As the revised spectrum above differs substantially from that currently in
use it follows that spectral indices calculated from existing ATCA observations
may include significant errors. In the worst case, spectral indices computed
between 4.8GHz and 8.6GHz are too steep by 
0.3. It is therefore
recommended that the above expression for the flux density for 1934-638
be implemented as soon as possible as the default in both the ATCA online
calibration software (CACAL) and the standard reduction packages (AIPS
and MIRIAD). [The new flux scale was implemented in August 1994.]
Although there is no evidence among the flux density measurements described here that 1934-638 is time-variable, it is intended to continue monitoring the source both with the ATCA and Parkes not only to check for variability but to improve the spectral sampling across the ATCA observing bands (eg 6.7GHz). An ATCA observing program (C280) is also underway to isolate additional compact Southern sources suitable for flux calibration, allowing the flux scale to be defined by a grid of calibration sources rather than relying solely on 1934-638.
Finally, it is worth pointing out that the emphasis of this analysis has been to reconcile the ATCA flux scale with those in use in the North, rather than define the spectrum of 1934-638 in an absolute sense. While the spectrum presented here is tied to the widely-used absolute spectrum of 3C286 and 3C295 defined by Baars et al. (1977), more recently Gough (1994) has found evidence above 8GHz for a significantly greater absolute flux density for 1934-638 than the figures given here. This question may possibly be resolved by more accurate single-dish measurements of 1934-638 and Virgo A, for which latter source the absolute spectrum has been measured directly.