Hints for power on very large scales were first detected by the deep BEKS (Broadhurst et al. 1990) pencil-beam survey extending to z ~ 0.5 in the direction of the Galactic poles. The survey used a collection of narrow probes to map the galaxy distribution over a linear scale of about 2,000 h-1 Mpc. The observed distribution shows a remarkable regularity exhibiting an alternation of peaks and voids with a typical scale of 128 h-1 Mpc. However, follow-up observations in other directions not only do not confirm this regularity but detect power on smaller scales (~ 60 h-1 Mpc), in agreement with the results from nearby surveys. In order to verify these claims of large scale power, the ESO-Sculptor redshift survey (ESS, Bellanger & de Lapparent 1995) was designed to have a transverse dimension larger than the galaxy correlation length at the median redshift of the survey (z ~ 0.3) to assure that the detected structures are not artifacts caused by small-scale clustering, one of the main criticisms to the original interpretation of BEKS results. The ESS provided the first detailed map of the galaxy distribution in the redshift interval 0.1 < z < 0.5, which confirmed the existence of voids bounded by thin structures over the entire redshift interval. More importantly, the ESS confirmed that the voids have a typical size of ~ 60 h-1 Mpc, finding no evidence for periodic structures on scales ~ 130 h-1 Mpc.
Other deep surveys, extending to even larger redshifts (z 1), have also been completed but have focused primarily on the direct study of the evolution of the luminosity function, star formation and clustering. Among them are: 1) the Autofib Redshift Survey (Ellis et al. 1996) which combines several pencil-beam surveys of magnitude-limited samples (1700 galaxies) spanning a wide range in apparent magnitude down to bj = 24, and reaching z ~ 0.75; 2) the Canada-France Redshift Survey (CFRS, Lilly et al. 1995) containing some 600 galaxies brighter than IAB = 22.5, with a median redshift of z ~ 0.56, and covering an effective solid angle of 112 arcmin2. These data have provided one of the first secure evidence of a physical association of galaxies at z ~ 1 (Le Févre et al. 1994); 3) The CNOC2 survey (Carlberg et al. 1998) which presently contains about 5,000 galaxies with R < 21.5 to z ~ 0.6, over a total area of 1.5 square degrees.
Like in the past, these first results have motivated different groups to plan ambitious surveys using 8-10m class telescopes such as Keck (DEEP) and VLT (VIMOS) to observe large samples of galaxies (~ 105), probing scales of ~ 100 h-1 Mpc at z ~1 (e.g., Le Févre et al. 1996a). Completion of these surveys will allow one to put together a coherent and self-consistent picture of galaxy and clustering evolution from z ~ 1 to the present.