Dr E. J. Cobbing writes: Following the discussion on the origin of granite magmas under Dialogue in the journal, I would like to make a few points which may clarify some of the issues raised, and also indicate other aspects of the problem which seem to me to be important. In his contribution, B, E. Leake made the point that large bodies of granite are always associated with the continental crust and are virtually absent from oceanic areas, which to him suggested that the granites were derived from continental crust. G. C. Brown agreed that continental crust was necessary for the formation of granitic melts but argued that the crust simply provided a structural trap where by the thermal regime was altered and mantle-derived granites could form and develop. This hypothesis is attractive because it resolves the paradox that granitic batholiths are invariably associated with continental crust but that the granites them selves may be mantle derived.
Our work in Peru has shown that the Peruvian Batholith has initial 87Sr/86Sr ratios of 0.7042, which suggests that all of its components, even the most acid ones, are mantle derived, and as the batholith is emplaced into thick continental crust these data support the hypothesis of Brown insofar as it relates to calc-alkali granites of I type (Chappell & White1974).
The association of granitoid rocks with volcanic rocks is a common feature of circum-Pacific fold belts and the phenomenon is commonly ascribed to subduction of oceanic crust at convergent plate boundaries.