The pattern of incompatible elements (K, Rb, Ba, r.e.e., Hf etc.) is the same for most samples from the lunar highlands. It is suspected that this pattern of incompatible elements is typical for the whole lunar crust. This seems to be a reasonable assumption as one can show from heat flow data that a large part of the Moon's total U (and consequently other incompatible elements) has to be concentrated in a thin crustal layer, which certainly contributes to the sampled highland rock types. It is supposed that a partial melting process of the major part of the Moon has extracted the trace elements from the interior into the crust. The patterns of incompatible elements of mare basalts are those expected if a second partial melting process were applied to the trace-element-depleted interior. Some consequences of this model are discussed. A relatively constant Sr and Eu distribution through the whole Moon is inferred, implying a positive Eu-anomaly in the lunar interior.