It seems to be a property of nature that objects composed of plasma vary on timescales as short as allowed by the laws of physics. The most basic limitation on timescale is the light-crossing time. Most compact extragalactic objects seem to vary by tens or hundreds of percent even faster than this, and those that do so are referred to as "blazars." The term includes both quasars and BL Lac objects.
The vast assortment of observations of spectral variability of blazars have revealed a diverse range in characteristic behavior. There are many similarities and differences both among sources and among flares within the same source. At the risk of oversimplifying, I list the major properties of spectral variability that I feel best exemplify the nature of the phenomenon, while recognizing that there are exceptions to these patterns.
(i) Flares usually begin at high frequencies, then propagate to lower frequencies. Some flares, however, develop simultaneously over a wide range of frequencies before propagating to lower frequencies, and still others do not seem to move to lower frequencies at all. Some events begin at optical-infrared wavelengths, while others begin at submillimeter, millimeter, or even centimeter wavelengths (see Epstein et al. 1982; Aller et al. 1985).
(ii) The flaring emission usually rises abruptly, propagates to lower frequencies while maintaining essentially a constant amplitude, and then dies, sometimes gradually (as in AO 0235+164; O'Dell et al. 1988) and sometimes rapidly (as in 1156+295; McHardy et al. 1990).
(iii) Superposed on the flares, and sometimes on the "quiescent state," are mini-flares and/or rapid flickering; viz., the light curves are not smooth.
(iv) To first order, the multi-waveband spectra connect smoothly from the radio to the infrared. However, when observed with sufficient frequency coverage, the spectra often show double peaks (Brown et al. 1989). The timescale of variability is shorter at the higher frequencies. The polarization behavior also tends to be much more erratic at these frequencies.
(v) In some sources, rapid flickering has been observed at the
10% percent
level at centimeter wavelengths by
Heeschen et al. (1987),
Quirrenbach et al. (1989),
and Krichbaum,
Quirrenbach, & Witzel (1992).
The flickering occurs on the same
timescales as found in the optical. In fact, in the BL Lac object
0716+714, the variations
are quasi-periodic, with the same quasi-period (although not the same
phase) at both
radio and optical wavelengths over several "cycles" of the light curve
(Quirrenbach et al. 1991).
The two most fundamental problems associated with the timescales of
variability are
superluminal flux variations (apparently correlated brightness changes
in two or more regions separated by a distance
c 
t) and brightness
temperatures that exceed the
inverse Compton limit of 1012 K derived by assuming that the
size of the emitting region is less than
c t. Cases of
apparent superluminal flux variations have been reported by
Mutel, Aller, & Phillips
(1981),
Pauliny-Toth et
al. (1987),
Götz et
al. (1987), and
Pauliny-Toth et
al. (1990).
The highest brightness temperatures inferred from variability observations are derived for low-frequency variations on timescales of months at meter wavelengths (Cotton & Spangler 1982) and intraday variations at centimeter wavelengths (Qian et al. 1991). The most troublesome low-frequency fluctuations, however, are now thought to be caused by refractive interstellar scintillations rather than variations intrinsic to the source (Rickett, Coles, & Bourgois 1984). The highest brightness temperature inferred from the intraday variations is ~ 2 x 1018 K (Qian et al. 1991).
The polarization of a compact extragalactic radio source is typically even more highly variable than the flux density, with rapid swings in position angle - by as much as 180° - often observed (e.g., Aller et al. 1985; Qian et al. 1991; Sillanpää & Takalo 1992).
Variability observations of extragalactic sources with compact jets at
X-ray and
-ray wavelengths
are not yet
extensive. BL Lac objects tend to be highly variable at X-ray energies (e.g.,
Maraschi 1992)
and tend to have steeper X-ray spectra than do quasars
(Worrall & Wilkes 1990).
In fact, the spectra of some BL Lac objects
appear to be continuous from radio to X-ray frequencies, indicating a
common emission
mechanism - presumably synchrotron radiation - in the compact jet. Quasars are
also variable at X-ray energies (e.g., 3C 273:
Tananbaum et al. 1979,
Courvoisier et al. 1987;
3C 279:
Makino et al. 1989;
NRAO 140:
Marscher 1988),
although not many
cases are well documented. The X-ray spectra of quasars are rather flat
(Wilkes & Elvis 1987),
too much so to be explained as continuations of the optical-uv emission.
The recent detections of a number of quasars and BL Lac objects containing
compact jets by the Compton Observatory at hard
-ray energies demonstrate that
a significant, in some cases dominant, fraction of the nonthermal
luminosity is emitted at extremely high energies
(Hartman et al. 1992
and various IAU Circulars). The
-ray
flux from the quasar 3C 279 was reported to be variable by a factor of
three over
a 4-month period and also to be significantly variable on a timescale of
a few days
(Kanbach et al. 1992).
The BL Lac object Mk 421 has also been detected at TeV
energies (1 TeV = 1.6 ergs = 2.4 x 1026 Hz!)
(Weekes 1992).
The overall multifrequency variability of nonthermal sources has yet to
be studied
extensively, although monitoring campaigns are planned from radio to
-ray frequencies
in late 1992 and 1993. There are, however, a number of studies that have
demonstrated that flares in nonthermal sources are often broadband in nature. A
direct correspondence between X-ray and radio-infrared variability has
been found in
3C 279
(Makino et al. 1989).
In the case of BL Lac,
Kawai et al. (1991)
found that the
X-ray flux was correlated with the submillimeter-wave flux. In NRAO 140,
Marscher (1988)
found a similar relationship between the X-ray flux and the flux of the radio
core as measured with VLBI.
Courvoisier et al. (1987),
on the other hand, found little
correspondence across wavebands in the variations of 3C 273, although
the sampling rate was somewhat sparse.
Bregman et al. (1990) and Hufnagel & Bregman (1992) have analyzed in detail the variations occurring at radio, infrared, and optical frequencies in BL Lac and several other blazars. They find that there is no significant time delay between features in the optical and infrared light curves, but that there is a delay of about 1 year between the weakly correlated optical and radio variations. They also find that the optical variations can be characterized by a combination of shot noise and flicker, whereas the radio variations have power spectra similar to shot noise. The conclusion is that flickering is a high-frequency phenomenon in the sources studied, and that the radio emitting region is larger than, but connected to, the site of the optical emission.