(b) Cliff dwellings, or buildings erected in caves.
From its topographic character it might be expected that the Canyon de Chelly ruins would hardly come within a scheme of classification based upon those found in the open country; and here, if anywhere, we should find corroboration of the old idea that the cliff ruins were the homes and last refuge of a race harassed by powerful enemies and finally driven to the construction of dwellings in inaccessible cliffs, where a last ineffectual stand was made against their foes; or the more recent theory that they represent an early stage in the development of pueblo architecture, when the pueblo builders were few in number and surrounded by numerous enemies. Neither of these theories are in accord with the facts of observation. The still later idea that the cliff dwellings were used as places of refuge by various pueblo tribes who, when
the occasion for such use was passed, returned to their original homes, or to others constructed like them, may explain some of the cliff ruins, but if applicable at all to those of De Chelly, it applies only to a small number of them.
The ruins of De Chelly show unmistakably several periods of occupancy, extending over considerable time and each fairly complete. They fall easily into the classification previously suggested, and exhibit various types, but the earliest and the latest forms are not found. In the descriptions which follow the classification below has been employed:
I—Old villages on open sites.
II—Home villages on bottom lands.
III—Home villages located for defense.
IV—Cliff outlooks or farming shelters.
[I—OLD VILLAGES ON OPEN SITES]
In the upper part of the canyon, and extending into what we may call the middle region, there are a number of ruins that seem to be out of place in this locality. They are exactly similar to hundreds of ruins found in the open country; such, for example, as the older villages of Tusayan, located on low foothills at the foot of the mesa, and the peculiar topographic characteristics of the location have not made the slightest impression on them. These ruins are located on gentle slopes, the foothills of the talus, as it were, away from the cliffs, and are now marked only by scattered fragments of building stone and broken pottery. The ground plans are in all cases indistinguishable; in only a few instances can even a short wall line be traced. They seem to have been located without special reference to large areas of cultivable land, although they always command small areas of such land. There is a remarkable uniformity in ruins of this type in character of site occupied, outlook, and general appearance. They are always close to the stream bed, seldom more than 10 or 12 feet above it, and the sites were chosen apparently without any reference to their defensibility. A typical example occurs at the point marked [60] on the detailed map ([plate XLIII]), another occurs at [58], and another at [52]. One of the largest examples is in the lower part of the canyon. At the junction of Del Muerto there is a large mass of rock standing out alone and extending nearly to the full height of the canyon walls. On the south it is connected with the main wall back of it by a low tongue of rock, sparsely covered in places by soil and sand, and on the top of this tongue or saddle there is a large ruin of the type described, but no ground plan can now be made out. Possibly the obliterated appearance of this ruin and of others of the same class is due to the use of the material, ready to hand and of the proper size, in later structures. It is known that a similar appearance was produced in Tusayan by such a cause. The old village of Walpi, on a foothill below the mesa point and the site of the village at the time of the Spanish conquest, presents an appearance of great antiquity, although it was partly occupied so late as fifty