Many galaxies that are not AGNs have strong emission lines in their spectra; only about 10% of the Markarian galaxies are Seyferts and the remaining 90% are nearly all HII region galaxies. A significant fraction of the latter have quite strong [O III] 4959, 5007, as in Seyfert 2 galaxies. To segregate the AGNs from the H II region galaxies is one of the main goals of emission-line galaxy spectral classification. A straightforward way to do this is by using numerical criteria based on ratios of emission line strength, as first proposed by Baldwin, Phillips and Terlevich (1981). They explored many different pairs of ratios; Veron (1981), at about the same time, used [O III] 5007 / H vs [N II] 6583 / H, which has the advantage that both ratios are nearly independent of the exact reddening, and of a highly accurate calibration of the sensitivity as a function of wavelength. Keel (1983b) used the three ratios [N II] 6583 / H, are [S II] (6716 + 6731) / H and [O I] 6300 / H for the same reason. Veilleux and Osterbrock (1987) obtained, collected and analyzed a large body of data involving all three of these ratios as functions of [O III] 5007 / H.
Figure 4. [O III] / H vs [O I] / H line ratios for HII region galaxies (open circles), AGNs (filled circles), and unclassified objects (asterisks). The half filled circles are objects for which the data or published classification is in doubt. Dashed line is the sequence of AGN photoionization models described in the text; solid line is adopted division between H II region galaxies and AGNs.
The results are shown in graphical form in Figures 4, 5, and 6. In all three diagrams to the solid and open circles are the previously classified AGNs and HII regions respectively. It can be seen that the two types of objects are well segregated in these diagrams. (The few objects marked by half-filled circles are cases in which there is some question about the data or the prior classification.) The [O I] / H and [S II] / [O I] vs [O III] / H segregations shown in Figures 4 and 5, respectively, can be understood directly in terms of the differences between the photoionizing spectra of O stars and of AGNs, while the [N II] / H vs [O III] / H segregation is less straightforward but is well reproduced by the models. The solid line in all three diagrams shows the best adopted division between HII region galaxies and AGNs. The galaxies not previously classified, indicated by stars, were put into one class or the other on the basis of these diagrams. For nearly all of these galaxies, all the ratios measured give the same classification. The [O III] / H vs [O I] / H diagram (Figure 4) is the best single criterion for separating AGNs from HII region galaxies, but because [O I] 6300 is weak even in AGNs, it requires the best to signal-to-noise ratio on the spectra. If possible it is best to use all three ratios; if they do not all agree the average classification should not be adopted, but instead the object should be regarded as peculiar and deserving further study.
Figure 5. [O III] / H vs [S II) / H line ratios for H II regions and AGNs. Symbols and curves as in figure 4.
Figure 6. [O III] / H vs [N II] / H line ratios for H II regions and AGNs. Symbols and curves as in figure 4.
The dashed lines in all three diagrams are from a series of AGN photoionization models calculated by Binette (1985) for a power-law input spectrum, of the form F -n, with n = 2. The ionization parameter decreases along the series, from = 10-2.2 at the top end of each of the dashed lines to = 10-8 at the bottom end. The abundances of the elements are solar, and the mean electron density is Ne 102.5 cm-3.