Initially, we shall suppose that the observed CIV line emitting region is seen directly rather than via scattered photons. Furthermore, suppose that the UV continuum and CIV line emission are subject to the same extinction as the optical non-stellar continuum/line regions. Since we have determined the optical reddening to be in the range E(B-V)=0.61-1.09, we can deredden the CIV line in order to estimate its intrinsic (i.e. dereddened) flux. Thus, given this supposition that the UV line emitting region is seen directly, we constrain the intrinsic CIV line flux to lie in the range
where we have included the 1- 
errors in the observed flux and used
the UV extinction law of Osterbrock (1989). This is a rather large line
flux, corresponding to an isotropic luminosity of 
or greater in the CIV line alone.
To quantitatively assess how large this line flux is, consider the lower
end of this range corresponding to E(B-V)=0.61. For this reddening, the
dereddened H 
flux is
leading to a lower limit on the intrinsic CIV/H flux ratio
of 15. This ratio is very sensitive to the reddening assumed and can
greatly exceed this value if E(B-V)>0.61. In unreddened Seyfert nuclei,
this ratio is often significantly smaller. For example, in the AGN Watch
Campaign on NGC 3783, the intrinsic CIV/H flux ratio is
(Reichert et al. 1994; Stirpe et al. 1994). Similarly, the CIV/H flux ratios found during the monitoring campaigns on
NGC 4151 (Crenshaw et al. 1996; Kaspi et al. 1996) and NGC 5548 (Korista et
al. 1995) are 
and 
, respectively.
We must conclude that the CIV line flux is unusually high compared
with the optical line fluxes, or that one of our assumptions has broken
down. There are three possible ways that our above argument might be
flawed. First, source variability during the 9 months separating the UV
and optical observations may produce an apparently unusual line ratio, even
if the intrinsic line ratio is normal. In our minimum reddening case
(E(B-V)=0.61), only mild variability ( 
per cent over 9 months)
is required to make the observed CVI/H ratio of 15
consistent with the that seen in other objects. As one postulates higher
reddening values, the more extreme is the inferred intrinsic line ratio and
the more violent the variability needed. Secondly, the reddening towards
the high-ionization BLR (including the CIV line emitting region)
may be different than that towards the low-ionization BLR (which includes
the Balmer line emitting region). Whilst this is clearly a viable
possibility (and one can imagine central-engine geometries that produce
such an effect) there is no precedent for the high-ionization BLR to be
less reddened than the low-ionization BLR. Thirdly, some fraction of the
photons from the BLR might be scattered around the material responsible for
the extinction. If the scattering fraction is wavelength independent
(e.g. electron scattering), the scattering will tend to preferentially
enhance the UV relative to the optical due to the fact that the
direct flux is heavily reddened. Since we know scattering to be an
important process in some other Seyfert nuclei, we now explore this last
possibility in more detail.
Suppose that the intrinsic UV/optical line spectrum is similar to that of
NGC 3783, with a CIV/H flux ratio of 10. We can write the
observed fluxes of both of these lines, 
, as a sum of the
direct (extinguished) flux and the scattered flux which is assumed not to
suffer any extinction beyond that due to Galactic material. If f is the
scattering fraction, then we have
where b is a parameter dependent on the extinction law used. The
standard interstellar extinction curve of Osterbrock (1989) gives b=3.2
for CIV 
and b=1.45 for H . The first term on
the right hand side of equation (8) represents the scattered flux including
the effects of extinction by Galactic material. We take 
(Berriman 1989). The second term of equation (8) gives the
contribution due to the direct (extinguished) flux. Dividing these
equations for CIV and H gives a relation
between the required scattering fraction f and the total line-of-sight
reddening E(B-V). This relationship is shown in Fig. 5 for interesting
values of E(B-V). It can be seen that scattering fraction of between
1-5 per cent (depending on the total reddening) are required in order to
make the observed line ratios consistent with an intrinsic CIV/H line ratio of 10.
Figure 5: Relationship between the required scattering fraction, f, and the
total line-of-sight reddening, E(B-V), assuming an intrinsic CIV /H flux ratio of 10.