If the African Rifts are incipient plate margins, it is noteworthy that the magmas associated with them are wholly different from the tholeiites of the oceanic ridges. They are among the most highly alkaline of any rocks known.

The magmatic activity at the subduction zones, where the plates are being destroyed, is very different. There are three varieties of these plate junctions: continental against oceanic, oceanic against oceanic, and continental against continental. In both the junctions involving oceanic crust the material being consumed includes a variable thickness of sediment, underlain by 5 or 6 km of tholeiitic basalt overlying the downgoing mantle. These rocks are much less refractory than the pyrolite of the mantle and must surely compose a large part of material parental to the magmas formed along the subduction zones, the andesites, granodiorites, and granites. There is nowhere the tremendous volume of intermediate rocks that would have had to be formed if these voluminous magmas had been products of crystallization differentiation from a basaltic magma.

The presently most active of the continent-continent junctions is along the Himalayas where India is underthrusting the continent of Asia; here there is no evidence of magmatism except along the transcurrent faults at either end of the main range.

But there are large volcanic and plutonic masses that have no obvious relation to the plate boundaries active in Mesozoic and Cenozoic time. The Eogene volcanics of the San Juans and the Neogene volcanics of the Yellowstone are more than 1,500 km from any obvious subduction zone, and these regions of magmatic activity seem no more closely related to subduction zones than are the Tertiary igneous rocks of West Texas, the Cretaceous tuffs and plutons of Arkansas, the Cretaceous intrusives of the Monteregian Hills, and the minor Tertiary intrusives of Virginia.