The above results, which of course by no means come exclusively from HDF (but have benefited enormously from it), point to a remarkably simple observational synthesis discussed by Fall et al. (1996), Madau et al. (1996) and Madau (1997a, b). This will no doubt be widely discussed during the meeting. The observational data point to a remarkably recent era of major star-formation as delineated by the radiation we can see. Although different techniques are used to delineate the total star formation rate per comoving volume and extrapolation is necessary beyond the magnitude limits probed at the various redshifts, progress is already being made to overcome these limitations. Supporting the argument that we have witnessed the construction of galaxies directly with HST over 1 < z < 3 are the flattening of the faint K-band counts (which precludes the existing of a significant population of highly reddened sources) and the rapid growth in mean physical size, and in the 0 < z < 1 era, the declining fraction of irregular and multi-component galaxies in conjunction with the luminosity function changes witnessed in the redshift surveys. Completely independent support comes from trends with redshift in the gas content (Storrie-Lombardi et al. 1996) and metallicity (Pettini et al. 1994) of the QSO absorption line population. There are many uncertainties in each of these measures but the synthesis of so many results from different directions is quite compelling.
We should remember that this an empirical picture; it does not guarantee an unique physical interpretation. Much attention has been given recently to a supposed theoretical triumph in explaining it on the basis of hierarchical models (Baugh et al. 1998). I think we have to put this result in perspective. Perhaps it is not surprising that hierarchical models can be arranged to approximately fit the above observations given the redshift of peak star formation activity is surely sensitive to the precise way in which feedback is implemented in the models. (3) However, there are some remaining puzzles. Foremost, in the hierarchical models, we expect the growth of large disks to have occurred relatively recently (Efstathiou, this volume) and this should mirror the decline in the abundance of systems with irregular morphology (Baugh et al. 1998). The presence of well-formed massive galaxies to at least z = 0.8, with approximately the present comoving density (Lilly, this volume), suggests a more complex interpretation may be required. And, of course, many suspect that the optically-detected trends may be only a lower limit to the star formation energy density that occurs over all wavelengths (Rowan-Robinson, this volume).
3 Actually the fit of
Baugh et al. (1998)
to the star
formation history discussed by
Madau (1997a)
is not particularly good, but the observations are hardly
completely determined at this point. Back.