years ago. When the movement to the summit of the mesa became general, the material of the old houses was utilized in the construction of the new ones, and at the present day it can almost be said that not one stone remains above another. So complete is the obliteration that no ground plan can be made out.

If similar conditions prevailed in De Chelly, there might be many more ruins of this class than those so far discovered. Even those found are not easily distinguished and might easily be passed over. Possibly there were small ruins of this type scattered over the whole canyon bottom. An example which occurs at the point marked [12] on the map, and shown in plate XLIV, presents no trace on the surface except some potsherds, which in this locality mean nothing. The site is a low hill or end of a slope, the top of which is perhaps 25 feet above the stream bed, but separated from it by a belt of recent alluvium carpeted with grass. The hill itself was formed of talus, covered with alluvium, all but a small portion of which was subsequently cut away, leaving an almost vertical face 15 or 18 feet high. In this face the ends or vertical sections of several walls can be seen; one of them is nearly 3 feet thick and extends 4 feet below the present ground surface.

Plate XLIV. Section of Old Walls, Canyon De Chelly

The filling of these ruins to a depth of 4 or 5 feet and the almost complete absence of surface remains or indications does not necessarily imply a remote antiquity, although it suggests it. During the fall and early winter months tremendous sand storms rage in the canyon; the wind sweeps through the gorge with an almost irresistible power, carrying with it such immense quantities of sand that objects a few hundred feet distant can not be distinguished. These sand storms were and are potent factors in producing the picturesque features of the red cliffs forming the canyon walls; but they are constructive as well as destructive, and cavities and hollow places in exposed situations such as the canyon bottom are soon filled up. The stream itself is also a powerful agent of destruction and construction; during flood periods banks of sand and alluvium are often cut away and sometimes others are formed. Yet there are reasons for believing that the old village ruins on open sites, now almost obliterated, mark the first period in the occupancy of the canyon, perhaps even a period distinctly separated from the others. Excavation on these sites would probably yield valuable results.

[II—HOME VILLAGES ON BOTTOM LANDS]

Ruins comprised in the second class are located on the bottom lands, generally at the base of a cliff, and without reference to the defensibility of the site. They are, as a rule, much broken down, and might perhaps be classed with the ruins already described, but there are some distinctive features which justify us in separating them. Ruins of this class are always located either at the base of a cliff or in a cove under it, on the level or raised but slightly above the bottom land, and sometimes at a considerable distance from the stream. The ground plans

can generally be distinguished, and in many instances walls are still standing—sometimes to a height of three stories. The ground plans reflect more or less the character of the site they occupy, and we would be as much surprised to find plans of their character in the open country as we are to see plans of class I within the canyon. Unlike the ground plans of class I, those of this group were laid out with direct reference to the cliff behind them, and which formed, as it were, a part of them.

In point of size, long period of occupancy, and position these villages were the most important in the canyon. The ruins often cover considerable areas and almost invariably show the remains of one or more circular kivas. Sometimes they are located directly upon the bottom land, more often they occupy low swells next the cliff, rising perhaps 10 feet above the general level and affording a fine view over it. Sometimes they are found in alcoves at the base of the cliff, but they always rest on the bottom land which extends into them; these merge insensibly into the next class—village ruins on defensible sites—and the distinction between them is partly an arbitrary one, as is also that between the last mentioned and the cliff ruins proper.