4 Emission

Emission lines which are present in the wavelength region covered by our spectra comprise H${\beta}$ (4861 Å), [O  I] (6300 Å, 6363 Å), [O  III] (4959 Å, 5007 Å), [N  I] (5199 Å), [N  II] (5755 Å, 6548 Å, 6583 Å), H$_{\alpha}$(6563 Å), Fe  II (5270 Å), [Fe  VII] (5721 Å), He I (5876 Å) & He  II (4686 Å), among others. Line-strength indices affected by these emission lines are H${\beta}$, Fe5015, Fe5270, Fe5709, Mg₁, Mg₂, Mgb & NaD. In the case of H${\beta}$and Fe5270 the emission lines are situated right on top of the line itself. The [N  I] emission line is found in the right continuum window of Mgb (see Goudfrooij & Emsellem 1996 for a more detailed discussion). In the case of Fe5015 the two [O  III] lines at 4959 Å and 5007 Å are found in both the left continuum and the line window. In all the other cases the emission lines are found very close to the edge of a window or just in the very narrow gap between the line window itself and the continuum window.

In all these cases the "true'' value of the index may be influenced.

The strongest emission feature which can be identified unmistakably is the [O  III] 5007 Å emission line, which falls on top of the Fe5015 index. The [O  III] 4959 Å feature is found to be weaker in all spectra, consistent with theoretical expectations (Osterbrock 1989), often falling below the detection limit. Often connected to [O  III] 5007 Å is the emission in H${\beta}$, although line ratios can vary (Osterbrock 1989).

The strength of [O  III] 5007 Å was used to subdivide the sample into three classes. Galaxies of class "0'' show no emission with EW[O  III] < 0.3 Å, in class "1'' the objects show weak/possible emission with 0.3 < EW[O  III] < 0.5 and in class "2'' galaxies have the strongest emission lines, with EW[O  III] >0.5. In general, for the intermediate class 1, only [O  III] 5007 Å is detected, different from emission class 2, where several other emission lines are also detected. Typical examples are given Fig. 8. There is no evidence for emission lines in the spectrum of NGC 2986 which is therefore called a class "0'' galaxy. The spectra of NGC 1549 and NGC 1316 contain instead emission lines. However, in NGC 1549 only a weak evidence for emission is found, located in the Fe5015 feature in the form of the [O  III] $\lambda = 5007$ Å line. This emission can not be verified at the positions of the other emission lines. Therefore the galaxy is classified as class "1''. A clear cut case for a galaxy with emission is NGC 1316. The emission is not only clearly visible in the [O  III] $\lambda = 5007$ Å line, but also at $\lambda = 4959$ Å ([O  III]), $\lambda = 4861$ Å (in the line center of H${\beta}$) and around $\lambda = 6563$ Å (H$_{\alpha}$). Therefore the galaxy belongs to class "2''.

Figure 8: Three examples of spectra with good signal-to-noise ratios plotted against the natural logarithm of the wavelength. Left panels: the full spectra transformed to zero redshift are shown. The flux is given in arbitrary units. Overlaid in light grey are the windows of the Lick/IDS indices measured in this work. Right panels: zoom of the spectra around the Fe5015 feature. In the upper left corner the emission class is indicated. Overlaid in light gray are the positions of emission lines. The dotted spectrum in the lower two panels is of NGC 2986, rescaled to match the spectra of NGC 1549 and NGC 1316.

Out of the 148 galaxies observed in this work, 56 objects (38 per cent) fall into class "0'', 45 objects (30 per cent) in class "1'' and 47 objects (32 per cent) in class "2''. Altogether about 62 per cent of the galaxies show signs for some degree of emission. This result is consistent with the findings of Phillips et al. (1986), Caldwell (1984) and Goudfrooij et al. (1994), but it is lower compared to the result of Macchetto et al. (1996), who find that $\approx$79 per cent of their sample galaxies have emission lines. This discrepancy may be explained by the fact that Macchetto et al. searched for line emission via narrow band imaging over the whole projected area of the galaxy while our data only refer to the galaxy centers.

We decided not to correct the line-strength indices for emission as the ratios between the various emission lines can vary significantly. Therefore, even a scaling of all emission lines, say, with the EW of [OIII], and subtracting a typical emission line pattern will not allow a consistent correction of all indices. However, the effect of emission is generally small. For the class 0 objects, the correction for emission is typically much smaller than the error bar. For class 1 objects, it is of the order of the error bar. Among the important indices, the H${\beta}$ index is most affected by emission. The correction can be estimated from González (1993) who gives $EW({\rm H}\beta) \approx 0.7\ EW([{\rm OIII}]5007~{\rm\AA})$, i.e. the correction is typically of the order 0.1 Å for class 0, 0.3 Å for class 1 and >0.4 Å for class 2 objects.