5 Discussion

Under the Unified Theory hypothesis, when we are looking at a Seyfert 2 we should expect to measure a high X-ray column density produced by a thick molecular torus which hides the BLR. According to the results detailed above, the column density measured in our Seyfert sample is consistent with the extinction obtained from the reddening of the lines produced in the NLR. This means that the presence of an obscuring torus is not required and, therefore, that the objects considered in our sample are somehow anomalous Seyfert 2 galaxies. A column density of $N_{\rm H}\sim 10^{22}$ cm^-2 or slightly under this value can be associated with the presence of a dust lane, a bar or an HII region. This could be the case for the objects in the present sample which in Fig. 4 populate the high X-ray luminosity - high column density region (IRAS 01428-0404, IC 1631, NGC 2992, NGC 5995, NGC 6221, NGC 6251 and IRAS 20051-1117); these have $N_{\rm H}$ values sufficient to cover the BLR. These objects could be seen perpendicularly to the torus and some other type of absorbing material obscures the BLR. To summarise, if the BLR of these galaxies is obscured, then a larger-scale dusty environment is likely to be responsible for their obscuration.

However, in Fig. 4 at lower X-ray luminosities and lower column densities we find those objects with $N_{\rm H}$very low and sometimes compatible with the Galactic absorption; i.e. a value not sufficient to obscure the BLR (IRAS 00317-2142, NGC 3147, NGC 4579, NGC 4594, NGC 4565, NGC 4698, NGC 5033, MRK 273x, NGC 7590 and NGC 7679). A dust-to-gas ratio different from the Galactic one could be a possible solution to the problem, i.e. a strong dust contribution obscures the BLR and it is not detectable in X-rays. It is worth noting that, except for MRK 273x and NGC 7679, these low $N_{\rm H}$ Seyferts are all LLAGNs. The nature of LLAGNs is a hotly debated issue, since it is not clear what powers such objects and it is not obvious that the unified model is valid at these low luminosities (Ho et al. 2001).

A few of our galaxies have very massive black holes (>10 $^{9}~M_{\odot}$for NGC 4594 and $6{-}13 \times 10^{7}~M_{\odot}$ for NGC 4579) and they accrete at extremely sub-Eddington rates ( $L/L_{\rm Edd}< 10^{-5}$, Ho 1999). It is possible that in these extreme conditions there is not sufficient ionization power to illuminate the BLR or there is insufficient gas to feed the clouds in this region or the structure of the accretion flow is uncapable of activating it (Barth 2002). Such objects would be preferentially detected in early type galaxies with massive black holes because a sub-Eddington accreting AGN would be too faint to be visible in a spiral with a low black hole mass. About 50% of our LLAGNs are in early type galaxies. There is observational evidence to support the above suggestion: several recent works have investigated the size-mass-luminosity relations in AGNs (Wandel et al. 1999; Kaspi 2000; Collin & Huré 2001) supporting the idea that there is a correlation between the luminosity of the AGN and its BLR size in the sense that the BLR size decreases as the AGN luminosity decreases.

There are also theoretical studies which lead us to believe that in intrinsically weak AGNs the BLR is extremely faint or absent. Nicastro (2000) suggests a model in which the FWHMs of the broad emission lines are related to the accretion rate, in particular at very low accretion rates the clouds of the BLR would cease to exist. Therefore for sufficiently weak AGNs no BLR is expected. This is the "pure'' Seyfert 2 model (Tran 2001).

Furthermore, we consider the hypothesis of "Fossil'' objects in which the central engine has turned off (Maiolino 2001). Shortly after this occurs, the BLR fades away and the nucleus appears as a type 2 object. If the nucleus remains quiescent for a sufficient time then even the echo of the NLR will fade away. Under this hypothesis unabsorbed Seyfert 2s are simply a transition state of an AGN. Since this phase should rapidly decay, the fraction of fossil AGNs may be very small.

The obscuring material around an active nucleus has been proven by some recent evidence to be complex and variable and the possibility that in some objects our line-of-sight might be intercepting a "hole'' in the absorber at a certain time must also be taken into account (Risaliti 2002).

To summarize, in the 17 unabsorbed Seyfert 2 galaxies the low absorption is not likely to be related to the torus but to material distributed over larger scale, very likely the NLR unless the BLR is particularly dusty. Some of these objects could be seen perpendicularly to the torus with some types of absorbing material obscuring the BLR; the material may be slightly dusty and associated with lanes, patches or starburst regions. In a few cases, the lack of evidence for the existance of a BLR suggests that this is weak, absent or has just faded away. In this case the nucleus is of low luminosity and this may explain the weakness or lack of a BLR.