|Annu. Rev. Astron. Astrophys. 1980. 18:
Copyright © 1980 by Annual Reviews. All rights reserved
In this section, we describe the general properties of the blazar class represented by Tables 1 and 2 in Section II. We begin with the characteristic strength and variability of the optical-infrared polarization followed by the wavelength dependence of polarization. The strong correlation of the optical polarization with photometric variability and a smooth optical-infrared continuum is discussed as are the general radio characteristics of the blazar class. We end this section pointing out correlations of the X-ray and radio properties with the presence of strong line emission.
GENERAL CHARACTERISTICS OF THE OPTICAL-INFRARED POLARIZATION
Strength and timescale of variability A primary property of the polarization is its variability; this adds richness to what can be learned from polarimetric studies, and provides job security for polarimetrists. The variations can be substantial and erratic on every time scale from a few hours on up. In addition, even for a single source the general type of variability can change. For instance, for the first year or so after its identification, OJ 287 showed polarization varying wildly in strength and angle. Since 1972, however, its polarization, although still strong, has nearly constant position angle, with the radio emission polarized at the same angle (e.g. Rudnick et al. 1978). In another case, 3C 454.3, the strong variable polarization observed a decade ago has currently vanished, and the source has settled to only 1-2% polarization (Moore & Stockman 1980). The best we can do then is to characterize the behavior of specific objects from the known data, and recognize that it may not persist.
The strength of polarization, whose maximum and minimum recorded values are given in column 7 of Table 1, has exceeded 30% in three objects (PKS 0735+178, OJ-131, OJ 287) and 20% in another nine. While it is often the most extensively measured objects that yield the highest polarization values, there are also some well-studied examples that never show very high polarization. Another polarimetric distinction that can be made among the blazars is by the range over which the position angle varies. Angel et al. (1978) found that among 12 BL Lac objects monitored polarimetrically, 5 appeared to have a definite preferred angle. Making use of all the published data referenced in Table 1, we can distinguish two types of position angle variability. One group shows no tendency for preferred angle, and data span all points of the compass. These objects are nearly all distant and extremely bright: the 3C sources 279, 345, 446, and 454.3 (during outburst), AO 0235+164, B2 1308+326, PKS 0735+178, OM 280, OJ 287 (during outburst), BL Lac. The group of objects for which repeated measurements show a restricted range of angles are 3C 66A, NGC 1275, 0422+004, 0752 + 258, OJ-131, OJ 287 (recent years), Mkn 421, ON 325, 1400+162, AP Lib Mkn 501, 3C 390.3, 3C 371. These are generally less luminous objects. In Section VI, we consider the possibility that the two types of variability may reflect the difference between sources with jets pointed straight at us and those that are inclined to the line of sight.
We next turn to the time scale for changes in the polarization. Polarization measurements afford a good method for exploring variability on time scales less than a day since the linear polarization Stokes parameters Q and U change by amounts comparable to the total intensity parameter I. Rather small changes of Q / I and U / I are detected differentially to a precision limited only by photon statistics. Small changes of the total intensity I over periods of hours can only be detected by reference to standard stars, and require excellent conditions and careful treatment of extinction. Thus photometry is rarely limited by photon statistics except for the faintest objects. Variations on a time scale of a few days were recognized from the first discovery of strong polarization in the violently variable QSOs, and substantial night to night changes in the polarization of 3C 345 were measured in 1967-1968 by Kinman et al. (1968) and Visvanathan (1973a). OJ 287 also showed large nightly fluctuations in the early data of Dyck et al. (1971) and Kinman & Conklin (1971). Fluctuations on these time scales were also well established from photometric data. The synoptic study of 12 BL Lac objects made by Angel et al. (1978) with repeated measurements of the same objects during each night found that erratic variations on a time scale of a day are not uncommon in several objects but that variations from hour to hour are small. When significant hourly changes are detected they are monotonic, with dP / dt no larger than required to explain the daily variations. Thus it appears that the power spectrum of variability falls sharply for frequencies higher than 1 d-1. These results have been confirmed by more recent work on BL Lac (see Section II). The variability of B2 1308 + 326, a source like AO 0235 + 164 that reaches extremely high brightness in outbursts of months duration, has recently been studied and is also discussed in Section II. The polarization changes very strongly (more even than BL Lac) on a time scale of 1 day in the rest frame of the source, but again there is no compelling evidence of much more rapid changes.
Not all strongly polarized sources show detectable variability; a few appear to be essentially constant in both strength and angle. OJ 287 has been virtually constant with 8% polarization at 145° over a two year baseline (Moore & Stockman 1980), 3C 66A shows slight changes in polarization, but virtually all data points lie within 12±2% in strength and 30±16° in angle. The double lobe radio source 1400 + 162 (see Section II) has polarization 12% at close to 90° for all but one measurement.
Wavelength dependence of polarization Another property of the polarization of considerable interest is its dependence with wavelength across the optical-infrared spectral range. It has long been known, particularly from the work of Visvanathan (1973a), that the polarization is usually wavelength independent at least across the optical spectrum. Sometimes, though, some objects do show small but significant rotations in position angle of up to 15° (Rieke et al. 1977, Moore et al. 1980), or changes in strength of polarization from blue to red (Kikuchi et al. 1976, Nordsieck 1976). The smoothness of the continuous emission from the optical to the infrared suggests that the infrared emission (where most energy is released) should share the polarimetric properties of the visible spectrum. This has recently been found to be the case. Simultaneous or nearly simultaneous observations of polarization over the range 0.4-2.2 µ have been reported for AO 0235, PKS 0735 + 178, OI 090.4, OJ 287, B2 1308 + 326, 3C 345, and BL Lac, in papers by Knacke, Capps & Johns (1976, 1979), Rieke et al. (1977), Puschell et al. (1979), Moore et al. (1980), and Puschell & Stein (1980). Generally the polarization is found to be the same in both strength and angle, but there are observations of substantial rotation of position angle (e.g. 35° in OI 090.4) from the optical to the infrared. Until recently, no marked color dependence in the strength of polarization had been found, but some recent observations of BL Lac (Puschell & Stein 1980) and 3C 345 (Knacke et al. 1979; Section II) showed a strong increase to the ultraviolet in the former, to the red in the latter.
CORRELATION OF STRONG OPTICAL POLARIZATION WITH OPTICAL-INFRARED CONTINUUM PROPERTIES There seems to be virtually a one-for-one correspondence between the occurrence of polarization and large fluctuations in the optical flux. Essentially every object in Table 1 that has been monitored extensively shows variations larger than a magnitude. Conversely, almost all objects known to exhibit large fluctuations in brightness are found to be strongly polarized. For instance, in Usher's (1975) table of variable QSOs and BL Lac objects, there are 13 objects with B > 2 magnitudes. Eleven of these are in common with Table 1, one is in the south and has not been measured for polarization, and only one, 3C 323.1, is not strongly polarized. The correspondence is not perfect and a few variable objects are found not to be polarized (e.g. 3C 120 with B = 1.7). Selection effects play some role in this correlation, as some outstanding examples were originally picked out for polarimetric study after their variability was discovered. 3C 446, the first QSO found to be polarized, was measured by Kinman et al. (1966) after variability was discovered; the identification of BL Lac as an inverted-spectrum radio and optical variable (MacLeod & Andrew 1968, Schmitt 1968) led to the discovery of optical polarization. Nevertheless, many high polarization objects were identified from radio surveys before their optical variability was studied.
A second striking correlation of optical polarization is with the shape of the optical-infrared continuum. The fact that the optical spectrum of polarized objects tends to be steeper than that of most emission line QSOs has been recognized for some time (e.g. Stein, O'Dell & Strittmatter 1976). Now that good photometry from 0.3-10µ has been obtained for many Seyfert nuclei. BL Lac objects, and QSOs, we find a division can be made between those that can be well represented by a simple power law or spectral shape, and those that have complex shapes (Rieke & Lebofsky 1979). The former class is virtually coincident with the class of highly polarized objects we are considering. The recent spectra of bright QSOs by Neugebauer et al. (1979) illustrate this very well. Picking out the straight spectra from the sample of 30 objects, one finds they are just the group of classical violent variable QSOs (3C 279, 3C 345, 3C 446, 3C 454.3). 3C 68.1 has the only other smooth spectrum in the sample, and its polarization has not yet been measured. Puschell (1980) remarks that there are no objects known showing correlated variability in the optical and infrared that are not highly polarized. Thus 3C 120 fails as a blazar not only in lacking polarization, but also in showing infrared radiation which does not track the optical on short time scales (Rieke & Lebofsky 1979).
RADIO EMISSION OF BLAZARS An outstanding property we see from Table 1 is that all entries except PHL 5200 show radio emission, a correlation supporting the idea that nonthermal processes are responsible for the polarization. Historic interest in radio sources has biased most polarization surveys toward the discovery of radio-loud objects. Nevertheless the correlation of radio emission with the optical properties of polarization, strong variability, and smooth spectra is not a result of selection effects. Among emission line objects large numbers of radio-quiet QSOs searched by Stockman & Angel (1978) and Stockman (1978) yielded PHL 5200 as the only strongly polarized radio-quiet QSO. Seyferts, which are nearly all very weak radio sources, in the study by Maza, Martin & Angel (1980), showed exclusively dust type polarization, with only one or two possible exceptions (see Section IV).
Since nearly all known BL Lac objects are from radio surveys (Condon 1978) the observational bias against radio-quiet BL Lacs is severe. Certainly in one case, I Zw 186, the unusual bright featureless nuclear continuum was discovered by Zwicky (1966) while observing compact galaxies, and optical variability was then found by Oke et al. (1967) and Sandage (1967), still before weak radio emission was detected (Altschuler & Wardle 1976). Mkn 501 is another object noticed first by its unusual featureless spectrum, in a survey of Markarian galaxies by Khachikian & Weedman (1974). It then proved to have a flat radio spectrum and strong variable polarization. X-ray emission is not well correlated with radio emission at the lower frequencies of most radio surveys (see below), so one might hope that X-ray surveys would find radio-quiet objects. All-sky X-ray surveys have discovered at least one source (2155-304) not previously recognized, but this again has proven to be a radio source. From these data, it seems likely that if radio-quiet BL Lac objects exist, then they must be considerably less numerous than radio-loud ones. This situation is of course the reverse of that for emission line QSOs, where the radio-quiet QSOs far outnumber the radio-loud.
Having noted the common feature of radio emission for the objects in Table 1, we must go on to point out that the character of the emission varies widely among different sources. Nearly all have a flat-spectrum component, but there is a large range in luminosity. Five objects are among the intrinsically brightest of all known sources at 2700 MHz. 3C sources 3C279, 3C345, 3C446, 3C454.3, and CTA 102 all have apparent luminosities (assuming isotropic emission) of ~ 1035 ergs Hz-1 s-1, and constitute about 20% of all sources known to be this luminous. By contrast the faintest objects such as Mkn 180 have luminosities of 1031 ergs Hz-1 s-1. It is interesting that the most direct evidence for bulk relativistic motion is only for the intrinsically brightest objects. The two certain cases of superluminal expansion that are known for objects in Table 1 are from this group, namely 3C 345 and 3C 279. Also, the only examples in Table 1 of low-frequency variability (which suggests bulk relativistic motion - see Section V) are 3C 454.3, CTA 102, and PKS 0420-014, which is also very bright. AO 0235 + 164 is so bright and rapidly variable at high radio frequencies that, again, brightness temperatures in excess of 1012 K are indicated. For the fainter objects with absent or weak emission lines (BL Lacs), VLBI data is reviewed by Shaffer (1978). Changes in structure are seen, but there is no strong case of superluminal expansion.
Radio structure on scales of arcseconds or larger, generally with steep spectra, has been measured in the following objects: NGC 1275 (Miley & Perola 1975), PKS 0521-365 (Fomalont 1968, Mills, Slee & Hill 1960), AO 0827+24 (Hazard, Gulkis & Sutton 1968), Mkn 180 (Kojoian et al. 1976), 1400+162 (Baldwin et al. 1977), AP Lib (Conway & Stannard 1972), 3C 345 and 3C 454.3 (Davis, Stannard & Conway 1977), 3C 371 (Fomalont & Moffet 1971), and 3C 390.3 (Harris 1972). Extended structure of 10-200 arcsec extent, generally of steep spectrum, is also known to be present in about half the sample of 27 BL Lac objects studied by Wardle (1978). Details of this structure are not known because the interferometer was used at only one position angle. Since symmetric extended radio emission, like the optical line emission, is almost surely isotropic, the geometry and strength of the extended sources can aid in distinguishing the degree to which relativistic beaming may be present, and will be considered further in Section VI.
Variability and polarization at radio wavelengths have quite different characteristics for different objects in Table 1. From the four year monitoring program by Altshuler & Wardle (1976) one finds that the optically violently variable quasars 3C 454.3, 3C 446, 3C 345, and CTA 102 generally have rather steady flux and polarization. By contrast, the BL Lac objects 0048-09, 0235+16, 0300+47, OJ 287, W COM, 1749+096, 2155-304, and BL Lac are strongly variable in both flux and polarization. Unfortunately there is little data for the nearby objects with z < 0.1, except for BL Lac. NGC 1275, whose radio emission is essentially unpolarized, and AP Lib, which has relatively steady polarization. Coordinated observations of blazars from radio through optical wavelengths by Rudnick et al. (1978) show the polarization is generally weaker at radio wavelengths. With the exception of OJ 287, for most blazars there is little correlation between the angles of optical and radio polarization.
CORRELATION OF RADIO AND X-RAY EMISSION WITH OPTICAL EMISSION LINES IN BLAZARS There is a strong correlation between the radio emission of blazars and their optical emission lines. Consider the group of all eleven objects in Table 1 with z < 0.1, all of which are found at the center of giant elliptical galaxies. The five of them (Mkn 180, 421, 501, I Zw 186, 0548-322) that have no detectable emission lines all show the weakest radio emission, all of similar strength with flat spectra at ~ 0.5 Jy scaled to a common z of 0.05. The remaining six objects that do show emission lines are 3C 84 (NGC 1275), 3C 371, 3C 390.3, PKS 0521-36, BL Lac, and AP Lib. All show stronger radio emission, at least 2 Jy at 1400 MHz, again scaled to z = 0.05, and all but BL Lac have a low-frequency steep-spectrum component.
The correlation between the presence of detectable emission lines and of a steep component to the radio spectrum holds for nearly all objects in Table 1, without regard to redshift. Another general correlation we have mentioned above is that the emission line objects tend to have relatively stable radio polarization and flux, in contrast to the line-free objects.
In contrast to the radio correlation, the X-ray emission observed from some of the same 11 objects is not stronger for those with lines, indeed there is a suggestion that the line-free objects may have stronger X-ray fluxes. The highest X-ray luminosities of > 5 µJy at 3.6 keV, referred to z = 0.05, have been observed in Mkn 521 and PKS 0548-322, both line-free objects. All the line-free objects except Mkn 180 have been detected at the ~ 1 UHURU count level, making the X-ray and optical luminosities comparable.