_Excavation of the kiva._--The tendency to depress or partly excavate the ceremonial chamber existed in Zuñi, as in all the ancient pueblo buildings which have been examined; but the solid rock of the mesa tops in Tusayan did not admit of the necessary excavation, and the persistence of this requirement, which, as I shall elsewhere show, has an important connection with the early types of pueblo building, compelled the occupants of these rocky sites to locate their kivas at points where depressions already existed. Such facilities were most abundant near the margins of the mesas, where in many places large blocks of sandstone have fallen out from the edge of the surface stratum, leaving nearly rectangular spaces at the summit of the cliff wall. The construction of their villages on these rocky promontories forced the Tusayan builders to sacrifice, to a large extent, the traditional and customary arrangement of the kivas within the house-inclosed courts of the pueblo, in order to obtain properly depressed sites. This accidental effect of the immediate environment resulted in giving unusual prominence to the sinking of the ceremonial room below the ground surface, but a certain amount of excavation is found as a constant accompaniment of this feature throughout the pueblo region in both ancient and modern villages. Even at Zuñi, where the kivas appear to retain but few of the specialized features that distinguish them at Tusayan, the floors are found to be below the general level of the ground. But at Tusayan the development of this single requirement has been carried to such an extent that many of the kivas are wholly subterranean. This is particularly the case with those that occupy marginal sites on the mesas, such as have been referred to above. In such instances the broken-out recesses in the upper rocks have been walled up on the outside, roughly lined with masonry within, and roofed over in the usual manner. In many cases the depth of these rock niches is such that the kiva roof when finished does not project above the general level of the mesa summit, and its earth covering is indistinguishable from the adjoining surface, except for the presence of the box-like projection of masonry that surrounds the entrance trap door and its ladder (see Pl. LXXXVII). Frequently in such cases the surface of the ground shows no evidence of the outlines or dimensions of the underlying room. Examples of such subterranean kivas may be seen in the foreground of the general view of a court in Oraibi (Pl. XXXVIII), and in the view of the dance rock at Walpi (Pl. XXIV). But such wholly subterranean arrangement of the ceremonial chamber is by no means universal even at Tusayan. Even when the kiva was placed within the village courts or close to the houses, in conformity to the traditional plan and ancient practice as evidenced in the ruins, naturally depressed sites were still sought; but such sites as the mesa margin affords were