Sources - Papers
A Pilot Study of High Precision Radiocarbon Dating at the Red Wing Locality
Clark A. Dobbs
A paper presented at the 58th Annual Meeting of the Society for American Archaeology St. Louis, Missouri
April 14 -18,1993
This research was supported by a grant from the National Science Foundation (NSF Project Number DBS-9011744). Copies of the final report for this project are available upon request.
The Red Wing Locality is a dense complex of Mississippian-related and Oneota village sites and earthworks at the junction of the Mississippi and Cannon Rivers in southeastern Minnesota. It has long been hypothesized that the Locality functioned as the northernmost node in a Middle Mississippian oriented extraction and magico-religious network. However, a clear understanding of the interaction between the Red Wing Locality and Middle Mississippian groups to the south, particularly at Cahokia, has been hampered by an imprecise chronology at the Locality.
Recent advances in radiocarbon analysis allow significantly greater precision in dating prehistoric events. This paper reports the results of a pilot study which utilizes high-precision radiocarbon dating to refine the chronology of the Mississippian presence at the Bryan Site (21GD4), a fortified Mississippian-related village along the Cannon River within the Red Wing Locality.
A series of 12 samples from 6 pit features excavated at Bryan in 1983 - 1984 were submitted to the Quaternary Research Center at Seattle, Washington, for high-precision radiocarbon assay. We suspected that the pit features were in use within approximately 50 to 75 years of one another. Corollary evidence (ceramic cross-fits and floral and faunal data) seemed to indicate that two of the features (F109 and F202) were in use within a few weeks or months of one another. These features contained Middle Mississippian-like and prairie-related ceramics and were presumably in use during the expansive period of Mississippian interaction with the Locality.
The results of the study suggest that the Bryan was occupied for no more than one or two generations and that the most expansive period of occupation at Bryan occurred between A.D. 1190 and 1223. This places Bryan toward the end of the Stirling Phase at Cahokia (see Hall 1991:10). The rapid growth of Bryan may have been associated with the expansive nature of the Stirling Phase and other socio-political developments at Cahokia and within the American Bottom.
This research demonstrates that high precision radiocarbon dating can be a valuable tool for developing more precise chronologies in archaeology. However, high precision dating is most useful within the broader context of a carefully defined program of chronology building. The results of this study suggest that there are several elements which should be included in archaeological research which employs high precision dating.
First, it is important to identify the materials being dated and to carefully differentiate between samples containing annual and perennial plants.
Second, it is essential to obtain several dates from a single and preferably short-lived provenience unit.
Third, a variety of corollary evidence must be considered in determining which samples to select for dating and which provenience units can be successfully dated.
In March of 1984, I received the results of a series of radiocarbon dates from the Bryan Site (21GD4), a Mississippian-related site in southeastern Minnesota. Extensive excavations at the site during the previous year had been particularly fruitful and I had high hopes that we might be able to develop a tighter chronology for the Mississippian occupation of the area using the Bryan data.
As I reviewed the dates, I was disconcerted to note that the ages for two features (F109 and F202) were apparently separated by several hundred years. Although I was prepared to accept the fact that such a lengthy occupation had taken place at Bryan, the radiocarbon determinations were at variance with other information we had obtained from these two particular pits.
Laboratory staff had discovered several rim sherds from these pits which were clearly from the same vessel. These sherds were from the lower levels of both pits and probably had been deposited at roughly the same time. The features were both similar in form and content, and subsequent analysis of the floral and faunal data indicated that both had been filled with refuse in the late spring or early summer. All of the evidence, except the radiocarbon dates, indicated that these pits were contemporary and had been filled within a few weeks or months of one another.
When the dates were calibrated and plotted at two standard deviations, they overlapped slightly at the extreme ends of their ranges (A.D. 1189 - 1215). It seemed reasonable to believe that the pits had been used during this brief period of time. This interpretation suggested that the occupation at Bryan had been both brief and intense. Moreover, it appeared that this approach to evaluating radiocarbon dates might provide a method for resolving the ongoing problems with the chronology of the Mississippian occupation of the region.
This problem was on my mind during the next several years and in 1988 I learned of the work Dr. Minze Stuiver (Quaternary Research Center, University of Washington) was conducting with high precision radiocarbon dating. It quickly became apparent to me that a series of high precision dates from several pits, combined with information about the relative contemporaneity of those pits, might be an ideal tool for addressing the problems posed by very rapid cultural change in a relatively small region. Dr. Stuiver agreed to process a series of samples from Bryan and the National Science Foundation subsequently funded the project (NSF Award Number DBS-9011744).).
This paper summarizes the results of that project. This research supports the initial hypothesis that the principal occupation at Bryan was relatively brief. The project demonstrates the utility of high precision radiocarbon dating for archaeological projects and makes clear the pitfalls of relying on a single radiocarbon determination for features and other similar depositional units. Finally, this research lays the groundwork for a more ambitious dating project which could evaluate the relative contemporaneity of several of the large Mississippian-related villages near the Bryan site.
This research would not have been possible without the support and assistance of several individuals and institutions. Dr. Minze Stuiver kindly agreed to process the radiocarbon samples from Bryan and has been gracious in assisting me in interpreting the results of these age determinations. The National Science Foundation provided funding for the project and I would like to particularly like to thank Dr. John Yellen for his assistance and patience with this study. Dr. Orrin C. Shane III (The Science Museum of Minnesota) has provided ongoing and stimulating counsel and encouragement about our Red Wing research and the problems posed in developing a chronology for the region. Dr. George Holley (University of Southern Illinois/Edwardsville) has been of tremendous assistance in evaluating the ceramics from the Bryan site and their relationship to sites in the American Bottom of southwestern Illinois. The late Prof. Widen Johnson also provided invaluable encouragement and direction for the entire Bryan project.
The Institute for Minnesota Archaeology provided ongoing support for this study and other aspects of our Red Wing research. Mr. John Redmann prepared the ceramic drawings and some of the figures in this report.
Finally, I wish to thank the Minnesota Dept. of Transportation, which provided funding for the data recovery project at the Bryan site. Without their assistance, most notably the continuing interest and support of Mr. C. P. Kachelmeyer, the materials on which this study are based would have never come to light.
Figure 1: Upper Mississippi Valley and the Red Wing Locality
At the end of the first millennium after Christ, a sweeping series of economic, technological, and social changes transformed aboriginal society throughout much of eastern North America. These changes resulted in the Mississippian Tradition, which represents the second great 'climax' in eastern North American prehistory (Hall 1980). In the Midwestern United States, the most complex manifestations of Mississippian culture are found in the central Mississippi Valley and the lower reaches of its tributary streams in Illinois, Indiana, Kentucky, and Missouri. Although there are hundreds of Mississippian sites scattered through this region, the premier Mississippian center is the site of Cahokia and the series of sites related to it in the American Bottom of southwestern Illinois. Cahokia is the largest prehistoric site in North America north of the Valley of Mexico and, during its most expansive period, appears to have functioned as an intense 'cultural reactor' that profoundly touched and influenced aboriginal groups throughout the central United States (cf. Baereis and Porter, eds... 1984).
The history of Cahokia and Mississippian cultures in the American Bottom is complex and spans more than 400 years (Hall 1991). Of particular interest is the period during the 12th century A.D.... when the northern and western frontiers of the Mississippian world expanded to include much of the Upper Mississippi and Missouri river drainages. During this period, evidence of Mississippian influence appears throughout this region and several focal points for Mississippian interaction appear to develop in the area including the Red Wing Locality of southeastern Minnesota (Figure 1) and the Mill Creek sites of northwestern Iowa (Anderson 1987).
Located some 500 miles upriver from Cahokia and the American Bottom, the Red Wing Locality (Figures 2, 3) is situated at the confluence of the Cannon and Trimbelle Rivers with the Mississippi (Dobbs 1985; Dobbs 1986; Dobbs and Breakey 1987; Gibbon and Dobbs 1991; Dobbs, in press) in Goodhue County, MN and Pierce County, WI. The Locality encompasses an area of some 58 square miles and contains more than 2,000 mounds and earthworks, eight major village sites, and dozens of smaller secondary sites (Figure 2). The Locality is the most northern center of Mississippian interaction in eastern North America and is (arguably) the largest cluster of Mississippian-related sites in the northern Mississippi Valley.
The most distinctive archaeological culture at the Locality during the period A.D. 1050 - 1300 is the Silvernale Phase, which has been defined on the basis of its Middle Mississippian-like ceramic forms and modes (e.g. vessel forms, rolled rims, surface treatment, design, etc.). These forms are unlike any Minnesota ceramics that precede them and document the expansion of Middle Mississippian forms into the region.
Within the Locality, there are four sites that are predominantly associated with the Silvernale Phase (Figure 2). One of these (Mero -Diamond Bluff [47PI02]) is in Wisconsin and situated on a high glacial outwash terrace overlooking the delta of the Trimbelle River at its' confluence with the Mississippi. The other three Silvernale sites are along the south side of the Cannon River in Minnesota. The Silvernale Site (21GD3) is situated on a low terrace near the mouth of the Cannon and overlooks its delta. The Energy Park Site (21GD158) is roughly a mile upriver from Silvernale and is situated on a high glacial outwash terrace overlooking the Cannon. The Bryan Site (21GD4) is roughly three-quarters of a mile beyond Energy Park and is situated on the same terrace, again overlooking the Cannon River.
The Locality was occupied for, at most, a period of some 250 years between about A.D. 1050 and 1300. However, the cultural change and interaction that occurred during this brief span of time was both rapid and profound. Two intertwined but distinct types of change can be distinguished.
First, the indigenous Late Woodland groups were transformed, within a very brief period of time, from hunting and gathering groups to populations relying on maize horticulture as well as hunting and gathering.
This transformation, which presumably occurred during the 11th century, formed the basis for the subsequent Oneota culture which emerges in its most distinctive form as a regional complex that persists until the time of European contact in the mid-17th century.
The second major type of culture change involved the development of long-distance patterns of interaction between the Red Wing Locality and Middle Mississippian groups to the south, as well as interaction with groups in the prairies of southern Minnesota and the eastern Dakotas at the same time.
The relationship between the Locality, situated on the northern frontier of Mississippian culture, and the American Bottom in the Mississippian heartland, remains unclear. Although it is obvious that the Locality was an integral part of the broader Mississippian system and that there were strong ties between the occupants of both areas, the character of this relationship, its evolution, and the subsequent appearance of Oneota culture remains problematic.
Although there has been vigorous debate on this topic (see Gibbon 1972, 1974; Stoltman 1983, 1986; Benn 1984; Dobbs 1982, 1984a:191-229; 1989; Gibbon and Dobbs 1991), only limited progress has been made in resolving these issues, particularly in the Red Wing region. In part, this is due to the fact that until recently archaeological research in the Red Wing Locality has relied on limited excavations primarily conducted during the 1940s and 1950s.
One of the greatest obstacles to investigating these questions is that the existing chronology for sites and artifact forms is coarse and imprecise. This chronological framework is not adequate to allow us to examine key research questions which involve changes in populations density, timing of contact and external relationships with other regional cultures, and interaction between local groups.
Recent advances in high-precision radiocarbon assay have reduced the standard deviation associated with radiocarbon dates to an order of 10 to 20 years (Stuiver, Robinson and Yang 1979; Sieh,, Stuiver and Brillinger 1989; Atwater, Stuiver and Yamaguchi 1991). Thus, it should now be possible to develop refined chronologies spanning considerably smaller intervals.
This paper presents the results of a pilot study which employs high-precision radiocarbon assay to refine our understanding of the precise period of both maximum Middle Mississippi interaction with the Red Wing region and the prairie area to the west. A secondary objective of the paper is to evaluate the utility of high-precision radiocarbon dating in archaeological projects. This study uses data recovered from pit features during excavations at the Bryan Site (21GD4) in 1983 and 1984 to test the hypothesis that many of these features were in use within approximately 50 to 75 years of one another.
The Bryan site (21GD4) is one of the largest and most dense sites within the Locality. Originally the habitation portion of this site covered more than 20 acres (Figure 4) but almost all of this has now been destroyed (see Yourd 1985:11-17). Although the dominant component at the site is referable to the Silvernale Phase, there is some evidence that there is an Oneota component that is temporally discrete from the Silvernale component.
In 1983 and 1984, the Minnesota Department of Transportation funded excavation site remnants in the northwestern portion of the site as part of a bridge replacement project (see Dobbs 1984b, 1987b). This excavation laid the groundwork for subsequent research at the Red Wing Locality.
Fieldwork at the site was conducted between April - October in 1983 and during April, May, and August 1984. A series of 1 x 2 meter excavation units were opened to obtain a sample of material from the surface of the site. Then, the entire site area was mechanically stripped and the stripped surface cleaned by hand. All possible features, postmolds, and other items were mapped and photographed. The northeast quadrant of each feature was excavated and, wherever possible, additional quadrants of each feature were removed. Priority was given to features that were relatively large, intact, and which contained significant quantities of cultural material. The fill from each feature was in most cases water-screened through 1/8" mesh. Numerous samples for flotation were taken to enhance the recovery of fine-scale botanical and faunal remains.
By the end of the 1984 field season, the Bryan team had investigated an area of some 1.6 acres in the northwestern portion of the site. The team had conducted detailed mapping of the site and all vandal's pits, laid out 94 excavation units and excavated 70 of these, identified, mapped and evaluated 558 pit features, and identified, mapped and tested more than 500 postmolds. The western, northern, and probable location of the southern edges of a palisade surrounding the site had been identified. Two well-defined structures and several possible structures had been located. A total of 558 pit features were examined and more than 300 of these were excavated. Excavation yielded more than 166,000 artifacts and more than 120 kg. of bone and charcoal. Detailed description of the Bryan data recovery project and its results are contained in Dobbs (1984b, 1987b, 1989b).
Variation in the ceramic assemblage both within the Locality and specifically at the Bryan Site has been an ongoing research problem. Earlier studies have identified different types (e.g. Wilford 1945:34-38; Gibbon 1978; Wilford 1985:24,29-31; Stortroen 1985:43-44). During the last several years, a comparative study of the larger ceramic rimsherds from Bryan has been underway in cooperation with Dr. George Holley, a specialist in the ceramics of the Cahokia Site (e.g. Holley 1988). The objective of this study is to refine the ceramic typology for the Red Wing Locality and to identify both differences and similarities between these ceramics and those from Cahokia. The initial sort of the ceramics used variation in rim form, particularly variation in rolled rims, as the principal criterion for grouping. Several preliminary types have been identified and of particular relevance to this study are three different types of rolled rim vessels. All of these are similar to Middle Mississippian forms at Cahokia, particularly during the Stirling Phase.
Type 1 vessels (Figures 5 - 7) do not have any traces of a neck, the rolled rims are relatively high, and the inside of the rim is characteristically faceted. The body/shoulder junction in these vessels is very sharp and decoration is confined to the shoulder itself. Design elements are generally rectilinear. All of the vessels are shell-tempered. Many are smudged and some are burnished although this is not a universal characteristic of Type 1 vessels.
Type 2 vessels have a short but distinctly noticeable neck below the rolled rim. The overall morphology of these vessels is the same as Type 1 vessels, but design motifs are generally curvilinear in form. These vessels are generally shell-tempered although there are a few grit-tempered pieces. None of these vessels are smudged or burnished.
Type 3 vessels represent 'true' rolled rims. The vessels have no necks, the rim is rounded and fully rolled over, and the interior of the rim is not faceted. Again, vessel form is essentially the same as Types 1 and 2, and decoration is similar to Type 2 vessels. Smudging and burnishing are not present.
Shell-tempered bowls are present at Bryan and at least 10 bowls are part of the assemblage excavated in 1983 and 1984. Although bowls are typical of Middle Mississippian complexes, they are not found in Oneota assemblages.
In addition to the rolled rim types, other ceramic forms in the assemblage include short and medium, straight-rimmed vessels and vessels with excurvate rims. The sample of these forms is smaller and the variation more complicated than for the rolled-rim ceramics. Likewise, there are several different kinds of grit-tempered ceramics that are related to prairie complexes such as Cambria, Mill Creek, Great Oasis, and some of the Middle Missouri Tradition materials. Because of the small number of these pieces and the small size of each piece, we hesitate to assign them to any specific archaeological complex at this time. Finally, one rimsherd is present at Bryan that is most probably affiliated with the Sandy Lake complex of northern Minnesota.
Prior to 1990, 22 radiocarbon dates had been obtained from the Bryan, Silvernale, Bartron, Mero (Diamond Bluff) and Double sites (Johnson 1965; Bender, Bryson and Baerreis 1971, 1978; Shane 1981; Dobbs 1982, 1984b:58; Steventon and Kutzbach 1987). These dates have standard deviations which range from 50 to 200 years and the 95% confidence interval for any date may be as small as 196 years (for a standard deviation of 50 years) or greater than 500 years (Figures 8, 9; Tables 1, 2). All of the dates are from different provenience units, making it impossible to compare or evaluate several dates from a single context. Using these dates, it is impossible to know whether all the major sites were occupied at one time, sequentially, or repeatedly reoccupied for relatively brief periods.
This chronological framework is inadequate to examine changes in population density, the timing of contact and external relationships with other regional cultures, and the interaction between local groups. To evaluate and test existing models of Oneota and Mississippian development and interaction, it is necessary to have some understanding of the rate of culture change and population density within the Locality. Using the existing chronological framework, it was impossible to determine whether the sites within the Locality represent contemporary villages which were all occupied at the same time (implying rapid population growth and high population density) or a series of villages that were occupied one at a time (implying low population density).
One essential element of the research at the Locality is to develop a more precise chronological framework that identifies the temporal relationships between the various sites and artifact forms. The principal tools used to achieve this goal will be high-precision radiocarbon assay and cross-dating of sites and features using ceramic evidence.
This research is a first step in developing such a chronology for the Red Wing Locality. The study focuses on one specific period and one specific set of artifact forms.
The primary research objective of the project is to more precisely date the period during which there is evidence for interaction with both Middle Mississippian groups to the south and plains groups to the west.
Secondary research objectives include:
* Obtain more precise dates for the Middle Mississippian ceramic types and modes present at the Bryan Site.
* Test the hypothesis that features containing these types of ceramics are contemporary and tightly bounded in time.
* Determine whether the numerous features scattered across the portion of the Bryan Site excavated in 1933 and 1984 are contemporary or not.
* Evaluate the potential of high-precision radiocarbon dating at late-prehistoric sites in the Midwestern United States.
Although a variety of dating methods are available to archaeologists (Bradley 1985), the materials available for dating at the Locality limited our choices to radiocarbon, thermoluminescence, AMS dating, and paleomagnetism. Until recently, none of these methods could produce the precision required to develop a chronology for a brief period of rapid cultural change.
Recent advances in conventional radiocarbon analysis now enable significantly greater precision in dating prehistoric events. Systematic high-precision measurements, using large proportional counters with extremely low backgrounds, have been conducted at the Quaternary Isotope Laboratory since 1973 (Stuiver, Robinson and Yang 1979; Atwater, Stuiver, and Yamaguchi 1991). These counters require fairly large samples (about 7 g of carbon) and have extremely low backgrounds. Whereas typical counters and accelerator mass spectrometric (AMS) analyses produce errors of the order of 50 - 100 years, the high-precision counters produce radiocarbon dates with standard errors of 12-20 years.
Radiocarbon dating is a valuable archaeological technique but is not a panacea for resolving all chronological problems (Shots 1992:202-203). Technical problems like sample contamination, inter-laboratory error, counting error and relatively large standard deviations, and the need to calibrate radiocarbon dates can all be sources of concern in dating projects. A particular problem in dating at the Bryan Site is that the primary material available for radiocarbon assay is wood charcoal.
Annual plants like corn and nuts contain only the radiocarbon from the single year that they were alive. Perennial plants absorb radiocarbon throughout their lifespan. In trees, the radiocarbon for a given year is deposited only in the annual ring for that year. A radiocarbon date taken from the innermost rings of a 300 year old oak tree would therefore be several hundred years older than a date obtained only from the outer few rings of the same tree. Since it is the time of death of the tree that is of interest, because the wood was presumably used shortly thereafter, this type of "old wood" problem has potentially serious implications for archaeological dating.
Wood charcoal is abundant both at the Bryan Site and at other large villages in the Red Wing Locality. Although a variety of woods are represented, oak is the most common species used. As Zalucha (1987:420) observes:
"All of the genera present except oak are present only as occasional specimens. The Bryan distribution suggests an almost single-minded reliance on oak, especially the white oak subgenus. Oak is dominant in every excavation unit. Although charred specimen counts cannot be directly related to amounts of wood fed into fires, the overwhelming preponderance of oak charcoal leaves no doubt that it was used well out of proportion to the other woods in the fires which contributed to this analysis. Differential breakdown favoring oak could certainly obscure the real frequency of other wood use in a few contexts but for oak to dominate in all contexts, it must have been the preferred fuel."
This problem may not pose insurmountable obstacles for chronology building in the Red Wing region. This conclusion is based on the following observations:
a.) Although oak is a long-lived tree, the oldest rings at the center of the tree are proportionally smaller than the outermost rings and therefore contributed smaller amounts of "old" radiocarbon. Wood charcoal dates are therefore 'composites' of the radiocarbon contributed each year to the various tree rings. Radiocarbon is present in proportionally larger amounts for more recent tree rings and this should tend to counterbalance the effect of the radiocarbon contained in the older rings. Unless people are intentionally selecting and burning only heartwood from large trees, the dates obtained from wood charcoal should approximate the date of death of the tree, although such dates will be somewhat older than the actual date.
b.) Oak is a particularly hard, tough wood to cut. The occupants of the Bryan Site did not have steel tools, but rather employed groundstone axes and other methods to obtain firewood. It seems highly unlikely that they sought out and felled very large, old oaks for firewood. Rather, it is probable that they focused their attentions on younger trees, dead branches, and windfalls. The potential for large amounts of very old wood in the charcoal sample seems unlikely.
c.) The wood charcoal in the trash pits at the site is the end product of a long process of use and modification. Logs, branches, and smaller pieces were mixed during the collection process, in the woodpile, and the charred remains were commingled in the ashes of the hearth. This process of commingling was continued when one or more hearths were cleaned and the contents tossed together into the trash pit. The net result of this type of commingling is that the various portions of the tree are 'homogenized', reducing the potential for serious error due to old wood.
Another potential problem is associated with the calibration of radiocarbon dates. One of the key assumptions of radiocarbon dating is that the production of radiocarbon in the atmosphere and other reservoirs has been constant through time. Research during the last 20 years has demonstrated that there has been fluctuation in radiocarbon production. Depending on what period of time is under consideration, dates expressed in radiocarbon years may be older or younger than the calender age. Moreover, there was considerable variation in the production of radiocarbon during the 11th and 12th centuries when the Bryan Site was occupied and a particular radiocarbon date from this time may actually represent one of several different calendar dates.
A third potential source of confusion is the counting error associated with each date. This error may add additional uncertainty about the interpretation of any series of radiocarbon determinations.
The overlapping dates from Features 109 and 202 obtained in 1984 suggested one possible approach to resolving these three problems. As noted above, Features 109 and 202 are probably contemporary, but two earlier dates from these features overlapped only at the extreme ends of their range (A.D. 1189 - 1215). Because the trash pits at the Bryan Site were short-lived and were used for only a few days or weeks, each pit itself would date to a single year. While one date from each pit might be misleading, it should be possible to obtain a more accurate age determination by examining the overlap between several dates from the same pit.
To test this assumption, two additional samples were submitted for dating from Feature 109 and 202. We reasoned that if the assumption is correct, then all of the dates from these two features fall within the general range around A.D. 1189 -1215 and this approach will be used in evaluating the other features in this study. If it is incorrect, then there should be wide variance in the pairs of dates from each feature.
Schiffer (1987) and others (see Shott 1992:203) have discussed the importance of evaluating all the available evidence when using radiocarbon. This approach is employed here and the following controls were built into the research design:
First, two samples from each of six pit features were submitted for radiocarbon assay.
Second, features that are either deep bell-shaped pits or deep irregularly shaped pits were selected for this study. These types of pits should be less susceptible to the possibility of contamination by later, intrusive deposits that were not recognized in the field. Only features that were completely excavated and for which the field and laboratory documentation is excellent were selected for this study.
Third, features that appear to be relatively contemporary were selected for this study. Features 109 and 202 appear have been used in the same year, and the cultural materials in the other features are sufficiently similar to suggest that they are all relatively contemporary.
Finally, one large sample of maize was submitted for radiocarbon assay. Samples of maize that are large enough for high-precision dating are rare at the Bryan Site. However, Feature 505 yielded both maize and wood charcoal samples. This maize sample served as a control to evaluate the relative difference between maize and wood charcoal and to help evaluate the period during which the site was actually occupied.
The samples were forwarded to Dr. Anthony Zalucha who identified the wood charcoal in each. As Zalucha began to work with the samples, he noted that corn was present in all but three of them (908-1093; 908-202-1-4a; 908-202-1-4b) and suggested that the presence of corn would bias the test results. He suggested a simple and cost effective procedure for removing the corn and this suggestion was incorporated into the research design. Zalucha's and correspondence, which contains a complete description of the sorting procedure and detailed wood identification for each sample, is presented in Dobbs (1992, Appendix Two) of this report.
After Zalucha completed the identification of the samples, they were forwarded to Dr. Minze Stuiver and processed at the Quaternary Isotope Laboratory at the University of Washington in Seattle. The processing of the samples was straightforward. According to Stuiver (Dobbs 1992, Appendix 1): "To achieve the quoted precision, we use large samples (7gC converted to CO2 for a single counter filling) in counters with low backgrounds (circa 2.0 counts per minute) and count for 4 days. Standards (background and oxalic acid, 3 day counts) are run on successive weekends.
Calibration curves have been constructed by measuring the radiocarbon age of tree rings of known age (see for example Stuiver 1982). All dates in this report were calibrated using the program CALIB prepared by Stuiver and Reimer (1986). The radiocarbon age dataset selected for calibration contains a ten year (decadal) record from 1950 to 2490 BC (Stuiver and Becker 1986).
Dates obtained from all sources prior to 1990 were calibrated without a laboratory error multiplier since this information was not available. The twelve dates for this study were calibrated using the lab error multiplier for the Quaternary Isotope Laboratory which is 1.6. Thus, the pre-1990 dates and the dates for this investigation are not, in the strictest sense, directly comparable. In certain instances, pre-1990 and high precision dates are combined in the same figures. These figures are for illustrative purposes and should not be considered precise combinations of the two sets of dates.
Many of the dates had several different possible calendar ages. All possible calendar ages are presented in the tables and figures presenting the radiocarbon data.
An overall plan map of the Bryan Site is shown in Figure 10. This map provides the location of the features described below.
Feature 95 is a deep, irregularly shaped pit which is roughly in the center of the portion of the site excavated in 1983 and 1984. It is surrounded by a series of other pits, most of which were quite shallow. Feature 95 was completely excavated.
Feature 95 was circular in plan view and roughly 90 by 90 cm in size. It was 85 cm in depth. The feature fill was predominantly dark silt and there were three well defined ash lenses within the pit (Figure 11).
There were a wide variety of tools and artifacts recovered from this feature, and the raw material of which stone tools and debitage were made is rather heterogeneous.
The ceramics from this feature include 7 rimsherds including a Type 2 rolled rim with a loop handle and two rims with flaring, excurvate rims. One fragment of a possible bowl is present. A red-slipped bodysherd and a very finely made, black burnished bodysherd are also present. This last sherd was clearly not made at the Red Wing Locality and is the only sherd from the excavation that is definitely of non-local origin.
The floral and faunal material from this feature have not been analyzed.
There were eight samples of wood charcoal from this feature weighing a total of 206.10 grams. Two samples (908-95-1-4 and 908-95-3-2) were submitted for radiocarbon assay.
Sample 908-95-1-4 (QL-4489) consisted of 39.45 grams of wood charcoal from Natural Level 4 of the northwest quadrant of Feature 95. This basal natural level is an ash lens containing Oneota-like rimsherds.
This sample was predominantly oak (41 fragments) but also contained willow (4 fragments), elm (1 fragment) and birch (1 fragment).
Sample 908-95-3-2 (QL-4490) consisted of 42.5 grams of wood charcoal removed from Natural Level 2 of the southeast quadrant of Feature 95. This level is an ash lens which is between 15 and 40 cm above the ceramic concentration containing Field Specimens 294 and 293.
This sample was predominantly oak (42 fragments) but also contained ash (5 fragments).
An earlier radiocarbon date (Beta 8844) of 840 n 70 RCY (calibrated at 95% confidence: A.D. 1050 - 1265) was obtained from the feature in 1983.
Features 109 and 202 are both deep, bell-shaped pits that were probably used for storage of corn and subsequently refilled with refuse. These features extend to a depth of ca. 1.4 meters (45 feet) below ground surface. Both features are in the southwestern portion of the site excavated in 1983-1984 and are immediately adjacent to the west palisade wall. We suspect that they are associated with a poorly defined structure in the vicinity of N10-20 W50-60.
Both features have internal natural stratigraphy that is complex (Figures 12, 13) and an unusually high number of artifacts. Faunal material is abundant and well preserved. Floral remains, although well preserved, are limited in number.
The artifact assemblage from Feature 109 is diverse. Three triangular projectile points, 2 notched projectile points, 3 endscrapers, 4 bifacial tools or knives, 1 piece of worked antler, 6 pieces of worked bone, 1 bone bead, 1 bone whistle or flute, 1 bone ornament, a mano used in processing corn and other plant foods, and a sandstone shaft abrader were recovered from the feature.
Seven different types of lithic raw material are present in the feature. Cedar Valley and Prairie du Chien are present in almost equal amounts and Hixton Silicified Sandstone is also present in significant quantities. Tongue River Silicified Sediment, a material normally associated with the prairie area to the west is also present. Tongue River is rare at the Bryan Site and the relatively high number of flakes of this material in Feature 109 is unusual.
The feature contains a minimum of three Type 1 rolled rim vessels. Each of these vessels is black as a result of smudging and has a burnished appearance. One rolled rim sherd from the lower level of the feature fits with two sherds from Feature 202. A portion of a bowl is also present, as is a small grit-tempered rimsherd.
Preservation of faunal material is excellent and 7,258 skeletal elements are present in the feature. Shane (1989) has analyzed the bone from this pit and indicates that 7,154 of the bones are identifiable to the class level only. Of these, 6,767 are fish.
The mammalian bone elements that were identifiable to genus or species level in Feature 109 represent at least 1 white-tail deer, 1 elk, 1 black bear, 2 mink, 1 grey fox, 1 wolf, 1 canid (i.e. canine), 1 beaver, 1 cottontail rabbit, 1 mouse, 1 13-lined ground squirrel and 1 short tail shrew were present. Three types of birds were identified, including 1 Canada Goose, 1 Pigeon Hawk, and the remains of three songbirds (Passerines). Nine types of fish were identified, including Dogfish, Gar, Catfish, Bullheads, Bass, Walleye/Sauger, Redhorse, Rockbass and 1 Sucker.
The floral material from Feature 109 has not yet been analyzed.
There were three samples of charcoal from this feature weighing a total of 133.8 grams. Two were submitted for radiocarbon assay (908-4-9 and 908-3).
Sample 908-109-4-9 (QL-4491) consisted of 27.25 grams of wood charcoal from Zone Q of the southwest quadrant of Feature 109. This feature is probably contemporary with Feature 202.
This sample was oak (59 fragments).
Sample 908-109-3 (QL-4492) consisted of 40 grams of wood charcoal from the southeastern quadrant of Feature 109. The total weight of the sample was 98 grams and 40 grams (consisting of the larger pieces of charcoal) were selected for dating. The remainder has been retained with the Bryan collection for future research. The sample is from the entire southeast quadrant of the feature which is not particularly desirable. However, since Feature 109 was probably filled rapidly, within weeks or perhaps a month, this composite sample from one quadrant should probably be suitable.
This sample was predominantly oak (57 fragments) but also contained maple (1 fragment) and elm (1 fragment).
An earlier radiocarbon date (Beta-8842) of 920 n 50 RCY (calibrated at 95% confidence: 930 - 965 or 1015 - 1235 A.D.) was obtained for this feature in 1983.
The artifact assemblage from Feature 202 is modest when compared to Feature 109. Two endscrapers, 1 bifacial tool, and 1 unifacial tool were recovered from the feature.
However, the lithic raw material profile is diverse. Seven different types of lithic raw material are present in the feature, Cedar Valley dominates the assemblage but Hixton Silicified Sandstone and Prairie du Chien are present in smaller amounts.
The feature contains one Type 1 rolled rim ceramic vessel, which is smudged black and appears to have a burnished surface. This vessel is represented by two rimsherds from the lowest level of the feature which fit with a rimsherd from Feature 109. Grit-tempered cord-impressed ceramics, and at least 4 bowls were recovered from this feature. Two small rolled-rim sherds with exterior beveling of the rim are also present.
Preservation of faunal material is excellent and a total of 12,324 skeletal elements are present in the feature. Shane (1989) has analyzed the bone from this pit and indicates that 12,097 of the bones are identifiable to the class level only. Of these, 9,646 are fish and 1,611 are indeterminate. The mammalian bone elements that were identifiable to genus or species level in Feature 202 represent, at minimum, 1 white-tail deer, 1 raccoon, 2 grey fox kits (babies), 1 fox (Vulpes ?), 1 carnivore (cf. Gulo - wolverine), 1 beaver, 2 mice, 1 chipmunk, 1 rock squirrel, and 1 small rodent. Three types of birds were identified including, at minimum, 1 Canada Goose, 1 Sparrow Hawk, and 1 Passerine (songbird). Thirteen types of fish were identified including, at minimum, 13 redhorse, 2 suckers, 3 Buffalo fish, 1 Quillback, 1 Channel Catfish, 1 bullhead, 2 bass (Micropterus spp.), 2 walleye/sauger, 1 Northern Pike, 1 perch, 1 Black Crappie, 1 White Bass, and 2 Freshwater Drums.
The botanical remains from Feature 202 have been identified by Zalucha (1987: Table 4). The charcoal from the feature is dominated by members of the white oak and red oak group. In addition, birch, ash, willow, maple, walnut or butternut, and elm are present. The diversity of tree species present in the charcoal indicates that the people occupying the site were obtaining wood from the upland forest, perhaps a mesic forest and a floodplain forest.
The density of charred seeds in Feature 202 is low. However, he (Zalucha 1987: Table 11) identified seeds of Amaranthus (Amaranth), Chenopodium (goosefoot), Rhus cf. glabra (smooth sumac), Galium (bedstraw), and Physalis/Solanum in the pit. Further, he discovered 3 wild rice seeds (Zizania aquatica) and 2 sunflower (Helianthus spp.) seeds. Charred corn fragments were also found in the feature.
This feature contains 5 charcoal samples weighing a total of 159.6 grams. Two (908-202-4b and 908-202-14b) were submitted for radiocarbon assay.
Sample 908-202-1-4a (QL-4493) was derived from natural level 4 in the northwest quadrant of Feature 202 which contained 85.5 grams of charcoal. This charcoal was divided into two samples for this study. Sample 908-202-4a contains the large pieces of charcoal which weigh 30.6 grams. The remainder was submitted as Sample 908-202-4b.
A wide variety of ceramics were recovered from Feature 202 and included both shell and grit tempered vessels and bowls. This sample is from a level which contained black, smudged, rolled rim ceramics. One of the rims from this level fits with a rim fragment from Feature 109 and is presumably from the same vessel.
This sample was predominantly oak (54 fragments) but also contained cottonwood (1 fragment).
Sample 908-202-1-4b (QL-4494) weighed 54.4 grams and is from the same provenience unit as Sample 908202-4a. It contained the smaller fragments of charcoal as described above.
This sample was predominantly oak (56 fragments) but also contained cottonwood (1 fragment).
An earlier radiocarbon date (Beta 8840) of 740 n 50 RCY was obtained using a sample of oak and ash charcoal from Feature 202. The calibrated date for Feature 202, expressed as a 95% confidence interval, is A.D. 1189- 1318.
Several kinds of independent data suggest that these features are contemporary and were in use at the same time during the late spring or early summer months at the end of the 12th or beginning of the 13th centuries.
The strongest evidence are the ceramic cross-fits between the two features. Fragments of a single Type 1 rolled rim vessel were found in the lower levels of both Feature 109 and 202 (Figure 5). The fact that these fragments were found in the lower levels of the features minimizes the possibility that one portion of the vessel was deposited in one of the pits much later from another setting.
The faunal evidence provides information on the season the pits were in use. As Shane (1989) comments:
"The occurrence of very substantial fish remains in Features 109 and 202 indicate fishing as an important activity, probably carried out during the warm months of the year. The occurrence of the kit [baby] foxes in F202, the goose and hawk remains in both F109 and F202, and the numerous songbird remains in both of these features suggest that they were open and used during the same season, probably late spring and summer. Although the faunal evidence cannot be used to prove or verify the contemporaneity of Features 109 and 202, the assemblages from these two features are in no way inconsistent with possible contemporaneity."
Although the botanical remains from Feature 109 have not yet been analyzed, those from Feature 202 are consistent with a late spring or early summer occupation. The presence of a few seeds of wild rice, charred corn, and two sunflower seeds probably represent remnants of food stored over the winter, since these plants are normally not available for harvest until early autumn.
The assortment of artifacts found in these two features are intriguing. The ceramic assemblage contains both grit and shell-tempered materials. The grit tempered vessels are very different from the Silvernale Phase vessels but are obviously contemporary with them. The Silvernale ceramics include three large vessels that have been smudged black and burnished. This surface treatment is quite rare at the Bryan Site and we suggest that these vessels were of some particular significance. Likewise, more bowls were found in these two features than in the entire remainder of the features excavated. Bowls are a distinctive component of the Silvernale assemblage and presumably were used for serving or presenting food (or other items?) in a non-utilitarian setting.
The assortment of non-ceramic artifacts in Feature 109 is unusual. The number of projectile points and scrapers is rather high. Moreover, the bone whistle or flute, the bone bead, and the mano for grinding corn (or other plant foods) are rare artifact forms at this site.
The lithic raw materials in both pits are unusual. There is a wide range of non-local material present and the numbers of Cedar Valley chert is high. Further, the relatively high number of Tongue River flakes in Feature 109 strongly suggests a link with the prairie area to the west.
Taken as a whole, the assemblage from these two features is characterized by a high level of non-local ceramic and lithic materials and high numbers (relative to other features) of non-utilitarian artifacts. The faunal remains are diverse and the number of individuals of various species is higher than one would expect for a single family to use during a short period of time.
Based on this description of Features 109 and 202, we hypothesize that these features represent the remains of a communal gathering of people from the Bryan Site and visitors from the west. These types of gatherings or congregations of people have been documented ethnographically and commonly took place from late spring to early autumn. This type of gathering would be consistent with the suggestion that the Red Wing Locality functioned as a northern node of trade and interaction between the prairies and the central Mississippi Valley. The presumed season of use of these pits is also consistent with a model in which groups from the west periodically visited the Red Wing region for trade and other activities.
The radiocarbon dates are problematic. The most parsimonious interpretation of these dates is that they indicate that the features were filled with refuse sometime between ca A.D.1189 to 1215.
This feature is a bell-shaped pit in the northwestern portion of the site. It is on the margin of the bluff overlooking the Cannon River and is surrounded by a series of other pit features.
Feature 411 was circular in plan view and roughly 90 cm by 90 cm in size. It is a bell-shaped pit 110 cm in depth. The feature f 11 was a very dark brown gravelly loam with several large ash lenses and a pottery concentration (Figure 14).
The artifact assemblage is relatively homogeneous and is dominated by shell-tempered rimsherds. The raw material profile is contains only a modest amount of Hixton Silicified Sandstone and is dominated by equal amounts of Cedar Valley and Prairie du Chien chert.
There are large rimsherds from a minimum of 2 (and more probably 3) Type 1 vessels. However, only one of these vessels is smudged and it is not burnished. Also, there are two Type 2 rimsherds present.
The faunal and floral remains from this feature have not been analyzed.
There were seven samples of charcoal from this feature weighing a total of 117.85 grams. Two samples (908-411-5-4 and 908-411-6-2) were submitted for radiocarbon assay.
Sample 908-411-5-4 (QL-4495) consisted of 25 grams of wood charcoal from Level 4 in the south half of Feature 411. This level was a deep concentration of refuse within the feature and ranged in depth from 46 to 170 cm below ground surface.
This sample was dominated by oak (42 fragments) but also contained ash (9 fragments) and elm (6 fragments).
Sample 908-411-6-2 (QL-4496) consisted of 205 grams of wood charcoal from arbitrary Level 2 in the north half of Feature 411. This level was between 20 and 40 cm below ground surface. This sample is stratigraphically above Sample 908-411-5-4.
This sample was more heterogeneous than the other samples. Oak (26 fragments) and ash (23 fragments) were present in almost equal amounts while white pine (4 fragments), elm (2 fragments) and cottonwood (2 fragments) were also present.
No dates had been previously obtained from this feature.
This feature is a bell-shaped pit in the northwestern portion of the site, inside and immediately adjacent to the north palisade line.
This feature was circular in plan view and was roughly 90 by 95 cm in size and 80 cm in depth. The feature fill was very dark brown silty loam and there was one large ash lens in the center of the feature (Figure 15).
The artifact assemblage is relatively homogeneous and is dominated by shell and grit-tempered rimsherds. The raw material profile contains a very small amount of Hixton Silicified Sandstone and is dominated by local Prairie du Chien chert with a smaller proportion of Cedar Valley chert.
Eighty percent of a large Type 2 vessel was recovered from this feature. Although this vessel is clearly of local manufacture, the decoration and vessel form are very similar to the Ramey Incised forms in the American Bottom. Also, five rimsherds from at least one grit-tempered vessel were recovered below the Type 2 vessel. These rimsherds have notched lips, smooth surfaces, and are typical of archaeological complexes found in southwestern Minnesota and the eastern Dakotas.
The floral and faunal material from this feature have not been analyzed.
Two samples of charcoal, weighing a total of 51.10 grams were available from the lowest levels of this feature. Two samples (908-450-5-2 and 908-450-5-3) were submitted for radiocarbon assay.
Sample 908-450-5-2 (QL-4497) consisted of 25.05 grams of wood charcoal removed from Natural Level 2 in the north half of Feature 450. This sample is from outside the boundaries of Feature 451 which intruded into Feature 450. Grit tempered, smoothed surfaced ceramics were recovered in this level. This level was stratigraphically above the level containing a largely complete vessel that is very similar to Ramey Incised.
This sample was predominantly oak (44 fragments) but ironwood (1) and a species like Morus rubra (5 fragments) were present.
Sample 908-450-5-3 (QL-4498) consists of 23.8 grams of wood charcoal from Natural Level 3 in the north half of Feature 450. This sample is outside the boundaries of Feature 451 which intrudes into Feature 450 and is stratigraphically above the level containing a largely complete vessel that is very similar to Ramey Incised.
This sample was almost exclusively oak (57 fragments).
No earlier radiocarbon dates were available from this feature.
This feature is a deep, irregularly shaped refuse pit immediately south of (and probably associated with) the square house (?) structure in the northeastern portion of the site.
This feature appeared as an irregularly shaped stain 100 by 135 cm in plan view and is 88 cm in depth. The internal stratigraphy was somewhat complex and a large ash lens was present at the top of the feature. Other distinctive zones within the feature could be defined (Figure 16).
The artifact assemblage is heterogeneous and contains a variety of ceramics and tools. The raw material profile is also heterogeneous and only 24% of the raw materials is local Prairie du Chien chert. The remainder is non-local Cedar Valley, Tongue River, and Hixton Silicified Sandstone.
There are numerous very small rolled rimsherds from this feature, and one larger Type 2 rimsherd. Also, there are two rolled rims with an exterior bevel on the rim. A complete miniature vessel was recovered from this feature and its form is almost identical to the Type 2 vessel from Feature 450. Another vessel is represented by four large rimsherds of a Type 2 vessel with nested chevrons (an Oneota decorative motif). Again, the vessel form is very similar to the Type 2 vessel from Feature 450.
The floral and faunal material from this feature have not been analyzed.
Two samples of charcoal were available from this feature weighing a total of 28.00 grams. Two samples (908-505-4-3 and 908-505-3/4-3) were submitted for radiocarbon assay.
Sample 908-505-4-3 (QL-4499) consisted of 14.45 grams of charred corn fragments from Natural Level 3 in the southwest quarter of Feature 505. Natural Level 3 contained the majority of cultural material in the feature and is stratigraphically above Natural Level 4, which contained diagnostic Oneota-like ceramics. Several large corn cobs were removed from this sample prior to submission so that they could be analyzed for paleo-botanical information. We submitted this sample of corn because corn grows in a single season, does not have the potential for the "old wood" problem, and to evaluate differences between wood charcoal and single season plant dates.
The sample consisted solely of corn kernels and fragments.
Sample 908-505-3/4-3 (QL-4500) consisted of 14.3 grams of wood charcoal from Natural Level 3 of Feature 505. It is a combination of 8.15 grams of charcoal from the southeastern quadrant and 6.15 grams of charcoal from the southwest quadrant. Neither of these samples by themselves was large enough for this study. Natural Level 3 is a distinctive natural level described in the fieldnotes as "lOYR2/1 black sticky fill with gravel". There is an intrusive zone in the southeast quadrant at the top of Level 3. However, the maps and field notes indicate that this zone was removed separately. Therefore, it seems relatively safe to combine charcoal samples from the same level in two adjacent quadrants.
This sample was predominantly oak (51 fragments) but also contained elm (1 fragment).
There are no previous radiocarbon dates from this feature.
Twelve high precision radiocarbon dates were obtained from six pit features at the Bryan site. These dates had standard deviations ranging between 13 and 21 years. The dates in radiocarbon years (RCY) and the calibrated calendar ages for these dates are presented at 1 sigma in Table 3 and at 2 sigma in Table 4. The dates are graphed in Figures 17 (1 sigma) and 18 (2 sigma).
There is variation in each pair of dates for each individual feature. One date in each pair tends to be older while the other tends to be younger. This variation is probably caused by variation in the calibration curve, charcoal from older portions of the tree, or both. However, it does not appear that the 'old wood' effect is dramatically skewing the dates. As expected, the date obtained from a sample of maize (QL-4499) was somewhat younger than the wood charcoal dates.
A variety of methods have been used to evaluate series of radiocarbon dates (e.g. Ward and Wilson 1978; Shott 1992. In this study, we explore another method which examines the overlap between the ranges of each pair of radiocarbon dates. This method may be useful in this particular case because each pair of dates is from a single pit feature that was filled with refuse in less than a year. Therefore, variation in the dates is not the result of cultural activity. Rather, it must be the result of the counting error (standard deviation), 'old wood', or because a particular date may be one of several possible dates on the calibration curve. While the range of a single date may not accurately reflect the age of the feature, the overlap in the ranges of several dates should allow us to more precisely evaluate the probable age of the pit.
To test this method, we use Features 109 and 202 which are probably contemporary.
In 1984 one date from each of these features was obtained from Beta Analytic (Feature 109: Beta-8842; Feature 202: Beta-8840). Initially it appeared that these two features were separated in time by approximately 150 years (Figure 19). Since we had evidence that the features were contemporary, it appeared that the radiocarbon dates were in error. However, when the dates were plotted at 2 sigma, the dates overlapped at the extreme ends of their ranges (Figure 19). Since the features were contemporary, it seemed possible that this overlap (A.D. 1189 - 1215) represented a more precise estimate of the age of this feature than either of the dates alone.
As part of this research, two high precision dates from Feature 109 (QL-4491 and QL-4492) and two high precision dates from Feature 202 (QL-4493 and QL-4494) were obtained. Like the earlier Beta dates, one date for each feature was older than the other. However, the central tendency of the high precision dates was within the range defined by the overlap of the Beta dates (Figure 20).
At 1 sigma, the dates for Feature 109 overlap between A.D. 1194 and 1216 and at 2 sigma they overlap between A.D.1159 and 1231 (Table 5, Figure 20). The dates for Feature 202 overlap at 1 sigma between 1190 and 1211 and at 2 sigma between A.D.1190 and 1223 (Table 5, Figure 20). These ranges are consistent with the initial estimate of A.D. 1189 - 1215 for the age of these pits. Further, the central tendency of the high precision dates themselves is around A.D. 1200.
It seems likely that wood from older trees was being burned and discarded into these features. The earlier dates may represent tree rings from the innermost portion of the tree while the later dates may represent more recent rings. This factor, when combined with multiple dates from the calibration curve, and the 1 and 2 sigma ranges of these dates, is the probable cause of the variation in the dates.
One date had been obtained for Feature 95 in 1984 (Beta-8844) and two additional high precision dates (QL-4489 and QL-4490) were obtained for this feature. The overlap between the high precision dates at 1 sigma is A.D.1190- 1208 and at 2 sigma is A.D.1157- 1221. The overlap between the high precision dates and the Beta date are consistent (Figure 21) and provide a tighter estimate of the age of this feature.
Based on this analysis of the overlap of dates from these features, it seems reasonable to conclude that this method is useful in evaluating several dates from the same feature when that feature is short-lived and/or contemporary with features.
The high precision radiocarbon dates from the Bryan Site are presented at 1 sigma ranges in Figures 22 and at 2 sigma in Figure 23. The overlap between each pair of dates is shown as the hatched region between each pair of dates. The numeric values for this overlap is given in Table 5.
The high precision dates for Features 109 overlap at 1 sigma between A.D. 1194 - 1216 and at 2 sigma between A.D. 1159 - 1231. The dates for Feature 202 overlap at 1 sigma between A.D.1190- 1211 and at 2 sigma between A.D.1190- 1223. The overlap of all four dates at 1 sigma is A.D. 1194 - 1211 and at 2 sigma is A.D.1190- 1223. The original estimate of the age of these pits based on the Beta dates was A.D.1189-1215. Using the high precision dates, the most conservative estimate of the age of these features is the 2 sigma overlap of A.D.1190- 1223 and this age estimate seems reasonable for these pits.
One of the dates from Feature 95 (QL-4489) appears to be older than the other (QL-4490) and has several possible values. The small earlier outlier of QL-4490 also overlaps with QL-4489 but most probably is the result of multiple dates in the calibration curve. Excluding this outlier, the overlap for these dates is A.D. 1190 - 1208 at 1 sigma and A.D. 1157 - 1221 at 2 sigma.
The dates for Feature 411 overlap between A.D.1191- 1217 at 1 sigma and A.D. 1157 - 1257 at 2 sigma.
The dates for Feature 450 overlap between A.D. 1189 - 1224 at 1 sigma and A.D. 1156 - 1261 at 2 sigma. The earlier outlier of QL-4497 is considered to be a product of variation in the calibration curve.
The dates for Feature 505 overlap between A.D. 1230 - 1242 at 1 sigma and between A.D. 1214 - 1263 at 2 sigma. This overlap is of particular interest since one of these dates (QL-4499) was derived from a sample of maize. The maize date is later than the wood charcoal date for this feature and the overlap between these two dates is skewed toward the later date. While the maize date is probably a more accurate representation of the actual time of use of Feature 505, the overlaps are not as helpful as in the other pairs of features.
There is little significant variation in the estimated age ranges of the features based on the overlap of dates for each pit. In examining Figures 22 and 23, it appears that all of the features are essentially contemporary. This is consistent with the other data available on these features.
At 1 sigma (Figure 22), the overlap of dates for Feature 450 is A.D. 1189 - 1224. This period encompasses all of the other features (excluding the maize date) and suggests that the most conservative estimate of the age of these pits at 1 sigma is A.D. 1189 - 1224.
At 2 sigma (Figure 23), the overlap of dates for Feature 450 again encompasses all of the other features (A.D. 1156 - 1261). However, the maximum overlap for the dates of Feature 202 at 2 sigma is A.D. 1190 1223 and this overlap includes portions of all of the other overlaps between pairs of dates. The most conservative estimate of the age of the pits at 2 sigma is A.D. 1156 - 1261. However, given the central tendencies of all the dates, the overlap of dates for each feature at 1 sigma, and corollary evidence that these pits are relatively contemporary, this longer period of time seems too conservative. A more reasonable estimate is the 2 sigma overlap of the dates for Feature 202 (AD. 1190 - 1223).
Based on this discussion, we conclude that a conservative estimate for the age of the six pit features in this study at 1 sigma is A.D. 1189 - 1224. At 2 sigma, a reasonable estimate is A.D. 1190 - 1223. These age ranges are essentially the same, suggesting that these pits were most probably in use during the last decade of the 12th century and the first two decades of the 13th century A.D. These estimates are fairly consistent with the earlier dates from the Bryan Site (Figures 24, 25).
The chronology for other sites within the Red Wing Locality remains coarse and imprecise. However, the results of this study demonstrate the utility of high precision radiocarbon dating and provide a baseline for comparing Bryan to other key sites in the region (Figures 26, 27).
Eleven of the high precision dates were obtained from samples of wood charcoal, dominated by oak. One date (QL-4499) from Feature 505 was obtained from a large sample of maize. As discussed above, radiocarbon dates from perennial plants probably understate the actual age of the feature being dated.
In this instance, the maize date (QL-4499) has a value of 770 RCY n 15. The wood charcoal date from Feature 505 has a value of 839 RCY n 21. There is a difference of 69 RCY between the two dates, suggesting a variation of roughly 8% between maize and charcoal dates at Bryan.
The average age for the eleven wood charcoal samples is 846 RCY. The uncalibrated maize date is, therefore, roughly 9% younger than the average uncalibrated date from the wood charcoal samples.
The calibrated values for the maize date are A.D. 1230 - 1242 or 1256 - 1277 at 1 sigma and A.D. 1214 - 1281 at 2 sigma. The wood charcoal dates tend to cluster around A.D. 1200. The calibrated maize date, in this instance, tends to be only about 4 to 6% younger than the wood charcoal dates.
Because maize is a single season plant, the maize date provides a more accurate estimate of the age of the pit than do the wood charcoal dates. Therefore, it seems reasonable to assume that the maximum actual period of use of Feature 505 is between roughly A.D. 1214 and 1281 (based on the 2 sigma calibrated values for QL-4499). The period of use of the features dated using wood charcoal should be somewhat earlier than that dated by the maize date, but extreme outliers are probably functions of variation in the calibration curve and are not indicative of the age of the features.
The difference between the wood charcoal and maize dates suggests that it may be helpful to think of 'relative' and 'absolute' chronologies at the Bryan Site. Relative chronologies can be built with wood charcoal dates and are useful in evaluating the relative contemporaneity of pit features and the probable length of occupation of the site. Dates obtained from samples of maize and other annual plants can be used to precisely date the actual period of occupation of the site. Such dates also provide a useful way to check the accuracy and validity of interpretations based on wood charcoal dates.
In this instance, the maize date indicates that the actual date of occupation of the Bryan Site was somewhat later than indicated by the wood charcoal dates. It also tends to support our interpretation that the pit features are relatively contemporary and that the principal occupation at the site was brief. Because most other Mississippian dates are obtained from wood charcoal samples, the estimate based on wood charcoal will be used in comparing the chronology of the Bryan Site to other Mississippian. sites.
This pilot study of high precision radiocarbon dating has refined the chronology of the Bryan Site and contributed to the interpretation of the role of Bryan in the broader Mississippian system during the 12th and 13th centuries A.D.
The features which were dated contained Mississippian ceramic types and modes that are very similar to ceramics from Cahokia during the Stirling Phase. The relatively short time during which the features in this study were in use suggests that the variation in Mississippian-like ceramics at the site is not primarily a function of change through time.
Three features also contained ceramics that are derived from the Plains area in association with Mississippian ceramics. The association of Plains-like and Mississippian forms in these features suggests that interaction between Bryan and sites to the west was occurring during the Stirling Phase at Cahokia.
Because Bryan was a complex entity and has been largely destroyed, discussion of the period of occupation of the site is somewhat speculative. However, the palisade surrounding the site provides a spatial boundary for one maximum period of occupation at the site and it seems reasonable to believe that it is this period of occupation that we have dated in this study. Based on our results, we suggest that the most expansive period of occupation at Bryan, represented by the portion of the site surrounded by the palisade, occurred between A.D. 1190 and 1223. We hypothesize that Bryan functioned as a central place for trade and interaction between more westerly groups in the Plains and Cahokia. We suspect that Bryan was first occupied during the last quarter of the 12th century, grew rapidly to its maximum extent between A.D. 1190 - 1223, and quickly declined thereafter. In this model, the occupation and use of the Bryan site encompassed no more than one or two generations.
Hall's revised chronology for Cahokia places (Hall 1991:10) places the Stirling Phase between A.D. 1100 and 1200. This fits well with the model presented here. We suspect that the rapid growth of Bryan was intimately associated with the expansive nature of the Stirling Phase and other socio-political developments at Cahokia and within the American Bottom.
This study also demonstrates that high precision radiocarbon dating can be a valuable tool for developing more precise chronologies in archaeology. However, high precision dating is most useful within the broader context of a carefully defined program of chronology building. The results of this study suggest that there are several elements which should be included in archaeological research which employs high precision dating.
First, it is important to identify the materials being dated and to carefully differentiate between samples containing annual and perennial plants. Identification of the wood(s) contained in samples is important since some species like oak are quite long-lived and can potentially provide misleading results. It is equally important to separate out annual plant remains from wood charcoal since dates from these materials can be interpreted in different ways. Finally, dating even one sample of annual plant material provides a valuable reference point to evaluate wood charcoal dates.
Second, it is essential to obtain several dates from a single and preferably short-lived provenience unit. If two dates had not been available from each pit feature in this study, the interpretation of the results would have been considerably less useful.
Third, a variety of corollary evidence must be considered in determining which samples to select for dating and which provenience units can be successfully dated. This study was strengthened by the knowledge that Features 109 and 202 were contemporary. This type of corollary information provided an important way to cross-check the results of the radiocarbon dates with other data sets. It also allowed us to refine our interpretation of dates which otherwise might have been confusing or equivocal.
© 1999 The Institute for Minnesota Archaeology
Email us: email@example.com
Updated 29 Jun 1999