1 Introduction

Blue Compact Galaxies (BCGs) are low-luminosity ($M_{B} \geq -18$) extragalactic objects characterized by a very high current star formation rate (0.1-1) $M_{\hbox{$\odot$}}$ yr^-1 [6, (Fanelli et al. 1988)], responsible for their optical appearance: one or several high surface brightness compact star-forming regions superimposed in the majority of cases on a more extended low surface brightness component [26, (Loose & Thuan 1985]; [20, Kunth et al. 1988]; [31, Papaderos et al. 1996)]. Their optical spectra show strong narrow emission lines on top of a stellar continuum which is rising towards the blue, similar to spectra of HII regions, and indicating the presence of a large population of massive OB stars. Much attention has been devoted to BCGs since their discovery by [41, Sargent & Searle (1970)], because of the realization that BCGs best approximate young galaxies, as implied by their low metallicity ($Z_{\hbox{$\odot$}}/50 < Z < Z_{\hbox{$\odot$}}/3$), and their very high gas content (see [54, Thuan 1991] for a review). Because the gas in the most metal-poor BCGs has not been processed through many generations of stars, it is nearly primordial. Thus such BCGs are the best objects in which to attack such problems as the determination of the helium abundance [13, (Izotov & Thuan 1998] and references therein). They are also excellent nearby laboratories for studying physical processes which occurred in the galaxy formation era, in a very metal-deficient environment. Many observational studies of BCGs have been carried out in a variety of wavelength domains, from spectrophotometric studies to derive metal abundances (e.g. [57, Thuan et al. 1995)] to UV spectral synthesis [6, (Fanelli et al. 1988)] and optical and near-infrared imaging to study stellar populations [52, (Thuan 1983)], to single-dish and interferometric 21 cm observations to examine the HI content and structure [55, (Thuan & Martin 1981]; [22, Lequeux & Viallefond 1980]; [60, Viallefond & Thuan 1983)]. In this paper, we present a large set of HI single dish observations for a new sample of BCGs assembled from the First [29, (Markarian et al. 1989)] and Second Byurakan [28, (Markarian et al. 1983)] objective prism surveys. Such observations are crucial for the understanding of BCGs for several reasons:

1) They yield accurate ($\pm$ 15 km s^-1) redshifts for the study of the large scale distribution of BCGs [36, (Pustilnik et al. 1995)].

2) The HI flux allows to determine the neutral atomic gas mass and the width of the HI profile yields an estimate of the total mass of the BCG. The gas mass fraction combined with abundance measurements are crucial for testing chemical evolution models of BCGs, such as closed-box models, models with galactic winds or infall, etc. [23, (Lequeux et al. 1979]; [35, Pilyugin 1993]; [27, Marconi et al. 1994]; [32, Peimbert et al. 1994]; [5, Carigi et al. 1995]; [37, Pustilnik et al. 1996)].

3) Combining with other data, the HI data permit to derive global parameters of BCGs (such as $M({\rm HI})/L_{B}$ or $M({\rm total})/L_{B}$, study general trends and correlations (for example $M({\rm HI})/L_{B}$ vs. L_B) suggested by various galaxy formation theories and compare with other types of galaxies. For example, one question of great interest is the relationship between BCGs and another class of dwarf galaxies, the low-surface-brightness (LSB) dwarfs. Star formation in BCGs is known to occur in bursts lasting $\leq 10^{8}$ yr, separated by long quiescent periods of $2-3 \ 10^{9}~$yr [54, (Thuan 1991)]. If LSB dwarfs are BCGs in their quiescent phases, then their HI properties should be statistically similar [53, (Thuan 1985)]. The HI data obtained here can be compared statistically with the large HI data set for LSB dwarfs assembled by [43, Schneider, Thuan and their colleagues (1990, 1992)] to test the above hypothesis. Finally, a HI single-dish survey allows to judge the feasibility of follow-up HI interferometric studies of particularly interesting objects in the sample.

The first comprehensive HI survey of BCGs was carried out by [55, Thuan & Martin (1981)]. These authors assembled a list of 115 blue compact dwarfs known at that time from the objective prism surveys of Markarian and Haro, with a few objects from Zwicky and other investigators. Other HI surveys of BCGs followed, such as that of [8, Gordon & Gottesman (1981)] which included mainly brighter BCGs (M_B < -18) from the Markarian, Haro and Zwicky lists, that of [12, Hoffman et al. (1992)] for BCGs in the Virgo cluster, and that of [45, Staveley-Smith et al. (1992)] for a small sample of nearby BCGs.

The BCG sample we are concerned with here was primarily assembled from objective prism survey plates obtained with the 1 m Schmidt Telescope at the Byurakan Observatory during the Second Byurakan Survey (SBS; [28, Markarian et al. 1983)]. The objective prism plates cover the sky area defined by $7^{\rm h}40^{\rm m} \leq \alpha \leq 17^{\rm h}20^{\rm m}$, $49^\circ \leq \delta \leq 61^\circ$, an area of about 1000 square degrees. All objects with prism spectra showing strong or moderate emission lines were observed spectroscopically with the 6m telescope of the Special Astrophysical Observatory. Then all emission-line galaxies with a HII region-like spectrum and an equivalent width of the [O III] $\lambda 5007$ emission line larger than $\approx$ 30 Å, were selected to constitute the BCG sample. The above criterion excluded massive nuclear starburst galaxies and galaxies with an active galactic nucleus such as Seyfert galaxies and quasars. In addition, we have also added $\sim$ 50 BCGs from the First Byurakan Survey [29, (Markarian et al. 1989)] and a few BCGs from the Case Survey [33, (Pesch & Sanduleak 1987)] in the same sky area and satisfying the same selection criterion. The resulting sample contains a total of 220 BCGs [14, (Izotov et al. 1993)]. The redshift distribution of the total sample along with the large-scale space distribution are given in [36, Pustilnik et al. (1995)]. The BCG sample is reasonably complete out to $\approx$ 10000 km s^-1, with increasing incompleteness beyond. The redshift distribution shows 3 peaks, one at $V \approx$ 1000 km s^-1 due to the Virgo cluster, and two additional peaks at $V \approx$ 3000 km s^-1 and $V \approx 7000-10000$ km s^-1. In order not to spend inordinately large amounts of telescope time on a single galaxy and still reach interesting upper limits for the BCG HI content with a given telescope sensitivity, we have extracted a complete subsample of 88 BCGs by further imposing a lower limit on the equivalent width of the [O III]$\lambda 5007$ emission line of 50 Å, and a velocity upper limit of 6000 km s^-1.

Of these 88 BCGs, 10 were already observed in earlier studies [55, (Thuan & Martin 1981]; [8, Gordon & Gottesman 1981)]. We have obtained new HI observations for 77 BCGs, and obtained measurements with better signal-to-noise ratio of two BCGs with published data. One BCG which is close to a HI rich galaxy with the same velocity was not observed.

This complete subsample will be used for statistical studies in a subsequent paper. For comparison, we have observed in addition 20 BCGs in the SBS zone with $V \leq 6000$ km s^-1 but with less strong emission lines, 47 BCGs not in the SBS zone with the same velocity limit, and 17 BCGs in the SBS zone with $6000 \le V \leq 14000$ km s^-1, which are of particular astrophysical interest. Finally 8 more BCGs outside the SBS zone with $V \ge 6000$ km s^-1 were observed. Altogether we have obtained HI parameters or upper limits for 171 BCGs.

We discuss the HI observations and the data reduction in Sect. 2. In Sect. 3 we describe the data tables and present the observed HI profiles. We give a preliminary discussion of the data in Sect. 4. A more complete discussion is deferred to a subsequent paper.