5 Preliminary analysis of the results

5.1 Some statistical trends and distributions

Our resulting sample of ELGs presented in Tables 2 and 3 is not artificially biased to any specific type of galaxies. Besides the selection criterion based on the strength of [OIII] nebular lines, we do not apply any other selection. However, as we observed all the ELGs left after earlier spectroscopy by other groups it appeared that the fraction of the Case catalog faint galaxies in our part is higher. The effect of this is seen on the histograms in Fig. 2a (apparent magnitude) and Fig. 2b (absolute magnitude). Our subsample (solid line) is on average fainter by about one magnitude in comparison with the sample of all other Case ELGs observed by other groups (dashed line). In this respect the correlations shown in Fig. 3 and Fig. 4 may be somehow biased relative to the similar correlations for the whole Case ELG sample.

  

Figure 2: Histograms of apparent a) and absolute b) photographic magnitudes (taken from the Case catalog) for our ELGs from Table 2 (solid line) in comparison with the similar histograms for the rest Case ELG in the studied sky region (short-dash line). The arrows indicate the mean magnitudes for each sample

Since we are looking for selection of a sample of Case BCGs, it is interesting to examine how the detected BCGs are connected with Case emission-line codes. In Fig. 3a we show the histogram with all our observed Case galaxies, distributed versus the 5 Case catalog emission-line codes, with crude separation in each column into the galaxies of 3 different types: BCGs (cross-hatched), all other ELGs (hatched), and galaxies without detectable emission lines (white). The galaxies with inconclusive spectra are not included in this histogram. It is evident that for the emission-line codes "vs'', "s'' and "m'' BCGs dominate all other types, however they represent only 25% of the Case galaxies marked as "w?'', and only one fifth, if "w?'' inconclusive spectra are also taken into account. Altogether BCGs comprise about 70% of all Case ELGs which display emission lines in our spectra.

Since as we have noticed our Case ELG subsample is biased to the fainter magnitudes relative to the whole sample, it is interesting to check the effect of the faintest ($m_{\rm CG} = 18$ mag) galaxies on these distributions. We show in Fig. 3b the same histogram for 46 faintest galaxies of our subsample. Again Case galaxies with the code "s'' and "m'' are dominated by BCGs. Among those with the code "w'' 60% are BCGs. And those, coded with "w?'', have again a fraction of BCGs about 38%. So, in general, the fractions of BCGs relative to the total number of Case galaxies in various Case codes are quite similar for the whole sample and for the subsample of the faintest galaxies.

  

Figure 3: a) The histogram of ELG types distribution versus the spectral-line code in the Case catalog. Cross-hatched -- galaxies, classified as BCGs, hatched ones are all other ELGs, white -- galaxies without emission lines. b) The same histogram as in a), but only for the faintest Case ELGs ($m_{\rm CG} = 18-18+$ mag)

However, despite this similarity, we notice that 91% of all the faintest Case galaxies with emission lines, are classified as BCGs, which is different from 70% for the whole sample of Case galaxies with some code of presence of [OIII]$\lambda\lambda$4959, 5007 emission. So, this should result in some enhancement of BCG fraction in our subsample, in comparison with the whole Case ELG sample.

  

Figure 4: a) The equivalent widths EW([OIII]$\lambda$5007) (Å) of all Case ELGs from this work versus the spectral-line code (only the objects from Table 2). Solid line joins mean values of equivalent width for corresponding spectral-line codes. There is a clear trend, but the scattering of the data is very large. The corresponding value of Spearman rank-order correlation coefficient is k = 0.90, and the confidence level of correlation is P = 0.96. b) The same distribution as in a), but only for the faintest Case ELGs. The same trend, but the scattering is significantly reduced. The corresponding value of Spearman rank-order correlation coefficient is k = 0.999, and the confidence level of correlation is P = 0.999

From our data we can also follow the correlation of the slit-spectroscopy parameters (EW([OIII] $\lambda$5007) with those from objective-prism spectra (the code of emission line strength). As can be seen in Fig. 4a, while there is some correlation of the code and the average for this code the value of EW([OIII]$\lambda$5007), the scattering is extremely large. The brighter galaxies in each code have systematically lower EWs, which is in agreement with assigning the emission-line code in the Case catalog according to the total line flux, and not to the relative strength of the line and underlying continuum. This fact was already noticed by Salzer et al. (1995). The same histogram for the faintest Case catalog galaxies, observed by us (Fig. 4b), clearly demonstrates the same correlation, but the scattering is much reduced, for most of the (brighter) galaxies with very low EWs have gone. The corresponding correlation coefficients and confidence levels are given in the figure legend.

5.2 Sky and redshift distribution

We illustrate in Fig. 5a the sky distribution of all our observed ELGs (filled circles) in comparison with the distribution of all other observed ELGs in the same zone (empty circles). Our ELGs comprise the majority of all Case emission-line galaxies in the upper half of the lane $\delta = +29^{\circ} \div +38^{\circ}$,and the most of ELGs in this lane at $\alpha \gt 13^{\rm h}$.The hole near $\alpha = 14^{\rm h}$ is partly due to the field coverage in the Case survey, and partly due to the absence in this region of Case galaxies which have in the catalog any code of [OIII] $\lambda\lambda$4959, 5007 emission lines strength.

The redshift distribution of our ELGs is shown in Fig. 5b (solid line) in comparison with the same distribution for all other Case ELG in this zone (short-dash line). Both distributions look very similar, with abrupt decline of the number of ELGs for z > 0.04. A few ELGs are at redshifts z > 0.06. This redshift distribution looks very similar to that of BCGs in the zone of the SBS (Pustilnik et al. 1995).

  

Figure 5: a) The sky distribution of all ELGs observed in this paper (filled circles) in comparison with the distribution of all other observed ELGs in the same zone (empty circles). b) The redshift distribution of our ELGs (solid line) in comparison with the same distribution for all other Case ELGs in this zone (short-dash line)

5.3 Accuracy of data, comparison with independent measurements

To get an estimate of the external accuracy of IPCS redshifts we compared our results for common galaxies with the redshifts from CfA catalog (Huchra et al. 1995). For 18 galaxies in common we have as mean difference $\langle\Delta v_0\rangle = -27$ kms^-1 with the error 16 kms^-1 and the standard deviation $\sigma(\Delta v_0) = 64$ kms^-1. This comparison indicates that most of IPCS redshifts have an uncertainty less than 100 kms^-1.