June 25, 2008

Middle Woodland Ceramic Exchange in the Lower Illinois Valley

By Fie, Shannon M

Abstract This paper examines the exchange of Middle Woodland ceramics within the Havana region of the Hopewell Interaction Sphere. Ceramics from six lower Illinois Valley sites and samples of surrounding days are examined using neutron activation analysis. Statistical evaluation of the elemental data reveals the presence of foreign ceramics in all six site samples. The local ceramics encompass a variety of both fine and coarse wares, including several sherds initially anticipated as exotic based on style. In contrast, the foreign sherds consist almost exclusively of coarse tempered, utilitarian types. Contrary to conventional wisdom, these data suggest that fine tempered, special purpose vessels were manufactured near their place of discovery, while utilitarian pottery may have circulated among Middle Woodland communities. In documenting the presence of foreign sherds in all of the study samples, this study confirms the regular transport of ceramic vessels into and within the lower Illinois Valley region. Introduction

Exchange is an invaluable tool in modeling prehistoric social interaction. Trade goods provide an important chronicle of past economic, social, and political connections that linked groups both within and between regions. Such interactions may account for a variety of changes within local cultural systems, ranging from the incorporation of minor traits to substantial reorganization of socioeconomic systems. It is not surprising, then, that Middle Woodland research has focused on the interregional circulation of goods, materials, and concepts recognized as the Hopewell Interaction Sphere.

As it is currently understood, the Hopewell Interaction Sphere consisted of a variable lattice of relationships that connected numerous cultural traditions through the exchange of highly visible raw materials, goods, and symbols. In the Havana region, the long distance circulation of Hopewell goods was accompanied by increased interaction among local groups as evidenced by the widespread elaboration of ceramic decoration (Braun 1991 ), as well as the use of common mortuary programs (Bullington 1988; Charles 1992:190- 192). Within these regions, the introduction of Hopewell goods and ceremonialism corresponded with increased sedentism and a greater reliance on intensive resource gathering and plant cultivation. Accordingly, explanations for the acquisition and exchange of these valued Hopewell goods relate them to negotiations within the regional social system.

Unfortunately, it has proven difficult to relate these two system scales. The exchange of exotics between different regions (interregional exchange) can often be identified and traced back to a few likely source areas. Comparable transactions conducted within regions (intraregional exchange) are more difficult to detect. In the case of Hopewell goods, these items are typically infrequent and varied, and their incidence rarely supports distributional study. At the same time, more abundant regional materials often lack sufficient contrast to permit determinations of exchange. Thus, models of Middle Woodland interaction rely heavily on the presence of Hopewell exotics to infer patterns of intraregional interaction, and the consequent integration of groups into larger social units. Within most regions, however, the frequency of Hopewell goods appears inadequate to support models of frequent or even periodic interaction. If local networks relied on exchange to sustain social relationships, these transactions likely included mundane regional commodities and the participation of the broader social group.

This paper examines Middle Woodland ceramic exchange within one region of the Hopewell Interaction Sphere. Ceramic sherds from six lower Illinois Valley sites and samples of regional days are examined using neutron activation analysis. Statistical evaluation of these elemental data reveals two intriguing patterns. First local sherds represent an array of wares and types, many of which were initially thought to be exotic based on stylistic properties. In contrast, sherds that do not source to either the site or regional groupings include a surprising number of utilitarian types. Contrary to conventional wisdom, these data suggest that many of the ritual vessels were manufactured at or near their place of discovery, while utilitarian pottery may have circulated within and between Havana regions. This study thus documents the presence of nonlocal ceramics in all of the study samples, and confirms the regular transport of vessels into and within the lower Valley region.


As a core area of the Hopewell Interaction Sphere, the Illinois Valley has been a focus of Middle Woodland research for much of the twentieth century. Research into Middle Woodland adaptations in the lower Illinois Valley developed through the activities of such scholars as James Griffin, Gregory Perino, John McGregor, and Stuart Struever. Investigations into the prehistory of the lower Illinois Valley region progressed with the formation of the Illinois Valley Archeological Program in 1960 (Struever 1960:v); support for archaeological research in this region continues through the efforts of the Center for American Archeology. This research documents more than 150 Middle Woodland sites within the lower Illinois Valley region.

The lower Illinois Valley region offered a particularly rich setting for Middle Woodland occupation. This region exhibits a striking topography consisting of a broad floodplain, steep bluffs, and heavily dissected uplands. The diverse landforms, along with variations in sediment drainage and microclimate, result in concentrations of plant and animal resources within a relatively confined corridor along the southern Illinois River. This rich environment provided Middle Woodland groups with a reliable resource base, and the high seasonal densities of migratory fowl, nuts and acorns, fish, deer, and seeds allowed for more concentrated populations within the major valleys (Asch 1976:63; Asch et al. 1979; Struever 1968a:295, 1968b:195-197).

Essentially vacant of human occupation throughout much of the Early Woodland period, the lower Illinois Valley experienced a substantial influx of people during the Middle Woodland (Farnsworth 1986:639-640; Farnsworth and Asch 1986:446-447). Once introduced into this area, settlement concentrated along the valley margins. Positioned near the entrance of tributary streams, these locales afforded access to all major foodstuffs (Struever 1968a:307, 1968b:215). Occupied much of the year by a handful of extended family households, these habitation sites were the primary location to which resources were conveyed, processed, and stored. Even with the likelihood of differential access and storage, base settlements exhibit little evidence of differential ranking or organization above the household level (Braun 1991:370).

This pattern stands in marked contrast to the complex expressions of social differentiation manifested in Middle Woodland mortuary contexts. Less elaborate than their Ohio counterparts, Illinois Middle Woodland mounds nevertheless evidence a complex program of mortuary ceremony which served to reinforce social integration among neighboring groups. Access to mound burial was apparently conferred on all members of the local community, although a few individuals received preferential treatment. While no obvious badges of position are evident, social ranking is suggested by the placement of some individuals (usually males) within the central tomb and their frequent association with exotic goods (Brown 1981; Buikstra 1976; Carr 2006).

The exotic goods used to express social differences entered the area via activities conducted periodically at transaction centers.1 Distributed along the Illinois River floodplain, these mound groups functioned as nodes in the receipt, concentration, manufacture, and redistribution of goods both within and between regions (Struever 1968a:308; Struever and Houart 1972:61). While securing access to important status-related goods, seasonal participation in political, economic, and ceremonial activities also served to integrate individual communities into larger sociopolitical units.

The functional differentiation of habitation and floodplain mound groups provides a rare opportunity to evaluate possible explanations for the presence of foreign materials. Within the lower Illinois Valley subsistence-settlement system, exchange interactions are more likely to occur at some site types than others. Base settlements, for example, functioned as the primary domestic residence (see Struever 1968a:295-297) for a handful of extended families (Braun 1987:167, 1991:369-370), whereas floodplain mound groups served as important locations for periodic intergroup ceremonies and exchange activities (Struever 1968a:300-308; Struever and Houart 1972:60- 64). Even though the role of transaction centers in exchange has yet to be demonstrated, comparative site studies document significant differences between these floodplain mound groups and base settlements. Whether maintained as a discrete site type or included as a subset of ritual camps, transaction centers are readily distinguishable from base settlements in their location, content, and internal structure (Cowan and Fie 1993:5; McGimsey and Wiant 1986:536). The basic distinction in site types is therefore incorporated into the current research design as a means of structuring possible interpretations of nonlocal material distributions. Ceramics constitute an unparalleled data set for understanding Middle Woodland adaptations in the Illinois Valley. Differences in ceramic types and attributes support fundamental reconstructions of the Middle Woodland culture sequence (Bluhm 1951; Cantwell 1980; Cole and Deuel 1937; Epstein 1958; Fowler 1952, 1955; Griffin 1952a, 1952b; McGregor 1952, 1958; Morgan 1985, 1986; Morse 1970; Powell 1957; Struever 1960, 1961; Taylorl958a, 1958b; Wray and MacNeish 1961). Yet studies of ceramic types and distributions also inform a host of research interests ranging from subsistence practices and settlement patterning (e.g., Braun 1983, 1985a, 1987; Struever 1965, 1968a) to social organization and interaction (Braun 1977, 1985b; Braun and Plog 1982; Houart 1973, 1975; Loy 1968; Morgan 1986; Neiman 1995). The importance of ceramics in Middle Woodland studies results, in large part, from their considerable abundance and diversity. Middle Woodland ceramics encompass a substantial range of variation in both decorative and technological traits. While supporting regional chronologies, these differences implicate functional differentiation in the use of vessels (e.g., Braun 1977:174, 1983; Griffin 1952a:106; Struever 1968b:162-163), as well as local and regional preferences in their manufacture and decoration (e.g., Loy 1968; Struever 1964, 1965).

Research Design and Methods

The identification of prehistoric exchange inevitably prompts speculation as to the purposes, functions, and outcomes of intergroup contact. Exchange, whether within or between societies, may occur for a variety of reasons including resource redistribution, alliance formation, and sociopolitical maneuvering. Researchers have invested considerable efforts in the development of methods and models to better understand the role of exchange in Middle Woodland interaction (Boszhardt and Gundersen 2003; Brose 1990; Carr and Komorowski 1995; Clark 1984; Emerson et al. 2005; Glascock 2005; Griffin et al. 1969; Hatch et al. 1990; Hughes 2006; Hughes et al. 1998; Morrow 1988; Ruby and Shriner 2006; Ruhl 2006; seeman 1979a, 1979b; Spence and Fryer 2006; Stevenson et al. 2004; Struever and Houart 1972; Walthall 1981).

Reconstructions of exchange interaction ultimately rely on the commodities believed to circulate and their presumed source. The recognition of foreign materials provides the basis for identifying trade, as well as information as to the distance moved, the parties involved, and the intensity of the interaction. Establishing this interaction requires both documenting the transport of goods within the region, and demonstrating that the transport of these goods results from exchange interaction.

For Middle Woodland groups, models of exchange traditionally focus on the highly visible exotic goods and raw materials associated with the Hopewell Interaction Sphere. Between regions, the display of Hopewell Interaction Sphere goods likely facilitated communication between socially distant parties (Braun and Plog 1982:510-511; Braun et al. 1982:63; seeman 1995). The stimulus for their procurement, however, originated within regional social systems where these objects were employed to create and maintain social alliances (Braun 1977:94; Brose 1979:7; Ford 1979:237-238), reinforce status (Struever and Houart 1972:49), and/or validate positions of leadership (Bender 1985:55; seeman 1992:8).

Yet, when the actual frequencies are examined, the quantities of Hopewell commodities are small in most contexts. Only within the Ohio earthwork complexes are conspicuous deposits of supralocal materials occasionally encountered. Within most regional traditions, such as Havana, exotics constitute the minority of mortuary accouterments while in others, such as Crab Orchard, they are rare or absent. These sporadic distributions of exotics limit the inferences that can be drawn.

Within regions, the distribution of exotics has still less explanatory power. Not only are Hopewell items infrequent, they originate outside the sphere of interaction, leaving the participants obscured. Within Illinois Havana contexts, more proximate materials, such as Cobden chert, are far more prevalent, and their distributions afford greater potential for delineating exchange within small networks. Yet Cobden chert also originates outside the intraregional sphere of interaction. The delineation of exchange within a region thus requires a commodity that circulated among local groups, but can also be sourced to areas within that region.

Ceramics remain a preferred artifact class in studies of Middle Woodland interaction (Houart 1973, 1975; Loy 1968; Morgan 1985, 1986). Yet, they are only rarely considered in investigations of exchange. Much of this oversight stems from difficulties in distinguishing foreign sherds from local imitations. Assuming such challenges can be overcome, ceramics provide several advantages over other artifact classes for studies of intraregional exchange: they are abundant artifacts, their use is not socially restricted, and as short lived tools, they presumably pass through few transactions. The difficulty, of course, is in finding a technique capable of distinguishing foreign sherds at the intraregional scale of analysis.

The identification of foreign ceramics requires a technique capable of differentiating materials that originate within restricted geographic areas. Previous identifications of such pottery in Middle Woodland assemblages rely heavily on the visual comparisons of decoration, form, and paste (Braun etal. 1982:62-63; Epstein 1958:37-40; Griffin et al. 1970:70-78; Stoltman 1979:135; Winters 1967:50-52). Often difficult to replicate, visual identifications are further complicated by possible local imitations of foreign styles. Such difficulties may be mitigated through the application of techniques that utilize mineral or elemental content to identify foreign materials.

Previous investigations of western Illinois days reveal considerable variability within the Illinois Valley regions. Compositional differences are reported in the percentage of major minerals including kaolinite, illite, montmorillonite, and chlorite (Parham and White 1963; White 1963; White and Lamar 1960; White and O'Brien 1964), and in the content of heavy minerals such as zircon, tourmaline, garnet, rutile, sillimantile, kyanite, hornblende, and epidote (White and O'Brien 1964:10-11). These differences occur both between formations (Frye et al. 1964; White 1963; White and O'Brien 1964) and as directional trends within particular formations (Parham and White 1963; White 1963; White and O'Brien 1964). Vessels manufactured from these different materials should reflect their respective mineral and/or elemental content. Thus, ceramics manufactured in the lower Illinois Valley are potentially sourceable to areas within the sphere of interaction.

To date, only a handful of studies employ compositional data to identify foreign ceramics in Midwest assemblages. Although still few in number, these studies document both mineral (Carr and Komorowski 1995; Ruby and Shriner 2006; Stoltman 1991; Stoltman and Mainfort 2002) and elemental (Carr and Komorowski 1995; Cogswell et al. 1995; Cogswell et al. 1998; Elam et al. 1992; Neff et al. 1994; Yeatts 1990) differences among archaeological assemblages, and attest to the considerable potential of compositional analysis as a means of sourcing ceramic materials. The geological variability present within the lower Illinois Valley suggests that comparable success is possible within this region.

The transport of ceramics does not, however, necessarily indicate that ceramics were exchanged. The occurrence of foreign or nonlocal vessels may result from several different behaviors related to residential mobility, intergroup visitation and/or marriage. A determination of exchange therefore requires a means of eliminating these alternative accounts. The current study uses functionally distinct sites to help distinguish vessels resulting from foreign visitation to social ritual centers from those purposefully obtained through exchange.

For the current study, trace elements offer the most promise as a means of distinguishing separate source areas within the lower Illinois Valley region. Elemental data provide important evidence of variation occurring in days that cannot be detected through mineralogical analysis (Stewart et al. 1990:620621). Many trace elements-but especially the rare-earths and actinides-are fixed within the crystalline structure of minerals. With weathering, these elements concentrate in the finer particle fraction and deplete in the coarser silts and sands, which makes them particularly useful in differentiating clay sources (Bishop et al. 1982:294-295; Elam et al. 1992:95). Moreover, most trace elements do not reflect significant weathering or substitution effects (Bishop et al. 1982:296), nor do they undergo significant changes with firing temperatures below 850[degrees] C (Rice 1987:423).

For the current study, instrumental neutron activation analysis (INAA) provides several advantages over other techniques. First, neutron activation analysis is a highly sensitive technique capable of simultaneously measuring numerous trace elements, as well as major elements. Thus the technique provides data appropriate to the discrimination of localized day deposits. In contrast to other available techniques, the suitability of INAA for delimiting ceramic source areas in the Midwest is already established (Clark et al. 1992; Elam et al. 1992; Mainfort et al. 1997).

Figure 1. The study sites.

The technique employed in an analysis of archaeological materials inevitably imposes some conditions on the selection of samples. Most compositional analyses are costly and, consequently, limited to a relatively small number of samples. Sampling is therefore directed at specimens with the greatest potential to explain patterning in the compositional data. Yet, the targeting of individual samples for analysis compromises any inferences that can be made regarding the frequency of foreign materials. Instead, arguments must be structured around their prevalence in specific artifact classes or contexts. For the current study, the occurrence of foreign ceramics within functionally distinct site assemblages provides the interpretive framework. The selection of site assemblages in the current analysis includes examples of base settlements-Macoupin, Apple Creek, Sandy Creek Church, and Smiling Dan-and transaction centers Peisker and Mound House (Figure 1). These contrasting occupational contexts provide an important backdrop against which possible explanations for the occurrence of foreign ceramics may be assessed. In all, 313 sherds from these six sites are included in the analysis. Ceramic decoration provides an opportunity to compare groups and outliers with previously identified distributions of styles. Sample selection therefore emphasizes large decorated sherds representing the ceramic wares regularly recovered at lower Illinois Valley sites, namely Havana, Hopewell, Pike, and Baehr. Thus, the resulting collection of sherds encompasses a considerable amount of variation in both decorative and technological traits.

The bulk of the ceramic samples consist of Havana and Havana- related sherds (Table 1). These 188 samples include examples of several well established Havana types, such as Havana Zoned, Naples Ovoid Stamped, Naples Stamped, Hummel Stamped, Neteler Stamped, Steuben Punctated, Havana Cordmarked, and Havana Plain. Pike-a second, more localized utilitarian ware found in the lower Valley (Struever 1968b:7) - is represented by 53 sherds. Ritual or special purpose vessels are represented by examples of Hopewell and the more localized Baehr ware. The Hopewell sherds include examples of several types, including Hopewell Zoned Dentate Stamped, Montezuma Punctated, Bluffdale Plain Rockered, Grigsby Dentate Rockered, and even a Brangenburg Rim.

Table 1. Breakdown of analyzed sherd samples by site and ware (Fie 2006:Table 23.1).

The compositional analysis also incorporates less common Woodland ceramics. These potential exotics comprise nine Middle Woodland FabricImpressed, a Simple-Stamped rim, and two Middle Woodland Type Indeterminate. Finally, a handful of Early Woodland and Late Woodland sherds are represented in the analysis. These ceramics consist of one Sister Creeks Punctated sherd, one Marion Thick sherd, two Schafner Pinched sherds, and two Liverpool/Black Sand sherds.

Figure 2. clay sample locations. Circles represent seven km radius around each habitation site.

The identification of foreign ceramics within these assemblages necessarily relies on contrasts with material identified with areas of the lower Illinois Valley. Following the "Criterion of Abundance" local reference groups are typically generated from sherd data (Bishop et al. 1982:301); however, such groups are subject to biases in the selection of samples. Ideally, determinations of local and foreign ceramics include comparisons with geologic samples; such samples provide a known of local day composition and variability.

For the current study, collections from 82 day deposits provide a comparative baseline of the days available within this region of western Illinois (Figure 2). These samples provide an important gauge of the compositional variability present in the lower Illinois Valley days. Additionally, day samples collected within approximately seven km of the habitation sites provide an important signature of local raw materials (e.g., Arnold 1985): Macoupin (SFC- 420, 422, 437, 447, 448, and 482), Apple Creek (SFC-401, 424, 426, 427, and 475), Sandy Creek Church (SFC-407, 450, and 451), and Smiling Dan (SFC-410, 452, 453, and 454).

Table 2. Clay sample contexts.

The clay collection universe consists of a 3600 km^sup 2^ area centering on the Illinois River drainage between the Spoon River and Illinois-Mississippi river confluence, and includes parts of Fulton, Schuyler, Cass, Menard, Brown, Morgan, Pike, Scott, Greene, Calhoun, Jersey, and Madison counties; sampling of the Mississippi drainage includes only a few deposits in southern Calhoun County. The majority of exposed clays occur within 15 km of the Illinois River, near the entrance of streams into the Illinois Valley trench.

The clay collection emphasized outcrops and developing surficial materials (Table 2). Several samples of alluvium were also included, but these were limited to tributary stream deposits. While extensive flats of river muds typically blanket the Illinois and Mississippi river floodplains, no suitable samples of such materials were obtainable at the time of the survey.2 The absence of these alluvial deposits from the major river systems represents an unfortunate shortcoming in the current assessment of the regional days. While far from exhaustive, these samples nevertheless represent a variety of depositional contexts, including ancient marine sediments, weathering surface sediments, redeposited surface sediments, alluvium, unknown materials, and one archaeological sample.3

Figure 3. Biplot of chromium and thorium concentrations for the sherd and day samples. The ellipse represents a 90% confidence interval for the group.

Analysis of the Ceramic Data

Neutron activation analysis yielded concentrations of 33 elements for the 313 sherds and 82 day samples. Yet not all of these elements are useful in identifying meaningful compositional groups. For nickel (Ni) and strontium (Sr), concentrations in ceramic materials often approach the minimal limits of detection and result in a significant number of missing values (Glascock 1992:19). At the same time, the presence of limestone or shell temper may elevate concentrations of calcium (Ca) (Cogswell et al. 1998; Steponaitis et al. 1996), whereas concentrations of hafnium (Hf) and zirconium (Zr) tend to correlate with the silt fraction of the paste (Blackman 1992:118; Elam et al. 1992:101; Neff et al. 1992:67). Lastly, elements such as barium (Ba), cesium (Cs), potassium (K), rubidium (Rb), manganese (Mn), sodium (Na), and uranium (U) are considered "mobile" and thus susceptible to postdepositional alteration (Bishop et al. 1982; Mainfort et al. 1997; Stewart et al. 1990; Stoltman and Mainfort 2002:27). In all, 13 elements are omitted from the analysis due to erratic detection (As, Ni, Sr), elemental enrichment (Ca), variable silt content (Hf, Zr), or the likelihood of post- depositional alteration (Ba, Cs, Mn, K, Rb, Na, and U). Concentrations for the remaining 20 elements are transformed to log^sub 10^ concentrations to compensate for differences in magnitude between major and trace elements.

Figure 4. Biplot of the first two principal components of the lower Illinois Valley habitation site sherds minus outliers. Principal components based on the variance-covariance matrix of 20 elements. Ellipses represents a 90% confidence interval for the groups.

Neutron activation analysis generates large data matrices. The identification of groupings within elemental data is typically achieved through a series of pattern recognition and group evaluation analyses. The current study utilizes a combination of principal components analysis (PC) and Mahalanobis distance analysis (see Glascock 1992:16-25). These analyses are performed using several computer programs written specifically to identify groups within compositional data4 (see Fie 2000).

The presence of ceramic outliers is suggested in the biplots of several logged elemental concentrations (Figure 3). However, interpretation of the compositional date requires a means of objectively evaluating group tendencies within the data matrix. Principal components analysis finds that several of the sherds exhibit only marginal association with the individual habitation site samples and their associated local clay samples; the habitation sherds as a group and their associated local days; the day samples; or a combined group consisting of the habitation site samples and all 82 days. The composition of these sherds differs from the general pattern of the lower Illinois Valley, and they likely represent vessels produced outside the study area (Fie 2006). These samples are not immediately relevant to the identification of source groups within the lower Illinois Valley. However, their indusion may adversely affect the multivariate analysis by extending the distribution of samples in particular directions and cause samples to appear more similar to each other relative to the outliers (Mainfort et al. 1997:47). Twenty-four sherds are initially omitted as provisional outliers: SFS-028, SFS-037, SFS-043, SFS-065, SFS- 066, SFS-114, SFS-125, SFS-126, SFS-148, SFS-155, SFS-161, SFS-170, SFS-176, SFS-179, SFS-200, SFS-226, SFS-227, SFS-233, SFS-240, SFS- 242, SFS-243, SFS-251, SFS-284, SFS-313. The principal components are then recalculated for the remaining sherd samples.

Table 3. Mean elemental concentrations for the lower Illinois Valley habitation site samples minus outliers.

Biplots of the remaining habitation site samples reveal considerable overlap among these four assemblages (Figure 4). Yet patterns, such as directional trending in the elemental data, suggest that source areas may be distinguishable. For example, the mean concentrations (Table 3) of several rare-earth elements- antimony, dysprosium, europium, lutetium, and samarium-are lowest in the southernmost samples (Macoupin) and generally increase in assemblages to the north. Conversely, the mean concentrations of cobalt and vanadium-both transitional metals-are highest at Macoupin.

With the outliers removed, the probability of membership in each of the four habitation groups is calculated using Mahalanobis distance analysis of all 20 principal components. It is important to underscore that, while the radiocarbon carbon dates overlap among these four Middle Woodland sites (see Fie 2000:Appendix B), they are not assumed to be contemporaneous. For this study, these assemblages serve as reference groups for particular stretches of the lower Illinois Valley. Attempts to refine the four habitation site groups met with only minimal success. The overlap in elemental concentrations and subsequent principal component calculations often produced near-comparable membership probabilities for multiple site groups. Refinement of these groups by removing samples with ambiguous classifications (Bishop et al. 1988:321; Neff et al. 1988:342-343) eventually led to the omission of a significant number of samples. Stoltman and Mainfort (2002:21-22) report a similar difficulty in the analysis of the Pinson Mounds NAA data, and question the effects such attenuation of samples may have on the resulting patterns. Rather than excluding much of the diversity that characterizes the lower Valley Middle Woodland ceramics, these groups are left intact for the posterior classification of the transaction centers and ceramic outliers previously set aside.

The assignment of individual sherds to reference groups is based on higher membership probabilities. Among the current data, sherds produce near comparable p-values within several reference groups. Thus, stronger membership probability in a reference group other than the parent group does not necessarily warrant its reassignment. The strategy utilized here relies on a low probability of membership (

Table 4. Membership probabilities for the compositional outliers.

Table 5. Membership probabilities for the nonlocal sherds identified in the habitation site samples.

Samples exhibiting low membership probabilities in all of the habitation reference groups (p

Not surprisingly, most of the habitation site sherds generate the greatest probability of membership in their parent group. For many sherds, group membership probabilities are comparable in a neighboring reference group. At the same time, several sherds exhibit considerably greater probability of membership in a group other than their parent site group (Table 5). For example, SFS-004, a Naples Dentate Stamped rim yields a membership probability of less than 1.0% in its parent site of Macoupin, but a considerably higher probability of membership in the Sk group (p=11.9%). In the case of several Smiling Dan samples (i.e., SFS-266, 282, and 295), membership probabilities are higher in multiple groups, suggesting a general southern affiliation. Several additional samples-including SFS-Ol 6, 022, and 106-yield only minimal membership probability within a habitation site group (

Both of the transaction centers are situated near one of the analyzed habitation sites. If these floodplain mound sites served to integrate local communities (Struever and Houart 1972:61), much of the ceramic assemblage should assodate with a proximate habitation group. In other words, the membership probabilities for the Mound House samples should produce greater membership probabilities in the Sandy Creek Church (Sk) and Apple Creek (Ap) groups, while the Peisker samples should generate greater membership probabilities in the Macoupin (Mp) group.

The resulting membership probabilities for the Mound House samples (minus outliers) generally support this expectation with 19 of the remaining 31 sherds yielding their highest membership probabilities in the Sk group. Of the remaining twelve samples, nine produce p-values of greater than 40% in Sk group, but somewhat higher membership values in another habitation site group. For example, the p-values for SFS-234, a Havana Plain rim, are 43.1% in the Sk group, but 48.6% in the Mp group, 5.0% in Ap, and 11.8% in the SMD group. At the same time, the remaining two sherds-SFS-221 and SFS-253-produce p-values suggesting a source elsewhere in the valley (Table 6). A third sherd, SFS-233, exhibits only minimal affiliation with any of the four base settlements or the LIV group, but a low affiliation with the lower valley clays.

Table 6. Membership probabilities for the nonlocal sherds identified in the transaction center samples.

The membership probabilities for the Peisker samples suggest a greater representation of source groups than Mound House. Comparable membership probabilities in multiple groups preclude the determination of a likely source area for several of the analyzed sherds. At the same time, they underscore those sherds that fall outside the compositional profile generated for their local habitation site. Four samples (SFS-142, SFS-144, SFS-165, and SFS- 169) exhibit very low affiliation with the Mp group, but substantially higher p-values in the Ap or Sk groups. Two additional samples (SFS-123 and SFS-146), also exhibit a stronger affinity for the Sk group even while generating somewhat slightly higher p- values in the Mp group. The affiliation of sherd SFS-167 is less obvious, as it generates comparable p-values in all three of the habitation groups north of Macoupin. Lastly, the membership probabilities generated for sherds SFS-124, SFS-126, SFS-137, and SFS-181 suggest little relationship to the habitation group samples, but some similarity for the lower Illinois Valley days.

Table 7. Sherd assignments for the six site samples.

Statistical analysis of the compositional data finds that most sherds do not differ significantly from the ceramic samples that comprise their parent or proximate habitation group. Of the 313 analyzed sherds, 253 samples remain assigned to their parent habitation site group or, in the case of transaction centers, the nearest habitation group (Table 7). Notably, these specimens include a diversity of ceramic wares and types, many originally believed to be exotic. Havana types dominate the local ceramics: Havana Zoned, Naples Dentate Stamped, Naples Cordwrapped-Stick Stamped, Hummel Dentate Stamped, Hummel Plain Stamped, Naples Ovoid Stamped, Neteler Dentate Stamped, Neteler Plain Stamped, Havana Cordmarked, and Havana Plain. Also included among the local ceramics are 19 of the 21 Hopewell sherds; all 20 of the Baehr samples, and 49 of the original 53 Pike and Pikerelated ceramics. Somewhat surprisingly, several of the compositionally local sherds represent types uncommon in the lower Valley: six of the nine Middle Woodland Fabric- Impressed, the lone Simple Stamped sherd, and both Middle Woodland Type Indeterminate.

In contrast to the local ceramics, a number of sherds exhibit only weak affiliation with their parent site while generating a substantially stronger probability of membership in another habitation site group, the day group, or a combined sherd-clay group. These sherds are therefore regarded as nonlocal to their respective assemblages and the possible products of intraregional transport.

Fifteen of the nonlocal sherds are reassigned to the Sk group. Further to the south, both the Ap and Mp group attract one and twelve, respectively. To the north, the SMD group attracts only one nonlocal sherd from another site. Eight sherds produce only low probabilities of membership within any of the four habitation site groups, but show substantially stronger values within the days or the combined lower Illinois Valley group. Unfortunately, the relationship of these samples to the analyzed sites or the lower Illinois Valley cannot be determined with the current data set. However, future analysis of the day samples may permit more accurate sourcing of these sherds.

Most of the reassigned samples represent common Havana types, such as Naples Dentate Stamped, Havana Zoned, Havana Cordmarked, and Havana Plain. The few exceptions include a Liverpool Incised (SFS- 179), a Neteler Dentate Stamped (SFS-266), and a Brangenburg Rim (SFS-218). Additionally, one fabric-impressed sherd (SFS-022), a Liverpool Cordmarked sherd (SFS-016), and the Sister Creeks Punctated (SFS-181) produce only minimal probabilities of membership within any of the four habitation groups. However, comparison to the day group suggests some affiliation with the lower Illinois Valley region.

Twenty-three sherds exhibit only minimal affiliation with any of the habitation reference groups, the clays or the combined LIV group. Contrary to conventional expectations, most of these outliers represent utilitarian ceramics. These spedmens indude several familiar Havana types: Naples Cordwrapped-Stick Stamped, Naples Dentate Stamped, Hummel Dentate Stamped, and Steuben Punctated, and Havana Plain. Also counted among the outliers are several types less commonly reported in lower Illinois Valley assemblages, such as Naples Ovoid Stamped, Neteler Plain Stamped, and Middle Woodland Fabric-Impressed.

Of these outliers, one Steuben Punctated rim (SFS-065) can confidently be sourced to a region outside the lower Illinois Valley based on its strong membership probabilities in the upper Illinois Valley region (Fie 2000:440). The remaining outliers also represent compositional profiles that contrast with the general patterns defined for the lower Illinois Valley. Thus, while only one of these sherds currently can be assigned to a region outside the lower Illinois Valley, the remaining samples are also considered likely products of interregional transport (Fie 2006). For occupants of the lower Illinois Valley, therefore, ceramic transport almost exclusively involved utilitarian vessels; that is Havana and Pike wares. In contrast, of the small ritual ceramics, only a Hopewell Montezuma Punctated (SFS-123) and the Brangenburg Rim (SFS-218) suggests an origin elsewhere in the lower Illinois Valley.

These observations may challenge previous notions of vessel transport, but they conform nicely to the reported distribution of these ceramic wares. For example, the production of limestone- tempered Hopewell vessels is identified with the lower Illinois Valley (Griffin 1952b:15) and, specifically with Scott, Pike and, later, Calhoun and Brown counties (Griffin 1952b; Taylor 1958b:204). Outside the lower Illinois Valley, the occurrence of these finely crafted vessels is often attributed to exchange (Braun et al. 1982:6263; Farnsworth 1973:31-32; Strueverand Houart 1972:74-77).

The production of Baehr vessels, like their Hopewell antecedents, is strongly associated with the lower Illinois Valley, and their occurrence in other parts of the Midwest is similarly attributed to exchange (e.g., Braun et al. 1982:62-63; Farnsworth 1973:31-32; Griffin et al. 1970:7-9; Struever and Houart 1972:74-77). Like Baehr, Pike ceramics also concentrate along the southern 115 km of the Illinois River and adjacent portions of the Mississippi River Valley (Struever 1968b:7), although Pike ware sherds are occasionally reported in sites of the central and upper Illinois Valley (Griffin et al. 1970:8). Given the strong association of Hopewell, Baehr, and Pike ceramics with the lower Illinois Valley, it is not surprising that most of the analyzed examples are assigned to lower Illinois Valley groups.

Several decorative and technological attributes offer potential signatures of local, nonlocal, and exotic manufacture. Yet, comparison of these attributes finds little difference between "local" sherds and those identified with intraregional and interregional transport. For example, orifice diameter for the local Middle Woodland utilitarian rims (i.e., Havana, Pike, MW Fabric Impressed, MW Simple Stamped) averages 26.2+-6.1 cm (n=119), as compared to 26.3+-5.5 cm for the nonlocal rims (n=22), and 29.2+- 5.5 cm for the outliers (n=14).

Data collected on the aplastic inclusions5 suggests some possible texture differences among the habitation sherd groups. Specifically, the Sk group contains more low-to-moderately tempered sherds as compared to other habitations, whereas the nonlocal/LIV groups and the outliers contain more moderately-heavily tempered sherds (Table 8). While these tendencies implicate some influence of aplastics, the reassignment of sherds to habitation site group does not appear to pattern according to particle density. The 15 sherds reassigned to Sk, for example, exhibit a considerable range of aplastic densities (10-55 percent), as do the 12 sherds reassigned to Mp (10- 45 percent). The eight sherds assigned to the combined LIV sherd and day group include a slightly narrower range of particle densities (15-50 percent), whereas the Ap and SMD groups each attract a single sherd of moderate particle density.

Table 8. Estimated density of aplastics in the utilitarian sherd samples.


The compositional analysis reported here reveals that nonlocal and exotic ceramics occur in both the transaction centers and habitation site contexts. Their presence at sites throughout the lower Illinois Valley may follow from exchange interaction. However, foreign ceramics may also result from the presence of foreign groups and the consumption of vessels during the occupation of socio- ceremonial sites (Seeman 1979a:376). Compositionally distinct vessels may also result from marriage alliances and the incorporation of foreigners and their wares into local groups (see Clark et al. 1992:265; Ruby and Shriner 2006:570-571), or possibly seasonal or intermittent mobility that occasionally afforded access to contrasting raw materials (e.g., Neff et al. 1994:7; Simms et al. 1997). A determination of exchange therefore requires some means of differentiating these behaviors archaeologically.

Studies of settlement, subsistence, technology, mortuary patterning, and material exchange within the lower Illinois Valley provide a framework within which to address questions of Middle Woodland social interaction. Consideration of the available mortuary and settlement data allows for the rejection of the two competing interpretations: marriage alliances and an extensive resource area. The likelihood of marriage alliances as a possible source of nonlocal ceramics, for example, may be evaluated using Middle Woodland skeletal data, whereas the possibility of extended resource areas may be considered in light of settlement patterns within the region.

The distribution of Middle Woodland sites in the lower Valley region suggests maintained territories and relatively confined areas of raw material exploitation. Spaced at intervals of 12 to 30 km (Struever and Houart 1972:61), the large floodplain mound groups coincide with stylistic boundaries noted in Middle Woodland ceramics, and suggest the presence of social borders (Houart 1975:16- 17; see also Morgan 1985:250-255). At a smaller scale, more localized territories may be inferred from the clustering of mound groups across the valley bluff tops (Charles et al. 1986:458) and the interment of relatively stable, isolated populations (Buikstra 1976:55-57, 1980:289).

A second line of reasoning may be drawn from ethnographic data which suggests that potters tend to utilize proximate raw material sources. For day, potters generally exploit territories within one km of their settlement with a maximum range of about seven km (Arnold 1985:36-57). However, this distance threshold may be considerably greater for semi-mobile groups or in instances of water transport. The acquisition of more distant sources may also be embedded within the procurement of other resources or familial visits (DeBoer 1984:545-549).

Previous analysis of lower Illinois Valley skeletal remains by Buikstra (Buikstra 1980, 1984) provides little support for extensive marriage networks during the Middle Woodland period. Rather, her study of metric and nonmetric skeletal traits from Gibson-Klunk, Bedford, and Peisker found significant difference between mortuary population groups; additional comparison of the Middle Woodland and Late Woodland interments at Gibson-Klunk revealed considerable stability in the frequency of skeletal traits over time. Together, these analyses suggests the persistence of localized and somewhat (biologically) isolated communities in the lower Illinois Valley through the Middle Woodland period (Buikstra 1980:295, 1984:227). The occurrence of significant differences in trait frequencies among populations separated by some 15 km raises serious doubt about whether marriageexchange regularly transpired over substantial distances. Given these data, it is unlikely that foreign vessels in the lower Illinois Valley reflect marriage dowers.

Implications and Conclusions

The results of the current analysis have important implications for models of Middle Woodland exchange. First and foremost, these results demonstrate that, with the careful selection of materials and methods, the intraregional scale of exchange can be detected through compositional analysis. In the current study, chemical analysis permits the identification of ceramic exchange among groups of the lower Illinois Valley. Because the analyzed assemblages are not contemporaneous, it is difficult to extend arguments of ubiquity. Still, the presence of nonlocal sherds in all six of the analyzed sites suggests some regularity to the circulation of ceramic containers within this region. The results of this finer scale of study compel reconsideration of several past perceptions regarding the scale, goods, character, and organization of Middle Woodland exchange.

Hopewell goods were of obvious importance in Middle Woodland social systems. Yet, such items are rare in the lower Illinois Valley, and their numbers appear insufficient to support models of frequent or even regular exchange. Evidence for more recurrent interaction among Middle Woodland groups may be inferred from the occurrence and distribution of more proximate, less valuable materials such as Cobden chert and other lithic raw materials.

The results of compositional analysis of Middle Woodland sherds indicate that ceramic vessels were also regular commodities in intraregional and interregional exchange and interaction. In contrast to the sporadic presence of Hopewell exotics, nonlocal and exotic vessels occur in each of the four base settlements and in both transaction center assemblages. While these results are consistent with distance-decay prediction (e.g., Renfrew 1977:72, 77), the significance of more proximate connections in models of Middle Woodland exchange has been overshadowed by more far-flung contacts.

Contributing to this oversight is an often implicit assumption that, while the conveyance of status goods connotes social or political alliances, the exchange of subsistence-maintenance materials implicates economic interdependence. For the Middle Woodland, such an inference disputes the widely held view that groups were economically self-sufficient (Smith 1992:240) and met their needs with local materials or through negotiated access to additional territories (Cantwell 1980:26). Consequently, the exchange of raw materials and foodstuffs is generally disregarded in models of Middle Woodland exchange.

Yet, the exchange of subsistence-maintenance goods does not necessarily implicate economic interdependency. Foreign cherts, for example, were regularly acquired by lower Illinois Valley groups despite the presence of high quality local alternatives (McGimsey 1995:Appendix A; Morrow 1988). The same may be true of ceramics; the abundance of locally made vessels notwithstanding, foreign ceramics occur in all six of the analyzed Lower Valley assemblages. Rather than signify a reliance on vessel exchange, these foreign ceramics may simply represent goods that were incorporated into exchange networks to sustain valued social relationships (Vickery 1996:122- 123). Additional analyses will no doubt clarify many of the relationships identified in this study, and resolve many of the difficulties encountered. It should be possible, for example, to evaluate the source areas represented at each of the reference groups as well as the relative importance of different types of days. While such problems may yet be addressed with the current data set, establishing likely source areas for the ceramic outliers will require comparable elemental analysis of ceramic materials from other regions of the Midwest, particularly the central Illinois Valley. Yet, as the current study demonstrates, such questions may be addressed through the deliberate selection and analysis of archaeological materials.


This research was conducted with support from the National Science Foundation (Doctoral Dissertation Improvement Grant) and the University of Missouri Research Reactor Center in Columbia, Missouri. I am particularly grateful to Dr. Hector Neff, Dr. Michael Glascock, Jim Cogswell, and Cindy Hayes for their comments on the original study, instruction in the MURR labs, and their overall patience. I would also like to thank the Center for American Archeology and the Illinois State Museum for access to the curated collections used in this study, and particularly Ken Farnsworth and Gail Anderson for contributing their expertise and time to the selection of the sherd samples. Lastly, I am indebted to Philip J. Arnold, III, Dan Shea, Robert Mainfort and an anonymous reviewer for their insightful comments and suggestions.


1. Transaction centers, as originally proposed by Struever ( 1968a:300, 1968b:208209; Struever and Houart 1972:61), served as locations for the receipt, concentration, manufacture, and redistribution of Hopewell Interaction Sphere items within and between regions. Since its introduction, the transaction center concept has prompted several proposed revisions (Asch et al. 1979:83; McGimsey and Wiant 1986:540; Staab 1984:252-253) and even calls for its outright rejection (Farnsworth 1990:126). While somewhat dated, the transaction center concept was employed in the current study to underscore the functional distinctions associated with the location, size, structure, and content of these floodplain mound groups in the lower Illinois Valley (Asch et al. 1979; Buikstra 1976; Cowan and Fie 1993; McGimsey and Wiant 1986).

2. The collection of clay samples was originally scheduled for the summer of 1993. However, extensive flooding of the region prevented access to many areas, and the survey was postponed until the following spring. In several areas of the lower Illinois Valley, scouring floodwaters eroded much of the accumulated talus on the bluff slopes and, fortuitously, exposed the underlying geologic strata. However, these rapidly moving floodwaters also carried heavy particle loads that, once the waters receded, resulted in extensive flats of sand and silt. Since it was unlikely that the modern chemical and mineral contaminants associated with agricultural and barge activity could be controlled for, no additional effort was made to acquire samples of these materials.

3. Sample SFC-482 was recovered from a clay-filled pit (W.M. 2004F1 ) during the 1976 excavations of the Late Woodland period Worthy-Merrigan site by the Center for American Archeology's Education Program.

4. A series of GAUSS language routines made it possible to compute replacement values, generate bivariate plots, calculate principal components, and calculate the probability of group membership based on Mahalanobis distances. These routines were written by Hector Neff for use at the University of Missouri Research Reactor Center.

5. Subsequent to the cutting of the sample for compositional analysis, the aplastic inclusions were identified and the density of these inclusions visually estimated across the planed surface of the sample profile using a binocular microscope (20/40 X) with reference to a density chart. The identified types of aplastics include both processed tempers and natural inclusions: crushed granite (grit), sand, as well as some limestone, darks (crushed mafic material), chert, limonite, and crushed ceramic (grog) or shale.

References Cited

Arnold, D. E.

1985 Ceramic Theory and Culture Process. Cambridge University Press, New York.

Asch, D. L.

1976 The Middle Woodland Population of the Lower Illinois Valley: A Study in Paleodemographic Methods. Scientific Papers Number 1. Northwestern University Archeological Program, Evanston, Illinois.

Asch, D. L., K. B. Farnsworth, and N. B. Asch

1979 Woodland Subsistence and Settlement in West-Central Illinois. In Hopewell Archaeology: The Chillicothe Conference, edited by D. S. Brose and N. Greber, pp. 80-85. Kent State University Press, Kent, Ohio.

Bender, B.

1985 Emergent Tribal Formation in the American Midcontinent. American Antiquity 50:52-62.

Bishop, R. L, R. L. Rands, and G. R. Holley

1982 Ceramic Compositional Analysis in Archaeological Perspectives. In Advances in Archaeological Method and Theory, Vol. 5, edited by M. B. Schiffer, pp. 275-330. Academic Press, New York.

Bishop, R. L, V. Canouts, S. P. De Alley, A. Qoyawayma, and C. W. Aikins

1988 The Formation of Ceramic Analytical Groups: Hopi Pottery Production and Exchange, A.C. 1300-1600. Journal of Field Archaeology 15:317-337.

Blackman, M. J.

1992 The Effect of Natural and Human Size Sorting on the Mineralogy and Chemistry of Ceramic clays. In Chemical Characterization of Ceramic Pastes in Archaeology, edited by H. Neff, pp. 113-124. Monographs in World Archaeology No. 7. Prehistory Press, Madison, Wisconsin.

Bluhm, E.

1951 Ceramic Sequence in the Central Basin and Hopewell Sites in Central Illinois. American Anthropologist 16:324.

Boszhardt, R. E, and J. Gundersen

2003 X-Ray Powder Diffraction Analysis of Early and Middle Woodland Red Pipes from Wisconsin. Midcontinental Journal of Archaeology 28:33-48.

Braun, D. P.

1977 Middle Woodland-(Early) Late Woodland Social Change in the Prehistoric Central Midwestern United States. Ph.D. dissertation, University of Michigan, Ann Arbor. University Microfilm, Ann Arbor.

1983 Pots as Tools. In Archaeological Hammers and Theories, edited by J. A. Moore and A. S. Keen, pp. 107-134. Academic Press, New York.

1985a Absolute Sedation: A Time Series Approach. In For Concordance in Archaeological Analysis: Bridging Data Structure, Quantitative Techniques, and Theory, edited by C. Carr, pp. 509- 539. Westport Publishers, Kansas City, Missouri.

1985b Ceramic Decorative Diversity and Illinois Woodland Regional Integration. In Decoding Prehistoric Ceramics, edited by B. A. Nelson, pp. 128-153. Southern Illinois University Press, Carbondale.

1987 Coevolution of Sedentism, Pottery Technology, and Horticulture in the Central Midwest, 200 B.C.- A.D. 600. In Emergent Horticultural Economies of the Eastern Woodlands, edited by W F. Keegan, pp. 153-181. Occasional Papers Number 7. Center for Archaeological Investigations, Southern Illinois University, Carbondale.

1991 Why Decorate a Pot? Midwestern Household Pottery, 200 B.C.- A.D. 600. Journal of Anthropological Archaeology 10:360-397.

Braun, D. P., and S. E. Plog

1982 Evolution of "Tribal" Social Networks: Theory and Prehistoric American Evidence. American Antiquity 47:504-525.

Braun, D. P., J. B. Griffin, and P. F. Titterington

1982 The Snyders Mounds and Five Other Mound Groups in Calhoun County Illinois. Technical Reports No. 14. Research Reports in Archaeology No. 8. Museum of Anthropology, University of Michigan, Ann Arbor.

Brose, D. S.

1990 Towards a Model of Exchange Values for the Eastern Woodlands. Midcontinental Journal of Archaeology 15:100-136.

1979 A Speculative Model of the Role of Exchange in the Prehistory of the Eastern Woodlands. In Hopewell Archaeology: The Chillicothe Conference, edited by D. S. Brose and N. Greber, pp. 3- 8. Kent State University Press, Kent, Ohio.

Brown, J. A.

1981 The Search for Rank in Prehistoric Burials. In The Archaeology of Death, edited by R. Chapman, I. Kinnes, and K. Ransborg, pp. 25-37. Cambridge University Press, Cambridge.

Buikstra, J. E.

1976 Hopewell in the Lower Illinois River Valley: A Regional Study of Human Biological Variability and Prehistoric Mortuary Behavior. Scientific Papers No. 2. Northwestern University Archeological Program, Evanston, Illinois.

1980 Epigenetic Distance: A Study of Biological Variability in the Lower Illinois River Region. In Early Native Americans: Prehistoric Demography, Economy, and Technology, edited by D. L. Browman, pp. 271-299. Mouton Publishers, The Hague.

1984 The Lower Illinois River Region: A Prehistoric Context for the Study of Ancient Diet and Health. In Paleopathology at the Origins of Agriculture, edited by M. N. Cohen and G. J. Armelagos, pp. 215-234. Academic Press, New York.

Bullington, J.

1988 Middle Woodland Mound Structure: Social Implications and Regional Context. In The Archaic and Woodland Cemeteries at the Elizabeth Site in the Lower Illinou Valley, edited by D. K. Charles, S. R. Leigh, and J. E. Buikstra, pp. 218-241. Kampsville Archeological Research Series Volume 6. Center for American Archeology Press, Kampsville, Illinois.

Cantwell, A.

1980 Dickson Camp and Pond: Two Early Havana Tradition Sites in the Central Illinois Valley. Reports of Investigation Number 36. Illinois State Museum, Springfield.

Carr, C.

2006 The Question of Ranking in Havana Hopewellian Societies: A Retrospective in the Light of Multi-Cemetery Ceremonial Organization. In Gathering Hopewell: Society, Ritual, and Ritual Interaction, edited by C. Carr, pp. 238-257. Springer Science+ Business Media, New York. Carr, C., and J. Komorowski

1995 Identifying the Mineralogy of Rock Temper in Ceramics Using X-Radiography. American Antiquity 60:723-749.

Charles, D. K.

1992 Woodland Demographic and Social Dynamics in the American Midwest: Analysis of a Burial Mound Survey. World Archaeology 24:175- 197.

Charles, D. K., J. E. Buikstra, and L. W. Konigsberg

1986 Behavioral Implications of Terminal Archaic and Early Woodland Mortuary Practices in the Lower Illinois Valley. In Early Woodland Archaeology, edited by K. B. Farnsworth and T. E. Emerson, pp. 458-474. Kampsville Seminars in Archaeology Volume 2. Center for American Archeology Press, Kampsville, Illinois.

Clark, C., H. Neff, and M. D. Glascock

1992 Neutron Activation Analysis of Late Woodland Ceramics from the Lake Superior Basin. In Chemical Characterization of Ceramic Pastes in Archaeology, edited by H. Neff, pp. 255-267. Monographs in World Archaeology No. 7. Prehistory Press, Madison, Wisconsin.

Clark, F.

1984 Knife River Flint and Interregional Exchange. Midcontinental Journal of Archaeology 9:173-198.

Cogswell, J. W, H. Neff, and M. D. Glascock

1995 Neutron Activation Analysis of Pottery and clays from Selected Sites in the Upper Illinois River Valley. Final report. University of Missouri Research Reactor, Columbia. Samples submitted by R. J. Jeske and M. Lynott.

1998 Analysis of Shell-Tempered Pottery Replicates: Implications for Provenance Studies. American Antiquity 63:63-72.

Cole, Fay-Cooper, and T. Deuel

1937 Rediscovering Illinois: Archaeological Explorations in and around Fulton County. University of Chicago, Illinois.

Cowan, F. L., and S. M. Fie

1993 Hopewell Transaction Centers Revisited: The Mound House case. Paper presented at the 58th Annual Meeting of the Society for American Archaeology. St. Louis, Missouri.

Deboer, W R.

1984 The Last Pottery Show: System and Sense in Ceramic Studies. In The Many Dimensions of Pottery, edited by S. E. van der Leeuw and A. C. Pritchard, pp. 527-571. University of Amsterdam, Amsterdam, Netherlands.

Elam, J. M., C. Carr, M. D. Glascock, and H. Neff

1992 Ultrasonic Disaggregation and INAA of Textural Fractions of Tucson Basin and Ohio Valley Ceramics. In Chemical Characterization of Ceramic Pastes in Archaeology, edited by H. Neff, pp. 93-111. Monographs in World Archaeology No. 7. Prehistory Press, Madison, Wisconsin.

Emerson, T. E., R. E. Hughes, K. B. Farnsworth, S. U. Wisseman, and M. R. Hynes

2005 Tremper Mound, Hopewell Catlinite, and Pima Technology. Midcontinental Journal of Archaeology 30:189-216.

Epstein, J. F.

1958 Ceramics and Ceramic Artifacts of the Pool Site. In The Pool and Irving Villages: A Study of Hopewell Occupation in the Illinois River Valley, edited by J. C. McGregor, pp. 31-63. University of Illinois Press, Urbana.

Farnsworth, K. B.

1973 An Archaeological Survey of the Macoupin Valley. Reports of Investigation No. 26, Research Papers Vol. 7, Illinois Valley Archaeology Program. Illinois State Museum, Springfield.

1986 Black Sand Culture Origins and Distribution. In Early Woodland Archaeology, edited by K. B. Farnsworth and T. E. Emerson, pp. 634-641. Kampsville Seminars in Archaeology Volume 2. Center for American Archeology Press, Kampsville, Illinois.

1990 The Evidence for Specialized Middle Woodland Camps in Western Illinois. Illinois Archaeology 2:109-132.

Farnsworth, K. B., and D. L. Asch

1986 Early Woodland Chronology, Artifact Styles, and Settlement Distribution in the Lower Illinois Valley Region. In Early Woodland Archaeology, edited by K. B. Farnsworth and T. E. Emerson, pp. 326- 457. Kampsville Seminars in Archaeology Volume 2. Center for American Archeology Press, Kampsville, Illinois.

Farnsworth, K. B., and A. L. Koski

1985 Massey and Archie: A Study of Two Hopewellian Farmsteads in the Western Illinois Uplands. Kampsville Archeology Research Series Vol. 3. Center for American Archeology, Kampsville, Illinois.

Fie, S. M.

2000 An Integrative Study of Ceramic Exchange During the Illinois Valley Middle Woodland Period. Ph.D. dissertation, State University of New York at Buffalo. University Microfilm, Ann Arbor.

2006 Visiting in the Interaction Sphere: Ceramic Exchange and Interaction in the Lower Illinois Valley. In Recreating Hopewell, edited by D. K. Charles and J. E. Buikstra, pp. 427-445. University of Florida Press, Gainesville.

Fowler, M. L.

1952 The Clear Lake Site: Hopewellian Occupation. In Hopewellian Communities in Illinois, edited by T. Deuel, pp. 133-174. Scientific Papers Volume 5, Number 4. Illinois State Museum, Springfield.

1955 Ware Groupings and Decoration of Woodland Ceramics in Illinois. American Anthropologist 20:213-225.

Frye, J. C., H. B. Willman, and H. D. Glass

1964 Cretaceous Deposits and the Illinois Glacial Boundary in Western Illinois. Illinois State Geologic Survey Circular Number 364. Urbana.

Glascock, M. D.

1992 Characterization of Archaeological Ceramics at MURR by Neutron Activation Analysis and Multivariate Statistics. In Chemical Characterization of Ceramic Pastes in Archaeology, edited by H. Neff, pp. 11-26. Monographs in World Archaeology No. 7. Prehistory Press, Madison, Wisconsin.

2005 Neutron Activation Analysis of Chert Artifacts from a Hopewell Mound. Journal of Radioanalytical and Nuclear Chemistry 262(1):91-102.

Griffin, J. B.

1952a Some Early and Middle Woodland Pottery Types in Illinois. In Hopewellian Communities in Illinois, edited by T. Deuel, pp. 95- 129. Scientific Papers Vol. 5, Illi