4 Discussion and summary

We present here the HI observations of 79 galaxies of a statistical sample of 88 BCGs selected from the First and Second Byurakan objective prism surveys to have a HII region-like spectrum, an equivalent width of the [O III] $\lambda$5007 line larger than $\approx$ 50 Å, and a heliocentric velocity less than or equal to 6000 km s^-1. This statistical BCG sample shall be used in a subsequent paper to: 1) study the physical parameters governing the burst mode of star formation in dwarf galaxies; 2) study the chemical evolution of dwarf galaxies and 3) compare statistically the properties of BCGs with those of other types of dwarf galaxies. For comparison, we have also observed an additional 20 BCGs in the SBS zone with weaker emission lines, 47 BCGs with $V \leq 6000$ km s^-1 not in the SBS zone and 17 BCGs in the SBS zone with $V \ge 6000$ km s^-1, which are of particular astrophysical interest. For the statistical sample which contains 61 non-confused galaxies, we detected 34 BCGs with $S/N\geq 3-4$ (Table 1) and 11 with $2 \leq S/N \leq 3$ and no confusion (Table 2), giving a detection rate of $\approx$ 74%. This is to be compared with the 80% detection rate of [55, Thuan & Martin (1981)].

  

Figure 3: Histogram of the number of all non-confused observed (dashed line) and detected (solid line) BCGs versus heliocentric velocity

Figure 3 shows the histogram of all non-confused observed (dashed line) and detected (solid line) BCGs as a function of heliocentric velocity. It is seen that the detection rate is roughly constant, equal to $\approx$ 75%, out to 6000 km s^-1, so that the complete statistical sample does not suffer from any bias due to telescope sensitivity drop off.

  

Figure 4: Histogram of profile widths W50 for the 45 detected BCGs in the statistical sample (dashed line) and for the 100 detected BCGs in the enlarged sample (solid line)

Figure 4 shows the distribution of W₅₀, the velocity width at half-maximum for both the statistical sample with $V \leq 6000$ km s^-1 (a total of 45 galaxies, dashed line) and all detected BGGs (a total of 100 galaxies, solid line). The peak at $\sim$ 50 km s^-1 is due mainly to the statistical sample and may be slightly artificially enhanced as a third of the objects contributing to this peak are from Table 2 with lower signal-to-noise ratio profiles and hence with less reliable widths. Most of the statistical sample shows widths $\leq$ 160 km s^-1. The mean width is 92 $\pm$ 48 km s^-1. These relatively small widths are characteristic of dwarf galaxies. The histogram of all detected BCGs shows a larger proportion of widths $\geq$ 60 km s^-1) with a high end extending to 300 km s^-1, implying rotational velocities comparable to those of large disk galaxies. Some of these high widths are contributed by more distant ($V \geq 6000$ km s^-1) and hence more luminous and massive BCGs.

  

Figure 5: Histogram of HI masses for the 45 detected BCGs in the statistical sample (dashed line) and for the 100 detected BCGs in the enlarged sample (solid line). The dash-dotted line shows the histogram for the upper limits for the 37 undetected BCGs listed in Table 3. A Hubble constant of 75 km s-1 Mpc-1 has been adopted

Figure 5 shows the distribution of hydrogen masses for both the statistical (dashed line) and the enlarged sample of all detected BCGs (solid line). The HI masses for the statistical sample range between $4 \ 10^{7} M_{\odot}$ and $5 \ 10^{9} M_{\odot}$ with the distribution peaking at $\sim 3 \ 10^{8} M_{\odot}$.The HI mass distribution for all detected BCGs extends from $\sim$ 10⁶ $M_{\odot}$ to 10¹⁰ $M_{\odot}$, with a broad peak at $\sim$ 10⁹ $M_{\odot}$. The shifted peak toward higher HI masses is again partly due to more distant and more massive galaxies than in the statistical sample. BCGs are gas-rich objects, not a surprising conclusion, as gas is needed to fuel their very intense star-forming activity. We have also shown in dashed-dotted lines the distribution of upper limits for M(HI) derived using the data in Table 3, and assuming that the HI flux is less than 2 rms $\times$100 km s^-1. The HI mass upper limit distribution is very similar to the HI mass distribution of the statistical sample, confirming again that there is no selection effect in the latter due to telescope sensitivities.

From the notes to Tables 1 and 2 we can estimate that at least $\sim$35% of the BCGs in the statistical sample show evidence for being member of a pair, a group, or for signs of tidal interaction with another galaxy. This percentage is similar to that found by [3, Campos-Aguilar & Moles (1991)] and [4, Campos-Aguilar et al. (1993)] for other BCG samples.

Acknowledgements

The Lyon-Meudon Extragalactic Database was very useful for the confusion analysis of our data and we acknowledge the efforts of the team of people who created and manage it. S.A.P. thanks the ESO C& EE program for partial financial support through Travel Grant No. B-02-047. T.X.T. has been partially supported by NSF grant AST-9616863, a visiting professorship at the University of Paris VII and the Centre National de la Recherche Scientifique. This collaboration was made possible by NATO collaborative research grant No. 921285. T.X.T. and S.A.P. are grateful to Dr. Chantal Balkowski for her hospitality at the Département d'Astronomie Extragalactique et de Cosmologie at the Observatoire de Paris. Finally, we thank the directors of the Nançay and Green Bank Observatories for generous allotments of observing time which allowed us to carry this project to completion. We thank Claude Carignan for useful comments on the manuscript.