The density of basalt samples from Mare Tranquillitatis and Oceanus Procellarum is 3.36 g/cm3. These basalts should undergo a phase change at a pressure of approximately 10 kbar (200 km lunar depth) that would increase the density to 3.74 g/cm3. This density is substantially higher than the average density of the moon; therefore, lunar maria must be surrounded by a lower density material. The gravity anomaly at Mare Serenitatis, the largest lunar mascon, is so small that only a slender neck of mare material can extend to a source region at depth. No large mass concentrations exist in Mare Tranquillitatis and Oceanus Procellarum; therefore, only a tenuous network of pipes or thin dikes, at most, can connect the vast expanses of basalt on the lunar near side to a source region at depth. A density reversal must be present beneath the maria and Oceanus Procellarum. A good candidate for highland material is a high-aluminum basalt, which has a mineralogical composition similar to that of anorthositic gabbro and a density of 3.0 g/cm3. In the pressure range from 4 to 10 kbar (80 to 200 km in depth on the moon), this highland basalt would undergo a phase transformation with a concomitant density increase to 3.4 g/cm3. With this phase change, a moon composed predominantly of this high-alumina basalt would have the correct density and moment of inertia. For the lunar model proposed in this paper, the maria are areas of 3.36-g/cm3-density material overlying a 3.0-g/cm3-density material that undergoes a gradual phase change to a density of 3.4 g/cm3. In previous work I have hypothesized a modified fission process that would result in the moon having the composition described in this paper.