The ``bread and butter'' of the universe is the conversion of gas
into stars. We know a lot about the modes of star formation in local
galaxies all along the Hubble sequence
(Kennicutt 1998).
We are also
getting a basic idea from deep surveys of the history of star formation
in the universe
(Madau, Pozzetti, &
Dickinson 1998).
However, it is also
clear that there are dim and unrecognized repositories of baryons. At low
redshift, the serendipitous discovery of Malin 1 - still that largest
and most gas-rich galaxy known - is a reminder that star formation in
disks can be slow and very inefficient
(Impey & Bothun 1989).
Figure 1 shows that the lowest surface
brightness galaxies have properties
quite distinct from normal stellar systems on the Hubble sequence.
At high redshifts, galaxies can escape detection due to the severe
effects of surface brightness dimming or due to dust obscuration.
We still do not know the full range of star formation histories in
the universe. The luminosity-weighted integral 
(L) L
dL appears to converge not far below L*, but there are
indications of a steepening fainter than MB = -16 (see the
discussion in
Impey & Bothun 1997).
The most abundant stellar system
in the universe is a gas-rich dwarf galaxy, examples of which have
rarely been studied outside the Local Supercluster. The motivation
to look for additional baryons is strong; 
lum from
the integral of the galaxy luminosity function is a factor of 2-3 below
the lower bound on baryon from nucleosynthesis arguments
(Copi, Schramm, &
Turner 1995;
Persic & Salucci 1992;
Bristow & Phillipps
1994).
|
| Figure 1. Collapsed structures in the universe, assuming dissipation by intercloud collisions, after Efstathiou & Rees (1983). |