6.2.6 New Terminology and a New Connection Between BL Lac Classes

There is now a viable interpretation of the observational data that is unrelated to jet collimation (Giommi and Padovani 1994; Padovani and Giommi 1995a). It is based on a single population of BL Lac objects characterized by a wide range of multiwavelength spectral shapes, with bolometric luminosities peaking at infrared/optical wavelengths for most RBL and at ultraviolet/X-ray for most XBL, as is observed (Giommi et al. 1995; Sambruna 1994). Before discussing this scenario further, we discuss new terminology that distinguishes clearly between the selection method used to find BL Lacs and the spectral characteristics of the sources themselves.

The new terms are motivated by the fact that the division of BL Lac objects into RBL and XBL can get garbled or confused. Strictly speaking, those terms are based on selection band rather than intrinsic physical properties, in which case some BL Lac objects already qualify as both RBL and XBL (e.g., Mrk 501), and many more ``double agents'' will appear as deeper radio and X-ray samples become available (Elvis et al. 1992). This leads to awkward terminology such as ``XBL-like RBL'' and ``RBL-like XBL'' when discussing the strong differences between the broad-band spectral properties of RBL and XBL in the historical samples. Since we wish to explore these differences - notably the peak luminosities in the infrared/optical or ultraviolet/X-ray, respectively - it is useful to define categories in terms of the ratio of X-ray to radio flux (Ledden and O'Dell 1985).

Padovani and Giommi (1995) have suggested dividing RBL-like and XBL-like objects into LBL and HBL, respectively (for ``low-energy cutoff BL Lacs'' and ``high-energy cutoff BL Lacs''), according to whether _rx (between 5 GHz and 1 keV) is greater (LBL) than or less (HBL) than 0.75. These names reflect the actual spectral characteristics of the two types of BL Lacs, allowing the terms ``XBL'' and ``RBL'' to be reserved strictly for sample membership (which is well-defined). BL Lac objects like Mrk 501 or OJ 287 which appear in both radio- and X-ray-selected samples can be uniquely categorized as HBL or LBL, respectively. Most XBL are HBL (OJ 287 is an exception) and most RBL are LBL (Mrk 501 is an exception). The lack of precision of ``high'' and ``low'' reflects the possibility that the synchrotron peak in BL Lacs occurs at a wide range of wavelengths, perhaps not fully probed in current samples, from infrared through X-ray.

The essence of the argument of Padovani and Giommi (1995a) is that X-ray selection favors objects with a peak at ultraviolet/X-ray wavelengths and thus finds fewer with peak at infrared/optical wavelengths. If radio rather than X-ray surveys are unbiased (because the radio emission does not ``know'' the wavelength of the peak luminosity) then HBL are relatively rare, about 10% in the 1 Jy plus S4 plus S5 radio-selected samples (Padovani and Giommi 1995). In effect, Giommi and Padovani take the opposite approach from Maraschi et al. (1986), assuming radio selection rather than X-ray selection is unbiased, and they find the opposite result: in complete contrast to the accelerating jet picture, they conclude that HBL constitute a minority of the BL Lac population.

Specifically, Giommi and Padovani (1994) argue that HBL outnumber LBL at a given X-ray flux, even though they are intrinsically less numerous, because the two classes sample different parts of the BL Lac radio counts (Giommi and Padovani 1994). As a consequence of their higher f_x / f_r ratios, HBL have lower radio fluxes (~ 10 mJy) and since fainter objects are more numerous than brighter ones (the radio counts are rising), their surface density is higher. Stated differently, X-ray surveys sample the BL Lac radio counts at low fluxes and mostly detect the ~ 10% of objects with high f_x / f_r ratios. This holds down to quite faint X-ray fluxes, well below the ROSAT deep survey limit, below which the fraction of LBL should increase slowly and eventually dominate by a factor of 10 (Padovani and Giommi 1995a). The Giommi and Padovani hypothesis explains most of the properties of HBL - e.g., X-ray luminosity function, X-ray number counts, and radio flux distribution - using only the observed properties of RBL (and no free parameters).

There are at present insufficient data to determine whether the number ratio of HBL to LBL (integrated over all luminosities) is ~ 0.1, as Giommi and Padovani conclude, or roughly the opposite! One persistent objection to the Giommi and Padovani picture is that the polarization characteristics of LBL and HBL differ in a way not obviously explained by a change in break frequency. Specifically, the LBL have higher polarization which varies more in both degree and position angle. According to Giommi and Padovani, the HBL are intrinsically less luminous - less extreme - so they should have lower polarization but why the polarization angle is more stable is not obvious in their picture.

The beaming picture, comparing BL Lacs and FR I LFs, suggests that HBL are more numerous than LBL by at least a factor of 3. It does seem intuitive that more luminous objects are more rare, i.e., that luminosity functions rise to low luminosities (which they do for all known galaxies, active or not). Thus one could ask how the bolometric luminosity functions of LBL and HBL compare; i.e., what the relative number densities are, at least in the range of overlapping luminosities. Using data from Giommi et al. (1995), we formed a close approximation to the bolometric LF using the luminosities at the peak of the emission in L. Figure 18 shows this pseudo-bolometric LF for the 1 Jy LBL (excluding the two HBL in the 1 Jy RBL sample) and the EMSS HBL (this sample includes no LBL). This is a bivariate LF so it does not compensate for the selection effects inherent in the 1 Jy and EMSS samples. The HBL appear to be more numerous in the range of overlap with number densities systematically above the LBL points. We note that the EMSS is effectively much deeper than the 1 Jy survey (its equivalent radio flux limit is ~ 1 mJy; Padovani and Giommi 1995a). We also note the uncertainties are very large. This kind of figure will be very useful to re-derive when larger samples of LBL and HBL are available.

Figure 18. The ``approximate bolometric'' luminosity function for 1 Jy LBL (filled circles) and EMSS HBL (open squares). This is the bivariate peak luminosity function, where the peak luminosity has been defined as the maximum in L from the data of Giommi et al. (1995) and no evolution has been assumed. Error bars correspond to 1 Poisson errors (Gehrels 1986). The HBL appear to be more numerous in the range of overlapping luminosity but selection effects inherent in the 1 Jy and EMSS samples are already present in these luminosity functions.

Clearly one would like to select samples in an unbiased way. Somewhere between the wavelengths where LBL and HBL have their spectral peaks, they must have roughly comparable fluxes. The optical has to be less biased than the X-ray or radio band, and indeed the optically brightest BL Lacs in the 1 Jy and Slew Survey samples have comparable optical magnitudes (Stickel et al. 1991; Perlman et al. 1995). The relative numbers of LBL and HBL in an optically-selected sample, then, should reflect their relative numbers globally, at least better than radio or X-ray surveys do. Unfortunately, the best optically-selected sample, the complete PG sample, has only 6 BL Lac objects (Green et al. 1986; Fleming et al. 1993). Of these, 1/3 are LBL and 2/3 are HBL, but the statistics are obviously poor. While formally there are more HBL than LBL in the PG sample, the observed ratio actually agrees well with what the Giommi and Padovani scenario would predict in the optical band (Padovani and Giommi, in preparation). Larger optical samples of BL Lac objects are important for addressing the fundamental differences between LBL and HBL.