Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.atnf.csiro.au/people/Matthew.Whiting/papers/comptel_quasars.ps.gz Дата изменения: Thu Mar 23 02:58:51 2006 Дата индексирования: Sun Dec 23 12:39:47 2007 Кодировка: Поисковые слова: m 63

The Optical Emission from Gamma-Ray Quasars
M.T. Whiting 1 , P. Majewski 2 and R.L. Webster 2
1 Department of Astrophysics and Optics, School of Physics, University of New South
Wales, Sydney NSW 2052, Australia
mwhiting@phys.unsw.edu.au
2 Astrophysics Group, School of Physics, University of Melbourne, Victoria 3010, Australia
rwebster@physics.unimelb.edu.au
Received 2001 November 29, accepted 2003 April 23
Abstract
We present photometric observations of six radio-loud quasars that were detected
by the COMPTEL gamma-ray telescope. The data encompass seven wavebands in the
optical and near infrared. After correction for Galactic extinction, we nd a wide range in
optical slopes. Two sources are as blue as optically-selected quasars, and are likely to be
dominated by the accretion disc emission, while three others show colours consistent with
a red synchrotron component. We discuss the properties of the COMPTEL sample of
quasars, as well as the implications our observations have for multiwavelength modelling
of gamma-ray quasars.
Keywords: galaxies: active | quasars: individual (PKS 0208 512, PKS 0506 612,
PKS 0528+134, PKS 1622 297, PKS 2230+114, PKS2251+158)
1 Introduction
Active galactic nuclei (AGN) are powerful emitters in all wavebands. Emission processes
from dierent parts of the spectrum in many cases appear to be linked. For example, the
electrons in the jet which produce the synchrotron radiation may at the same time upscatter
photons into the X-ray and gamma-ray regime via the inverse Compton (IC) process (see for
example the analysis by Bloom & Marscher 1996). In order to unravel the contribution from
each emission process, the shape of each distinct emission component needs to be determined,
and the associated physical parameters measured.
The IC emission is produced by seed photons (internal, from the jet itself, or external,
from the accretion disc or the emission-line clouds) which are upscattered by the electrons
(or other relativistic particles) in the jet. The energy of the photons increases by a factor
of  2 , where is the Lorentz factor of the relativistic particle. Consequently, there are
two crucial factors that determine the nature of the IC emission: the maximum energy of
the jet electrons, which determines the turnover frequency of the synchrotron component;
and the typical energy of the seed photons, which is governed by the relative strengths of
the thermal and non-thermal components. The optical regime is where these components
1

Table 1: Sample description. E(B V ) values are the Galactic extinction values from Schlegel
et al. (1998).
Name Other RA DEC E(B V ) z
name (J2000) (J2000) (mag)
PKS 0208 512 | 02:10:46.20 51:01:01.9 0.020 0.999
PKS 0506 612 | 05:06:44.00 61:09:41.0 0.025 1.093
PKS0528+134 | 05:30:56.36 +13:31:55.0 0.860 2.060
PKS 1622 297 | 16:26:06.02 29:51:27.0 0.454 0.815
PKS2230+114 CTA 102 22:32:36.40 +11:43:51.3 0.072 1.037
PKS2251+158 3C 454.3 22:53:57.75 +16:08:53.6 0.105 0.859
are both energetically important, and so a good understanding of the emission at optical
frequencies is crucial for accurate modelling of the high-energy emission.
Before COMPTEL was shut down in June 2000, this telescope detected nine blazars in
the range 0.75 { 30 MeV which were also detected by EGRET in the gamma-ray regime 
100 MeV (Schonfelder et al. 2000; Collmar 2001). For many of the sources in this sample,
the data coverage in the optical/near infrared (NIR) region is very poor. To better under-
stand this important part of the spectral energy distribution (SED), six blazars from the
COMPTEL sample, chosen subject to visibility during an observation period from 27 Au-
gust to 2 September 2001, were observed in the photometric bands B, V , R, I, J , H, and
K n . Basic information for this sample is shown in Table 1. We note that the redshift for
PKS 1622 297 (z = 0:815) is taken from the Parkes Catalogue (Wright & Otrupcek 1990,
hereafter PKSCAT90), wherein no reference is given. No other determination of its redshift
has been made in the literature, and, while we use this value herein, it should be regarded as
potentially suspect.
Detailed models of the multiwavelength spectra of several of the sources in this sample
have been published in the literature. For example, Mukherjee et al. (1999) describe a model
with both synchrotron and IC emission components tted to nine epochs of monitoring data
for the source PKS 0528+134. While considerable eort has been expended in tting the
high-energy spectrum (X-rays and above) of this source, the information in the low-energy
part of the spectrum, from optical to radio energies, has largely been ignored. Rather than
modelling one source in detail, by considering a modest sample of six sources more general
conclusions can be drawn about the nature of the emission processes.
As synchrotron emission causes linear polarisation, optical polarisation measurements
can be used to check the predictions from the modelling. Optical polarimetry of most of the
sources in this sample is available from the literature. It will be shown that the sources with
high optical polarisation are generally distinguished by their radio to optical spectral shapes,
and are consistent with having a synchrotron component extending to optical wavelengths.
Section 2 describes the photometric optical and NIR data for the six sources. The SEDs
in each case are adequately modelled by a power law in Section 3. Finally, in Section 4 we
examine the implications of our observations for multiwavelength modelling of gamma-ray
sources, as well as the demographics of the sample as a whole.
2

2 Observations and Data
2.1 Observations
The six blazars were observed in optical bands B, V , R, and I on the night of 29 August
2001 with the ANU 40 in telescope at Siding Spring Observatory (SSO). The exposure time
in each band was 300 seconds. The seeing was typically 2:5 00 3 00 , but conditions were
generally photometric (although the weather did force some breaks in the observing, and the
loss of the B band data for PKS 1622 297). Cross-checking of eld stars from Raiteri et
al. (1998) enables verication of the photometric conditions. Observations of standard stars
from Graham (1982) enabled calibration to an absolute magnitude/ ux scale.
Observations in the NIR bands J , H, and K n were made on the nights of 31 August and 1
September 2001 with the CASPIR (Cryogenic Array Spectrometer/Imager|see McGregor et
al. 1994) instrument on the ANU 2.3 m telescope, SSO. Each image is a median-combination
of four (or, in the case of the fainter sources, 20) dithered images of 60 seconds exposure time
each (these individual exposures were: 2 cycles of 30 seconds for J ; 6 cycles of 10 seconds in
H; and 12 cycles of 5 seconds in K n ).
Poor weather on the rst night allowed us to observe only PKS 1622 297 and two standard
stars, but the second night had good conditions, allowing us to observe the remaining sources.
The seeing was better than for the optical (typically  1:5 00 2 00 ) and conditions were generally
photometric. Individual frames where the background increased due to the presence of cloud
were discarded prior to co-addition. Observations of IRIS standard stars from the Carter
system (Carter & Meadows 1995) were used for calibration.
2.2 Data
The data reduction was performed in a standard way using IRAF software. Optical images
were bias- and overscan-subtracted, and then at-elded. The at-eld images were generated
from a combination of night sky ats (created from all object images such that any sources
are excluded) and ats made from exposures of a uniformly cloudy sky. NIR images were
bias- and dark-subtracted, and then at-elded using ats derived from the dierence of
exposures of the dome with and without illumination. This removes telescope emission and
improves photometric accuracy. The sky emission was then subtracted from the individual
images using a median of all dither positions (excluding those with high background due to
cloud), and then the individual dither positions were added together to produce the nal
image.
The photometry was performed using the IRAF daophot package. The resulting optical
and NIR magnitudes are shown in Tables 2 and 3 respectively. The errors quoted are 1
errors resulting from the sum in quadrature of the random photometric errors, as calculated
by the daophot package, and the systematic errors that result from the spread in zero points
for the various standard star observations over the course of the night. The photometry
includes corrections for the varying atmospheric extinction due to the dierent airmasses
of the sources. For two sources (PKS 0506 612 and PKS 1622 297), subtraction of nearby
stars was required to provide accurate background subtraction for the aperture photometry
(although in no case were the images of the star and quasar overlapping). Figure 1 plots the
observed photometric SED for each source as a function of observed wavelength.
The magnitudes were converted into uxes to enable modelling and plotting of the data.
The zero magnitude uxes for the optical bands were taken from Bessell, Castelli, & Plez
3

Table 2: Optical magnitudes and dates of observation. For each source, the rst row shows
the observed magnitudes, while the second row shows the magnitudes after correction for
Galactic extinction (see text).
Source B V R I UT date
PKS 0208 512 16.270.04 15.800.03 15.430.03 14.940.03 29 Aug 2001
16.180.04 15.730.03 15.380.03 14.900.03
PKS 0506 612 17.670.07 17.250.05 16.880.04 16.620.05 29 Aug 2001
17.560.07 17.170.05 16.820.04 16.570.05
PKS0528+134 20.760.23 19.530.15 18.980.11 17.940.11 29 Aug 2001
17.040.23 16.740.15 16.720.11 16.250.11
PKS 1622 297 | 18.840.33 18.290.25 17.880.24 29 Aug 2001
| 17.370.33 17.090.25 16.990.24
PKS2230+114 17.200.05 16.820.03 16.430.04 16.030.04 29 Aug 2001
16.890.05 16.590.03 16.240.04 15.890.04
PKS2251+158 15.340.05 14.830.01 14.430.03 13.880.03 29 Aug 2001
14.880.05 14.490.01 14.150.03 13.670.03
(1998), while the CASPIR zero magnitudes are obtained from a 11200 K blackbody nor-
malised to F  (555nm) = 3:44  10 12 Wcm 2 m 1 (Bersanelli, Bouchet, & Falomo 1991).
The values of the zero points are shown in Table 4. Note that although the observations
used the K n lter, they were calibrated according to the K magnitudes of the IRIS standards
without a colour-correction term, and so have been normalised using the K band zero point.
2.3 Galactic Extinction
Values for the Galactic extinction, taken from the maps of Schlegel, Finkbeiner, & Davis
(1998) are included in Table 1. For two objects, PKS0528+134 and PKS 1622 297, the
extinction is large due to their location close to the Galactic plane. We note that Zhang et
al. (1994) found evidence for excess absorption over the Galactic value for PKS 0528+134
from X-ray observations, which would imply a larger E(B V ) value than that from Schlegel
et al. (1998) (and hence a larger correction to the observed magnitudes). For our analysis,
however, we just use the Galactic value.
The values of E(B V ) were transformed to an absorption A  for each photometric band
using the conversion factors given in Schlegel et al. (1998). These absorption values were
then subtracted from the observed magnitudes to give the extinction-corrected magnitudes,
and both sets of data (raw and corrected) are included in Tables 2 and 3. The photometric
extinction-corrected SEDs for each object have also been plotted in Figure 1.
4

Table 3: NIR magnitudes and dates of observation. For each source, the rst row shows
the observed magnitudes, while the second row shows the magnitudes after correction for
Galactic extinction (see text).
Source J H K UT date
PKS 0208 512 13.770.03 12.930.01 12.180.02 1 Sept 2001
13.750.03 12.920.01 12.170.02
PKS 0506 612 16.680.04 16.070.03 15.160.05 1 Sept 2001
16.660.04 16.060.03 15.150.05
PKS0528+134 16.990.09 16.160.06 15.260.07 1 Sept 2001
16.210.09 15.660.06 14.940.07
PKS 1622 297 16.450.03 15.940.03 15.310.03 31 Aug 2001
16.040.03 15.680.03 15.140.03
PKS2230+114 15.250.05 14.480.04 13.730.06 1 Sept 2001
15.180.05 14.440.04 13.700.06
PKS2251+158 13.060.03 12.140.01 11.310.03 1 Sept 2001
12.960.03 12.080.01 11.270.03
Table 4: Filters used, with their central wavelengths and the zero magnitude uxes used for
ux conversion.
Filter Flux of zero magnitude 
star (F  , W cm 2 m 1 ) (m)
B 6:32  10 8 0.44
V 3:63  10 8 0.55
R 2:18  10 8 0.70
I 1:13  10 8 0.88
J 3:11  10 9 1.239
H 1:15  10 9 1.649
K 4:10  10 10 2.192
5

Table 5: Results from power law ts to the extinction-corrected data. An asterisk indicates
a t rejected at the 99% condence level. Also shown are polarisation measurements from
Impey & Tapia (1990). Both their measurements, and the maximum polarisation that they
found in the literature, are shown.
Source   2 = p (%) p max (%)
PKS 0208 512 0:82 0:27 11.5 11.5
PKS 0506 612 2:05 11:21* 1.1 1.1
PKS0528+134 2:13 2:52 0.3 0.3
PKS 1622 297 1:90 0:55 | |
PKS2230+114 1:36 0:74 7.3 10.9
PKS2251+158 1:12 1:15 2.9 16.0
3 Model Fitting
Each of the extinction-corrected SEDs has been tted by a power law, given by f  () /   ,
using a  2 -minimisation method. The t is deemed `acceptable' if the reduced  2 (i.e.  2 =,
where  is the number of degrees of freedom in the tted model) is below the 99% condence
level. For the power-law model tted to seven data points,  = 5 and the acceptance threshold
is  2 = 13:08. A t with a  2 value greater than this is rejected (with 99% condence). The
resulting ts are shown in Figure 1, and the tted parameter values|the spectral index 
and the reduced chi-squared value  2 =|are shown in Table 5.
We nd that the tted power laws can be separated into two groups. Three sources|
PKS 0506 612, PKS 0528+134, and PKS 1622 297|are quite blue (although PKS 0506 612
is a poor t, as discussed below, and should probably not be included in this group). These
slopes are the same as those found at the blue end of the distribution of tted slopes for the
Parkes Half-Jansky Flat-spectrum Sample (PHFS) quasars in Whiting, Webster, & Francis
(2001). A natural interpretation of such blue slopes is that they are due to emission from the
accretion disc. This would be expected for high-redshift sources such as PKS 0528+134, as
the observed optical emission is probing the rest-frame UV.
The remaining three sources have much redder optical slopes. Again comparing to the
PHFS sources in Whiting et al. (2001), we nd that these slopes are in the middle of the
distribution, corresponding to sources with synchrotron-dominated spectra.
One source had a poor t: PKS 0506 612. There is a large oset between the optical and
NIR points, giving what appears to be substantial spectral curvature. However, it is likely
that the source has faded in the three days between the two observations|if the spectrum
is a smooth continuum then a reduction in ux of  1 mag is implied. This also implies the
intrinsic spectrum is a red power law of slope  1:3. A similar, albeit much weaker eect
is seen in PKS 0528+134, although this could also be due to the I band data having excess
emission over the power law continuum. Contamination by the Mg ii emission line (which
would appear at  obs = 0:86m) is a likely origin of this excess.
We can use data from the literature to support the interpretation of the dierent optical
slopes. Synchrotron radiation is intrinsically highly polarised, and so a signicant net po-
larisation from a source is a good indication that the emission is dominated by synchrotron
emission. Impey & Tapia (1990) measured polarisation for ve of the sources (excluding
PKS 1622 297), and their values are also shown in Figure 5. It can be seen that the three
sources with red optical slopes are the sources that have been observed to exhibit high optical
6

polarisations, indicating a likely synchrotron origin for the emission.
4 Discussion
4.1 Demographics
All the sources we have studied here are radio-loud, at-spectrum quasars (FSRQs). We
would like to know how typical these objects are compared to other FSRQs, or whether they
are exceptional in some way that would explain their gamma-ray emission. We examine
here two properties|the redshift and the radio luminosity|and compare it to a sample of
at-spectrum radio-loud objects. For this comparison sample we use the PHFS (Drinkwater
et al. 1997), a sample of southern, radio-bright, at-spectrum objects. We also include in
this discussion the COMPTEL sources that we did not observe: PKS1222+216 (z = 0:435),
PKS1226+023 (3C 273, z = 0:158), and PKS 1253 055 (3C 279, z = 0:536). Note that the
latter two sources are also in the PHFS.
The cumulative redshift distributions for both samples are plotted in Figure 2a. Note that
we have excluded from the PHFS distribution the 33 objects with no known redshift. These
objects include a few BL Lac objects, which have no emission line from which to measure a
redshift, but are mostly optically faint, meaning no suitable spectrum has been obtained. The
COMPTEL sample tends to be biased towards lower redshift values, with only one source
at high redshift z  1 (PKS0528+134 at z = 2:07, Hunter et al. 1993). This is likely to be
a consequence of the sensitivity of the COMPTEL detector, with sources at suitably high
redshift not being bright enough at gamma rays for detection. We do note, however, that a
Kolmogorov-Smirnov (K-S) test shows that the distributions are dierent only at the 90%
condence level (the probability that they are the same is 9.9%).
Figure 2b shows the cumulative distributions of the radio luminosity for both samples.
This is calculated from the 2.7 GHz uxes from PKSCAT90 (assuming H 0 = 75 km s 1 Mpc 1
and q 0 = 0:5). We see that the COMPTEL sources tend to have, on average, higher radio
luminosities than the PHFS sources (there is a dierence in average luminosity of 0.56 dex, or
a factor of  3:6). Again, however, a K-S test shows this is not very signicant (the probability
that the distributions are dierent is 12.6%). We can, however, discuss the trends shown in
the graphs.
Even though, as we see from the redshift distribution, the COMPTEL sample lacks the
high-redshift sources (which one would expect to be more luminous than low-redshift sources
in a ux-limited sample), the average radio luminosity of this sample is still greater than that
of the PHFS. This implies that the sample selected at gamma rays lacks sources of low radio
luminosity. This is probably an indication that a certain radio luminosity is required for a
source to be active at high-energy gamma rays.
4.2 Multiwavelength SEDs
In Figure 3 we plot the multiwavelength (radio to gamma-ray) SEDs constructed from pub-
lished data, as well as the new data presented herein. Two epochs of radio observation are
shown. One set of data (indicated by the asterisks) is from PKSCAT90, while the second (in-
dicated by crosses) is from Kovalev et al. (1999). The latter does not include PKS 0208 512
and PKS 0506 612. The optical/NIR data (from this paper) have been corrected for Galactic
extinction. The X-ray data come from Brinkmann, Yuan, & Siebert (1997). The `bowtie'
7

represents the twin constraints given by the errors on the total 0.1{2.4 keV ux and the
spectral slope. (Note that the constraints on the slope for PKS 1622 297 are rather poor.)
The COMPTEL data, taken from Schonfelder et al. (2000), are indicated by the four points
between 10 20 Hz and 10 22 Hz. The horizontal error bars indicate the energy range of each
bin, while the vertical error bars indicate the ux error. Upper limits (2) are indicated by
arrows. The nal point is the EGRET datum from the Third EGRET Catalog (Hartman et
al. 1999), using their approximation to generate a 400 MeV ux density by multiplying the
catalogued ux by 1.7. Note that the observations presented here are not simultaneous.
The SEDs provide an interesting glimpse of the range of dierent spectral shapes present in
radio-loud quasars. The optically-red sources, particularly PKS 0208 512 and PKS 2251+158,
show optical{NIR SEDs that can quite easily be connected to the radio data, as would be
expected if both regimes are dominated by synchrotron emission from the relativistic jet.
For other sources, however, it is less obvious what, if any, connection exists between these
two regimes. For instance, the K band ux of PKS 1622 297 is approximately equal (on a
F  scale) to the 20 GHz ux from Kovalev et al. (1999), while the spectral index in the two
regimes is approximately the same. This, however, should not be surprising if we attribute
the optical emission to the accretion disc, as it will then not be the same component as the
radio emission. In such a picture, the synchrotron emission from the jet has turned over
somewhere in the infrared, leaving the accretion disc emission to dominate the optical. Note
that the ux level at radio frequencies is approximately equal for all sources, whereas the
redder sources tend to be brighter in the optical/NIR, indicating the presence of an additional
emission component.
Clearly, while the non-thermal processes (i.e. synchrotron and IC) are undoubtedly im-
portant for the radio and high-energy emission, the optical regime is not always going to be
dominated by them. This was shown in the detailed study of the optical/NIR properties
of radio-loud quasars by Whiting et al. (2001), which found that a large fraction of sources
showed no evidence for non-thermal synchrotron emission at optical wavelengths. The obser-
vations of strong emission-line spectra and low polarisation in the optical for the blue objects
in this sample also back this up.
This raises an important issue when it comes to modelling the multiwavelength SEDs of
such objects. Detailed models do exist in the literature, and are used with good eect to
model the high-energy emission (X-rays through gamma rays). See, for instance, Mukherjee
et al. (1999) for modelling of the multiwavelength spectra of PKS 0528+134. However, many
of these models do not take the lower-energy (i.e. radio through optical) data fully into
account. As an example, we refer the reader to Figures 8 and 9 of Mukherjee et al. (1999),
where the model ux in the optical goes in the opposite direction to the data. This seems to
be due to the tted models not fully taking into account the presence of signicant emission
from the accretion disc.
This is an important issue, since if the thermal accretion disc emission is dominating
the optical, the synchrotron emission in that frequency range is much less than previously
assumed. This will then aect the derived values for the Lorentz factors, which aect the
generation of the gamma rays via the IC process.
The photon density at the jet of the seed photons is another key parameter for calculating
the IC luminosity. For a given observed total optical ux, this parameter will be dierent for
synchrotron-dominated and thermally-dominated spectra, and its spectral shape will dier
as well. This will alter the predicted shape of the IC emission.
It can be seen, therefore, that a good understanding of the makeup of the optical emission
8

is crucial for accurate modelling of the high-energy emission. This is something that needs
to be considered for future multiwavelength campaigns involving high-energy observations
and subsequent modelling, particularly with the advent of new gamma-ray telescopes such
as INTEGRAL and GLAST.
5 Conclusions
We have observed six COMPTEL-detected quasars at optical and NIR wavelengths, and we
present photometry in seven wavebands, corrected for Galactic extinction.
We nd large dierences in the optical properties between the sources. Two of the sources
have intrinsically blue optical SEDs, and are likely to be dominated by thermal emission from
an accretion disc. A further three have much redder optical slopes, and are more likely to be
dominated in the optical by non-thermal synchrotron emission from the relativistic jet.
This identication of accretion disc emission in some gamma-ray sources has important
implications for the sources' broad-band modelling. It implies both that the thermal emis-
sion is stronger and the non-thermal emission weaker in the optical regime than previously
considered, indicating that broad-band models used to explain the high-energy emission will
need to be revisited, particularly as far as the optical regime is concerned.
Acknowledgments
The authors would like to thank A. Melatos for some very helpful discussions. We also want
to thank the MSSSO TAC for granting us time for the observations at the 40 in and 2.3 m
telescopes at the SSO.
This research has made use of the NASA/IPAC Extragalactic Database (NED) which is
operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract
with the National Aeronautics and Space Administration.
9

References
Bersanelli, M., Bouchet, P., & Falomo, R. 1991, A&A, 252, 854
Bessell, M.S., Castelli, F., & Plez, B. 1998, A&A, 333, 231
Brinkmann, W., Yuan, W., & Siebert, J. 1997, A&A, 319, 413
Bloom, S.D., & Marscher, A.P. 1996, ApJ, 461, 657
Carter, B.S., & Meadows, V.S. 1995, MNRAS, 276, 734
Collmar, W. 2001, in High Energy Gamma-Ray Astronomy, AIP Proceedings Volume 558,
ed. F.A. Aharonian & H.J. Volk (Melville, NY: American Institute of Physics), 656
Drinkwater, M.J., et al. 1997, MNRAS, 284, 85
Graham, J.A. 1982, PASP, 94, 244
Hartman, R.C., et al. 1999, ApJS, 123, 79
Hunter, S.D., et al. 1993, ApJ, 409, 134
Impey, C.D., & Tapia, S. 1990, ApJ, 354, 124
Kovalev, Y.Y., Nizhelsky, N.A., Kovalev, Yu.A., Berlin, A.B., Zhekanis, G.V., Mingaliev,
M.G., & Bogdantsov, A.V. 1999, A&AS, 139, 545
McGregor, P., et al. 1994, in Infrared Astronomy with Arrays: The Next Generation, ed. I.S.
McLean (Dordrecht: Kluwer), 299
Mukherjee, R., et al. 1999, ApJ, 527, 132
Raiteri, C.M., Villata, M., Lanteri, L., Cavallone, M., & Sobrito, G. 1998, A&AS, 130, 495
Schlegel, D.J., Finkbeiner, D.P., & Davis, M. 1998, ApJ, 500, 525
Schonfelder, V., et al. 2000, A&AS, 143, 145
Whiting, M.T., Webster, R.L., & Francis, P.J. 2001, MNRAS, 323, 718
Wright, A., & Otrupcek, R. 1990, Parkes Catalogue (Australian Telescope National Facility)
(PKSCAT90)
Zhang, Y.F., Marscher, A.P., Aller, H.D., Aller, M.F., Terasranta, H., & Valtaoja, E. 1994,
ApJ, 432, 91
10

Figure 1: Photometry and model ts for sources. The observed photometry is shown by the
red solid symbols, whereas the photometry corrected for Galactic extinction is shown with
the open symbols. The error bars shown are 1. The t to the corrected photometry is shown
by the line, with the parameters of the t (power law index and reduced- 2 ) indicated in the
corner of each plot. Note that each plot has a vertical scale spanning two decades, although
the range is dierent for each source.
11

Figure 2: Comparison of the COMPTEL sample with the PHFS, with the black line rep-
resenting the PHFS and the red line and points representing the COMPTEL sources. a)
Cumulative redshift distributions. b) Cumulative distributions of 2.7 GHz radio luminosity.
12

Figure 3: Broad-band SEDs for the six sources discussed in this paper. See text for references
for each dataset. Upper limits (i.e. non-detections) for the high-energy data are indicated by
arrows. Note that the dierent sets of observations are in general not simultaneous.
13