INFORMATION TO USERS

This manuscript has been reproduced from the microfihn master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Ifigher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor Ml 48106-1346 USA 313/761-4700 800/521-0600

LITHIC EVIDENCE OF PREHISTORIC ROCKSHELTER USE IN EASTERN KENTUCKY

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree of Doctor of Philosophy in the Graduate

School of The Ohio State University

By

Darlene Applegate, M.A.

The Ohio State University 1 9 9 7

Dissertation Committee:

Dr. William S. Dancey, Advisor Approved by

Dr. Kristen J. Gremillion

Dr. Paul J. Sciulli ^ Advisor

Dr. Richard W. Yerkes Department of Anthropology UMI N um ber: 9 7 3 1 5 8 4

UMI Microform 9731584 Copyright 1997, by UMI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 Copyright by Darlene Applegate 1997 ABSTRACT

Eastern Kentucky's rockshelters gained archaeological prominence in the 1920s

and 1930s because of their characteristically thick ash deposits containing a wide

variety of well-preserved, prehistoric organic remains. Explanations of the role that

rockshelters played in the settlement strategies of prehistoric people in eastern

Kentucky have been related closely to subsistence economy reconstructions. From the

beginning, archaeological studies of the rockshelters placed a relatively low priority on

the analytical potential of inorganic materials like lithics, especially lithic assemblages

dominated by chipped-stone waste material. The present study addresses both issues, shelter use and lithic analysis, by examining the physical and spatial attributes of debitage-dominated lithic assemblages from two eastern Kentucky rockshelters. Cold Oak

Shelter and Rock Bridge Shelter, in order to assess diachronic variability in prehistoric occupational intensity. Gold Oak Shelter is a multicomponent shelter with occupations from the Terminal Archaic, Early Woodland, and Middle Woodland periods. Rock Bridge

Shelter, on the other hand, was occupied during the Late Woodland period only. Indicators of occupational variability include lithic density, tool diversity, use of exotic cherts, heat treatment of cherts, toolrdebitage ratio, flake fragment types, and debitage types.

The results of the lithic analyses indicate that, despite the effects of formation processes that affect the distribution and condition of lithics. Terminal Archaic and Early Woodland occupations at Cold Oak Shelter were relatively more intense than later occupations in terms of duration and frequency of occupation. Comparable ranges of lithic-related

11 activities are associated with Terminal Archaic, Early Woodland, and Late Woodland occupations, although chipped-stone tool manufacture was more intense during the earlier periods and chipped-stone tool maintenance was more common during the Late

Woodland period. Nonlithic remains from the two shelters support these conclusions. The results concur with previous explanations positing that Late Archaic and Early Woodland shelter use was more intense than other occupations in terms of duration and frequency of occupation.

111 Dedicated to my parents, John and Margaret

I V ACKNOWLEDGMENTS

I gratefully acknowledge the support of my adviser, Dr. William Dancey, for his thoughtful comments on theoretical and methodological issues related to the research project. I am thankful for his encouragement and guidance throughout my graduate career, especially completion of this research project.

I am indebted to Dr. Kristen Gremillion for introducing me to the wonders of eastern Kentucky's rockshelters and for the opportunities to work at Rock Bridge and

Cold Oak. I am thankful for her patience, support, and guidance while conducting the lithic analyses and preparing the lithic reports.

I sincerely thank Dr. Paul Sciulli and Dr. Richard Yerkes for their insightful comments on the dissertation proposal and manuscript and for the great classes I took with them at OSU.

Dr. Loren Babcock of the Department of Geology at The Ohio State University provided helpful suggestions regarding geological issues related to the dissertation, for which I am very appreciative.

Dr. Bede Clay, Kentucky State Archaeologist, generously provided me with access to site forms and cultural resource management reports for the counties of the study area. I also wish to thank USDA Forest Service archaeologists Cecil Ison, Johnny

Faulkner, and Tom Fouts for their assistance in chert identification.

Dr. Gremillion's work at Rock Bridge Shelter was supported financially by a

University Seed Grant Award from the Office of Research, The Ohio State University, and financially and technically by the USDA Forest Service. Excavations and analyses conducted by Dr. Gremillion at Cold Oak Shelter were supported by a grant from the

National Geographic Society. VITA

March 18, 1964 ...... Born - Marion, Ohio

1 9 8 6 ...... B.A. Anthropology & Geology, Miami University

1991 ...... M.A. Anthropology, The Ohio State University

1990 - 1994 ...... Graduate Teaching and Research, Associate, The Ohio State University

1993 - 1994 ...... Archaeology Intern, Bureau of Environmental Services Ohio Department of Transportation

PUBLICATIONS

1. D. Applegate, "Lithic analysis at Rock Bridge Shelter (15Wo75), Wolfe County, Eastern Kentucky." In Current Archaeological Research in Kentucky: Volume Four, pp. 32-68. Edited by Sara L. Sanders, Thomas N. Sanders, and Charles Stout. Kentucky Heritage Council, Frankfort, Kentucky (1996).

2. D. Applegate, "Lithic artifacts." In Archaeological and Paleoethnobotanical Investigations at the Cold Oak Shelter, Kentucky, pp. 89-217. Written by Kristen J. Gremillion, Department of Anthropology, The Ohio State University. Report submitted to the National Geographic Society (1995).

3. D. Applegate, "Lithic analysis." In A Phase III Cultural Resources Assessment of Sites #33-Vi-391, #33-Vi-392, #33-Vi-393, »33-Vi-394, and ff33-Vi-395 for the VIN-50-8.07 (PID 5213) Bridge Replacement and Road Realignment Project, Richland Township, Vinton County, Ohio, pages 10-17. Submitted by Ohio Department of Transportation, Bureau of Planning and Environmental Services, Cultural Resources Unit (1994).

4. D. Applegate, "Lithic analysis." In A Phase III Cultural Resources Assessment of Site #33-Li-573 for the LIC-70-27.55 (PID 10479) Road Relocation Project, Bowling Green Township, Licking County, Ohio, pages 9-15. Submitted by Ohio Department of Transportation, Bureau of Planning and Environmental Services, Cultural Resources Unit (1993).

V I 5. D. Applegate, "Lithic artifacts." in Archaeological Investigations at the Rock Bridge Shelter (15Wo75), Wolfe County, Kentucky, pages 54-133. Written by Kristen J. Gremillion, Department of Anthropology, The Ohio State University. Report submitted to Cecil Ison, Field Archaeologist, Daniel Boone National Forest, Stanton, Kentucky ( 1 9 9 3 ) .

FIELDS OF STUDY

Major Field: Anthropology

V I I TABLE OF CONTENTS

Abstract ...... ii

Dedication ...... iv

Acknowledgments ...... v

Vita ...... vi

List of Figures ...... xi

List of Tables ...... xiii

Chapters:

1. Research problem ...... 1

2. Environmental and cultural milieu of the study area ...... 11

2.1. Environmental background ...... 11 2.2...... Culture history ...... 16 2.2.1. Paleoindian period ...... 18 2.2.2. Archaic period ...... 21 2.2.3. Woodland period ...... 29 2.2.4. Mississippian period ...... 35 2.2.5. Historic period ...... 39 2.2.6. Summary ...... 40 2.3. Previous archaeological research ...... 41 2.3.1. Historical context ...... 42 2.3.2. Previous research by county ...... 45 2.3.3. Explanations of prehistoric shelter use ...... 58

3. Cold Oak and Rock Bridge shelters: Geology and archaeological stratigraphy ...... 67

3.1. The geology of rockshelters ...... 63 3.1.1. Rockshelter development ...... 68 3.1.2. Rockshelter deposition ...... 70 3.2. Cold Oak Shelter ...... 77 3.3. Rock Bridge Shelter ...... 88 3.4. Summary ...... 94

VIII 4 . Cold Oak and Rock Bridge lithic assem blages ...... 98

4.1. Classification ...... 99 4.2. Raw material identification ...... 106 4.3. Lithic assem blage descriptions ...... 107 4.3.1. Cold Oak Shelter lithic assemblages ...... 108 4 .3 .2 . Rock Bridge Shelter lithic assem blage ...... 124 4.4. Summary ...... 132

5. Formation processes and the integrity of the lithic assem blages ...... 135

5.1. Rockshelters and formation processes ...... 138 5.2. Formation processes at Cold Oak and Rock Bridge shelters ...... 143 5.2.1. Processes affecting substrate properties ...... 143 5.2.2. Processes affecting vertical distribution of artifacts ...... 144 5.2.3. Processes affecting lateral distribution of artifacts ...... 155 5.2.4. Processes affecting artifact condition: Tram pling 160 5.3. Summary ...... 166

6. Analysis of lithic production s y s te m ...... 169

6.1. Thermal alteration of chert ...... 172 6.1.1. Methods ...... 172 6.1.2. Results ...... 174 6.1.3. Discussion ...... 178 6.2. Raw material utilization ...... 179 6.2.1. Raw material diversity ...... 179 6.2.2. Raw material selection ...... 183 6.2.3. Exotic and local cherts ...... 185 6.2.4. Discussion: Patterns in raw material use ...... 187 6.3. Tool manufacture and maintenance ...... 191 6.3 .1 . Intensity of lithic production ...... 191 6.3.2. Reduction stage ...... 192 6.3.3. Reduction techniques ...... 202 6.3.4. Summary ...... 202

7. Synthesis and extension: Diachronic patterns of rockshelter use 205

7.1. Indicators of occupational intensity ...... 206

I X 7. Synthesis and extension: Diachronic patterns of rockshelter use (continued)

7.2. Lithic assemblages and occupational variability ...... 213 7.2.1. Duration of occupation ...... 213 7.2.2. Frequency of occupation ...... 216 7.2.3. Range of lithic-related activities ...... 216 7.2.4. Group size ...... 222 7.2.5. Discussion ...... 222 7.3. Other artifactual remains ...... 224 7.3.1. Cold Oak Shelter ...... 224 7.3.2. Rock Bridge Shelter ...... 230 7.4. Comparison with select shelters ...... 233

8. Summary and conclusions ...... 239

List of References ...... 254

Appendix A: Chert Descriptions ...... 271

Appendix B: Figures and Tables ...... 278 LIST OF FIGURES

Etanr.? Page

1 Major drainages of eastern Kentucky and the Upper Kentucky and Licking River Management Area ...... 279

2 Map of the ten-county study area ...... 280

3 Physiographic provinces of the eastern United States, illustrating the lateral extent of the Appalachian Plateau Province ...... 281

4 Physiographic diagram of Kentucky, illustrating the Cumberland Plateau, Pottsville Escarpment, and Knobs ...... 282

5 Bedrock geology of Kentucky ...... 283

6 Physiographic provinces of Kentucky ...... 284

7 Cultural chronology for Kentucky ...... 285

8 Parts of a typical rockshelter ...... 286

9 Stages of rockshelter development ...... 287

1 0 Map showing the locations of Rock Bridge Shelter (15Wo75) and Cold Oak Shelter (ISLeSO) ...... 288

1 1 Plan view of Cold Oak Shelter showing the locations of the 1984 and 1994 excavation trenches ...... 289

1 2 Stratigraphie profile of the 1994 excavation trench at Cold Oak Shelter ...... 290

1 3 Plan view of Rock Bridge Shelter showing 1992 excavation areas ...... 291

1 4 Stratigraphie profile of the main (southern) excavation block at Rock Bridge Shelter ...... 292

X I Figure

1 5 Lateral distributions of chipped-stone tools from Rock Bridge Shelter ...... 293

1 6 Distribution of flakes by stratum (ash:middenzsubsoil), main excavation block, Rock Bridge Shelter ...... 294

1 7 Average weight (grams) of flakes by unit, main excavation block. Rock Bridge Shelter ...... 295

1 8 Percentages of flakes more than one gram in weight, main excavation block. Rock Bridge Shelter ...... 296

1 9 Percentage of flakes greater than 1.5 cm in diameter, main excavation block. Rock Bridge Shelter ...... 297

2 0 Relationship between late-stage debitage and debitage:tool ratio with respect to tool manufacture, raw material conservation, exportation, rejection, and discard ...... 298

XII LIST OF TABLES

JabJ£ Rage

1 Cultural affiliations and functional designations of features for Cold Oak and Rock Bridge shelters ...... 299

2 Lithic artifacts by feature, Cold Oak Shelter ...... 302

3 Lithic artifacts by feature. Rock Bridge Shelter ...... 305

4 Composition of the Cold Oak and Rock Bridge shelter lithic assem blages 306

5 Raw materials for the Cold Oak Shelter lithic assem blages ...... 307

6 Horizontal distribution of lithics by excavation unit, Cold Oak Shelter ...... 308

7 Vertical distribution of lithics by cultural zone. Cold Oak Shelter ...... 309

8 Vertical distribution of lithics by cultural period Cold Oak Shelter ...... 310

9 Chipped-stone artifacts for the Cold Oak and Rock Bridge shelter lithic assem blages ...... 311

1 0 Horizontal distribution of chipped-stone tools and tool fragments by excavation unit, Cold Oak Shelter ...... 312

1 1 Vertical distribution of chipped-stone tools and tool fragments by cultural zone. Cold Oak Shelter ...... 313

1 2 Vertical distribution of chipped-stone tools and tool fragments by cultural period. Cold Oak Shelter ...... 314

1 3 Chert distribution for chipped-stone tools and tool fragments. Cold Oak Shelter ...... 315

1 4 Metric attributes of projectile points, perforator, and point fragments from Cold Oak Shelter and Rock Bridge Shelter ...... 316

XIII Table Page

1 5 Raw material distribution for Cold Oak Shelter flake sample, listed by flake fragment type ...... 317

1 6 Distribution of flake fragment types for the Cold Oak Shelter flake sample ...... 318

1 7 Horizontal distribution of lithic artifacts by excavation block and by method of recovery, Rock Bridge Shelter ...... 319

1 8 Vertical distribution of lithic artifacts by excavation block, Rock Bridge Shelter ...... 320

1 9 Raw material distribution by lithic category. Rock Bridge Shelter ...... 321

2 0 Distribution of chipped-stone tools and tool fragments. Rock Bridge Shelter ...... 322

2 1 Distribution of chipped-stone flakes and debris, Rock Bridge Shelter ...... 323

2 2 Raw material distribution for Rock Bridge Shelter flake sample, listed by flake type ...... 324

2 3 Flake fragment and debris percentages for Cold Oak, Rock Bridge, and experimental assem blages ...... 325

2 4 Summary of formation processes findings. Cold Oak and Rock Bridge shelters ...... 327

2 5 Collins' (1975) stages of chipped-stone tool manufacture and seven product groups ...... 328

2 6 Local and exotic chert distribution by occupation period, Cold Oak and Rock Bridge shelters ...... 331

2 7 Predicted lithic and nonlithic residues for activities or tasks (after Ledbetter and O'Steen 1991) and observed residues from from Cold Oak and Rock Bridge shelter assemblages ...... 332

2 8 Lithic indicators of occupational intensity for Cold Oak and Rock Bridge shelters ...... 336

2 9 Summary of lithic indicators of occupational intensity and statistical significance ...... 337

3 0 Comparison of stratigraphy, features, and artif actual remains for selected shelters in the study area ...... 338

XIV CHAPTER 1

RESEARCH PROBLEM

Archaeologically, the Cumberland Plateau area of eastern Kentucky is well known for its rockshelter resources. The context of early archaeological investigations in the area involved questions about the emergence of agriculture. Since the 1930s, researchers have recovered from eastern Kentucky's rockshelters some of the earliest evidence of cultigens in the midcontinent. The material supports Linton's (1924) contention that the Eastern Woodlands, a culture area bounded by the Mississippi River,

Gulf Coast, Atlantic Coast and Great Lakes, was a hearth of indigenous plant domestication. The rockshelters and their well-preserved organic remains gained the archaeological community's attention following the work of Funkhouser and Webb

(1929, 1930) and Webb and Funkhouser (1936) in Lee, Wolfe, Powell, and Menifee counties and the publication of Jones' (1936) landmark study of vegetal remains from

Newt Kash Hollow Shelter in Menifee County.

Prehistoric organic materials, including moccasins and overshoes, bags, cordage, leather, cradleboard, wooden tools, grass beds, grass-lined pits with food remains, and uncarbonized plant remains, brought eastern Kentucky rockshelters into the archaeological spotlight. Recovery of these artifacts was an important development in early archaeology because, due to the relative rarity of such remains, we knew little about prehistoric vegetal diets, clothing, basketry, and textiles. Although well- preserved organics had been discovered in some of Kentucky's cave systems, including

1 Mammoth Cave and Salts Gave, the cave remains did not compare in volume to those in the rockshelters of the eastern part of the state.

Additionally, discoveries of the rich organic resources in the shelters determined, to some extent, the course of subsequent archaeological research in eastern

Kentucky. Research projects and field methods designed to maximize the recovery of plant materials, especially small seeds, are becoming more common. Questions about the timing of plant domestication, the sequence in which indigenous and tropical cultigens were adopted into the diet, and the replacement of wild foods with crops remain under investigation, although the picture is becoming clearer. One might argue that the study of paleodiet through examination of human fecal remains has been spurred by the discoveries at eastern Kentucky's rockshelters. Debates still rage about the role of rockshelters in the agricultural cycle of the prehistoric groups living in the area. Were the shelters used to store the crops grown on narrow flood plains, or did indigenous people live in the shelters as they cultivated crops on the surrounding slopes?

What roles have lithic artifacts played in these sixty years of research in eastern

Kentucky's rockshelters? Few archaeologists would argue that lithics have taken a back seat to the organics. For the most part, analyses of rockshelter lithic assemblages have focused on descriptive matters such as identification and quantification of functional tool types, questions of raw material identification and chert use, and the use of projectile points for chronological control. In the early reports of Funkhouser and Webb (1929,

1930), for instance, analysis of lithic artifacts involved only classification of complete or nearly complete tool specimens into functional categories based on ethnographic analogy. Of the 36 reports consulted during this dissertation project, 80% limit the lithic analysis to issues of diagnostic projectile point descriptions, inventories of tool types, and summary statements about lithic waste. In some cases, lithic waste or debitage is classified into categories and counted. Common categories of debitage are flakes and debris, with the former often subdivided into types related to the technology of

chipped-stone tool manufacture.

This disregard for the research potential of lithic artifacts certainly has not been

intentional. During the early decades of work at eastern Kentucky rockshelters,

researchers were concerned with recovering and analyzing the well-preserved organics that are so rare at other sites, like the shell mounds and earthen mounds being

investigated elsewhere in Kentucky and the Ohio Valley. These organic remains were

unprecedented in Eastern Woodlands archaeology and rivaled some of the spectacular finds made in caves and shelters in the arid Southwest. In addition, because of the huge volumes of sediment moved during typical excavations in the 1930s and 1940s,

investigators probably had to be selective about the types of lithic artifacts (e.g., tools versus waste) they recovered.

As research foci shifted in the 1960s with the rapid growth of culture resource management work in eastern Kentucky, archaeologists found themselves under considerable time and resource constraints. Charged with surveying huge tracts of land for evidence of prehistoric occupations that might be impacted by timber sales, pipeline projects, and reservoir flooding, archaeologists were not afforded the opportunity to extensively excavate shelters as had earlier researchers. They also did not have time to conduct detailed lab analyses of lithics and other artifactual remains. Their vital contributions were documentation of hundreds of prehistoric occupations in the shelters, flood plains and ridge tops and determination of site distribution patterns in eastern

Kentucky.

In addition to the shadow cast by inorganics and the constraints experienced by surveyors, another possible reason for the relative lack of attention to lithics is the disturbed nature of most rockshelter deposits. Because of the protection rockshelters

afford, they have been m agnets of human use for millennia. This, coupled with the confined nature of shelters, has resulted in the disturbance of archaeological materials

in shelters. Humans have impacted the location and condition of artifacts in most

rockshelters to varying degrees. For instance, in a ten-county area of eastern Kentucky,

about 55% of nearly 500 rockshelters show readily apparent evidence of disturbance;

26.5% are not extensively disturbed and the nature of disturbance is not indicated for

the remaining 18.5% (Applegate 1997).

The integrity of prehistoric archaeological deposits in the shelters has been

compromised by a number of forces. Native American groups using the shelters impacted

earlier deposits by excavating pits or hearths, building fires, and walking on remains.

According to Wyss and Wyss (1977), Euro-Americans have contributed significantly to

the disturbance of rockshelter remains. Perhaps the earliest source of disturbance in

eastern Kentucky was treasure hunters. In the nineteenth and twentieth centuries, and

continuing today, locals in search of the legendary silver mines of John Swift have dug in

rockshelter deposits in their quest for riches. Nitrate miners working in the 1800s to

early 1900s produced large amounts of debris that displaced, damaged or covered archaeological materials. Lumber industry workers of the 1880-1920 period often

used the shelters as campsites or stables for their oxen. Early twentieth century

moonshiners housed illegal stills in many of the secluded shelters. For decades, farmers have removed the loose, ashy sediments of shelters for use as fertilizers and have used some shelters as livestock quarters (Funkhouser and Webb 1929, 1930; Webb and

Funkhouser 1932, 1936). Visitors to recreational areas have also impacted rockshelter deposits, but the primary threat to the integrity of archaeological deposits in eastern

Kentucky's shelters today is illegal digging by looters. Rockshelters "are a favored target

of relic collectors who in their search for artifacts can destroy a site in a single day"

(Ison 1988:205). Looters are a special problem because they are able to locate the most

concentrated cultural deposits and the burials at shelters. The disturbance of lithic artifacts is a problem because, unlike organic remains that can be radiocarbon dated, undiagnostic chert artifacts like waste material and some broken tools cannot be directly dated at this time. So if these lithics are found out of context and cannot be dated by association with other artifacts, the material is of limited use for interpretive endeavors. In some cases, a pessimistic view of the research potential of materials from more intensely disturbed contexts emerged.

One thing that previous rockshelter research has shown us is that the shelters abound in lithic artifacts. For example, Railey (1991) recovered an average of 5,534 flakes/m ^ from the Conley-Greene Rockshelter in Elliott County. Chipped-stone w aste material is particularly common in most shelter lithic collections. For example, the collection obtained by Wyss and Wyss (1977) from dozens of sites in the Red River

Gorge consists of 93% chipped-stone debitage and 7% chipped-stone tools. Powell

County's Abner Shelter yielded a 250-piece collection of lithics, 66% of which are chipped-stone debitage, 32% are chipped-stone tools, and 2% are ground-stone artifacts (Meadows 1977). About 97% of the chipped-stone artifacts from Ratliff

Shelter in Lee County is debitage: only 3% is chipped-stone and ground-stone tools

(Kluth 1992). At Cloudsplitter Shelter in Wolfe County, chipped-stone debitage makes up 99% of the lithic assemblage, with the remaining 1% being chipped-stone and ground-stone tools (Cowan et al. 1981).

Archaeologists have documented hundreds of prehistoric occupations in the eastern Kentucky rockshelters since the early twentieth century. Despite the extensive disturbance these locations have experienced at the hands of treasure hunters, miners, loggers, farmers and relic collectors, the wealth of organic and inorganic material remains recovered from the rockshelters has contributed significantly to our understanding of prehistoric subsistence and technology. Nevertheless, there are serious methodological hurdles to clear before the full research potential of rockshelters can be realized. In particular, while used to estimate period of occupation through projectile

point typologies, lithics are underutilized in rockshelter analyses. This is especially

true of lithic collections dominated by debitage remains and relatively deficient in

finished lithic objects.

Previous work has demonstrated diachronic changes in rockshelter plant

assemblages, but because chipped-stone lithic technologies are not systemically related

to plant cultivation (except perhaps chipped-stone hoes), one cannot assume that

changes in plant use were accompanied by changes in chipped-stone lithics. Questions of

rockshelter use cannot be answered with organic remains only. This dissertation is an

attempt to make up for this deficiency by presenting a detailed analysis of the lithic

assemblages from two rockshelters in eastern Kentucky, Cold Oak Shelter and Rock

Bridge Shelter. The study is important because it demonstrates the research potential of

lithic collections composed primarily of debitage, an artifact class that, with few

exceptions, has not been used to its fullest potential for hypothesis testing in studies of

eastern Kentucky prehistory.

The current research is guided by three research questions that seek to

demonstrate what can be done with shelter lithic assemblages dominated by chipped-

stone debitage. First, although most deposits in eastern Kentucky rockshelters are

disturbed to some extent, we can still derive useful information from the lithic

assemblages contained therein. This proposition is an extension of what several

archaeologists have argued for some time. Ison (1988:205) asserted over a decade ago that "although the cultural deposits within many of [the] shelters have been extensively

disturbed, substantial amounts of data often remain intact." Similar statements have

been made about non-shelter sites. Speaking of Zilpo Cemetery, a ridge top site in Bath

County, Knudsen (1988:231) stated that "surface indications do not always adequately

reflect the contents, depth, and integrity of [a] site's deposits." Based on his work at Wickliffe Mounds in Ballard County in western Kentucky, Wesler (1991:131) argued that "mixed deposits may represent different types of mixing, and offer different challenges and opportunities for analysis."

The first proposition is based on Schiffer's (1987:11) contention that there are knowable, detectable laws that govern the operation of formation processes, or extraneous processes responsible for artifact modification: "it has been shown that formation processes . . . exhibit regularities that can be expressed as (usually statistical) laws." Experimental studies that replicate the effects of various processes on lithic artifacts, which are described in Chapter 5, provide expectations that are evaluated using lithic assem blages from Cold Oak and Rock Bridge shelters. Lithics, therefore, are used to identify agents of disturbance in the shelters and to assess the impact of the agents on the integrity of archaeological deposits.

The second proposition is that lithic assemblages can be used to test hypotheses about prehistoric rockshelter use in eastern Kentucky. In other words, the lithic remains left by shelter occupants can be used to determine some of the activities that took place there in prehistory. Ericson's (1984) model of lithic production systems provides a context for this type of study, and middle-range approaches like Collins'

(1975) model of chipped-stone lithic reduction and experimental studies (e.g., Ingbar, et al. 1989, Johnson 1975, Johnson 1989, Prentiss and Romanski 1989, Sullivan and

Rosen 1985, Tomka 1989) serve as the means to link artifactual remains to the behaviors that produced them. As described in Chapter 6, one can use these approaches to learn about such cultural activities as raw material use and tool manufacture. These approaches are used because they focus on lithic debitage, which is the predominant form of lithics in the Cold Oak and Rock Bridge assemblages, rather than on lithic tools as other approaches do. Mass analysis of the lithic debitage is not used because the sample sizes of the lithic assem blages from Cold Oak and Rock Bridge shelters are not unmanageably large and because attributes such as raw material types per debitage class are considered in assessing occupational variability at the shelters.

The final proposition is that one of the most compelling and yet unresolved issues about eastern Kentucky rockshelters is diachronic differences in how humans used them in prehistory. How did shelter occupations vary over time? As demonstrated in the next chapter, archaeologists continue to debate the role that rockshelters played in prehistoric settlement systems in eastern Kentucky. Resolution of this issue requires fine-grained temporal control of occupations, consideration of site contemporaneity, and determination of the occupational histories of the shelters. The present research, which emphasizes detailed analyses of lithic assem blages from two shelters, contributes to the latter.

In this study, rockshelter occupations are described in a relative manner according to occupational variability. There are four dimensions of occupational variability: (1) the duration of each occupation, (2) the frequency with which occupants revisited the location, (3) the range of activities that occurred at the location, and (4) the number of people using the location at a given time. Lithic indicators of occupational variability include artifact densities, tool diversity, thermal alteration, reduction strategies, raw material use, and stratigraphie context. Relatively intense occupations (i.e., long occupational duration, frequent occupation, and/or use by large groups of people engaging in a wide range of activities) are indicated by lithic assemblages with relatively more remains, a high diversity of tool types, more evidence of heat treatment of cherts, local cherts and unmodified exotic cherts, and evidence of a variety of reduction activities found within relatively thick and temporally indistinct strata. On the other hand, assemblages with relatively few remains, a low diversity of tool types, little evidence of heat treatment of cherts, local cherts and modified exotic cherts, and evidence of limited reduction activities found within relatively thin and temporally distinct strata represent relatively less intense occupations (i.e., short occupational duration, infrequent occupation, and/or use by small groups of people engaging in a limited range of activities). These hypotheses, which are detailed in

Chapter 7, may be independently evaluated using the non-lithic artifactual remains from Cold Oak and Rock Bridge shelters.

To recapitulate, previous rockshelter research in eastern Kentucky has focused on archaeobotanical remains and the questions of plant domestication and subsistence change. For the most part, lithics have played a limited role in the investigation of these and other questions; lithic analyses have focused on functional categories of finished tools and on the use of diagnostics, especially projectile points, to estimate periods of shelter occupation. For a number of reasons, researchers have not fully utilized the abundance of chipped-stone debitage that characterizes most shelter assemblages. The present study seeks to demonstrate the research potential of shelter lithic assemblages dominated by waste material. So what can one learn by studying such assemblages? Chipped-stone debitage assemblages can be used to evaluate the integrity of rockshelter deposits, to identify cultural activities that occurred at the shelters, and to document diachronic patterns of shelter use.

This research contributes to the resolution of other issues in eastern Kentucky prehistory. One issue to which this work relates is subsistence change. If we have a better understanding of how people used the shelters, we may be better able to determine why the early cultigens are there. We can also determine if the transition to horticulture was accompanied by changes in lithic systems. Another issue to which this work relates is the utility of traditional temporal divisions of Eastern Woodlands prehistory. In terms of rockshelter use, do the lithic samples from different temporal occupations indicate differences in lifestyles or not? Is the distinction between Late Archaic and Early Woodland periods, for example, a valid or useful one in terms of how

humans were using rockshelters?

To put the work reported here in context, the next chapter presents a description

of the study area. Environmental characteristics are outlined, a brief cultural history of

the area is presented, and previous rockshelter research is summarized. In Chapter 3,

the geological and archaeological milieu of Cold Oak and Rock Bridge shelters is covered.

Chapter 4 describes the lithic assemblages of the two shelters. In Chapter 5, Schiffer's

(1987) behavioral model of formation processes is outlined, lithic indicators of a

number of processes are described, and the integrity of the Cold Oak and Rock Bridge

assemblages is evaluated. Ericson's (1984) model of lithic production systems and

Collins' (1975) model of lithic reduction are described and applied to the shelter

assemblages in Chapter 6. Chapter 7 addresses the issue of diachronic patterns of

prehistoric rockshelter use at Cold Oak and Rock Bridge and considers if the patterns are generally applicable to other shelters in the area. The final chapter summarizes the

research and presents the conclusions.

1 0 CHAPTER 2

ENVIRONMENTAL AND CULTURAL MILIEU OF THE STUDY AREA

This research project uses lithic assem blages from two eastern Kentucky rockshelters, Cold Oak Shelter in Lee County and Rock Bridge Shelter in Wolfe County, to evaluate hypotheses related to shelter formation processes and prehistoric shelter use.

These counties are in “the heart of rockshelter country" and fall within the Upper

Kentucky and Licking River Presen/ation M anagement Area of eastern Kentucky (Figure

1). The following summary of the physical and cultural milieu of this area focuses on previous research in Wolfe and Lee counties as well as the eight contiguous counties of

Breathitt, Estill, Jackson, Magoffin, Menifee, Morgan, Owsley, and Powell (Figure 2).

This ten-county are is referred to as the "study area" throughout the dissertation.

In this chapter, the environment of the study area is described in terms of physiography, bedrock, hydrology, vegetation, and fauna. The next section presents a general culture history of the study area. Finally, previous archaeological research conducted in Wolfe, Lee, and surrounding counties is summarized.

2.1. ENVIRONMENTAL BACKGROUND

Cold Oak and Rock Bridge shelters are located in the Cumberland Plateau section of the Appalachian Plateau physiographic province (Figure 3). In general, the Appal­

1 1 achian Plateau province is a shallow synclinorium with relatively soft shales in the

center and more resistant rocks on the eastern and western borders (Fenneman 1938;

Hunt 1974). The plateau is "an elevated tract of nearly horizontal or gently folded

strata" (Hunt 1974:256). Resistant sandstones, which are interbedded with the shales,

crop out sporadically within the plateau, as do lesser amounts of limestone and coal. The

plateau extends from New England to the Gulf Plain states and from the Valley and Ridge

province to just east of the Mississippi River. Geologists divide the Appalachian Plateau

province into seven sections based on variations in the degree of dissection, the type of

bedrock, altitude, and the effects of glaciation. The west-central of these sections is the

Cumberland Plateau (Figure 4).

In the north-south direction, the Cumberland Plateau extends from the Kentucky

River drainage basin in Kentucky south to the Gulf Coastal Plain province. The Cumber­

land Front marks the eastern edge of the Cumberland Plateau. The western edge of the

Cumberland Plateau, which corresponds to the western edge of the Appalachian Plateau province as well, is conspicuously marked by the Pottsville Escarpment (Figure 4). The cliff marks the contact between differentially weathering Mississippian (to the west) and Pennsylvanian (to the east) bedrock (Figure 5). In Kentucky, the west-facing escarpment is composed of sandstones and conglomerates of the Rockcastle series of the

Pottsville Group. As Fenneman (1938:334) noted, "the strong conglomerates and sandstones at the base of the Pottsville underlie and support the margin of the plateau" in this area. Speaking of the western escarpment, Fenneman (1938:335) explained that

"its course is straight in general but ragged in detail, made so by younger valleys."

As a whole, the Cumberland Plateau is a dissected peneplain at a submature level of development, due in part to the thickness and strength of certain sandstone beds. In the

Kentucky River basin, however, the plateau is more rugged than usual. This maturely dissected portion of the Cumberland Plateau is, according to Fenneman (1938:334),

1 2 "some of the roughest land in the United States." The presence of less resistant sedimentary rocks in the study area accelerated dissection. "All divides are sharp, and the general level of the hilltops is being lowered" (Fenneman 1938:341). Of the

Cumberland Plateau, Fenneman (1938:337) explained

in the Kentucky portion, where the [Appalachian Plateau] province broadens, immaturity is seen only in a belt some 25 miles wide along the western edge ... The surface in this 25-mile strip is plainly a dip slope, apparently stripped since the last peneplaining. Its altitude of 1,300 to 1,500 feet at the province boundary [the Pottsville Escarpment] declines southeastward to 1,100 or 1,200 feet in east-central Kentucky, which is only 100 or 200 feet above the lower province to the west [the Bluegrass area of the Lexington Peneplain]. From this low level the surface rises eastward over formations of shale and sandstone ... The basin of the Upper Kentucky River is sharply and maturely dissected, with a relief of 500 to 1,000 feet.

The bedrock of the Cumberland Plateau in the study area is interbedded Paleozoic sedimentary rocks that weather differentially, with sandstones creating topographically high ridges (Figure 5). Resistant Corbin Sandstone caps the narrow ridges and forms the steep cliff line; the thickness of this member is proportional to the number of rockshelters and rockshelter development. The "Cliff Section" is the narrow strip of conglomeratic sandstones ranging 8 to 40 km wide on the westem edge of the plateau.

"This area is characterized by high precipitous escarpments ringing narrow, dendritic stream valleys with discontinuous floodplains" (Ison 1988:2). Strata dip slightly to the east off the eastern flank of the Cincinnati Arch. Narrow valleys with steep slopes and poorly developed flood plains, narrow ridge tops, entrenched streams, cliff lines pitted by rockshelters, and rugged, high-relief topographies characterize the Cumberland

Plateau in the study area.

McFarlan (1943) identified two sections of the Cumberland Plateau. The study area falls into one of these sections, the "Ridge Top (and Limestone Valley) Settlements."

McFarlan (1943:177) explained that this section corresponds to

1 3 the region of outcrop of the Rockcastle conglomerate, the region of the westward- facing Pottsville cuesta with its deeply dissected eastward sloping strip plain. Streams have trenched themselves in steep-sided and precipitous-walled valleys, capped with cliffs often several hundred feet high. It is a region of rockshelters and natural bridges ... . In the westem portion near the Escarpment ■ divides are narrow but they involve considerable summit area eastward. Valleys in the western portion tend to be of fair width with considerable bottom land. Here streams have widened their valleys in the outcropping Ohio and Waverly shales. To the east as the Mississippian limestones and Lee conglomerate come down to stream level, the bottom land disappears in narrow gorge-like valleys, to widen again eastward as these beds pass below drainage.

The study area encompasses two physiographic regions of the state: The Eastern

Mountains and Coalfields region and The Knobs region (Figure 6). The former is divided into three parts, the boundaries of which trend in a northeast-southwest direction. To the east is the Mountain and Creek Bottom Area. The Plateau Area is in the center of the

Eastern Mountains and Coalfields. The westernmost area, the Escarpment Area, marks the boundary of the Cumberland Plateau.

The western counties in the study area are in The Knobs region on the western edge of the Cumberland Plateau (Figure 6). Forming a "narrow belt of hilly land around the Bluegrass region," the knobs are steep-sided remnants of the Pottsville Escarpment

(Weinland and Sanders 1977:4). "Scattered buttes and promontories" are common in the

Knob Belt (Hunt 1974:256). Knobs with resistant cap rocks are flat-topped while others are conical. Knob elevations vary: in Powell County, for example, the hills range from 600 to 1,300 feet (amsi) in elevation. Rockshelters and overhangs are common.

The bedrock is Silurian, Devonian and Mississippian sedimentary rocks, especially the relatively non-resistant Ohio Shale and Borden Formation. Compared to the Cumberland

Plateau to the east, major stream valleys in The Knobs area are broad and flat (McFarlan

1 9 4 3 ).

McFarlan (1943) divided The Knobs region into three parts. The portion falling in the study area is "an eastern belt in front of the Pottsville Escarpm ent with its

1 4 Rockcastle conglomerate cap, flat-topped while the cap remains, conical after its removal" (McFarlan 1943:197). The Lexington peneplain lies to the west of The Knobs.

The main drainage systems of the study area are the Kentucky River and the Red

River (Figures 1 and 10). The Kentucky River is divided into middle, north and south forks, and tributaries in the study area include Red River, Devil Creek, Crystal Creek,

Contrary Creek, Hell Creek, Walker Creek, Bear Creek, Red-Eye Creek, Big Sinking

Creek, Little Sinking Creek, Cave Fork Creek, Little Ash Creek, Bald Creek, Sore-Heel

Creek, and Billings Fork. Tributaries of the Red River in the study area include Chimney

Top Creek, Camp Swift Creek, Stillwater Creek, Gilmore Creek, and Rose Fork.

Florally, the Cumberland Plateau area is diverse owing to differences in slope and slope direction, soil acidity, soil texture, moisture, and disturbance. Virgin forest remains in only the most remote parts of the study area. Ridge top vegetation is xeric, being characterized by oak-pine or dry pine forests on the margins and oak-hickory or oak-chestnut forests on the broader expanses. Below the sandstone cliffs, however, the forest is mixed mesophytic with beech, yellow poplar, tulip poplar, yellow buckeye, sugar maple, basswood, yellow birch, hemlock, oak, chestnut, and hickory. Here the understory often contains rhododendron, dogwood, maple, redbud, sounwood, iron wood, hornbeam, holly, sumac, hazel, sassafras, woodbine, poison ivy, fern, mountain laurel, and magnolia. Eighteen nut-bearing trees, such as black walnut, have been documented in the study area. Other economically important plants include fleshy fruits, tubers, and weedy annuals (Cowan 1979; Gremillion 1993; Weinland and Sanders 1977).

Animal life is equally diverse. In the Red River area alone, researchers have identified 59 species of fish, 31 species of amphibians, 30 species of reptiles, 105 species of resident birds, and 36 species of mammals (Cowan 1979). Economically important species in the past included white-tailed deer, turkey, opossum, and raccoon.

1 5 Shellfish and many fish, although available, probably did not contribute significantly to diets of prehistoric humans in the study area.

The diversity of wild plants and animals in the study area probably was appealing to prehistoric hunting and gathering populations. Other aspects of the environment that likely facilitated human occupation include the availability of chert raw material in sedim entary outcrops and alluvial deposits, the proximity to water resources, and the protection afforded by rockshelters. The rugged nature of the terrain, on the other hand, probably impeded to some extent the movement of people across the eastern Kentucky landscape. Steep cliff lines, the scarcity of gaps across divides, and the narrow ridge tops of the Cumberland Plateau in the study area presumably limited travel from one valley to another. It seems likely that people moved in sinuous pathways along the narrow stream valleys. Transportation of supplies or raw materials was possibly a challenge in this environment because many of the streams are not navigable by canoe, and foot travel along the hill slopes can be arduous. Yet, despite the challenges posed to inhabitants by

"some of the roughest land in the United States" (Fenneman 1938:334), humans have occupied the flood plains, rockshelters, and ridge tops of eastern Kentucky for millennia.

An overview of the culture history of the study area is presented next.

2.2. CULTURE HISTORY

In this limited space it is impossible to present a thorough account of 10,000 years of eastern Kentucky culture history. The following overview, then, is necessarily general. For more detailed treatments of the culture history of the study area, see Lewis

(1996), Niquette and Henderson (1984), and Schwartz (1967). Respected summaries of Eastern Woodlands culture history include Brown (1977), Caldwell (1962), Dragoo

1 6 (1976), Ford and Willey (1941), Griffin (1946, 1967), Steponaitis (1986), and

Willey and Phillips (1958).

Archaeologists traditionally divide Eastern Woodlands prehistory into several time periods (Figure 7). The most widely used chronological scheme is based on the

Midwestern Taxonomic System (McKern 1939), which originally delineated two periods. Woodland and Mississippian\ According to the initial formulation of the culture history units, the periods were equated with cultural traditions^ or stages of cultural development reflecting different ways of life. Subsequent modifications have involved the addition of the Archaic and Paleoindian periods, the addition of chronometric dates for the period boundaries, expansion of trait lists, refinement of intraperiod chronologies, and descriptions of local variations. In addition, many archaeologists now characterize the periods as temporal units rather than cultural stages. As Yerkes (1988:345-346) noted, "there has been a movement away from the view that Woodland and Mississippian complexes are examples of developmental stages." Alternate chronological schemes have been proposed by Ford and Willey (1941), Stoltman (1978), and Yerkes (1988).

The following culture history is based on Ahler (1987), Bush (1988),

Gremillion (1993), Henderson and Turnbow (1987), Jefferies (1988a, 1988b,

1996), Lewis (1988, 1996), McBride (1991), McBride and McBride (1996),

McGowan (1988), Niquette and Henderson (1984), O'Steen, Gremillion and Ledbetter

(1991), Railey (1990, 1991b, 1996), Rossen and Edging (1987), Sanders (1988),

Schenian (1988), Sharp (1996), and Tankersley (1990, 1996).

^ Unless otherwise noted, throughout the rest of the dissertation Mississippian refers to a cultural time period from A.D. 1,000 to 1,550; in the previous section, Mississippian referred to a geological time period of the Paleozoic Era. 2 A tradition is "a distinctive group of assemblages or artifacts that is found in a region during a long time interval" (Lewis 1996:230).

1 7 2.2.1. PALEOINDIAN PERIOD

The Paleoindian Period was not included in the original formulation of the

Midwestern Taxonomic System. But with the discovery of pre-Archaic materials in the

Eastern Woodlands, the idea of an "Early Hunting" period developed. This pre-Archaic period is now known as the Paleoindian Period, which is dated from 9,500 to 8,000 B.C.

Compared to most later periods, the Paleoindian Period is poorly known in

Kentucky and in the Eastern Woodlands in general. This is probably due to the relatively small number of Paleoindian sites, which apparently reflects the low population densities and high residential mobility of the first inhabitants of the area. About 200

Paleoindian sites have been documented in the state, but most materials are surface finds, isolated finds, or from multicomponent sites (Tankersley 1990). Most archaeological evidence of the Paleoindian way of life falls in the areas of stone tool technology, settlement, and faunal subsistence. We know less about social organization, plant use, organic tool manufacture, and trading practices. Because evidence of

Paleoindian occupations in eastern Kentucky is sparse, the following descriptions are based on work throughout the state.

Tankersley (1996) used environmental, artifactual, and subsistence information to delineate three divisions of the Paleoindian period. The Early Paleoindian or Clovis period is dated from 9,500 to 9,000 B.C. based on radiocarbon dates from other Clovis sites in the Eastern Woodlands. The post-glacial climate was cooler and moister than today, and Kentucky was covered by a "mixed coniferous forest/parkland vegetational mosaic" (O'Steen et al. 1991). In Kentucky, Early Paleoindian sites are associated with Clovis fluted projectile points, unifacial chipped-stone tools, and blades and blade cores made of high-quality cherts^ (Tankersley 1996). Artifacts have been

3 Fluted projectile points have relatively long, narrow channels originating at the base and extending toward the tip of the point: the flutes aided in halting the points to spears. Unifacial

1 8 found in association with the remains of now-extinct fauna, but the relationships are

tenuous. Regardless, most archaeologists propose that Early Paleoindian people were

specialized big-game hunters who relied heavily on large mammals but probably

exploited small mammals, fish, and wild plants as well.

Early Paleoindian groups were likely seasonally mobile, following herds of large

game animals. Camps, quarries, lithic reduction sites, and kill sites are typically located

near chert outcrops or water sources where game might aggregate (Tankersley 1996).

Early Paleoindian sites in north-central Kentucky are Big Bone Lick (Boone County),

Clays Ferry Crevice (Fayette County), and Adams Mastodon (Harrison County). Western

Kentucky Paleoindian sites are Parrish (Hopkins County), Adams (Christian County),

and Savage Cave (Logan County). The Adams site is different from many Paleoindian sites

in the Eastern Woodlands because there is little evidence of repeated use of the location

over time (Sanders 1988). Nearly all artifacts found at this single component site are

made of locally available chert. The site is interpreted as "a manufacturing site [of

Clovis fluted points] and a base camp" (Sanders 1988:18).

The Middle Paleoindian period is dated from 9,000 to 8,500 B.C., again based on correlations with remains found in surrounding states (Tankersley 1996). Climatic warming was accompanied by grassland expansion, a shift to hardwood forests, and extinction of at least some of the megafauna. Diagnostic projectile point types are

Cumberland and Gainey fluted points. Blade production ceases, and chipped-stone

implements tend to be made of poorer quality cherts. A variety of scraping tools,

including spurred end scrapers and slug-shaped scrapers, are ubiquitous at Middle

Paleoindian sites (Tankersley 1996). The subsistence strategy of Middle Paleoindian groups was likely a mixture of hunting and gathering, with reduced reliance on big game.

A relatively mobile way of life likely continued. Kentucky sites dated to this period are tools are modified on one side. Blades are elongated lithic flakes with a central ridge on one side; they are detached from specially prepared cores of raw material.

1 9 Henderson (Lyon County) in western Kentucky (Tankersley 1996) and the Savage Cave

Site where Cumberland-like points were recovered but not /r? s/Yu (Schenian 1988).

Dated from 8,500 to 8,000 B.C., the Late Paleoindian period is marked by the appearance of unfluted projectile points. The warming trend continued, accompanied by continued shifts in the flora and fauna, most notably the extinction of the remaining

Pleistocene megafauna. Tankersley (1996:33) reported that Late Paleoindian artifact assemblages are more diverse than those of the earlier periods and contain "large, bipointed, alternately beveled bifaces; backed bifaces; proximal end and side scrapers; asymmetrical end scrapers; narrow end scrapers; hafted perforators; and backed and snapped unifaces" in addition to projectile points. Point styles diversified and include lanceolate and Dalton Cluster types. In terms of subsistence, the shift to generalized foraging was firmly established by the Late Paleoindian period. Because of the widespread occurrence of game animals like deer, bear, and turkey. Late Paleoindian groups probably were not as mobile as their predecessors.

Kentucky sites dating to the Late Paleoindian period include Morris (Hopkins

County) and Roach (Trigg County) in western Kentucky (Tankersley 1996). The Savage

Cave Site yielded Quad-like points of the Dalton cluster (Schenian 1988). Two sites in eastern Kentucky yielded Late Paleoindian diagnostics. Dalton Cluster points were recovered from surface contexts at the Rhondle Lee site, which is located on a terrace along the Red River in Powell County (Applegate 1995b). A lanceolate biface was excavated from a subsurface context at Enoch Fork Shelter in Perry County (Bush

1988). The morphology of the artifact suggests the presence of a Late Paleoindian component at the shelter, but it is associated with an early radiocarbon date of 9010 -

240 B.C. and a retouched blade, both of which suggest an Early Paleoindian affiliation

(Bush 1988).

20 2.2.2. ARCHAIC PERIOD

The term Archaic was coined by Ritchie (1932) to describe pre-ceramic artifact suites from sites in New York. This period began with the Holocene Epoch about 8,000

B.C. and lasted seven thousand years. Because it covers a large time interval, the Archaic

Period encompasses considerable cultural change. For this reason, the period is divided into Early, Middle, and Late periods, as described below.

Throughout the Archaic period in Kentucky, artifact assemblages become more diverse and regional differentiation in artifact styles develops. The Archaic record in

Kentucky is known from about 1,400 sites; this constitutes over 70% of the archaeological sites documented in the state (Jefferies 1990). About 3% of the Archaic sites are located in the Upper Kentucky and Licking River Management Area (Jefferies

1 9 9 0 ).

2.2.2.a. Early Archaic

The Early Archaic period spans two thousand years, from 8,000 to 6,000 B.C.

Almost 30% of the Archaic sites documented in the state are dated to this early period

(Jefferies 1990). "Most Early Archaic artifacts ... have been recovered from culturally mixed deposits or surface collections" throughout the state (Jefferies 1988b:97).

Plant and pollen remains from eastern Kentucky indicate that climatic conditions during the Early Archaic were cooler and moister than today. "Although deciduous trees were present, hemlock and probably spruce were still growing in the upper elevations of eastern Kentucky" (Jefferies 1988b:97). Modern species of mammals, birds, and reptiles were available.

Although there is little botanical or artifactual evidence of Early Archaic plant use in eastern Kentucky, archaeologists presume that a generalized, hunting-gathering- fishing economy was practiced. Animals continued to be the predominant subsistence

2 1 resource, and faunal remains from eastern Kentucky sites include deer, elk, beaver, turtles, and birds. Fish and mussels were used as well.

The Early Archaic settlement pattern in eastern Kentucky is described as "small mobile hunting bands [exploiting] relatively large territories" (Jefferies 1988b:94).

Population densities continued at relatively low levels. Few sites show evidence of long­ term habitation. "The widespread distribution of corner and basal notched points, ... the presence of tools made from nonlocal materials, and the general absence of middens, features, and burials in Early Archaic sites suggest that most camps were used only for a short time" (Jefferies 1996:40).

Early Archaic upland sites in the eastern Kentucky mountains were occupied at least seasonally (McGowan 1988). Based on lithic remains from the Pine Fork site in

Floyd County, McGowan (1988:160) concluded that Early Archaic groups "used the site as part of a very mobile seasonal round." Nonlocal chert artifacts, which are dominated by debitage from the later stages of chipped stone tool reduction, suggest "considerable movement and/or trade within local prehistoric populations" (McGowan 1988:157).

Early Archaic components are present at both rockshelter and open sites in the

Red River Gorge area (Cowan 1976, Jefferies 1988b, Wyss and Wyss 1977). Red River shelters were probably the loci of seasonal occupations as were ridge top sites, which were rather rare during this time. In the Cave Run Reservoir area along the Licking

River, on the other hand, no Early Archaic components have been identified at open sites, and rockshelters were probably used for long-term or repeated occupation (Dorwin et al. 1970, Jefferies 1988b, O'Steen et al. 1991).

Ground-stone tools were added to the cultural repertoire during the Early

Archaic, but the diagnostic artifacts are notched chipped-stone projectile points. Typical early Early Archaic point types in the eastern part of the state are of the Kirk Corner

Notched and Kirk Stemmed clusters: Kirk Corner Notched, Stilwell, Palmer, Charleston,

22 Pine Tree, Decatur, Kirk Stemmed, and Kirk Serrated (Jefferies 1988b, O'Steen et ai.

1991). Thebes Cluster and Large Side Notched Cluster points are also documented from eastern Kentucky (Applegate 1995b, Jefferies 1996). Late Early Archaic point types are bifurcate-base points of the Rice Lobed (MacCorkle, St. Albans) and LeCroy

(LeCroy, Kanawha) clusters (Jefferies 1988b, O'Steen et al. 1991). A wide variety of raw materials were used for chipped-stone tool production (Jefferies 1988b). Few

Early Archaic burials are documented in eastem Kentucky, but grave goods from Early

Archaic burials in western Kentucky include necklaces of animal teeth and chipped-stone tools (Jefferies 1996).

In eastern Kentucky, Early Archaic artifacts or components are documented at a number of rockshelters; Big Ash Cave and Zachariah Shelter in Lee County: Dillard

Stamper Shelter #1 and Cloudsplitter Shelter in Wolfe County; Kay Shelter in Breathitt

County; Leaf Pile Site in Estill County; and JA60 in Jackson County (Applegate 1997,

Cowan 1976, Cowan et al. 1981, Fiegel et al. 1992, Jefferies 1988b, O'Steen et al.

1991, Webb and Funkhouser 1932). The Deep Shelter In Rowan County is unique in that numerous superimposed Early Archaic zones are present as intact deposits (Dorwin et al. 1970). Subsurface Early Archaic deposits are recorded from Enoch Fork Shelter in

Perry County and 15CY24^ in Clay County (Bush 1988). Open sites include Rhondle Lee

(Applegate 1995b) and two ridge top sites in Floyd County, 15FD46 and Pine Fork

(Jefferies 1988b, 1996; McGowan 1988). The latter are significant because m id d en ^ deposits and features were uncovered.

^ By convention, archaeological site designations in the United States follow the Smithsonian Trinomial System. The first part is a one- or two-digit number that indicates the state in which the site is located, in this case, 15 designates Kentucky. The second part is a two-letter abbreviation that stands for the county in which the site is located. The third part is a one- to three-digit number that indicates the number assigned to the site when it was entered into the state's files. From this point on, the 15 will be omitted from site designations to enhance readability of the descriptions. ^ A midden is "the accumulation of garbage and other living debris that marks the location of a former camp or viallage" (Lewis 1996:228).

23 2.2.2.b. Middle Archaic

In Kentucky, the boundaries of the Middle Archaic period are placed at about

6,000 to 3,000 B.C. In correspondance with climatic changes of the Hypslthermai

Interval. The lower boundary differs from that In other regions of the Eastern Woodlands where the lower boundary Is at 4,000 B.C. Although over 200 Middle Archaic sites are documented throughout the state, this period "Is not well understood In much of

Kentucky" due to definitional problems and/or low site visibility (Jefferies

1988b:102).

With the warmer and drier conditions of the Hypslthermai, the nature of the biosphere changed In Kentucky. The mixed mesophytic hardwood forests gave way to a mixed oak-hlckory canopy of reduced spatial extent (O'Steen et al 1991). Grasslands expanded In some areas (Jefferies 1996).

Middle Archaic subsistence Involved generalized collecting of a greater diversity of wild plant and animal resources. Deer and turkey were the primary sources of animal protein, but aquatic resources were also used. "Hickory and other nuts were Important plant foods, and starchy seeds, greens, and sap were also eaten" (Jefferies 1996:49).

Toward the end of the Middle Archaic, there Is evidence that humans began to cultivate some plants (Jefferies 1988b, MIckelson and Gremillion 1995).

According to Jefferies (1996:50), In eastern Kentucky "Middle Archaic settlement patterns were similar to those of the Early Archaic" but. In general, settlement strategies changed over the Middle Archaic period (Jefferies 1988b). During the early part of the period, locations near water sources were the foci of repeated or long-term occupations. Toward the end of the period, populations dispersed In response to environmental changes associated with the end of the Hypslthermai.

Specialized tools for acquiring and processing a variety of resources are found at

Middle Archaic sites. New ground-stone Implements Included atlatl weights, axes.

24 pestles, pitted anvils, and grinding stones. Bone and antler tools became more common

(Jefferies 1996). Diagnostic chipped-stone projectile point styles in eastern Kentucky are Morrow Mountain, Matanzas, Godar, Raddatz, and Big Sandy (Jefferies 1988b,

1996). In the Red River Gorge, most chipped-stone tools are made of chert. Middle

Archaic burials have not been found in the eastern part of the state, but burials at the

KYANG site in north-central Kentucky contained a variety of artifacts, including

"engraved bone pins; wolf, deer, and bear tooth necklaces; red ochre; groundstone pendants and beads; a deer antler atlatl hook; and a variety of chert tools" (Jefferies

1 9 9 6 :5 4 ).

Middle Archaic sites in the Upper Kentucky and Licking River M anagement Area include Mounded Talus Shelter in Lee County (Mickelson and Gremillion 1996), Dillard

Stamper Shelter #2 in Wolfe County (Webb and Funkhouser 1932), and Rhondle Lee

(Applegate 1995b) and P046 (Cowan et al. 1981) in Powell County.

2.2.2.C. Late Archaic

The Late Archaic period is dated from 3,000 to 1,000 B.C. With the end of the

Hypsithermal interval, the climate of Kentucky stabilized, and by about 1,000 B.C. essentially modern faunal and floral communities were established. Evidence of Late

Archaic occupations in the state is considerably more visible than earlier periods. Even so, the Late Archaic represents "a continuation of a trend toward greater regional specialization and adaptation first evident in the Middle Archaic period" (Jefferies

1988b: 106). Some archaeologists recognize a Terminal Archaic period, dated from

1,500 B.C. to 1,000 B.C.

Archaeological research in Kentucky has contributed significantly to our understanding of the Late Archaic period in the Eastern Woodlands. In the early twentieth century, excavations at Mammoth Cave, the Green River shell mounds in western

Kentucky, and eastern Kentucky's rockshelters produced volumes of information about

25 Late Archaic sites, even though the Archaic concept had not yet been formulated

(Jefferies 1988a). During the New Deal era of large-scale excavations (1935 to

1941), Webb and others investigated twenty sites with significant Late Archaic components along the Tennessee and Green Rivers in western Kentucky and the

Cumberland River in southern Kentucky (Jefferies 1988a).

Late Archaic occupants of eastern Kentucky practiced a mixed subsistence strategy, as had Middle Archaic groups. Deer and hickory nuts formed the core of the diet

(Jefferies 1988b, 1996). Chestnut use increased. Wild fruits, seeds and small mammals were eaten. Unlike other parts of the state, eastern Kentucky sites provide limited evidence of mussel exploitation. A significant subsistence change toward the end of the Late Archaic period involved cultivation of squash, gourd, and native starchy seed- bearing plants like goosefoot, sunflower, knotweed, and marshelder (Cowan et al. 1981;

Gremillion 1995, 1996, 1997; Ison 1988; O'Steen et al. 1991).

"The size, number, and distribution of Late Archaic sites" differs from previous periods (Jefferies 1988b:106). Many Late Archaic sites are smaller than Middle

Archaic sites (Jefferies 1996). Over half of the Archaic sites documented in Kentucky are of Late Archaic affiliation (Jefferies 1990). Late Archaic components are common at both rockshelter and flood plain sites. The Late Archaic marks the "first substantial flood plain occupations" in the Red River Gorge area (Jefferies 1988b:111).

The trend toward increased technological specialization, which began in the

Middle Archaic, continued in the Late Archaic period (Jefferies 1988b). New types of chipped-stone, ground-stone, bone, and wood tools developed. Stemmed points and knives are diagnostic Late Archaic chipped-stone tools in eastern Kentucky. Stylistic types include Merom-Trimble, Brewerton, McWhinney, Benton, Lamoka, Pickwick,

Ledbetter, Savannah River, Cogswell, Delhi, and Wade. A wider range of materials was used to manufacture chipped-stone tools; "siliceous shale, quartz, siderite, and

26 terrigenous sandstone" (Jefferies 1988b:107) were locally available, but non-local cherts were used as well. Ground-stone and polished-stone tools are more abundant than in preceding periods.

While the existence of long-distance trade networks is induced for earlier periods, good evidence of Late Archaic exchange takes the form of marine shell artifacts, mica (Sampson Spencer Shelter), and non-local cherts in eastern Kentucky

(Funkhouser and Webb 1930, Jefferies 1996). It is not surprising that copper is not found at eastem Kentucky sites because copper deriving from the Great Lakes area is rarely found beyond a 560 km radius from the sources, with the exception of sites along the Green River in western Kentucky (Goad 1980).

In some parts of Kentucky like the Green River, evidence suggests that Late

Archaic groups developed some degree of social inequality. Certain burials contain bone hair pins, red ochre, shell cups, and other "exotic grave goods made from Great Lakes copper and marine shells, suggesting special treatment of high-status individuals"

(Jefferies 1996:54). While rockshelter burials are found in the eastern part of the state, large cemeteries and differentiated grave goods are generally lacking.

Throughout the state, the Late Archaic period is divided into a number of phases.®

Four phases are recognized in eastern Kentucky. The Skidmore Phase (2,400-1,650

B.C.) of the Red River-Licking River area is associated with flood plain base camp sites like Seldon Skidmore and Bluestone (Jefferies 1988b, 1996; O'Steen et al. 1991).

Upland sites were used less intensely. Diagnostic attributes of the Skidmore Phase are contracted or straight-stemmed bifaces with ovate blades, chert adzes, grooved axes, grinding stones, thick midden deposits, fire-cracked rock, circular earth ovens, and roasting pits (Jefferies 1988b, 1996; O'Steen et al. 1991).

® A phase is "a detailed description of the human communities of a region during a relatively short time interval, for instance one or two centuries" (Lewis 1996:229).

27 The Slone Phase (1,900 B.C.) and the Maple Creek Phase (1,650-1,250 B.C.) are Late Archaic manifestations whose geographic ranges fall partially within eastern

Kentucky. The Slone Phase, originally defined by Dunnell (1972), was centered in southeastern Kentucky and western W est Virginia and Virginia. Diagnostics include siderite tools, nut-processing tools, chipped-stone axes, and “bifacially asymmetrical knives" (Jefferies 1996:67). The Maple Creek Phase of northern Kentucky-southern

Ohio, which was defined by Vickery (1980), is demarcated by Merom-Trimble points in addition to McWhinney, Brewerton, Lamoka, and Ashtabula (O'Steen et al. 1991). Other diagnostics are micro-perforators, gravers, drills, manos, and the scarcity of ground- stone tools. Maple Creek sites in northeastern Kentucky are Grayson (Carter County),

Hansen (Greenup County), and Glacken (Boone County).

The Cogswell Phase (1,250-800 B.C.) was defined by Ison (1988) to describe

Terminal Archaic-Early Woodland occupations in eastern Kentucky. Cogswell and Wade points made of local cherts are diagnostic of this phase. Sites with Cogswell Phase components include Cold Oak Shelter, Grayson, and Zilpo (Bath County).

Numerous shelters with Late Archaic components are found in the study area.

Examples are Cold Oak, Big Ash Cave, Buckner, Pine Crest, and Ratliffe shelters in Lee

County: Sampson Spencer, Worth Creech, Rhoda Smith, George W. Spencer, Dillard

Stamper #1 and #2, Green Gentry, and Cloudsplitter shelters in Wolfe County; Newt

Kash Hollow Shelter in Menifee County; Steven DeHart Shelter in Powell County; Kay

Shelter in Breathitt County; The Pitch Pine Rockshelter and The Hawk-View Shelter Site in Owsley County; and Silver Mine Hollow Rockshelter, JA73, and Wheeler Rockshelter in Jackson County (Applegate 1997; Cowan and Wilson 1975; Fiegel et al. 1992;

Gremillion 1995; Knudsen et al. 1983; O'Steen et al. 1991; Turnbow 1975; Webb and

Funkhouser 1932). Open sites with Late Archaic components include Skidmore and

Rhondle Lee in Powell County. Elsewhere in eastern Kentucky, important sites have been

28 documented at Fishtrap Reservoir in Pike County, Paintsville Reservoir in Johnson

County, and Cave Run Reservoir (Deep Shelter and Zilpo site) in Bath County. Additional sites are the Bluestone Complex in Rowan County, Grayson, and Hansen.

2.2.3. WOODLAND PERIOD

The Woodland period covers two thousand years of Eastern Woodlands prehistory.

The lower boundary of the period, dated at about 1,000 B.C. in most parts of the Eastern

Woodlands, is associated with a number of cultural changes. Diagnostic Woodland traits include the development of ceramic technology, construction of earthworks, special mortuary treatments, extensive networks for the exchange of raw materials and finished goods, intensification of horticultural practices, and the development of social stratification.

In Kentucky, the term Woodland was initially used by Webb in the 1930s to describe artifact assemblages recovered from eastern Kentucky rockshelters (Schwartz

1967). It later became associated with earthen mound sites across the state. Woodland period sites in the Upper Kentucky and Licking River Management Area number 160 and are concentrated in the Gorge section (Railey 1990). The three divisions of the Woodland period are described below.

2.2.3.a. Early Woodland

Over much of the Eastern Woodlands, the Early Woodland period is dated from

1,000 to 200 B.C. Several regional cultural manifestations are associated with the

Early Woodland. Eastern Kentucky falls within the range of the Adena complex of the

Middle Ohio Valley and excavations at a number of Adena sites in Kentucky contributed to the formulation of the Adena description (Schwartz 1967). Railey (1996:79), however, argued that Adena "is probably more a Middle Woodland culture than an Early

29 Woodland one." The Cogswell Phase, discussed in the Late Archaic section above, is a more localized early Early Woodland manifestation.

Early Woodland groups in eastern Kentucky exploited wild and domesticated resources. There were no dramatic differences in the types of foods consumed by Early

Woodland groups compared to Late Archaic populations. However, there is evidence of increased use of cultigens (Cowan et al. 1981; Gremillion 1995, 1996, 1997).

Extractive territories may have decreased in size and resulted in reduced residential mobility during the Early Woodland. Early Woodland settlement strategies in eastern Kentucky differ from Late Archaic strategies in two ways. First, the number of rockshelter occupations increases. Second, ritual sites (nonmound inhumation and earthen burial mounds) located away from habitation sites develop (Railey 1990,

1991a, 1996). In the Gorge section of eastem Kentucky, Early Woodland populations occupied rockshelters and stream terrace locations; shelter occupations not only outnumber bottom land sites, but also represent more intense occupations in many cases

(Cowan 1985; O'Steen et al. 1991; Railey 1990, 1996).

The Early Woodland period is associated with conical burial mounds, cremated inhumations, engraved stone tablets, tubular and elbow pipes of stone and ceramic, plain or cordmarked ceramic vessels with limestone or grit temper, straight or contracting stemmed bifaces, stone gorgets, and copper and mica artifacts. In eastern Kentucky, diagnostic Early Woodland point types are Cogswell, Wade, Turkeytail, Motley, Savannah

River, Kramer, Adena, Cypress Creek, Little Bear Creek, and Robbins (Railey 1990,

1 9 9 6 ).

Eastern Kentucky rockshelters with Early Woodland components include Cold

Oak, Red-Eye Hollow, Little and Big Ash Caves, Big Turtle, and Pine Crest shelters in Lee

County; Worth Creech, Rhoda Smith, George W. Spencer, Dillard Stamper #1 and #2, and Cloudsplitter shelters in Wolfe County; Newt Kash Hollow and Hooton Hollow shelters

30 in Menifee County: Steven DeHart Shelter in Powell County; Jake Fork Rockshelter in

Breathitt County; and JA49 and JA122 in Jackson County (Applegate 1997; Cowan

1975; Cowan et al. 1981; Gremillion 1995; Knudsen et al. 1983; O'Steen et al. 1991;

Turnbow 1975, 1981; Webb and Funkhouser 1932, 1936). Open sites Include Rhondle

Lee, ESI 9 in Estill County, and Grayson. Mound sites are less common in eastern

Kentucky than to the west, but an example is the C & O Mounds in Johnson County (Webb

1 9 4 2 ).

2.2.3.b. Middle Woodland

Dated from 200 B.C. to A.D. 500, the Middle Woodland period is associated with the Hopewell complex in the Middle Ohio Valley. The Hopewell is considered one of two

"cultural climaxes" in the Eastern Woodlands and is associated with "an intensification of ceremonialism and long distance trade" (O'Steen et al. 1991:29). The most visible expression of Hopewell in Kentucky is along the Ohio River and the extreme northeast.

Elsewhere, the Middle Woodland is poorly documented compared to the earlier and later

Woodland periods.

"Middle Woodland subsistence was based on hunting, gathering, and gardening"

(Railey 1996:90). Middle Woodland groups continued to cultivate native starchy-seed plants, squash, and gourds, and "maize is unknown from Middle Woodland sites in

Kentucky" (Railey 1996:90).

Within the Middle Ohio Hopewell core, settlement strategies are described as dispersed farmsteads organized around earthworks (Dancey 1992). "Early Middle

Woodland groups in central and eastern Kentucky probably lived in small, scattered settlements with ritual space such as burial mounds and earthen enclosures serving as focal points of group ritual and social integration" (Railey 1996:90). In eastern

Kentucky and in the Bluegrass area, groups congregated in large nucleated settlements as mound construction declined during the late Middle Woodland (Railey 1996).

3 1 In the Upper Kentucky and Licking River Management Area, however, evidence of domestic and ritual activities occurs at both rockshelters and bottom land sites, and earthworks are not present (O'Steen et al. 1991, Railey 1996). In the Red River Gorge,

Middle Woodland sites are located closer to the river than in other periods (Wyss and

Wyss 1977), and "there was some degree of re-establishment of bottom land settlements" (O'Steen et al. 1991). "Numerous large and small sites have been found dating to this period, suggesting periodic aggregation and dispersion, or some kind of a village/base camp - specialized resource extraction station settlement dichotomy"

(O'Steen et al. 1991:30). "Middle Woodland use of rockshelters ranged from short-term camps that left few traces to extended, possibly sedentary, occupations with thick middens and rich assemblages" (Railey 1996:110).

Diagnostic Middle Woodland artifacts in eastern Kentucky are bifaces and ceramics. Middle Woodland projectile point types documented at eastern Kentucky sites are Snyders, Robbins, Lowe Flared Base, Chesser Notched, and Copena Triangular. Plain, cordmarked, brushed, and checkmarked ceramics are characteristically limestone tempered, although a variety of tempering materials was used. Early Middle Woodland vessel forms include plain jars with thick outflaring rims, while cordmarked jars characterize the late Middle Woodland (Railey 1996). Grave goods from eastern

Kentucky sites are less elaborate than those from core Hopewellian sites, lacking copper and mica artifacts, blades, and effigy or platform pipes, for example.

Eastern Kentucky rockshelters with Middle Woodland components or artifacts include Little Sinking Cave, Pine Crest, LE68, Crystal Creek shelters in Lee County:

Worth Creech, Rhoda Spencer, George W. Spencer, and Trinity shelters in Wolfe County;

MF125, MF127, and MF129 in Menifee County; and JA49 in Jackson County (Applegate

1997, Funkhouser and Webb 1936, Knudsen et al. 1983, O'Steen et al. 1991, Turnbow

1981, Webb and Funkhouser 1932, Wyss and Wyss 1977). Examples of Middle

32 Woodland open sites are Rhondle Lee and Anderson in Powell County (Applegate 1995b,

Cowan 1976) and the Calloway site in Martin County (Railey 1996). Brisbin Mound in

Boyd County, the C & O Mounds in Johnson County, an Adena site that Railey (1996)

placed in the Middle Woodland Period, and the Old Fort Earthwork and the Biggs site in

Greenup County are examples of Middle Woodland earthworks in eastern Kentucky.

2.2.3.C. Late Woodland

The Late Woodland period spans five hundred years from A.D. 500 to 1,000 in

eastern Kentucky (A.D. 900 in western and southern Kentucky). "By the beginning of the

Late Woodland period, earthwork construction and the long-distance exchange of goods had declined sharply" for uncertain reasons (Railey 1996: 110). There was considerable cultural continuity in eastern Kentucky, though, because Hopewellian influences there had been minor (O'Steen et al. 1991).

Cultigens became a larger portion of the diet during the Late Woodland, but the subsistence economy remained a mix of hunting, gathering, and horticulture (Ahler

1987, O'Steen et al. 1991, Railey 1996). Evidence of Late Woodland subsistence change is absent from the Hansen site in Greenup County (Ahler 1987). Maize use increased toward the end of the Late Woodland in the western part of the state.

Northeastern Kentucky witnessed the development of nucleated, circular villages with central plazas during the early Late Woodland. Examples of these Newtown Complex village sites are Hansen and Bentley in Greenup County (Ahler 1987; Railey 1991b,

1996). "By terminal Late Woodland times (ca. A.D. 700-1,000), Newtown villages in central and northeastern Kentucky appear to have been abandoned, and a more dispersed settlement pattern, similar to that of Adena-Hopewell times, returned" (Railey

1996:116). Smaller sites associated with households or household clusters, such as

Niebert and Grayson, replaced the nucleated villages (Ahler 1987, Ledbetter and O'Steen

1 9 9 1 ).

33 In east-central and southeastern Kentucky, nucleated villages did not develop at any time during the Late Woodland. Rockshelters continued to be used; shelters with

Newtown Phase Late Woodland components include Rock Bridge Shelter in Wolfe County and Rogers and Haystack shelters in Powell County. Upland sites continued to be occupied at least seasonally (McGowan 1988). Based on lithic remains from the Pine Fork site in

Floyd County, McGowan (1988) concluded that Late Woodland groups used the site to gather resources that supplemented the horticultural component of the diet. Nonlocal chert artifacts, which are dominated by debitage from the later stages of chipped-stone tool reduction, suggest "considerable movement and/or trade within local prehistoric populations" (McGowan 1988:157).

Little evidence of changes in material culture occurred during the early Late

Woodland (Ahler 1987, Railey 1996). "During the terminal Late Woodland, however, there began to emerge strong regional differences in ceramic styles" and the bow and arrow was introduced (Railey 1996:111). In eastern Kentucky, diagnostic Late

Woodland biface types are Chesser, Lowe, Steuben, and Bakers Creek. These Lowe Cluster types are characteristic of both the Newtown and Sims Creek phases in eastern Kentucky.

Other point types associated with the Late Woodland are pentagonal. Jacks Reef, and large triangular points (Ahler 1987).

The Late Woodland period in eastern Kentucky is divided into a number of phases based on ceramics (O'Steen et al. 1991). The Newtown Phase, dated from A.D. 300 to

800 in the study area, is considered a Middle Woodland-early Late Woodland manifestation. Diagnostic Newtown ceramics are grit or limestone tempered, cordmarked or check stamped, subconical or subglobular jars with angular shoulders. Lowe Cluster, pentagonal. Jacks Reef, and large triangular points are found at Newtown Phase sites

(Ahler 1987).

34 The Sim's Creek Phase (Dunneli 1972), dated from A.D. 400 to 700, is characterized by sandstone-tempered pottery vessels with cordmarking and non-angular shoulders (Ahler 1987). Sim's Creek sites like Slone (Pike County) and Dow Cook

(Lawrence County) probably represent "seasonal camps or small hamlets" (Railey

1 9 9 6 :1 1 7 ).

Rockshelter sites in the Upper Kentucky and Licking River Management Area with Late Woodland components are Little Sinking Cave and Big Turtle Shelter in Lee

County: Worth Creech, Rhoda Smith, George W. Spencer, Trinity, and Rock Bridge shelters in Wolfe County; MF129 in Menifee County; Rogers and Haystack shelters in

Powell County; Mud Dauber Shelter and Kay Shelter in Breathitt County; The Pitch Pine

Rockshelter and The Hawk-View Shelter Site in Owsley County; and Barking Dog Site,

Dead Cow Site, and Hungry Hound Shelter in Jackson County (Applegate 1997; Cowan

1974, 1975, 1979; Cowan et al. 1981; Fiegel et al. 1992; Gremillion 1993; Knudsen et al. 1983; O'Steen et al. 1991; Turnbow 1981; Webb and Funkhouser 1932; Wyss and

Wyss 1977). Late Woodland points were recovered from surface contexts at the bottom land Rhondle Lee site (Applegate 1995b).

2.2.4. MISSISSIPPIAN PERIOD

The Mississippian period is archaeologically defined as both a temporal period, dated from A.D. 900 to 1,700 in the Eastem Woodlands, and a cultural tradition or way of life (Lewis 1988, Yerkes 1988). In major river valleys of the Eastern Woodlands,

Mississippian "chiefdoms" developed a complex settlement hierarchy, constructed flat- topped platform mounds in planned towns, and were socially stratified. Such groups were concentrated in the western part of Kentucky, where "their remains are among the largest, most complex archaeological sites in Kentucky" (Lewis 1996:125-126). The

35 Mississippian period in eastern Kentucky, on the other hand, is associated with somewhat different cultural complexes: PIsgah and Fort Ancient.

Pisgah complex sites are identified by diagnostic ceramics in southeastern

Kentucky, along the upper Cumberland River in Perry, Knox, and other southeastern

Kentucky counties that border Virginia. Pisgah is associated with the southern

Appalachians, and Pisgah populations in Kentucky probably were influenced by contemporaneous Mississippian groups in Tennessee (Lewis 1996, Niquette and

Henderson 1984, O'Steen et al. 1991). An example of a Pisgah site in southeastern

Kentucky is HL304 in Harlan County; the site represents a small hamlet with structural remains from two domiciles (Lewis 1996).

Fort Ancient refers to Mississippian manifestations in the Middle Ohio Valley, including northeastern Kentucky, dated between A.D. 1,000 and 1,550. Fort Ancient is considerably more visible and widespread than Pisgah in eastem Kentucky. The Fort

Ancient complex is often divided into early (A.D. 1,000 to 1,200), middle (A.D. 1,200 to 1,400), and late (A.D. 1,400 to 1,550) based on ceramic and projectile point stylistic types (Henderson and Turnbow 1987, Sharp 1996).

Subsistence economies during the Mississippian period, including those of the

Fort Ancient in eastern Kentucky, were based on maize agriculture and continued exploitation of wild plants and animals. Compared to earlier occupants in eastern

Kentucky, Fort Ancient groups relied more heavily on domesticated plants, although nut collecting, wild plant collecting, and hunting continued. According to Henderson and

Turnbow (1987:217), Fort Ancient subsistence reflects "relatively diverse plant exploitation and a multiple-plant oriented subsistence strategy" as new resources were added to the diet while others were dropped. In contrast with western Kentucky Middle

Mississippian groups. Fort Ancient populations of eastern Kentucky made greater use of corn as a dietary staple, exploited beans to a greater extent, consumed native cultigens

36 and nuts to a lesser extent, but used wild fruits and berries (Rossen and Edging 1987).

Besides com and beans, squash and tobacco were cultivated. The only native cultigen recovered from Fort Ancient sites is Chenopodium (Henderson and Turnbow 1987).

Hickory was the major nut used by Fort Ancient people. A variety of wild plants were exploited: pawpaw, grape, bedstraw, smartweed, pokeberry, and morning glory

(Henderson and Turnbow 1987).

Evidence of animal use at Fort Ancient sites in northeastern Kentucky indicates that large terrestrial animals, such as deer, elk and bear, and wild turkey were used to a greater extent than smaller mammals and aquatic resources. This pattern differs from other Fort Ancient sites in the Middle Ohio Valley, where deer is the primary faunal resource and elk is more ubiquitous than bear (Henderson and Turnbow 1987).

"Fort Ancient settlements were more nucleated, and larger, than Late Woodland settlements. Village sites tend to be located in valley bottoms, while smaller sites that may represent seasonal camps are found throughout tributary drainages" (O'Steen et al.

1991). Like Mississippian sites, larger Fort Ancient sites were palisaded and arranged around central plazas. But they differ in that public earthwork constructions were absent and comparable settlement hierarchies did not develop (Sharp 1996).

Diagnostic artifacts of the late prehistoric period, including the Fort Ancient complex, are shell-tempered ceramics and small triangular projectile points. The characteristic Fort Ancient ceramic vessel form is the jar. While this form first appeared during the Late Woodland Period, Fort Ancient people began tempering with crushed shell and added strap handles or lugs. Fort Ancient projectile points are characteristically triangular in shape. Flared-base triangular points are associated with early Fort Ancient, coarsely serrated triangular points are found at middle Fort Ancient sites, and late Fort Ancient is associated with concave-base triangular, straight-sided triangular, and short excurvate-sided points (Henderson and Turnbow 1987). Chipped-

37 stone tool inventories of late Fort Ancient sites are more diverse than those of early and middle Fort Ancient sites (Henderson and Turnbow 1987). Compared with their predecessors, Fort Ancient groups used copper is smaller quantités; however, they excelled in shell work (Schwartz 1967).

The only Fort Ancient phase designated in the Eastem Mountains is Woodside

(Sharp 1996). Woodside sites are documented from Pike (Slone site), Breathitt (BR9),

Johnson (Mayo site), and nearby counties along the Kentucky and Big Sandy River drainages. Knot-roughened pottery is more abundant at Woodside sites than at other Fort

Ancient sites, and projectile point morphologies differ (Sharp 1996).

Four Fort Ancient phases are identified for along the Ohio River based on radiocarbon dates, ceramic styles, and chipped-stone projectile point types (Henderson and Turnbow 1987, Sharp 1996,). In northeast Kentucky, the Croghan phase correlates with the early Fort Ancient, is focused on the Ohio River near the confluences with

Tygarts Creek and Little Sandy River, and is characterized by Baum ceramics and flared-base triangular points. The Manion (middle Fort Ancient), Gist and Montour (late

Fort Ancient) phases are centered in the bordering Bluegrass area.

Rockshelter sites with Fort Ancient components include the Crystal Creek and

Ratliffe shelters in Lee County; Green Gentry, Cloudsplitter, and Lindon Fork shelters in

Wolfe County; Webb Shelter and at least twenty unnamed shelters in Menifee County;

Mud Dauber and Kay shelters in Breathitt County; Barking Dog Site, Dead Cow Site,

Wheeler Rockshelter, and Hungry Hound Shelter in Jackson County; Bobby Pendleton and

Dewy Rowland shelters in Morgan County; and The Pitch Pine Rockshelter and The

Hawk-View Shelter Site in Owsley County (Applegate 1997; Cowan 1974; Cowan and

Wilson 1975; Fiegel et al. 1992; Hand 1992; Kluth et al. 1992; Niquette and Hughes

1992; Sanders 1976, 1991; Sanders and Sanders 1975; Turnbow 1975, 1981; Webb and Funkhouser 1932; Wyss and Wyss 1977).

38 2.2.5. HISTORICAL PERIOD

The Historical period in Kentucky witnessed the widespread depopulation of

Native Americans and the establishment European-American and African-American influences in the area. Native populations were displaced by immigrants and reduced in number by foreign diseases. Dispersed groups of Shawnee, Cherokee, Chickasaw,

Delaware, and Miami lived in eastern Kentucky during the early historic period

(Cotterill 1954, Rice 1975).

Between the middle eighteenth century and the early twentieth century, four phases of European-American and African-American occupations are identified

(McBride and McBride 1996). Mountain settlement by new groups did not begin until the 1780s and 1790s, during the middle of the Exploration and Early Settlement phase

(1749-1810). European-Americans and African-Americans entered eastern Kentucky by the Ohio River or through mountain cols like Cumberland Gap in the extreme southeastern part of the state. Early settlers included hunters/trappers and farmers who lived on small farms or isolated plantations (McBride and McBride 1996). Many were of Scotch-lrish or German descent (O'Steen et al. 1991). In the Robinson Forest area of Breathitt and Knott counties, most sites dating from this period are farmsteads, along with some mills and gas wells (McBride 1991).

Antebellum phase (1810-1860) research in eastern Kentucky has focused on nitre mining and other industries. "These studies have presented valuable information on the extraction and processing of nitre or saltpeter, and on the equipment used in these operations. ... these nitre mining sites [have] been the major means for archaeologists to study Kentucky's role in the War of 1812" (McBride and McBride 1996:201-202).

Sites associated with nitre mining are mostly rockshelters, which have been located in

Wolfe, Menifee, Jackson, Carter and other counties (Funkhouser and Webb 1936,

39 Gremillion 1993, McBride and McBride 1996). Most oil drilling in the eastern part of

the state occurred during the Antebellum phase (O'Steen et al. 1991). Iron furnaces

such as the Red River Iron Works in Powell County produced wrought iron during this

time (O'Steen et al. 1991).

Civil War phase (1861-1865) research has focused on forts and other military

sites, which are not common in eastern Kentucky. In Robinson Forest, sites dated to this

period are farms and some industrial sites (McBride 1991).

The Postwar phase (1865-1915) in eastern Kentucky is associated with the

lumbering industry (McBride 1991, McBride and McBride 1996). The period of

commercial logging between the 1880s and 1923 in Robinson Forest saw the

abandonment of farmsteads, the appearance of temporary domestic sites, and the

expansion of sawmills, dams, railroad structures, and other industrial sites (McBride

1991). Frugal consumption economies characterize farmstead sites of this period in

eastern Kentucky, such as Prater in Floyd County (McBride and McBride 1996). Large timber and mining cities developed in some parts of eastern Kentucky, such as

Cumberland Gap in the southeast (McBride and McBride 1996).

2.2.6. SUMMARY

In many ways, eastern Kentucky culture history has been cyclical. Subsistence

economies shifted between focal or specialized and diffuse or generalized (Cleland

1976). Early inhabitants had a narrow diet, possibly focusing on megafauna resources.

A relatively generalized diet persisted a considerable length of time over the Archaic and

Woodland periods. With the late development of maize agriculture. Late Woodland,

Mississippian, and historic groups again became specialized. Yet, continuity in the

40 subsistence base of prehistoric Kentuckians is seen in the exploitation of wild resources over the millennia.

Settlement cycling was more frequent than swings in subsistence. Small, dispersed sites characterize the early-middle Paleoindian, Early Archaic-early Middle

Archaic, Early Woodland-Middle Woodland, and Historic periods. Relatively larger and, in some cases, more nucleated sites are evidenced during the late Paleoindian, late Middle

Archaic, Late Archaic, and Late Woodland-Mississippian periods.

One might argue that technological changes have been sporadic as well. Chipped- stone tool inventories are common during the earliest periods, but ground-stone tools become widespread as well during the Middle Archaic and later periods. The development or diffusion of ceramic technology seems to have profoundly influenced native Woodland and subsequent cultures. Shell work and the bow and arrow are rather late developments.

Other technologies, such as earthwork construction and copper working, were not as common in much of eastem Kentucky as elsewhere in the Eastem Woodlands.

The study of eastern Kentucky culture history makes evident a pattern that is noted in many parts of the Eastern Woodlands. Different aspects of prehistoric culture underwent change at different times and in different directions. Traditional boundaries between culture-historical periods may correspond to changes in one aspect of culture but not others. There are few neat and tidy cultural-temporal packages.

2.3. PREVIOUS ARCHAEOLOGICAL RESEARCH

The following summary of previous archaeological research in eastern Kentucky rockshelters is based on site reports, survey manuscripts, state archaeology files, and comprehensive summaries of Kentucky archaeology like Lewis (1996) and Schwartz

(1967). This section focuses on three issues: the historical context of archaeological

4 1 research in eastern Kentucky, summaries of shelter work in Wolfe, Lee, and

surrounding counties, and explanations of prehistoric rockshelter use.

2.3.1. HISTORICAL CONTEXT

Kentucky archaeologists typically recognize two phases of archaeological

research in the eastern part of the state. Extensive archaeological investigations of

eastern Kentucky's rockshelters and Adena mound sites began in the 1920s and 1930s

with the work of Webb, Funkhouser, and their associates. After a relative hiatus in

research during the next two decades, the second peak began in the 1960s with culture

resource management (CRM) related work. Shelter research that generated information

about prehistoric subsistence is viewed by Schwartz (1967) as one of two major

contributions Kentucky archaeology made to our understanding of Eastem Woodlands

prehistory.

Though criticized for their lack of stratigraphie controls and field documentation,

Webb and Funkhouser were instrumental in the early years of Kentucky archaeology

because they recorded dozens of prehistorically occupied rockshelters and made known the preservation potential of these locations (Funkhouser and Webb 1928, 1929,

1930; Webb and Funkhouser 1932, 1936). Webb and Funkhouser excavated extensively a number of shelters in Lee, Wolfe, Powell, and Menifee counties, including

Newt Kash Hollow Shelter, Little Ash Cave, Big Ash Cave, and Dillard Stamper Shelter.

Funkhouser and Webb's first published report on the archaeological resources in

eastern Kentucky rockshelters appeared in 1928. In a chapter entitled "Cliff Dwellers,"

Funkhouser and Webb described the cultural traits that distinguished a cliff-dwelling

population from other prehistoric groups in Kentucky. The three unique traits are graves, hominy holes and associated stone pestles and plugs, "kitchen-midden" deposits

42 containing discarded flints, sherds, animal bone, shell and bone tools, charcoal, and plant materials. Prehistorically inhabited rockshelters offered the cliff dwellers "camping- places and protection," and many of the shelters are "found along the old 'Indian Trails' across the mountains" (Funkhouser and Webb 1928:148).

Webb and Funkhouser's (1932) archaeological survey of Kentucky documented

1,255 prehistoric sites, including 108 rockshelters. They evaluated the archaeological resources for each Kentucky county in separate chapters of the publication, reporting the results of their surveys as well as information provided by informants.

Funkhouser and Webb also conducted county-wide surveys. In 1929, they published the results of their assessment of rockshelter resources in Lee County.

Funkhouser and Webb (1930) conducted research in Wolfe and Powell counties in the following year. Descriptions of rockshelters in Menifee County were published in 1936.

Haag's work in eastern Kentucky rockshelters introduced more rigorous field methods (Haag 1947, 1960, 1974). One of the most important sites excavated by Haag is Hooton Hollow Shelter. Although the field notes from the excavations were lost, Haag

(1974) indicates that the shelter contained a significant Adena component.

Works Progress Administration archaeological survey and excavation projects were limited in eastern Kentucky to the counties of Menifee, Montgomery, and Bath.

Overall, WPA research in the Bluegrass region and in select counties along the Ohio,

Green, and Tennessee Rivers overshadowed work in eastem Kentucky. Archaeologists paid special attention to Adena sites, especially earthworks, during the WPA era (Haag

1974). Work on Adena focused on static material culture and trait lists, although the researchers "made some fundamental contributions in terms of house patterns or settlement patterns" (Haag 1974:142).

Archaeologists undertook relatively few projects in the study area in the 1940s and 1950s. Government reservoir projects and culture resource management surveys

43 related to pipeline projects and timber sales spurred a second phase of research beginning in the late 1950s and continuing through the 1980s. These studies were primarily salvage in nature. A proliferation of publications documented hundreds of additional rockshelter sites.

In the 1960s and 1970s, archaeologists conducted several surveys related to the proposed Red River Reservoir project along the upper course of the stream. Because the researchers demonstrated "a continuous utilization of both the floodplain and rockshelter sites beginning during the Early Archaic and continuing through the Woodland and Fort

Ancient periods, " officials decided to move the dam project downstream (Weinland and

Sanders 1977:19).

In the 1970s, the Kentucky Heritage Commission (KMC) sponsored county surveys, including several in eastern Kentucky: Floyd County (Sanders and Gatus

1977), Perry County (Gatus and Sanders 1978), Bell County (DeLorenze 1979),

Greenup County (Maynard and Gatus 1979), and Knox County (DeLorenze and Weinland

1980). One of these surveys, Weinland and Sanders' (1977) work in Powell County, falls in the present study area.

Archaeology in eastem Kentucky since the CRM boom has been a mixed bag.

Cultural resources assessments of relatively large tracts of land, which are often related to timber sales and pipeline projects, continue. Reports assessing the eligibility of rockshelters for nomination to the National Register of Historic Places are published frequently. Rockshelter studies conducted by archaeologists in academic circles address questions of prehistoric subsistence, settlement, and technology.

44 2.3.2. PREVIOUS RESEARCH BY COUNTY

In this section, the nature and general results of previous investigations at rockshelters are summarized by county. Descriptions of individual shelters are not given in this limited space because the number of rockshelters in the study area is too large. Refer instead to Applegate (1997), which summarizes attributes for shelters in

Wolfe, Lee, and surrounding counties. County locations are indicated on Figure 2.

2.3.2.a. Lee County

"Lee County is in the heart of the 'rock shelter' culture and the possibilities of discoveries of this type of prehistoric habitation is [sic] almost unlimited" (Webb and

Funkhouser 1932:219). Site forms for over 50 Lee County rockshelters were examined at the Office of the State Archaeologist in Lexington (up to site LEI 50). About half of them show evidence of disturbance. Cultural components identified at the shelters range from Early Archaic to historic periods. Southerly and easterly aspects dominate. Shelter elevations cluster around 900 to 1,200 m amsi (Applegate 1997).

In 1929, Funkhouser and Webb published a report on the "ash caves" in north­ western Lee County along the Big Sinking Creek and Little Sinking Creek valleys. They documented six shelters and excavated the first three of these: Red Eye Hollow Shelter

(LEI), Little Ash Cave (LE2), Big Ash Cave (LE3), Cave Fork Hill Cliff (LE4), Buckner

Hollow Rockshelter (LE5), and Great Rock House (LE6). Diagnostic bifaces pictured in the report span Early Archaic to Woodland periods, but the majority are typical of

Terminal Archaic and Early Woodland forms as defined today (of. Justice 1987).

Most of the excavated Lee County rockshelters documented by Webb and

Funkhouser contained dry, stratified, ashy deposits and preservation of organics was exceptional. Webb and Funkhouser described the cultural assemblages from many of the

45 shelters as unspectacular “kitchen-midden material." Unfortunately, relic collectors,

farmers, and others had extensively disturbed many of the shelters even in the 1930s.

Knudsen, Kellar, and Simpson (1983) conducted an inventory to locate, record

and analyze cultural resources in a 2350-acre tract the Big Sinking Creek Oil Field in

Lee County. They documented 60 shelter and open-air sites with prehistoric

components, but only about half of them yielded diagnostic artifacts from surface

contexts. The archaeologists revisited two shelters previously recorded by Funkhouser and Webb in 1929 (LES, LE9) and documented an additional 28 previously unknown shelters (LE45 through LE51, LE53 through LESS, LE59 and LEGO, LE63 and LE64,

LESS, LESS through LE79). The number of rockshelter components associated with the following periods are: two Archaic, IS general Woodland, two Early Woodland, two

Middle Woodland, two Late Woodland, and five Fort Ancient. This study shows that the number of Woodland occupations at the shelters exceeds that of Archaic occupations, suggesting that perhaps Woodland occupations were more intense in this respect.

Using the cultural management plan outlined by Knudsen et al. (1983), O'Steen et al. (1991) conducted Phase II investigations at five of the shelters in the Big Sinking

Creek drainage: Little Sinking Creek (LE9), Zachariah Shelter (LE44), Cold Oak Shelter

(LE50), Big Turtle Shelter (LESS), and Pine Crest Shelter (LE70). It was determined that all the shelters qualified for nomination to the National Register of Historic Places.

Radiocarbon dates for the shelters range from Late Archaic to Middle Woodland. Lithic artifacts are composed of local materials.

Other studies conducted at Lee County shelters include Gremillion (199S), Kluth et al. (1992), Mickelson and Gremillion (199S), Mickelson and Mickelson (1997),

Sanders (1991), and Turnbow (197S, 1976, 1981).

46 2.3.2.b. Wolfe County

The shelters of Wolfe County have received a great deal of the archaeological community's attention due to the work of Funkhouser and Webb and, later, the proposed

Red River dam project. Like Lee County, Wolfe County is "in the heart of rockshelter country. " Site forms for 44 Wolfe County shelters were examined at the Office of the

State Archaeologist (up to site W 0107). Almost half of these shelters exhibit evidence of disturbance. Early Archaic to Fort Ancient occupations are noted. Southerly and easterly exposures are reported for some of the shelters. Shelter elevations in Wolfe County are similar to those in neighboring Lee County (Applegate 1997).

Funkhouser and Webb published the first survey of Wolfe County archaeological resources in 1930. As expected, Wolfe County was very rich in rockshelter sites, which typically contained ash beds and provided good preservation of prehistoric materials.

They documented nineteen shelters in the southwestern part of the county (W01 through

W019), a number of which were excavated. Lithic bifaces pictured in the report are associated with Early Archaic through Mississippian periods, but most are of Terminal

Archaic-Early Woodland affiliations. In their state-wide survey, Webb and Funkhouser

(1932:401) concluded that "unquestionably aborigines occupied [Wolfe County] in considerable numbers and over a long period of years, and left many evidences of their occupation."

A flurry of research in Wolfe County occurred in the late 1960s and early

1970s in association with the proposed Red River dam project. Fryman (1967) led a

University of Kentucky survey that identified three new rockshelters (W021 through

W023) with evidence of prehistoric occupation. Two shelters (W025 and W026) were documented by Cowan (1975) in his survey along the Red River; the latter of these is the Middle-Late Woodland Trinity Shelter. Turnbow (1975, 1976) surveyed additional forested areas of Wolfe County and located thirteen new shelters (W028 through W032A

47 and B, W033 through W037, W071, W072). Cowan and Wilson (1977) continued the work in the Red River area to determine if the Red River Gorge was eligible as a National

Historic District on the National Register of Historic Places. In their pedestrian survey of the north and south sides of the river, they documented 42 previously recorded and 16 new prehistoric sites, three of which are in Wolfe County (W036, W038, W039).

Additional research at individual shelters in Wolfe County includes Hand

(1992), Gremillion (1993), and Niquette and Hughes (1992).

2.3.2.C. Powell County

"A transition zone between the rock shelter region on the east and the mound area on the west," Powell County encompasses two culture areas defined by the topography

(Webb and Funkhouser 1932:335). Site forms for 50 Powell County shelters with evidence of human occupation were studied at the Office of the State Archaeologist (up to

PC110). At least three-fifths of these shelters show evidence of disturbance. Cultural affiliations identified at some of the shelters range from Late Archaic to historic periods; the Powell County shelters have yet to produce materials or dates from the Early and

Middle Archaic periods, as shelters in Lee and Wolfe counties have. Shelter aspects and elevations have not been consistently reported, but several shelters have southerly exposures and many are located between 1,100 and 1,240 m amsI (Applegate 1997).

In their survey of Powell County archaeological resources, Funkhouser and Webb

(1930) located two shelters (P01 and P02) in the southeastern part of the county, near the border with Wolfe County. They excavated the dry, rich deposits of the former,

Steven DeHart Shelter; diagnostic lithic bifaces pictured in the report are of Terminal

Archaic-Early Woodland affiliations. They documented four additional shelters in Powell

County during their state-wide survey published in 1932; the shelters are P04, P08,

P09, and one without a site number.

48 Twenty-seven shelters were located during surveys associated with the proposed

Red River dam project. In his survey of the proposed impoundment area, Fryman

(1967) located one shelter (P011) in Powell County. This shelter was revisited by

Cowan (1974) during his survey of the proposed dam area; Cowan also recorded the well-known Haystack Shelters (P047A and B) at this time. In the next year, Cowan

(1975) recorded four new shelters along the Red River in Powell County (P044, P067,

P068, P069). The U.S. Forest Service commissioned the University of Kentucky to sun/ey additional forested areas of Powell, Menifee and Wolfe counties in 1975;

Turnbow (1975, 1976) recorded sixteen new shelters in Powell County (P023, P024,

P052 through P054, P056 through P066). Cowan and Wilson (1977) continued the work in the Red River area to determine if the Red River Gorge was eligible as a National

Historic District on the National Register of Historic Places. In their pedestrian survey of the north and south sides of the river, they documented one previously unrecorded shelter in Powell County (PO70) and revisited three others (P067 through P069).

In their KHC-sponsored county survey, Weinland and Sanders (1977) located 39 prehistoric sites in Powell County: five of these sites had been recorded previously.

Though they focused on identification of open sites, since many rockshelters had been located earlier by previous researchers, Weinland and Sanders identified five new rockshelter sites (P086, P087, PO103, PCI 07, PO108) and revisited three shelters

(P025, P069, PO106). They examined the surface of each shelter, collecting all diagnostics and a sample of any debitage. Prehistoric materials recovered from three of the shelters (P086, P087, PO103) include biface fragments, modified flakes, unmodified flakes, chunk, cores, and potsherds. The other five shelters are associated with petroglyphs.

Other archaeological studies at Powell County shelters include Cowan (1979),

Cowan et al. (1981), Meadows (1977), and Turnbow (1981).

49 2.3.2.d. Menifee County

Menifee County is also in the core area where extensive prehistoric use of rockshelters has been documented. Over 140 shelters have been recorded since the

1930s (site forms up to number MF163). Relatively significant disturbance is evident in at least 57% of these. Though only a small proportion of the shelters in Menifee

County are associated with cultural periods, components range from Terminal Archaic to

Fort Ancient: this is similar to the pattern in Powell County, but unlike that of Lee and

Wolfe counties where earlier occupations are noted. Common shelter orientations range from southwest through east. Shelter elevations are slightly lower than those in the preceding three counties, ranging from 900 to 1,100 m amsi (Applegate 1997).

Menifee County, though it lies in rockshelter country, was "virgin territory" to

Webb and Funkhouser when they conducted their 1932 state-wide survey. They noted, however, that local residents reported ash beds and prehistoric remains in many shelters. Webb and Funkhouser returned to Menifee County in 1935 to survey more carefully the archaeological resources in the rockshelters. They published their results in 1936. The study encompassed only the southern side of the Red River, as the northern side was largely inaccessible. Webb and Funkhouser documented eleven shelters (MF1 through MF7A, B and C, MF8 through MF10), but they extensively tested only a few of these. The best known of the sites is Newt Kash Hollow Shelter (MF1); Terminal Archaic and Woodland points are pictured among the remains recovered from Newt Kash.

Jones (1936) conducted a landmark analysis of the vegetal remains from Newt

Kash Hollow Shelter. He identified a variety of edible plants in the dry deposits and fecal samples from the shelter; corn, squash, gourd, and cultivated varieties of pigweed or goosefoot, sunflower, marsh elder, giant ragweed, and possibly canary grass. In the absence of radiocarbon dating, Jones (1936:165) provisionally dated the indigenous plant material to "the pre-prairie transition about 4,000 years ago." Subsequently,

50 cultigens dated as early as the Late Archaic period have been recovered from other shelters like Cloudsplitter and Cold Oak, confirming Jones' suspicions (Cowan et al.

1981; Gremillion 1995, 1996, 1997; O'Steen et al. 1991). Archaeologists have come to learn, though, that the plants identified by Jones were not incorporated into the diet at the same time; for instance, evidence of dietary use of corn postdates by several hundred years the first documented use of sunflower, pigweed, and squash.

Like Wolfe and Powell counties, Menifee County was part of the area investigated for the proposed Red River dam project. Twenty-two shelters were documented during eight surveys in the 1960s and 1970s. Fryman's (1967) survey recorded two shelters

(MF27, MF30) in Menifee County, and Cowan's (1974) study documented Webb

Shelter, MF32. In the next year, Cowan (1975) recorded the well-known Hooton Hollow

Shelter (MF10). Turnbow (1975, 1976) recorded six new shelters in Menifee County:

MF33 through MF35 and MF53 through MF55. In a survey of the north and south sides of Red River, Cowan and Wilson (1977) located twelve new shelters in Menifee County:

MF36 through MF41A and B, MF42A, B and 0 , MF43, and MF45. Of these, Cloudsplitter

(MF36) has been investigated extensively.

In 1977, Wyss and Wyss conducted an extensive survey of the cliff lines, ridge tops, and bottom lands on the northern side of the Red River in Menifee County. The study area covered 26,000 acres, or approximately one-fifth of the Red River Gorge area.

Wyss and Wyss evaluated over 70 miles of the northeast-southwest trending cliff line.

They located 117 prehistoric sites, several of which were previously recorded, whose temporal-cultural affiliations range from Archaic to Fort Ancient. Of these, 109 are rockshelters: MF56 through MF120, MF121A-D, MF122 through MF140A and B,

MF141 through 149A and B, MF150 through MF153, MF155, MF156, MF158,

MF161, MF162, MF163.

5 1 Wyss and Wyss (1977) recovered a collection of 4404 chipped-stone artifacts

from select rockshelter, bottom land, and ridge top sites: 25 projectile points, 75

bifaces and biface fragments, 18 biface or point fragments, 167 modified flakes, 63

cores, 4002 flakes, and 54 pieces of chunk (only a sample of debitage was collected from

some sites). Notably, about 91% of the collection is chipped-stone flakes. Most of the

specimens are made from locally available cherts. Haney accounts for 83% of the

collection, St. Louis comprises 6%, and 4% of the artifacts are Paoli. An exotic chert,

Boyle, makes up the remaining 7%. Fairly equal numbers of projectile points are made of the four cherts. Other lithic artifacts are pecked stone, ground stone, limestone, and

limonite. Wyss and Wyss (1977) explain that six shelters (MF-82, MF-84, MF-90,

MF-91, MF-104, MF119) found near sources of chert had relatively high percentages

(>50% compared to 10-15% at other sites) of cortex flakes, suggesting that core reduction was at least one activity that took place in the shelters. Five rockshelters

(MF82, MF83, MF105, MF106, MF153) contained evidence that limonite or ironstone quarrying or processing had occurred there.

2.3.2.e. Contiguous Counties

Over 50 shelters have been recorded in Breathitt County, which borders Lee and

Wolfe counties to the southeast. Unlike the aforementioned counties, the majority of shelters in Breathitt County (85%) lack apparent evidence of disturbance. Cultural components identified at the shelters span the Early Archaic to Fort Ancient, but no

Middle Archaic and Middle Woodland components have been identified. Shelter elevations range from 880 to 1,450 m amsi, but most are found at 1,100 to 1,300 m amsi. The most common aspects are west, southeast, and south (Applegate 1997).

Archaeological resources in Breathitt County were first assessed during Webb and Funkhouser's (1932) state-wide survey. They documented only one rockshelter.

52 BR8, and attributed the general paucity of prehistoric sites to the terrain, rough streams, and thick forests.

Boisvert's (1981) archaeological survey of Robinson Forest in Breathitt and bordering counties documented four prehistoric sites. Three of these sites are rockshelters in Breathitt County: BRI7, BR18, and BRI9. Two are in the Clemons Fork drainage and one is in the Coles Fork drainage. Lithics recovered from the shelters include flakes and fire-altered rock. Boisvert found no diagnostic lithics during limited subsurface testing.

Sussenbach (1990) revisited the archaeological resources of Robinson Forest in

Breathitt and Knott counties. Research focused on reconnaissance to determine site distributions and land use patterns, assessment of biases in field methodology, and description of chert outcrops in the study area. Sussenbach located 234 components in the 7222-acre study area, 137 of which contained prehistoric remains.

Sussenbach (1990) divided prehistoric site locations into four categories: chert outcrops or chipping stations, single artifact locations, open-air sites, and rockshelters.

Almost half of the sites are shelters. Fifty-four shelters contained prehistoric materials, and seven shelters are of undetermined affiliation (site numbers are not given in the report). Sussenbach noted that rockshelters are more common and their locations more diverse than anticipated, despite the lack of thick sequences of resistant sandstone outcropping as cliffs in the area. Also unusual is "the near absence of disturbed rockshelters" (Sussenbach 1990:149).

Sussenbach (1990) evaluated relationships between prehistoric rockshelter use and eight variables intrinsic and extrinsic to the shelters. On average, prehistorically occupied shelters are located almost 200 m from permanent wafer sources, usually small streams. Distances to stream junctions averaged over 500 m. With respect to elevation, two shelter clusters at 1040-1,100 amsi and 1,180-1,280 amsi

53 correspond to bedrock stratigraphy. Southern and western aspects are more common

than eastern and northern exposures. Rockshelter size averages 12 m long by 4 m deep

by 2.7 m high, with floor areas ranging from 12 m2 to 144 m2. Four floor area shapes were observed; elliptical, oval, rectangular, and U-shaped. Ceiling height decreased toward the backwall in all but two of the prehistorically occupied shelters. In general, the am ount of roof fall was inversely proportional to shelter occupation in that most shelters with prehistoric remains had relatively less roof fall. Shelters with small pieces of rock fall over the floor tended to be avoided, while shelters with just large boulders showed evidence of prehistoric use?. Rockshelter modifications, the final variable, are associated with rockshelter use at a number of locations. Sussenbach

(1990) observed three types of modification: alteration of boulders, clearing of small roof debris, and fire rings.

Recovered through surface collection, shovel and trowel probes, and limited screening, artifact assemblages from all types of prehistoric sites are dominated by chipped-stone debitage. Lithic raw materials are locally abundant Breathitt, which constitutes 94% of the chipped-stone sample, and several nonlocal cherts: Boyle, St.

Louis, Newman [Haney], Brush Creek, Paoli, Lost K, and Kanawha. Among the prehistoric shelters, one yielded a triangular biface, eight had estimated artifact counts greater than 100, and two were estimated to have over 1,000 artifacts. On the whole, however, the small sample sizes for the prehistoric sites make it difficult to determine site activities and site functions.

Lithic collections from the open sites and rockshelters differ in the Robinson

Forest study area. Compared to open-site assemblages, "rockshelter collections are characterized by a higher incidence of artifacts from early in the reduction stage.

? It may be difficult to assess the relationship between roof fall and prehistoric rockshelter occupation because of historic modifications of shelters during niter mining. Sussenbach did not indicate if niter mining, which contributes debris to shelter surfaces, occurred in the past in his study area.

54 greater frequencies of cortex on tfie artifacts, and fiigher frequencies of utilized

artifacts" (Sussenbach 1990:207). While he could not specify activities or functions,

Sussenbach (1990:209) explained that "the difference in these site assemblages is

suggestive of different types of activities taking place at these site types, or different

levels of intensity of various activities."

Estill County borders the western edge of Lee County. Sixteen rockshelters are

documented at the Office of the State Archaeologist. Of these, 69% show evidence of

disturbance, while 19% are apparently undisturbed and 13% were not evaluated.

Cultural components are identified at only a few shelters and indicate Early Archaic and

Woodland occupations. Elevations range from 980 to 1,420 m amsi, but two clusters are

apparent at 1,100 to 1,240 m amsi and about 1,400 m amsi. Roughly equal numbers of

south- and north-facing shelters are documented. Webb and Funkhouser did little

research in Estill County, though they posited that rockshelter sites should be numerous

(Applegate 1997).

Jackson County borders the southwestern tip of Lee County. Like Lee, Wolfe,

Menifee, and Powell counties, Jackson County has numerous rockshelters with evidence of prehistoric occupations. Site forms for 110 shelters were examined at the Office of the State Archaeologist (up to JA185). About 70% of these shelters appear to be disturbed, 28% lack such evidence, and 2% are not described in terms of disturbance.

Components associated with Early Archaic through Fort Ancient periods, except Middle

Archaic, are represented in the sample of shelters. Shelter elevations range from 900 to

1,420 m amsi but cluster between 1,100 and 1,300 m amsi. Over half of the shelters have southerly to easterly aspects (Applegate 1997).

Webb and Funkhouser (1932:194) were surprised to find "little or no evidence of prehistoric occupation" of Jackson County, and local informants reported no sites.

Subsequent research projects, many of which were related to timber sales, located that

55 evidence. Fryman et ai. (1967) documented eight shelters in Jackson County during his

Red River Gorge survey: JAB through JA9, JA14, JA16, JA18, and JA20. Knudsen (no

date) and Knudsen et ai. (no date) located eight shelters (JA29, JA30, JA 74 through

JA76 and JA79 through JA81) in association with proposed timber sales. Ison and

Knudsen (1985) documented JA50, JA157 through JA159, and JA164 during a timber

sale assessment.

Magoffin County borders the easternm ost tip of Wolfe County. Although little

work had been conducted in Magoffin County, Webb and Funkhouser expected that the

area would be rich in prehistoric resources since it lies in rockshelter country. They

reported three shelters from this county in their 1932 publication: MAGI, MAG2, and

MAGS.

Magoffin County abounds in rock shelters, locally known as 'caves' or 'rock- houses' and we are advised that many of these shelters especially along Coon Creek and Middle Fork contain ash beds and the usual evidence of prehistoric occupation. It is apparent that this region is part of the great rock shelter culture which has been reported for Wolfe, Powell and Lee counties and well worth careful archaeological investigation (Webb and Funkhouser 1932:266).

Morgan County borders the northeastern edge of Wolfe County. Webb and

Funkhouser (1932) did not investigate Morgan County, and no archaeological resources were reported by local informants. But because the county is on the edge of rockshelter country, Webb and Funkhouser expected that evidence of prehistoric occupations would be found. Sanders and Sanders (1975) and Sanders (1976) found that evidence. They documented seven shelters with Woodland and Fort Ancient components (M037 through

M 043).

Owsley County lies south of Lee County. As “expected from its location and its topography," Owsley County yielded little evidence of prehistoric use and no prehistorically occupied rockshelters (Webb and Funkhouser 1932:333). While little work had been done, Webb and Funkhouser suggested that the county appeared "unrich" archaeologically. Since then, ten rockshelters with prehistoric materials have been

56 recorded at the Office of the State Archaeologist. Eight of these have not been disturbed

extensively and two show such evidence. Late Archaic, Woodland, and Fort Ancient components have been identified. Shelter elevations range from 920 to 1,100 m amsi.

Over half of the shelters have southerly aspects (Applegate 1997).

2.3.2.f. Summary

In summary, over sixty years of archaeological research at eastern Kentucky rockshelters has involved extensive excavations of the drier and more deeply stratified shelters, distributional studies of prehistorically occupied shelters, and reconstruction of subsistence changes based on organic remains from dry shelters. Lithic analyses from previous archaeological research projects In the study area focused largely on descriptive matters, such as chert identification and stylistic or functional type counts, with the analytical potential of lithic collections being largely unrealized.

Archaeologists have carefully tested and excavated numerous rockshelters in the study area, including Cloudsplitter, Haystack, and Rogers. Much of the prior research in the Cumberland Plateau area of Kentucky may be characterized as large-scale survey and reconnaissance. The primary purposes of the work were site documentation, assessment of disturbance, and development of management proposals. Necessarily, the survey projects have focused on site distributions. The work of Sussenbach

(1990:261), for instance, is noteworthy in its attempt to identify through multivariate analysis "significant [environmental] predictors of site occurrence."

Over 58% of the 531 shelter components examined in this study are undesignated prehistoric occupations. About 18% of the components are Woodland components and

11% are late prehistoric/Fort Ancient components. Over 6% of the components are associated with each of the Archaic and Historic periods. (Applegate 1997).

Analyses of rockshelter material collections have focused on the perishable artifacts, especially the remains of foodstuffs. Understandably, archaeologists are

57 excited about the excellent preservation of organics in many of the eastern Kentucky rockshelters, and assessment of these important collections has contributed significantly to our understanding of prehistoric plant use. Analyses of the organic remains led to a number of hypotheses regarding the role of rockshelters in prehistory, as explained in the next section.

2.3.3. EXPLANATIONS OF PREHISTORIC ROCKSHELTER USE

The summation of previous work indicates that rockshelters were used throughout prehistory in eastern Kentucky. Over the decades, researchers have advanced several interpretations about the role of rockshelters in the regional settlement systems of prehistoric inhabitants of eastern Kentucky. While the research was in its infancy, archaeologists advanced hypotheses about prehistoric utilization of the shelters. Because the issue of prehistoric rockshelter use remains unresolved, archaeologists continue to examine shelter artifact assemblages and the nature of shelter deposits.

As explained more fully in this section, archaeologists have proposed that shelters were the loci of temporary visits or were utilized as relatively sedentary base camps during certain time periods. Artifact diversity, seasonal plant indicators, raw material types, and the extent and composition of midden deposits are used most often to make the distinctions. It is recognized that this dichotomy may oversimplify the various perspectives on rockshelter use, but ideas are pigeonholed into categories for simplicity's sake. It is also realized that generalizations about rockshelter use over a long period of time and a sizable geographic space may be problematic.

Some archaeologists contend that a majority of the shelters were used on a relatively temporary basis. Such interpretations are linked to presumed rockshelter functions within subsistence-settlement systems. Within a horticultural economy.

58 shelters were used for storing crops (Struever 1973). Within a hunting-gathering economy, shelters were occupied by groups making resource forays into the uplands

(Funkhouser and Webb 1930, Webb and Baby 1957). Funkhouser and Webb (1930) suggested that those shelters with relatively thin or discontinuous ash beds, fewer artifacts, and "unspectacular" artifact types were used on a relatively temporary basis.

For instance, speaking of Brashford Shelter in Menifee County, the researchers explain

"it is probable that this shelter, like the Dark Hollow Shelter, was merely a transient site " (Funkhouser and Webb 1930:134).

Other archaeologists have argued that rockshelters in the study area were the loci of "base camps" in prehistory (Dorwin et al. 1970, Ison 1991). The definition of base camp implied in the studies consulted follows Binford's (1980) formulation of the concept. In Binford's model of settlement-subsistence patterns, a base camp is the most permanent form of habitation among foraging and collecting groups, although year-round residence did not occur. Residential base camps represent the central base or hub from which groups exploited resources on a daily basis. The types of material remains associated with base camps depend on the season of occupation and the resources exploited.

Whereas temporary rockshelter use is often associated with hunting or storage functions, archaeologists who view rockshelter use as relatively intense often associate such trends with environmental conditions or resource concentrations, be they wild or domesticated. Groups spent more time at shelters and performed a broader range of activities there. Shelter occupation was not necessarily limited to the winter season.

Because most archaeologists specify certain time intervals for which their interpretations of rockshelter use apply, the following descriptions are arranged temporally. Publications focus on the nature of shelter use during the Early Archaic

(Dorwin et al. 1970) and the Late Archaic - Early Woodland (0‘Steen et ai. 1991,

59 Railey 1991). These studies tend to deal with the nature of intersite temporal change in

rockshelter use, or changes among sites over time. Researchers like Webb and

Funkhouser discussed intrasite temporal change in shelter use, meaning changes in the

intensity of occupations within one shelter.

In the Red River Gorge area, Cowan (1976), Jefferies (1988b), and Wyss and

Wyss (1977) noted that Early Archaic remains from many shelters suggest that occupations were seasonal during this interval. This pattern presumably represents a continuation of earlier trends in rockshelter use. Although Cloudsplitter Shelter in

Wolfe County yielded stratified Early Archaic deposits, post molds, and hearths, the shelter is interpreted as a “temporary, fall-season camp, perhaps being occupied for only a few days and nights before resource availability dictated movement" (Cowan et al.

1 9 8 1 :7 4 ).

A different scenario is evidenced to the north of the Red River Gorge. Shelters in the Cave Run Reservoir area along the Licking River yielded evidence that Early Archaic occupations were of long duration or high frequency, whereas open sites were not used at all. At Deep Shelter in Rowan County, for instance, Dorwin et al. (1970) docum ented a number of distinct Early Archaic occupation levels, features, and reliance on local cherts. Dorwin et al. (1970) explained that the shelter was used intensely because of its proximity to resources and diverse habitats.

Much has been written about Late Archaic and Early Woodland shelter occupations. With respect to Adena traits and rockshelter materials only, Webb and Baby

(1957:38) contended that evidence points to transient use of shelters during Adena occupations. The Adena used these locations as hubs for winter hunting-gathering activities. “Occupation in each case apparently was temporary, the shelters being used from time to time by hunting parties" (Webb and Baby 1957:33-34).

60 It is apparent that the artifacts that have been found in the Kentucky rock shelters are generally of a type that would have been used by transient parties who lived in them while they were hunting and gathering food and dressing skins, which they probably carried to their permanent settlements. The evidences of long-time residence are lacking. For example, there are relatively few burials [but rockshelters are relatively confined spatially] and only one deposited cremation. No copper has been discovered in shelters, and only one small piece of mica, one cradleboard, and one cache of leaf-shaped blades have been found (Webb and Baby 1957:34).

In addition, Webb and Baby contended these transient inhabitants were Adena migrants from the north.

Located less than 150 miles from Chillicothe, Ohio, and the Scioto River Valley, the great center of Adena residence in the Ohio Valley, the rock shelters were occupied by the Adena Indians during the winter hunting season when they were away from their main villages. The shelters apparently were used year after year, thus producing an accumulation of refuse and ashes on the floor (Webb and Baby 1957:34).

Struever and Vickery (1973) argued that in several Midwestern areas, including eastern Kentucky, early horticultural plots were located on bottom lands along rivers and tributaries. The only reason the earliest remains of cultigens are recorded in nearby rockshelters in eastern Kentucky is because crops stored there were preserved.

The more permanent settlements were located in the bottoms. In this scenario, rockshelters functioned as short-term camps or as storage locations for preserving crops grown along the flood plains. According to Ison (1988), Brown (1984) came to a similar conclusion in his study of Ozark Bluff sites.

Another interpretation of rockshelter use suggests that the shelters were used by a mobile group of people who resided only in the shelters. Based on their study of rockshelter deposits and assemblages in the London Ranger District, Carmean and Sharp

(1995) argued that during the Woodland Period two settlement adaptations were employed in eastern Kentucky. Distinct but interacting groups of people practiced village-based sedentism and residential mobility. Bottom land sites are associated with a more sedentary lifestyle, and rockshelters are related to a more mobile settlement strategy. Shallow deposits, low feature diversity, high percentages of exotic cherts, and

6 1 low artifact densities at three shelters in Laurel County indicate temporary, short-term use by small groups. Similarly, 85% of 62 shelters in the London Ranger District yielded evidence of temporary shelter use. As they explained.

Clearly, there was interaction between groups over broad areas as the overall similarities in material culture (e.g. pottery and points) indicate, but perhaps changes in lifestyles (e.g. from nomadic to sedentary) were differentially accepted or never accepted in some areas.... a mobile Woodland settlement system , such as the one we describe, is implicit in Dunnell's (1972) interpretation of Woodland Period developments of the Fishtrap Reservoir, also located in Eastern Kentucky (Carmean and Sharp 1995:16).

According to Ison (1991), in eastern Kentucky hillslope gardening of Eastern

Agricultural Complex® crops in upland zones was associated with more intensive use of nearby rockshelters compared to bottom land locations. Garden plots associated with

Early Woodland horticultural activities were located close to shelters, which yield evidence of more intense occupation during this period. Ison (1991:218) noted a common misconception regarding shelter utilization is "that rockshelters rarely served as base camps." Ison (1991:3) indicated that "recent work in Eastern Kentucky has demonstrated that rockshelters situated considerable distances from major river bottoms did function as base camps," citing Cloudsplitter Rockshelter (Turnbow 1976,

Cowan et al. 1981) as an example.

Knudsen et al. (1983) and O'Steen et al. (1991) have commented on the nature of shelter use in the Big Sinking Creek area of Lee County. Based on a survey that located over 20 prehistorically occupied shelters, Knudsen et al. (1983:67) determined that the Big Sinking Creek area "became settled as the shift in subsistence base became more pronounced during the Archaic" and that "the same life way continued from the Archaic through the Woodland with the usual changes in lithics, the introduction of ceramics, and some social changes." O'Steen et at. (1991:157) concluded that the Terminal Archaic

® The Eastern Agricultural Complex is a collection of plants, native to eastern North America, that were domesticated by indigenous human groups. The EAC includes sunflower, Chenopodium (pigweed or goosefoot), amaranth, maygrass, and sumpweed.

62 through Middle Woodland were the periods of most intensive shelter use. Prehistoric groups were not mobile during the times when there was significant rockshelter use

(i.e., many components and/or sites). "The predominance of locally available cherts in the assemblages from all of the tested [shelter] sites strongly suggests that the Late

Archaic and Early Woodland inhabitants were an indigenous population and were not highly mobile." A wide range of activities is evidenced by the Terminal Archaic and Early

Woodland assemblages at several sites in the Big Sinking Creek area (O'Steen et al.

1 9 9 1 ).

Railey (1991), on the other hand, indicated that in regions of the Cumberland

Plateau, the Late Archaic to Early Woodland transition is associated with a shift in settlement strategies, including dispersal from nucleated open sites and increased use of rockshelters. Late Archaic occupations in the Red, Licking, and Little Sandy River drainages are typified by substantial, "nucleated sites or base camps,' characterized by thick midden deposits and located primarily within alluvial environments" (Railey

1991:97). Examples of such sites are Bluestone, Seldon Skidmore, Zilpo, Grayson, and

CR73. Ison (1988:215) agrees that, "based strictly on the occurrence of Cogswell points, [Terminal Archaic-Early Woodland Cogswell phase] settlement appears to be primarily along the major drainages of northeastern Kentucky." According to Railey

(1991:97), "this pattern of aggregated settlements remained rather stable until ca.

1,000 B.C." Depopulation or abandonment of the "base camps'" occurred during the

Terminal Archaic Cogswell Phase, and by the Early Woodland Period, "groups in the region shifted to a dispersed mode of settlement that would persist throughout the Early

Woodland period and into its immediate aftermath" (Railey 1991:97).

The dispersal from lowland base camps was accompanied by “intensified utilization of upland rockshelters" as evidenced at Cloudsplitter, Newt Kash, and Hooton

Hollow (Railey 1991:97). Not only does the number of Early Woodland rockshelter

63 components increase in parts of the Cumberland Plateau, but Early Woodland occupations

at many shelters are more “intense." Railey (1991:97) reported that at the Conley-

Greene Rockshelter (EL4), for example, initial occupations were less intense than

occupations after 500 B.C., and that the shelter probably served as “a home base for a

nuclear family or similarly small domestic unit."

Railey (1991:97) suggested that the Late Archaic to Early Woodland transition

in parts of the Cumberland Plateau witnessed "considerable fragmentation of Late

Archaic macrobands Into smaller social units with correspondingly smaller extractive

territories during the Early Woodland Period." Group fissioning led to territorial

constriction that, coupled with the shift to horticulture, resulted in settlement dispersal

and increased use of rockshelters. However, these changes in settlement strategies were

not necessarily coupled with a shift in the degree of sedentariness. As Railey (1991:97)

indicated, “Early Woodland inhabitants of the region were perhaps just as sedentary as

their Late Archaic ancestors, but, rather than congregating in large base camps, they

apparently resided in smaller, scattered habitations" at shelters and elsewhere.

Wyss and Wyss (1977) examined temporal patterns in site location in the Red

River Gorge. Their results are comparable to Railey's (1991) results. Generally

speaking. Archaic sites tend to be located on the bottom lands while Woodland and Fort

Ancient sites are predominantly rockshelters.

Although Funkhouser and Webb's interpretation of the occupational nature and

history of the eastern Kentucky rockshelters changed over the course of their careers,

they were the first to conclude that the intensity of use at some rockshelters changed over time. While Railey (1991) and Wyss and Wyss (1977) cited evidence that

shelters were used less intensely during the Late Archaic than the Early Woodland,

Funkhouser and Webb contended that, within shelters, pre-pottery [Archaic] occupations are more intense than pottery [Woodland] occupations.

64 Based on their excavations in Lee and Wolfe counties, Funkhouser and Webb

(1929, 1930) concluded that the major occupations at shelters like Red Eye Hollow and

Dillard Stamper involved more intense use by pre-pottery groups followed by less intense use by pottery-using groups. Speaking of the Lee County shelters, Funkhouser and Webb (1929:108) argued that "two distinct cultures are represented - one which may be very old and which is responsible for the ash beds, the hominy holes, the flints and the burials; the other a later culture which is represented by the pottery and the fabrics." Referring to the shelters investigated in Wolfe and Powell counties, the researchers argued that

the surface material may represent a later and entirely different culture or [many of the shelters] may show a transition from a very primitive to a higher stage of culture in the same people. However, the occupation seems to have been continuous from the lowest levels of the ash bed to the comparatively thin surface layer containing pottery (1930:300).

Relatively little has been written about the Middle Woodland, Late Woodland, and

Fort Ancient shelter occupations in the study area. Middle Woodland components are identified at a number of shelters in the Red River and Big Sinking Creek areas.

According to Railey (1996), groups used shelters both temporarily and intensely during

Middle Woodland. Late Woodland shelters such as Haystack and Rogers may have been used year-round, implying intensive use (Cowan 1979). Following Woodland occupations. Fort Ancient components at shelters are the most numerous at a large sample of shelters in the study area (Applegate 1997).

The brief summary of research about prehistoric rockshelter use shows a lack of consensus among archaeologists. Early Archaic use of shelters is characterized as temporary in the Red River Gorge area, but Early Archaic shelters in the Cave Run

Reservoir area apparently were used more intensely. The differences may be explained by geographic factors. However, disagreement about shelter use within one area like the

Red River Gorge also exists. Some archaeologists have contended that shelters were used

65 more intensely beginning in the Late Archaic period and continuing at least through the

Early Woodland period. Others have argued that Late Archaic shelter use was temporary, while shelter use intensified during the Early Woodland. Intrasite evidence also points to intense pre-pottery [Archaic] occupations and temporary use by pottery-using

[Woodland] groups at some shelters.

These ideas about prehistoric rockshelter use in eastern Kentucky set the stage for the analysis and interpretation of lithic assemblages from Cold Oak and Rock Bridge shelters. The question of Early Archaic shelter use cannot be addressed based on these shelters, but the Late Archaic-Early Woodland issue is a relevant issue. Do the Gold Oak and Rock Bridge assemblages indicate diachronic differences in shelter use at this important transition? What is the nature of diachronic differences? Do the assemblages support the interpretation of Railey (1991b), of Knudsen et al. (1983) and O'Steen et al. (1991), or of Funkhouser and Webb (1929, 1930)? What of patterns of shelter use within the Woodland period? In preparation for the hypothesis testing aspect of this research, the next two chapters describe the geology and archaeological stratigraphy of the sites and the lithic assem blages from the two shelters.

66 CHAPTER 3

COLD OAK AND ROCK BRIDGE SHELTERS: GEOLOGY AND ARCHAEOLOGICAL STRATIGRAPHY

Chapter 2 characterized the environment of the study area, outlined the culture history of eastern Kentucky, and described previous archaeological research at rockshelters in the study area. The record of prehistoric shelter use in eastern Kentucky is substantial. Occupations as early as the late Paleoindian period have been documented, and human use of shelters continued through the historic period. Within the core of rockshelter territory, the current study area, shelter components span the Early

Archaic through historic periods. It is unlikely, though, that the nature of shelter use was similar over this interval. Is this the case at Rock Bridge and Cold Oak shelters as w ell?

In order to understand the nature of human occupations at Rock Bridge and Cold

Oak shelters, it is necessary to consider the geology and archaeological stratigraphy of the sites. That is the goal of this chapter, which begins with a discussion of the formation and depositional history of rockshelters in general. Cold Oak and Rock Bridge shelters are then described in terms of stratigraphy, features, and artifact assemblages.

67 3.1. THE GEOLOGY OF ROCKSHELTERS

Rockshelters are "naturally formed recesses within bedrock" or “shallow niches

or ledges under overhanging bedrock" (Waters 1992:240). They usually form In

sedimentary rocks, especially In limestone or sandstone. Compared to caves, the extent

to which rockshelters recess Into the bedrock Is limited.

Figure 8 Illustrates the main components of a typical rockshelter. The floor,

backwall, and celling define the physical space within a shelter. "The edge of the bedrock overhang of a rockshelter Is called the brow. Directly below the brow, on the lower ledge of the rockshelter. Is the dripllne" (Waters 1992:241). Debris piles commonly form below the dripllne on the shelter floor. Downslope and outside the shelter, talus and other colluvlum accumulate over time.

3.1.1. ROCKSHELTER DEVELOPMENT

The stages of rockshelter development and the processes responsible for the formation of shelter cavities depend to a great extent on the nature of the bedrock and the topographic setting of the shelter. Generally speaking, "the evolutionary sequence Is critical In determining ... when the shelters were first available to human occupants, ... what the physical environment was like within the confines of the shelter and Its

Immediate environs," and the distribution of artifacts and features produced by human occupants (Donahue and Adovaslo 1990:249).

According to most researchers, hydrologie processes are primarily responsible for shelter development, whether In clastic or nonclastic sedimentary rocks. Fluvial, lacustrine, and marine process are responsible for the erosion of most shelter recesses

(Schmid 1963). Hydrologie processes responsible for the formation of rockshelter

68 cavities include erosive stream scour at cliff bases, lateral stream erosion, wave action, groundwater flow, and cryoclastism or frost weatfiering in temperate regions (Barton and Clark 1993, Donahue and Adovaslo 1990, Nash 1993, Schmid 1963). Solution is important in the development of cavities in limestone and dolostone bedrock. In order for hydrologie processes to form shelter recesses, joints and/or faults must be present in the bedrock to allow for the movement of water.

Differential weathering of bedrock is another important factor in the development of rockshelters. “Rockshelters ... tend to result from differential erosion rates ... or weathering over a restricted area (e.g., and especially the formation of an overhang or cornice through differential erosion of relatively soft [less resistant to weathering] strata overlain by or interstratified with harder, more resistant ones)”

(Barton and Clark 1993:34).

Donahue and Adovaslo (1990) divided the evolution of sandstone shelters into two stages, (1) development of an overhang, and (2) sedimentation; the latter is covered in the next section. In the initial stage of shelter development, a combination of hydrologie processes and differential weathering form the shelter cavity. At least five conditions are necessary for the initial development of sandstone rockshelters (Donahue and

Adovaslo 1990). Vertical exposures of sandstone and conglomerate outcrops or cliffs must be present, and the sandstone unit must be of adequate thickness. The sandstone must be underlain by or interbedded with relatively less resistant rocks such as shale;

Donahue and Adovaslo (1990) interpreted this condition as "critical." The outcrops must be located in youthful stream valleys, and the sandstone outcrops must be located on valley slopes or walls. In the eastern United States, the conditions were met during the late Pleistocene-Holocene epochs.

Donahue and Adovaslo (1990) indicated that the development of overhangs can be a rapid process. For instance, an artificially diverted stream about 1 m wide and 0.2 m

69 deep in Mississippi carved a 0.5 m overhang in only 26 years (Donahue and Adovasio

1990:234). Streams typically found in minor tributary valleys, then, are capable of producing shelters in relatively short periods of time. Knox (1983) argued that most river valleys in the southeastern United States were cut between 6,000 and 4,000 years ago, but Donahue and Adovasio (1990) contended that 10,000 to 8,000 years ago is a more accurate estimate.

Rockshelters are dynamic in that their morphology, dimensions, and contents change constantly. For example, as the overhang recedes due to weathering and erosion, the dripline recedes and sediments blanket the floor (Waters 1992). The cliff lines in which shelters form also evolve, with colluvial deposits accumulating below rockshelters recording cliff line retreat (Donahue and Adovasio 1990). A general pattern of rockshelter development is described by Farrand (1985:23) and is illustrated in Figure 9:

The general pattern involves deepening of the shelter as the backwall is attacked by weathering, either freeze/thaw or solution or both, because of the damper microclimate found there. Eventually the brow or part of the ceiling of the shelter collapses because of the increasing lack of support. Thus, large blocks of rock, several meters across, fall onto the shelter floor. Since the brow is particularly susceptible to such collapse, a reduction of the depth of the shelter results. During this time, sediments of varied provenience have been accum­ ulating on the shelter floor, and now they protect the bedrock of the floor from further attack by weathering or erosion. This cycle repeats itself as the weathering of the back wall continues, but at a slightly higher level, creating a step on the bedrock floor. Ultimately collapse occurs again, and so forth, leaving a buried bedrock floor that rises in steps. In the final stages the overhang is progressively reduced, and the shelter is choked with its own sedim entary fill to the point where it merges with the normal talus/colluvial slope of the valley side.

3.1.2. ROCKSHELTER DEPOSITION

Sediments are "any particulate matter on the surface of the earth that has been deposited by some process under normal surface conditions" (Stein 1985:6). Sediments

70 are distinguished from soil in that "soils develop in sediments near the surface of the earth through weathering under the influence of plants, other biological elements, and atmospheric conditions. Soils exhibit vertical differentiation within the sediments, with horizons reflecting changes in mineralogy, texture, and chemistry" (Stein 1985:6).

There are several common aspects of rockshelter sedimentation. In general, rockshelters have "a consistently aggradational depositional regime, a slow rate of deposition, and the protection of deposits (and their "cultural content") from erosion, weathering, and other forms of alteration " (Barton and Clark 1993:44). Barton and

Clark (1993:34) suggested that ""a considerable amount of deposition in rockshelters is episodic and catastrophic in nature." Similarly, Farrand (1993) warned that archaeologists cannot assume rockshelter deposition is a continuous process, and he described how erosion or nondeposition may produce unconformities in stratigraphie successions. According to Donahue and Adovasio (1990:235), sandstone shelters may experience “continuous sedimentation," but the rate varies considerably. Barton and

Clark (1993:44) described rockshelter depositional histories as "complex and idiosyncratic." Shelter reuse, especially reuse of those limited areas within shelters amenable to human occupation, usually results in superposition and disturbance of cultural materials and complex stratification of deposits (Barton and Clark 1993).

Older deposits may be located outside the current dripline of a shelter due to evolution of the shelter through brow fall and retreat of the upper cliff face (Waters 1992).

The types of deposits that accumulate in rockshelters depend on a number of factors. "The sedimentary ensembles within ... rockshelters may vary because of bedrock variability, exposure, local relief, size and shape, and, of course, the intensity of human habitation" (Farrand 1985:23). For example, chemical precipitates such as flowstone and dripstone are common in limestone shelters but rare in sandstone shelters. On the

7 1 other hand, sand is a common component of sandstone shelter deposits but is less prevalent in the sediments of limestone shelters.

Deposits that accumulate in rockshelters derive from a number of sources.

Sediment sources are classified in a number of ways, most commonly dichotomized as internal versus external or as cultural versus noncultural. Noncultural sediment sources may be geomorphic or biological. One of the earliest references to the difference between internal and external sediments is in Ford and Cullingford (1976:51). They distinguished between “endogenetic sediments" and "exogenetic sediments," referring to sediments deriving from processes internal and external, respectively, to the shelter or cave. Rockshelter deposits are a highly variable mixture of endogenous and exogenous sediments that accumulate at different rates (Barton and Clark 1993, Waters 1992).

Farrand (1985:22) noted that sediments are "the result of all kinds of deposition at a given site; they are deposited by natural geological processes, by the activity of animals, and by the activities of man during his habitation of the site."

With few exceptions, endogenous sediments are noncultural, since they derive from the shelter itself. However, one might argue that endogenous sedimentation processes are cultural if the processes are initiated or accelerated by human intervention. For instance, the burning of fires within shelters may lead to or increase the rate of bedrock deterioration and deposition of weathered bedrock material.

Nevertheless, endogenous sediments include elastics and nonclastics deriving from physical and chemical weathering of the shelter ceiling and walls. Silt, sand, and

éboulis are the most common forms of clastic sediments, and these materials occur in both sandstone and limestone shelters. Éboulis is "coarse angular rubble" (Waters

1992:242). Nonclastic sediments include chemical precipitates of carbonates that, in limestone shelters, may form stalactites, stalagmites, and flowstone deposits. Chemical weathering in the presence of water may also lead to clay illuviation (deposition of a

72 zone of leached clay), clay éluviation (removal of clay by leaching), and mineral precipitation (Barton and Clark 1993).

Internal geomorphic processes contributing to sediment deposition in shelters are frost shattering or cryoclastism, dripping water, hydration, thermoclastic weathering, consolidation or brecciation, precipitation, roof exfoliation, wall exfoliation, and various mass wasting processes (Bar-Yosef 1993, Koetje 1993, Nash

1993). Some researchers refer to the large debris produced by these internal geomorphic processes as éboulis. Éboulis materials differ from lithic artifacts in that the former lack negative dorsal bulbs and well-defined platforms^ (Nash 1993).

In sandstone shelters, the dominant endogenous processes leading to sediment deposition are rockfall and attrition (Donahue and Adovasio 1990). During the early history of a shelter, rockfall is the predominant process contributing to shelter deposits. Attrition fills in the interstices between rockfall debris.

Rockfall, or the free fall of rock from shelter roofs and walls in response to gravity, is an active erosional-depositional process at shelters. Rockfalls result from water flow, biological processes, and frost weathering at joints and bedding planes in the shelter bedrock. Material may fall from the roof or the walls of the shelter. Slab failure, rock avalanche, and rock free-fall are forms of rockfall in shelters. Rockfall contributes relatively large fragments to the sediments of shelters, and the sediments underlying rockfall debris are often deformed. Rockfall is more common during the initial period of shelter development, when debris tends to be large, equant, and angular.

Over time, rockfall debris becomes more tabular in shape and smaller in size. In conjunction with the retreat of the shelter overhang, the size grade of rockfall debris decreases from the shelter apron toward the backwall (Donahue and Adovasio 1990).

9 Negative dorsal bulbs refer to concave depressions on one side of a piece of debris; a negative bulb on the dorsal side of one piece of debris indicates where another piece of debris was detached previously. A platform is the surface where stress is applied to a rock in order to break it.

73 Attrition, the granular breakdown of shelter bedrock due to mechanical and

chemical weathering, is an important sedimentation process in sandstone shelters that,

in and of itself, leads to the accumulation of massive, homogeneous, well-sorted deposits

of sand. The sand, which is often bioturbated, accumulates at the edge of and within the

shelter, filling in gaps between éboulis and creating a horizontal surface over time

(Donahue and Adovasio 1990).

Farrand (1985) discussed other endogenous processes that form shelter

sediments. Cryoclastism creates large sediments like frost slabs and spalls, or it results

in grain-by-grain sedimentation. Collapse produces large blocks and shattered

fragments, the latter of which may resemble artifacts of lithic production (see Nash

1993). Solution processes cause hydration spalls, grain-by-grain disintegration, and

deposition of dripstone and travertine.

Turning to external sediment sources, material may derive from cultural

processes, biological processes, and geomorphic processes. Humans introduce a wide variety of exogenous materials to shelters. Cultural components of shelter sediments

include artifacts, lithic debris, mineral soil, garbage, body wastes, plant matter, and

animal products (Butzer 1982, Farrand 1985). Humans introduce to shelter sediments

both altered and unmodified lithic materials. According to Butzer (1982), excluding

roof fall, imported lithics often comprise the majority of shelter sediments larger than

2 mm in size. Mineral components of shelter deposits may be introduced by humans on their feet, on the fur and hides of food animals, in the shells of mollusks carried to a shelter, and as fecal material. Human use of plants for food and raw materials adds to shelter deposits a variety of plant materials, including carbon, "organic colloids, amino acids, cellulose, resins, phosphates, nitrates, potash, and manganese" (Butzer

1982:81). In a similar manner, human use of animals for food and raw materials contributes to shelter deposits "protein, bone, shell, horn, feathers, tissues, and feces,"

74 which often decompose and add "bone phosphorus, calcium, nitrogen or potassium

compounds, organic acids, carbonates, silica colloids" and carbon (Butzer 1982:81).

Humans contribute to sediment deposition in shelters by importing inorganic and organic

materials and by intentionally or unintentionally moving objects (Koetje 1993).

Biological sources of shelter sedimentation are diverse. Carnivores, insects,

worms, birds and rodents may introduce or remove material from rockshelters.

Exogenous biological sediments include dung, castings, mineral soil, and organic

material such as bone and vegetal debris left from animals' meals. Skeletal elements and

tissue may be incorporated into shelter deposits upon an organism's death inside a

shelter. Insects that build mud nests within shelters introduce fine-grained sediments,

and the decomposition of bird nests contributes organics and fine-grained sediments.

External geomorphic processes that introduce material to rockshelter deposits

include aeolian transport, fluvial deposition, colluvial deposition due to gravity, mass wasting processes, and glacial transport (Bar-Yosef 1993). Winds deposit well-sorted and cross-bedded layers of sand and silt in shelters. Aeolian deposits are not as common

in temperate climates as in arid regimes. Flowing water introduces flood deposits, karstic imports, and surface runoff deposits, which tend to be fine grained and well sorted. Colluvial processes deposit debris of mixed size (often including large particles) and result in the deposition of talus or scree in poorly sorted shelter strata. Colluvial deposits in shelters may indicate relatively rapid deposition (Barton and Clark 1993).

Solifluction may deposit hillslope sludge in shelters.

In sandstone shelters, the dominant exogenous processes leading to sediment deposition are flooding and sheetw ash (Donahue and Adovasio 1990). Following rockfall and attrition, flooding continues to contribute sediments until the stream level drops well below the shelter's elevation. When reenterants in the cliff line develop, sheetwash begins to contribute sediments to the shelter floor, especially on the lateral margins.

75 and continues to deposit material as the cliff overhang retreats. Flooding produces subhorizontal, silt and clay deposits that may be massive or layered. Differences in stream characteristics led to variations in flood deposits in rockshelters. Sheetwash deposits form by the flow of water within or outside the shelter. This process produces well-sorted laminae and poorly sorted unlaminated deposits that tend to have a sloping attitude. Rates of sheetwash deposition in sandstone shelters are often higher than attrition rates (Donahue and Adovasio 1990).

3.1.3. DISCUSSION

Knowledge of rockshelter formation and sedimentation processes is helpful because it allows one to make predictions and advance explanations about cultural materials recovered from shelters. For example, one might predict, based on the principles of shelter deposition, that older cultural deposits would be located at or outside the dripline. This may explain why relatively few Early Archaic and Middle

Archaic components are found at rockshelters as few archaeologists excavate beyond the dripline.

One cannot separate artifacts from ecofacts without an understanding of the formational and depositional history of a shelter. If one can estimate the rate of deposition at shelters, artifact densities per stratum or occupational episode become more meaningful culturally. An understanding of rockfall. attrition, and other shelter processes allows one to distinguish culturally bounded features from deposits that accumulated between rockfall debris. As a final example, one might predict that in sandstone shelters, artifacts found within zones of sand and rockfall debris may represent relatively early occupations at the shelters.

76 3.2. COLD OAK SHELTER

Cold Oak Shelter (LE50) is in an eastern-facing sandstone cliff above an intermittent stream that drains into Cold Oak Hollow, a tributary of Sinking Caves Creek in the Kentucky River drainage of Lee County (Figure 10). Measuring 40 m long and 15 m wide with a ceiling height of 30 m, the entire shelter falls within the drip line. The area of usable space measures about 20 m by 10 m. Deposits are at least 90 cm deep, but excavations were not carried to the shelter's bedrock floor.

Cold Oak clearly is a multicomponent site, yielding evidence of Terminal Archaic and Woodland occupations. Many of the deposits at Cold Oak shelter are dry, and preservation of normally perishable materials is exceptional. Roof breakdown limits the usable area of the shelter to the central 200 m^. Digging by relic collectors has seriously impacted the shelter, the "deep lootholes and high backdirt piles [creating] the effect of a lunar landscape" (O'Steen et al. 1991). Due to the multiple occupations and disturbance, the stratigraphy at Cold Oak Shelter is very complex.

Knudsen et al. (1983) originally recorded Cold Oak Shelter during their sun/ey of the Big Sinking Creek Oil Field. Archaeological excavations at Cold Oak Shelter began in 1984 under the direction of O'Steen and Ledbetter of Southeastern Archaeological

Services. The primary goals of this research were to assess the integrity of the disturbed shelter and to evaluate the shelter's eligibility for National Register status.

Researchers excavated an east-west trending trench, oriented perpendicular to the backwall, in the center of the shelter from the backwall to the drip line. They divided the trench into seven 1x1m units. Excavation proceeded according to natural strata, despite the complex nature of the deposits. O'Steen et al. (1990) concluded that the shelter retained sufficient integrity and research potential to make it eligible for

National Register status.

77 Gremillion resumed excavations at Cold Oak during the summer of 1994; the author was a member of the field crew. Documentation of undisturbed deposits and recovery of subsistence remains were the fundamental goals of the project. Gremillion laid out a 5 X 1 m trench, divided into five 1x1m units, south of and parallel to the

1984 trench (Figure 11). Excavation proceeded by arbitrary 5 cm increments, and distinct sediment zones within each level were excavated and bagged separately. Features were also excavated separately.

3.2.1. STRATIGRAPHY AND DEPOSITIONAL HISTORY

Ashy deposits, discontinuous and truncated beds, and subtle differences between feature and nonfeature fill characterize the sediments at Cold Oak Shelter. Features were found in all zones. Ison (1988:206) suggested that the “numerous discontinuous strata identified at the site may represent short term phenomena such as hearth cleaning and the disposal of meal refuse." Five cultural-stratigraphic zones were identified during the 1984 and 1994 excavations at Cold Oak Shelter (Gremillion 1995, Ison 1988,

O'Steen et al. 1991). Figure 12 shows a stratigraphie profile of the south wall of the

1994 trench.

Zone 1 represents a combination of disturbed deposits resulting from illegal excavations and historic use of the shelter. The loose, sandy loam and ash layer contains bovine dung, leaves, axe wood chips, and other plant materials. The thickness of Zone I varies across the shelter from 10 cm to 65 cm, and the lower portions are "strati- graphically continuous with Zone II but impacted by disturbance" (Gremillion 1995:4).

Averaging 20 cm in thickness. Zone 11 represents prehistoric deposits of Early

Woodland and early Middle Woodland affiliations. In general. Zone II is "a powder dry, ashy loam with numerous light colored ash lenses and features scattered throughout"

78 (O'Steen et al. 1991:80). Gremillion (1995) subdivided Zone II into three subzones based on sediment color, texture, and inclusions. Limited in lateral extent and stratigraphically uppermost. Zone lia is a thin, brown, compact sandy loam that may be partially disturbed. Also limited to the central portion of the excavation block, “Zone lib is a loose, grayish brown loam with a substantial admixture of ash” (Gremillion

1995:4). Zone lie is an ashy loam with "numerous features, burned areas, and shallow basins" (Gremillion 1995:4).

Calibrated radiocarbon dates from Zone II are 260 B.C. (2210 ± 60 B.P.) and

520 B.C. (2470 ± 60 B.P.) for features, A.D. 40 (1950 ± 50 B.P.) for Zone lia, 110

B.C. (2060 ± 60 B.P.) for Zone lib, and 240 to 640 B.C. (2190 ± 80 to 2590 ± 90

B.P.) for Zone lie (Gremillion 1995, O'Steen et al. 1991). Diagnostic artifacts recovered from Zone II are a Jacks Reef Pentagonal point, a triangular point, and plain and cordmarked limestone-tempered ceramics. While the radiocarbon dates suggest

Early Woodland and early Middle Woodland occupations, some artifacts indicate a Late

Woodland or later component in Zone II as well (Gremillion 1995, Ison 1988, O'Steen et al. 1991).

Averaging 14 cm thick. Zone III is characterized as “a compact sandy loam with numerous ash lenses" (Gremillion 1995:4). Artifacts and radiocarbon dates (760 to

950 B.C., 2710 ± 60 to 2900 ± 100 B.P.) indicate a Terminal Archaic affiliation

(Gremillion 1995). This zone is subdivided by O'Steen et al. (1991) into two subzones.

Zone Ilia and Zone III. Zone Ilia, dated to 980 B.C. (2930 ± 20 B.P.), is a thin vegetal deposit of "nut fragments, small seeds, grasses, leaves, wood, and bits of discarded cordage" (Ison 1988:208). A Cogswell point, cultigen seeds, wooden tools, and leather fragments were recovered from Zone Ilia. Zone Ilia may be an example of the vegetal beds discovered at other shelters in the study area^° (O'Steen et al. 1991). Funkhouser and

For example, Red-Eye Hollow (LEI), Big Ash Cave (LE3), Worth Creech Shelter (W02), Green Gentry Shelter (W 014), Steven DeHart Shelter (POI), and Newt Kash Shelter (MF1).

79 Webb (1929, 1930) and Webb and Funkhouser (1936) interpreted strata of this type as sleeping mats or layers of plant material laid down to stabilize the ashy surfaces of the shelters. The underlying Zone 111, dated to 880 B.C. (2830 ± 60 B.P.), is "a series of crossbedded and irregular ash and sandy loam lenses" (O'Steen et al. 1991:80).

Artifacts recovered from Zone III include Cogswell and Wade points, bone, wood, debitage, worked and un worked gourd rind, and squash seeds. Investigators found no ceramics in Zone III deposits excavated during both the 1984 and 1994 seasons.

Cultural remains recovered from Zone IV suggest a Terminal Archaic occupation.

The loose, sand to sandy loam deposit contained pockets of ash and decaying sandstone.

Thicknesses ranged from 10 cm to 15 cm in the 1984 trench. Features originating in this stratum include small pits and post molds. Excavators recovered bone, shell, nuts, wood, and debitage from the deposits. During the 1984 season, workers only encountered

Zone IV deposits in the two easternmost units. Zone IV deposits were not distinguished from Zone III strata during the 1994 field season.

Though not completely sterile. Zone V yielded no diagnostics and few artifacts.

Gremillion (1995) suggested a tentative Terminal Archaic assignment. The "saprolitic" deposits of Zone V are sandy with decaying sandstone. The few pieces of debitage and charcoal found in Zone V may have sifted down from overlying deposits. Zone V is up to

55 cm thick (O'Steen et al. 1991).

The developmental history of Cold Oak Shelter probably began with the formation of the recess by stream erosion during the late Pleistocene-Holocene epochs (Donahue and Adovasio 1990). The earliest depositional sequence is unknown due to the limited vertical extent of archaeological investigations. The lowermost identified zone. Zone V, consists almost exclusively of sand that likely derived from attrition of the sandstone walls and ceiling. Overlying zones contain a combination of sand from attrition, ash from human fires, and various organic materials introduced by humans. Rocky inclusions of

80 sandstone likely derived from roof fall, while humans introduced limestone and chert

clasts.

3.2.2. CULTURAL FEATURES

Gremillion (1995) excavated 44 Cold Oak Shelter features during the 1994

season, four of which had been encountered during the earlier excavations (Table 1).

Feature functional classifications are based on artifactual remains as well as contextual

evidence. Temporal designations are assigned using artifactual content, radiocarbon

dates, and stratigraphie context. Thirty-two of the features identified during 1994

yielded lithic artifacts. During the 1984 excavation season at Cold Oak, O 'Steen et at.

(1991) identified 30 features, 26 of which they described in the field report.

Feature 38 is identified as a historic hearth (Gremillion 1995). The feature was

encountered in Zone I deposits. One unintentionally heated flake and one pot lid recovered

from the feature support the hearth designation.

One feature is associated with Middle Woodland occupation of the shelter.

Gremillion (1995) interpreted Feature 31 as a hearth or hearth cleanings. Wood charcoal from the feature was radiocarbon-dated at A.D. 4 (1910 ± 50). Among the nine

lithic artifacts recovered from Feature 31, two are fire-altered sandstone and two are thermally altered cherts, supporting the hearth interpretation. Feature 31 also fits the profile of a secondary deposit resulting from manual pick-up cleaning as defined by

Nielsen (1991a).

Gremillion (1995) and O'Steen et al. (1991) identified a total of 40 Early

Woodland features at Cold Oak. Functional designations for the features are eleven pits, three of which may be storage pits, one grass-lined pit, two ash-filled pits, six pits or post molds, two post molds, one "ash bed" comparable to features described by

8 1 Funkhouser and Webb (1929), two hearths, one pit/earth oven, one pit and intrusive

post combination, one grass-lined pit or post, one charcoal concentration, and eleven

indeterminate or postoccupation features (Gremillion 1995; O'Steen et a. 1991).

Lithics were recovered from 20 of the 26 Early Woodland features excavated in

1994 (Table 2). The lithic artifacts confirm some of Gremillion's (1995) functional

designations but contradict others. Of course, functional designations based on a variety

of evidence are more reliable, especially since lithic sample sizes tend to be rather

small. Lithics were not recovered from six Early Woodland features: 36, 40, 42, 44,

50, and 68.

Feature 8, identified as a pit or hearth, yielded 27 lithics for the third highest

feature density of 1.59 lithics/liter. The hearth interpretation is supported by one piece of fire-altered sandstone, one pot lid, and seven thermally altered cherts. O'Steen et al.

(1991) placed Feature 8 with the Terminal Archaic features.

Feature 10, interpreted as a post mold by Gremillion (1995) and O'Steen et al.

(1991), contained only seven lithics for a density of 1.00 per liter. Five flakes and two pieces of chert shatter were recovered. Two unintentionally heated flakes suggest that

Feature 10 was a hearth or hearth cleanings.

The density of lithics from Feature 22 is relatively low, 0.27 per liter, with 14 specimens recovered. Although most of the lithics are flakes from early- and late-stage lithic reduction, the low density supports the interpretation that the feature was used as a pit. Only one chert artifact was heat treated, and one pot lid was recovered.

Only one primary flake came from Feature 34, giving a density of 1.00/liter.

Because that specimen shows evidence of unintentional thermal alteration, this feature may have functioned as a hearth or may represent hearth cleanings. Gremillion (1995) interpreted this feature as a postoccupational disturbance.

82 The three lithics from Feature 35 yield a density of 0.75/liter. One broken flake, one piece of fire-altered sandstone, and one piece of fire-altered limestone derived from this feature. Because all lithics show evidence of heating. Feature 35 may

have been a hearth or a deposit of hearth cleanings. Gremillion (1995) interpreted

Feature 35 as a looter’s shovel probe.

Late-stage debitage and fire-altered rock are the predominant lithic artifacts from Feature 37. Lithic artifact density is 0.97/liter. Two of the chert artifacts are unintentionally thermally altered. The fire-altered sandstone (n=2) and limestone

(n=1) specimens are relatively large. Feature 37 may be a hearth or hearth cleanings, but comparison with Nielsen's (1991a) experimental data indicate it is a secondary deposit resulting from manual pick-up cleaning. Gremillion (1995) classified Feature

37 as a disturbed feature. A radiocarbon date of 389 B.C. (2170 ± 70) was obtained from charred nutshell in Feature 37 (Gremillion 1995).

Two unaltered flakes, one heat spall, and three pieces of fire-altered sandstone derived from Feature 41. The density of lithics is 0.33/liter. A hearth or hearth cleanings interpretation for this feature seems plausible based on the lithic remains, but Gremillion (1995) questioned if this anomaly is a cultural feature.

Feature 43 was interpreted as a pit (Gremillion 1995). Only two flakes came from Feature 43 fill. One shows evidence of unintentional heating. Lithic density for this feature is 0.36/liter. While the sample size is very small, the thermally altered specimen suggests that Feature 43 may have been a hearth or may represent hearth cleanings.

That Feature 49 is a hearth deposit is supported by the eight pieces of large fire- altered sandstone and limestone recovered. Two unaltered flakes were also found in the feature fill. The density of lithics is 1.43/liter. Gremillion (1995) classified Feature

49 as a refuse dump or redeposited looting material.

83 A feature function is not given for Feature 51 (Gremillion 1995). Three flakes derived from Feature 51 fill. One exhibits indicators of unintentional heating, suggesting tentatively that the feature may be a hearth deposit. At 1.50/liter, the lithic density for this feature is relatively high. Gremillion (1995) reported that wood charcoal from

Feature 51 yielded a radiocarbon date of 196 B.C. (2060 ± 60).

Feature 52 yielded seven flakes, mostly from the early stages of reduction, and one small piece of fire-altered sandstone. Two flakes are thermally altered, one intentionally and one unintentionally. This feature was interpreted as a hearth or hearth cleanings based on the lithics, but the lithic artifacts do not refute Gremillion's (1995) pit designation either.

Feature 53 produced five flake specimens for a density of 0.69/liter. Based on

Nielsen's (1991a) findings, this feature may represent a secondary deposit resulting from manual pick-up cleaning, although the sample size is small. Gremillion (1995) was unable to attribute a function to this feature.

Four flakes were found in Feature 54 fill, giving a density of 1.33/liter. Because half of these were thermally altered, the feature may represent a hearth or hearth cleanings. Gremillion (1995) interpreted Feature 54 as a pit.

Feature 55 lithics fit Nielsen’s (1991a) description of secondary deposits resulting from manual pick-up cleaning. Of the thirteen lithics recovered, eleven are flakes and two are debris. The lithic density for this feature is 1.73/liter, the second highest of all Cold Oak features. One flake is thermally altered. Gremillion (1995) suggested that Feature 55 was used as a pit or represents a post mold.

Four flakes derived from Feature 56, yielding a density of 0.89/liter. Although the flakes suggest lithic reduction activities, the sample size is too small to be conclusive. According to Gremillion (1995), the function of Feature 56 is unknown.

84 Early- and late-stage reduction debitage comprise 16 of the 17 lithics recovered from Feature 57. Of these, four flakes show evidence of thermal alteration. Coupled with the piece of fire-altered limestone, the lithics suggest that Feature 57 functioned as a hearth or represents hearth cleanings. Lithic density is 1.10/liter. Gremillion (1995) did not propose a functional designation for Feature 57. A radiocarbon date of 387 B.C.

(2190 ± 80) was obtained from uncharred wood (Gremillion 1995).

Excavators recovered only one unaltered broken flake in Feature 58 fill, making it difficult to interpret the feature's functional significance. The density for the feature is only 0.12/liter. Gremillion (1995) interpreted Feature 58 as a pit.

Of the five flakes recovered from Feature 59, one is heat treated. A thermally altered Adena-like projectile point stem derived from this feature. Lithic density is

1.00/liter. Most of the flakes are secondary and thinning flakes, suggesting that the feature is a deposit of late-stage reduction debris: however, the feature could represent a hearth or hearth cleanings. Gremillion (1995) interpreted Feature 59 as a pit.

With a density of 1.00/liter, Feature 60 yielded two unaltered flakes. There is insufficient lithic evidence to propose a functional designation. Gremillion (1995) interpreted the feature as a pit.

Feature 67 contained only one piece of chert debris for a density of 0.17/liter.

The paucity of lithics makes it impossible to suggest a functional designation, which agrees with Gremillion's (1995) interpretation of the feature.

Turning now to the Terminal Archaic features, a total of 26 such features were encountered in the excavation trenches at Cold Oak Shelter in 1984 and 1994. Feature designations are ten pits, one fire-cracked rock-filled pit, one hearth or storage pif ^, three hearths, one hearth/earth oven, one pit or post mold, five post molds (one of which was encountered during both field seasons), and two casually constructed refuse

^ The hearth/storage pit is Feature 8, which Gremillion (1995) identified as Early Woodland and G'Steen et al. (1991) designated Terminal Archaic.

85 pits or natural depressions filled with midden. One feature is a midden-filled natural depression and one feature is of unknown function (Gremillion 1995, G’Steen et al.

1991). Sixteen of the Terminal Archaic features were excavated during the 1994 field season (Table 1).

Lithics were recovered from 10 of the 16 Terminal Archaic features excavated in

1994 (Table 2). As with the Early Woodland features, the lithic artifacts confirm some of Gremillion's (1995) functional designations but contradict others. Lithics were not recovered from six Terminal Archaic features: 32, 33, 45, 46, 64, and 70.

Feature 24, interpreted by Gremillion (1995) and G'Steen et al. (1991) as a post mold, contained 14 lithics for a density of 0.52/liter. Gniy flakes and debris derived from Feature 24 fill, and late-stage debitage is more abundant than early-stage debitage. Eight of the cherts are heated. The size distribution of the small flake sample corresponds to Nielsen's (1991a) experimental data for secondary deposits resulting from manual pick-up cleaning, but the thermally altered lithics support a hearth or hearth cleanings interpretation.

Twelve lithics derived from Feature 39, yielding a density of 1.09/ liter. Most of these are early and late-stage debitage, and two are debris. Almost half of these cherts show signs of unintentional heating. Feature 39 may be a hearth or hearth cleanings, but the size distribution of flakes corresponds to Nielsen's (1991a) data for secondary manual pick-up cleaning deposits. Gremillion (1995) interpreted Feature 39 as a pit.

Three small pieces of fire-altered sandstone derived from Feature 47, yielding a relatively high density of 1.50 lithics/liter. This feature may have been a hearth or may represent hearth cleanings. Feature 47 was used as a pit, according to Gremillion

(1 9 9 5 ).

Although only one flake was recovered from Feature 48, the lithic density is rather high at 1.33/liter. There is insufficient evidence to posit a functional

86 interpretation for this feature based on lithics alone, but Gremillion (1995) identified it as a pit.

Feature 61 contained three lithics for a density of 0.50/liter. All are flakes, two of which show signs of thermal alteration. Perhaps this feature is a hearth deposit.

Gremillion (1995) designated Feature 61 as a natural depression or a refuse pit.

Twenty-two lithics derived from Feature 62 fill, giving a density of 0.97/ liter.

Most of the artifacts are early and late-stage flakes, one of which is thermally altered.

Five very large pieces of fire-altered sandstone, along with the heated chert, indicate that Feature 62 may have been a hearth or hearth cleanings. Feature 62 was interpreted as a natural depression by Gremillion (1995).

While only one unaltered broken flake was found in Feature 63 fill, the highest density of lithics for all Cold Oak features, 2.00/liter, is recorded for this feature. The author cannot propose a functional interpretation for Feature 63 based on lithics alone.

The feature represents a post mold, according to Gremillion (1995).

Feature 65 also contained only one lithic, a piece of unaltered debris, and the density is very low at 0.14/liter. There is not enough lithic data to make a functional identification for this feature, but Gremillion (1995) identified it as a pit.

Two large pieces of fire-altered sandstone derived from Feature 66, suggesting it functioned as a hearth or represents hearth cleanings. The density is 0.50/liter.

Gremillion (1995) designated Feature 66 as a pit.

Feature 69, identified by Gremillion (1995) a s a pit, contained only one unaltered flake. The density of lithics for the feature is 0.50/liter. A functional designation based on lithics cannot be proposed because of the small sample size.

87 3.2.3. LITHIC ASSEMBLAGES

The present study uses lithic materials recovered from the 1994 excavations at

Cold Oak Shelter. Chipped-stone artifacts, made mostly of locally available cherts, dominate the 2241-piece assemblage and account for 84% of the specimens; only 1% of the chipped-stone material is tools, while about 99% is chipped-stone debitage. About

10% of the assemblage is thermal debris, and 5% is miscellaneous rocks and minerals that were not physically modified but are foreign to the rockshelter matrix. Only three ground-stone fragments were found, representing less than 1% of the entire assemblage.

Lithic specimens were recovered through surface collection, 1/4" and 1/16“ screenings of unit and feature fill, and processing of 0.6 mm paleoethnobotanical samples.

In terms of percentages and densities, there is a fairly equitable distribution of lithics from the four easternm ost excavation units of Cold Oak Shelter, with percentages ranging from 15% to 25% and densities from 0.57/liter to 0.83/liter. Vertically, Zone

I yielded the highest percentage of lithics with 38%, followed by Zone II with 26%.

Although over half of the specimens could not be assigned to a cultural-temporal period,

25% derived from Early Woodland strata and features and 14% from Terminal Archaic deposits. The highest density of lithics, 0.92/1, derived from Early Woodland deposits.

For the entire shelter, the density of lithics is 0.75/liter. Chapter 4 describes the lithic assemblages in more detail.

3.3. ROCK BRIDGE SHELTER

Rock Bridge Shelter (W075) is located north of Cold Oak Shelter in the Red River

Gorge of Wolfe County (Figure 10). Deeply entrenched along the sandstone cliff line, the river reaches its maximum gradient, dropping 200 feet over a few-mile stretch, in the

88 "Roughs of the Red" valley. Bottomlands along the Red River and its tributaries are

significant in lateral extent only below the Roughs near Gladie Creek. However, bottom

land width never exceeds the local relief. Steeply sloping valley walls and sandstone

cliffs impede travel between bottom lands and uplands. Few gaps in the cliff line permit

access to the ridge tops, resulting in the relative isolation of different tributary valleys.

Though numerous, ridge tops are narrow and sinuous. The gorge provides a protected

environment for plants, though there is some vertical zonation in floral distribution.

Mixed mesophytic forests with numerous conifers are typical, and nut-producing trees

are abundant (Gremillion 1993, Wyss and Wyss 1977).

Identified in 1989 by U.S. Forest Service archaeologist Johnny Faulkner, Rock

Bridge Shelter is located in an unnamed hollow about 80 m above Rock Bridge Fork, a

tributary of Camp Swift Creek in the Red River drainage. Cut into the Corbin Sandstone

Member of the Lee Formation, the shelter measures approximately 40 m long, 8 m wide, and 1-3 m high and has a westerly aspect. The maximum depth of deposits is 20-25 cm.

While the northern end of the shelter lies within the dripline, the southem end does not.

Deposits are damp due to ceiling leakage.

The floor of Rock Bridge Shelter, which slopes slightly to the north, forms the ceiling of another shelter located directly below. Talus from niter mining covers much of the lower shelter's floor, drastically reducing the amount of testable surface area. The only documented archaeological testing of the lower shelter (Gremillion 1993) involved limited shovel testing and yielded lithic debitage, sherds, bone, charcoal, and uncarbonized plant remains from a shallow, stratified midden deposit.

Compared to Cold Oak Shelter to the south and to other rockshelters in the Red

River Gorge area, Rock Bridge Shelter is unique in a number of ways. Rock Bridge appears to be a single component site, with deposits dating to the Late Woodland period based on radiocarbon dates and diagnostic artifacts. The shelter is protected and hardly

89 accessible by humans: the southern end is blocked by dense vegetation, only experienced climbers can scale the northern end, and few gaps afford access from above. B ecause of its inaccessibility. Rock Bridge apparently has remained undisturbed since the prehistoric period. The floor of the shelter is clear of niter mining talus, and there is no evidence of illegal excavation by relic collectors.

The Forest Service conducted a brief survey and limited surface collection upon the shelter's discovery in 1989. Gremillion implemented more intensive investigations in 1992 in order “to obtain data relevant to the subsistence and settlement patterns of

Late Woodland populations occupying the Red River Gorge area" (Gremillion 1993:ii).

The author served on the field crew. Gremillion tailored field methods to maximize recovery of subsistence debris and to identify spatial patterns in the material remains. A systematic surface collection of most of the floor was conducted in 2 x 2 m units. Using the distributions of surface artifacts, Gremillion delineated three excavation blocks

(Figure 13). Four contiguous units were placed over a charcoal-producing feature at the northern end of the shelter. To the southeast, a pair of units partially quartered an apparent hearth feature. Gremillion examined the central portion of the shelter, where surface artifacts were most numerous, with 16 contiguous units. In all, twenty-two 1 x

1 m units were excavated to a depth of about 20-25 cm, for a total of 4.4 m^ of excavated sediments. Excavation proceeded at arbitrary 5 cm levels except when differences in natural strata within one level were encountered, in which case the strata were removed separately. Features were excavated and bagged separately.

3.3.1. STRATIGRAPHY AND DEPOSITIONAL HISTORY

Little horizontal variability was observed in the sediments, though sediment thickness increased slightly downslope. Vertical transitions among deposits were gradual

90 and generally Indistinct. In addition to numerous pockets and lenses of sediments,

Gremillion identified three stratigraphie zones. Zone I is a "grayish-brown loose, ashy, unconsolidated silty sand mixed with artifacts, bone and charcoal that extends from one to four cm below the ground surface" (Gremillion 1993:20). Gremillion (1993:43) reported that wood charcoal from this zone produced a corrected radiocarbon date of A.D.

640 (1310 ± 60 B.P.). A 5-10 cm transitional zone. Zone II, separates Zones I and III; it is a rockier, more compact, yellowish or brownish sandy silt or silty sand with more artifacts than the upper midden. Workers recovered few artifacts from Zone III, a 5-6 cm thick, yellowish, compact rocky sand with decaying sandstone. A layer of "stiff orange-yellow sandy clay" lacking cultural remains lies below Zone III in some portions of the shelter (Gremillion 1993:21). Figure 14 shows a stratigraphie profile for the main excavation block of Rock Bridge Shelter.

The developmental history of Rock Bridge Shelter is probably similar to that of

Cold Oak Shelter, beginning with the formation of the recess by stream erosion during the late Pleistocene-Holocene epochs (Donahue and Adovasio 1990). The sediments in

Rock Bridge Shelter could have accumulated during and after the recess formed. The lowermost clay deposits might represent water-borne sediments that were laid down as the shelter formed, or they might be the illuviated products of chemical weathering by hydrolysis of minerals in the sandstone bedrock. Attrition of the sandstone bedrock probably accounts for the sand component of the shelter deposits. The sand- to boulder­ sized sandstone particles throughout the shelter's sediments likely derived from roof fall, while hum ans introduced chert particles. Fine-grained material in the upper zones may be the result of human activities such as burning fires.

9 1 3.3.2. CULTURAL FEATURES

Eighteen cultural features were Identified during field investigations at Rock

Bridge Shelter. Based on artifactual remains and stratigraphie properties, Gremillion

(1993) concluded that three features represent hearth deposits, one is a basin, one is a

cylindrical pit, one is a charcoal concentration, one is a mix of cultural materials and

pack rat debris, and 11 represent redeposited midden that settled in low portions of the

shelter (Table 1). Gremillion (1993) reported that charcoal from the surface of

Feature 6 yielded an uncorrected date of A.D. 612 (1380 ± 50 B.P.). Nine of the 18

Rock Bridge features yielded lithic materials. No lithic artifacts derived from features

2, 4, 6, 7,8,9, 10, 11, and 12 (Table 3).

Feature 1 produced eight lithic specimens, seven flakes and one piece of debris, giving it a fairly low lithic density of 0.11/liter. A relatively wide range of chert materials derived from this feature. One flake shows evidence of thermal alteration.

Gremillion (1993) interpreted Feature 1 as a mixture of cultural deposits and pack rat debris, and the lithics neither confirm nor discount the designation.

Feature 3 had the largest number of lithic artifacts, 97 specimens. These lithics account for three-quarters of the feature lithic sample and over one-tenth of the entire shelter assem blage. With a lithic density of 1.36 specim ens per liter. Feature 3 has the second highest density of the features. One utilized flake, possibly a side scraper, came from Feature 3. Only two of the specimens have evidence of thermal alteration. Much of the chert material is representative of the later stages of reduction, suggesting that perhaps Feature 3 is a deposit of lithic production waste. The distribution of lithic reduction classes suggests association with primary reduction and/or secondary reduction and retooling. While the proportions of debitage size classes does not correspond to the expected signatures for deposits resulting from the cleaning of lithic

92 activity areas (Nielsen 1991a), Gremillion (1993) interpreted Feature 3 as redeposited hearth materials.

Ten unaltered flakes were recovered from Feature 5 fill, and the density of lithics is rather low at 0.12/liter. The percentages of large and small microdebris for the feature resemble the signature for a residual primary deposit left after sweeping

(Nielsen 1991a), but the small sample size makes this interpretation tenuous.

However, based on other artifactual evidence, Gremillion (1993) interpreted Feature 5 as hearth cleanings or a disturbed hearth.

Feature 13 yielded two lithic specim ens, both of which are utilized flakes composed of Paoli chert. Neither shows signs of heat alteration. The lithic density for this feature is 0.43/liter. It is not possible to propose a functional designation for

Feature 13 based on lithics, and Gremillion (1993) proposed that the feature is redeposited midden.

Feature 15 has the highest lithic density of the excavated features at 3.0/liter.

None of the nine specimens were heat altered. The size distribution of these items resembles that of a primary residual deposit left after sweeping a lithic activity area

(Nielsen 1991a). Gremillion (1993) proposed a comparable interpretation, that

Feature 15 represents redeposited midden.

Feature 17, a shallow basin feature (Gremillion 1993), was encountered in the portion of the shelter with the highest concentration of lithic remains. The feature yielded two hafted bifaces, a Chesser Notched point and a perforator, and three pieces of debitage. The lithic density is 0.25/liter. It is difficult to assign a function to this feature based on the limited lithic data. Feature 17 was one of only two features with distinct lower boundaries.

Features 14 and 18 yielded only one lithic specimen each. Two small flakes were recovered from the fill of Feature 16. None has evidence of heat alteration. There is

93 insufficient evidence to propose functional types for these features based on lithics only.

Gremillion (1993) indicated that charcoal from Feature 16 produced a corrected

radiocarbon date of A.D. 780 (1170 ± 70 B.P.). Features 14 and 16 were classified by

Gremillion (1993) as redeposited midden, and Feature 18 was a cylindrical pit with distinct lower boundaries.

3.3.3. LITHIC ASSEMBLAGE

The lithic assemblage from Rock Bridge Shelter consists of 755 specimens recovered from systematic surface collections, 1/4" and 1/16" screening of unit and feature fill, and 0.6 mm screening of sediment samples. Over 99% of the assemblage is chipped-stone material, including 26 tools and tool fragments and 723 pieces of debitage. Most of the chert artifacts are made of locally available material. Few ground- stone artifacts and fracture lids were recovered, accounting for the other 1% of the assemblage. About 90% of the lithics are from the main excavation block in the center of the shelter. Vertically, 65% of the lithics derived from Zone II and 32% from Zone I;

3% came from Zone III. Overall, the density of lithics is 0.42/liter. The lithic assemblage is described in more detail in Chapter 4.

3.4. SUMMARY

Rockshelters form in sedimentary rocks largely as a result of hydrological processes. Differential weathering of interbedded lithologies results in the formation of a recess in the sedimentary cliff line and the retreat of the shelter overhang over time.

Rockshelters develop in young river valleys where the erosive power of streams is

94 greater. In eastern Kentucky, rockshelters formed In interbedded sandstones and shales of the Pennsylvanian Lee Formation (Donahue and Adovasio 1990).

In general, rockshelter sedimentation is a more-or-less continuous, aggradational process. However, the sediments that accumulate in rockshelters vary greatly in derivation and, therefore, structure and composition. In sandstone shelters, the most important endogenous sedimentary processes are rockfall and attrition, which contribute to shelter deposits large debris and sand (Donahue and Adovasio 1990).

Attrition contributes only sand, but rockfall contributes particles of a variety of size classes. Exogenous processes that introduce sediments to shelters are biological, geological, and cultural; they include, for example, carnivores transporting prey into shelters, sheetwash from the slopes above shelters, and the importation of resources to shelters by humans.

According to Donahue and Adovasio (1990), eastern Kentucky's rockshelters like

Cold Oak and Rock Bridge developed by stream erosion during the late Pleistocene-

Holocene epochs. Sandstone boulders found throughout Cold Oak and Rock Bridge shelters likely derived from rockfall of the ceilings and walls. Basal sand deposits at Cold Oak

Shelter are probably the result of attrition relatively early in the life of the shelter.

The paucity of cultural artifacts in these sandy zones suggests that humans did not consistently use Cold Oak Shelter during its early developmental history. The lowermost deposits at Rock Bridge Shelter, while also largely devoid of cultural material, are mostly clay and may derive from sheetwash or flooding events or chemical weathering.

As human occupations become more evident in the strata at Cold Oak and Rock Bridge, the sedim ents become more silty and ashy. Lithic objects m ade of chert and limestone are probably artifacts imported to the shelters by humans.

Though the shelters share a somewhat similar developmental history. Cold Oak differs from Rock Bridge in other ways. Cultural deposits at Cold Oak Shelter are thicker

95 and more complexly stratified, evidence indicates use of tfie slielter over several time periods, preservation of organic materials is better due to dry conditions, the shelter is easily accessible, and deposits were considerably disturbed by several historic groups.

Zone II strata, which immediately underlie the disturbed deposits at Cold Oak Shelter and possible mark the predisturbance surface of the shelter, contain a high density of artifacts that may have been visible to people visiting the site. The fact that most artifacts at Rock Bridge Shelter derived from subsurface Zone II contexts may explain why the shelter was left relatively undisturbed.

Besides strata, another difference between Cold Oak and Rock Bridge shelters is the number and diversity of cultural features. About 300 liters of feature fill was excavated from the 42 features during the 1994 season at Cold Oak, which is an average of seven liters of fill per feature; feature fill accounts for about 11% of the total volume of sediments excavated at Cold Oak. The cultural features (excluding midden-filled depressions) at Rock Bridge yielded at least liters of fill for an average of over 50 liters per feature; about 14% of the total volume of sediments excavated at Rock Bridge derived from these features. There is greater diversity in the types of features encountered at Cold Oak, pits are relatively uncommon at Rock Bridge, and no post molds were encountered in the shallow deposits at Rock Bridge. Rock Bridge features tend to be extensive in the lateral direction. Features may be more extensive vertically at Cold Oak because a thicker sequence of sediments (Zones IV and V) underlay the zones of occupation, allowing more room for excavating features like pits into the subsurface of the shelter.

Despite differences in the lateral extent of excavations at Cold Oak and Rock

Bridge, the lithic assemblages from the two shelters are generally comparable in terms of numbers and composition. Lithic densities from Cold Oak and Rock Bridge shelters fall below one specimen per liter, and the assemblages are dominated by chipped-stone

96 artifacts. The ratio of tools to debitage in both assemblages is small. Do the assemblages represent differential shelter use at Cold Oak and Rock Bridge? Superficially, the answer is no. This issue is closely examined in the next chapter, which describes in detail the lithic assemblages from Cold Oak and Rock Bridge shelters.

9 7 CHAPTER 4

COLD OAK AND ROCK BRIDGE LITHIC ASSEMBLAGES

As indicated in Chapter 3, Cold Oak and Rock Bridge differ in terms of stratigraphy and features, but they share a similar developmental history. A cursory examination of the lithic assemblages from the shelters also reveals a number of similarities. The purpose of this chapter is to provide more detailed descriptions of the lithic assemblages that serve as the database for the current research. In this study, the lithic assem blages from Cold Oak and Rock Bridge shelters are used to evaluate the three research propositions outlined in Chapter 1, that useful information can be derived from shelter lithic assem blages, that lithic assem blages can inform us about prehistoric rockshelter use, and that lithics should reflect diachronic differences in shelter use.

Therefore, special attention is paid to similarities and differences among lithic assemblages from different occupations at the shelters. Complete descriptions of the Cold

Oak and Rock Bridge lithic assemblages cannot be presented here due to spatial constraints: see Applegate (1993) and Applegate (1995a) instead. The chapter begins with a discussion of lithic classification and raw material identification.

98 4.1. CLASSIFICATION

In this study several concepts fundamental to archaeological research are used: artifact, feature, attribute, context, and assemblage. Some of these terms have been alluded to already and they will be used throughout the rest of the dissertation. Most archaeologists view artifacts as the building blocks of archaeological collections, and assemblages are the basic analytical units of behavioral studies (Villa and Courtin

1983:269). Such is the case in the present study. This section also outlines the classification scheme used to subdivide the lithic assemblages from Cold Oak and Rock

Bridge Shelters.

The archaeological concept of artifacts has changed somewhat over time. Most early definitions of artifact were narrowly conceived and focused on objects, modification, and human agencies of production. Definitions such as "an object made by human workmanship" (Braidwood 1960:190), "any object showing manufacture or use by humans" (Meighan 1966:173), and "man-made objects classified by function, date, and culture or context" (de Paor 1967:36) typify this perspective. Braidwood

(1960:174) went so far as to define non-artifactual material as "all of the things which men utilized without fashioning, or the raw materials which were fashioned into artifacts, if found yet unworked." Artifacts are produced through physical modification by humans, not natural processes.

Biek (1970:567) defined an artifact as "every piece of evidence of early human activity." Although he viewed artifacts as the product of intentional human behavior or

“human purpose impressed on inanimate material," Biek (1970:567) allowed for the possibility that artifacts may be "material remains essentially unaltered by working."

Identification of unmodified material (raw material) as artifacts requires consideration

99 of the surrounding environment. At the other extreme, artifacts “may be completely

altered" and only discernible by traces or outlines (Biek 1970:569).

The present study employs a relatively broad definition of artifacts, although the

human element is still stressed. An artifact is anything, modified or unmodified, that

owes any of its attributes to human activity. Artifacts may be the product of intentional

human activities, such as shaping clay into a vessel, with "intentional" meaning

behaviors that are intended to produce a product. Unintentional activities, such as

transporting burrs on one's clothing from one place to another, may also result in an

object being classified as an artifact. Artifacts may be physically modified by human

action, or they may be unmodified but in different locations because of human activities.

Artifacts may be cultural items, or objects with discrete boundaries such as a ceramic

vessel, or they may be cultural features.

The feature concept is prevalent in archaeology and is dominated by a functional

perspective. A feature is an aggregate of meaningfully related cultural items and natural

items that often lacks discrete boundaries. Because of their indistinct nature, features

are often difficult to collect and transport from a location. Features are typically

classified according to their supposed functions; common designations are hearth, fire oven, burial, cache, post mold, storage pit, and refuse pit.

An attribute is defined as a measurable quantity or property of something. Two types of attributes are physical or formal attributes and locational attributes. The

former refers to the physical or chemical properties of an artifact or feature, such

length, color, and composition. Locational attributes are the three-dimensional spatial coordinates of an artifact or feature. Locational attributes are sometimes referred to as

context. Artifacts or features found in primary contexts are in the places where they were originally deposited, whereas artifacts or features located in places other than the original sites of deposition are in secondary contexts.

100 For the purposes of this study, an assemblage is defined as a collection of cultural items, lithic artifacts in this case, recovered from a particular context. Cultural items recovered from features may be included in an assemblage, but features themselves are not. With respect to context, an assem blage may be the artifact collection from a particular geological stratum (strata), cultural occupation zone(s), or geographic location.

The lithic assemblages from Cold Oak and Rock Bridge shelters are divided by stratigraphie context, with the stratigraphie units delineated by a combination of natural (sediment type) and cultural (radiocarbon dates, diagnostic artifacts) criteria.

For Cold Oak, the lithic remains from Zones III, IV, and V constitute one assemblage of

Terminal Archaic age, the remains from Zones lib and lie are an assem blage of the Early

Woodland period, the Zone lia remains are a Middle Woodland assemblage, and Zone I is a disturbed assemblage. The remains recovered from all strata at Rock Bridge are considered one assemblage dated to the Late Woodland period.

As recognized by Braidwood (1960), an assemblage is rarely, if ever, a complete representation of all artifactual remains left by a person or group of people. Differential preservation and recovery influence the representativeness of an assemblage. In most environments, durable inorganic remains that are relatively resistant to decomposition dominate assemblages. Some rockshelters of eastern Kentucky are unique, and therefore well known, because of the exceptional preservation of organic artifacts in their assemblages.

Most definitions of assem blages explicitly or implicitly delineate a temporal dimension. Meighan (1966:190), for instance, defined an assemblage as "a collection of contemporaneous specimens." Villa and Courtin (1983:268-269) defined assemblages as sets of material grouped by their association, with association meaning that the material "is referred to a single episode of occupation or, at least, a single mode of site

101 use by a specific group of people. " Contemporaneity is a relative term; in otfier words, artifacts are contemporaneous at what temporal scale - a month, a year, a decade, a century? In most cases, the issue of contemporaneity of artifacts in an assemblage is an assumption that must be tested because of the potential for artifact and sediment mixing.

Michels (1973) and Thomas (1979) distinguished between homogeneous assemblages, which represent related samples of the residues of a group of people, and mixed assemblages, which are the combined residues of once-distinct events. If assemblages are the unit of behavioral analyses in archaeology, as in this study, then consideration of extraneous processes that may mix assemblages is critical. This issue is considered in

Chapter 5.

The lithic assemblages from Cold Oak and Rock Bridge shelters are categorized using a combination of criteria that facilitate answering the research questions sketched in Chapter 1: raw material, function, mode of genesis, and reduction class. A primary dichotomy of lithic artifacts, ground stone versus chipped stone, takes into account raw material and technological attributes. Specific ground-stone tools are identified according to presumed function and chipped-stone artifacts are divided into two techno­ functional groups, tools and debitage. Tools include bifacial tools, marginally modified flakes, or utilized flakes, which are defined by the nature and extent of modification.

Three categories of debitage, distinguished by size and morphology, are cores, flakes, and debris. In addition to ground-stone and chipped-stone artifacts, a third category of lithics is thermal debris, which includes fire-altered rock and fire-altered chert. The fourth category, miscellaneous lithics, includes unmodified lithic materials.

Ground-stone artifacts are lithics shaped by preliminary pecking and chipping of the raw material, followed by grinding of the surface to finish the object.

Generally, ground-stone artifacts are manufactured from a wide range of raw materials,

102 including fine-grained rocks and minerals like limestone, basalt, and hematite and coarse-grained rocks such as granite, andésite, and sandstone.

Chipped-stone artifacts are lithics formed by the direct or indirect application of pressure to a brittle raw material in order to detach pieces of the material. Knappers may modify the initial piece of raw material or the detached debitage.

Cherts are typical raw materials used in chipped-stone tool manufacture. Quartz, quartzite, and obsidian are other microcrystalline to cryptocrystalline materials suitable for chipped-stone tool production.

Chipped-stone tools and tool fragments include any artifacts specially modified in order to perform some funotion(s) or altered during the course of usage. Bifacial tools are modified on two sides, ventral and dorsal, and have special edge treatment such as hafting. Modification of such artifacts proceeds well beyond simple shaping of edges.

Biface fragments are broken pieces of bifacial blanks, bifacial preforms, or finished bifaces representing various stages of the biface reduction sequence. Flake blanks are bifaces whose lateral margins are not completely worked. Preforms have completely worked lateral margins, but those margins are not straightened. Finished bifaces have straight, completely worked lateral edges (Johnson 1989). Biface fragments do not retain sufficient attributes for confidently assigning them to one of these categories.

Marginally modified flakes are lithics that “exhibit uniform flake removal along one or more edges; they may also exhibit ground or crushed edges ' (Ledbetter and G'Steen

1990:78). These tools do not have special edge treatment. Utilized flakes are artifacts with edges altered by use rather than intentional shaping. As such, “edge wear resulting from the use of an unmodified flake for activities such as cutting and scraping" characterizes utilized flakes (Ledbetter and O'Steen 1991:79).

Chipped-stone debitage refers to material left over from the manufacture of chipped-stone tools. Three categories of debitage are cores, flakes, and debris. Cores are

103 the remains of raw materials from which flakes were detached during the course of chipped-stone tool manufacture. They may be stream cobbles, nodules, or quarried material with negative bulbs of force and flake scars; cores may or may not have cortex or weathered rinds. Cores are different from bifacial blanks and preforms in that the former lack lateral margin reworking. Three types of cores are distinguished by the nature of flake removal (Crabtree, no date). Unifacial cores have one surface from which flakes were detached. Bifacial or bi-directional cores possess two surfaces from which flakes were removed. Cores with more than two surfaces of flake removal are multifacial cores.

Flakes are chipped-stone debitage with a single interior (ventral) surface indicating where they were detached from cores by the application of pressure. Several criteria are used to assign flake specimens to reduction classes so that the shelter assemblages can be compared with the experimental models described in Chapter 6 and work at other shelters in the study area; the criteria are amount of cortex, number of dorsal scars, platform morphology, bulb of percussion, and relative size. Primary decortication flakes are the first flakes removed from a core with cortex during the lithic reduction sequence. As such, they have cortex on 100% of the dorsal surface and lack dorsal scars. Although they occur in a relatively wide range of sizes, they are generally large and thick. Secondary decortication flakes have dorsal scar(s) from previously detached primary decortication flakes and hence have less than 100% cortex on the dorsal surface. Flakes with cortex remaining on the platform only were not included in this category because they can be produced throughout the lithic reduction process. Primary flakes lack dorsal cortex, have pronounced bulbs of force, have single or few dorsal flake scars, are often triangular in transverse cross section, and tend to be longitudinally curved (convex-concave). While large primary flakes are common, they may be relatively small depending, in part, on the size of the core from which they were

1 04 detached. Primary flakes are produced during modification of a decorticated core.

Secondary flakes also lack dorsal cortex and have rather pronounced bulbs of force, but they have multiple flake scars, are biconvex in transverse cross section, and are straight longitudinally. They are produced during modification of a decorticated core.

Thinning flakes are produced during biface manufacture and tool reworking. They characteristically lack cortex, have parallel surfaces, multiple dorsal flakes scars, faceted platforms, and may exhibit dorsal platform retouch. They tend to be small and thin. This category includes both bifacial thinning flakes and resharpening flakes. Flakes lacking platforms and other diagnostic attributes are broken flakes. They could be secondary decortication flakes whose dorsal surfaces with cortex were detached, the relatively thin medial-to-distal portions of primary flakes, or fragments of secondary or thinning flakes. While not considered a reduction category, a final flake type is blades

(bladelets), which are unmodified flakes with parallel sides and one or two dorsal ridges. They are typically twice as long as wide.

Debris is chipped-stone debitage lacking a single interior surface. Debris with bulbs of percussion probably formed during lithic reduction while debris lacking bulbs may have formed by chert failure and explosion as a result of heating (Luedtke 1992;

Purdy 1975) or by frost action (Luedtke 1992). Several categories of debris are identified based on size and cortex. Blocky irregular chunk refers to angular debris greater than two cm in maximum dimension and lacking cortex, while blocky irregular decortication chunk is angular debris greater than two cm in maximum dimension and having cortex. Angular shatter is less than two cm in size and lacks cortex, and angular decortication shatter is less than two cm in size and has cortex.

The fracturing of chert and other materials by heating produces thermal debris. There are two types of fire-altered chert. Pot lids are generally small (less than one cm), circular chert debitage with plano-convex dorsal and ventral sides. Heat

105 spalls are relatively large (greater than one cm), irregularly shaped chert fragments with plano-convex sides. For both of these types of thermal debris, the ventral surface is often granular in appearance, and the dorsal surface may or may not have cortex and/or flake scar ridges. The second major category of thermal debris, fire-altered rock, shows one or more of the following effects of heating: reddening, blackening, ash residue, shattering, vitrification (fusion of crystals), or loss of cement resulting in greater friability. Common lithologies of fire-altered rock are sandstone and limestone.

The fourth major category of lithics, miscellaneous, refers to unmodified lithics imported to a site. These artifacts are included in the analysis because, although they show no signs of cultural modification, their occurrence at the site must be the result of human activity. These are materials that are foreign to the bedrock matrix of the shelters. Miscellaneous lithics include frost lids, metamorphic rock fragments, sedimentary rock fragments, river cobbles, natural chert, and other minerals.

4.2. RAW MATERIAL IDENTIFICATION

Identification of raw material types is important in this study because the relative proportions of local and non-local raw materials is one indicator of occupational intensity and because diachronic patterns of raw material use are investigated. Raw material types were identified by their physical properties (color, luster, hardness, fracture, texture, and streak, the color of the powdered specimen), reaction with weak hydrochloric acid, reference to published descriptions, and comparison with chert samples provided by the National Forest Service. Moistening the surfaces of the specimens with water (especially those with patinas that could not be dissolved) aided in raw material identification.

106 Nonchert materials represented in the assemblages are both rocks and minerals.

The former includes limestone, dolostone, and sandstone. Limestone has a dull luster, hardness less than five (steel knife), and strong reactions with acid. Dolostone exhibited similar characteristics, except it has relatively weaker reactions with acid. Texture and composition distinguished the sandstone.

Five minerals present in the assemblages are hematite, limonite, siderite or jaspilite, quartz, and gypsum. The first two are iron oxide minerals characterized by a dull luster and a hardness less than five. Hematite has a red to brown color and streak while limonite is usually yellow-brown in color and streak. Siderite or jaspilite is characteristically dull, brown to gray in color, harder than five, and breaks with conchoidal fractures. Quartz has a vitreous luster, hardness greater than five, conchoidal fracture, and colors ranging from clear to reddish, yellowish, or brownish.

Gypsum varies in color and luster but it has a characteristic hardness of two.

Eight types of chert represented in the assemblages are Boyle, Muldraugh, St.

Louis, Paoli, Haney, Breathitt, Kanawha, and Ste. Genevieve. Appendix A gives descriptions of these cherts, which are based on Black (1978), Ettensohn et al.

(1984), Gatus (1987), Graham (1990), Haney (1976), Ledbetter and O'Steen

(1991), Meadows (1977), Rice (1984), Rice and Weir (1984), Sable and Dever

(1990), and Yerkes and Pecora (1991) as well as the author's obsen/ations from the

Cold Oak and Rock Bridge shelter assemblages (Applegate 1993, 1995a). Applegate

(1995b) provided quantitative descriptions of chert and cortex colors.

4.3. LITHIC ASSEMBLAGE DESCRIPTIONS

In this introductory section, the Cold Oak and Rock Bridge lithic assemblages are described in general terms, focusing on the composition and spatial attributes of each

1 07 assemblage. The spatial distributions, raw material distributions, and other properties of each lithic artifact type are then described.

4.3.1 COLD OAK SHELTER LITHIC ASSEMBLAGE

4.3.1.a. Overview

Table 4 summarizes the composition of the lithic assemblages recovered from

Cold Oak Shelter. Chipped-stone artifacts dominate the 2241-piece assemblages, accounting for about 84% of the specimens. Thermal debris is the next most common artifact class, representing over 10% of the assemblages. About 5% of the lithics are miscellaneous rocks and minerals that probably did not derive from the sandstone and conglomeritic-sandstone matrix of the shelter. Ground-stone artifacts constitute only a minor proportion of the assemblage at 1%.

The density of lithic artifacts from the 1994 excavations at Cold Oak is

0.75/liter; this number excludes lithics recovered from the backfill of the 1984 excavation trench and surface collected artifacts from the 1994 season because soil volumes are incomplete. At 0.62/liter, the density of chipped-stone artifacts differs considerably from the densities of thermal debris (0.08/liter), miscellaneous lithics

(0.04/liter), and ground-stone artifacts (less than 0.01/liter).

Table 5 summarizes the distribution of raw material types in the Cold Oak lithic assemblages. About 86% of the artifacts are made of chert, 13% are rocks, and less than

1% are minerals. All ground-stone artifacts are made of carbonate rocks. As one might expect, 99% of the chipped-stone artifacts are made of chert, with minerals and limestone accounting for the other 1%. Three-fourths of the thermal debris is sedimentary rocks, 24% is chert, and less than 1% is minerals. Miscellaneous lithics are 78% rocks, 14% chert, and 8% minerals. These percentages are referenced

108 throughout the lithic descriptions and when discussing the lithic production system of the shelter in Chapter 6.

4.3.1.b. Spatial Distributions

Table 6 presents the horizontal distribution of lithic artifact types by excavation unit at Cold Oak Shelter. In terms of percentages, units 510.5N 498E and 510.5N 500E each contained 25% of the lithic specimens, and unit 510.5N 497E yielded about 20%.

The central unit, 510.5N 499E, had 15% of the lithics, while the easternm ost unit,

510.5N 501E, only yielded about 8%. The remaining 7% of the lithics derived from backfill of the 1984 excavation trench and from surface collections outside the 1994 excavation area.

In terms of lithic densities, a more equitable distribution of lithics among the units is evident (Table 6). The highest density of lithics (0.89/liter) derived from unit

510.5N 500E. Although it has a relatively low percentage of lithics, unit 510.5N 499E had the second highest density at 0.82/liter. Units 510.5N 498E and 510.5N 497E yielded densities of 0.77 and 0.62, respectively. The easternmost unit, 510.5N 501E, had the lowest density (0.59/liter) of lithics.

The vertical distribution of lithics by cultural zone in given in Table 7.

Disturbed Zone I clearly yielded the highest percentage of lithics at 38%, but Zone II also contained a relatively high percentage at 26%. Eleven percent and 10% derived from Zone III and from features, respectively. The other cultural zones contained very small percentages of lithics at 1% total. About 14% could not be assigned to a cultural zone because they were recovered by cleaning of profiles and excavation level floors, collected from 1984 backfill or the surface of the shelter, or recovered at the interface of two zones.

Adjusting for differences in sediment volumes, a somewhat different pattern of vertical distribution by cultural zone emerges (Table 7). The highest density.

1 09 2.10/liter, is recorded for Historic deposits. Zone II yielded a relatively high density at

0.94/liter. Zone I, which had the highest percentage of lithics, Zone III, and the features contained densities comparable to the overall shelter density. Again, the density of lithics from unknown contexts (1.20/liter) excludes lithics found outside the 1994 excavation block because soil volumes are not available.

Table 8 outlines the vertical distribution of lithic artifacts by period. A majority of the specimens (60%) could not be assigned to a cultural period because they were recovered from 1984 backfill, from the surface of the shelter, or while cleaning unit floors and profiles. Of the remaining 40%, about 26% are from Early Woodland layers and about 14% derived from Terminal Archaic strata. Middle Woodland contexts yielded less than 1% of the lithics.

The density of lithics in the Early Woodland deposits is higher than the shelter average at 0.92/liter. The density of Terminal Archaic lithics is below the overall shelter density at 0.53/liter, as are the densities for the Historic (0.41/liter) and

Middle Woodland (0.29/liter) assemblages.

4.3.1.0. Chipped-stone Artifacts

Most of the lithic remains of the Cold Oak Shelter assemblage are chipped-stone artifacts (Table 4). The density of chipped-stone specimens at the shelter is 0.62/liter.

Chipped-stone artifacts consist of two groups: tools and tool fragments and debitage. The most common form of chipped-stone artifacts in the Cold Oak lithic assemblage is debitage, which accounts for almost 99% of the chipped-stone material; the other 1% is tools and tool fragments (Table 9). Within the debitage category, the most prevalent type is flakes. Chipped-stone flakes constitute nearly 91% of the debitage, with the remaining 9% being debris and cores.

The lateral and vertical distributions of chipped-stone artifacts follow closely the distribution for the entire lithic assemblage from Cold Oak Shelter (Table 6). One-

1 1 0 fourth of the chipped-stone artifacts came from excavation unit 510.5N 498E, and

nearly that much came from units 510.5N 500E and 510.5N 497E. Adjusting for

sediment volume differences, the highest density derives from 510.5N 500E at

0.72/liter. Units 510.5N 499E and 510.5N 498E also had above-average densities of

chipped-stone lithics.

Vertically, 38% of the chipped-stone artifacts derived from disturbed Zone I

(Table 7). Zone II yielded 27% and Zone III, 11%. Over 9% came from feature fill. The

remaining 15% derived from Zone V and historic and unknown contexts. The highest

density is for Historic deposits at 1.25/liter, followed by Zone II (0.81/liter), Zone I

(0.65/liter), feature fill (0.58/liter), and Zone III (0.51/liter).

Sixty percent of the chipped-stone specimens could not be assigned to a cultural

period (Table 8). Early Woodland deposits contained 26% of the chipped-stone material,

and Terminal Archaic layers about 13%. About 1% derived from Middle Woodland and

Historic contexts. When normalizing for differences in sediment volumes, the vertical distribution by period is similar. The density of chipped-stone artifacts in Early

Woodland contexts is 0.78/liter, followed by Late Archaic at 0.42/liter, Historic at

0.34/liter, and Middle Woodland at 0.26/liter.

Ninety-nine percent of the chipped-stone artifacts are composed of chert; the other percent is minerals and rocks (Table 5). About 28% of the chert specimens are

untyped, but 27% are Haney chert and 18% are St. Louis. Boyle, Breathitt, Kanawha,

Muldraugh, Paoli, and Ste. Genevieve each account for 10% to less than 1% of the chipped-stone specimens. The chipped-stone sample includes several limestone, iron ore, and quartz specimens.

Twenty-three specimens are classified as chipped-stone tools and tool fragments based on edge treatment and edge wear (Table 9). Most chipped-stone tools were recovered from units 510.5N 498E (43%) and 510.5N 500E (26%), on either side of

11 1 the large rock in the central unit. The highest densities also came from these units. Unit

510.5N 499E yielded 9%, units 510.5N 497E and 510.5N 501E yielded 4%, and the remaining 13% derived from outside the 1994 excavation block (Table 10).

Vertically, 57% of the tools were recovered from disturbed Zone I, followed by

22% in Zone II, 13% from unknown contexts, and 4% each from Zone III and features

(Table 11). The highest density, 0.012/liter, is from Zone I, followed again by Zone II with 0.008/liter. Only 30% of the chipped-stone tools from Cold Oak Shelter be assigned to cultural periods (Table 12). Of these, 26% are from Early Woodland contexts and 4% are from Terminal Archaic deposits.

As shown in Table 13, all tools are made of chert. Boyle is the most common chert type in the tool sample at 22%, followed by Kanawha and Paoli (13% each), Haney, St.

Louis and Breathitt (9% each), and Pennsylvanian cherts (undifferentiated Kanawha and

Breathitt) at less than 4%. About 22% of the tools are made of unidentified chert.

Eleven bifacial tools and tool fragments were recovered from Cold Oak Shelter during the 1994 excavations. Lateral and vertical distributions of the bifaces do not differ significantly from the distributions of the entire chipped-stone tool sample

(Tables 10, 11, 12). No one chert type dominates the bifacial tool and tool fragment sample (Table 13). Haney, Boyle, Breathitt, Kanawha, and Paoli cherts are represented, with one to two specim ens per chert type.

Six of the bifacial tools and tool fragments are classified as projectile points or knives. These specimens retain sufficient diagnostic attributes to classify them according to historically sensitive typological schemes (cf. Justice 1987).

Classification of these specimens as projectile points/knives is somewhat provisional as no use-wear analyses, which can inform one about the type of action or materials modified by the tool, were conducted. It is possible that people used the artifacts for purposes other than killing, wounding, or processing game. It is also possible that the

1 12 specimens are multi-functional implements. Table 14 summarizes the quantitative

properties of these six lithic specimens. Unfortunately, five of the six projectile

points/knives were recovered from disturbed contexts.

Specimen 650, recovered from Zone I in unit 510.5N 4985, resembles a Gary

Contracting Stemmed (or Cogswell) point of the Dickson Cluster. It has straight to

slightly excurvate blade edges, squared to slightly upsloping shoulders, and a biconvex

cross section. Although the stem appears snapped, a characteristic attribute of Cogswell

points in eastern Kentucky (C.R. Ison, pens, comm.), it is more parallel-sided than the

stems of Gary Contracting Stem type specimens described in Justice (1987). It is made

of Kanawha chert. Little use w ear is apparent on the edges under low magnification.

Specimens 894 and 1309 derived from disturbed Zone I, the former in unit

510.5N 498E and the latter in unit 510.5N 5005. Both are Adena-like points of the

Dickson Cluster. Specimen 894 exhibits characteristically heavy stem grinding, weak side notches, a biconvex cross section, and a contracting stem, but it is rather poorly flaked and generally smaller in size than most Adena points described by Justice

(1987). The blade and stem are partially broken, but there is no evidence of blade reworking. Ridges separating flake scars on both surfaces are smoothed or polished.

There is no indication of edge damage or use wear. Specimen 1309 is also an incomplete specimen. It is the heavily ground, contracting stem of a hafted biface broken at the shoulder. Breakage is thermal in nature. Specimen 894 is made of Boyle chert, while specimen 1309 is fashioned from Kanawha.

Specimen 1237 is a Paoli projectile point recovered from Zone I in unit 510.5N

5005. The expanding stem, angular shoulders, flattened blade face, and lateral stem grinding indicate the specimen is a Lowe Flared Base point of the Lowe Cluster. Because the blade edges are broken, there is no indication of use wear on the specimen.

1 13 Specimen 574 appears to be a rather crudely flaked triangular point.

Typologically it most closely resembles a Madison point, although it is rather thick and the maximum width is not at the base. It might be that the tool was not finished. The tip is snapped, and there is little edge dam age. Workers recovered this Boyle point from 1/4 inch screening of backfill from the western half of the 1984 excavation trench.

Field sample 54412 yielded the only projectile point fragment from an un­ disturbed context. This Breathitt stem fragment, provisionally identified as part of a

Little Bear Creek point, derived from Feature 59, an Early Woodland feature originating in Zone II. Broken where the stem begins to flare out to the shoulder, the contracting stem is ground laterally.

Spatially, Cold Oak Shelter falls within the recorded ranges of the five point types. Justice (1987) reported Gary Contracting Stemmed points from Kentucky, southern Illinois and Indiana, Missouri, Tennessee, Alabama, Louisiana, Arkansas,

Mississippi, and Texas. Adena Stemmed points are found over most of the eastern United

States, but the core area is the Ohio Valley. Sites in Kentucky, Tennessee, Alabama, southern Indiana, and southern Illinois yield Little Bear Creek points. Justice (1987) indicated that Lowe Flared Base points derive from Kentucky, Tennessee, Ohio, Indiana,

Illinois, and eastern Missouri. Madison points have been reported in all states east of the

Mississippi River and into Canada (Justice 1987).

The projectile points/knives are important as temporally sensitive artifacts.

According to Justice (1987), Gary Contracting Stem (Cogswell) points are diagnostic of the Terminal Archaic to Middle Woodland periods, from about 1500 B.C. to A.D. 100.

Ison (1988) and O'Steen et al. (1991) note that Gary Contracting Stemmed (Cogswell) points are diagnostic of the Cogswell Phase in eastern Kentucky, dated at 1250 to 800

B.C. (O'Steen et al. 1991:25). Ison (1988) first defined the Cogswell Phase based on the

12 This artifact has no specimen number. It was recovered from a 0.6 mm soil sample after the lithic lab analysis was completed.

1 1 4 1984 excavations at Cold Oak Shelter. O'Steen et ai. (1991) associated Gary Contracting

Stemmed (Cogswell) points with the earlier Skidmore Phase (2750-1750 B.C.) as

well, but the radiocarbon dates for Cold Oak Shelter (Gremillion 1995) are more in line

with the Cogswell Phase interpretation.

Little Bear Creek points date to the Terminal Archaic-Early Woodland transition,

from 1500 to 500 B.C. (Justice 1987). Most Adena Stemmed points from other sites date to the Early Woodland Period, from about 800-300 B.C. (Justice 1987).

Lowe Flared Base points are diagnostic of the terminal Middle Woodland-Late

Woodland Period, with dates ranging from A.D. 200 to 600 (Justice 1987). O'Steen et al. (1991:30) noted that "expanded stemmed and side notch points, including Lowe

Flared Base ... became the most common projectile point/knife forms during the latter part of the Middle Woodland" in eastern Kentucky. People continued to manufacture the bifaces into the Late Woodland period in the study area. Ahler (1988) argued that Lowe points are characteristic of the Early and Late Newton Phases of the Late Woodland period

(A.D. 500 to 1000) in eastern Kentucky.

Justice (1987) attributed triangular points to the Late Woodland and

Mississippian Periods, with dates ranging from A.D. 800 to 1350 at other sites.

Based on this information, people could have occupied Cold Oak Shelter as early as

1500 B.C. to as late as A.D. 1350. The dates for the bifaces correspond nicely to radio­ carbon dates for the shelter (see Chapter 3), especially if one excludes from consideration the questionable Madison point. Two-sigma calibrated dates from the shelter range from 1400 B.C. to A.D. 400 (Gremillion 1995). Unfortunately, most of the diagnostics derived from disturbed deposits, and radiocarbon dates are not available for Feature 59, which produced an Adena-like biface fragment.

The five hafted biface fragments from Cold Oak Shelter lack sufficient physical attributes for assigning them to temporally sensitive types. They do, however, retain

1 1 5 evidence of hafting elements. The five stem fragments are made of Paoli, Breathitt,

unidentified, Pennsylvanian (Breathitt or Kanawha), and Haney cherts (Table 13). The

biface stems derived from three units of the excavation trench; 510.5N 497E, 510.5N

498E, and 510.5N 500E (Table 10). Stem specimens were recovered from Zones I, II,

and III (Table 11), corresponding to disturbed. Early Woodland, and Terminal Archaic

period zones (Table 12).

Ten chipped-stone tool specimens are fragments of blanks, preforms, or finished

bifaces. It is possible that they functioned as projectile points, knives, scrapers, or

other implements. It is also possible that knappers did not work them into usable tools.

Without microscopic use-wear analysis, it is difficult to interpret the functional life-

histories of the biface fragments. The broken bifaces were recovered from all excavation

units except 510.5N 497E (Table 10). The only stratigraphie zones yielding biface

fragments were Zones I and II (Table 11). Only two could be assigned to a cultural

period. Early Woodland (Table 12). Five chert types are represented: Boyle, Haney, St.

Louis, Kanawha, and Paoli (Table 13).

One of the biface fragments, specimen 904, is a utilized flake blank. Made of St.

Louis chert, this artifact shows clear signs of use wear in the form of rounding,

grinding, and step fractures. The use damage occurs unifacially on one lateral margin of the blank. Another lateral edge is broken. The utilized blank was recovered from Zone I of unit 510.5N 499E.

The Cold Oak assemblage contains one marginally modified flake. Specimen 1071 is a nearly complete secondary flake made of St. Louis chert. Both lateral edges of the flake are broken, and one is bifacially reworked. The flaked edge is ground and dulled, presumably from use-wear. This specimen derived from unit 510.5N 498E in Early

Woodland Zone lie deposits.

11 6 Specimen 1280 is a utilized flake. Made of unidentified cliert, it is ttie medial portion of a primary flake. Use-wear damage is present on one lateral edge of tfie specimen. The specimen was excavated from Early Woodland Zone lib deposits in unit

510.5N 499E.

The 23 tools and tool fragments make up a small but important part of the chipped-stone lithic sam ple. But debitage is the most common form of chipped-stone lithics in the Cold Oak Shelter assem blages (Table 9), and 1856 specim ens or alm ost

83% of the assemblages are debitage. Debitage density for the shelter is 0.62/liter, excluding specimens found outside the 1994 excavation area.

The horizontal distribution of debitage is almost identical to that of the entire chipped-stone sample, as one might expect since 99% of the chipped-stone material is debitage. Most of the debitage derived from disturbed Zone I, Zone II, and unknown deposits. Although over half of the debitage could not be assigned to a cultural period, the remaining specimens are clearly concentrated in Early Woodland and Terminal Archaic deposits, with about half as much in the latter than the former.

Though most debitage is made of unidentified chert, all identifiable chert types are represented. Haney debitage, followed by St. Louis and Boyle, are the most abundant chert types. Limestone, iron ore, and quartz debitage is also recorded.

Six worked cores were recovered from Cold Oak Shelter (Table 9). All are multidirectional, and none show evidence of use-wear. Only two specimens came from identifiable contexts, in Early Woodland Zone II deposits of unit 510.5N 498E. Specimen

1109a, which workers recovered from Early Woodland Zone lib, is made of unidentified chert and less than 25% of the surface retains cortex. Recovered from Early Woodland

Zone lie in unit 510.5N 498E, specimen 2005a is a St. Louis core retaining no cortex.

The other four cores derived from undated contexts. Specimen 642a is a St. Louis core recovered from the surface of looter's backfill dirt near the 1994 excavations. Less

1 1 7 25% of the surface retains cortex. Two Muldraugh cores were found on the surface of the shelter near the dripline. Specimens 552a and 552b retain less than 50% cortex.

Specimen 1027 came from the surface of the shelter near the 1984 excavation trench.

This Pennsylvanian chert core has less than 25% cortex and is the only core showing evidence of thermal alteration.

Flakes are the most common type of chipped-stone debitage, and flakes account for about three-quarters (n=1695) of the lithic assemblages (Table 9). Average dimensions of the flake sample are: 1.33 cm long, 1.29 cm wide, 0.28 cm thick, and

0.64 grams in weight. Flake distributions are similar to the chipped-stone debitage distributions discussed previously. As with the debitage, there are twice as many chipped-stone flakes from Early Woodland deposits than Terminal Archaic strata, though about half of the flakes could not be assigned to a cultural period.

Tables 15 and 16 summarize the distribution of flake types. The most common reduction class is broken flakes, which comprise 27% of the flake sample. Secondary flakes and thinning flakes account for 22% and 20% of the flake debitage, respectively.

About 16% of the flakes are secondary decortication flakes, and 14% are primary flakes. Primary decortication flakes and blades each constitute 1% or less of the sample.

Table 16 shows the distribution of flake fragment types for each reduction category. Overall, proximal flakes are by far the most abundant flake fragment type, accounting for 46% of the sample. Medial flakes (25%) are the next most common, followed by complete flakes and distal flakes at 16% and 14%, respectively. Primary decortication flakes and blades are fairly evenly distributed across the four flake fragment types. Most of the secondary decortication flakes are proximal. Most primary flakes, secondary flakes, and thinning flakes clearly are proximal pieces as well. Broken flakes are, by definition, medial and distal exclusively. These distributions are not unexpected when one considers how the flake reduction classes are defined.

1 1 8 All chert types are recorded in the flake sample, but Haney, unidentified, and St.

Louis cherts are most common (Table 13). Over half of the primary decortication flakes

are made of St. Louis chert. St. Louis is also the most common chert in the secondary decortication sample, but the percentage drops to 27%. This relative abundance of St.

Louis flakes with cortex corresponds nicely to the core sample, as two of the four

identified cherts are St. Louis. Haney and unidentified cherts are also abundant in the secondary decortication sample. Haney clearly dominates in the primary flake sample,

accounting for almost half of the specimens. Unidentified, Haney, and St. Louis cherts occur in fairly equal proportions in the secondary flake sample. Most broken flakes are

made of unidentified chert, and thinning flakes are dominated by approximately equal

numbers of unidentified and Haney cherts. Half of the blades are made of Haney.

Over half of the Boyle chert occurs as secondary flakes and thinning flakes. One- third of the Lower Pennsylvanian (Breathitt and Kanawha) specimens are broken flakes, followed by secondary decortication at 21%. Muldraugh flakes are fairly evenly distri­ buted across secondary decortication, secondary, thinning, and broken flake samples. The highest percentage of Paoli flakes are thinning flakes. Only three flake types, secondary decortication, thinning, and broken flakes, are made of Ste. Genevieve.

Debris is the third type of chipped-stone debitage. Debris accounts for over 8%

(n=156) of the debitage specimens and about 7% of the entire lithic assemblage (Table

9). Angular shatter debris (n=109) is more abundant than blocky irregular chunk debris (n=47). About 43% of the debris retains cortex. The lateral and vertical distributions of debris follow the general patterns of the chipped-stone lithic sample.

Haney, Boyle, Breathitt, and St. Louis together account for m ost of the identified chert debris artifacts. Chunk is made of a wide variety of materials, but unidentified chert and Haney chert predominate. Decortication chunk is also made mostly of Haney and

unidentified chert, but the range of materials is limited. All raw material types except

1 1 9 quartz are present in the shatter sample, with unidentified, Haney, and Boyle cherts

being the most common. Nearly half of the decortication shatter is made of unidentified

cherts, followed by Haney and Breathitt.

4.3.1.d. Thermal Debris

The 1994 investigations at Cold Oak Shelter yielded over 200 pieces of lithic

debris produced by heating (Table 4). Most of the debris (76%) is fire-altered rock;

the remaining 24% is fire-altered chert. The density of thermal debris in the 1994

excavation block is 0.08/liter.

Tables 6, 7 and 8 summarize the distribution of thermal debris at the shelter. As

indicated in Table 6, the highest percentage (29%) and density (0.11/liter) of thermal

debris derived from unit 510.5N 500E. About 24% was recovered from unit 510.5N

498E, 17% from 510.5N 499E, and 12% each from units 510.5N 497E and 510.5N

501E; about 6% derived from non-unit contexts.

The distribution of thermal debris by cultural zone closely follows the pattern for the entire assemblage (Table 7). The highest percentages are recorded for Zones I

(38%) and 11 (21%), followed by 14% each from Zone 111 and features, and 13% from

Historic and unknown contexts. Historic deposits yielded the highest density of thermal debris at 0.25/liter.

As shown in Table 8, the vertical distribution of thermal debris by period also parallels that for the entire lithic assemblage. About 56% of the specimens could not be assigned to a period, but the remaining specimens are concentrated in the Early

Woodland (24%) and Terminal Archaic (18%) layers and features. The remaining 2% derived from Middle Woodland and Historic contexts. The highest density of thermal debris, 0.09/liter, is associated with Early Woodland deposits.

About 45% (n=56) of the thermal debris is fire-altered chert. Fire-altered chert is represented by equal numbers of pot lids and heat spalls. All but four of these

120 specimens were found during excavation of test units. The two westernmost units.

510.5N 497E and 510.5N 498E, contained the highest percentage of chert thermal debris. Adjusting for differences in sediment volumes, the two end units (510.4N 497E and 510.5N 501E) yielded the highest densities of fire-altered chert at 0.02/liter each.

Vertically, the distribution of fire-altered cherts follows that of the entire thermal debris sample.

Excluding unidentified specimens (39%), the most common chert type represented in the altered chert sample is Haney, which accounts for 27% of the specimens. Fire-altered cherts made of Breathitt, St. Louis, Boyle, Muldraugh,

Kanawha, and Paoli cherts together account for the remaining 34%. None of the thermal debris is made of Ste. Genevieve chert.

Non-chert thermal debris is fire-altered rock and miscellaneous materials.

These specimens comprise over 75% of the thermal debris, and about 8% of the entire lithic assemblage from the shelter. Fire-altered sandstone outnumbers fire-altered limestone by a ratio of about 17:1. Other remains are a fire-altered concretion and a fire-altered quartz pebble. In terms of weights, the fire-altered sandstone totaled

4,117.6 g, while the fire-altered limestone totaled 144.4 g. Taking into account sample sizes, the average weights are 24.8 g and 14.4 g, respectively.

Horizontally and vertically, the distribution of fire-altered rock follows that of the entire thermal debris sample. The average weight of specimens, however, is slightly larger for Terminal Archaic layers and features than for other deposits. Fire-altered limestone is concentrated in Early Woodland deposits.

4.3.I.e. Miscellaneous Lithics

The 125 specimens in this category comprise over 5% of the Cold Oak Shelter lithic assemblages (Table 4). Sedimentary rock fragments comprise about 60% of the miscellaneous lithics in the Cold Oak assem blages (Table 7). In descending order of

121 abundance, river cobbles make up about 17% of the miscellaneous lithics, natural chert accounts for 14%, 8% is minerals, and 1% is metamorphic rock fragments. Overall, the density of miscellaneous lithics at the shelter is 0.04/liter, excluding those specimens found outside the 1994 excavation block.

Sedimentary rock fragments identified as artifactual differ in composition from the shelter bedrock. Of the sedimentary rock fragments, most are limestone. Chalk-like specimens, dolostone, and petrified wood were also recovered.

River cobbles and cobble fragments have one or more rounded surface(s). Most are composed of limestone. The sandstone river cobbles are often cracked: the origin of the fractures is uncertain, but they probably did not arise from heating. The nearest stream from which the river cobbles and cobble fragments could have derived is Big

Sinking Creek, which is less than one mile from the shelter. Water-worn cobbles presently are observed in that stream.

Natural chert specimens show no signs of cultural modification. Most are composed of unidentified chert. The typed natural chert specimens are Haney, Boyle, and

Breathitt. Minerals include hematite, limonite, and gypsum. The two metamorphic rock fragments recovered from the shelter are quartzite and mica schist.

Horizontally, units 510.5N 500E and 510.5N 497E have the highest percentages and densities of miscellaneous lithics; the former yielded 30% and a density of 0.06/ liter while the latter had 29% and a density of 0.05/liter (Table 6). These units also contained the highest percentages of river cobbles and fragments, metamorphic rock fragments, and sedimentary rock fragments. Unit 510.5N 498E yielded 22% of the miscellaneous lithics, 510.5N 499E produced 11%, and 4% derived from both 510.5N

50IE and contexts outside the excavation block.

Disturbed Zone I contained the highest percentage of miscellaneous lithics at 44%

(Table 7). Zone II and Zone 111 yielded 23% and 12%, respectively, and 15% could not

122 be assigned to a cultural zone. Feature fill contained the remaining 6%. Most of the river cobbles, sedimentary rock fragments, and natural chert specimens came from Zones I and II.

In terms of cultural periods, 65% of the specimens could not be assigned to a period (Table 8). Otherwise, the highest percentages of miscellaneous lithics are from

Early Woodland (21%) and Terminal Archaic (14%) strata and features. Lithic densities follow a similar pattern. River cobbles, minerals, and sedimentary rock fragments are fairly evenly distributed in Early Woodland and Terminal Archaic deposits. Metamorphic rocks were found in Terminal Archaic and unknown contexts, and natural chert predominates in unknown and Early Woodland deposits.

4.3.1 f. Ground-Stone Artifacts

Three lithic specimens are fragments of ground-stone artifacts. All derive from subsurface deposits of Cold Oak Shelter, and none show signs of thermal alteration.

Specimens 584 and 585 appear to be fragments of the same ground-stone artifact, perhaps an atlatl weight or gorget, although they do not fit together. Composed of fairly soft dolostone, the specimens exhibit shallow striations on one side and intentional squaring of edges. Specimen 584 is an edge piece with striations oriented in two directions: adjacent to the edge the striations run parallel to it, but away from the edge the striations are oriented at a 45° to 50° angle. It is 1.95 cm x 1.70 cm x 1.19 cm in size and weighs 4.33 grams. Specimen 585 is a corner piece with striations running approximately parallel to one edge. It is 3.30 cm x 3.14 cm x 1.31 cm in size and weighs 13.51 grams. Both fragments derived from unit 510.5N 500E at 59-60 cm below datum. Specimen 584 was recovered at the interface of disturbed Zone I and

Terminal Archaic Zone III, specimen 585 was found in Zone III.

A ground-stone fragment with only one intact edge was excavated from disturbed

Zone I of unit 510.5N 501E. Specimen 797i is a striated fragment apparently detached

1 23 from the surface of a limestone artifact. The dorsal surface exhibits shallow striations oriented in three directions in an obtuse-chevron manner, while the ventral surface retains a bulb of force indicating detachment from the surface of a larger artifact. There is, however, no striking platform, suggesting the detachment was accidental or post- depositional. This fragment is 1.30 cm x 1.48 cm x 0.27 cm in size and weighs 0.54 grams.

4.3.2. ROCK BRIDGE SHELTER LITHIC ASSEMBLAGE

4.3.2.8. Overview

Table 4 summarizes the composition of the Rock Bridge Shelter lithic assemblage. Like the assemblage from Cold Oak Shelter, chipped-stone artifacts dominate the Rock Bridge assem blage, accounting for over 99% of the lithics. Ground-stone artifacts, thermal debris, and miscellaneous lithics together represent only 1% of the assemblage. The lithic density for the entire shelter is 0.42/liter.

4.3.2.b. Spatial Distribution

Table 17 sum m arizes the distribution of lithic artifacts recovered by system atic surface collection and excavation of various areas of Rock Bridge Shelter. The largest proportion (90%) of the lithic assemblage came from the southern (main) excavation block. About 3% and 2% of the assemblage derived from the northern and middle excavation blocks, respectively. Almost 5% of the lithics were recovered from the surface of areas outside the excavation blocks.

Adjusting for the differences in sediment volumes among the excavation blocks, a similar pattern emerges. The densest concentration of lithics (0.48/liter) came from the southern excavation block. The density of lithics in the excavation units of the

124 northern block is about half of this at 0.21/liter. Only 0.05 lithics/liter were

recovered from the two excavation units of the middle block.

Not only was the greatest total number of lithics recovered from the southern

excavation block, but the largest numbers of most lithic types derive from here as well.

About 81% of the chipped-stone tools came from the main block. Over 91% of the debitage and all of the thermal debris and miscellaneous lithics derived from surface and subsurface deposits of the southern block.

Table 18 presents the vertical distribution of artifacts from Rock Bridge

Shelter. For the shelter as a whole, 65% of the lithics derived from the silty midden layer of Zone II. Thirty-two percent derived from the upper ashy midden stratum of

Zone I, while only 3% came from the sandy subsoil of Zone III. This pattern is repeated for the southern excavation block as well. The middle excavation block, on the other hand, has a high percentage (91%) of lithics from the subsoil layer. The remaining 9% derived from Zone I. In the northern excavation block, all lithics derived from the ashy midden and silty midden layers in roughly equal proportions.

One must evaluate the vertical lithic distributions at Rock Bridge with caution since only 69% of the specimens could be vertically provenienced. Many artifacts were difficult to place stratigraphically because some of the five cm excavation levels cut across two strata and deposits are relatively thin compared to those at other shelters like Cold Oak. Any specimens fitting these criteria were omitted from stratigraphie consideration, but all lithics from feature fill are included in the silty midden layer if vertical provenience data are lacking. This is feasible for features that were not defined until the upper ashy midden layer was removed.

Table 19 summarizes the distribution of raw material types for each lithic category and for the entire lithic assemblage. Over half of the lithic specimens are composed of Haney chert, and 22% are made of Paoli chert. Other cherts account for an

1 25 additional 16%. The remaining specimens are made of sandstone, limestone, hematite,

quartz, and unidentified materials.

4.3.2.C. Chipped-stone Artifacts

Almost all the lithic remains of the Rock Bridge Shelter assemblage are chipped-

stone artifacts of two groups; tools and tool fragments and debitage. The former accounts for 3% of the chipped-stone artifacts, the latter, 97%. Because the chipped-stone artifacts, and the debitage for that matter, make up such a large proportion of the entire

lithic assemblage, descriptions of the spatial and raw material distributions for the chipped-stone artifact sample follows those of the entire Rock Bridge assemblage and are not repeated here.

Twenty-six chipped-stone tools and tool fragments are identified by generalized edge treatment and edge wear characteristics. Most of the tool specimens (81%) were recovered from the southern excavation block, 11 % were found outside excavation blocks, and 4% each derived from the middle and northern blocks (Figure 15, Table

19). Zone II yielded 54% of the tool specimens, 15% each derived from Zones I and III, and 15% could not be placed confidently in one layer (Table 19).

Seven bifacial tools and tool fragments were recovered from Rock Bridge in

1992, six from the main excavation block and one that could not be provenienced. Of these six, five came from the Zone II midden layer and one derived from ashy Zone I

(Table 20). Within the main excavation block, the bifaces are concentrated at the northern and southern ends (Figure 15).

Paoli chert appears to have been used more often than other cherts for biface manufacture (Table 19). Five of the seven (70%) specimens are made of this chert type; one is Haney and one is Breathitt (15% each). Three of the four Paoli bifaces for which provenience data is available derived from the south-southwestern part of the main excavation block (Figure 15).

126 Three of the seven bifaces are complete specimens; the other four are broken fragments of (presumably) tools. Table 12 outlines some physical attributes of the complete bifaces. Two of the complete hafted bifaces might be classified as projectile points/ knives and the third may have been used as a perforator.

Specimen 71, recovered from the southern end of the main excavation block, closely resembles a Lowe Cluster point, probably Bakers Creek. Low magnification examination of the edges revealed some evidence of use wear. Specimen 251 was found in the northern end of the main excavation block. It is a Breathitt biface, typologically classified as a Chesser Notched projectile point. This biface exhibits the most edge wear of the three complete biface specimens; the tip and margins are very chipped and worn.

Based on experimental and microwear analyses, Yerkes and Pecora (1991:105) proposed that Lowe and Chesser hafted bifaces in the Putnam County, West Virginia area may have functioned as dart points.

These two hafted bifaces are important as temporally sensitive artifacts.

According to Justice (1987:214), Chesser Notched points are diagnostic of the terminal

Middle Woodland to Late Woodland periods, from about A.D. 300 to 700. Most Bakers

Creek points from other sites are dated from A.D. 150 to 600, the terminal Middle

Woodland Period (Justice 1987:211-212). Based on this information. Rock Bridge

Shelter could have been occupied as early as A.D. 300 to as late as A.D. 600. The dates for the bifaces agree well with the radiocarbon dates for the shelter (Gremillion 1995; see

Chapter 3). Ahler (1988) noted that Lowe and Chesser points are characteristic of

Early and Late Newton Phase sites in eastern Kentucky. Spatially, Rock Bridge Shelter falls within the recorded ranges of these two point types. Chesser Notched points derive from eastern Kentucky, south-central Ohio, West Virginia, Pennsylvania, and eastern

New York sites. Archaeologists reported Bakers Creek points from Kentucky, Tennessee,

1 27 Georgia, Mississippi, Louisiana, West Virginia, southern Indiana, and southern Ohio

(Justice 1987:212-214).

Specimen 250 is a hafted perforator. It is more than twice as long as wide, has roughly parallel blade edges, and tapers to a point. Made of Haney chert, the specimen derived from the northern end of the main excavation block in the vicinity of the northern half of Feature 17. Lateral margins are serrated and exhibit some indication of use wear.

Eleven lithic specimens are marginally modified flakes (Table 20). About 64%

(7 of 11) of the marginally modified flakes are primary (n=1) and secondary (n=6) flakes. The other four marginally modified flakes are tertiary and bifacial thinning flakes that probably represent the later stages of lithic reduction. One marginally modified flake resembles a spokeshave and several others may have served as side or end scrapers. Compared to bifacial tools, the marginally modified flakes are made of a wider range of raw materials (Table 19).

Marginally modified flakes derived from all three excavation blocks at Rock

Bridge Shelter, with 73% from the main excavation block, and 9% each from the middle block, northern block, and outside the block (Table 20). Many of the marginally modified flakes in the main block were found at the northern end (Figure 15).

Vertically, six of nine (67%) of the marginally modified flakes that could be provenienced cam e from the Zone II, with two (22%) from Zone III and one (11%) from

Zone I (Table 20). No patterning is apparent in the horizontal distribution of marginally modified flakes by raw material type in the main excavation block (Figure 15).

Eight lithic specimens show evidence of use wear but no form of intentional edge modification. In general, the utilized flake tools are smaller than the marginally modified flake tools. There is also a predominance of tertiary and bifacial thinning flakes as opposed to initial reduction flakes. Compared to marginally modified flakes, utilized

128 flakes are made of fewer raw material types (Table 19). The distribution of utilized

flakes by raw material for the main excavation block shows no apparent pattering

(Figure 15). Utilized flakes were found in the main excavation block only, where there appear to be clustered at the northern, middle and southern parts of the block (Table 19 and Figure 15). Vertically, utilized flakes are fairly evenly distributed in the three

major strata.

Debitage is the most common form of chipped-stone lithics in the Rock Bridge

Shelter assemblage. Over 95% (n=723) of the assemblage is debitage. Tables 17 and 21 indicates how the distributional patterns of debitage follows that of the entire lithic assemblage. Most specimens were recovered from Zone II of the main excavation block.

As indicated in Table 19, Haney and Paoli cherts are the most common raw materials represented in the debitage sample. These cherts account for 74% of the debitage sample. Boyle and unidentified materials each make up 9% of the debitage, and other cherts and materials account for the remaining 8% .

Two of the debitage specimens are worked cores recovered from surface contexts at Rock Bridge Shelter. A core of Paoli chert was found during systematic surface collection of the shelter in unit 40N52E, south of the main excavation block. Forest

Service archaeologists found a pot-lidded Breathitt core at the shelter; provenience data is not available. Both specimens have scars indicating the removal of cortical flakes.

About 15% of the Paoli core surface has cortex, while the Breathitt core is about 60% cortex.

The main excavation block at Rock Bridge Shelter yielded 238 pieces of microdebitage, or lithic waste too small to be classified into types. All microdebitage came from 1/16 inch and 0.6 mm screenings of unit and feature fill. There were three concentrations of microdebitage, at the northern, southern, and middle parts of the block. Vertical provenience data are available for 216 of the 238 pieces of

1 29 microdebitage, and this microdebitage was vertically distributed in Zone II (72%) and

Zone I (28%). Despite fine screening, no microdebitage was recovered from Zone III.

While nearly one-quarter of the sample could not be securely typed due to small

fragment sizes (despite the use of magnification), there seems to be a predominance of

Haney fragments (Table 19).

About 94% macrodebitage is flakes (Table 4). Most derived from the southern

excavation block and the silty midden layer. When one plots the horizontal distribution

of flakes by specimen count and specimen density for the main block, three clusters that

correspond to those for all debitage are evident in the southern, middle and northern

portions of the block (Table 21). Average dimensions of the flake sample are: 1.53 cm

long, 1.48 cm wide, 0.29 cm in thick, and 1.09 grams in weight. About 16% of the

flakes are complete, 41% are proximal, 26% are medial, 14% are distal, and 3% could

not be classified since they lacked two intact lateral margins.

As Table 19 indicates, the flakes are made of a wide range of raw materials.

Haney accounts for 51% of the flake specimens and 25% are Paoli. About 10% of the

flakes are made of Boyle, and the remaining 14% are other materials.

In terms of reduction classes, broken flakes are the most common at 38% of the

flake sample (Table 22). Secondary and bifacial thinning flakes are also relatively

abundant and account for 21% and 23% of the sample, respectively. About 11% of the

flakes are secondary decortication flakes and 6% are primary. Only one primary decortication flake was identified.

Table 22 also summarizes the distributions the flake reduction classes by raw

material for Rock Bridge Shelter. Locally available Haney and Paoli, the most abundant

cherts in the assemblage, account for the highest percentages of early-stage flakes.

Boyle, Breathitt, St. Louis, and Kanawha are predominantly late-stage flakes. Ste.

Genevieve flakes are exclusively late stage.

130 Thirty-one pieces of debris represent about 6% of the debitage sample (Table

5). As with most lithic categories, the majority of the debris derived from the southern excavation block and Zone II. Although several raw material types occur in the debris sample, three cherts comprise the majority of specimens - Paoli, Haney, and Boyle

(Table 19).

4.3.2.d. Ground-stone Artifacts

Three lithic specimens are ground-stone artifacts and fragments. U.S. Forest

Service archaeologists recovered the specimens during systematic surface collection of the rockshelter to the west of excavation unit 49N50E in a niche along the back wall.

None of the ground-stone artifacts shows evidence of heat alteration.

Two lithic specimens are ground-stone hammerstones. Both are asymmetrically spherical with flattened tops and bottoms and smoothed surfaces. Specimen 26 is made of limestone and has numerous chips on its edges and small pits on one of the flattened surfaces. It is 10.78 cm x 11.02 cm x 5.39 cm and weighs over 400 grams. Specimen

27 is a hematite concretion with some chips along its edges. It is 9.95 cm x 12.30 cm x

5.02 cm and weighs over 400 grams. Because its shape is probably the result of geologic rather than cultural processes, the hematite concretion may not qualify as a ground- stone artifact as defined in this study. Perhaps it should be categorized as an unaltered but utilized lithic artifact. Nonetheless, Specimen 27 is included in this lithic category since its function was likely similar to that of Specimen 26. The third specimen is the fragments of the broken hematite hammerstone.

4.3.2.e. Thermal Debris

Two small, circular, plano-convex lithic specimens are identified as pot lids.

Both derived from the northern end of the main excavation block. They are not associated with a specific stratum due to insufficient provenience data. One pot lid, made of

131 Breathitt chert, has a pot lid fracture on its dorsal surface. The sandstone pot lid shows no additional signs of thermal alteration.

4.3.2.f. Miscellaneous Lithics

The frost lid recovered from the main excavation block is a relatively large

(1.55 X 1.83 X 0.24 cm), elongate, plano-convex fragment of Haney chert. Its dorsal surface has one pot lid fracture, indicating unintentional heat alteration. Since it was recovered during general cleaning of the southern excavation block, the frost lid cannot be provenienced to a specific excavation unit or stratum.

4.4. SUMMARY

Chapter 3 demonstrated similarities and differences in the geology and archaeological stratigraphy of Cold Oak and Rock Bridge shelters. In this chapter, which describes the nature of the lithic assemblages from the two shelters, it is evident that the assemblages themselves share many characteristics but also differ in some ways.

The lithic assemblages from both shelters are composed of chipped stone, thermal debris, miscellaneous lithic, and ground stone specimens. The former is the predominant lithic artifact type at the shelters, accounting for 84% of the Cold Oak assemblage and

99% of the Rock Bridge assemblage. Thermal debris is more abundant in the Cold Oak assemblage, accounting for 10% of the specimens, whereas less than 1% of the Rock

Bridge lithics is thermal debris. Almost 6% of the Cold Oak specimens are made of miscellaneous lithic materials, whereas less than 1% of the Rock Bridge artifacts are miscellaneous lithics. Less than 1% of both assemblages is ground-stone artifacts. Lithic densities for Cold Oak and Rock Bridge are 0.75/liter and 0.42/liter, respectively.

Within the chipped-stone samples from both shelters, debitage (especially flakes) is the

1 32 predominant artifact type. Besides broken flakes, secondary and tfiinning flakes are the most common flake types in both assemblages.

Chert is the most common raw material in the shelter assemblages, accounting for 87% of the Cold Oak specimens and about 89% of the Rock Bridge specimens. About

13% of the Cold Oak lithics are various types of rocks, including sedimentary and metamorphic types, compared to less than 2% of the Rock Bridge assemblage. Minerals make up less than 1% of each assemblage. About 9% of the raw materials in the Rock

Bridge assemblages were unidentifiable.

Lithics from the 1994 excavations at Cold Oak, including the chipped-stone tool sample, were concentrated in the two units (510.5N 498E and 510.5N 500E) flanking the large rock in the center unit. Vertically, the number and density of lithics from

Early Woodland contexts is almost double that of Terminal Archaic deposits. The ratio of chipped-stone tool fragments in Early Woodland compared to Terminal Archaic contexts is 1:6. Only a small percentage of the lithics derived from Middle Woodland and Historic deposits.

At Rock Bridge Shelter, lithics were concentrated in the southern (main) excavation block, as were the chipped-stone tool sample. The largest proportion of specimens, including the chipped-stone tool sample, were recovered from the silty midden deposits of Zone II, but a significant percentage derived from ashy Zone I.

Diagnostic lithics were recovered from both shelters. Gary Contracting Stemmed

(Cogswell), Little Bear Creek, Adena Stemmed, Lowe Flared Base, and possibly Madison triangular projectile points were found during the 1994 excavations at Cold Oak Shelter.

These bifaces are associated with Terminal Archaic through Mississippian periods. At

Rock Bridge Shelter, Chesser Notched and Bakers Creek points, both of which are associated with the terminal Middle Woodland-Late Woodland periods, were recovered.

1 33 Before the lithic assemblages from Cold Oak and Rock Bridge shelter can be used to test hypotheses about diachronic patterns of prehistoric rockshelter use, the integrity of the assemblages must be evaluated. The next chapter examines this issue in detail, using Schiffer's (1987) behavioral model of formation processes as a guideline, with special emphasis on formation processes typical in rockshelter settings.

134 CHAPTER 5

FORMATION PROCESSES AND THE INTEGRITY OF THE LITHIC ASSEMBLAGES

Formation processes encompass a variety of cultural and noncuitural processes that influence the deposition, accumulation, distribution, and condition of material remains in both systemic and archaeological contexts. The systemic context refers to

“artifacts when they are participating in a behavioral system" while the archaeological context deals with “artifacts that interact only with the natural environment" (Schiffer

1 9 8 7 :3 -4 ).

Archaeologists have studied for decades formation processes and their effects on the archaeological record. Over the years, general conceptions of formation processes and procedures for investigating them have changed. Generally speaking, one might argue that a relatively optimistic perspective of formation processes has gradually replaced a more pessimistic view. Whereas in the past archaeologists might have concluded that formation processes limit drastically what we can learn from some parts of the archaeological record, archaeologists are less likely to shy away from studying heavily impacted deposits. This is a welcome development since, as indicated in Chapter 2, rockshelters in the study area tend to be disturbed more often than not.

More specifically, three notions about formation processes have evolved

(Schiffer 1987). First, archaeologists are replacing the "entropy" concept of formation processes, or the idea that alteration of the archaeological record due to formation

135 processes is proportional to the length of time the materials were deposited, with the contention that alteration depends on the types of processes operating in a particular situation. Second, whereas archaeologists had focused on how formation processes influence what materials are available for recovery, they are increasingly concerned with retrieving information from severely impacted deposits and from ecofacts as well as artifacts. Third, archaeologists have demonstrated that most formation processes impart identifiable patterns on the archaeological record, so the prior contention that formation processes hopelessly transform or distort the archaeological record is perhaps overly despairing. Schiffer (1987:11) concluded that "it has been shown that formation processes (1) transform items formally, spatially, quantitatively, and relationally, (2) can create artifact patterns unrelated to the past behaviors of interest,

[but that they] (3) exhibit regularities that can be expressed as (usually statistical) laws."

Formation processes may alter artifacts in four ways (Schiffer 1987). Formal modifications change an artifact’s physical or chemical properties, including dimensions, morphology, and composition. Formation processes may alter the spatial attributes of artifacts, such that formation processes disturb previous patterns and/or create new patterns. Frequency modifications occur when formation processes alter the numbers of each type of artifact. Because formation processes may affect the relational attributes of artifacts, one cannot assume that “patterns of co-occurrence of artifacts" in an archaeological context "are determined by activity patterns” alone (Schiffer

1987:19-20). This is why a consideration of processes that may have affected the lithic assemblages at Cold Oak and Rock Bridge shelters is important.

Archaeologists document the operation of formation processes with artifacts, collections of artifacts, and deposits in which artifacts are found (Schiffer 1987).

Patterns in artifact size, sorting, density, shape, orientation and dip, accretions, and

1 36 damage suggest the action of specific formation processes. Additional indicators are artifact quantity, inventory, vertical distribution, horizontal distribution, diversity, disorganization, reassembly, and representation of parts. Sediment color, texture, composition or chemistry, fabric, and compaction as well as ecofacts and site morphology indicate the operation of certain formation processes.

Because regular causes and consequences govern formation processes, one can predict the effects of their influence and express the effects as low-range "experimental laws and empirical generalizations" linked to middle-range and high-range theories in anthropology and other sciences (Schiffer 1975:22). It is these laws for patterned regularities of formation processes that form the basis for evaluating formation processes at Cold Oak and Rock Bridge Shelters.

Archaeologists commonly dichotomize formation processes in terms of genesis.

According to Schiffer (1987:7), cultural formation processes are "the processes of human behavior that affect or transform artifacts after their initial period of use in a given activity." Noncuitural processes, on the other hand, are "any and all events and processes of the natural environment that impinge upon artifacts and archaeological deposits" (Schiffer 1987:7). Cultural processes encompass human activities such as introduction of material to the site, alteration of human imports, acceleration of natural physiogenic processes (Butzer 1982:77), artifact discard, artifact reuse, post­ occupation site disturbance (Butzer 1982:98-99), artifact loss, ritual caching, treatment of the dead, abandonment, reclamation, earth moving, trampling, plowing, and cleaning (Schiffer 1987). Some researchers (e.g., Butzer 1982:77) divided noncuitural formation processes into physiogenic and biogenic components. The former includes such general geologic processes as erosion, deposition, and mass wasting that are normal for the area of the site. Specific physiogenic processes are patination, hydration, oxidation, ionic dissociation, ionic substitution, leaching, creep, slump.

137 water erosion, wind erosion, volcanism, salt erosion, rising damp, freeze-thaw, cryoturbation, and thermal shock (Butzer 1982; Schiffer 1987). Biogenic processes refer to plant and animal activities such as burrowing, gnawing, root growth, bacterial decay, fungal decay, and insect decay (Butzer 1982; Schiffer 1987).

Because of the variety in formation processes, and because the interaction of any two processes may counteract or enhance each other, evaluation of the integrity of archaeological deposits is a complex endeavor. However, some suggestions are offered and illustrated in this chapter. The chapter begins with a discussion of formation processes typical of most rockshelters, especially sandstone shelters. The remainder of the chapter outlines a methodological program for investigating rockshelter formation processes using lithic assemblages and presents the results of the Cold Oak and Rock Bridge analyses. Cold Oak Shelter is obviously disturbed; evidence of post-depositional changes at Rock Bridge Shelter is less apparent but the potential exists. How might the operation of extraneous processes affect comparisons of shelter lithic assemblages?

5.1. ROCKSHELTER FORMATION PROCESSES

Rockshelters are dynamic landscape features representing somewhat unique foci of human occupation due to the limited physical space available and the barriers to movement imposed by their configuration and contents. Spatial constraints on human activities in shelters affect site structure, discard patterns, and artifact morphology

(Barton and Clark 1993).

Based on research at rockshelters around the world, particularly in Europe and the Middle East, archaeologists have identified a number of formation processes that affect sediment deposition (see Chapter 3) and artifact distributions in rockshelters. It is important to remember that shelter formation processes can change over time with

138 changes in physical environment and/or human occupation. Additionally, each shelter

represents a complex and unique suite of formation processes that may vary in terms of

the processes operating, the rates of deposition or alteration, and the predominance of

erosional versus depositional processes (Barton and Clark 1993).

5.1.1. FORMAL, DISTRIBUTIONAL AND RELATIONAL ALTERATIONS

Cultural and noncuitural processes affect the formal, distributional, and

relational properties of archaeological materials deposited in rockshelters. In sandstone shelters, human activities account for many of the postdepositional changes in sediments and inclusions (Donahue and Adovasio 1990). Human activities that damage or displace artifacts in shelters are pit excavation, hearth excavation, tool production and use, dwelling construction, dumping, cleaning, trampling, burning, maintenance, recycling, artifact loss, and pavement construction (Bar-Yosef 1993; Butzer 1982; Koetje

1 9 9 3 ).

Despite the limits on human occupation due to physical space, people frequently reused rockshelters, making them useful in chronology building. However, reuse often alters or obscures previous site structures and material remains. "An important implication of greater reuse is that lithic assem blages from cave and rockshelter contexts may differ systematically from those at open sites, even though an identical activity suite may have taken place in both c a se s' (Barton and Clark 1993:48).

Noncuitural biological processes involve the activities of carnivores, insects, worms, rodents, and bacteria that move artifacts, trample artifacts, and alter artifacts by gnawing or consumption (Bar-Yosef 1993). Animal bioturbation, like cultural activities, is responsible for many of the postdepositional changes in sediments and artifact distributions in sandstone shelters, according to Donahue and Adovasio (1990).

139 A number of geomorphic processes alter or move artifacts in rockshelters. Based on their research at 25 sandstone shelters in the eastern United States, Donahue and

Adovasio (1990) concluded that, in addition to bioturbation and human activity, the deformation caused by rockfalls creates considerable postdepositional change and that freeze-thaw, cryoturbation, solution, and pedogenesis are not as important as they are in limestone shelters.

Geomorphic agents that move artifacts vertically and laterally are wind, water, gravity, subsidence, plastic deformation, cryoturbation, debris flows and other structural collapse, slump, and creep (Bar-Yosef 1993; Barton and Clark 1993).

Gravity and sheetwash often result in horizontal movement of materials and reorientation of artifacts prior to burial (Petraglia 1993). Solifluction displaces earlier deposits inside shelters (Farrand 1985). Petraglia (1993) noted that boulders deposited within shelters may constrain artifact deposition and reduce lateral movement of materials on slopes.

Both chemical and mechanical weathering of cultural materials occur in shelters.

Chemical weathering in the presence of water flow or water seepage may lead to leaching or corrosion of artifacts, cementation of artifacts, mineral alteration, artifact deterioration, and chemical precipitation such as patination on lithics (Bar-Yosef

1993; Barton and Clark 1993). Cryoclastism can split or fissure stones, smaller spalls and debris, and cryoturbation may round artifacts. Collapse can crush debris of all kinds. Pedogenesis leads to chemical or mechanical changes, leaching, cementation, and root disturbance (Farrand 1985).

140 5.1.2. EVALUATION OF FORMATION PROCESSES

Archaeologists need a variety of data to a sse ss the operation of formation processes in rockshelters. The critical question in such studies is how did cultural and natural formation processes result in the distribution and condition of artifacts observed in a shelter (Petraglia 1993). Information must derive from artifacts, ecofacts, and stratigraphie deposits. One must study artifacts as well as ecofacts because formation processes affect both. Nash (1993) argued for collection of all materials from shelters, not simply "artifacts," so one can distinguish flakes and other materials produced by knapping from those resulting from rockfall and other natural processes.

Spatial attributes of artifacts are very important because they indicate the operation of numerous processes that alter the relational and distributional properties of cultural remains. This information is vital in distinguishing patterns that result from behavioral practices originating in the systemic context from those due to other processes of the archaeological context. Artifact orientation or inclination is important because horizontally oriented artifacts suggest little disturbance while vertical or inclined orientations indicate post-depositional movements. Evidence of vertical movement processes are size sorting by weight or size ratios, differences in the percentages of debitage types per stratum, vertical separation of refittable objects, differential patination of artifacts by stratum, rodent microfauna and the skeletal remains of burrowing reptiles, artifact breakage relative to artifact durability, artifact breakage relative to depth of burial, and the nature of artifact breakage (Petraglia

1993). Evidence of horizontal disturbance includes size sorting by weight or size ratios, percentages of debitage types per unit, lateral separation of refittable objects, and differential artifact breakage by unit.

1 4 1 Indicators of formal alterations in the archaeological context include artifact completeness, lithic edge damage, and the nature of artifact damage (Koetje 1993).

Patterns of burning, patination, and artifact breakage and refitting are additional clues

(Petraglia 1993:100). Evidence of limited occupation span, rapid burial, and limited disturbance in occupied shelters includes reconstructable cores, low frequencies of retouched lithics, and lack of edge-damaged artifacts (Barton and Clark 1993:44).

Microartifacts are important in the study of formation processes. Artifactual remains ranging in size from 2 cm to 0.25 mm provide information about human activities at shelters (e.g., food preparation, lithic manufacture), trampling, cleaning, activity areas, and water sorting (Nielsen 1991a, 1991b; Rosen 1993). With microartifacts, one must consider grain size distributions and relative proportions among artifact classes (e.g., bone, lithics, ceramics).

Sedimentary clues to depositional and postdepositional formation processes include total granulometry, roundness, porosity, calcium carbonate content, and heavy mineral assemblages (Farrand 1985:31-35). Evidence of postdepositional movements of sediments and/or inclusions are burrows, water channels, slumps, rock fall, sediment compaction, the distinctness of layer boundaries, and disruption of feature boundaries (Koetje 1993). In sandstone shelters, one may distinguish the mechanisms of sedimentation by particle-size analysis and examination of sedimentary structures.

Lack of sedimentary structure is one indicator of postdepositional disturbance of shelter deposits and, possibly, of artifactual inclusions as well (Barton and Clark 1993). In addition to calcium carbonate analysis, chemical testing for formation processes involves pH, phosphate, and trace mineral analysis.

142 5.2. FORMATION PROCESSES AT COLD OAK AND ROCK BRIDGE SHELTERS

Lithic assemblages and sediment characteristics are used to evaluate the

influence of several formation processes on lithic distribution and condition at Cold Oak

and Rock Bridge shelters. To assess the influences of gravity, creep, bioturbation, and trampling, flake and debris samples are used because they represent the largest lithic samples and because most experimental studies of formation processes focus on these forms of debitage.

5.2.1. PROCESSES AFFECTING SUBSTRATE PROPERTIES

Trampling may increase the penetrability and homogeneity of the substrate of a rockshelter, resulting in the formation of loose layer of sediment overlying a more compact zone (Nielsen 1991b). At Rock Bridge and Cold Oak, one would expect to find layers of loose sediment in more accessible areas of the shelters, especially away from the back wall and near rock outcrops. The thickness of such layers may be proportional to the degree of trampling since, as Nielsen (1991b:488) suggested, the thickness will vary according to “intensity of treadage." Zones of loosened sediment facilitate assemblage mixing.

Loose sediments, perhaps resulting from trampling, characterize the current surface of Cold Oak Shelter examined in 1994. This stratum corresponds to the upper levels of Zone I. The loosened sediments vary in thickness from five to ten cm and are thickest in the proposed zone of heavy traffic, away from the back wall of the shelter in the westernmost end of the 1994 excavation trench (Figure 12).

As the shelter stratigraphy reveals, there is also a loose top layer at Rock Bridge that may have formed as a result of trampling. This gray, ashy midden layer varies in

143 thickness from one to eight cm, with the thickest sequence near the dripline in the western side of the main excavation block, the proposed zone of heavy traffic (Figure

14).

One implication of these observations is that, if trampling affected the nature of the shelter substrates at Cold Oak and Rock Bridge, trampling may have affected the distribution and condition of artifacts as well. This is considered in the next sections of the paper. Another implication is that, if the substrates were modified throughout the occupational histories of the shelters, then artifact movement may have been enhanced in loose zones but artifact positions may have stabilized, relatively speaking, in compacted zones as the shelter deposits accumulated.

5.2.2. PROCESSES AFFECTING VERTICAL DISTRIBUTIONS OF ARTIFACTS

Several formation processes that operate in rockshelters can influence the vertical distributions of lithics and other artifacts. The specific processes considered here are historic disturbances, trampling, and bioturbation. Because chronological studies at shelters, like the current study of diachronic patterns of rockshelter use, rely on interpretations of the vertical distribution of artifact assemblages, evaluation of processes that affect such distributions is important.

5.2.2.3. Historic Disturbances

Historic disturbances including digging and trampling can alter the vertical distribution of artifacts in rockshelter deposits. Groups and activities that have impacted shelters in the study area, which were discussed in more detail in Chapter 1, include treasure hunters, nitrate miners, lumberjacks, farmers, moonshiners, relic collectors, and recent visitors to recreational areas.

144 The issue of historic disturbance is most relevant at Cold Oak Shelter, as Rock

Bridge Shelter apparently was not drastically disturbed at the time of excavation. While some deposits at Cold Oak appear undisturbed, several sediment layers identified during the 1994 excavations were obviously disturbed by post-nineteenth century visitors to the shelter. Evidence of the historic disturbance is bovine dung, which presumably derived from oxen kept in the shelter by niter miners and lumberjacks, and an aluminum soda can in the deposits. The integrity of other deposits is unclear. In the case of these ambiguous deposits, properties of the lithic assemblages were used to determine if the strata are disturbed or undisturbed.

Chipped-stone flake samples were used to distinguish disturbed and undisturbed strata because flakes are the most abundant class of lithics and the specimens are well distributed in most layers at Cold Oak Shelter. Analysis focused on the following attributes of the flake samples for several strata known to be disturbed, layers apparently undisturbed, and ambiguous deposits; density of flakes, average flake volume, average flake length, average flake width, average flake thickness, average flake weight, percentage of complete flakes, and percentage of heated flakes. Because one might expect that flakes would become more concentrated due to disturbances like screening by looters, the density of flakes in different deposits was examined. The six dimensional attributes and the percentage of complete flakes assess the fragmentation effects of recent disturbances. The percentage of heated flakes was examined simply because it is easy to quantify.

The following deposits containing flake artifacts were studied. The 23 disturbed strata are: 510.5N 497E layers A, B, C, D, E, G, J, K, N, P, and S (Feature 24);

510.5N 498E layers A, Ab, Ac, and B; 510.5N 499E layers A, C, H (Feature 35), and I

(Feature 36); 510.5N 500E layers A and B; and 510.5N 501E layers A and B. The six presumably undisturbed strata are: 510.5N 500E layers C and E and 510.5N 501E

145 layers C, D, E, and F. The four ambiguous strata are: 510.5N 498E layer H and 510.5N

499E layers B, D, and G.

The procedure for assessing ambiguous deposits based on the flake samples involves two steps. First, it was determined if there is a statistically significant difference between disturbed and undisturbed deposits with respect to the eight flake properties. Pooled t-score null hypothesis testing was employed during this step.

Second, K-means cluster analysis was used to partition the disturbed, undisturbed, and ambiguous strata to determine if the latter are most like the disturbed or undisturbed deposits.

There are three reasons why pooled t-score null hypothesis testing was used: the small sample sizes of the disturbed and undisturbed deposits, the unequal sample sizes of the deposits, and the unknown standard deviations of the samples. Two-tailed t-tests of the null hypothesis (H q :K1 - K2 = 0) used three confidence levels: 95% (p=0.05),

90% (p= 0.10), and 80% (p=0.20). Observed t-scores were calculated for the eight flake properties in the disturbed and undisturbed samples and compared to the expected t-scores of the three confidence levels. Attributes used to cluster the disturbed, undisturbed, and ambiguous samples are those for which the null hypothesis is rejected, meaning there is a statistically significant difference in the means for disturbed and undisturbed deposits.

Clustering of deposits employed SPSS for Windows software (version 6.0). The

K-means clustering technique (using Euclidean distances between samples) is a non- hierarchical test for splitting samples into a designated number of groups maximizing variation between to within clusters. For each of the significant flake attributes, three clustering tests to yield three, six, and nine clusters were run.

The observed t-scores for the eight flake properties in the disturbed and undisturbed samples from Cold Oak Shelter were calculated and compared to the expected

146 t-scores at three confidence levels. At the 95% confidence level, the null hypothesis was accepted for all flake attributes, indicating no significant difference between disturbed and undisturbed deposits. At the 90% confidence level, only the percentage of heated flakes in the disturbed and undisturbed deposits showed significant differences

(te=±1.70; to=1.84).Two additional flake attributes, density (te=±1.32; to=1.6S) and average length (te=±1.31; to=1.40), show significant differences at the 80% confidence level. Therefore, only three flake attributes were appropriate for clustering the disturbed, undisturbed, and ambiguous sediment layers: density, average length, and percentage of heated specimens.

The clustering tests yielded interesting results. For both flake density and average flake length, all four ambiguous deposits clustered with the disturbed strata. For the percentage of heated flakes, however, layer B in 510.5N 499E and layer H in

510.5N 498E clustered with the undisturbed strata. Considering that the attribute of flake heating was statistically different in the disturbed and undisturbed strata with a higher level of confidence, those results are interpreted as more reliable than the flake density and flake length clusters. Therefore, layers D and G in 510.5N 499E are interpreted as disturbed deposits, and layers B in 510.5N 499E and H in 510.5N 498E may represent undisturbed deposits. Artifacts recovered from the undisturbed deposits are placed in the appropriate assemblages for further analysis.

5.2.2.b. Trampling

Trampling may result in vertical migration of artifacts and vertical sorting by weight. By pushing surface materials into the substrate up to several cm in depth, trampling may preserve the horizontal but not vertical provenience of artifacts

(Nielsen 1991a, 1991b; Villa and Courtin 1983). Trampling also may displace artifacts upward (Villa and Courtin 1983).

147 In rockshelters, where movement is somewhat restricted, one would expect that

trampling occurred in the past as well as during the course of archaeological

investigation. One would expect a greater degree of trampling damage to occur in areas of

relatively easy access, such as where the ceiling is high enough to accommodate

comfortable movement, in areas next to rock outcrops that are convenient spots for

sitting, and/or in areas next to well-defined features. Near the back wall of the cave,

where the ceiling is quite low, one would expect less displacement due to trampling.

Flake samples are used to determine if trampling vertically displaced lithic

artifacts at Cold Oak and Rock Bridge since the samples are large enough for use in

hypothesis testing and previous experimental research is based on debitage. Trampling

analyses omit lithics recovered from outside the excavation block at Cold Oak because the

lithics cannot be assigned to cultural zones or horizontal units. The analytical samples do include lithics recovered from the surface outside Rock Bridge's excavation blocks as the shelter is probably single component.

Villa and Courtin (1983) conducted experimental trampling research using chert debitage, chert blades, bone, shell, sherds, and limestone pebbles on a dry, loose, well-sorted, silty sand substrate. They determined that the degree of vertical movement of artifacts depends on four factors. Two factors are “the intensity of trampling" and "the degree of compaction of the sediments" (Villa and Courtin 1983:275). Another variable is the weight and size of the artifacts. Villa and Courtin (1983) found that artifacts weighing less than 50 grams could move up, move down, or remain stationary: artifacts heavier than 50 grams tended not to move vertically. Stockton (1973) reported similar results, noting that mean artifact weights decreased with depth after trampling.

The thickness of the deposits covering the artifacts is the fourth factor. The distance between the surface and the artifact relates to the amount of vertical displacement due to trampling. In uncovered test squares, sixteen days of trampling

148 displaced artifacts from +0.9 cm to -6.9 cm (Villa and Courtin 1983). Over half of the artifacts moved downward from one to three cm. In test squares where two to four cm of sand covered the artifacts, movement ranged from +2.9 cm to -6.9 cm. Almost 70% of the artifacts moved between +0.9 cm and -0.9 cm. "Upward displacement, of 1 cm or more above the original position, is limited" (Villa and Courtin 1983:274).

Nielsen (1991b) discussed the effects of trampling on the vertical displacement of lithics only. Based on experimental studies, Nielsen (1991b) demonstrated that relatively small lithic specimens (less than two cm) are more likely to be vertically displaced than larger ones as a result of trampling. Moreover, movement is confined to the upper one to two cm zone of loose sediment described previously. These results are comparable to Villa and Courtin's (1983) experimental findings. Nielsen (1991b: 490) explained that

If the surface was buried after a period of dry trampling (as can be assumed in the case of roofed areas), the less-disturbed evidence will be found in a thin (20 mm at most), loose level overlaying a hard, compact, and probably sterile one (unless previous occupations exist in the site). Holding constant other factors, the artifacts recovered in that upper layer should be very small and could be considered primary refuse.

If trampling did occur at Rock Bridge, one would expect the average length of flakes in the upper ashy midden layer of Zone I to be smaller than that of the surface collected flakes. One would also expect the average flake weight of the surface collected specimens to be greater than that of the flakes excavated from Zone I.

For the southern excavation block of Rock Bridge Shelter, the average length of the 19 flakes collected from the surface is 1.79 cm and the average length of the 80 flakes from Zone I is 1.33 cm. The subsurface flakes in the Late Woodland assemblage are, on average, smaller than the surface collected flakes for this portion of the shelter and this difference is statistically significant (te=±2.00; to =2.60; p=0.05). Surface collected flakes weigh, on average, 0.94 g while the average weight of the subsurface

149 flakes in the loose zone is 0.49 g. This difference, too, is statistically significant

(te=±2.00; to=2.22; p=0.05). These data suggest that the relatively small flakes in

Zone I may have been displaced downward as a result of trampling. Further, trampling

probably preserved the horizontal provenience of these Late Woodland artifacts (Nielsen

1 9 9 1 b ).

For the multicomponent Cold Oak Shelter, the possibility that trampling affected

each assemblage is considered. Trampling analyses assess the effects of recent as well as

prehistoric trampling. If trampling did occur at Cold Oak, one might expect three

patterns in the flake sample distribution.

(1) The average length and weight of flakes on the surface will be greater than those of flakes recovered 0-3 cm below the surface.

(2) In the 0-3 cm zone, the proportion of the flakes smaller than two cm should be greater than the proportion larger than two cm.

(3) On the surface, the proportion of flakes larger than two cm should be greater than the proportion smaller than two cm.

Testing these hypotheses, however, is complicated by the size of excavation levels at the shelter, which were five or ten cm or more. This is larger than the two cm trampling zone noted by Nielsen (1991b) and the three cm trampling zone identified by

Villa and Courtin (1983). One also must consider that large flakes are more likely to be detected and recovered from the surface because they are relatively more visible than smaller flakes. In addition, several Cold Oak assemblage sizes are small in terms of statistical significance (e.g., n < 25). Any of these factors might bias the conclusions regarding recent trampling at Cold Oak.

In order to evaluate the possibility that prehistoric trampling vertically displaced lithic artifacts according to the above hypotheses, the distributions of flakes in

Early Woodland and Terminal Archaic zones are examined. To conduct this analysis it is necessary to somewhat hesitantly assume that each zone, which is delineated from others by morphological changes in the sediments, was deposited as a continuous unit and then

150 trampled, rather than being trampled as it was deposited. Except for the Early Woodland strata, the analysis does not consider the possibility that there exist subzones that could have been deposited and trampled individually.

In this analysis, the sizes and proportions of large and small flakes in the first level in which each zone appears are compared to the proportions in the second level, assuming that the former represents the prehistoric "surface" of the deposit and the latter represents Nielsen's "vertical displacement zone." These "levels" represent relative units that allow one to account for differences in the absolute depth of zones across the excavation block.

There is little evidence that trampling vertically displaced lithics in Early

Woodland strata at Cold Oak. Zone lib probably was not trampled based on the vertical distribution of flakes. Twenty-seven flakes in the first level, on average, are somewhat smaller than the 30 flakes in the second level. The averages are 1.13 cm and 1.16 cm, respectively: this difference is not significant at a p=0.05 level (te=±2.02; to=-0.24).

The average weight of flakes from the second level is greater than the average flake weight from the second level; this is not a significant difference (te=±2.02; to=-1.11, p=0.05). Both lengths and weights are opposite the expected patterns. The second expectation is supported in that over 80% of the flakes in the second level are smaller than two cm in size. Only 13% of the flakes in the first level are larger than two cm, a pattern not predicted by the model.

Only the second hypothesis is supported for Early Woodland Zone lie as well.

Average flake sizes in the first (1.28 cm, n=30) and second (1.20 cm, n=71) levels follow the expected pattern, but the difference is not significant (te =±2.00; to=0.67; p=0.05). The expected weight distribution is not observed, as the average weight of level-one flakes (0.38 g) is less than that of level-two flakes (0.42 g); the difference is not significant (te =±2.00; to=-0.38, p=0.05). As predicted if trampling occurred.

1 5 1 over 90% of the flakes in the second level are smaller than two cm. However, only 10% of the flakes in the first level are larger than two cm.

The flake samples suggest that trampling did not vertically displace Terminal

Archaic materials either. Only one trampling expectation is supported for Zone III. As expected, flakes smaller than two cm comprise 80% of the flake sample from the second level. But the average size of nine flakes in the first level (1.11 cm) is smaller than the average for four flakes in the second level (2.04 cm), a difference that is opposite the expected and not statistically significant (te=±2.23; to=-1 .75, p=0.05). Level-one flakes also weigh less than those in the second level, and the difference is not significant

(te =±2.23; to=-2.08, p=0.05). That only 11% of the flakes in the first level are larger than two cm counters the third hypothesis.

The sizes of the lithic samples from Middle Woodland Zone lia and Terminal

Archaic Zone V at Cold Oak Shelter are too small (n=6 and n=4, respectively) to use in assessing the effects of trampling.

5.2. 2 .0 . Bioturbation

Turning now to bioturbation, the growth of plant roots may affect the vertical distribution of artifacts, resulting in mixing and movement of remains from one stratum to another. As Butzer (1982:113) noted, "the fine networks of roots of various plants can actually dislodge and distort archaeological distributions in unconsolidated matrices." Further, the voids left from rotten roots may fill with younger sediment and artifacts. Because many plant roots of various sizes were encountered during excavations at Rock Bridge Shelter, this form of bioturbation is evaluated. No roots were found in the dry deposits at Cold Oak Shelter.

Evaluation of plant bioturbation uses the flake sample from the Rock Bridge main excavation block to see if there is a relationship between root location and the vertical distribution of specimens. If root bioturbation occurred, one might expect a larger

152 proportion of lithics in the sandy substratum of Zone III (which is probably not an occupation deposit) of the excavation units where roots were encountered compared to those units lacking roots.

Excavation units of the Rock Bridge main block containing roots are indicated on

Figure 16. The percentage of flakes recovered from each stratum is expressed as a ratio for all units (Zones 1: 2: 3). Some of the units’ percentages do not add up to 100% since all flakes could not confidently placed into a particular stratum. The abbreviation "NP“ marks those units lacking stratigraphie profiles. Only two units show a concentration of flakes in the substratum. These units with relatively high proportions of flakes in the subsoil are toward the southern end of the block. This may indicate some vertical movement of artifacts. The northern end of the block has high proportions of flakes in the upper ashy midden layer of some units. Perhaps this is the result of upward migration of artifacts due to root action, as vertical movement may be great in a sandy substrate. However, since other processes, especially trampling, may affect the vertical distribution of artifacts at a site, it is difficult to conclusively state that root bioturbation operated as a formation process displacing Late Woodland artifacts at Rock

Bridge Shelter.

Although Funkhouser and Webb (1929) suggested that ash deposits, such as those at Cold Oak, are “not suitable for burrowing animals," there is some evidence of animal bioturbation at Cold Oak Shelter. It takes the form of Terminal Archaic Feature 24 in unit 510.5N 497E, which Gremillion (1995) reported was impacted by a rodent burrow. Bioturbation may lead to the movement and concentration of artifacts as the animal excavates the burrow. If this is the case, one would expect that the density of lithics in the burrow deposits would exceed that of the surrounding matrix.

At 0.523/liter, the density of lithics (flakes and debris) in Feature 24 significantly exceeds densities in the surrounding deposits (less than 0.061). On one

153 hand, these figures might indicate that bioturbation by burrowing resulted in the movement of lithics from other deposits and the concentration of those materials in the burrow sediments. Or, the high density of lithics derived from the feature deposits of

Feature 24. The lithic data are inconclusive, but recent skeletal remains of burrowing rodents like mice, moles, and shrews are documented from Cold Oak (Holm 1995).

While no animal burrows were found during excavations at Rock Bridge, it is possible that animal activity impacted the distribution of lithic artifacts. Evidence of this includes skeletal remains of rodents (mice or voles) in the faunal collection and a pack rat deposit over Feature 1.

5.2.2.d. Implications of Vertical Distributional Analyses

The preceding evaluation of processes that might lead to vertical mixing of lithic assemblages indicates that trampling and bioturbation did not cause significant vertical displacement of lithic artifacts at Cold Oak Shelter. At Rock Bridge Shelter, trampling likely displaced downward the relatively small flakes in ashy Zone I, preserving their horizontal provenience once they became embedded; vertical displacement due to plant bioturbation is not evidenced. Therefore, the temporally defined assemblages at the shelters may be used in testing hypotheses about diachronic differences in rockshelter use at the shelters. Lithics from Zones lib and lie represent Early Woodland occupations, and specimens from Zone III are from Terminal Archaic occupations. Although artifact movement likely occurred at Rock Bridge, diagnostic artifacts and radiocarbon dates indicate that the shelter was used during the Late Woodland period.

The absence of evidence of downward displacement of lithics at Cold Oak Shelter corresponds to some extent with the distribution of ceramics artifacts. Gremillion

(1995) found evidence of limited upward movement of ceramics but no downward displacement. “The accumulation of artifacts from a broad time span in Zone I and the general absence of mixing in lower deposits indicates that artifacts are more likely to be

154 moved upward than downward as a result of disturbance (most of the sherds from 'mixed' contexts come from the surface or profiles rather than from deep intrusions)"

(Gremillion 1995:18). No ceramics were found in the lower Terminal Archaic deposits at the shelter.

5.2.3. PROCESSES AFFECTING LATERAL DISTRIBUTIONS OF ARTIFACTS

Several formation processes that operate in rockshelters influence the lateral distributions of lithics and other artifacts. The specific processes considered here are gravity, creep, and trampling. A consideration of such processes is important because the lateral distributions of lithic artifacts are used to identify activity areas at the shelters.

5.2.3.a. Gravity

The force of gravity causes relatively heavy objects to move further downslope than lighter objects (Schiffer 1987; Butzer 1982; Rick 1976). Owing to their large sample sizes and roughly similar morphology, shelter flake samples were used to assess the effects of gravity on lithic distributions. To determine if gravity affected the observed horizontal distribution of lithics at the shelters, the average weights of flakes recovered from each unit or zone were plotted. If gravity did move objects, one would expect to find greater average weights toward the downslope end of the excavation blocks.

Another indicator of lateral displacement by gravity is the percentage of "heavy" flakes, which are defined as those greater than the average weight of flakes in the assemblages, per unit or zone. The average flake weight for Cold Oak is 0.6 g and for Rock Bridge, 1.0 g. A downslope increase in the percentage of "heavy" flakes indicates that gravity affected the horizontal distribution of lithics at the shelters.

155 There is little indication that gravity affected the spatial distribution of lithics in the Late Woodland assemblage from Rock Bridge Shelter. The data on Figure 17 indicate that there is no trend toward increasing average weights downslope in the main block.

The larger averages are randomly scattered throughout the excavation block. Some of the units contained heavy flake specimens; these outliers were removed from the unit samples and recalculated and plotted the average weights to see if there is a downslope increase in weight. Again, the larger average weights are randomly distributed in the main block. Figure 18 shows the percentage of flakes greater than 1.0 gram in weight per unit in the main excavation block at Rock Bridge. There is no downslope increase in the percentage of "heavy" flakes for the main block, indicating that gravity did not greatly affect the horizontal movement of lithics at the shelter.

Because of the limited lateral extent of the 1994 excavations at Cold Oak and the disturbed nature of the shelter, the effects of gravity on the spatial distributions of lithic artifacts could not be assessed. Post-depositional processes have impacted the original surface so significantly that one cannot reconstruct accurately topographic contours and downslope-upslope relationships.

5.2.3.b. Creep

Creep is the slow, downslope movement of soil and artifacts under the influences of frost formation and gravity. In cold climates, frost forms under dense objects like rocks, pottery and lithic artifacts because they have higher conductivity than soil

(Butzer 1982; Wood and Johnson 1978). The ice crystals push the objects upward since water expands when it freezes. When the ice melts, the objects shift slightly downslope as they settle back due to gravity. Over time, the objects gradually move downslope.

According to Butzer (1982:103), the net result is that larger objects sort from fine sediment into "circular patterns of level surfaces and elongated ones on slopes. " Based on case studies, Butzer concluded that on slopes of 2° to 5°, circular rings of larger objects

156 form due to creep; the internal diameters of the rings are proportional to the size of the objects. At these lower inclinations, creep rearranges artifacts but the objects retain their "basic associations." The rings of large objects are ellipsoidal in shape on slopes of

5° to 10". Artifacts on slopes greater than 8° are "effectively dispersed"" (Butzer

1 9 8 2 :1 0 4 ).

It is unlikely that creep affected lithic distributions at Cold Oak Shelter because of the dry nature of the deposits. But because Rock Bridge is not a dry shelter, creep resulting in the movement of relatively large lithic artifacts is a potential formation process. To evaluate this possibility, the locations of large flake specimens were plotted on an elevation map for the main excavation block. “Large" lithics are those flakes with length and width values greater than the average of 1.5 cm x 1.5 cm or those flakes greater in weight than the average of 1.0 g. The locations of flakes greater than 2.0 cm x

2.0 cm were also plotted. Since the slope of the main block is less than 5", one might expect circular rings of the relatively large artifacts to be present.

Evidence that frost, an agent in soil creep, formed at Rock Bridge Shelter takes the form of the five frost-pitted lithics and the one frost lid fragment recovered. Hence, creep resulting in the movement of relatively large lithic artifacts is a potential formation process. Figure 18 shows the distribution of percentages of relatively heavy flakes in the main block of Rock Bridge. Dashed lines indicate three concentrations of these specimens. Without more extensive east and west coverage it is difficult to say if these represent rings, but the southernmost cluster is three meters wide and may be part of a linear to curvilinear trend.

Figure 19 shows the distribution of flakes greater than 1.5 cm x 1.5 cm as percentages of the total flake count per unit. The three clusters indicated for heavy flakes are also evident for the relatively large flakes. If one plots the distribution of

157 percentages of 2.0 cm x 2.0 cm flakes for the main block, a similar pattern of three clusters emerges.

These three "rings" of large flakes correspond to the locations where the densest concentration of lithics and the highest numbers of heat-treated and heat-altered cherts were recovered. Based on this evidence, then, it is suggested that creep may have affected the observed horizontal distribution of Late Woodland lithics at Rock Bridge Shelter.

5.2.3.0. Trampling

Besides gravity and creep, trampling can laterally displace artifacts (Villa and

Courtin 1983). The horizontal displacement of artifacts due to trampling is a function of size: Nielsen (1991b) expresses these relationships in three hypotheses. First, items less than two cm in size are not likely to be displaced horizontally since trampling often pushes them into the loose sediment layer created by treadage. Second, trampling randomly moves artifacts less than 50 cm^ but larger than 2 cm^ such that they are shifted outside the zone of heavy trampling. Third, trampling displaces objects greater than 50 cm3 further into marginal zones of low traffic. "Even moderately trampled areas will be composed of a "marginal zone" characterized by a high proportion of bulky artifacts, and a "traffic zone" with small- and medium-size items randomly scattered and very small ones buried close to their original spot of deposition" (Nielsen 1991b:500).

At sites where occupants practiced systematic maintenance or cleaning, an activity that usually results in the movement of large artifacts, the contrast between the two zones may be partially obscured. In their experimental study. Villa and Courtin (1983) noted that the maximum amount of horizontal displacement of artifacts due to trampling was

85 cm. They indicated that uncovered artifacts are more likely to be displaced than ones covered by sediment. Villa and Courtin (1983) found no relationship between artifact weight and horizontal displacement due to trampling.

158 Because the average flake sizes of the Cold Oak and Rock Bridge assemblages fall within Nielsen's “very small" size category of less than two cm, trampling probably did not drastically affect tfie horizontal positions of most of the flakes. If trampling resulted in horizontal movement of lithics at Rock Bridge, one might expect that lithics greater than two cm recovered from the surface are concentrated in the low traffic areas of the shelter, especially in the westernmost units of the main excavation block.

Unfortunately, the number of surface collected specimens greater than two cm in size is relatively small for Rock Bridge Shelter, and a plot of their provenience does not suggest a preferred distribution. Therefore, there is little evidence that trampling led to horizontal displacement of Late Woodland lithic artifacts at Rock Bridge.

One possible factor complicating the analysis of horizontal displacement at Cold

Oak Shelter is the limited horizontal extent of the 1994 excavations. Only the effects of trampling resulting from human movement in a north-to-south direction can be evaluated at Cold Oak Shelter because the excavation units lie in an east-west orientation. As with the vertical movement analysis, evaluation of horizontal movement of flakes due to trampling is hindered by the disturbed nature of the shelter as well as the multiple occupations.

The large rock uncovered in the middle excavation unit at Cold Oak Shelter may have affected the flow of traffic if it was deposited prehistorically, which seem s likely based on the shelter stratigraphy. If this was the case, one would expect that the proportion of flakes between 2 cm^ and 50 cm^ in size would be greatest in unit 510.5N

497E and less in other units. Because all of the flakes larger than 50 cm^ in size were recovered from outside the 1994 block at Cold Oak, the distributional hypothesis related to that size class cannot be evaluated.

Like Rock Bridge, the Cold Oak assemblages do not indicate that trampling laterally displaced larger lithics. The Early Woodland lithic assemblage from Cold Oak

159 does not exhibit the expected lateral distributions. Only 5.0% of the lithics from unit

510.5N 497E are flakes between 2 cm^ and 50 cm^ in size. The largest proportions,

8.3% and 6.9%, derived from those units where the smallest percentages are expected,

510.5N 498E and 510.5N 500E, respectively. Terminal Archaic deposits yielded

similar results, with 8.5% of the lithics from 510.5N 497E falling in the 2 cm^ and

50 cm^ range, but 10.9% and 25.0% in the presumably higher traffic areas of 510.5N

498E and 510.5N 501E, respectively. The Middle Woodland assemblage was not assessed

due to the small sample size.

5.2.3.d. Implications of Lateral Displacem ent Analyses

The preceding evaluation of processes that laterally displace lithic artifacts

suggests that creep may be at least partly responsible for the concentrations of

relatively large artifacts in the main excavation block at Rock Bridge Shelter. Gravity

apparently was not a significant force at Rock Bridge, and there is little evidence that

trampling laterally displaced large lithics at either shelter. Therefore, identification of

activity areas at Cold Oak should not be influenced by the potential effects of trampling, but at Rock Bridge one must consider the effects of creep on the distribution of artifacts

in apparent activity areas.

5.2.4. PROCESSES AFFECTING ARTIFACT CONDITION: TRAMPLING

Trampling often results in lithic damage, including fracture or breakage, abrasion, and random edge scarring (Nielsen 1991a, 1991b; Prentiss and Romanski

1989; Villa and Courtin 1983). According to Villa and Courtin (1983), artifact breakage due to trampling depends on the density of artifacts, with more damage occurring in deposits with high artifact densities, and the degree of compaction of the substrate, with more damage occurring over compact substrates. Of course, some flakes

160 break during the course of lithic reduction or maintenance; this factor may affect the

results of the analysis by accentuating the apparent effects of trampling on the condition

of the samples. Conversely, assessing the damaging effects of trampling is important

because the degree of fragmentation of a flake sample greatly affects one's

interpretations of lithic reduction and maintenance strategies, which are discussed in

the next chapter. Spatial distributions and experimental studies are means of distinguishing between these two possibilities. The following indicators assess the potential effects of trampling on lithic damage at Rock Bridge and Cold Oak shelters.

The first approach focuses on the distribution of edge-damaged flakes. Edge- damaged flakes are those flakes with randomly chipped margins that do not appear to have been intentionally altered by humans. If trampling did contribute to fragmentation of the Cold Oak and Rock Bridge lithic assemblages, one would expect edge-damaged flakes to cluster in those areas more likely to have experienced prehistoric or historic human traffic, particularly the eastern sides of the shelters and around the large rock in the center of the Cold Oak excavation block.

Distributions of complete flakes are the second means of assessing the damaging effects of trampling. If trampling converts complete flakes into proximal, medial and/or distal flakes due to lateral, bending or compression fractures (Prentiss and Romanski

1989), one would expect higher proportions of complete flakes in areas of low traffic.

For Rock Bridge, one would expect relatively low percentages of complete flakes closer to the drip line and rock outcrops and relatively high percentages toward the back wall of the shelter. At Cold Oak, relatively low percentages of complete flakes should be found in the easternmost units and around the large rock in the center of the excavation block, and relatively high percentages should be found toward the back wall in the westernmost units or in the unit with the rock. The percentage of complete flakes is similar to a m easure of trampling dam age used by Villa and Courtin (1983). Their ratio of flake

161 fragments is the number of flake fragments from a unit or stratigraphie zone

(multiplied by 100) divided by the total number of flakes from the unit or stratum. For

both shelters, one would expect higher ratios of flake fragments in the high traffic areas

on the easternmost sides of the shelters; the ratio should decrease toward the backwalls.

Because of the redundancy in these two measures, only complete flake distributions are

considered.

The distribution of flake counts by weight is the third means of assessing

trampling. The amount of flake breakage is evaluated by looking at the distribution of

flake counts normalized to one gram of flake weight for each excavation and collection

unit. One would expect relatively high flake to weight ratios (small flakes) in units

where one expects trampling to have been prevalent. The expected distribution of flake

counts per gram of flake weight involves decreasing numbers of flakes per gram (bigger

flakes) as one approaches the back walls of the shelters or in the central unit of Cold Oak

where the large rock is located.

The basis of the previous three approaches to evaluating the operation of

trampling as a formation process at the shelters is the horizontal distribution of different measures of fragmentation. The last approach, an aggregate analysis, makes use of the entire flake and debris samples for the shelter assemblages. The total numbers of complete, proximal, medial and distal flakes plus debris in the shelter samples are compared with data reported by Prentiss and Romanski (1989:91).

Prentiss and Romanski (1989) conducted trampling experiments using four

experimentally produced debitage assemblages generated by block core reduction, spheroidal core reduction, biface manufacture, and end scraper manufacture. They used

Morrison chert from northwestern Wyoming, a very fine-grained chert with incipient fracture planes and occasional inclusions, in all replicative knapping procedures.

Prentiss and Romanski subjected each of the four experimental debitage assemblages to

1 62 trampling on a loose sand substrate. They then compared the percentages of complete, proximal, medial/distal, and debris specimens for each assemblage before and after tram pling.

In their studies, trampling of the two experimental core reduction assemblages generally resulted in fewer complete flakes and debitage specimens and increased numbers of proximal and medial/distal flakes. Trampling of the two tool reduction assemblages led to fewer complete flakes and more proximal and medial/distal flakes

(Table 23). If trampling occurred at either of the shelters, one would expect the flake samples to mimic either the trampled core reduction or tool reduction samples that

Prentiss and Romanski (1989) described.

Results of these analyses indicate that trampling led to some lithic damage and fragmentation of the Late Woodland assemblage from Rock Bridge Shelter. At least two of the four tests yielded positive results. First, although surface collection units at Rock

Bridge have small sample sizes and one must interpret them cautiously, there is a general orientation of the complete-flake contours roughly parallel to the back wall of the shelter, and the larger percentages of complete flakes are toward the wall as expected if trampling caused fragmentation. The percentage of complete flakes per flake sample of each unit, then, supports the conclusion that trampling may be responsible for at least part of the fragmentation of the Rock Bridge lithic assemblage.

Second, the Rock Bridge flake and debris samples resemble the trampled samples reported by Prentiss and Romanski (Table 23). The macrodebitage and microdebitage samples from Rock Bridge are similar to Prentiss and Romanski's trampled tool reduction assemblages, especially the end scraper assemblage. Taking the Rock Bridge data as a whole, the percentages are still much closer to the trampled tool reduction assemblages than to any untrampled assemblage or the trampled core reduction assemblage.

163 Third, the horizontal distribution of all edge-damaged flakes recovered by excavation and surface collection at Rock Bridge Shelter offer mixed indications of trampling. Although some sample sizes are very small (less than three specimens), there are relatively high percentages of edge-damaged flakes in the main excavation block, an area of the shelter classified as a potentially high traffic area in a roughly north to south direction. The percentages show a generally decreasing trend toward the back wall, but the expected north-south orientation of the contour lines does not hold for the southern end of the main block. The contour lines near the middle excavation block show decreasing percentages of edge-damaged flakes toward the back wall as expected, but the pattern near the northern excavation block diverges from the expected.

Finally, the distribution of flake counts per gram of flake weight for Rock Bridge

Shelter does not correspond to the expected distribution if trampling fragmented the specimens. Toward the northern end of the shelter, the number of flakes per gram of flake weight increases toward the back wall, indicating greater fragmentation in the less accessible part of the shelter. In the area of the main excavation block, there are two concentrations of high flake counts per gram that form ellipsoidal clusters trending northwest to southeast. The expected decrease in flake counts in zones parallel to the back wall is not evident. The surface collection area south of the main block is problematic as well, indicating an increase in flake count per gram toward the back wall, but those units have very small lithic samples that are biased toward large flakes since they were surface collected rather than 1/4" and 1/16" screened.

Looking at the main excavation block at Rock Bridge in more detail, the contours do not run north-south as expected, but are clustered on the western side of the block at the northern and southern ends. If one omits the southern cluster in unit 49N50E, as one probably should since the sample size is only two, then the southern end of the block comes closer in line with the expected decrease toward the back wall.

164 Taking into account the data for the three features in the main block containing

flakes (Features 3, 5 and 14), the observed distribution diverges even more from the expected. There are subcircular to ellipsoidal concentrations of fragmented flakes in the

north, middle and southern portions of the block. The cluster corresponding to Feature

14 is dismissed due to small sample size (n=1), and Feature 5, part of the southern cluster, only has one flake specimen. But one cannot discount the cluster of small flakes

indicated by the high flakerweight ratio for Feature 3 since the sample size is 40. In conclusion, the distribution of the ratio of flake count to flake weight does not support the expected distribution of relatively large and small flakes one might encounter as a result of trampling.

The lithic indicators suggest that trampling fragmented or damaged at least one of the lithic assemblages from Cold Oak Shelter. First, as indicated in Table 23, the percentages of debitage types for the Early Woodland and Terminal Archaic debitage samples very closely resemble Prentiss and Romanski’s trampled end scraper sample.

The Cold Oak samples also show some similarities with the experimentally trampled biface sample. Clearly, the Cold Oak samples differ considerably from all of Prentiss and

Romanski's untrampled samples. The Middle Woodland sample was not assessed due to the small sample size.

The distribution of complete flakes in the Early Woodland assemblage at Cold Oak does not follow the expected pattern if trampling resulted in fragmentation. Two of the three highest percentages are in units defined as probable high traffic areas: 510.5N

498E and 510.5N 501E. However, Terminal Archaic samples follow more closely the expected pattern in that the central and westernmost units contained the highest percentages of complete flakes.

The percentages of edge-damaged flakes for Cold Oak Shelter samples, on the other hand, do not follow the expected patterns if trampling occurred. The Woodland and

165 Archaic samples have percentages of edge-damaged flakes that increase toward the rock

and back wall, which is the opposite of the expected pattern. The fourth measure, flake-

weight ratio, also provides little evidence that trampling fragmented lithics at Cold Oak.

For the Early Woodland sample, the units that should have the lowest ratio have the

highest, and there is no patterning in the ratios by unit for the Terminal Archaic sample.

The implications of the analysis of lithic damage and fragmentation due to

trampling are as follows. Two and possibly three of the four damage indicators provide

evidence that trampling at Rock Bridge Shelter affected the condition of lithic artifacts.

Therefore, flake fragmentation at the shelter, especially on the eastern side, may be

explained in part by trampling rather than or in addition to systemic human behavior

like lithic reduction activities. The same is true for the Terminal Archaic assemblage

from Cold Oak Shelter. The Early Woodland assemblage from Cold Oak, on the other hand,

does not indicate that trampling damaged the flake samples. Therefore, patterns of flake

fragmentation in this assemblage may indicate lithic reduction activities rather than the

effects of trampling.

5.3. SUMMARY

Formation processes are the cultural and noncultural processes that influence

the deposition, accumulation, distribution, and condition of material remains. The study of formation processes in archaeology has expanded over the past decades, yielding a

wealth of descriptive and experimental information about a variety of processes. Some of

this information was used to assess the operation of select formation processes at Cold

Oak and Rock Bridge shelters.

A variety of formation processes affect sediments and artifacts at rockshelters.

Cultural processes, which result in artifact damage and movement, include the

166 importation and alteration of raw materials, excavation of pits, dwelling construction, cleaning, and trampling. Artifacts can be damaged and moved by biological processes such

as burrowing, gnawing, and root action. Rockfall, sheetwash, gravity, and creep are geological processes that can alter or displace artifacts at rockshelters.

Using sediment properties and debitage distributions, the current study assessed the operation of trampling, bioturbation, gravity, and creep at Cold Oak and Rock Bridge shelters (Table 24). Trampling likely resulted in the formation of a loose upper zone at both shelters, and this zone could have facilitated the movement of artifacts. Vertical displacement of small lithics by trampling is evidenced in the Late Woodland assemblage from Rock Bridge but not in any Cold Oak assemblages. There is no evidence that bioturbation vertically displaced lithics at either shelter. It is likely that creep horizontally displaced Late Woodland lithics at Rock Bridge, resulting in curvilinear concentrations bearing east-west across the main excavation block. Gravity and trampling apparently did not affect the lateral distribution of lithics at either shelter.

Flake fragmentation in the Late Woodland assemblage from Rock Bridge and the Terminal

Archaic assemblage from Cold Oak may be due, in part, to trampling.

Results of the formation process studies have implications for the study of shelter use over time. The Woodland and Archaic assemblages from Cold Oak and Rock

Bridge probably represent relatively unmixed assemblages that may be used to assess diachronic differences in shelter use. Although artifact movement likely occurred at

Rock Bridge, diagnostic artifacts and radiocarbon dates indicate that the materials derived from occupations during one period, the Late Woodland.

In the previous chapter, descriptions of the Cold Oak and Rock Bridge lithic assemblages showed many similarities among Woodland and Archaic assemblages but also some differences. One difference was the density of lithics in each assemblage, where the

Early Woodland density was double that of the Late Woodland and Terminal Archaic

167 densities. Based on the results presented in this chapter, trampling is unlikely to have fragmented the Early Woodland assemblage, so that density is not inflated. But the densities of the Late Woodland and Terminal Archaic assemblages may be inflated because of trampling, increasing the differences with the Early Woodland assemblage.

Lithic reduction activities, which are covered in the next chapter, may differ among the Cold Oak and Rock Bridge assemblages, indicating differences in shelter use over time. The results of the formation process analyses have implications for interpreting evidence of lithic production. Flake fragmentation may be an indicator of reduction activities, but this analysis demonstrated that flake fragmentation in Late

Woodland and Terminal Archaic assemblages from Rock Bridge and Cold Oak may be due to trampling as well. In addition, comparisons of raw material use over time may also be influenced by the trampling results because densities of different raw materials in the assemblages may differ, in part, because of processes extraneous to raw material selection patterns, as trampling probably inflated the sample sizes of Late Woodland and

Terminal Archaic assemblages.

168 CHAPTER 6

ANALYSIS OF LITHIC PRODUCTION SYSTEM

A lithic production system is the totality of processes used by humans to modify lithic material through a series of activities and at a series of locations for use as manufactured objects in a social context (Ericson 1984). Lithic production systems are highly variable, and the factors that influence the character of a lithic production system include "the regional lithic resource base, the modes of procurement, social distance between knappers and consumers, labor investment, modes of transportation and social organization" (Ericson 1984:1). According to Ericson (1984:1), "the analysis of the quarry, debitage analysis at sites within the study region, the use of production indices and spatial analysis, chemical characterization and chronometric dating of artifacts and debitage will play roles in reconstructing lithic production systems."

While one cannot use rockshelters alone to track the entire lithic production system of a group of people, the research reported here considers several of the lithic production system factors identified by Ericson (1984), namely the lithic resource base and procurement strategies, debitage analysis, and spatial analysis. Assessment of diachronic differences in these components of the lithic production systems employed by prehistoric inhabitants of Rock Bridge and Cold Oak shelters are emphasized. Because

"debitage analysis is a basic technique used in the reconstruction of a lithic production

1 69 system" (Ericson 1984:3), the lithic assemblages of the two shelters are well suited to the task.

Ericson (1984:3) asserted that “during the course of the many stages of production of [lithic] material, debitage will be created at the sites of production, which will be indicative of the stag es of production." The stage view of lithic production with its attendant material remains is defined in Collins' (1975) oft-cited model of lithic reduction.

Drawing analogies from experimental studies, Collins (1975) linked the activities of chipped-stone tool manufacture with the specific types of material produced. His goal was to provide a means of inferring processual information from archaeological lithic assemblages. "Examination of the attributes imparted to the discarded waste and rejects as well as to the end products provide the basis for inferring manufacturing steps and techniques from archaeological evidence" (Collins 1975:23).

A fundamental proposition of Collins' approach is that "there are certain basic and unavoidable reductive step s involved in producing useful [lithic] objects"

(1975:16). While the process of "generalized" or "etic" lithic reduction represents a continuum of activities, Collins pragmatically distinguished five steps whose activity sets result in the production of distinct collections of material remains or "product groups." Lithic product groups contain both "waste by-products and objects destined for further reduction or for use" (Collins 1975:17).

The five steps of lithic reduction identified by Collins (1975) are acquisition of raw material, core preparation and initial reduction, optional primary trimming, optional secondary trimming and shaping, and, subsequent to use, optional maintenance and modification. Table 25 summarizes the activities, controlling variables, and product groups Collins associated with each step.

1 70 Collins (1975) applied the lithic reduction model in interpreting lithic assemblages from Arenosa Shelter, Texas and Laugerie Haute Ouset, France. Using the model, Collins inferred intrasite diachronic changes in raw material acquisition, manufacturing techniques, stages of reduction, and tool use. The analyses described below share similar goals.

Evaluation of lithic production systems focuses on chipped-stone artifacts in the

Rock Bridge and Cold Oak shelter assemblages, especially the flake samples. Part of the reason for this emphasis is the numerical predominance of flake artifacts in the assemblages. Other justification is provided by Collins (1975). First, Collins

(1975:17) argued that "if isolated, product groups can be described in terms of their technological attributes and inferences can be drawn concerning the specific activities by which the particular manufacturing step was accomplished. The waste, or debitage, is particularly amenable to this technological analysis." In addition, debitage, and especially small debitage, may be less likely to be moved [at least by the knappers or tool users] so it may be more helpful than tools in reconstructing some aspects of lithic production systems. As Collins (1975:19) noted, "except in unusual circumstances, debitage remains at the locus of manufacture and finished objects are subject to relocation with use."

Evaluation of the lithic production system s for the Terminal Archaic, Early

Woodland, and Late Woodland assemblages considers thermal alteration, raw material utilization, the processes of tool manufacture and maintenance, and differences in these activities over time; Middle Woodland data are not commented on here due to the small size of the assemblage. While these activities represent a continuum of processes, they are treated as groups of activities. Thermal alteration considers intentional heat treatment as it relates to the process of tool manufacture. The incidence of thermal alteration is one indicator of occupational intensity at a site. Raw material utilization

171 takes into account raw material acquisition, the first step in Collins' (1975) model of lithic reduction, and lithic utilization strategies. The relative proportion of local and non-local raw materials is another indicator of occupational intensity. Tool manufacture and maintenance encompasses core preparation and initial reduction, primary trimming and reduction, secondary trimming and shaping, and retooling and resharpening (Collins

1975). Analyses focus on identifying the reduction stages and techniques that characterize each occupation at the shelters in order to characterize the range of lithic- related activities that occurred during each occupation, which is a component of occupational variability.

6.1. THERMAL ALTERATION OF CHERT

According to Luedtke (1992:99), "heat treating, also called thermal alteration, thermal pretreatment, and annealing, involves the intentional heating of chert to bring about desired changes, usually to improve its flaking properties" or perhaps to change its color. The process of heat treatment usually involves slowly heating chert to the appropriate temperature at which chert alteration occurs, holding the chert at that temperature for some period of time, and then slowly cooling the chert. Luedtke

(1992:101-105) described the geochemical and mechanical effects of heat treatment on chert.

6.1.1. METHODS

Four indicators are used to evaluate the use of heat treatment by prehistoric knappers at Rock Bridge and Cold Oak shelters. For the purposes of this study, the visible effects of heat treatment are most relevant. They are color change, loss of

1 72 translucency, and increase in luster on post-heating flake scars. One effect of heat treatment on chert mechanical properties, rippling of flake scars, is also tabulated

(Luedtke 1992:103). Only the chipped-stone specimens in the shelter assemblages were used in this analysis.

One aspect of the analysis involves examining the sample of heat-treated cherts from each shelter with respect to reduction class. While any sized chert fragment, from cobble to flake, can be heat treated, Luedtke (1992:100) noted that relatively thin flakes are less susceptible to thermal shock and fracture because they heat more evenly.

Since knappers often used heat treatment to produce cherts suitable for bifacial tool manufacture, one might expect preforms or large flakes to show evidence of heat treatment (Luedtke 1992:100). Such a pattern might be indicative of earlier stage reduction at the shelters. Secondary or bifacial thinning flakes exhibiting heat treatment indicators may have been detached from larger cherts that were heated prior to secondary reduction or retooling; this might indicate later stage reduction.

The spatial distributions of heat-treated cherts may be indicative of prehistoric occupational variability. Laterally, one might expect that heat-treated specimens would be associated with activity areas or hearth features. Vertically, one might also expect that a higher percentage of heat-treated specimens would be associated with cultural zones representing relatively more intense occupational episodes. Johnson (1989) argued that at more intensely used sites, one would expect a relatively high incidence of heat alteration since this process requires adequate time and care to be carried out effectively. Signs of heat alteration should be more prevalent on reduction debris at such sites. At less intensely used sites, on the other hand, one might expect to find heat- altered tools and resharpening or reworking debitage.

Another component of the analysis examines the sample of heat-treated cherts with respect to raw material type. Correcting for differences in sample sizes, one might

1 73 expect that indicators of heat treatment are more common for the poorer quality cherts, if knappers used heat treatment to improve knapping quality. Alternatively, one might expect a relatively high incidence of heat treatment for certain cherts to alter their appearance. To evaluate these expectations, the proportions of heat-treated specimens for each chert type are compared.

Anderson (1979) outlined archaeological signatures that may indicate the reasons why knappers thermally altered cherts. If knappers employed heating to alter the specific appearance of cherts, one might expect differential distribution of altered versus unaltered cherts in some chert categories, or a high incidence of intentionally heat-altered specimens in particular lithic classes in the shelter assemblages. If Rock

Bridge or Cold Oak knappers used thermal alteration to improve the quality of cherts, one would expect an overall high incidence of alteration on specific chert types, regardless of lithic artifact category. The Rock Bridge and Cold Oak assemblages lack sufficient data to evaluate the possibilities that knappers heated cherts to sharpen the edges of cutting tools, to improve soft hammer or pressure flaking efficiency, or to conserve raw materials.

6.1.2. RESULTS

Of the 1860 chert chipped-stone specimens examined in the Cold Oak assemblages, about 6% show evidence of heat treatment. For the shelter, the density of heated cherts is 0.04/liter. None of the chipped-stone tools or cores show evidence of intentional heating. However, heat treatment indicators were recorded for about 6% of the flakes and 7% of the debris. All types of flakes exhibit the characteristic change(s) associated with intentional heating except blades and primary decortication flakes, and percentages per assemblage are fairly comparable. Flakes considered to be diagnostic of

1 74 primary and/or secondary reduction as opposed to initial core reduction dominate the sample of heat-treated flakes. Excluding broken flakes, about 71% of the heated flakes indicate primary or secondary reduction while the remaining 29% are diagnostic of core reduction. Looking at the percentages another way, 10% of all late-stage debitage were heated, while only 5% of all early-stage were so altered, excluding broken flakes.

Because heat treatment indicators occur most commonly with the smaller debitage of the

Cold Oak Shelter assemblages, it is possible that knappers heated the cherts prior to secondary or bifacial thinning flake removal.

All chert types except Kanawha, Muldraugh and Ste. Genevieve show evidence of heat treatment by the Cold Oak inhabitants. About 44% of the heat-treated specimens are made of unidentified chert, and 24% are Breathitt. Boyle and Haney chert account for an additional 12% and 10%, respectively. The remaining 10% are St. Louis, Lower

Pennsylvanian, and Paoli cherts. Excluding the unknown chert specimens, one might conclude that there is a high richness (6/9 chert types) but low evenness (42% are one chert type, Breathitt) of heat-treated cherts in the Cold Oak chert artifact sample.

Horizontally, workers recovered heat-treated cherts from Cold Oak Shelter from all units as well as outside the excavation block. The highest percentage (41%) and density (0.07/liter) of heated cherts came from unit 510.5N 500E. About 10% of the chert artifacts recovered from that unit were heat treated. Because this is also the unit that produced the highest percentage and density of all lithic artifacts, sample size effects may partially explain the predominance of heated cherts.

Vertically, roughly equal percentages of heat-treated cherts derived from Early

Woodland (42%) and unknown (44%) deposits. Terminal Archaic strata yielded an additional 13%, and the remaining 1% derived from historic contexts. Oddly, the highest density of heat-treated cherts, 0.34/liter, is from historic contexts (which may reflect mixing with prehistoric deposits), followed by 0.07/liter from Early Woodland layers

1 75 and 0.02/liter from Terminal Archaic deposits. Ten percent of the ten chipped-stone lithics recovered from historic contexts are heat treated, and 9% of 485 chipped-stone artifacts in the Early Woodland assemblage are heat treated. These figures about double the 5% of 246 Terminal Archaic lithics that are heat treated.

Turning to Rock Bridge Shelter, 26 flake specimens, one piece of debitage, and three chipped-stone tools/tool fragments show evidence of heat treatment. This amounts to 6% of the sample examined and a density of 0.02/liter. Neither of the cores exhibit the expected changes associated with heat treatment. Of the 30 specimens exhibiting signs of heat treatment, 23 have changed color, 10 have reduced translucency, 8 show more lustrous flake scars, and 8 have ripple marked flake scars.

Three chert raw materials were heat treated by the Rock Bridge inhabitants:

Haney (n=15), Paoli (n=12), and Breathitt (n=2). One heat-treated lithic specimen is made of unidentifiable chert. Unlike Cold Oak Shelter, the Rock Bridge sample has low richness as only three of the eight cherts in the assemblage are heat treated. The shelters, however, are similar in terms of low evenness, as 50% and 40% of all heat- treated specimens from Rock Bridge are of Haney and Paoli cherts, respectively:

Breathitt and unknown cherts account for the other 10%.

Flakes diagnostic of primary and/or secondary reduction (85%) as opposed to initial core reduction (15%) dominate the sample of heat-treated flakes from the Late

Woodland assemblage at Rock Bridge. Cores and primary decortication flakes do not exhibit signs of intentional heat treatment. Over half of the heat-treated flakes are bifacial thinning and broken flakes. However, adjusting for differences in sample sizes, the highest proportion of heat-treated specimens are primary flakes, followed by secondary decortication flakes. Almost 8% of all "early-stage " flakes are heat treated and about 6% of all ""late-stage"" flakes are so altered.

1 76 Some patterns in the spatial distribution of heat-treated cherts at Rock Bridge are noted. Looking at the southern excavation block, which yielded 83% of the heat- treated specimens, there are three concentrations of heat-treated artifacts in the south, middle, and north. When one plots the horizontal distribution of heat-treated cherts in the southern excavation block according to density, three concentrations separated by units lacking heat-treated lithics are still evident. These clusters of heated cherts correspond to clusters of all lithic artifacts at Rock Bridge.

The vertical distribution of heat-treated cherts from Rock Bridge's Late

Woodland assemblage indicates that the largest number (n=13, 43%) derive from the silty Zone II. Ten cherts, or 33% of the sample, came from the ashy Zone I. Ten percent

(n=3) and 13% (n=4) were, respectively, from Zone III and from unprovenienced contexts.

Anderson (1979) also outlined archaeological signatures that may indicate the reasons why knappers thermally altered cherts. If they used heat to alter the specific appearance of cherts, one might expect differential distribution of altered versus unaltered cherts in some chert categories, or a high incidence of intentionally heat- altered specimens in particular lithic classes. There is evidence to support the former contention for the Cold Oak assemblages. Twenty percent of all Breathitt specimens are heat treated, compared to the average of 6% for all cherts. On average, 8% each of the

Boyle and Paoli specimens are heat treated. However, there is no evidence to support the latter contention that particular artifact types in the Cold Oak assemblages have higher incidence of heat-treated specimens. For the Rock Bridge assemblage, both expectations are supported. Thirteen percent of the Breathitt specimens are heat treated, compared to

6% for the entire sample; 8% of Paoli specimens are heat treated. Support for the second expectation is the significant percentage (43%) of thermally altered bifacial tools and tool fragments in the Rock Bridge assemblage.

177 If thermal alteration was employed to improve the quality of cherts, one would

expect an overall high incidence of alteration on specific chert types, regardless of lithic

artifact category. This appears to be true for all assemblages. In the Cold Oak

assemblages, high percentages (from 35% to 51%) of Lower Pennsylvanian cherts have evidence of thermal alteration. In the Rock Bridge assemblage, Haney and Paoli cherts are the predominant raw material types exhibiting signs of thermal alteration.

6.1.3. DISCUSSION

Overall, the percentages of heat-treated cherts from Cold Oak and Rock Bridge are similar at about 6% each. The densities are also comparable at 0.04/liter for Cold

Oak and 0.02/liter for Rock Bridge. Heat-treated specimens are consistently dominated by late-stage debitage at both shelters. One difference between the shelters is the richness of chert types showing signs of heat treatment: six of nine chert types in the

Cold Oak assemblages were heated compared to three out of seven chert types in the Rock

Bridge assemblage. Evidence indicates that Breathitt chert was heat treated by occupants of both shelters in order to alter its appearance. Breathitt chert was heated by Cold Oak inhabitants to improve knapping quality and Haney and Paoli was so altered by Rock

Bridge inhabitants.

Comparing the shelter assemblages, heat treatment appears to have been slightly more prevalent during Early Woodland occupations at Cold Oak Shelter. Nine percent of all lithics recovered from Early Woodland contexts are heat treated, compared to 6% of the Late Woodland sample from Rock Bridge and 5% of the Terminal Archaic sample from

Cold Oak. The density of heat-treated cherts in the Early Woodland assemblage is about three times that of the Terminal Archaic and Late Woodland assemblages.

1 78 6.2. RAW MATERIAL UTILIZATION

Three issues related to raw material utilization are considered in this section:

raw material diversity, use of local and exotic raw materials, and raw material

selectivity. These factors of raw material use are compared for the four assemblages

from Cold Oak and Rock Bridge in order to assess temporal patterns in how shelter

occupants used raw materials.

8.2.1. RAW MATERIAL DIVERSITY

Because chipped-stone artifacts are the predominant form of lithic remains

recovered from the shelters, chert is overwhelmingly the most common form of raw

material represented in the assemblages. Minerals and miscellaneous rocks are also

present in some of the assemblages. The number of raw materials used by the shelters'

inhabitants (richness) and the relative proportions of these materials (evenness) are two measures that allow one to describe the raw material utilization strategy.

Chert accounts for 86% of the 2240 lithic specimens in the Cold Oak assemblages

and 89% of the 755 lithic artifacts recovered from Rock Bridge (Tables 5 and 19).

Artifacts made of sedimentary and metamorphic rocks account for 13% and 1% of the

Cold Oak and Rock Bridge assemblages, respectively. About 1% and 2% of the lithics from Cold Oak and Rock Bridge are minerals. Over 8% of the lithic specimens from Rock

Bridge could not be typed by raw material.

There are up to eight varieties of chert in the lithic assemblages. They are - in decreasing order of abundance at Cold Oak Shelter - Haney, St. Louis, Boyle,

Breathitt, Kanawha, Muldraugh, Paoli, and Ste. Genevieve. The former accounts for one- quarter of both the chert specimens and the combined assemblages (Table 5). St. Louis

1 79 makes up 18% of the chert artifacts and 16% of the assemblages. The other chert types

each account for about 9% or less of the lithic assemblages and chert samples. The

author could not identify about one-quarter of the cherts. For the Rock Bridge

assemblage, the chert types identified are - in decreasing order of abundance - Haney,

Paoli, Boyle, Breathitt, Kanawha, St. Louis, and Ste. Genevieve (Table 19). The most

common of these, by far, is Haney, which accounts for about half of the lithic assem blage

and 58% of the chert sample. Paoli makes up about 22% of the assemblage and 25% of

the chert artifacts. Almost 9% of the lithic specimens and 10% of the cherts are Boyle.

The remaining 19% of the lithics and 7% of the cherts are made of Breathitt, Kanawha,

St. Louis, Ste. Genevieve, and unidentified cherts.

With twenty raw material types evidenced at Cold Oak, the lithic assemblages

have high richness. Of the nine chert types known to have been used in the Cold Oak area,

including Renfro (Meadows 1977), eight of these occur in the Cold Oak assemblages.

Because almost half of the chert artifacts are made of two chert types, Haney and St.

Louis, the assemblages have low evenness. When one excludes from consideration the

unidentified cherts in the assemblages, the percentage increases to 64%.

With eleven raw material types evidenced at Rock Bridge, the lithic assem blage

has low richness compared to Cold Oak. Of the eight chert types known to exist in the area of Rock Bridge, including Renfro (Meadows 1977), seven of these occur in the Rock

Bridge assemblage. But considering that about 73% of all specimens and 80% of all

identifiable specimens are made of two cherts, Haney and Paoli, the assemblage has low

evenness.

There are several potential explanations for the relatively low evenness of raw

material use at the shelters (Leonard et al. 1989). First, when few raw materials with the necessary properties (for knapping or other uses) are available, reliance on only a few types may ensue. This explanation, however, is unlikely since other quality cherts

180 (e.g. Boyle, Breathitt) suitable for knapping and tool manufacture are available in the area (Graham 1990; Meadows 1977). A second explanation is that reliance on a few raw material types may result when few raw materials are available. This, too, is unlikely since several other cherts occur in outcrops and/or alluvial deposits in the area

(Graham 1990; Meadows 1977). A third possibility, and one that is difficult to test archaeologically, is that there was a personal and/or cultural preference for the aesthetic qualities of Haney, St. Louis, and Paoli. Perhaps burial goods data for other

Archaic-Woodland sites in the area would help to assess this possibility. Another explanation posits that high procurement costs of some cherts may lead to reliance on a few, lower cost types if they are available. One may measure costs in terms of time expenditures, energy expenditures, transportation requirements, and/or extraction expenditures. While it is possible that Haney, St. Louis and/or Paoli cherts required relatively small expenditures for utilization, there is not enough data about these variables to adequately assess this possibility. It is also possible that the low evenness is due to the relative ease in identifying Haney, Paoli, and St. Louis cherts. The percentage of artifacts made from these cherts may be high because their diagnostic attributes make them easier to identify than some other chert types.

Raw material richness may be a function of sample size (Grayson 1984, 1981;

Jones et al. 1983; Leonard 1989; Leonard et al. 1989). To test this possibility, the relationship between raw material richness and sample size for each of the lithic categories containing chert specimens (tools, cores, flakes, debris, thermal debris) is measured with Pearson's correlation coefficient. A moderate to weak relationship between sample size and richness indicates there is little chance that raw material richness for the lithic categories is a function of sample size rather than, or in addition to, selectivity. If the data indicate patterns in raw material use, they probably reflect personal preferences or lithic availability. If the correlation coefficient is moderate to

1 8 1 high, sample sizes and behavioral strategies, to unknown degrees, both influence richness.

Pearson's correlation coefficient for the Cold Oak assemblages is 0.46. This indicates that a moderate positive relationship exists between sample size and raw material richness. If patterns of raw material use are indicated, they probably will reflect personal preferences or raw material availability to some extent, although these results should be interpreted with caution. The Rock Bridge assemblage yielded a

Pearson's correlation coefficient of 0.85, indicating a strong positive relationship between sample size and raw material richness. Raw material richness, then, may be a function of sample size rather than, or in addition to, selectivity. The issue of raw material selectivity is covered in more detail in the next section.

The previous paragraphs indicate that there are some differences in raw material use at Cold Oak and Rock Bridge shelters. But, are there differences among the separate assemblages from the shelters? The Terminal Archaic assemblage from Cold Oak Shelter is characterized by high richness and low evenness, as seven of the eight chert types are identified but 51% of the specimens are made of two chert types, St. Louis and Haney.

The percentage of St. Louis chert is larger in the Terminal Archaic assemblage than in the other assemblages (Table 26). All eight chert types are found in the Early Woodland assemblage from Cold Oak, and the high richness is accompanied by low evenness. Haney and St. Louis are still the most abundant but at smaller percentages (36% total) compared to the Terminal Archaic assemblage. Compared to the other assemblages, a significantly higher percentage (15%) of Breathitt is noted in the Early Woodland assemblage. Muldraugh chert is only present in the Early Woodland assemblage (Table

26). The Terminal Archaic and Early Woodland assemblages from Cold Oak Shelter differ somewhat from those of other Cogswell Phase sites, where Ison (1988) noted preferential chert use of Haney and Paoli.

182 Rock Bridge's Late Woodland assemblage, which contains seven of the eight chert

types, has more Haney specimens than the other assemblages. At 53%, the percentage of

Haney in the Late Woodland assemblage is more than double the percentages of Haney in the other three assemblages. Other differences are that, compared to the Early Woodland and Terminal Archaic assemblages, the percentage of Paoli is 25 times greater and the percentage of St. Louis is eight to 13 times smaller.

6.2.2. RAW MATERIAL SELECTION

Another question related to raw material utilization at the shelters concerns selectivity in chert use for tool manufacture and flake tool use. If people selectively used certain cherts for the tool manufacture and use, then there should be a low degree of evenness in the chert distribution for chipped-stone tools. However, one must consider whether or not raw material richness by category is a function of sample size. If

Pearson's correlation coefficient between raw material richness and sample size for the seven lithic categories employed in this study is high, there is a strong relationship between sam ple size and richness. Hence, raw material richness for the lithic categories may be a function of sample size rather than, or in addition to, selectivity.

The relationship between raw material richness and sample size for the five lithic categories containing chert specimens (tools, cores, flakes, debris, thermal debris), as measured by Pearson's correlation coefficient, is 0.46 for Cold Oak and 0.85 for Rock Bridge. These indicate that moderate to strong relationships exist between sample size and richness for the samples. Hence, there is some chance that raw material richness for the lithic categories is a function of sample size rather than, or in addition to, selectivity for Cold Oak and a greater chance for Rock Bridge. If data indicate patterns

1 83 in raw material use, the data may reflect personal preferences or lithic availability for

Cold Oak but may not for Rock Bridge.

The chipped-stone tools and tool fragments (n=23) from Cold Oak are composed of six identifiable chert types (Table 13). Curiously, the most common chert type is

Boyle, not Haney or St. Louis, which are the most common chert types in the Cold Oak assemblages. Although Boyle makes up 23% of the tool sample, compared to 9% of the entire chipped-stone sample, the percentage is not that high. The projectile points/ knives (n=6) and hafted biface fragments (n=5) are made of different combinations of four chert types. A different combination of five types of chert comprise the ten biface fragments. The two flake tools are made of other cherts. There seem s to be no clear preference of chert types used for tool manufacture at Cold Oak Shelter. This interpretation must be viewed as provisional because of the small sample sizes, and it is not possible to assess tool raw material types for each assem blage from Cold Oak for the same reason.

Chipped-stone bifaces from Rock Bridge (n=7) are made of three chert types:

Haney, Paoli, and Breathitt. The eight utilized flakes are com posed of Haney and Paoli only, while the 11 marginally modified flakes are made of Haney, Paoli, Kanawha, and

St. Louis (Table 19). The common chert types for all three categories of tools are Haney and Paoli. This suggests that knappers selected the two cherts for tool manufacture over other raw materials. This may be a function of availability and/or personal preferences of the knappers, or it may be due to sample size effects, as discussed in the previous section of this chapter. In addition, this interpretation is somewhat tentative since the total number of tool specimens for the shelter is relatively small. The distribution of raw materials in the debitage, though, support the interpretation.

184 6.2.3. EXOTIC AND LOCAL CHERTS

Geologic maps, published data, and reports of local informants are used to classify cherts as either local or exotic. The Zachariah (Black 1978) and Cobh ill (Haney

1976) quadrangles indicate that the Renfro Formation, which may contain Renfro chert, crops out about one mile west of Cold Oak Shelter along Big Sinking Creek as well as along Billey Fork at the border of Lee and Estill counties about four miles north- northwest of the shelter. Chert-bearing strata of the Slade (Newman limestone) formation crop out near Cold Oak Shelter: less than one mile north and south of the shelter along Big Sinking Creek, and less than two miles north and one mile west of the shelter along Little Sinking Creek. These rocks may contain St. Louis, Ste. Genevieve,

Paoli, and Haney cherts. Deposits of Quaternary alluvium, which contain "angular to rounded fragments of chert from the Newman Limestone" (Haney 1976), occur less than one mile from Cold Oak to the east and south along Big Sinking Creek, Caves Fork, and

Bald Rock Fork.

Of the Big Sinking Creek area, where Cold Oak Shelter is located, O'Steen et al.

(1991:9) stated that chert

was non-existent in the Pennsylvanian rocks of the area. Quaternary age alluvium deposits contain chert gravels that are available from several river and stream sources in the region. Paoli and Haney outcrops in Mississippian [the geological time unit] age Newman limestones are another available source. The Middle Devonian Boyle dolomite produced a fossiliferous Boyle chert that was also utilized by prehistoric people.

The Pomeroyton quadrangle (Weir and Richards 1974) indicates that several chert-bearing strata of the Slade (Newman limestone) formation crops out to the north of Rock Bridge along the Red River; these strata may contain St. Louis, Ste. Genevieve,

Paoli and Haney cherts. Alluvial deposits south and southwest of the shelter may contain chert pebbles from the Slade (Newman limestone) formation.

185 Archaeologists from the National Forest Service (pers. comm.) indicated that

Pennsylvanian-aged Breathitt chert may be found in alluvial deposits in nearby Estill

County; Breathitt-bearing strata crop out in Breathitt County to the east. Meadows

(1977:103-105) notes that Haney, Paoli, Boyle, St. Louis, and Breathitt cherts occur in Powell County in colluvium, in alluvial deposits, and/or in outcrops. It is likely, then, that six of the eight cherts found at Cold Oak and Rock Bridge were locally available to prehistoric inhabitants: Haney, Paoli, St. Louis, Boyle, Breathitt, and Ste. Genevieve.

Nonlocal cherts are Muldraugh and Kanawha. Although the Borden Formation crops out in the study area (Black 1978; Haney 1976), the member from which

Muldraugh chert derives does not. Sable and Dever (1990) indicated that the Muldraugh

Member is exposed in west-central Kentucky. Kanawha chert apparently is exposed only in northwestern West Virginia in Braxton and Lewis Counties and is found as pebbles in alluvial deposits along the Kanawha River and several of its tributaries (Yerkes and

Pecora 1991).

The six locally available cherts comprise three-quarters of the lithic assem blages from Cold Oak and Rock Bridge shelters (Table 26). Exotic cherts make up only 3% and the remaining 22% are unidentified cherts. Because six of the eight chert types present in the assemblages are of local derivation and these local cherts make up a large percentage of the assemblages, one may conclude that the inhabitants of the shelters made use of predominantly local chert materials.

Table 26 shows the percentages of local, exotic, and unidentified cherts in the four assemblages from Cold Oak and Rock Bridge. Although the Late Woodland chert sample contains the highest percentage of local cherts, the statistic is misleading because most of the unidentified cherts in the other samples are probably of local derivation. The percentage of exotic chert in the Early Woodland sample (4%) is slightly higher than that of the Terminal Archaic (3%) and Late Woodland (2%) samples.

186 6.2.4. DISCUSSION: TEMPORAL PATTERNS IN RAW MATERIAL USE

Because the Cold Oak and Rock Bridge lithic assemblages represent occupations over several periods in the same geophysical zone, one can evaluate temporal patterns of raw material use among rockshelter occupants. These data are also important in evaluating the intensity of shelter use over time.

When one considers each chert type individually, some temporal differences in chert use are noted (Table 26). Haney chert dominates each period sample, but shows a somewhat increasing trend over time. The use of Paoli chert increases over time and clearly peaks during the Late Woodland; the percentage of Paoli specimens for Late

Woodland contexts is about 20 times greater than the percentages in Early Woodland and

Terminal Archaic deposits. St. Louis chert shows an opposite trend, with usage decreasing over time. There is a peak in the percentage of Breathitt chert associated with the Early Woodland assemblage. The percentage of Boyle chert is fairly even over time.

Ste. Genevieve occurs in low percentages in all deposits. Though only a small proportion of the samples, Kanawha is slightly more abundant in early deposits than later ones.

Muldraugh specimens are noted in the Early Woodland assemblage only.

The relative proportions of local versus exotic cherts in an assemblage may be indicators of occupational permanence (Ledbetter and O’Steen 1991; O'Steen et al. 1991;

Yerkes 1989). According to this perspective, groups that move less often around the landscape would have reduced access to exotic cherts, barring importation, stockpiling or trading, so the lithic assem blages they produce should have lower percentages of exotic cherts compared to less permanent groups. Although lithic procurement tactics are difficult to distinguish in the archaeological record, Leonard et al. (1989:101) suggested that direct procurement, or "scheduling of procurement of lithic materials ...

187 independent of other activities," is most cost effective when the distances to raw

material sources are relatively small. If groups were relatively sedentary, one might

expect a direct lithic procurement system and nearly exclusive use of local cherts,

assuming no trade or stockpiling of exotics. Embedded procurement, defined as

acquisition of “lithic raw materials ... in the context of another continually functioning

system" (Leonard et al. 1989:101), might be expected if distances are relatively large

and/or the population is mobile. If groups were relatively mobile, one might expect an

embedded lithic procurement system and use of local and nonlocal cherts. The use of both

local and nonlocal cherts may also indicate sedentism and stockpiling or trading of exotic

cherts. Of course, complete evaluation of the lithic procurement strategies of the Cold

Oak and Rock Bridge inhabitants requires more data from rockshelter, quarry, and

open-air sites of the same time periods in the Big Sinking Greek and Red River Gorge

areas.

If there was a shift in rockshelter occupation at the Archaic-Woodland

transition, as several researchers have suggested (see Chapter 2.3.3.), one might expect

to see a corresponding shift in chert raw material usage. In terms of the types of cherts

used, there are differences. The percentage of St. Louis chert specimens in the Terminal

Archaic assemblage is higher than the percentage in the other assemblages. The percentage of Breathitt chert specimens is noticeably higher in the Early Woodland

assemblage than in the other assemblages. Both Haney and Paoli cherts make up higher percentages of the Late Woodland assemblage compared to the other assemblages. For each chert type, the number of lithic specimens of that chert type was compared to the size of each lithic assemblage (Terminal Archaic, Early Woodland, Late Woodland) to yield

Pearson's correlation coefficients (Table 26). Sample size effects, such that there is a strong positive relationship between the two variables, suggest that larger samples yield larger percentages of certain cherts. Sample size effects may explain the high

188 percentages of Haney cherts in the Late Woodland assemblage; Pearson's correlation

coefficient is 0.64. St. Louis, Paoli, and Breathitt cherts have smaller coefficients, so

interassemblage differences may reflect cultural patterns of chert use. The possibility

that interassemblage differences are due to trampling, which could have inflated

specimens counts, cannot be ruled out for the Haney and Paoli chert percentages in the

Late Woodland assemblage and the St. Louis chert percentage in the Terminal Archaic

assemblage.

The Early Woodland assemblage has more exotic cherts and is the only assemblage

with Muldraugh chert specimens, an exotic chert. Tentatively speaking, the higher

percentage of exotics in the Early Woodland assemblage may represent less intense

occupation of Cold Oak Shelter compared to the other periods. If this is the case, it lends

credence to Funkhouser and Webb's (1929, 1930) interpretation that pre-pottery

[Archaic] occupations at some shelters were more intense than pottery [Woodland] occupations. Or, the Early Woodland assemblage may represent more intense occupation with stockpiling of exotic raw materials, which supports Railey's (1990) contentions about shelter use in the Cumberland Plateau area during these time periods.

One way to distinguish between these interpretations is local:exotic ratios by flake reduction classes. Unfortunately, the small sample sizes of exotic specimens limit the analysis (Table 26). If the Early Woodland occupations at Cold Oak were more intense than the Terminal Archaic occupations and exotics were stockpiled, one would expect that the percentage of early-stage lithics made of exotic cherts be greater in the

Early Woodland assemblage. If the Early Woodland occupations were less intense and exotics were obtained by embedded procurement, the percentage of early-stage lithics made of exotic cherts should be comparable to the Terminal Archaic assemblage. The ratio of early to late stage debitage for exotic cherts in the Early Woodland assemblage is

0.4:1, compared to 0.2:1 for the Terminal Archaic assemblage. These data support the

1 89 former contention of more intense shelter use with stockpiling during the Early

Woodland. In fact, the early:late ratio for the exotic specimens in the Early Woodland assemblage (0.4:1) is even higher than that of the local specimens in that assemblage

(0.3:1), lending further support to the conclusion that Early Woodland shelter use was more intense.

When one compares the combined Terminal Archaic and Early Woodland assemblages with the Late Woodland assemblage, more drastic differences are noted. The percentage of St. Louis chert drops off drastically, while Paoli specimens are much more abundant. Exotic cherts are slightly less abundant. The percentages of early:late stage debitage made of exotic cherts is 0.1:1 for the Late Woodland assemblage, compared to

0.2:1 and 0.4:1 for the Terminal Archaic and Early Woodland assemblages, respectively.

These data suggest that earlier shelter occupations at Cold Oak were more intense than the later occupation at Rock Bridge. Of course, the differences may also be related to the geographic locations of the shelters.

Despite these findings, a Pearson's correlation coefficient of 0.68 for exotic cherts per assemblage suggests that differences in percentages are due to sample size effects (Table 26). In addition, the percentages of exotic cherts in all assemblages is somewhat low at less than 5%. Relative to other some sites in the area, these percentages are relatively low. At Cloudsplitter (MF36), the percentages of exotic cherts in

Terminal Archaic and Early Woodland assemblages are 16% and 25%, respectively

(Cowan et al. 1981). At three shelters in Laurel County, the percentages of exotic cherts in Woodland deposits ranged from 19% to 32% (Carmean and Sharp 1995).

In sum, the examination of raw material utilization indicates there are some differences in shelter use over time at Cold Oak and Rock Bridge. The previous section of the paper indicated that there are also temporal differences in shelter use reflected in the incidence of thermal alteration of cherts. In the next section, the third aspect of the

190 lithic production system, tool manufacture and maintenance, is examined for evidence of diachronic differences in shelter use.

6.3. TOOL MANUFACTURE AND MAINTENANCE

This section describes the nature of tool manufacture and maintenance at Cold Oak and Rock Bridge shelters in order to investigate diachronic trends in shelter use and ranges of lithic-related activities. Did different activities related to lithic reduction occur at the shelters over time? How does this third component of the prehistoric lithic production system compare with evidence of thermal alteration and raw material utilization? Three aspects of tool manufacture and maintenance are considered: the intensity of lithic production, reduction stages, and reduction techniques.

6.3.1. INTENSITY OF LITHIC PRODUCTION

First, the debitage index is an indicator of general production at a site. Ericson

(1984) defined the debitage index as the quotient of debitage (excluding retouch and sharpening flakes) and total tools and debitage. “Debitage" includes the decortication, primary, and secondary flakes recovered from the shelters. A relatively high debitage index indicates considerable lithic production, while a low index suggests little lithic production. The debitage indexes for all Cold Oak and Rock Bridge assemblages indicate considerable, and comparable, lithic production at the shelters. For the Terminal

Archaic assemblage, the debitage index is 51%, but it could be as high as 76% if one assumes the broken flakes are not retouch and sharpening flakes and includes them in the numerator. The debitage index for the Early Woodland assemblage is 50% minimally.

191 but could be as high as 79%. The debitage index for the Late Woodland assemblage from

Rock Bridge is at least 46% or as high as 80%.

According to Ericson (1984), the "cortex index" is an indicator of the impor­ tation of lithic raw materials to a site. The cortex index is the number of primary and secondary decortication flakes divided by the total number of debitage, excluding retouch and resharpening flakes, in an assemblage. A relatively high cortex index suggests extensive importation of raw materials that were subsequently reduced, while a relatively small cortex index indicates little importation of unmodified cores. The cortex indices for three assemblages are moderate to low. For the Terminal Archaic, Early

Woodland, and Late Woodland assemblages, the cortex indices range from 0.34 to 0.45,

0.35 to 0.49, and 0.33 to 0.23, respectively."'^ One might conclude, then, that shelter inhabitants imported small numbers of unmodified cores to the shelters. Perhaps, instead, people carried more blanks or preforms than cores to the shelters for subsequent reduction at the shelters.

6.3.2. REDUCTION STAGE

In general, the types of lithic artifacts recovered from the shelters suggest which stages of lithic reduction the assem blages represent. If initial reduction occurred, one would expect to find hammerstones or other tools used to deliver pressure to lithic raw materials, worked cores, and unmodified decortication debitage. Evidence of primary biface reduction includes marginally modified flakes, unmodified debitage, blanks, and preforms. Bifacial tools with specialized edge treatment, bifacial thinning and other flake debitage, and dorsal surface platform retouch suggest secondary reduction or retooling. The composition of the Terminal Archaic, Early Woodland, and Late Woodland

^ The cortex index ranges correspond to whether or not the broken flakes are included in the sample sizes.

192 assemblages suggest that core reduction, primary biface reduction, and secondary reduction or retooling occurred during those occupations at Cold Oak and Rock Bridge.

This section considers if one or more of the lithic reduction activity sets predominated during the other occupations.

Several lines of evidence indicate the predominance of certain lithic reduction activities. The variables and measures utilized for this study are: reduction-class frequency distribution, flake fragmentation and flake-type frequency distribution, platform lipping and faceting, dorsal surface platform retouch, biface index, ratio of modified lithics to debitage, relationship between late-stage debitage and debitage to tool ratio, and platform cortex. While it is recognized that there is considerable variation in these attributes for different manufacturing stages and materials, and that actualistic studies using the raw materials present in the shelter assemblages would be useful as a baseline for comparison, the Cold Oak and Rock Bridge lithic assemblages are evaluated without the benefit of original replication experiments designed to address the present research questions and particular raw material suite.

6.3.2.0. Flake Reduction Classes

The relative proportions of debitage reduction classes may distinguish between early and late reduction stages. Assuming that larger debitage with cortex characterize initial reduction, one might expect high percentages of decortication and primary flakes and decortication debris with initial reduction. Assuming that small debitage lacking cortex characterize the later stages of reduction, one might expect high percentages of secondary and bifacial thinning flakes and cortex-free debris with primary and/or secondary reduction.

Primary and/or secondary reduction are indicated for all of the lithic assemblages, though the ratios of early- to late-stage debitage vary somewhat. The Late

Woodland assemblage had the lowest percentage of early-stage debitage with a ratio of

193 0.1:1. The Terminal Archaic ratio is 0.2:1, and the Early Woodland ratio is 0.3:1.

Therefore, the ratios of flake reduction classes, as expressed by the ratio of early- to late-stage debitage, suggest that the later stages of lithic reduction predominated and that lithic reduction did not differ drastically over time.

6.3.2.b. Flake Fragmentation

Several researchers (Baumler and Downum 1989; Ingbar et al. 1989; Prentiss and Romanski 1989; Tomka 1989; Sullivan and Rosen 1985) have proposed that the relative proportions of flake fragment types and debris distinguish lithic reduction sequences. Sullivan and Rosen (1985) defined flake types according to criteria independent of reduction stage. Instead of terms like primary flake and secondary flake, they use three non-technological dimensions of flake variability (single interior surface, platform, and margins) to identify four types of flakes based on completeness: complete flakes, broken flakes (proximal), and flake fragments (medial and distal).

Sullivan and Rosen (1985) suggested that debitage assemblages with high proportions of broken flakes and flake fragments indicate shaped stone-tool manufacture, while high percentages of complete flakes and debris represent core reduction (Table 23). Thus, flake fragmentation serves as a means of broadly distinguishing initial reduction from primary or secondary reduction. Sullivan and Rosen (1985) based their propositions on chert fracture principles rather than experimental replications.

Ingbar et al. (1989) used flake completeness to assess biface production experiments using chert and quartzite. For both materials, the percentages of complete flakes ranged from 45% to 60%; broken (proximal) flakes ranged from 10% to 20%, flake fragments (medial and distal) ranged from 25% to 30%, and debris represented

5% of the samples (Table 23). These data diverge somewhat from Sullivan and Rosen's

(1985) expectations. Sullivan and Rosen (1985) suggested that assemblages produced during shaped-tool manufacture should have high percentages of proximal flakes and

1 94 flake fragments, ingbar's et al. (1989) data, however, indicate that the percentages of these flake types in the experimental assemblages range from only 35% to 50%.

However, the difference may be attributed to differences in the end results of shaped- tool manufacture (Sullivan and Rosen 1985) and biface production (Ingbar et al.

1 9 8 9 ).

Tomka (1989) conducted three reduction experiments: core reduction to produce blades, flake reduction to produce a biface, and flake reduction to produce a projectile point. These studies are important because they aid in distinguishing primary (biface production) and secondary (point production) reduction, reduction stages that are lumped together in the work of Sullivan and Rosen (1985). In Tomka's (1989) experiments, core reduction assemblages are characteristically 50% complete flakes,

7% broken (proximal) flakes, 38% flake fragments, and 6% debris. Assemblages deriving from biface production contain 34% complete flakes, 20% broken (proximal) flakes, 45% flake fragments, and 2% debris. Point production assemblages are typically

58% complete flakes, 14% broken (proximal) flakes, 28% flake fragments, and 1% debris (Table 23). These data do not support Sullivan and Rosen's (1985) expectations.

Tomka's (1989) data also differ from Ingbar's et al. (1989) in terms of complete flakes and flake fragments.

Baumler and Downum (1989) conducted two replication studies, core reduction

(6 experiments) and end scraper production (16 experiments), in order to assess the relationship between reduction strategy and waste flake type. Using the flakes between two mm and four mm in size as the sample, Baumler and Downum (1989) categorized the chert and obsidian specimens as complete flakes, broken flakes (proximal, medial, distal) and shatter. They reported the percentages of each category for the six core reduction replications and sixteen end scraper experiments. The average percentages of complete flakes, broken flakes and shatter for core reduction are 10.9%, 57.2% and

195 31.8%. For the end scraper replications the average percentages were 50.4%, 45.1%

and 4.5% (Table 23). There are similarities and differences with Baumler and

Downum's (1989) data and the other researchers' data.

A crucial assumption built into the preceding models is that trampling does not

affect the proportions of flake types in the lithic assemblages. Trampling, however, can

lead to fragmentation of flakes. Because it is likely that trampling fragmented lithic

artifacts in the Terminal Archaic and Late Woodland assemblages, one must consider the

possibility that this formation process affected the proportions of flake types.

Trampling, instead of lithic production activities, may explain the relatively low

numbers of complete flakes relative to broken flakes and flake fragments.

To assess the relationship between flake fragmentation due to lithic production versus trampling, the lithic assemblages from Cold Oak and Rock Bridge shelters are compared to experimentally trampled assemblages produced by different lithic reduction strategies (Prentiss and Romanski 1989). Their work was summarized in Chapter 5.

Trampling of their two experimental core reduction assemblages generally resulted in fewer complete flakes and debitage specimens and increased numbers of proximal and medial/distal flakes. Similarly, trampling of their two tool reduction assemblages led to fewer complete flakes and more proximal and medial/distal flakes.

Table 23 presents the proportions of complete flakes, broken flakes, flake fragments and debris for the Cold Oak and Rock Bridge assemblages. The Terminal

Archaic, Early Woodland, and Late Woodland assemblages resemble Prentiss and

Romanski's trampled biface and trampled end scraper manufacture assemblages and

Sullivan and Rosen's hypothetical shaped-tool manufacture assemblage. The Terminal

Archaic and Late Woodland distributions are so similar to the Early Woodland distribution that, even though the latter was not fragmented by trampling, there is a good indication that the assemblages represent similar reduction activities. Based on

1 96 these data, one might conclude that lithic reduction activities varied little over the late

Archaic-Woodland periods. While the Prentiss and Romanski (1989) data seem very reliable and relevant to the current study, because the lithic reduction experiments of

Tomka (1989), Ingbar et al. (1989), and Baumler and Downum (1989) yielded different results, it is necessary to look at other indicators of lithic reduction stage.

6.3.2.C. Platform Morphology

In addition to flake fragment studies, Sullivan and Rosen (1985) identified two types of evidence that may confirm or counter the conclusions drawn from the distribution of flake fragment types and debris. If the proportions of debitage categories indicate tool manufacture, as is the case in this analysis, then one would also expect high incidence of platform lipping and faceting on the complete and proximal flakes. According to Sullivan and Rosen (1985:764), "collections should be characterized by abundant evidence of faceting and lipping, if these collections did ... result from tool manufacture."

Platform attributes support the contention that the shelter lithic assemblages derived from tool manufacture more so than initial reduction. Platform lipping was recorded for 88% of the Terminal Archaic complete and proximal flakes, 84% of the

Early Woodland sample, and 55% of the Late Woodland sample. Platform faceting is less prevalent, but the percentages of complete and proximal flakes with faceted platforms are comparable for the assemblages: 28% for the Terminal Archaic, 26% for the Early

Woodland, and 32% for the Late Woodland.

6.3.2.d. Dorsal Platform Retouch

Another potential indicator of reduction stages is dorsal surface striking platform retouch or damage. This refers to flake removal or crushing on the dorsal side of the striking platform of a flake. The damage occurs before a knapper removes the flake exhibiting the retouch from a larger piece of material. Johnson (1975) argued that this type of platform morphology indicates core preparation or retooling.

197 The relatively low percentages of flakes with dorsal platform retouch for all cultural zones of Cold Oak and Rock Bridge suggest that little to some retooling and/or core preparation occurred at the shelters. The Terminal Archaic and Early Woodland percentages are comparable at 19% and 17%, respectively. About 29% of the Late

Woodland specimens show evidence of retouch or damage on the dorsal platform edge.

This suggests that retooling and/or core preparation were slightly more common during the later occupation.

6.3.2.e. Biface Index

The biface index is another means of assessing reduction strategies. It is the quotient of bifacial thinning flakes and tool debitage. Ericson (1984) contends that the biface index measures biface production at a location. The biface indexes for the Cold Oak and Rock Bridge assemblages suggest varying but comparable degrees of biface production over time. Using the flake sample and excluding broken flakes and blades, the biface index is 0.24, 0.22, and 0.28 for the Terminal Archaic, Early Woodland, and Late

Woodland assemblages, respectively. The indexes, however, could be higher if some or all of the broken flakes are bifacial thinning flakes: 0.45, 0.44, and 0.56, respectively.

6.3.2.f. Modified Lithlcs-.Oebltage

Ahler (1988) suggested that the ratio of modified lithics to debitage in an assemblage indicates reduction activities. A relatively high ratio represents tool use and discard while a low ratio suggests tool manufacture and breakage. Approximately equal numbers of modified items and debitage indicates that "activities associated with tool manufacture, use and discard took place in similar parts of the site" (1988:469).

Following Ahler's (1988) model, the ratios of modified lithics to debitage for the

Gold Oak and Rock Bridge assemblages indicate that tool manufacture and breakage were more common during the earlier periods, but tool use and discard were more prevalent

198 in the late Woodland. The tool:debitage ratio is 1:246 for the Terminal Archaic

assemblage, 1:81 for the Early Woodland, and 1:19 for the Late Woodland.

6.3.2.g. Late-State Debitage and DebltagerTool Ratio

Magne (1989) used two lithic measures to characterize assemblage formation

and reduction stages, the percentage of late-stage debitage and the debitage:tool ratio.

Late-stage debitage is “debitage produced in finishing complex tools and in resharpening

and maintenance" (Magne 1989:20). Magne (1989) dichotomized both lithic measures

into high and low and uses them as axes to delineate four quadrants of lithic strategies

(Figure 20). Assemblages with low percentages of late-stage debitage and low

debitage:tool ratios indicate tool and blank manufacture coupled with a high rejection

rate. The low to moderate percentage of late-stage debitage coupled with the high ratio of

debitage to tools corresponds to tool/blank manufacture and high export rate.

Assemblages with high percentages of late-stage debitage and low debitage;tool ratios suggest tool maintenance, high discard rates, and low conservation rates. A high percentage and high ratio characterize tool maintenance, low discard rate, and high conservation.

In addition to these quadrants, Magne (1989) outlined three special case situations. An extremely low proportion of late-stage debitage and an equally low debitage:tool ratio characterize situational repair coupled with adequate raw material availability. Situational repair coupled with raw material scarcity correlate with an extremely high percentage of late-stage debitage and an equally high debitage:tool ratio.

Percentages and ratios that are about mid-scale suggest reoccupation of sites and high reuse or scavenging rates.

The low to moderate percentage of late-stage debitage coupled with the high ratio of debitage to tools in the Terminal Archaic and Early Woodland assemblages correspond to the quadrant Magne (1989) associates with "tool/blank manufacture and high export

199 rate.” Taking late-stage debitage to be bifacial thinning flakes, the percentage of such material in the two assemblages ranges from 20% to 49%, depending on how many of the broken flakes are thinning flakes. The debitage to tool ratio for the two assem blages is

1.0. The Late Woodland assemblage suggests "tool/blank manufacture and high export rate” or "tool maintenance with low discard rates and high conservation;" the percentage of late-stage debitage ranges from 17% to 56% and the debitage:tool ratio is 0.95. That knappers reworked one of the complete hafted biface tools supports the proposition that

Late Woodland knappers practiced conservation to some degree.

6.3.2.h. Platform Cortex

Tomka (1989) contended that the percentage of flakes with platform cortex in an assemblage distinguish among three reduction processes: (1) multi-directional core reduction to produce flake blades, (2) bifacial cobble reduction to make bifaces

(secondary trimming), and (3) bifacial flake reduction to make hafted bifaces. Tomka

(1989) based his conclusions on experimental replications using Edwards Plateau

(Texas) chert. The percentage of flakes with platform cortex for the first experimental sample is almost 2%, with platform cortex absent on the other 98%. Tomka (1989) did not indicate if he divided the number of specimens with or without cortex by the total num ber of pieces in the assemblage or by the total number of flakes with platforms in the assemblage. It is assumed in this analysis that the latter is the case. For the second experiment, bifacial core reduction, the percentage of flakes with platform cortex increases to over 15%. About 9% of the sample produced by the third experiment, bifacial flake reduction, had platform cortex.

Variable results were obtained when examining the percentages of flakes with platform cortex in the four assemblages. Eleven percent and 10% of the Terminal

Archaic and Early Woodland complete and proximal flakes retain cortex, respectively, which is similar to Tomka's third experiment, production of a hafted biface from a flake.

200 The percentage of flakes with platform cortex is 6% in the Late Woodland assemblage,

which falls between Tomka's percentages for blade production and hafted biface

production from a flake and therefore is inconclusive.

6.3.2.1. Summary

Based on the eight indicators of lithic reduction stage, it is concluded that the

nature of lithic reduction differed little over time at Cold Oak and Rock Bridge shelters.

The distributions of flake reduction classes suggest that primary and/or secondary

reduction predominated at the shelters over the four time periods. Flake fragment types

suggest that knappers produced shaped tools, bifaces, and/or end scrapers during the

Archaic and Woodland periods. Tool manufacture is indicated by the platform morphology

of complete and proximal flakes in the four assemblages. Few differences in platform

retouch are observed during the earlier periods, indicating that retooling and core preparation were not common activities at Cold Oak; although the percentage of platform

retouch increases in the Late Woodland, it still is relatively low. The biface index is comparable for the assemblages.

Several measures suggest temporal differences at the shelters, and in all cases, the two early assemblages are similar to each other but different from the later assemblage. One indicator that suggests diachronic differences in reduction activities is the tools:debitage ratio; it indicate that tool manufacture and breakage predominated during the earlier periods, and tool use and discard was more common during the Late

Woodland. Similarly, the percentage of late-stage debitage and the debitagertool ratio show similarities in the early assemblages, during which times tool or blank manufacture and high export rates predominated. The Late Woodland assemblage is suggestive of tool or blank manufacture and high export rates or tool maintenance, low discard rates, and high conservation; the tool maintenance interpretation is supported by

201 the higher percentage of platform retouch in the Late Woodland assemblage compared to the other assemblages.

6.3.3. REDUCTION TECHNIQUES

Lithic indicators of hard-hammer versus soft-hammer percussion techniques are platform lipping and bulb fissures. Several authors (Mauldin and Amick 1989;

Newmann and Johnson 1979) proposed that small waste flakes with lipped platforms result from soft-hammer percussion that occurs near the end of the manufacturing process. Johnson (1989) suggested that large flakes with fissures on the bulb of percussion characterize hard-hammer percussion.

As previously indicated, platform lipping is common in all the assem blages. The average dimensions of the lipped flakes is equal to or smaller than the averages for all flakes, and most of the lipped flakes are secondary and bifacial thinning flakes. This attribute, then, suggests that soft-hammer percussion was a reduction technique employed by knappers during all shelter occupations at Cold Oak Shelter. Bulb scars and fissures are not common on the lithics in the four assemblages, ranging from 13% in the

Terminal Archaic to 8% in the Early Woodland, to 20% in the Late Woodland. These data suggest that knappers did not use hard-hammer percussion extensively at the shelters over time.

6.4. SUMMARY

Evaluation of the lithic production system at Cold Oak and Rock Bridge shelters focused on heat treatment of cherts, the nature of raw material utilization, and tool manufacture. There is some indication of diachronic differences in thermal alteration of

202 cherts. Heat treatment was slightly more common during the Early Woodland period, suggesting that shelter use during that time was more intense, and heat was used to alter the appearance and knapping properties of Breathitt chert. Haney and Paoli cherts may have been heated to alter knapping qualities during the Late Woodland. The densities of heat-treated lithics in the Terminal Archaic and Late Woodland assemblages are comparable.

Temporal differences in raw material use are also noted. Though Haney is present in all the assemblages, the use of Haney is considerably higher in the Late Woodland period. Specimens of St. Louis chert are more common in the earlier assemblages, especially the Terminal Archaic. Use of Breathitt chert peaks in the Early Woodland, and

Paoli use peaks in the Late Woodland. Cold Oak differs from other Cogswell Phase sites in the low incidence of Paoli cherts in the Terminal Archaic and Early Woodland assemblages. Use of exotic cherts is most common during the Early Woodland, and the ratio of early- to late-stage exotic debitage suggests that exotic cherts were stockpiled by groups during this time.

There are few indications that lithic reduction activities differed dramatically over time at Cold Oak and Rock Bridge. The four assemblages appear to have been formed as a result of primary and secondary reduction as opposed to core reduction. Biface production and shaped tool manufacture appear to have been the most common activities over time. However, there is some evidence that tool maintenance was more prevalent in the Late Woodland period. Soft-hammer percussion was employed across the Archaic-

Woodland periods.

These analyses suggest that people transported mostly local raw materials in an altered form, perhaps blanks or preforms, to Cold Oak and Rock Bridge and further modified them at the shelters through thermal alteration and reduction. There was little variation in lithic reduction activities over time at the shelters. But thermal alteration

203 and raw material use patterns suggest that Early Woodland occupations were more intense than the other occupations. In the next chapter, the results presented in this chapter and the two previous chapters are combined to address comprehensively the issues of diachronic differences in rockshelter use at Cold Oak and Rock Bridge. In addition, the lithic evidence is compared to other artifactual remains from the sites, and the findings from Cold Oak and Rock Bridge shelters are compared to other shelters to see if the patterns apply elsewhere.

204 CHAPTER 7

SYNTHESIS AND EXTENSION: DIACHRONIC PATTERNS OF ROCKSHELTER USE

Before launching into a consideration of diachronic differences in the nature of

occupations at Cold Oak and Rock Bridge, it may be helpful to recap what has been learned

thus far. It was noted in Chapter 4 that the four lithic assemblages from Cold Oak and

Rock Bridge shelters, while similar in terms of artifact composition, differ in density,

tool ratios, and tool diversity. The Early Woodland assem blage has higher values for

these measures. The analysis of formation processes in Chapter 5 revealed that, among other things, the sample sizes of the Terminal Archaic and Late Woodland assemblages

may be inflated because of artifact fragmentation due to trampling. The implication of this finding is that density differences between these assemblages and the Early Woodland assemblage may be even greater than the raw densities indicate.

Chapter 6 presented the results of the lithic production system analysis for the

Cold Oak and Rock Bridge assemblages. Intentional heat treatment of chert appears to

have been more common during the Early Woodland period and may have been employed to alter the appearance or improve the knapping qualities of some cherts. Patterns of

raw material utilization shifted over time, with St. Louis use peaking in the Terminal

Archaic, Breathitt use peaking in the Early Woodland, and the use of Haney and Paoli being most prevalent during the Late Woodland. While shelter inhabitants relied on locally available cherts over time, exotic cherts are slightly more abundant in the Early

Woodland assemblage. Coupled with the relatively substantial proportion of early-stage

205 exotic debitage from Early Woodland contexts, the percentage of exotic cherts suggests

that Early Woodland groups used Cold Oak Shelter more intensely than in other periods

and that they may have stockpiled exotic materials for use in chipped-stone tool

manufacture. Despite differences in heat treatment and raw material utilization,

however, there is little indication that the nature and intensity of lithic production

activities varied considerably over time at Cold Oak and Rock Bridge.

The goals of this chapter are three-fold. First, while it is recognized that differences do not necessarily indicate change, the statistical significance of the differences documented among the lithic assem blages is evaluated with respect to the question of diachronic shifts in shelter use. Lithic artifacts recovered from 1984 excavations at Cold Oak Shelter are included in this discussion. Second, the results of the lithic analyses are compared with shelter stratigraphy, features, organic materials, and other artifactual remains from Cold Oak and Rock Bridge in order to assess occupational variability at the shelters. Finally, the Cold Oak and Rock Bridge occupations are compared with select other shelters in the study area to see if the patterns of shelter use at Cold Oak and Rock Bridge apply to other shelters. Before addressing these issues, a review of lithic and other indicators of occupational intensity is presented.

7.1. INDICATORS OF OCCUPATIONAL INTENSITY

In this study, rockshelter occupations are described in a relative manner according to occupational variability. There are four dimensions of occupational variability: (1) the duration of each occupation, usually measured in seasons, (2) the frequency with which occupants revisited the site, (3) the range of activities that occurred at the location, and (4) the number of people utilizing the site at a given time.

Rockshelter occupations can vary in one or more or these dimensions over time. Group

206 size may have certain limits at rockshelters, especially the smaller ones or the ones

with extensive roof fall, as the number of people who can comfortably use the shelters is

limited by the amount of usable space or ceiling height.

Lithic indicators of occupational variability have been alluded to throughout the

dissertation, but it is important to emphasize them here before applying them to the Cold

Oak and Rock Bridge assemblages. Differences in occupational intensity at the shelters

should be reflected in archaeological signatures of lithics and other artifactual remains,

features, and stratigraphy. Non-lithic indicators are considered because it is reckless to

evaluate the nature of shelter occupation on the basis of just one artifact class. It is

advisable to use multiple indicators because sometimes there are multiple explanations for lithic patterns. It is also important to note that the following indicators of occupational intensity are assumptions that should be tested independently, perhaps at single occupation sites.

Several lithic and non-lithic indicators are employed to assess duration of

occupation. The number or density of lithic artifacts at a location may be proportional to the duration of site use. Of course, this measure may be influenced by a number of factors that one must control when comparing sites or components. Formation processes that fragment artifacts, such as trampling, may skew the relationship between assemblage size and site use. Differential rates of sediment accumulation can affect lithic densities, such that strata with higher rates of sediment accumulation may have lower artifact densities. Site function may also influence lithic abundance and density. For example, a location used for plant processing is likely to have a lower abundance and density of certain lithic artifacts than a tool manufacturing location.

Tool diversity at a site or during an occupational episode is proportional to the length of stay. Yerkes (1989) noted, but did not necessarily advocate, that tool diversity

is a traditional index of occupational duration. According to this view, the longer people

207 stay at a location, the more activities they will participate in to fulfill their subsistence and nonsubsistence needs, if occupants complete the activities with task-specific tools, in other words one tool or few tools are not used to perform all tasks, longer occupations are indicated by a relatively wide range of tool types (Beardsley et al. 1956).

Based on an archaeological survey of sites in the southeastern United States,

Johnson (1989) argued that evidence of chert thermal alteration may help to distinguish lengths of occupation. At sites occupied for relatively longer periods of time, one would expect a relatively high incidence of heat alteration since this process requires adequate time and care to be carried out effectively. Signs of heat alteration should be more prevalent on reduction debris at such sites. At sites occupied for shorter periods of time, on the other hand, one might expect to find heat-altered tools and resharpening or reworking debitage. Of course, thermal alteration of chert is not practiced by all groups, occupational permenance notwithstanding: even if they are staying at a location for long periods of time, knappers who are pleased with chert properties they may opt not to alter cherts thermally.

Another indicator related to length of stay is the proportion of exotic lithic material at a location (Binford 1979; Meltzer 1984-85). Simply stated, when groups spend longer periods of time in one location, they are less likely to obtain exotic lithics through direct procurement and, consequently, there will be fewer exotic materials in the assemblages. However, one must rule out the possibility that "exotics" derived from locally available gravel cherts (Meltzer 1984-85), and one must consider the types of lithics made of exotic cherts (Yerkes 1989). Yerkes (1989) indicated that assemblages containing preforms or finished tools of exotic materials are likely produced by more

"mobile" groups, who shaped the artifacts before transporting them, whereas assemblages with exotic tools and debitage are probably produced by more "sedentary"

208 groups who stockpiled foreign cherts and reduced the lithics on site. Trade may also

relate to the proportions of exotic cherts found at a site.

Non-lithic indicators of occupational duration are features, midden deposits, and

organic remains. One might expect a positive correlation between duration, feature

diversity, and the number of features. As occupations increase in duration, and people

engage in more activities associated with different seasons, the number and diversity of

features should increase. The types of features may vary as well. For example,

multiseason occupations (or cold season occupations) might be associated with

structural remains like post molds.

In terms of midden deposits, one might expect that thicker midden accumulations

will mark occupations of longer duration. The longer a site is occupied, the more

material will accumulate. If midden thickness is used as an indicator of occupational duration, one must consider the origin of the sediments, as only those components deriving from anthropogenic sources should be used to estimate duration. This is another reason why formation processes resulting in sediment deposition must be considered.

Organic materials are commonly used as seasonal indicators and can inform about duration of site occupation. Certain plants are only available during select seasons of the year and, barring storage, these materials can be used to infer season(s) of occupation.

Seasonal fruits such as blackberry, dewberry, and persimmon as well as nuts are used to estimate occupational duration (Gremillion 1993). Wood charcoal, an indicator of the production of fires at a site, may inform about occupational duration (Gremillion

1995). Faunal seasonal indicators include age distributions, which require large skeletal samples, and evidence of horn development, which is somewhat uncommon at most sites.

Frequency of occupation is often more difficult to tie down than duration of occupation. It is difficult to differentiate archaeologically the frequency and duration of

209 occupations. Lithic and artifact densities may indicate frequency of occupations, but they also may indicate duration of occupation and they are subject to the effects of formation processes and differential sedimentation rates. If frequent use of a shelter occurred during the same season, low artifact diversity, low feature diversity, and seasonally- sensitive organic remains may distinguish such occupations from long duration occupations that spanned several seasons. Frequent use in different seasons, though, may leave similar traces to long duration occupations: perhaps the lateral distribution of materials would differ. It is difficult to distinguish infrequent use from short duration of use.

The caching of lithic and other tools may indicate (intended) site reuse.

Stratigraphie indicators may be used to infer frequency of occupation at a site.

Superposition of distinct occupational levels or living floors, which may be distinguished by differences in sedimentological properties, artifactual remains, or chronometric dates, suggests frequent reuse of a site. Nielsen (1991b) argued that the loose layer of sediment formed by trampling at a site should be underlain by a hard, compact, sterile stratum unless previous occupations exist at the location.

There are several indicators for the third component of occupational variability, range of activities. In general, it is likely that more intense use of a site will be associated with a wider range of activities. Chapman (1981), Ledbetter and O'Steen

(1991), and O’Steen et al. (1991) suggested that different activities leave predictable archaeological residues that may be used to reconstruct site function (Table 27). Of course, differential preservation within and between locations will affect functional interpretations of archaeological residues.

Lithic indicators of various site functions are relevant in this study. Manos, m eta tes, hominy holes, pitted cobbles, and fire-altered rock are indicators of food processing and food storage. If fabrication and processing of organic materials such as

21 0 bone or wood occurred, one would expect to find hafted bifaces, fire-altered rock, blades,

drills, end scrapers, spokesfiaves, wedges, axes, celts, adzes, and utilized flakes. Hafted

bifaces, utilized flakes, blades, bifacial flake knives or cutting implements, and

cfioppers suggest that butchering and hide preparation may have occurred.

Indicators of lithic maintenance are hafted bifaces, abraders, hammerstones,

bifacial thinning flakes, and flake tool fragments, while indicators of lithic manufacture

are cores, debitage, unfinished bifaces, hammerstones, abraders, anvils, pitted cobbles,

groundstone, and preforms. If hunting activities occurred in association with the

shelters, one would expect to find hafted bifaces, atlatl weights, and utilized flakes.

Utilized flakes may indicate that fishing activities occurred in association with the

shelters. Celts and grooved axes suggest non-lithic procurement, and axes, picks, and

test cores indicate lithic procurement. Indicators of personal status maintenance and

social activity are ochre, hematite, and lithic burial goods.

Several indicators can be used to evaluate lithic-related activities that occurred

at a location. The intensity of lithic reduction is measured by lithic densities (in relation to the type of reduction activities that occurred, in that tool maintenance may produce

less lithic material than quarrying, for example), the debitage index, and the cortex

index. Reduction activities are indicated by thermal alteration, the distribution of lithic

reduction classes, flake fragment types, flake platform morphology, the biface index, the

ratio of modified lithics to debitage, and the relationship between late-stage debitage and the debitageitool ratio. Indicators of reduction techniques include flake platform

morphology and bulb fissures.

Like frequency of occupation, group size is difficult to a ssess archaeologically and

little has been written on this subject. Lithic and other artifact densities or artifact

diversity may be used to infer relative group size during shelter occupations. One might

expect that artifact densities would increase as group size increases. Artifact diversity

2 1 1 might also Increase with increasing group size if people were engaging in different activities.

In summary, the proposed relationships between the occupational history of a location and the resulting archaeological assemblages are as follows. Rather thin occupation strata, the lack of temporally distinct strata, the relative paucity of lithic and other artifactual remains, the lack of thermally altered cherts, reliance on local cherts and the presence of modified exotic cherts, and the somewhat specialized nature of the lithic tools suggest short duration occupations that involved relatively small groups of individuals. Infrequent but long-term use of the shelters is indicated by temporally distinct strata, diverse tool types, chert thermal alteration, local cherts and/or unmodified exotic cherts, and more abundant remains. Temporally distinct strata, specialized tools, little thermal alteration of chert, local and modified exotic cherts, and abundant remains suggest infrequent, short-term use of shelters by large social groups, whereas temporally distinct strata, diverse tool types, abundant remains, local and/or unmodified exotic cherts, and substantial dwellings indicate infrequent, long-term use by large social groups. Frequent, short-term use by small groups is indicated by the lack of temporally distinct strata, specialized tools, local and modified exotic cherts, and low artifact densities. The lack of temporally distinct strata and heat-treated chert, specialized tools, local and modified exotic cherts, but more remains characterize frequent, short-term use of shelters by larger groups. Frequent and long-term use of shelters by large or small groups is indicated by the lack of temporally distinct strata, diverse tool types, more remains, local and/or unmodified exotic cherts, and evidence of chert thermal alteration. Caching of tools suggests repeated use.

212 7.2. LITHIC ASSEMBLAGES AND OCCUPATIONAL VARIABILITY

Overall, when one compares the entire lithic collections of Cold Oak and Rock

Bridge, there is little indication of radical differences in lithic activities and occupation.

But when one compares the lithic assemblages of the different cultural components, there are some differences, as summarized in Table 28. The significance of these differences with respect to diachronic patterns of rockshelter use, are evaluated in this section. Lithic indicators of occupational variability are compared for the assemblages with sufficient sample sizes (Terminal Archaic, Early Woodland, and Late Woodland).

In most cases, the Cold Oak assemblage data are based on those lithics recovered during

1994 excavations. Then, as best as one can do considering differences in lithic classification, the 1984 data from O'Steen et al. (1991:81) are combined with the

1994 data for some indicators. Nonlithic indicators of occupational variability are covered in the next section.

7.2.1. DURATION OF OCCUPATION

Considering first duration of occupation, some lithic indicators suggest that occupation of Cold Oak during the Early Woodland was of longer duration compared to the

Terminal Archaic period and compared to the Late Woodland occupation of Rock Bridge

Shelter. These indicators are lithic density, tool diversity, heat alteration, and raw material use (Table 28).

Table 28 summarizes the lithic densities of the assemblages. Based on the 1994 collection, the lithic densities of Terminal Archaic (0.496/liter) and Late Woodland

(0.422/liter) deposits are nearly equal, and both are only half that of the Early

Woodland (0.878/liter). The data suggest that the Early Woodland occupation was of

213 longer duration. In addition, trampling probably inflated the Terminal Archaic and Late

Woodland sample sizes, which would also inflate their densities.

However, with the combined 1994 and 1984 collections, the density difference between Terminal Archaic and Early Woodland decreases and the difference between

Terminal Archaic and Late Woodland increases. Indications are that the Terminal Archaic lithic density is likely higher than that of the Early Woodland in the 1984 collection.

The number of lithics recovered from Zone II (Early Woodland) in 1984 is 497 and the number from Terminal Archaic contexts (Zones III, IV, V) is 500. These values are nearly identical, but the volume of Zone II sediments, based on the 1984 trench profile, is likely higher than that of the lower zones. This would result in a lower density for the

Early Woodland assemblage, and the difference in densities for the two periods would be smaller than it is for the 1994 sample.

Tool diversity, measured as the quotient of tool types and number of tool specimens, is rather low for all assemblages (Table 28). Because of the small samples sizes of ground-stone artifacts, the tool diversity represents chipped-stone tools only.

The maximum value of tool diversity is 1.0. Considering both 1984 and 1994 samples, the ratio ranges from 0.15 for the Terminal Archaic, to 0.23 for the Late Woodland, to

0.31 for the Early Woodland (Table 28). The differences may be due to sample size effects (r=0.86), or they may suggest that, relatively speaking, the Early Woodland occupation was of longer duration than the occupations during other periods and the Late

Woodland occupation was of longer duration that the Terminal Archaic occupation.

Examination of the proportion and density of heat-treated cherts suggests longer durations of occupations during the Early Woodland compared to the other periods. Based on the 1994 sample only (O'Steen et al., 1991, did not evaluate heat treatment for the

1984 sample), 9% of the chipped-stone artifacts from Early Woodland contexts are heat treated, whereas the Late Woodland and Terminal Archaic deposits yielded 6% and 5%,

214 respectively (Table 28). Density differences are more obvious. The density of heat- treated specimens in Early Woodland deposits is three to five times that of the Terminal

Archaic and Late Woodland deposits.

The proportion of exotic cherts in the lithic assemblages are comparable. About

4% of the Early Woodland specimens are made of exotic cherts, compared to 3% for the

Terminal Archaic and 2% for the Late Woodland (Table 26). Excluding the untyped chert specim ens, the highest percentage of exotic cherts, 6.5%, is still associated with Early

Woodland deposits; 4.7% of the typed cherts from Terminal Archaic contexts are exotic and 2.3% of the Late Woodland cherts are nonlocal (Table 28). These statistics are not distinctive, but when one considers the reduction classes of exotic chert specimens more obvious interassemblage differences are noted.

One Muldraugh decortication flake was recovered from Early Woodland contexts, and Early Woodland deposits yielded the highest proportion of early-stage exotic debitage at 29%, which is larger than the proportion of early-stage local debitage in the Early

Woodland assemblage and twice that of the Terminal Archaic deposits (Table 28). Almost

5% of the Terminal Archaic cherts are exotic and 14% of the exotic chert is early stage, and over 2% of the Late Woodland cherts are exotic and 10% of the exotic chert is early stage. If the relative use of exotic and local cherts is an indicator of residential permanence and the presence of early-stage exotic debitage is associated with stock piling, one might conclude that Early Woodland occupations were of somewhat longer duration than Terminal Archaic occupations, which in turn were of longer duration than

Late Woodland occupations.

215 7.2.2. FREQUENCY OF OCCUPATION

In terms of frequency of occupation, tfiere are indications tfiat, relative to Late

Woodland occupations, tfie Terminal Arcfiaic and Early Woodland occupations occurred more frequently. Evaluation of tfiis dimension of occupational variability depends on stratigrapfiy and the stratigraphie distribution of lithics at the shelters.

The various discontinuous beds and lenses of sedim ent associated with Zones lib and lie (Early Woodland) and Zones III, IV and V (Terminal Archaic), coupled with the proportions and densities of lithics recovered from these contexts, suggest frequent use of Cold Oak Shelter during these periods. Late Woodland occupations at Rock Bridge

Shelter could have been repeated but relatively less frequently. While cached lithics discovered at Rock Bridge and associated with the Late Woodland occupation suggest intended reuse, the absence of sedimentologically and chronometrically distinct strata contrasts with the Terminal Archaic and Early Woodland deposits at Cold Oak Shelter. In addition, the loose layer of sediment with artifacts underlain by a hard, compact, sterile zone at Rock Bridge suggests little occupation prior to the Late Woodland occupation

(Nielsen 1991b).

7.2.3. RANGE OF LITHIC-RELATED ACTIVITIES

In terms of the range of lithic-related activities that occurred at the shelters over the Archaic and Woodland periods, there are some differences in raw material use.

There are temporal differences in the lithic samples in terms of lithic reduction strategies and possibly intensity of lithic production.

216 7.2.3.a. Raw Material Use

As noted previously, the frequencies of exotic cherts in the assemblages are comparable (Table 28). A primary decortication flake of exotic chert derived from

Early Woodland contexts, in contrast with the other samples. Early-stage debitage of

exotic materials is also most prevalent in the Early Woodland assemblage. Apparently,

people transported un worked and worked exotic cherts to Cold Oak during the Early

Woodland occupation, possibly stock piling the materials.

Ison (1988) proposed that Cogswell-phase groups in the study area differentially used locally available Paoli and Haney cherts. The frequency of Haney and

Paoli cherts compared to other local cherts, however, is considerably higher for the Late

Woodland (78%) assemblage than for the Terminal Archaic (26%) and Early Woodland

(20%) assemblages from the two shelters considered here (Table 26). In addition, there are indications that Late Woodland knappers used Haney and Paoli chert preferentially in the manufacture of chipped-stone tools.

The lithic assemblages suggest that people used St. Louis chert more extensively during the Terminal Archaic and Early Woodland periods than during the Late Woodland.

The percentage of chert specimens made of St. Louis is 26% for the Terminal Archaic assemblage and 17% for the Early Woodland assemblage, com pared to 2% for the Late

Woodland sample (Table 26). Another difference in chert use is the peak in Breathitt use during the Early Woodland period, in which 15% of all cherts are made of Breathitt, compared to 2% of the Terminal Archaic assemblage and 3% of the Late Woodland assemblage.

7.2.3.b. Intensity of Lithic Production

There could be slight differences in the intensity of lithic production, as measured by lithic densities and debitage indexes, represented in the shelter assemblages. Lithic densities were discussed previously. While the combined 1984 and

217 1994 samples from Cold Oak indicate that the Terminal Archaic and Early Woodland densities are greater than the density of the Late Woodland assemblage (Table 28), the

Terminal Archaic and Late Woodland densities may be distorted to some extent by the effects of trampling.

The debitage indexes for the four assemblages are difficult to interpret because they represent ranges of values, but they indicate few differences among the assemblages. Looking at the 1994 collection, the lower endpoints of the indexes are similar for all assemblages; they range from 46% to 51% (Table 28). Similarities are also noted in the higher endpoints, which range from 76% to 80% (Table 28). When one considers the combined 1984 and 1994 collections, differences are only evident in the lower estimates of the debitage index (Table 28).

7.2.3.C. Reduction Stages

Turning to lithic reduction strategies, all assemblages indicate the predominance of chipped-stone tool manufacture and maintenance as opposed core reduction (decorti­ cation of a core to produce flakes). Interassemblage differences in platform lipping and platform preparation are noted, and the biface index also differs among the samples.

Other lithic indicators of reduction stage that differ among the assemblages are early- and late-stage debitage, platform faceting, and platform cortex.

Based on the 1994 sample alone, platform lipping is comparable for the two earlier periods at 88% to 84%, but it is significantly lower for Late Woodland contexts at 55% (Table 28). This suggests that tool manufacture was not as intensive during the

Late Woodland compared to the earlier periods.

Retooling and/or core preparation was more intense during the Late Woodland compared to the other periods based on the frequency of platform preparation, which is

29% for the Late W oodland but ranges from 17% to 19% for the two earlier assemblages (Table 28).

218 The ratio of modified lithics to debitage is significantly different among the occupations (Table 28). The ratio is dramatically lower for Late Woodland deposits

(1:19), followed by Early Woodland (1:81) and Terminal Archaic (1:246). When one looks at the 1984 and 1994 combined samples, the differences among samples decrease considerably, with Late Woodland retaining the lowest ratio at 1:19, followed by

Terminal Archaic (1:28) and Early Woodland (1:60) deposits. The ratios for the Late

Woodland and the Terminal Archaic combined samples are on the order for two open-air sites (1:22 and 1:11) reported in Cowan et al. (1981). None of the ratios for 1994 or for the combined samples is similar to the ratios for two other shelters in the study area: Haystack (1:122) and Cloudsplitter (1:129). The Early Woodland 1994 and combined ratios are midway between the ratios for the open-air and shelter sites reported in Cowan et al. (1981).

Comparable values of the biface index, which indicates the prevalence of biface manufacture, suggest that there was little difference in biface manufacture during the

Terminal Archaic, Early Woodland, and Late Woodland periods. The values range from

22% to 28% for the low estimates and from 44% to 56% for the high estimates (Table

28). Considering the combined 1994 and 1984 samples, however, the upper biface index estimates of the two earlier periods (17% to 21%) are considerably lower than that of the Late Woodland (56%). This suggests that biface manufacture was more intensive during the Late Woodland occupation of Rock Bridge Shelter, which contradicts the findings based on platform retouch and the ratio of modified lithics to debitage.

Overall, the proportions of early-stage and late-stage debitage in the 1994 assemblages and the combined 1984 and 1994 assemblages indicate only a slight difference, with early-stage debitage slightly more prevalent in the Terminal Archaic and Early Woodland assemblages (18% and 20%, respectively) compared to the Late

Woodland at 12% (Table 28). For local cherts only, the proportions of early-stage and

219 late-stage debitage for each period follow the same pattern. For exotic cherts, the Early

Woodland assemblage has a higher percentage of early-stage debitage (Table 28).

As indicated in Table 28, differences in flake fragment types for the 1994 sample suggest that the proportions of initial reduction and shaped-tool manufacture debitage are similar for the occupations. Similarly, the frequencies of platform faceting in the assemblages indicate that biface production occurred with similar intensity during the three occupations (Table 28).

There seem to be some differences in the nature of reduction and conservation among the assemblages based on Magne's (1989) scheme (Table 28). There might be a higher incidence of conservation and tool maintenance during the Late Woodland, based on the percentages of late-stage debitage and debitageitool ratios. These data agree with the statistically high percentage of platform retouch in the Late Woodland assemblage, which suggests that retooling was more common during this period, and the low Late Woodland ratio of modified lithics to debitage.

7.2.3.d. Reduction Techniques

Platform lipping and bulb fissures suggest soft-hammer and hard-hammer percussion techniques, respectively. As indicated above, the differences in platform lipping in the assemblages suggest that soft-hammer percussion was less prevalent in the Late Woodland period compared to the other occupations (Table 28). Likewise, the incidence of bulb fissures, with the highest proportion (20%) associated with the Late

Woodland sample (Table 28), suggests that hard-hammer percussion was more common during that occupation.

7.2.3.e. LIthic-Related Activities

Due to small tool sample sizes and differential preservation conditions at the two shelters, it is difficult to identify all activities that might have occurred during the occupations of Cold Oak and Rock Bridge shelters, but the following observations are

220 made (using the combined 1994 and 1984 lithic samples from Cold Oak). There is little indication of variation in lithic-related activities at the shelters over time.

The Terminal Archaic occupation yielded 26 chipped-stone tools, seven ground- stone tools, and almost 1.1 kg of fire-altered rock. The identifiable chipped-stone artifacts include a core, projectile points/knives, a flake graver, drills, and utilized flakes. One ground-stone tool may be an atlatl weight or gorget fragment, one is a celt, and one is a nutting stone. These artifacts suggest that the following activities or tasks took place at Cold Oak Shelter during the Terminal Archaic occupation: providing shelter, food processing and food storage, fabrication and processing of organic materials, butchering and hide preparation, lithic maintenance, lithic manufacture, hunting, and nonlithic procurement (Table 27).

Deposits associated with the Early Woodland occupation yielded 16 chipped-stone tools, four ground-stone tools, and over 4.3 kg of fire-altered rock. The identifiable chipped-stone artifacts include projectile points/knives, cores, a flake knife, a flake scraper, a flake graver, drills, a hoe, and a hammerstone. A pin or plummet is the only identifiable ground-stone tool. These artifacts suggest that the following activities or tasks took place at Cold Oak Shelter during the Early Woodland occupation: providing shelter, food processing and food storage, fabrication and processing of organic materials, butchering and hide preparation, lithic maintenance, lithic manufacture, and hunting (Table 27).

Twenty-six chipped-stone tools, two ground-stone tools, and an unrecorded number of fire-altered rocks derived from Late Woodland deposits at Rock Bridge.

Points/knives, flake scrapers, cores, a flake knife, a spokeshave, and a perforator are among the chipped-stone artifacts. The only ground-stone tool type recovered is hammerstones. These artifacts suggest that the following activities or tasks took place at

Rock Bridge Shelter during the Late Woodland occupation: providing shelter, food

221 processing, fabrication and processing of organic materials, butchering and hide preparation, lithic maintenance, lithic manufacture, and hunting (Table 27).

7.2.4. GROUP SIZE

It is unlikely that group size varied drastically over the course of occupations at

Cold Oak and Rock Bridge shelters. Ceiling height, shelter width, and roof fall limited the usable living space of the shelters. Considering lithic densities, it is possible that group sizes were larger during the Early Woodland and Terminal Archaic occupations of Cold

Oak than during the Late Woodland occupation of Rock Bridge Shelter (Table 28). But the density measures for the Terminal Archaic and Late Woodland assemblages are likely distorted due to trampling.

7.2.5. DISCUSSION

Cursory examination of interassemblage differences in the percentages and ratios of lithic indicators of occupational variability suggest that there were some differences in shelter occupations at Cold Oak and Rock Bridge shelters over time. But when one considers the influence of extraneous processes on som e of the lithic indicators, temporal differences in some aspects of shelter occupation become less obvious or more difficult to interpret (Table 29).

Four lithic indictors of occupational duration show that Early Woodland occupations were of longer duration than occupations during the other periods. Terminal

Archaic and Late Woodland occupations are difficult to rank in terms of duration because two lithic indicators point to longer durations of occupation during the Terminal Archaic than the Late Woodland, but two others suggest the opposite.

222 Some differences are noted in the lithic production system. Thermal alteration of

cherts may have been more prevalent during the Early Woodland occupations. In terms of

raw material utilization, St. Louis use is pronounced during the Terminal Archaic

period, use of exotic cherts and Breathitt is more prevalent in the Early Woodland

period, and Haney and Paoli use is more prevalent in the Late Woodland period. Though

the intensity of lithic production may have been greater during the earlier periods,

lithic reduction activities remained essentially the same with shaped-tool manufacture

predominating. Possible differences in lithic reduction patterns are increased retooling,

maintenance, and conservation during the Late Woodland, which is an interesting finding

because the Haney and Paoli cherts relied on so heavily during this period are readily

available in the vicinity of Rock Bridge Shelter. While tool maintenance may have been

more common during the Late Woodland, this was not the only lithic reduction activity as cores and early-stage debitage are found at Rock Bridge Shelter. Soft-hammer percussion appears to have been more common during the earlier occupations and hard-

hammer percussion, during the Late Woodland. The range of activities that occurred at the shelters is similar over time, with evidence of sheltering, food processing and food storage, fabrication and processing of organic materials, butchering and hide preparation, lithic maintenance, lithic manufacture, hunting, and fishing.

As expected, it turns out that frequency of occupations and group size are difficult to assess with lithic data alone. While there are few lithic indicators of the frequency of shelter occupation during a given time period, the stratigraphy and lithic densities of the

Terminal Archaic and Early Woodland zones may represent more frequent occupation compared to the later periods. Group sizes may have been comparable over time or slightly larger in the earlier periods.

Based on the lithic analyses, one might conclude that rockshelter occupation differed somewhat over time, such that earlier occupations were somewhat more intense

223 in terms of duration and possibly frequency than later occupations. Terminal Archaic and

Early Woodland shelter use was probably comparable. However, the differences are not drastic or consistent. In terms of lithic activities, it is unlikely that people performed radically different tasks at the shelters over time. Perhaps there were changes in other aspects of lifestyles (e.g. subsistence, non-lithic material culture) over time. An examination of the other remains from each occupation may shed some light on this possibility as well as permit further evaluation of the sometimes inconsistent differences in the lithic sam ples.

7.3. OTHER ARTIFACTUAL REMAINS

The results of the diachronic lithic analysis are evaluated in light of other artifactual remains from Cold Oak and Rock Bridge. Using data from both the 1994 and

1984 excavations, the stratigraphy, features, organics, and artifact assemblages and densities of each period are compared with respect to occupational variability.

7.3.1. COLD OAK SHELTER

7.3.1.a. Stratigraphy

The Cold Oak Shelter stratigraphy suggests intense occupations in terms of frequency of use, as indicated previously. The stratified ash deposits punctuated by a vegetal mat indicate repeated use of the shelter over time. The ash bed below the plant layer (Zone III) yielded temporally early remains (Cogswell and Wade points, no ceramics) and early radiocarbon dates. The ash bed above the plant stratum (Zone II) produced temporally later remains (Adena and Cresap points, blades, ceramics) and later radiocarbcn dates (Gremillion 1995; O'Steen et al. 1991).

224 The strata and artifacts from Cold Oak support Funkhouser and Webb's

(1930:279) contention that the ash deposits separated by vegetal and/or sand-clay beds represent “different periods of occupancy" of the shelters in Lee County. Gremillion

(1995:73) cited additional evidence of the temporal distinction between Zones II and III.

Because of the obvious differences in color and texture between Terminal Archaic and Woodland deposits at Cold Oak (particularly on the less disturbed parts of the excavated area), it was suspected that a major change in deposition patterns took place between the latest Terminal Archaic occupation (ca. 800 B.C.) and the earliest Early Woodland activity (ca. 600 B.C.). The 1994 data support the earlier conclusion (O'Steen et al. 1991) that Cold Oak had witnessed repeated occupations beginning in the Terminal Archaic, a pattern that contrasts with earlier Archaic evidence from sites in the region.

Within the main cultural zones of Cold Oak, lenses of ash and plant materials with limited lateral extent were frequently encountered during the 1984 and 1994 excavations. "Many of the numerous discontinuous strata encountered at Cold Oak ... probably represent very short term phenomena such as hearth cleaning and the disposal of meal refuse" (O'Steen et al. 1991:77), which suggest shelter reuse or longer durations of occupation.

7,3.1.b. Features

Based on all artifactual remains as well as contextual evidence, Gremillion

(1995) proposed the following functional interpretations for the 44 Cold Oak Shelter features excavated during the 1994 season. Some of the designations differ from those based on the lithics alone. During the 1984 excavation seaso n at Cold Oak, O'Steen et al.

(1991) identified 30 features, 26 of which they described in the field report.

One feature is associated with Middle Woodland occupation of the shelter.

Gremillion (1995) interpreted it as a hearth or hearth cleanings.

Gremillion (1995) and O'Steen et al. (1991) identified a total of 40 Early

Woodland features at Cold Oak: 23 from the 1994 season, 14 from the 1984 season, and three encountered during both seasons. Functional designations for the 26 Early

Woodland features excavated in 1994 are nine pits, two of which may be storage pits,

225 one "ash bed" comparable to features described by Funkhouser and Webb (1929), four pits or post molds, one post mold, and eleven indeterminate or postoccupation features

(Gremillion 1995). Sixteen features originated in Zone II during the 1984 field season and O'Steen et al. (1991) assigned them Early Woodland ages. Gremillion (1995) encountered two of these, a post mold and a pit, during 1994 excavations and included them in that feature count as well. The remaining 14 features include two hearths, one post mold, one grass-lined pit, one grass-lined pit/post mold, one charcoal concentration, one pit and intrusive post combination, one pit/post mold, two storage pits, two ash-filled pits, one earth oven/pit, and one unspecified feature (O'Steen et al.

1991). There are eight feature functional types in the Early Woodland deposits.

A total of 26 Terminal Archaic features representing five functional types were encountered In the excavation trenches at Cold Oak Shelter during both excavation seasons; one of these, a post mold, was found during both field seasons and one. Feature

8, was assigned a Terminal Archaic age by O'Steen et al. (1991) only. The 1994 excavations revealed 16 features associated with Terminal Archaic occupations at Cold

Oak. Ten of these are pits, two are post molds (one of which workers uncovered in

1984), and two resemble casually constructed refuse pits or natural depressions filled with midden. One feature is a midden-filled natural depression and one feature is of unknown function (Gremillion 1995). Terminal Archaic features excavated in 1984 number 10 and include three post molds (one of which is part of the 1994 feature count), one post mold or pit, one fire-cracked rock-filled pit, one hearth or storage pit

(Feature 8), three hearths, and one hearth/earth oven (O'Steen et al. 1991).

Although features are more abundant in Early Woodland deposits, Gremillion

(1995) indicated this may be a function of the relatively thicker sequence of these deposits compared to Terminal Archaic strata. The character of the features does differ,

226 with storage pits confined primarily to the Early Woodland zone, and the diversity of functional types is slightly greater in the Early Woodland deposits.

Ison (1988) cited the post mold features as one line of evidence that people used

Cold Oak Shelter as a year-round base camp. O'Steen et al. (1991:110) implied a similar view: "postmolds indicate the presence of structures in the shelter, an apparently common situation in rockshelters since posts have been found at other Lee

County rockshelter sites." Post mold features occurred in both Early Woodland and

Terminal Archaic deposits at Cold Oak.

7.3.I.e. Organic Remains

A variety of plant remains occur in the paleoethnobotanical assemblages from the

1984 and 1994 field seasons. Evidence of the basic components of a gardening strategy based on indigenous cultigens derived from Terminal Archaic deposits and intensified during subsequent Early Woodland occupations. Mast exploitation seems to have changed over time at Cold Oak, as hickory replaced acorn as the primary nut resource (O'Steen et ai. 1991). Plants associated with disturbance as well as seed plant remains increased in abundance in Woodland contexts. Seasons of occupations based on plant remains are not given.

There is little difference in wood charcoal density in Terminal Archaic and

Woodland deposits. If people used the shelter during the same seasons during these occupations, occupation is unlikely to have intensified in term s of group size or duration. If, on the other hand. Archaic occupations predominated in colder months and

Woodland occupations in warmer seasons. Woodland occupations could have been more intense, as indicated by some nonbotanical evidence, without a corresponding increase in wood charcoal (Gremillion 1995).

Over one dozen animal genera/species occur in the Cold Oak faunal collection, including invertebrates, amphibians, reptiles, birds, and mammals. Fish remains are

227 conspicuously lacking in the collection, and one cannot attribute the absence to inadequate sampling or poor preservation conditions. Edible meat derived primarily from large mammals, especially deer and bear, during both Terminal Archaic and Early

Woodland occupations. Species diversity in somewhat higher in the Early Woodland deposits, and "smaller mammals may have played a somewhat more important role in the

Archaic than Woodland diet" (O'Steen et al. 1991:109). Overall, however, "comparison of Terminal Archaic and Woodland vertebrate remains from both field seasons does not reveal any changes in hunting patterns. Throughout the occupation of the site, hunting

(or at least consumption and processing of animal carcasses) does not seem to have been a major activity" (Gremillion 1995:73). Ison (1988) contended that the faunal remains support a year-round base camp interpretation for the Cold Oak occupations.

7.3.1.d. Artifact Assemblages and Densities

O'Steen et al. (1991) did not report artifact densities for the 1984 excavations, but one may compare the sample sizes of Terminal Archaic (Zones III, IV, and IV) and

Early Woodland (Zone II) occupations. Excluding weighed artifacts like fire-altered rock, 515 of the 1444 artifacts derive from Terminal Archaic deposits, representing

36% of the 1984 collection. Early Woodland contexts yielded 522 artifacts, accounting for an additional 36% of the collection (O'Steen et al. 1991). Given that the Zone II

(Early Woodland) deposits in the 1984 trench profile are considerably thicker than the

Zone III (Terminal Archaic) deposits, and therefore probably represent greater volumes of sediment, the Terminal Archaic artifact density may be greater than the Early

Woodland density.

7.3.1.e. Discussion

The results of the 1984 and 1994 excavations converge and diverge on different points related to the prehistoric occupation of Cold Oak Shelter. Researchers reached similar conclusions with respect to stratigraphy, paleoethnobotanical remains, and

228 faunal assemblages. A wider range of feature types were uncovered during the 1984 season than in 1994, but for both excavation seasons Early Woodland features outnumbered Terminal Archaic features. Neither study addressed seasonality of shelter use based on plant or animal remains.

The Terminal Archaic and Early Woodland occupations at Cold Oak do not differ significantly in terms of mineral content of the deposits, amounts of wood charcoal, and the types of faunal remains. The proportions of nuts and seeds did change over these occupations and indicate increased reliance on the latter during the Early Woodland period. Using feature number and diversity alone, one might conclude that the Early

Woodland occupation of Cold Oak Shelter was somewhat more intense than the Terminal

Archaic.

With respect to shelter function, Gremillion (1995:73) concluded, based on organic remains and lithics, that "Cold Oak functioned as a b ase camp used primarily for plant procurement accompanied by limited hunting." The plant exploitation strategies involved use of wild resources such as nuts as well as cultivation of indigenous crops.

The relative reliance on these two resources shifted somewhat over time, weighted toward the former during the Terminal Archaic and the latter during the Early Woodland.

Gremillion (1995) concluded that the Early Woodland occupation of Cold Oak represents increased occupational intensity in terms of frequency and duration of use.

Cold Oak and other shelters in the area

were used more frequently and perhaps for longer durations than they had been during the preceding Archaic period. Differences in site use based on the frequency and type of feature excavation also seem indicated between the Terminal Archaic and Early Woodland occupations of Cold Oak. These shifts in residential patterns and seasonal mobility are correlated with the integration of domesticated plants into the diet (Gremillion 1995:30-31).

O'Steen et al. (1991) contended that both Terminal Archaic and Early Woodland occupations were relatively intense at Cold Oak Shelter.

229 The variety of features, artifact types, and structural remains suggest that Terminal Archaic and Early Woodland occupation of Cold Oak was fairly intensive and probably year-round. This is also supported by the floral and faunal remains from the site that provide evidence of year-round hunting, collecting and gardening activities during the period between approximately 950 B.C. and 250 B.C. The shelter was probably inhabited by a small group of people, e.g., an extended family, at any given time (O'Steen et al. 1991:110).

So, depending on the significance attributed to the features at Cold Oak, the lithic analysis might lend support to either the interpretation stated by Gremillion (1995) or the conclusion of O'Steen et al. (1991). Concerning duration of occupations, the former contention seems plausible, but the latter seems appropriate considering frequency of occupations.

7.3.2. ROCK BRIDGE SHELTER

7.3.2.a. Stratigraphy

In contrast to Cold Oak, the strata at Rock Bridge Shelter suggest relatively less intense occupation in terms of the frequency of use. There was, essentially, only one thin cultural deposit at Rock Bridge; the differences in the upper and lower levels of the midden are probably due to trampling rather than multiple occupations over time. The presence of a compact, sterile substratum under the midden suggests one period of occupation (Nielsen 1991b).

7.2.3.b. Features

Upon examination of artifactual remains and stratigraphie characteristics,

Gremillion (1993) interpreted many of the sedim ent anom alies at Rock Bridge as redeposited midden that settled in low portions of the shelter: Features 4, 8, 9, 10, 11,

12, 13, 14, 15, 16 . A number of these features proved to be poorly bounded, darkly stained sediment with or without cultural remains. Feature 1 is a deposit of pack rat material.

230 Features 3, 5 and 6 were interpreted as hearths, disturbed hearths, or redeposited hearth cleanings. While not excavated. Feature 2 superficially resembled a charcoal concentration. Only two features had distinct lower boundaries; Feature 17 is a shallow basin feature and Feature 18 is a cylindrical pit. Workers recovered few artifacts from these features, however.

There were no storage pits or post mold features at Rock Bridge Shelter.

Gremillion (1993: 151) concluded that

the absence of storage pits or excavated hearths is hardly surprising given the shallow depth of sediments over most of the site and their unconsolidated, sandy character. Caching of plant foods in such a damp environment would in any case have been an inefficient means of storage, especially given the local availability of more suitable alternatives in the form of dry shelters. These findings agree well with interpretations that Rock Bridge functioned as a short term extractive camp.

7.3.2.0. Organic Remains

Conclusions about shelter use based on organic remains are problematic because the damp deposits probably adversely affected the preservation of faunal and floral remains at Rock Bridge. Taking this into account, Gremillion (1993) concluded that, compared to other early Late Woodland sites in the Red River area, occupations at Rock

Bridge were not intense.

Animal species represented in the Rock Bridge collection include box turtle, snapping turtle, painted turtle, bear, white-tailed deer, and turkey. There is little evidence that prehistoric inhabitants used aquatic resources, a deficiency that may reflect prehistoric resource use or poor preservation. Therefore, occupants may have used the utilized flakes recovered from the shelter for activities other than fishing

(Table 27). One can determine little about seasonality of shelter occupation based on the fauna alone. With a low diversity of animal types, relatively fragmentary remains, and a small sample, the faunal collection "is consistent with the interpretation of the site as a hunting/collecting encampment occupied for a relatively short period of time"

231 (Gremillion 1993:134). The faunal assemblage "reflects the subsistence activities of small foraging parties who may have dispersed into the uplands for brief periods to hunt" (Gremillion 1993:139-140).

The botanical collection contains carbonized nut and seed remains as well as wood charcoal. Small quantities of a variety of nuts suggest opportunistic exploitation, and prehistoric inhabitants likely used only a portion of the seeds as food (Gremillion

1993). The botanical collection indicates fall and summer use of the shelter.

7.3.2.d. Artifact Densities / Activity Areas

Lithics concentrated in three parts of the main excavation block at Rock Bridge resemble activity areas, but it is possible that creep rather than human action is responsible to some degree for the concentrations. However, statistics confirm a concentration of bone, lithic, pottery, and charcoal remains in the northern part of the main excavation block (Gremillion 1993). Some of these materials may be less susceptible to displacement by creep due to differential conductivity.

The density patterning does clearly indicate a northward trend in general evidence of human activity. The greatest concentration of cultural material occurs in a reasonably comfortable area where the ceiling is high enough for passage, the floor is relatively wide, and large sandstone boulders are available for seating. This would be a logical focus for a variety of activities, including food preparation and consumption, tool maintenance and manufacture, and social interaction (Gremillion 1993:37).

In addition, “the fact that there is relatively good correspondence between surface and subsurface archaeological assemblages indicates either that surface and subsurface deposits represent the same episode of occupation, or that activity was centered on the same parts of the site each time it was used" (Gremillion 1993:43).

7.3.2.e. Discussion

There is good correspondance between lithic and nonlithic evidence of occupational intensity at Rock Bridge. When one compares the stratigraphie, feature, and organic remains from the Late Woodland occupation at Rock Bridge to Cold Oak

232 occupations, the Late Woodland occupation seems relatively less intense in terms of duration and frequency of use; this supports the conclusions drawn from the lithic analysis. The lithic artifacts support the conclusion that during the Late Woodland period, Rock Bridge served as a hunting and/or collecting camp. And given the damp nature of the sediments at Rock Bridge, one would not expect that food storage would have been common at the shelter (Gremillion 1993).

Plant remains from Rock Bridge suggest most strongly that people used the shelter during the fall and/or summer, indicating relatively less intense duration of occupation. Compared to other Late Woodland settlements in the area, "the relatively low density of plant remains and other cultural materials indicates a low intensity of site occupation" (Gremillion 1993:146). While this pattern may reflect differential preservation, it might also reflect shelter function; "a seasonal foray carried out for a specific task such as obtaining chert, hunting, or plant collecting would account for the assemblage of cultural materials found at Rock Bridge" (Gremillion 1993:147).

Gremillion (1993) proposed that perhaps dry and wet shelters served different functions for the Late Woodland period occupants of the area.

7.4. COMPARISON WITH SELECT SHELTERS

This section considers whether or not the results obtained from lithic analyses at

Cold Oak and Rock Bridge apply to other shelters in the study area. While it is recognized that such comparisons must be viewed with caution due to differences in recovery and analytical methods and problems with temporal control, shelters are compared in terms of lithic assemblages, other artifactual remains, features, and stratigraphy. Funkhouser and Webb (1930:301) posited that the rockshelters in the area of Lee, Powell, and

Wolfe counties have "a similar cultural history." Do Cold Oak and Rock Bridge shelters

233 share a common culture history with other shelters in the study area? Table 30

presents a summary of stratigraphie, artifactual, and feature information for 40

shelters in the study area (and one in nearby Elliott County).

The first dimensions of occupational variability considered here are duration and

frequency. As indicated previously, there are some indications that the earlier

occupations at Cold Oak were more frequent and perhaps of longer duration than later

occupations there and at Rock Bridge Shelter. Archaeological evidence from several other

shelters suggests that Terminal Archaic and Early Woodland occupations were frequent

and/or of relatively long duration. Red-Eye Hollow (LEI) contained stratified ash

deposits separated by vegetal beds and a variety of features. Based on illustrations in

Funkhouser and Webb's (1929) report, diagnostic points are Terminal Archaic

(Terminal Archaic Barbed Cluster) and Early Woodland (Gary, Adena, Robbins) types.

Little Ash Cave (LE2) yielded stratified deposits and several types of features in

association with Late Archaic (Brewerton and Matanzas) and Early Woodland (Kramer,

Little Bear Creek, Gary, Adena, Robbins) bifaces (Funkhouser and Webb 1929).

Funkhouser and Webb (1929) noted that artifacts were more dense at these two shelters compared to other shelters they investigated in Lee County. At Steven DeHart Shelter

(POI), relatively long-term and/or frequent use is indicated by stratified ash, plant, and sand-clay deposits, a variety of features, and abundant artifactual remains; Table

Rock Cluster and Cresap points point to Late Archaic and Early Woodland occupations

(Funkhouser and Webb 1930). Other shelters excavated by Funkhouser and Webb

(1929, 1930) produced stratigraphie, feature, and artifactual indicators of relatively intense use, but diagnostic bifaces show these shelters were used during periods in addition to the Terminal Archaic and Early Woodland.

Though occupations at Cloudsplitter (MF36) spanned a number of periods, a cluster of Early Woodland storage pits and complexly stratified deposits of ash, vegetal

234 material, and sediment suggest long-term and/or frequent use of ttie sfielter (Cowan et al. 1981). Cowan et al. (1981) noted that Early Woodland deposits are considerably ashler and contain significantly more features than earlier strata. Post-mold patterns indicate the presence of two windbreaks. Burned material is pervasive throughout the extensive ash deposits. Despite this evidence, Cowan et al. (1981) concluded that during the Early Woodland, Cloudsplitter functioned as a temporary extractive camp for small groups, possibly nuclear families. Occupations occurred primarily during the fall and could have been relatively longer in duration but were still seasonal. But relative to earlier and later occupations. Early Woodland use of the shelter was more intense.

Looking outside the study area, the Conley-Greene Rockshelter (EL4) in Elliott

County produced Adena points diagnostic of the Early Woodland period (Railey 1991a).

Evidence of long-term and frequent use of the shelter includes complexly stratified wood-ash deposits up to 35 cm thick, a variety of tool types, and high lithic densities at

5,534 flakes/m^.

Although Big Ash Cave (LE3) had lower densities of artifacts than Red-Eye

Hollow and Little Ash Cave, the complexly stratified ash, vegetal, and clay deposits and the variety of features in association with Late Archaic and Early Woodland (Cresap) points may indicate at least frequent use of the shelter during these periods (Funkhouser and Webb 1929). Stratified ash, vegetal, and sand-clay beds that produced Terminal

Archaic Barbed Cluster and Early Woodland Stemmed Cluster points are reported for

Worth Creech Shelter (W02) by Funkhouser and Webb (1930); these data suggest frequent use of the shelter.

There is evidence from Late Archaic and Early Woodland occupations at other shelters that points to infrequent occupations or short duration of occupations. For instance, thin deposits, low lithic density, and low tool diversity at Ratliff Shelter

235 (LE 149) suggest that Late Archaic (and possibly Woodland) occupations were of short duration (Kluth 1992).

Published information about Late Woodland components at shelters in the study area is less abundant compared to Terminal Archaic and Early Woodland occupations.

Excavations conducted by Cowan (1979) at Haystack Rockshelter B (P047B) yielded over 43 kg of materials from 232 kg of deposits. Lithic remains from Haystack include two triangular points, one flake scraper, one biface fragment, one ground-stone fragment, 15 utilized flakes, and 1783 nonutilized flakes. Cowan (1979) based his reconstruction of shelter occupation largely on the organic remains. He concluded that, like Rogers Shelter (P026 and P027), Haystack was probably "occupied by a nuclear family, or extremely small extended family group," during mid to late summer, through the fall, and possibly into the winter. He could not rule out year-round occupation. This differs from Late Woodland use of Rock Bridge Shelter.

Turning now from duration and frequency to the range of activities, it was demonstrated that diachronic differences in the lithic production system at Cold Oak and

Rock Bridge involved raw material utilization and, in the Late Woodland period, perhaps increased tool maintenance and conservation. The types of lithic-related activities, however, differed little over time as shaped-tool manufacture predominated. Similar patterns are documented at other shelters in the study area.

At Ratliff Shelter (LEI49), Kluth (1992) documented reliance on locally available Boyle, Haney, and St. Louis cherts during the Late Archaic (and possibly

Woodland) occupations, with smaller amounts of Paoli. This is similar to the nature of chert use during the Terminal Archaic period at Cold Oak Shelter. At the Early Woodland

Conley-Greene Rockshelter (EL4), the majority of lithic artifacts are made of St. Louis,

Haney, and Paoli cherts (Railey 1991a); the incidence of St. Louis and Haney is similar to Cold Oak.

236 At Cold Oak Shelter, exotic cherts are slightly more common in the Early

Woodland and Terminal Archaic strata, and early-stage debitage of exotic cherts is more abundant in the Early Woodland assem blage. Similarly, about 25% of the Archaic lithic assemblage and 16% of the Early Woodland assemblage from Cloudsplitter is made of cherts originating outside the Red River Gorge area; nonlocal cores are only associated with Early Woodland occupations (Cowan et al. 1981). However, the Cloudsplitter percentages are considerably higher than the Cold Oak percentages.

Cloudsplitter differs from the early occupations at Cold Oak, however, in terms of lithic reduction activities. Cowan et al. (1981:73) contended that "curation of already existing tools rather than production of new ones was the focus of lithic activities at

Cloudsplitter" during the Early Woodland. Another Early Woodland shelter, Conley-

Greene, is more similar to Cold Oak (and Rock Bridge) shelter:

it is obvious that chipped stone tool making and refurbishing was an important activity at the site. ... The wide variety of waste by-products ... indicates that all stages of the chipped stone reduction trajectory were carried out at the site. The vast majority of flakes are, however, the result of primary or secondary reduction, the by-products of bifacial manufacture and tool maintenance. Overall, the flake assemblage suggests a strategy involving transport to the site of partially finished bifaces, finished tools, and some initially reduced chunks of raw material, with most flint knapping activities at the site involving tool finishing and maintenance (Railey 1991a:82).

One might conclude that, based on the excavated shelters considered here, there seems to be some truth to Funkhouser and Webb's (1930) contention that many of the shelters in the study area had similar cultural histories. For the most part, shelter occupations of the Late and Terminal Archaic periods were relatively intense in terms of duration and frequency, and this pattern continued or increased during the Early

Woodland period. However, evidence that Late Woodland occupations at Rock Bridge may have been relatively less intense than earlier occupations at Cold Oak differs from research at Late Woodland shelters in Powell County. Lithic reduction activities, on the other hand, probably did not differ drastically.

237 On the other hand, this picture of shelter use may relate to the attention paid to shelters with abundant remains and stratified deposits. These shelters are the ones that researchers have excavated and studied more intently, resulting in the conclusions about increased occupational intensity during certain periods of time. Archaeologists have not studied as extensively shelters with less obvious indications of occupation.

238 CHAPTER 8

SUMMARY AND CONCLUSIONS

The Cumberland Plateau of eastern Kentucky is characterized as an area of rugged terrain with high relief, steep slopes, narrow stream valleys, and poorly developed flood plains. A prominent feature of the landscape is rockshelters, or recesses carved into sandstone outcrops along valley walls. Because of the protection they afford, the rockshelters have been used by humans for thousands of years. For this reason, and because preservation at many shelters is exceptional, eastern Kentucky's rockshelters represent an invaluable resource in the study of Eastern Woodlands prehistory.

Archaeological investigations at the shelters began in the 1920s and 1930s with the work of Funkhouser and Webb at shelters in Lee, Wolfe, Powell, and Menifee counties. The wealth of well-preserved, uncarbonized organic remains recovered from the shelters had a significant impact on our understanding of prehistoric subsistence and technology, and the materials helped to chart the direction of future archaeological research in the area. Early studies focused on questions of prehistoric plant use, subsistence, and the development of plant domestication and agriculture. Subsequent studies sought to document prehistoric occupations at hundreds of shelters and to maximize retrieval of subsistence-related data. Archaeologists formulated hypotheses about prehistoric rockshelter use phrased in relation to subsistence economies, first positing that shelters were used as short-term encampments for hunting parties or as places to store crops grown on the flood plains. Subsequent research led to the

239 development of an alternate hypothesis, that indigenous groups lived in the shelters while cultivating crops on the surrounding hill slopes.

Compared to the organic remains from eastern Kentucky's rockshelters, lithic materials have received relatively less attention. In most studies, lithic analyses have focused on chronology building with diagnostic projectile points and functional classifications of tools in order to describe activities that might have occurred at shelters. Although a number of shelter lithic assemblages are dominated by chipped- stone lithic debitage, these materials have not been the focus of intensive, hypothesis- driven research for a number of reasons. And while organic remains and lithic tools have contributed significantly to our understanding of prehistoric rockshelter use, they cannot tell the entire story. For these reasons, this dissertation sought to demonstrate how lithic assemblages composed mostly of chipped-stone debitage can be used to test hypotheses about prehistoric rockshelter use.

Lithic assemblages from two rockshelters were used in the study. Cold Oak

Shelter is located in the Kentucky River drainage of Lee County. Cold Oak Shelter is a dry shelter that produced evidence of Archaic and Woodland period occupations. Although the site has been impacted considerably by illegal digging, O'Steen et al. (1991) and

Gremillion (1995) demonstrated that significant information can still be derived from the cultural deposits. Sediments are at least 90 cm thick at Cold Oak and are divided into five stratigraphie zones. Zone I is a loose, sandy loam and ash layer varying in thickness from 10 to 65 cm and deriving from historic occupations at illegal excavations at the shelter. Zone II is a 20-cm thick ashy loam associated with Middle and Early Woodland occupations at Cold Oak; it is divided into three subzones. Averaging 14 cm thick. Zone III is a compact sandy loam with ash lenses that was deposited during the Terminal Archaic period; it is divided into two subzones. Terminal Archaic Zone IV ranges in thickness from 10 to 15 cm and is characterized as loose, sand to sandy loam with decaying

240 sandstone. Zone V is up to 55 cm thick and is a deposit of sand and decaying sandstone.

Although it contained few artifacts, Zone V is associated with Terminal Archaic

occupations. Seventy features identified at Cold Oak Shelter include post molds, hearths,

and a variety of pits.

Excavations at Cold Oak Shelter were conducted in 1984 and 1994. The lithic

assemblages recovered during 1994 excavations include 2240 specimens. Four broad

categories of lithics, in increasing order of abundance, are ground-stone artifacts,

miscellaneous rocks and minerals, thermal debris, and chipped-stone artifacts. The

density of lithics for those portions of the shelter investigated in 1994 is 0.75/liter.

Lithic artifacts are fairly evenly distributed laterally in the portions of the shelter

excavated in 1994, but vertically lithics are most abundant in Zones I and II. For the

purposes of this study, the Cold Oak lithics were divided into three assemblages: a Middle

Woodland assemblage deriving from Zone lia, an Early Woodland assemblage associated

with Zones lib and lie, and a Terminal Archaic assemblage recovered from Zones III, IV,

and V. Unfortunately, the Middle Woodland assemblage was too small to use in the lithic

analyses.

Rock Bridge Shelter is located in the Red River Gorge area of Wolfe County.

Unlike Cold Oak, Rock Bridge is a wet shelter that was likely occupied during one time period only, the Late Woodland. Access to the shelter is severely limited by dense vegetation and sheer cliffs, and little evidence of disturbance is apparent at Rock Bridge.

Archaeological resources at the shelter were first assessed in 1989, and excavations were undertaken in 1992. Deposits across the shelter are up to 20 cm thick and are divided into three zones. The upper Zone I is a loose, ashy, silty sand from one to four cm thick. Zone II is a rockier, more compact, sandy silt from five to ten cm thick. The lowermost Zone III is five to six cm thick and is composed of compact sand with decaying

24 1 sandstone. It is underlain by a sandy clay deposit. Eighteen features identified during

excavations include hearths and pits.

The Rock Bridge lithic assemblage is composed of 755 specimens for a density of

0.42/liter. There are few thermal debris, miscellaneous rocks and minerals, and

ground-stone artifacts in the assemblage. Most lithic specimens are chipped-stone

artifacts, especially debitage, composed of locally available cherts. Lowe Cluster points

are among the diagnostic artifacts recovered from Rock Bridge. Almost all of the lithic artifacts derived from Zones I and II and from the main excavation block.

Three propositions guided the current research. The contention that lithic assembiages from disturbed rockshelters can provide useful information about shelter

use was evaluated using Schiffer's (1987) behavioral model of formation processes and experimental studies of the effects of various cultural and noncultural processes on artifact distribution and condition. Trampling, bioturbation, gravity, and creep at Cold

Oak and Rock Bridge shelters were assessed using sediment properties and debitage distributions (Table 24). Trampling likely resulted in the formation of a loose upper zone at both shelters, and this zone could have facilitated the movement of artifacts.

Vertical displacement of small lithics by trampling is evidenced in the Late Woodland assemblage from Rock Bridge only. There is no evidence that bioturbation vertically displaced lithics at either shelter. It is likely that creep horizontally displaced Late

Woodland lithics at Rock Bridge, resulting in curvilinear concentrations bearing east- west across the main excavation block. Gravity and trampling apparently did not affect the lateral distribution of lithics at either shelter. Flake fragmentation in the Late

Woodland assemblage from Rock Bridge and the Terminal Archaic assemblage from Cold

Oak may be due, in part, to trampling.

Results of the formation process studies indicate that the Woodland and Archaic assemblages from Cold Oak and Rock Bridge probably represent relatively unmixed

242 assemblages that may be used to assess diachronic differences in shelter use. Although

artifact movement likely occurred at Rock Bridge, diagnostic artifacts and radiocarbon

dates indicate that the materials derived from occupations during one period, the Late

Woodland.

The second proposition that lithic assemblages, including tools and debitage, can

inform about activities that took place at rockshelters in the past was evaluated using

Ericson's (1984) model of lithic production systems, Collins' (1975) model of

chipped-stone lithic reduction, and experimental studies that attempt to link artifactual

remains with the behaviors that produced them. Analysis of the lithic production

systems at Cold Oak and Rock Bridge shelters focused on thermal alteration of chert,

patterns of raw material utilization, and tool manufacture and maintenance. The

debitage-dominated assemblages proved helpful in addressing these issues.

Heat treatment of cherts may have been more common during the Early Woodland

period, when heat was used to alter the appearance and knapping properties of Breathitt

chert. The densities of heat-treated lithics in the Terminal Archaic and Late Woodland

assemblages are comparable. Haney and Paoli cherts may have been heated to alter

knapping qualities during the Late Woodland.

Though Haney is present in all the assemblages, the use of Haney chert is

considerably higher in the Late Woodland period. Specimens of St. Louis chert are more

common in the earlier assemblages, especially the Terminal Archaic. Use of Breathitt

chert peaks in the Early Woodland, and Paoli chert use peaks in the Late Woodland. Cold

Oak differs from other Cogswell Phase sites in the low incidence of Paoli cherts in the

Terminal Archaic and Early Woodland assemblages. Use of exotic cherts is most common

during the Early Woodland, and the ratio of early- to late-stage exotic debitage suggests

that exotic cherts may have been stockpiled during this time.

243 There are few indications that lithic reduction activities differed dramatically over time at Cold Oak and Rock Bridge. The assemblages appear to have been formed as a result of primary and secondary reduction as opposed to core reduction. Biface production and shaped tool manufacture appear to have been the most common activities over time. However, there is some evidence that tool maintenance was more prevalent during the Late Woodland period.

These analyses suggest that people transported mostly local raw materials in an altered form, perhaps blanks or preforms, to Cold Oak and Rock Bridge and further modified them at the shelters through thermal alteration and reduction. There was some variation in lithic reduction activities over time at the shelters, in that thermal alteration and raw material use patterns suggest that Early Woodland occupations were more intense than the other occupations.

The third proposition contends that lithic assemblages can be used to assess diachronic differences in rockshelter use. While the current research is not the final word on the issue of shelter occupational variability and it is recognized that differences in assemblages do not necessarily mean that change in shelter use occurred, the study contributes methodologically and substantively to this aspect of eastern Kentucky prehistory. Occupational variability was evaluated using lithic and nonlithic indicators of duration of occupation, frequency of occupation, the range of activities that occurred during an occupation, and group size. Lithic density, tool diversity, heat treatment, exotic cherts, feature number and diversity, midden accumulation, and seasonal faunal and floral remains indicate duration of shelter occupations. Lithic densities, shelter stratigraphy, and caching vary with the frequency of shelter occupations. The range of lithic-related activities is indicated by both tool and nontool lithic artifacts. Lithic densities may be suggestive of group size.

244 Previous rockshelter research provided hypotheses about shelter use that were evaluated using the lithic assemblages and ancillary data from Cold Oak and Rock Bridge.

If shelters were used on a relatively sporadic basis by prehistoric groups during a certain time period (Carmean and Sharp 1995, Cowan 1976, Jefferies 1988b,

Struever and Vickery 1973, Webb and Baby 1957, Wyss and Wyss 1977), then one would expect to find low lithic and artifact densities, low tool diversity, little evidence of heat treatment, more exotic raw materials or only finished tools made of exotic materials, few and limited types of features, organic remains available during specific times of the year, few cached tools, and thin and temporally indistinct strata. On the other hand, if shelters were used on a relatively intense basis by prehistoric groups during a certain time period (Funkhouser and Webb 1930, Ison 1991, Knudsen et al.

1983, O'Steen et al. 1991, Railey 1991, Wyss and Wyss 1977), then one would expect to find high lithic and artifact densities, high tool diversity, abundant evidence of heat treatment, few exotic raw materials or lithic reduction debris made of exotic materials, many features of a variety of types, organic remains available during several seasons of the year, cached tools, and thicker, temporally distinct strata.

The current study found that Early Woodland occupations of Cold Oak Shelter may have been of slightly longer duration than Terminal Archaic occupations and longer than

Late Woodland occupations. Early Woodland lithic densities are higher, the tool diversity is higher, heat treatment is more prevalent, and exotic cherts, including early-stage debitage and a core, are slightly more abundant. However, density differences must be viewed carefully because of the damaging effects of trampling on Terminal Archaic and

Late Woodland assemblages. Regarding the non-lithic evidence. Early Woodland deposits are thicker, contain more features, and have a slightly higher diversity of features than the Terminal Archaic strata. However, the density of all artifacts is probably higher in the Terminal Archaic, and the abundance of wood charcoal is comparable in both deposits.

245 Post molds are found in both Terminal Archaic and Early Woodland deposits, but storage

pits are confined to the latter. It is concluded that, while there is an inkling that Early

Woodland occupations were slightly more intense than earlier Terminal Archaic

occupations, the safest conclusion is that the occupations were comparable in terms of

duration.

Obvious differences in occupational duration obtain when comparing the two

earlier assemblages with the Late Woodland assemblage. Terminal Archaic and Early

Woodland lithic assemblages from the complexly stratified deposits of Cold Oak Shelter

are more dense and have more exotic cherts with early-stage debitage. Middle Woodland

and Late Woodland deposits are thin, undifferentiated, and contain few features of limited

diversity: post molds and storage pits were not encountered. Faunal and floral remains

from Late Woodland contexts at Rock Bridge suggest short-term use of the shelter.

Compared to occupational duration, frequency of occupations at Cold Oak and Rock

Bridge was more difficult to a sse ss with the lithic assem blages. Stratigraphie distributions of lithics suggest that Terminal Archaic and Early Woodland shelter occupations were more frequent than the Late Woodland occupations. While the cached tools found at Rock Bridge indicate that the Late Woodland habitants at least intended to return to the shelter, the sediments lack the internal differentiation evident in the

Terminal Archaic and Early Woodland zones of Cold Oak Shelter.

Temporal variations in the lithic production system are not accompanied by drastic differences in lithic-related activities at Cold Oak and Rock Bridge shelters. Heat treatment of chert may have been more prevalent in the Early Woodland period. During this occupation, inhabitants may have used heat to alter the appearance and knapping properties of Breathitt and other cherts. Heat-treated cherts are also common in the

Late Woodland assemblages of Rock Bridge Shelter and indicate that inhabitants used heat to alter the knapping properties of Haney and Paoli cherts.

246 Although prehistoric groups relied heavily on locally available raw materials,

there are differences in chert use at Cold Oak and Rock Bridge shelters over time. Not

only do exotic cherts account for a slightly higher proportion of specim ens in the Early

Woodland sample, but a primary decortication flake of Muldraugh chert as well as other

early-stage debitage of exotic cherts derived from Early Woodland contexts. This suggests that people transported both worked and unworked exotic chert material to Cold

Oak during this period. There is no indication that Cogswell phase groups inhabiting Cold

Oak during the Terminal Archaic-Early Woodland made differential use of Paoli chert as some authors have suggested. In fact, Paoli is more abundant in the Middle Woodland and

Late Woodland assemblages. The use of St. Louis chert is prevalent during early occupations and decreases over time, while Breathitt use peaked during the Early

Woodland occupations.

For all periods of occupation, lithic manufacture and maintenance, as opposed to core reduction, were the predominant activities, even taking into account the damaging effects of trampling. The intensity of lithic production, as indicated by debitage densities, debitage indexes, and the incidence of platform lipping, was possibly greater in the earlier periods compared to the Late Woodland occupations. Differences in platform morphology, tooLdebitage ratios, and early- to late-stage debitage ratios suggest that primary and secondary reduction predominated during the earlier periods, but tool maintenance or retooling was more common during the Late Woodland period.

Platform morphology suggests that soft-hammer percussion may have been more common in the earlier periods compared to the Late Woodland, as hard-hammer percussion likely predominated during that occupation.

Lithic-related activities vary little among the occupations. During the three occupations, the lithics indicate that providing shelter, food preparation, lithic manufacture, fabrication and processing of organic materials, butchering and hide

247 preparation, lithic maintenance, and hunting occurred. Non-lithic procurement is evidenced during the Terminal Archaic only.

Due to spatial constraints, it is unlikely that group size varied considerably over the course of occupations at Cold Oak and Rock Bridge shelters. Based on research at other shelters, some researchers (Cowan et al. 1981; O'Steen et al. 1991; Railey

1991) have suggested that nuclear families or extended nuclear families used the shelters prehistorically. The present study could not confirm nor discount this scenario.

In sum, with respect to the question of diachronic differences in the variability of rockshelter occupations in the study area, the lithic assem blages from Cold Oak and

Rock Bridge indicate clearly that Terminal Archaic and Early Woodland use of the shelters was more intense than Late Woodland use in terms of duration and possibly frequency of use. While raw material utilization varied over time, lithic-related activities and group size probably did not differ significantly. Non-lithic remains from the shelters support this conclusion. Therefore, rockshelter occupations varied diachronically in at least two of the four dimensions of occupational variability.

The differences in shelter utilization during the Terminal Archaic and Early

Woodland, on the other hand, are more difficult to interpret. The duration and frequency indicators of occupational intensity are inconsistent or insignificant in many ways.

Differences in the non-lithic remains from these occupations are somewhat ambiguous as well. While it is tempting to conclude that Early Woodland occupations were more intense in terms of duration and frequency of use, the author chooses to err on the safe side and conclude that there was little difference in shelter use during these time periods.

The current research supports the interpretation of rockshelter use advanced by

Knudsen et. al. (1983) and O'Steen et. al. (1991), that there were few differences in shelter use over the Terminal Archaic and Early Woodland periods. Railey's (1991b)

248 observations, that Terminal Archaic shelter use in the eastern mountains was limited in

intensity as base camps were located in the lowlands and that population dispersal and

increased use of the shelters ensued during the Early Woodland period, are not supported

by the data from Cold Oak Shelter. Similarly, Funkhouser and Webb's (1929, 1930)

contention that pre-pottery (Archaic) occupations at some shelters were more intense

than later pottery occupations is also not upheld by the current study. Of course,

research at other shelters may yield different results. Less has been written about the

nature of Middle and Late Woodland occupational intensity, but the record at Rock Bridge

is not comparable to Haystack and Rogers shelters (Cowan 1979) in terms of shelter

use.

In terms of lithic indicators of rockshelter use, the absence of differences in

lithic reduction activities across the Terminal Archaic-Early Woodland boundary suggests that the nature of occupation over these times did not drastically change and the distinction between these two time periods is indistinct at Cold Oak Shelter. The utility of this traditional temporal distinction is limited with respect to lithic assemblages from the shelters. The distinction between Early and Middle Woodland, however, is difficult to interpret because of the small Middle Woodland sample size and because the Middle

Woodland assemblage may be particularly unrepresentative as the deposits may have been truncated by twentieth century disturbances.

This study illustrates the analytical potential of lithic collections recovered from rockshelter contexts. It also demonstrates the need to consider the effects of formation processes on artifact assemblages and the importance of statistical manipulations of data.

Replication of the procedures followed here should be relatively simple. Appropriate collections to which one may apply the approach should have the following requisites.

First, for collections recovered from multicomponent sites, the excavation techniques must allow for adequate documentation of artifact provenience and identification of

249 distinct strata. This was the case for the Cold Oak assemblages but may not be so for

collections recovered in the earlier part of this century or those acquired by collectors.

Second, field or lab techniques that allow for the recovery of at least a representative

sample of microlithics (less than 1/4 in) is recommended to allow for evaluation of

formation processes. The use of 1/16 in and 0.6 mm screening techniques at Cold Oak

and Rock Bridge contributed in this respect. Third, the scheme employed to classify

lithic artifacts must be relevant to experimental work and explanatory models of

formation processes and lithic reduction activities.

Future analysis of the lithic assemblages of Cold Oak and Rock Bridge shelters

should involve replicative experiments to serve as a baseline for evaluating chert

alteration, formation processes, and lithic production. Locally available cherts should be

thermally altered to determine the physical and mechanical changes that occur at various

temperatures. Knapping experiments using the local cherts would help to build

guidelines for identifying different reduction strategies and techniques in archaeological

assemblages. Trampling and cleaning studies using different substrates (especially

mixtures of wood ash) would help in assessing their effects on artifact assemblages and substratums.

A distributional study of the sources of various cherts would be useful for better

assessing the lithic procurement strategy of the Cold Oak and Rock Bridge inhabitants. It would be helpful to know more about where each type of chert is located and in what form

(outcrop, colluvium, alluvium) in Lee, Wolfe, and the surrounding counties.

Researchers should also attempt to document quarry sites and to relate them to lithic

reduction sites in the area.

Microwear analysis would allow one to identify with more confidence and specificity the functions of lithic artifacts. This, in turn, would allow for more accurate characterization of the activities that took place at the shelters prehistorically and the

250 spatial patterning of those activities (Yerkes 1989). Refitting studies are another

means that could be pursued in the future to assess the movement of artifacts by

formation processes and to evaluate the activities that took place at the shelters in the

past.

Examination of the talus slopes below the shelters, as was done by Cowan et al.

(1981) at Cloudsplitter, would be informative in terms of refuse discard. While it

might not be possible to control for time in such deposits, which would be problematic at

multicomponent sites like Cold Oak Shelter, the lithic and nonlithic materials recovered from talus contexts might complement nicely the information derived from studying these remains within the shelters. For example, expended cores, broken tools, and axial animal bones (if preserved) would contribute to our understanding of the nature of occupation at shelters.

A survey of other sites in the Big Sinking Creek and Red River areas is necessary to place Cold Oak and Rock Bridge into a larger cultural system to better understand the nature of shelter use with respect to other sites on the landscape. With more sites, one could better document the lithic procurement system, and one might use the composition of lithic assem blages to characterize sites as short-term or long-term foci of the overall settlement pattern. Researchers must consider how shelters fit into the settlement strategy in comparison to open sites. This is the only way to determine if shelter occupations represent “habitation sites” or "temporary locations" or if shelter occupants were "sedentary" or “mobile."

One factor that may influence the nature of lithic and non-lithic assemblages is the length of time over which the materials accumulate. In the present study, time was controlled to some extent by the cultural periods used to stratify the collections. Based on radiocarbon dates, the Terminal Archaic, Early Woodland, Middle Woodland, and Late

Woodland occupations at the shelters were of comparable length. It may be necessary to

251 estimate sedimentation rates at shelters in the future and to incorporate differences in sedimentation rates into analytical and explanatory models. This will be especially critical if researchers attempt to compare shelters and/or open sites in terms of artifact densities.

The issue of reasons for differences in shelter use over time was not addressed in this study but makes for interesting future research. Why did shelter use differ in certain ways? Factors that may explain differences include shelter function

(habitations, ritual activities, burials, storage), environmental differences like topography or shelter conditions (e.g., wet versus dry and shelter aspect), changes in other aspects of culture like subsistence, or the spatial extent of excavations at the shelters.

Another issue not pursued in this research, but one that may bear on interpretations of rockshelter use, relates to the scarcity of lithic tools compared to lithic debitage. Why are many shelter lithic assemblages dominated by chipped-stone debitage but lacking in chipped-stone and other tools? The answer may relate to seasonality of occupations, the nature of activities that occurred at the shelters, the geographic foci of tool use, group size, or the nature of the tool conservation practiced by the inhabitants.

Future research should also consider the validity of the lithic indicators used to assess occupational intensity. Lithic indicators such as density, diversity of tool types, thermal alteration, and exotic materials are untested assumptions that one must independently determine are related to occupational intensity. Work at single occupation sites might be useful in this respect. More research to identify indicators that distinguish duration and frequency of occupations is needed, as is more research on indicators of group size.

252 There appears to be relatively good correspondence between the results of the lithic analysis and the conclusions about shelter use drawn from other artifactual features and remains. This begs the question, why spend extra time on the intensive lithic analyses if one could arrive at the same conclusions using other evidence? The main reason archaeologists need intensive lithic analyses, moving beyond simple classifications, is because lithic collections will be the foundation of future studies that assess prehistoric settlement strategies in the study area. Unlike ceramics, people used lithics throughout the period of prehistoric habitation of the Cumberland Plateau. And unlike organics, lithics preserve in a variety of depositional contexts, including protected shelters and open-air sites on ridge tops and alluvial bottoms. Lithic assemblages deficit in finished tools can be used to address some issues of site use, especially duration of occupations and range of activities.

253 LIST OF REFERENCES

Ahler, S. R. 19 87 Middle-Late Woodland Occupation at the Hanson Site, Greenup County, Kentucky. In Current Archaeological Research in Kentucky Volume 1, edited by D. Pollack, pp. 44-77. Kentucky Heritage Council, Frankfort.

19 8 8 Excavations at the Hansen Site in Northeastern Kentucky. Archaeological Report 173, Program for Cultural Resource Assessment, University of Kentucky, Lexington.

Anderson, D. G. 1979 Prehistoric Selection for Thermal Alteration: Tests of a Model Employing Southeastern Archaeological Materials, Mid Continental Joumal of Archaeology 4(2):221-241.

Applegate, D. 1993 Lithic Artifacts, In Archaeological Investigations at the Rock Bridge Shelter (15Wo7S), Wolfe County, Kentucky, pp. 54-133. Written by Kristen J. Gremillion, Department of Anthropology, The Chio State University. Report submitted to Cecil Ison, Field Archaeologist, Daniel Boone National Forest, Stanton, Kentucky.

1995a Lithic Artifacts, In Archaeological and Paleoethnobotanical Investigations at the Cold Oak Shelter, Kentucky, pp. 89-217. Written by Kristen J. Gremillion, Department of Anthropology, The Chio State University. Report submitted to the National Geographic Society.

1995b The Research Potential of Surface Collected Lithic Assemblages: The Rhondle Lee Collection, Powell County, Kentucky. Paper presented at the Twelfth Annual Kentucky Heritage Council Archaeology Conference, February 25-26, Richmond, Kentucky.

1997 Inventory of Rockshelters in Lee, Wolfe, Powell, Menifee, Breathitt, Es till, Jackson, Magoffin, Morgan, and Owsley Counties, Kentucky. Unpublished manuscript in possession of the author.

Bar-Yosef, C. 19 93 Site Formation Processes from a Levantine Viewpoint, In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 13-32. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

254 Barton, C. M. and Clark, G. A. 1993 Cultural and Natural Formation Processes in Late Quaternary Cave and Rockshelter Sites of Western Europe and the Near East, In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 33-52. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

Baumler, M. F. and Downum, C. E. 198 9 Between Micro and Macro: A Study in the Interpretation of Small-Sized Lithic Debitage. In Experiments in Lithic Analysis, edited by D. S. Amick and R. P. Mauldin, pp. 101-116. B.A.R., Oxford.

Beardsley, R., Holder, P., Krieger, A., Meggers, B., Rinaldo, J. and Kutsche, P. 1956 Functional and Evolutionary Implications of Community Patterning. In Seminars in Archaeology: 1955 Memoir 11. Society for American Archaeology, Salt Lake City.

Biek, L. 19 70 Artifacts. In Science in Archaeology: A Survey of Progress and Research, edited by D. Brothwell and E. Higgs, pp. 567-570. Praeger, New York.

Binford, L. R. 1 9 79 Organization and Formation Processes: Looking at Curated Technologies. Journal of Anthropological Research 35:255-273.

19 80 Willow Smoke and Dog's Tails: Hunter-Gatherer Settlement Systems and Archaeological Site Formation. American Antiquity 45:1-17.

Black, D. F. B. 19 78 Geologic Map of the Zachariah Quadrangle, East-Central Kentucky. Map GO-1452. U.S. Geological Survey, Reston, VA.

Boisvert, R. A. 1981 Archaeological Survey of Robinson Forest. Memorandum to Drs. Thielges and Muller, Forestry Department, University of Kentucky from the Office of State Archaeology. Report on file at the Office of State Archaeology, Lexington.

Braidwood, R. J. 19 60 Archaeologists and What They Do. Franklin Watts, New York.

Brown, J. A. 19 77 Current Directions in Midwestern Archaeology. Annual Reviews of Anthropology 5:161-179.

1 9 84 Prehistoric Southern Ozark Marginality: A f^yth Exposed. Special Publication No. 6. Columbia, Missouri Archaeological Society.

255 Bush, D. R. 19 8 8 New Evidence of Paleoindian and Archaic Occupations in Clay and Perry Counties, Kentucky. In Paleoindian and Archaic Research in Kentucky, edited by C. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 47-66. Kentucky Heritage Council, Frankfort.

Butzer, K. 1982 Archaeology as Human Ecology. Cambridge University Press, Cambridge.

Caldwell, J. R. 19 62 Eastern North America. In Courses Toward Urban Life, edited by R. J. Braidwood and G. R. Willey, pp. 288-308. Aldine Publishing Company, Chicago.

Carmean, K. and Sharp, W. E. 19 95 Not Quite Newt Kash: Three Small Rockshelters in Laurel County. Paper presented at the 12th Annual Kentucky Heritage Council Meeting, Eastern Kentucky University, Richmond.

Chapman, J. 1981 Bacon Bend and Iddens Sites. University of Tennessee, Department of Anthropology Report of Investigations 31.

Cleland, C. E. 1976 The Focal-Diffuse Model: An Evolutionary Perspective Cn The Prehistoric Cultural Adaptations Cf The Eastern United States. Mid-Continental Journal of Archaeology 1:59-75.

Collins, M. B. 1975 Lithic technology as a means of processual inference. In Lithic Technology: Making and Using Stone Tools, edited by E. Swanson, pp. 15-34. Mouton, P aris.

Corbyn, R. C. 1975 Final Report Concerning Eight Select Archaeological Sites and Other Involved Cultural Resources in the Area of the Proposed Corps of Engineers' Red River Lake Project. Kentucky Heritage Commission, Frankfort.

Cotterill, R. S. 1954 The Southern Indians. University of Oklahoma P ress, Norman.

Cowan, C. W. 1974 Archaeological Resources in the Proposed Red River Lake Area. Museum of Anthropology, University of Kentucky, Lexington.

19 75 An Archaeological Survey and Assessment of the Proposed Red River Reservoir in Powell, Wolfe, and Menifee Counties, Kentucky. Museum of Anthropology, University of Kentucky, Lexington.

19 76 Test Excavations in the Proposed Red River Lake, Kentucky: 1974 Season. Museum of Anthropology, University of Kentucky, Lexington.

256 Cowan, C. W. 1979 Excavations at the Haystack Rockshelters, Powell County, Kentucky. Midcontinental Journal of Archaeology 4(1):3-35.

Cowan, C. W. and Wilson, F. T. 19 7 7 An Archaeological Survey of the Red River Gorge Area in Menifee, Powell, and Wolfe Counties. Kentucky Heritage Commission, Frankfort.

Cowan, C. W., Jackson, H. E., Moore, K., Nickelhoff, A., and Smart, T. L. 1981 The Cloudsplitter Rockshelter, Menifee County, Kentucky: A Preliminary Report. Southeastern Archaeological Conference Bulletin 24:60-76.

Dancey, W. S. 1992 Village Origins in Central Ohio: The Results and Implications of Recent Middle and Late Woodland Research. In Cultural Variability in Context: Woodland Settlements of the Mid-Ohio Valley, edited by M. F. Seeman, pp. 24-29. Mid-Continental Journal of Archaeology Special Paper No. 7. Kent State University Press, Kent, Ohio. de Paor, L. 196 7 Archaeology: An Illustrated Introduction. Penguin Books, Baltimore.

DeLorenze, G. N. 1979 A Reconnaissance and Evaluation of Archaeological Sites in Bell County, Kentucky. Kentucky Heritage Commission Archaeological Survey Reports 13. Kentucky Heritage Commission, Frankfort.

DeLorenze, G. N. and Weinland, M. 198 0 A Reconnaissance and Evaluation of Archaeological Sites in Knox County, Kentucky. Kentucky Heritage Commission Archaeological Sunrey Reports 13. Kentucky Heritage Commission, Frankfort.

Donahue, J. and Adovasio, J. M. 199 0 Evolution of Sandstone Rockshelters in Eastern North America: A Geoarchaeological Perspective, In Archaeological Geology of North America, edited by N. P. Lasca and J. Donahue, pp. 231-251. The Geological Society of America, Boulder, Colorado.

Dorwin, J. T., Henson, E., Meadows, L. and Warholic, D. T. 1970 Archaeological Investigation of the Deep Shelter, Cave Run Reservoir Area, Rowan County, Kentucky. Museum of Anthropology, University of Kentucky, Lexington.

Dragoo, D. W. 19 76 Some Aspects of Eastern North American Prehistory: A Review 1975. American Antiquity 41:3-27.

Dunnell, R. C. 1972 The Prehistory of Fishtrap, Kentucky. Publications in Anthropology 75. Yale University, New Haven.

257 Ericson, J. E. 1984 Towards the Analysis of Lithic Production Systems. In Prehistoric Quarries and Lithic Production, edited by J. E. Ericson and B. Purdy, pp. 1-9. Cambridge University Press, Cambridge.

Ettensohn, F. R. et al. 1984 Slade and Paragon Formations - New Stratigraphie Nomenclature for Mississippian Rocks along the Cumberland Escarpment in Kentucky. U.S. Geological Survey Bulletin 1605-B. Contributions to Stratigraphy. U.S. Geological Survey, Boulder, CO.

Evans, S. and Gould, P. 1982 Settlement Models in Archaeology. Journal of Anthropological Archaeology 7:275-304.

Farrand, W. R. 1985 Rockshelter and Cave Sediments, In Archaeological Sediments in Context, edited by J. K. Stein and W. R. Farrand, pp. 21-41. Peopling of the Americas Volume 1. Center for the Study of Early Man, Institute of Quaternary Studies, University of Maine at Orona, Orono, Maine.

1993 Discontinuity in the Stratigraphie Record: Snapshots from Franchthi Cave, In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 85-96. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

Fenneman, N. M. 1938 Physiography of Eastern United States. McGraw-Hill, New York.

Fiegel, K. H., McGraw, B. J. and Nixon, J. L. 1992 Archaeological Investigation of the Kay Shelter in Breathitt County, Kentucky. In Current Archaeological Research in Kentucky: Volume Two, edited by D. Pollack and A. G. Henderson, pp. 43-54. Kentucky Heritage Council, Frankfort.

Ford, J. A. and G. R. Willey 1941 An Interpretation of the Prehistory of the Eastern United States. American Anthropologist 43:325-363.

Ford. R. I. 1974 Northeastern Archaeology: Past and Future Directions. Annual Review of Anthropology 3:385-413

Ford, T. D. and Cullingford, C. H. D. 1 9 76 The Science of Spelology. Academic Press, London.

Fryman, F. B. 1 9 6 7 An Archaeological Survey of the Red River Reservoir in Wolfe, Powell, and fVenifee Counties, Kentucky. Report submitted to the U.S. Department of the Interior, National Park Service, Richmond, Virginia. Museum of Anthropology, University of Kentucky, Lexington.

258 Fryman, F. B., Fryman, M. L., and Rodeffer, M. J. 19 6 7 A Preliminary Survey of the Archaeological and Historical Resources of the Proposed Parker Branch Reservoir, Laurel, Jackson, and Rockcastle Counties, Kentucky. U.S.D.A. Forest Service.

Funkhouser, W. D. and Webb, W. S. 1928 Ancient Life in Kentucky. Geologic Reports, Series 6. Kentucky Geological Survey, Lexington.

1929 The So-called "Ash Caves" in Kentucky. Reports in Anthropology and Archaeology 1{2):37-tt2. University of Kentucky, Lexington.

1930 Rock Shelters of Wolfe and Powell Counties, Kentucky. Reports in Anthropology and Archaeology 7(4):239-306. University of Kentucky, Lexington.

Gatus, T. W. 198 7 Patterns of Chert Utilization at the Danville Tank Site. In A Phase III Archaeological Examination of the Danville Tank Site (15Bo16) Boyle County, Kentucky, edited by R.D. Boedy and C.M. Niquette, pp. 147-186. Cultural Resource Analysts, Inc., Lexington.

Gatus, T. W. and Sanders, T. N. 1978 A Reconnaissance and Evaluation of Archaeological Sites in Perry County. Kentucky Heritage Council Archaeological Report 10. Kentucky Heritage Council, Frankfort.

Goad, S. I. 1980 Patterns of Late Archaic Exchange. Tennessee Anthropologist 5:1-16.

Graham, C. Douglas R. 1990 Geology. In Archaeological Site Distributions on the Cumberland Plateau of Eastern Kentucky, edited by T. Sussenbach, pp. 13-24. University of Kentucky Press, Lexington.

Grayson, D. K. 1981 The Effects of Sample Size on Some Derived Measures in Vertebrate Faunal Analysis. Journal of Archaeological Science 8:77-88.

19 8 4 Quantitative Zooarchaeology: Topics in the Analysis of Archaeological Faunas. Academic Press, New York.

Gremillion, K. J. 19 9 3 Archaeological Investigations at the Rock Bridge Shelter (15Wo75), Wolfe County, Kentucky. Report submitted to Cecil Ison, Forest Archaeologist, Daniel Boone National Forest, Stanton, Kentucky. The Ohio State University, Columbus.

1 9 95 Archaeological and Paleoethnobotanical Investigations at the Cold Oak Shelter, Kentucky. Report submitted to the National Geographic Society. The Ohio State University, Columbus.

259 Gremillion, K. J. 1996 Early Agricultural Diet in Eastern North America: Evidence from Two Kentucky Rockshelters. American Antiquity 67:520-536.

199 7 New Perspectives on the Paleoethnobotany of the Newt Kash Shelter. In People, Plants, and Landscapes: Studies in Paleoethnobotany, edited by K. J. Gremillion, pp. 23-41. University of Alabama Press, Tuscaloosa.

Griffin, J. B. 194 6 Cultural Change and Continuity in Eastern United States Archaeology. In Man in Northeastern North America, edited by F. Johnson, pp. 37-95. Phillips Academy, Andover, M assachussetts.

1967 Eastern North America Archaeology: A Summary. Science 756:175-191.

Haag, W. G. 1 94 7 Preliminary Appraisal of the Archaeological Resources of Wolfe Creek Dam Reservoir. Museum of Anthropology, University of Kentucky, Lexington.

19 60 A Proposed Sequence of Occupation of Ceramic Sites in the Wolfe Creek Dam Reservoir. Museum of Anthropology, University of Kentucky, Lexington.

19 74 The Adena Culture. In Archaeological Researches in Retrospect, edited by G. R. Willey, pp.119-148. Winthrop Publishing, Cambridge, MA.

Hand, R. B. 19 92 A Coal Mine Survey Along Lindon Fork, Wolfe County, Kentucky. Contract Publication Series 92-20. Cultural Resource Analysts, Inc., Lexington.

Haney, D. C. 19 78 Geologic Map of the Cobhill Quadrangle, East-Central Kentucky. Map GQ-1347. U.S. Geological Survey, Reston, VA.

Henderson, A. G. and C. A. Turnbow 1987 Fort Ancient Developments in Northeastern Kentucky. In Current Archaeological Research in Kentucky Volume edited 1, by D. Pollack, pp. 205-224. Kentucky Heritage Council, Frankfort.

Holm, M. A. 1 995 Faunal Remains, In Archaeological and Paleoethnobotanical Investigations at the Cold Oak Shelter, Kentucky, pp. 82B-88. Written by Kristen J. Gremillion, Department of Anthropology, The Ohio State University. Report submitted to the National Geographic Society.

Hunt, C. B. 19 74 Natural Regions of the United States and Canada. W. H. Freeman, San Francisco.

260 Ingbar, E. E., M. L. Larson, and B. A. Bradley 19 89 A Nontypoiogicai Approach to Debitage Analysis. In Experiments in Lithic Analysis, edited by D.S. Amick and R.P. Mauldin, pp. 117-136. B.A.R., Oxford.

Ison, C. R. 19 88 The Cold Oak Shelter: Providing a Better Understanding of the Terminal Archaic. In Paleoindian and Archaic Research in Kentucky, edited by 0. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 205-220. Kentucky Heritage Council, Lexington.

1991 Prehistoric Upland Farming along the Cumberland Plateau. In Studies in Kentucky Archaeology, edited by C. D. Hockensmith, pp. 1-10. Kentucky Heritage Council, Frankfort.

Ison, C. R. and Knudsen, G. D. 19 85 A Cultural Resource Assessment of the Smith and Isaacs Land Exchange Tracts, Jackson County, Kentucky, Berea District, Daniel Boone National Forest. U.S.D.A. Forest Service.

Jefferies, R. W. 1988a The Archaic in Kentucky: New Deal Archaeological Investigations. In New Deal Era Archaeology and Current Research in Kentucky, edited by D. Pollack and M. L. Powell, pp. 14-25. Kentucky Heritage Council, Frankfort.

1988b Archaic Period Research in Kentucky: Past Accomplishments and Future Directions. In Paleoindian and Archaic Research in Kentucky, edited by C. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 86126. Kentucl^ Heritage Council, Frankfort.

1990 Archaic Period. In The Archaeology of Kentucky: Past Accomplishments and Future Directions, edited by D. Pollack, pp. 143-246. State Historic Preservation Comprehensive Plan Report No. 1. Kentucky Heritage Council, Frankfort.

1996 Hunters and Gatherers after the Ice Age. In Kentucky Archaeology, edited by R. B. Lewis, pp. 39-78. The University Press of Kentucky, Lexington.

Johnson, L. L. 1 9 75 Graph Theoretic Snalysis of Lithic Tools from Northern Chile. In Lithic Technology: fAaking and Using Stone Tools, edited by E. Swanson, pp. 63-95. Mouton Publishers, Paris.

Johnson,J. K. 1 9 89 The Utility of Production Trajectory Modeling as a Framework for Regional Analysis. In Alternative Approaches to Lithic Analysis, edited by D. O. Henry and G. H. Odell, pp. 119-138. Archaeological Papers of the American Anthropological Association No. 1. American Anthropological Association, Washington, D. C.

26 1 Jones, G. T., Beck, C. and Grayson, D. K. 1983 Artifact Class Richness and Sample Size in Archaeological Surface Assemblages. In Lulu Linear Punctuated: Essays in Honor of George irving Quimby, edited by R. C. Dunnell and D. K. Grayson, pp. 55-73. Anthropological Papers 72. Museum of Anthropology, University of Michigan, Ann Arbor.

Jones, V. H. 1936 The Vegetal Remains of Newt Kash Hollow Shelter. University of Kentucky Reports in Archaeology and Anthropology 3(4):147-165. University of Kentucky, Lexington.

Justice, N. 1987 Stone Age Spear and Arrow Points of the Midcontinentai and Eastern United States. Indiana University Press, Bloomington.

Kluth, R. A. 1 992 Phase a investigations at the Ratliff Shelter (15Le149) in West Central Lee County, Kentucky. Archaeological Report Number 13. Kentucky Department for Surface Mining Reclamation and Enforcement, Frankfort.

Knox, J. 0. 1 983 Response of River Systems to Holocene Climates. In Late Quaternary Environments of the United States, Volume edited2, by H. E. Wright, Jr., pp. 26-41. University of Minnesota Press, Minneapolis.

Knudsen, G. D. n.d. Cultural Resource inventory. Walker Branch F485 Timber Sale, Berea Ranger District, Jackson County, Kentucky. U.S.D.A. Forest Service.

1 98 8 Ubiquitous Lithic Scatters: You Can't Tell a Site by its Disturbance. In Paleoindian and Archaic Research in Kentucky, edited by C. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 221-232. Kentucky Hertiage Council, Frankfort.

Knudsen, G. D., et al. n.d. Cultural Resource Assessment Number 448 Timber Sale, Jackson County, Kentucky, Daniel Boone National Forest. U.S.D.A. Forest Service.

Knudsen, G. D., Kellar, J., and Simpson, S. 1 9 8 3 Cultural Resource Inventory, Big Sinking Creek Oil Field, Daniel Boone National Forest, Kentucky. USDA, Forest Service, Daniel Boone National Forest.

Koetje, T. A. 1 993 Site Formation Processes and Behavioral Deposition Episodes: The View from Le Flageolet II. In Formation Processes In Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 113-124. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

262 Ledbetter, R. J. and L. D. O'Steen 1992 The Grayson Site: Phase III Investigations of 15Cr73, Carter County, Kentucky. Report prepared by Southeastern Archaeological Services, Atlanta, Georgia.

Leonard, R. D. 198 9 Resource Specialization, Population Growth, and Agricultural Production in the American Southwest. American Antiquity 54:491-503.

Leonard, R. D., Smiley, F. E. and Cameron, C. M. 198 9 Changing Strategies of Anasazi Lithic Procurement on Black Mesa, Arizona. In Quantifying Diversity in Archaeology, edited by R. D. Leonard and G. T. Jones, pp. 100-108. Cambridge University Press, New York.

Lewis, R. B. 198 8 New Deal Archaeology at Mississippian Sites in Kentucky. \nNew Deal Era Archaeology and Current Research in Kentucky, edited by D. Pollack and M. L. Powell, pp. 26-45. Kentucky Heritage Council, Frankfort.

Lewis, R. B. (Editor) 1996 Kentucky Archaeology. The University Press of Kentucky, Lexington.

Linton, R. 1924 The Significance of Certain Traits in North American Maize Culture. American Anthropologist 26:345-359.

Luedtke, Barbara E. 1992 An Archaeologists Guide to Chert and Flint. University of California Institute of Archaeology, Los Angeles.

Magne, Martin P.R. 198 9 Lithic Reduction Strategies and Assemblage Formation Processes. In Experiments in Lithic Analysis, edited by D.S. Amick and R.P. Mauldin, pp. 15-31. OB.A.R., Oxford.

Marquardt, W. H. 19 70 Archaeological Investigations in the Cave Run Reservoir, Kentucky: 1969 Season. Ms. on file. Department of Anthropology, University of Kentucky, Lexington.

Mauldin, Raymond P. and Daniel S. Amick 198 9 Investigating Patterning in Debitage from Experimental Bifacial Core Reduction. In Experiments in Lithic Analysis, edited by D.S. Amick and R.P. Mauldin, pp. 67-88. Oxford: B.A.R.

Maynard, D. R. and Gatus, T. W. 19 79 A Reconnaissance and Evaluation of Archaeological Sites in Greenup County, Kentucky. Unpublished manuscript, Kentucky Heritage Council, Frankfort.

263 McBride, K. A. 1991 Historical Settlement Patterns within Robinson Forest, Breathitt and Knott Counties, Kentucky. In The Human Landscape in Kentucky's Past: Site Structure and Settlement Patterns, edited by C. Stout and 0. K. Hensley, pp. 8-29. Kentucky Heritage Council, Frankfort.

McBride, K. A. and W. S. McBride 19 9 6 From Colonization to the 20th Century. In Kentucky Archaeology, edited by R. B. Lewis, pp. 183-212. The University Press of Kentucky, Lexington.

McFarlan, A. C. 1943 Geology of Kentucky. University of Kentucky, Lexington.

McGowan, K. P. 19 8 8 Settlement Patterns in the Kentucky Mountains. In New Deal Era Archaeology and Current Research in Kentucky, edited by D. Pollack and M. L. Powell, pp. 144-160. Kentucky Heritage Council, Frankfort.

McKern, W. C. 1939 The Midwestern Taxonomic System as an Aid to Archaeological Culture Study. American Antiquity 4:301-313.

Meadows, Larry G. 1977 Chert Resources in Powell County in Concurrence with Aboriginal Use. In A Reconnaissance and Evaluation of Archaeological Sites in Powell County, Kentucky, edited by M.K. Weinland and T.N. Sanders, pp. 98-122. Kentucky Heritage Commission, Lexington.

Meighan, C. W. 1966 Archaeology: An Introduction. Chandler, San Francisco.

Meltzer, D. J. 1984-85 On Stone Procurement and Settlement Mobility in Eastern Fluted Point Groups. North American Archaeologist 6:1-24.

Michels, J. W. 1 9 73 Dating Methods in Archaeology. Seminar Press, New York.

Mickelson, K. R. and Gremillion, K. G. 1 996 Archaeological Investigations at the Mounded Talus Shelter (15Le77). Paper presented at the Thirteenth Annual Kentucky Heritage Council Archaeological Conference, March 2-3, Louisville, Kentucky.

Murdock, G. P. 1967 Ethnographic Atlas: A Summary. Ethnology 6:109-236.

Nash, D. T. 1993 Distinguishing Stone Artifacts from Naturefacts Created by Rockfall Processes. In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 125-138. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

264 Newmann, T. W. and Johnson, E. 19 79 Patrow Site Lithic Analysis. Midcontinental Journal of Archaeology 4:79-93.

Nielsen, Axel E. 1991a Where Do Microartifacts Come From? Paper presented at the 56th annual meeting of the Society for American Archaeology, New Orleans, April 11-14, 1991.

1991b Trampling the Archaeological Record: An Experimental Study. Am erican A n tiq u ity 56(3): 483-503.

Niquette, C. M. and Henderson, A. G. 1984 Background to the Historic and Prehistoric Resources of Eastern Kentucky. Cultural Resource Series No. 1. Environmental Consultants, Inc., Lexington.

Niquette, C. M. and Hughes, M. A. 1992 A National Register Evaluation of the Lindon Fork Rockshelter (15Wo107), Wolfe County, Kentucky. Contract Publication Series 92-77. Cultural R esource Analysts, Inc., Lexington.

O'Steen, L. D., Gremillion, K. J. and Ledbetter, R. J. 1991 Archaeological Testing of Five Sites in the Big Sinking Creek OH Field, Lee County, Kentucky. Report submitted to Cecil Ison, Forest Archaeologist, Daniel Boone National Forest, Stanton, Kentucky. Southeastern Archaeo­ logical Services, Inc., Atlanta.

Petraglia, M. D. 1993 The Genesis and Alteration of Archaeological Patterns at the Abri Dufaure: An Upper Paleolithic Rockshelter and Slope Site in Southwestern France. In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 97-112. Monographs in World Archaeology No. 17. Prehistory P^ess, Madison, Wisconsin.

Prentiss, William C. and Eugene J. Romanski 1 989 Experimental Evaluation of Sullivan and Rosen's Debitage Typology. In Experiments in Lithic Analysis, edited by D.S. Amick and R.P. Mauldin, pp. 89-100. B.A.R., Oxford.

Puetz, C. J. 1985 Kentucky County Maps. County Maps, Lyndon Station, Wl.

Purdy, 8. A. 19 75 Fractures for the Archaeologist. In Lithic Technology: Making and Using Stone Tools, edited by E. Swanson, pp. 133-141. Mouton, Paris.

Rafferty, J. E. 1985 The Archaeological Record of Sedentariness: Recognition, Development, and Implications. In Advances in Archaeological Method and Theory Volume 8, edited by M. B. Schiffer, pp. 113-156. Academic Press, Orlando.

265 Rafferty, J. E. 19 94 Gradual or Step-wise Change: The Development of Sedentary Settlement Patterns in Northeastern Mississippi. American Antiquity 59:405-425.

Railey, J. A. 1 9 90 Woodland Period. In The Archaeology of Kentucky: Past Accomplishments and Future Directions, edited by D. Pollack, pp. 247-374. State Historic Preservation Comprehensive Plan Report No. 1. Kentucky Heritage Council, Frankfort.

1991a The Conley-Green Rockshelter (15Le4): An Early Woodland Occupation in the Cumberland Plateau of Northeastern Kentucky. In Studies in Kentucky Archaeology, edited by C. D. Hockensmith, pp. 66-101. Kentucky Heritage Council, Frankfort.

1991b Woodland Settlement Trends and Symbolic Architecture in the Kentucky Bluegrass. In The Human Landscape in Kentucky's Past: Site Structure and Settlement Patterns, edited by C. Stout and C. K. Hensley, pp. 56-77. Kentucky Heritage Council, Frankfort.

1996 Woodland Cultivators. In Kentucky Archaeology, edited by R. B. Lewis, pp. 79-126. The University Press of Kentucky, Lexington.

Rice, C. L 1984 Sandstone Units of the Lee Formation and Related Strata in Eastern Kentucky. Contributions to the Geology of Kentucky Geological Survey Professional Paper 1151-G. U.S. Government Printing Office, Washington.

Rice, C. L. and Weir, G. W. 1984 Lee and Breathitt Formations Along the Northwestern Part of the Eastern Kentucky Coal Field. In Sandstone Unites of the Lee Formation and Related Strata in Eastern Kentucky. Contributions to the Geology of Kentucky Geological Survey Professional Paper 1151-G. U.S. Government Printing Office, Washington.

Rice, O. K. 1975 Frontier Kentucky. The University Press of Kentucky, Lexington.

Rick, J. W. 19 76 Downslope Movement and Archaeological Intra-site Spatial Analysis, American Antiquity 41:^33-^44.

Ritchie, W. A. 1932 The Lamoka Lake Site. Researchers and Transactions of the New York State Archaeological Association, Volume 7, Number 4. New York State Archaeological Association, Rochester.

266 Rosen, A. M. 1993 Microartifacts as a Reflection of Cultural Factors in Site Formation. In Formation Processes in Archaeological Context, edited by P. Goldberg and D. T. Nash, pp. 114-148. Monographs in World Archaeology No. 17. Prehistory Press, Madison, Wisconsin.

Rossen, J. and R. Edging 19 87 East Meets West; Patterns in Kentucky Late Prehistoric Subsistence. In Current Archaeological Research in Kentucky Volume edited 1, by D. Pollack, pp. 225-234. Kentucky Heritage Council, Frankfort.

Sable, E. G. and Dever, G. R., Jr. 1990 Mississippian Rocks in Kentucky. U.S. Geological Survey Professional Paper 1503. U.S. Government Printing Office, Washington.

Sanders, S. L. 1991 An Archaeological Survey and Assessment of the David Ratliff Proposed Surface Coal Mine in Lee County, Kentucky. Archaeological Report Number 12. Kentucky Department for Surface Mining Reclamation and Enforcement, Frankfort.

Sanders, T. N. 1976 Archaeological Survey and Testing Excavations in the Proposed Paintsville Lake Reservoir Project. Museum of Anthropology, University of Kentucky, Lexington.

1988 The Adams Site: A Paleoindian Manufacturing and Habitation Site in Christian County, Kentucky. In Paleoindian and Archaic Research in Kentucky, edited by C. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 1-24. Kentucky Heritage Council, Frankfort.

Sanders, T. N. and Gatus, T. W. 19 77 A Reconnaissance and Evaluation of Archaeological Sites in Floyd County, Kentucky. Kentucky Heritage Council, Frankfort.

Schenian, P. A. 1988 An Oven/iew of the Paleoindian and Archaic Period Occupations of the Savage Cave Site. In Paleoindian and Archaic Research in Kentucky, edited by C. D. Hockensmith, D. Pollack, and T. N. Sanders, pp. 67-84. Kentucky Heritage Council, Frankfort.

Schiffer, M. B. 19 87 Formation Processes of the Archaeological Record. University of New Mexico Press, Albuquerque.

Schmid, E. 1963 Cave Sediments and Prehistory. In Science in Archaeology, edited by D. Broth we 11 and E. Higgs, pp. 123-138. Basic Books, New York.

Sharp, W. E. 199 6 Fort Ancient Farmers. In Kentucky Archaeology, edited by R. B. Lewis, pp. 161-182. The University Press of Kentucky, Lexington.

2 6 7 stein, J. K. 1 9 85 interpreting Sediments in Cultural Settings. In Archaeological Sedimer^ts in Context, edited by J. K. Stein and W. R. Farrand, pp. 5-20. Peopling of the Americas Volume 1. Center for the Study of Early Man, Institute of Quaternary Studies, University of Maine at Orona, Orono, Maine.

Steponaitis, V. P. 19 86 Prehistoric Archaeology in the Southeastern United States, 1970-1985. Annual Review of Anthropology /5:363-404.

Stockton, E. D. 1 9 73 Shaw's Creek Shelter: Human Displacement of Artifacts and Its Significance. Mankind 9:112-117.

Struever, S. and Vickery, K. D. 1 9 73 The Beginnings of Cultivation in the Midwest-Riverine Area of the United States. American Anthropologist 75(5):1197-1220.

Sullivan, Alan P. Ill and Kenneth C. Rosen 1985 Debitage Analysis and Archaeological Interpretation. American Antiquity 50: 755-779.

Sussenbach, T. 1990 Archaeological Site Distributions on the Cumberland Plateau of Eastern Kentucky. University of Kentucky, Program for Cultural Resource Assessment, Archaeological Report 218. University of Kentucky, Lexington.

Tankersley, K. B. 1990 Paleoindian Period. In The Archaeology of Kentucky: Past Accomplish­ ments and Future Directions, edited by D. Pollack, pp. 73-142. State Historic Presen/ation Comprehensive Plan Report No. 1. Kentucky Heritage Council, Frankfort.

Tankersley, K. B. 1996 Ice Age Hunters and Gatherers. In Kentucky Archaeology, edited by R. B. Lewis, pp. 21-38. The University Press of Kentucky, Lexington.

Thomas, D. H. 1 97 9 Archaeology. Holt, Rinehart & Winston, New York.

Tomka, Stephen A. 1989 Differentiating Lithic Reduction Techniques: An Experimental Approach. In Experiments in Lithic Analysis, edited by D.S. Amick and R.P. Mauldin, pp. 137-162. B.A.R., Oxford.

Turnbow, C. 1 9 75 Site documentation on file at the Office of the State Archaeologist, Lexington.

268 Turnbow, C. 19 76 An Archaeological Survey of the Red River Gorge Geological Area in Daniel Boone National Forest in Powell, Wolfe, and Menifee Counties, Kentucky. Report submitted by the University of Kentucky Museum of Anthropology to the United States Forest Service, Winchester, Kentucky.

19 81 Cultural Radiocarbon Determinations of Kentucky. Occasional Papers in Anthropology No. 3, Department of Anthropology, University of Kentucky. University of Kentucky, Lexington.

Vickery, K. D. 19 8 0 Preliminary Definitions of Archaic 'Study Unit' in Southwestern Ohio. Unpublished manuscript prepared for the State Archaeological Preservation Plan Meeting held on April 25, 1980, Columbus, Ohio.

Villa, P. and Courtin, J. 1983 The Interpretation of Stratified Sites: A View from Underground. Journal of Archaeological Science, 10, 267-281.

Waters, M. R. 1992 Principles of Geoarchaeology: A North American Perspective. The University of Arizona Press, Tucson.

Webb, W. S. 1942 The 0 and O Mounds at Paintsville, Sites Jo2 and Jo9, Johnson County, Kentucky. Reports in Anthropology and Archaeology 5:297-372. University of Kentucky, Lexington.

Webb, W. S. and Baby, R. 1957 The Adena People No. 2. Ohio Historical Society, Columbus.

Webb, W. S. and Funkhouser, W. D. 1932 Archaeological Survey of Kentucky. University of Kentucky Reports in Anthropology and Archaeology 2. University of Kentucky, Lexington.

1936 Rock Shelters in Menifee County, Kentucky. University of Kentucky Reports in Anthropology and Archaeology 3(4):105-167. University of Kentucky, Lexington.

Weinland, M. and Sanders, T. N. 1977 A Reconnaissance and Evaluation of Archaeological Sites in Powell County, Kentucky. Kentucky Heritage Commission Report No. 2. Kentucky Heritage Commission, Frankfort.

Weir, G. W. and Richards, P. W. 1974 Geologic Map of the Pomeroyton Quadrangle, East-Central Kentucky. U.S. Geological Survey, Reston, VA.

Wesler, K. W. 1991 On the Analytical Utility of Disturbed Deposits: Moundfill and Backfill at Wickliffe Mounds. In Studies in Kentucky Archaeology, edited by C. D. Hockensmith, pp. 119-132. Kentucky Heritage Council, Frankfort.

269 Willey, G. R. and Phillips, P. 1958 Method and Theory in American Archaeology. University of Chicago Press, Chicago.

Winters, H. D. 1969 The Riverton Culture: A Second Millenium Occupation in the Central Wabash Valley. Illinois Archaeological Survey Monograph 1. Illinois State Museum, Springfield.

Withington, W. A. and G. Pauer, III 1980 Kentucky in Maps, [s.n.], Lexington.

Wood, W. R. and Johnson, D. L. 1 978 A Survey of Disturbance Process in Archaeolgical Site Formation. Advances in Archaeological Method and Theory 7:315-381.

Wyss, J. D. and Wyss, S. K. 19 77 An Archaeological Assessment of Portions of the Red River Gorge Geological Area, Menifee County, Kentucky. Ohio Valley Archaeological Research Associates, Lexington.

Yerkes, R. W. 1988 The Woodland and Mississippian traditions in the prehistory of Midwestern North America. Journal of World Prehistory 2:307-358.

1989 Lithic Analysis and Activity Patterns at Labras Lake. In Alternative Approaches to Lithic Analysis, edited by D. O. Henry and G. H. Odell, pp. 183-212. Archaeological Papers of the American Anthropological Association Number 1. American Anthropological Association, Washington.

Yerkes, R. W. and Pecora, A. M. 1991 Lithic Analysis. In Late Woodland Archaeology at the Parkline Site (46PU99) Putnam County, West Virginia, edited by C. M. Niquette and M. A. Hughes, pp. 51-108. Cultural Resource Analysts, Inc., Lexington.

270 APPENDIX A

Chert Descriptions

Boyle

Boyle chert derives from the Boyle Member of the Devonian Borden Formation.

Within Its dolomltic matrix, Boyle chert occurs as nodules or discontinuous layers. The

Boyle Member Is stratlgraphlcally the lowest chert-bearing stratum to crop out In the study area. While It Is not exposed In the Immediate vicinity of Cold Oak and Rock Bridge shelters, It Is found In the area as colluvlal chert In streams and it crops out In bordering counties (Meadows 1977).

Boyle chert ranges in color from white to tan to gray to greenish-gray; pink, red and blue varieties are also documented. The cortex Is usually yellowish-brown . It Is opaque with a dull to seml-vltreous luster. Mottling Is common, and banding also may be present. Fossil Inclusions are crinolds and, less commonly, sponge spicules.

Meadows (1977) noted that Boyle was the chert most commonly used by prehistoric knappers In Powell County. The abundance of Poyle artifacts at the Rhondle

Lee site In Powell County attests to this observation (Applegate 1995b). In the western part of the state, Boyle Is characterized as "a common chert source for prehistoric tool manufacturers" (Gatus 1987: 153), as It Is the most abundant chert In the Danville

Tank site area.

Muidraugh

Muldraugh chert derives from the Lower Mississippian Muidraugh Member of the Borden Formation. Outcrops of this deposit occur In west-central Kentucky (Sable and Dever 1990). Since It does not crop out or occur as colluvlum In the study area.

27 1 Muidraugh is considered an "exotic" chert"'. According to Ison (pers. comm.), Muidraugh chert most often occurs as finished tools in eastern Kentucky.

Muidraugh chert is usually a combination of gray, blue, and white hues dis­ tributed in a swirled manner. The cortex is yellow-red. White sponge spicules are a diagnostic inclusion in this chert type. A sugai-y luster is typical.

St. Louis

St. Louis is one of several chert types of the Upper Mississippian Period Slade

Formation (Ettensohn et al. 1984), formerly the Newman Limestone. Cherts are more abundant toward the top of the member. According to Ettensohn et al. (1984:8),

[cjhert of various colors occurs abundantly as irregularly shaped masses, as spheroidal and discoidal nodules, and as thin to very thin discontinuous beds. Silicification of fossils is common. Micritic crusts and stringers in an exposure zone at the top of the member commonly are silicified selectively and form wavy bands of light-colored chert in the dark-gray altered limestone.

In addition, Sable and Dever (1990:59) reported that within the St. Louis Limestone

Member, "chert is a common to abundant accessory and occurs as ... replacement layers

("scraggy" or "scraggly" chert) of limestone and dolostone. In some areas chert occurs in fairly discrete widespread "zones"."" According to Black (1978), St. Louis chert also takes the form of pebble- to boulder-sized blocks. In addition, it may be obtained from stream beds or alluvial deposits in the study area.

This semi-vitreous chert is cryptocrystalline to microcrystalline. Banding or slight mottling may be present. White spheroidal inclusions are noted in some specimens. Fossil brachiopods, bryozoans, corals, crinoids, and sponge spicules are common inclusions in St. Louis chert. An unidentified, cylindrical, segmented invertebrate also occurs in some specimens. Fossils occur isolated or in very dense patches. In the Cold Oak assemblage, the maximum dimension of the largest fossils approaches one cm. Fossils may be oriented in a linear manner.

^ Exotic is defined as cherts whose sources are greater than 30 km from a site.

272 Meadows (1977) identified two varieties of St. Louis cfiert based on color. The olive green to greenish-gray variety was often used by knappers owing to its high quality and fracturing properties. The red to reddish-brown variety is of poor quality and was not extensively utilized prehistorically for chipped-stone tool manufacture.

Gatus (1987) also noted the occurrence of white, grayish-blue, and tan mottled varieties of St. Louis chert. The blue to green varieties commonly develop a porous, orange-brown cortex while the white variety has a white-orange, chalky cortex. Green, red, gray, and tan varieties of St. Louis chert are found in the Cold Oak and Rock Bridge assemblages.

Ste. Genevieve

Stratlgraphlcally intermediate to St. Louis and Paoli, Ste. Genevieve chert originates in the lower part of the Ste. Genevieve Member of the Upper Mississippian

Slade Formation (formerly Newman Limestone). “Chert is a minor constituent mainly associated with selective silicification of fossils and of micritic crusts and stringers in vadose- and subaerial-exposure zones" (Ettensohn et al. 1984:10). Sable and Dever reported that “chert is a relatively minor accessory, occurring as nodules, irregular masses, and siliceous replacements of fossiliferous beds such as the Lost River Chert

Bed" (1990:62) and that “northwest of Somerset, a bed of cherty limestone is at the top of the mapped St. Louis; it weathers to a chert similar in appearance to the Lost River chert" (1990:65). At the base of the limestone member, "pebbles of ... red chert as much as 2 cm across" have been documented (Black 1978); it is possible these are inclusions from the underlying St. Louis Member.

Characterized as having poor chipping qualities, Ste. Genevieve chert is typically fine-grained to microcrystalline, subvitreous, fossil-free, and dusky blue in color.

Gray-black to gray to brownish varieties are also observed. Banding may be present. The cortex is typically orange-brown and limonite-like.

273 Paoli

Paoli chert derives from the Paoli-Beaver Bend Member of the upper Slade

Formation (formerly Newman Limestone), which is Upper Mississippian in stratigraphie position. The chert occurs as discontinuous beds, thin chert lenses, and nodules.

Paoli chert is opaque, cryptocrystalline, highly silicified, and vitreous to subvitreous. Banding and mottling may be present in some specimens. Fossil sponge spicules are observed but rare. Paoli varies in color from red to brown to various shades of gray. The cortex is red to yellow-red in color.

Second only to Boyle chert as the preferred raw material of prehistoric knappers in Powell County as a whole, Paoli is the most common chert type recovered from rock­ shelters in the north-northeastern portions of that county (Meadows 1977). Meadows added that "this chert was used chiefly in the production of projectile points" (1977:

108). Paoli is sometimes referred to as Elkhorn chert.

Haney

Haney chert derives from the Ramey Creek Member of the Upper Mississippian upper Slade Formation. This stratum was previously classified as the Haney Limestone

Member of the upper Newman Limestone Formation (Ettensohn et al. 1984). Haney chert is not usually associated with alluvial deposits since it pits and cracks relatively easily. Hence, it is more common to encounter this chert in outcrops. The Ramey Creek

Member contains "abundant chert" and "widespread replacement of megafossils by red chert" (Ettensohn et al. 1984:16). Sable and Dever (1990:74) explain that Haney chert is “characteristically a whitish-weathering replacement of oolitic and fossil fragmental limestone and is conspicuous in weathering residuum." It also may occur as thin lenses of chert.

274 According to many authors, the most diagnostic attribute of Haney chert is light- colored oolites. In the Cold Oak and Rock Bridge assemblages, however, some specimens of Haney chert do not have oolites and some show a gradation from having to lacking oolites. Fossil sponge spicules also occur but are not common. Haney chert is usually cryptocrystalline and fairly translucent. The luster ranges from vitreous to subvitreous to dull, the latter being associated with darker specimens. Various colors of Haney chert are documented: reddish-brown, brown, yellow-green, blue, and gray. Banding and mottling are less common properties of Haney chert. Flake scars on Haney artifacts are often indistinct and difficult to discern.

Meadows (1977) reported that Haney chert is very common in rockshelter sites of Powell County, especially the eastern side of the county. Haney material dominates the lithic assemblage of the Rhondle Lee site in southwest Powell County (Applegate 1995b).

Breathitt

Breathitt is one of two chert types from Lower Pennsylvanian formations, in this case the Breathitt Shale Member of the Lower Tongue of Breathitt Formation. Breathitt chert crops out in only two counties of eastern Kentucky: Breathitt and Magoffin. It does, however, occur in alluvial deposits and stream beds in other parts of eastern Kentucky.

It is considered to be high quality chert. Breathitt is also referred to as "Flint Ridge of

Morse " chert but is not correlated with the well-known Flint Ridge deposits of east- central Ohio.

Diagnostic attributes of Breathitt chert include cryptocrystalline to micro- crystalline texture, occasional thin banding or brown mottling, and sponge spicule inclusions. Fossil inclusions may be oriented in a liner fashion. Colors range from black to dark gray to grayish-brown, but reddish and yellowish varieties are also reported.

Breathitt chert is opaque with a subvitreous to dull luster. Cortex colors range from dark brown to reddish black.

275 Kanawha

Kanawha is the other chert of Lower Pennsylvanian derivation. It crops out in

West Virginia near Charleston in Braxton and Lewis Counties, but is also found as pebbles in alluvial deposits along the Kanawha River and several of its tributaries

(Yerkes and Pecora 1991). Therefore, Kanawha chert is considered an "exotic" chert in the Cold Oak and Rock Bridge assemblages.

Kanawha chert is similar to Breathitt chert in color (dark gray to black) and fossil content (sponge spicules). It tends, however, to be more dull, opaque and coarse grained than Breathitt owing to minor amounts of sand, silt or mica inclusions. Kanawha also may have light-colored crescents and subspherical portions internally. It may exhibit mottling. The cortex is commonly white, cream or honey colored, and weathering may produce iron-stained fracture planes.

Yerkes and Pecora (1991) note that heat treatment improves the quality of

Kanawha chert and makes flake scars slightly more vitreous, but it has little affect on the color. They also report that Kanawha chert was used extensively for chipped-stone tool manufacture in the Kanawha and Teays river valleys of northwestern West Virginia.

Other Cherts

There are several additional chert-bearing strata of Mississippian and

Pennsylvanian ages in east-central and northeastern Kentucky. They were not noted in the Cold Oak and Rock Bridge assemblages since the author is not familiar with them and they are not in the comparative collection used in this analysis. More chert-bearing strata are exposed in other parts of the state (see Sable and Dever 1990).

According to Ettensohn et al. (1984), several members of the Mississippian

Slade Formation (formerly Newman Limestone) contain authigenic or foreign chert. The

Renfro Member, which previously had been classified as the upper part of the Borden

Formation, can be cherty. The Warix Run Member, formerly known as the northern unit

276 of the Ste. Genevieve Limestone Member of the Newman Limestone Formation, may

contain clasts from lower strata (such as the St. Louis Limestone) of the formation. The

Mill Knob Member of the Slade Formation was previously classified as the Ohara

Limestone Member of the Ste. Genevieve Member of the Newman Limestone Formation.

Ettensohn et al. (1984:12) note that "chert is relatively common in the Mill Knob

Member and is associated generally with the secondary silicification of exposure

features such as micritic crusts and stringers and teepee structures."

The Pennsylvanian Lee Formation is largely a sandstone unit. Therefore, chert is

not common to rocks in this formation. The Livingston Conglomerate Member, however, contains "sparse chert fragments and pebbles derived from the underlying Mississippian carbonates" (Rice and Weir 1984:G29). Livingston paleovalley channel fill "contains pebbles and cobbles as much as 10 in. in diameter of chert derived from the Newman

[Slade Formation] Limestone" (Rice and Weir 1984:G30). Further, some paleochannel deposits may contain a basal layer of "poorly sorted pebbles and boulders of locally derived chert, sandstone, and limestone as much as 3 ft thick" (Rice and Weir

1984:G32). As previously noted, the Corbin Sandstone Member contains "minor amounts of gray chert" as rounded to subrounded pebbles in local conglomeritic layers

(Haney 1976), but these are probably not of sufficient size for chipped-stone tool manufacture.

277 APPENDIX C

Figures and Tables

278 Foils A rea Louisville /salt ^

IV)

CD

Laka Kentucky Barkley Lake

Lond Between the L okes

#00

Figure 1. Major drainages of Kentucky (modified from Jefferies 1988b;89) and the current study area (rectangle). ro 00 o

miles

Figure 2. Kentucky counties, showing the ten-county study area (modified from Kentucky Chamber of Commerce 1964). sM j lUÇüSlOjW

A dirondack ^ f *Acad>o ftP Province New Englond ^ P rovince

ALtw'^'feC Bosior i

I ClgvtlOrd ■ New C.ty Appalachian ‘!9» Trenton P lateau rtsourq "Z.l#Philodeiohio ilm in g to n Cincimctt 8/f/

LduiSv Ricnmon Peiersfiuf^ P rovince "‘^Norfolk

K n o iv ille Piedmcnt Province Nosnv

ChcttQ

Cdumb'O Blue Ridge Province Atlanta Aoqu^tQ

Macon C*'CrtcSTon 400 M.f

Figure 3. Physiographic provinces of the eastern United States, illustrating the lateral extent of the Appalachian Plateau Province (modified from Hunt 1974:254). Study area is shown as a rectangle.

2 8 1 C in c ln n a n

hUn«

CvanavilU

ro CO ro

«NVROVAV. •Cwm-lMrlSU'CXp' —

40

mi les

Figure 4. Physiographic diagram of Kentucky (modified from fVlcFarlan 1943:3), illustrating the Cumberland Plateau, Pottsville Escarpment, The Knobs, and the study area (rectangle). Key to ages of bedrock:

ΠCenozoic K Cretaceous P Pennsylvanian M Mississippian D/S Devonian / Silurian O Ordovician D/S

D/S

ro 00 0 3

CZ 40 mi les D/S

Figure 5. Bedrock geology of Kentucky (modified from Withington and Gyuia 1980:11). Study area is shown as a rectangle. 1. Eastern Mountains and Coal Fields la. Escarpment Area 1b. Plateau Area 1c. Mountain and Creek Bottom Area 2. The Knobs

4 c

4d

4b

6# miles

Figure 6. Physiographic provinces of Kentucky (modified from Withington and Gyuia 1980:10). Study area is sho\wn as a rectangle. TIME CULTURAL PERIOD SUBDIVISION TIME

A.D. 1550 A.D. 1550 Late A.D. 1400 FOfTT ANCIENT Middle A.D. 1200 Early A.D. 1000 A.D. 1000

Late LATE WOODLAND A.D. 700 Early

A.D. 500 A.D. 500 Late MIDDLE WOODLAND A.D. 200 Early 200 B.C. 200 B.C.

EARLY WOODLAND

1000 B.C.

TERMINAL ARCHAIC

1500 B.C.

LATE ARCHAIC

3000 B.C.

MIDDLE ARCHAIC

6000 B.C.

EARLY ARCHAIC

8000 B.C. 8000 B.C. Late 8500 B.C. PALEOINDIAN Middle 9000 B.C. Early 9500 B.C. 9500 B.C.

Figure 7. Cultural chronology for eastern Kentucky (not to scale).

285 Rubble from hillside and ■erosion of brow- Former position of overhang Dripstonne (StaladitM) N) 00 oi . I 1 Sediments transported into Dripline Sediment from ceiling and walls shelter (exogenous) (endogenous) Rowstone

Talus and Large blocks from Colluvium Eboulls celling fall Paleo-drlpline Large blocks from brow collapse

Figure 8. Parts of a typical rockshelter (Waters 1992:241). I\] 00 I II

Figure 9. Stages of rockshelter development (Farrand 1985:24). Red River

Rock Bridge WOLFE V Shelter

Cold Oak Shelter '

K) 00 0 0 Kentucky Riven N. Fork

LEE miles Middle Fork Fork

Figure 10. Map showing the locations of Rock Bridge Shelter (15W075) and Cold Oak Shelter (15LE50). sm oll cavern

hominy hoi#

S20N500C 1984 trench

1994 trench

t\3 CD ID 500N500E

m / /

/ Nm

X Figure 11. Plan view of Cold Oak Shelter showing the locations of the 1984 and 1994 excavation trenches (modified from Gremilllon 1995:3). Ng = geographic north, Nm = magnetic north. 0 cm

20 ro 40 ID III o 60 rock 59 III 80

100

5 0 0 . 5N 0 . 5N 5 0 0 . 5N 5 0 0 . 5N50 5 0 0 . 5N 5 0 0 . 5N 501E 500E 4 99E 498E 497E 496E 0 k rock < 5 1 2 5 3 7 plant concentration 1 m

Figure 12. Stratigraphie profile of the 1994 excavation trench at Cold Oak Shelter; north is coming out of the paper (after Gremillion 1995:12). %-----

ro CD

R ocks E xcavated Nm

Ng

Figure 13. Plan view of Rock Bridge Shelter showing the 1992 excavation areas (modified from Gremillion 1993:16). Ng = geographic north, Nm = magnetic north. r 0 cm

20

48N 49E 49N 49E 50N 49E ho ID lO r 0 cm

■ 20 loose sand

48N 51E 48N 50E

^ 0 rock 1 m O root

Figure 14. Stratigraphie profiles of the main (southern) excavation block at Rock Bridge Shelter; for B-B‘, north is coming out of the paper (after Gremillion 1993:6). 56N 50E 56N 52E Sm Hm

Pm Bb Hm Hm Hu

Hb

rock

52N 49E

Hm Hm Pu

Pb Pb Pb Pm 48N 49E 48N 52E

Figure 15. Lateral distribution of chipped-stone tools in the main excavation block at Rock Bridge Shelter. Cherts are: B = Breathitt, H = Haney, P = Paoli, 8 = St. Louis. Tools types are: b = biface, lake. The two other chipped-stone tools were found outside the main excavation block.

293 56N 50E 56N 5 2 E 57:43:0 15:85:0

2:98:0 21:79:0

no profile available 80:15:5

rock 61:39:0

52N 49E 30:60:10 16:84:0

no profile 0:40:60 available

no profile 0:0:100 available

20:70:10 100:0:0 11:57:22 48N 49E 48N 5 2 E

Figure 16. Distribution of flakes by stratum (Zones 1:2:3) for the main excavation block at Rock Bridge Shelter. Bold print indicates units with evidence of large roots.

294 56N 50E 56N 52E 0.63 0.64

99.7

0.79 1.56

99.8 0.84 0.23

rock 0.77

52N 49E 99.9 1.99 0.99

1.05 100.0

0.42 1.48

2.47 0.58 48N 49E 48N 52E 100.1

Figure 17. Average weight in grams of flakes by excavation unit for the main excavation block at Rock Bridge Shelter.

295 56N 50E 56N 52E 17% 17%

99.7

26% 29%

99.8 17% 0%

rock 11%

52N 49E 99.9 60% 24%

40% 25% 100.0

0% 50%

30% 22 % 48N 49E 48N 52E 100.1

Figure 18. Percentage of flakes more than one gram in weight for the main excavation block at Rock Bridge Shelter.

296 56N 50E 56N 52E 26% 26%

99.7

46% 66%

99.8

5%

rock 33%

52N 49E 99.9 60% 38%

2 0 % 38% 100.0

0% 100%

40% 42% 56% 48N 49E 48N 52E 100.1

Figure 19. Percentage of flakes greater than 1.5 cm in diameter for the main excavation block at Rock Bridge Shelter.

297 High

Tool M aintenance Tool / Blank Manufacture Low Discard Rate High Export Rate High Conservation

S Reoccupied Sites 0> High Reuse/Scavenging Rate a> S a 0) a

Tool M aintenance Tool / Blank Manufacture High Discard Rate High Rejection Rate Low Conservation

Situational Repair Situational Repair Raw Material Raw Material Available Scarce Low Low % Late Stage Debitage High

Figure 20. Relationship between iate-stage debitage and debitage.tool ratio with respect to tool manufacture, conservation, exportation, rejection, and discard (after Magne 1989:20).

298 Feature Cultural Function Based on Function Based on Number Affiliation All Artifacts Lithic Artifacts

COLD OAK SHELTER FEATURES

38 Historic hearth hearth or hearth cleanings. lithic reduction 31 Middle Woodland hearth hearth, manual pick up cleaning 8 Early Woodland pit or hearth hearth or hearth cleanings 10 Early Woodland postmold hearth or hearth cleanings 22 Early Woodland pit indeterminate 34 Early Woodland postoccupational disturbance hearth or hearth cleanings 35 Early Woodland looter's shovel probe hearth or hearth cleanings 36 Early Woodland disturbed feature no lithics recovered 37 Early Woodland disturbed feature manual pick up cleaning. hearth or hearth cleanings 40 Early Woodland pit or postmold no lithics recovered 41 Early Woodland may not be a feature hearth or hearth cleanings 42 Early Woodland pit or postmold no lithics recovered 43 Early Woodland pit hearth or hearth cleanings 44 Early Woodland pit no lithics recovered 49 Early Woodland refuse dump or redeposited hearth or hearth cleanings looting material 50 Early Woodland looter's shovel probe no lithics recovered 51 Early Woodland indeterminate hearth or hearth cleanings 52 Early Woodland pit hearth or hearth cleanings 53 Early Woodland indeterminate manual pick up cleaning 54 Early Woodland pit hearth or hearth cleanings 55 Early Woodland pit or postmold manual pick up cleaning 56 Early Woodland indeterminate indeterminate 57 Early Woodland indeterminate hearth or hearth cleanings 58 Early Woodland pit indeterminate 59 Early Woodland pit hearth, lithic reduction 60 Early Woodland pit indeterminate 67 Early Woodland indeterminate indeterminate

Table 1. Cultural affiliations and functional designations for features at Gold Oak and Rock Bridge shelters. Functional designations based on all artifacts are from Gremillion (1993, 1995) (continues).

299 Table 1 (continued).

Feature Cultural Function Based on Function Based on Number Affiliation All Artifacts Lithic Artifacts

COLD OAK SHELTER FEATURES (continued)

68 Early Woodland pit or postmold no lithics recovered 24 Terminal Archaic disturbed post mold manual pick up cleaning, hearth or hearth cleanings 32 Terminal Archaic pit no lithics recovered 33 Terminal Archaic pit no lithics recovered 39 Terminal Archaic pit manual pick up cleaning, hearth or hearth cleanings 45 Terminal Archaic pit no lithics recovered 46 Terminal Archaic natural depression or no lithics recovered refuse pit 47 Terminal Archaic pit hearth or hearth cleanings 48 Terminal Archaic pit indeterminate 61 Terminal Archaic natural depression or hearth or hearth cleanings refuse pit 62 Terminal Archaic natural depression hearth or hearth cleanings 63 Terminal Archaic post mold indeterminate 64 Terminal Archaic pit no lithics recovered 65 Terminal Archaic pit indeterminate 66 Terminal Archaic pit hearth or hearth cleanings 69 Terminal Archaic pit indeterminate 70 Terminal Archaic indeterminate no lithics recovered

ROCK BRIDGE SHELTER FEATURES

1 Late Woodland mixed hearth residues and indeterminate pack rat nesting debris 2 Late Woodland surface charcoal no lithics recovered concentration 3 Late Woodland redeposited hearth and other lithic reduction materials 4 Late Woodland redeposited midden no lithics recovered 5 Late Woodland redeposited hearth sweepings residual primary deposit left or disturbed hearth after sweeping

3 0 0 Table 1 (continued).

Feature Cultural Function Based on Function Based on Number Affiliation All Artifacts Lithic Artifacts

ROCK BRIDGE SHELTER FEATURES (continued)

6 Late Woodland redeposited hearth sweepings no lithics recovered or ash 7 Late Woodland redeposited midden no lithics recovered 8 Lave Woodland redeposited midden no lithics recovered 9 Late Woodland redeposited midden no lithics recovered 1 0 Late Woodland redeposited midden no lithics recovered 1 1 Late Woodland redeposited midden no lithics recovered 1 2 Late Woodland redeposited midden no lithics recovered 1 3 Late Woodland redeposited midden indeterminate 1 4 Late Woodland redeposited midden indeterminate 1 5 Late Woodland redeposited midden residual primary deposit left after sweeping 16 Late Woodland redeposited midden indeterminate 17 Late Woodland shallow basin indeterminate 1 8 Late Woodland cylindrical pit indeterminate

30 1 FEATURE FLAKES BIFACIAL NUMBER 1 Decor 2Decor Prim Second Thin Broken Total TOOLS

8 0 2 2 3 5 12 24 0 1 0 0 0 1 1 1 2 5 0 22 0 2 3 1 2 2 1 0 0 24 0 1 2 2 3 2 10 0 31 0 1 3 0 1 1 6 0 32 0 0 0 0 0 0 0 0 33 0 0 0 0 0 0 0 0 34 0 0 1 0 0 0 1 0 35 0 0 0 0 0 1 1 0 36 0 0 0 0 0 0 0 0 37 0 1 0 2 2 1 6 0 38 0 2 2 1 3 1 9 0 39 0 3 1 1 3 2 10 0 40 0 0 0 0 0 0 0 0 41 0 0 1 0 0 1 2 0 42 0 0 0 0 0 0 0 0 43 0 1 0 0 0 1 2 0 44 0 0 0 0 0 0 0 0 45 0 0 0 0 0 0 0 0 46 0 0 0 0 0 0 0 0 47 0 0 0 0 0 0 0 0 48 0 1 0 0 0 0 1 0 49 0 0 0 0 2 0 2 0 50 0 0 0 0 0 0 0 0 51 0 0 2 0 0 1 3 0 52 0 1 1 2 0 3 7 0 53 0 0 1 1 0 3 5 0 54 0 1 0 1 0 2 4 0 55 0 0 4 5 0 2 1 1 0 56 0 1 0 0 2 1 4 0 57 0 6 1 0 3 6 1 6 0 58 0 0 0 0 0 1 1 0 59 0 0 1 2 1 1 5 1 60 1 0 0 0 1 0 2 0 61 0 0 1 1 1 0 3 0 62 0 1 6 1 4 5 1 7 0 63 0 0 0 0 0 1 1 0 64 0 0 0 0 0 0 0 0 65 0 0 0 0 0 0 0 0 66 0 0 0 0 0 0 0 0 67 0 0 0 0 0 0 0 0 68 0 0 0 0 0 0 0 0 69 0 0 0 0 0 1 1 0 70 0 0 0 0 0 0 0 0 TOTAL 1 24 33 24 34 53 169 1

Table 2. Lithic artifacts by feature, Cold Oak Shelter (continues).

3 0 2 Table 2 (continued).

FEATURE DEBRIS FIREALT SANDSTONE NUMBER Chunk Chunk Dec Shatter Shat Dec Total n wt avg

8 0 0 1 0 1 1 19.5 19.5 1 0 0 0 0 2 2 0 0 0 22 1 0 1 1 3 0 0 0 24 1 1 1 1 4 0 0 0 31 0 0 1 0 1 2 7.8 3.9 32 0 0 0 0 0 0 0 0 33 0 0 0 0 0 0 0 0 34 0 0 0 0 0 0 0 0 35 0 0 0 0 0 1 2.9 2.9 36 0 0 0 0 0 0 0 0 37 0 0 0 0 0 2 74.3 37.2 38 0 0 0 0 0 0 0 0 39 0 0 0 2 2 0 0 0 40 0 0 0 0 0 0 0 0 41 0 0 0 0 0 3 70.1 23.4 42 0 0 0 0 0 0 0 0 43 0 0 0 0 0 0 0 0 44 0 0 0 0 0 0 0 0 45 0 0 0 0 0 0 0 0 46 0 0 0 0 0 0 0 0 47 0 0 0 0 0 3 2.6 0.9 48 0 0 0 0 0 0 0 0 49 0 0 0 0 0 7 327.7 46.8 50 0 0 0 0 0 0 0 0 51 0 0 0 0 0 0 0 0 52 0 0 0 0 0 1 9.7 9.7 53 0 0 0 0 0 0 0 0 54 0 0 0 0 0 0 0 0 55 0 0 1 1 2 0 0 0 56 0 0 0 0 0 0 0 0 57 0 0 0 0 0 0 0 0 58 0 0 0 0 0 0 0 0 59 0 0 0 0 0 0 0 0 60 0 0 0 0 0 0 0 0 61 0 0 0 0 0 0 0 0 62 0 0 0 0 0 5 377.0 75.4 63 0 0 0 0 0 0 0 0 64 0 0 0 0 0 0 0 0 65 0 0 1 0 1 0 0 0 66 0 0 0 0 0 2 90.4 45.2 67 0 0 1 0 1 0 0 0 68 0 0 0 0 0 0 0 0 69 0 0 0 0 0 0 0 0 70 0 0 0 0 0 0 0 0 TOTAL 2 1 7 7 1 7 27 982.0 36.4

3 0 3 Table 2 (continued).

FEATURE FIREALT UMESTONE ARE POT HEAT TOTAL SOIL LITHIC NUMBER n w t avg CHERT UOS SPALLS LITHICS VOLUME DENSTTY

8 0 0 0 7 1 0 27 1 7 1.59 10 0 0 0 2 0 0 7 7 1.00 22 0 0 0 1 1 0 14 52.3 0.27 24 0 0 0 8 0 0 14 26.8 0.52 31 0 0 0 2 0 0 9 23.5 0.38 32 0 0 0 0 0 0 0 0 0 33 0 0 0 0 0 0 0 0 0 34 0 0 0 1 0 0 1 1 1.00 35 1 15.0 15.0 1 0 0 3 4 0.75 36 0 0 0 1 0 0 0 5 0 37 1 44.0 44.0 2 0 0 9 9.2 0.98 38 0 0 0 1 1 0 10 21.5 0.47 39 0 0 0 5 0 0 12 1 1 1.09 40 0 0 0 0 0 0 0 4 0 41 0 0 0 0 0 1 6 18 0.33 42 0 0 0 0 0 0 0 1 0 43 0 0 0 1 0 0 2 5.5 0.36 44 0 0 0 0 0 0 0 0.4 0 45 0 0 0 0 0 0 0 2.1 0 46 0 0 0 0 0 0 0 6.2 0 47 0 0 0 0 0 0 3 2 1.50 48 0 0 0 0 0 0 1 0.8 1.25 49 1 38.1 38.1 0 0 0 10 7 1.43 50 0 0 0 0 0 0 0 0 0 51 0 0 0 1 0 0 3 2 1.50 52 0 0 0 2 0 0 8 18 0.44 53 0 0 0 0 0 0 5 7.2 0.69 54 0 0 0 2 0 0 4 3 1.33 55 0 0 0 1 0 0 1 3 7.5 1.73 56 0 0 0 0 0 0 4 4.5 0.89 57 1 2.6 2.6 7 0 0 17 15.5 1.10 58 0 0 0 0 0 0 1 8 0.13 59 0 0 0 1 0 0 6 6 1.00 60 0 0 0 0 0 0 2 2 1.00 61 0 0 0 2 0 0 3 6 0.50 62 0 0 0 1 0 0 22 22.8 0.96 63 0 0 0 0 0 0 1 0.5 2.00 64 0 0 0 0 0 0 0 1 0 65 0 0 0 0 0 0 1 7.2 0.14 66 0 0 0 0 0 0 2 4 0.50 67 0 0 0 0 0 0 1 6 0.17 68 0 0 0 0 0 0 0 0 0 69 0 0 0 0 0 0 1 2 0.50 70 0 0 0 0 0 0 0 0.5 0 TOTAL 4 99.6 24.9 49 3 1 222 349 0.6 4

3 0 4 TOOLS FLAKES DEBRIS MICRODEB TOTAL % OF SAMPLE VOLUME (1) DENSITY

Feature 1 0 7 1 0 8 5.8% 74 0 0.11

Feature 3 1 40 0 59 100 72.5% ZZJi 1.39

Feature 5 0 1 0 9 1 0 7.2% 80.5 0.12

Feature 13 2 0 0 0 2 1.4% 7.0 0 29

Feature 14 0 1 0 0 1 0.7% 2.0 0 50 w o U1 Feature 15 0 0 0 9 9 6.5% 3.0 3.00

Feature 16 0 0 0 2 2 1.4% 2.0 1.00

Feature 17 2 0 0 3 5 3.6% 20 0 0 25

Feature 18 0 0 0 1 1 0.7% 6.0 0.17

TOTAL 5 49 1 83 138 100.0% 266.5 0.52

Table 3. Lithic artifacts by feature, Rock Bridge Shelter. Underlined feature fill volumes are estimated. COLD OAK SHELTER ROCK BRIDGE SHELTER

Chipped Stone 1879 83.8% 749 99.2%

Thermal Debris 234 10.4% 2 0.3%

Miscellaneous 125 5.6% 1 0.1%

Ground Stone 3 0.1% 3 0.4%

TOTAL 2241 100.0% 755 100.0%

Table 4. Composition of the Gold Oak and Rock Bridge shelter lithic assem blages.

3 0 6 MINERALS ROCKS HEMA LIMON GYP FEORE ÛT2 ALL UME SAND DOLO CHALK PWOOO CONOR META ALL

CHIPPED STONE n 0 0 0 5 2 7 1 1 0 0 0 0 0 0 1 1 P e r c e n t 0 3% 0 1% 0 .4 % 0 6% 0 .6 %

THERMAL DEBRIS 0 Û 0 Ü 1 1 1 0 1 6 6 0 0 0 1 0 1 7 7 P e r c e n t 0 4% 0 .4 % 4 3% 70 9% 0 4% 7 5 .6 %

MISCELLANEOUS n B 1 1 0 0 1 0 a 1 3 1 9 1 0 2 9 7 P e r c e n t 6 4% 0 8% 0 B"o 8 .0 % 64 8% 2 4% 0 a% 7 2% 0 8*-b 1 6% 7 7 .6 %

GROUND STONE n 0 0 0 0 0 0 1 0 2 0 0 0 0 3 P e r c e n t 33 3% 6 6 7*!o 1 0 0 %

TOTAL n a 1 1 5 3 1 8 1 0 3 1 6 6 3 9 1 1 2 2 8 8 w P e r c e n t 0 .4 % 0 .0 % 0 .0 % 0 .2 % 0 .1 % 0 .8 % 4 .6 % 7 .5 % 0 .1 % 0 .4 % 0 .0 % 0 .0 % 0 .1 % 1 2 .9 % o CHERT TOTAL ? HANSTL BOY BR KAN PEN MUL PAO GEN ALL

CHIPPED STONE 53Ü 6U9 3 4 3 1 70 1 2 5 6 2 4 3 3 6 3 7 6 1 8 6 1 1 6 7 9 P e r c e n t 2/ 1% 18 3% 9 0% 6 7*0 3 3*,o 2 a^^o 1 9*0 2 0% 0 3% 9 9 .0 % 1 0 0 %

THERMAL DEBRIS n 2 2 1 5 5 3 5 1 2 2 1 0 5 6 2 3 4 P e r c e n t 9 4% 6 4% 2 1% 1 3% 2 1% 0 4% 0 9% 0 9% 0 4% 0% 2 3 .9 % 1 0 0 %

MISCELLANEOUS 1 t 4 0 1 2 0 0 0 0 0 1 8 1 2 5 P e r c e n t 8 8% 3 2% 0% 0 8% 1 6% 0% 0% 0% 0% 0% 1 4 .4 % 1 0 0 %

GROUND STONE n 0 0 0 0 0 0 0 0 0 0 0 3 P e r c e n t 1 0 0 %

TOTAL n 5 6 3 5 2 8 3 4 8 1 7 4 1 3 2 6 3 4 5 3 6 3 6 6 1 9 3 5 2 2 4 1 P e r c e n t 2 5 .1 % 2 3 .6 % 1 5 .5 % 7 .6 % 5 .6 % 2 .6 % 2 .0 % 1 .7 % 1 .7 % 0 .3 % 6 6 .3 % 1 0 0 %

Table 5. Raw materials for the Cold Oak Shelter lithic assem blages. Minerals = hematite, limonite, gypsum, iron ore (siderite /jaspilite), quartz. Rocks = limestone, sandstone, dolostone, chalk, petrified wood, concretion, metamorphic rock. Cherts=unknown, Haney, St. Louis, Boyle, Breathitt, Kanawha, Pennsylvanian, Muldraugh, Paoli, Ste. Genevieve. EXCAVATION UNIT 510.SN 497E 510.5N 498E 510.5N 499E 510.5N 500E 510.5N 501E OTHER TOTAL VOLUME (liters) 737 725.25 412.25 623 294.5 2792

CHIPPED STONE n 395 472 285 447 139 141 1879 Percent 21.0% 25.1% 15.2% 23.8% 7.4% 7.5% 100.0% Density 0.54 0.65 0.69 0.72 0.47 0.62**

THERMAL DEBRIS n 28 57 40 68 28 13 234 Percent 12.0% 24.4% 17.1% 29.1% 12.0% 5.6% 100.0% Density 0.04 0 08 0.10 0.11 0.10 0.08**

CO MISCELLANEOUS o œ n 36 27 14 38 5 5 125 Percent 28.8% 21.6% 11.2% 30.4% 4.0% 4.0% 100.0% Density 0 05 0.04 0.03 0.06 0.02 0.04**

GROUND STONE n 0 0 0 2 1 D 3 Percent 66.7% 33.3% 100.0% Density 0.003 0.003 0.001**

TOTAL n 459 556 339 555 173 159 2241 Percent 20.5% 24.8% 15.1% 24.8% 7.7% 7.1% 100.0% Density 0.62 0.77 0.82 0.89 0.59 0.75**

Table 6. Horizontal distribution of lithics by excavation unit, Cold Oak Shelter. OTHER includes lithics found outside the 1994 excavation block. * Densities cannot be calculated for lithics found outside the excavation block as sediment volumes were not recorded. ** Densities in the total column exclude those lithics found outside the 1994 excavation block. LEVEL FEATURES ZONEV ZONE III ZONE II ZONEI HISTORIC UNKNOWN* TOTAL* VOLUME (liters) 304.25 98 386.5 626.25 1103 8 266 2792

CHIPPED STONE n 1 75 6 197 509 715 1 0 267 1879 Percent 9.3% 0.3% 10.5% 27.1% 38.1% 0.5% 14.2% 100.0% Density 0.58 0.06 0.51 0.81 0.65 1.25 0.47 0.62

THERMAL DEBRIS n 32 0 32 48 88 2 32 234 Percent 13.7% 13.7% 20.5% 37.6% 0.9% 13.7% 100.0% Density 0.11 0.08 0.08 0.08 0.25 0.07 0.08 w o

GROUND STONE n 0 0 2 0 0 0 1 3 Percent 66.7% 33.3% 100.0% Density 0.005 0.004 0.001

TOTAL n 214 6 246 586 858 12 319 2241 Percent 9.5% 0.3% 11.0% 26.1% 38.3% 0.5% 14.2% 100.0% Density 0.70 0.06 0.64 0.94 0.78 1.50 0.60 0.75

Table 7. Vertical distribution of I it flics by cultural zone, Cold Oak Stielter. * Densities exclude those lithics found outside the 1994 excavation block. PERKX) TERM. ARCHAIC EARLY WOODLAND MID. WOODLAND HISTORIC UNKNOWN* TOTAL* VOLUME (liters) 587.25 625 27.5 29.5 1522.75 2792

CHIPPED STONE n 247 490 7 1 0 1 125 1879 Percent 13.1% 26.1% 0.4% 0.5% 59.9% 100.0% Density 0.42 0.78 0.25 0.34 0.63 0.62

THERMAL DEBRIS n 43 57 1 2 131 234 Percent 18.4% 24.4% 0.4% 0.9% 56.0% 100.0% Density 0.07 0.09 0.04 0.07 0.08 0.08

G3 MISCELLANEOUS o n 1 8 26 0 0 81 125 Percent 14.4% 20.8% 64.8% 100.0% Density 0.03 0.04 0.05 0.04

GROUND STONE n 2 0 0 0 1 3 Percent 66.7% 33.3% 100.0% Density 0.003 0.001 0.001

TOTAL n 31 0 573 8 1 2 1515 2241 Percent 13.8% 25.6% 0.4% 0.5% 67.6% 100.0% Density 0.53 0.92 0.29 0.41 0.76 0.75

Table 8. Vertical distribution of lithics by cultural period, Cold Oak Shelter. * Densities exclude those lithics found outside the 1994 excavation block. COLD OAK SHELTER ROCK BRIDGE SHELTER n % n %

TOOLS AND TOOL FRAGMENTS 23 1.2% 26 3.5%

BIFACIAL TOOLS 1 1 0.6% 7 0.9% BIFACE FRAGMENTS 10 0.5% 0 0.0% MARGINALLY MODIFIED FLAKES 1 0.1% 1 1 1.5% UTIUZED FLAKES 1 0.1% 8 1.1%

DEBfTAGE 1856 98.8% 723 96.5%

CORES 6 0.3% 2 0.3%

FLAKES 1694 90.2% 452 60.3% Primary Decortication 17 0.9% 1 0.1% Secondary Decortication 276 14.7% 51 6.8% Primary 240 12.8% 25 3.3% Secondary 369 19.6% 95 12.7% Thinning 336 17.9% 106 14.2% Broken 448 23.8% 174 23.2% Blade / Bladelet 8 0.4% 0 0.0%

DEBRIS 156 8.3% 31 4.1% Decortication Chunk 24 1.3% 3 0.4% Chunk 23 1.2% 4 0.5% Decortication Shatter 43 2.3% 3 0.4% S h atter 66 3.5% 21 2.8%

MICRODEBfTAGE 0 0.0% 238 31.8%

TOTAL 1 8 7 9 100.0% 749 100.0%

Table 9. Chipped-stone artifacts for the Cold Oak Shelter and Rock Bridge Shelter lithic assem blages.

31 1 510.5N 497E 510.5N 498E S1G.5N 499E S10.5N SOGE 51G.5N SG1E OTHER* TOTAL

BIFACIAL TOOLS

Projectile Points n 0 3 0 2 0 1 6 Percent 50% 33% 17% 100%

Hatted Biface Frags n 1 3 0 1 0 0 5 Percent 20% 60% 20% 100%

All Bifacial Tools

03 n 1 6 0 3 0 1 1 1 Percent 9% 55% 27% 9% 100% ro BIFACE FRAGMENTS n 0 3 1 3 1 2 1 0 Percent 30% 10% 30% 10% 20% 100%

MARG. MOD. FLAKES n 0 1 0 0 0 0 1 Percent 100% 100%

UTIUZED FLAKES n 0 0 1 0 0 0 1 Percent 100% 100%

TOTAL TOOLS n 1 10 2 6 1 3 23 Percent 4% 43% 9% 26% 4% 13% 100%

Table 10. Horizontal distribution of chipped-stone tools and tool fragments by excavation unit, Cold Oak Shelter. " Other includes specimens found outside the 1994 excavation block. FEATURES ZONE III ZONE II ZONEI UNKNOWN* TOTAL

BIFACIAL TOOLS

Projectile Points n 1 0 0 4 1 6 Percent 17% 67% 17% 100%

Hafted Biface Fragments n 0 1 1 3 0 5 Percent 20% 20% 60% 100%

All Bifacial Tools n 1 1 1 7 1 1 1

CO Percent 9% 9% 9% 64% 9% 100%

CO BIFACE FRAGMENTS n 0 0 2 6 2 10 Percent 33% 100% 33% 100%

MARG. MOD. FLAKES n 0 0 1 0 0 1 Percent 100% 100%

UTILIZED FLAKES n 0 0 1 0 0 1 Percent 100% 100%

TOTAL TOOLS n 1 1 5 13 3 23 Percent 4% 4% 22% 57% 13% 100%

Table 11. Vertical distribution of chipped-stone tools by cultural zone, Cold Oak Shelter. No tools were recovered from Zone V or Historic deposits. * Unknown includes specimens found outside the 1994 excavation block and from cleaning unit floors and profiles. TERM. ARCHAIC EARLY WOODLAND UNKNOWN* TOTAL

BIFACIAL TOOLS

Projectile Points n 0 1 5 6 Percent 17% 83% 100%

Hafted Biface Fragments n 1 1 3 5 Percent 20% 20% 60% 100%

All Bifacial Tools n 1 2 8 11 Percent 9% 18% 73% 100%

BIFACE FRAGMENTS n 0 2 8 10 Percent 20% 80% 100%

MARG. MOD. FLAKES n 0 1 0 1 Percent 100% 100%

UTIUZED FLAKES n 0 1 0 1 Percent 100% 100%

TOTAL TOOLS n f 6 16 23 Percent 4% 26% 70% 100%

Table 12. Vertical distribution of chipped-stone tools by cultural period, Cold Oak Shelter. No tools were recovered from Middle Woodland and Historic contexts. " Unknown includes specimens found outside the 1994 excavation block and from cleaning unit floors and profiles. ? BOYLE KAN HANEY ST LOUIS BREATH PENN PAOU TOTAL

BIFACIAL TOOLS

Projectile Points n 0 2 2 0 0 1 0 1 6 P erc en t 33% 33% 17% 17% 100%

Hafted BIface Frags n 1 0 0 1 0 1 1 1 5 P e rc en t 20% 20% 20% 20% 20% 100%

All Bifacial T ools n 1 2 2 1 0 2 1 2 1 1 w P e rc en t 9% 18% 18% 9% 18% 9% 18% 100% m BIFACE FRAGMEhfTS n 3 3 1 1 1 0 0 1 10 P e rc e n t 30% 30% 10% 10% 10% 10% 100%

MARG. MOD. FLAKES n 0 0 0 0 1 0 0 0 1 P e rc en t 100% 100%

UTIUZED FLAKES n 1 0 0 0 0 0 0 0 1 P e rc en t 100% 100%

TOTAL TOOLS n 5 5 3 2 2 2 1 3 2 3 P e rc en t 22% 22% 13% 9% 9% 9% 4% 13% 100%

Table 13. Chert distribution for chipped-stone tools and tool fragments, Cold Oak Shelter. Breath = Breathitt, Kan = Kanawha, Penn = Pennsylvanian (Breathitt and Kanawha), IVIuldr = Muldraugh, Ste Gen = Ste. Genevieve. COLD OAK SHELTER ROCK BRIDGE SHELTER SPEOMEN NUMBER 574 650 894c 1237 1309a 544*" 71 250 251

POINT TYPE Madison? Gary/Cogs Adena? Lowe Adena? Ltl Bear Ck? Bakers Ck Perforator Ctiesser

TOTAL LENGTH 3.28" 5.13 3.53" 2.84" 2.89" - 3.49 4.59 3.24

BLADE LENGTH - 4.09 2 23" 1.82" -- 2.26 3.86 2.15

STEM LENGTH - 1.05 1.28 0.95 - 1.21 1.23 0.73 1.09

BLADE EDGE LENGTH 6 6" 8.40 - 3.8" -----

BLADE THICKNESS 1.00 0.85 0.80 0.66 -- 0.62 0.87 0.85

STEM THICKNESS - 0.54 0.55 0.50 0.60" 0.52 ---

NECK THICKNESS - 0.60 0.73 0.56 -----

TIP THICKNESS 0.31" 0.27 - 0.24" - - ---

BLADE WIDTH - 2.14 1.80 1.89 -- 2.74 1.18 1.79

STEM WIDTH - 1.43 1.32 1.25 2.32 1.28 ---

BASE WIDTH 1.49 -- 1.30 - - 2.48 1.67 1.52

NECK WIDTH - 1.51 1.39 1.12 - - 2.01 1.20 0.98

WEIGHT 3.67" 10.94 4.87* 3.46" 3.32" 0.92" 5.68 4.54 4.75

Table 14. Metric attributes of projectile points, perforator, and point fragments from Cold Oak Slieiter and Rock Bridge Stielter. All dimensions are in cm; weigtits are in grams. * Measurements do not represent maximum dimensions or weigtits due to breakage of specimens. " Ttiis is a field sample number; ttie artifact was not assigned a specimen number. ? HANEY ST L BOYLE BREA KAN PEt#t MULD PAOU STE G LS FEORE TOTAL

PRIMARY DECORT n 2 2 8 0 3 2 0 0 0 0 0 0 17 P e rc e n t 0.1 % 0.1% 0.5% 0.2% 0.1% 1.0%

SECONDARY DECORT n 65 54 75 21 25 12 6 8 6 2 1 1 2 7 6 P e rc e n t 3.8% 3.2% 4.4% 1.2% 1.5% 0.7% 0.4% 0.5% 0.4% 0.1% 0.1% 0.1% 16.3%

PRIMARY n 39 105 51 9 8 12 8 1 4 0 3 0 2 4 0 P e rc e n t 2.3% 6.2% 3.0% 0.5% 0.5% 0.7% 0.5% 0.1% 0.2% 0.2% 14.2%

SECONDARY w n 103 94 83 41 21 6 5 7 4 0 3 2 3 6 9 N P e rc e n t 6.1% 5.5% 4.9% 2.4% 1.2% 0.4% 0.3% 0.4% 0.2% 0.2% 0.1% 21.8%

THINNING n 99 97 52 39 14 1 2 3 8 10 1 0 1 3 3 6 P e rc e n t 5.8% 5.7% 3,1% 2.3% 0.8% 0.7% 0.2% 0.5% 0.6% 0.1% 0.1% 19.8%

BROKEN n 160 114 55 36 36 1 1 17 8 7 2 2 0 4 4 8 P e rc e n t 9.4% 6.7% 3.2% 2.1% 2.1% 0.6% 1.0% 0.5% 0.4% 0.1% 0.1% 26.4%

BLADE/BLADELET n 2 4 2 0 0 0 0 0 0 0 0 0 8 P e rc e n t 0.1% 0.2% 0.1 % 0.5%

TOTAL FLAKES n 4 7 0 4 7 0 3 2 6 1 4 6 1 0 7 5 5 3 9 3 2 31 5 9 4 1694 P e rc e n t 27.7% 27.7% 19.2% 8.6% 6.3% 3.2% 2.3% 1.9% 1.8% 0.3% 0.5% 0.2% 100%

Table 15. Raw material distribution for Cold Oak Shelter flake sample, listed by flake type. St L = St. Louis, Brea = Breathitt, Kan = Kanawha, Penn = Pennsylvanian, Muld = Muldraugh, Ste G = Ste. Genevieve, LS = Limestone, FE Ore = Iron Ore. FLAKETYPE COMPLETE PROXIMAL MEDIAL DISTALTOTAL

PRIMARY DECORTICATION Sample Size 3 4 5 5 17 Percentage 0.2% 0.2% 0.3% 0.3% 1.0%

SECONDARY DECORTICATION Sample Size 54 98 68 56 276 Percentage 3.2% 5.9% 4.1% 3.4% 16.6%

PRIMARY Sample Size 61 144 18 1 7 240 Percentage 3.7% 8.6% 1.1% 1.0% 14.4% w SECONDARY 00 Sample Size 69 288 7 5 369 Percentage 4.1% 17.3% 0.4% 0.3% 22.1%

THINNING Sample Size 86 235 8 7 336 Percentage 5.2% 14.1% 0.5% 0.4% 20.2%

BROKEN Sample Size 0 0 311 137 448 Percentage 18.7% 8.2% 26.9%

BLADE/ BLADELET Sample Size 1 3 2 2 8 Percentage 0.1% 0.2% 0.1% 0.1% 0.5%

TOTALS Sample Size 274 772 419 229 1694 P e rc e n ta g e 16.2% 45.6% 24.7% 13.5% 100.0%

Table 16. Distribution of flake fragment types for tfie Cold Oak Stielter flake sample. NORTHERN BLOCK MIDDLE BLOCK SOUTHERN BLOCK OUTSIDE BLOCKS TOTAL ______Surface Excavation Surface Excavation Surface Excavation ______S u rface ______

CHIPPED-STONE 0 ^ 0 1 0 21 3 26 TOOLS

CHIPPED-STONE 4 19 0 13 2 0 6 4 0 27 723 DEBITAGE

TOTAL CHIPPED 4 20 0 14 20 661 3 0 749 STONE

GROUND 000000 3 3 CO STONE ID FIRE-ALTERED 000002 0 2 CHERT

MISCELLANEOUS 0 0 0 0 0 1 0 1 LITHICS

BLOCK 4 2 0 0 14 2 0 664 33 755 TOTAL

PERCENT OF 1% 3% 0% 2% 3% 88% 4% 100% ASSEMBLAGE

ESTIMATED 115.5 265 1407.5 1788.0 VOLUME (liters)

DENSITY 0.21 0.05 0.49 0.42

Table 17. Horizontal distribution of lithic artifacts by excavation block and by method of recovery, Rock Bridge Shelter. NORTHERN BLOCK MIDDLE BLOCK SOUTHERN BLOCK ALL BLOCKS Count Percent Count Percent Count Percent Count Percent

ZONEI 1 0 42% 1 9% 1 55 32% 1 6 6 32% Ashy Midden

ZONE II 14 58% 0 0% 320 66% 334 64% GO IV) Silty Midden O

ZONE III 0 0% 10 91% 8 2% 1 8 3% Sandy Subsoil

ALL ZONES C ount 2 4 11 4 8 3 5 1 8 P e rc en t 5% 2% 93% 100%

Table 18. Vertical distribution of lithic artifacts by excavation block, Rock Bridge Shelter. The sample size is 518 (69% of the total assemblage), which includes the lithics that could be assigned to a sediment zone. LITHIC CATEGORY n HANEY PAOU BOYLE BREA KAN ST L STE G HEMA SS QTZ LS ?

GROUND STONE 3 • ■ • • 67% • - 33%

FIRE-ALTEHED CHERT 2 -- 50% ■ ■ ■ 50%

MISC. LITHICS 1 100% ■ - • ■ ■

CHIPPED STONE 7 4 9 53% 21% 9% 2% 2% 1% <1% <1% <1% <1% 9%

T ools 2 6 31% 58% 4% 4% 4% • -•

Bilaces 7 14% 71% 14% • - w ro Marg. Mod. Flakes 1 1 45% 36% 9% 9% • ■

Utilized Flakes a 25% 75% - -•

Debltage 723 54% 20% 9% 2 % 2% 1% <1% <1% <1% <1% 9%

Cores 2 50% 50% • ■

MIcrodebitaga 238 60% 8% 6% 2% ■ • ■ 2% 2% - 20%

Macrodebltage 483 51% 26% 10% 3% 2% 2% 1% 1% <1% • 4%

Flakes 452 53% 25% 10% 3% 2% 2% 1% <1% <1% • 4%

Debris 31 28% 29% 23% 6% 3% • • 3% • • 10%

TOTAL 755 51% 22% 9% 2% 2% 2% 1% 1% 1% 1% <1% 9%

Table 19. Raw material distribution by lithic category, Rock Bridge Shelter. Brea = Breathitt. Kan = Kanawha, St L = St. Louis, Ste G = Ste. Genevieve, Hema = Hematite, SS = Sandstone, Qtz = Quartz, LS = Limestone, ? - Unknown. BIFACIAL MARGINALLY UTIUZED TOTAL % OF SAMPLE TOOLS MOD. FLAKES FLAKES

HORIZONTAL PROVENIENCE

Southern Block 6 8 7 21 80.8%

Middle Block 0 1 0 1 3.8%

Northern Block 0 1 0 1 3.8%

CJ Outside Block 1 1 1 3 11.5% ro ru T otal 7 t 1 8 26 100%

VERTICAL PROVENIENCE

Zone 1 (Ashy Midden) 1 1 2 4 15.4%

Zone II (Silty Midden) 5 6 3 14 53.8%

Zone III (Subsoil) 0 2 2 4 15.4%

Unknown 1 2 1 4 15.4%

T otal 7 11 8 26 100%

Table 20. Horizontal and vertical distribution of chipped-stone tools and tool fragments, Rock Bridge Shelter. FLAKES DEBRIS TOTAL % OF SAMPLE

HORIZONTAL PROVENIENCE

Southern Block 394 28 4 2 2 87.4%

Middle Block 1 3 1 1 4 2.9%

Northern Block 2 2 1 23 4.8%

Outside Block 16 0 16 3.3%

Unknown 7 1 8 1.7%

Total 4 5 2 31 4 8 3 100%

VERTICAL PROVENIENCE

Zone 1 (Ashy Midden) 9 8 9 107 22.2%

Zone II (Silty Midden) 139 20 159 32.9%

Zone III (Subsoil) 1 4 0 14 2.9%

Unknown 201 2 2 0 3 42.0%

Total 4 5 2 31 4 8 3 100%

Table 21. Horizontal and vertical distribution of chipped-stone flakes and debris, Rock Bridge Shelter.

3 2 3 ? HANEY PAOU BOYLE BREA S T L STE G KAN SS HEMA TOTAL %

PRIMARY DECORT n 0 1 0 0 0 0 0 0 0 0 1 0.2% P e rc e n t 0.1% 0.1%

SECONDARY DECORT n 1 23 17 5 2 2 0 1 0 0 51 11.3% P e rc e n t 0.1% 1.4% 1.0% 0.3% 0.1% 0.1% 0.1% 3.0%

PRMARY n 1 13 8 1 1 1 0 0 0 0 2 5 5.5% P e rc e n t o.r/o 0.8% 0.5% 0.1% 0 1% 0.1% 1.5% CO ro SECONDARY n 4 53 19 1 0 1 2 0 3 0 3 9 5 21.0% P e rc e n t 0.2% 3.1% 1.1% 0.6% 0.1% 0.1% 0.2% 0.2% 5.6%

THINNING n 3 52 31 7 3 3 3 4 Ü 0 1 0 6 23.5% P e rc e n t 0.2% 3.1% 1.8% 0.4% 0.2% 0.2% 0.2% 0.2% 6.3%

BROKEN n 7 96 37 20 5 2 3 2 1 1 1 74 38.5% P e rc e n t 0.4% 5.7% 2.2% 1.2% 0.3% 0.1% 0.2% 0.1% 0.1% 0.1% 10.3%

TOTAL FLAKES n 16 2 3 8 112 4 3 12 1 0 6 1 0 1 4 4 5 2 100% P e rc e n t 3.5% 52,7% 24.8% 9.5% 2.7% 2.2% 1.3% 2.2% 0.2% 0.9% 100% 100%

Table 22. Raw material distribution for Rock Bridge Shelter flake sample, listed by flake type. Brea = Breathitt, Kan = Kanawha, Ste Gen = Ste. Genevieve, SS = Sandstone, Hema = Hematite. C om plete Proxim al Medial/Distal Debris

COLD OAK SHELTER

Entire Shelter 15% 42% 35% 8%

Early Woodland 13% 40% 39% 9%

Terminal Archaic 16% 45% 34% 5%

ROCK BRIDGE SHELTER

Macrodebltage 15% 40% 38% 7%

Microdebitage 15% 47% 38% 1%

Combined Samples 15% 42% 38% 5%

PRENTISS & ROMANSKI

Biface Manufacture Untrampled 36% 23% 37% 4% Trampled 16% 29% 50% 4%

End Scraper Manufacture Untrampled 31% 23% 36% 10% Trampled 17% 37% 37% 9%

Block Gore Reduction Untrampled 27% 17% 23% 33% Trampled 13% 19% 45% 23%

Spheroidal Core Reduction Untrampled 18% 20% 28% 35% Trampled 21% 21% 47% 12%

SULLIVAN AND ROSEN

Core Reduction high low low high

Shaped-Tool Manufacture low high high low

Table 23. Flake fragment and debris percentages for Cold Oak and Rock Bridge assemblages, Prentiss and Romanski's (1989) experiments, Sullivan and Rosen's (1985) model, Ingbar's et. al. (1989) experiments, Tomka's (1989) experiments, and Baumler and Downum's (1989) experiments (continues).

3 2 5 Table 23 (continued).

Complete Proximal Medial/Distal Debris

INGBAR ET. AL.

Biface Production 45%-60% 10%-20% 25%-30% 5%

TOMKA

Core to Blade 49% 7% 38% 6%

Core to Biface 34% 19% 45% 2 %

Flake to Hafted Biface 58% 13% 28% 1%

BAUMLER AND DOWNUM

Core Reduction 11% 57% combined 32%

Scraper Manufacture 50% 45% combined 5%

3 2 6 DAMAGE ASSEMBLAGE SUBSTRATE VERTICAL LATERAL % Com plete Edge Damage FlakeiWeight Experimental

Late Woodland yes yes yes yes possible no yes

CO N) Middle Woodland unknown no unknown unknown unknown unknown unknown

Early Woodland unknown no no no no no yes

Terminal Archaic unknown no no yes no no yes

Table 24. Summary of formation processes findings for the Terminal Archaic, Early Woodland, and Middle Woodland lithic assem blages from Cold Oak Shelter and the Late Woodland lithic assem blage from Rock Bridge Shelter. VARIABLES THAT CONTROL MANUFACTURING ACTIVITIES ACTIVITIES AND NATURE OF PRODUCT GROUP STAGE OR STEP PRODUCT GROUP

Acquisition of selective collecting material visibility or exposure selected raw material, such as: Raw Material

quarrying nature of matrix river cobbles

importation kinds of raw material + upland chert nodules

acquisition techniques selective criteria + auxiliary manipulations such 05 as heat treatment, storage, ru 00 submersion +

Core Preparation and shape the core and discard size of acquired material flakes, including cortical flakes Initial Reduction the debltage detach and retain flakes and quality of acquired material cores discard the core shape and use both core and shape of acquired material "nondescript debris" like chunk debltage and nature of surfaces and shatter abundance of acquired material + "rejects"

condition of acquired material

knappers' objectives

Table 25. Collins' stages of chipped stone tool manufacture and seven product groups; based on Collins (1975:19-32). + Factors under consideration in this study. Not indicated by Collins (1975); added by author for clarification (continues). Table 25 (continued).

VARIABLES THAT CONTROL MANUFACTURING ACTIVITIES ACTIVITIES AND NATURE OF PRODUCT GROUP STAGE OR STEP PRODUCT GROUP

Optional Primary outline shaping properties of materials + preforms Trimming

transection shaping desired properties in products rejects / rejected bifaces

edge shaping flakes and other debltage discard finished implements

CO unshaped flake Implements ro «D unshaped core Implements biface fragments

Optional Secondary bifacial edge trimming nature of preforms from debltage-llke bifacial retouch Trimming and Shaping processes, such as: previous step flakes desired product and product notching rejects / rejected bIfaces properties serrating finished objects denticulating shaped implements

beveling bIfaces

straightening edges biface fragments Table 25 (continued).

VARIABLES THAT CONTROL MANUFACTURING ACTIVITIES ACTIVITIES AND NATURE OF PRODUCT GROUP STAGE OR STEP PRODUCT GROUP

(Intervening step) artifact use not given worn objects

recycling damaged objects

recycled objects untrimmed blades suitable for use

CO CO Optional Maintenance rejuvenation like resharpening o nature of dam age + rejuvenated objects and Modification edges ^ modification by trimming and nature of wear ^ + modified objects shaping ^ debitage such as scraper renewal knappers' purposes or needs flakes raw material availability +

(intervening step) specialized disposal not given burial goods

potlatch goods

offerings ? ALL ? ALL HANEY ST LOUIS PAOU BOYLE BREATH STE GEN KAN MULDR ? TOTAL LOCAL LOCAL EXOTIC EXOTIC

LATE WOODLAND n 238 10 112 43 12 6 0 421 10 0 0 10 16 447 Percent 53.2% 2.2% 25.1% 9.6% 2.7% 1.3% 94.2% 2,2% 2.2% 3.6% 100%

MIDDLE WOODLAND n 2 0 0 1 0 0 0 3 0 0 0 0 4 7 Percent 28.6% 14.3% 42.9% 57.1% 100% w EARLY WOODLAND w n 86 74 4 33 68 1 9 2 7 5 1 1 3 5 1 9 149 443 Percent 19,4% 16.7% 0.9% 7.4% 15,3% 0.2% 2.0% 62.1% 2.5% 0.7% 1.1% 4.3% 33.6% 100%

TERMINAL ARCHAIC n 56 59 2 14 5 2 4 142 6 0 1 7 77 226 Percent 24.8% 26.1% 0.9% 6.2% 2.2% 0.9% 1.8% 62.8% 2.7% 0,4% 3.1% 34,1% 100%

TOTAL n 382 143 118 91 85 9 13 841 27 3 6 3 6 246 1 123 Percent 34.0% 12.7% 10.5% 8.1% 7.6% 0.8% 1,2% 74.9% 2.4% 0.3% 0.5% 3.2% 21,9% 100.0%

PEARSON'S r 0.64 -0.31 0.53 0.95 0.57 0.34 - 0.86 0.98 0.49 0,68 -

Table 26. Local and exotic chert distribution by occupation period. Late Woodland sample is from Rock Bridge Shelter; all other samples are from Cold Oak Shelter. Breath = Breathitt, Ste Gen = Ste. Genevieve, Kan = Kanawha, Muldr = Muldraugh, ? = Unknown. ACTIVITY PREDICTED TERMINAL EARLY MIDDLE LATE OR TASK RESIDUE ARCHAIC WOODLAND WOODLAND WOODLAND

Providing FOR FCR pen pen S h elter Structures Structures Structures Daub

Food FCR FCR FCR FCR FCR Processing Pitted Cobbles Pitted Cobbles and Food Manos. Metates S to rag e Soapstone Bowls

Food Remains Food Remains Food Remains Food Remains Food Remains Storage Pits Storage Pits Storage Pits Are Pits Hominy Holes Ceramic Vessels

Fabrication FCR R R FCR RR FCR and ProjPts/KnIves ProjPts/Knlves ProjPts/Knlves ProjPts/Knlves Processing Utilized Flakes Utilized Flakes Utilized Rake Utilized Flakes organic Blades Blades Blades materiais End Scrapers End Scrapers (bone or wood) Spokeshaves Spokeshave Gravers, Burins Graver Graver Perforator Drills Drills Wedges, Axes Celts, Adzes Celt

Organic Tools Organic Tools Organic Tools Organic Tools Cordage/Leattier Cordage Cordage Cordage Basketry Basketry Basketry

B utchering FCR FCR R R FCR R R and Hide ProjPts/Knlves ProjPts/Knlves ProjPts/Knlves ProjPts/Knlves Preparation Utilized Flakes Utilized Flakes Utilized Flakes Utilized Flakes Blades Blades Blades Bifacial Knives Choppers

Faunal Remains Faunal Remains Faunal Remains Faunal Remains Faunal Remains Smudge Pits

Table 27. Predicted lithic and nonlithic residues for activities or tasks (after Ledbetter and O’Steen 1991), and observed residues from the Cold Oak Shelter Terminal Archaic, Early Woodland, and Middle Woodland assemblages for both the 1994 and 1984 seasons and from the Rock Bridge Shelter Late Woodland assemblage (continues).

3 3 2 Table 27 (continued).

ACTivrrv PREDICTED TERMINAL EARLY MIDDLE LATE OR TASK RESIDUE ARCHAIC WOODLAND WOODLAND WOODLAND

Lithic PP/K Fragments PP/K Fragments PP/K Fragments PP/K Fragments Maintenance Bit. Thin Flakes Bit. Thin Flakes Bit. Thin Flakes Bit. Thin Flakes Bit. Thin Flakes Hammerslones Hammerslones Hammerslones Flake Tool Frag s Abraders

Litttic Cores Cores Cores Cores Manufacture Debitage Debitage Debitage Debitage Debitage Untinished Bitaces Untinished Bitaces? Untinished Bitaces? Untinished Bitaces? Hammerslones Hammerslones Hammerslones Abraders Anvils Pitted Cobbles Pitted Cobbles Ground Stone Ground Stone Ground Stone Ground Stone Pretorms Pretorms? Pretorms? Pretorms?

Hunting ProjPts/Knives ProjPts/Knives ProjPts/Knives Atlatl/Bola Wts Atlatl Fragment?

Fishing Utilized Flakes Utilized Flakes Utilized Flakes

Cordage Cordage Cordage Rsh Remains

Non-Lithic Celts Celt Procurement Grooved Axes

Lithic Axes, Picks Procurement Test Cores

Personal Burials, Pipes S ta tu s Ochre, Hematite Maintenance Ornaments and Social Burial Goods A ctivity Public Construct'n

3 3 3 Terminal Archaic Early Woodland Middle Woodland Late Woodland Cold Oak Shelter Cold Oak Shelter Cold Oak Shelter Rock Bridge Shelter

ALLUTHtCS n 291 549 8 755 Density 0.50 0.88 0.29 0.42

ALL LITHICS - 1994 + 1984 n 806 1071 8 755 Density > 0.50 < 0.88 0.29 0.42

CHIP TOOLS - 1994 +1984 n 26 16 0 26 03 03 Density > 0.002 < 0.010 0.015 # Tool Types 1 3 6 Tool Diversity 0.15 0.31 0.23

HEATTREATMENT Percent 5.3% 9.1% 0.0% 6.0% Density 0.02 0.07 0.02

RAW MATERIAL DIVERSITY Richness 13/20 11/20 3/20 1 1/20 Evenness low low high low

CHERT DIVERSITY Richness 7 /9 8 /9 2 /9 7/9 Evenness low low high low

Table 28. Lithic indicators of occupational intensity for Cold Oak Sfieiter's Terminal Archaic, Early Woodland, and Middle Woodland assemblages and Rock Bridge Shelter's Late Woodland assemblage. Cold Oak samples are for 1994 assemblages only, unless otherwise indicated (continues). Table 28 (continued).

Terminal Archaic Early Woodland Middle Woodland Late Woodland Cold Oak Shelter Cold Oak Shelter Cold Oak Shelter Rock Bridge Shelter

LOCAL CHERTS n 142 275 3 421 Percent 95.3% 93.5% 100.0 % 97.7%

EXOTIC CHERTS n 7 19 10 Density 4.7% 6.5% 2.3%

CO CO %EARLY:%LATE STAGE c n Local Cherts - 1994 only 18:82 20:80 33:67 12:88 Exotic Cherts - 1994 only 14:86 29:71 0 10:90 All Cherts - 1994 only 18:82 20:80 33:67 12:88 All Cherts - 1994 + 1984 17:83 17:83 33:67 12:88

DEBITAGE INDEX - 1994 ONLY Low 51% 50% 43% 46% High 76% 79% 57% 80%

DEBITAGE INDEX - 1994+1984 Low 80% 78% 43% 46% High 85% 86% 57% 80%

FLAKE FRAGMENT TYPE Initial Reduction Flakes 21% 23% 0% 15% Shaped-tooi Manuf. Flakes 79% 77% 100% 85% Table 28 (continued).

Terminal Archaic Early Woodland Middle Woodland Late Woodland Cold Oak Shelter Cold Oak Shelter Cold Oak Shelter Rock Bridge Shelter

PLATFORM MORPHOLOGY Lipping 88% 84% 83% 55% Faceting 28% 26% 33% 32% Preparation 19% 17% 17% 29% Cortex 11% 10% 0% 6%

U) BIFACE INDEX - 1994 ONLY w Low 24% 22% 50% 28% High 45% 44% 57% 56%

BIFACE INDEX - 1994 + 1984 Low 11% 12% 50% 28% High 17% 21% 57% 56%

TOOL:DEBITAGE RATIO 1994 only 1:246 1:81 0 1:19 1994 + 1984 1:28 1:60 0 1:19

REDUCTION/CONSERVATION Percent Late-Stage Debitage 24% to 49% 20% to 49% 43% to 57% 17% to 56% Debitage.-Tool 1.000 1.000 1.000 0.095

BULB FISSURES/SCARS 13% 8% 0% 20% OCCUPATION LTTHIC LTTHIC DIMENSION INDICATOR RESULTS

DURATION Density EW > TA > LW

Tool Diversity EW > LW > TA

Heat Treatment EW > LW > TA

Exotic Percentages EW > TA > LW

FREQUENCY Stratigraphy EW and TA > LW and Densities

RANGE OF ACTIVITIES

Thermal Alteration Heat Treatment EW > LW > TA

Raw Materials Exotic Percentages EW > TA > LW

Haney and Paoli LW > TA > EW

Intensity of Production Density EW > TA > LW

Debitage Index EW and TA > LW

Reduction Stage Platform Lipping TA and EW > LW

Platform Preparation LW > TA and EW

ToohDebitage (1994) LW > EW > TA

ToohDeb (1994+1984) LW > TA > EW

Biface Index LW > TA and EW

EarlyzLate Debitage no differences

EarlyzLate Local Cherts EW > TA > LW

EarlyzLate Exotics EW > TA > LW

Flake Fragment Types no differences

Reduction Technique Platform Lipping TA and EW > LW

Bulb Fissures LW > EW and TA

GROUP SIZE Density EW > TA > LW

Table 29. Summary of lithic indicators of occupational intensity. TA = Terminal Archaic, EW = Early Woodland, LW = Late Woodland.

3 3 7 IT) O ) S i < T i - ’-CNjco'

STRATA Ash Deposits + + + + + + + + Vegetal Beds + - + + + Sand-Clay Beds - - - + + Kitchen-Midden + + + + + + +

FEATURES Burials - - - + + Post Molds + - + - Hominy Holes + - - + + Fire Pits - - + - + Hearths/Hearth Deposits + + - - Storage Pits + Miscellaneous Pits +

LITHIC ARTIFACTS Projectile Points/Knives + + + + + + Other Bifaces + + + + + + Scrapers + + + Ground Stone + + + Fire-altered Rock + + + Awls + + + Hoes + Hammerstones + + Celts + - +

OTHER Ceramics + + + + + + Uncarbonized Botanicals + - + + + + Carbonized Botanicals + + + + + + Faunal Remains + + + + + +

Table 30. Comparison of stratigraphy, features, and artifactual remains for selected shelters in the study area (continues).

3 3 8 Table 30 (continued). f-ojco'a-incot^oo o) tnmininmiotnin m

STRATA Ash Deposits + + + + Vegetal Beds + Sand-Clay Beds - - + Kitchen-Midden + + +

FEATURES Burials + Post Molds + Hominy Holes + Fire Pits Hearths/Hearth Deposits + Storage Pits + Miscellaneous Pits +

LITHIC ARTIFACTS Projectile Points/Knives + + Other Bifaces + + Scrapers + + Ground Stone + Fire-altered Rock + Awls + Hoes + Hammerstones Celts +

OTHER Ceramics + + Uncarbonized Botanicals + Carbonized Botanicals + + Faunal Remains + +

3 3 9 Table 30 (continued).

STRATA Ash Deposits + + + + + + + Vegetal Beds + + + + Sand-Clay Beds - + + Kitchen-Midden + + + + + +

FEATURES Burials Post Molds + Hominy Holes Fire Pits Hearths/Hearth Deposits + Storage Pits + Miscellaneous Pits +

LITHIC ARTIFACTS Projectile Points/Knives + 4- + Other Bifaces + + Scrapers + Ground Stone + + Fire-altered Rock Awls Hoes Hammerstones Celts

OTHER Ceramics + + + Uncarbonized Botanicals + + + Carbonized Botanicals + + Faunal Remains + +

3 4 0 Table 30 (continued). in (O CO 05 o o o o o O o o o § 5 <: ^ ^ ^ lO in m in to in U5 m in

STRATA Ash Deposits Vegetal Beds Sand-Clay Beds Kitchen-Midden

FEATURES Burials + Post Molds Hominy Holes 4- Fire Pits + Hearths/Hearth Deposits Storage Pits Miscellaneous Pits

LITHIC ARTIFACTS Projectile Points/Knives + + 4- + Other Bifaces Scrapers Ground Stone 4- Fire-altered Rock Awls Hoes Hammerstones Celts

OTHER Ceramics Uncarbonized Botanicals Carbonized Botanicals Faunal Remains

3 4 1 Table 30 (continued). CO in (O o o o o in5 in in in

STRATA Ash Deposits + + + Vegetal Beds + Sand-Clay Beds + Kitchen-Midden + + +

FEATURES Burials Post Molds Hominy Holes Fire Pits Hearths/Hearth Deposits Storage Pits Miscellaneous Pits

LITHIC ARTIFACTS Projectile Points/Knives Other Bifaces Scrapers Ground Stone Fire-altered Rock Awls Hoes Hammerstones Celts

OTHER Ceramics Uncarbonized Botanicals Carbonized Botanicals Faunal Remains

3 4 2