4.3. Tests and Uncertainties

It should soon be possible to test the results based on NV for some QSOs using either the intercombination lines or other permitted line ratios such as NIII 990 / CIII 977 (Hamann, Korista & Ferland 1999). However, the main conclusion for Z Z inferred from NV appears to be robust. It is not, for example, sensitive to uncertain parameters used in the photoionization calculations. In fact, adopting other (less than optimal) parameters only strengthens the case for large Z's (Ferland et al. 1996). Photoionized gas is simply too cool to produce the large observed NV / HeII ratios with solar or lower N/He abundances. We therefore considered non-radiatively heated clouds, where higher temperatures might enhance the collisionally-excited NV line relative to the HeII recombination transition. We found that collisionally ionized clouds can indeed match the high NV / HeII ratios with roughly solar abundances if the temperatures are above 10⁵ K. However, this situation still underpredicts the NV / CIV ratio - because both lines are collisionally excited - and overpredicts several other lines (e.g. SiIV 1397 and OIV] 1402). Such clouds are also thermally unstable and would tend to runaway to coronal temperatures (where no NV would be present) on thermal timescales of order tens of seconds (Hamann et al. 1995a, Ferland et al. 1996). Finally, we also explored the possibility that the NV emission is enhanced by scattering in an outflowing BAL region rather than by abundance effects (Krolik & Voit 1998, Turnshek et al. 1996, Turnshek 1988). Explicit calculations, which use observed BAL profiles for input and include scattering of both the underlying continuum and Ly line radiation, indicate that the flux scattered by NV in BAL winds should contribute negligibly to the measured NV emission lines (Hamann & Korista 1996, Hamann et al. 1999, HF99).