Note: This paper is dedicated to Aaron and Elizabeth Waters on the occasion of Dr. Waters' retirement.

The eastern flank of the central and southern Cascade Mountains is bordered by a belt of shield volcanoes that appears to be a subprovince of the Oregon high-alumina plateau basalt petrologic province. Most of the volcanoes in this belt are low shields in which differentiation from the parent high-alumina basalt magma has been relatively slight, but several are large complex shield centers where differentiation has been extreme. The location of these large centers, and of some of the smaller volcanoes as well, was largely determined by intersecting concentrations of faults and fault-fissures of three regional fault systems.

One of the largest of the complex volcanic centers is Newberry Volcano in central Oregon, a shield volcano with a big caldera at its summit. The stratigraphy of the caldera walls and of features on the caldera floor at Newberry allows detailed interpretation of the history of the younger parts of the volcano and caldera. The formation of Newberry Caldera was apparently a slow process controlled largely by faulting along the three regional fault systems. The magma conduits were probably a gridlike plexus of intersecting dikes and fissures, with larger “magma pockets” at the grid intersections. The magma was trapped in shallow chambers and periodically released by faulting. The entrapment of the magma allowed differentiation in the shallow chambers.

The stratigraphy and petrology of the wall sequence also allows determination of the relative time at which the caldera had grown large enough to hold a caldera lake.

On differentiation plots, chemical analyses of the Newberry rocks show two trends: rocks erupted before the presence of a lake in the caldera trend toward slight iron enrichment, whereas rocks erupted after water was present in the caldera generally trend toward alkali enrichment. These different trends are attributed to differences in the oxygen fugacity of the magma which, in turn, are related to the presence or absence of large volumes of water in the caldera lake. The interpretation is supported by field, petrographic, petrologic, chemical, trace-element, and isotopic data. Plots of existing data for the Medicine Lake Highland Volcano, another large complex shield center in the belt, show the same type of two-trend relation as those of Newberry Volcano.