The image resulting from the SHEVE+VLBA observations of 1992 November 22 can be seen in Figure 5.16. The structure can still be seen to consist of the familiar core component and components C1 and C2, with an underlying smoother jet. However, an additional component appears to the south-west of the core component which is not seen in the images from SHEVE data alone (Figures 5.1 and 5.7).
Figure 5.16: SHEVE+VLBA image of Centaurus A from 1992 Nov. 22. Map peak, 1.6 Jy/beam. Contours, -0.5, 0.5, 1, 2, 4, 8, 16, 32, 64% of peak. Beam FWHM, 2.2 
1.7 mas @ 5.4 
.
Figure 5.17: SHEVE+VLBA image of Centaurus A from 1993 Oct. 20. Map peak, 1.8 Jy/beam. Contours, -0.5, 0.5, 1, 2, 4, 8, 16, 32, 64% of peak. Beam FWHM, 3.3 1.9 mas @ 27.3 .
In the image resulting from the 1993 October 20 observations (Figure 5.17) even more structure can be seen to the south-west of the core component. In this image two components lie along the position angle of the jet.
The structure to the south-west of the core appears weaker than the main emission and this is essentially the reason why it was missed in imaging SHEVE-only datasets. The improvement in u-v coverage gained by combining the SHEVE and VLBA data allowed the weak components to appear strongly during the imaging process, becoming the brightest features in the residual dirty images following each iteration of clean and self-calibration, after the bright emission had been cleaned. During the imaging of the SHEVE-only data, these weak residual features appear only marginally above the noise level and can easily be missed or ignored if the imager is conservative and the clean windows are restricted to the core and the components north-east of the core.
To test the reality, or otherwise, of the features south-west of the core, many trial images of these data were made in an attempt to produce a good image without the additional features, but all attempts were unsuccessful. The features remained the strongest after each iteration near the end of the imaging procedure and the inclusion of the features in the clean model always greatly improved the agreement between model and data.
Additional trials were conducted which involved the use of datasets produced with the Caltech VLBI task FAKE. Simulated data were produced using various combinations of SHEVE and VLBA telescopes and a variety of different source models. These trials were undertaken on a blind basis with one person producing the FAKE data and another producing the images with no knowledge of the source model used for the data. In each case, the trials showed that no spurious features were added and all real features were found.
A final trial was made, allowing three independent investigators to image the real datasets. All three agreed in the existence of the features.
The two images in Figures 5.16 and 5.17 nonetheless have a difference. At 1992 November 22, only one additional component was revealed, but at 1993 October 20, two additional components were revealed. This difference is primarily due to the difference in data quality between the two epochs, with the 1993 October 20 data being of a higher quality. Two telescopes also needed to be removed from the 1992 November 22 dataset due to equipment failures, leading to a difference in the u-v coverage. However the second additional component, furtherest from the core toward the south-west at 1993 October 20 is weakly apparent in the 1992 November 22 image if its existence is assumed a priori. Consistent with the discussion of the imaging philosophy adopted here (see Chapter 2), that each data set be considered strictly upon its merits, the second component was not well enough constrained to appear in the 1992 November 22 image and was therefore excluded from the clean windows.
The additional features, seen best in Figure 5.17, can be interpreted as evidence for a sub-pc-scale counterjet emerging from the core toward the south-west. The counterjet appears much weaker than the main jet. This may be due to a combination of the effects of relativistic beaming and the effects of the possible free-free absorbing structure discussed earlier. Presumably the images show just the brightest portions of the counterjet emission (analagous to the bright components C1 and C2 in the main jet) and the smoother underlying jet emission may be too faint to be detected.
Using Figure 5.17 and correcting for the free-free absorbing structure discussed in 
5.2, an estimate of the intrinsic jet to counterjet surface brightness ratio (R) can be found. The jet to counterjet surface brightness ratio is defined here as the ratio of the brightest jet feature to the brightest counterjet feature. For Centaurus A on the sub-pc-scale, the intrinsic (corrected) ratio is 
.
The equation describing the jet to counterjet surface brightness ratio due to relativistic beaming is
for a spherical component of emission. 
is the speed of the radiating material, 
is the angle the motion of the material makes to our line of sight, and 
is the spectral index ( 
). Using the assumption of 
as in 5.2 and the corrected jet to counterjet brightness ratio, R, then 
. Now, with the various estimates for the speed of material in the jet from the model-fitting analysis in 5.3.2, the jet angle to the line of sight can be estimated.
With the subluminal ( 
) speed for component C2, a solution for can just be found, the minimum speed required for a solution being 
. However, the angle to the line of sight for the jet in this case is 
, not in good agreement with the large-scale structure of the radio source. For the faster speeds implied from the episodic rapid evolution, 
and 
, solutions for of 
and 
respectively can be found, with an upper limit of 
(for 
). These large values for the angle to the line of sight are in agreement with the extended morphology of the kpc-scale radio source and the optical morphology of the host galaxy.
Therefore, it may be that the rapid evolution occasionally seen in Centaurus A is indicative of the motion of material in the jet, with the subluminal components perhaps representing the motion of slower and longer lived patterns in the jet. It follows that the temporal frequency at which the sub-pc-scale structure of Centaurus A has been monitored is such that this rapid evolution may be under-sampled and that other episodes of rapid variability may have gone unobserved.