The lengthy paper on this subject by M. J. O’Hara Tompkins, 1999) and studies of crustal thickness (e.g. (2000) contains an interesting reassessment of lunar evoluNeumann et al., 1996). These papers are also not cited. tion, although it differs little in its conclusions from his This new evidence has led to a revision of the Al2O3 original views published over a quarter of a century ago. content of the bulk lunar crust from 24·6% [the value To answer all the arguments in detail would demand a used by Taylor (1975) that O’Hara discusses] to 30% paper of equal length. However, it is possible to discern (Lucey et al., 1995). some basic underlying assumptions among the extensive To account for this amount of feldspar (plagioclase has discussion and I have chosen to concentrate on these to 36% Al2O3), close to 50% of the Al (and Eu) in the bulk display some of the fundamental fallacies that underlie Moon has to be concentrated in the crust, shortly after O’Hara’s case. the formation of the Moon. Flotation of feldspar in a dry Crucial to his argument is his statement that ‘there is magma ocean is the only viable hypothesis, accounting, no positive europium anomaly in the average lunar incidentally for the parallel REE patterns (excluding Eu) highland crust’ (p. 1545, see also p. 1551) and this leads observed in most highland samples. Thus much new him to deny the existence of a magma ocean from which evidence ignored by O’Hara has only served to subthe crust formed by plagioclase flotation. Consequently, stantiate the traditional interpretation. ‘there is no negative europium anomaly in the average O’Hara wishes in fact to form the thick feldspathic mantle to be inherited by later mare basalts’ (p. 1545, crust by partial melting of wet peridotite and claims that see also p. 1555). ‘partial melting in the presence of water, followed by As these two factors, an enrichment of Eu in the near-surface fractionation and volatile losses, can explain highland crust and a corresponding depletion in the the feldspathic character, high incompatible element conmantle, inherited by the later mare basalts, are corcentrations and lack of Eu anomaly in the lunar highlands’ nerstones of the standard model for lunar evolution (p. 1545, see also p. 1553). This view, heavily based on based on a magma ocean, the remainder of O’Hara’s terrestrial analogues, is unfortunately negated by the conclusions stand or fall on their validity. completely anhydrous nature of all lunar minerals and His claim that there is no overall enrichment of Eu in total absence of any evidence for water on the Moon, the average highland crust is based on the data from the even at ppb levels (the famous rusty rock 66095 was Apollo 16 site as interpreted by Korotev & Haskin (1988). hydrated in the terrestrial atmosphere; Epstein & Taylor, The fundamental difficulty is that the Apollo 16 site 1974). is more basic than the average highlands. Subsequent O’Hara also seems unwilling to accept the data that extensive coverage of the farside highlands by the Clemshow that there is a great depletion of other volatile entine and Lunar Prospector missions has revealed that elements in the Moon, preferring a scenario, as in much they are dominated by Ca-rich anorthosite, as discussed of the rest of the paper, based heavily on terrestrial by Lucey et al. (1995), a paper that does not appear in experience. However, he also appeals to the example of the extensive reference list of O’Hara’s paper. Other Io as a model for getting rid of volatiles such as Na and evidence of the highly feldspathic nature of the highlands sulphur. That unfortunate body, in the close embrace of Jupiter (six Jupiter radii distant from a body of 318 Earth comes from studies of the lunar farside (e.g. Pieters &