Anatomy of a radio to gamma
ray outburst between the
black hole and jet core in the active galactic nucleus BL
Alan P. Marscher
, Svetlana G. Jorstad
, Francesca D. D’Arcangelo
, Paul S. Smith
G. Grant Williams
, Valeri M. Larionov
*, Alice R. Olmstead
, Margo F.
, Hugh D. Aller
, Ian M. McHardy
, Anne Lähteenmäki
, Merja Tornikoski
, Vladimir A. Hagen
, Eugenia N. Kopatskaya
, Walter K. Gear
, Lorand Sigua
& Wesley T. Ryle
1. Institute for Astrophysical Research, Boston University, 725 Commonwealth Ave.,
Boston, MA 02215, USA
2. Astronomical Institute of St. Petersburg State University, Universitetskij Pr. 28,
rets, 198504 St. Petersburg, Russia
3. Steward Observatory, University of Arizona, Tucson, AZ 85721
4. MMT Observatory, University of Arizona, Tucson, AZ 85721
5. Isaac Newton Institute of Chile, St. Petersburg Branch, St. Petersburg St
University, Universitetskij Pr. 28, Petrodvorets, 198504 St. Petersburg, Russia
6. Astronomy Department, University of Michigan, 830 Dennison, Ann Arbor, MI
7. Department of Physics and Astronomy, University of Southampton, Highfield,
outhampton, SO17 1BJ, UK
8. Metsähovi Radio
Helsinki University of Technology TKK,
Metsähovintie 114, FIN
540 Kylmälä, Finland
9. Tuorla Observatory
ntie 20, FI
21500 Piikkiö, Finland
10. Department of Physics, University of Turku, FI
20014 Turku, Finland
11. School of Physics and Astronomy, Cardiff University, Cardiff CF24 3YB, Wales, UK
12. Department of Physics, University of Perugia
Via A. Pascoli, 06123 Perugia, Italy
13. Abastumani Astrophysical
14. Department of Physics and Astronomy, Georgia State University, Atlanta, GA
* Present address: Department of Physics, University of California, Berkeley, CA 94720
Active galactic nuclei (AGN) are famous for violent outburst
s of radiation across
the electromagnetic spectrum. These giant flares are especially prominent in the
subclass termed ‘blazars,’ whose most prominent members include the AGN BL
Lacertae (BL Lac). However, until now we have been unable to specify the locat
of the flares in the jet and the physical mechanisms that cause dramatic increases
in brightness. Such AGN contain compact jets of highly energetic, magnetized
plasma that emanate from accreting supermassive black holes. According to
, such jets are launched by dynamic magnetic fields twisted by
the differential rotation of the accretion disk or inertial
ergosphere surrounding the black hole
. The flow velocity increases down the jet
in an acceleration and collimation
zone (ACZ) containing a tightly wound helical
. Here we report sequences of images and polarisation
measurements of BL Lac that reveal a bright feature in the jet causing a double
flare of radiation from optical frequencies to TeV
O慴愠楮d楣慴e⁴桡琠 he楲獴污牥瑡 瑳t
we en瑩t礠睩yh⁴桥⁁䍚Ⱐ,hee⁴桥汯l f潬oow猠獰楲慬瑲eam汩ne猠
passing through a magnetic field with a toroidal component. The f
lux rises due to
an increase in the relativistic beaming of the radiation as the feature accelerates.
The second flare and radio outburst occur as the feature encounters a standing
shock wave corresponding to the bright ‘core’ seen on VLBI images.
1 displays a sequence of Very Long Baseline Array (VLBA) radio images of the
jet of BL Lac. The jet approaches us within 8
of the line of sight at a flow speed of
, corresponding to a Lorentz factor of 7.7
Relativistic aberration and the
er effect strongly beam the radiation so that the apparent luminosity is hundreds
of times higher than if the emitting plasma were at rest. An essentially identical
counterjet is presumably present, but too faint to detect because of beaming in the
e direction. The stationary core lies at the northern end. Bright knots emerge
from the core at a rate of 1
2 per year and move southward at apparent superluminal
speeds, an illusion caused by their relativistic motion
Figure 2 presents the radio, optica
l, and X
ray light curves of BL Lac over a two
year period, with the addition of optical polarisation data during late 2005, when our
observations are most intensive. As is evident from the light curves, we observe a
double flare during this period. The hi
ghly significant detection
of > 0.2 TeV
from 2005.819 to 2005.831 during the first X
ray flare implies that acceleration of sub
energy electrons was particularly efficient at this time. These same electrons both
rays from synchrotron
radiation and scatter the X
ray photons to GeV
energies that are boosted to the TeV range by relativistic motion of the jet plasma.
The location of such flares has been controversial, with some observations
suggesting that they occur downstream o
f the core and theoretical models requiring that
they take place well upstream of this region, where the plasma is more compact. As we
explain below, our data indicate that the first flare in late 2005 corresponds to a
disturbance passing through the upstr
eam zone where the jet flow is still accelerating,
while the second is caused by interaction between the disturbance and a standing shock
system in the core.
The identification of the initial flare with the ACZ is significant, since previous
f jet collimation are quite limited. For example, a radio image
at 7 mm
wavelength of the radio galaxy M87 appears to reveal an initially broad outflow that
narrows into a nearly cylindrical jet. This is consistent with gradual collimation by
either a to
roidal magnetic field
or external confining gas pressure that declines with
distance from the black hole
. The flow seen in M87 could include a “sheath” that
moves more slowly and is less focused than the “spine.”
In the case of BL Lac, the
nt superluminal motions of bright knots in the jet and its pronounced
variability at all wavelengths suggest that the observed radiation arises exclusively from
the spine where special relativistic effects dominate.
The primary observational indicator of m
agnetic collimation requiring a helical
magnetic field in the spine of the jet is the evolution of the polarisation. Synchrotron
radiation from a circularly symmetric jet with a helical field observed at an angle to its
axis displays a net polarisation ori
ented either parallel or perpendicular to the projected
. Such polarisation can be confused with shock waves or velocity shear,
respectively, that can produce the same polarisation patterns. However, in a model
where magnetic forces gradually acc
elerate and focus the jet, the flow velocity is
directed along streamlines that follow a helical trajectory with a different, wider pitch
angle than that of the magnetic field
. The rotation of the flow traces back to the
footprint in the orbiting accretio
n disk or differentially rotating ergosphere where the
spin of the black hole drags the inertial frames. A shock wave or other condensation
propagating down the jet traces a spiral path along the streamlines that cycles through
the orientations of the heli
cal field (see Fig. 3 and ref. 5). This should manifest itself as
a rotation of the position angle of linear polarisation as the feature moves outward. The
degree of polarisation should drop to a minimum in the middle of the rotation, when the
c field in the flaring region is transverse to that of the previously existing
. As Fig. 2 demonstrates, we see both effects.
The optical polarisation shown in Fig. 2 rotates steadily by about 240
over a 5
day interval before settling at an EVPA
. The sequence of images (Fig. 1)
reveals a bright, superluminal knot that first appears upstream of the core. It
subsequently moves past the core and proceeds down the jet along position angle ~190
and with an EVPA that is parallel to the jet to
within the uncertainty.
correspondence of the optical and 7 mm EVPAs after 29 October implies that the knot is
the emitter of the polarised optical emission during the flare.
have suggested that rotations of the polaris
occur in BL Lac and the similar object OJ287. These earlier observations were more
poorly sampled than ours, which allowed multiple interpretations owing to the ±180º
ambiguity of the EVPA. Notwithstanding this last point, the model that we ad
quite similar to one of those proposed in these previous papers
, with the location of
the emission region and connection with high
energy flares now specified by our
sequences of VLBA images and multi
waveband light curves.
We interpret the
event in the following manner, as illustrated in Fig. 3. Explosive
activity at the inlet of the jet near the black hole injects a surge of energy into the jet
across part of its cross
sectional area. This disturbance forms a shock wave that
g a subset of streamlines down the acceleration and collimation zone of
the jet. The shock front compresses the ambient magnetic field and energizes electrons,
while the Doppler beaming increases as the knot accelerates along a spiral path that
ut with distance down the jet. These effects cause the flux of synchrotron
radiation of the knot to rise until it dominates the optical, X
ray, and (via inverse
ray emission as the knot exits the zone of helical magnetic field.
and therefore the peak in the light curve of the first flare
during the last spiral when the Lorentz factor of the jet is near its asymptotic value and
the velocity vector of the shock points most closely toward our line of sight. The pea
can be quite sharp
, as observed. When the flare dominates the optical flux, we see the
optical polarisation vector rotate before the shock exits the ACZ. The ACZ is opaque at
radio wavelengths, hence the first flare is absent in the radio light curves.
Beyond the ACZ, the shock encounters a region of turbulence that is possibly
driven by velocity shear across the jet
downstream of the point where the magnetic and
particle energy densities reach rough equipartition
. The ambient magnetic field in the
has a chaotic structure in this region. Since the shock front amplifies only the
component of the field that is parallel to the front, the EVPA becomes transverse to this
direction and therefore essentially parallel to the velocity vector of the knot alon
PA~190º. During this phase, the flux declines as the knot proceeds down the broadening
jet, where there is a gradient of decreasing magnetic field strength and electron density.
If the model we propose is correct, then the variation of EVPA with time sho
deviate from a strict linear dependence owing to projection effects, since the circular
section has an elliptical shape from our vantage point. We have calculated this
effect, including relativistic aberration, and show in Fig. 2 that the optical
do indeed follow the corresponding curve. The small number of brief excursions of the
EVPA from the curve, the deviations from the mean EVPA before and after the rotation,
and irregularities in the light curves can all be explained by local flar
ups of emission
momentarily amplify both the polarisation along a particular direction and the flux
at various wavebands.
The smoothness of the EVPA vs. time curve eliminates the possibility
rotation is actually a random walk of the po
larisation vector owing to a chaotic magnetic
field. In this case, our numerical simulations (see ref. 18) indicate that the curve should
be much more jagged than observed when the degree of polarisation ~ 5%. In the
simulations, this level of polarisation
corresponds to synchrotron emission from ~200
independent cells, each with randomly oriented magnetic field. In such a model,
apparent rotations by ~240º are very rare in the simulations, whereas they are common
during flares of BL Lac and similar objects
Both synchrotron radiation and inverse Compton scattering contribute to the X
ray emission from BL Lac, with synchrotron dominating when electrons are accelerated
to energies in the TeV range
. This generally causes the X
ray flux density (
) spectrum to steepen such that the spectral index
> 1, where
ray spectral steepening occurs during the first flare. In contrast, the X
spectrum becomes harder (
< 1) during the second flare, as expected if the X
enerated by inverse Compton scattering of optical and infrared photons.
The second flare, which starts at 2005.89, is simultaneous with the passage of the
knot through the core seen on the VLBA images. If the core is a standing conical shock,
from simultaneous radio and optical polarisation variability in the case of
the quasar PKS 0420
, the emission would increase as the knot undergoes
compression by the shock front. The flare dies at optical and X
ray frequencies as the
away from the core down the expanding jet. However, it lives much
longer at 43 GHz, at which the synchrotron radiation requires lower energy electrons
that have longer energetic lifetimes than those emitting at higher frequencies.
The observed time delay
of 18 days between the end of the optical EVPA rotation
and the coincidence of the knot with the core in the VLBA images corresponds to a
linear distance of ~2 light
years ~ 2
cm between the end of the ACZ and the core
at 7 mm wavelength. The flatne
ss of the radio spectrum down to ~0.7 mm wavelength
implies that the jet is a self
similar cone with constant Lorentz factor downstream of the
point where the jet cross
section becomes ~ 0.1 times its value at the 7 mm core
Although the cross
l radius of the jet at the position of the 7 mm core is too
small to resolve on our VLBA images, we can estimate a value ~ 1
cm from the
timescale of radio flux variability
brightness temperature arguments.
If the site
where the jet becomes opti
cally thin at 0.7 mm wavelength is at the outer radius of the
, the cross
sectional radius at that point ~ 1
cm, equivalent to ~200
Schwarzschild radii (
) for a black
hole mass M ≈ 2
solar masses, as estimated for
The observed rot
ation rate of the polarisation, 50° day
, corresponds to a
single full twist of the streamline occurring across an axial distance of ~ 1 light
cm, since the apparent speed in the plane of the sky is 5.0
and the jet is
foreshortened by a fac
tor of 8 as projected on the sky. The rotational velocity of the
streamlines is then
at the outer edge of the ACZ. At smaller distances from
the black hole, down to the Alfvén radius where the flow speed equals the local Alfvén
should vary as 1/
to conserve angular momentum. In this case, the
is a fraction
of the speed of light
when the cross
cm, or ~20
. Inside the
the flow should co
with the foo
tpoint of the magnetic field in the accretion disk or ergosphere, so that
In the simulations of Vlahakis and Königl
. If we set the rotational speed at
the footpoint as
cm) = (
, we can estimate that the outer
magnetic field of the jet is anchored at
. If instead
≈ 1, as might be expected
for almost pure Poynting flux
jets launched from the ergosphere of a rotating black
, we arrive at a value
The uncertainties in both the numerical value of
are of order a factor of 2. Hence, these distances from the black hole are
consistent with the expec
tations of models in which the jet is driven by twisting
magnetic fields from the vicinity of an accreting black hole.
We constructed the images at 7 mm wavelength from data obtained with the Very Long
Baseline Array (VLBA). The optical po
larimetric data resulted from observations at
Steward Observatory and the Crimean Astrophysical Observatory. The optical flux
density points are from photometry at these two sites plus Lowell Observatory, Perugia
University Astronomical Observatory, and Ab
ustumani Astrophysical Observatory. All
of the optical telescopes are equipped with CCD cameras. We measured the X
and continuum spectrum via a monitoring program with the
Explorer (RXTE), a NASA space observatory. We obtained t
he radio flux density
measurements at the University of Michigan Radio Astronomy Observatory (UMRAO)
and the Metsähovi Radio Observatory
. Descriptions of the telescopes and data analysis
are available in the online supplementary information.
R. D. & Payne, D. G. Hydromagnetic flows from accretion discs and the
production of radio jets.
Mon. Not. R. Astron. Soc
2. Blandford, R. D. & Znajek, R. “Electromagnetic extraction of energy from Kerr black
Mon. Not. R. Astro
3. Meier, D. L., Koide, S. & Uchida, Y. Magnetohydrodynamic production of
4. Vlahakis, N. & Königl, A. Magnetic driving of relativistic outflows in active galactic
nuclei. I. Interp
retation of parsec
5. Vlahakis, N. in
Blazar Variability Workshop II: Entering the GLAST Era
, vol. 350 (eds Miller, H. R., Marshall, K., Webb, J. R. & Aller, M.
stronomical Society of the Pacific, San Francisco, 2006).
6. Jorstad, S. G.
Polarimetric observations of 15 active galactic nuclei at high
frequencies: jet kinematics from bimonthly monitoring with the Very Long Baseline
7. Freedman, W. L.
Final results from the Hubble Space Telescope key project to
measure the Hubble constant.
8. Albert, J.
. Discovery of very high energy
ray emission from the low
d BL Lac object BL Lacertae.
9. Jorstad, S. G.
epoch VLBA observations of EGRET
detected quasars and
BL Lac objects: connection between superluminal ejections and gamma
ray flares in
10. Lähteenmäki, A. & Valtaoja, E.
Testing of inverse Compton models for active
galactic nuclei with gamma
ray and radio observations.
11. Junor, W. , Biretta, J. A. & Livio, M.
Formation of the radio je
t in M87 at 100
Schwarzschild radii from the central black hole.
12. Melia, F., Liu, S. & Fatuzzo, F.
Is Thermal expansion driving the initial gas ejection
in NGC 6251?
13. Laing, R. A., Parma,
P., de Ruiter, H. R. & Fanti,
R. Asymmetries in the jets of
weak radio galaxies
Mon. Not. R. Astron. Soc.
14. Lyutikov, M., Pariev, V. I. & Gabuzda, D.
C. Polarisation and structure of
scale AGN jets.
Mon. Not. R.
15. Sillanpää, A., Takalo, L. O., Nilsson, K. & Kikuchi, S. Photopolarimetry of BL
Astrophys. & Space Sci.
16. Moore, R. L.
The noise of BL Lacertae.
Kikuchi, S., Inoue, M., Mikami, Y., Tabara, H. & Kato, T. A synchronous variation
of polarization angle in OJ 287 in the optical and radio regions.
18. D’Arcangelo, F. D.
Rapid Multiwaveband Polarization variab
ility in the
quasar PKS 0420
014: optical emission from the compact radio jet.
19. Böttcher, M.
observations of BL Lacertae in
20. Impey, C. D.
& Neugebauer, G. Energy distributions of blazars.
Königl, A. Relativistic jets as X
ray and gamma
22. Lähteenmäki, A.,
Valtaoja, E. & Wiik, K.
Total flux density variations
extragalactic radio sources. II. Determining the limiting brightness temperature for
23. Marscher, A. P
. Relativistic jets and the continuum emission in QSOs.
. Woo, J.
H. & Urry, C. M. Active galactic nucleus black hole masses and Bolometric
25. McKinney, J. C. & Naraya, R. Disk
jet coupling in black hole accretion systems II:
free electrodynamical models.
Mon. Not. R. Astron. Soc.
26. Hawley, J. & Krolik, J. H. Magnetically driven jets in the Kerr metric.
The authors thank Dr. N. Vlahakis for enlightening discussions. The research reported here is ba
work at Boston University supported by the U.S. National Science Foundation under grant AST
and NASA grant NNX06AG86G, at St. Petersburg State University supported by grant no. 05
from the Russian Fund for Basic Research, and at the
University of Michigan supported by the U.S.
National Science Foundation under grant AST
0607523 and a series of NASA grants. P.S.S.
acknowledges support from NASA contract 1256424.
AL and MT acknowledge support from the
Academy of Finland for the Metsäho
vi observing projects.
The VLBA is an instrument of the National
Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperative
agreement by Associated Universities, Inc. The authors declare that they have no competin
Correspondence and requests for materials should be addressed to A.P.M. (e
Figure 1 | Sequence of VLBA images of BL Lac at a wavelength of 7 mm
(frequency = 43 GHz).
The images are convolved with a circular
function of FWHM=0.1 milliarcseconds (mas), the resolution of the longest
baselines of the array, which corresponds to 0.12 parsecs (pc) at the distance
of 291 Mpc derived from the redshift,
=0.069, and the Hubble Law with an
72 km s
(ref. 7). Contours indicate total intensity,
starting at 2.30 Jy/beam and increasing by factors of 2. Colour scale represents
polarised intensity, with peak (yellow) at 0.215 Jy/beam. The yellow line
segments indicate the electric vector po
sition angle (EVPA) of the polarisation,
which is uncertain by ±7° in each image. The approximate path of the centroid
of the knot is given by the slanted red line, the definition of which includes a
point from an image obtained on 12 March 2006 in additio
n to those shown. The
observed proper motion of the knot of 1.2 mas yr
is equivalent to an apparent
speed of 5.0
after correction for time dilation owing to the cosmological
Figure 2 | Flux density at various wavebands and optical polarisati
BL Lac as a function of time.
Dependence on time of the flux of radiation
BL Lac over a two
at the indicated wavebands.
uncertainties in the values plotted.
exponent of the power
law dependence of
ray flux density on frequency is
up of the
year time interval marked by vertical dotted lines in the lefthand panels,
with optical R
band EVPA and degree of polarisation replacing X
index and radio flux density. The ti
me range of highly significant detections
photon energies > 0.2 TeV is indicated by the width of the head of the upward
arrow in the top panel. The rotation in optical
band EVPA near the time of the
peak of the first optical and X
ray flare is appare
nt. Since there is an ambiguity
of ±180º in the value of the EVPA, we have selected the quadrant of each value
that provides a consistent overall trend of rotation between 2005.81 and
2005.83. As is discussed in the main text, the solid curve in the EVPA p
corresponds to the pattern expected according to the model shown in Figure 3
when relativistic aberration is included. The vertical arrow (with error bar) in the
bottom panel denotes the time when the superluminal knot is coincident with the
y core seen in the images displayed in Fig. 1.
Figure 3 | Sketch of the inner jet of BL Lac according to the scenario we
propose to explain the behavior during the last quarter of 2005.
propagates down the jet along a spiral streamline. The fir
st flare occurs during
the last 240
twist of the streamline before the flow straightens and becomes
turbulent. The passage of the feature through the mm
wave core stimulates the
second flare. A logarithmic scale of distance from the black hole, shown in t
bottom panel in terms of Schwarzschild radii, is employed to illustrate
phenomena on various scales.
We observed BL Lac at 7 mm wavelength (frequency of 43 GHz) with the Very
Array (VLBA) at the epochs displayed in Fig. 1 as well as the following
dates: 24 and 28 October 2005, 2 November 2005, and 12 March 2006. After
correlation at the Array Operations Center of the National Radio Astronomy
Observatory (NRAO) in Socorro, New
Mexico, we passed the data through the VLBI
routines of the Astronomical Image Processing System (AIPS) software supplied by
NRAO for initial calibration and followed the procedures described in ref. 6 to create
and analyze the resultant images. We refer
the EVPA measurements to a stable feature
in the VLBA polarised intensity image of the quasar CTA102. There is good agreement
between the thus
calibrated EVPAs of several objects with archival data and with
contemporaneous Very Large Array measurements (a
http://www.vla.nrao.edu/astro/calib/polar/). In our analysis, we subtract 16
polarisation angle in the pseudocore of BL Lac to compensate for Faraday rotation
We collected optical polarimetric data at Steward
Observatory and the Crimean
Astrophysical Observatory and photometry at these two sites plus three others. At the
Steward Observatory 1.55 m Kuiper telescope
at Mt. Bigelow, Arizona, USA, we used
the SPOL spectropolarimeter
, making a total of 34
, each covering
800 nm during 12 contiguous
nights from 2005 October 23 to November 3.
Details of the reduction procedure can be found in, e.g., ref. 29. At the Crimean
Astrophysical Observatory in Crimea, Ukraine, we performed
band photometry an
polarimetry on the AZT
8 70 cm telescope, with a prime
equipped with two Savart plates.
We also measured the optical R
band magnitude of BL Lac with the following
telescopes: the 1.83 m Perkins telescope at Lowell Observator
y in Flagstaff, Arizona,
USA, the 0.4 m Automatic Imaging Telescope at Perugia University Astronomical
Observatory in Perugia, Italy, and the 70 cm meniscus telescope
Astrophysical Observatory in Abastumani, Georgia. All telescopes are equi
CCD cameras. For all flux measurements, we
determined the magnitude of BL Lac
relative to field stars calibrated by
Smith et al.
We observed BL Lac an average of 3 times per week for ~2000 s per pointing
y Timing Explorer (RXTE). For each X
ray observation, we
determined the X
ray flux in photon counts per second over the energy range 2 to 10
keV by subtracting an X
ray background model (supplied by the RXTE Guest Observer
Facility) from the raw spectrum,
using the standard X
ray data analysis software
package FTOOLS. We fit the photon spectrum (observed flux
versus photon energy
E from 2.4 to 10 keV) with a model consisting of a power
law intrinsic spectrum with
, plus ph
otoelectric absorption along the line of sight using
the program XSPEC. For the latter, we adopted a hydrogen column density of 2.7
(see ref. 32).
Radio flux density measurements.
We determined the flux density of BL Lac at a wavelength of
2 cm (frequency of
14.5 GHz) with the 26
m diameter antenna of the University of Michigan Radio
Astronomy Observatory (UMRAO) near Ann Arbor, Michigan, USA and at 8 mm (37
GHz) with the 14 m antenna of the Metsähovi Radio Observatory in Kylmälä, Finland.
he data displayed in Fig. 2 are nightly averages. Detailed description of the
instrumentation and reduction procedures can be found from Aller et al.
and Teräsranta, H. et al.
Jorstad, S. G.
Multiwaveband polarimetric observations of 15 active galactic
nuclei at high frequencies: correlated polarization behavior.
28. Schmidt, G. D., Stockman, H. S. & Smith, P.
Discovery of a sub
magnetic white dwar
f through spectropolarimetry
29. Smith, P. S., Schmidt, G. D., Hines, D. C. & Foltz, C.
stellar objects discovered by the two
micron all sky survey
30. Kurtanidze, O. M. & Nikolashvili, M. G. in
Blazar Astrophysics with BeppoSAX
and Other Observatories
, ASI Special Publication (eds Giommi, P., Massaro, E. &
Palumbo, G., 189
ESRIN, Frascati, 2002).
Smith, P.S., Balonek, T.J., Heckert,
P.A., Elston, R. & Schmidt, G.D. UBVRI field
comparison stars for selected active quasars and BL Lacertae objects.
32. Madejski, G.
ray observations of BL Lacertae during the 1997 outburst and
association with quas
Aller, H. D., Aller, M. F., Latimer, G.E. & Hodge, P.E. Spectra and linear
polarizations of extragalactic variable sources at centimeter wavelengths.
Fifteen years monitoring of extragalactic radio sources at 22, 37
and 87 GHz.
Astron. Astrophys. Suppl.
, 305 (1998)