Initial conditions set by Big Bang nucleosynthesis are Y = .24, Z = 0 for helium (e.g. Pagel 2000) and heavy elements respectively, D/H = 4 × 10-5 (Levshakov, Tytler & Burles 1998) and 7Li/H = 1.7 × 10-10 (Bonifacio & Molaro 1997). The D/H ratio is the best indication of the overall density of baryons in the universe, which can be expressed as 0.03 B h702 0.04, similar to the density of Lyman- forest gas at red-shifts 2 to 3 (Rauch et al. 1998), whereas the mass in visible stars in galaxies is given by * h70 0.0035 (Fukugita, Hogan & Peebles 1998), i.e. only 1/10 as much. Thus 90 per cent of baryonic matter is unseen and of unknown chemical composition, although it is reasonable to speculate that most of it is still intergalactic gas with Z now somewhere between 0.3 Z (Mushotzky & Loewenstein 1997) and 0.1 Z or less (Cen & Ostriker 1999).
The remainder of cosmic chemical evolution is the result of star formation, the history of which has been extensively studied by Madau and others (Madau et al. 1996; Steidel et al. 1999; Pettini 1999) using data from red-shift surveys, Lyman break galaxies etc. (see Figure 1). Thus we now have a fair idea about global star formation rates since z = 4.5, but ironically we do not know how to associate them with particular types of galaxies. The good news is that the integral over this version of the SFR history does come close to the estimated cosmic density of stars as given above, and it seems that about 1/4 of the stars were formed at red-shifts greater than 2.5, over 1010 years ago. This raises the question of what happened to all the metals they made, to which I return in the last section.
Figure 1. Global comoving star formation rate density vs. lookback time compiled from wide-angle ground-based surveys (Steidel et al. 1999 and references therein) assuming E-de S cosmology with h = 0.5, after Pettini (1999). Courtesy Max Pettini.