It may be well in this connection to cite a theory entertained by Mr.
Victor Mindeleff, of the Bureau of Ethnology, whose wide experience
among the southwestern ruins entitles his judgment to high
consideration. In his opinion the rectangular form of architecture,
which succeeds the type under discussion, must have been evolved from
the circular form by the bringing together, within a limited area, of
many houses. This would result in causing the wall of one circular
structure to encroach upon that of another, suggesting the partition
instead of the double wall. This partition would naturally be built
straight as a twofold measure of economy. Supposing three such houses
to be contiguous to a central one, each separated from the latter by a
straight wall, it may be seen that (as in the accompanying plan) the
three sides of a square are already formed, suggesting the
parallelogramic as a convenient style of sequent architecture.
[Illustration: FIG. 494.--Evolution of rectangular forms in primitive
architecture.]
All this, I need scarcely add, agrees not only with my own
observations in the field but with the kind of linguistic research
above recorded. It would also apparently explain the occurrence of the
circular semisubterranean _kí wi tsi we_, or estufas. These being
sacred have retained the pristine form long after the adoption of a
modified type of structure for ordinary or secular purposes, according
to the well known law of survival in ceremonial appurtenances.
In a majority of the lava ruins (for example those occurring near
Prescott, Arizona), I have observed that the sloping sides rather than
the level tops of _mesa_ headlands have been chosen by the ancients as
building-sites. Here, the rude, square type of building prevails, not,
however, to the entire exclusion of the circular type, which, is
represented by loosely constructed walls, always on the _outskirts_ of
the main ruins. The rectangular rooms are, as a rule, built row above
row. Some of the houses in the upper rows give evidence of having
overlapped others below. (See section, Fig. 495.)