6.1.4 H I Clouds
Because it was observationally difficult to confirm the recurrent burst picture against the idea that blue compact dwarf galaxies are young, several H I surveys have provided independent clues. In their decisive early H I survey, Lo and Sargent (1979) showed that the space density of protogalactic clouds required by the youth hypothesis must be at least 8 Mpc-3, each cloud having a mass of about 5 x 108 M which is 2 to 3 orders of magnitude higher than what they found. H I-selected sample of clouds unavoidably turn out to have optical counterparts. In all cases, possible local H I primeval clouds have been found to be associated with stars (Djorgovski 1990, Impey et al. 1990, McMahon et al. 1990, Salzer et al. 1991, Chengalur et al. 1995). Since then, other surveys have been carried out to find isolated H I clouds but without success (Briggs 1997). Some interesting examples of H I without coincident optical emission were however provided by ``off-scans'' in 21cm line studies (Schneider 1989, Giovanelli and Haynes 1989, Chengalur et al. 1995, Giovanelli et al. 1995) but they are likely to be associated with (or bridge) nearby large visible galaxies. Tyson and Scalo (1988) have suggested that the majority of dwarf galaxies may be in a quiescent state, not forming young stars, and that they might consequently be missed by optical selection methods. A recent analysis of H I-selected galaxies (Szomoru et al. 1994) shows that H I searches do not yield a population of optically underluminous galaxies. Some faint nearby galaxies with strong emission lines remaining undetected in H I must have very little gas or store their gas in ionised or molecular form (Doublier et al. 1999). Interestingly, the recent work by Schneider et al. (1998) shows that the H I mass function may become steeper at faint masses, in analogy with an upturn in the optical luminosity function at faint luminosities. However detection limits for H I surveys remain quite high (NHI ~ 1018 cm-2) hence smaller pockets may still be hidden. On the other hand, isolated H I clouds with masses and sizes comparable to present dwarf galaxies would be easily seen with modern radio telescopes hence they must be very rare. This conclusion is reinforced by the damped QSO absorption lines that tend to occur in the haloes of bright galaxies and not in smaller H I clouds (Lanzetta et al. 1995; however see discussion in Sect. 8.2). It has been noted that the diffuse ionising background could drive H I clouds under the detection limits of current surveys (see Corbelli and Salpeter 1993a, b).