Academic journal article Southeastern Archaeology

A K-Means Analysis of Late Period Ceramic Variation in the Central Mississippi Valley

Academic journal article Southeastern Archaeology

A K-Means Analysis of Late Period Ceramic Variation in the Central Mississippi Valley

Article excerpt

K-means analysis is a nonhierarchical multivariate technique for recovering structure from datasets. K-means analysis of large ceramic assemblages from 39 late period sites in the central Mississippi Valley suggests that some previously defined phases represent analytically defensible units while others are much less robust. Some problems regarding the use of plainwares in grouping site assemblages are discussed.

The late period archeological phases (c. A.D. 1400-1600) of the central Mississippi Valley (Morse and Morse 1983; Figure 1) are among the best-known culture historical constructs in the Southeast. The Kent, Nodena, Parkin, and Walls phases, all located within this region, were singled out by Phillips (1970:930) as being "valid culture-historical units, if we may judge by the comparison of their pottery complexes." More recently, however, questions have been raised about the analytical validity of these and other existing phase formulations in the region (e.g., Fox 1992, 1998; Lewis 1990; Lipo 2001; Mainfort 1995, 1999, 2001a, 2001b; O'Brien 1994, 1995). Here I further investigate variation between late period ceramic assemblages in the central Mississippi Valley, which serve as the primary criterion for creating culture-historical groups (Phillips 1970; Smith 1990), using K-means analysis, a nonhierarchical clustering technique. The analysis presented here is a response to Phillips's (1970:923-924) call for a rigorous testing of his phases that moves beyond the largely intuitive method he used in formulating these units (Phillips 1970:523-524).

As was unfortunately the case at the time Phillips was writing, much of our knowledge about the sites under consideration derives from surface collections of potsherds. The situation is improving somewhat with the ongoing analysis of burial data and associated ceramic vessels from 1930s excavations in the study area (e.g., Brandon 1999; Brown 2002; Fisher-Carroll 2001a, 2001b; Fisher-Carroll and Mainfort 2000; Cannon 2002; Mainfort and Fisher-Carroll 1996).

Archaeological Background

Although some earlier studies included discussions of late period culture-historical units in the study area (e.g., Griffin 1952; Phillips, Ford, and Griffin 1951; Williams 1954), Phillips's (1970) landmark synthesis provides the baseline for late period phase definitions throughout the central Mississippi Valley. Phillips assigned most of the sites included in this analysis to the Nodena, Parkin, Kent, and Walls phases, which represent perhaps the best-known and best-documented late period culture-historical units in the lower Mississippi Valley (House 1991; D. Morse 1989, 1990; P. Morse 1981, 1990; see also O'Brien 1995). Phillips (1970), however, had access to data from virtually none of the western Tennessee sites discussed here and scarcely mentions this area. More recently, Smith (1990) proposed several late period phases that encompass the majority of the Tennessee sites used here and suggested new and/or modified phase definitions for other portions of the study area.

The Nodena phase includes a group of sites located primarily within the Mississippi River counties of northeastern Arkansas. Phillips (1970:933-936) states that "the outstanding feature" of this phase is "the low frequencies of Barton Incised and Parkin Punctated, which makes for a very different pattern from either the Parkin or Walls phases" (1970:935). Phillips (1970) included Carson Lake, Bell, Notgrass, Upper Nodena, Shawnee Village, Bradley, Golightly, and several other sites in this phase, while D. Morse (1989, 1990) attributes over 60 sites to the Nodena phase. Bradley, which actually includes at least three major late period sites (Morse and Morse 1983), has been interpreted as representing the capital of Pacaha, as recorded in the De Soto chronicles (Morse and Morse 1990). Only Upper Nodena, Notgrass, and Carson Lake have produced sufficiently large ceramic samples to be used here (Figure 1 and Table 1). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.