Following Garrington et al. (1991), we have measured for each
source (and for each component in the source) the mean scalar polarization,
m', defined as the integrated polarization flux over the total intensity.
For the lobes, typical values of m' (at 6 cm) range from 10% to 30%,
with a median value 
. The m' values of the two lobes are
rather similar, contrary to what was found at 20 cm (see Paper I), where the
lobe containing the brighter jet was significantly more polarized than the
other lobe.
The old maps at 20 cm and the new at 6 cm have been analysed to determine
the depolarization, 
, between these two frequencies. The
depolarization
is defined as the ratio of the mean scalar fractional polarization, m', at
20 cm over that at 6 cm: 
. The m' values we find at 6 cm are larger than the values
of m'20, implying depolarization between the two frequencies.
The values of are presented in Col. 10 of
Table 3 (click here). They are peaked around 0.7 and extend down to 
.
Only about 25% of the objects show little or no depolarization
(
). Head-tail sources seem to be more depolarized,
although the statistics is small.
The differences between the position angle of the polarization at 6 and
20 cm (
) are presented in Col. 12 of Table 3 (click here). The values
obtained cover a broad range with a number of sources showing little or no
rotation between the two wavelengths. We used the component polarization
position angles to derive two-point Faraday rotation measures (RMs). Since
position angles are ambiguous by 
, the calculated RMs are
ambiguous by 
. In practice we have:
with in degrees.
For 22 sources the integrated polarization data in Parma & Weiler
(1981) at 2.7 GHz and of Mack et al. (1994) at 10.8 GHz
were used to resolve the ambiguity (see Appendix). We find n= 0 in 13
sources (0034+25, 0828+32, 0836+29, 1102+30, 1113+29, 1122+39,
1141+35, 1243+26, 1316+29, 1322+36, 1357+28, 1422+26,
1455+28), 
in seven cases and n= 2 in two cases (see
Appendix). In this subsample, which we consider free of ambiguity, two
thirds of the sources have RMs less than 50 rad 
. These figures
are mostly consistent with Faraday rotation in the galactic foreground
screen (Simard-Normandin et al. 1981).
We note that of the three sources with 
larger than 100
rad , two (1254+27 and 1626+39) are in rich Abell clusters and
the third (2116+26) is at low galactic latitude (
).
For the remaining sources for which we do not have additional data we choose the n value which minimizes the absolute value of RM. For them we expect, statistically, a comparable fractional number of ambiguities.
We have compared the RMs of the two lobes for the double sources (38 objects).
Following the result of Conway et al. (1983), we assume that
the differences in RM (
) are free of ambiguity. The rms
difference is 
. About two
thirds of the sources have 
, in
agreement with what Simonetti & Cordes (1986) found. Only in
three cases did we find differences in the rotation angles of the two lobes
above 
, implying RM differences above 20 rad . These
few cases are mentioned in the Notes on individual sources in the Appendix.
As for the sample of PRG studied by Garrington et al. (1991), we do not find any correlation between rotation measure and depolarization.
On the basis of the obtained RM, we conclude that the
polarization position angle at 6 cm is within 
of the
intrinsic position angle and we derive the magnetic field projected
geometry from the maps at this wavelength by a rotation of 
.
In the double sources, the magnetic field appears in general very well oriented along the outer borders of the component, in agreement with earlier findings (e.g. Bridle et al. 1991). The best examples of this geometry are: 0755+37, 0828+32, 1141+37, 1322+36, 1441+26, 1455+28, 1658+30.
In general the direction of the magnetic field in jets is perpendicular to the jet orientation, in agreement with what found in Bridle & Perley (1984), for the typical range of radio powers of our sample. In a number of double sources the jet shows up in the polarisation map for its magnetic field orientation which contrast with that of the radio lobes (see, for example, 0755+37, 1322+36, 1613+27, 1658+30). In some double sources the jet, which is not well distinguished from the lobe in the total-intensity map, emerges clearly in the polarization map (0828+32, 1457+29 1609+31).
There are, however, situations where we see a parallel magnetic field. The first is at the jet sides, as in 1357+28 and, most notably, in 1553+24. This characteristic has already been found in 3C 31 (R.A. Laing et al. in preparation). The second is in the final part of the jets, in sources dominated by jet emission (see 0915+32 and 1243+26).
For the hot-spots the situation is not very clear. In general we find low fractional polarization and an unclear field geometry. We believe that the generally inadequate resolution of our observations mixes regions with different field geometry and causes beam depolarization.
In the head-tail sources the magnetic field tends to run parallel to the tail.
As discussed in Laing (1988) and Garrington & Conway
(1991), the depolarization observed in PRG is likely to be caused by
an inhomogeneous, unresolved, foreground Faraday screen, which rotates the
polarization position angle randomly across the observing beam. The more
depolarized lobe is, therefore, the lobe seen through a larger halo depth.
The depolarization ratio 
is related to the standard
deviation of the Faraday depth, 
, by the relation
where K= 0.81, is in units of 
and
and 
are in metres (Garrington & Conway
1991).
From this we derive for each source the lobes' Faraday dispersion parameters,
and 
corresponding to the lobe with
higher and lower depolarization, respectively. The values we find range mostly
from 4 to 
. These values are similar to those
of the low redshift (z < 0.2) sources of the Garrington & Conway
(1991) sample, and are much lower than the values of the sources at
high redshift. For each source with lobes, we have also estimated 
defined as the ratio of the highest to the lower depolarization
(
). Using this parameter, we have
searched for correlation between and source size. Although we
find a marginal tendency for large in large sources, no obvious
relation has been found between and source size.
In their work on PRG, Garrington & Conway (1991) conclude that
the major contribution to depolarization arises from a halo of hot gas
surrounding the radio source. By using a Montecarlo simulation, we have
derived the parameters of such a halo. From the distributions of the and of the ratio 
we
find the following parameters:
where B0 and n0 are the halo's central magnetic field and density. These
parameters agree well with those found by Garrington & Conway
(1991) for the low redshift (z < 1) radio galaxies.
A discussion on the depolarization asymmetry in the source components is presented in a companion paper (Morganti et al. 1996).