5 Discussion

We have revisited the $\sp1$D$-\sp1$S quadrupole transition of Ar III, looking in detail at the different factors that affect the accuracy of the effective collision strength. It is found that differences with the data computed in IP-X are not greater than 25% at the upper limit of the common temperature range. This is consistent with their original accuracy rating and with the general level of reliability that is usually attained in the type of massive computations where several ionic systems are treated simultaneously. Furthermore, a high-temperature limit of the present reduced effective collision strength has been estimated and is in good agreement (15%) with the high-energy Coulomb-Born limit of BCT. We are therefore confident about the accuracy ranking for this transition (20%). It appears that the suspicion of BCT arose because they overstretched the behaviour of the IP-X reduced effective collision strength at low temperatures to obtain an extrapolated high-temperature limit that appeared to be grossly discrepant with the Coulomb-Born limit. It has been shown here that for this transition the high-temperature trend is only reached at relatively high temperatures and differs significantly from those at the low- and intermediate-temperature regimes. Finally, it is hoped that we conveyed the computational difficulties that must be addressed in order to ensure reliable data throughout a chosen integration region.

Acknowledgements

Calculations were carried out on the IBM SP2 computer at CeCalCULA, Universidad de Los Andes, Mérida, Venezuela. The present collaboration also benefited from funding by the CNRS, IVIC, the Observatoire de Paris and the Ministère des Affaires Etrangères. This research has been supported by CONICIT under contract No. S1-95000521.