Just three years ago, the first galaxy was found that had a higher redshift than the then highest redshift QSO; such an event was expected given that galaxies presumably predate QSOs, but this was the first time since the discovery of QSOs in the 1960s that this had happened. This object was at z = 4.92 (Franx et al. 1997). It identified z > 5 as the time when we might begin to see the development of substantial baryonic potential wells. Since then, the highest redshift QSO has crept over z = 5, but the highest redshift galaxy has jumped to at least z = 5.74, and possibly even to z = 6.68.
The three best determined z > 5 objects are at z = 5.34 (Dey et al. 1998), z = 5.60 (Weymann et al. 1998), and z = 5.74 (Hu et al. 1999). All these redshifts were measured from Keck LRIS spectra, and all show Ly, though all are faint with integrations ranging from 4 to 10 hours, as required to get adequate S/N in the emission line. The continuum fluxes are low, since these sources typically have IAB magnitudes around 26-27, and hard to detect against the bright night sky (Ly falls at 800 nm at z = 5.6). The detections are almost certainly real, since the objects are measurable in deep imaging data, and the expected flux decrements due to the continuum breaks at Ly 1216 Å and the Lyman limit at 912 Å have been seen. In addition, the line profiles display the expected asymmetry due to blue edge absorption in Ly in the outflowing ISM associated with the starburst (see e.g., Franx et al. 1997). The highest redshift object, at z = 6.68 (Chen, Lanzetta and Pascarelle 1999), is a less certain detection, given that it is faint and the data are relatively low S/N.
It is clear that we are pushing to z ~ 6 and beyond, but it is also clear that further progress really requires good near-IR imagers and spectrographs. At z = 6.68, Ly falls at 940 nm, where typical CCDs have quite low quantum efficiency. Another interesting issue with these high redshift objects is what will we actually do with them? They are so faint that it will be extremely difficult to to obtain high S/N data, even in images. As noted, they are fainter than IAB = 26, with typical line fluxes that are very low, and, even with Keck, integrations of 4-10 hours provide little more than redshift detections. If we are to get more detailed information about such high redshift objects then a different approach must be used.