It commonly happens that the less factual information we have on a problem, the more argument is generated. The lower crust is no exception. In fact the main thrust of our review paper on the lower crust was to demonstrate just how little factual data was available on which to construct reliable models for rustal growth, but that the data which was available (largely on Precambrian lower crustal granulites) did not fully support popular concepts that the lower continental crust is predominantly made up of cumulates or the refractory residues from partial melting. We did not set out to criticize Taylor's andesitic model for crustal growth directly, although in view of the important role the model has played over the last decade we could not avoid referring to it. Taylor & McLennan's defence of the andesite model, coupled with their critique of our own paper, nevertheless behoves us to comment on the andesite model both in relation to their specific points and in relation to our further observations since the paper was written.
With the andesite model there are basically three components to the equation: island arc = upper crust + lower crust. Knowledge of the geochemical characteristics of each is vital in order to test whether the equation will balance. Of the three, only the upper crust composition is known sufficiently well, and as far as the mean rare-earth distribution is concerned it can be approximated by the average sedimentary pattern given by Taylor (1977). There is much less