3 Results

Radio images of the compact structure of 3C395 are shown in Fig. 1. They were obtained by applying natural weighting to the data. Total intensity maps are displayed with superimposed vectors representing the electric field.

3.1 Total intensity

At 8.4 GHz, we can identify the "classical'' components A, B, and C (Fig. 1a), although A shows an elongated structure in the S-E direction as an indication of its now-known complex composition. The jet can be traced continuously from A to B, although we find no evidence of the radio jet beyond the latter component (Saikia et al. 1990; Lara et al. 1997, 1999). There is a sharp decrease in the intensity profile beyond component A. The total flux density recovered in the VLBI map is $1.70\pm 0.09$ Jy.

The images at 15.4 GHz from 1995 and 1998 (Figs. 1b-c) show only emission from components A and B, while component C is resolved out and is too faint to be detected. The observations at this frequency, separated by almost three years, confirm the constant separation between components A and B. Moreover, the structure of B has remained essentially invariable during this period of time. On the other hand, A shows a bent core-jet structure with clear changes between the two epochs, probably resulting from a new ejected component (see also Fig. 3). The total flux density recovered in the VLBI maps at 15.4 GHz is the same in both epochs, $1.12\pm 0.06$ Jy, despite the structural variations in component A. Close-in-time space-VLBI observations at 4.85 GHz (1 May 1998; Lara et al. 1999), which provide an angular resolution comparable to our 15.4 GHz observations, show a similar bent core-jet structure.

At 22.2 GHz (Fig. 1d), the detailed structure of component A is even more evident, consisting of an unresolved component at the western end of the brightness distribution (assumed to be the true core), a new component at a distance of 0.8 mas from the core at position angle PA $\sim 110^{\circ}$ and a jet-like feature directed along PA $\sim 130^{\circ}$- $140^{\circ}$. The flux density in the VLBI map is $0.94\pm 0.05$ Jy.

We have fitted simple elliptical Gaussian components to the visibility data using a least square algorithm within the Difmap package in order to obtain a quantitative description of the milliarcsecond structure of 3C395. Parameters describing the Gaussian components for each epoch and frequency are given in Table 1. Component A requires a rather complex model, three Gaussian components, to satisfactorily reproduce the uv-data: A1 stands for the core; A2 describes the new component which is clearly separated from the core in 1998.50; A3 describes the outer jet emission beyond component A2 (see Fig. 1d). Component B is described in terms of 2 Gaussian components to account for its compact and extended emission, respectively. Finally, component C, only detected at 8.4 GHz, is rather weak and extended and it is represented in terms of a single and elongated Gaussian component. We also include in Table 1 a similar fit obtained at 4.85 GHz in 1998.33 (Lara et al. 1999), but modified to describe B with two components, instead of the three used in the quoted publication.

   

Table 1: Gaussian components derived from the model fitting

	IDa	Sb	Dc	PAd	Le	rf	$\Phi^g$
		(mJy)	(mas)	($^{\circ}$)	(mas)		($^{\circ}$)
A1	X95	741	-	-	0.22	?	133
	U95	691	-	-	0.20	0.17	106
	U98	617	-	-	0.16	0.28	115
	K98	570	-	-	0.14	?	107
	C98	124	-	-	0.59	?	106
A2	X95	353	0.56	113	0.63	0.12	134
	U95	179	0.61	115	0.56	0.30	132
	U98	314	0.76	112	0.32	0.49	131
	K98	211	0.77	111	0.26	0.49	129
	C98	629	0.66	109	0.30	0.74	132
A3	X95	323	1.25	135	0.66	0.50	119
	U95	142	1.30	135	0.50	0.59	101
	U98	73	1.50	132	1.09	0.38	131
	K98	84	1.06	126	1.71	0.23	140
	C98	245	1.57	126	0.58	0.71	160
C	X95	25	6.97	119	4.60	0.44	127
	C98	48	7.60	118	8.70	0.22	118
B1	X95	168	15.86	120	1.46	0.80	38
	U95	71	15.88	120	1.48	0.77	45
	U98	72	15.92	120	1.53	0.79	64
	K98	31	15.96	120	1.62	0.57	53
	C98	270	16.00	120	1.54	0.97	25
B2	X95	82	16.15	118	4.47	0.55	138
	U95	30	15.81	117	3.55	0.40	137
	U98	28	15.81	118	3.95	0.46	150
	K98	34	15.78	119	2.63	0.47	137
	C98	79	16.83	117	6.18	0.40	137

^a Frequency and year of observation. C, X, U and K stand for 4.85, 8.4, 15.4 and 22.2 GHz, respectively.
^b Flux density of the Gaussian component.
^c Angular distance from the western-most component A1.
^d Position angle with respect to A1, measured north through east.
^e Length of the major axis of the Gaussian component.
^f Ratio between the major and minor axis of the Gaussian component.
^g Orientation of the major axis, defined in the same sense as the position angle.
A question-mark indicates that the parameter involved cannot be well constrained by our data.

If we combine the close-in-time data at epoch 1998 (4.85 GHz on 1998.33 and 15.4/22.2 GHz on 1998.50) with the results shown in Table 1, we can derive the spectra for the different components of 3C395 (Fig. 2). The spectral decomposition shows that the core peaks at a frequency between 5 and 15 GHz in 1998. On the other hand, the other jet components (A2, B1) show a typical steep spectrum with spectral index $\alpha_{\rm A2}=-0.68$, $\alpha_{\rm B1}=-1.35$ (the spectral index $\alpha$ is defined so that the flux density $S\propto \nu^{\alpha}$). Considering the data at 8.4/15.4 GHz on 1995.91, we can confirm the steep spectrum for A2 and B1, and a flat spectrum for the core ( $\alpha_{\rm A1}=-0.12$), consistent with a shift of the turnover frequency towards higher frequencies at the later epoch. It is interesting to note that, in both epochs, the spectrum steepens with increasing core separation.

The components A3 and B2 show also steep spectra in both epochs. We have not included them in Fig. 2 since they fit the extended emission of the inner jet and have different angular sizes in our model fitting, depending on the observing frequency.

Figure 2: Spectrum of components A1 (circles), A2 (triangles) and B1 (squares) of 3C395 at epoch 1995 (dashed line) and epoch 1998 (solid line). See Table 1 for numerical values. The lines represent linear fits to the data of each component, excepting A1, for which the line helps to visualize the spectral dependence of its flux density. Spectral index values of B1 and A2 correspond to epoch 1998

3.2 Polarized emission

The polarized emission of 3C395 comes predominantly from component A. The orientation of the magnetic field is essentially aligned with the radio jet, except at the compact core (A1), where we observe a strong dependence of the polarized intensity and of the magnetic field orientation with the frequency and epoch of observations. We further discuss this issue in the next section.

The polarized structure of component B is complex. Since this component is probably the result of the Doppler boosting of the radiation from a portion of a relativistic jet which is sharply directed towards the observer (Lara et al. 1994), the changes in the orientation of the magnetic vector in this region might reflect the change of orientation of the jet trajectory with respect to the line of sight. It is interesting to note that the configuration of the magnetic field suggests that the interaction of the jet with the surrounding medium is not significant, otherwise the magnetic field would be compressed and disposed perpendicularly to the jet direction. This result supports the idea that component B is a consequence of a geometry effect in the jet. Moreover, the fact that component B does not show any dependence of the magnetic field orientation and magnitude with time is consistent with its stationary character.

Component C is only detected at 8.4 GHz with polarized emission slightly above the noise level. However, we observe that the magnetic field tends to be parallel to the jet trajectory in this component. This fact, joined with its large size, does not lend support to the interpretation of C as a moving shocked component, favoring the hypothesis of C being mainly the result of the underlying jet emission.