The closeness of correspondence of all these features in the round ruins to those in the cliff ruins (features which in the round ruins appear less in place than in at least the older cliff ruins) would seem to justify my conclusion, earlier stated, that the round towns were simply outgrowths of the cliff villages, transplanted, as it were, into the plains; for all of these features, as they occur in the old cliff ruins, can, with but a single exception (that of the circular form of the kivas or assembly chambers, which, as will presently be shown, were survivals of a yet older phase of building), be accounted for as having originated from necessity, whereas in the round ruins they could not have originated even as possible expedients, since they were unsuitable save by having become customary through long usage. I have reasserted this fact because the theory that all cliff dwellings were but outlying places of refuge or the hunting and farming stations of larger pueblos in their neighborhood, strongly fortified by position in order that the small parties occupying them now and then for longer or shorter seasons might find safe retreat in them, has been advanced quite successfully. As this theory is not unlikely to gain a considerable hearing, it is necessary to demonstrate even more fully the fact that at least the round towns did not give their structural characteristics to such of the northern cliff ruins as resembled them in plan, and that therefore the latter are to be regarded as actual cliff-dweller remains. In the southern portions of New Mexico and Arizona, as on the upper Salado and in canyons of the Sierra Madre, still farther south, all the cliff dwellings and villages were built without reference to the curved forms of the caverns in which they occurred.[2] That is, they rigidly retained the rectangular pueblo form of arrangement characteristic of the larger ruins in the valleys and plains around them. Hence for this and for other reasons they may be regarded as pueblos transferred to the cliffs, such outposts of the larger pueblos of the plains as it is claimed all cliff dwellings were. So, also, as hitherto intimated, many of the later cliff dwellings, even of the north, have rectangular pueblo additions below them in the canyons or above them on the mesas, and some of the village ruins in the cave shelters themselves are almost faithful miniature reproductions in general plan of the large pueblos of the plains near at hand; but in the one case the pueblo additions above and below were comparatively modern, and indicate either that the cliff dwellings they are adjacent to continued to be occupied down to the time of later true pueblo building, or that they were reoccupied from comparatively modern pueblos and that all additions made were constructed according to customary later forms of building. In the other case, that of the rectangular structures in semicircular cave shelters, either a return