The position of the cavate dwellings is a disadvantageous one to reach any cultivated fields if defenders were necessary. When the Tusayan Indian today moves to his _kisi_ or summer brush house shelter he practically camps in his corn or near it, in easy reach to drive away crows, or build wind-breaks to shelter the tender sprouts; but to go to their cornfields the inhabitants of the cavate dwellings I have described were forced to cross a river before the farm was reached. That these cavate dwellings were lookouts none can deny, but I incline to a belief that this does not tell the whole story if we limit them to such use. It is not wholly clear to me that they were not likewise an asylum for refuge, possibly not inhabited continuously, but a very welcome retreat when the agriculturist was sorely pressed by enemies. Following the analogy of a Hopi custom of building temporary booths near their fields, may we not suppose that the former inhabitants of Verde valley may have erected similar shelters in their cornfields during summer months, retiring to the cavate dwellings and the mesa tops in winter? All available evidence would indicate that the cavate dwellings were permanent habitations.[19] There are several square ruins on top of the mesa above the cavate dwellings. The walls of these were massive, but they are now very much broken down, and the adobe plastering is so eroded from the masonry that I regard them of considerable antiquity. They do not differ from other similar ruins, so common elsewhere in New Mexico and Arizona, and are identical with others in the Verde region. I visited several of these ruins, but made no excavations in them, nor added any new data to our knowledge of this type of aboriginal buildings. The pottery picked up on the surface resembles that of the ruins of the Little Colorado and Gila. The dwellings which I have mentioned above are said[20] to be duplicated at many other points in the watershed of the Verde, and many undescribed ruins of this nature were reported to me by ranchmen. I do not regard them as older than the adjacent ruins on the mesa above or the plains below them, much less as productions of people of different stages of culture, for everything about them suggests contemporaneous occupancy. From what little I saw of the village sites on the Verde I believe that Mindeleff is correct in considering that these ruins represent a comparatively late period of pueblo architecture. The character of the cliff houses of the Red-rocks shows no very great antiquity of occupancy. While it is not possible to give any approximate date when they were inhabited, their general appearance indicates that they are not more than two centuries old. There is, however, no reference to them in the early Spanish history of the Southwest. [Illustration: BUREAU OF AMERICAN ETHNOLOGY SEVENTEENTH ANNUAL REPORT PL. XCIII BOWLDER WITH PICTOGRAPHS NEAR WOOD'S RANCH]