582 PROCEEDINGS OF THE NATIONAL MUSEUM vol. ss 



Despite the accumulating; wealth of archeological materials in 

 subsequent years, there has been a notable absence of recorded arti- 

 facts that could with certainty be regarded as tools for producing 

 indented pottery decoration. A brief search of the literature suggests 

 a tacit assumption by some archeologists that a notched wheellike 

 object was used, as when they speak of ''roulette" impressions (e. g., 

 McKern, 1931; Setzler, 1933; Cole and Deuel, 1937). Others, per- 

 haps with less faith in Holmes' experiments, have preferred a less 

 committal term such as "curved dentate stamp," "notched rocker," 

 etc. Willoughby (1922, p. 92), describing pottery from the Turner 

 mound group, speaks of "zigzag patterns which were not made with 

 a roulette, as suggested by Holmes, but with a tool more or less gouge- 

 shaped, having a plain or notched edge, which was pressed against the 

 soft clay with a rocking motion, each opposite corner being raised 

 and slightly advanced alternately, the tool not being wholly lifted 

 from the vessel." 



All these various terms and interpretations suggest devices by 

 which clay vessels could have been ornamented, but so far as we are 

 aware none rests on incontrovertible archeological evidence derived 

 from the finding of the envisioned implements. More recently it has 

 been demonstrated in the laboratory that certain pottery markings 

 found at Marksville, La., in all probability represent impressions from 

 the edge of a bivalve marine mollusk of the scallop or pecten family 

 (Setzler, 1934, fig. 44, middle row; and unpublished notes). Fewkes 

 (1937, p. 148) has described a flat elliptical end-notched stone object 

 from Minnesota "that is unquestionably a stone stamp used for 

 imprinting the roulette design on unfired clay." The same writer, 

 scouting Holmes' theory of a roulette because of lack of supporting 

 evidence, offers no alternative explanation for the even continuous 

 lines of indented impressions that sometimes encircle, or partially 

 cover the decorated surfaces of, vessels from the Hopewellian area. 



In 1939 the junior writer was engaged in excavations at a small pre- 

 historic village site about a mile south of the Missouri River near 

 Bethel, Wyandotte County, Kans. Occupational debris, including 

 sherds of varying types, worked and unworked flints, bone artifacts, 

 limestone fragments, broken animal bones, charcoal, and burnt clay, 

 occurs here in a dark horizon about 18 inches thick and in circular 

 pits, overlain by about 22 inches of culturally barren colluvial soils. 

 Up to the present time digging has been confined to the gullied margin 

 of the site overlooking a small unnamed creek. Owing to the thickness 

 of overburden, modern cultivation has not as yet disturbed the greater 

 portion of the site along the creek bank. In most particulars the 

 remains so far foimd appear to parallel others recovered by the United 

 States National Museum in 1937 at the Renner site 5 or 6 miles to 



