Some LINERs are known to exhibit broad H emission (FWHM of a few thousand km s-1), reminiscent of the broad emission lines that define type 1 Seyfert nuclei (Khachikian & Weedman 1974). This subset of LINERs suffers from the least degree of ambiguity in physical origin and can be most safely regarded as representing genuine low-luminosity AGNs. These objects are analogous to the so-called intermediate Seyferts (types 1.8 and 1.9) in the terminology of Osterbrock (1981), except that their narrow-line spectra have low ionization and satisfy the definition of LINERs. The luminosities of broad H can be orders of magnitude fainter than those in classical Seyfert 1 nuclei. The well-known example of the nucleus of M81 (Peimbert & Torres-Peimbert 1981; Shuder & Osterbrock 1981; Filippenko & Sargent 1988), for example, has a broad H luminosity of only 1.8 x 1039 ergs s-1 (Ho et al. 1996b), and a number of other even less conspicuous cases have been recognized by Filippenko & Sargent (1985).
Searching for broad H emission in nearby galaxy nuclei is a nontrivial business, because it entails measurement of a (generally) weak, low-contrast, broad emission feature superposed on a complicated stellar background. Thus, the importance of careful starlight subtraction cannot be overemphasized. Moreover, even if one were able to perfectly remove the starlight, one still has to contend with deblending the H + [N II] 6548, 6583 complex. The narrow lines in this complex are often heavily blended together, and rarely do the lines have simple profiles (see Section 6.4). The strategy adopted by Ho et al. (1996f) for the Palomar survey makes use of the line profile of the [S II] 6716, 6731 doublet to model [N II] and the narrow component of H.
Some examples of the line decomposition are shown in Figure 10. As was already known from previous studies (Heckman 1980b; Blackman, Wilson, & Ward 1983; Keel 1983b; Filippenko & Sargent 1985), broad H is unmistakably present in the LINER NGC 3998 (Fig. 10a); the line has FWHM 2150 km s-1 and FWZI 5000 km s-1 (Ho et al. 1996f). Though much weaker than the component in NGC 3998, a broad component of H also seems necessary in order to adequately model the H + [N II] complex in NGC 4036 (Fig. 10b). Broad H emission has long been known to exist in NGC 4579 (Stauffer 1982b; Keel 1983b; Filippenko & Sargent 1985), but its strength is substantially weaker than that deduced by assuming that the narrow lines can be represented by single Gaussians. The extended, asymmetric bases of the [N II] lines, visible in the [S II] doublet, largely account for most of the broad wings in the H + [N II] blend (Fig. 10c). Finally, I pick NGC 4594 (the Sombrero galaxy; Fig. 10d) to illustrate the pitfalls that can potentially afflict data of insufficient S/N or spectral resolution. Judging by the similarity of its H + [N II] blend to that of NGC 4579, one might be led to believe that NGC 4594 also has broad H emission. However, careful inspection of the line profiles indicates that the [S II] lines have large widths (FWHM 500 km s-1) and extended wings (FWZI 3000 km s-1), and if one assumes that all the narrow lines have identical profiles, no broad H component is required to achieve a satisfactory fit in this object. Note that such subtleties would easily have escaped notice in previous surveys.
Figure 10. Examples of LINERs with (NGC 3998, 4036, 4579) and without (NGC 4594) broad H emission. [N II] 6548, 6583 and the narrow component of H are assumed to have the same shape as [S II] 6716, 6731, and the broad component of H is specified with a single Gaussian. Residuals of the fit are shown on the bottom of each panel.
Faint broad H emission has been discovered or confirmed for the first time in numerous nuclei. The overall statistics of the survey can be summarized as follows: of the 223 emission-line nuclei classified as LINERs, transition objects, and Seyferts, 33 (15%) definitely have broad H, and an additional 16 (7%) probably do. Questionable detections were found in another 8 objects (4%). Thus, approximately 20%-25% of all nearby AGNs, corresponding to ~ 10% of all nearby, bright (BT 12.5 mag) galaxies, can be considered ``Seyfert 1'' nuclei, if one adopts the definition that a Seyfert 1 nucleus contains a visible BLR (see Table 3). These numbers, of course, are merely lower limits, since undoubtedly there must exist AGNs with even weaker broad-line emission to which we are insensitive. The fraction of galaxies hosting Seyfert 1 nuclei, therefore, is much higher than previously thought (Weedman 1977; Huchra & Burg 1992; Maiolino & Rieke 1995). Of the 33 objects with definite detections of broad H, only 9 are well-known Seyfert 1 nuclei; the majority have substantially lower H luminosities and can truly be regarded as ``dwarf'' Seyfert 1 nuclei.
Excluding previously classified (Véron-Cetty & Véron 1993) Seyfert 1 nuclei (retaining only NGC 4395; Filippenko & Sargent 1989), the broad H line of the remaining 40 objects has a median luminosity of ~ 1.2 x 1039 ergs s-1 and FWHM = 2150 km s-1 (Ho et al. 1996f). Five of them have broad H luminosities as low as (1-3) x 1038 ergs s-1, and the probable detection in NGC 4565, if real, has a luminosity of only 8 x 1037 ergs s-1!
It is illuminating to consider the detection rate of broad H emission as a function of spectral class (Table 3). Among objects formally classified as Seyferts (according to their narrow-line spectrum), approximately 40% are Seyfert 1s. The implied ratio of Seyfert 1s to Seyfert 2s (1:1.6) has important consequences for several models concerning the evolution and small-scale geometry of AGNs (e.g., Osterbrock & Shaw 1988; Lawrence 1991), but such a discussion is beyond the scope of this paper and will be considered elsewhere. In the present context, of greatest interest is the fraction of LINERs showing broad H emission. If we first consider ``pure'' LINERs, nearly 25% of them have a BLR. The detection rate among transition objects, however, drops drastically. The cause for this dramatic change is unclear, but a likely explanation is that the broad-line component is simply too weak to be detected in the presence of substantial contamination from the H II region component. Supposing for the moment that the ratio of LINERs with and without BLRs is similar to that in Seyferts, and furthermore that the statistics of the presence of broad H in all LINERs (i.e., ``pure'' LINERs + transition objects) are intrinsically the same as those of ``pure'' LINERs, one would conclude that at least 60% of all LINERs are genuine AGNs.