Continental arcs, such as the Andes, parts of the Himalaya and Tibetan Plateau and the Sierra Nevada of the western USA, are stuffed with granite intrusions. Large volumes coalesce to form classic batholiths. It is now well-accepted that very little of the granitic magma originated by melting of older continental crust, but by processes of fractionation from more mafic parent magmas. That presupposes a layer of dense, mafic to ultramafic cumulates below and complementing up to 30 km of batholithic crust. The overall density of the continental arc crust would be high relative to that of the granites themselves. So the fact that many batholithic cordilleras are topographically high suggests one of several processes: either the granitic part of the crust has become tectonically thickened relative to its denser root, or that root has separated from the continental lithosphere as a whole, and sunk into the mantle. Such decoupling, or delamination, would induce the remaining lithosphere to rise dramatically. Also, its descent could result in partial melting to produce peculiar potassium-rich basaltic magmas. The latter occur in Tibet and their presence there has been linked to foundering of deep lithosphere, that may have triggered the relatively recent surge in Himalayan uplift. Proving the existence of a descending lump of lithosphere is not easy, but developments in seismic processing can make a crucial contribution, if sufficient data are available for a suspected zone of delamination. The western USA is blessed with lots of seismic stations, so is a natural place to try out the new techniques as a test of the hypothesis. George Zandt of the University of Arizona, and other US colleagues have come up with interesting results (Zandt, G. et al. 2004. Active foundering of a continental arc root beneath the southern Sierra Nevada in California. Nature, v. 431, p. 41-46). Their analyses of seismic data shed light on a late stage in the development of the Sierra Nevada. During the Mesozoic Era, subduction beneath North America of the now disappeared Farallon plate of Pacific ocean lithosphere built up the Sierra Nevada batholith. About 10-16 Ma ago, subduction stopped and the plate margin became one of transpression, the most prominent feature of which is the San Andreas Fault. At that stage, a “drip” of dense cumulates began to form, and subsequently separated to descend into the mantle. Cruustal rebound was not simple but included zones of extension, as well as tell-tale high-K volcanism during the Pliocene.