576 EXPEDITION TO ARIZONA IN 1895 (ETH. ANN. 17 
that account unsafe, it may be stated provisionally that no circular 
kivas have yet been found in any ruins of the Gila-Salado watershed. 
This form of kiva, however, is an essential feature of the cliff dwellings 
of Rio Colorado, especially of those along its affluents in southern 
Colorado and northern New Mexico. Roughly speaking, then, the cir- 
cular kiva is characteristic of the ruins of this region and of certain 
others in the valley of the Rio Grande, where they still survive in 
inhabited pueblos. 
Circular ruins likewise are limited in their distribution in the South- 
west, and it is an interesting fact that the geographic distribution of 
ancient pueblos of this form is in a general way the same as that of 
circular kivas. There are, of course, maty exceptions, but so far as I 
know these can readily be explained. No ruins of circular dwellings 
occur in the Gila-Salado drainage area, where likewise no circular 
kivas have been observed. Moreover, the circular form of dwelling 
and kiva is distinctively characteristic of prehistoric peoples east of 
Tusayan, and the few instances of their occurrence on its eastern 
border can readily be explained as extra- Hopi. 
The explanation of these circular kivas advanced by Nordenskiéld 
and the Mindelefts, that they are survivals of round habitations of 
nomads, has much to commend it; but whether sufficient or not, the 
geographic limitation of these structures tells in favor of the absence 
of any considerable migration of the prehistoric peoples of the upper 
Colorado and Rio Grande watersheds southward into the drainage area 
of the Gila-Salado. Had the migration been in that direction it may 
readily be believed that the round kiva and the circular form of dwelling 
would have been brought with it. 
The round kiva has been regarded as a survival of the form of the 
original homes of the nomad, when he became a sedentary agriculturist 
by conquest and marriage. 
The presence of rectangular kivas in the same areas in which round 
kivas occur does not necessarily militate against this theory, nor does 
it oblige us to offer an explanation of a necessarily radical change in 
architecture if we would derive it from a circular form. It would 
indeed be very unusual to find such a change in a structure devoted to 
religious purposes where couservatism is so strong. The rectangular 
kiva is the ancient form, or rather the original form; the round kiva is 
not a development from it, but an introduction from an alien people. It 
never penetrated southward of the Colorado and upper Rio Grande 
drainage areas because the element which introduced it in the north 
was never strong enough to influence the house builders of the Gila- 
Salado and tributary valleys. 
