A summary of the arguments both in favour of, and against, the mantle and crustal origin of granitic magmas is presented, with reference to the literature for detailed and fuller exposition of the relevant criteria.
B.E.L.: I find it difficult to understand why granitic bodies appear to be totally absent from the oceanic crust if granitic magma is commonly derived from the mantle. Not only are oceanic granites unknown, but geophysical evidence does not support the existence of buried batholiths in the oceanic crust that have not risen high enough to be exposed. Moreover, as it is generally agreed that the upper mantle is dominantly ultrabasic and rich in olivine, it seems that this is the least likely sort of composition to yield substantial quantities of quartz-rich granitic magma by partial melting. Indeed, Wyllie et al. 1976 argued that any melt formed from a peridotite must be in equilibrium with the residual olivine, and such melts cannot also be silica-rich. I believe, therefore, that most granitic magma is dominantly derived by partially melting the lower continental crust, sometimes by uprise of basaltic or other mantle-derived magma, sometimes by deep faults that bring hot mantle against already nearly melting basal crust, and sometimes simply by deep faults uplifting segments of the base of the crust, resulting in isothermal melting under reduced pressure. The detailed exposition of this view is in Leake (1978).
I know that the rather low initial 87Sr/86Sr ratios of many granites, such as the Caledonide suite of Britain,