Broad emission lines provide the most readily available
abundance diagnostics because they are present in all QSOs.
Many BELs can be measured in large QSO samples using medium-resolution
spectra. However, the strengths of the metal lines, such as CIV 
1549,
relative to Ly
are surprisingly
insensitive to the global
metallicity (HF99,
Hamann & Ferland
1993b).
QSOs could actually have a very wide range of
metallicities while displaying qualitatively similar BEL spectra
(also Korista et
al. 1998).
Nonetheless, some line ratios are
sensitive to the relative abundances.
The challenge is to identify observable lines that have
significant abundance sensitivities above their unavoidable
dependences on other factors. In general, we must rely
computational models to quantify the various parameter
sensitivities. Since the BEL region (BELR) spans a range of radii
(Peterson 1993)
and probably has a wide range of densities and ionizations
(Ferland et al. 1992,
Baldwin et al. 1995,
Hamann et al. 1998),
the most important
criterion for abundance-sensitive line ratios is that they form
under similar conditions (e.g. in the same or overlapping regions).
Line ratios involving nitrogen should be particularly useful for
constraining the total metallicity and evolutionary state of QSO
environments (Section 2.4).
Boyle (1990),
Laor et al. (1995)
and Osmer & Shields (this volume) present
labeled plots showing the UV emission lines that are generally
available for abundance studies.