The culture-historical dichotomy drawn between the Woodland and Mississippian cultural traditions in the Southeast is quite distinct. Woodland peoples have been viewed as egalitarian groups who lived in scattered, nucleated villages along the main rivers or in small hamlets located in the uplands, making sand- and/or grog-tempered pottery and primarily hunting, with some cultivation of native crops for food (Anderson and Mainfort 2002:1–19). In contrast, the Mississippian period (following Rafferty and Peacock 2008a:6) is most often characterized by inferred hierarchies of mound complexes with surrounding villages whose inhabitants subsisted primarily upon maize agriculture and made mussel shell-tempered pottery. As drawn, this change in cultural patterns is so pronounced that it is often taken to indicate an influx of foreign “Mississippian” peoples from other regions of the continent into a given area (e.g., the Tombigbee and Black Warrior River valleys [Jenkins and Krause 1986:90]).
A body of research has begun to refute the stark contrasts drawn between these two periods (Anderson and Mainfort 2002; Bozeman 1982; Hogue and Peacock 1995; Mistovich 1988, 1995; Peacock 1997, 2003; Peacock and Rafferty 1996; Rafferty 1996, 2001, 2002, 2003). As Rafferty has pointed out (1996, 2001; Rafferty and Peacock 2008b), in the Black Prairie region of eastern Mississippi, just to the west of the Tombigbee River valley (Figure 10.1), there are indications of continuity in settlement patterns between the two periods. Quite often the same landform or area was either continuously settled or repeatedly resettled over extended periods of time that encompassed both the Woodland and Mississippian cultural traditions.
Figure 10.1. The Ackerman Unit of the Tombigbee National Forest and surrounding physiographic provinces of Mississippi.
To the west of the Black Prairie is the North Central Hills physiographic province (Figure 10.1), within which lies most of the 44,000 acres of the Ackerman Unit of the Tombigbee National Forest. The Noxubee River and its principal tributaries, the Little Noxubee River and Mill Creek, drain the area within the boundaries of the Ackerman Unit before eventually emptying via the Noxubee River into the Tombigbee River farther south.
Numerous archaeological surveys conducted on the Ackerman Unit indicate a long history of occupation. A dramatic increase in the number of occupations concurrent with the Woodland period (Figure 10.2; see also Parrish 2006 and Peacock 1997) may indicate a switch from a mobile existence during the Paleo-Indian and Archaic periods to a sedentary settlement pattern coincident with the adoption of ceramics (Bacon-Schulte 2008; Rafferty 1994) and brought about by range compression caused by increases in population density in the region (Rafferty 1994:420). Sites containing Woodland-period occupations are found on all landforms on the Ackerman Unit, something also likely a product of increased population density as earlier sites are found primarily along the larger waterways. Twelve of the 13 Mississippian components (represented by the presence of shell-tempered pottery; see Feathers 2006) recorded on the Ackerman Unit as of 2007 (Triplett 2008a) are located at sites with much larger Late Woodland-period occupations, and all but one are concentrated along the Noxubee River or one of its main tributaries (Figure 10.3).
Figure 10.2. Number of prehistoric occupations at sites within the Ackerman Unit of the Tombigbee National Forest. “Woodland” = undifferentiated component dating from ca. 200 B.C.–A.D. 900.
Figure 10.3. Sites containing Mississippian-period occupations within the Tombigbee National Forest, Ackerman Unit.
The dramatic decrease in the number of Mississippian-period components within the Ackerman Unit, and the concentration of sites with such components along major streams, could indicate population consolidation (i.e., nucleation) or the movement of some part of the population out of the area late in prehistoric times. Research in the western portion of the Black Prairie has revealed numerous small sites containing mussel shell-tempered pottery. While some have suggested that these occupations represent expansion away from the Tombigbee River and into the uplands during the Protohistoric period (Futato 1989; Johnson 1996; Johnson and Sparks 1986; Johnson et al. 1994), others believe them to represent continuity in upland settlement patterns from the Woodland through the Mississippian periods (Hogue and Peacock 1995; Peacock and Rafferty 1996; Rafferty 1996, 2001, 2003; Rafferty and Peacock 2008b). This issue is complicated because of the equivocal evidence (plain mussel shell-tempered pottery) being used to place these occupations chronologically (Johnson 1996:244; Peacock and Rafferty 1996:249).
Recent work using a combination of absolute dates and frequency seriations suggests that there was an expansion of settlements marked by shell-tempered pottery in the Black Prairie beginning ca. A.D. 1200 (Rafferty and Peacock 2008b). This apparent increase in site numbers is attributed by Rafferty and Peacock (2008b:258) to “selective pressures related to the onset of climate change, possibly coupled with adoption of maize-based subsistence.” It has also been hypothesized that by about A.D. 1100 the North Central Hills were abruptly abandoned (Blitz 1984; Peacock 2003; Peacock et al. 2008), suggesting a place of origin for the movement of early agricultural groups into the Black Prairie (Rafferty and Peacock 2008b). This hypothesis is bolstered by data from Stinking Water (22WI515/516), a large, multicomponent site located beside the Noxubee River in the Ackerman Unit. Diagnostic artifacts indicate that this locale was occupied from the Archaic into the Mississippian period, but was abandoned around A.D. 1000 (Peacock 2003:47).
Two hypotheses can explain the co-occurrence of Woodland- and Mississippian-period diagnostics in the Ackerman Unit area. The first hypothesis is that the Woodland and Mississippian materials represent continuous occupation, with the sites being abandoned early in the Mississippian period. The second hypothesis is that the components do not represent continuous occupations but that the shell-tempered pottery indicates resettlement in the Mississippian period. These hypotheses were tested via a program of systematic shovel testing at eight sites to determine the location and size of the occupations with shell-tempered pottery, followed by test excavations at three sites designed to measure the degree of co-occurrence between earlier and later pottery types.
An occupation is a “spatial cluster of discrete objects which can reasonably be assumed to be the product of a single group of people at that particular locality deposited over a period of continuous residence comparable to other such units in the same study” (Dunnell 1971:151). As such, an occupation can be considered an artifact at the scale of assemblage (Parrish and Peacock 2006; Rafferty 2008). A class of artifact at this scale is needed to examine culture change and artifact distributions because it “allows artifacts at smaller scales to be associated in meaningful assemblages” (Rafferty 2008:102).
Determining the spatial boundary of an occupation is dependent upon field and proveniencing methods, because spatial associations are based upon propinquity (Rafferty 2008). Demonstrating that individual artifacts, including features, are associated is of utmost importance. Horizontal boundaries can be detected via spatially extensive methods such as controlled surface collection or shovel testing at even intervals (e.g., Parrish and Peacock 2006). Stratigraphic excavation can be used in determining the vertical (temporal) boundaries of an occupation, as artifacts from the same occupation would be expected to be associated in a depositional layer (Rafferty 2008).
Eight sites were chosen for investigation based upon the documented presence of Mississippian-period artifacts (see Triplett 2008a). Because disparate methods of artifact collection had been used previously, systematic shovel testing (Roskams 2001:49) was employed to delineate concentrations of shell-tempered pottery within the larger site (see also Bozeman 1982; Lorenz 1996). Sites were shovel tested on 10-meter grids following the orientation and shape of the landforms (see Triplett 2008a for details).
Shovel testing confirmed that shell-tempered ceramics are present at the eight sites investigated, albeit in relatively small proportions (Table 10.1). Mixed grog- and shell-tempered sherds may represent the transition from grog to mussel shell pottery tempers in this area. There is some debate as to the chronological position of grog/shell-tempered pottery. Jenkins (Jenkins and Krause 1986:93) places this pottery early in the Mississippian period, during the Summerville I phase. Steponaitis (1983:158), though, believes that it is present throughout the Mississippian period. On the Ackerman Unit, it is usually found at sites with Late Woodland and Mississippian components, suggesting continuity of settlement.
As will be discussed further below, these ceramic data were employed in frequency seriations. To further investigate intrasite spatial relationships, three sites were chosen for test excavations, the results of which are discussed first.
A paradigmatic classification (Table 10.2) was developed for determining which of the shovel-tested sites to excavate. The paradigm was based on several mutually exclusive dimensions: total area containing Mississippian materials, type of landform on which the site was located, and the number of components (identified by diagnostic pottery tempers and surface decorations) present. This classification was constructed to incorporate the greatest amount of variability among sites based upon the information gathered through systematic shovel testing. Testing sites on different landforms, with varying degrees of Mississippian occupation and with varying lengths of habitation, was considered the best way of determining the timing of the Mississippian-period habitation of the area (see Triplett 2008a for more detailed discussion of dimensions).
As can be seen in Table 10.2, sites 22CH515 and 22WI666 are both included in the class defined as being located on a terrace, having three or more components present, with an estimated area producing Mississippian materials of between 101 and 500 square meters. Of the two, site 22WI666 was chosen for excavation because it was judged to be less disturbed and therefore had a higher possibility of containing features.
In spite of only a single shell-tempered eroded sherd being recovered at 22CH814 (Table 10.1), the site was chosen for further investigation based upon several factors. First, it was the only terrace location with just two components represented. Also, the Mississippian occupation, according to the initial survey in 2006, was very small (approx. 25 square meters), a finding that was not contradicted by systematic shovel testing. This was the only site with a Mississippian occupation under 100 square meters in size in which the location of the shell-tempered pottery was known, providing an interesting contrast to the larger sites being considered.
Site 22WI865 is the only site found to date on the Ackerman Unit that has two spatially separate Mississippian occupations, i.e., two distinct concentrations of shell-tempered ceramics as revealed by systematic shovel testing. It is also distinct in its paradigmatic classification: the estimated total size of the Mississippian occupation is over 500 square meters; it is the only site under consideration located on a bluff; and it contains four identifiable components.
Excavations were conducted by the 2007 Mississippi State University field school under the direction of Evan Peacock. Fieldwork consisted of topographic mapping and the excavation of 1-×-1-m test units using natural levels subdivided into maximum 10-cm increments as necessary. Unless otherwise noted, soils were dry-screened through quarter-inch (.64 cm) hardware cloth.
During systematic survey of the site, shell-tempered pottery was recovered from a shovel test (Figure 10.4) on a slight downhill slope in the southeastern area of the site. It was thought that any artifacts found in this area had most likely eroded down the slope, and also that the most probable area to encounter features would be on the level area above. Consequently, a block of 12 test units was placed on the level area approximately 5 m south of this particular shovel test (Figure 10.5). The soils in the unit were rather disturbed, and no features or postholes were encountered in any of the excavated units.
Figure 10.4. 22CH515, shovel-test placement and pottery distribution.
Figure 10.5. 22CH515, topographic map showing excavation unit placement.
The systematic shovel testing conducted prior to excavation indicated a previously unknown Gulf Formational component present, as evidenced by fiber-tempered pottery (Table 10.1). This finding was further substantiated by the recovery of eight additional fiber-tempered sherds from the excavation units. (Analyses do not include any pottery pieces smaller than ca. 20 mm.) Several sand-tempered plain and eroded sherds were also present (Table 10.3). Due to the lack of diagnostic surface decorations (e.g., cord marking or fabric marking), it could not be determined if the sand-tempered pottery represented a Middle Woodland component. Most of the sherds recovered are grog-tempered, including two fabric-marked sherds, and are assignable to the Late Woodland period. A Mississippian component was marked by the recovery of 19 shell-tempered sherds.
The original shovel-test survey of this site (Figure 10.6), conducted in the summer of 2006, recovered a single shell-tempered sherd. It was decided to open a block unit approximately 2 m south of the shovel test that produced this sherd (Figure 10.7), in the hope that additional shell-tempered sherds would be recovered and features would be encountered. Nine adjacent units were excavated to subsoil, forming a 3-×-3-m block. It became evident, in light of the very thin soil horizons (Triplett 2008a), that this site had been impacted by previous activities. No features or postholes were encountered.
Figure 10.6. 22CH814, shovel-test placement and pottery distribution.
Figure 10.7. 22CH814, topographic map showing excavation unit placement.
Due to the lack of features and the paucity of pottery in the initial block unit, it was decided to excavate two other 1-×-3-m blocks in close proximity, one 6 m to the west and the other 6 m south (Figure 10.7). Unfortunately, as with the original test units, these units produced very few ceramic artifacts and no features were encountered.
A total of 34 grog-tempered and three grog/shell-tempered sherds was recovered (Table 10.4). The grog-tempered sherds, particularly the cordmarked ones, indicate a Late Woodland occupation. No shell-tempered pottery was recovered from any of the excavation units at this site.
Excavated soil at this site was water-screened through quarter-inch and sixteenth-inch stacked screens in order to recover small artifacts. The latter material has not been processed or analyzed, but has been retained for future research.
Systematic shovel testing suggested two spatially separate areas of occupation (Figure 10.8). One was in a midden in the northernmost area, closest to the bluff above the Little Noxubee River, which contained diagnostic pottery tempers and surface finishes representing all of the known pottery-producing cultural periods of the region. Two separate 1-×-1-m test units were placed in this area, as well as an irregular block unit that eventually consisted of six 1-×-1- and one 1-×-.50-m test units (Figure 10.9). The other pottery concentration, on the southern end of the site, indicated Late Woodland and Mississippian components. Soils in this area were much shallower and had been heavily disturbed. A large block unit was laid out in approximately the middle of the several shovel tests that contained shell-tempered pottery. Initially, this block consisted of 20 contiguous 1-×-1-m units; four additional 1-×-1-m units were opened up when no features were encountered in the initial units.
Unit 15S11E was located in the deepest part of the midden and is the only unit at this site reported by zone/level. The pottery recovered there includes three fiber-tempered, three sand-tempered, and 19 grog-tempered sherds (Table 10.5). It appears that, even though this is a relatively deep midden, it has been somewhat disturbed either by bioturbation or human activities. The majority of the sherds were approximately the size of a US 25-cent piece (ca. 24 mm), and the distribution of artifacts within the unit shows signs of disturbance, with a grog-tempered sherd being recovered from level B4 along with fiber-tempered sherds. The block unit near the midden contained the greatest diversity of pottery tempers and surface finishes at this site. Diagnostic pottery tempers and surface decorations/finishes from all of the known pottery-producing cultural periods for this area are present (Table 10.6). The pottery recovered from 37S11E consisted of three sand-tempered, 14 grog-tempered, and one shell-tempered sherd. The pottery recovered from the southern block unit consisted almost exclusively of either grog-tempered or shell-tempered pottery (Table 10.7).
Figure 10.8. 22WI865, shovel-test placement and pottery distribution.
Figure 10.9. 22WI865, topographic map showing excavation unit placement.
Spatial information on the distribution of pottery tempers recovered during excavation was used in measuring the tendency of pottery classes to agglomerate about a central location within each excavated site, using the arithmetic mean (Thomas 1986:65). If the mean location of each pottery class is significantly different, it is an indication that they are from separate occupations. If the mean locations are similar, this suggests that the pottery classes are from a single occupation. Also, the number of grog- and shell-tempered sherds recovered from the excavation units are displayed in line graphs. Co-occurrence of both types of pottery, with similar proportions in each unit, would indicate a single occupation. Alternatively, a mutually exclusive pattern of occurrence, in which the pottery tempers are not present within the same units, is an indication that two separate occupations are present. Due to small sample sizes, 22CH814 is not included in these analyses.
Figure 10.10. 22CH515, arithmetic mean of pottery classes.
Figure 10.11. 22WI865, midden units, arithmetic mean of pottery classes.
Figure 10.12. 22WI865, southern block, arithmetic mean of pottery classes.
Results from 22CH515 (Figure 10.10) show the arithmetic mean for three of the four pottery classes to be 2.5. The mean for the shell-tempered pottery is 2.27, which shows that the central tendency for all four pottery classes is very similar, suggesting that a single occupation is present. Results similar to those from 22CH515 were encountered when computing the arithmetic mean for the midden block and southern block units at 22WI865 (Figures 10.11 and 10.12). In both blocks, the shell-tempered pottery has a slightly different arithmetic mean from the rest of the pottery classes represented, but there is still a high coincidence of occurrence between the grog-tempered and shell-tempered pottery. In the midden block, shell-tempered pottery occurred in six of the seven units (85.7 percent) that contained grog-tempered pottery. Shell-tempered pottery was present in 21 of the 23 units (91.3 percent) that contained grog-tempered pottery in the southern block unit as well.
The results from the midden block unit seem to indicate that this area of the site, much like the excavated portion of 22CH515, was continuously occupied over an extended period of time. The southern block unit appears to have been a spatially separate occupation, as suggested by the shovel-test data. The pottery recovered from this area was almost exclusively grog- or shell-tempered, with only two sherds of sand-tempered pottery having been recovered, suggesting occupation during the Late Woodland and into the Mississippian period.
As can be seen in Figure 13, there is co-occurrence of grog- and shell-tempered pottery from 22CH515 (Figure 10.13), with a general pattern in which, once shell-tempered pottery is established, it increases in numbers along with grog-tempered pottery. This accords well with expectations that the proportions of the two tempers would not vary much if the Late Woodland and Early Mississippian materials are part of one occupation.
The line graphs for the two block units at 22WI865 (Figures 10.14 and 10.15) also indicate a general pattern of shell-tempered pottery increasing in numbers along with the grog-tempered pottery. The units that have very few shell-tempered sherds in relation to the number of grog-tempered sherds may be exceptions to this and may represent the earliest part of the occupation. The data gathered through shovel testing and excavations, as well as the statistical analyses, suggest that two spatially separate occupations are present at this site. The area closest to the Little Noxubee River exhibits a well-developed midden, an abundance of artifacts, and in some areas, deep cultural deposits; all of these have been shown to be indicators of long-duration sites (Rafferty 1994). The distribution of pottery as seen in both shovel testing and excavations shows that this area was occupied frequently—and possibly continuously—over a long duration of time. Conversely, the southern area of the site appears to contain a distinct, single occupation with both grog- and shell-tempered pottery present.
Figure 10.13. 22CH515, co-occurrence of grog-tempered and shell-tempered pottery.
Figure 10.14. 22WI865, midden block, co-occurrence of grog-tempered and shell-tempered pottery.
Figure 10.15. 22WI865 south block unit, co-occurrence of grog-tempered and shell-tempered pottery.
Each of the sites under consideration contains occupations that potentially span hundreds to thousands of years. Another way to test for continuity of occupations is through the use of frequency or occurrence seriations (Dunnell 1970; Lipo et al. 1997). Artifact classes, the members of which are the result of homologous transmission, will display a unimodal frequency distribution through time. If artifact classes do not produce a unimodal frequency distribution, homologous transmission has not occurred (Lyman and O’Brien 2003:270). There are three requirements of artifact groups for seriations to work (Dunnell 1970). The first is that the groups must be of comparable duration; if they are noncomparable, they will not seriate together. The second is that all of the assemblages come from the same area. This is because heritable continuity deals with both the spatial and temporal aspects of the artifact groups (Lipo et al. 1997). The final requirement is that all of the assemblages must come from the same cultural tradition or lineage. If these three requirements are met, then heritable continuity is demonstrated by display of a deterministic seriation (Lyman and O’Brien 2003:270; see also Parrish and Peacock 2006; Rafferty 2008). Assemblages from sites with multiple, non-sequent occupations should not seriate with those from single-occupation sites.
Pottery was classified using temper and surface finish types, allowing for other local assemblages classified in the same manner (e.g., Parrish 2006; Peacock 2003) to be included in the seriation. Only assemblages with 35 or more identifiable (non-eroded) sherds were included in the seriations; eroded sherds were not used in frequency calculations. For ease of display, some classes (e.g., sand-tempered punctate, grog-tempered rocker stamped, bone-tempered plain) were not used, as they had so few occurrences that their omission had no impact on the frequency percentages used in the ordering of the seriation. The frequencies for each class of pottery were calculated, and the assemblages were then ordered using the Seriationmaker macro (Lipo 2001). An error factor of 5 percent was chosen because sample size varies greatly between assemblages.
Figure 10.16. Frequency seriations I, II, and III (from top) of ceramic assemblages from the Ackerman Unit of the Tombigbee National Forest.
Figure 10.16 shows that not all of the assemblages are of comparable duration. Three separate seriations are constructed, indicating that some sites were occupied for much different periods of time than others. Seriation I is populated by short-duration assemblages, as indicated by the relatively small number of pottery classes represented and the minimal changes in frequencies over time. It appears that the sites in this seriation were generally occupied later than those represented in the other seriations, with mostly Late Woodland and Mississippian occupations indicated. The assemblage recovered from the southern block unit at 22WI865 (22WI865S) is included in this seriation.
Seriation II consists of assemblages with a greater number of pottery classes, probably indicating that these sites were occupied for longer durations. The sites represented seem to follow the general pattern of settlement in the North Central Hills, with primarily Middle and Late Woodland period occupations. The few Mississippian components present in this seriation are all associated with much larger Late Woodland components; none are found at sites with only a Middle Woodland component. These patterns strengthen the assertion that the Late Woodland and Mississippian-period components share a cultural lineage and are likely part of the same occupations.
Seriation III includes the excavated portions of the two longest-duration sites. It was decided to use the excavated units instead of the entire sites because, as noted earlier, systematic shovel testing at 22WI865 indicated that it contained two spatially separate occupations. Accordingly, this site was divided, for the purpose of seriation, into assemblages consisting of the ceramics found in the northern, or “midden,” occupation (22WI865M) and the southern occupation (22WI865S). The excavated units at 22CH515 (22CH515E) and the midden at 22WI865 contain diagnostic pottery tempers for all of the known ceramic periods in this region. While these sites could represent many smaller, short-duration occupations spanning several cultural periods (Rafferty 2003:172), they more likely represent continuous occupation over a long time span at locations closest to the water (presumably the most favorable area of the sites).
Even though three separate seriations were created from these assemblages, the implications of all are the same. All of the seriations indicate that their respective assemblages belong to the same cultural tradition, thus, “heritable continuity is assured and phylogenetic affinities between the seriated assemblages are guaranteed” (O’Brien and Lyman 2000:287). The short-duration sites appear to have been occupied later than the others, primarily during the Late Woodland period, possibly because of settlement shifts away from uplands to terraces along the major waterways. The seemingly abrupt end to the grog-tempered plain curve in all three seriations accords well with the hypothesis that this area was rapidly abandoned at the very beginning of the Mississippian period.
The purpose of this research was to ascertain if cultural lineage continuity could be shown between the Late Woodland and Mississippian occupations at selected sites on the Ackerman Unit of the Tombigbee National Forest. Two hypotheses were tested for this purpose. The first was that the Late Woodland and Mississippian components found together at these sites indicate continuous occupation, with the sites being abandoned early in the Mississippian period. The second hypothesis was that these components did not represent continuous occupations but resettlement of Woodland-period site locales later in the Mississippian period. The best way to determine this was through delineating the occupations present at the sites. The spatial aspect of the occupations was addressed through systematic shovel testing, while the temporal aspect was met through stratigraphic excavations, and frequency seriations were used to determine if there was cultural lineage continuity present.
Statistical analysis of sherd distributions shows clear evidence of spatial and temporal continuity, indicating that one Late Woodland/Mississippian occupation was present at 22CH515, 22WI865M, and 22WI865S. At each site, the coincidence of shell-tempered pottery and grog-tempered pottery was strong. At sites 22CH515 and 22WI865, shell-tempered pottery occurred with grog-tempered pottery in 38 of the 42 test units (90.47 percent).
The data gathered through systematic shovel testing and excavation were used to construct three separate seriations, with differences between them relating to the duration over which the sites were occupied. In each seriation, regardless of duration, the sites showed clear continuity, and thus cultural inheritance, between the assemblages. From these results, a strong case can be made for lineage continuity between the Late Woodland and Mississippian components at these sites; i.e., the components are simply arbitrary divisions of single occupations.
It appears that this area was relatively heavily inhabited during the Woodland period in general. During the Late Woodland period, the population seems to have begun nucleating along the terraces of the major waterways (Triplett 2008a:Figure 1.2). At some point between A.D. 1000 and 1100 the area was abandoned, but not before shell-tempered pottery started being made. The length of occupation during the period in which shell-tempered pottery was made seems to vary. At several sites, the Mississippian end of the occupation was so ephemeral that no additional shell-tempered pottery could be located during the systematic shovel testing (e.g., 22CH514, 22CH516, 22CH814, and 22WI508). Other sites, though, such as 22CH515 and 22WI865, seem to have had more substantial occupations at this relatively late date. It may be that these were some of the last sites in the area to have been abandoned.
While more absolute dates from excavation contexts would be helpful in further testing these hypotheses, the case for settlement continuity across traditional culture-historical boundaries and abandonment of the area early in the Mississippian period cannot be falsified based on the results of this research.
