Prev--Paleoindian sites in Kansas || Next--Appendices
Paleoindian sites have been considered to be rare in the state of Kansas (see Logan and Brown, this volume). A stream-terrace analysis of the Neosho drainage suggests that paleoindian sites are relatively abundant on stream terraces of Wisconsinan age. Archeological sites with diagnostic paleoindian artifacts represented approximately 28% of all sites on the Wisconsinan terraces, and this figure should be considered a minimal estimate.
Kansas archeologists have been puzzled for many years by the relative lack of paleoindian sites in the state; whereas, a number of important paleoindian sites have been excavated in the neighboring states of Colorado, Wyoming, Nebraska, Missouri, and Oklahoma, and a significant number of paleoindian surface sites have been found in some of these areas. The only buried paleoindian site from Kansas is the 12 Mile Creek site, excavated in 1895 (Williston, 1902; Rogers and Martin, 1984). A few surface indications of a Kansas paleoindian presence have been found (Witty, 1964; Yaple, 1968; O'Brien, 1972, 1984; Reichart, 1972; Glover, 1974; and Rogers and Martin, 1982, 1983), but these sites represent only a very small proportion of the total number of prehistoric sites identified in the state.
An analysis of stream terraces was undertaken to see if a geomorphological explanation existed for the apparent lack of paleoindian material in Kansas. The area analyzed was the Neosho River drainage (Arkansas River drainage) in southeastern Kansas (fig. 1A). Because much of the archeological field work in the state of Kansas has been done in stream valleys, it was thought appropriate that stream-terrace analysis should be used to examine the cause of the scarcity of paleoindian sites in Kansas.
The author wishes to thank Wakefield Dort, Jr., L. D. Martin, and Edward Kost for their assistance.
The stream terraces of the Neosho River drainage of southeastern Kansas were surveyed for archeological sites. One hundred and one sites were found on four terraces. These terraces chronologically span the Wisconsinan and the Holocene. The floodplain (Terrace 0) and Terrace 1 were formed during the Holocene and contain artifacts from the Ceramic Period and the Archaic Period (including the early Archaic). These Holocene terraces contain the remains of Bison bison but not the remains of extinct Pleistocene megafauna. The remains of Mammuthus sp. and Equus sp. were recovered from both Terrace 2 and Terrace 3 but not the remains of Holocene Bison bison. Fluted projectile points were recovered from sites on Terrace 2. The presence of Pleistocene fauna and fluted points on Terrace 2, but not on Terrace 1, strongly indicates that the break between Terrace 2 and Terrace 1 marks the Wisconsinan-Holocene boundary. Additional support for this interpretation was the absence of Holocene Bison bison on Terrace 2 and the presence of early Holocene cultural material on Terrace 1. The heights of the terraces and what diagnostic taxa of fauna were found in the terraces are illustrated in fig. 1B. Terrace 4 was not intensively surveyed for archeological sites and therefore was not included in the analysis. The dental morphology of a mammoth molar recovered from Terrace 4 suggested a middle Pleistocene date. For further information on the dating, geomorphology, and other details of this terrace system, see Rogers (1984).
Figure 1--A) Map of Kansas illustrating extent of Neosho River drainage (solid area) and larger Arkansas River drainage (stippled area) of which Neosho River drainage is part. B) An idealized cross section of half a stream valley in Neosho River drainage, illustrating heights of terraces and typical fauna recovered from each terrace.
The inhabitants of the area at any particular time period could camp on the active floodplain or on any of the higher terraces but not on floodplain surfaces that had yet to form. Thus, older archeological sites will be found on or within higher terraces and not on or within lower terraces whose surfaces or fills had not come into existence at the time of occupation. This applies to both surface sites and buried sites. The break between the Wisconsinan and Holocene occurs between Terrace 2 and Terrace 1. This provides an opportunity to examine two different categories of archeologically relevant land surfaces: Terrace 2 and Terrace 3, which could have been occupied either during the Wisconsinan or the Holocene, and the floodplain (Terrace 0) and Terrace 1, which could have been occupied only during the Holocene.
A study of the paleoindian occupation in this terrace system obviously would focus on Terrace 2 and Terrace 3, the only terrace surfaces that would be relevant chronologically. This would be a stratified sample (Binford, 1964).
Diagnostic artifacts used to indicate the existence of a paleoindian component were fluted projectile points and spurred end scrapers. The fluted projectile points all were Clovis projectile points except for one Folsom projectile point, and all probably would date prior to 10,000 yrs B.P. Spurred end scrapers are a probable diagnostic paleoindian artifact type (Frison, 1978). ne number of sites examined on each terrace was 21 on the floodplain (Terrace 0), 51 on Terrace 1, 11 on Terrace 2, and 18 on Terrace 3.
Eight of the 101 archeological sites examined yielded either fluted projectile points, spurred end scrapers, or both. This is approximately 8% of the total number of sites. None of the sites with these diagnostic paleoindian artifacts was found on Terrace 1 or the floodplain. However, they represented approximately 45% of the sites on Terrace 2, approximately 17% of the sites on Terrace 3, and approximately 28% of the Terrace 2 and Terrace 3 sites combined. All fluted projectile points and spurred end scrapers were recovered from the surfaces of the sites. All were found on the scarps of the terraces, which leaves open the possibility that they may have been eroding from buried components in the terraces. Future excavation is needed to determine the positions of these artifact classes relative to the terrace fills. Terrace analysis can chronologically sort archeological data from either surface or buried sites. Additional information on the artifacts recovered from the sites found on the terraces can be found in Rogers (1984).
The difference in terrace distribution of sites with diagnostic paleoindian artifacts can be examined statistically. The Kolmogorov-Smirnov One Sample Test (Siegel, 1956, p.47-52) was used for statistical analysis. The actual proportion of sites on a terrace with diagnostic paleoindian artifacts is the number of such sites found on a terrace compared to the total number of such sites on all terraces. The expected proportion of sites on a terrace with diagnostic paleoindian artifacts (assuming random distribution) is the proportion of the number of sites on the terrace to the total number of sites on all terraces.
The actual proportion of sites with diagnostic paleoindian artifacts compared to the expected proportion for the floodplain was .0000 compared to .2079; for Terrace 1, .0000 compared to .5050; for Terrace 2, .6250 compared to .1089; and for Terrace 3, .3750 compared to .1782. These differences are statistically significant at the .05 level when analyzed by the Kolmogorov-Smirnov One Sample Test.
The absence of sites with diagnostic paleoindian artifacts on the floodplain (Terrace 0) and Terrace 1 is in agreement with the dating of these terraces as Holocene. These Holocene terraces comprise the bulk of the area of the stream valleys of the Neosho River drainage. This suggests that stream erosion during the Holocene destroyed much of the Wisconsinan land surfaces in the stream valleys where the paleoindian inhabitants would once have camped. This is particularly true of Terrace 2, which is not extensively preserved. Because stream valleys have been the focus of much of the archeological effort in the state of Kansas, it is not surprising that paleoindian sites have seemed to be rare, especially when concerted efforts were not made to locate and to survey Wisconsinan terraces.
The survey in this study indicated that 28% of the sites on the Wisconsinan terraces had paleoindian affinities. This should be considered a minimal estimate because many of the sites occupied by paleoindians may not have yielded diagnostic artifacts. Paleoindian sites would not seem to be rare in this area if the chronologically appropriate terraces were searched.
Particularly interesting is the observation that 45% of the sites on Terrace 2 yielded diagnostic paleoindian artifacts. Terrace 2 is composed of remnants of the floodplain that was active during the end of the Wisconsinan when populations known to have used fluted projectile points and spurred end scrapers were living in the area. Proportionately fewer sites (17%) with these diagnostic artifacts are found on Terrace 3. This suggests that paleoindians, when residing in these stream valleys, may have preferentially camped on the lowest terrace surface, perhaps to keep as near to water as possible. The limited technology that hunter-gatherers would have possessed to contain and to transport water would favor this interpretation.
Different problems present themselves when the paleoindian record outside the stream valleys is analyzed. Extensive Wisconsinan land surfaces may be preserved on the uplands. Unfortunately, almost all the upland areas in the Neosho River drainage are in pasture and any of the possible paleoindian sites there will tend to be "invisible" to standard archeological surveys.
Binford, L. R., 1964, A consideration of archeological research design: American Antiquity, v. 29, p. 425-441.
Frison, G. C., 1978, Prehistoric hunters of the High Plains: New York, Academic Press, 457 p.
Glover, G., 1974, An analysis of early paleoindian projectile points with new data from southwestern Kansas: M.S. thesis, Wichita State University, Wichita, Kansas, 111 p.
O'Brien, P. J., 1972, A clovis point from the Waterville, Kansas, area: Plains Anthropologist, v. 17, p. 60-64.
O'Brien, P. J., 1984, The Tim Adrian site (14NT604)--a Hell Gap quarry site in Norton County, Kansas: Plains Anthropologist, v. 29, p. 41-55.
Reichart, M., 1972, A Plainview-type point from the Delaware: Kansas Anthropological Newsletter, v. 18, p. 8-9.
Rogers, R. A., 1984, Kansas prehistory--an alluvial geomorphological perspective: Ph.D. dissertation, University of Kansas, Lawrence, Kansas, 321 p.
Rogers, R. A., and Martin, L. D., 1982, A Clovis projectile point from the Kansas River: Transactions of the Kansas Academy of Sciences, v. 85, p. 78-81.
Rogers, R. A., and Martin, L. D., 1983, American Indian artifacts from the Kansas River: Transactions of the Nebraska Academy of Sciences, v. 11, p. 13-18.
Rogers, R. A., and Martin, L. D., 1984, The 12 Mile Creek site-a reinvestigation: American Antiquity, v. 49, p. 757-764.
Siegel, S., 1956, Nonparametric statistics: New York, McGraw-Hill, 312 p.
Williston, S. W., 1902, An arrowhead found with bones of Bison occidentalis Lucas, in western Kansas: The American Geologist, v. 30, p. 313-315.
Witty, T. A., 1964, Appraisal of the archeological resources of the Perry reservoir, Jefferson County, Kansas: Kansas Anthropological Association Newsletter, September 1964.
Yaple, D. D., 1968, Preliminary research on the paleoindian occupation of Kansas: Kansas Anthropological Association Newsletter, v. 13, p. 1-7.
Prev--Paleoindian sites in Kansas || Next--Appendices
Kansas Geological Survey
Comments to firstname.lastname@example.org
Web version updated April 7, 2010. Original publication date 1987.