The X-band (8.4-GHz) and L-band (1.4-GHz) data used in this paper were described in Paper I. Data from a short (0.5 h) C-band (4.9-GHz) observation with the NRAO Very Large Array (VLA) in its C configuration, taken on 1984 Jun 11, were kindly provided by Alan Bridle. The U-band (15-GHz) images presented here are the result of a 3-hour observation with the VLA in its D configuration taken on 1999 Mar 11. All data were reduced and analysed using the AIPS software package.
2.1. Flux scales
As described in Paper I, both 3C 48 and 3C 286 were used as primary flux calibrators for the L- and X-band observations, which were taken between 1994 Nov 10 and 1995 Nov 28. Specifically, 3C 286 was used for the A-configuration X-band observations and 3C 48 for all others. The data analysis in Paper I used the older version of the SETJY AIPS task, which incorrectly rounded coefficients in the analytic expression for the flux densities of calibrator sources. However, the effect is very small when combined with the change between the old (1990) values of the coefficients and the more appropriate new (1995.2) values. L-band fluxes should be reduced by 0.3% and the X-band B, C and D-configuration fluxes by 2%; the A-configuration X-band flux, based on 3C 286, should be reduced by 2.5% when the effect of the partial resolution of the source is incorporated, but we use 2% for all the X-band data in what follows. The primary flux calibrator for the C-band observations was 3C 48; the fluxes derived from this are increased by 2% to take account of the variation in 3C 48's flux density between the epoch of observation and 1995 (Perley & Taylor 1999). The primary flux calibrator for the U-band observations was 3C 48, and it is assumed that 3C 48 has not varied significantly between 1995 and 1999, so that the flux densities derived from this are correct.
The long, multi-array X- and L-band observations of Paper I sample the uv plane more densely and have a much broader range of baselines than the short, single-configuration C-band observations. The U-band observations are long and so well-sampled, but still cover a narrow range of baselines. In order to try to correct for this and ensure that fluxes from maps were directly comparable I mapped the source at all four frequencies with the CLEAN-based AIPS task IMAGR using only baselines between 1.8 and 50 k; the uv plane was acceptably sampled in this range at all four frequencies, though the C-band data are still sparse (Fig. 1). Reweighting the data (using different values of the `robustness' parameter in IMAGR) ensured that the fitted beams were similar. The same 4.0-arcsec restoring Gaussian was then used for each map, scaling residuals appropriately. (Experiment showed that, in spite of folklore to the contrary, there was no significant difference between the results of this procedure and those of restoring clean components with the Gaussian which was the best fit to the dirty beam and then convolving to the required resolution.) Primary beam correction was applied with the AIPS task PBCOR. Small shifts were applied to each image so that the unresolved cores were aligned to better than 0.05 arcsec.
© European Southern Observatory (ESO) 1999
Online publication: September 2, 1999