Recent results on meteor ablation are used as input to a model in which ablation (or evaporation) is the source and deposition on dust particles is the sink. The dust comes from recondensation of the rest of the meteor vapor. An excellent match is found to the observed characteristics of the sodium layer. It is argued that the high-latitude winter maximum of abundance (and layer height) is due to the much lower rate of ionization there and over the polar cap. The computations show that required factor of 4--5 and also match the height variation. To keep the ion density from being too large, downward electrodynamic transport at ?50 cm/sec must be invoked. The different behavior and low abundance of potassium require an additional sink. The one proposed in Penning ionization by metastable excited O2. Inherent in the model is that sodium and its compounds reside almost entirely on dust particles below 80 km. Large abundances in the free form at stratospheric heights are very unlikely and are not a good explanation of observed ambient ions.