5 Discussion

5.1 Survey efficiency and fraction of BCGs

The principles, methods, selection procedures and first follow-up spectroscopy results of the Hamburg/SAO survey described in the previous sections are intended to create a new large and deep sample of BCGs. Summarizing the data and their preliminary analysis presented in Sects. 3 and 4 we conclude that the selection technique used in this paper is efficient in search for objects with strong and moderate [OIII]$\lambda\lambda$4959, 5007 Å emission lines. However, the overall detection efficiency of emission-line objects is rather low for this first sample (about 31%). In Sect. 5.3 we discuss additional selection criteria which can improve the ELGs detection efficiency.

The fraction of BCGs (as our main goal) among all emission-line objects is $\sim$63% (47 BCGs among 74 in total). This is significantly larger than the fraction of all HII-type objects among ELGs in the UCM (33%) (Gallego et al. 1997) and larger than $\sim$50% in the UM sample (DHIIH, HIIH and SS-objects in S89a).

The preliminary estimate of the mean surface number density of BCGs found in the HSS is about 0.21 BCG/square degree, accounting all HII-type objects selected as first and second priority candidates in the band centered at $\delta = +47.5\hbox{$^\circ$}$ with the range of $\alpha = 12^{\rm h} - 17^{\rm h}$.The estimate incorporates 38 BCGs in the indicated region from this paper and 21 objects obtained independently from the HSS, which either are known as HII-galaxies from earlier publications (Vogel et al. 1993; Popescu et al. 1996; Schneider et al. 1994), or have been observed later by us and will be presented in forthcoming papers, or are among several of our unpublished BCGs in and near the SBS zone. This estimate is still a lower limit since not all first priority HSS candidates in this region are observed. This value is similar to that for the BCG sample in the zone of the SBS survey (0.24 BCG/square degree; Thuan et al. 1994) and for the sample of BCGs from the region 3 of Popescu et al. (1996).

Due to our selection criteria cut-off of luminous galaxies, we reduce significantly the fraction of Sy galaxies in comparison with the above mentioned surveys (< 2% here versus 10% in the UM sample and 8% in the UCM sample) and pick up only the low-luminosity tail of AGN (possible LINERs) mainly as interlopers and second priority candidates. The detection of a significant fraction of possible LINERs (6 galaxies or 8%) is surprising, and has no clear explanation. The remaining discovered ELGs are probable SBN and DANS (22%).

It is interesting to note that due to the limit in redshift, the faintest ELG candidates are mainly also the least luminous, and thus they are expected to be typical BCGs. This is reflected by the statistics of all detected ELGs. For the ranges of $m_B = 16^{\rm m} - 17^{\rm m}$, $17^{\rm m} - 19^{\rm m}$ and $19^{\rm m} - 20^{\rm m}$the fraction of BCGs is about 57%, 67% and 80%, respectively. This was also noticed by Ugryumov et al. (1998) in the study of Case ELG candidates. It gives us an additional impetus to search for emission-line candidates down to the very limits of the plates.

5.2 Aspects of spectral classification, chemistry and metallicity

We already discussed in Sect. 4.1.1, that the ELGs are classified according to their positions in the line ratio diagnostic diagrams.

A comparison of the positions of the new HSS ELGs in these diagrams shows that our classification appears to be self-consistent for all BCGs, and, therefore, is rather reliable.

It is interesting to compare the positions in diagnostic diagram of the new HSS BCGs with those of the well studied BCGs from the SBS sample (Izotov et al. 1994, 1997b; Thuan et al. 1995). Most of the SBS BCGs populate the region with log([NII]$\lambda\,6583$ Å/H$\alpha$) < -1.2 (Fig. 4). About 20% of the HSS BCGs are located in the same region. The real fraction of such BCGs with high excitation HII-regions will be quantified after better quality spectroscopic observations. At least 2 HSS BCGs with high excitation HII-regions are present in the SBS list ($\delta = 49\hbox{$^\circ$}- 50\hbox{$^\circ$}$) and they are independently rediscovered in the present study. In particular, our line flux ratios for HS 1249+4919 (log(O/H)+12 = 7.72) are very close to those from Thuan et al. (1995).

 

Figure 6:   Sky projection of the BCGs from the HSS sample (o) together with BCGs from the zones of SBS (x) and Case surveys (+). The dashed box delineates one of the regions of the Heidelberg void survey (Popescu et al. 1996, 1997)

 

Figure 7:   Histograms of apparent a) and absolute b) photographic magnitudes for three BCGs samples: new HSS BCGs from this paper, BCGs from the SBS sample (Pustilnik et al. 1995) and Case BCGs from Ugryumov et al. (1998), Ugryumov (1997). The arrows indicate the mean magnitude for each sample

 

Figure 8:   Histograms of redshift distributions for the same BCG samples, as for the previous figure

The region with log([NII]$\lambda\,6583$ Å/H$\alpha$) < -2 contains the BCGs with the lowest metallicities. We discover two new HSS BCGs located in this region. Hence low metallicity objects can be selected by our search method.

5.3 Improvement of selection criteria

Having quite low efficiency at this pilot stage it is natural to try to improve the selection criteria. The analysis of all accumulated data discussed here, shows that in order to improve the selection of BCGs the following additional criteria and selection procedures should be applied to the preselected lists of candidates:

1.

Second priority candidates have very low detection rates for BCGs (< 20%) and pick up mainly ELGs of other types which are not in the focus of our project. So, these objects should not be observed further except as backup sample. This will result in an underestimate of the BCG surface density of  20%.

2.

It appears that the majority of bright (m_B < $18\hbox{$.\!\!^{\rm m}$}0$) stellar-like objects are either blue or M-dwarf stars. Thus, additional examination of candidates on direct images in order to discriminate between stellar and fuzzy images will allow to remove most of the obvious brighter star-type interlopers.

3.

The results of follow-up spectroscopy have shown that even part of the prominent first priority candidates turned out non-emission objects. Subsequent check of the plate material have shown that some defects or strong noise peaks looked like strong emission features on objective prism spectra. Thus, we concluded that it is necessary to make a careful check of all first priority candidates on the original spectral plates (scanning of 2 plates of the same region with cross-check -- in the majority of cases) or additional visual inspection of the candidate spectra -- in case of only one plate -- to remove possible dust grains and noise hits.

These improvements allow to increase the discovery rate of BCGs by a factor of $\sim$2, as it is shown in a forthcoming paper by Pustilnik et al. (1998).

5.4 The statistical properties of HSS BCGs

The sky distribution of the BCGs from the HSS sample is shown in Fig. 6 together with BCGs from the SBS and Case surveys (Izotov et al. 1993b; Pustilnik et al. 1995; Ugryumov 1997; Ugryumov et al. 1998). Our aim is to create a large sample (the Northern BCG sample) in this whole area of about 3000 square degrees. For this it is important to understand how the properties of the objects in these three samples compare to each other. First, we compare the apparent magnitude distributions of the three samples. For the HSS BCG sample we use the data of the 47 BCGs found in this first part of the survey. We consider it as representative (excluding the brighter part of the sample, which will be discussed later). An abrupt decrease of the number of objects in the magnitude range between $m_B = 18\hbox{$.\!\!^{\rm m}$}0$ and $19\hbox{$.\!\!^{\rm m}$}0$ is present in the distributions of all three samples (Fig. 7) although the fraction of galaxies in this magnitude range is different in each sample. As it was shown by Pustilnik et al. (1995) the BCG sample in the SBS zone can be considered as rather complete up to $m_{\rm pg} = 18\hbox{$.\!\!^{\rm m}$}0$.The similar behaviour of the faint end of the apparent magnitude distributions infers that the Northern BCG sample may have a similar completeness down to this limiting magnitude. It is evident from Fig. 7 that there is a significant deficit in the HSS of galaxies with $m_B \le 16\hbox{$.\!\!^{\rm m}$}0$ caused by the adopted limits in the parameter space of the selection criteria. Additionally, several known bright emission-line galaxies are found in the HSS, but not observed in this work. These selection constraints lead to a difference between the mean apparent magnitudes of HSS BCGs and the BCGs of the two other samples. The real brightness distribution of HSS BCGs will be discussed in more detail in later papers.

Despite of the differences in the apparent magnitude distributions, the absolute photographic magnitude distributions for the same BCG samples shown in the histograms of Fig. 7 look much more similar and have similar mean luminosities. This may indicate that the contents of all three BCG samples are the same.

The radial velocity distributions shown for the same BCG samples in Fig. 8 have similar break values of the redshift (about 0.035), at which a significant fall-off takes place. This again reflects the similar limiting apparent magnitudes of all three samples. The absence of nearby HSS objects in this histogram is artificial, because the bright BCGs with known velocities were not observed here. For the sake of clarity it should be noted that galaxies were selected in SBS both through UV-excess and emission lines. The BCG sample however was exclusively selected from emission-line candidates. Therefore the redshift distribution of the SBS is similar to the other two samples. The similarities of BCG properties in the HSS, SBS and Case surveys imply that the galaxies from these surveys can be combined into a larger sample to study the spatial distribution of BCGs in this area. A first comparison of the luminosity function which can be derived from these different samples can be found in Hopp (1998).