1 Introduction

The blue compact galaxies (BCGs) are characterised by their blue colour, strong nebular emission lines, compact appearance and low chemical abundances (e.g. Searle & Sargent 1972; Lequeux et al. 1979; Kunth & Sargent 1983; Masegosa et al. 1994; Izotov & Thuan 1999). The low chemical abundances and high star formation rates (SFRs) opened up the possibility that these were genuinely young galaxies presently forming their first generation of stars (Searle & Sargent 1972). Since these BCGs were found at low red-shift, this would mean that galaxies could still be forming in the local universe. If that would be the case, it would have considerable implications on theories on galaxy and structure formation, and on the nature of dark matter.

Since this was proposed, many BCGs have been found to contain a population of old stars. Evidence comes mainly from the studies of colours and morphologies of the faint halos seen around most BCGs (Loose & Thuan 1985; Kunth et al. 1988; Papaderos et al. 1996; Telles & Terlevich 1997; Doublier et al. 1999; Bergvall & Östlin 2001). Regular red halos are taken as evidence that there is an underlying old population present. Despite a few remaining young galaxy candidates it is evident that most BCGs are in fact old (see Kunth & Östlin 2001 for a review). What is less evident, however, is the reason why many BCGs seem to be forming stars at such high rates.

For many BCGs, the gas consumption time scales are significantly shorter than a Hubble time (e.g. Fanelli et al. 1988), which indicates that the current high SFRs have to be transient. The general picture has been that these galaxies undergo a few or several short bursts of star formation followed by longer more quiescent periods (Searle et al. 1973; Gerola et al. 1980). The explanation for this could be gas dynamical; the gravitational field has to compete with galactic winds produced by mass loss and supernovae (SNe), which may expel the gas and prevent further star formation. Depending on the mass of the galaxy and the pressure in the surrounding intergalactic medium, the gas may cool and later accrete back on the galaxy producing a new burst (Babul & Rees 1992). It is believed though that the number of bursts has to be few, from gas consumption arguments and in order not to exceed the observed metallicities. However, the latter argument is not so strong in view of the possibility that BCGs lose enriched gas in supernovae winds created shortly after the onset of the starburst. On the other hand, as winds travel through the galaxy, the newly synthesised metals will mix with the surrounding ISM. Thus an outflow of gas is not necessarily more metal rich than the ISM in the galaxy (Ferrara & Tolstoy 2000). Other explanations for the starburst phenomenon involve tidally triggered gas infall through interactions with companion galaxies (Lacey & Silk 1991). In this case one could imagine a disturbed gas-rich dwarf galaxy, or even an unevolved protogalactic cloud, as the progenitor of a BCG. However, some investigations show that most BCGs are fairly isolated: Campos-Aguilar et al. (1993) investigated the environments of H II-galaxies from the spectrophotometric catalogue by Terlevich et al. (1991) by comparing their location in space with those of galaxies in the CfA catalogue. They found that most H II-galaxies were isolated, and those that were not still had an average distance to its nearest neighbour of several hundreds of kpc. Similar results were obtained by Telles & Terlevich (1995). Salzer (1989) found that emission line galaxies were less clustered then normal giant galaxies and that they tended to avoid regions with high galactic density (of massive galaxies), similar to the results obtained for UV excess galaxies by Iovino et al. (1988), and recently by Telles & Maddox (2000).

On the other hand, Taylor et al. (1995, 1996a, erratum 1996b; Taylor 1997) made H I surveys for companion galaxies around H II-galaxies and LSBGs. They found that nearly 60% of the H II-galaxies in their sample had H I companions. The important difference with respect to the studies above is that this survey was a direct search sensitive also to faint gas-rich dwarf galaxies, while the others used catalogues with bad completeness for dwarfs. Thus, while BCGs tend to avoid giant luminous galaxies, they appear to have dim low-mass neighbours in many cases.

In any case, it is likely that the BCGs hold some key information about the early evolution and formation of dwarf and intermediate sized galaxies. If it can be understood how BCGs evolve on a longer time scale and under what circumstances an "object" can be transformed into a BCG, much will have been learned about the evolution of galaxies. Some have proposed a relationship with faint blue galaxies (FBGs) at $z\sim 0.5$ (Cowie et al. 1991; Babul & Rees 1992), others with low surface brightness galaxies (LSBGs), which may be the precursors and/or the successors of BCGs (Bergvall et al. 1999a, 1999b; Telles & Terlevich 1997). Further, an understanding of BCGs may help in interpreting observations of star forming objects at high red-shifts. The studies of intermediate-high red-shift compact galaxies in the Hubble deep field (Guzmán et al. 1996, 1997; Phillips et al. 1997) reveal that these include galaxies with properties very similar to those of local luminous BCGs.

Fully understanding the evolution of BCGs requires detailed knowledge about their star formation histories (SFHs), dynamics, and possible merger history. Most BCGs are far too distant to be resolved into stars even with the Hubble Space Telescope (HST). Instead, the SFHs can be examined with the help of spectral evolutionary synthesis models (SEMs) and optical surface photometry (Östlin & Bergvall 1994; Östlin et al. 1996; Bergvall & Östlin 2000). This can put constraints both on star formation processes, including the initial mass function (IMF) of stars and the number of old stars present. The tricky part is to constrain the mass of the old, cool stellar component, since the light in the optical is totally dominated by hot/massive stars. The old population is easier approached in the halos, outside the starburst region, and by adding observations in the near infrared (NIR) (Bergvall & Östlin 2001). A lot could also be gained with knowledge of the dynamics of BCGs, from which one can then put interesting constraints on the amount of old stars and dark matter present, and provide important information on the triggering mechanism behind starbursts in BCGs.

Previous studies of the dynamics of BCGs include optical and radio investigations. The latter target the 21 cm emission from neutral hydrogen and is very useful for studying the large scale dynamics, while the inferior spatial resolution (as compared to the optical regime) make them of limited use for studying the central dynamics and the kinematics of the star forming regions. Optical studies have probed either the full two dimensional velocity field, e.g. by utilising Fabry-Perot interferometry (Thuan et al. 1987; Marlowe et al. 1995; Petrosian et al. 1997; Östlin et al. 1999) or limited parts from slit spectroscopy (e.g. Gil de Paz et al. 1999). The general picture is that of large scale rotation (Östlin et al. 1999) and smaller scale distortions e.g. in the form of expanding bubbles (Marlowe et al. 1995; Martin 1998; Kunth et al. 1988; Östlin et al. 1999; Gil de Paz et al. 1999). Radio observations provide further evidence for large scale rotation in BCGs (Viallefond et al. 1987; Bergvall & Jörsäter 1988; Meurer et al. 1996, 1998; van Zee et al. 1998c).

The notion "starburst'' is frequently used in the literature to describe galaxies, or regions of galaxies, with varying degrees of star formation activity. In this paper, we will adopt a more strict definition by requiring that a starburst involves a global SFR which is unsustainably high. The SFRmay be unsustainable because the gas consumption time scale, or the time scale to build up the observed stellar mass, is significantly shorter than the Hubble time, i.e. the time averaged SFR is an order of magnitude lower than the present. Many galaxies have SFRs fluctuating with time, and/or have impressive bright H II regions, but this does not necessarily imply that the SFR is unsustainable over a Hubble time. Indeed, many galaxies classified as BCGs in the literature are not starburst according to our definition.

1.1 Sample selection and Fabry-Perot observations

In this paper we will discuss the results from a study of the dynamics of BCGs. In 1995, we obtained Fabry-Perot interferometric observations, targeting the H$\alpha$ emission line, for a sample of six luminous BCGs ( $M_B \in [-17,-20]$) and two companion galaxies. Four of the BCGs were taken from the catalogue of compact galaxies by Bergvall & Olofsson (1986). One galaxy (ESO 185-13) was taken from an extension of this catalogue (Bergvall et al. unpublished). The last BCG (Tololo 0341-407) was taken from the spectrophotometric catalogue of H II galaxies by Terlevich et al. (1991). The galaxies were selected to be actively star forming, as judged by the equivalent width and luminosity of H$\beta$ or H$\alpha$. The four galaxies from the Bergvall & Olofsson (1986) catalogue are the ones with the highest the equivalent widths in H$\beta$ from that list. The two other BCGs were in addition chosen because they had the right coordinates with respect to the allocated observing time. The reason for choosing intrinsically luminous BCGs was to get a more homogeneous sample, in view of the fairly small number of galaxies observed. Two of the BCGs selected from the Bergvall & Olofsson (1986) catalogue have confirmed physical star forming companions, and these were also included in the target list.

The observations and reductions are thoroughly described in Paper I (Östlin et al. 1999, hereafter Paper I), where we also present the velocity fields, rotation curves and rough mass estimates. The velocity fields obtained appear to be very peculiar, showing non axi-symmetric distortions and in several cases evidence for multiple dynamical components. In this paper we will continue the discussion and combine the dynamics with surface photometric data to model the dynamical and photometric masses of the galaxies. Moreover, morphologies will be discussed in detail.

The absolute visual magnitudes of BCGs range from -12 to -21, and there is of course no guarantee that our sample of relatively luminous BCGs is representative for all BCGs. Nor is there any guarantee that BCGs with similar mass or luminosity have similar physical origin. Morphologically there exist different types of BCGs (Loose & Thuan 1985; Salzer et al. 1989; Kunth et al. 1988; Telles et al. 1997). Apart from the varying intrinsic luminosities, the relative intensity of star formation varies as well and not all BCGs are starbursts. Many BCGs are simply distant dwarf irregulars with moderately active star formation that have been picked up in emission line surveys. Telles et al. (1997) found that BCGs which have irregular morphology at faint isophotal levels, are on average more luminous than those with regular morphology. Most of the BCGs in our sample have irregular morphology and thus may have a physically different evolutionary history as compared to low mass BCGs with regular morphology.

In 1999 and 2000 we carried out observations of some 15 new BCGs at La Silla, extending our sample to lower luminosities, and observing time for spectroscopic and photometric followup has been obtained. The new sample was selected to extend our luminosity range downwards, and include galaxies with a minimum H$\beta$ equivalent width (sometimes estimated from H$\alpha$) of $\sim$30 Å. These observations will be discussed in forthcoming papers. In addition, we obtained complimentary observations of the H$\alpha$ line widths for the BCGs in the current sample, which will be partly used in the present paper.

Throughout the paper we use $H_0 = h_{75} \cdot 75$ kms^-1/Mpc, with h₇₅=1. In general, absolute photometric properties like masses, luminosities and SFR, scale with h₇₅^-2. Kinematical properties are not sensitive to the choice of h₇₅, but the kinematical mass estimates scale with h₇₅^-1 since they involves a length scale. Relative properties like mass to light ratios and time-scales do not depend on h₇₅.