Abstract

Virtually all accretion disk models predict that QSOs observed from nearly edge-on should show extremely high equivalent-width emission lines. These are not seen. Either accretion disks must be significantly non-planar, or most edge-on QSOs must be concealed by an obscuring torus.

Model

If the UV-optical continuum emission of QSOs comes from an accretion disk, it will be emitted anisotropically. If in addition the line radiation is either isotropic, or anisotropic in a different way from the continuum radiation, then identical QSOs observed from different orientations will show different emission-line equivalent widths.

I assume that all QSOs have the same intrinsic line-to-continuum flux ratio, and that the line radiation is isotropic. Any magnitude-limited sample is strongly biased towards face-on QSOs, and this bias is taken into account using luminosity function information. A wide variety of both thick and thin disk models have been used.

Results

A typical predicted equivalent-width distribution is compared with an observed distribution in the figure. Both are taken from. Two discrepancies are evident. Firstly, the observed distribution has a broader, smoother peak than the prediction. This can easily be explained if there is an intrinsic dispersion in QSO equivalent-widths. Secondly, the model has a tail of very high equivalent-width QSOs not seen in the observations. This tail is significant at the 99% confidence level, for most accretion disk models, and for Lyman-α, C III] and Mg II as well as C IV.