Based on the most recent QSO ultraviolet spectra, the covering factor of th
e clouds of the Broad Line Region (BLR) is about 30%, or larger. This value
would imply that in at least 30% of the QSOs our line of sight crosses one
, or more, BLR clouds and, in the latter case, the UV spectrum should show
a sharp Ly-edge in absorption. This Ly-edge in absorption is never observed
. This paradox is solved if, as suggested by various authors, the BLR is fl
attened and the dusty gas in the outer parts, on the same plane, prevents t
he observation along the lines of sight passing through the BLR clouds. The
objects observed edge-on (with respect to the flattened BLR) would be clas
sified as obscured QSOs or, within the framework of the unified model, type
2 QSOs. The covering factor of the BLR constrains the fraction of obscured
QSOs to be QSO2/QSO1 > 0.5. This lower limit is already high with respect
to the number of candidate type 2 QSOs claimed so far. We discuss this cons
traint in relation to recent AGN surveys.