The model-fitting analyses have shown that the prominent, discrete components in the source can consistently be identified despite changes in the details of the VLBI array used and variations in the resulting u-v coverage. The model-fitting has also shown that the component of the source which is most sensitive to variations in the array and its u-v coverage is the component used to represent the smooth, diffuse and extended jet emission underlying the discrete components C1, C2, and C3.
For example, on 1991 March 6 when four telescopes were available, only one baseline was observing in the range of hour angles when the source position angle was aligned with the primary interferometer fringe, allowing the full flux density of the source to be measured. This lack of constraint on the underlying jet emission caused it to be essentially absent from the image and the model (Figure 5.4, Table 5.2). The discrete components in the source produced structure in the visibilities at other hour angles and were, therefore, well constrained in the image and model.
Conversely, at 1992 November 22 and 1993 October 20, when five telescopes were available and almost all baselines observed the source to cross the primary interferometer fringe, many constraints on the total flux density were obtained, allowing the underlying jet to be imaged and modelled (Figures 5.7 and 5.9, Tables 5.5 and 5.7). Thus the appearance and prominence of the jet-like component in both the images and the models is strongly related to details of the u-v coverage.
To achieve a uniform comparison of the data at each epoch, Figure 5.15 shows the 8.4 GHz images, each rotated by 39 
. The images have been aligned on the core component and the vertical space between each pair of images is proportional to the temporal difference between epochs. Each of the eight SHEVE images have been convolved with a beam (8.7 
3.2 mas @ 
) which is the average of the eight original beams, and plotted on a common flux density scale, with a peak flux density of 3.3 Jy/beam. The ninth epoch, with VLBA telescopes only, requires a completely different beam (12.2 2.3 @ 
). The peak flux density in this last image is 4.1 Jy/beam. The lowest contour in the montage was chosen to be 5% of the peak since most of the effects of differing u-v coverage on the underlying jet appear below this level and the discrete components are prominent above this level. The contour levels are 5, 10, 15, 20, 25, 35, 45, 55, 65, 75, 85, and 95% of 3.3 Jy/beam for the first eight images and of 4.1 Jy/beam for the ninth image.
Marked on Figure 5.15 is the position of the core and the 0.12c motion of component C2. For comparison, a 0.12c motion is also shown for component C1. The internal evolution in C1 is visible from epoch to epoch. At the end of the series of images, the new component C3 can be clearly seen as an extension to the core in the VLBA image.
Figure: 4.3 year sequence of 8.4 GHz Centaurus A observations