On the ground, meanwhile, the advent of the VLTs, and the prospect of NGST in space, have not stopped the progress of ambition. Even in the early '90s consideration of possible 25-m class ground-based facilities was developing (c.f., for example, Ardeberg et al. 1993, Owner-Petersen et al. ,1994 and references therein) and technical studies progressing (e.g. Ardeberg et al. 1996). By 1996 also, serious exploration of alternative technologies (Bash et al. 1996) was producing cost estimates as low as $150M for some 25-m telescope designs, while current science drivers and technical options were being closely examined. Mountain (1996) set what is now a benchmark scientific objective: to be able to secure spectra of the faintest objects in the Hubble Deep Fields. This requires a 50-m telescope working close to its diffraction limit. The possibility of 100-m telescopes was soon explored (Gilmozzi et al. 1998); such a facility could perform the same service for a large fraction of the objects expected to be seen in deep images with the NGST.
Even more recently, the critical technique of Multi-Conjugate Adaptive
Optics (MCAO)
, originally proposed by Beckers (1988, 1989) has been demonstrated
to work (Ragazzoni et al. 2000a). This promises (c.f. Ellerbroek
& Rigaut 2000) to liberate Adaptive Optics (AO) from the limitations
of classical single-guidestar techniques, which constrain fields of view
(FOVs) to a few arcsec, across which the point-spread function (PSF) varies
greatly. MCAO has the potential to provide, at least in the important NIR
wavelength range 1 - 2.5 um, quite uniform, near-diffraction-limited,
images over FOVs of order arcminutes. The combination of MCAO with
extremely large (>20 m) telescopes promises facilities of such power as
to offer a discontinuous change in astronomical capability ``comparable
to that of the invention of the telescope itself" (Gilmozzi
et al. 1998,
section 6).