Galaxy surveys 333Figure 11.9.Comparison between the spectroscopic redshift (zspec) and the photometric
redshift (zB) in the HDF-N (Benitez 2000).
et al1999). The standard procedure is to apply statistical corrections, which use
empirical correlations of the UV slopeβwith the extinction derived from the
Balmer decrement in nearby starburst galaxies (Calzettiet al1994).
A collection of (mostly dust corrected) estimates of the SFR density over
a broad range of redshifts is shown in figure 11.11, which illustrates the great
progress made in recent years. This picture seems to suggest that a large fraction
of the stars had already been formed byz∼3. However, global average SFR
densities over large cosmic volumes, even in the hypothesis that we can correct
for dust extinction, tell us very little about the processes which modulate the star
formation (e.g. merging events) and lead to build galaxy masses over time. Future
space-based far-infrared (5–30μm) observations, by providing rest-frame near-
IR radiation (which is well correlated with the stellar and dynamical mass) and by
measuring the thermally reradiated dust emission in distant galaxy, hold the best
promise to shed new light on these issues.