424 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1919. 
Juan in Utah and Colorado. The essential difference is the existence, 
in the latter, of a room specialized in form and structure for a distinct 
purpose, of which no homologue appears in the cliff houses of the 
Gila area. The differences that give us two great groups are so funda- 
mental that when we examine the implements and domestic objects 
found in groups characterized in this way we find they also differ. 
There are large cliff dwellings in the mountains of Chihuahua, 
Old Mexico. These resemble in their sites the cliff dwellings of the 
San Juan and those of the Gila, but differ from them in structure. 
They have no signs of circular rooms as the former, and, unlike both, 
have a peculiarly formed granary resembling a large inverted vase. 
The cliff houses reached their highest development, so far as va- 
riety of rooms and excellency of masonry goes, in the Mesa Verde, 
Colorado. These cliff" houses (pi. 4) are also distinguished from 
others in the Southwest by the form and character of specialized 
rooms. Here we find the most complicated form of specialized rooms, 
called kivas, with the best masonry and the most elaborate roofs. 
Moreover, a kiva of this form and structure is prehistoric, and practi- 
cally disappeared before the Avritten history of the Indians began. Its 
presence indicates the extent of one type of culture area, its absence 
another, and on the periphery where the two culture types overlap 
both become obscured or degenerated. 
Taking, then, the circular form of the kiva as a characteristic 
feature, we may say that the southwestern cliff house has two forms ; 
one with a kiva with a domed roof, the other with a flat roof; one 
constructed of regular horizontal, the other of irregular horizontal 
masonry; both forms were evolved from antecedent houses built of 
mud plastered on earth, or vertical undressed slabs of stone. 
A second great group of cliff houses in the Southwest is that in 
which no circular kiva exists; where, in fact, no kiva of any form 
has yet been detected (pi. 5). As the circular kiva is limited to a 
definite geographic area, so the kivaless cliff houses are likewise 
confined to others. So far as site goes, they are alike, but their 
masonry is poorer and the objects in them are different. All these 
differences point to another kind of cliff dwelling, that of southern 
Arizona and southern New Mexico. 
The deterioration of the circular kiva, or shall we say the arrested 
development of the same, appears in the region north of the Hopi, at 
Marsh Pass, and in the great ruin called " Kitseel " in the Navajo Na- 
tional Monument, Here the kiva loses the distinctive character of a 
vaulted roof and other features, still retaining, however, certain 
morphological elements, as the ceremonial floor opening, the venti- 
lator, and the fire screen. It is simpler in form, but it is still a cir- 
cular kiva. 
