Previous
abstract 
Next abstract
\def\eg{e.g. \ } \def\La{{Ly$\alpha$} \ }
One way to improve our understanding of galaxy formation is to study the properties of the very youngest galaxies in the Universe. However, no definitive observations of galaxies in their earliest collapse and star formation phases (primeval galaxies or PGs) have yet been made. It follows that the detection of PGs is one of the most important tasks confronting modern observational astronomy.
Model-independent considerations lead to predictions of many of the properties of PGs. Principal among these is the fact that PGs must be quite numerous; predicted surface densities typically lie in the range 10$^3$--10$^5$ deg$^{-2}$. Luminosities are more model dependent, but most (dust-free) models predict that PGs should be detectable optically with ground-based 4m-class telescopes.
Although many methods for searching for PGs have been proposed, one of the most sensitive is searching for faint emission line (redshifted Ly$\alpha$) objects. A number of groups have searched for \La emission line objects at redshifts from 1.8 to 6, without success. These surveys have placed constraints on the brightness and surface density of PGs that are in marked conflict with a variety of ``standard'' models.
Recently, we have obtained CFHT observations in a narrow spectral region near 9000\AA\ that is characterized by an exceptionally low sky background. (The narrow band that was chosen lies between strong OH lines, and possesses a sky flux that is close to the limit set by the zodiacal light.) Since our PG detection thresholds are set by sky photon noise, we are able to place very strong limits on the numbers of \La emitting PGs at $z=7$.
The conclusion from our work at ``high'' redshift ($z > 1.8$) is either that (i) most galaxy formation is very recent ($z < 1.5$), or that (ii) the \La emission that should accompany star formation activity is shrouded by dust. The latter possibility leads to several interesting search strategies for the future.
