Our maps of total intensities (Figs. 1 and 3) show a large number
of unresolved (point-like) sources. As most of these have been
catalogued before (Walterbos et al. 1985; Braun 1990), we do not give
a new source list in this paper
.
Differences in flux density may result from the different telescope
beams or from source variability. With the larger beam of our survey,
we find significantly larger flux densities for 37W45, 50, 51, 175,
180, 201, 207A and 207B. Except for 37W207 these sources were
classified as extended by Walterbos et al. (1985). Lower flux
densities (possibly due to variability) are found for the sources
37W91, 115, 131, 168 and 200.
As this is the first M 31 survey at 
20 cm including linearly
polarized emission, we give a list of polarized sources in Table 2.
These sources may be used e.g. to search for systematic variations of
Faraday rotation in the disk and halo of M 31.
Many of the unpolarized sources seen on M 31 are 
regions
or weak supernova remnants. In the northern half of M 31 Braun (1990)
identified 103 regions and Braun & Walterbos (1993) detected
52 SNRs and SNR candidates.
In order to obtain a map of M 31 largely free of unrelated point
sources we subtracted all sources brighter than 2 mJy. As 25% of
the identified regions and 10% of the SNRs and SNR candidates
are brighter than 2 mJy, this means that the brightest regions
and SNRs were also subtracted. The resulting combined map of total
intensities is shown in Fig. 4.
Table 2: Polarized point-like sources in the M 31 field
After subtraction of unresolved sources brighter than 2 mJy we
integrated the flux densities of our VLA maps (Figs. 1 and 2) in
ellipses around the centre out to a radius of 16 kpc in the plane of
the galaxy (inclined by 78
). The total flux density is 
Jy in total intensity and 
Jy in linear
polarization. The combined map in total intensity (Fig. 4) includes
Jy within the same radius. This demonstrates the
significance of the effect of missing interferometer spacings in the
case of an extended object like M 31.
To the total flux density at 20 cm derived here we should
add about 300 mJy of subtracted regions and SNRs in
M 31 brighter than 2 mJy. The value of 
Jy then is in good
agreement with that expected from the spectrum between 74 cm
and 11 cm (Beck & Gräve 1982), 
Jy. We note,
however, that our value at 20 cm is a lower limit as along
the major axis the combined map extends only to 12 kpc radius
(compared to 16 kpc at 11 cm).
Our Effelsberg map (Fig. 3) shows the full extent of M 31. After
subtraction of the Galactic foreground emission and of the unresolved
background sources (visible in our VLA map) we obtained a total
flux density of 
Jy within 16 kpc radius.
The extended total emission seen in the VLA map (Fig. 1) emerges
from many small-scale features which often seem filamentary.
These filamentary structures are seen in the "ring'', i.e.
around the spiral arms and in interarm regions, and are typically
about 3
(= 600 pc) long. The contribution of the filamentary
emission to the total extended emission seen in Fig. 4 may be estimated
from the total flux densities given above as 
.
Figure 4: Total emission of M 31 at 20 cm observed with
the VLA-D array (Fig. 1) combined with that observed with the
Effelsberg telescope (Fig. 3) to correct for the missing
interferometer spacings. The angular resolution is 45
. The
rms noise generally is 
increasing to 
at the map edges. All point sources brighter than 2 mJy are
subtracted. Blank squares indicate regions where subtraction of
Gaussian profiles was insufficient. The coordinate system is centred
on M 31 (
, position angle of major axis = 37 )
We checked our M 31 field for polarized features in the Galactic
foreground of the kind discovered by Wieringa et al. (1993) with the
WSRT. At 327 MHz these authors detected polarized filaments, typically
4 wide and up to 1 long, which are not visible in the
total emission. They interpreted these features as modulations
of a smooth polarized Galactic background by filaments in an ionized
foreground. The filaments are generally parallel to the Galactic
plane.
Figure 5: Combined VLA-Effelsberg map of the total emission of
M 31 at 20 cm with E-vectors of polarized emission
(VLA only) superimposed. The contour levels are 0.4, 0.8, 1.2, 1.6,
2.0 and 4.0 mJy/beam. The angular resolution is 45 . Point
sources brighter than 2 mJy are subtracted. A vector plotted with
1 length corresponds to a polarized intensity of 
.
As in Fig. 4 the coordinate system is centred on M 31
In Fig. 5 (click here) we show the polarization E-vectors (VLA only) on a contour map of the total (VLA+Effelsberg) emission. Neither in this figure nor in Fig. 2 we see filaments of the kind observed by Wieringa et al. running across M 31 parallel to the Galactic plane (which is practically parallel to constant declination in these figures). The absence of polarized foreground filaments in our maps at 1465 MHz is consistent with the experience of Wieringa et al. that the filaments are not visible at 610 MHz or 1420 MHz.
The polarized emission of M 31 shown in Figs. 2 and 5 is clearly
concentrated in the area of the bright "ring'' seen in total power. Out
to 12 kpc radius the average degree of polarization is 
(see Sect. 4.2). This value is a lower limit because large-scale
polarized emission from M 31 is missing in Figs. 2 and 5. However,
small-scale variations in the degree of polarization can still be
studied.
The highest polarized intensities occur not on the centre of the
"ring'' but on the inner and outer edges, giving the impression of
filaments. This phenomenon is most clearly visible in Fig. 2 at about
20 on either side of the minor axis. Many polarized filaments
exist on scales of about 
to 10\
running preferentially along the spiral arms.
The same effect is visible in the distribution of the percentage polarization shown in Fig. 6 (click here), on which the positions of the OB associations (van den Bergh 1964) are superimposed. On the spiral arms as delineated by the OB associations the degree of polarization is generally low, but across the arms it increases towards the interarm regions. Relatively low degrees of polarization on spiral arms also occur in other galaxies (Beck et al. 1996) and are due to Faraday effects and turbulence in the interstellar medium in the arms causing depolarization. The very low degrees of polarization inside some OB associations in Fig. 6 support this interpretation.
In Fig. 6 the polarization angles of the electric vector seem rather
chaotic. Closer inspection, though, reveals cells of coherent
polarization angles of 3 - 7 in extent corresponding
to about 1 kpc in the plane of M 31. A sudden change of polarization
angles often occurs near the border of an OB association suggesting
that supernova remnants in the association deform the regular magnetic
field causing variations in Faraday rotation. Brinks & Bajaja (1986)
showed that many OB associations are surrounded by an 
hole
which needs 1049-1053 erg to be
produced. The largest holes have diameters of 
kpc and
thus mean energy densities of 
,
enough to deform magnetic fields of 
G strength.
Figure 6: Distribution of the degree of polarization in M 31 (vectors)
with positions of OB associations catalogued by van den Bergh (1964)
(ellipses) superimposed. Cells of coherent polarization angles are
clearly visible. For the coordinate system centred on M 31 a position
angle of the major axis of 37 was used. No corrections for
Faraday rotation were applied
Figure 7: Distribution of the degree of polarization in M 31 shown in
grey-scale with contours of the surface brightness in H
(Devereux et al. 1994) superimposed, smoothed to a beam size of
45 . The contour levels are 0.5, 1.5, 5 and 25 in units of 
. The
M 31 coordinates are based on a position angle of the major axis of
37
Figure 7 (click here) shows the distribution of the percentage polarization in
grey-scale with contours of surface brightness in H
(Devereux et al. 1994) superimposed. Especially in the northern part of the
bright "ring'' a clear anticorrelation between degree of polarization
and H intensity exists, but like with the OB associations
there is little
one-to-one correlation. As apart from the density of thermal electrons other factors are involved in depolarization (Sokoloff et al. 1998), this is understandable.
Detailed discussions of the distribution of rotation measure and of
depolarization across M 31 derived from data at 20 cm (this
paper), 11 cm (Beck 1982) and the new Effelsberg survey at
6 cm (Beck et al. in preparation) will be
given in forthcoming papers.
Acknowledgements
We wish to thank Dr. E. Hummel and Dr. A.P. Rao for assistence
in the data reduction, Dr. N. Devereux for putting the H-map
at our disposal, and the referee Dr. S. Sukumar for useful comments
on the manuscript. The VLA staff is acknowledged for help in
preparing the observations and for performing them.