The Wineglass welded tuff is therefore regarded as the product of glowing avalanches, erupted as an emulsion of finely divided magma spray enclosing larger incandescent clots of viscous glass in a medium of highly heated and compressed gas. The mass must have been extremely fluid, and its path was controlled by pre-existing hills and valleys. Where it came to rest in shallow depressions, as at the Wineglass, it accumulated to greater thickness than elsewhere and therefore retained its heat and gases for a longer time. Cooling slowly in these depressions, the particles of fine, plastic glass firmly adhered to one another while the larger and hotter lumps slowly collapsed and spread under the material above. On the low divides between the depressions, conditions were quite different. There the tuff was thinner, cooled more rapidly, and lost its gas earlier. For these reasons, the constituent fragments show neither welding nor flattening, but remain incoherent, and the deposits clearly betray their pyroclastic origin. In a word, the nature of the tuff is determined by the topography over which it flowed. In the valleys it resembles streaked lava; on the higher ground, it is thinner and the fragmental character is unmistakable.
Beneath the welded tuff forming the brim of the Wineglass lies a layer of loose, coarse lump pumice, the exposed thickness of which is approximately 20 feet (plate 15, figure I). Probably similar ejecta extend downward beneath the talus for another 100 feet, to the neck of the Wineglass. The bulk of this deposit is made up of lumps of pumice between I inch and 2 feet across, characterized by extreme vesicularity and a shredded, silky appearance. Mixed with the pumice, and comprising about 5 per cent of the deposit, are angular lithic blocks of andesite; these were doubtless torn from the sides of the conduits as the frothy magma was expelled. In its lower part the pumice is pale cream or white, but within 3 or 4 feet of the overlying welded tuff it becomes pink or brick red. At first one is likely to attribute this discoloration to reheating by the welded tuff, since the pulverulent base of the tuff is also pink. Closer examination and comparison with other pumice deposits which can never have been reheated in this way show that this interpretation is false. Similar reddening may be seen close to the top of the latest pumice deposits filling the canyons around Crater Lake. It is apparently caused by near-surface oxidation of iron-bearing fumes escaping from the pumice as it slowly cools. This explanation is supported by the fact that within the white pumice, far below the base of the welded tuff, there are many bombs with pink or brown crusts and others which show a concentric pink layer within a millimeter or two of the surface. Tsubo3 has described similar pumice lumps among the products of the 1929 eruption of Komagatake and has shown that the pink color may be produced artificially when the pumice is kept at 730° C. for a period of I to 4 hours.
With regard to the Wineglass section, one more question calls for, an answer. Was the welded tuff erupted immediately after the coarse lump pumice, or did a long interval of quiescence intervene? The pumice, as we have noted already, is incoherent. It would therefore have been particularly liable to erosion. Yet throughout a length of 4 miles, the contact with the welded tuff is perfectly smooth and comfortable. There is no hint of any channeling. Presumably, therefore, the eruption of the welded tuff followed immediately on that of the lump pumice. Possibly the pumice was still hot and giving off gas when it was buried by the tuff, and this may explain why the pulverulent base of the latter shares the pink color of the topmost pumice.
Sections northwest of the Wineglass. It is instructive to make a traverse along the rim of the caldera wall in either direction from the Wineglass, if only to observe the remarkably rapid lateral variations that take place in the welded tuff. As the tuff and underlying lump pumice are traced uphill toward Roundtop, the pumice thins to approximately 50 feet and the tuff thins to about 10 feet, at the same time losing its welded character. The tuff is no longer streaked with black obsidian, but gradually merges into a yellow, orange, or brown, friable pumiceous lapilli tuff. If the tuff was once continuous over the entire top of the Roundtop lava, it has since been stripped from the higher marginal parts and is exposed only in the saddle-shaped depression over the center, where it is approximately 8 feet thick and is overlain by bouldery drift. The underlying silky lump pumice, on the other hand, can be followed uninterruptedly over the Roundtop flow, resting directly on the lava near its margins and on an intervening lens of morainic debris in the lower, central part.
In the descent from Roundtop to the shallow depression separating it from the Palisades flow, the tuffaceous dacite begins at once to thicken and become compacted again, and within a short distance it shows the same firmly welded, streaky appearance which it displays in the brim of the Wineglass. In the center of the valley it forms a 30-foot cliff underlain by some 50 feet of incoherent, white lump pumice. The two layers then continue westward over the top of the Palisades flow.
