6 The properties of the sample

In this section we present some statistics on the relevant spectroscopic parameters of the sample as well as of subsamples of different morphological subtypes.

Our sample was selected in certain fields which contain huge underdensities. One concern is that its galaxy content is thus different from the galaxy content of samples selected from the general field environment. In passing we should note that Popescu et al. (1997) found that most of our sample galaxies belong to the normal field features of the large scale structure. Nevertheless, we make an extra check, by comparing the frequency distribution of the Salzer et al. (1989b) types (see Tables 3, 5, and 6) with those from the University of Michigan (UM) survey, which is a general field survey. The comparison is given in Table 10 and indicates little significant differences in the type distribution of the two samples. If at all, we have some more DHIIH galaxies and fewer SBN objects in our sample. This is to be expected due to the selection criteria we adopted, namely bright objects were excluded from the survey (Popescu et al. 1996). The motivation was that bright objects were already included in other catalogues and we were mainly interested in dwarf galaxies. Since the frequency distribution of different morphological types is very close to that of the UM survey, we also expect a correlation between the Salzer et al. (1989) morphological types of our objects and their absolute blue magnitude. We verified that this correlation exists, but with a reasonable scatter in each sub-class. This scatter (peak-to peak) can be as large as 4 mag in most of the type bins. For a detailed discussion of the sample properties as a function of the galaxy density see Popescu et al. (1999).

The histogram of the H${\beta }$ luminosities of the whole sample is given in Fig. 3a. The luminosities were calculated for a Hubble constant H₀ = 75 km s^-1 Mpc and they were corrected for internal extinction. The distribution is asymmetric, with a sharp cut-off towards the brighter end. This is due to the selection criteria we adopted, as mentioned above (see also Popescu et al. 1996). The incompleteness at the bright end is also obvious when compared with the H${\beta }$ distribution of other similar samples from the literature. For example, in Fig. 3d we show, for comparison, the H${\beta }$ luminosity distribution of a sample of 99 ELGs (Salzer et al. 1989b) from the UM objective-prism survey and in Fig. 3e the same distribution from the ELG sample of Terlevich et al. (1991). In comparison with our distribution, the latter ones are more symmetrical and span over more orders of magnitude. They include at the bright end also Seyfert galaxies, which were not included in our study. The H${\beta }$ luminosity distributions of different morphological subtypes are shown in Fig. 3b (for SS and DHIIH galaxies) and in Fig. 3c (DANS and HIIH), respectively. The differences between them show the known trend of increasing H${\beta }$ luminosities from the SS to DANS class, though a significant overlap exist, too.

The expected correlation between the H${\beta }$ luminosities and the blue magnitudes is shown in Fig. 3f, where different morphological subtypes are plotted with different symbols. A linear least square fit to the log(L(H${\beta }$))=f(M_B) data (plotted with solid line) gives a slope of -0.388 and a correlation coefficient of 0.85. This is close to the the slope (-0.391) of a similar correlation found by Salzer et al. (1989b) for the ELGs of the Michigan Survey. The slope of -0.391 was derived when the Seyfert and SS galaxies were excluded from the correlation. This is consistent with our set of data, which does not contain Seyfert galaxies. Also the SS galaxies of our sample seem to follow in a better way the correlation, while those found by Salzer are lying mainly in the upper part of the correlation. As remarked by Salzer, this slope suggests that, within the uncertainties, the H${\beta }$ luminosity scales directly with the blue luminosity: L(H ${\beta})\propto L_B$. This would indicate that the recent star-formation (in the last $\sim10$ Myr) and the integrated star formation are related. On the other hand the scatter from the correlation suggests variations in the global equivalent widths, W(H${\beta }$), which means different stages of activity. If the total blue magnitude is a good measure of the average past star formation, then the scatter indicates an intrinsic variation in the ratio between the present star formation rate and the average past.

The distribution of the H${\beta }$ equivalent widths is also given in Fig. 4.

Figure 4: The distribution of the H${\beta }$ equivalent widths (in units of Å)