During the late 1980's, successively deeper redshift surveys carried out with
multi-object spectrographs on 4-m class telescopes produced growing evidence for evolution in the galaxy population. While some evolution had been expected from analysis of the galaxy number counts, the surprising indication from the first deep redshift surveys was that this appeared to involve moderate luminosity galaxies lying at moderate redshifts (Broadhurst et al. 1988, Colless et al. 1990, Cowie et al. 1991). However, while the results were suggestive, these early surveys suffered a number of significant problems that hampered their interpretation:
(a)the samples were small, especially at the faintest levels, so the statistical weight was limited and analysis was based on crude parameterizations of the data such as the median redshift of samples;
(b)the typical redshifts were small (z << 0.5), so that evolutionary effects could only be seen against “local” populations whose selection was often quite different - indeed the local luminosity function of galaxies is still poorly defined (Loveday et al. 1992, Marzke et al. 1994);
(c)the samples were selected in the observed B-band, so that comparison with local samples was based on the poorly constrained ultraviolet properties and relative numbers of galaxies of different types.