Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.eso.org/~ppadovan/xray_spectra/1jy_paper_II.ps Äàòà èçìåíåíèÿ: Wed Jan 5 14:06:36 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 09:26:20 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

Mon. Not. R. Astron. Soc. 000, 000--000 (0000) Printed 10 October 2003 (MN L A T E X style file v2.2)
BeppoSAX observations of 1­Jy BL Lacertae objects ­ II
Paolo Padovani 1,2# , Luigi Costamante 3,4 , Paolo Giommi 5 , Gabriele Ghisellini 6 ,
Annalisa Celotti 7 , Anna Wolter 8
1 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore MD. 21218, USA
2 A#liated with the Space Telescope Division of the European Space Agency, ESTEC, Noordwijk, The Netherlands
3 Universit‘a degli Studi di Milano, Via Celoria 16, I­20133 Milano, Italy
4 Max­Planck Institute f˜ur Kernphysik, Postfach 10 39 80, D­69029 Heidelberg (current address)
5 ASI Science Data Center, ASDC, c/o ESRIN, Via G. Galilei, I­00044 Frascati, Italy
6 Osservatorio Astronomico di Brera, Via Bianchi 46, I­23807 Merate, Italy
7 SISSA, via Beirut 2­4, I­34014 Trieste, Italy
8 Osservatorio Astronomico di Brera, Via Brera 28, I­20121 Milano, Italy
Accepted , Received
ABSTRACT
We present new BeppoSAX LECS and MECS observations, covering the energy range
0.1 - 10 keV (observer's frame), of four BL Lacertae objects selected from the 1 Jy
sample. All sources display a flat (# x # 0.7) X­ray spectrum, which we interpret as
inverse Compton emission. One object shows evidence for a low­energy steepening
(## x # 0.9) which is likely due to the synchrotron component merging into the
inverse Compton one around # 2 keV. A variable synchrotron tail would explain why
the ROSAT spectra of our sources are typically steeper than the BeppoSAX ones
(## x # 0.7). The broad­band spectral energy distributions fully confirm this picture
and model fits using a synchrotron inverse Compton model allow us to derive the
physical parameters (intrinsic power, magnetic field, etc.) of our sources. By combining
the results of this paper with those previously obtained on other sources we present a
detailed study of the BeppoSAX properties of a well­defined sub­sample of 14 X­ray
bright (f x (0.1 - 10) keV > 3 â 10 -12 erg cm -2 s -1 ) 1­Jy BL Lacs. We find a very
tight proportionality between nearly simultaneous radio and X­ray powers for the 1­Jy
sources in which the X­ray band is dominated by inverse Compton emission, which
points to a strong link between X­ray and radio emission components in these objects.
Key words: galaxies: active -- X­ray: observations
1 INTRODUCTION
BL Lacertae objects constitute one of the most extreme
classes of active galactic nuclei (AGN), distinguished by
their high luminosity, rapid variability, high ( >
# 3 per cent)
optical and radio polarization, radio core­dominance, appar­
ent superluminal speeds, and almost complete lack of emis­
sion lines (e.g., Kollgaard 1994; Urry & Padovani 1995). The
broad­band emission in these objects, which extends from
the radio to the gamma­ray band, appears to be dominated
by non­thermal processes from the heart of the AGN, undi­
luted by the thermal emission present in other AGN. There­
fore, BL Lacs represent the ideal class to study to further
our understanding of non­thermal emission in AGN.
Synchrotron emission combined with inverse Compton
scattering is generally thought to be the mechanism respon­
# Email: padovani@stsci.edu
sible for the production of radiation over such a wide energy
range (e.g., Ghisellini et al. 1998). The synchrotron peak
frequency, #peak , ranges across several orders of magnitude,
going from the far­infrared to the hard X­ray band. Sources
at the extremes of this wide distribution are referred to as
low­energy peaked (LBL) and high­energy peaked (HBL) BL
Lacs, respectively (Giommi & Padovani 1994; Padovani &
Giommi 1995). Radio­selected samples include mostly ob­
jects of the LBL type, while X­ray selected samples are
mostly made up of HBL.
This scenario makes clear and strong predictions on the
X­ray spectra for the two classes. In the relatively narrow
ROSAT band di#erences between the two classes became
apparent only when very large BL Lac samples (# 50 per
cent of the then known objects) were considered (Padovani
& Giommi 1996; Lamer, Brunner & Staubert 1996). The
BeppoSAX satellite with its broad­band X­ray (0.1 - 200
keV) spectral capabilities, is particularly well suited for a
c
# 0000 RAS

2 P. Padovani et al.
detailed analysis of the individual X­ray spectra of these
sources.
In Padovani et al. (2001; Paper I) we presented Bep­
poSAX observations of seven 1­Jy BL Lacs (plus an ad­
ditional ``intermediate'' source), six of them LBL. We dis­
cuss here BeppoSAX observations of four additional 1­Jy BL
Lacs, all of the LBL type.
In Section 2 we present our sample, Section 3 discusses
the observations and the data analysis, while Section 4 de­
scribes the results of our spectral fits to the BeppoSAX data.
Section 5 deals with the ROSAT data for our sources, while
Section 6 presents the spectral energy distributions. Section
7 discusses the X­ray properties of all 1­Jy BL Lacs observed
by BeppoSAX , while Section 8 presents our conclusions.
Throughout this paper spectral indices are written S# #
# -# and the values H0 = 50 km s -1 Mpc -1 and q0 = 0 have
been adopted.
2 THE SAMPLE
The 1­Jy sample of BL Lacs is presently the only sizeable,
complete sample of radio bright BL Lacs. It includes 34 ob­
jects with radio flux > 1 Jy at 5 GHz and # r # 0.5 (Stickel
et al. 1991). All 1­Jy BL Lacs have been studied in detail in
the radio and optical bands; all objects have also soft X­ray
data, primarily from ROSAT.
We selected for BeppoSAX observations all 1­Jy BL
Lacs with 0.1 - 10 keV X­ray flux larger than 2 â 10 -12
erg cm -2 s -1 (estimated from an extrapolation of the single
power­law fits derived for these objects from ROSAT data;
Urry et al. 1996). This included twenty 1­Jy BL Lacs (or
# 60 per cent of the sample). Observing time was granted
for 11 of them, while three more sources have been included
in other BeppoSAX programs (MKN 501: Pian et al. 1998;
S5 0716+714: Giommi et al. 1999; PKS 0537-441: Pian et
al. 2002). We present here the results obtained for the four
objects observed in Cycle 2, all LBL. The object list and ba­
sic characteristics are given in Table 1, which presents the
source name, position, redshift, R magnitude, 5 GHz radio
flux, and Galactic NH .
3 OBSERVATIONS AND DATA ANALYSIS
A complete description of the BeppoSAX mission is given by
Boella et al. (1997a). The relevant instruments for our obser­
vations are the coaligned Narrow Field Instruments (NFI),
which include one Low Energy Concentrator Spectrometer
(LECS; Parmar et al. 1997) sensitive in the 0.1 -- 10 keV
band; two (three before 1997 May) identical Medium Energy
Concentrator Spectrometers (MECS; Boella et al. 1997b),
covering the 1.5 -- 10 keV band; and the Phoswich Detec­
tor System (PDS; Frontera et al. 1997), coaligned with the
LECS and the MECS. The PDS instrument is made up of
four units, and was operated in collimator rocking mode,
with a pair of units pointing at the source and the other
pair pointing at the background, the two pairs switching on
and o# source every 96 seconds. The net source spectra have
been obtained by subtracting the `o#' to the `on' counts. A
journal of the observations is given in Table 2.
The data analysis was based on the linearized, cleaned
event files obtained from the BeppoSAX Science Data Cen­
ter (SDC) on­line archive (Giommi & Fiore 1997). The data
from the two MECS instruments were merged in one single
event file by the SDC, based on sky coordinates. The event
file was then screened with a time filter given by SDC to
exclude those intervals related to events without attitude
solution (i.e., conversion from detector to sky coordinates;
see Fiore et al. 1999). This was done to avoid an artificial
decrease in the flux. As recommended by the SDC, the chan­
nels 1--10 and above 4 keV for the LECS and 0--36 and 220--
256 for the MECS were excluded from the spectral analysis,
due to residual calibration uncertainties. None of our sources
was detected by the PDS.
Spectra and lightcurves were extracted using the stan­
dard radius of 4 arcmin for the MECS. As regards the LECS,
the recommended value of 8 arcmin was used only for S5
1803+784, while 6 and 4 arcmin were used for AO 0235+164
and OQ 530, respectively, due to their very low signal­to­
noise ratio (S/N). A 4 arcmin radius was used for 3C 371,
due to the presence of a serendipitous source at # 7 arcmin
(see below).
The spectral analysis was performed using the matrices
and blank­sky background files released in 1998 November
by the SDC, with the blank­sky files extracted in the same
coordinate frame as the source file, as described in Paper I.
Because of the importance of the band below 1 keV to as­
sess the presence of extra absorption or soft excess (indica­
tive of a low­energy steepening of the spectrum), we have
also checked the LECS data for di#erences in the cosmic
background between local and blank­sky field observations,
comparing spectra extracted from the same areas on the de­
tector (namely, two circular regions outside the 10 arcmin
radius central region, located at the opposite corners with
respect to the two on­board radioactive calibration sources).
No significant di#erences were found.
Using the software package XRONOS we looked for
time variability in every observation, binning the data in
intervals from 500 to 3600 s, with null results.
4 SPECTRAL FITS
Spectral analysis has been performed with the XSPEC 10.00
package, using the response matrices released by the SDC
in 1998. Using the program GRPPHA, the spectra were re­
binned with more than 20 counts in every new bin and using
the rebinning files provided by SDC. Various checks using
di#erent rebinning strategies have shown that our results are
independent of the adopted rebinning within the uncertain­
ties. The data were analyzed applying the Gehrels statistical
weight (Gehrels 1986) in case the resulting net counts were
below 20 (typically 12­15 in the low energy band of LECS).
The LECS/MECS normalization factor was left free to vary
in the range 0.65--1.0, as suggested by SDC (see Fiore et al.
1999). The X­ray spectra of our sources are shown in Fig. 1.
4.1 Single Power­law Fits
At first, we fitted the combined LECS and MECS data with
a single power­law model with Galactic and free absorption.
The absorbing column was parameterized in terms of NH ,
the HI column density, with heavier elements fixed at solar
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 3
Table 1. Sample Properties.
Name RA(J2000) Dec(J2000) z R a
mag F 5GHz Galactic NH
(Jy) (10 20 cm -2 )
AO 0235+164 02 38 38.9 +16 36 59 0.940 16.5 2.9 7.60 b
OQ 530 14 19 46.6 +54 23 15 0.152 14.5 1.1 1.20 c
S5 1803+784 18 00 45.7 +78 28 04 0.684 16.5 2.6 3.65 d
3C 371 18 06 50.6 +69 49 28 0.051 14.0 2.3 4.38 d
a Mean R magnitude from Heidt & Wagner (1996).
b From Elvis et al. (1989).
c Dickey & Lockman (1990).
d From Murphy et al. (1996).
Table 2. BeppoSAX Journal of observations.
Name LECS LECS MECS MECS Observing date
Exp. Count rate a Exp. Count rate a
(s) (cts/s) (s) (cts/s)
AO 0235+164 24575 0.006 ± 0.002 29071 0.011 ± 0.001 1999 Jan 28
OQ 530 16312 0.006 ± 0.002 39680 0.008 ± 0.001 1999 Feb 12­13
S5 1803+784 18359 0.013 ± 0.002 40303 0.028 ± 0.001 1998 Sep 28
3C 371 13090 0.018 ± 0.002 33760 0.024 ± 0.001 1998 Sep 22­23
a Net count rate full band.
abundances and cross sections taken from Morrison and Mc­
Cammon (1983). The NH value for AO 0235+164 was also
fixed at a value larger than the Galactic one to take into ac­
count absorption by an intervening galaxy (Madejski et al.
1996). NH was also set free to vary for all sources to check
for internal absorption and/or indications of a ``soft­excess.''
Our results are presented in Table 3, which gives the
name of the source in column (1), NH in column (2), the
energy index #x in column (3), the 1 keV flux in µJy in
column (4), the unabsorbed 2 - 10 keV and 0.1 - 2.4 keV
fluxes in columns (5)­(6), the LECS/MECS normalization in
column (7), the reduced chi­square and number of degrees
of freedom, # 2
# /(dof) in column (8), and the F --test proba­
bility that the decrease in # 2 due to the addition of a new
parameter (free NH) is significant in column (9). The errors
quoted on the fit parameters are the 90 per cent uncertain­
ties for one and two interesting parameters, for Galactic and
free NH respectively. The errors on the 1 keV flux reflect the
statistical errors only and not the model uncertainties.
Two results are immediately apparent from Table 3.
First, the fitted NH values do not agree with the Galactic
ones for AO 0235+164 (as expected from previous ASCA
and ROSAT observations; see Sect. 4.2.1) and possibly OQ
530. An F ­test shows that the addition of NH as a free pa­
rameter results in a (marginally significant) improvement in
the # 2 values only for AO 0235+164 (column 9 of Table
3). Second, the fitted energy indices are flat, #x < 1, with
##x# = 0.67±0.11 and a weighted mean equal to 0.59±0.09.
One of our sources, OQ 530, appears to show a low­
energy excess, as illustrated by Fig. 1 and by the fact that
the best fit NH in Table 3 is below the Galactic value (and ac­
tually consistent with zero). We then fitted a broken power­
law model to the data. The results are presented in Table
4, which gives the name of the source in column (1), NH in
column (2), the soft energy index #S in column (3), the hard
energy index #H in column (4), the break energy in column
(5), the 1 keV flux in µJy in column (6), the unabsorbed
2 - 10 keV and 0.1 - 2.4 keV fluxes in columns (7)­(8),
the LECS/MECS normalization in column (9), the reduced
chi­squared and number of degrees of freedom, # 2
# /(dof), in
column (10), and the F --test probability that the decrease
in # 2 due to the addition of two parameters (from a single
power­law fit to a broken power­law fit) is significant in col­
umn (11). Although the fit is improved by using a double
power­law model, an F --test shows that the improvement is
more suggestive than significant, with a probability # 90 per
cent. The best­fit spectrum, however, points in the direction
of a flatter component emerging at higher energies, as was
the case for four sources in Paper I. In fact, the spectrum is
concave with a large spectral change ##S - #H# # 0.9, and
an energy break around E # 2 keV. Evidence for a concave
spectrum comes also from the shape of the ratio of the data
to the single power­law fits, shown in Fig. 1.
4.2 Notes on individual sources
4.2.1 AO 0235+164
This source is in a crowded field, with three absorbing sys­
tems at z = 0.524, 0.851, and 0.94 in a 10x10 arcsec region
and a probable Seyfert 1 galaxy at less than 5 arcsec (see
Burbidge et al. 1996). ROSAT and ASCA spectra show ab­
sorption above the Galactic value, which our data confirm,
likely due to the intervening absorption system at z = 0.524
(Madejski et al. 1996). A fit with free absorption, in fact,
gives an NH value slightly higher than the corresponding
ASCA one (3.7â10 21 cm -2 vs. 2.8 â10 21 cm -2 ) but still
consistent within the errors. Since the ASCA data provide
a better constraint on the column density, we have also fit­
ted our data with total NH fixed at the ASCA value. The
BeppoSAX X­ray spectral index and flux are consistent with
c
# 0000 RAS, MNRAS 000, 000--000

4 P. Padovani et al.
Table 3. Single power­law fits, LECS + MECS a .
Name NH #x F 1keV F [2-10] F [0.1-2.4] Norm # 2
# /dof F ­test, notes
(10 20 cm -2 ) (µJy) (c.g.s.) (c.g.s.) (L/M) fixed­free NH
AO 0235+164 7.6 fixed 0.79 +0.24
-0.24 0.16 +0.06
-0.05 8.54e­13 1.08e­12 0.70 0.97/14
37 +85
-33 1.01 +0.47
-0.41 0.22 +0.16
-0.08 8.43e­13 1.71e­12 0.82 0.78/13 94 per cent
28 b fixed 0.96 +0.27
-0.26 0.20 +0.08
-0.06 8.42e­13 1.51e­12 0.79 0.74/14 NH at ASCA and ROSAT values
OQ 530 1.20 fixed 0.55 +0.27
-0.32 0.09 +0.03
-0.03 6.74e­13 5.24e­13 0.74 1.12/15
0.0(< 7.38) 0.55 +0.32
-0.39 0.09 +0.03
-0.03 6.75e­13 5.24e­13 0.71 1.09/14 74 per cent
S5 1803+784 3.65 fixed 0.45 +0.12
-0.12 0.24 +0.04
-0.04 2.21e­12 1.41e­12 0.67 0.97/22
10(< 33) 0.51 +0.20
-0.20 0.26 +0.07
-0.05 2.20e­12 1.57e­12 0.67 0.97/21 65 per cent
3C 371 4.38 fixed 0.72 +0.16
-0.16 0.30 +0.07
-0.07 1.76e­12 1.93e­12 0.72 0.86/30
5.6 +19
-4.6 0.74 +0.24
-0.24 0.30 +0.08
-0.08 1.76e­12 1.99e­12 0.73 0.89/29 30 per cent
a Errors are at 90 per cent confidence level for one (with fixed NH ) and two parameters of interest.
b Absorption including the contribution of an intervening system at z=0.524 (Madejski et al. 1996).
Table 4. Broken power­law fits, LECS + MECS a .
Name NH # S #H E break F 1keV F [2-10] F [0.1-2.4] L/M # 2
# /dof F ­test b
(10 20 cm -2 ) (keV) (µJy) (c.g.s.) (c.g.s.)
OQ 530 1.20 fixed 1.3 +4.0
-0.7 0.4 +0.4
-0.4 1.8 +1.1
-1.5 0.12 +0.03
-0.07 7.31e­13 1.10e­12 0.66 0.91/13 90 per cent
a Errors are at 90 per cent confidence level for two parameters of interest.
b The values of the F --test refer to the comparison with a single power­law model with Galactic column density.
the values of Madejski et al. (1996), under the same assump­
tions.
Given the BeppoSAX point spread function, it is nearly
impossible to assess the likelihood of a contribution from the
Seyfert galaxy to the X­ray flux of our source. However, the
MECS profile is the one expected from a point­like source,
with no hints of extension. Moreover, using its observed op­
tical magnitude (# 20.5) and assuming an optical­X­ray ef­
fective spectral index typical of Seyfert 1s (#ox # 1.2), one
can estimate an X­ray flux from this nearby source roughly
a factor of 10 smaller than that observed for AO 0235+164.
4.2.2 OQ 530
This source shows some evidence of a steeper spectrum at
low energies, but the F ­test significance is only marginal,
due to the low statistics and narrow range a#ected. Taglia­
ferri et al. (2003) have also presented evidence for a low­
energy excess, with a very steep #S # 6.7 and #H = 0.4±0.2
(2000 March 3­4) and #H = 0.75±0.20 (2000 March 26­27).
In their case the F ­test probability for improvement for the
double power­law model is > 99.9 per cent and # 80 per
cent, respectively. The fluxes for the two observations are
similar, f(2 - 10keV) # 10 -12 erg cm -2 s -1 , # 50 per cent
larger than what we find.
4.2.3 3C 371
A serendipitous source identified as RIXOS F272 023, a
Seyfert 1.8 galaxy at z = 0.096, is present in the LECS and
MECS images, at a distance of # 7 arcmin (Puchnarewicz
et al. 1997). The extraction of LECS data was then done by
using a radius of 4 arcmin.
5 ROSAT PSPC DATA
In order to compare our results with previous (soft) X­ray
observations and especially to take advantage of the higher
resolution and collecting area at low energies, we used data
from the ROSAT Position Sensitive Proportional Counter
(PSPC). The 1­Jy BL Lac ROSAT data had been originally
published by Urry et al. (1996). In order to ensure a uni­
form procedure for the whole sample, we have re­analyzed
all ROSAT data, obtaining results consistent within the er­
rors with those already published.
The journal of the ROSAT observations is given in Table
5. The basic event files from the archive have been corrected
for gain variations on the detector surface with the program
PCSASSCORR in FTOOLS, when not already done by the
Standard Reduction process (SASS version 7 8 and later,
M. Corcoran, private communication). Since all the sources
were ROSAT targets a standard extraction radius of 3 # (2.5 #
when serendipitous sources were present in the field or when
the source was particularly weak) was used, to avoid the
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 5
Table 5. ROSAT journal of observations.
Name Exposure Full band net countrate Observing Date
(s) (cts/s)
AO 0235+164 17949 0.170 ± 0.003 1993 Jul 21 -- Aug 15
OQ 530 11474 0.223 ± 0.005 1990 Jul 19 -- 23
S5 1803+784 6781 0.081 ± 0.004 1992 Apr 7
2773 0.105 ± 0.008 1992 Jul 25
2827 0.108 ± 0.008 1992 Dec 10
3C 371 10348 0.127 ± 0.004 1992 Apr 9
possible loss of soft photons due to the ghost imaging ef­
fect. We have used the appropriate response matrices for
the di#erent gain levels of the PSPC B detector before and
after 14 Oct. 1991 (gain1 and gain2, respectively). The back­
ground has been evaluated in two circular regions (of radius
# 20 - 30 pixels) away from the central region and from
other serendipitous sources, but inside the central rib of the
detector. The spectra have been rebinned (using GRPPHA)
with more than 20 counts in every new bin, to validate the
use of # 2 statistics. Channel 1­11 and 212­256 have been
excluded from the analysis, due to calibration uncertainties.
As for the BeppoSAX data, we fitted the ROSAT PSPC
data with a single power­law model with Galactic and free
absorption. Our results are presented in Table 6, which gives
the name of the source in column (1), NH in column (2),
the energy index #x in column (3), the 1 keV flux in µJy in
column (4), the unabsorbed 0.1-2.4 keV flux in column (5),
the reduced chi­squared and number of degrees of freedom,
# 2
# /(dof) in column (6), and the observing date in column
(7).
Table 6 shows that the fitted NH values are consistent
with the Galactic ones for OQ 530 and S5 1803+784. AO
0235+164 shows evidence of excess absorption, as was the
case for our BeppoSAX data, while 3C 371 displays a soft­
excess, suggestive of a steeper soft component. A broken
power­law fit, in fact, improves the fit significantly (99.7
per cent), as shown in Table 7. The spectrum is obviously
concave with quite a large spectral change, with ##S-#H# #
2, and an energy break around E # 0.3 keV.
The fitted energy indices are relatively steep. The mean
ROSAT value for our sources is ##x# = 1.34 ± 0.18, while
the weighted mean is ##x# = 1.21 ± 0.04.
5.1 Notes on individual sources
5.1.1 AO 0235+164
Nine ROSAT observations are available, spaced by about
# 3 days, between July 21 and August 15 1993. The source
shows some interesting variability, but with no evident spec­
tral variations, as the hardness ratio is roughly constant.
Therefore, the data have been summed together (see Made­
jski et al. 1996 for a discussion of the single observations).
5.1.2 OQ 530
The ROSAT observations are relatively old and done with
the PSPC C detector. The spectrum has many ''wiggles'',
which suggest that perhaps the calibration is not optimal.
The resulting reduced # 2 is not very good, but alternative
models (free NH , broken power­law) do not improve the fit.
5.1.3 S5 1803+784
Three ROSAT observations are available. The best­fit
NH agrees with the Galactic value. The source displays in­
teresting spectral and intensity variations, with a ``steeper
when brighter'' behaviour, which suggests properties typi­
cal of ``intermediate'' BL Lacs. These are possibly associ­
ated with a shift of the synchrotron peak or a hardening
of the injected electron spectrum in higher states, so that
synchrotron emission becomes dominant in the soft X­ray
band.
5.2 Comparison between BeppoSAX and
ROSAT results
Figure 2 shows the 0.1 - 2.4 keV BeppoSAX flux versus
the corresponding ROSAT flux. Our sources display mild
X­ray variability: the median value of fBeppoSAX /fROSAT is
0.4. Fig. 2 includes also the BL Lacs studied in Paper I. Most
sources have fBeppoSAX < fROSAT , as shown by the fact that
the median value for all 12 sources is fBeppoSAX /fROSAT is
0.4. Fig. 2 should be compared with Fig. 2 of Wolter et al.
(1998; see also Beckmann et al. 2002), which plots the Bep­
poSAX 1 keV fluxes versus the corresponding ROSAT fluxes
for a sample of 8 HBL. There the two fluxes are within 30
per cent for most sources and the points follow more closely
the line of equal fluxes. Note that the median value of the
flux ratio at 1 keV is # 0.5 for the 4 sources studied in this
paper (and # 0.6 if one includes those studied in Paper I).
Figure 3 shows the BeppoSAX spectral index (0.1 - 10
keV) vs. the ROSAT spectral index (0.1 - 2.4 keV). The
larger BeppoSAX error bars for most of our sources, as
compared to ROSAT, are due to the worse photon statis­
tics. (The PSPC count rates, in fact, are typically a fac­
tor of 20 larger than the LECS ones.) All sources stud­
ied in this paper occupy the region of the plot where
#x(BeppoSAX) < #x(ROSAT ). The interpretation of this
plot is complicated by variability e#ects, which a#ect the
shape of the X­ray spectrum, and possibly by ROSAT mis­
calibrations (e.g., Iwasawa, Fabian & Nandra 1999). Keep­
ing this in mind, in agreement with our previous results
the figure suggests a concave overall X­ray spectrum for our
sources, with a flatter component emerging at higher ener­
gies. We find #x(ROSAT)-#x(BeppoSAX) = 0.7±0.1. This
di#erence can hardly be explained by miscalibration e#ects
which, if present, should only steepen the ROSAT slopes
c
# 0000 RAS, MNRAS 000, 000--000

6 P. Padovani et al.
Table 6. ROSAT PSPC, single power­law fits a .
Name NH #x F 1keV F [0.1-2.4] # 2
# /dof Observing Date
(10 20 cm -2 ) (µJy) (erg cm -2 s -1 )
AO 0235+164 7.60 fixed 0.50 ± 0.08 0.64 ± 0.02 3.84e­12 3.22/30 1993 Jul 21 -- Aug 15
33 +12
-10 2.1 ± 0.7 1.4 +0.4
-0.3 4.45e­11 0.79/29
28 b fixed 1.88 ± 0.11 1.21 ± 0.04 2.37e­11 0.79/30
OQ 530 c 1.20 fixed 1.07 ± 0.05 0.30 ± 0.02 2.47e­12 1.32/76 1990 Jul 19 -- 23
1.2 +0.4
-0.3 1.1 ± 0.2 0.30 ± 0.02 2.50e­12 1.34/75
S5 1803+784 3.65 fixed 0.84 ± 0.15 0.23 ± 0.02 1.61e­12 0.89/21 1992 Apr 7
4.2 +1.9
-1.6 1.0 ± 0.4 0.24 ± 0.02 1.79e­12 0.91/20
3.65 fixed 1.2 ± 0.2 0.27 ± 0.03 2.40e­12 0.68/10 1992 Jul 25
3.5 +2.4
-2.0 1.1 ± 0.7 0.27 ± 0.04 2.28e­12 0.76/9
3.65 fixed 1.2 ± 0.2 0.26 ± 0.03 2.38e­12 1.32/10 1992 Dec 10
4.6 +2.8
-2.4 1.5 ± 0.8 0.26 ± 0.04 3.20e­12 1.40/9
3C 371 4.38 fixed 1.21 ± 0.10 0.35 ± 0.02 3.20e­12 1.36/25 1992 Apr 9
2.7 +0.9
-0.8 0.8 ± 0.3 0.33 ± 0.02 2.21e­12 0.85/24
a Errors are at 90 per cent confidence level for one (with fixed NH ) and two parameters of interest.
b Absorption including the contribution of an intervening system at z=0.524 (Madejski et al. 1996).
c PSPC C
Table 7. Broken power­law fits, ROSAT a .
Name NH # S #H E break F 1keV F [0.1-2.4] # 2
# /dof F ­test b
(10 20 cm -2 ) (keV) (µJy) (c.g.s.)
3C 371 4.38 fixed 3.0 +2.4
-2.2 0.9 ± 0.2 0.34 +0.64
-0.06 0.3 ± 0.1 5.60e­12 0.88/23 99.7 per cent
a Errors are at 90 per cent confidence level for two parameters of interest.
b The values of the F --test refer to the comparison with a single power­law model with Galactic column density.
by # 0.2 - 0.3 (see also Mineo et al. 2000). Fig. 3 includes
also the BL Lacs studied in Paper I. Note that only one BL
Lac has #x(ROSAT ) < #x(BeppoSAX) and that for all 12
sources we find #x(ROSAT)-#x(BeppoSAX) = 0.53±0.16
(0.53 ± 0.18 excluding the two HBL).
Again, this figure should be compared with Fig. 3 of
Wolter et al. (1998), which shows the same plot for a sam­
ple of 8 HBL. In that case the BeppoSAX and ROSAT spec­
tral indices agree within the errors for all but one source. A
sample of 10 HBL studied by Beckmann et al. (2002) shows
more scatter but the mean values for the spectral indices are
still similar.
Looking at the di#erences between BeppoSAX and
ROSAT spectra in more detail, OQ 530 shows, as discussed
above (see Tab. 4), evidence of a steeper BeppoSAX spec­
trum at low energies (very significant for the data presented
by Tagliaferri et al. 2003), consistent with the ROSAT data.
Similarly, the ROSAT spectrum of 3C 371 is best fitted
by a broken power­law with a hard component consistent
with the BeppoSAX spectrum and a steep, soft component
which dominates only for E <
# 0.3 keV. Could a ROSAT­
like component be also present in the BeppoSAX spectra of
our other two sources but not have been detected? The an­
swer is: no. We have simulated this by assuming a broken
power­law model with a break at 2.4 keV (the end of the
ROSAT band) and a hard component with spectral index
equal to that measured by BeppoSAX. A soft component
with spectral index equal to that seen by ROSAT can be
excluded with very strong significance (> 99 per cent) for
both AO 0235+164 and S5 1803+784. If the flux were also
constrained to be the same as the ROSAT one the signifi­
cance of the exclusion would be even higher. In other words,
BeppoSAX did not detect a steep, ROSAT­like component
in these two sources because it was not there.
6 SPECTRAL ENERGY DISTRIBUTIONS
To address the relevance of our BeppoSAX data in terms of
emission processes in BL Lacs, we have assembled multifre­
quency data for all our sources. The main source of informa­
tion was NED, and so most data are not simultaneous with
our BeppoSAX observations. For all our sources, however,
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 7
we were able to find nearly­simultaneous (typically within a
month) radio observations in the University of Michigan Ra­
dio Astronomy Observatory (UMRAO) database. These are
reported in Table 8, which also gives the nearly­simultaneous
radio­X­ray spectral index, #rx with its error. This is defined
between the rest­frame frequencies of 4.8 GHz and 1 keV,
and has been K­corrected using the X­ray spectral indices
given in Tab. 3 and radio spectral indices between 4.8 and
Figure 1. BeppoSAX data and fitted spectra for our sources,
and ratio of data to fit. Data are from the LECS and MECS
instruments. The data are fitted with a single power­law model
with Galactic absorption, apart from AO 0235+164, for which
the absorption includes also the contribution of an intervening
system at z = 0.524 (Madejski et al. 1996).
8.0 GHz derived from the UMRAO data. One of our sources
(AO 0235+164) has been detected by EGRET so its energy
distribution reaches # 5 â 10 24 Hz. The EGRET data come
from the compilation of Lin et al. (1999), which include the
first entries in the Third EGRET Catalog.
The spectral energy distributions (SEDs) for our
sources are shown in Fig. 4, where filled circles indicate Bep­
poSAX data and the nearly­simultaneous radio data, and
open symbols represent non­simultaneous literature (mostly
NED) data. The BeppoSAX data have been converted to
#f# units using the XSPEC unfolded spectra after correct­
ing for absorption. ROSAT (from Tab. 6 and 7) and EGRET
data are shown by a bow­tie that represents the spectral in­
dex range. The plotted ROSAT data correspond to the fixed
galactic column density fits (for AO 0235+164 the one in­
cluding the extra absorption), and to the highest and lowest
flux observed. For 3C 371, we plotted the spectral index
range from the broken power­law fits (above 0.3 keV; Tab.
7). We also show the available ASCA data for AO 0235+164
(Madejski et al. 1996) and 3C 371 (Donato et al. 2001).
To derive the intrinsic physical parameters that could
account for the observed data, we have fitted the SED of our
sources with an homogeneous, one--zone synchrotron inverse
Compton model as developed in Ghisellini, Celotti & Costa­
mante (2002). This model is very similar to the one described
in detail in Spada et al. (2001; it is the ``one--zone'' version
of it), and is characterized by a finite injection timescale, of
the order of the light crossing time of the emitting region (as
occurs, for example, in the internal shocks scenario, where
the dissipation takes place during the collision of two shells
of fluid moving at di#erent speeds).
In this model, the main emission comes from a single
zone and a single population of electrons, with the particle
energy distribution determined at the time t inj , i.e., at the
end of the injection, which is the time when the emitted lu­
minosity is maximized. Details of the model can be found
c
# 0000 RAS, MNRAS 000, 000--000

8 P. Padovani et al.
Figure 2. The 0.1-2.4 keV X­ray flux from our BeppoSAX data
vs. the corresponding ROSAT flux (filled points). The dashed line
represents the locus of f BeppoSAX = fROSAT . For S5 1803+784,
which has multiple ROSAT observations, we took the observation
with the largest X­ray flux (1992 July). Open squares represent
the BL Lacs studied in Paper I.
Figure 3. The BeppoSAX spectral index (0.1 - 10 keV) vs. the
ROSAT spectral index (0.1 - 2.4 keV) for our sources (filled
points). The dashed line represents the locus of #x (BeppoSAX ) =
#x (ROSAT ). For S5 1803+784, which has multiple ROSAT ob­
servations, we took the observation with the largest X­ray flux
(1992 July). Open squares represent the BL Lacs studied in Pa­
per I.
in Ghisellini et al. (2002), who have applied it successfully
to both low--power, high­peaked BL Lacs and powerful flat­
spectrum radio quasars. A summary of the model main char­
acteristics and a discussion of its application to the SEDs of
other BeppoSAX sources can also be found in Padovani et
al. (2001; 2002).
Some of our sources have (weak) broad lines and there­
fore the contribution of the disc to the SED might not be
completely negligible. Furthermore, photons produced in the
broad line region could contribute to the seed photon dis­
tribution for the inverse Compton scattering. We accounted
for this by assuming that a fraction # 10% of the disc lu­
minosity L disc is reprocessed into line emission by the broad
line region (BLR), LBLR , assumed to be located at RBLR .
LBLR estimates for three of our sources were found in Celotti
et al. (1997). RBLR was assumed to scale as L 0.7
disc , following
Kaspi et al. (2000) and disc emission was assumed to be a
simple black--body peaking at 10 15 Hz.
The source is assumed to emit an intrinsic luminosity
L # and to be observed with the viewing angle #. The input
parameters are listed in Table 9, which gives the name of
the source in column 1, L # in column 2, L disc in column 3,
RBLR in column 4, the magnetic field B in column 5, the
size of the region R in column 6, the Lorentz factor # in
column 7, the angle # in column 8, the slope of the particle
distribution n in column 9, the minimum Lorentz factor of
the injected electrons #min in column 10, the Lorentz factor
of the electrons emitting most of the radiation #peak in col­
umn 11, and finally # syn
peak in column 12. Note that #peak and
# syn
peak are derived quantities and not input parameters.
The model fits are shown in Fig. 4 as solid lines. The
applied model is aimed at reproducing the spectrum origi­
nating in a limited part of the jet, thought to be responsible
for most of the emission. This region is necessarily compact,
since it must account for the fast variability shown by all
blazars, especially at high frequencies. The radio emission
from this compact regions is strongly self--absorbed, and the
model cannot account for the observed radio flux. This ex­
plains why the radio data are systematically above the model
fits in the figures. For AO 0235+164 we have also modelled
a high state corresponding to the hardest EGRET spec­
trum, which can be accounted for by assuming also a high
X­ray state (ROSAT data). This is not compatible (in our
model) with the BeppoSAX data, which instead correspond
to the lowest X­ray state for this source, and can be fitted
well together with the low­state EGRET data, taken quasi­
simultaneously with the ASCA observation (whose flux and
spectrum are very similar to the BeppoSAX ones; Madejski
et al. 1996).
As shown in Table 9, the source dimensions, magnetic
field, bulk Lorentz factors, and viewing angles are quite sim­
ilar for all sources. The need for external seed photons for
some sources, while indicative of a broad line region, is not
extremely compelling, since the disc luminosities are much
smaller than those required in radio--loud quasars (see, e.g.,
Ghisellini et al. 1998). The main di#erence between sources
are in the derived value of #peak and intrinsic luminosities.
Fig. 4 shows that the BeppoSAX band is dominated
by inverse Compton emission for S5 1803+784 and 3C 371,
while it is in between the synchrotron and inverse Compton
regions for AO 0235+164 and OQ 530. These results are
consistent with the BeppoSAX data, as discussed in Sect. 4.
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 9
Figure 4. Spectral energy distributions for our sources. Filled symbols indicate BeppoSAX data and nearly­simultaneous radio data
from UMRAO, while open symbols indicate data from NED. The solid lines correspond to the one--zone homogeneous synchrotron and
inverse Compton model calculated as explained in the text, with the parameters listed in Table 9. The dashed line represents the fit to
the high state of AO 0235+164. ROSAT, ASCA (for AO 0235+164 and 3C 371), and EGRET (high­state) data are shown by a bow­tie
that represents the spectral index range.
Table 8. Nearly­simultaneous Radio Observations.
Name F 4.8GHz Observing date F 8.0GHz Observing date F 14.5GHz Observing date #rx
(Jy) (Jy) (Jy)
AO 0235+164 1.54 ± 0.03 1999 Jan 30 1.55 ± 0.08 1999 Jan 26 1.37 ± 0.03 1999 Jan 22 0.86 ± 0.02
OQ 530 0.59 ± 0.08 1998 Sep 24 0.66 ± 0.06 1999 Feb 15 0.41 ± 0.01 1999 Mar 17 0.88 ± 0.02
S5 1803+784 2.67 ± 0.03 1998 Sep 25 2.88 ± 0.17 1998 Sep 18 2.29 ± 0.20 1998 Sep 30 0.90 ± 0.01
3C 371 1.53 ± 0.03 1998 Sep 26 1.82 ± 0.18 1998 Sep 18 1.28 ± 0.03 1998 Sep 9 0.87 ± 0.01
6.1 X­ray Spectral Index and the Synchrotron
Peak Frequency
One of the aims of this project was to study the depen­
dence of the X­ray spectral index on the synchrotron peak
frequency #peak found by Padovani & Giommi (1996) and
Lamer et al. (1996) from ROSAT data by using the broader
BeppoSAX energy band. Padovani & Giommi (1996) found a
strong anti­correlation between #x and #peak for HBL (i.e.,
the higher the peak frequency, the flatter the spectrum),
while basically no correlation was found for LBL. This was
interpreted as due to the tail of the synchrotron compo­
c
# 0000 RAS, MNRAS 000, 000--000

10 P. Padovani et al.
Table 9. Model Parameters.
Name L # L disc RBLR B R # # n #min # peak # syn
peak
erg s -1 erg s -1 cm G cm Hz
AO 0235+164 1.1e43 1.3e45 4.2e17 3.8 4.1e16 16 2.9 3.5 20 276 2.1e13
(high state) 6.3e43 1.3e45 4.2e17 1.7 2.3e16 13 2.6 3.9 3600 3600 1.5e15
OQ 530 8.0e41 ... ... 3.8 1.25e16 12 4.9 3.5 70 1410 3.3e14
S5 1803+784 6.8e42 3.9e45 8.8e17 5.0 2.0e16 11 3.5 3.5 230 373 3.9e13
3C 371 2.2e41 4.8e42 1.1e16 1.8 9.0e15 13 5.0 3.3 10 5000 1.9e15
Figure 5. The BeppoSAX spectral index (0.1 - 10 keV) vs. the
logarithm of the peak frequency for our sources (open pentagons),
those studied in Paper I (filled circles), the HBL studied by Wolter
et al. (1998) and Beckmann et al. (2002) (open squares), PKS
0537-441 (filled pentagon), ON 231 (star), S5 0716+714 (open
circle), PKS 2155-304 (filled triangle), MKN 421 (cross), 1ES
2344+514 (open triangle), and MKN 501 (filled square). See text
for details and references.
nent becoming increasingly dominant in the ROSAT band as
#peak moves closer to the X­ray band (see Fig. 7 of Padovani
& Giommi 1996).
The updated BeppoSAX version of this dependence is
shown in Fig. 5, which plots the BeppoSAX spectral index
(0.1 - 10 keV) vs. the logarithm of the peak frequency for
our sources (open pentagons), the other 1­Jy sources stud­
ied in Paper I (filled circles), the HBL studied by Wolter
et al. (1998) and Beckmann et al. (2002) (open squares),
and other BL Lacs studied by BeppoSAX. These include,
in order of increasing peak frequency: PKS 0537-441 (filled
pentagon; Pian et al. 2002), ON 231 (star; #x in the 0.1-3.8
keV range; Tagliaferri et al. 2000), S5 0716+714 (open circle;
Giommi et al. 1999), PKS 2155-304 (filled triangle; Giommi
et al. 1998), MKN 421 (cross; #x in the 0.1 - 1.6 keV; Fos­
sati et al. 2000), 1ES 2344+514 (open triangle; Giommi,
Padovani & Perlman 2000), and MKN 501 (filled square;
Pian et al. 1998). When more than a value of #x was avail­
able for these BL Lacs we picked the one corresponding to
the largest #peak . The #peak values for the sources studied
in this paper have been taken from Sambruna, Maraschi &
Urry (1996), who fitted a parabola to the #f# broad­band
spectra. The #peak values for the HBL studied by Wolter et
al. (1998) and Beckmann et al. (2002) are taken from those
papers and similarly the values for the additional sources
are taken from the referenced papers.
Fig. 5, although with less statistics, basically confirms
the ROSAT findings, namely a strong anti­correlation be­
tween #x and #peak for HBL and no correlation for LBL. A
few di#erences, however, are worth mentioning. First, the
range in #x is somewhat smaller (# 1.5 vs. # 3). This is
likely due to the larger energy range over which #x is mea­
sured (0.1 - 10 keV for BeppoSAX vs. 0.1 - 2.4 keV for
ROSAT). Objects with very steep ROSAT #x , in fact, are
those in which synchrotron emission is nearing the exponen­
tial cut­o#; by having a larger band BeppoSAX includes flat­
ter, higher energy emission due to inverse Compton. Second,
the wide 0.1 - 100 keV coverage of BeppoSAX has allowed
the detection of spectacular spectral variability with #peak
reaching >
# 10 keV. As predicted by Padovani & Giommi
(1996), these sources display very flat #x (# 0.5 -0.8), since
BeppoSAX is sampling the top of the synchrotron emission.
Note that in this case the flat X­ray spectrum is not associ­
ated with inverse Compton emission, although extreme HBL
(objects to the far right in Fig. 5) have X­ray spectra as flat
as extreme LBL (objects to the far left of the figure).
7 THE 1­JY BL LAC SAMPLE: THE BeppoSAX
VIEW
BeppoSAX data are now available for fourteen 1­Jy BL Lacs:
four studied here, seven studied in Paper I, plus MKN 501
(Pian et al. 1998), S5 0716+714 (Giommi et al. 1999), and
PKS 0537-441 (Pian et al. 2002). Based on their #rx and
#peak values, 11 objects can be classified as LBL and 3 as
HBL, including S5 0716+714 which is an HBL/intermediate
source. We note that these sources include all 1­Jy BL Lacs
with 0.1 -10 keV X­ray flux larger than 3â10 -12 erg cm -2
s -1 (estimated from an extrapolation of the single power­
law fits derived for these objects from ROSAT data; Urry et
al. 1996). This sub­sample is therefore well­defined, makes
up a sizeable fraction (# 41 per cent) of the full sample,
and can therefore be used to obtain robust results on the
hard X­ray emission of X­ray bright 1­Jy BL Lacs. Radio
data from the UMRAO database are available for ten out
of fourteen sources, including MKN 501 and S5 0716+714.
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 11
Therefore, # 71 per cent of our BL Lacs (# 73 per cent of
LBL) have nearly simultaneous high­frequency radio data
which allow us to study the relationship between X­ray and
radio emission almost independently of variability.
7.1 The hard X­ray spectra of 1­Jy BL Lacs
The mean value of the LECS/MECS spectral indices, cov­
ering the 0.1 - 10 keV range, for the fourteen 1­Jy BL Lacs
studied by BeppoSAX is ##x# = 0.83 ± 0.07. As already in­
dicated by Fig. 5, however, this value reflects two broadly
di#erent types of sources and X­ray emission processes.
On one side, in fact, we have the eleven LBL, namely
PKS 0048-097, AO 0235+164, PKS 0537-441, OJ 287,
PKS 1144-379, OQ 530, PKS 1519-273, S5 1803+784, 3C
371, 4C 56.27, and BL Lac. For these sources the X­ray
spectrum is relatively flat, with #x <
# 1 within the errors and
##x# = 0.79 ± 0.07. Five of these sources show evidence for
a low­energy steepening. While the single F --test probabili­
ties that the improvement provided by a double power­law
model is significant range between 87 and 93 per cent, by
adding up the # 2 values we derive a 98 per cent probabil­
ity for the five sources together. The BeppoSAX picture for
LBL, supported also by their SED, is then that of an in­
verse Compton dominance, with the steep synchrotron tail
coming in around # 1 - 2 keV in about half the sources.
ROSAT data, modulo variability and calibration e#ects, cor­
roborate the evidence for overall concave spectra. We find,
in fact, ##x(ROSAT ) - #x(BeppoSAX)# = 0.55 ± 0.17 (and
a median value of 0.73). The mean flux ratio (0.1 -2.4 keV)
is #fBeppoSAX /fROSAT # # 0.4 (# 0.3 median), which is again
consistent with the picture of a variable synchrotron tail
contributing at low energy, thus causing both steeper spec­
tra and higher fluxes when present. This is confirmed also
by considering only the five LBL with indication of low­
energy steepening: using the broken power­law fit values,
#fBeppoSAX /fROSAT # approaches one (# 0.75, with three ob­
jects having values # 1 and two # 0.4 - 0.5).
On the other side, we have the three HBL, namely S5
0716+714, MKN 501, and PKS 2005-489. The mean value
in this case is ##x# = 1.0 ± 0.2, which increases to ##x# =
1.2±0.1 if we exclude MKN 501, found in an extremely high
state, with very large #peak and a flat synchrotron spectrum,
as predicted by the correlation shown in Fig. 5. The evidence
for spectral curvature here is mixed. While MKN 501 shows
flattening at lower energies (##x # -0.2 below # 2 keV;
Pian et al. 1998), S5 0716+714 shows steepening (##x # 0.7
below # 2 - 3 keV; Giommi et al. 1999). This di#erence can
be explained by looking at Fig. 5. While S5 0716+714 is an
HBL with #peak # 10 15 Hz, and therefore relatively close
to LBL values, MKN 501 was observed by BeppoSAX in
outburst, with #peak >
# 100 keV or >
# 2 â 10 19 Hz. In one
case we are then seeing a mix of synchrotron and inverse
Compton radiation, while in the other we are looking at
the overall steepening of pure synchrotron emission. The
BeppoSAX picture for HBL, supported also by their SED, is
then that of a synchrotron dominance, with the flat inverse
Compton component making a contribution for intermediate
sources. As regards ROSAT data, we get ##x(ROSAT) -
#x(BeppoSAX)# = 0.89±0.03 (and a median value of 0.92).
The mean flux ratio is #fBeppoSAX /fROSAT # # 1.5 (# 1.5
median), a factor # 4 larger than for the LBL. The mean
flux ratio at 1 keV is a factor # 7 larger for HBL than for
LBL.
We note that large variability has been seen in the syn­
chrotron component of several BL Lacs observed with Bep­
poSAX (Ravasio et al. 2002, and references therein) and pos­
sibly only once, to the best of our knowledge, in their inverse
Compton component (in the case of PKS 1144-379: Paper
I). Comparing single power­law fits, however, one should be
aware of another factor that could contribute to di#erent
values of fBeppoSAX /fROSAT for HBL and LBL. As pointed
out by Beckmann et al. (2002), the di#erent X­ray spectral
shapes of HBL and LBL could play a role, given the di#er­
ent spectral response of the two instrument. LBL have flat
inverse Compton spectra but synchrotron emission, with a
steep spectrum, sometimes dominates at soft energies (in
# 45 per cent of our sources). The fits to their BeppoSAX
LECS/MECS spectra, which sample a much larger range
towards high energies than the ROSAT PSPC spectra, may
then be mostly sensitive to the flat inverse Compton com­
ponent, with the result that the fitted flux at 1 keV (and
integrated in the soft band) is lower than that inferred from
the ROSAT fits. This e#ect is less important when a bro­
ken power­law is adopted (see above), but the relatively low
statistics of the LECS data may not always justify a sta­
tistical preference of this model with respect to the single
power­law one.
7.2 X­ray and radio powers and #rx
As mentioned above, since # 70 per cent of the objects have
nearly simultaneous X­ray and radio data, we can address
the relation between X­ray and radio emission almost in­
dependently of variability. Fig. 6 plots the radio power L r
vs. the X­ray power Lx for our sources. As expected, the
HBL, having larger #peak values, have larger Lx at a given
L r and therefore occupy the bottom­right part of the dia­
gram. The correlation between L r and Lx for LBL is sur­
prisingly tight, strong, and linear. We find L r # L 0.99±0.08
x ,
significant at the > 99.99 per cent level. A partial corre­
lation analysis (e.g., Padovani 1992), shows that the corre­
lation is still very strong (P # 99.7 per cent) even when
the common redshift dependence of the two powers is sub­
tracted o#. Equivalently, #rx has a very small dispersion for
LBL. We get ##rx# = 0.861 ± 0.008 (dashed line in Fig. 6).
Given the tightness of the correlation, and the importance
that the suggested strong link between the radio and X­ray
band would have for LBL, we have looked in the UMRAO
database for radio observations at 4.8 GHz nearly simulta­
neous with the ROSAT data presented by Urry et al. (1996).
This provided us with 11 more sources, for a total of 22 1­Jy
LBL, # 86 per cent of them with nearly simultaneous X­
ray and radio data. The results obtained for our BeppoSAX
sources are confirmed with this enlarged sample. Namely,
we find now L r # L 0.97±0.08
x , significant at the > 99.99
per cent level. As before, the correlation is still very strong
(P # 99.97 per cent) when the common redshift dependence
is subtracted o#. We get ##rx# = 0.89 ± 0.01 (dotted line in
Fig. 6) for the 11 ROSAT sources, not significantly di#erent
from our previous value, and a mean for the 22 1­Jy LBL
of ##rx# = 0.875 ± 0.008. This means that the X­ray power
(or flux) can be extrapolated from the radio power (or flux)
c
# 0000 RAS, MNRAS 000, 000--000

12 P. Padovani et al.
within # 15 per cent over almost eight orders of magnitude
in frequency.
It is interesting to note that the strong L r - Lx cor­
relation displayed by the 1­Jy LBL is not only due to the
nearly simultaneous radio and X­ray data. Even if we use
the original 1­Jy data, in fact, we find a strong correlation
(P > 99.99 per cent) for the 22 sources in Fig. 6, still very
significant (P # 99.97 per cent) when the redshift e#ect is
subtracted o#. The slope of the correlation is slightly flat­
ter and ##rx# is slightly larger (although not significantly so:
0.890 ± 0.008), as on average the 1 Jy fluxes are larger than
the UMRAO ones, but the dispersion is still very small. We
note that the relatively small range in ##rx# for LBL, al­
though never studied with nearly simultaneous radio and
X­ray data, has been previously noted, amongst others, by
Padovani & Giommi (1995) and Fossati et al. (1998). To the
best of our knowledge, however, all previous studies of the
L r -Lx correlation for 1­Jy BL Lacs did not distinguish be­
tween LBL and HBL, thereby reducing the significance of
the correlation.
The interpretation of the strong linear link between X­
ray and radio emission components in LBL suggested by
the L r - Lx correlation is not straightforward within the
currently accepted scenario. As illustrated in Fig. 4, in fact,
our one­zone model cannot reproduce the observed radio
flux, thought to originate in a region of the jet larger than
the (more compact) one responsible for most of the emission,
including X­rays. On the other hand, since the radio spectra
in these sources are invariably flat, there is a correlation
between the radio flux at the self­absorption frequency (#
10 2
-10 3 GHz) of the smaller X­ray emitting region and the
flux at 5 GHz. This may account, at least in part, for the
L r -Lx correlation, even if the powers in the two bands are
produced in di#erent parts of the jet.
8 CONCLUSIONS
We have presented new BeppoSAX observations of four BL
Lacertae objects selected from the 1­Jy sample, all LBL,
i.e., characterized by a peak in their multifrequency spectra
at infrared/optical energies. We have then used these data
plus data previously published by us (7 sources; Paper I) and
others (3 sources) to study the properties of the hard X­ray
spectra of an X­ray flux­limited 1­Jy sub­sample (containing
11 LBL and 3 HBL) and constrain the emission processes.
A relatively simple picture comes out from this paper: a
dominance of inverse Compton emission in the X­ray band of
LBL, with # 50 per cent of the sources showing also a likely
synchrotron component. Our main results are as follows:
(i) The BeppoSAX spectra of our LBL sources are rel­
atively flat (#x # 0.8). For five sources broken power­law
models improve the fits at the 98 per cent level when the
# 2 values for the single sources are added up. The result­
ing best­fit spectra all concur in indicating a flatter com­
ponent emerging at higher energies, with spectral changes
##x # 0.8 around 1 - 2 keV.
(ii) Our LBL have a typical di#erence between BeppoSAX
and ROSAT spectral slopes #x(ROSAT )-#x(BeppoSAX) #
0.6. The interpretation of this e#ect is complicated by pos­
sible ROSAT miscalibrations (which, if at all present, could
explain a di#erence # 0.3) and variability e#ects, as the
Figure 6. Nearly simultaneous X­ray and radio luminosities of 1­
Jy BL Lacs objects observed by BeppoSAX (circles) and ROSAT
(squares). Lower limits indicate sources for which no redshift is
available but z > 0.2 was assumed, based on the fact that the
host galaxies are optically unresolved. The dashed line indicates
a value of #rx = 0.861, the mean value for the BeppoSAX LBL,
while the dotted line indicates a value of #rx = 0.89, the mean
value for the ROSAT LBL. Labels identify the HBL (open circles).
mean flux ratio is #fBeppoSAX /fROSAT # # 0.4, but is con­
sistent with the picture of a variable synchrotron emission
contributing in the soft X­ray band, leading to high fluxes
and steeper spectra when present.
(iii) Despite the non­simultaneity of the multifrequency
data (UMRAO radio data excluded) it is apparent that the
BeppoSAX spectra indicate a di#erent emission component
in the SEDs of our LBL sources, separate from that respon­
sible for the low energy emission. In fact, the extrapolation
of the relatively flat BeppoSAX slopes cannot be extended
to much lower frequencies since the predicted optical flux
would be orders of magnitude below the observed value. A
sharp steepening towards lower frequencies is then necessary
to meet the much higher optical (synchrotron) flux.
(iv) Our interpretation of the #x - #peak diagram is the
one originally proposed by Padovani & Giommi (1996) for
the ROSAT data. Namely, #x steepens moving from LBL to
HBL as synchrotron replaces inverse Compton as the main
emission mechanism in the X­ray band. The spectral index
then flattens again as the synchrotron peak moves to higher
energies in the X­ray band, eventually converging to the
relatively flat value characteristic of synchrotron emission
before the peak. Again, this fits perfectly with a dominance
of inverse Compton emission in our LBL.
(v) We find a very tight proportionality between nearly
simultaneous radio and X­ray powers for our LBL sources
and 11 additional LBL with ROSAT data, such that X­ray
power can be predicted within # 15 per cent from the radio
c
# 0000 RAS, MNRAS 000, 000--000

BeppoSAX observations of 1­Jy BL Lacertae objects ­ II 13
power. This points to a strong link between X­ray and radio
emission components in LBL.
(vi) The data for the (small number of) HBL are consis­
tent with other studies based on larger samples of sources
and confirm the dominance of synchrotron emission in the
BeppoSAX band.
ACKNOWLEDGEMENTS
We thank Andrea Comastri, Giovanni Fossati, Franco Man­
tovani, Laura Maraschi, Carlo Stanghellini, Gianpiero Ta­
gliaferri, and Meg Urry for their contribution at an early
stage of this project. LC acknowledges the STScI Visitor
Program. This research has made use of data from the Uni­
versity of Michigan Radio Astronomy Observatory which is
supported by funds from the University of Michigan and of
the NASA/IPAC Extragalactic Database (NED), which is
operated by the Jet Propulsion Laboratory, California Insti­
tute of Technology, under contract with the National Aero­
nautics and Space Administration.
REFERENCES
Beckmann V., Wolter A., Celotti A., Costamante L., Ghisellini
G., Maccacaro T., Tagliaferri G., 2002, A&A, 383, 410
Boella G. et al., 1997a, A&AS, 122, 299
Boella G. et al., 1997b, A&AS, 122, 327
Burbidge E. M., Beaver E. A., Cohen R. D., Junkkarinen V. T.,
Lyons R. W., 1996, AJ, 112, 2533
Celotti A., Padovani P., Ghisellini G., 1997, MNRAS, 286, 415
Dickey J. M., Lockman F. J., 1990, ARAA, 28, 215
Donato D., Ghisellini G., Tagliaferri G., Fossati G., 2001, A&A,
375, 739
Elvis M., Lockman F. J., Wilkes, B. J., 1989, AJ, 97, 777
Fiore F. et al., 1999, Handbook for NFI Spectral Analysis, avail­
able at http://www.asdc.asi.it/bepposax/software/
Fossati G., Maraschi L., Celotti A., Comastri A., Ghisellini G.,
1998, MNRAS, 299, 433
Fossati G. et al., 2000, ApJ, 541, 166
Frontera F. et al., 1997, A&AS, 122, 357
Gehrels N., 1986, ApJ, 303, 336
Ghisellini G., Celotti A., Fossati G., Maraschi L., Comastri A.,
1998, MNRAS, 301, 451
Ghisellini G., Celotti A., Costamante L., 2002, A&A, 386, 833
Giommi P., Fiore, F., 1997, 5th Int. Workshop on Data Analysis
in Astronomy, Erice. World Scientific Press, Singapore, p. 93
Giommi P., Padovani P., 1994, MNRAS, 268, L51
Giommi P., Padovani P., Perlman E., 2000, MNRAS, 317, 743
Giommi P. et al., 1998, A&A, 333, L5
Giommi P. et al., 1999, A&A, 351, 59
Heidt J., Wagner S. J., 1996, A&A, 305, 42
Iwasawa K., Fabian A. C., Nandra K., 1999, MNRAS, 307, 611
Kaspi S., Smith P. S., Netzer H., Maoz D., Jannuzi B. T., Giveon
U., 2000, ApJ, 533, 631
Kollgaard R. I., 1994, Vistas Astron., 38, 29
Lamer G., Brunner H., Staubert R., 1996, A&A, 311, 384
Lin Y. C. et al., 1999, ApJ, 525, 191
Madejski G., Takahashi T., Tashiro M., Kubo H., Hartman R.,
Kallman T., Sikora, M., 1996, ApJ, 459, 156
Mineo T. et al., 2000, A&A, 359, 471
Morrison R., McCammon D., 1983, ApJ, 270, 119
Murphy E. M., Lockman F. J., Laor A., Elvis M., 1996, ApJS,
105, 369
Padovani P., 1992, A&A, 256, 399
Padovani P., Giommi P., 1995, ApJ, 444, 567
Padovani P., Giommi P., 1996, MNRAS, 279, 526
Padovani P. et al., 2001, MNRAS, 328, 931 (Paper I)
Padovani P., Costamante L., Ghisellini G., Giommi P., Perlman
E., 2002, ApJ, 581, 895
Parmar A. et al., 1997, A&AS, 122, 309.
Pian E. et al., 1998, ApJ, 492, L17
Pian E. et al., 2002, A&A, 392, 407
Puchnarewicz E. M. et al., 1997, MNRAS, 291, 177
Ravasio M. et al., 2002, A&A, 383, 763
Sambruna R., Maraschi L., Urry C. M., 1996, ApJ, 463, 444
Spada M., Ghisellini G., Lazzati D., Celotti A., 2001, MNRAS,
325, 1559
Stickel M., Padovani P., Urry C. M., Fried J. W., K˜uhr H., 1991,
ApJ, 374, 431
Tagliaferri G. et al., 2000, A&A, 354, 431
Tagliaferri G. et al., 2003, A&A, 400, 477
Urry C. M., Padovani, P., 1995, PASP, 107, 803
Urry C. M., Sambruna R. M., Worrall D. M., Kollgaard R. I.,
Feigelson E. D., Perlman E. S., Stocke J. T., 1996, ApJ, 463,
424
Wolter A. et al., 1998, A&A, 335, 899
c
# 0000 RAS, MNRAS 000, 000--000