537-441 was on at least one occasion larger than 5 Å (Wilkes 1986), though not when the 1 Jy sample was being defined. This object has also been suggested as a lensing candidate, meaning it might actually be a quasar, although recent observations do not confirm this (see Sec. 7.2.3). B2 1308+326 may have an FR II radio morphology (Kollgaard et al. 1992), while its VLBI polarization properties seem to be more similar to those of quasars than of BL Lacs (Gabuzda et al. 1993).

Apart from these two cases, the Mg II luminosities of BL Lac objects are clearly lower than for FSRQ across the entire redshift range sampled. Still, we can not exclude a continuous distribution of line luminosities from BL Lacs to quasars, which could result simply from the definition of BL Lacs in terms of their low equivalent widths. In particular, the radio selection in the 1 Jy and 2 Jy samples is similar, so the radio luminosity distributions of these BL Lacs and FSRQ at a given redshift must be similar, suggesting their total luminosities may be comparable. Division in terms of equivalent width would then translate into the observed division in line luminosity. It remains to be seen if this explanation is viable quantitatively when more line luminosities become available.

The apparent difference between the evolutionary properties of radio-selected BL Lacs (V / V_max = 0.60 ± 0.05) and X-ray-selected BL Lacs (V_e / V_a = 0.36 ± 0.05) has been ascribed to contamination of the RBL sample from strongly evolving FSRQ (Morris et al. 1991). The comparison of line luminosities in Figs. 11 and 19 shows that RBL are clearly not quasar-like for z 0.5. If we divide the 24 1 Jy BL Lac objects having certain redshift information at z = 0.5, there is no evidence that the V / V_max distributions or mean values for the low- and high-redshift subsamples are different, as shown by a KS and Student's t-test, respectively. More simply, there is no significant correlation between V / V_max and redshift in the 1 Jy sample. For the five objects with FR II or FR I/II morphology (Kollgaard et al. 1992), which might be considered the most quasar-like, < V / V_max > is also not significantly different from the < V / V_max > for the rest of the sample. In short, the evolutionary properties of BL Lac objects do not depend on redshift and do not indicate contamination by strongly evolving quasars.

A number of other results suggest that low- and high-redshift BL Lacs are more alike than the latter are like FSRQ. In particular:

(1) The distributions of extended radio power for low- and high-redshift BL Lacs are not significantly different, while those of high-redshift BL Lacs (the 1 Jy with z > 0.3; the largest measured redshift is z = 1.048; Stickel et al. 1991) and FSRQ (the 2 Jy sample from Wall and Peacock 1985) differ at the 99.9% confidence level (Padovani 1992b).

(2) The ratio between compact and extended radio emission is not significantly different for low- and high-redshift BL Lac objects. This is in spite of the possibility that some high-redshift BL Lacs are FR IIs with low-excitation optical spectra (Sec. 5.4), and therefore might have systematically different radio properties.

(3) The X-ray spectral shapes of quasars and of BL Lac objects observed with the Einstein Observatory (Worrall and Wilkes 1990) and with ROSAT (Urry et al. 1995) are systematically different, with the FSRQ having flatter spectral indices. Even BL Lacs with faint but broad emission lines have X-ray spectra more similar to the other BL Lacs than to the FSRQ (Worrall and Wilkes 1990), although those with the broad lines tend to have flatter-than-average (i.e., more FSRQ-like) X-ray spectra (Urry et al. 1995).

(4) The characteristic VLBI polarization structure of BL Lac objects, which implies a magnetic field perpendicular to the jet axis, is quite different from that of quasars, which have magnetic field parallel to the jet (Gabuzda et al. 1992), with one possible exception (the 1 Jy BL Lac B2 1308+326; Gabuzda et al. 1993). Furthermore, the milliarcsecond polarization structures of low- and high-redshift BL Lacs, divided at z = 0.3 as suggested by Burbidge and Hewitt (1989), are indistinguishable. It is perhaps possible to create the distinction between quasars and BL Lacs if the latter are systematically more aligned with the line of sight because of the sensitive dependence of polarization on aspect at small angles (Gopal-Krishna and Wiita 1993).

We have asked two simple questions: (1) whether there is any evidence requiring separate high- and low-redshift BL Lac populations, and (2) whether the evidence shows that high-redshift BL Lacs are similar to quasars. We believe that, based on presently available data, the answer to both questions is ``no.'' BL Lacs selected on the basis of their equivalent width represent a fairly homogeneous class, with no strong differences between objects with redshift below and above z ~ 0.3, z ~ 0.5 or other, or between objects with broad and with narrow emission lines. There may be a few intermediate objects, characterized by FR II radio morphology, high emission-line luminosities, unusually weak continuum or quasar-like VLBI polarization, but their small number is unlikely to affect unified schemes significantly.