A&A 420, 115-123 (2004)
DOI: 10.1051/0004-6361:20034165
M. Krause - A. Löhr
Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
Received 6 August 2003 / Accepted 31 January 2004
Abstract
We present 2
4 resolution, high sensitivity radio continuum
observations of the nearby spiral galaxy NGC 4258 in total intensity
and linear polarization obtained with the Very Large Array at
3.6 cm (8.44 GHz).
The radio emission along the northern jet and the center of the galaxy
is polarized and allows investigation of the magnetic field. Assuming
energy-equipartition between the magnetic field and the relativistic
particles and distinguishing between (1) a relativistic electron-proton
jet and (2) a relativistic electron-positron jet, we obtain average
magnetic field strengths of about (1) 310 G and (2) 90
G. The
rotation measure is determined to range from -400 to -800 rad/m2in the northern jet. Correcting the observed E-vectors of polarized
intensity for Faraday rotation, the magnetic field along the jet turns
out to be orientated mainly along the jet axis. An observed tilt with
respect to the jet axis may indicate also a toroidal magnetic field
component or a slightly helical magnetic field around the northern jet.
Key words: galaxies: spiral - galaxies: indvidual: NGC 4258 - radio continuum: galaxies - polarization - galaxies: magnetic field - galaxies: jets
The nearby galaxy NGC 4258 (M 106) is a bright SAB(s)bc spiral (de Vaucouleurs et al. 1976) at a distance of 7.2 Mpc (Herrnstein et al. 1999). It seems to possess a small bright nucleus with a highly excited emission line spectrum (Burbidge et al. 1963) and has been classified as a weakly active Seyfert 2-type galaxy.
Most striking are the two so-called "anomalous arms'', which are not
visible in the optical and were first detected in H
by
Courtès & Cruvellier (1961) in the inner region of the
galaxy. Van der Kruit et al. (1972) detected these anomalous
arms in the radio range where they extend out to the optical periphery.
Spectral index studies indicate that their radio emission is of
non-thermal origin (de Bruyn 1977; van Albada 1978).
The anomalous arms of NGC 4258 have been extensively discussed in terms of ejection of matter from the nucleus. The detection of a water-maser (Claussen et al. 1984; Henkel et al. 1984) and an accretion disk around a supermassive central object (Miyoshi et al. 1995; Herrnstein et al. 1998), and the fact that inner anomalous arms are orientated parallel to the rotation axis of the accretion disk and can be traced even on subparsec scale (Herrnstein et al. 1997) indicate that the anomalous arms indeed are jets. Whereas in the inner regions the anomalous arms clearly reveal their jet character, many of the features at a greater distance from the nucleus (e.g. their bifurcation) remain unexplained.
The three-dimensional geometry of the galaxy and its jets has been
discussed for a long time (cf. e.g. van Albada & van der Hulst 1982; Hummel et al. 1989). The detection of an
accretion disk revealed directly for the first time that the central
part of the galaxy has a significant tilt with respect to the galactic
disk: the accretion disk itself has an inclination angle of
and a position angle (PA) of
(Miyoshi et al. 1995),
i.e. it is oriented nearly east-west. The PA of the galactic disk,
however, is
,
which is nearly north-south, and its
inclination is
(van Albada 1980). Thus, the plane of
the galactic disk and the plane of the accretion disk have a
significant angle to each other. As the jets emerge perpendicular to
the accretion disk, they have to pass the galactic disk at a rather
small angle. Due to the projection of the whole system with respect to
the Earth, the inner jet direction is almost parallel to the major axis
of the galactic disk.
To further investigate the jet geometry and especially the
magnetic field along the jets we obtained data with the Very Large
Array (VLA) at
3.6 cm and compared these with already published but
reprocessed VLA data by Hummel et al. (1989) at
6.2 cm
(4.86 GHz) and
20 cm (1.49 GHz) and with observations at
2.8 cm (10.55 GHz) made with the Effelsberg 100-m telescope.
The observations and data reduction procedures are described in Sect. 2. In Sect. 3 we present the results and examine the measurements of the linearly polarized emission in terms of Faraday rotation, magnetic field strength and direction. The discussion of the results and the summary follow in Sects. 4 and 5, respectively.
We observed the radio continuum emission from NGC 4258 with the Very
Large Array (VLA) in its C-configuration at 8.44 GHz (3.6 cm)
in March 1996 for 14 h in total and linearly polarized intensity.
The observations were done with two independent IFs, each with a
bandwidth of 50 MHz and separated by 50 MHz. The two IFs were combined
afterwards. The resulting observational parameters are given in
Table 1. For a detailed description of the VLA see Thompson et al. (1980) and Napier et al. (1983).
Table 1:
Observational parameters of the VLA measurements. The rms
noise values are given for total intensity (I) and polarized intensity (). Further explanations are given in the text below.
The data reduction was performed using the Astronomical Image Processing System (AIPS) of the NRAO.
The flux density scale was calibrated by observing 3C 138 and 3C 286
and is based on the Baars et al. (1977) scale. We assumed flux
densities of 2.52 Jy for 3C 138 and 5.21 Jy for 3C 286.
We used an angle of
for the polarized emission of 3C 138
and
for that of 3C 286. The phase calibrator 1216+487,
whose position is known with an accuracy of
,
was also
used to correct for the instrumental polarization.
The edited and calibrated visibility data were cleaned interactively,
self-calibrated and Fourier transformed to obtain maps of the
Stokes parameters I, U and Q. We produced uniformely weighted maps
(with ROBUST = 0) to obtain the best resolution and sensitivity
compromise with a HPBW of
.
Maps even more
sensitive to weak extended structures and the polarized emission were
made with natural weights (ROBUST = 4) at the expense of a slightly lower
HPBW of
.
We used the zero flux correction
provided by AIPS as a first-order correction for the missing
large-scale flux. The U- and Q-maps were combined to obtain maps of the
linearly polarized intensity
(corrected for the positive
zero level offset) and the position angles of the polarized emission
(the E-vectors). Values of the respective noise levels are summarized in
Table 1, the full resolution map is shown in Figs. 1
and 2.
![]() |
Figure 1:
The 8.44 GHz high resolution ( HPBW =
![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
The 8.44 GHz high resolution ( HPBW =
![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
NGC 4258 was observed with the Effelsberg 100-m telescope
at 10.55 GHz (2.8 cm) in July 1995. The multi-beam receiver
has 4 horns and 16 channels. Each horn is equipped with two total-power
amplifiers and an IF polarimeter. The bandwidth is 300 MHz, the system
noise temperature about 50 K and the resolution is
HPBW.
For pointing and focusing we observed regularly the sources 3C 48 and
3C 286. The calibration was done with 3C 286 according to the flux
values of Baars et al. (1977). The observing procedure is
essentially the same as described by Klein & Emerson (1981). We
obtained 28 coverages of NGC 4258 which led, after restoration (Emerson
et al. 1979) and combination (Emerson & Gräve 1988),
to an rms noise in the final maps of Jy/b.a. in total
power and
Jy/b.a. in linear polarization.
Figure 1 shows the full resolution map of NGC 4258 in total
intensity at 8.44 GHz as obtained with the VLA in its C-configuration.
The HPBW is
and the inner region of the
galaxy is well resolved. The S-shaped feature of the anomalous arms is
clearly visible (marked as AN
and AN
in
Fig. 2) with the southern jet being fainter than the
northern one. The jet direction in the central part is approximately north
to south, whereas it changes to about east-west orientation in the outer
parts. The northern arm bifurcates at least two times (as first
detected by van Albada & van der Hulst 1982).
Maximum radiation is emitted from the galactic center. The normal
spiral arms of the disk are also visible to the north and
south (marked as SP
and SP
in Fig. 2).
Along them, three very bright regions can be observed at
,
and
,
where the intensity is
Jy/b.a. (
),
Jy/b.a.
(SP
1) and
Jy/b.a. (SP
2).
We observe a ridge of high intensity of
Jy/b.a. along the northern jet. The southern jet shows a maximum
brightness of
Jy/b.a. at large distance from the center. Between the
center and this region the southern jet is very weak. The northern region of
high brightness (
Jy/b.a.) at
and
(marked as "A'' in
Fig. 6) has been interpreted as a radio hot spot of the northern
jet by Cecil et al. (2000). The same source has been classified as
a supernova remnant candidate by Hyman et al. (2001), noted as
source 7 therein.
The corresponding hot spot in the southern jet (marked as "B'' in Fig. 6)
has been reported to be located
south of the nucleus at
and
(Cecil et al. 2000). In this region we also detected a local maximum in brightness
of
Jy/b.a. which coincides with the change of the jet
direction as indicated by the solid line in Fig. 6.
The total flux of the central source was measured by fitting a
two-dimensional Gaussian to the central emission in the map with highest
resolution (Fig. 1). It is derived to be
mJy.
In Fig. 2 the contour plot of the total intensity is
superimposed on an H
map of the galaxy (observed at the
Hoher List Observatory of the University of Bonn, (courtesy of N.
Neininger)). In H
the central region as well as the normal spiral arms to the north
and south show the highest intensity. The superposition assigns
clearly which part of the radio emission belongs to the spiral arms
and which belongs to the jets. The clumps of strong emission along the
spiral arms mentioned above are well correlated with the regions of
highest intensity in H
.
They are most probably star-forming
regions as they emit unpolarized thermal radiation, whereas the radio
emission along the anomalous arm is polarized and of nonthermal origin
(cf. e.g. Hummel et al. 1989; Hyman et al. 2001).
![]() |
Figure 3:
The 4.86 GHz map in total intensity with HPBW = 14
![]() ![]() |
Open with DEXTER |
![]() |
Figure 4:
The 8.44 GHz map of the total intensity at HPBW = 14
![]() ![]() |
Open with DEXTER |
Figure 3 shows the observations of NGC 4258 at
4.86 GHz (6.2 cm) as observed with the VLA in its C-configuration.
The contours give the total intensity, the length of the E-vectors is
proportional to the linearly polarized intensity. The rms noise
values are given in Table 1 and the HPBW is 14
.
To compare the
3.6 cm map with the
6.2 cm
map we smoothed the
3.6 cm to the resolution of
HPBW. This map is shown in Fig. 4. The shape of the
total intensity is very similar at both wavelengths.
The observations at 2.8 cm made with the Effelsberg 100-m telescope are presented in Fig. 5. The HPBW is 69
,
hence
the distribution is much smoother and the bifurcation in the western
arm is barely resolved. The normal spiral arms, notably the southern
one, are indicated.
To investigate the jet geometry and in particular the outflow
direction from the nucleus, Fig. 6 shows the central region in
detail up to
away from the nucleus. The location of the circumnuclear accretion disk and the observed direction of the ejected
matter are indicated. From our observations we determined the
position angle of the jet in the central region to be
.
This is in full agreement with the value given
by Cecil et al. (2000) from their radio observations and along
the projected spin axis of the accretion disk as determined by Miyoshi
et al. (1995).
At larger distances from the nucleus the jets change their direction
from a north-south orientation to northwestern to southeastern
as visible on larger scales (cf. also Fig. 1). This
change is not smooth but can be described by symmetric kinks on
both sides at a projected distance of
from the galactic
center to
.
![]() |
Figure 5:
The 10.55 GHz map of NGC4258 in total intensity (contours)
with HPBW
![]() ![]() |
Open with DEXTER |
![]() |
Figure 6:
The 8.44 GHz high resolution ( HPBW =
![]() ![]() |
Open with DEXTER |
In addition to the total intensity we obtained the linearly polarized
intensity at a resolution of
HPBW. This is
the first time that extended linear polarization has been detected in
NGC 4258 with arcsec resolution. In Fig. 7 the E-vectors of
the linearly polarized emission are superimposed on the total
intensity maps at 8.44 GHz. The length of the vectors is proportional
to the polarized intensity.
As can be seen, polarized intensity has been detected exclusively along the jets. The highest polarized emission occurs at the center and along the northern jet, there mainly to the northwest of the bifurcation. The map shown in Fig. 7 will be used later to calculate the magnetic field strength (Sect. 3.6).
The degree of linear polarization (polarized intensity/total intensity) along the jets is high and lies in the range between 35% and 65%. In the center the degree of linear polarization is below 2% and may be partly instrumental. A high degree of polarization of about 55% was detected at the presumed hot spot A, whereas no polarization was detected at the position of source B. This will be discussed further in Sect. 4.
In the smoothed map (Fig. 4) the highest polarized emission
was detected along the jets and is particularly strong along the
northern jet. The polarization percentage varies between 20% and 45%
there. These values are lower than those of the higher resolution map
(Fig. 7) which indicates that the polarized intensity is
patchy and no longer resolved at
HPBW. The smoothed map
also reveals several small regions of low percentage polarization
(20%) in the southern jet.
A comparison of the 3.6 cm map (Fig. 4) to the
6.2 cm map (Fig. 3) shows very different
orientation of the E-vectors in both maps, indicating strong Faraday
rotation.
![]() |
Figure 7:
The 8.44 GHz map of the total intensity with HPBW =
![]() ![]() |
Open with DEXTER |
The observed electric vectors are rotated by Faraday effects. The
amount of the Faraday rotation can be determined by calculating the
rotation measure RM between different wavelengths. Correction of the
observed electric vectors according to these RMs and rotation by
leads to the intrinsic direction of the magnetic field
in the sky plane. The RM value itself depends on the strength of the
magnetic field component parallel to the line of sight, its sign
indicates the direction of this parallel field component.
Although also we had 20 cm data, we could not use
them for the calculation because at this wavelength and resolution
polarized intensity was only detected at large distances from the
center. At
3.6 cm and
6.2 cm the polarized intensity
was detected closer to the nucleus and along the jets. Thus the
polarized regions detected do not coincide with those at
20 cm.
We determined the RM between 3.6 cm and
6.2 cm.
The calculated RM varies between 400 and
.
The
ambiguity (i.e. the RM value that corresponds to a Faraday
rotation of
and hence is undistinguishable by
observations at only two wavelengths) between these two wavelengths is
as high as
for n=1. A substraction of one
rotation (n=-1) from these values leads to RM between -800 and
which is in absolute value not distinguishable from the
values above. Futhermore, a rotation of the observed vectors at
3.6 cm by e.g.
or
(which corresponds to the
ambiguity) leads to an angle
difference of
,
hence about perpendicular to each other.
We tried to solve the RM ambiguity using the 2.8 cm
observations. Therefore we smoothed the
3.6 cm map
(Fig 4) out to 69
HPBW, the angular resolution of
the
2.8 cm map (Fig. 5), and determined the RM between
these two wavelengths. The
ambiguity between these
wavelengths is as large as about
.
The vectors
between
2.8 cm and
3.6 cm rotate clockwise by
about 40-
which correspond to negative RM between about
-1500 and
.
We conclude that the vectors rotate
further clockwise towards the E-vectors observed at
6.2 cm. This corresponds to negative RM values. Hence we
consider the values for n=1 between
3.6 cm and
6.2 cm
as more probable. The corresponding RM ranges from -800 to
and is presented graphically in Fig. 8.
![]() |
Figure 8:
The rotation measure RM between ![]() ![]() ![]() |
Open with DEXTER |
To verify the consisitency of the derived values we estimate
the depolarization DP between 3.6 cm and
6.2 cm
using
However, the observed RM may also be due to a layer of higher thermal density but fewer relativistic particles around the jet that does not contribute to the polarized emission as proposed by Bicknell et al. (1990). This layer would act as a foreground screen and rotate the observed vectors without depolarizing the emission.
The observed depolarization of the emission can also be explained by beam depolarization and Faraday dispersion (Burn 1966; Sokoloff et al. 1998) which is expected to occur in the emitting region, especially in the case of a toroidal magnetic field (see Sect. 3.5) that is not resolved by the beam size.
With the observed large values for the RM we expect strong
depolarization at 20 cm in all scenarios mentioned above.
This can explain the lack of linearly polarized emission in the inner
part of NGC 4258 as observed by Hummel et al. (1989).
To obtain the direction of the intrinsic magnetic field, we corrected
the E-vectors (Fig. 4) for Faraday rotation with the help of
the RM (Fig. 8). The resulting magnetic field is shown in
Fig. 9 superimposed on the smoothed 3.6 cm map.
The vectors generally follow the jet direction, also along the main
bifurcation of the northern jet. They are, however, somewhat tilted
with respect to the jet direction, especially in the outermost part of
the northern bifurcation of the northern jet and along the southern
part of this bifurcation. We infer that the magnetic field is mainly
poloidal along the jet axis with a (weaker) toroidal component
especially in the outer part of the northern bifurcation and along the
southern bifurcation. This can be a superposition of a poloidal and
(weaker) toroidal magnetic field in different layers around the jet
axis or is consistent with a slightly helical magnetic field
around the jet axis.
![]() |
Figure 9:
The 8.44 GHz map of the total intensity ( HPBW = 14
![]() ![]() |
Open with DEXTER |
To calculate the magnetic field strength we used the polarized data at
(Fig. 7). As the polarized
emission is not homogeneously distributed but varies strongly along the
jets, we only calculate the magnetic field in the regions of strongest
polarized emission, which are: region I northeast of the bifurcartion,
region II northwest of the bifurcation and region III just south of the
bifurcation.
For the nonthermal spectral index along the jets we used the latest
value derived by Hyman et al. (2001) which is
.
This value is in good agreement
with the average spectral index in jets of
and also with the previously derived spectral index between
6.2 cm and
20 cm by Hummel et al. (1989) of
.
The jet is assumed to be cylindrical. For the line of sight Lthrough the jets we assume L=50 pc as we do not resolve the jet
width with our linear resolution of 90 pc. Based on our previous
results in Sect. 3.5 we assume a magnetic field mainly along the
jet axis. As the inclination of the jet we considered values between
and
(as argued in Sect. 4).
We calculate the magnetic field strengths assuming energy equipartition between the magnetic field and the relativistic particles. The calculations consider two different models concerning the composition of the ejected jet-matter in active galactic nuclei:
Table 2: Magnetic field strengths for an electron-proton jet. Region I is south of the northern bifurcation, II is the southern part of the N bifurcation; III is the northern part of the N bifurcation.
Table 3: Magnetic field strengths for an electron-positron jet. Regions I, II, and III are chosen as in Table 2.
The large-scale radio and H
features of NGC 4258 could only
be interpreted as jets (Falcke & Biermann 1999; Yuan et al.
2002) after the accretion disk around a super-massive central
object had been discovered from observations of water-maser emission (Claussen
et al. 1984; Henkel et al. 1984). Previously they were
called "anomalous arms'' and explained in terms of ejected matter that
interacts strongly with the disk gas and hence is compressed (van der
Kruit et al. 1972) or as jet-like outflows (Sanders et al.
1982).
The detection of a water-maser implies a thin, rotating Keplerian disk and is strong evidence for a black hole in the center of NGC 4258 (Miyoshi et al. 1995). Jets require enormous amounts of energy and it is generally assumed that they are produced by accretion disks which rotate around black holes and eject matter along their rotation axis.
The jets in NGC 4258 are, however, extraordinary jets, as they are not strongly collimated but show a diffuse structure and even bifurcation. They are also strongly bent.
The three-dimensional geometry of this galaxy is somewhat unusual in that the accretion and galactic disk are almost perpendicular to each other (cf. Sect. 1). If the jets emerge almost perpendicular to the nuclear disk, they have to pass the galactic disk and seem to interact with it at least in the inner 4 kpc (cf. Krause et al. 1990).
Our data are the first that are sensitive enough to detect the magnetic
field in the northern jet regions also in the inner
from
the galactic center. We determined the rotation measure and corrected
the polarization angles for Faraday rotation. The derived intrinsic
magnetic field orientation is mainly along the jet direction.
As the determined RMs are negative, the uniform magnetic field component along the jet axis points away from us, hence it is directed towards the nucleus in the northern jet. The few RM values that are found in the southern jet are also negative and may indicate that the magnetic field there is directed away from the nucleus.
The magnetic field in the northern jet however is also somewhat tilted towards the jet direction, especially in the outermost part of the northern bifurcation and along the southern part of the bifurcation. This can indicate an additional (weaker) toroidal component that is either located in different layers around the jet axis or may be due to a partly helical magnetic field around the jet axis. This is found to be typical of extragalactic jets (Begelmann et al. 1984).
The longitudinal field (along the jet axis) can either be in the inner part of the jet near the jet axis, in the so-called beam, with a toroidal field further away from the jet axis (e.g. Roland & Hermsen 1995) or be part of a helical magnetic field (e.g. Lesch et al. 1989) that may be amplified by dynamo action such as the screw dynamo (Shukurov & Sokoloff 1993).
As
(the sign
of RM is determined by the direction of
), the observed
high RM values require a uniform magnetic field component
parallel to the line of sight with uniform direction.
It has been argued from the geometry of the accretion disk (see
Sect. 1) that the inner jet orientation is almost parallel to the major
axis of the galactic disk of NGC 4258. A magnetic field along the jet
that is parallel to the major axis of the disk would not contribute to
a field component parallel to the line of sight, and hence not
cause Faraday rotation. However, even an angle of the jet to the line of
sight of only
yields
,
where
is the field component along the jet axis.
Additionally, we observe several kinks in the northern jet as described
in Sect. 3.2. The first appeared at
projected
distance from the nucleus where the jet changes direction by as
much as
in the plane of the sky. We consider it highly
improbable that this change of direction happens only in the
plane of the sky. It will take place at an arbitrary angle to the
line of sight.
Let us assume that we have a kink of equal strength along the
line of sight. With the uniform magnetic field strength of about
G we estimate
of about
G in the
case of an electron-proton jet. We calculate the thermal electron
density
using
If the jet consists of relativistic particles (electron-positron jet), the expected rotation measure in the jet is RM=0 and the Faraday rotation is produced in a cocoon around the jet (e.g. Bicknell et al. 1990) with a correspondingly smaller line of sight, hence an even higher thermal electron density and/or magnetic field strength there. In this case the expected depolarization is smaller (as described in Sect. 3.4.) but still compatible with our observations.
Previous investigations on the magnetic field were only able
to determine the magnetic field orientation in the outermost parts of
the jets (outside
)
(cf. Krause et al. 1984; Hummel et al.
1989). The magnetic field there was also derived to be along
the jets.
Concerning the hot spots detected by Cecil et al. (2000), we can
confirm that we also detected high emission at both locations. Hot
spots are usually detected when the relativistic jet matter hits
intergalactic gas and undergoes strong interaction. The observed electric
vector at 3.6 cm fits, however, into the general pattern of
the northern jet. As we have no high-resolution polarization
information at
6.2 cm we cannot determine the RM and the intrinsic magnetic field direction at the presumed hot spot A.
On the other hand, the observed high degree of linear polarization of about 55% as averaged over the whole source makes it rather improbable that source A is a supernova remnant as has been proposed by Hyman et al. (2001).
We present interferometer data obtained with the VLA in its
C-configuration at 3.6 cm (8.4399 GHz) in total power and
linear polarization. For comparison and to obtain quantities
like rotation measure and depolarization we also reprocessed VLA data at
6.2 cm (4.8851 GHz) and
20 cm (1.4899 GHz)
that were previously published by Hummel et al. (1989) and
observations at
2.8 cm (10.55 GHz) made with the Effelsberg
100-m telescope.
The high resolution maps (
HPBW) at
3.6 cm as
obtained with the C-configuration of the VLA are able for the first time
to resolve the jets in the central region of the galaxy. Detailed
inspection shows that they emerge from the galactic center along the
projected spin axis of the accretion disk as determined by Miyoshi et al. (1995). At a distance
away from the center
they change direction towards the previously seen northwest to
southeast and bifurcate at
from the nucleus in
the northern jet and at
in the southern jet. The multiple
splitting of the northern jet is clearly visible at this resolution.
The polarized emission was detected exclusively along the jets and
allowed the calculation of the magnetic field strength in the inner
region of the northern jet. Energy equipartition considerations lead
to magnetic field strengths of
G assuming a relativistic
electron-proton jet and
G assuming an electron-positron
jet. The rotation measure could be determined between
3.6 cm
and
6.2 cm, at a linear resolution of
HPBW. It is
derived to be -400 to -800 rad/m2 in the northern jet. Correcting
the observed E-vectors of polarized emission for Faraday rotation, the
magnetic field is mainly along the jet axis in the
central region and tends to become somewhat tilted with respect to the
jet direction in the outer part of the northern jet. This may be
consistent with a slightly helical magnetic field around the northern
jet or may indicate a superposition of a longitudinal magnetic field
near the jet axis and a toroidal magnetic field away from the axis.
Acknowledgements
We thank N. Neininger for providing the Himage and M. Ehle for helping us with the Effelsberg observations. We acknowledge fruitful discussions with H. Falcke and A. Shukurov and are grateful for helpful comments by R. Perley.