of the terrace. The most reasonable hypothesis, therefore, is that the
space between the base of the mounds and the edge of the terrace was
occupied by rooms of one story. This would also help to explain the
steepness of the slopes of the mounds themselves. The walls of the
structures they represent, being protected by the adjacent low walls of
the one-story rooms, would not suffer appreciably by undermining at the
ground level, and if the central room or rooms of each cluster were
higher than the surrounding rooms, as is the case in the Casa Grande
ruin, the exterior walls, being usually heavier than the inner walls,
would be the last to succumb, the clusters would be filled up by the
disintegration of the inner walls, and not until the spaces between the
low one-story walls surrounding the central cluster were nearly filled
up would the pronounced disintegration of the outer walls of the
structures commence. At that period the walls were probably covered and
protected by debris dropping from above, and possibly the profile of the
mounds was already established, being only slightly modified by surface
erosion since.

[Illustration: Pl. LII: Ground Plan of Casa Grande Ruin.]

About the center of the eastern side of the terrace, and also on the
western side, the water which falls on the surface of the structure is
discharged through rather pronounced depressions at these points. These
depressions are not the work of running water, though doubtless
emphasized by that agency, but represent low or open spaces in the
original structure, probably passageways or gateways. Furthermore,
before or inside each gateway there is a slightly depressed area, just
where we would expect to find it under our hypothesis, and showing that
the process of filling in is not yet completed. If the structure were to
remain undisturbed for some decades longer these spaces would doubtless