After the main cone of Mazama had been constructed by long-continued emission of hypersthene andesite, the magma changed to dacite on the one hand and to olivine-bearing basaltic andesite and olivine basalt on the other. More the culminating eruptions which led to the formation of the caldera, the basic magma escaped from parasitic cinder cones far down the sides of the mountain, and the acid magma was also erupted in the main from parasitic vents, either in the form of viscous flows and domes or as pumice.
The culminating eruptions were likewise characterized by the presence of opposed types of magma, but now they escaped from the summit vents instead of from fissures on the flanks of the cone. The acid magma was erupted as dacite pumice, essentially like the lavas erupted from the Northern Arc of Vents, but somewhat richer in hornblende. On the other hand, the basic magma was blown out in the form of scoria exceptionally rich in hornblende and basic feldspar and relatively poor in pyroxene and olivine. Yet despite the abundance of hornblende in the products of these final scoria eruptions, the chemical character of the magma is almost identical with that of the hornblende-free parasitic cinder cones and the Pliocene, pre-Mazama lavas.
From a petrological standpoint there is perhaps no more striking feature of the Crater Lake rocks than this extraordinary abundance of hornblende in the basic scoria of the culminating eruptions and the paucity of the mineral among all the earlier products of the volcano. Apparently before the culminating eruptions began, the upper part of the magma chamber was composed of dacite in which the stable ferromagnesian minerals were mainly hypersthene and augite. In this pan, hornblende was rare. At greater depth, the reverse was the case.
At Montserrat, MacGregor was led to the conclusion that “the pumiceous and most glassy rocks characterized by green hornblende . . . represent highly gas-charged magma that initiated great volcanic explosions.” He pictured this magma as overlain by semiconsolidated material in which the hornblende was brown. This was incapable of initiating great explosions. He suggested further that the pyroxene lavas represent hornblendic magma in which the amphibole has been completely or almost completely altered. Finally, he noted that in general the hornblende of the basic autoliths is brown and more or less resorbed, and he therefore supposed that they were derived from the semiconsolidated upper part of the magma chambers and were blown out with the green hornblende-bearing pumice rising from greater depth.
To what extent are these deductions applicable at Crater Lake? In several respects there appear to be close analogies. In the first place, as we have seen, the hornblende-rich scoria of the culminating eruptions is chemically almost identical with the olivine-bearing basaltic andesites erupted by the parasitic cinder cones. From the distribution of these cinder cones, it may be supposed that they were fed from fissures connected with the main reservoir. Perhaps small satellitic chambers underlay each one. In such shallow chambers, the hornblende which had been stable at greater depths in the principal chamber may have suffered complete breakdown to pyroxene and olivine.