1 Introduction

Blue compact galaxies (BCGs) with strong emission lines or H II-galaxies are low-mass objects with the current star-formation (SF) rate exceeding many times this parameter averaged over cosmological time. The observed SF rates for most BCGs are so high that star formation could be sustained at the current level only on a timescale significantly lower than 1/$H_{\rm0}$ before their neutral gas reservoir will be completely depleted (e.g., Thuan 1991). This implies that either these objects are rather young, with typical ages of less than one or a few Gyr, or SF in BCGs is highly variable and proceeds in short episodes (bursts) with a typical duration of 10-100 Myr, intermittent with long periods of SF activity at levels much lower than that during the SF burst. In the course of such SF bursts, BCGs experience significant brightening relative to their quiescent state - an average of 0 $.\!\!^{\rm m}$75 in the small BCG sample from Papaderos et al. (1996), and up-to ${\sim}2$ $.\!\!^{\rm m}$0 in a large sample from Lipovetsky et al. (2001). Therefore being in general low-mass and underluminous systems with a wide range of baryon (stars+gas) mass (10 $^{\rm 7} < M_{\rm bar} < 10^{\rm 10}~M_{\odot}$), BCGs sometimes can be as bright as M_B=-20 $.\!\!^{\rm m}$0.

In the great majority of properly studied BCGs, the traces of old low-mass populations are detected (e.g., Thuan 1983). This implies that, in general, BCGs are rather old objects, and their evolution on the Hubble time scale can effectively be determined by short, but very intensive SF, bursts. The option of truly young galaxies is still probable for very small fraction of BCGs (${\sim}1\%$) with extremely low metallicities (e.g., Izotov & Thuan 1999; Kniazev et al. 2000; Pustilnik et al. 2001a).

While the mechanisms that regulate and terminate SF bursts on the "short'' time scales are more or less known and partly understood (SNe and stellar winds), the nature of the mechanisms which trigger the collapse of gas clouds and the onset of SF bursts in low-mass galaxies has been debated since the early 70-s.

A general approach to the question should take into account that even though BCGs do not comprise a homogeneous class of objects (since they possess a wide range of observed morphologies), to a first approximation they all are gas-rich galaxies with high enough specific angular momentum, and therefore more or less flattened rotating "disks" (e.g. Taylor et al. 1993,1995; Chengalur et al. 1995; van Zee et al. 1998). In the least massive BCGs, the rotation energy is not much greater than that of random motion, and their "disks" are thicker. On the other hand the BCGs similar to Mkn 996 that is, with the light distribution typical of elliptical galaxies are quite rare (Doublier et al. 1997). The latter is probably indicative of their post-merger nature (Thuan et al. 1996).

The concept of equilibrium and gravitational instability in a rotating disk is based on the local stability criterion, introduced by Toomre (1964). There is a threshold surface density (depending on the local epicyclic frequency and the dispersion of the random velocity component, Toomre 1964; Kennicutt 1989), above which gas is unstable. Positions of SF regions in disk galaxies, examined on a variety of galaxy types (Sa-Sm, Im, LSB) are consistent with the criterion of Toomre-Kennicutt (Kennicutt 1989; van der Hulst et al. 1993). Positions of SF regions in BCG/H II galaxies also fulfill this criterion (e.g., Taylor et al. 1994; van Zee et al. 1998).

On the other hand, most of the observed cases of enhanced SF in massive gas-rich galaxies are associated with various interactions from other galaxies. Keel (1993), from the kinematical study of a well-selected sample of interacting galaxies has drawn the important conclusion that gravitational instability in their disks is driven initially by external perturbations (see also Bernloehr 1993).

At a first glance this mechanism could play an important role for BCGs as well. However, the studies of the spatial distribution of BCGs indicate that only a minority are tightly connected to massive galaxies or their systems (e.g., Salzer 1989; Campos-Aguilar & Moles 1991; Vilchez 1993; Pustilnik et al. 1995).

This fact does not imply that interactions are of low significance as SF triggers in BCGs. However many researchers seem to favor some internal process as the main trigger of enhanced SF. In particular, the mechanism suggested by Gerola et al. (1980), which is based on the stochastic collisions and merging of large gas clouds is noted. Another type of internal trigger is connected with the hypothesis of cyclic gas re-processing, with the characteristic time between the subsequent SF bursts related to the fall-off back to the "disk" of the gas lifted to the halo during the previous SF burst (Salzer & Norton 1999).

The two alternative models have been suggested to overcome the apparent difficulties of the hypothesis in explaining the interaction-induced SF burst in BCGs, as evidenced by their weak spatial connection to massive galaxies. The first, mentioned by Melnick (1987) and clearly formulated by Brinks (1990), incorporates the tidal action from nearby low-mass galaxies. It was tested in the VLA search for HI-rich companions of 19 nearby (V < 2500 km s$^{\rm -1}$) "isolated'' H II-galaxies (Taylor et al. 1995; see also Taylor 1997). Surprising, 10 of the 19 target galaxies revealed low-mass HI companions, most of which were identified later with faint galaxies. However, the statistics are still rather sparse. Some additional indications of the potentially important role of low-mass galaxies appeared after the deep CCD imaging of the large BCG sample by Lipovetsky et al. (2001).

Another model of an external trigger for SF bursts in BCGs was suggested by Östlin et al. (1999) and Kunth & Östlin (2000). Based on morphology and gas kinematics data for the small sample of luminous BCGs, they consider that mergers of low-mass galaxies are the most important factor affecting the group of BCGs. Similar results were reported by Sung (2000), based on studies of a sample of 115 BCG/H II-galaxies.

Thus, during the last 5-7 years, an increasing amount of evidence has appeared which imply the importance of gravitational interaction of low-mass galaxies with current SF bursts, in particular with galaxies of comparable or lower mass. At the same time, neither new simulation has demonstrated the potential of an internal spontaneous onset of SF bursts in gas-rich galaxies, nor do any concrete models indicate the physical conditions which are necessary for this mechanism to apply.

Therefore the relative role of external and internal trigger mechanisms of SF in BCG/H II-galaxies seems far from to being settled. It is high time to pose the question: what are the main trigger mechanisms of SF burst in BCGs? This is important both for the general understanding of the BCG phenomenon and the cosmological evolution of BCGs.

In this paper, we use the well-selected BCG sample of 86 galaxies in the zone of the Second Byurakan Survey to get the quantitative estimate of the relative frequency of gravitational interaction as a probable trigger of SF bursts in BCG progenitors. Based mainly on UZC (Falco et al. 1999) and the NED, we perform an extensive search for nearby disturbing galaxies, both more massive and those of comparable or lower luminosity than that of the studied BCGs. For BCGs without such probable disturbers in UZC and NED and in recent publications on galaxy redshifts, we have measured the radial velocities of some candidate disturbers with the SAO 6m telescope.

The paper is organized as follows. In Sect. 2 we briefly review tidal trigger of SF, discuss in more detail the most efficient examples, and describe the criteria we apply to assign a nearby galaxy as a disturbing neighbour. In Sect. 3 we describe our BCG sample, while in Sect. 4 we present the results of the "companion'' search and explain the sources we have used. A brief analysis of the properties of galaxies, associated with BCGs, the discussion of results obtained and our conclusions are presented in Sect. 5. The results of spectroscopy of 3 BCGs and 24 candidate companions in the vicinity of 15 studied BCGs are presented in the Appendix.