Diachronic variation in late archaic lithic technology at the Big Pine Tree site, Allendale County, South Carolina

Citation
Diachronic variation in late archaic lithic technology at the Big Pine Tree site, Allendale County, South Carolina

Material Information

Title:
Diachronic variation in late archaic lithic technology at the Big Pine Tree site, Allendale County, South Carolina
Creator:
McIntosh, Thomas I.
Place of Publication:
Tampa, Florida
Publisher:
University of South Florida
Publication Date:
Language:
English
Physical Description:
xvi, 204 leaves : ill. (some col.) ; 29 cm.

Subjects

Subjects / Keywords:
Stone implements -- Analysis ( lcsh )
Stone implements -- South Carolina -- Allendale County ( lcsh )
Dissertations, Academic -- Applied Anthropology -- Masters -- USF ( FTS )

Notes

General Note:
Thesis (M.A.)--University of South Florida, 2001. Includes bibliographical references (leaves 187-204).

Record Information

Source Institution:
University of South Florida
Holding Location:
Universtity of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
028871051 ( ALEPH )
50727871 ( OCLC )
F51-00158 ( USFLDC DOI )
f51.158 ( USFLDC Handle )

Postcard Information

Format:
Book

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

DIACHRONIC VARIATION IN LATE ARCHAIC LITHIC TECHNOLOGY AT THE BIG PINE TREE SITE, ALLENDALE COUNTY, SOUTH CAROLINA by THOMAS I. MCINTOSH A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Anthropology University of South Florida December 2001 Major Professor: Brent Weisman, Ph.D.

PAGE 2

Office of Graduate Studies University of South Florida Tampa, Florida CERTIFICATE OF APPROVAL Master's Thesis This i s to certify that the Master's Thesis of THOMAS I. MCINTOSH with a major in Anthropology has been approved for the thesis requirement on December 14, 2001 for the Master of Arts degree. Examin ing Committee: Major Professor: Brent Weisman, Ph.D. Member: Robert Tykot, Ph.D. Kennethlfassaman, Ph. D.

PAGE 3

Thomas I. Mcintosh 2001 All Rights Reserved

PAGE 4

ACKNOWLEDGMENTS I wish to express my gratitude for the helpful, scholarly advice of Dr. Kenneth Sassaman and Dr. Albert Goodyear, the editorial advice of Dr. Robert Tykot, and the continued optimism and encouragement of Dr. Brent Weisman. I would also like to acknowledge the technical expertise and encouragement of Susan Yewell, who helped me get the final drafts out the door.

PAGE 5

DEDICATION Th i s thesis is dedicated to those who encouraged me to finish it, especially my mom and dad; and to all those other students now early in the process of their papers, who doubt the possibility that it can be done. ii

PAGE 6

TABLE OF CONTENTS LIST OF FIGURES v LIST OF TABLES ix LIST OF PLATES xii LIST OF SYMBOLS AND ACRONYMS xiii ABSTRACT xiv CHAPTER 1. INTRODUCTION 1 Project Overview 1 Research Questions 6 Organization of Text 6 CHAPTER 2. THEORETICAL PERSPECTIVE 8 The Nature of Technological Change 8 The Hunter-Gatherer Model 11 Theory of Debitage Analysis 15 CHAPTER 3. ENVIRONMENTAL CONTEXT 22 Climatic and Botanical Trends 23 Paleohydrology 29 Lithic Resources 33 CHAPTER 4. CULTURAL CONTEXT 42 Paleoindian Period (11,500-9,900 BP) 45 Archaic Period (9,900-3,000 BP) 50 Early Archaic (9,900-8,000 BP) 51 Middle Archaic (8,000-5,000 BP) 52 Late Archaic (5,000-3,000 BP) 57 CHAPTER 5. BACKGROUND OF ARCHAEOLOGICAL RESEARCH 63 Pen Point Site 64 Site 38BR34 71 Big Pine Tree Site 72 Summary of Previous MALA Research 78 iii

PAGE 7

CHAPTER 6. METHODOLOGY 85 Field MethodsData Recovery 85 Standardization of Data 86 Statistical Analysis 87 Analysis of Variance {ANOVA) 100 Post hoc Testing 119 All Whole Flakes 119 Cortical Whole Flakes 121 Flakes with !S50/o Cortex 124 Flakes with >50/o Cortex 132 Riverine Quarry Cortical Flakes 139 Upland Quarry Cortical Flakes 148 Flakes with Heavy Weathering 157 Thermally Altered Flakes 166 CHAPTER 7. SUMMARY OF INFERENCES AND CONCLUSIONS 175 APPENDICES 180 Appendix 1: Summary Statistics of MALA Bifaces Excavated at the Pen Point Site. 181 Appendix 2: Lithic Artifacts Catalogue from the Big Pine Tree Site 1985 Underwater Recovery. 182 Appendix 3: Ceramic Artifacts Catalogue from the Big Pine Tree Site 1985 Underwater Recovery 183 Appendix 4: Regression Analysis for All Whole Flakes in Level 1. 184 Appendix 5: Regression Analysis for All Whole Flakes in Level 2 185 Appendix 6: Regression Analysis for All Whole Flakes in Level 3. 186 REFERENCES 187 iv

PAGE 8

LIST OF FIGURES 1. Locational Map of the Big Pine Tree Site. 2 2. Sea Level Change Curve for the South Carolina Coast 27 3. Soils Stratigraphy at the Big Pine Tree Site. 37 4. Five Coastal Plain Sites with Stratigraphy Similar to The Big Pine Tree Site. 39 5. Cultural-Ethnic Chronology for the Southeastern United States. 46 6. Middle and Late Archaic Bifaces from the Savannah River Region. 55 7. Allendale County Chert Outcrops and Quarries Showing the Big Pine Tree Site and Site 38AL135. 73 8. Topographic Site Map of the Big Pine Tree Site. 76 9 Plan Map of the Big Pine Tree Site, 1993-1996, Showing Test Units A-G. 79 10. Ogive Comparison of Flake Size from Debitage Assemblages at the Big Pine Tree Site and the Pen Point Site 82 11. Comparison of Thermal Alteration Between the Big Pine Tree Site, the Pen Point Site, Site 38BR34, and the Early and Middle Archaic Periods. 83 12. Frequency per Cubic Meter of Thermally Altered Flakes at the Big Pine Tree Site in Levels 1-3. 83 13. Cumulative Size Frequency Distribution of Adjusted Raw Data for All Whole Flakes in Levels 1 3. 93 v

PAGE 9

(LIST OF FIGURES continued) 14. Cumulative Size Frequency Data from the Big Pine Tree Site and Patterson's 1990 Experimental Biface Reduction Data. 97 15. Cumulative Size Frequency Data from the Big Pine Tree Site and Three Other Experimental Data Sets. 97 16. Sample Means and Standard Deviations for Flake Size Frequency in Levels 1-3. 99 17. Bell Curve of Normally Distributed Data Set. 102 18. Comparison of Logarithmic and Regression Trajectories for All Whole Flakes in Levels 1-3. 103 19. Regression Lines for All Whole Flakes in Levels 1-3. 111 20. LOG10 (fjm3) Values for All Whole Flakes in Levels 1-3. 112 21. Rate of Resource Depletion During MALA Occupation at the Big Pine Tree Site. 113 22. F-Distribution Indicating the Outer Limits of Significance, or F-Critical Value. 116 23. Adjusted Size Frequency Data for Whole Flakes with Cortex in Levels 1-3. 127 24. Cumulative Size Frequency for Whole Flakes with Cortex in Levels 1-3. 127 25. LOG10 (ffm3) Values for Whole Flakes with Cortex in Levels 1-3. 128 26. Regression Lines for Whole Flakes with Cortex in Levels 1-3. 128 27. Adjusted Size Frequency Data for Whole Flakes with >50/o Cortex in Levels 1-3. 135 vi

PAGE 10

(LIST OF FIGURES continued) 28. Cumulative Size Frequency for Whole Flakes with >50/o Cortex in Levels 1-3. 135 29. LOG10 (f/m3 ) Values for Whole Flakes with >50/o Cortex in Levels 1-3. 137 30. Regression Lines for Whole Flakes with >50/o Cortex in Levels 1-3. 137 30. Adjusted Size Frequency Data for Whole Riverine Quarry Flakes in Levels 1-3. 144 32. Cumulative Size Frequency for Whole Riverine Quarry Flakes in Levels 1-3. 144 33. LOG10 (f/m3 ) Values for Whole Riverine Quarry Flakes in Levels 1-3. 145 34. Regression Lines for Whole Riverine Quarry Flakes in Levels 1-3. 145 35 Adjusted Size Frequency Data for Whole Upland Quarry Flakes in Levels 1-3. 152 36 Cumulative Size Frequency for Whole Upland Quarry Flakes in Levels 1-3. 152 37 LOG10 (f/m3 ) Values for Whole Upland Quarry Flakes in Levels 1-3. 154 38 Regression Lines for Whole Upland Quarry Flakes in Levels 1-3. 154 39. Adjusted Size Frequency Data for Whole Heavily Weathered Flakes in Levels 1-3. 162 40. Cumulative Size Frequency for Whole Heavily Weathered Flakes in Levels 1-3. 162 vii

PAGE 11

(LIST OF FIGURES continued) 41. LOG10 (f/m3 ) Values for Whole Heavily Weathered Flakes in Levels 1-3. 163 42. Regression Lines for Whole Heavily Weathered Flakes in Levels 1-3. 163 43. Adjusted Size Frequency Data for Whole Thermally Altered Flakes in Levels 1-3. 171 44. Cumulative Size Frequency for Whole Thermally Altered Flakes in Levels 1-3. 171 45. LOG10 (f/m3 ) Values for Whole Thermally Altered Flakes in Levels 1-3. 172 46. Regression Lines for Whole Thermally Altered Flakes in Levels 1-3. 172 viii

PAGE 12

LIST OF TABLES 1. Summary of Hanson et al's. 1981 Model of Prehistoric Savannah River Valley Ecology. 24 2. Whole Flake Size Frequency Data (not standardized) from the Big Pine Tree Site and the Pen Point Site. 80 3. Volume in Cubic Meters of Undisturbed Midden Soil Excavated Per Unit. 87 4 Frequency Distribution for All Whole Flakes in Level 1. 92 5. Frequency Distribution for All Whole Flakes in Level 2. 92 6. Frequency Distribution for All Whole Flakes in Level 3. 93 7. T-Test Output of Regression Values for All Whole Flakes in Levels 1-3. 106 8. ANOVA Output for All Whole Flakes in Levels 1-3. 115 9. ANOVA F-Statistics for All Attributes in Levels 1-3. 118 10. T-Statistics of Segmented and Paired Sample Data for All Whole Flakes in Levels 1-3. 120 11. Adjusted Size Frequency Data for All Whole Flakes with 50/o Cortex in Level 1. 125 12. Adjusted Size Frequency Data for All Whole Flakes with 50/o Cortex in Level 2. 125 13. Adjusted Size Frequency Data for All Whole Flakes with 50/o Cortex in Level 3. 126 ix

PAGE 13

(LIST OF TABLES Continued) 14. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with <50/o Cortex in Levels 1-3. 130 15. Frequency Distribution for Whole Flakes with >50/o Cortex in Level 1. 133 16. Frequency Distribution for Whole Flakes with > 50/o Cortex in Level 2. 133 17. Frequency Distribution for Whole Flakes with >50/o Cortex in Level 3. 134 18. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with >50/o Cortex in Levels 1-3. 139 19. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 1. 141 20. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 2. 141 21. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 3. 142 22. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with Riverine Cortex in Levels 1-3. 146 23. Frequency Distribution for Whole Flakes with Upland Cortex in Level 1. 149 24. Frequency Distribution for Whole Flakes with Upland Cortex in Level 2. 149 25. Frequency Distribution for Whole Flakes with Upland Cortex in Level 3. 150 26. T Statistics of Segmented and Paired Sample Data for Whole Flakes with Upland Cortex in Levels 1-3. 156 X

PAGE 14

(LIST OF TABLES Continued) 27. Frequency Distribution for Whole Heavily Weathered Flakes in Level 1. 159 28. Frequency Distribution for Whole Heavily Weathered Flakes in Level 2. 159 29. Frequency Distribution for Whole Heavily Weathered Flakes in Level 3 160 31. T-Statistics of Segmented and Paired Sample Data for Whole Heavily Weathered Flakes in Levels 1-3. 164 31. Frequency Distribution for Whole Thermally Altered Flakes in Level 1. 168 32. Frequency Distribution for Whole Thermally Altered Flakes in Level 2. 168 33. Frequency Distribution for Whole Thermally Altered Flakes in Level 3. 169 34. T-Statistics of Segmented and Paired Sample Data for Whole Thermally Altered Flakes in Levels 1-3. 173 xi

PAGE 15

LIST OF PLATES 1. MALA Preform Fragments and Bifaces Recovered from the Big Pine Tree Site. 5 2. Flake Morphology. 19 3 Morrow Mountain Stemmed Point Recovered from the Big Pine Tree Site. 56 4. MALA Stemmed/Notched Point Recovered from the Big Pine Tree Site. 56 5. MALA Stemmed/Notched Bifaces from the Pen Point Site. 67 6 MALA Biface Morphology Terminology Using an Example From the Pen Point Site Feature 14. 68 7 Flake Sizing Grid Used to Determine the Size Category Of Each Flake. 89 8. Large Size Flake with Riverine Cortex. 123 9. Medium Size Flake with Upland Cortex. 123 10. Heavily Weathered Flakes. 157 11. Thermally Altered Coastal Plain Chert Flakes. 167 xii

PAGE 16

df a m X y AD ANOVA BHT BP cmbs DOE MALA SCIAA SRS S d. LIST OF ACRONYMS AND SYMBOLS degrees of freedom critical value sum population means sample means square root independent variable dependent variable Anno Domini Analysis of Variance Backhoe Trench Before Present Centimeters Below Ground Surface United States Department of Energy Middle-Archaic Late-Archaic South Carolina Institute of Archaeology and Anthropology Savannah R iver Site Standard Deviation xiii

PAGE 17

DIACHRONIC VARIATION IN LATE ARCHAIC LITHIC TECHNOLOGY AT THE BIG PINE TREE SITE, ALLENDALE COUNTY, SOUTH CAROLINA by THOMAS I. MCINTOSH An Abstract A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Anthropology University of South Florida December 2001 Major Professor: Brent Weisman, Ph.D. xiv

PAGE 18

The goal of this thesis research is to examine a Late Archaic (5,000-3,000 years BP) lithic tradition by statistical analysis of debitage attributes. It is expected that variances in the analysis reflect variances w ithin the technological tradition. Such techno lo gical variances may be stylistic or functional or both, and may reflect adaptations to environmental or social conditions. The research site is the Big Pine Tree quarry site (38AL143) in the South Carolina Coastal Plain of the Savannah River Valley. Data recovery was conducted within a dark band of midden soil lying between Middle and Late Archa ic deposits. This midden came to be known as the "MALA Midden." MALA is an acronym for "M iddle Archaic-Late Archaic, first used in 1984 by Dr. Kenneth Sassaman to refer to the technological tradition that produced disti nctive, f i nely crafted lanceolate spear points discovered at the Pen Point site (38BR383) at the United States Department of Energy (DOE) Savannah River Site (SRS) in the Coastal Plain of South Caro l ina (Sassaman 1985b). The discovery of anomalies within MALA technology will contribute to the on-going refinement of Sou theastern Late Archaic studies, specifically related to site selection, site utilization, volume of production, and intensity of occupat io n. During Spring 1995 research excavations at the Big P ine Tree site (38AL143), Alle ndal e County, South Carolina, l ithic debitage was XV

PAGE 19

collected from three arbitrary, contiguous levels of the MALA Midden. Quantitative statistical observations of whole flakes were analyzed using Analysis of Variance (ANOVA) for homogeneity of flake size-frequency and seven attributes related to decortication, quarry source, degree of weathering, and thermal alteration. Statistically significant diachronic variation among these data supports a hypothesis that associates MALA lithic resource management with specific environmental and social phenomena. Although biface manufacture was clearly a primary focus of lithic technology between 4,820 years BP and 3,980 years BP, statistical analysis shows significant technological variation among and between the groups of data, particularly between the upper-most sample and the lower two. Discussion of these findings follows, concluding that the Late Archaic quarriers of the Big Pine Tree Site were delimited by nature as to the availability of raw material. Sp ecifically, sea level changes that coincided with early and middle occupations of the MALA midden inundated the riverine chert outcrops and rendered them inac ce ssible. Major Professor: Brent Weisman, Ph.D. Prof esso r, Department of Anthropology Date Approved: I 'f #o v 2 co' 1 xvi

PAGE 20

CHAPTER 1 INTRODUCTION Project Overview The problem of identifying Late Archaic lithic technological variation and related environmental and social factors at the Big Pine Tree quarry site (38AL143) in Allendale County, South Carolina (Figure 1) is addressed in this paper. This thesis sets out to discover whether statistically significant variation in the frequencies of nine size categories within eight attribute classes exists in the socalled MALA (Middle Archaic-Late Archaic) Midden of the Big Pine Tree Site. The term "significance" is described within the parameters of inferential quantitative statistical analysis: that variations from random phenomena may be measured within specified confidence levels, and deemed "significant" within certain confidence levels. The MALA component of the Big Pine Tree site, referred to as the "MALA midden," actually post-dates the Middle Archaic Period, and is represented by a dark layer of organically rich midden soil approximately 60-90 em below surface (cmbs). Lithic attributes related to decortication, quarry source, degree of weathering, and thermal

PAGE 21

Figure 1. Locational Map of the Big Pine Tree Site. 60 so'trTH P.tCJ.FIC O ,CA.N a.t.j)i!f TENNESSEE A I ---.. .......... GEORGIA 0 lOMite' 0 lO Kolometen --_] ----.---v 7 I I .,( ../ ..,,r ..-..-0 40" so= 120 160' l2"N C>GeoSystems \ c: ATLANTIC OCEAN 2

PAGE 22

alteration are the focus of this study. Specifically, attribute categories include the following cortical and non-cortical categories: flakes with cortex, flakes With > 50/o cortex, flakes derived from the riverine quarry at the Big Pine Tree Site, and flakes derived from the upland quarry at the Big Pine Tree Site, heavily weathered flakes, and thermally altered flakes. Data from these categories are normalized and statistically tested for size frequency variations. Attributes selected for analysis in this thesis are visually recognizable without magnification. It is proposed that by measuring the frequency of whole flakes within size categories for each attribute within each of three arbitrary and contiguous levels, statistical analysis of key debitage attributes will produce a faithful, reconstructive view of stone tool manufacturing activity during the one or more occupations indicated by the midden feature. If whole-flake attribute distribution throughout the midden is statistically homogeneous, then no significant change in tool manufacturing techniques, tool use, or tool re-use may be inferred. If whole-flake attribute distribution throughout the midden is heterogeneous, then variation in technological organization may be inferred. Technological change may be related to environmental or social causes. 3

PAGE 23

This research is important to the study of the archaeological record because it identifies a debitage profile within a stratigraphic context that is shared with a specific in-situ tool tradition and radiocarbon dates. This correlation presents a distinct signature for one or more temporal or cultural entities. Site utilization and level of stone tool production may also be inferred from this model of debitage analysis. These data offer a new perspective to the body of knowledge derived from previous MALA studies. The scope of tool morphology within the MALA midden at the Big Pine Tree Site includes a wide variety of long, slender, excurvate, stemmed, hafted, lanceolate points and large quantities of hand-size, roughed out preforms. Preforms are the bifacially flaked, early stage form from which bifacial tools may be crafted (Plate 1). These were found in direct association with thousands of debitage flakes. A distinctive characteristic of the MALA midden is the large quantity of debitage, suggesting large scale production activity. This characteristic is consistent with the nearby Pen Point site (38BR383), a stratified flood plain site containing deposits representative of Early Archaic through Woodland occupations (Sassaman 1985b). The relationship of stone tool manufacture to settlement and subsistence strategies has been studied extensively in the Southeast and elsewhere (Daniel 1982:34). The earliest stages of manufacture 4

PAGE 24

Plate 1. MALA Preform Fragments and Bifaces Recovered from the B i g Pine Tree Site. may be expected to be found at sites established at the source of the raw material, or quarry sites, which were utilized for extraction, thermal alterati on of raw material, and biface production. The voluminous level of biface production at Big Pine Tree, the accessibil ity to high quality chert from both riverine and upland sources, and the suggested longevity of occupation, suggest such site utilization. Technologically distinct from horizons above and below the midden, the unique signature of MALA debitage attributes indicates a distinctive technological tradition. 5

PAGE 25

Research Questions Although the discovery of the systemic components of a specific lithic technical tradition may eventually have an integral effect on the regional archaeological record, its immediate effect is on contributions to the local archaeological record, and more accurately on contributions to the specific site record. Therefore questions formulated within the context of research design are site specific, drawing on a body of theory that relates subsistence strategies to environmental resources (Goodyear 1989): 1) Is significant variation present in the debitage sample excavated from test units at Big Pine Tree, and if so, how does such variation manifest itself in the attributes studied? 2) If variation exists, do parallel events in the environmental record offer clues for the reasons for change? Organization of Text The initial discussion (Chapter 2) considers the theoretical posture of processualism inherent in the present study, and the nature of technological change. This posture views the basis of technological change as rooted in response to environmental change Recent 6

PAGE 26

literature concerning debitage analysis is reviewed Chapter 3 develops background information essential to the understanding of prehistoric lifeways in the Southeast, by considering environmental and cultural contexts. Chapter 4 chronicles recent MAlA research by focusing on the Pen Point and Big Pine Tree sites. Data derived from the Big Pine Tree Site are compared with data derived from the Pen Point Site. Chapter 5, Methodology, discusses implementation of testing, research design, and analytical methods. Chapter 6 discusses standardization of data and application of statistical analysis. Analysis results are discussed. Chapter 7 concludes thi s thesis by reviewing research questions and their implications. The importance of this study to the archaeological record is evaluated on both local and regional levels. 7

PAGE 27

CHAPTER 2. THEORETICAL PERSPECTIVE The Nature of Technological Change A fundamental focus of anthropology is the discovery of cultural change and the processes from which it develops. The discovery of cultural change among Late Archaic populations of the Savannah River Valley requires the enmeshing of data obtained from archaeological artifacts, features, and sites, and the subsequent discovery of the systematic relationships within and between them. The discovery of such relationships is most readily achieved by using inferential statistical analysis. Much can be learned about the behavior of hunter-gatherers from statistical analysis of the technological processes employed by them, particularly stone tool manufacture. Choices regarding subsistence, settlement, and technological strategies were limited then, as now, by a finite range of possibilities. The demeanor of decision-making helped to develop culture and its variations, and is now reflected in the discovery of archaeological artifacts and features. Environmental and social factors affect individual and group decision-8

PAGE 28

making, which led to variations in technological organization. As opportunities to strengthen the economy were identified, new improved tool types and manufacturing methodologies were developed. Departures from old styles and traditions employed to better accommodate improved lifeways are reflected as variations within and among debitage attributes. With increased acceptance of the processual paradigm evoked by the New Archaeology of the 1960s, lithic studies evolved to new levels of acceptance by professional archaeologists in concordance with increasing applications of lithic data to behavioral models (Johnson 1993). Theories relating to human behavioral change may be formulated by debitage analysis. Processual lithic theory may be viewed as having two sets of systematically connected, verifiable principles: the set of physical principles pertaining to stone tools themselves, their source, manufacture, and use; and the set of behavioral principles determining how stone tools were used in prehistory (Odell 1996:4). "Design theory" has been the formalization of this sentiment, in response to the growing frustrations that lithic studies rarely address questions relevant to the field of anthropology as a whole (Torrence 1989: 1-2). Design theory postulates specific behavioral responses to problems or constraints encountered in a natural or cultural context. 9

PAGE 29

Thus, such responses are inextricably related to environmental factors. As applied to stone tool manufacture, behavioral responses to procurement or production constraints may be postulated. Response to constraints results in variation. The presence or absence of constraints and responses may be identified through the discovery of variation in the production of lithic tools and corresponding production of waste flakes. Constraints may include 1) adequate task performance; 2) availability and cost of materials; 3) technologies available; 4) the introduction of new technologies; and 5) the return on investment of various production and use alternatives, including tool-life and cost of repair considerations (Hayden et al. 1996: 10). Responses may have been : 1) flexibility, or changes in tool form for different uses (Shott 1986:19; Nelson 1991:70), as well as variations in size and weight, raw material used, and thermal alteration; 2) versatility, or changes in the number of uses for a tool (Shott 1986:19; Nelson 1991:70); 3) development of higher precision tools (Aidenderfer 1991:207); 4) production for durability requiring less maintenance (Bleed 1986; Torrence 1989), or for less durability requiring more maintenance (i.e. resharpening or retooling) in an environment with abundant raw materials; and 5) imposed limitations of time availability (Torrence 1983). 10

PAGE 30

The Hunter-Gatherer Model Processual theory views technology as a cultural subsystem that may be quantitatively analyzed. Such analyses may help to gain a finer-grained perspective of culture change. Throughout most of this period, people relied primarily on an economy of hunting, plant gathering, and fishing-an economy in which technology was fundamental. Culture changes in the Southeastern United States among Archaic people were encouraged by environmental, social, and political agents. Technology is a fundamental part of the economic infrastructure on which society and ideology rest (Harris 1979:55-56; Austin 1997:8). Technology is the study of techniques and the science of interpreting and implementing the methods inherent in individual techniques. A lithic technique is a method of procedure used in employing flintknapping skills. It implies a systematic control of minute and distinguishable detail (Crabtree 1982:50). Technology responds to environmental changes such as resource depletion and climatic variations, in concert with its response to the social and subsistence adaptations that accompany environmental changes. 11

PAGE 31

As populations increased in areas of optimal resource availability, notably riverine and coastal regions, social and subsistence pressures encouraged the invention of pottery, cooking stones, and other technologies. Although identification of technological change through statistical analysis does not readily reveal specific insight into the causal effects of cultural change, such inferences may be made utilizing these data and those of other fields such as paleobotany, geoarchaeology, and environmental studies. The earliest people inhabited the Savannah River valley at least as early as 11,500 BP, when small bands of Paleoindians moved about the region, exploiting a wealth of late-Pleistocene resources. They utilized a h ighly developed stone tool technology now associated with the Clovis point. By 8,000 BP, a more temperate climate caused an increase in the growth of deciduous, food-bearing vegetation. Populations grew as they prospered under these favorable post glacial conditions People became less mobile and exploited a wider range of plants and animals. Early and Middle Archaic tool typologies, in which the development of the tang, or stem becomes apparent, suggest an increased dependence on the hunting of small game (Sassaman et al. 1990:6). By the mid-Holocene, more permanent settlement strategies 12

PAGE 32

had developed. Small shellfish had been added to the diet. Ceram i c technology emerged by the close of the fourth mil l ennium BP. With the increasing sedent ism, stone tool maintenance and recycling became prioritized. Prior cultural assessments carried out on and near the DOE Savannah River site (SRS) i n the M i ddle and Lower Coastal Plain regions of South Carolina by the University of South Carolina Institute of Archaeology and Anthropology, have shown that changes in technology during the increasingly sedentary Archaic Periods are instructive in the reconstruction of settlement range, organization, and subsistence economies (sassaman et al. 1990:319). Of particular interest to archaeologists are behavioral changes related to adjustments in lithic production, design, and procurement, made necessary by the increasing demands of a sedentary lifestyle. SRS lithic assemblage analyses have identified three broad categories of change. Responses to decreased settlement range and limitations of raw material affected 1) raw material selection, 2) production organization, and 3) hafted biface design. Early Archa i c hunter-gatherers depended heavily on accessibility to high quality isotropic Coastal Plain chert, much of which is found in the chert quarries of Allendale County (Goodyear and Charles 1984). Access to these sources was facilitated by organized strategies throughout their expansive settlement range. 13

PAGE 33

By the Middle Archaic, raw material procurement strategies focused on local sources including Allendale County, and production strategies had been modified to adjust for spatial inconsistencies extant between decreasing availability of raw materials and settlement locales Debitage concentrations indicative of intensive secondary biface (cores, preforms, blanks) manufacture on the SRS suggest that chert quarries were exposed and being quarried. Late Archaic evidence of this production and its transport for seasonal provisioning has been found in the upland Sandhills region some 50 km away Production strategies on the SRS exhibit a trend from quarry based stage production, to surplus manufacture on a quantitative level previously unparalleled in the region. This level of production may have been preparatory for seasonal journeys and camps or exchange motivated. Regardless of motivation, labor specialization on at least a part-time basis is indicated (Sassaman et al. 1990:320). Stone tool design in the Archaic not only reflects responses to environmental changes, but also to decreased raw material availability (Goodyear 1989; Bamforth 1986). Limited raw materials for tool making encouraged the manufacture of durable tools that would endure long treks between sparsely located quarries. For those inhabiting quarry areas, a higher degree of flakability at the expense of durability was desirable, since availability of raw materials was not 14

PAGE 34

problematic. The practice of thermal alteration enhanced flakability as demonstrated by previous MALA studies. As decreased mobility placed heightened demands of resource availability on Late Archaic populations, the large, broad-bladed Savannah River Stemmed biface was adopted. This hafted tool functioned as a saw or knife, and could be reduced to produce different tool forms, functioning like a portable core. By the end of the Late Archaic and the onset of the Woodland period, chert sources had been exploited to the point that lithic production was limited to site provisioning and situational manufacturing. Evidence of the scavenging of Late Archaic lithic sites by Woodland period people is extant on the SRS, as documented at 38AK158 (Sassaman et al. 1990:319) and 38AK157 (Sassaman 1993). The large size of Late Archaic lithic materials and their presence at or near the surface made them ideal for Woodland period people to recycle and utilize. Theory of Debitage Analysis Debitage is an important medium for analysis since it most often occurs in primary context. Waste flakes, debris, and manufacturing failures were less likely to be curated than finished tools, during 15

PAGE 35

moves. By using the artifact class of debitage as a component of lithic tool analyses, the issue of context is strengthened. Furthermore, issues surrounding technological innovations, site use, and level of production may be addressed by debitage analysis. In general, flake size decreases as tools undergo the reduction process (Newcomer 1971; White 1981 :93). Sassaman has underscored the importance of flake size considerations in a summary of his Pen Point site study (Sassaman et al. 1990: 133). Citing a previous study which attempted to predict site function by flake size frequency analysis (House and Ballenger 1976), Sassaman reminds us that a preponderance of larger flakes resulting from the manufacture of tools is expected to be found at maintenance camps where tool and preform manufacture are primary occupations, and which are generally located close to quarries. Furthermore, a preponderance of smaller flakes is expected to be found at camps located further away from the quarries where flakes are created primarily from use and maintenance of tools (Sassaman et al. 1990: 137). These studies indicate a high level of confidence for models which relate flake size to site function and/or raw material availability Debitage at Big Pine Tree is found in large quantities throughout the MALA midden Because of its presumed primary context, as well as 16

PAGE 36

the abundance with which it occurs, debitage offers the archaeological analyst direct evidence of lithic technological activity. Crabtree's (1982) description of the lithic reduction process underscores the diagnostic importance of debitage. In order to better understand Crabtree's description, a brief discussion of flake morphology follows. Flake morphology is best understood by picturing the reduction sequence of manufacturing a stone tool from a rough cobble. A flake is detached from a rough cobble or core, and an organized series of flakes is detached from the core. A flake is detached from the core by striking the core with a hammerstone on a flat, discreet, striking platform, which may occur naturally or may be prepared by the knapper. The platform typically detaches with the flake, and may be intact or crushed, depending on the intensity of the hammer strike and the integrity of the core. With respect to impurities which may exist in the core, a relatively consistent amount of force is imparted from the "platform" to the interior of the core. This force is evidenced by tiny arched "lines of force" which become visible on the ventral (interior) surface of the flake upon detachment. Additionally, a "bulb of percussion" develops immediately below the platform on the ventral surface. The dorsal (exterior) surface of a flake may be partially or totally composed of "cortex," a rind-like skin of rough, outer parent material. 17

PAGE 37

Such flakes are expected to be among the earliest detachments in the reduction process, and are often referred to as primary flakes. At the Big Pine Tree Site, cortex appears as two distinct types, each representing a different but adjacent source of raw material. A flake may also contain no cortex. Such flakes, typically coming from the interior of the core, are considered "tertiary" and are produced later in the reduction process The edges of a flake terminate in a sharp, thin, tapered, "feathered" manner, barring manufacturing failure which can produce thick, blunt edges resultant of snaps. Since accurate size data cannot be obtained from broken flakes, only whole-flakes are used for size related analysis. Whole-flakes are qualified by the presence of a platform, lines of force or bulb of percussion, and margins to the extent that maximum width of the flake is discernable (Plate 2). A flake that is not whole, that is lacking in one or more of the preceding attributes, is classified as broken, fragmented, or debris Only whole-flakes are considered for analysis in this thesis. The exclusive use of whole-flakes for analysis is consistent with previous MALA debitage studies (Sassaman 1984). The reduction of a core to a tool requires variously defined stages of manufacture. Waste flakes are discarded during every phase of the process. These debitage flakes are usually more diagnostic than flake 18

PAGE 38

Plate 2. Flake Morphology. scars, for their size, thickness, and shape, and degree of curvature can reveal several manufacturing steps. They can indicate the technique, for they retain the bulb of applied force (platform area), show the method of platform preparation and innumerable other characteristics which indicate the technique. For this reason, a careful study of the flaking debris is a prime requisite in determining the manufacturi ng technique (Crabtree 1982: 1). Different reduction strategies result in characteristically distinct patterns of debris (Morrow 1996:354). For example, a blade industry has a relatively high tool -todebris ratio, whereas the manufacture of a single bifacial tool can produce 19

PAGE 39

hundreds of waste flakes (Newcomber 1971; Henry et al. 1976; Patterson and Sollberger 1978). However, debitage is subject to post-depositional disturbances. During the process of lithic reduct i on debitage would have either been left in its primary context, utilized as tools and curated to a secondary context, or swept up and carried to a dumping place in a post-primary context. The term "post-primary context" is used to differentiate between the removal of flakes from one part of a site to another, and the removal of flakes from the site altogether. Morrow (1996:355) has discussed the curation of flakes utilized as tools Larger flakes are more suitable for utilization than smaller flakes because they are easier to resharpen and maintain, while smaller flakes would have remained largely unused (Kelly and Todd 1988). Vertical displacement of flakes caused by trampling is limited by the penetrability of the soil matrix (Schiffer 1987: 126) and flake size (Gifford-Gonzalez et al. 1985; Stevenson 1991:271-272). Larger items are more susceptible to lateral displacement than smaller items. Schiffer (1972 : 162, 1987) has discussed the intentional removal of flakes to reduce clutter, and noted a correlation between increasing s ite population and decreasing refuse discard locations. Refuse clearing may be either expedient, where debris is simply brushed aside, or systematic, where debris is collected removed, and dumped 20

PAGE 40

elsewhere (Stevenson 1991 :273 276) Limited activity locations such as kill and butchering sites, quarry sites such as the Big Pine Tree Site, and many seasonally occupied sites consist largely of primary refuse (Wilmsen 1970). Primary refuse refers to material discarded at its location of use (Schiffer 1972: 162). With the exception of curated utilized flakes and systematic refuse discard, the post-depositional agents discussed place debitage in a primary or intra-site, post-primary context. Curated utilized flakes are not classified as debitage, but as tools. Systematically discarded refuse may find itself within the site of its origin, in another place, or may be removed from the site of origin. If the method of lithic manufacturing consists of the rules, mechanics, and procedures employed in flintknapping, then technique is the application of method to the execution of a specific plan. Debitage analysis offers keen insight into both method and technique. By registering key attributes of a debitage assemblage, the analyst may produce a resultant signature that is culturally unique. By monitoring variations from that signature, the analyst may infer technological changes that mirror cultural change. 21

PAGE 41

CHAPTER 3 ENVIRONMENTAL CONTEXT The Coastal Plain of the Savannah River Valley has a tradition rich in archaeology, and an ever-changing geophysical environment. Fluvial-estuarine systems continuously dump and remove sediment onto extant surface soils, providing an uninterrupted chronicle of landscape evolution and well-preserved artifacts from past human occupations Much of what is known about the paleo-environment is the result of paleohydrology and geoarchaeology. Pollen studies offer the researcher the ability to reconstruct paleo-environmental climates and botanical trends that provide ecological niches favorable for human adaptation. Long-term fluvial changes influenced the stability of alluvial landforms, the availability and productivity of food and lithic resources, and the suitability of sites for human occupation. Archaeological features and artifacts provide temporal markers for deposition, sedimentation, and pedology which may indicate former utilized land surfaces of past human occupations, and offer opportune conditions for the study of resource availability and site utilization. These data enhance the researcher's ability to view diachronic variations in technology. 22

PAGE 42

Climatic and Botanical Trends Well-preserved datable material, stratigraphy, and pollen in alluvial settings of the Carolinas, Georgia, and Florida document prehistoric climatic and botantical trends in the South Atlantic Slope region. Research focusing on 25 millennia of ecology (Hanson et al. 1981 :26) has been conducted on the Savannah River Site (SRS), and is summarized in Table 1. Late Pleistocene pollen studies in the Piedmont and Coastal Plain provinces indicate a period of full glaciation between 25,000 BP and 15,000 BP. Cold and dry conditions were conducive to vegetation consisting mainly of spruce and jack pine, with some oak and ironwood. These climatic conditions are documented in studies from White's Pond, South Carolina (Watts 1980), Bob Black and Quicksand Ponds in northwest Georgia (Watts and Stuiver 1980), Pigeon Marsh in northwest Georgia (Watts 1975), and Singletary and Bladen Lakes (Whitehead 1965, 1973). During the late glaciation (ca. 15,000 BP), spruce and jack pine declined (Watts 1975, 1980; Watts and Stuiver 1980; Whitehead 1965), and deciduous species such as oak, beech, hickory, black walnut, hemlock, hazelnut, and iron wood dominated 23

PAGE 43

the forests (Watts 1980). These species are indicative of a warmer, moister climate, favorable to human habitation. Regional forests were Table 1. Summary of Hanson et al's. 1981 Model of Prehistoric Savannah River Valley Ecology (Hanson et al. 1981). Episode Climate Vegetation Sites and Dates Full-Glacial Much colder and drier Jack pine, spruce, herbs White's Pond sc 25,000-15,000 BP than present few deciduous species 19,100-12,810 BP (Watts 1980) Late Glac i al Warmer and moister than Oak, hickory, beech, White's Pond, SC 15,000-10,000 BP Full Glacial; cooler and hemlock 12,810-9,500 BP moister than present (Watts 1980) Pigeon Marsh, GA 13,00010,800 BP (Watts 1980) Singletary Lake, NC 11,000 BP (Watts 1975) Bladen Lake, NC 11 ,000 BP (Whitehead 1965, 19 Post Glacial Early (10,000-7,000 BP)oak and hickory predominate ; White' s Pond, SC 10 ,000 BP-pres ent warming and Increase d beech declines sharply; 9,500-7,000 BP moisture gums in c rease (Watts 1980) Late (7,000 BP-present)oaks gradually decrea se; Bladen Lake, NC warming and gradual pine increases. (Whitehead 1965) desiccat i on mod e rn vegetation evi d ent by 7,000 BP Okefenokee Swamp, 5,200 BP (Bond 1971) changing from a patchy, coarse-grained, boreal forestparkland, to a denser, more homogenous, fine-grained, mesic oak and hickory forest. This transition was complete not later than 9,000 BP to 10,000 BP (Watts 1971, 1980; Davis 1983; Larsen 1982). Delacourt and 24

PAGE 44

Delacourt (1987) suggest that mature hardwood forestation in Georgia was in place well before this time. After 10,000 BP, oak and hickory disappear from the record. Hickory and ironwood decline in the record around 9,500 BP and were replaced by sweet gum, black gum, and oak (Watts 1980; Watts and Stuiver 1980). Periods of rapid warming and increased moisture on the South Atlantic Slope are suggested by these data, and indicate an environment favorable for continued human exploitation. As selective environmental factors changed through time, the reproductive successes of human adaptation provided an index for the limiting conditions of a given environment. Hunter-gatherer adaptation may thus be equated with a "lowest common denominator strategy" (Sassaman and Anderson 1995:5), often lacking the means to overcome short-term natural environment change, and therefore adapting to the worst possible conditions in order to avoid the requirements of such sudden changes. The continuity of human occupation at Big Pine Tree since Paleoindian times (11,500 BP-10,500 BP) supports a model of post-glacial warming prior to 10,000 BP, which produced more moisture and resulted in a least common denominator of increased game populations, deciduous vegetation, and mast. Reconfigured landforms resulting from a rising sea-level 25

PAGE 45

and subsequent river channel migration localized lithic resources that enabled human technology to evolve. Between 8,000 BP and 6,000 BP, Coastal Plain oak forests were reduced to less that 40/o, and replaced by pine (Delacourt and Delacourt 1987). The result was a diminished supply of hardwood nuts and habitats of low resource potential for human subsistence. A period of global warming, known as the Altithermal, Hypsithermal, or Climatic Optimum, had dramatic effects of vegetation and fauna, and redefined the relationship between human settlement and techno l ogical requirements to survive (Sassaman and Anderson 1995). Drier climatic conditions that led people to populate riverine areas where fish and shellfish could be exploited in the Tennessee River Valley have been dated as early as 7,500 BP (Carmichael 1977; Sassaman and Anderson 1995). However, others (e.g., Goodyear et al. 1979) contend that mid-Holocene settlement and technological changes in the South Carolina Coastal Plain cannot be conclusively tied to global warming. In the 1,000 years that followed, moist conditions resulting from increased precipitation and sea level rise (Figure 2) led to the development of extensive coastal salt marshes, interior wetlands, cypress swamps, and river flood plains (Brooks et al. 1986; Colquhoun and Brooks 1986; Davis 1983; Delacourt 1985; Delacourt and Delacourt 1983, 1987; Foss et al. 1985; Wright 1976). With regard to 26

PAGE 46

varied elevations on the southeastern Atlantic Slope, it is likely that these environmental changes were first evident in the lower elevations of the Coastal Plain along major drainages that were primarily Figure 2. Sea Level Change Curve for the South Carolina Coast (Sassaman et al. 1990:24). 8 7 6 s 3 2 1 '\ ... ,._ \ 0 0 ,' ,, ,. 1 4 lj 1 i\ : \ ; if I '\'' I\/ .... \i I I --.-I r / v ./, 8 influenced by sea level fluctuations or departures from contemporary environmental conditions. The mild, temperate climate known today in the Coastal Plain region of the Savannah River Valley is comparable to the climate of the Middle and late Archaic periods. Bland (1995) has synthesized contemporaneous climatic conditions for the region. Seaward of the Appalachian Mountains, the region is protected from the severe 27

PAGE 47

winters of the Tennessee Valley to the west (USDOE 1990). Weather systems from the west stall in the Appalachians, while moist, warm weather comes up from the Gulf of Mexico and interacts with Atlantic weather systems moving westward, producing a "rain shadow territory" (Plummer 1983). The annual mean humidity is 70 percent, and average monthly temperatures vary between 48 and 81 degrees Fahrenheit (Langley and Marter 1973). Average rainfall is 47 inches (120.5 em), and is most heavily concentrated during the spring, summer, and early fall (Semlitsch and Gibbons 1991). These conditions provide for a growing season optimum to 220 days per year. Although rainfall is sufficient during the growing season to promote growth, it is less consistent during the crucial periods of the growing season. Between 1904 and 1951, an overall drying trend was evident, and rainfall was not consistent enough to promote the proper growth of either corn or cotton in the Upper Coastal Plain region of the Savannah River Site (Sassaman et al. 1990), which is bordered to the south by Big Pine Tree. These data suggest that the Upper Coastal plain region is not a prime agricultural area from a climatic perspective. 28

PAGE 48

Paleohydrology An understanding of the changing Savannah River fluvial system is imperative for the reconstruction of archaeological site formation there. Large fluvial changes had a direct bearing on the availability of lithic resources, food, and suitable habitation sites. Atlantic and Gulf coastal sea levels were at least 70 meters lower than present during the end of the last glaciation between 11,000 and 15,000 years ago. The melt and northward retreat of the glaciers resulted in inundated coastal areas, and by 9,000 BP sea levels were within several meters of their present position (Colquhoun and Brooks 1986). The rapid rise in sea level resulted in a reduced gradient of the Savannah River in the Coastal Plain region, accompanied by aggradation (the process of building up of a surface deposition), increased channel migration, and associated landforms such as point bars, levees, and terraces. The river channel evolved from a braided to a meandering pattern during this time. Macklin (1956) has suggested that erosion of banks and channel constraint directly effect braiding and meandering. If a channel has relatively weak, low-cohesion banks incapable of withstanding the 29

PAGE 49

hydraulic shear of flowing water, a wide, shallow channel forms, promoting channel disorder and braiding Sandbars and islands with little stability are randomly placed by rapidly shifting channels, and offer few opportunities for cultural resource stability, site formation, and preservation. Channels with strong cohesive banks, able to withstand the hydraulic shear of flowing water, become narrow and deep, promoting a helical flow pattern, which has the form of a spiral or helix. The effect of helical flow is to establish an orderly flow pattern that localizes deposition and erosion and promotes development of the river bend Sediment eroded from the outside of the bend due to greater flow velocity there, is carried downstream and deposited on point bars on the inside of bends, eventually leading to a meandering pattern (Easterbrook 1993). Point bars are generally flood-stage depos its characterized by cross-bedded sands that fine upward and downstream. Successive over bank flooding may result in levees; fine grained, depositionally structureless flood-stage deposits that are often cap bar sequences. Point bar-levee complexes may ultimately form terraces; landforms which are effectively segregated from the active channel, often enhancing landform stability and preservation. Sea level fluctuations influenced variations in channel morphology and related landform stability in the Savannah River 30

PAGE 50

Valley. Holocene sea level change studies by Colquhoun and Brooks (1986) in the coastal-estuarine zone of South Carolina include both archaeological and geological data sets Archaeological data sets include age, location, and elevation of basal portions of shell middens relative to the local datum of the high marsh surface. Geological data sets include marsh stratigraphy, sedimentology, clay mineralogy, dated peats, dated root tree stumps, palynology, and diatoms. Colquhoun and Brooks (1986) found that sedimentation and aggradation, influenced by sea level changes, proceed in an upriver direction causing floodplain development. In coastal areas, deltas form where fluvial sediments are discharged into ocean water. As sediments accumulate, the delta front eventually grades seaward, increasing in elevation with progressive deposition. During periods of raised base sea level, deposition aggrades in the upstream direction and is effectively removed from further downstream transport (Easterbrook 1993). The upstream trend of sedimentation results in the infilling of bays, lagoons, and estuaries that gives way to floodplain development is progressively more recent in the upstream direction, with cohesive, fine-grained, less erodable floodplain sediments influencing the local hydraulic base levels for associate tributaries upstream. 31

PAGE 51

These archaeological and geological data suggest that while modern estuarine development was initiated around 6,000 BP at the time sea level fluctuations began to stabilize, floodplain development in the Upper Coastal Plain was initiated as recently as 4,000 BP (Sassaman et al. 1990). This concept is supported by net sedimentation studies constructed from alluvial terrace site data in the Upper Coastal Plain and Fall Zone of the Savannah River Valley (Brooks et al. 1986). Sedimentation curves from these studies show successively more recent sedimentation in the upriver direction, and resemble the sea-level curve with high rates of sedimentation during the early Holocene, and lower rates during the last 6,000 years when landform stability increased. Paleohydrologic work in the Upper Coastal Plain has focused mainly on the formation of alluvial terraces and point bars, using archaeological data to calibrate the times and duration of alluvial landforms stability. Archaeologically significant alluvial landforms in this region are located on the Aiken Plateau on a series of three terraces (T1a, T1b, T2) formed by down cutting and lateral migration of the Savannah River. Age and elevation of these terraces increases with distance from the present course of the river. The lowest, most recent of these terraces (T1a) contains substantial evidence of continued human occupation beginning with the Early Archaic period 32

PAGE 52

(10,000-8,000 BP), indicating relatively stable landform stability since that time. Lithic Resources Local lithic resources available to early and mid-Holocene humans living in the Savannah River Valley were concentrated in the Piedmont, Fall Zone, and Coastal Plain regions (Sassaman et al. 1988). While this thesis focuses on the Upper Coastal Plain, an awareness of the availability of stone suitable for tool making in nearby provinces is important for understanding technological trends that evolved at Big Pine Tree in the Middle and Late Archaic periods (5,000-4,000 BP). From the Piedmont, early people obtained raw material primarily from igneous and metamorphic formations that yielded quartz, quartzite, and basalts. Limited amounts of low-grade siliceous cherts and jaspers were also available The Fall Zone roughly corresponds to the Carolina Slate Belt formation which contains thick beds of volcanic rocks such as amphibolite, argillite, muscovite and chlorite schist, and limited amounts of rhyolite. Daniel and Butler (1991) have shown that the flow-banded variety of rhyolite seen occasionally at Big Pine Tree and in assemblages across the Carolinas originates at Morrow Mountain, Stanley County, North Carolina (Sassaman and Anderson 33

PAGE 53

1995 : 14). However, these are coarse-grained materials with poor fracture properties, and generally less desirable for tool manufacture than the cryptocrystalline Coastal Plain chert (CPC) found in outcrops near Big Pine Tree in Allendale County. Coastal Plain chert is entirely sedimentary and is derived from two major formations. The Oligocene Flint Ridge formation extends northward from upper Tampa Bay to the South Carolina Coastal Plain, terminating in Allendale County, and westward to southeastern Alabama. These Tertiary cherts are classified as silicified grainstones (Upchurch 1984) and have superior conchoidal fracture properties, compared to the Piedmont and Fall Zone materials. Lesser quality chert sources of fossiliferous and poorly cemented materials are found in the Black Mingo Formation of the Santee River Valley in central South Carolina. Soapstone, important for its use in cooking technology as slabs and bowls, and also for bannerstones (atlatl weights) and ornaments, is found in the Savannah River Valley in the Georgia counties of Stephens, Elbert, Wilkes, Lincoln, and Columbia; and in the South Carolina counties of Pickens, Oconee, Abbeville, and Edgefield (Wood et al. 1986: 305). Goodyear and Charles (1984) recorded eleven quarry sites in Allendale County during a survey related to the Flint Ridge formation, and concluded that "Allendale chert," known by archaeologists as a 34

PAGE 54

preferred tool making resource among Paleoindian and Archaic people of the region, is silicified grainstone and related to the Flint Ridge formation. During this survey, sites along Smith's Lake Creek were discovered to have great potential for quarry site investigation due to the presence of numerous chert outcrops and artifacts, and the stratigraphic integrity of the alluvial terraces there (Figure 3) Site 38AL23 yielded quarry debris and lithic tools buried as deep as 150 em below the surface. Site 38AL135 yielded highly weathered, discolored chert bifaces, cores, and debitage in a Paleoindian horizon (11,5000-9,900 BP) between 110 em and 120 em below the surface Brightly colored heat-treated chert was discovered in a middle Archaic horizon (8,000-5,000 BP) between 50 em and 90 em below the surface. Weathered lithics were also discovered at this depth. Underwater archaeological dredging in Smith's Lake Creek adjacent to these sites yielded a large amount of lithic debris and tools with a river-smoothed, thin, dark brown "riverine" cortex, suggesting the presence of previously utilized quarry sites now under water, most likely inundated by the westward migration of the Savannah River channel. "Upland" cortex associated with terrestrial chert outcrops nearby, is also plentiful on debris and tools from these sites, and is lightly colored, thick, and rough, having a foam-like appearance, with various stages of weathering. 35

PAGE 55

A milder Pleistocene-Holocene transitional environment that favored human adaptation is supported by numerous geoarchaeological studies in Savannah River Valley dra i nages. Excavations in 1995 and 1996 at Big Pine Tree (38AL143) yielded Paleoindian (11,500-9,900 BP) and Early Archaic (9,900-8,000 BP) lithic material including fluted preforms, Dalton points, and side notched Taylor points in a yellow, coarse, loosely bedded sand, 115-135 em below surface. This horizon lies atop a highly weathered soil zone of accumulation, or horizon, that is void of artifacts and cultural features (Figure 3). A comparable stratigraphy was documented in a cutbank profile by Goodyear and Charles (1984) at 38AL135, 1 km downstream of Big Pine Tree on Smith's Lake Creek. They recorded the presence of Paleoindian lithic artifacts associated with a 4C horizon, a soil zone composed of mostly unweathered, disintegrated parent materials of the overlying horizons. This horizon consisted of pedogenically unmodified coarse yellow to grayish-brown sand approximately 135 em below surface. The artifacts rested in the upper portion of a 4C horizon, atop 10 em of artifact-free sand, suggesting an initial Pleistocene-Holocene flood, followed by at least one other flood that covered the artifacts with more sand. Two backhoe test trenches were dug 1.5 meters and 30 meters from the cutbanks. In all three profiles, 36

PAGE 56

Figure 3 Soils Stratigraphy at the B i g Pine Tree Site (Bland 1995:55) . . . . . . Meters Modem, Disturbed ........... Apt . . .. . . . . . . . . . . . . . . . . . . . . ... ...... ... 0.07 . . . . . 2008P -Ap2 7 5 YR 4/4, SYR 0.15 4/6 Woodland Pertod (1,100 BP-3,000 BP) Bw Welldv developed Refuge Ceramics B tlortzal Sendy l.olm 7.5YR 4/4, 4/6, 516 l,OOOBP 4,500 BP 0.60 Bw/A Trlnsltkln MALA Lanceolate 0.65 Brier Creek Lanceolate 6,000 BP Bw/ A SarwJv 1J1tJ1n Morrow Mountain 7.5YR 3/4,4/4 Kirk Comer-Notched 0.90 2Bw/AB/C Transition loamy Sand Taylor Side-Notched 1.00 7.5YR4/6 10,500 BP Paleoindian 28/C Loemv saoo 10YR613 Fluted Preforms 1.15 3C Ane Sand UWR 1.35 7/7, 8/1, Thin Lamelae 11,000 BP (7.5YR5/6) 12.000 BP 1.40 4Bt1 b Sandy CJay 1.60 7.5YR5/6 Pleistocene Terrace 4Bt2b cay 7.SYR 516 37

PAGE 57

a 4C sand horizon overlies two red and gray argillic B horizons which contain no indications of human occupation, and are assumed to be Pleistocene in age. Based on attributes of the lithic artifacts, human habitation on the 4C surface probably dates to between 10,900 and 10,000 BP (Goodyear and Foss 1992). Five other sites within the Savannah River Valley (Figure 4) substantiate these findings. The Gregg Shoals site (9Eb259) lies on a high terrace-levee and contains approximately 6 meters of Holocene alluvium overlying bedrock. Three radiocarbon dates of peat overlying bedrock indicate an age of 10,370 to 10,000 BP (Segovia 1985). Early Archaic artifacts were found 3 meters above the bedrock (Tippitt and Marquardt 1984). The radiocarbon dates, relevant to the dating of the onset of aggradation in the Piedmont, suggest an earlier onset of upper valley infilling than that provided by Sassaman et al. (1990). It is significant that as much as three meters of alluvium was deposited in this area within the approximately 1,000 years that span the earliest human occupation and the onset of alluvial deposition, and is suggestive of Pleistocene-Holocene flooding as inferred from the stratigraphy at the Smith's Lake Creek sites. Rucker's Bottom (9EB91), located on a terrace-levee landform atop a relict Pleistocene terrace, also yielded Paleoindian and Archaic artifact assemblages within a one meter deposit of Holocene alluvium 38

PAGE 58

Figure 4. Five Coastal Plain Sites with Stratigraphy Similar to the Big Pine Tree Site. "' ' F' ------r --(Anderson and Schuldenrein 1985). A contextually ambiguous Clovis point (Anderson and Joseph 1988) and Early Archaic side and corner notched points were located 80-100 em below surface, overlying 20 em of coarse to medium bedload sands designated IIIC2, which was void of artifacts. This horizon lay atop an argillic B horizon in clayey, silty sand described as an argillic IVB2T (Anderson and Schuldenrein 39

PAGE 59

1985), overlaying gravel and coarse bedload sediments to 2.1 meters below surface. Simpson's Field (38AN8), located on a slight ridge of an eroded Pleistocene terrace, yielded prehistoric artifacts within 20 em of silty loam deposition, overlying a light reddish-brown sand clayey loam (Wood et al. 1986). A "Clovis-like" point, two unifacial lithic tools, and an early Archaic notched point were found at the interface of the silty loam deposits and the clay horizon. Here, like 38AL135 and Rucker's Bottom, coarse sediment associated with early Archaic artifacts overlay an argillic B horizon. Rae's Creek (9RI327) is also located on a terrace remnant of the Savannah River floodplain. Alluvial sand to 4.6 meters in depth had accumulated over a 9,000 year span through point bar formation (Matthews 1990). An Early Archaic Kirk corner notched lithic assemblage was located between 4.0 and 3.9 meters below surface, and midden material there was radiocarbon dated to around 9,060 BP (Crook 1990). The Kirk horizon was overlying sterile, mottled orange and tan, dense, sandy clay, that increased in density for more than 60 em, suggestive of early Holocene sands overlying a scoured late Pleistocene, weathered argillic B horizon (Goodyear 1999a:455). The Pen Point Site is located on a point bar on the first terrace (T1a) of the Savannah River valley. Late Paleoindian Dalton points and 40

PAGE 60

Early Archaic points lie atop as much as 2.5 meters of unmodified coarse point bar sediments (Brooks and Sassaman 1990: 185; Sassaman 1985b). As demonstrated by archaeological studies conducted at the Big Pine Tree Site and other sites with similar stratigraphic context, the intact, deeply stratified alluvial terrace deposits of the Savannah River Valley offer opportune conditions for the study of resource availability and site utilization. 41

PAGE 61

CHAPTER 4. CULTURAL CONTEXT In this chapter I first review mobility, subsistence, settlement, and population data for cultural periods leading up to and including MALA. Secondly, I emphasize the organizational strategies of lithic technology apparent in cultural peri ods leading up to and including the MALA tradition found at the Big Pine Tree Site. Thirdly, I identify specific cultural contexts at the Big Pine Tree Site. The organizational approach to understanding Southeastern lithic technology is explored by Daniel Amick and Philip Carr (1996). This approach focuses on variation of production, design, and tool use in response to demands of the environment, settlement mobility, subsistence strategies, and social organization. Under the organizational approach, behavioral and dynamic aspects of technology are emphasized. Three definitive perspectives concerning study of the organization of technology prevail in current thinking. First, technological strategies focus on "the selection and integration of strategies for making, using, transporting, and discarding tools and the materials needed for their manufacture and maintenance" (Nelson 1991: 57). Secondly, the successful organization 42

PAGE 62

of a technology must consider: a) types of raw materials available; b) distance to raw material sources; c) basic food procurement strategies; d) the movement and predictability of biotic resources; e) the seasonal availability and accessibility of biotic and mineral resources; f) group mobility; and g) the social relations with neighboring groups that enable access to the resources of other groups (Koldehoff 1987:154). Thirdly, the goal of the research of technological organization must "elucidate how [technological] changes reflect large-scale behavioral changes in prehistoric society" (Kelly 1988:717). In this thesis I focus on technological strategies for selection of raw materials and making stone tools. The roots of the organizational approach are found in ethnographic studies of extant hunter-gatherers. Particularly informative is the ethnographic research of Lewis Binford (1983:243-386) developed from his work among the Nunamiut Eskimos, which relates technological organization to settlement and mobility strategies. In prehistoric lithic technology, the effects of mobility are most readily apparent in the organization of procurement and transport of raw materials Reduction of energy expenditure and risk reduction are also implicated in the organization of technology (Amick and Carr 1996:42). 43

PAGE 63

Traditional New World (Western Hemisphere) origins models have postulated that Clovis people entered North America by traversing the Bering land bridge at a time when a nonglaciated corridor existed between the Cordilleran and Laurentide ice sheets, between 25,000 and 10,000 BP (Cordell 1984: 122). From a geological perspective, human migration across the Bering land bridge would have been possible at several times during this period. However, the absence of substantial numbers of well-defined sites with antiquity exceeding 12,000 BP thus far refutes such a theory. Recent dialogue has suggested multiple migrations from different directions, including the Bering land bridge model, the Pacific coast model, the Australian model, and the trans-Atlantic Iberian model (see Rose 1999). Sites such as Monte Verde in southern Chile (Dillehay 1984; Meltzer et al. 1997; Cactus Hill in Virginia {McAvoy and McAvoy 1997), and the Topper Site in Allendale County, South Carolina (Goodyear 1999a) lend some credence to pre-Clovis human occupation of the New World. Native American settlement in the Savannah River Valley dates to at least 12,000 BP during the final stages of the Pleistocene Epoch when small bands of Paleoindian hunter-gatherers relied on elaborate lithic technology and high mobility to exploit the Pleistocene environment. As the environment became more temperate during the 44

PAGE 64

early Holocene, Archaic human populations expanded and placed larger demands on plant and animal food resources. By the mid Holocene, a pattern of diminished mobility emerged, leading to the discovery of pottery and increased exploitation of shellfish and plant foods (Sassaman et al. 1990:5). Corn was util ized in the region by AD 1000. Fluctuations in settlement permanence and in the level of sociopolitical complexity characterized the next millennium. By the time Europeans made contact in the region, the Native American population was ethnically diverse and characterized by rapid demographic turnover (Sassaman et al. 1990:6). The prehistory of this continuum is well represented by multiple cultural components at the Big Pine Tree Site (Figure 5). The ascent of this cultural legacy leading to the MALA midden suggests that the chert quarries of Smith's Lake Creek have been favored for human settlement since Paleoindian times. Paleoindian Period (11,500-9,900 BP) The Southeastern United States Paleoindian sequence consists of a pre-Clovis Period (pre-11,500 BP), Clovis Period (11,500-10,900 BP), Middle Paleoindian Period (10,900-10,500 BP), and Dalton Period 45

PAGE 65

Figure 5. Cultural-Ethnic Chronology for the Southeastern United States. TY1E) ( 400-0 EP) MssissiJ:t)C11 TY1E) (fm-400 EP) V\txxlcrd TY1E) (3,(Xl}ffi) EP) Savan:tl Rive-(S,cm-3,(Xl) EP) EP) BiaOe:k (6,cm-5,(Xl) EP) G.ilfad (6,-B,:m EP) T8'ficr (9,ID-9,:aJ EP) Ditm (10,:00..9,g:x) EP) 9r t fiD1-9.MEriEE (11,
PAGE 66

(10,500-9,900 BP). The discovery of pre-Clovis cultural deposits at Monte Verde, Chile, and the recent peer-acceptance of the antiquity of these deposits (Meltzer et al. 1997), has placed new emphasis on the discovery of other ''pre-Clovis" sites in the New World. At the Cactus Hill site ( 44SX202) in Virginia, radiocarbon dates associated with an archaeological manifestation pre-dates the Clovis layer (Goodyear 1999a:435). Pre-Clovis artifacts associated with a "feature-like concentration of charcoal" at Cactus Hill include quartzite flakes and quartzite prismatic blades. The pre-Clovis layer has been radiocarbon dated to 15,070 BP (McAvoy and McAvoy 1997). The Topper site (38AL23), located on Smith's Lake Creek less than 1,000 meters downstream from the Big Pine Tree Site, has revealed artifacts in a "pre-Clovis zone" 100-150 cmbs. Artifacts include small chert flakes, chert cobbles, microblades, an end-scraper, and a burin spall (Goodyear 1999b). Historically, the earliest well-documented colonists of the New World are generally believed to be of the Clovis culture. These groups are believed to have been highly mobile, exploiting Pleistocene megafauna (Mason 1962; Kelly and Todd 1988) as well as small game and plant foods (Meltzer and Smith 1986). As Paleoindian populations moved into North America, they quickly identified key locations and used these as "staging areas" for subsequent population expansion. 47

PAGE 67

Under this model, the Savannah River Valley was an area of secondary colonization and home to a relatively stable subregional population (Anderson 1989). The presence throughout Paleoindian sites of biface and uniface lithic tools made from high quality cryptocrystalline raw materials suggests a settlement strategy of mob ility and specialized raw material scheduling (Goodyear 1989). The Middle Paleoindian Period is chronologically post-Clov i s and pre-Dalton. Its cultural markers in the Southeast are the Cumberland, Beaver Lake, Quad points in the middle regions; and Simpson and Suwanee points in the coastal plains. Stylistically, a large degree of variation exists among these tools, a remarkable aspect of the Middle Paleoindian lithi c tradition given its brief span of only 400 years. The cultural significance of stylistic variations of these Middle Paleoindian Period tools lies in their demographic diversity, indicating that demographic associations were in place by 10,500 BP if not earlier The Dalton Period marks the end of the Paleoindian lanceolate point tradition. The lithic technology of Dalton assemblages is clearly Paleoindian in character (Goodyear 1974; Morse 1973, 1997) with the addition of significant attributes such as serrated edges and hafting tang, or stem. Available faunal evidence indicates that modern plants and animals were the focus of subsistence strategies by Dalton times. Population is estimated to have increased by a factor of five to ten 48

PAGE 68

from the Clovis and Middle Paleoindian Periods (Goodyear 1999a:441). The Dalton technology has been regarded as a Paleoindian adaptation to a new Holocene environment, and the beginning of the early Archaic Period in the Southeast. The Paleoindian lanceolate complex at the Big Pine Tree Site lies between 115-135 cmbs in loamy sand (BC or C) atop a sterile Pleistocene clay terrace, and is dominated by bifacial preforms and blanks exhibiting strong basal fluting. This geoarchaeological-cultural environment is consistent with other Paleoindian sites in the Southeast, with the presence of a Paleoindian lithic assemblage associated with the first Holocene fluvial sands overlying a weathered argillic Pleistocene terrace (Goodyear 1999:462). Emphasis on percussion fluting of blanks in the early reduction stage suggests Clovis presence. Ten Dalton points have been recovered from the site above the zone containing the fluted blanks: four in situ from 100-115 cmbs, and six from Smith's Lake Creek. A large number of unifacial scrapers, flake knives, prismatic blades, and blade cores have been found in this zone. Only lithic artifacts have been recovered from this zone at Big Pine Tree. The Charles site lies about 300 meters downstream from the Big Pine Tree Site. The Paleoindian component lies between 100-125 cmbs in unmodified fluvial sands (C) or slightly weathered sandy loam (BC) 49

PAGE 69

atop two sterile argillic Pleistocene horizons ( 4Btl, SBt2). Basally thinned, bifacially fluted preforms are the predominate artifact, intermixed with flake tools and highly weathered debitage. Extensive research excavations on the Charles site have yet to be performed, and the extent of cultural implications remains unknown. Archaic Period (9,900-3,000 BP) The Archaic Period was first used to refer to a pre-ceramic, non agricultural, hunter-gatherer period following the Paleoindian period (Ritchie 1932). The onset of the Holocene, generally considered to have begun around 10,000 BP, presented human groups with radical adaptive challenges in the face of post-Pleistocene warming. The focus of subsistence activities moved to hunting smaller game and gathering plant foods. Changes in lithic technology may be noted by a reduction in projectile point size and the introduction of the notched biface, a projectile point style that assured a snugger fit on the smaller shafts of Archaic spears, darts, and knives. The Archaic Period in the Southeast is divided into three chronological-developmental periods: Early Archaic (9,900-8,000 BP), Middle Archaic (8,000-5,000 BP), and Late Archaic (5,000-3,000 BP). 50

PAGE 70

The Early Archaic (9,900-8,000 BP) The Early Archaic typically refers to a period of adaptation to the initial warming trend of the Holocene Epoch. Changes in settlement patterns and technologies evolved as temperate climatic conditions produced and maintained an increasingly homogeneous resource base (Claggett and Cable 1982). A settlement strategy of residential mobility characterized by expedient tool manufacture and situational technology was preferred over logistic mobility (Anderson and Schuldenrein 1983; Binford 1980). Some scholars argue for a more sedentary model in which base camps were positioned in areas of greatest resource diversity and density (O'Steen 1983). Others argue for a settlement model in which "small bands engaged in a mixed forager-collector strategy of watershed-wide seasonal mobility" (Anderson and Hanson 1988). Settlement was delimited by seasonal and spatial structure of food resources, mating requirements, information exchange, and demographic structure. 51

PAGE 71

Middle Archaic (8,000-5,000 BP) Historically, the Middle Archaic is viewed as a period of human adaptation to the mid-Holocene warming trend expanding demographics. Although the effects of this warming trend on Southeastern Archaic people are poorly understood (Goodyear et al. 1979) this climatic anomaly triggered the eastward expansion of prairie in the Midwest which led to dramatic changes in mobility and settlement strategies among Middle Archaic people there (Butzer 1978; Carmichael 1977; McMillan and Klippel 1981). It is expected that climatic changes occurring in the Midwest would cause an effect on populations in the Southeast and other regions (Sassaman 1990:10). Historically, Archaic studies in the Southeast have projected the position that very little technological change occurred during the Archaic in the Southeast (Smith 1986:21). More recently, this view has been tempered by contrasting data yielded by organizational studies of lithic assemblages and settlement patterns in the North Carolina Piedmont. These studies have recognized a shift from an Early Archaic 52

PAGE 72

logistically mobile organization strategy with curated technology to residential mobility and expedient technology (Cable 1982, 1992). Increasing population growth accompanied the warming trend of the mid-Holocene. Accelerated population growth appears in the archaeological record as diverse regionally inspired styles, indicating growing territoriality (Walthall 1980: 58). Mid-Holocene population growth may have caused band territories to become more compact and more densely populated, resulting in the more frequent reoccupation and possibly longer occupation of sites (Steponaitis 1986:372). Additionally, regional population increases may have constrained mobility (Amick and Carr 1996:44), resulting in the accelerated exploitation of local raw materials. Technological innovations relating to expedient technology are expected in the Middle Archaic. For example, evidence of systematic scavenging of Early Archaic sites by Middle Archaic people has been noted in the Duck River Basin {Amick 1985:30; Hofman 1986:81). Archaeological lithic scatters become exposed by erosion to later human occupants who utilized these resources, especially larger pieces of debris, flake tools, and bifaces Lithic scavenging is recognized as an important factor affecting assemblage variation and the organization of lithic technology (Camilli and Ebert 1992; Ebert 1992; Kelly 1988; Wandsnider 1989). Scavenging of early sites by Woodland people has 53

PAGE 73

been noted in the South Atlantic Coastal Plain (Sassaman and Brooks 1990; Sassaman et al. 1990:319-320). Hafted biface technology indicates a shift away from strategies that extend tool-life such as the beveling of edges which may have extended tool-life by facilitating resharpening. Thermal alteration as an effort to optimize local resources represents an expected innovation in the Middle Archaic. The transition from Early Archaic to Middle Archaic in the Savannah River region is characterized by replacement of a notched hafting technology with a stemmed hafting technology. Middle Archaic stemmed phases include the Kirk Stemmed, Stanly, Morrow Mountain, Guilford, Brier Creek, and MALA (Figure 6). The stemmed point tradition at the Big Pine Tree Site is most readily evident in the subregional Morrow Mountain stemmed point tradition (Plate 3), and later in the MALA stemmed-notched tradition (Plate 4). The Morrow Mounta i n phase was first identified by Coe (1964) in North Carolina Absolute dates for this type derive from Tennessee (5,490 BP) where they have been found in association with Benton points, and South Carolina (5,477 BP) (Sassaman et al. 1990:151). While Morrow Mountain artifacts have been discovered throughout much of the Southeast, the Morrow Mountain phase in the Savannah River region is representative of cultural divergence between the Piedmont and Coastal Plain, based on site density and 54

PAGE 74

Figure 6. Middle and Late Archaic Bifaces from the Savannah River Region: a. Kirk Stemmed; b, c. Stanly; d-f, i. Morrow Mounta in; g, h. Guilford; j. Brier Creek; k-m. MALA; n-r. Late Archaic Stemmed Bifaces. (after Sassaman 1993:32, drawn by Stephanie Brown) 55

PAGE 75

Plate 3. Morrow Mountain Stemmed Point Recovered from the Big Pine Tree Site. Plate 4 MALA Stemmed/ Notched Point Recovered from the Big Pine Tree Site. assemblage data (Anderson and Schuldenrein 1985; Blanton 1983; Blanton and Sassaman 1989; Sassaman 1983). Data from Piedmont sites reflect a seemingly random locational pattern, lack of interassemblage functional variation, exploitation of loca l raw materials (primarily quartz), and expedient technology. A strategy of small group size, frequent residential mobility, generalized hunter-forager subsistence, low-investment technological organization, and 56

PAGE 76

social flexibility has been inferred from these data (Sassaman 1983; 1985a). While Morrow Mountain site density in the Coastal Plain is sparse compared to that of the Piedmont, Coastal Plain variability data for this phase suggest diverse organization of lithic technology there. For example, the Morrow Mountain point is a triangular blade with a tapered stem and the Guilford point is a thick, lanceolate blade with a straight, rounded, or concave base (Coe 1964 : 37 ,43). A high incidence of thermal alteration has been noted on Morrow Mountain specimens made of Coastal Plain chert (Sassaman et al. 1990: 150). Late Archaic (5,0003,000 BP) The Late Archaic is among the most studied prehistoric archaeological periods in the Southeast, due in large part to rich site and artifact assemblage records. Suggested by these data are patterns of tribal formation, territorial resource rights, long-distance exchange, ceremonial burials activities, and surplus production (Sassaman et al. 1990: 11). Late Archaic population mobility in the Southeast has been characterized as "entrenched" (Graham and Roberts 1986), where systematic reoccupation at a ser ies of select sites is expected (Amick 57

PAGE 77

and Carr 1996:45). Entrenched mobility systems may develop in response to mobility constraints resulting from population growth. The Late Archaic Period refers to a period of increased settlement permanence, population growth, subsistence intensification, and technological innovation (Smith 1985). An emphasis on riverine habitats prevailed during the Late Archaic Period in the Coastal Plain (Stoltman 1974), while seasonal movement from uplands to bottomlands was practiced in the Piedmont (Sassaman 1983; White 1983). Exploitation of shellfish occurred as an adaptation to the demands of explosive population growth in the Coastal Plain, and coincided with the development of the first fiber-tempered pottery around 4,500 BP. As the modern Savannah River floodplain developed in an upriver, "time regressive" direction (Brooks et al. 1986), increased attention was placed on upriver, tributary sites such as those found in the Smith's Lake Creek vicinity. Exploitation of shellfish and the utilization of pottery did not occur above the Fall Line until after 3,700 BP. Regional population density may have peaked during this time, indicated by increased richness of pottery designs, elaborate non subsistence material culture, exchange in soapstone and other items, and the formation of huge shell-middens and coastal shell rings (Sassaman et al. 1990: 11). The roots of this period reach back into 58

PAGE 78

the Middle Archaic Period in areas of high population density relative to available resources. The onset of the Late Archaic is marked by the appearance of the large, broad-bladed, square-stemmed, bifacial Savannah River Stemmed point. This point type was originally defined by Claflin (1931) from excavations at Stallings Island in the Middle Savannah River Valley. The early phases of this lithic technology are referred to as the Savannah River Stemmed phase or the Stallings I phase (5,000-4,500 BP). Stallings I sites in South Carolina are aceramic More recently, Coe (1964:44, 123) has provided a formal definition of the point type, regarding it as a descendant of the Middle Archaic Stanly phase. The large blade of the Savannah River Stemmed point provided for a versatile, multipurpose tool, and is a notable deviation from the specialty tool production of the Middle Archaic. Additionally, Late Archaic technology included increasingly elaborate polished and ground stone tools, stone vessels, and ceramic vessels. Late Archaic ceramic technology in the Savannah River Valley begins around 4,500 BP with fiber-tempered vessels of the Stallings and Thorn's Creek series (Sassaman 1990: 184-190). Savannah River Stemmed points have been found in contexts with and without associated ceramic assemblages. Aceramic contexts in the Piedmont which have rendered Savannah River Stemmed points 59

PAGE 79

span dates between 4,900 BP and 4,100 BP (Anderson et al. 1985; Bullen and Greene 1970; Wood et al. 1986). While absolute dates for such aceramic contexts are not available for the Coastal Plain, Coastal Plain dates of 4,500 BP (Stoltman 1974), and 4,290 BP (Sassaman et al. 1990: 158) have been reported for Savannah River Stemmed phase contexts containing fiber-tempered pottery. MALA specimens (n=26) were found at the Middle Coastal Plain sites of Pen Point (38BR34) (Appendix 1) on the SRS (Sassaman 1985b), and Big Pine Tree (Appendix 2). These are often thermally altered and exemplify a corner-removed, stemmed-notched hafting technology. Stratigraphically, MALA is found above a Morrow Mountain layer and below a Late Archaic Savannah River Stemmed layer at both sites. This form only occurs in the Coastal Plain of the Savannah River valley region with no apparent local antecedent. The origins of the MALA phase have yet to be determined. The MALA phase at the Big Pine Tree Site is represented by a dark band of organically stained midden soil that has rendered radiocarbon dates of 3,980 BP; 4,430 BP; and 4,820 BP for Levels 1, 2, and 3 respectively (Goodyear 1998:20). It is probable that the makers of the original MALA points occupied Level 3 (85-95 cmbs), and that Level 2 (75-85 cmbs) is pre-Stallings Late Archaic. Level 1 is clearly Late Archaic with a context containing fewer numbers 60

PAGE 80

of heat-treated stemmed points than the lower two-thirds of the midden Level 1 also contained steatite disks indicating a culinary technology. The MALA technology has been related to that of the Benton phase (Sassaman et al. 1990: 104,153), found mainly in the Tennessee River Valley (Futato 1983; Thorne et al. 1981) and in the upper Tombigbee River Valley (Bense 1983; 1987). The Benton point shares morphological similarities with the MALA point. The Benton point was originally described as basic triangular with excurvate lateral margins and an incurvate or straight stem base (Kneberg 1956:25). Radiocarbon dates of 4,130 BP and 4,190 BP have been reported for the Benton points found in association with known Late Archaic point types in the Tennessee River Valley (Amick 1987: 12), although Benton points have been recorded in the Tennessee River Valley that date as early as 5,700 BP (Futato 1983), and in the Upper Tombigbee that date as early as 6,000 BP (B ense 1983, 1987). Based on occurrences of Benton-like caches found as far as 150 km from the Benton Ft. Payne chert sources (Johnson and Brookes 1987), an argument for a "Benton exchange sphere" has surfaced. The Late Archaic Period in the Savannah River Valley may be summarized by the following cultural attributes: explosive population 61

PAGE 81

growth, entrenched mobility, intensive site and resource use, resource innovation and invention, surplus production, and exchange. Technological attributes which overlap with Late Archaic cultural attributes may be summarized as: innovation in lithic tool design such as seen in the multifunctional Savannah River Stemmed point, development of culinary technologies which utilized a variety of new resources and technological innovations such as steatite (soapstone) and fiber-tempered ceramic paste. Based on the recovery of culinary artifacts such as steatite disks in association with heat treated, stemmed lithic tools, the upper third of the MALA midden at the Big Pine Tree Site probably relates directly to this transitional preceramicto-ceramic period. This inference is supported by radiocarbon dates that fall between 3,700 BP and 2,900 BP for well-preserved pit features that intrude from above. These features are probably transitional between Late Archaic and Early Woodland periods (Goodyear 1998:20). 62

PAGE 82

CHAPTER 5. BACKGROUND OF ARCHAEOLOGICAL RESEARCH This chapter reviews the archaeological work which has focused on the MALA phase of the Late Archaic Period in the Savannah River Valley. Available MALA data are limited to studies from the Pen Point Site, 38BR34, and the Big Pine Tree Site. Data recovery and debitage analysis from the MALA component at the Big Pine Tree Site are represented by this thesis. Field identification of bifacial tools from the Big Pine Tree Site has revealed morphological similarities with the Pen Point Site and 38BR34 biface assemblage. This is also true of debitage attributes between the sites. That the types recovered from the three sites represent the same technological phase is unequivocal. The MALA phase was originally identified by cultural resource management researchers on the SRS in the mid-1980s (Sassaman 1985 b). The identity of the phase was based on the discovery of heattreated, stemmed-notched projectile points within a midden feature at the culturally stratified Pen Point Site. The biface typology was without precedent in the Savannah River Valley. For this reason, coupled with the aspects of large scale production, a preference of Coastal Plain chert, high frequency of thermal alteration, and high 63

PAGE 83

morphological variability, the MALA phase has piqued the interest of Coastal Plain researchers. Pen Point Site The presence of long, slender, stemmed, notched, lanceolate bifaces, intermixed with a dense concentration of mildly weathered, thermally altered debitage, first attracted the attention of scientific researchers at the Pen Point Site in the 1980s (Sassaman 1985). This typology was designated MALA. The Pen Point MALA assemblage occupied the position traditionally filled by the Guilford horizon in the Piedmont, and the Brier Creek or Guilford horizon in the Coastal Plainabove a Morrow Mountain zone (7,500-6,000 BP) and below a Savannah River Stemmed zone (4,900-4,100 BP). The Guilford point is a lanceolate with either concave or rounded base, and the Brier Creek is a stemmed lanceolate (Figure 5). While the stratigraphic position of this distinctive point type between Morrow Mountain and Savannah River components had been suspected (Charles 1981:29; Goodyear and Charles 1984:76-77,85,89-90), the phenomenon was not scientifically documented until the Pen Point excavations. Artifact-bearing deposits at the Pen Point Site extend from surface to one meter in depth The sandy loam point bar deposits of 64

PAGE 84

the Pen Point Site are geologically undifferentiated, but culturally stratified in that temporally assigned biface types were chronologically superimposed. The site contains a continuum of cultural components representative of the Early Archaic Period (9,900-8,000 BP) through the Woodland Period (3,000-800 BP). MALA specimens were recovered from 30-55 cmbs. Morrow Mountain hafted bifaces were recovered directly below the MALA zone 50-75 cmbs. Savannah River Stemmed biface fragments were identified in a zone 25-35 cmbs. MALA specimens differ from Morrow Mountain and Savannah River artifacts primarily in the employment of a "notched/stemmed" technology (Sassaman 1985b), in which both stem and side or corner notching are developed. While slight overlap occurs in the upper and lower MALA levels, the zone between 40-50 cmbs contained only MALA specimens. Within the 40-50 cmbs zone and extending to 55 cmbs, a lithic production feature (Feature 14) was identified. Feature 14, measuring approximately 1.6 square meters, contained a dense concentration of debitage (n = 18,357), preform fragments (n= 131), notched hafted bifaces (n=9), and fire-cracked rock. This feature is important because it represents a discrete record of lithic tool production. The MALA assemblage at Pen Point is limited to the tools and debitage of hafted 65

PAGE 85

biface production. Groundstone, unifaces, and flake tools are poorly represented. Whole, or nearly whole, MALA hafted biface specimens (n=14) (Plate 5) from the Pen Point Site provided the basis for preliminary typological assessment (Plate 6). The MALA biface morphology was originally described by Sassaman (1985b: 1) as "stemmed/notched hafted bifaces having no clear typological precedent in the region," with the footnote that definitive modes of variation should be avoided due to limitations of sample size. Further, the sample was noted to consist of lanceolate forms exhibiting a diversity of haft and shoulder morphology. Blades are long and slender with straight or excurvate margins. Shoulders are either square or barbed depending on the notching technique. Bases are flat, indented, or excurvate, lacking basal grinding. Cross sections are bi-convex to plano-convex, the latter indicative of manufacture from flake blanks. Blanks are usable stone flakes from which tools may be manufactured. Initial shaping was by percussion techniques which left broad, shallow flake scars across most of the blade. Pressure retouch along edges appears to have been the usual technique of final edge preparation and rejuvenation. Several examples show long lamellar scars resulting from rejuvenation, and running perpendicular to the blade margin, some ending in step fractures (Sassaman 1985b:2). 66

PAGE 86

Plate 5. MALA Stemmed/Notched Bifaces from the Pen Point Site (Sassaman et al. 1990: 154) 34H-2 42J-2 511-1 52l-1 461-1 30J5 47J-3 4614 3SG-5 461-12 5113 Step fractures are flake scars that terminate abruptly in a right angle break at the point of truncation, and result from a dissipation of force or the collapse of the flake (Crabtree 1982:53). 67

PAGE 87

Plate 6 MALA Biface Morphology Terminology Using an Examp l e from the Pen Point Feature 14 (Specimen 42-12; Sassaman 1985: 11). distal end or tip straight blade margin pressure flake retouch barbed shoulder hafting element -( / corner notch t excurvate blade margin lamellar flaking expanding stem straight to excurvate base proximal end or base Technological attributes of MALA technology at these sites underscore Late Archaic technological attributes discussed in the last chapter: preference for Coastal Plain chert (CPC) large scale production, high inc i dence of thermal alteration slight fad i ng from weathering of flake and tool artifacts, and wide range of variabil ity 68

PAGE 88

among bifacial tools. With the exception of three biface specimens in the Pen Point MALA assemblage which were manufactured from quartz, all were manufactured from CPC. Furthermore, the Pen Point Site MALA assemblage exhibits among the highest percentage of thermally altered CPC known. While the incidence of thermal alteration among Early and Middle Archaic sites in the Coastal Plain is around 42/o for chert thinning flakes and 12.5/o for retouched stone tools (Anderson et al. 1979:51, 55), the incidence of thermally altered whole thinning flakes recovered from Feature 14 at the Pen Point Site is 67 .2/o, and 90.6/o for all retouched stone tools (Sassaman 1985b:14). Summary statistics of MALA hafted bifaces from Pen Point are indicative of variation found among specimens. Most variation within the MALA typology appears in the shoulders and hafting elements, which consist of stems and notches. Such variation may be resultant of historical adaptive design modifications While environmental adaptations may be causal to such variation, social aspects also may have contributed to haft variation. For example, spear points were probably crafted to fit an individual's shaft; thus variation in hafting element attributes varies with shaft dimensions Based on compelling evidence from Feature 14, variation within the Pen Point MALA assemblage has been interpreted as a relatively 69

PAGE 89

contemporaneous phenomena (Sassaman 1985b: 12). This interpretation rules out the possibility of historical variations implemented as adaptations to environmental or social change. As related to the organization of technology, bifacial tool variation may have been the result of retooling or repairing existing tools. Indicators of such activities within the Pen Point tool assemblage are variable working-edge angles, blade widths, blade lengths, and tip angles. Tool discard, inferred from the high frequency of broken tools, occurred as new tools were manufactured. The high frequency of preform fragments supports this notion. Accordingly, the debitage assemblage is expected to reflect this activity. Based on morphological and stratigraphic similarities between the Halifax Side-Notched biface typology (Coe 1964; South 1959) from the Gaston Site in North Carolina which was radiocarbon dated to 5440 BP, and MALA morphology ascertained from samples at the Pen Point Site, a tentative inferred date of 5,500 BP was placed on the MALA horizon (Sassaman 1985a: 12). Radiocarbon dates ranging between 4,820 and 3,980 years BP recorded for the Big Pine Tree Site some ten years later would refine that assessment (Goodyear 1998). 70

PAGE 90

Site 38BR34 Sassaman et al. (1990: 109) also reported the discovery of 38BR34 situated on an alluvial terrace of Four Mile Creek, approximately 4.5 km NNE of the Savannah River on the SRS. Test units at this site contained a dense concentration of thermally altered debitage, substantial amounts of fire-cracked rock, and bifacial preforms made of CPC, as was the case at the Pen Point Site and the Big Pine Tree Site. Vertically, the highest density of artifacts occupied levels between 20 cmbs and 40 cmbs. Two 2 x 2 m test units yielded over 35,000 flakes ( 42/o whole, 58/o broken), 87 preform fragments, and 9 bifaces including two MALA/Benton bifaces. Average whole-flake density in two 2 x 2m units equals 11,056 flakes per cubic meter. Bifaces were found in the peripheral areas of the debitage concentrations, suggesting that the test units may not have been placed in tool discard areas. Thermal alteration was noted on 80 .2/o of the debitage assemblage and cortical attributes were noted on 3.3/o of the assemblage. Additional testing and analysis of this site will undoubtedly reveal a larger tool discard assemblage. 71

PAGE 91

Big Pine Tree Site While the Pen Point Site and site 38BR34 appear to have been retooling, tool discard, and tool production locales, about 10 km and 15 km respectively from the Allendale chert quarries, the Big Pine Tree Site is an Allendale chert quarry site in which preform manufacture appears to have been a major occupation. Early in the Pen Point Site excavations, Sassaman noted s imilarities between broken bifaces recovered from the Pen Point Site and side-notched examples recovered by Albert C. Goodyear and Tommy Charles (1984:76-77, 85, 89-90) duri ng their quarry site survey of Allendale County. Goodyear and Charles, both of the South Carolina Institute of Archaeology and Anthropology (SCIAA), d i scovered the Big Pine Tree Site in 1983, while surveying prehistoric chert quarries along the Savannah River near Martin in Allendale County, South Carolina (Figure 7). The purpose of the survey, funded by the South Carolina Department of Archives and History and the U. S. Department of the Interior, was to locate Paleoindian sites with possible eligibility for nomination to the National Register of Historic Places. Eligibility would be based on the integral relationship between 72

PAGE 92

Figure 7. Allendale County Chert Outcrops and Quarries (hatched areas) Showing the Big Pine Tree Site and Site 38AL135 (mapped by Goodyear and Charles, 1983). OOE Savannah River Plant Big Pine Tree Site I ) l ------=-:a..-....c. ---0 ...--------------73

PAGE 93

the Paleoindian requirement for high-grade siliceous stone and site selection (Goodyear 1989). In August 1983, Goodyear and Charles discovered a large quantity of chert debitage that had washed out of an eroded river bank along Smith's Lake Creek, a tributary of the Savannah River The bank was on property then owned by Sandoz Chemical Company, now by Clarion Corporation, in the Coastal Plain of South Carolina. In January 1984, they conducted a series of five auger tests that yielded bifacially worked chert resembling Paleoindian and Early Archaic types bifaces similar to Brier Creek lanceolate, MALA, and Guilford types dating from the late Middle Archaic; and diagnostic Woodland period projectile points. At the time of this survey, MALA was not recognized as a discreet technological phase. However, a distinction was made the following year. In July and August 1985, underwater archaeological investigations were conducted in Smith's Lake Creek adjacent to the Big Pine Tree Site and site 38AL135, 300 meters downstream. The Underwater Division of SCIAA performed this work, which was funded by the Venture Fund of the University of South Carolina. Artifacts from nearly all archaeological time periods were recovered from the floor of the creek us i ng an airlift water pump system (Lindeman 1990:25). From the 93 diagnostic lithic artifacts including 18 (19/o) diagnostic 74

PAGE 94

MALA bifaces (Appendix 2) and 709 ceramic artifacts (Appendix 3) recovered from underwater operations in 1985, it was concluded that a large portion of the site had eroded into the creek during modern times, an act probably associated with recent upstream reservoir construction. No further archaeological work proceeded at the Big Pine Tree Site until July 1992 when the landowner discovered numerous artifacts while digging a small recreational boat ramp in the immediate vicinity. This discovery prompted further archaeological testing, and six trenches were excavated by backhoe. The excavation of Backhoe Trench Three (BHT 3) revealed a charcoal lens which was collected for radiocarbon dating and later placed in a Late Archaic (5,000 3,000 BP) context. The trench was then backfilled In July 1993 Goodyear reopened the trench to a depth of 90 cmbs. They recorded the presence of numerous CPC flakes intermixed with concentrations of carbonized plant remains (Bland 1995:53). In February and March 1994, Goodyear excavated 18 square meters to a maximum depth of 155 cmbs at the Big Pine Tree Site, designated by corner posts E 92, E 94, and E 96 (Figure 8). This investigation "revealed exceptionally high quality geological and archaeological evidence for a variety of early prehistoric cultures" (Goodyear 1994:2), 75

PAGE 95

Figure 8. Topographic Site Map of the Big Pine Tree Site (38AL143) (Goodyear 1994). 38AL 143 BIG PINE TREE SITE AC G<)()(WFA,'I 0 1DII II II I II II I IIIII II Ill S 10 IS V') X 1---N .n 76

PAGE 96

including several hundred thousand pieces of CPC debitage which were catalogued and curated at SCIAA (Bland 1995:53). Investigations at the Big Pine Tree Site revealed a rich multi component site based on data obtained from the analysis of several thousand stone artifacts obtained during the underwater recovery in 1985, subsequent excavations, and additional underwater recovery in 1995. A substantial MALA component occupied the levels between 60 cmbs and 90 cmbs. This component was identified by a dense concentration of thermally altered debitage, substantial amounts of fire-cracked rock, bifacial tools and fragments, and bifacial preforms, as was the case at the Pen Point Site and site 38BR34. A total of 95 square meters had been excavated by May 1995 when data for this analysis were recovered. In May 1995, Goodyear opened an additional 31 square meters contiguous with the previously excavated 18 square meters. The top 50 em, composed mostly of the previous year's backfill, was scraped off by backhoe. A crew of eight archaeologists then shovel-skimmed another 7 to 10 em to the top of the dark band of organically stained silty, clayey soil known as the MALA Midden. From the top of the midden, contiguous two square-meter units were excavated in 10 em levels. These units were eventually excavated through Archaic and Paleoindian levels to an artifact-free zone below 120 cmbs. 77

PAGE 97

The research plan for the 1995 season provided for two weeks of excavating the MALA layer which lay approximately 60 to 100 em below the surface of the project, and two weeks of Archaic and Paleoindian excavations. Ken Sassaman, also of SCIAA, assisted through the MALA phase of excavations, during which time the debitage for this thesis was collected from Test Units C, D, E, and G (Figure 9). Test units were 1 x 1 meter square and excavated in three arbitrary 10 em levels (study units). A total of 7,839 whole-flakes (22,683 whole flakes/m3 ) was recovered from these test units for analysis. Summary of Previous MALA Research Similarities between MALA assemblages at the Big Pine Tree Site, the Pen Point Site, and site 38BR34 strengthen the argument for a unique technological phase that overlapped the final three centuries of the Savannah River Stemmed (Stallings I) phase which lasted from "roughly 5,000 to 4,500 BP" (Sassaman et al. 1990:161), and suggest a previously unrecognized cultural anomaly. Flake size data for all whole-flakes from the Big Pine Tree Site are compared with data from the Pen Point Site (Table 2) An inherent flaw when comparing the Big Pine Tree Site and the Pen Point site lies 78

PAGE 98

Figure 9. Plan Map of the Big Pine Tree Site, 1993-1996, Show i ng Test Units A G. E87 E89 1995 1993-south wall level 1 E91 E93 E95 E97 E99 ElOl 38AL143 Big Pine Tree Site 1993-1996 N 1 8 8 N18 6 N184 N182 N1 7 9 N177 1 meter charcoal recovery 79

PAGE 99

Table 2. Whole Flake Size Frequency Data (not standardized) from the Big Pine Tree Site and the Pen Point Site. Big Pine Tree Site Pen Point Site size n 0/o cum n 0/o cum variation 0/o 0/o in cum 0/o 1 4,855 61.9 61.9 2,623 18.9 18. 9 43.0 2 1,385 17.7 79.6 4,540 32.7 51.6 28.0 3 578 7.4 87.0 3,877 27. 9 79. 5 7.5 4 462 5.9 92.9 2,050 14.8 94.3 -1.4 5 268 3.4 96.3 609 4.4 98. 7 -2.4 6 176 2.2 98.5 140 1.0 99.7 -1.2 7 66 0.8 99.3 33 0 2 99.9 -0. 6 8 27 0 3 100.0 12 0.1 100.0 0 9 22 0 3 100.0 4 0 0 100.0 0 within Size 1, which includes flakes obtained by 1/Sth inch field screening at the Big Pine Tree Site, and by 1/4th inch field screening at the Pen Point Site. Thus, data for Size 1 at the Big Pine Tree Site is enriched when compared with data from the Pen Point site. This flaw is minimized when size categories are paired or otherwise grouped for statistical analysis. For example, more than 90/o of the flakes in both 80

PAGE 100

sites accumulate within size categories 1, 2, 3, and 4, ( <2.5 em length). Ogive comparisons of whole-flakes between the Big Pine Tree Site and the Pen Point Site provide a graphic perspective of size frequency data accumulations, and indicate that debitage assemblages from the two sites are similar in middle and large size categories (Figure 10). Ogive graphic representation considers frequency distribution in which every ordinate represents the sum of frequencies in preceding intervals, here expressed as cumulative percent. Ogive comparisons are especially useful when comparing samples with a wide range of values, since they standardize each sample to a scale ranging from 0 to 100/o. Ogive representations are useful in comparisons of frequency accumulation values between samples. The wide range of difference between cumulative percentages at each site is best rationalized by the field screening differences. The intersection of data points at the midpoint of Size 4 suggests a common level of production aimed at manufacturing one or more tool types, from which a high frequency of size 4 flakes were derived. The similar data curves represented by size categories 4-9 indicate similar biface manufacturing activities at each site. 81

PAGE 101

The Big Pine Tree Site, Pen Point Site, and site 38BR34 exhibit thermal alteration data that are similar (Figure 11). Thermal alteration provided a means of increasing tool production by facilitating the flaking process, while also facilitating the retooling p r ocess in the field. Figure 10. Ogive Comparison of Flake Size from Debitage Assemblages at the Big Pine Tree Site and the Pen Point Site 100.0 80.0 -Bi g Pine Tree -Pen Poin t 0 Q) > 60.0 :p ltl ::J E 40. 0 ::J u 20.0 0.0 1 2 3 4 5 6 7 8 9 size Durability of tools, made so because of the frequent l ong-range treks that took the hunter away from the source of raw material, became less important as the range of subs i stence resources shru n k. The average incidence of thermal alteration among whole thinning flakes for the three is 80/o. This contrasts sharply w ith the 42/ o average 82

PAGE 102

Figure 11. Comparison of Thermal Alteration Between the B i g Pine Tree Site, the Pen Point site, Site 38BR34, and the Early and Midd l e Archaic Periods. 1 "0 Ql .... Ql ttl ttl E .... Ql 0 0 0% 9 0 0/o 8 0 0/o 7 0 % 6 0 0/o 5 0 % 4 0 0/o 3 0 % 2 0 0/o 1 0 0/o 0 % Big Pine T r e e P e n Point 38Br34 Earl y M I d d I e Archaic Figure 12. Frequency per Cubic Meter of Therma l ly Altered Flakes at the B i g Pine Tree Site in Levels 1-3. 9.0 80 7.0 ,...... 60 Vl 0 0 5.0 0 .-I .._ 4.0 ("'") E ........ 30 ...... 20 10 83

PAGE 103

found in Early and Middle Archaic sites in the Coastal Plain (Anderson et al. 1979:51, 55), and is indicative of a preference for settlement near chert outcrops with material suitable for tool manufacture. Thermal alteration was integral in the organization of lithic technology at the Big Pine Tree Site well into the Late Archaic Period (Figure 12). 84

PAGE 104

CHAPTER 6. METHODOLOGY In this chapter I describe field methods for data recovery, standardization of data, and application of inferential statistical analysis to data from the Big Pine Tree Site. Implications derived from statistical analysis are discussed. Field MethodsData Recovery Debitage recovered for analysis in this thesis came from four one-square meter units designated C, D, E, and G. Each unit was dug to approximately 30 em in depth. Charcoal samples from the large quantity of burnt nutshell (Bland 1995) intermixed with the debitage were extracted for radiocarbon dating from Test Unit F (Goodyear 1998:20). Debitage was extracted from excavated midden soil using an 1/8th inch (0.3175 em) wet-screening technique, which employed the use of a 5 horsepower water pump placed at the edge of Smith's Lake Creek. Hoses were attached to the pump to supply a constant flow of water to the screens in order to wash the excavated flakes as 5-gallon 85

PAGE 105

buckets of excavated material were emptied into the screens. The use of 1/8th inch screen provi ded a substantially larger sample size and correspondingly enriched data than had been obtained at the Pen Point Site where 1/4th inch screen was used. As sample size increases it approaches population parameters, having the affect of bringing the means of variable samples together. This effect may disguise variation. Additionally, as sample size increases, the probability that rare classes of material will be included into the sample increases; there is a positive correlation between sample size and richness. However, this correlation is not sufficient to dismiss behavioral factors as cause for variation in richness (Piog and Hegmon 1993:489). Standardization of Data During the May 1995 excavations, debitage was separated from the excavated soil of the test units for the purpose of a debitage analysis thesis project. The test units were one-meter squares, each consist i ng of three arbitrarily established ten-centimeter levels. The actual unit volumes excavated varied due to the intrusion of disturbed soil from Backhoe Trench 1 dug in 1994 along the south wall of the test units. Also, an intrusion of yellow sand at the northwest corner of 86

PAGE 106

Test Unit A, as well as mechanical erosion events caused variation in midden thickness Each whole unit level consisted of 0.1 cubic meter of midden soil However, because of the disturbed provenience the levels were not equal in volume. Therefore, flake frequency values are expressed as number of whole-flakes per cubic meter (ffm3). The actual volume of each unit was calculated by plotting the test units on graph paper and then counti ng the number of squares in each level for the nondisturbed portion each unit (Table 3). Table 3. Volume in Cubic Meters of Undisturbed Midden Soil Excavated Per Unit Unit C Unit D UnitE Unit G Total Level 1 0.08000 0.09000 0.09000 0.10000 0.36000 Level 2 0 .09275 0 .09350 0.10000 0.10000 0.38625 Level 3 --------0.09725 0.10000 0.10000 0.29725 Statistical Analysis Statistical analysis of standardized data from the Big Pine Tree Site is addressed u t ilizing inferential statistical methods. Whi le descriptive statistics allow the researcher to formulate inferences for data samples by visual inspection inferential analytical methods allow 87

PAGE 107

the researcher to accept or reject specific data-based inferences utilizing the concept of statistical significance. Statistical tests employed in this research are performed at the 95/o confidence level. Formulation of hypothesis and theory follow Data for each level in each unit are categorized by size. Size is determined by placing each flake in a sizing-grid (Plate 7). The smallest flakes are placed in Size 1 which includes all whole flakes less than 1.0 mm in length. Each size category is delineated by a 0.5 em increase in length. The largest size category is 9, which includes flakes larger than 4.5 em in length. Since 1/8th-inch screening was used to extract flakes from the field, flakes in Size 1 measure no less than 0.3 em. Since Size 1 is thus encumbered, and Size 9 is open-ended, these two size categories encompass a larger dimensional range than other size categories. However, this inequity is not addressed since the skewed nature of Size 1 and Size 9 is equally inherent in all three samples (levels). Descriptive comparisons are made between data size frequencies from each level. Radiocarbon dating of charcoal samples from the Big Pine Tree Site recovered from each level in Test Unit F suggests three separate statistical populations (Goodyear 1998:20). Statistical correlation between radiocarbon dates and debitage data is addressed by graphic representation, followed by discussion. 88

PAGE 108

Plate 7 Flake Sizing Grid Used to Determine the Size Category of Each Flake With statistically normal data, variation is typically measured by tests of variability. A test ideally suited to testing the variability between two or more samples of normally distributed data is ANOVA To understand ANOVA, the concept of variability must first be understood. Variability is a quantitative measure of the degree to which scores (frequencies per size category) in a distribution are spread out or clustered together. A distribution may be plotted 89

PAGE 109

graphically, where size is the independent (X) variable and frequency is the dependent (Y) variable. ANOVA is an inferential procedure which uses sample data to make inferences about populations. The purpose of ANOVA is to discern whether differences between sample means are the result of chance or systematic treatment. Treatment in this study is represented by each of three levels. Each level represents an arbitrarily different time phase. The goal of ANOVA is to test the null hypothesis that no statistical variation exists within and between the three samples of data statistical test. ANOVA is performed on the size frequency data for all whole flakes. The three levels are compared for significant variation in size frequency values. The same procedure is followed in analyzing nominal attributes. Samples are segmented into contiguous, paired size categories and compared across the three levels. Nominal attributes consist of whole flakes characterized by cortex," ">50/o cortex," "riverine cortex," "upland cortex," "heavy weathering," and "thermal alteration." The perspective of technology related to these selected attributes becomes fine-grained when data are segmented, illuminating areas of "statistical noise" where change in technology might be found. Statistical noise is identified by anomalies in the continuity of statistical test-result values. If significant differences 90

PAGE 110

exist, then an inference of variation in size-frequencies between the levels may be advanced. More specifically, different levels of lithic production may be inferred from these data Cumulative (ogive) size distribution for the whole-flake assemblage within each level and between the three levels is first considered. Size data are best represented by line graphs where size is the independent (X) variable and cumulative frequency (cf 0/o) is the dependent (Y) variable. An ogive distribution with a relatively flat trajectory indicates a high concentration of flakes in the small size categories, while a steep trajectory indicates a high concentration of flakes in the large size categories. A profile of these data is represented tabularly in Tables 4-6 and graphically in Figure 13. Variation in slope indicates variable proportions of size frequency data across the three levels Inconsistencies in the proportion of size frequencies indicate variable production activities. For example, a preponderance of small flakes is indicative of retooling and late-stage reduction activities, while proportionately larger numbers of medium and large flakes are indicative of preform biface production Greater frequencies of flakes in size categories 1-3 may characterize the Level 1 sample, while greater frequencies of flakes in size categories 4-9 may characterize Level 2 and Level 3 samples. Such representation indicates a decrease in flake frequency over time, from Level 3 to 91

PAGE 111

Table 4. Frequency Distribution for All Whole Flakes in Level 1. Level 1 (0.36000 m3). Frequency D i stribution of Raw and Adjusted Data by Size fo r all Whole Fla ke s size f f/m3 0/o cf 0/o LOG1 0 predicted Y 1 1278 3550. 0 61/o 61/o 3.6 3.4 2 387 1075. 0 18/o 79/o 3.0 3 1 3 160 444.4 8/o 87/o 2 6 2 8 4 134 372.2 6/o 93/o 2 6 2 5 5 68 188. 9 3/o 96/o 2 3 2 3 6 46 127.8 2/o 98/o 2 1 2 0 7 15 41.7 1/o 990/o 1.6 1.7 8 6 16.7 OO!o 990/o 1.2 1.4 9 7 19.4 oolo 990/o 1.3 1.1 TOTALS 2101 5836.1 100/o 100/o Table 5. Frequency Distribution for All Whole Flakes i n Level 2. Level 2 (0. 3 8 625 m 3 ) F r e quency D istributi on of Raw and Adju s ted Data by Si ze for all Whole Flakes size f f/ m3 0/o cf 0/o LOG10 predicted Y 1 1967 5092.6 62/o 62/o 3.7 3.5 2 563 1457. 6 18/o 80/o 3.2 3.2 3 233 603. 2 7/o 87/o 2.8 3.0 4 185 479.0 6/o 93/o 2.7 2.7 5 111 287.4 4/o 97/o 2.5 2.4 6 64 165.7 2/o 990/o 2.2 2.1 7 25 64.7 1 /o 100/o 1.8 1.8 8 12 31.1 OO/o 100/o 1.5 1.6 9 9 23. 3 0/o 100/o 1.4 1.3 TOTALS 3169 8204.5 100/o 100/o 92

PAGE 112

Table 6. Frequency Distribution for All Whole Flakes in Level 3 Level 3 (0.29725 m3). Frequency D istri butio n of Raw and Adjusted Data by Size for all Whole Flakes size f f/m3 0/o cf 0/o LOG10 pred i cted Y 1 1610 5416.3 63/o 63/o 3 7 3.6 2 435 1463.4 17/o 80/o 3 2 3.3 3 185 622.4 7/o 87/o 2.8 3.0 4 143 481.1 6/o 93/o 2.7 2.7 5 89 299.4 3/o 96/o 2.5 2.4 6 66 222.0 3/o 990/o 2.3 2.2 7 26 0.0 1/o 100/o 1.9 1.9 8 9 30.3 0/o 100/o 1.5 1.6 9 6 0.0 0/o 100/o 1.3 1.3 TOTALS 2569 8534.9 100/o 100/o Figure 13. Cumulative Size Frequency Distribution of Adjusted Data for All Whole Flakes in Levels 1-3. 120% 100% -80% 60% 7 ,,, .< 40% J -Level 1 -Level 2 20% .. <; ... ' Level 3 0% 1 2 3 4 5 6 7 8 9 93

PAGE 113

Level 1. Such a decrease may be viewed as reduced levels of production. Reasons for reduced levels of production may include reduced availability of raw materials, due to environmental changes which limited access to raw materials, and a shift in priorities from lithic technology to ceramic technology in the early part of the 4th millennium BP. As regional sources of raw materials became less accessible due to geographic circumscription during the post-ceramic phase of the Late Archaic period, use of local raw materials increased (Sassaman et al. 1990: 161). Although the Big Pine Tree Site was historically a regional quarry for high quality Coastal Plain chert, it became the focus of localized procurement during the Late Archaic. Graphic representation of size frequency values for all whole flakes indicates a strong correlation between Level 2 and Level 3, suggesting homogeneity of production activities during the occupation of these levels. While the slopes of Level 2 and Level 3 are similar, the slope of the Level 1 sample departs slightly from this similarity. Flake frequencies in all size categories are proportionately larger in the Level 3 sample, and decrease systematically in Level 2 and Level 1. Whole flake frequencies in Level 1 are proportionately less in size categories 1-3, indicating a departure from the routine activities of the earlier occupants of Level 2 and Level 3. Furthermore, a higher proportion of Size 6 flakes was found in Level 3. It may be inferred from that 94

PAGE 114

production focus was on a larger type of biface, specifically preforms, during the earliest occupations of the midden. This inference is consistent with the large quantities of preforms recovered within the midden and in Smith's Lake Creek. Finally, graphic representation of size frequency values for all whole flakes indicates a non-normal distribution (non-linear) of size frequency data. Since ANOVA requires that data are distributed normally, these data are later transformed mathematically to render a normal distribution prior to testing by ANOVA. Then, cumulative flake size frequency data for all whole flakes from the Big Pine Tree Site are compared graphically with experimental biface reduction data (Patterson 1990:551). When plotted as graphic trajectories (Figure 14), these data show strong similarities in slope, with the exception of size-category 1 which is skewed for the Big Pine Tree Site, as previously discussed in this chapter. These data are significant because they support the inference of biface production at the Big Pine Tree Site. The Big Pine Tree Site exhibits substandard size frequency data in size categories 2-4, and proportionately larger frequencies in size categories 4-8, when compared to Patterson's experimental data. These data suggest emphasis on the manufacture of larger bifacial tool types such as preforms. 95

PAGE 115

Size frequency data from the Big Pine Tree Site are then compared by cumulative percentage with experimental data derived from recent doctoral dissertation work at the University of Florida (Austin 1997:269) (Figure 15). With the exception of the high frequency in Size 1, graphic configuration of data from the Big P i ne Tree Site most closely resembles experimental biface reduction data. Graphic plots between the experimental biface reduct i on assemblage and the Big Pine Tree Site assemblage from the midpoint of Size 3 to the midpoint of Size 9 are similar. The consistency of m id-to-large size flake accumulat i ons represented by Figure 15, and the slight gain in accumulation at the Big Pine Tree Site between sizes 5 and 8, suggest a focus on the manufacture of larger bifaces, such as preforms. This inference is supported by the fact that preforms, by definition, lack the finishing touches of pressure flaking which enhance frequencies in small size categories. Standardized frequency data (ffm3) from each level are used to compute mean and standard deviation for each of the nine size categories. Mean size frequency data are important because they provide an "average value" which may be used to characterize a whole sample Mean values, or averages, provide "signatures for frequency distributions, and are readily comparable with data from other 96

PAGE 116

Figure 14. Cumu lative S ize Freq u ency Da t a from t h e Big Pine Tree S ite and Patt erson's 1990 Expe rimental B i face Reduct i o n Da t a. 1 2 0% 1 0 0 % >-u Big Pin e Tree S it e c Q) 8 0% / / Debitage Sam p i e :::l w .... Experimental Bifac e 0 6 0% Reduction (Patterson Q) I 1 9 9 0) > :;:; rtl :5 4 0% E I :::l u 2 0% 0 % 1 2 3 4 5 6 7 8 9 size F igure 15. Cumula tive Size Freq u e n cy Dat a f rom the B i g P i ne Tree S ite and Three Other Exper imental Data Sets. 120.0% () 100.0% c Q) :::l 80.0 % / 7 -Big Pine Tree o5 ite Q) L-..... 60.0% ;/' Experimental 0 B if ace Q) j Re duction > :;::::; 4 0 0 % / -Experimental U n iface :::l E Reduc tion :::l u 20.0% -Experimental Core Reduction 0.0 % 1 2 3 4 5 6 7 8 9 1 0 size 9 7

PAGE 117

statistical populations. Mean values are computed by summing the scores of each size category in each sample, and dividing by the number of scores. Standard deviation, the square root of variance (s2), is calculated by determining a score's deviation from the sample mean within each level. Standard deviation size frequency data are important because they are the basis from which inferential statistical tests such as ANOVA are derived. The computational formula for standard deviation is: s = ys2 = Y[L(x-Xm)2/n-1], where Y[L(x-Xm)2] equals the sum of the squares (55) of the standard deviations of each score (Drennan 1996: 174). Squared values of standard deviations are used to eliminate the cancellation effect of negative values added to similar positive values. Variance is the basis of ANOVA, which is expressed as a ratio (F) of variance between the samples and variance within the samples of two or more samples: F = s2 between I s2 within. 98

PAGE 118

With these statistical tools, variations between levels may be identified. Size frequency variations in the assemblage of "all whole flakes" may indicate an increase or decrease in volume of tool production. Size frequency variations between levels within a particular attribute class may indicate a shift in production strategy or change in raw material availability. Mean and standard deviation of data from the Big Pine Tree Site are graphically summarized in Figure 16. Because of the skewing affect of extreme values of Size 1 whole flakes in each of the three levels, Size 1 data were eliminated for this calculation. It may be seen Figure 16. Sample Means and Standard Deviations fo r Flake Size Frequency in Levels 1-3. 700 devi 600 500 >-400 u c ., :::> CT 1'! -300 200 100 0 2 3 leve l 99

PAGE 119

from the graph that in Level 1, standard deviation departs from its mean by 32/o, while in Level 2 and Level 3, standard deviation departs from its mean by 59/o and 71 /o respectively, indicating a closer association of size frequency data between Level 2 and Level 3. Furthermore, mean values for Levels 2 and 3 vary by Jess than 1 /o, from which a close correlation in debitage production may be inferred. Clearly, the greatest departures in size frequency data for all whole flakes is evident in the Level 1 assemblage. Analysis of Variance Analysis of Variance {ANOVA) is used to test theY-values of the linear regressions of each sample (level) for acceptance of the null hypothesis that statistically significant variation does not exist within or between the samples. This procedure follows Sakal and Rohlf (1981 :499-509). Regression determines the best-fitting straight line for a set of data. The term "best fit" defines a straight line to which all data points are separated from the line by a distance equal to the least number of squared distances from the line. The best fit is derived from the predicted Y values (frequency values) rendered by regression analysis. Distances are squared so that negative and positive values 100

PAGE 120

do not cancel each other out. The equation for linear regression is the basic linear equation Y = bX =a. Statistically significant values are considered those which are tested at 95/o confidence-that is, 95/o of all the possible populations in a normal distribution have a similar or the same mean. Data requirements for testing by ANOVA include: 1) statistical normalcy; 2) homogeneity of variances of each sample; 3) independence of samples; and 4) random sampling. A normal frequency distribution is often represented graphically by a "bell curve" with the greatest frequency and single highest peak in the middle of the distribution, and least frequencies in and beyond the "tail" of the distribution (Figure 17). A normal frequency distribution where scores are equally distributed may be represented by a relatively straight line. To test abnormal data, such as data frequencies that appear as multi-peaked, asymmetrical distributions, special actions must be taken to normalize the data. Such actions include segmenting the samples so that each peak represents a single normal distribution. The present analysis considers treatments 101

PAGE 121

Figure 17. Bell Curve of Normally Distributed Data Set. tail 0 tail (levels) within segmented, paired size categories for significant effect. Thus, ANOVA is utilized to test the samples for variation within columns (levels) and between columns. Normalizing data is accomplished by transforming the non-linear (non-normal) data to predicted linear values that would occur under statistically normal conditions. This is done by calculating the natural logarithms (LOG10) of each of the three samples. Logarithms of each frequency value for the three levels are used to create a linear relationship (Austin 1997:226; Shennan 1990: 146-147), to which a regression line is fitted (Figure 18). Linear regression analyzes data by using the "least squares" method to fit a line through a set of observations. Data from linear regressions are then tested by ANOVA. 102

PAGE 122

Figure 18. Comparison of Logarithmic and Regression Trajectories for All Whole Flakes in Levels 1-3. ;:;) 4 3.5 3 E 2.5 2 ... (.!) 1 5 g ;;;E <.!:> 0 -' ;;;E ::::: 0 (3 g 0.5 0 4 3 5 3 2.5 2 1 5 0 5 0 4 0 3 5 3 0 2 5 2 0 1 5 1 0 0 5 0 0 1 2 3 ........ -2 3 :::-.... I 2 3 Level 1 Level 1 LOG 10 -Levell Regression -l 4 5 size 6 Level 2 -7 ---8 9 Level 2 LOG 1 0 -Level 2 Regression I ..._, 4 5 6 7 8 9 s lze Level 3 -Level 3 Reg ression Level 3 --L 0 G 1 0 --I 4 5 6 7 8 9 s lz e 103

PAGE 123

Testing the three predicted Y values is preferable to testing the LOG10 values since the regression lines provide a more normal (linear) distribution. The juxtaposition of LOG10 values and predicted Y values from linear regression offers a unique perspective of variation in LOG10 values from the predicted frequency values derived from linear regression. This perspective focuses on variations within each size category that represent increases and decreases in flake size frequency. Homogeneity of variances is verified by Hartley's F-max test which establishes a ratio between the sample with the largest variance (s 2 ) and the sample with the smallest variance. While many different statistical methods may be used to test homogeneity of variances, Hartley's F-max test has the advantage of testing more than two samples at one time. This test is based on the principle that a sample variance provides an unbiased estimate of the population estimate. Therefore, if the population variances are the same, the sample variances should be the same. A relatively large F-max ratio indicates a large difference between the sample variances. A small (near 1.00) ratio indicates that the sample variances are similar (homogeneity of variances). The formula for computing this test is, F-max = s2 (largest)/ s 2 (smallest), where s 2 = L(x-Xm)2/n-1. 104

PAGE 124

Computation of the F-max statistic yields a ratio of 1.036, with critical values of 6.000 (a=O.OS) and 9.900 (a=0.01) respectively, indicating homogeneity of variances. Statistical independence of the samples is verified by applying a paired two-sample t-test to the regression values of the size frequency data of paired combinations of each sample (ie: Level 1 versus Level 2, Level 1 versus Level 3, Level 2 versus Level 3). This test performs a paired two-sample t-test to determine whether a sample's means are distinct. Predicted Y values derived from regression of the samples are tested against each other to affirm independence of the sample means (Table 7). To satisfy the ANOVA requirement of statistical independence, rejection of the null hypothesis that the means are the same is anticipated. Statistical independence between all combinations of levels is verified for both one-tail and two-tail tests (Table 7). This is to say that in the one-tail test, S01o of the distribution is located in the tail beyond the critical value of 1.860. In the two-tail tests, S01o lie beyond the critical value of 2.300 and S01o beyond the critical value of -2.300. Since the distributions of samples considered in this thesis are non-parametric, the one-tailed test is considered most rigorous. All ttests are calculated using the "analytical tools" program in the Small Business Edition of Microsoft Excel (1997). 10S

PAGE 125

Table 7. T-Test Output of Regress i on Values for All Whole Flakes in Levels 1 3. t -Test: Pai red Two Sample for Means, Level 1 versus Level 2 Variable 1 Variable 2 Mean 2.257 2.409 Variance 0.602 0 .586 Observations 9.000 9.000 Pearson Correlation 1.000 Hypothesi zed Mean 0.000 Difference df 8.000 t Stat -44.513 P(T=::;t) one -tail 0.000 t C r itical one-tai l 1.860 P(T=::;t) two-tail 0.000 t Critical two-tail 2.306 t-Test: Paired Two Sample for Means, Level 1 versus Level 3 Variable 1 Variable 2 Mean 2.257 2.436 Variance 0.602 0.588 Observations 9 .000 9.000 Pearson Correlation 1.000 Hypothesized Mean 0 .000 Difference df 8 .000 t Stat -61.962 P(T=:;t) one-tail 0.000 t Critical one-tail 1.860 P(T=:;t) two-tail 0 .000 t Critical two-tail 2.306 106

PAGE 126

(Table 7 continued) t-Test: Paired Two Sample for Means, Level 2 versus Level 3 Variable 1 Variable 2 Mean 2.409 2.436 Variance 0.586 0 588 Observations 9.000 9.000 Pearson Correlation 1.000 Hypothesized Mean 0.000 Difference df 8.000 t Stat -52. 843 one-tail 0.000 t Critical one-tail 1.860 two-tail 0.000 t Critical two-tail 2.306 The t-statistic for the comparison of Level 1 versus Level 2 i s -44.513. Since the absolute value of the t-statistic is greater than the critical values, the null hypothesis that the sample means are the same must be rejected. A similar conclusion is reached for Level 1 versus Level 3 (t-statistic = -61.962), and Level 2 versus Level 3 (t-statistic = -52.843). The absolute value of the t-statistic is greater than critical values for one-tail tests, therefore indicating reject i on of the null hypothesis that the means are the same, and confirming independence of the samples. 107

PAGE 127

A Pearson Correlation statistic (r-statistic), also displayed in the tables for each relationship, considers the quotient of the degree to which the X-variable and theY-variable vary together and the degree to which they vary separately. It is expressed as a measure of the strength and direction of the relationship between the X-distribution (independent variables) and theY-distribution (dependent variables). The computational formula for the r-statistic considers variability between the sum of products (SP) of two samples divided by the square root of the product of the sum of squares (SS) of the samples: r = SP I v'SSXSSY = I v'SSXSSY. When there is a perfect linear relation, every change in the X-variable is accompanied by a corresponding change in the Y-variable. The result appears graphically as a straight line of increasing values (rising from left to right). The closer the r-statistic is to the absolute value of 1, the stronger the correlation. A correlation is considered positive if it decreases from left to right on a line graph, and negative if it increases from -left to right. Based on the Pearson Correlation values, the three samples tend towards 1, decreasing from left to right, indicating a relatively strong, positive correlation between size (X) and frequency (Y). This correlation is represented by nearly parallel slopes (Figure 108

PAGE 128

19), which indicate strong similarity in the general patterns of lithic reduction through time. This inference is expected from the MALA assemblage at the Big Pine Tree Site, where the debitage assemblage is intermixed with large numbers of biface tools and preforms. At all three levels, bifacial manufacturing was the primary occupation. Random sampling refers to methods of data recovery that insure every member of a population has an equal chance of being collected. Additionally, a constant probability for selection must exist throughout the test area for sampling to be considered random (Gravetter and Wallnau 1991:124). The field approach to random sampling at the Big Pine Tree Site involved establishing a linear grid of seven contiguous 1 m x 1 m archaeological units (Units A-G). Unit F was devoted exclusively to the recovery of charcoal samples from which rad i ocarbon dates were obtained. All debitage i n the other 6 units was recovered by 1/8t h inch wetscreening, a technique using water hoses to wash the flakes as they are dumped from 5-gallon buckets into the screen. Artifacts from Units A and B were recovered and curated, however, these two units were not included in this analysis because they were rendered incomplete to a degree of fault by an intrusion of disturbed soils from the previous year's backhoe work. Units C D, E and G yielded approximately 80 pounds of debitage, whi ch was first sorted by wholeness, then by attribute 109

PAGE 129

classification, then by size for each attribute. By the end of the May 1995 field season, 40 square meters of the MALA midden had been excavated. The four square meters represented by Test Units C, D. E, and G is equivalent to 10/o of the known MALA component. The requirement of random sampling is satisfied by virtue of a 100/o collection strategy employed in the test units. To transform the data into a normal distribution, predicted linear regressions of the logarithmic values of the size frequency data are calculated for each level. The LOG10 function transforms a non-linear distribution (non-normal) to a relatively linear distribution (normal). The LOG10 transform also diminishes the scale of frequency values so that finer resolution of variation may be achieved. Linear expression of predicted Y values from regressions is ideally suited for ANOVA. Once the LOG10 values are calculated for all size categories in each level, regression values are calculated from the LOG10 values. This is done to find the best predicted slope of the residuals, those y-axis logarithmic values that do not naturally fall on the regression line. Since residuals account for eight out of nine observations (size), they form the best plot for statistical analysis of these data. Regression lines (Figure 19) allow three assumptions to be made about the statistical populations from which the samples are derived 110

PAGE 130

Figure 19. Regression Lines for All Whole F l a k e s in Level s 1 -3. -t.e;od ll 1 2 3 4 5 6 7 s size First, the independence of each populatio n i s c onfirme d Second, there is a closer correlation between Levels 2 a n d 3 (thos e with the greatest antiquity) than between Leve l 1 and either Leve l 2 o r L eve l 3 Third .. a decrease in the flake frequency occurs over time. These ass umptiioos may be put in finer perspective by uti lizing t h e l ogarithmic data whfidhl have the advantage of being nearly linear ( n orma l ) while i ndicating variation within size categories (Figure 20 ) B etwee n Leve l s 2 and 3 frequency correlation is strongest within si z e cate g ories 1 through 5 .. 111

PAGE 131

Figure 20. LOG10 (f/m3 ) values for all whole flakes for Levels 1-3. Le v el l 0 5 +---------------------------------------------2 3 4 5 s i z e 6 7 8 9 The conformation of the slopes i n all three levels i s similar for smaller sizes. A higher frequency of larger flakes is characteristic of Level 3 Smaller size flakes are expected to indicate the bifacial reduction of finished tools, and larger flakes are expected to indicate the bifacial reduction of larger tools such as preforms, blanks, and cores. MALA points were crafted from the preforms. From the strong correlation evident in small flake size categories between Levels 2 and 3, it may be inferred that the makers of the MALA point occupied these levels. 112

PAGE 132

The people who occupied the upper level (Level 1) produced a similar pattern of debitage, but much less of it. By projecting radiocarbon dates onto a linear graph, juxtaposed with the mean values of the logarithmic data, a comparison between the rate of decline in flake frequency and time may be made (Figure 20). While variations in the LOG10 (frequency) line between Levels 2 and 3 are barely distinct, a definite downturn in the line occurs Figure 21. Rate of Resource Depletion During MALA Occupat i on at the Big Pine Tree Site. 6000 5000 4000 3000 2000 1000 0 -------3,980 years BP ,. 1 4,430 years BP 2 level 4,820 years BP C-14 dates (years ago before present) mean LOG (f/m3) x 1000 3 113

PAGE 133

between Level 2 and Level 1. Since real values are exponential in proportion to LOG10 values, even a slight variation in the line graph of logarithmic values can indicate a large change in frequency. The point on the graph at which the downturn occurs roughly corresponds with the invention of pottery in the Savannah River Valley. Decrease in flake frequency may be indicative of environmental and social causes. Depleted raw materials or the reconfiguring the natural or social environment may reduce accessibility to raw materials. Decrease in flake frequency may also be indicative of an improved technology in which less raw material was used. Furthermore, the introduction of ceramic technology undoubtedly diverted focus from lithic production. Finally, the linear regressions for all whole flakes of each level are tested by ANOVA. Since ANOVA delivers indications of variance within the three samples, the post-hoc application of the segmented, paired sample t-test provides finer resolution to variation. As discussed, predicted residual Y-values (frequency) are calculated for each X-value (size) by regression analysis for the three levels and tested by ANOVA (Table 8). Regression values render parallel, straight lines, ideally suited for ANOVA. In Table 8, 55 refers to "sum of the squares," df to "degrees of freedom" (the number of scores in a sample minus 1), and M5 to the 114

PAGE 134

Table 8. ANOVA Output for All Whole Flakes in Levels 1-3. ftnova: Single Factor for predctej reg-ession residuals Grouos CoU1t Sum Average Variance Colurm 1 9 20.314 2.257 0 602 Colurm 2 9 21.600 2.409 0 586 Colurm 3 9 21.927 2.436 0.588 f:N:NA Sourre of Variation ss elf MS F P-value Fait Betw:!en GrouJ:s 0.168 2 0 084 0.142 0.869 3.403 WthinQoups 14.209 24 0.592 TcA:al 14 377 26 "mean of the sum of squares." The F statistic is a ratio of the mean squares between groups to the mean squares within groups It must be less than the F-critical value, which refers to the outer-most limits of significance in the tai l of an F-distribution (Figure 22), in order to accept the null hypothesis that the squared means are the same. The F-critical value is found in the table of cri tical values for the F-d i stribution. That the F-value is less than the F-critical value allows us to accept the null hypothesis that no significant variation exists between the columns at the 95/o confidence level: F = 0.14172 < F-critical = 3.4028 Accept H o : (F = 26, df = 2, 24, p < 0.05). 115

PAGE 135

This homogeneity of size-frequencies between levels for all whole flakes is significant because it indicates similar biface reduction activities occurring in all three levels. This is to say that biface manufacture was a primary focus between 4,820 years BP and 3,980 years BP. The number of flakes for each size remains relatively proportionate between levels. To gain a finer perspective of variances within and between each level, samples are segmented in contiguous paired size categories. Figure 22. F-D i stribution Indicating the Outer Limits of Significance, or F-Critical Value. C ritical F 116

PAGE 136

Data derived from this procedure are summarized in Table 9. Even though the category "all whole flakes" indicates no significant variation at the 95/o confidence level, analysis of size frequency for segmented paired size categories within each attribute class indicates "noise" in two attribute areas: "> 50/o cortex" and "riverine cortex." Noise in the attribute class of"> 50/o cortex" located in size category 1-2 is suppressed by the closeness of the F-statistic (10.47) to the critical value (9.55). Such noise may be the result of size frequency variation. However, the weighted nature of Size 1 may encourage divergence of the means. Noise in the attribute class of riverine cortex is better defined. While noise in size category 1 2 of the attribute class riverine cortex" may be attributable to the weighted nature of Size 1, variation may be inferred in size category 2-3 and size category 3-4 with more confidence. Frequency mean variations within contiguous, paired size categories between levels may be verified by using the segmented sample paired t-test. Such post hoc testing typically follows ANOVA in order to identify specific aspects of variation. 117

PAGE 137

Table 9. ANOVA F-Statistics for All Attributes in Levels 1-3. ,ro cv zo u 0 u 0 ltl"t:: 1\ 8 "0 >-GI ... -Ql >.c .,..., Gin> :I:QI >=-u IOQI E.._ o.....l!l Ql.cn> 1118

PAGE 138

Post Hoc Testing A fine-grained view of the segmented samples of data for all whole flakes is obtained by comparing the means of paired subsets of size frequency data from each level utilizing the t -test. Specifically, each subset is composed of paired, overlapping size categories (i.e.: 1-2, 2-3, 3-4, etc.). Each attribute class is tested this way. Since the ttest is designed to test variation in means between two samples, subsets from Level 1 are tested against those from Level 2, then against those from Level 3; and subsets from Level 2 are tested against those from Level 3. Thus, 24 t-tests are performed at the 95/o confidence level for each attribute class. This methodology is helpful in discerning more specific information about the degree and nature of size frequency anomalies. All Whole Flakes T-statistics in size categories 4 5 and 5-6 in all whole flake comparisons of Level 1 versus Level 2 and Level 1 versus Level 3, indicate inconsistencies in mean variation (Table 10). The data within 119

PAGE 139

these size categories represent a "threshold of variation" in which anomalies occur within normal distributions. Such anomalies are not present in the comparison of Levels 2 and 3. From these data it may be inferred that variation from the earliest level (Level 3) of technological organization occurred during middle and later occupations of the midden (Level 1 and at least part of Level 2) within size categories 5-6 and 6-7, in an assemblage that considers sizefrequencies of all whole flakes. This argument is strengthened by graphic representation of LOG10 (f/m3 ) values for each level (Figure 20). Table 10. T-Statistics of Segmented and Paired Sample Data for All Whole Flakes in Levels 1-3. While flake frequencies in size categories 1-4 of Level 2 and Level 3 are relatively consistent, flake frequency in the same size categories of Level 1 indicate a substantial decrease. Furthermore, 120

PAGE 140

size categories 5-7 indicate a substantial decline in frequency values over time for all levels. Size 8 of Level 1 indicates a substantial decrease in flake frequency from an apparent equal expenditure of the same size in Level 2 and Level 3. Size 9 indicates equal expenditure in all levels. Raw data frequencies in size categories 8 and 9 are disproportionately small (i.e.: n = 6-12) in all levels. Small sample sizes may result in data which are not normal and may skew analysis results. However, such anomalies may represent variable levels of raw material expenditures, resultant of changes in technological organization. Cortical Whole Flakes By definition, cortical flakes must contain part of the weathered, rough, external surface of the parent stone. The rough, weathered rind that often appears as part of the external surface of a flake is referred to as cortex. To the craftsman of fine stone tools, cortex is considered undesirable. When a cobble is detached from a larger repository of parent stone, cortex often comes off with it. Such a cobble is further reduced to preform size, with additional detachments of cortex occurring throughout the reduction process. Visual determinations of cortex on the dorsal side of flakes are made, placing cortical flakes into either a 50/o" classification or a 121

PAGE 141

"> 50/o" classification. Since a flake detached either early or late in the reduction process may have =::; 50/o cortex, or no cortex, any position in the reduction sequence assigned to flakes in the "$ 50/o" class is equivocal. However, flakes with "> 50/o" cortex may be presumed to have originated in an early stage of the reduction process. It is expected that a preponderance of larger flakes with larger cortical constituents is to be found at a quarry site such as the Big Pine Tree Site. Cortex at the Big Pine Tree Site derives from two distinct quarry sites, both adjacent to the site, and each with its own distinctive signature. Riverine cortex derives from the now-inundated quarry site that lies submerged in Smith's Lake Creek. It is visually recognizable by its thin, smooth, dark brown surface (Plate 8). Even though the riverine quarry may have been inundated through much of prehistoric times, this assemblage accounts for nearly half of the cortical raw material recovered at the Big Pine Tree Site, and may have been the preferred material when accessible. Upland cortex is derived from numerous chert outcrops that protrude from slopes that lie immediately east of the Big Pine Tree Site. It is visually identified by its thick, light colored, foamy appearance (Plate 9). Since this resource lies above the floodplain of Smith's Lake Creek, it is presumed that access to it has remained 122

PAGE 142

Plate 8. Large Size Flake Riverine Cortex. Plate 9. Medium Size Flake with Upland Cortex. uninterrupted through t i me. Even so, material from the upland quarry was not preferred when the riverine source was available as i ndicated 123

PAGE 143

by increases in flakes from the riverine quarry during times of low riparian levels. Flakes with 50/o Cortex Flakes with 50/o cortex represent a class of cortical debitage in which at least half of the dorsal side of a flake is free of cortex. Since cortex offers no functional value, it is removed from the raw material early in the biface reduction process. Cortex may be removed by a lithic craftsman, or by natural erosive agents Furthermore, a "stage" in the biface reduction process may not be confidently inferred when analyzing flakes with 50/o cortical content on the dorsal side, since such a flake may occur early or late in the biface reduction process. Flakes in this class serve to differentiate flakes with > 50/o cortex, those flakes which are unquestionably among the earliest flakes to be removed in the reduction process. Flakes with 50/o cortex account for 76/o of all whole cortical flakes. Adjusted size frequency data are displayed in Tables 11-13. adjusted data show a disassociation of Level 1 from Level 2 and Level 3 (Figure 23) in a non-normal distribution of size frequency values. While disassociation between Level 1 and Level 2 is relatively distinct, except where data intersect in size categories 7 and 9, the 124

PAGE 144

Table 11. Adjusted Size Frequency Data for All Whole Flakes with Cortex in Level 1. Level 1 (0.36000 m3). Frequency Distribution of Data by Size for Whole Cortical Flakes with 50% cortex. size f f/ m3 % cf% LOG10 predicted Y 1 17 47.2 28% 28% 1.7 1.6 2 8 22. 2 13% 41% 1.3 1.5 3 4 11.1 7% 48% 1.0 1.4 4 11 30.6 18% 66% 1.5 1.3 5 7 19.4 12% 78% 1.3 1.1 6 7 19.4 12% 90% 1.3 1.0 7 2 5.6 3% 93% 0.7 0.9 8 3 8.3 5% 98% 0.9 0.8 9 1 2.8 2% 100% 0.4 0.7 totals 60 166. 6 100% 100% Table 12. Adjusted Size Frequency Data for All Whole Flakes with Cortex in Level 2. Level 2 (0.38625 m3). Frequency Distribution of Data by Size for Whole Cortical Flakes with <50% cortex. size F f/ m3 % cf% LOG10 predicted Y 1 23 59.5 22% 22% 1.8 1.8 2 21 54.4 20% 42% 1.7 1.7 3 10 25.9 10% 52% 1.4 1.6 4 18 46.6 17% 69% 1.7 1.5 5 10 25.9 10% 79% 1.4 1.3 6 12 31.1 12% 91% 1.5 1.2 7 4 10.4 4% 95% 1.0 1.1 8 1 2 6 1% 96% 0.4 1.0 9 5 12.9 4% 100% 1.1 0 8 totals 104 269.3 100% 100% 125

PAGE 145

Table 13. Adjusted Size Frequency Data for All Whole Flakes with Cortex in Level 3. Level 3 (0.29725 m3). Frequency Distribution of Data by S iz e for Whole Cortical Flakes with =:; SO% cortex. size f f/ m3 % cf% LOG10 predicted Y 1 17 57.2 24% 24% 1.8 1.9 2 10 33.6 28% 52% 1.5 1.7 3 16 53.8 16% 68% 1.7 1.6 4 9 30.3 16% 84% 1.5 1.4 5 8 26.9 8% 92% 1.4 1.2 6 7 23.5 8% 100% 1.4 1.0 7 1 3.4 0% 100% 0 5 0.9 8 1 3.4 0% 100% 0.5 0.7 9 1 3.4 0% 100% 0.5 0.5 totals 70 235.5 100% 100% disassociation between Level 1 and Level 3 is less distinct, especially in size categories 4, 7, and 9 (points on the graph that intersect). Ogive comparison of cumulative adjusted frequency percentage values (Figure 24) indicates substantially higher frequency accumulation in Level 3, and similar frequency values between Level 2 and Level 3, albeit frequency values for the most recent level, Level 1, are slightly lower than those of Level 2. Logarithmic values for these data (Figure 25) display a non normal pattern, even though the logarithmic scale is greatly reduced. Only after application of regression values do the data display a normal distribution. In the graph of regression values for each level, Level 1 and Level 2 display similar slopes, but distinct means. This is 126

PAGE 146

Figure 23. Adjusted Size Frequency Data for Whole Flakes with Cortex in Levels 1-3. 70. 0 -Level t -Le"Wel2 60. 0 so. o <0. 0 i .. 30. 0 20. 0 0 0 0 0 alze Figure 24 Cumulative Size Frequency for Whole Flakes with Cortex in Levels 1-3. 120% .,----------------------------, 100% ----80% .., E ::::-.,. 60% E B 40% -Levell 20% -Level2 -Levell 0 % 2 3 6 8 9 size 127

PAGE 147

Figure 25. LOG10 (f/m3 ) Values for Whole Flakes with Cortex in Levels 1-3. F igure 26. Regression Lines for Whole Flakes with Cortex in Levels 1-3. 2 0 1.8 1.6 1.4 1.2 ,.. '2 );1 1.0 0 8 0 6 0 4 0 2 0 0 6 8 size 128

PAGE 148

slope from which a different rate of use may be inferred. The intersection of the Level 3 line on the graph with Level 2 in Size 2 is an anomaly in the frequency distribution Likewise, the intersection of the Level 3 line on the graph with Level 1 in Size 6 is an anomaly. Anomalies such as these may represent a "threshold of change" in a normal distribution which is i nterrupted by another normal distribution A fine-grained perspective of these anomalies is obtained by t-testing segmented samples. A fine-grained perspective of the relationsh i p between the size frequency means of paired size categories for each level is obtained by reviewing the t statistics for flakes with 50/o cortex (Table 14). Rejection of the null hypothesis for the t-test confirms independence of the means, and supports an assumption of variation between Level 1 and Level 2 in all size categories. However t-testing of Level 1 versus Level 3, and Level 2 versus Level 3 show strong similarities between the size frequency means. Variation is expected between Levels 1 and 3 since they are not contiguous levels. However, similarities in the means of flake size frequency values between Level 1 and Level 3 may indicate anomalies such as disruption(s) in the ava i lability of raw material. For example material derived from the "riverine quarry" may have become unavailable due to inundati on caused by sea level r ise. A stati stical 129

PAGE 149

Table 14. T-Statistics of Segmented and Paired Sample Data for Whole Flakes w ith <50/o Cortex in Levels 1 3 size 1 2 2-3 3-4 4 5 5 6 6-7 7-8 8-9 > t sta t -46-6 8 -44.68 -42.68 -40.68 -38 .68 -36.68 -34. 6 8 -32.68 .- u acceot .J ,o', :EJ re ie ct 0 0 In V I > t-s tat -1.55 0.45 -2.45 4 .45 6.45 8.45 10.45 12.45 a c ceot l'L':t-': .. ..'k : .c r eiect relationship indicating similar means between Level 1 and Level 3 may also indicate the abandonment of a technical process and subsequent return to it. Although inequality of means indicated in size categories 1-2 and 2-3 of the Level 1 versus Level 3 comparison, the t statistics are tantalizingly close to the cr i tical value (6. 31) of the distributi on, and the inference of variation is not vigorously defended. Additionally, size categories which include S ize 1 flakes are skewed by the wider size range of category 1. Raw data for Size 1 of whole flakes with 50/o cortex indicate equal frequency values (n = 17) for both Level 1 and Level 3 which have the effect of drawing the means together (Tables 11 and 13). Thus, reject i on of the null hypothesis that the means are the same for whole flakes with 50/o cortex for size categor i es 1-2 and 2-3 in the Level 1 versus Level 3 comparison, is equ i voca l. Likewise, t-statistics in the Level 2 versus Level 3 comparison size 130

PAGE 150

categories 5-6, 6-7, 7-8, and 8-9 are close enough to the critical value to argue rejection of the null hypothesis in those categories. While independence of the means is confirmed in the sample of all flakes with 50/o cortex, the greatest variation is evident in the comparison of Level 1 and Level 2, with absolute values oft-statistics ranging between 46.68 (size category 1-2) and 32.68 (size category 8-9). These statistics contrast sharply with the t-statistics of comparisons between the other two level relationships, and greatly exceed the critical value (6.31) of the one-tailed t-test for paired samples. The relationship between Level 1 and Level 3 with the range oft-statistics falling between -9.54 and 4.46, indicates substantially less size frequency variation, as does the relationship between Level 2 and Level 3, where the range oft-statistics falls between -1.45 and 12.45. Thus, a significant amount of variation in flake frequency for flakes with 50/o cortex may be inferred for the most recent level (Level 1). This variation is indicative of diminished frequency in all size categories across Level 1. While variation between Level 2 and Level 3 is indicated, the degree of variation is greatly reduced. Thus, a close relationship for the sample of flakes with 50/o cortex may be inferred for Level 2 and Level 3, and a substantially more distant relationship may be inferred for Level 1. Given the vast differences 131

PAGE 151

between t-statistics in the Level 1 versus Level 3 comparison and the Level 2 versus Level 3 comparison, this inference may be advanced with confidence. Flakes with > 50/o Cortex Descriptive statistics in the attribute class of > 50/o cortex display variation are supportive of that found in the attribute class 50/o cortex. The attribute class of> 50/o accounts for 0.9/o of the whole flake population. These are flakes with cortex covering more than half of the dorsal side. Flakes with > 50/o cortex are expected to be found at or near a quarry site such as the Big Pine Tree Site. Most often, they are presumed to be primary" decortication, referring to their presumed primary position in the reduction process. Frequency distribution tables for each level indicate small raw sample size (Tables 15-17). While small sample size may influence statistical deduction from descriptive data, and cause errors in statistical testing of such data, raising each size frequency value to its projected frequency per cubic meter (f/m3 ) provides less volatile graphic values. Graphic representation of adjusted size frequency data (Figure 27) shows a highly irregular distribution, with obscure indications of flake 132

PAGE 152

Table 15. Frequency Distribution for Whole Flakes with > 50/o Cortex for Level 1. Level 1 (0.36000 rn3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Cortical Rakes with > 50% cortex. size f f/ m3 % cf/% LOGlO 1 7.0 19.4 30% 30% 1.3 2 4 0 11.1 17% 47% 1.0 3 1.0 2.8 4% 51% 0.4 4 4.0 11.1 17% 68% 1.0 5 2 0 5 6 9% 77% 0.7 6 1.0 2 8 4% 81% 0.4 7 0 0 0.0 0% 81% 0.0 8 0 0 0.0 0% 81% 0.0 9 4 0 11.1 17% 98% 1.0 TOTALS 23. 0 63.9 100% 100% Table 16. Frequency Distribution for Whole Flakes with >50/o Cortex for Level 2. predicted Y 1.0 1.0 0.9 0.8 0 7 0 6 0.5 0.4 0.3 Level 2 (0.38625 m3) Frequency Distribution of Raw and Adjusted Data by Size for Whole Cortical Flakes with > 50% cortex size f f/ m 3 % cf% LOG10 predicted Y 1 6.0 15.5 25% 25% 1.2 1.1 2 4.0 10.4 17% 42% 1.0 1.0 3 2.0 5.2 8% 50% 0 7 0 9 4 4.0 10.4 17% 67% 1.0 0.9 5 3 0 7 8 13% 80% 0 9 0.8 6 2.0 5.2 8% 88% 0 7 0 7 7 1.0 2.6 4% 92% 0.4 0.6 8 1.0 2.6 4% 96% 0.4 0.5 9 1.0 2.6 4% 100% 0.4 0.4 TOTALS 24.0 62.1 100% 100% 133

PAGE 153

Table 17. Frequency D i stribution for Whole Flakes with > 50/o Cortex for Level 3. Level 3 m3) Frequency D istribution of Raw and Adjusted Data by S i ze for Whole Cortical F l akes w ith > 50% cortex Size f f/ m3 % cf% LOG10 pred i cted Y 1 6 0 20.2 24% 24% 1.3 1.5 2 7 0 23 5 28 % 52% 1.4 1.3 3 4 0 13.5 16% 68% 1.1 1.1 4 4 0 13.5 16% 84% 1.1 0 9 5 2 0 6.7 8% 92% 0 8 0.7 6 2.0 6.7 8% 100% 0 8 0.5 7 0 0 0 0 0% 100% 0.0 0.3 8 0 0 0.0 0% 100% 0.0 0.1 9 0.0 0 0 0% 100% 0.0 -0.1 TOTALS 25. 0 84.1 100% 100% frequencies decreas i ng over time. Because the sample size is small, irregularities in the distribution of whole flakes with > 50/o cortex may be caused by sampling error. Ogive compari sons of the three levels (Figure 28) provide a more definitive representation of accumulat i on of flakes within size categories over time. These comparisons suggest that similar emphasis was being placed on similar production activities wh ich resulted in similar volumes of waste flakes being produced particularly i n size categor i es 1 5 Similarit i es in slope and cumulative frequency values between Level 1 and Level 2 support the previously suggested inference of strong correlation between these levels. Simi l arities in 134

PAGE 154

Figure 27. Adjusted Size Frequency Data for Whole Flakes w ith >50/o Cortex in Levels 1-3. 25. 0 ,--------------------------1 -Levell 6 8 s ize: Figure 28. Cumulative Size Frequency for Whole Flakes with >50/o Cortex in Levels 1-3. 120'11. ,----------------------------, 8 0 % e ;:::, ,. 60% a Leve l t -Level 2 level 3 2 0 % 0 % 6 8 9 1 0 size 135

PAGE 155

slope between these levels and Level 3 suggest similar manufacturing activities (i.e.: biface reduction). The disassociation of Level 3 from the two levels above it is due to a higher frequency of flakes with > 50/o cortex in Level 3. This result is consistent with observations made in analysis of other attribute classes. Graphic representation of logarithmic values (Figure 29) supports assumptions made about decreasing size-frequencies with time. This representation also supports the notion of irregularities in the data. Graphic representation of predicted regression values (predicted Y) provides a more lucid rendering of the data (Figure 30). Regression lines clearly demonstrate strong correlation of flake frequency per size category between Level 1 and Level 2; and the divergent nature of Level 3. As in the analysis of whole flakes with cortex, the divergent slope of the regression line for Level 3 whole flakes with >50/o cortex may indicate a different type of activity in which flakes with >50/o cortex were produced as a by-product at different rates in different size categories, than similar flakes in the upper two levels. While the weighted demeanor of Size 1 flakes influence slope to a limited degree, this influence is not enough to discount the inference that different manufacturing activities may have occurred during the earliest occupation (Level 3), than those of later occupations. 136

PAGE 156

F igure 29 LOG10 (f/m3 ) Values for Whole F l akes w ith >50/o Cortex in Levels 1-3. Figure 30. Regression Lin es for Who l e Flakes w ith >50/ o Corte x i n Levels 1-3. 1.8 Level I L evel2 1.6 -Level l 1.4 1.2 1.0 ::::---._ > "0 ., 0 8 a. 0 6 0 4 ----.:::: 0 2 0 0 1 2 3 4 s 6 7 8 2 saze 1 3 7

PAGE 157

Points of intersection between the Level 3 regression line and the other two levels indicate thresholds of change in which in which there is variation within a normal distribution. The distribution that occurs following the variation is also normally distributed. While the variation does not alter the results of the t-test, it is indicative of size frequency variation which should be noted. Such anomalies are verified by ttesting segmented size frequency data for whole flakes with > 50/o cortex. Similarities in size frequency means are indicated by acceptance of the null hypothesis fort-tests between Level 1 and Level 3, and Level 2 and Level 3 (Table 18). Association between the upper two levels indicates similar activities: the slope for Level 1 appears to be equal to the slope of Level 2. Variation indicated by t-statistics for the paired size categories in the comparison of Level 1 and Level 2 is due to diminished frequency. Furthermore, anomalies recognized in graphic representations of the data are supported by data derived from t-testing of segmented samples. For example, an anomaly in the distribution develops in the comparison of the lower two levels between size categories 4-5 and 5-6. The consistency oft-statistics on both sides of this "threshold of variation" indicates statistically significant variation. A similar anomaly exists in the Level 1 versus Level 3 test between size categories 5-6 138

PAGE 158

Table 18. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with >50/o Cortex in Levels 1-3. > iii u :e 0 u > ::1'1 0 0 IJ') A > and 6-7. Again, the stronger association is lodged in the lower two levels. The high t-statistic values evident in the comparison of size frequency means in Level 1 and Level 2 indicate more distance from the critical value and stronger disassociation in size-frequencies than in Level 1 and Level 3, and Level 2 and Level 3. The adhesion discernable in comparisons of Level 1 and Level 3, and Level 2 and Level 3 is indicative of shared means at points of intersection on the regression graph. However, this does not preclude the inference of different manufacturing activities in Level 3. Riverine Quarry Cortical Flakes Inferences derived from cortical flakes are further supported by descriptive data of the attribute class "riverine quarry." "Riverine quarry" is another cortex-based attribute which accounts for 2/o of the 139

PAGE 159

total of all whole flakes, and 48/o of the whole cortical flake assemblage These flakes were not differentiated by percent of dorsal cortex. With the results of ANOVA testing of segmented samples, it is expected that flakes with riverine cortex display significant variation in size categories 1-2, 2-3, and 3-4. While values related to Size 1 may be slightly skewed, associations with Size 2 and Size 3 clearly exceed the critical value for ANOVA, where degrees of freedom equal 2 between groups and 3 within groups (df = 2, 3). Frequency distribution tables (Tables 19-21) indicate the smallest frequency in this class occurring in Level 1 (n = 28). Raw data indicates an assemblage with severely diminished frequency values in size categories 2 -9 of Level 1. The greatest frequency (n = 62) lies within Level 2 which is similar in frequency to Level 3 (n = 57). While graphic representation of adjusted size-frequencies for whole flakes with riverine cortex in Level 1, Level 2, and Level 3 supports the inference of a close relationship between Level 2 and Level 3, the degree to which size frequency decreases in Level 1 is extraordinary (Figure 31). A substantially quicker reduction in flake size frequency is clearly evident in Level 1 for flakes with riverine cortex, compared to other cortical attribute classes. Such reduction may be resultant of 140

PAGE 160

Table 19. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 1. Level 1 (0.36000 rn3 ) Frequency Distribution of Raw and Adjusted Dat a by Size for Whole Riverine Flakes size f f/m3 % cf% LOG10 Pred i cted Y 1 12 33.3 43% 43% 1.5 1.2 2 4 11.1 14% 57% 1.0 1.1 3 3 8.3 11% 68% 0.9 1.0 4 2 5.6 7% 75% 0.7 0.8 5 2 5.6 7% 82% 0.7 0.7 6 1 2 8 4% 86% 0.4 0.6 7 0 0.0 0% 86% 0.0 0 5 8 0 0.0 0% 86% 0.0 0.4 9 4 11.1 14% 100% 1.0 0.2 TOTALS 28 77. 8 100% 100% Table 20. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 2. Level 2 (0. 38625 rn3 ) Frequ e ncy Distribution of Raw and Adjusted Da t a by Size for Whole Riverine Flakes size f f/m3 % cf% LOG10 predicted Y 1 14 36 2 23% 23% 1.6 1.7 2 13 33 7 21% 44% 1.5 1.5 3 8 20 7 13% 56% 1.3 1.4 4 10 25 9 16% 73% 1.4 1.3 5 7 18. 1 11% 84% 1.3 1.1 6 5 12.9 8% 92% 1.1 1.0 7 2 5.2 3% 95% 0.7 0 8 8 1 2.6 2% 97% 0.4 0.7 9 2 5 2 3% 100% 0 7 0.6 TOTALS 62 160. 5 100% 100% 141

PAGE 161

Table 21. Frequency Distribution for Whole Flakes with Riverine Cortex in Level 3. Level 3 (0.29725 rn3 ) Frequency Distribution of Raw and Adjusted Data by S i ze for Whole Riverine Flakes size f f/ m3 % cf% LOG10 predicted Y 1 15 50.5 26% 26% 1.7 1.9 2 10 33 6 18% 44% 1.5 1.7 3 13 43.7 23% 66% 1.6 1.5 4 8 26.9 14% 80% 1.4 1.3 5 5 16.8 9% 89% 1.2 1.0 6 5 16.8 9% 98% 1.2 0.8 7 0 0.0 0% 98% 0.0 0.6 8 1 3.4 2% 100% 0 5 0.4 9 0 0.0 0% 100% 0.0 0.1 TOTALS 57 191.8 100% 100% environmental or social change Furthermore, the apparent shared means of Level 2 and Level 3 in size categories 2, 4, 5, 6, and 8 suggest shared usage of the riverine quarry by Level 2 and Level 3 populations. These data are consistent with sea level change data discussed in Chapter 3, and consistent with results oft-testing of segmented data discussed later in this chapter Graphic representation of ogive comparison of the three levels indicates similar slopes for Level 2 and Level 3, albeit reduced frequency (Figure 32). Since slope represents a pattern of production (i.e.: similar proportions of frequencies across size categories), the 142

PAGE 162

inference of similar production between Level 2 and Level 3 may be advanced. Logarithmic values of adjusted raw data highlight the irregularity of the distribution (Figure 33). Additionally, similarities in slope between Level 2 and Level 3 are indicated; and equality of frequency is apparent in Size 4 and Size 5. These results support the inference of a similar technology that extended from Level 3 occupations into Level 2 occupations with time. Both slope and frequency of the Level 1 data depart radically from slope and frequency of Level 2 and Level 3 strongly suggesting variation in product ion technique. With the exception of a score of 0 in Size 9 of Level 3, which may be the result of sampling error, the elevated frequency indicated in Size 8 of Level 3 suggests that the occupants of Level 3 did most of the mining from the riverine quarry. These data suggest a pattern of decreasing mining activity from the riverine quarry ove r time. Pronounced separation of the regression line for Level 1 supports inferences made thus far about the frequency of flakes utilizing chert derived from the riverine quarry (Figure 34). The exaggerated decrease of flake frequency across all size categories may indicate a differential in the availability of raw material. However, other inferences related to a sharp decrease in frequency may relate to technological improvements that conserved waste flakes, elim i nation 143

PAGE 163

Figure 31. Adjusted Size Frequency Data for Whole River i ne Quarry Flakes in Levels 1-3. 60.0 ..----:-::---------,.-----------,----.......f""--=::-:-l= . ::- :-1 -, -Ltvel2 i i 30. 0 I 9 lite Figure 32 Cumulative Size Frequency for Whole Riverine Quarry Flakes i n Levels 1-3. 120% 100% 80% 'E i 60% I 40% 20% 0 % 8 9 slzt 144

PAGE 164

Figure 33. LOG10 (f/m3 ) Values for Whole Riverine Quarry Flakes in Levels 1-3. 1.8 0 0 8 Figure 34. Regression Lines for Whole R iverine Quarry Flakes i n Levels 1-3. --Level 1 -Level 2 -Level 3 2 3 4 s 6 8 9 stze 145

PAGE 165

of the need for MALA tools, o r reduction in populat i on caused by soc ial or economic division. Inferences derived from descriptive data of r iveri ne flakes are further supported by t-testing of segmented samples (Table 22). Table 22. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with Riverine Cortex in Levels 1-3. > X (I) t 0 u > (I) c: ;;: (I) > > This attribute class is particularly interesting in its close association of size frequency means between Level 2 and Level 3 and then again between Level 1 and Level 3. Access to the river ine source of raw material at the Big Pine Tree Site is dependent on the water level of Smith's Lake Creek. The riverine quarry is currently inundated, preventing access to it. Acceptance of the null hypothesis for the paired t-test in the comparison of Levels 1 and 3 indicates common frequencies of raw material. From this result, an inference o f similar 146

PAGE 166

availability of raw materials during oc c u pations of lletel :11. may be developed. Sea level change studies discussed i n Chapter 3 iiiM1fiitCallte tt1tllrrtef periods of low sea levels during which t ime the riveriine S(l)IUirncte have been available to the inhabitants of the MALA 1nllle periods of availability are interrupted by two period s of innuurndlattiioo which occurred between 5,000 radiocarbon years B P and 4,000 radiocarbon years BP, within the range of radioca rbon dates oitJtt:afil!lledl from Test Unit F at the Big Pine Tree Site. Therefo r e both eartiy (Levell 3) and late (Level 1) inhabitants of the midden ha d acc ess to the riverine quarry. The riverine quarry was ava il able to the MALA during middle (Level 2) and early (Level 3) occupatio n s. In the compari son of Level 1 and Level 2, support f o r the inference of variable resource availability is derived from the extreme t-statistics in all size categories that indicate confide n t re j e cti o n of the null hypothesis for the paired t-test that the means are simila r Th ese data suggest that the occupants of Level 1 had different ra w m ate ri a l access options than the inhabitants of Level 2 and Level 3 Whil e at least two interruptions in resource availability occurred during the MALA phase, a stronger association of similar resource availability exists between Level 2 and Level 3. 147

PAGE 167

Upland Quarry Cortical Flakes Flakes derived from the upland quarry account for 52/o of the cortical flake assemblage, and 2/o of the whole flake population Frequency distribution tables indicate the greatest frequency occurring within Level 2 (n = 72), and the smallest frequency occurring within Level 3 (n = 38) (Tables 23-25). It is expected that data from this attribute class will provide an image of data that varies as a function of availability to the riverine quarry. These cursory data support an inference of variable access to the riverine quarry, where early inhabitants (Level 3) had access to the riverine quarry and preferred its raw material to the upland quarry which was always accessible. During the most recent occupation (Level 1) of the midden, people relied on the upland quarry. During the period Level 2 was occupied, people had variable access to the riverine quarry, therefore relying on the upland quarry for raw material. At different times, variable amounts of the riverine quarry may have been accessible, depending on the dynamics of tributary development. That the riverine variety of coastal plain chert (CPC) seems to contain fewer crystalline inclusions (flaws) supports an inference of preferred use. However, without documented geological and experimental studies, the 148

PAGE 168

Table 23. Frequency Distribution for Whole Flakes with Upland Cortex in Level 1. Level 1 (0.36000 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes, Upland Quarry size f f/ m3 % cf% LOG10 predicted Y 1 12 33 3 22% 22% 1.5 1.5 2 8 22.2 15% 37% 1.3 1.4 3 2 5.6 4% 41% 0.7 1.3 4 13 36.1 24% 65% 1.6 1.2 5 7 19.4 13% 78% 1.3 1.1 6 7 19.4 13% 91% 1.3 1.0 7 2 5.6 3% 94% 0.7 0.9 8 3 8.3 5% 99% 0.9 0.8 9 1 2.8 1% 100% 0.4 0.7 TOTALS 55 152. 8 100% 100% Table 24. Frequency Distribution for Whole Flakes with Upland Cortex in Level 2. Level 2 (0.38625 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes, Upland Quarry size f f/ m3 % cf% LOG10 predicted Y 1 15 38.8 21% 21% 1.6 1.6 2 12 31.1 17% 38% 1.5 1.5 3 4 10.4 6% 44% 1.0 1.4 4 12 31.1 17% 61% 1.5 1.3 5 14 36.2 19% 80% 1.6 1.2 6 9 23.3 13% 93% 1.4 1.1 7 3 7.8 4% 97% 0.9 0 9 8 1 2.6 1% 98% 0.4 0.8 9 2 5.2 2% 100% 0.7 0.7 TOTALS 72 186.4 100% 100% 149

PAGE 169

Table 25. Frequency Distribution for Whole Flakes with Upland Cortex in Level 3. Level 3 (0.29725 m3 ) Frequency Distribution of Raw and Adjuste d Data by Size for Whole Flake!>, Upland Quarry size f f/m3 % cf% LOG10 predicted Y 1 8 26.9 21% 21% 1.4 1.6 2 7 23.5 18% 39% 1.4 1.5 3 7 23.5 18% 57% 1.4 1.3 4 5 16 8 13% 70% 1.2 1.1 5 5 16.8 13% 83% 1.2 1.0 6 4 13. 5 11% 94% 1.1 0 8 7 1 3.4 3% 97% 0 5 0.7 8 0 0.0 0% 97% 0.0 0 5 9 1 3.4 3% 100% 0.5 0 .3 TOTALS 38 127.8 100% 100% assumption that better quality chert is derived from the riverine quarry is offered with caution. From sea level change data discussed in Chapter 3, it is clear that multiple sea level fluctuations occurred during the 4th millennium BP. Additionally, studies indicate that the dynamics of modern Savannah River floodplain tributary development began around 4,000 years ago (Brooks et al. 1986). That these environmental dynamics coincide with variable availability of the riverine quarry is no coincidence. Graphs of the frequency of adjusted data support inferences of variable availability and preferred use (Figure 35). Clearly, adjusted data in Level 3 indicate lower frequencies in most size categories, and 150

PAGE 170

a more normal distribution of data. While frequency values of flakes with upland cortex in Level 1 and Level 2 are similarly higher than those for Level 3, they both indicate a non-normal distribution. Such interference in a normal distribution may indicate variable access to raw materials. Frequency values for upland quarry material occur in inverse proportion to those values for riverine quarry material. The close correlation between Level 1 and Level 2 that is apparent in statistical descriptions of flakes derived from the upland quarry is also inversely contrasted by similar data from the riverine quarry. Ogive comparisons which graphically depict the rate of flake accumulation across size categories provide a "site signature" which relates to a particular pattern of frequency (Figure 36). Frequency, when measured across time, correlates directly to a rate of usage of raw material, which implies technological organization. Level 1 and Level 2 have a strong association in this regard, in contrast to data from the attribute class "riverine cortex" where the stronger relationship exists between Level 2 and Level 3. Such correlation is based on similarities in frequencies for each size category, and is visually interpreted as the slope of a line graph. Variation in slope, indicating variable rates of usage through time, is an indicator of variable technological organization. Such variation may manifest in technological changes designed to 151

PAGE 171

Figure 35 Ad j us t ed S ize Freque n cy Data fo r Whole Upland Quarry F l akes in L evels 1-3. Figu r e 36 Cumulat i ve Size Frequency fo r Whole Upla n d Q uarry F l akes i n Leve l s 1-3. e ;:,. E a -Levell Level2 -level 3 3 6 8 9 slzt 152

PAGE 172

accommodate environmental changes such as tributary development. Subtle technological variations, such as the renegotiation of hammer force or the redesign of striking platform may have become necessary when alternating between the different grades of raw material available at the Big Pine Tree Site. Historically, flake size and frequency have demonstrated correlation with hammer force (Bordes and Crabtree 1968; Patterson 1981) and platform characteristics (Dibble and Pelcin 1995). Furthermore, changes in the organization of tool manufacture may result from the formation of social alliances, conflict, division, -or tribute. Fluctuations in population, creating more or less demand for tool production, also impact flake frequency data. LOG10 transformations indicate non-normal distributions of adjusted data (Figure 37). While diminished frequencies are indicated for flakes derived from the upland quarry in Level 3, similarities in slope between Level 1 and Level 2 are discernable. Similarity of slope between Level 1 and Level 2 becomes more apparent when eliminating the "0" value assigned to Size 8 of Level 3. The elevated frequency indicated in Size 9 of Level 3 suggests that the occupants of Level 3 did most of the mining from the upland quarry. These data suggest a pattern of decreasing mining activity from the upland quarry over time. However, given the small size of the samples within this attribute class, these data may be 153

PAGE 173

Figure 37. LOG10 (f/m3 ) Values for Whole Upland Quarry Flakes in Levels 1 3 Figure 38. Regression Lines for Whole Upland Quarry Flakes in Levels 1-3. 1.8 -Levell size 154

PAGE 174

skewed by sampling error. Since the predicted Y values are derived from logarithmic values, they display a normal perspective of the data (Figure 38). While the rates of flake production are variable for all levels as indicated by variation in slope, slopes are more similar between Level 2 and Level 3. This graph supports the inference of a secondary reliance on the upland quarry by the occupants of Level 3. Points of intersection indicate a threshold of change, where the value of change between segmented sample means shows variation within a normal distribution. Such a threshold exists where Level 1 crosses Level 3 in Size 3. The nature of this threshold becomes more apparent when t-testing segmented samples. Equal descending intervals oft-statistics in all size categories of the Level 1 versus Level 2 comparison support the assumption of normalcy. The larger half of the size categories shares means, and the means of the smaller size categories depart (Table 26). However, similarities between upland quarry mean data for Level 1 and Level 3 may indicate a cycle of variable accessibility to the riverine quarry, since these two levels are not sequential in time. An anomaly in the data occurring between size categories 2-3 and 3-4 of the comparison between Level 1 and Level 3 indicates similar mean values but of a slightly different organization. This anomaly is represented as variation in slope where the regression lines for Level 1 and Level 3 intersect 155

PAGE 175

Table 26. T-Statistics of Segmented and Paired Sample Data for Whole Flakes with Upland Cortex in Levels 1-3. > > X > cv t: 0 u -o c Ill > c. ::> (Figure 38). This inference is supported by the departure of similar means between Level 2 and Level 3. Since t-statistics associated with rejection of the null hypothesis in all three levels do not depart substantially from the critical value of the t-distribution, an assumption of significant variation between the three levels in the technological organization associated with raw material derived from the upland quarry is not advanced with vigor. While the organizational aspects of the occupants of Level 1 seem to depart from the organizational aspects employed by the occupants of at least part of Level 2 and all of Level 3, the conservative range oft-statistic values supports an inference of uninterrupted access to and continual use of the upland quarry. 156

PAGE 176

Flakes with Heavy Weathering Whole flakes with heavy weathering account for 11 /o of the whole flake population, and are identified by the presence of erosive material on their dorsal surfaces (Plate 10). Heavy weatheri ng r es ults Plate 10. Heavily Weathered Flakes. 157

PAGE 177

from age and prolonged exposure to the natural erosional processes (chemical, eolian, alluvial) inherent in the wet, temperate environment of the middle Savannah River Valley. Exposure of chert resources to erosional forces was amplified by the abrupt geological changes that were taking place during the initial stages of middle coastal flood plain development between 4,000 and 5,000 BP. It is expected that the highest frequency of flakes with heavy weathering is derived from the oldest level (Level 3). However, an inspection of the raw size frequency data tables for each level (Tables 27-29) shows the largest frequency value in Level 1 (n = 345). This unexpected excess is explained in part by the extraordinary value of Size 1 in Level 1 (n = 152), accounting for 44/o of the total whole flakes with heavy weathering in Level 1. Additionally, the frequent pitdigging by Late Archaic people inadvertently brought the debitage of earlier occupations to the surface of Level 1 (Sassaman, personal communication). While this large value may be the result of sampling error, it may also be an indicator of the scavenging of unused raw materials and discarded tools of earlier occupants. To reduce the possibility of sampling error related to the large value of Size 1 in Level 1, Size 1 in all levels was eliminated from analysis. With thi s correction to the data 158

PAGE 178

Table 27. Frequency Distribution for Whole Heavily Weathered Flakes in Level 1. Level 1 (0.36000 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes, Heavy Weathering size f f/ m3 % cf% LOG10 predicted Y 1 152 422.2 44% 44% 2.6 2 7 2 60 166.7 17% 61% 2.2 2.4 3 48 133.3 14% 75% 2.1 2.1 4 50 138.9 14% 90% 2.1 1.8 5 17 47.2 5% 95% 1.7 1.5 6 13 36.1 4% 98% 1.6 1.2 7 4 11.1 1% 100% 1.0 0.9 8 0 0.0 0% 100% 0.0 0.6 9 1 2.8 0% 100% 0.4 0 3 TOTALS 345 958. 3 100% 100% Table 28. Frequency Distribution for Whole Heavily Weathered Flakes in Level 2 Level 2 (0.38625 m3) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes, Heavy Weathering size f f/ m 3 % cf% LOG10 predicted Y 1 65 168.3 29% 29% 2.2 2.4 2 37 95.8 16% 45% 2.0 2.2 3 28 72.5 12% 58% 1.9 1.9 4 65 168.3 29% 87% 2.2 1.7 5 10 25.9 4% 91% 1.4 1.5 6 14 36.2 6% 97% 1.6 1.3 7 4 10.4 2% 99% 1.0 1.0 8 2 5 2 1% 100% 0 7 0.8 9 1 2.6 0% 100% 0.4 0.6 TOTALS 226 585.1 100% 100% 159

PAGE 179

Table 29. Frequency Distribution for Whole Heavily Weathered Flakes in Level 3. Level 3 (0.29725 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes, HE!avy Weathering size f f/ m3 % cf% LOG10 predicted Y 1 83 279.2 30% 30% 2.4 2.7 2 83 279.2 30% 60% 2.4 2.4 3 41 137.9 15% 75% 2.1 2.2 4 25 84.1 9% 84% 1.9 1.9 5 14 47.1 5% 89% 1.7 1.6 6 21 70.6 8% 97% 1.8 1.4 7 3 10.1 1% 98% 1.0 1.1 8 5 16.8 2% 100% 1.2 0.8 9 0 0.0 0% 100% 0.0 0.6 TOTALS 275 925.1 100% 100% in place, the overall value of Level 1 still exceeds the total of whole heavily weathered flakes for Level 3 by one flake. The conspicuous presence of heavily weathered flakes in Level 1 is an anomaly in the assemblage of all whole flakes. If the inference that late occupants of the MALA midden scavenged raw materials of earlier populations holds true, then several assumptions related to the inference of scavenging by later populations may be advanced: 1) during the Level 1 occupation, the availability of coastal plain chert (CPC) suitable for tool making was limited; 2) because of the limitations of availability, occupants scavenged pieces of chert left behind by previous occupants; 3) intensive pit-digging activities of more recent occupation(s) recycled 160

PAGE 180

heavily weathered flakes to the surface rendering these flakes available for reuse; 4) heavily weathered chert found in Level 1 may have been mined over 800 years earlier than non-heavily weathered chert found in Level 1. Graphic representation of adjusted size frequency data of whole heavily weathered flakes indicates a close correlation between Level 1 and Level 3 (Figure 39). Frequency data from Level 2 converges with data from Level 1 and Level 3 in Size 5, and the thr ee levels remain closely aligned in size categories 6-9. This anomaly may be explained by a facet of a technological organization based on selective scavenging, where larger pieces of chert were scavenged first. Graphic representation of predicted Y values from ogive, LOG10, and regression of the LOG10 {f/x) values projects different slopes for each level, signifying variable rates of usage of whole flakes with heavy weathering (Figure 40-42). Intersections of the regression lines signify anomalies in the data where frequencies between the levels become equal, although usage is occurring at a different rate. Anomalies such as these may represent shared aspects of technological organization between study units. T-tests of segmented data from the whole heavily weathered flake assemblage indicate the greatest differences in the means of size frequency exists between Level 2 and Level 3 {Table 30). S light 161

PAGE 181

Figure 39. Adjusted Size Frequency Data for Whole Heavily Weathered Flakes in Levels 1-3. 4$0. 0 L a \ -Levt12 \ Levell ,. . -\\ \ \ "' /\ ............... \. \...----=-=-----400. 0 350.0 300. 0 E zso. o i f 200. 0 ISO. O 100. 0 so o 0 0 J lr.e Figure 40. Cumulative Size Frequency for Whole Heavily Weathered Flakes in Levels 1-3. e i 60'111 f "' 40% / 7 y ...... I l 2 ........ ) 20% 0% 6 8 9 162

PAGE 182

Figure 41. LOG10 (f/m3 ) Values for Whole Heavily Weathered Flakes in Levels 1-3. Figure 42. Regression Lines for Whole Heavily Weathered Flakes in Levels 1-3. 3 0 .. Level 1 -Ltvt12 -level 3 2 5 2 0 .... ... ;:;E :::, "' 1.5 g 1.0 0 5 ............. 0.0 I 2 ) 4 5 6 7 8 9 site 163

PAGE 183

departures of the means are evident in size category 1-2 of the comparison between Level 1 and Level 2, and size categories 6-7, 7-8, and 8 9 of the comparison between Level 1 and Level 3. Interestingly, the comparison of heavily weathered whole flakes between Level 1 and Level 3 is almost identical to the comparison of whole flakes derived from the upland quarry between the same levels. Th i s s imilarity i s reversed in the same comparison for flakes derived from the riverine quarry. These data further support the inference of interrupted accessibility to the riverine quarry: the occupants of Level Table 30. T-Statistics of Segmented and Paired Sample Data for Whole Heavily Weathered Flakes in Levels 1-3. size 1 2 2 3 3 4 4 5 5 6 6 7 7-8 8 9 > t-stat 8 2 0 6 20 4.20 2 2 0 0 .20 -1.80 -3.80 ..- .c acce p t illol;,. .. ..... rc Q) r e iect :;: ;; t s t a t -8.20 6 .20 -4.20 -2. 2 0 0 .20 rc > accept Q) :I: reiect .. I" 1 and Level 3 had to rely in part on the upland qua r ry for raw material due to inundation of the riverine quarry. Uninte r rupted exposure of the upland quarry material to erosive elements supports an assumption of greater frequencies of highly eroded material deri ved from the upland 164

PAGE 184

quarry. If the riverine quarry was inundated during much of the Level 1 and Level 3 occupations, then much of the material scavenged by the occupants of Level 1 would have been upland quarry material mined during the occupation of Level 3. This inference is consistent with sea level change data previously discussed. Additionally, intensive pit-digging activities of Late Archaic people utilizing Level 1 accounts for the presence of heavily weathered flakes appearing in Level 1. Similarities in the means of size frequency data between Level 1 and Level 2 and between Level 2 and 3 indicate continued usage of the upland quarry material, even when the riverine material was available. During times of recessed sea levels which roughly occur during the beginning of the Level 3 occupation and the middle-to-latter part of the Level 2 occupation, raw material was accessible from the riverine quarry. During these times of recessed water levels, chert continued to be mined from the upland quarry, albeit in lesser quantities. During times of recessed water levels, scavenging for raw materials was not emphasized as a part of technological organization, since raw material was plentiful. Occupants of Level 1 experienced inundation of the riverine quarry and a supply shortage. This shortage of raw material from the riverine quarry required the incorporation of scavenging into the organization of lithic technology. Furthermore, as culinary pits were dug from the upper level, underlying heavily weathered material 165

PAGE 185

was brought into Level 1. Much of the material scavenged or dug up was upland quarry material originally mined by occupants of Level 2 and Level 3. Thermally Altered Flakes Thermally altered whole flakes account for 91 /o of the whole flake assemblage. In the organ i zation of lithic technology, raw material is thermally altered to enhance its flaking qualities. Thermally altered CPC is primarily identified by a lustrous finish that results during the thermal alteration process (Plate 11). Additionally, various coloration changes may occur which can be confused with the original raw material color. Extreme heat, or temperatures that are increased too fast can cause failures in the process. Since this attribute class is so large, it is expected to closely resemble graphic and analytical results obtained from the study of the attribute class, .. all whole flakes ... The technology of thermal alteration became increasingly popular as the constraints of increasingly sedentary life ways demanded a more economical use of raw materials. As previously discussed, the use of thermal alteration increased dramatically during the Late Archaic, then decreased with the advent of ceramics in the early 4th Millennium BP (Sassaman et al. 1990: 161) This trend is 166

PAGE 186

Plate 11. Thermally Altered Coastal Plain Chert Flakes. expected to be reflected in graphic depictions of the data for whole thermally altered flakes. Increased frequencies of thermally altered whole flakes are indicated by adjusted data frequency values (f/m3 ) until the most recent provenience, Level 1 (Tables 31-33), the occupants of which are presumed to have shifted technological focus from lithic production to the initial stages of ceramic production. As discussed, a radiocarbon 167

PAGE 187

Table 31. Frequency Distribution for Whole Thermally Altered Flakes in Level 1. Level 1 (0.36000 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for Whole Flakes with Thermal Alteration size f f/m3 % cf% LOG10 predicted Y 1 1200 3333.3 64% 64% 3.5 3.3 2 324 900.0 17% 81% 3 0 3 1 3 137 380.6 7% 88% 2.6 2.8 4 110 305 6 6% 94% 2.5 2.5 5 57 158. 3 3% 97% 2.2 2.2 6 37 102.8 2% 99% 2.0 1.9 7 14 38.9 1% 100% 1.6 1.6 8 6 16. 7 0% 100% 1.2 1.3 9 4 11.1 0% 100% 1.0 1.0 TOTALS 1889 5247.2 100% 100% Table 32. Frequency Distribution for Whole Thermally Altered Flakes in Level 2. Level 2 (0.38625 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for all Whole Flakes with Thermal Alteration size f f/m3 % cf% LOG10 predicted Y 1 1917 4963.1 64% 64% 3.7 3.5 2 522 1351.5 18% 82% 3.1 3.2 3 203 525.6 7% 88% 2.7 2.9 4 155 401.3 5% 94% 2.4 2.6 5 94 243.4 3% 97% 2.4 2 3 6 49 126.9 2% 98% 2.1 2.0 7 21 54.4 1% 99% 1.7 1.7 8 9 23.3 0% 99% 1.4 1.4 9 5 12.9 0% 100% 1.1 1.1 TOTALS 2975 7702 3 100% 100% 168

PAGE 188

Table 33. Frequency Distribution for Whole Thermally Altered Flakes in Level 3. Level 3 (0.29725 m3 ) Frequency Distribution of Raw and Adjusted Data by Size for all Whole Flakes with Thermal Alteration size f f/m3 % cf% LOG10 predicted Y 1 1446 4864.6 63% 63% 3.7 3.5 2 375 1261.6 16% 79% 3 1 3.2 3 163 548.4 7% 87% 2.7 2.9 4 125 420.5 5% 92% 2.6 2.7 5 76 255 7 3% 95% 2.4 2.4 6 59 198 5 3% 98% 2.3 2.1 7 23 77.4 1% 99% 1.9 1.8 8 9 30 3 0% 99% 1.5 1.6 9 6 20 2 0% 100% 1.3 1.3 TOTALS 2282 7677 0 100% 100% date of 4,430 90 years BP was produced from Level 2, and a radiocarbon date of 3,980 80 years BP was produced from Level 1. Graphic representation of adjusted raw frequency yields much of the same information as did the attribute class "all whole flakes," with strong correlation between frequency values of Level 2 and level 3 (Figure 43), with a general decrease in frequency over time. However the class of thermally altered whole flakes indicates a slightly increased frequency value in Level 2 when compared with Level 3. This increase in Level 2 is perhaps indicative of the pinnacle of thermal alteration use in the Late Archaic, just prior to the invention of pottery. Ogive comparison of the 3 levels indicates a shift from low frequency accumulation in smaller size categories (Level 3), to a rapid 169

PAGE 189

increase in smaller size categories (Level 1 and Level 2) (Figure 44). Level 1 and Level 2 are nearly equal with regard to rate of accumulation from small to large size categories, while Level 1 indicates a slightly stronger focus on the production of flakes between size categories 3 and 7. Perhaps this is indicative of a deferred focus on the manufacture of preforms, which produces a more rapid accumulation of larger flakes, during occupations of Level 1 and Level 2. Graphic representation of LOG10 (f/m3 ) values for thermally altered whole flakes supports the inference of decreased frequency over time (Figure 45). Since slopes representative of data from all levels are similar, the inference of variation in organization is not apparent in this attribute class. Thermal alteration was practiced by the occupants of all three levels, with little variation in proportions of frequency across the size categories. However, while independence of the means is indicated for most of the assemblage, predicted Y values of thermally altered whole flakes derived from regression analysis reveal a slight variation in the slope of Level 3 data, indicating a slightly different organization for Level 3 (Figure 46). These data support the inference of a shift from primary focus on preform manufacture during the occupation of Level 3, to a 170

PAGE 190

Figu r e 43. Adjusted S i z e Frequen c y Data for Whole Thermally Alt e red Flakes in L e vels 1 3 6 8 size Figure 4 4 Cumulative S ize F r e qu enc y for Whole Th e rm ally Altere d F la kes i n L e v els 1-3. Oglve Comparison o t Whol e F lakes w i t h T h erma l A l t eration 120% 100% 80% M E 60% <:T 40% 20% 0% 5 6 8 s fz e 9 9 171

PAGE 191

Figure 45. LOG10 (f/m3 ) Values for Whole Thermal l y A ltered F l a k es i n L e v els 1-3. 4.0 -Levell L e vel 2 3 5 Level 3 3.0 2.5 ;:,.. ;; 2.0 1.5 1.0 0.5 0 0 1 2 3 4 5 6 7 8 9 size Figu r e 46. Regression Lines for Whole Therma lly Altered F l a k es in Lev els 1-3. 4.0 -Level l -Level2 3 5 Level 3 3 0 2.5 ,. "2 .\1 2 0 .., .. 1.5 1.0 0.5 0 0 1 2 3 4 5 6 7 8 9 s i ze 172

PAGE 192

focus on biface manufacture and maintenance during the Level 1 and Level 2 occupations. The t-test of segmented paired samples for whole flakes with thermal alteration yields results consistent with the expectation that this attribute class is similar to the class of "all whole flakes" by virtue of the fact that nearly all flakes recovered from the MALA midden are thermally altered (Table 34). While independence of the means is indicated for nearly all size categories between levels, a correlation is indicated in the smaller size categories of the Level 2 versus Level 3 comparison. Table 34. T-Statistics of Segmented and Paired Sample Data for Whole Thermally Altered Flakes in Levels 1-3. "' > Q) ltl 16:: -o Q) .... Q) > ... -m ltl E > Q) These results strengthen the inference of strong correlation between Level 2 and Level 3 in size categories 1-2, 2-3, and 3-4. Correlation of frequency means in Level 2 and Level 3 may relate to a level of lithic 173

PAGE 193

technological organization (i.e. preform production) that was unique to earlier times in the Late Archaic Period, one in which thermal alteration was clearly integral. Disparity of the frequency means in later times may indicate a dismantling or disruption of this level of organization. Nonetheless, thermal alteration was a technique employed increasingly throughout the Late Archaic Period. 174

PAGE 194

Chapter 7. Summary of Inferences and Conclusions Inferences derived from statistical analysis of MALA flake data from the Big Pine Tree Site may be summarized as follows: 1) biface reduction was the primary lithic technological activity at the Big Pine Tree Site, and the focus of this activity shifted from the manufacture of preforms and bifaces to tool mai ntenance over time; 2) a decrease in production occurred over time, which was probably related to a shift in technological focus from lithic tool manufacturing to ceramic production duri ng the first half of the 4 t h millennium BP; 3) a closer correlation of manufacturing activities is indicated by data from Level 2 and Level 3, than by data from Level 1 and Level 3 and data from Level 1 and level 2; 4) the greatest variation is inherent in cortical flakes with > 50/o cortex derived from the riverine quarry, suggesting that access to the riverine quarry was variable; and 5) variability of access to the riveri ne quarry was dependent on fluctuation of water level of Smith's Lake Creek, which varied with sea level fluctuations and inundated the quarry at various times during the 4th Millennium BP. 175

PAGE 195

A preponderance of larger flakes found in Levels 2 and 3 suggests that biface manufacturing activities focused on the production of a larger type during the earliest phases of the MALA population. This inference is supported by large numbers of bifacial preforms recovered from the lower levels of the MALA midden. A preponderance of smaller flakes characterizes Level 1 suggests that tool maintenance and recycling was the primary focus of this occupation. The inference of recycling is further supported by a high percentage of heavily weathered flakes recovered from Level 1, suggesting that both pit digging and scavenging had become an integral part of technological organization by the end of the Late Archaic. These conclusions support the notion of depleted or inaccessible resources. All data suggest a decrease in lithic production activity over time. This decrease appears to have occurred gradually during occupations of Level 3 and much of Level 2, and accelerated during occupations of the latter part of Level 2 and all of Level 1. The time during which reduced production appears to have accelerated coincides with radiocarbon dates that point to a period in time when ceramic production was first becoming organized, suggesting a shift in technological focus from lithic production to ceramic production. These 176

PAGE 196

conclusions are supported by a close correlation of data from Level 2 and Level 3, and a departure of data similarity inherent in Level 1. Flakes derived from both the upland quarry and riverine quarry were recovered from all levels at the Big Pine Tree Site. Frequencies of flakes derived from the upland quarry vary in inverse proportions to those derived from the riverine quarry. While raw material derived from the upland quarry is presumed to have been constantly accessibility due to its location above the flood plain of Smith's lake Creek, increased proportions of material derived from the riverine quarry characterize Level 3 and Level 1, suggesting a preference for the riverine material when it was accessible. However, sea level studies indicate fluctuations in sea level that coincide with high frequencies of riverine flakes. Fluctuations in sea level have been directly associated with tributary development in the Savannah River floodplain. Such development undoubtedly relates to the development of Smith's Lake Creek, and to fluctuating water levels there. Fluctuations of 2-3 meters invariably inundated the riverine quarry located in Smith's Lake Creek, rendering it inaccessible to quarriers during at least two periods of time during the 4th Millennium BP. In conclusion, large scale preform manufacturing was the focus of technological organization during the occupations of Level 3 and 177

PAGE 197

part of Level 2. Variable access to the source of preferred raw material, together with geological changes driven by fluctuations in sea level stifled the high level of production to which these lithic technologists had become accustomed. Furthermore, the introduction of ceramic technology caused a reevaluation of technological organizational priorities, realized foremost by the occupants of Level 1. Inferential statistics applied to lithic debitage data recovered from the Big Pine Tree Site have generated these conclusions with confidence. While flaws in data collection such as disproportionate ranges for size categories 1 and 9 have been noted, they are not considered fatal to these conclusions, since they occurred consistently in all levels Future studies to further support conclusions derived from this thesis might address sourcing of raw materials inherent in finished tools recovered from the Big Pine Tree Site, to determine proportions of upland and riverine quarry material used. It is expected that such sourcing analysis will indicate a prefe r ence for material derived from the riverine quarry. Additionally, analysis of tools recovered from the upper levels of the site, equivalent to Level 1 discussed in this thesis, are expected to reveal a shift in technological organization that placed emphas i s on lithic retouch and recycling. Finally, the strength of inferences derived from this thesis by utilizing statistical applications such as ANOVA coupled with the post-hoc testing of segmented paired 178

PAGE 198

samples suggests that such a model for analysis is useful for the testing of lithic attributes. 179

PAGE 199

APPENDICES 180

PAGE 200

APPENDIX 1. Summary Statistics of MALA Bifaces Excavated at the Pen Point Site (after Sassaman 1985b: 10). n mean S.d. maximum minimum maximum length 7 62.45 9.02 53.20 75.80 maximum width 17 25.55 2.58 19.90 29.00 maximum thickness 19 8 .80 1.72 5.00 12.00 blade length 9 54.07 8.19 44.00 66.00 haft element length 16 9.21 0.66 8 .00 10.60 proximal haft element width 13 18.80 2.46 14.80 24.30 distal haft element width 18 18.00 2 .26 14. 00 23.20 mean edge angle 15 67.04 6.16 53.50 79.70 tip angle 10 64.30 9 .66 49.00 77.00 181

PAGE 201

APPENDIX 2 Lithic Artifacts Catalogue from the Big Pine Tree Site 1985 Underwater Recovery. Chronology /Type N Total 0/o Total Cum. 0/o Paleoindian 7 7.5 7 5 Fluted preforms 7 Early Archaic 8 8 6 16.1 Kirk point 5 Taylor point 1 Waller knife 1 Beveled blade 1 Middle Archaic 28 30.1 46. 2 Morrow Mountain point 9 Gui lford po int 1 MAlA point 18 Late Archaic 12 12. 9 59. 1 Savannah River Stemmed poil 12 Woodland 34 36. 6 95. 6 Woodland Stemmed point 34 Mississippian 4 4 3 100. 0 Tri angular point 4 Total 93 93 100.0 182

PAGE 202

APPENDIX 3. Ceramic Artifacts Catalogue from the Big P i ne Tree Site 1985 Underwater Recovery Period/Type N %Total Cum% Late Archaic 15 3 15 3 Stallings Plain 4 Stallings Punctated 2 Thorn's Creek 1 Thorn's Creek Punctated 5 Steatite disk fragments 3 Steatite vessel fragments 1 Woodland 64. 4 79 7 Refuge Simple Stamped 22 Deptford Check Stamped 32 Deptford Cord Marked 1 Deptford Linear Check 5 Ocumulgee Cord Marked 7 Mississippian 15 3 95 0 Savannah Cord Marked 5 Savannah Check Stamped 6 Savannah Plain 2 Other 4 8 99 8 Late prehistoric complicated sta 1 Unknown clay tempered 1 Unknown punctated clay tempe 1 Unknown incised 2 Eroded decorated and plain war 605 Total 709 99.8 183

PAGE 203

App e ndix 4 Regres s i on Ana l ysis for Level 1 All Who le F la k es. Level 1 Regress ion Analysis Regressio n Statistics Multip l e R 0 985 R Squa r e 0 970 Adj u sted R Sq u are 0 966 Standar d Error 0. 1 45 Observations 9 RESIDUAL OUTPUT PROBABIUlY OUTPUT Observation Predicted Y Resi dua l s Standard Residuals Percentile y 1 3 390 0.160 1.181 5.55 6 1.223 2 3 107 0 075 -0.556 16 .6 67 1.288 3 2. 824 0 1 7 6 -1.297 2 7.778 1.620 4 2 540 0 030 0.225 38. 889 2 107 5 2.257 0 019 0 141 50.000 2 .276 6 1.974 0 133 0.979 61.111 2 571 7 1.691 -0. 070 -0. 520 72.222 2.648 8 1. 4 07 0 185 -1.362 83.333 3 0 3 1 9 1.124 0 164 1.208 94.444 3 5 5 0 1 8 4

PAGE 204

Appendix 5 Regress i on Ana l ysi s for All Who l e Flak e s i n Level 2. Level 2 Regression Analysis Regression Stati stics M.Jitiple R 0.990 RSquare 0.979 Adjusted R Square 0 .976 S tandard Error 0.119 Observat io n s 9 RESIDUAL OUTPUT PROBABILTIY OUTPVT Observation Predicted y Residual s Standard Residuals Percentil e y 1 3 .527 0.180 1.621 5 .556 1.367 2 3 .247 0 .084 -0 .755 16.667 1.493 3 2.968 -0 187 -1.689 27. 778 1.811 4 2 .688 0 .008 -0 .073 38.889 2 .219 5 2 .409 0.050 0.447 50.000 2.458 6 2.129 0 .090 0 .810 61.111 2 680 7 1.850 0 039 0.351 72.222 2 7 80 8 1.570 -0 .078 0.699 83.333 3.1 64 9 1.291 0 .077 0 .689 94.444 3 .707 185

PAGE 205

Appendix 6. Regression Analysis for All Whole Flakes in Level 3. Level 3 Regression Analys is Regression Statistics Multiple R 0.985 R Square 0 .971 Adjusted R Square 0.967 Standard Error 0 .142 Observations 9 RESIDUAL OUTPUT PROBABIUlY OUTPUT Observation Predicted Y Residual s Standard Residual s Percentile y 1 3.557 0.177 1.333 5.556 1.305 2 3.277 -0.111 -0.837 16.667 1.481 3 2.996 -0.202 -1.524 27.778 1.942 4 2.716 -0.034 -0.257 38.889 2.346 5 2.436 0.040 0.301 50.000 2.476 6 2.156 0 .190 1.431 61.111 2.682 7 1.876 0 .066 0.496 72.222 2.794 8 1.596 -0.115 -0.863 83.333 3.165 9 1.316 -0.011 -0.080 94.444 3.734 186

PAGE 206

REFERENCES Aldenderfer, M. 1991 The Analytical Machine: Computer S imulation and Archaeological Research. In Archaeological Method and Theory vol. 3, edited by M Schiffer, pp. 195-247. University of Arizona P r ess, Tucson. Ahler, S A. 1971 1989 Projectile Point Form and Function at Rodgers Shelter, Missouri. Missouri Archaeological Society Research Series 8. Mass Analysis of Flaking Debris: Studying the Forest Rather than the Tree In Alternative Approaches to Lith i c Analysis: Proceedings of the University of Tulsa Conference on Lithic Analysis, edited by Donald 0. Henry and George H Odell, pp. 85-118. Archaeological Papers of the American Anthropological Association Number 1. University of Tulsa Amick, D. S. 1985 Buried Late Holocene Terrace Site Testing in the Central Duck River Basin. In Exploring Tennessee Prehistory: a Dedication to Alfred E. Guthe, edited by T. Whyte, C. Boyd, and B Riggs, pp. 23-38. University of Tennessee Department of Anthropology Report of Investigations 42, Knoxville. 1987 Lithic Raw Material Variability in the Central Duck River Basin: Reflections of Middle and Late Archaic Organization Strategies. Tennessee Valley Authority Publications in Anthropology 50, and University of Tennessee Department of Anthropology Report of Investigations 46, Knoxv i lle. 187

PAGE 207

Amick, D. S., and P. J. Carr 1996 Chang i ng Strategies of Lithic Technological Organization. In Archaeology of the Mid-Holocene Southeast, edited by Kenneth E. Sassaman and David G. Anderson, pp. 41-56. University Press of Florida, Gainesville. Anderson, David G. 1989 The Paleolndian Colonization of Eastern North America: a View From the Southeastern United States. Paper Presented at the 49th Annual Meeting of the Southeastern Archaeological Conference, Tampa, Florida. 1990 The Paleoindian Colonization of Eastern North America: a View from the Southeastern United States. In Research in Economic Anthropology: Early Paleoindian Economies of the Eastern United States, edited by Kenneth B. Tankersley and Barry L. Isaac, pp. 163-216. Supplement 5 JAI Press, Greenwich, Connecticut. Anderson, D. G., and J. W. Joseph 1988 Prehistory and History Along the Upper Savannah River: Technical Synthesis of Cultural Resource Investigations. Richard B. Russell Multiple Resource Area. Interagency Archaeological Services Division, National Park Services Division, National Park Service, Atlanta. Anderson, D. G. and G T. Hanson 1988 Early Archaic Settlement in the Southeastern United States: a Case Study from the Savannah River Valley American Antiquity 53 :262-286. Anderson, D. G., and J. Schuldenrein 1983 Early Archaic Settlement on the Southeastern Atlantic Slope: a View from the Rucker's Bottom Site, Elbert County, Georgia. North American Archaeologist 4(3):177210. 1985 The Rucker's Bottom Site (9Eb91) Archaeological Record. In Prehistoric Human Ecology Along the Upper Savannah River: Excavations at the Rucker's Bottom, Abbeyville and Bullard Site Groups, assembled by David G. Anderson and J. Schuldenrein, pp. 251-590. The Russell Papers, Atlanta Interagency Archaeological Services Division, National Park Service, Atlanta. 188

PAGE 208

Anderson, D. G., C. E Cantley, and J. Schuldenrein 1985 The Rocky River Site (38AB91) Archaeological Record. In Prehistoric Human Ecology Along the Upper Savannah River: Excavations at the Rucker's Bottom, Abbeville and Bullard Site Groups, assembled by D. G., Anderson and J. Schuldenrein, pp. 215-249. The Russell Papers, Atlanta Interagency A r chaeological Services Division, National Park Service, Atlanta. Anderson D G., S. T. Lee, and A. R. Parker 1979 Cal Smoak: Archaeological Investigations Along the Edisto River in the Coastal Plain of South Carolina. Occasional Papers 1. Archaeological Society of South Carolina, Columbia. Austin, R. J. 1997 The Economics of Lithic-Resource Use in South-Central Florida Ph.D. dissertation. Department of Anthropology, University of Florida, Gainesville 1999 Technological Characterization of Lithic Waste-Flake Assemblages: Multivariate Analysis of Experimental and Archaeological Data. Lithic Technology 24(1):55-68. Bamforth, D. B. 1986 Technological Efficiency and Tool Curation American Antiquity 51:38-50. Bense, J. A. (editor) 1983 Archaeological Investigations in the Upper Tombigbee Valley, Mississippi, Phase I U. S. Army Corps of Engineers, Mobile District. Office of Cultural and Archaeological Research, University of West Florida, Pensacola. 1986 The Midden Mound Project. The University of West Florida, Office of Cultural and Archaeological Research Report of Investigations 6, Pensacola Binford, L. R. 1980 Willow Smoke and Dogs' Tails: Hunter-Gatherer Settlement Systems and Archaeological Site Formation. American Antiquity 45:4-20. 1983 Working at Archaeology Academic Press, New York. 189

PAGE 209

Bland, M. C. P. 1995 Late Archaic Plant Use at the Big Pine Tree Site (38AL143), Allendale County, South Carolina Master's thesis, Department of Anthropology, University of South Carolina, Columbia Blanton, D. B. 1983 Lithic Raw Material Procurement and Use During the Morrow Mountain Phase in South Carolina. Master's thesis, Department of Anthropology, Brown University. Blanton, D. B and K. E. Sassaman 1989 Pattern and Process in the Middle Archaic Period of South Carolina. In Studies in South Carolina Archaeology in Honor of Dr. Robert L. Stephenson, edited by A. C. Goodyear and G. T Hanson, pp. 53-72. Anthropological Studies 9. South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Bleed, P. 1986 The Optimal Design of Hunting Weapons: maintainability and Reliability. American Antiquity 51:737-747. Bordes, F., and D. E. Crabtree 1969 The Corbiac Blade Technique and Other Experiments. Tebiwa 11:2:2-1. Idaho State University Museum, Pocatello, Idaho. Bradbury, A P and P. J. Carr 1995 Flake Typologies and Alternative Approaches: an Experimental Assessment. Lithic Technology 20(2): 100115. Brooks, M. J., D. J. Colquhoun, J. G. Brown, and P. A. Stone 1989 Sea Level Change, Estuarine Development and Temporal Variability in Woodland Period Su, bsistence-settlement Patterning on the Lower Coastal Plain of South Carolina. In Studies in South Carolina Archaeology in Honor of Robert L. Stephenson, edited by A. C. Goodyear and G. T. Hanson, pp. 91-100. Anthropological Studies 9. South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. 190

PAGE 210

Brooks, M. J., and K. E. Sassaman 1990 Point Bar Geoarchaeology in the Upper Coastal Plain of the Savannah River Valley, South Carolina: a Case Study. In Archaeological Geology of North America, edited by N., P. Lasca and J. E. Donahue. Centennial Special Volume 4, Geological Society of America, Boulder. Brooks, M. J., P A. Stone, D. J. Colquhoun, J. G. Brown, and K. Steele 1986 Geoarchaeological Research in the Coastal Plain Portion of the Savannah River Valley. Geoarchaeology 1:293-307. Bullen, R. P. 1975 A Guide to the Identification of Florida Projectile Points. Kendall Books, Gainesville. Bullen, R. P., and H. B. Greene 1970 Statigraphic Testing at Stalling's Island, Georgia. Florida Anthropologist 23:8-28. Butzer, K. W. 1978 Changing Holocene Environments at the Koster Site: a Geoarchaeological Perspective. American Antiquity 43:408-413. Cable, J. S. 1982 Organizational Variability in Piedmont Hunter-Gatherer Lithic assemblages. In The Haw River Sites: Archaeological Investigations at Two Stratified Sites in the North Carolina Piedmont, edited by Stephen R. Claggett and John S Cable, pp. 637-688. Report 2386, Commonwealth Associates, Inc., Jackson, Michigan. 1992 Haw River Revisited: Implications for Modeling Terminal Late Glacial and Early Holocene Hunter-Gatherer Settlement Systems in South Carolina. In Paleoindian and Early Archaic Period Research in the Lower Southeast: A South Carolina Perspective, edited by David G. Anderson, Kenneth E. Sassaman, and Christopher Judge, pp. 96-142. Council of South Carolina Professional Archaeologists, Columbia 191

PAGE 211

Camilli, E L., and J. I. Ebert 1992 Artifact Reuse and Recycling in Continuous Surface Distributions and Implications for Interpreting Land Use Patterns. In Time, Space, and Archaeological Landscapes, edited by J. Rossignol and L. Wandsnider, pp. 113-136. Plenum Press, New York Charles, T. 1981 Dwindling Resources: an Overture to the Future of South Carolina's Archaeological Resources Institute of Archaeology and Anthropology Notebook 13:1-13. Carmichael, D. L. 1977 Preliminary Archaeological Survey of Illinois Uplands and Some Behavioral Considerations. Midcontinental Journal of Archaeology 2:219-251. Claggett S. R., and J. S. Cable 1982 The Haw River Sites : Archaeological Investigations at Two Stratified Sites in the North Carolina Piedmont. Report 2386, Commonwealth Associates, Inc., Jackson, Michigan. Claflin, W. H., Jr. 1931 The Stalling's Island Mound, Columbia County, Georgia Peabody Museum of American Archaeology and Ethnology Papers 14(1). Coe, Joffre 1964 The Formative Culture of the Carolina Piedmont. Transactions of the American Philosophical Society 54( 5), Philadelphia. Colquhoun, D. J., and M. J. Brooks 1986 New Evidence from the Southeastern U. S. for Eustatic Components in the Late Holocene Sea Levels. Geoarchaeology 1:275291. Cordell, L. S. 1984 Prehistory of the Southwest. Academic Press, Orlando Crabtree, D E. 1982 An Introduction to Flintworking, Second Edition Idaho Museum of Natural History, Pocatello. 192

PAGE 212

Crook, M. R., Jr. 1990 Rae's Creek, a Multicomponent Archaeological Site at the Fall Line Along the Savannah River. Department of Anthropology, Georgia State University, Athens. Daniel, I. R., Jr. 1982 Test Excavations at the Deerstand Site (8Hi483A) in Hillsborough County, Florida. Interstate 75 Highway Phase II Archaeological Reports Number 2. Florida Department of State, Tallahassee. Daniel, I. R., Jr., and J. R. Butler 1991 Rhyolite Sources in the Carolina Slate Belt, Central North Carolina. Current Research in the Pleistocene 8:64-66. Davis, M. B. 1983 Holocene Vegetational History of the Eastern United States. In Late Quaternary Environments of the United States, Part 2: The Holocene, edited by H. E. Wright, Jr. pp. 166-181. University of Minnesota Press, Minneapolis. Delacourt, P. A. 1985 The Influence of Late-Quaternary Climatic and Vegetational Changes on Paleohydrology in Unglaciated Eastern North America. Ecologia Mediterranea Tome 11 (Fascicule 1):17-26. Delacourt, P. A., and H. R. Delacourt 1983 Late-Quaternary Vegetational Dynamics and Community Stability Reconsidered. Quaternary Research 13 :111-132. 1987 Long-Term Forest Dynamics in the Temperate Zone: a Case Study of Late-QuaternaryForests in Eastern North America. Ecological Studies 63. Springer-Verlag, New York. Dibble, H. L., and A. Pelcin 1995 The Effect of Hammer Mass and Velocity on Flake Mass. Journal of Archaeological Science 22:429-439. Dillehay, T D. 1984 A Late Ice-Age Settlement in Southern Chile. Scientific American 251:106-117. Drennan, R. D. 1996 Statistics For Archaeologists. Plenum Press, New York. 193

PAGE 213

Easterbrook, D. J. 1993 Surface Processes and Landforms. Macmillan Publishing, New York. Ebert, J. I. 1992 Distributional Archaeology. University of New Mexico Press, AI buquerq ue. Foss, J. E., D. P Wagner, and F. P. Miller 1985 Soils of the Savannah River Valley: Richard B. Russell Multiple Resource Area, Elbert County, Georgia. Russell Papers 1985. Interagency Archeolgical Services Division, National Park Service, Atlanta. 1987 Long-term Forest Dynamics of the Temperate Zone: A Case Study of Late-Quaternary Forests in Eastern North America. Ecological Studies 63. Springer-Verlag, New York. Futato, E. M. 1986 Patterns of Lithic Resource Utilization in the Cedar reek Reservoir Area. Southeastern Archaeology 2:118-131. Gifford-Gonzalez, D. P., D. B. Damrosch, D. R. Damrosch, J. Pryor, and R. L. Thunen 1985 The Third Dimension in Site Structure: an Experiment in Trampling and Vertical Dispersal. American Antiquity 50:803-818. Goodyear, A. C. 1974 The Brand Site: A Techno-Functional Study of a Dalton Site in Northeast Arkansas. Arkansas Archaeological Survey, Research Series 7, Fayetteville. 1982 The Chronological Position of the Dalton Horizon in the Southeastern United States. American Antiquity 47:382-395. 1989 A Hypothesis for the Use of Cryptocrystalline Raw Materials Among Paleoindian Groups of North America. In Eastern Paleoindian Lithic Resource Use, edited by C. J. Ellis and J. C. Lothrop, pp. 1-9. Westview Press, Boulder. 1998 Recent Findings of the Allendale Paleoindian Expedition. In Features and Profiles, edited by M. Groover, pp. 18-23. Archaeological Society of South Carolina, Columbia. 194

PAGE 214

1999a The Early Holocene Occupation of the Southeastern United States: A Geoarchaeological Summary. In Ice Age People of North America, edited by R. Bonnichsen and K. L. Turnmire, pp. 432-481. Oregon State University Press, Corvallis. 1999b Results of the 1999 Allendale Paleoindian Expedition. Legacy 4(1-3):8-13. Goodyear, A. C., and T. Charles 1984 An Archaeological Survey of Chert Quarries in Western Allendale County, South Carolina. Research Manuscript Series 195. Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Goodyear, A C., and J. E. Foss 1993 The Stratigraphic Significance of Paleosols at Smith' s Lake Creek, 38AL135, for the Study of the Pleistocene-Holocene Transition in the Savannah River Valley. In the Proceedings of the First International Conference on Pedo-Archaeology, edited by J. E. Foss, M. E. Timpson, and M. W. Morris, pp. 27-40. The University of Tennessee, Knoxville. Goodyear A. C., J. H. House, and N. W. Ackerly 1979 Laurens-Anderson: An Archaeological Study of the Interriverine Piedmont. Anthropological Studies 4. South Carolina Institute of Archaeology and Anthropology, University of South Carolina Columbia. Goodyear, A. C., J. L. Mitchie, and T. Charles 1989 The Earliest South Carolinians. In Studies in South Carolina Archaeology in Honor of Robert L. Stephenson, edited by A. C. Goodyear and Glen T. Hanson, pp. 19-52. Anthropological Studies 9. South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Goodyear, A. C., S. B. Upchurch, T. Charles, and A. B. Albright 1985 Chert Sources and Paleoindian Lithic Processing in Allendale County, South Carolina. Current Research in the Pleistocene 2:4749. Graham, M., and A. Roberts 1986 Residentially Constrained Mobility: A Preliminary Investigation of Variability in Settlement Organization. Haliska'a: UNM Contributions to Anthropology 5:105-116. 195

PAGE 215

Gravetter, F. J., and L. B. Wallnau 1991 Essentials of Statistics For the Behavioral Sciences. West Publishing St. Paul, Minnesota. Griffin, J. B. 1967 Eastern North American Archaeology: a Summary. Science 156:175-191. Hanson, G. T., R. D. Brooks, and J. W. White 1981 The Human Occupation Along the Steel Creek Floodplain: Results of an Intensive Archaeological Survey For the L Area Reactivation Project, Savannah River Plant, Barnwell County, South Carolina. Research Manuscript Series 134, South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Harris, M. 1979 Cultural Materialism: the Struggle For a Science of Culture. Random House, New York. Hayden, B., N. Franco, and J. Spafford 1996 Evaluating Lithic Strategies and Design Criteria. In Stone Tools: Theoretical Insights into Human Prehistory, edited by George Odell, ed., pp. 9-45. Plenum Press, New York. Henry, D. 0., C. V Haynes, and B. Bradley 1976 Quantitative Variations in Flaked Stone Debitage. Plains Anthropologist 21:57-61. Hofman, J. L. 1986 Eva Projectile Point Breakage at Cave Spring: Pattern Recognition and Interpretative Possibilities. Midcontinental Journal of Archaeology 11:79-95. House, J. H., and D. L. Ballenger 1976 An Archaeological Survey of the Interstate 77 Route in the South Carolina Piedmont. Research Manuscript Series 104, South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Johnson, J. K 1993 North American Biface Production Trajectory Modeling in Historic Perspective Plains Anthropologist 38(1143)151-162. 196

PAGE 216

Johnson, J. K., and S. 0. Brookes 1987 Benton Points, Turkey Tails, and Cache Blades: Middle Archaic Exchange in the Southeast. Paper Presented at the 52nd Annual Meeting of the Society for American Archaeology, Toronto. Kelly, R. L. 1988 The Tree Sides of a Biface. American Antiquity 53:717-734. Kelly, R. L., and L. C. Todd 1988 Coming into the Country: Early Paleoindian Hunting and Mobility. American Antiquity 53:231-244. Kneberg, M. 1956 Some Important Projectile Point Types Found in the Tennessee Area. Tennessee Archaeologist 12(1): 17-28. Koldehoff, B. 1987 The Cahokia Flake Tool Industry: Socioeconomic Implications for Late Prehistory in the Central Mississippi Valley. In The Organization of Core Technology, edited by J. K. Johnson and C. A. Morrow, pp 151-185. Westview Press, Boulder. Krieger, A. 1953 New World Culture History: Anglo-America. In Anthropology Today, edited by A L. Kroeber, pp. 238-264. University of Chicago Press, Chicago. Langley, T. M., and W. M. Marter 1973 The Savannah River Plant Site. Report DP-1323. Savannah River Laboratory, Aiken, South Carolina. Larsen, C. S. 1982 The Anthropology of St. Katherine's Island 3. Prehistoric Human Biological Adaptation. Anthropological Papers of the American Museum of Natural History 57(3): 157-276, American Museum of Natural History, New York. Lindeman, M. W. 1990 Late Paleo-Indian and Early Archaic Utilization of Two Quarry Sites in Allendale County, South Carolina. Master's thesis, Department of Anthropology, University of South Carolina, Columbia. 197

PAGE 217

Macklin, J. H. 1956 Cause of Braiding by a Graded Stream. Geological Society of America Bulletin 67:1717-1718. Mason, R. J. 1962 The Paleo-Indian Tradition in Eastern North America. Current Anthropology 3(3):227-278. Matthews, J. M. 1990 Surficial Geology of the Rae's Creek Archaeological Site on the Savannah River, Augusta, Ga. In Rae's Creek, A Multicomponent Archaeological Site at the Fall Line Along the Savannah River, edited by M. Crook, pp. 176-186. Department of Anthropology, Georgia State Un iversity, Athens. McAvoy, J. M., and L. D. McAvoy 1997 Archaeological Investigations of Site 44SX202, Cactus Hill, Sussex County, Virginia Virginia Department of Historic Resources, Research Report Series No 8 Richmond McMillan, R. B., and W E. Klippel 1981 Post-Glacial Environmental Change and Hunting-gathering Societies of the Southern Prairie Peninsula. Journal of Archaeological Science 8:215-245. Meltzer, D. J., D. K. Grayson, G. Ardila, A. W. Barker, D. F Dincauze, C. V. Haynes, F. Mena, L. Nunez, and D. J. Stanford 1997 On the Pleistocene Antiquity of Monte Verde. American Antiquity 62:659-663. Meltzer, D. J., and B. D. Smith 1986 Paleoindian and Early Archaic Subsistence Strategies in Eastern North America. In Foraging, Collecting, and Harvesting: Archaic Period Subsistence and Settlement in the Eastern Woodlands, edited by S. Neusius, pp. 3-31. Occasional Paper 6, Center for Archaeological Investigations, Southern Illinois University at Carbondale. Morrow, T. A. 1996 Bigger is Better: Comments on Kuhn's Formal Approach to Mobile Tool Kits. American Antiquity 61:581-590. 198

PAGE 218

Morse, D. F. 1973 Dalton Culture in Northeast Arkansas. Florida Anthropologist 26:23-28. 1997 Sloan, a Paleoindian Dalton Cemetery in Arkansas. Smithsonian Institution, Washington, D. C. Nelson, M. C. 1991 The Study of Technological Organization. In Archaeological Method and Theory, vol. 3, edited by M. B. Schiffer, pp. 57-100. University of Arizona Press, Tucson. Newcomer, M. H. 1971 Some Quantitative Experiments in Handaxe Manufacture. World Archaeology 3 :85-93. Odell, G. H., ed. 1996 Stone Tools: Theoretical Insights Into Human Prehistory. Plenum Press, New York. O'Steen, L. D. 1983 Early Archaic Settlement Patterns in the Wallace Reservoir: An Inner Piedmont Perspective. Wallace Reservoir Project Contribution No. 25, Department of Anthropology, University of Georgia, Athens. Patterson, L. W. 1981 A Chert Cobble F l aking Experiment. La Tierra 8(4):29-34. Southern Texas Archaeological Association, San Antonio, Texas 1990 Characteristics of Bifaciai-Reduction Flake-Size Distribution. American Antiquity 55(3): 550-558. Patterson, L. W., and J. B. Sollberger 1978 Replication and Classification of Small Size Lithic Debitage. Plains Anthropologist 23:103-112. Plog, S., and M. Hegmon 1993 The Sample Size-Richness Relation: The Relevance of Research Questions, Sampling Strategies, and Behavioral Variation. American Antiquity 58(3):489-496. Plummer, G. L. 1983 Georgia Rainfall: Precipitation Patterns at 23 Places 1734-1982. Georgia Academy of Science, Athens. 199

PAGE 219

Prentiss, W. C. 1998 The Reliability of a Lithic Debitage Typology: Implications for Archaeological Interpretation. American Antiquity 63( 4): 635-650. Ritchie, W. A. 1932 The Lamoka Lake Site: the Type Station of the Archaic Algonkin Period in New York. Researches and Transactions 7(4). New York State Archaeological Association, New York. Rose, M. 1999 Beyond Clovis: How and When the First Americans Arrived. Archaeology 52(6):80-84. Sassaman, K. E. 1984 Middle and Late Archaic Settlement in the South Carolina Piedmont. Master's thesis, Department of Anthropology, University of South Carolina, Columbia. 1985a The Middle Archaic Period in the Savannah River Valley: Patterns of Adaptive Flexibility. Paper presented at the 42nd Annual Meeting of SEAC, Birmingham, Alabama. 1985b A Preliminary Typological Assessment of MALA Hafted Bifaces from the Pen Point Site, Barnwell County, South Carolina. South Carolina Antiquities 17: 1-17. 1993 Twenty-five Lessons in Twenty-five Years of Middle and Late Archaic Archaeology. South Carolina Antiquities 25( 1-2):30-42. 1994 Production for Exchange in the Mid-Holocene Southeast: A Savannah River Valley Example. Lithic Technology 19(1):42-51. Sassaman, K. E., and D. G. Anderson 1995 Middle and Late Archaeological Records of South Carolina: a Synthesis For Research and Resource Management. Savannah River Archaeological Research Papers 6, South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. Sassaman, K. E., and M. J. Brooks 1990 Cultural Quarries: Strategies For Scavenging and Recycling Lithic Refuse. Paper presented at the annual meeting of SEAC, Mobile, Alabama. 200

PAGE 220

Sassaman, K. E., M. J. Brooks, G. T. Hanson, and D. G. Anderson 1990 Native American Prehistory of the Middle Savannah River Valley. Savannah River Archaeological Research Papers 1. South Carolina Institute of Archaeology and Anthropology. University of South Carolina. Sassaman, K. E., G. T. Hanson, and T Charles 1988 Raw Material Procurement and the Reduction of HunterGatherer Range in the Savannah River Valley. Southeastern Archaeology 7:79-94. Schiffer, M. B. 1972 Archaeological Context and Systemic Context. American Antiquity 37:156-165. 1987 Formation Processes of the Archaeological Record. University of New Mexico Press, Albuquerque. Segovia, A. 1985 Archaeological Geology of the Savannah River Valley and Main Tributaries in the Richard B. Russell Multiple Resource Area. Russell Papers. National Park Service, Archaeological Services Branch, Atlanta. Semlitsch, R. D., and J. W. Gibbons 1991 Guide to Reptiles and Amphibians of the Savannah River Site. University of Georgia Press, Athens. Shennen, S 1990 Quantifying Archaeology. (2nd printing with corrections.) Academic Press, San Diego Shott, M. J. 1985 Technological Organization and Settlement Mobility: An Ethnographic Examination. Journal of Anthropological Research 42:15-51. Smith, B. D. 1986 The Archaeology of the Southeastern United States: from Dalton to De Soto, 10,500-500 BP. In Advances in World Archaeology, vol. 5, edited by F. Wendorf and A. Close, pp. 1-92. Academic Press, Orlando. 201

PAGE 221

Sokal, R. R., and F. J. Rohlf 1981 Biometry: The Principles and Practice of Statistics in Biological Research. 2nd Edition. W. H. Freeman and Company, New York South, S. 1959 A Study of the Prehistory of the Roanoke Reports Basin. Master's thesis. University of North Carolina, Chapel Hill. Steponaitis, V. P. 1986 Prehistoric Archaeology of the Southeastern United States, 1970-1985. Annual Review of Anthropology 15:363-404. Stevenson, M. A. 1991 Beyond the Formation of Hearth-Associated Artifact Assemblages. In Method and Theory for Activity Area Research: an Ethnoarchaeological Approach, edited by S. Kent, pp. 569-300. Columbia University Press, New York. Stoltman, J. B. 1974 Groton Plantation: an Archaeological Study of a South Carolina Locality. Monograph of the Peabody Museum No. 1, Cambridge. Thorne, R. M., B. J. Broyles, and J. K. Johnson 1980 Lithic Procurement and Utilization Trajectories: Archaeological Survey and Excavations, Yellow Creek Nuclear Power Plant site, Tishomingo County, Mississippi, Vol. II. Archaeological Papers No. 1. Center for Archaeological Research, University of Mississippi. Tippitt, V. A., and W. H. Marquardt 1984 Archaeological Investigations at Gregg Shoals, a Deeply Stratified Site on the Savannah River. The Russell Papers, Archaeological Services Branch, National Park Service, Atlanta. Torrence, R. 1983 Time Budgeting and Hunter-Gatherer Technology. In Hunter Gatherer Economy in Prehistory: a European Perspective, edited by G. Bailey, pp. 11-12. Cambridge University Press, Cambridge. 1989 Tools as Optimal Solutions. In Time, Energy, and Stone Tools, edited by R. Torrence, pp. 1-6. Cambridge University Press, Cambridge. 202

PAGE 222

Upchurch, S. B. 1984 Petrology of Selected Lithic Materials from the South Carolina Coastal Plain. In An Archeological Survey of Chert Quarries in Western Allendale County, South Carolina, by A. C. Goodyear and T. C. Charles. Research Manuscript Series 195, South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia. USDOE (United States Department of Energy) 1990 Draft Environmental Impact Statement: Continued Operation K. L. and P Reactors, Savannah River Site, Aiken, South Carolina. Report No. DOE/EIS-0147D, Savannah River Site Operations Office, Aiken, South Carolina. Walthall, John A. 1980 Prehistoric Indians of the Southeast: Archaeology of Alabama and the Middle South. University of Alabama Press, Tuscaloosa. Wandsnider, L. 1.989 Long-term Land Use, Formation Processes, and the Structure of the Archaeological Landscape: A Case Study From Southwestern Wyoming. Ph.D. dissertation, University of New Mexico, Albuquerque. Watts, W. A. 1971 Postglacial and Interglacial History of Southern Georgia and Central Florida. Ecology 52:666-690. 1974 Vegetation Record for the Last 20,000 Years from a Small Marsh on Lookout Mountain, Northwestern Georgia Geological Society of America Bulletin 86:287-291. 1980 Late Quaternary Vegetation History at White Pond on the Inner Coastal Plain of South Carolina. Quaternary Research 13:187-199. Watts, W A., and M. Stuiver 1980 Late Wisconsin Climate of Northern Florida and the Origin of Species Rich Deciduous Forests. Science 210:325-327. White, J. W. 1981 An Integration of Late Archaic Settlement Patterns For the South Carolina Piedmont. Master's thesis. Department of Anthropology, University of South Carolina, Columbia. 203

PAGE 223

Whitehead, D. R. 1965 Palynology and Pleistocene Phytogeography of Unglaciated Eastern North America. In The Quaternary of the United States, edited by W. E. Wright, Jr. and D. G. Frey. Princeton University Press, Princeton. 1973 Late-Wisconsin Vegetational Changes in Unglaciated Eastern North America. Quaternary Research 3:621-631. Willey, G. R., and P. Phillips 1958 Method and Theory in American Archaeology. University of Chicago Press, Chicago. Wilmsen, E. N. 1970 Lithic Analysis and Cultural Inference: a PaleoIndian Case. University of Arizona, Anthropological Papers 16. Wood, W. D., D. T. Elliott, T. P. Rudolph, and D. B. Blanton 1986 Prehistory of the Richard B Russell Reservoir: The Archaic and Woodland Periods of the Upper Savannah River. The Russell Papers, Atlanta Interagency Archaeological Services Division, National Park Service, Atlanta. Wright, H. E., Jr. 1976 The Dynamic Nature of the Holocene Vegetation Problem of Paleoclimatology, Biogeography, and Stratigraphic Nomenclature. Quaternary Research 6:581-596. 204


printinsert_linkshareget_appmore_horiz

Download Options

close
No images are available for this item.
Cite this item close

APA

Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.

MLA

Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.

CHICAGO

Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.

WIKIPEDIA

Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.