3.1. Total intensity image of 3C 119 at 8.4-GHz
The total intensity structure at one of our three frequencies (8.86 GHz) is shown in Fig. 1; this image has an angular resolution of 1.6x1.2 mas. The I images for the other frequency channels were very similar. The three brightest components, marked A, B, and C in Fig. 1, correspond to features in the image of Nan et al. (1991), and are mapped here in more detail than in the earlier 1.6-GHz image. The observations of Nan et al. (1991) indicated that the core is A, the northernmost feature in Fig. 1. In our observations, A is compact, but shows a slight extension towards the other two bright components to the southwest. The middle component B, separated from A in position angle , also shows a compact feature, but has two-sided extensions elongated along the direction joining the core and C, the brightest knot in our image. A small fraction of the flux density of the very resolved southern component D, which was well defined at 5 GHz (Fanti et al. 1986) and 1.6 GHz (Nan et al. 1991), is visible on this map. Other extended, steep-spectrum structures detected at lower frequencies are not visible here. The map accounts for about 62% of the total flux density of 3C 119 indicated by our VLA measurements (2.86 Jy).
We estimated the flux density present in each component by summing the flux density corresponding to the component using the AIPS task IMSTAT. We present these flux density estimates for components A, B, C, and D in Table 2. Note that the observing frequencies for IF34 and IF56 are the same, as pointed out above; we can see that our flux density estimates for these two independent measurements are quite consistent, and suggest that the uncertainty in our flux density estimates for individual components is a few mJy. We calculated approximate spectral indices () for components A, B, C, and D by obtaining linear least-squares fits on a log -log plot. We used an average of the two measurements at 8.52 GHz for the flux density value at that frequency. Although our measurements span a relatively small frequency range, they have the advantage of being simultaneous in time and of having the same resolution at each frequency; for this reason, we prefer to consider only our three measurements, rather than to try to add information from previous observations at other frequencies. The resulting spectral indices are given in Table 2. We can see that the spectral index of A is close to zero, while the spectra of components B, C, and D are steep, with spectral indices . This confirms that A is the core, while B, C, and D are optically thin jet components.
Table 2. Total intensity flux densities for VLBI components in 3C 119.
3.2. Linear polarization
A contour map of polarized flux density p and a grey-scale I map of 3C 119 at the same frequency as that shown in Fig. 1 are superimposed in Fig. 2. We detected polarization in components B and C, which are both polarized . The polarization structure of component B is knotty and one-sided, and follows the extension of the total intensity elongation toward the west. More than 200 mJy of polarized flux was detected in component C; the peaks of the I and p images are well aligned. Both the I and p distributions of C are quite resolved. Fig. 3 shows a superposition of a contour I image and a grey-scale distribution of the degree of polarization m derived from the superposition of p and I shown in Fig. 2. Although caution must be employed when interpreting m distributions derived in this way, this figure suggests that the leading edges of B and C are more highly polarized than other parts of these features, that the degree of polarization there reaches 25%, and that there is a fairly smooth increase in degree of polarization from east to west across component C. Parameters for components A, B, and C derived from the P maps at the four frequency channels are listed in Table 3. In the case of B and C, the polarization parameters were obtained by summing the Q and U flux densities attributed to those features using the AIPS task IMSTAT. Again, the frequencies measured by IF34 and IF56 are the same; comparison of the p and values for these two independent measurements suggests that the uncertainty in p is 2-3 mJy and the uncertainty in is about . We have used the level in the P images at which the noise contours become approximately uniform over the map (so that they could hide a weak polarized component) as an upper limit for the polarized flux density of component A; this probably corresponds to approximately a 2- limit. The degree of polarization inferred for A in this way is , showing that the core is quite weakly polarized. It is difficult to place meaningful limits on the polarization of D since it is so heavily resolved; we cannot exclude the possibility that it is fairly highly polarized (up to ) if the polarized-flux-density distribution is as diffuse as the I distribution.
Table 3. Polarization parameters for 3C 119.
Table 3 also shows the vector sum of the polarizations for components B and C measured in each frequency channel. A very interesting point is that the sum of the milliarcsecond-scale polarized flux density for B and C averaged over the three frequencies is 242 mJy in position angle , which is quite close to the integrated polarized flux density for 3C 119 indicated by our VLA data, 238 mJy in ; this demonstrates that we have mapped essentially all the integrated polarization in the source at these frequencies, and that only a very small amount of polarized flux is located on more extended scales, beyond our VLBI images.
3.3. Rotation measures and intrinsic magnetic fields
As noted above, the two features for which significant polarization was detected were the two knots B and C. Fig. 4 shows the values measured for the total polarization of these two knots at each of the three frequencies we observed, plotted as a function of . If variations of at our different observing frequencies are associated with Faraday rotation, we expect the values to form a straight line in this plot. We can see that the variations for both components B and C can be described well by a dependence, indicating the presence of Faraday rotation; the rotation measure of C (about 1590 rad/m2) is much larger than that of B (about 260 rad/m2). This demonstrates clearly that material associated with the large integrated rotation measure of 3C 119 is concentrated in the region of component C. If the uncertainty in our measurements is roughly (as suggested by a comparison of the values for IF34 and IF56 in Table 3), the corresponding uncertainty in the rotation measure is rad/m2. This, again, shows that our estimate for the rotation measure of C, which dominates the polarized flux on milliarcsecond scales, is consistent with the integrated value of 1728 rad/m2 observed by Kato et al. (1987). No correlations between the local RM and total intensity of the sort found for the high RM source 3C194 (Taylor et al. 1992) are apparent in our images.
The rotation measures were mapped by performing a weighted fit of the position angle to a -squared dependence using the AIPS task RM. Fig. 5 shows the magnitude of the rotation measure in the region where the solution errors are below rad/m2, which is equivalent to in at each pixel in the map. The area shown is smaller than that in Figs. 2 and 3; the area of low rotation-measure uncertainty in Fig. 5 covers a region about 2.5 x 4 mas in size around component C. The rotation measure distribution smoothly increases from 1200 rad/m2 in the northeast of component C to 2100 rad/m2 in the southwest of C, with an average gradient of 2300 rad/m2 per mas along the direction of elongation of the source structure. This gradient may not be sufficient to cause beam depolarization.
We can now take into account the rotation measure distribution in 3C 119 to derive the intrinsic direction of the vectors for the milliarcsecond-scale polarization. Fig. 6 displays an I contour plot with the intrinsic magnetic field vectors in component C after the output of the task RM (Fig. 5) has been used to "derotate" the observed VLBI values. The B vectors follow the direction of elongation of the structure, and are roughly perpendicular to the rotation-measure contours in Fig. 5. The B field appears to bend toward the south, curving roughly toward the diffuse component D.
© European Southern Observatory (ESO) 1999
Online publication: March 18, 1999