Abstract
We have combined deep starcount data with Galaxy model predictions to investigate how effectively such measurements probe the faint end of the halo luminosity function, We have tested a number of star/galaxy classification techniques using images taken in 0.5 arcsec seeing with LRIS on the Keek telescope, and we iind that different combinations of these techniques can produce variations of 10% in the inferred starcounts at R = 22.5 and 30% at R = 24.5 mag. The decreasing average angular size of galaxies with fainter magnitude effectively limits ground-based work to R < 25.5 mag. The higher angular resolution provided by HST allows one to probe at least 2 mag fainter, but the small field size is a significant limitation, In either case, our models show that the contribution from halo subdwarfs is effectively limited to colours of (R-1)<1.0. with the redder stars being members of the Galactic disk, The apparent increase in number density for M(V) > 10 in the derived luminosity function is a result of contributions from disk stars at fainter absolute magnitudes and does not provide evidence for an upturn in the halo subdwarf mass function. Indeed, starcount data alone are not an effective method of probing the shape of the halo luminosity function close to the hydrogen-burning limit, Finally, we examine how the Hubble Deep Field observations can be used to constrain the contribution of various stellar components to die dark-matter halo. (C) 1996 American Astronomical Society.