onsequent employment of the covered way.

A further contrast between the general plans of Oraibi and Zuni is
afforded in the different manner in which the roof openings have been
employed in the two cases. The plan of Zuni, Pl. LXXVI, shows great
numbers of small openings, nearly all of which are intended exclusively
for the admission of light, a few only being provided with ladders. In
Oraibi, on the other hand, there are only seventeen roof openings above
the first terrace, and of these not more than half are intended for the
admission of light. The device is correspondingly rare in other villages
of the group, particularly in those west of the first mesa. In
Mashongnavi the restricted use of the roof openings is particularly
noticeable; they all are of the same type as those used for access to
first terrace rooms. There is but one roof opening in a second story. An
examination of the plan, Pl. XXX, will show that in Shupaulovi but two
such openings occur above the first terrace, and in the large village of
Shumopavi, Pl. XXXIV, only about eight. None of the smaller villages can
be fairly compared with Zuni in the employment of this feature, but in
Oraibi we should expect to find its use much more general, were it not
for the fact that the defensive site has taken the place of the close
clustering of rooms seen in the exposed village of Zuni, and, in
consequence, the devices for the admission of light still adhere to the
more primitive arrangement (Pls. XL and XLI).

The highest type of pueblo construction, embodied in the large communal
fortress houses of the valleys, could have developed only as the
builders learned to rely for protection more upon their architecture and
less upon the sites occupied. So long as the sites furnished a large
proportion of the defensive efficiency of a village, the invention of
the builders was not stimulated to substitute artificial for natural
advantages. Change of location and consequent development must
frequently have taken place owing to the extreme inconvenience of
defensive sites to the sources of subsistence.

The builders of large valley pueblos must frequently have been forced to
resort hastily to defensive sites on finding that the valley towns were
unfitted to withstand attack. This seems to have been the case with the
Tusayan; but that the Zuni have adhered to their valley pueblo through
great difficulties is clearly attested by the internal evidence of the
architecture