4 Discussion

4.1 Structural variability

The quasar 3C395, considered in the 1980's as being among the fast superluminal sources (Waak et al. 1985; Simon et al. 1988), showed, however, a stationary structure in observations made during the 90's (Lara et al. 1997), even though pronounced flux density variations are usually observed in this source (UMRAO database). Lara et al. (1999) found a large bend in the inner region of the jet, close to the core, and suggested that it could explain why previous Earth-based cm-VLBI observations did not detect the ejection of the new moving components expected from the flux density variability: a large bend in the inner jet would make it very difficult to correlate flux density variability within A with structural variations beyond this component because the effect of relativistic time-delay makes the time scales of these two events very different, and the decrease in the Doppler factor after the bend produces a large diminution in the flux density of possible moving components. As a step forward in the understanding of this source, the observations presented here show clear evidence of structural variability close to the core of 3C395 between 1995 and 1998, in agreement with the previous scenario. In Fig. 3 we display the brightness distribution in total intensity and polarization of component A, as observed at 15.4 GHz in 1995.91 and 1998.50, respectively. Changes are evident, both in total intensity and polarization. The structural variability is consistent with a new component (labeled A2 in Table 1) which has been ejected from the core and is traveling along the jet. From the results of the model fitting at 15.4 GHz we estimate an apparent velocity of $\beta_{\rm app} = 1.2\pm 0.8 c$ for this component, where the error is calculated assuming a conservative uncertainty in the position equal to one fifth of the beam size.

Figure 3: Structural variations in total intensity (left) and polarization (right) in the inner structure of 3C395, observed at 15.4 GHz between 1995.91 and 1998.50. The contours are spaced by factors of $\sqrt {2}$ in all maps. Vectors represent the same as in Fig. 1. For each map we list the Gaussian beam size (in mas), the first contour level (mJybeam-1), the peak of brightness (Jybeam-1) and in polarization maps also the polarized flux corresponding to 1 mas $\vec{E}$-vector length (mJybeam-1). U-95 left: beam = $0.67\times 0.45$ PA $-9.0^{\circ }$; 1st cntr = 1.5; peak = 0.652. U-95 right: beam = $0.94\times 0.64$ PA $-6.1^{\circ }$; 1st cntr = 3; peak = 0.016; 1 mas $\equiv$ 10. U-98 left: beam = $0.63\times 0.46$ PA $-4.9^{\circ }$; 1st cntr = 1.5; peak = 0.588; 1 mas $\equiv$ 2.5. U-98 right: beam = $0.92\times 0.70$PA $-1^{\circ }$; 1st cntr = 3; peak = 0.014; 1 mas $\equiv$ 10

4.2 Polarization in the inner jet region

In Table 2 we display, for the different frequencies and epochs of observations, the total and polarized flux densities, the mean fractional polarizations and the mean EVPAs of the components of 3C395. Total flux densities are taken from Table 1 while polarized flux densities were measured from the polarization images, defining a polygonal area containing all the component emission. We were not able to discriminate between the polarized emission from components A2 and A3, or B1 and B2, since they present similar EVPA. Thus, we give values for the blending of these components.

   

Table 2: Polarization in 3C395

Component	IDa	S $_{\rm tot}^b$	S $_{\rm pol}^c$	$p_{\rm m}^d$	$\chi^e$
		(mJy)	(mJy)	(%)	($^{\circ}$)
A1+A2+A3	X95	1417	49	3.5	22
A1	U95	691	11	1.6	158
A1	U98	617	3	0.5	$\sim$93
A1	K98	570	15	2.6	64
A2+A3	U95	321	25	7.8	40
A2+A3	U98	387	24	6.2	42
A2+A3	K98	295	19	6.4	45
B1+B2	X95	248	27	10.9	19
B1+B2	U95	101	10	9.9	21
B1+B2	U98	100	11	11.0	22
B1+B2	K98	65	6	9.2	15?

^a Frequency and year of observation: X, U and K stand for 8.4, 15.4 and 22.2 GHz, respectively.
^b Total flux density of the component(s).
^c Polarized flux density of the component(s).
^d Mean fractional polarization.
^e Mean EVPA of the component(s).

Figure 4: Frequency dependent polarization variations in the very compact structure of 3C395, observed at 15.4 GHz and 22.2 GHz in 1998.50. The contours are spaced by factors of $\sqrt {2}$ in all maps. Vectors represent the same as in Fig. 1. For each map we list the Gaussian beam size (in mas), the first contour level (mJybeam-1), the peak of brightness (Jybeam-1) and the polarized flux corresponding to 1 mas $\vec{E}$-vector length (mJybeam-1). Top: beam = $0.92\times 0.70$PA $-1^{\circ }$; 1st cntr = 3; peak = 0.014; 1 mas $\equiv$ 10. Bottom: beam = $0.67\times 0.51$ PA $-2.9^{\circ }$; 1st cntr = 1.5; peak = 0.015; 1 mas $\equiv$ 10

The core component, A1, shows strong variability in the polarized structure at 15.4 GHz (see Fig. 3). The polarized flux density of this component decreases from 11 to 3 mJy between 1995.91 and 1998.50. Although A1 is weak in the latter epoch, we are confident that it is a real feature in our map, at a level above $5\sigma$. The EVPA also rotates significantly from 158$^{\circ}$ to $\sim$ $93^{\circ}$ in this time period.

Comparing the results at 15.4 and 22.2 GHz from 1998 (Fig. 4), we find that component A1 presents a strongly inverted spectrum in polarization (15 mJy at 22.2 GHz and 3 mJy at 15.4 GHz). Since in 1998 the new component A2 was completely ejected from the core region, we suggest that i) the "true" core is almost unpolarized at frequencies lower than 15 GHz, as expected from synchrotron self-absorption near the core, and ii) the polarization variations observed between 1995 and 1998 at 15.4 GHz in A1 are most plausibly due to the process of ejection of component A2 and to the resulting opacity changes in the jet. From the 1998 data, a RM of $\sim$+2500 rad m^-2 can be estimated in component A1, in agreement with Taylor (2000), which is consistent with the depolarization of the core region at low frequencies. We derive an intrinsic orientation of the electric vector of $35^{\circ}$, in agreement with the orientation of A2+A3, and therefore with a magnetic field oriented along the jet. The origin of the Faraday rotation is unclear to us. While it might have an origin internal to the jet, the effects produced by a possible external ionized screen in the core region cannot be disentangled. We note that the effect of bandwidth depolarization is negligible with the frequencies and bandwidths used in our observations, even with the high RM involved in the core of 3C395.

Component A2+A3 shows an EVPA which has remained essentially constant ($\sim$ $40^{\circ}$) at 15.4 GHz from 1995 to 1998. On the other hand, the overall degree of polarization decreases from $\sim$$8\%$ to $\sim$$6\%$ in this time period. Moreover, the gradual rotation of the electric vector along component A2+A3 is also noticeable (see Fig. 3). This is probably related to the curvature present in the jet, as suggested by Lara et al. (1999). At 8.4 GHz it is not possible to discern the different components within A in polarization. However, since the peak of the polarized emission is coincident with the position of component A2 and assuming that A1 is almost unpolarized at this frequency, we can ascribe the observed EVPA to A2+A3. Comparing with results from 1995 at 15.4 GHz we estimate for this component a RM of $\sim$-350rad m^-2, which is consistent with the small variation of the EVPA observed between 15.4 and 22.2 GHz in 1998.

If we interpret the moving component A2 as a planar shock wave, then the compression of the shock enhances the component of the underlying magnetic field parallel to the shock front (transverse to the jet trajectory). This could explain the different field orientation of the core component A1 with respect to A2+A3 at 15.4 GHz in 1995. However, since the orientation of the magnetic field in 1998 is along the jet (once corrected of Faraday rotation), we should conclude that the enhancement of the field produced by the shock wave is not strong enough to dominate over the underlying parallel field component. This is consistent with the slight decrease in the degree of polarization at 15.4 GHz observed between 1995 and 1998, and in agreement with the general tendency in quasars to have a net magnetic field orientation parallel to the jet (Cawthorne et al. 1993; Wardle 1998).

Finally, we note the presence of a strong gradient in the RM along the inner jet of 3C395, in agreement with Taylor (2000). Strong gradients in relativistic jets have also been found in other compact sources, such as OJ287 (Gabuzda & Gómez 2000) and BL-Lac (Reynolds & Cawthorne 2000).

4.3 Polarization in component B

Component B1+B2 has a degree of polarization similar for all frequencies and observing epochs, which indicates that Faraday rotation is not significant in this region of 3C395. We note that Taylor (2000) finds a moderate RM of $68 \pm 40$ rad m^-2in component B, which is fully consistent with our results. Moreover, the observed EVPA implies that the magnetic field is oriented along the jet. As previously mentioned (Sect. 3.2), the configuration of the magnetic field suggests that this component is not the result of a possible interaction of the jet with the external medium, and argues in favor of a change in the jet geometry.