152 BUREAU OF AMERICAN ETHNOLOGY [Bull. 189 



wood; only an occasional piece of juniper was used. Probably the 

 use of Cottonwood was due to the fact that it grew readily at hand in 

 the nearby bottom lands, whereas juniper had to be transported from 

 a greater distance. The general practice probably was to choose the 

 material closer at hand. 



Within the floor area we found the same variation as to the place- 

 ment of pits. In Feature 22 there were a number of pits. Some were 

 undercut and bell shaped ; others were straight sided, and still others 

 were what we have called pocket caches, for they were small and shal- 

 low. All appeared to have been dug while the house was occupied, 

 for they all originated at floor level. In Feature 34, which was only 

 partially exposed, we found no floor pits outside the centrally placed 

 firepit. 



Lehmer (1954, p. 31) suggests that 



other holes in the floor may have contained wooden mortars after the fashion 

 of the historic Arikara and Mandan. These holes were cylindrical, somewhat 

 larger than the largest postholes, and were always located more or less on a 

 line between the firepit and the entrance. They were generally slightly offset 

 toward the entrance from a line connecting the eastern pair of primary roof 

 posts. 



Of all the structures investigated, none was burned. It would ap- 

 pear that they were abandoned and allowed to decay gradually, leav- 

 ing no tangible evidence as to the arrangement of the elements of 

 their superstructure. 



On the other hand. Feature 32 was characterized by a rectangular 

 pit of considerable size with rounded corners and outlined with small 

 individual posts more or less uniformly spaced. The west wall was 

 completely excavated, and also parts of the north and south walls. 

 Again, there was no evidence of the use of plaster on the walls of the 

 pit. The floor of native material was firmly packed. Entrance into 

 the structure was gained by a walled and roofed passageway from the 

 west-northwest, overlooking the river and the bottom land. There 

 was no evidence of any antechamber. Several hearth areas were un- 

 covered within the area of the structure. No arrangement of central 

 roof supports could be found. Whether all of these hearths were 

 coeval with the structure could not be determined, for there was some 

 indication that the whole area had been worked rather intensively by 

 the Indians after all traces of the structure had disappeared. There 

 were several large deep refuse and cache pits present in the area. 



The wall posts, when first found, appeared as small soft circular 

 spots along the periphery of the house floor. Wlien cleaned of their 

 contents they were found to be cylindrical with a rounded base aver- 

 aging 0.3 foot in diameter and 0.9 foot in depth. A number of them 

 did contain some midden material. 



