Very low Ti basalts andd green glass samples from the moon show high Lu/Hf ratios and low Hf concentrations. Low-Ti lunar basalts show high and variable Lu/Hf ratios and higher Hf concentrations, whereas high-Ti lunar basalts show low Lu/Hf ratios and high Hf concentrations. KREEP basalts have constant Lu/Hf ratios and high but variable Hf concentrations. Using the Lu-Hf behavior as a constraint, we propose a model for the mare basalts evolution. This constraint requires extensive crystallization of the primary lunar magma ocean prior to formation of the lunar mare basalt sources and the KREEP basalts. Mare basalts are produced by the melting of the cumulate rocks, and KREEP basalts represent the residual liquid of the magma ocean.

Lu and Hf concentrations and the Hf isotopic data of lunar rocks suggest that assimilation cannot be accepted as a major process to explain the diversity of the lunar mare basalts. The urKREEP hypothesis is also unnecessary. Both high- and low-Ti basalts show enough iron enrichment to be regarded as melting products of the last stage cumulate rocks from the lunar magma ocean. The KREEP basalts are also rich in iron and may be regarded as the final, residual liquid left after the crystallization of the major portion of the primary lunar magma ocean.