4 Discussion

Unified models have recently received much emphasis in AGN research (see e.g. Antonucci 1993). These models postulate that the distinction between broad and narrow lined AGN is simply due to our viewing angle. All Seyferts have the same basic structure, but in Seyfert 2s the plane of a geometrically and optically thick dusty molecular torus lies close to our line of sight and it blocks our direct view of the nuclear source and the BLR. On the other hand, in Seyfert 1s we look along the axis of the obscuring torus, and directly see the nucleus.

Briefly, the evidence in favour of the unified models includes the detection in a growing number of Seyfert 2s of: 1) polarised broad emission lines (e.g. Antonucci & Miller 1985; Miller & Goodrich 1990; Tran 1995b; Young et al.1996), interpreted as scattered BLR emission by warm material above the torus (e.g. Krolik & Begelman 1988; Pier & Krolik 1992); 2) broad NIR lines, revealing highly obscured BLRs (e.g. Blanco et al. 1990; Goodrich et al. 1994; Ruiz et al.1994; Veilleux et al.1997); 3) the biconical geometry of the high excitation gas (e.g. Pogge 1989; Tadhunter & Tsvetanov 1989; Haniff et al. 1991; Wilson & Tsvetanov 1994; Mulchaey et al.1996a), indicating that the ionizing nuclear radiation escapes anisotropically along the torus axis; 4) large X-ray absorbing column densities and a strong Fe K$\alpha$ emission line (e.g. Awaki et al.1990; Matt et al.1996; Malaguti et al.1998); and 5) a deficit of directly observed ionizing photons compared to that seen by the ENLR (e.g. Wilson et al. 1988; Kinney et al.1991; Binette et al. 1993). The ENLR (cone) axis, the radio source (torus) axis and the optical nuclear continuum axis are usually significantly aligned with each other, while there is no correlation with the host galaxy axis at larger scales (e.g. Pogge & De Robertis 1993; Wilson & Tsvetanov 1994; Mulchaey & Wilson 1995; KW97; this paper).

The colour maps presented in this paper and in KW97 offer a new, independent method to test the unified models of AGN. What is the origin of the blue elongated or double structures visible in the colour maps of the circumnuclear regions of several Seyfert 2 galaxies? In Mkn 3, Mkn 573 and NGC 1068 (KW97), and Mkn 533, Mkn 607, Mkn 1066, NGC 5347, NGC 5929, NGC 5953 and NGC 7319 (this paper), these structures are closely parallel to the radio and ENLR emission (to within 20$^\circ$). In the case of a stellar bar, we would expect red continuum colours from an old stellar population, clearly not seen. Optical synchrotron emission from the radio jets is also unlikely, because there is no detailed correspondence between the blue maxima and the radio structure, and the blue continuum is more extended than the radio continuum. Alternatively, the blue structures may be due to an intrinsically extended nonstellar continuum, e.g. emission from high velocity shock waves generated from the interaction of a radio jet with the ENLR gas (e.g. Sutherland et al. 1993). Such an extended component has been proposed to explain the constant H$\beta$ EW over five decades of optical-UV continuum luminosity (Binette et al.1993), and the larger polarisation of broad lines than continuum in many Seyfert 2s (Tran 1995b). However, a very close morphological correlation between the continuum and the high-velocity ionised gas is expected, but not seen in the colour maps.

Can any of the blue features be due to star formation? There is now increasing evidence for circumnuclear starbursts in many Seyfert 2 galaxies, e.g. strong far-IR and CO emission from cool dust (Heckman et al.1989), strong extended mid-IR emission and spectral features from warm dust (Maiolino et al.1995), and large NIR light-to-mass ratios (Oliva et al.1995). Also, strong optical CaII triplet 8600 Å absorption with respect to weak MgI 5100 Å absorption (Cid Fernandes & Terlevich 1995), and UV spectral properties (Heckman et al.1995) indicate that hot massive stars in a dusty metal-rich starburst can make a significant contribution to the nuclear optical/UV energetics of Seyfert 2s. In most Seyferts studied here, the blue elongations and double features are closely aligned with the linear radio structure and the emission line morphology. Although this could indicate that they are intrinsically blue regions of (jet-induced) star formation, we consider it unlikely because there is no direct evidence for star forming regions in the continuum images, and because the morphology of the regions is diffuse, unlike the sharp boundaries and knotty morphology usually seen in star forming regions. Furthermore, excitation maps indicate that the circumnuclear gas in Seyferts is mainly of high excitation, and does not originate from star formation (Mulchaey et al.1996a). Finally, the detected polarised broad lines, heavily absorbed X-ray sources, strong Fe K$\alpha$ lines and ionisation cones in an increasing number of Seyfert 2s argue for a hidden Seyfert 1 in them.

Therefore, the remaining and most attractive explanation for the blue features spatially coincident with the high-excitation circumnuclear gas is scattering of the nuclear light along the radio axis by dust or electrons (see also Pogge & De Robertis 1993; KW97). In either case the scattered continuum appears blue, but more so for dust scattering because of the wavelength dependence. The reason for the alignment between the continuum and the line emission in this case is that both the ionising and optical photons escape along the torus polar axis, and the optical continuum is scattered by dust and/or electron "mirrors'' associated with the ionised gas. For most of the sample galaxies, the scattering is likely due to electrons, because of the shape of the polarised flux spectrum, the correlation of the H$\beta$ luminosity with soft X-rays, and the strong Fe K$\alpha$ emission and lack of low-energy X-ray absorption (Tran 1995b). However, dust is strongly favoured for Mkn 533, NGC 1068 and NGC 7212 (Tran 1995b; see also KW97). Note that the blue structures in Mkn 1, NGC 788 and NGC 7212 are only weakly aligned with the radio and ENLR axes (to within 20-40$^\circ$), and their relation with the scattering of the nuclear light remains uncertain.

Finally, in the context of unified models, we might expect to detect red regions across the nuclei associated with the dusty torus and perpendicular to the radio/ENLR axes (cf. red nuclear structures in NGC 5252 and Mkn 348; Kotilainen & Prieto 1995; Simpson et al.1996). Such structures are clearly not seen in this sample. This can be understood if the true scale of the obscuring material is much smaller than our spatial resolution (as expected in the unified model; Mulchaey et al.1996a). In that case, regions of low reddening will be included, and the nuclear reddening will be severely underestimated. Thus, while nuclear reddening may be important in Seyferts, it remains undetected because of the limited spatial resolution, while the larger scale dust lane features become visible, as in many objects of this sample. Indeed, if we were to detect reddening by dusty tori in the colour maps, these features would correspond to hundreds of pc of scale, much larger than expected in unified models.