Stellar population models with abundance distributions determined from the analytic Simple model of chemical evolution fail to match observations of the nuclei of bulge-dominated galaxies in three respects. First, the spectral energy distribution in the mid-ultraviolet range 2000 < lam < 2400 exceeds observation by ~ 0.6 mag. Most of that excess is due to metal-poor main sequence stars. Second, the models do not reproduce metal-sensitive optical absorption features that arise mainly from red giant stars. Third, the strength of a Ca II index sensitive to hot stars does not jibe with the predicted number of A-type horizontal branch stars. The number of metal poor stars in galaxies is at least a factor of two less than predicted by the Simple model, exactly similar to the ``G Dwarf problem'' in the solar cylinder. Observations at larger radii in local group galaxies indicate that the paucity of metal poor stars applies globally, rather than only in the nuclei. Because of the dominance of metal rich stars, primordial galaxies will have a plentiful dust supply early in their star formation history, and thus will probably have weak Lyman-alpha emission, as is apparently observed. We confirm that early-type galaxies cannot have been formed exclusively from mergers of small all-stellar subsystems, a result already established by dynamical simulations. The constraint of peaked abundance distributions will limit future chemical evolution models. It will also make age estimates for the stellar populations in early type galaxies and bulges more secure.

14 pages, LaTeX includes 3 postscript figures. Uses AAS LaTeX v 4.0 and times.sty. Accepted for publication in the Astronomical Journal. Postscript available at http://shemesh.gsfc.nasa.gov/~dorman/Ben.html