Prehistoric Settlement Patterns at Lake Fork Reservoir

PREHISTORIC SETTLEMENT PATTERNS AT LAKE FORK RESERVOIR

by

James E. Bruseth and Timothy K . Perttula contrihutions by Barbara Butler Gary Rutenherg Bob Skiles Marilyn W estbury S. Alan Skinner, Principal Investigator

Southern Methodist University and Texas Antiquities Committee

Texas Antiquities Permit Series, Report No. 2

Report submitted to the Sabine River Authority of Texas under provisions of SMU contract number 5-28303 and Texas Antiquities Permit Number 168.

Distributed by: Archaeology Research Program, Department of Anthropology, Southern Methodist University and Texas Antiquities Committee, Box 12276, Capitol Station, Austin, Texas 7871 I

1981

•

Archaeology Research Program Press, SMU

•

Dallas


•'

·~'

,.,,

The Lake Fork Archaeological Project Crew. From left to right, top row: Jim Bruseth, Sara Murphy, Snowball, Bob Skiles, Tim Perttula, James Wilson; bottom row: Floyd Kent, Lynn Cronk, MaryAnn McBride, and Lynn Toburen. (Photo printing by Bud Mills)

ABSTRACT This report is the third and final in a series describing the results of archaeological research at Lake Fork Reservoir. The Reservoir is a 30,000 acre Jake being constructed in Wood, Rains, and Hopkins counties, Northeast Texas. Extensive testing of nine sites is discussed and the analyses of lithic and ceramic artifacts, floral and faunal remains, and Jocational data from intra-site and inter-site perspectives are presented. These analyses are used to devise four cultural phases for the Reservoir prehistory, and settlement patterns of each phase are discussed in detail. A model of change in settlement patterns and overall prehistoric adaptation through time is presented

v

ACKNOWLEDGMENTS

Successful completion of the 1978 field season and subsequent analysis and report writing was dependent upon a number of individuals. There are far too many to name all of them individually; however a few offered outstanding support in one form or another and are mentioned below. First and foremost, we thank Mr. Billy Roberts of Holloway Construction Company, the major Reservoir contractor, for his generous and patient assistance of the fieldwork. At various times heavy equipment and other more portable items were needed by the project, and these were readily lent to us by Mr. Roberts. Clearly, without his continued aid the fieldwork undertaken at Lake Fork Reservoir during the 1978 field season would have been considerably less successful. The Sabine River Authority, the agency responsible for completion and operation of Lake Fork Reservoir, offered support by funding this, as well as the previous two, field seasons and by supplying heavy equipment as it was available. Dr. Alan Skinner served as Principal Investigator and managed to achieve a careful balance between offering the authors considerable freedom in the execution of the project but at the same time offering constructive criticism. Dr. LaVerne Herrington of the Texas Antiquities Committee also offered constructive criticism and is primarily responsible for the work accomplished at the Spoonbill and Sandhill Sites. Dr. Herrington not only found financial assistance for work at these two sites after the 1978 field season had ended but also contributed her personal time to field investigation. Moreover, Dr. Herrington arranged for the Texas Antiquities Committee to help towards the cost of publication of this report. Thanks also go to the field crew: Lynn Cronk. Floyd Kent, MaryAnn McBride, Sara Murphy, Bob Skiles, Lynn Toburen, and James Wilson. Particularly, Mr. Skiles offered assistance far above that required by the job. Mr. Skiles spent numerous weekends and late evenings on the project, worked on the Spoonbill and Sandhill sites with no compensation, and coauthored two sections of this report. Mr. Skiles contributed all this to the project through his dedication to the archaeology of Wood County.

vii

Appreciation also goes to the following members of the Dallas Archeological Society who spent two weekends working at the Spoonbill Site: Ruth Ann Ericson. Doris Friedell, Doyle Granberry, Bud and Mary Mills, Elizabeth Perry, Norma Potter, Bill Schurrman, and Tommy Tucker. Back in Dallas several individuals contributed their time or advice on the project. Kate Huckabay, Jim Garber, and Jay Newman helped analyze the lithic material, and Kate Huckabay helped design and execute the lithic tool morphological classification. Mr. Larry Banks, of the Southwest District U.S. Corps of Engineers in Dallas, offered his time freely in describing and classifying the lithic raw material types present in the assemblages. Without Mr. Banks' effort the raw material section of this report could not have been accomplished. Initial editing of the manuscript and the superb drafting of figures was accomplished by Susan Burton. Toni Turner Bruseth, the senior author's wife, deserves special thanks for helping bring the draft manuscript to published form. She offered unending time to edit further and proof the innumerable drafts of this manuscript that were prepared prior to publication. June Schelling, aided by the senior author, typeset the manuscript and overcame the technological "future shock" of the typesetting machine. Several of the photographs, including the frontispiece, were printed by Bud Mills. Dr. L. Mark Raab allowed the senior author to spend considerable time on completing this manuscript and bringing it to the final printed form. The junior author would like to thank further the following: Chet Shaw, Jeff Richner, David Kelly, and G. Tom Jones, who listened, suggested, and discussed at length various portions of this report. Drs. Robert C. Dunne) and Alex D . Krieger offered ideas on the prehistoric archaeological record, and Ann Ramenofsky and Dr. Hiram F. Gregory offered suggestions on Caddoan archaeology and ethnohistory. Kathy Boswell typed certain sections of the manuscript and assisted in the compilation of the bibliography. Drs. James E. Price and Burt Purrington allowed time for the junior author while away from Southern Methodist University to prepare portions of this report.

TABLE OF CONTENTS

ABSTRACT

v

viil

ACKNOWLEDGMENTS LIST OF FIGURES LIST OF TABLES

xi xiii

CHAPTER I: INTRODUCTION Project Goals

1

1

CHAPTER II: CULTURAL AND ENVIRONMENTAL SETTING Culture History Natural Environment

••

s 5 7

CHAPTER III: THE SITES Bracheen Site (X41RA20) Gilbreath Site (X41WD83) Grimes Site (X41WD95) Hines Site (X41WD87) Killebrew Site (X41WD64/ 69) Osborn Site (X41WD16) Sandhill Site ( 41 WD 108) by Bob Skiles Spoonbill Site (41WD109) by Bob Skiles Taddlock Site (X41WD39)

11 11 12 17 21 26 31 37 40 48

CHAPTER IV: EVALUATION OF THE SYMAP SURFACE ARTIFACT DISTRIBUTIONS Methodology of Map Production Comparison of the SYMAPs with Subsurface Data Conclusion CHAPTER V: CERAMIC ANALYSIS Sherds Vessels Pipes Impressed a.nd Nonimpressed Fired Clay

55 55

CHAPTER VI: LITHIC ANALYSIS Morphological, Stylistic, and Functional Considerations Lithic Raw Material Utilization CHAPTER VII: SUBSISTENCE ANALYSIS Introduction Faunal Remains by Barbara Butler and Timothy Perttula Floral Remains Mussel Shell Interpretations CHAPTER VIII: SITE LOCATIONAL ANALYSIS Spatial Distribution of Prehistoric Occupation Density of Prehistoric Occupation Site Locational Behavior: Conclusions From Locational Analysis CHAPTER IX: CONCLUSION: PREHISTORIC SETTLEMENT PATTERNS AT LAKE FORK RESERVOIR Archaic Cultural Phase Lone Oak Cultural Phase Pecan Grove Cultural Phase Forest Hill Cultural Phase

ix

56 66 69 69 91 91 96 101 101

110 117 117 117 125 129

131 133 133 137 137 139 139 141 141 142

TABLE OF CONTENTS (Continued)

APPENDIX I: DEFINITION OF LITHIC ARTIFACT MORPHOLOGICAL CLASSES APPENDIX II: BONE ARTIFACT ANALYSIS by Barbara Butler APPENDIX III: HUMAN SKELETAL ANALYSIS by Gary Rutenberg and Marilyn Westbury BIBLIOGRAPHY

X

145 145 147 149

LIST OF FIGURES

1-1 1-2 2-1 3-1 3-2 3-3 3-4 3-5 3-6 3-7 3-8 3-9 3-10 3-11 3-12 3-13 3-14 3-15 3-16 3-17 3-J 8 3-19 3-20 3-21 3-22 3-23 3-24 3-25 3-26 3-27 3-28 3-29 3-30 3-31 3-32 3-33 3-34 3-35 3-36 3-37 3-38 3-39a 3-39b 3-40 3-41 3-42 3-43 3-44 3-45 3-46 3-47

General Location of Lake Fork Reservoir. Location of Sites in Lake Fork Reservoir Discussed in This Report. Major Biotic Associations in the Lake Fork Reservoir Area. Bracheen Site (X41 RA20) with Respect to Local Topography (I 0 ft. contour intervals). Bracheen Site (X41RA20). Gilbreath Site (X41 WD83) with Respect to Local Topography (10ft. contour intervals). Gilbreath Site (X41 WD83). Hearth (Feature l) at the Gilbreath Site. Burial 1 at the Gilbreath Site. Grimes Site (X41 WD95) with Respect to Local Topography (I 0 ft. contour intervals). Grimes Site (X4l WD95). Artifact Cluster Locations at the Grimes Site. Subsurface Rock Concentration (Feature l) at the Grimes Site. Hearth (Feature 2) at the Grimes Site. Burial 1 at the Grimes Site. Hines Site (X41 WD87) with Respect to Local Topography (I 0 ft. contour intervals). Hines Site (X41WD87). Artifact Cluster Locations at the Hines Site. House Patterns (Feature l) at the Hines Site. Partially Exposed House Patterns (Feature 1) at the Hines Site. Pit (Feature 2) at the Hines Site. Killebrew Site (X41WD64/ 69) with Respect to Local Topography (10ft. contour intervals). Killebrew Site (X4l WD64/69). Artifact Cluster Locations at the Killebrew Site. Hearth (Feature l) at the Killebrew Site. Subsurface Rock Concentration (Feature 2) at the Killebrew Site. Subsurface Rock Concentration (Feature 4) at the Killebrew Site. Osborn Site (X41WD16) with Respect to Local Topography (10ft. contour intervals). Osborn Site (X41WD16). Artifact Cluster Locations at the Osborn Site. Pit (Feature l) at the Osborn Site. Hearth (Feature 2) at the Osborn Site. Pit (Feature 3) at the Osborn Site. Pit (Feature 4) at the Osborn Site. Pit (Feature 5) at the O sborn Site. Pit (Feature 6) at the Osborn Site. Relative Horizontal Position of Features at the Osborn Site. Burial 1 at the Osborn Site. Sandhill (41WD108) and Spoonbill (41WDJ09) Sites with Respect to Local Topography (10ft. contour intervals). Sandhill Site ( 41 WD 108). Spoonbill Site (41WDI09). Soil Color Contours Compared to Subsurface Features at the Spoonbill Site. Soil Phosphate Contours Compared to Subsurface Features at the Spoonbill Site. Members of the Dallas Archaeological Society Troweling Bulldozed Path. House Posthole Pattern Observable in Test Unit Floors at the Spoonbill Site. Hearth (Feature I) at the Spoonbill Site. Hearth (Feature 2) at the Spoonbill Site. Burial 1 at the Spoonbill Site. Burial 2 at the Spoonbill Site. Burial 3 at the Spoonbill Site. Taddlock Site (X4l WD39) with Respect to Local Topography ( 10 ft. contour intervals).

xi

2 3

8 12 13 14 15 16 16 17 18 20 21 21

22 22 23 24 25 26 26 27 28 29 30 31 31 32 33 34 34 35 35 35 36 36 36 37 37 38 41 42 42 43 44 44 45 47 47 48 48

LIST OF FIGURES (Continued)

3-48 3-49 3-50 3-51 3-52 3-53 4-1 4-2 4-3 4-4 4-5 4-6 4-7 4-8 4-9

4-10 4-11 4-12 4-13 4-14 4-15

4-16 4-17 5-1 5-2 5-3 5-4 5-5 5-6 5-7 5-8 5-9 5-10 5-11 5-12

6-1 6-2 7-1 8-1 8-2 8-3 8-4

Taddlock Site (X4l WD39). Artifact Cluster Locations at the Taddlock Site. (a) Profiles of Midden A at the Taddlock Site; (b) Relative Horizontal Positioning ofFeatures at the Taddlock Site. Hearth (Feature 1) at the Taddock Site. Features 2, 3, and 4 at the Taddock Site. Pit (Feature 5) at the Taddock Site. Vertical Distributions of Artifacts at the Yarbrough Site. Comparison of Ceramic SYMAP and Subsurface Artifact Densities for Test Units at the Grimes Site. Comparison of Lithic SYMAP and Subsurface Artifact Densities for Test Units at the Grimes Site. Comparison of Ceramic SYMAP and Subsurface Artifact Densities in Test Units at the Osborn Site. Comparison of Lithic SYMAP and Subsurface Artifact Densities in Test Units at the Osborn Site. Comparison of Ceramic SYMAP and Subsurface Artifact Densities for Test Units at the Taddock Site. Comparison of Lithic SYMAP and Subsurface Artifact Densities in Test Units at the Taddock Site. Comparison of Ceramic SYMAP and Distribution of Subsurface Features at the Killebrew Site. Comparison of Lithic SYMAP and Distribution of Subsurface Features at the Killebrew Site. Comparison of Ceramic SYMAP and Distribution of Subsurface Features at the Taddlock Site. Comparison of Lithic SYMAP and Distribution of Subsurface Features at the Taddlock Site. Comparison of Ceramic SYMAP and Distribution of Subsurface Features at the Hines Site. Comparison of Lithic SYMAP and Distribution of Subsurface Features at the Hines Site. Comparison of Clay Daub SYMAP and Distribution of Subsurface Features at the Hines Site. Comparison of Surface Artifact Density Based on Ceramic SYMAP and Subsurface Chemical Values Along Transect X-X' at the Killebrew Site. Comparison of Surface Artifact Density Based on Ceramic SYMAP and Subsurface Chemical Values Along Transect Y-Y' at the Killebrew Site. Comparison of Surface Artifact Density Based on Ceramic SYMAP and Subsurface Chemical Values Along Transect Z-Z' at the Killebrew Site. Graph of Artifact Cluster Frequency by Percentage of Rim Sherds. Seriation of Artifact Clusters by Selected Decoration Methods. Seriation of Artifact Clusters by Collapsed Temper Categories. Ceramic Design Element Categories for the Type Category, Canton Incised. Ceramic Design Element Categories for the Type Category, Davis/East Incised. Ceramic Design Element Categories for the Type Category, Sanders Engraved. Ceramic Design Element Categories for the Type Category, Ripley Engraved. Ceramic Design Element Categories for the Type Category, McKinney Plain. Ceramic Design Element Categories for the Type Category, LaRue/Nash Neck Banded. Caddoan Ceramic Areas in the General Region of Lake Fork Reservoir. Ceramic Pipes from Artifact Clusters. (a) Impressed Clay Daub from Artifact Cluster WD87A; (b) Impressed Clay Daub from Artifact Cluster WD64/69A. Projectile Point Forms Discussed in Text. Lithic Raw Material Source Areas Within and Adjacent to the Caddoan Area. Seasonal Indicators as Derived From the Behavior of the Vertebrate Remains Present at the Taddlock Site. Location of All Known Archaic Components Within Lake Fork Reservoir. Location of All Known Caddoan Components Within Lake Fork Reservoir. Location of Sites Considered for Intra-Landform Occupational Patterning. Occupational Patterning by Temporal Period on Selected Sites.

xii

49 50 52 52 53 53 57 59 59 60 60 61 61 63 63 64 64 65 65 65 67 67 67 72 75 82 83 84 85 86 88 89 90 97 99 105

110 124 134 134 135

136

LIST OF TABLES 3-1 3-2 4-1 4-2 4-3 5-1 5-2 5-3 5-4 5-5 5-6 5-7 5-8 5-9 5-l 0 5-11 5-12 6-1 6-2 6-3 6-4 6-5 6-6 6-7 6-8 6-9 6-lO 6-11 6-12 6- 13 7-1 7-2 7-3 7-4 7-5 7-6 7-7 7-8 7-9 8-1 9-l

Radiocarbon Dates from the Spoonbill Site. Radiocarbon Dates from Midden A, Taddlock Site. Subsurface Ceramic and Lithic Artifact Densities from Test Units at the Grimes Site. Subsurface Ceramic and Lithic Artifact Densities from Test Units at the Osborn Site. Subsurface Ceramic and Lithic Artifact Sensities from Test Units at the Taddlock Site. Sherd Location and Vessel Form by Artifact Cluster. Orifice Diameter by Artifact Cluster. Decoration Method by Artifact Cluster. Surface Preparation for Rim and Decorated Sherds by Artifact Cluster. Lip ProfUe by Artifact Cluster. Temper for Rim and Decorated Sherds by Artifact Cluster. Temper for All Sherds by Artifact Cluster. Design Element Frequencies by Artifact Cluster. Design Element Frequencies Ordered According to Decoration Method Seriation by Artifact Cluster. Vessel Attribute Observations. Morphological Attributes of Ceramtc Pipes by Artifact Cluster. Impressed and Nonimpressed Fired Clay by Artifact Clusters. Variables Used in Stylistic Analysis. Variables Used in Functional Analysis. Projectile Point Forms by Artifact Cluster. Morphological Class Frequencies by Artifact Cluster. Wear Class Defmitions. Wear Class Frequencies by Temporal Period. Lithic Raw Material Types. Summary Characteristics of Model Expectations and Inferences for the Western Gulf Coastal Plain. Local Versus Exotic Raw Material by Temporal Period. Raw Material Types by Temporal Period. Local Raw Material Types by Temporal Period. Tools Made From Exotic Raw Materials by Artifact Clusters with Sufficient Sample Sizes. Presence of Exotic Raw Materials by Temporal Period. Distribution of Vertebrate Remains Vertically and Horizontally from Midden A, Taddlock Site. Comparison ofVertebrate Class Frequencies Between Upper and Lower Units from Midden A, Taddlock Site. Vertebrate Species Identified from Midden A, Taddlock Site. Relative Frequencies of Selected Anatomical Areas for Deer, Squirrel, and Jackrabbit from Midden A, Taddlock Site. Vertebrate Species Identified from the Steck Site. Carbonized Seeds from the Taddlock, Spoonbill, and Steck Site. Carbonized Nut Remains by Artifact Cluster. Carbonized Cultigens by Artifact Cluster. Minimum Numbers of Mussel Species by Artifact Cluster. Between Period and Phase Comparison of Component Area and Frequency by Site. Summary of Settlement Pattern Evidence for Lake Fork Reservoir Cultural Phases.

xiii

47 53 58 58 62 71 72 74 76 77 78 79 80 81 92 96 98 102 I 03 I 04 I 06 108 109 112 113 113 114 114 115 116 118 118 119 121 123 126 127 128 130 138 140

CHAPTER I INTRODUCTION

Lake Fork Reservoir, located in Wood, Rains, and Hopkins Counties, Texas (Fig. 1-1), is currently under construction by the Sabine River Authority of Texas and is expected to be completed by 1980. It will have surface area of 27,690 acres and a normal pool elevation of 403 ft. above sea level (Environment Consultants, Inc. 1974). This report, which summarizes data obtained from fieldwork in 1978, is the third, and fmal , in a series describing archaeological research at the Reservoir. Archaeological in· vestigations began in 1975 with an intensive survey of the lake area. One-hundred-thirty prehistoric and historic sites were located during the survey, and preliminary testing was conducted at half of the sites. Based on this work, a mitigation plan was prepared for the prehistoric archaeological resources of the Reservoir and presented to the Sabine River Authority (Bruseth and others 1977). Sites were selected for investigation for two reasons. First, subsistencesettlement models presented in the survey report were to be tested by work at the sites, and second, data, representative of presently observable site variability, were to be obtained for future researchers. In 1976, the mitigation plan was partially implemented (Bruseth, Perttula, and Skiles 1980). At this time the Howle Site (X41WD19) was excavated and the Bracheen (X41 RA20), Glen (X41WD99/100), Osborn (X4JWD16), Saddler (X41WD50), and Yandell (X41 RA23} Sites were tested (Fig. 1-2). Further study was recommended for the Bracheen and Osborn Sites. The mitigation program was completed in the Reservoir during the spring of 1978. Research during this season concentrated on the Bracheen, Osborn, Gilbreath (X41WD83), Grimes (X41 WD95 ), Hines (X41WD87), Killebrew (X41WD64/69), and Taddlock (X41WD39) Sites (Fig. 1-2). (Note : The Gilbreath Site was not included in the original mitigation plan but was selected because of a local amateur's discovery of a grave at the site.) Work at the Bracheen and Osborn Sites consisted of additional testing to determine if full-scale excavation was warranted. limited funds were available for large-scale excavation, and before funds were committed to Bracheen and Osborn exclusively, the potential of these two sites relative to the other five was evaluated. The testing indicated that the Bracheen Site warranted no further work, whereas the Osborn Site did warrant further excavation. Additional fieldwork was conducted during the spring of 1979 at two sites, Sandhill ( 41 WD 108) and Spoonbill (41 WD109), when these were scheduled for removal as part

of a Reservoir-associated road relocation (Fig. 1-2). Through fmancial assistance from the Texas Antiquities Committee, these sites were tested to determine their significance. PROJECT GOALS

Project goals were formulated prior to and during fieldwork and during laboratory analysis. Three goals were devised before beginning fieldwork.The first, and primary, goal was part of a continuing effort since the Reservoir survey (Bruseth and others 1977) to understand prehistoric settlement patterns at Lake Fork Reservoir. A model of settlement patterns for the Archaic and Caddoan periods had been developed during the survey of the Reservoir, and this model was modified during the 1976 testing and excavation field season (Bruseth, Perttula, and Skiles 1980). Data collected from the 1978 field season were to be used to refine further the settlement pattern model of previous seasons. The second goal devised prior to fieldwork was the use of cultural components instead of sites as the basic unit of field investigation and later laboratory analysis. Other studies in Northeast Texas (Anderson and others 1974; Sullivan 1977; McCormick 1973; Hyatt and Doehner 1975), as well as previous studies in the Reservoir (Bruseth and others 1977; Bruseth, Perttula, and Skiles 1980) had used sites as a whole for cultural and temporal comparison despite the fact that many of these sites were multicomponent. To achieve component separation, the third goal of the field season was devised, namely the use of surface artifact distributions to separate components. Previous field research in the Reservoir had shown that surface collections offered a quick visual picture of intrasite structure (Bruseth 1980). Since many sites possessed horizontally separate components, surface collections were found to be useful component delimeters. During fieldwork two goals were developed as a result of the recovery of unexpected data types. At the Taddlock Site an excellent faunal and floral sample was recovered from a midden. Although faunal and floral samples had been recovered from the Steck Site outside the Reservoir (Hockensmith 1977), research inside the Reservoir had indicated that floral and faunal remains were poorly preserved. With the data recovery obtained from the Taddlock Site, the study of prehistoric subsistence patterns was added as a new goal. Although a single site provides a small sample upon which to base generalizations, the Steck Site sample from outside the Reservoir offers comparative data to strengthen subsistence interpretations. The second goal,

2


OKLAHOMA

Surpllur River

I

II I

I

__ I

_...__-

~

TEXAS

N

I Toledo8end

Ruervoir

0I

mil II

20 I 20

0 ltm

Figure 1-1.

General Location of Lake Fork Reservoir.

developed during fieldwork, was based on the surface artifact distributions. The SYMAP surface density maps, which had successfully identified numerous horizontal components at several of the sites, indicated that the components tended to be situated on different landform types through time. Since these data offered the opportunity to examine a new aspect of prehistoric settlement pattern behavior, another research goal was included.

Finally, two goals were developed during the analysis phase of research. The first was use of the lithic and ceramic artifacts to develop a finer chronology for the project area than previously existed. The extant chronology was based on decorated sherd types and on projectile point styles generalized from Central and East Texas. The samples of artifacts from the Reservoir sites offered an opportunity to refine this chronology.

3

INTRODUCTION

•

NOR TH

0

......

Figure 1-2.

•

TESTEO OUlltiNO 1171

e

TESTED OUIUNG 1171

•

T£STED OUJII NG 1171

Location of Sites in Lake Fork Reservoir Discussed in This Report.

The second goal was part of a continuing attempt at Lake Fork Reservoir to produce meaningful results from lithic artifact analysis. Most problem-oriented researchers in Northeast Texas have used lithic analysis partially to determine site function (Anderson and others 1974; McCormick 1974; Sullivan 1977; Hyatt, Butler, and Mosca 1974; Hyatt and Doehner 1975 Doehner and Larson 1978). Despite these persistent attempts, little explanatory information has been obtained. Analysis of the 1976 lithic data was directed toward overcoming this difficulty by examining lithic artifact curate behavior to separate sedentary from nonsedentary occupations (Perttula and Mitchum 1980). However. this effort also met with little success. and a new orientation was developed for the 1978 analysis. Three approaches were used. The first was a stylistic analysis to define lithic artifact style changes over time that could be used to refine the existing chronology for the area. The second

was the use offunctional and morphological classifications to examine lithic assemblage changes over time, and to see if any changes observed could inform on prehistoric subsistence. The final approach was the use of lithic raw materials to examine interaction between the Reservoir and surrounding localities. All three approaches were to be related to Reservoir settlement patterns in the final chapter of this report. The foregoing goals served as the primary guiding force behind the 1978 field season and subsequent artifact analy· sis. The problems are based on the potential of the extant data base. Too often research in Northeast Texas has been based on idealistic and unobtainable goals. and the net result has been unsubstantiated conclusions based on limited data. An attempt has been made to avoid this situation in the Lake Fork analysis.

CHAPTER II CULTURAL AND ENVIRONMENTAL SETTING

This chapter provides information on the culture history and environment of the Reservoir area. This information is intended to familiarize the reader with basic concepts and terminology used in later chapters. The Culture History section only presents a skeleton outline and does not make reference to data obtained during the 1978 season to avoid duplication in reporting. It does, however, include data from previous seasons of work in the Reservoir. The fmal chapter of this report integrates the 1978 data into a new cultural historic framework for the Reservoir and is the final version . CULTURE HISTORY

PALEOINDJAN (10,000 - 6000 B.C.) With one possible exception, there are no known in situ Paleolndian components in Northeast Texas, and principal evidence of occupation during this time consists of scattered fmds of diagnostic projectile point types. The exception is a site located in Delta County where buried lenses of ash, charcoal, and burned clay were exposed in the cutbank of the North Sulphur River. Artifacts found included an antler tool and quartzite flake s, all recovered in the immediate area of the lenses. A single radiocarbon date of 7600 ± 375 B.C. (SMU-532) was obtained from one of the lenses (Slaughter and Hoover 1965 :351-2). Paleolndian points similar to the Scottsbluff type have been collected from the stream bed of the Sulphur River (H. Kitchens, personal communication), and it is possible other buried occupations exist within the flood plain. San Patrice sites, late Paleolndian or Early Archaic manifestations, are known from Northeast Texas as well as northwestern Louisiana, eastern Oklahoma, and southern and central Arkansas (Webb 1946; Schambach 1972; Webb, Shiner, and Roberts 1971; Duffield 1963). San Patrice Points, or Dalton var. San Patrice (Briar 1971 ) , have been found in the Lake Fork Reservoir area (Skiles, Bruseth, and Perttula 1980), but other Paleolndian types are uncommon. Bryant and Shafer (1977:19-20) have made a strong case for a generalized hunting/gathering subsistence pattern for the Paleolndian period of East Texas instead of the more traditionally accepted pattern of big-game hunting: We seriously doubt that big-game hunting would have been economically sound as a major subsistence pattern in any of the parkland and woodland areas of eastern Texas where the species of big game were probably dispersed or not present in large numbers. Indeed. to our knowledge,

there is no confirmed association of artifacts of late Pleis· tocene age associated with the remains of big-game faunal remains in any part of Texas outside the Plains and its associated environments. Therefore, the obvious implication of these data is that hunting and gathering adaptations, in all probability, represented the initial and the continuous prehistoric adaptive patterns in large areas of Texas.

ARCHAIC (6000B.C. - A.D. 1) Archaic occupations are common throughout Northeast Texas (Anderson and others 1974; McCormick 1974; Johnson 1962) and consist of dart points and other lithic artifact deposits, often in doubtful primary context. Abundant evidence exists for Archaic occupation in Lake Fork Reservoir; however, only one Archaic component (Howle Site, Block 2; X41 WD19) has been extensively investigated (Bruseth, Perttula, and Skiles 1980). This site, located on a flood plain rise along Lake Fork Creek, consists of an Archaic component with a veneer of Early Caddoan ceramic material in the upper levels of the deposit. The component consists of lithic debris, lithic tools (including dart points}, tire-cracked rock, and ground stone artifacts. During the Archaic Period there was a continued reliance on hunting and gathering of a variety of plant and animal foods. Archaic adaptation can be characterized as "primary forest efficiency" (Caldwell 1958:VII). Through time, as exploitation of resources became more efficient, it is hypothesized that Archaic group territories became smaller (Grady 1978; Lynott 1978).

EARLY CERAMIC (A.D. 1 - 800) Early Ceramic occupations, a term used to designate preCaddoan ceramic sites, are not well known in East Texas in comparison to Oklahoma, Arkansas, and Louisiana. The Early Ceramic Fourche Maline Phase, defined on the basis of Williams Plain, has been extensively investigated in eastern Oklahoma (Bell and Baerreis 1951 ; Galm 1978; Galm and Flynn 1978) and western Arkansas (Hoffman 1977; Schambach 1970), and seems to date from A.D. 1 to 800. Analogous occupations are present in East Texas along the Sulphur River (Jelks 1961; Doehner and Larson 1978), at Cypress Creek Lake (Dee Ann Story, personal communication; Stems n.d.), at Toledo Bend Reservoir, and at McGee Bend Reservoirs. Early Ceramic components are known from three Lake Fork Reservoir sites: Bracheen (X41 RA20), Howle Block I (X41WD19), and Osborn (X41 WD16).

6


Available evidence, although weak, suggests that subsistence practices consisted primarily of hunting and gathering of wild plant and animal foods, with some supplement from horticulture. Maize and squash are known from the temporally comparable Tchefuncte and Marksville periods in the Lower Mississippi River Valley (Struever and Vickery 1973; Byrd 1976), but the actual presence of these cultigens in Early Ceramic contexts in the Caddoan area is yet to be established.

EARLY CADDOAN (A.D. 800 - 1300) Early Caddoan occupations in Texas have been divided into the Sanders and Alto Foci (Suhm, Krieger, and Jelks 1954). On the basis of stylistic point and ceramic types, the Early Caddoan material in the Lake Fork area can be in· cluded within the Sanders Focus. Unfortunately, this focus was defmed on the basis of a single, and apparently rather unique, site (i.e., Sanders Site). No site excavated since has included all of the cultural traits recognized at the type site. Early Caddoan components in the Reservoir include the Sadler (X41 WDSO) and Yandel (X41 RA23) Sites (Bruseth, Perttula, and Skiles 1980}, and many others have been located by amateur archaeologists south of the Reservoir along Lake Fork Creek (Skiles, Bruseth, and Perttula 1980). These occupations are also found in the headwaters of the Sabine on the west and along Cypress Creek on the east. tittle is known about their northern distribution. Cemeteries belonging to this period are commonly found (Skiles, Bruseth, and Perttula 1979) and fail to show an organized burial plan, instead representing haphazardly interred individuals. Unlike other parts of the Caddoan area (Webb 1959; Krieger 1946, 1948), no mound construction i~ known from the Reservoir, but the Keith and Hale mounds are located about 40 km to the east and the Colony Church Site is situated in proposed Carl Estes lake about 25 km to the south (Malone 1972:25). Horticulture, in the form of maize, is directly evidenced from sites excavated by SMU in the Reservoir (Bruseth and Carter 1980), and it is likely that flood plain cultivation was an integral part of the subsistence system. Faunal and floral evidence from other areas suggests a continued reliance on a wide variety of wild plant and animal foods.

LATE CADDOAN (A.D. 1300- 1650) Late Caddoan period sites within the Lake Fork area belong to the Titus Focus, a cultural complex primarily known from cemetery excavations during the 1930s in Titus, Hopkins, Wood, Franklin, Camp, and Upshur Coun· ties. late Caddoan sites within and around the Reservoir are located along Dry and Caney Creeks and to a limited extent along Lake Fork Creek between the confluences of these two creeks. Site density is high with 15 known and

excavated Late Caddoan sites in the Dry Creek watershed and eight within the Caney Creek watershed (B. Turbeville. personal communication). Eight cemeteries of this period are known from amateur excavations. There is an average of l 0 graves per cemetery and all graves contain grave goods including ceramic vessels, elbow pipes, celts, and arrow point caches. The cemeteries are usually found in deep sandy areas and associated with midden deposits. The small number of graves per cemetery suggests a short-term sedentary occupation, perhaps on the order of 20-50 years. Although not present in the Reservoir, Late Caddoan mounds are noted from Lake of the Pines (Jelks and Tunnell 1959; Davis 1958), on Cypress Creek, and from the McKenzie Site near Big Sandy Creek (D. Granberry, personal communication). Direct evidence of agriculture occurs at these sites in the form of maize, and the well drained flood plain soils would have been most suitable for agriculture. Umited faunal data suggest partial dependence upon deer and other animal resources.

HISTORIC INDIAN (A.D. 1650- 1800) Historic Indian occupation in the Lake Fork area is represented by the Gilbert Site (Jelks 1967) and at nearby Lake Tawakoni by the Pearson Site (Duffield and Jelks 1961). European manufactured artifacts such as glass beads, gun parts, iron axes, and other iron and brass tools have been found, along with native manufactured artifacts such as Womack Engraved ceramics, arrow points, and end scrapers. These materials, particularly the glass beads, suggest late 18th century occupations. Both the Gilbert and Pearson Sites have been assigned to the Norteno Focus, a designation for sites associated with the Historic Wichita tribes of the Southern Plains. The simi· larities, however, between the Gilbert and Pearson Sites and the late Caddoan occupations are sufficient to suggest that these two sites are actually Caddoan. None of the native manufactured artifacts necessarily indicates a Wichita ori· gin, and Womack Engraved, a definitive ceramic type of the Nortel'io Focus, has been recovered from late Caddoan cemeteries within the area (Scurlock 1962:296). Most other ceramics from the Gilbert Site are of Late Caddoan afflliation (Story and others 1967:187). Moreover, end scrapers found at the Gilbert and Pearson Sites are unlike those from the Nortel'io sites further west and indicate deer hide preparation rather than bison hide preparation (Shiner and Harris 1973). Possibly, the Gilbert and Pearson Sites repre· sent the remaining Caddoan peoples in the Lake Fork area, and their material culture reflects European acculturation and not Wichita migration.

CULTURAL AND ENVIRONMENTAL SETTING NATURALENN1RONMENT

CLIMATE The climate of the project area is classified as humid subtropical (Environment Consultants 1974). The average annual temperature is 65°F, with the winter low mean at 47° F (January) and the summer high mean at 83° F (July and August). Rainfall is abundant, with the project area receiving an annual mean of 41 inches (based on data from the Emory and Quitman gauging stations; U.S. Department of Commerce 1965). The general precipitation pattern consists of two rainy seasons. A majority of the precipitation falls during April and May, at which time monthly means reach five or six inches. The second rainy season is from October to December. During this period, rainfall approaches a monthly mean of four inches. Relative drought conditions exist during the summer from June to September. Stream-flow characteristics are strongly correlated with rainfall patterns. Based on the 21 years during which records are available for Lake Fork Creek at the Quitman gauging station (4.8 miles south of the dam), a generally smooth annual trend is seen. Lake Fork Creek overflows its banks at 300 cubic feet per second (estimated by the U.S. Geological Survey , Fort Worth Office). Average monthly flow rates are above this stage from December until June. Thus, the Lake Fork Creek flood plain is inundated to some degree for seven months each year.

7 The Blackland Prairie to the west represents a substantially different vegetative environment consisting of medium-tall and rather dense bluestem grassland (Kuchler 1964:76). The Oak-Hickory-Pine association to the east of the project area is similar to the Oak-Hickory Forest with the exception that shortleaf and loblolly pine are also dorni· nant species. The presence of pine is thoug!H to be due to the less mature nature of this association (Mahler 1973:5). The biotic resources within the Reservoir area have been elaborated upon in detail in the survey report (Bruseth and others 1977), and only a summary of that discussion is provided here. Compensating for the effects of modem dis· turbance such as lumbering and field clearing, the prehis· toric bi~tic environment can be divided into two 'biotic communities (Cleveland n.d.). The first is the upland community, characterized by an overstory of tall trees and an understory of grasses. The trees of this zone would have shaded between 15 and 30% of the ground surface (U.S. Department of Agricuhure 1974) creating a Savannah envi· ronment. Numerous small prairies, grassland openings in the Savannah, would have been present (Jordan 1973). The second zone, the riverside-flood plain, would have been similar in basic composition except that a greater diversity of plants and animals would have been present, due to variable states of biotic succession caused by fre· quent creek overflow. The riverside-flood plain zone is char· acterized by a dense overstory with a less compact under· story of shrubby trees and herbaceous plants. Scattered open areas also occurred throughout this zone.

TOPOGRAPHY FAUNA The fauna of the Lake Fork area is reflective of its transitional position between biotic associations and contains species adapted to xeric and mesic communities. The prehistoric fauna was similar to that observed today (Bruseth and others 1977: 15). A large variety of mammals, including white-taileo deer, gray fox, beaver, and gray squirrel, as well as molluscs, arthropods, fish , amphibians, reptiles, and birds would have been available. Other faunal species, such as bison and pronghorn antelope, would have been present in the Blackland Prairie to the west (Shaw 1978:29-30).

FLORA Lake Fork Reservoir is located in the Oak-Hickory Forest biotic association. This association is a natural transition zone between the Blackland Prairie to the west and the Oak-Hickory-Pine Forestto the east (Fig. 2-1 ). Medium-tala to tall broad leaf deciduous forests characterize the OakHickory forest jlgsociation, with post oak and blackjack oak most common (Gould l 969:9). Prehistorically, grasses, including little bluestem and Indian grass, would have been common understory vegetation.

The topography of the project area is characterized by sharply rolling to gently undulating relief, separated on occasion by broad flood plains. The elevation differential between the flood plain and the upland ranges from 40 to 100 vertical feet. Generally the upland slopes north ofLake Fork Creek are steeper and higher in elevation. Many of these slopes are covered with quartizite cobbles or ferruginous sandstone nodules, which makes this side of the valley less susceptible to erosion. The western margin of the valley lacks comparable deposits, and consequently it has a gentler slope which is cut by numerous creeks.

ENVIRONMENTAL STABILITY The prairie/forest ecotone has apparently fluctuated and changed continually since its postglacial establishment in response to both short and long term climatic changes. As Berabo and Webb (1977:8) note: From 7000 to 2000 BP, the prairie/forest border receded westward, out at a slower pace tnan it mao expanoeo 1ll the early Holocene. The prairie/forest ecotone has thus been in motion during most of the Holocene, advancing east· ward relatively rapidly during the early Holocene and then retreating westward more slowly during the late Holocene.


8

...

""'·.: .-

......

~

·::

·.

...

,

.·

U

B•CK•'oM p,o.,.t

t: 41

Oo• . H.c.-ory

4 Fo,ut

m~?:.- ~~~!~~y -

Figure 2-1.

Major Biotic Associations in the Lake Fork Reservoir Area.

Evidence of change for the prairie/forest ecotone in the Caddoan area is meager by comparison with data from the midwestern states (e.g., .Berabo and Webb 1977; Webb and McAndrew 1976; Bryant and Shafer 1977). After the onset of the Postglacial Period (8000 B.C.), paleoenvironmental changes were not sudden or radical, but were gradual. Recent new studies from southwestern Missouri (Wood and McMillan 1976) and northcentral Oklahoma (Henry 1977, 1978) provide important paleoenvironmental information which have bearing on the prairie/forest transition in the Caddoan area. Because of the possibility of regional variations in paleoenvironmental change, however, reconstructed climatic sequences for the Caddoan area should be

viewed tentatively. Paleoenvironmental evidence from Rodgers Shelter, located along the Prairie Peninsula, suggests two periods of grassland expansion, 6500 to 4500 B.C. and 100 B.C. to A.D. 1, alternating with returns to a forest edge environment (McMillan 1976b:227-229). Based on pollen and land snail evidence from Painted Shelter, Cut Finger Cave, and Big Hawk Shelter in the cross-timbers of Osage County, Oklahoma, arid conditions at A.D. 800 and wetter conditions at A.D. 1200 have been proposed (Heury 1978:88-89). Exactly how this evidence relates to climatic changes in the Oak-Hickory to prairie border of the Caddoan area is not clear. Climatic and ecological changes can be projected,

CULTURAL AND ENVIRONMENTAL SETTING but as Bryson and others (1970) have suggested " ....the direction of....change will not be the same everywhere and there must (have been) many climatic-biotic core areas with very little change." Acknowledging these problems, three periods of grassland extension into the forest can be

9 projected for the Caddoan area. The first was between 6500 to 4500 B.C., the second was between 1000 B.C. and A.D. 1, and the third was between A.D. 800 and 1200. Until further evidence is available to refme these speculations, they remain extremely tentative.

CHAPTER III THE SITES

Nine sites form the primary data base for this report : the Bracheen (X41 RA20), Gilbreath (X41WD83), Grimes (X41 WD95), Hines (X41 WD87), Killebrew (X41 WD64/69), Osborn (X41WDI6), SandhiJI (41WDI08). SpoonbiJI (41WDI09), and Taddlock (X41WD39) Sites. A three-stage research methodology was employed at the Grimes. Hines, Killebrew, Osborn, and Taddlock sites. The first stage involved plowing and systematic surface collection. The surface collected artifacts were tabulated and input into a synagraphic computer program (SYMAP) to produce artifact density contour maps for separate artifact classes and for combinations of artifact classes. Based on SYMAP artifact density maps, sites were then trenched with a backhoe. Trenches were aligned to bisect both the major SYMAP artifact clusters and intervening areas. Trenches were profiled along one wall to locate cultural features. To obtain artifact samples and to explore subsurface features , various sized test units were also excavated at each site on the basis of the SYMAP artifact clusters and backhoe trench profiles. Excavation units varied in siLc from 0.5 m square to 2 m square. In some instances the methodological stages were implemented out of order. For example, a few test units at the Osborn, Taddlock. Killebrew. and Grimes Sites were excavated prior to the placement of backhoe trenches. This was done because it was felt that backhoe trenching in what appeared from surface evidence to be artifact clusters with unique qualities (e.g., exceiJcnt faunal preservation) might do more harm than good. Work at Midden A at the TaddIock Site exemplifies this. Prior to backhoe testing, a 1 m x I m unit was hand excavated in the midden. The recovery of abundant floral, faunal. and artifactual material from the midden suggested that it should be excavated solely under controlled conditions. A backhoe trench had been planned to bisect the midden and would have removed an estimated 15% of its content, thus significantly reducing the potentially recoverable material. Because of this. the trench was not excavated. Variations on the three-stage methodology were employed at the Gilbreath , Bracheen, Sandhill, and Spoonbill Sites. The Gilbreath Site was not scheduled for excavation in the mitigation plan outlined in the Lake Fork survey report (Bruseth and others 1977). However, a Late Caddean burial was discovered at the site by an amateur before the 1978 field season began, and this find suggested that the site might contain a cemetery. an important feature of some Late Caddoan sites in Wood County. Since this type

of feature had not been recovered by previous professional excavations in the Reservoir, the site was included for testing. Since it was doubtful that plowing and systematic surface collecting would be of value in locating subsurface graves, hand excavated test units and trenches together with backhoe trenches were used instead. The Bracheen Site was not plowed and surface collected since the entire site was thickly forested. Instead of clearing the trees, test units and postholes were placed over the site. The Sandhill and Spoonbill Sites were located during clearing operations associated with a road relocation along Caney Creek. These two sites were discovered after the 1978 field season had been completed and were subsequently tested with a skeleton crew and on a limited budget. The Spoonbill data arc fully integrated into the analyses presented in this report, whereas the Sandhill data arc only fully reported in this chapter and summarily included in other chapters. Insufficient funds are responsible for the Sandhill data not being fully interwoven into this report. Both sites were tested by a variety of techniques including test pits. postholes, and mechanical scraping. The following site descriptions provide pertment information from each site. Except in the discussion of certain features, interpretations are kept to a minimum. The remaining chapters of this report present interpretations and conclusions concerning the nine sites. BRACHEEN SITE (X4l RA20) The Bracheen Site is located at the northern end of the Reservoir on a flood plain rise about 50 m south of the present channel of Lake Fork Creek and 30 m west of Texas Highway 19 (Fig. 3-1 ). The rise is 1.8 m above the flood plain and 5000 sq m in area {Fig. 3-2}. The rise is steepest along its northern side and slopes gently along its other sides. Second-growth vegetation of immature oaks and sweetgums indicates that most of the site was timbered in the past 30 years. The soil profile at Bracheen is a 50 em thick, dark brown loam to yellowish brown silty loam A horizon over an undetermined depth, yellowish-brown, clayey silt B horizon. Average soil pH is 6.0, or slightly acidic.


12

and Skiles 1980), and if Bracheen was found to offer no new information, mitigation of the Howle Site was felt to be adequate compensation for loss of this site. Two 1 x 1 m units were placed within the area of the suspected buried midden. These units were excavated in 10 em arbitrary levels to a depth of 50 em and dirt was screened through 1/4 inch hardware cloth. In conjunction with the two test units excavated in the north-central area of the site, two other strategies were utilized. First, a 1 x 1 m unit was placed on the south backslope of the rise to investigate the spatial and depositional variability of the site. Second, following O'Malley's (1976:499) recommendations concerning soil chemical analysis, three posthole transects were laid out across the site. The postholes excavated at 1 m mtervals along each transect, and soil samples and Munsell color readings were taken at 20 em depth intervals within each po~thole. Soil samples were taken with trowels, and every effort was made to insure that soil horizon mixture was minimal.

Soil Chemical Analysis Figure 3-1.

Bracheen Site (X41RA20) with Respect to Local Topography ( 10 ft. contour intervals).

RESEARCH METHODOLOGY Research at Bracheen during the 197 8 season was dictated by the results and recommendations of the 1976 fieldwork at the site (Bruseth, Perttula, and Skiles 1980). The Bracheen Site was originally tested during 1976, and at this time three 1 x 1 m test units were excavated in the north-central (the highest) portion of the site. These excavations and subsequent soil chemical analysis suggested the possibility of a buried midden 20 to 40 em below the surface. A 55 ppm increase in phosphate content characterized the midden as compared to underlying and overlying soil. Posthole diggers were used to excavate 13 postholes around this area to determine the extent of the possible midden and generally to investigate the subsurface deposits. No soil chemical analyses were conducted on these samples. Three additional 1 x 1 m units were placed in other areas of the site. The artifact assemblage recovered during testing was predominately lithic; only 2% of the assemblage was ceramic. The 23 arrow points and 9 dart points recovered suggested that the site was a product of both Archaic and Caddoan occupation. Research at the Bracheen Site for the 1978 field season was designed to evaluate the possible buried midden. The investigation was aimed at establishing whether further excavations were necessary to adequately mitigate the loss of the site. The site was very similar in content and situation to the Howle Site (X41WD19), a flood plain rise site located in the southern part of the Reservoir (Bruseth, Perttula,

Because no distinct buried midden horizon could be visually identified from the soil profiles in test units, the primary means of rrudden evaluation became soil chemical analysis. The natural soil chemical content (i.e., the content not influenced by cultural factors) was derived from a con· trolled sample point without cultural materials at the north· ern edge of the rise. Phosphate chemical analysis was the main test undertaken to determine the presence of the mid· den. This analysis assumes that the relative phosphate values in soil horizons reflect past cultural activities, w 1th such activity enriching the natural soil phosphate content (Eidt 1973, 1977). The results indicated that phosphate values tend to increase in depth over the entire site with the lowest levels from all posthole samples (60-80 em) registering the highest phosphate readings. However, a midden zone of enriched phosphate was not detected . Apparently, the cultural activity responsible for artifact deposition did not result in chemical enrichment of the soil. which is visi· ble today. The 1976 results suggesting a midden were pro· bably due to phosphate analysis error.

GILBREATH SITE (X41WD83) The Gilbreath Site is located adjacent to Glade Creek, a secondary tributary of Lake Fork Creek. The site is situated partially on a sandy rise, about a meter in height, and partially on the slope of a swale along the western side of the rise (Figs. 3-3 and 3-4). The site is presently covered by pasture and widely scattered trees. A considerable portion of the rise has been removed through borrowing for sand, thus making it impossible to determine the original extent

13

THE SITES

L AK£

,.ORK CRl'l'K ,.LOOO PL A I N

--.

---------------- ----------+--

---

---

• •

--.......... •

\ l ) •

\

_..... ..-

•

lU G

N OR T H

•

~

me te rt SO c"'

•

con tour t nlerwot

TES T •

·~

UN ITS

POST HOLE

TESTS

Figure 3-2.

Bracheen Site (X41RA20).

of the site. It is very likely that much of the cultural deposit was removed by this activity. The site was discovered in 1975 during the survey of the Reservoir after recent borrowing had badly distwbed a burial (Burial 2). The sandy rise is part of the Wilcox Formation, an Eocene deposit consisting of unconsolidated sands, silts, and clays. The swale along the western side of the rise serves as a natural drainage for the area, and clays from the sandy rise have collected in the swale. The soil proflle of the sandy rise consists of an A horizon high in organic content and extending 40 em below the surface. Below this is a clay-rich B horizon ; since no cultural materials or features extended into the B horizon, no attempt was made to defme its vertical extent.

RESEARCH METHODOLOGY The Gilbreath Site, not originally scheduled for mitigation (Bruseth and others 1977), was brought to the authors' attention when a Caddoan period burial was discovered by an amateur archaeologist. The burial (Burial 1) was left intact for professional excavation, and this prompted the site's inclusion in the 1978 field season. Prior to the 1978 season, no intact burials had been encountered during professional excavations in the Reservoir. Burial 1 offered for the first time an opportunity to obtain osteometric and mortuary data. This burial, together with the presence of Burial 2, suggested the possible existence of an organized cemetery at the site. This possibility was further impetus to include the site in the mitigation program.

t

•

l

14


t N

ted. The first (Midden A) was found over Burial 1. and the 3 x 3 m unit used to expose this burial was also used to obtain an artifact sample from Midden A. Midden B was found on the slope of a swale IS m east of Midden A. A 1 x I m unit and a 2 x 2 m unit were excavated in Midden B to obtain an artifact sample.

FEATURES Although a cemetery was not located, the testing of the site did uncover the known burial and locate several other features, which are described below.

Midden A

0

300

li"""'P"""l!

m Figure 3-3.

Gilbreath Site (X4l WD83) with Respect to Local Topography (10ft. contour intervals).

The field methodology for the site was designed primarily to determine if a cemetery existed. Since plowing and surface collecting were not felt to be useful for detecting graves, this phase of testing was omitted. Plowing and surface collecting are only useful when subsurface cultural materials have a surface manifestation, and since Caddoan graves in the area are often located considerably below the plow zone, a surface manifestation could not be assumed to exist. Burial 1 was initially investigated by a 3 x 3 m unit. Based on collector information from known cemeteries in the area, this unit site was felt to be large enough to encounter surrounding graves if the known burial were actually part of a cemetery. Ten centimeter levels were employed during the excavation of the unit with all dirt screened through 1/4 inch mesh. In addition, a metal probe was employed to locate other graves. When potential areas were isolated, excavation units were used to investigate them. Vertical levels were not employed in these units although soil was screened. The main concern was to determine if the anomalies encountered by probing were graves; tlte results were uniformly negative. A more systematic search for the possible cemetery was accomplished through a senes of parallel trenches excavated from the known grave across the sandy rise. At first the trenches were hand dug to minimize disturbance in the event a grave was located However, this proved to be time consuming, and after more than 70 m of trenching had failed to locate a grave a backhoe was brought in to continue digging trenches. A total of 260 m of trenches was dug. and this effort failed to locate additional graves. During investigations at the site two middens were loca-

This midden, (also referred to in later chapters as WD83A), located above Burial 1, is situated on the western side of the sandy rise (Fig. 3-4 ). Although part of the midden has been removed along the western margin for fill. it is estimated to have been 8 m in diameter and today is clearly discernible because of its dark brown color which increases in darkness (maximum 10 YR 3/3) towards the center. Depth of the midden is 30 centimeters. Highest phosphate level is 5 and pH is 6.0 (Note : Phosphate value is based on scale presented by Eidt 1973). In addition to the 3 x 3 m unit excavated to uncover the burial. seven linear meters of trench were excavated in Midden A. A total of 29% of the midden's 50 sq m area was excavated by the trenches and the test unit; the test unit alone accounted for an 18% sample of the deposit.

Midden B This midden (also referred to in later chapters as WD83B; Fig.3-4), which is considerably larger than Midden A, is approximately 14 m in diameter and has an area of 155 square meters. It is located 15 m west of Midden A on the slope of the western edge of the sandy rise. The deposit consists of a culturally stained, silty dark brown soil over· lying a natural clayey soil. Elevation readings were taken through the midden on an east-west transect to examine its elevation in relation to the natural slope of the swale. The results showed that the midden actually overlies the natural slope of the swale.

Feature I This feature (Fig. 3-4) is a hearth located in the center of Midden A. It begins 10 to 15 em below the surface and extends vertically for another 10 to 12 centimeters. It has an oval horizontal plan and measures 35 em by 45 centimeters. It consists of two layers. The upper layer is an ashy, compacted soil about 5 em thick. Below this is a reddish clay, colored red from intense heat. The clay is from 5 to 7 em thick. The upper layer was saved for flotation analysis (see Chapter 7 for flotation results).

THE SITES

15

6LAD£

CR££1(

FLOOD

PLAIN

Borrowed

Area

M AG. NORT H

0

e

5

10

,

me t e rs

250m

~

0 •

BURIAL M lODEN TEST

AREAS

UNITS

. . . BACKHOE TRENCHES -

HAND DUG

TRENCHES

Figure 3-4.

Gilbreath Site (X41 WD83).

Burial!

This feature was a Caddoan burial of a child between 2 and 3 years of age (see Appendix III for details on Skeletal Analysis). Five ceramic vessels were found with the skele-

ton and were placed on either side of the chest and alongside the legs (see Table 5-10 for vessel descriptions). A conch shell effigy pendant was recovered from the chest area. The skeleton was extended and oriented with head to the west (Fig. 3-6). The bottom of the grave was 40 em

16


dork yeffowosh brown sondy

A-HORIZON

sandy loom !IOYR3/61 dork greyish brown sandy loom/osh uovo3nl reddosh

yeffowosh cloy

~;r-:;~-.,..-r-r--r--r-~.......,-...,1

17 HRI/61

brown sond IIOY.4/3)

Figure 3-5.

Hearth (Feature l) at the Gilbreath Site.

Figure 3-6.

Burial 1 at the Gilbreath Site.

below the surface. Although great care was taken during excavation to locate a pit outline, none was observed.

Buria/2 This burial was located during the 1975 survey of the Reservoir shortly after sand borrowing activities had disturbed the grave (Fig. 3-4). Five partially restorable vessels were found (see Table 5-10 for vessel descriptions) in a

small cluster 50 em in diameter in an area that prior to borrowing was the extreme southern portion of the sandy rise. No other artifactual remains were found in the area. The vessels all were badly crushed from borrowing activities and most of the tops of the vessels were missing. This burial had no skeletal material, and no grave pit was observed. Although the absence of skeletal material means that the feature is not absolutely proven to be a burial, the occurrence of the partially restorable vessels away from the middens and features at the site and in an area devoid of other cultural items strongly implies this context.

17

THE SITES GRIMES SITE (X41WD95) The Grimes Site is located on a terrace which projects into the lake Fork Creek flood plain (Fig. 3-7). The terrace covers about 40,000 sq m, and the surface is level except along the periphery which has been modified by erosion. The terrace is over 3 m in height above the flood plain and has gently sloping sides (Fig. 3-8). Formation of the terrace took place during the Pleistocene (Durler 1976), and the well developed soil profiles at the site reflect this antiquity. Since the site does not flood as a result of overflow from lake Fork Creek, and since the uplands to the west are too far for colluviation to be appreciable, no present day source of substantial soil aggradation exists. Th~ soil profile at the site consists of a brown silty loam A horizon varying in thickness from 40 to 80 centimeters. Below this is a clay-rich sandy and silty B horizon ; depth of this horizon is undetermined. Along the eastern edge of the site, erosion has caused the A horizon to be removed and the clayey B horison to be exposed. Although the predominant use of the site locale has been for pasture, during the winter of 1977-1978, it was cultivated in watermelons. This type of cultivation is particularly destructive to archaeological resources. It requires alternating paths to be plowed repeatedly, creating a series of parallel ridges and swales across the site. Destruction to archaeological resources occurs mainly in that the deeper, repeatedly plowed paths disturb more subsurface artifacts than normal plowing does.

WD95

r-'. '--"·

.1\..r' Figure 3-7.

r 6

¥

3oo ~

.

Grimes Site (X41WD95) with Respect to Local Topography (10 ft. contour intervals).

The northern third of the terrace, about 15,000 sq m, is wooded. No field work was undertaken in this area, but considering its close position to lake Fork Creek, it pro· bably has an artifact density similar to the tested portion of the site.

RESEARCH METHODOLOGY The Grimes Site was plowed and collected during the initial stage of testing. Five 1 x 1 m units were used for surface collection and the site was collected in 11.25 mandays. Only the portion of the site presently in pasture was plowed and collected, since removal of the forest over the remaining part was considered too time consuming and costly. Several SYMAPs were produced for various artifact classes and combinations of classes. The two found to be the most useful for planning subsequent testing were SYMAPs for ceramics and lithics (Figs. 4-2 and 4-3). One major and several minor lithic clusters were observed on the SYMAP for lithics. The major cluster occurs along the northeastern part of the site and runs along the eastern side of a line separating the pasture and forested parts of the site. The cluster ends abruptly where the forested part of the site begins, indicating that it probably extends into this area. Ceramics, like lithics, occur in a single major cluster on the northeastern part of the site and several minor clusters occur over much of the remainder of the site. After plowing, several dark stains were observed on the site surface and were suspected to be middens. However, SYMAP artifact clusters failed to coassociate with all the stains, and doubts were expressed about whether the dark stains were actually the product of cultural activity. Backhoe trenching, which followed the SYMAP generation process, cut through several of the dark stains and permitted a better evaluation of their origin. Based on the presence of subsurface features, subsurface artifact densities, and comparisons with middens from other sites, three stains were then designated as cultural middens, with the remainder considered natural phenomena. The three middens were designated A through C and are shown on Fig. 3-8. As discussed in Chapter IV, two of the middens (A and C) roughly correspond to SYMAP clusters; Midden B is located in the center of the large ceramic cluster. Although Middens A and C are located near SYMAP ceramic clusters, the correspondence is not exact. This lack of agreement, especially in light of the good correspondence at other surface collected sites, is difficult to explain, but may be due to the type of cultivation undertaken at the site. The site was plowed for watermelons, and this activity may have horizontally dispersed the artifacts. Midden B's location in the midst of a large lithic cluster is interpreted to be the result of a ceramic period occupation overlying an earlier, nonceramic period midden. The earlier occupation produced the large lithic cluster observed on the SYMAP. The backhoe trenches were placed to bisect the dark


18

I

r

)

I

THE SITES

19

stains and artifact clusters observed on the SYMAPS. An initial trench was cut through the longitudinal axes of the large ceramic and lithic cluster located along the boundary between the pasture and forest. Other trenches were then dug perpendicular to the first trench to bisect the middens and other artifact clusters. Additional short trenches were dug at various locations for further investigation. Test units, varying in size from 1 x 1 m to 2 x 2 m, were excavated either to investigate the middens and obtain an artifact sample or to investigate features encountered in the backhoe trenches. All units dug to obtain a sample of artifacts from the middens were excavated in 10 em levels, and all dirt was screened through 1/4 inch mesh. Units excavated to uncover features were not consistently screened, depending on the nature of the feature, and in two instances a backhoe was used for quick removal of overburden. On the basis of the surface collection data, feature locations, and subsurface artifact densities, 5 artifact clusters (AC) were defmed and labeled WD95A through WD95E. These AC will be used as the basic units of analyses in later chapters of this report. Figure 3-9 shows the location of these clusters on the site.

FEATURES The following feature descriptions include the three middens. Also described are hearths and burials encountered at the site by backhoe and test unit excavations. For details on whole vessels associated with burials or skeletal materials, see Table 5-10 and Appendix ill, respectively.

Midden A This midden (part of artifact cluster WD95A) is located in the center of the terrace portion presently in pasture {Fig. 3-8). Midden A measures 11 m by 13 m, has a maximum depth of 50 em, and has the highest artifact density of any midden at the site. The soil is a dark yellowish brown {10YR3/4) color, has a pH range of 6.2 to 6.4, and possesses a phospate reading of 3, based on the value scale of Eidt {1973). Perpendicular backhoe trenches produced proftles of the midden's subsurface structure and a 2 x 2 m hand excavated unit provided additional data. A total of 110 sq m comprises the total area of the midden. Approximately 21% of the midden was removed by the backhoe, with hand excavation accounting for 4 percent.

Feature 1

This feature is a rock concentration found between 70 and 80 em below the surface along Trench B {Fig. 3-1 0). The rock consisted of quartzite and ferruginous sandstone,

most of which appeared to have been burned. All rocks lay on a common horizontal plane. The layout of the feature was irregular, with the greatest density in the center and scattered rocks along the periphery. The absolute limits of the feature were difficult to determine since occasional isolated rocks were found up to 50 em from the major cluster. The approximate outer dimensions are 1.3 m by 0.5 meter.

Feature 2 This feature is a hearth located along a trench through Midden C (Fig. 3-11 ). The hearth consisted of a 10 em thick, ashy layer beginning between 28 and 30 em below the surface and lensing toward the periphery. Located partially to the east of and partiaJly below the ashy layer was a red, burned clay layer. The clay layer was located 40 em below the surface and was 10 em thick, also lensing out toward the periphery. The clay apparently represents soil fired red from use of the hea.rth. The entire feature was located at the bottom of the midden stain. It is not clear why the feature is below the midden, since the site does not seem to have had appreciable soil aggradation . The hori· zon tal displacement between the ashy and burned red areas is also confusing. In other respects, such as color and content, this feature is similar to hearths found at other sites.

Feature 3 During profiling of a trench through Midden C, a possible posthole was encountered and labeled Feature 3. The posthole located in the north wall of the trench, was observable only below the midden in a lighter tan sand. It may have extended higher into the midden but was obscured by the dark midden color. A I x 1 m test unit was opened above the feature. The soil above the visible portion of the posthole was removed by a backhoe and the final stages of excavation were carried out with shovel and trowel. The posthole was 15 em in diameter and extended verticaJly below the midden for 45 centimeters. Its cross-section was parallel-walled with a rounded bottom. Time did not permit the expansion of the unit to locate other postholes.

Burial I This grave was encountered during the profiling of the west wall of a backhoe trench (Fig. 3-12). A vessel and some poorly preserved teeth and mandible fragments were encountered in the wall. A 2 x 2 rn unit was opened above the burial with all soil screened through 1/4 inch mesh. Density of artifacts above the grave was found to be no different from that of the immediately surrounding area. Two additional vessels were found in the unit, but no additional skeletal material was located. A slight trace of a grave outline was observable at the level of the vessels but was not visible above this. The outline indicated a portion of


20

WD95C

~

Ga£c ~W095B

I

MAG. NORTH

-- --

0

10

20

30

m eters

Figure 3-9.

Artifact Cluster Locations at the Grimes Site.

the grave has been removed by the backhoe trench. Screening of the backhoe trench backdirt produced a single fragment of human bone and two additional vessels reconstructable from sherds. Since the grave did not extend into the opposite trench wall, its maximum possible longitudinal extent was 1.45 meters. The observed width was 0.5 m, and orientation of the grave was east-west. The mandible and teeth fragments were located in the approximate center of the grave; orientation of the head could not be determined.

Possible Burial 2 A small vessel partially reconstructed from sherds found in backdirt from trenches placed through Midden A is the only evidence for Burial 2. The sherds all had fresh breaks and suggested that the vessel was complete before testing. Despite examination of the trench wall and testing around the locality of the vessel fragments, no other evidence of a burial could be found. The burial may have been entirely

mE SITES

21 on the northern side which grades to a 7% slope on the southern side (Fig. 3-14). The flood plain is 0.7 km wide in the vicinity of the site. The upland projection is part of the Wilcox Formation, an Eocene deposit. Since its formation, the surface of the site has been relatively stable, with the exception of soil erosion along the slopes due to recent agricultural practices. The projection does not flood today and, considering its height, probably did not in the past; consequently, no source of soil aggradation exists. Where completely intact, the A horizon is approximately 30 em deep and consists of an organically stained sandy loam. The B horizon consists of clays derived from illuviation of the overlaying A horizons; B horizon depth is undetermined. The site is partially in pasture and partially in timber. Trees are located along the northern and western slopes, and pasture is confmed to the top and the eastern slopes. A few oak trees are located on the crest of the site landform.

Figure 3-10.

Subsurface Rock Concentration (Feature 1) at the Grimes Site.

removed by the trench, or alternatively, the vessel may have been located within the midden unassociated with a burial.

HINES SITE (X41WD87) The Hines Site is located along Burke Creek, a major tributary of Lake Fork Creek. The site is situated on an upland projection into the Burke Creek flood plain (Fig. 3-13). This upland projection is about 40,000 sq min size, with artifactual material scattered over about half of this area. The projection is about 8 m high and has a 15% slope

RESEARCH METHODOLOGY The first phase of research at the site was plowing and surface collection. Surface collection was accomplished in 7 mandays, and the portion of the site over the crest and along the upper part of the southern slope was collected in 5 x 5 m units. The lower portion of the southern slope was collected as a single unit because of the timber line which presented an artificial boundary to collection. The area on the upper slope and around the crest was unrestricted enough to permit meaningful artifact clusters to be delineated on the SYMAPs. Although a few trees are present on the crest, these were sparse enough not to seriously bias the results. However, on the lower slope the tree line represented a very irregular boundary which would have had marked effect on SYMAP artifact clusters.

ver y da r k brown sandy loom IOOYR21Zl

}"''"" """ A- HORIZON

No modd en stoon bro wn loom UOYR4/3>

Figure 3-11.

Hearth (Feature 2) at the Grimes Site.


22

0 10 t..u.u.J

em

------------------ -------

1

(IOYA4/31

very dork Qreyish brown sandy loam llaYA3tll

dark yellow!Sh brown sand I10YA3141

GJUV[

PIT OUTLI N(

Und tslurbed matrix

1- A-HORIZON

dor~

brown sandy toom tiOYA!I!J

-+----+-----1'' 'I VUS[ \

,

~

L

W

.

0

'RtG W£NTS

//

/

/

/

//

/

,'

by backhoe trench

/ ,I / / / / / / / /

////,1 ,I ////////////

//

/Removed by backhoe trench

/

Level of hornontol plan

ferruo no us

l:tTHTH & MANOIIL£ Otslurbed

-,..-- -r-'>-;..--,. ~ -,.....- .,..., 7 /

I

vnuL Posllton of a ~'i----~,---44-.verl tcol sectton

1

sandsta'e

oyer

1-r-r--r-r-:r---r-?---r--T---r-r-"..,...--,..--,r-r-7~ J

/

////////////

PLAN VIEW Figure 3-12.

VERTICAL SECTION Burial 1 at the Grimes Site.

Several SYMAP artifact density maps were produced for the portion of the site collected in S x S m umts. Maps found to be most useful were lith1c debris, daub, and ceramics (Figs. 4 12, 4-13, 4-14). Four artifact clusters were defined in the basis of surface collection (Fig. 3-1 S). The largest, labeled W087 A, is located from the crest of the

Figure 3-13.

Hines Site (X4l WD87) with Respectto Local Topography ( 10ft. contour intervals).

upland projection downslope to the east for a distance of S meters. Artifact cluster WD87 A corresponds to a very slight midden stain observed on the surface after plowing. The second cluster, WD87B, is located IS m west of cluster WD87 A and IS generally situated on the crest of the proJection. The third, undesignatcd cluster, is located S m west of WD87 A and 15 m and north ofWD87B. The third cluster is small in comparison with WD87 A and, considering its position on the periphery of the surface collected area, apparently extends into the wooded portion of the site. A fourth cluster, WD87C, was defined as the area on the lower slope that could not be collected by a grid system but nonetheless possessed a large artifact sample. Backhoe trenches were excavated to bisect the three artifact clusters. WD87 A was further bisected by a second trench perpendicular to the first. The trenching was conducted to check for subsurface features. The trenches through WD87 A were extended several meters beyond the associated midden to check for house patterns. At the De Shazo Site, an early Historic Caddoan site in cast central Texas, house patterns were found around a midden (D.A. Story, personal communication), and the trenche~ through WD87 A were extended to check for this possibility. The backhoe trenches through WD87 A were productive in that three overlapping house patterns (Feature 1) and a pit (Feature 2) were found. However, the patterns were found below and not around the midden stain. Trenches through the other two clusters failed to indicate any subsurface features. The overlapping house sequence in WD87 A was encountered because the last house in the sequence had burned and left a compacted mass of house debris. It is possible,

23

THE SITES

/

I

I 1ft

Figure 3-14.

c o ntour l n te r•o l

Hines Site (X41WD87).

based on surface cluster similarities, that houses were also located in the areas of WD87B and the third, unlabeled cluster, and that these houses, not having burned, were not observable in the trenches.

FEATURES Feature 1 Feature I initially appeared as a compacted mass of house debris in a backhoe trench placed through the center of cluster WD87 A (Fig. 3-16, C-C'). This feature appeared most prominently in the west wall and was scarcely visible

in the east wall. A 2 x 2 m unit was excavated to expose a horizontal surface over the feature. This unit was located in a very slight midden stain observed on the surface, and the midden soil, a dark brown (IOYR3/3) sandy loam, formed the first level of the unit. The burned layer, 10 to 20 em below the surface, was to be the next natural level, but was too hard for hand excavation. The layer consisted of a compacted mass of charcoal, clay daub with wattle impressions, and numerous artifacts. Before removal of the burned layer, its horizontal extent was defined. Since the amount of time that could be spent at this site was limited, mechanical scraping by a front-end loader was used to remove the midden that covered the burned mass.


24

~...

.,, S .. ftoee

CCJ•IeC:flOI'I

L 'l'tlt

I

p MAG

0

~

NORTH

10

15

~ me. len

Figure 3-15.

Artifact Cluster Locations at the Hines Site.

The burned area was found to be approximately 4 m in diameter, thickest and most compacted in the center and lenticular and less compacted along the periphery. After careful mspection. the burned feature was removed by mechanical stripping to the underlying clayey B horizon to look for postholes, since it was thought the mass mtght represent debris from a house. This removed an additional 10 to 15 em of soil and exposed a number of charred and noncharred but dark stained postholes The new surface was trowelled 3 separate times before the overlapping house patterns became clearly visible. The patterns were best observed when the soil had become water saturated and the sky was overcast. The house patterns were designated Feature 1 (Figs. 3-16 and 3-17). Due to limited recording time, a number of important attributes of the postholes (such as crosssections, composition, and depth} were not taken. Instead, a detailed horiwntal map was made and cross-sections of superimposed postholes were made to help unravel the construction sequence. The cross-sections were only partially useful for establishing the sequence since only two sets of two postholes were superimposed. However, from the available data, the following building sequence is proposed. The circular house was butlt first This is based on crosssections A-A' and B-B· (Fig. 3-16); both of these secuons show that the circular house "interior support" postholes were later cut through by posts of the small and large rectangular houses. This assumes that the circular house "interior supports" arc in fact correctly identified as belonging to this structure. The rationale for including these posts with the circular house is that 4 of the 5 interior support postholes are positioned about midway between the perim-

eter and the center of the circular house. The fifth interior support posthole is the central post. Moreover, all 5 posts are about equal in size and depth and have the same soil characteristics (brown sandy loam with numerous charcoal flecks). The or dcr of the last two house~ rs based on the interior supports of the large rectangular house Two postholes arc posuroned at pomts cqurdtstant from each other and from the walls of the large rectangular house. strongly implying that they arc in tenor supports tor this structure The southernmost mterror support of the large rectangular house was well charred and incorporated into the burned feature found above the pattern Since no evidence of later occupation over the burned mass was found the rectangular house apparently burned and was the last house in the sequence. Through indirect reasoning, the small rectangular house must have been built between the ctrcular and large rectan· gular houses. Further evtdencc supporting thts order is the entrance of the small rectangular and the .:ucular houses. These two en trances correspond exactly. suggestmg de tailed knowledge of the circular house layout during con'>t•uctron of the small rectangular house. Moreover, the large rt•ctan· gular house entrance cuts across the small rectangular house entrance. A total of 217 postholes was uncovered. Thu ty-four be· long to the circular house, 81 to the small rectangular house, and 57 to the large rcdangular house Forty-nine postholes were not associated directly with any of the house patterns and probably represent posts for interior dividers or supports. Spacing between posts varied significantly among the houses. The circular house had a mean spacing distance of 44 em (s = 5 6 em), the small rectangular house had a mean of I 0 em (s == 4.1 em), and the Jar gc rectangular house had a mean of ~3 em (s == 6.1 em). The posrtronmg of the houses strongly argues for their rcprescntrng part of a scqucntral building episode. The shgh t ~hi ft of each successive structure is similar to the pattern observed at the De Shato Site in east e;entral Texas, where house pattern shifts were thought to be due to house rebuilding and the lesser difficulties of digging holes for new posts (D. A. Story, personal communication). If we assume a maximum of 20 years occupation per house (D. A. Story, personal communication), then we can project a 60 year occupation length for the total house sequence. Feature 2

The second feature located at the Hines Site was a pit positioned 2 m directly north of the circular structure. The pit was visible by its darker color (1 OYR2/2 in relation to 1OYR4/3 of the surrounding soil) and by its basin-shaped bottom. It was 2 m in diameter, and its depth was difficult to determine because the pit top blended into the A horizon {Fig. 3-18). Estimated depth was 30 centimeters. The

I

THE SITES

25

M AG

0

I

NOR T H

...

50

5ZSl!!!

- - - - - BOU N DARY OF EXCAVATED

All [I• - - --

--\

0

0

0

0 5

0

:~ 0

@

0

...• •..

0

6'

0 0

0

0

•• •

A'

0

0

0 0

·t ····;

•

•

0

0

~

: e>8~~

iPo 0

0

~

G)

®

0 •

® 0

0

S M ALL REC TA NGUL A R H OUSE LARGE RECTANGULAR HOUSE

~

G

Cl4 ADI040!50

•

® ®

®

a•

•• •••••

0

0

~ @

0

~ ~

0

~

0 ~

• :0

••••

0

(l)

®

®

0

8 ~

0

Q

~

••

0

0

•

~

•• •• •• 0 • •

0

0

®

•• • ••

0

0

0

®

@

0

® 0 ®

~

0

0

CIRCULAR H OUSE

0

0

U N ASSOCIATEO POSTHOLE

c' wery dork brown

Gophe r borrow

c

t=~~::::;:;~~;:.~~~=~~~~~~,SIItM

(IO'ntiiU

Burnt lit

"'iddoootolo}

hon e d t bri t

L tu l o f -- uroptd t urfoct )tiiO W ith

brow ft (bvn e d) IOftci/CIIJ UOY . tlt )

Figure 3-16.

House Patterns (Feature 1) at the Hines Site.

CIOJ

l~hori J Oft

26


Figure 3-17.

Partially Exposed House Patterns (Feature 1) at the Hines Site.

contents of the pit were sherds (many larger than found elsewhere on the site), lithic tools and debris, and poorly )j)reserved faunal remains. A flotation sample was taken from the lower portion of the pit (see Chapter VII for results of the floral analysis).

dark 1tllowosh brown Fo•nt pll , aond1 loom UOYU/41 out11ne Y '

,I

, A-horuon B-hoflzon

(10YA2/21

Figure 3-18.

Pit (Feature 2) at the Hines Site.

KILLEBREW SITE (X41WD64/69) The Killebrew Site is located adjacent to Caney Creek, a major tributary of Lake Fork Creek. The site is situated on a terrace surrounded by the uplands to the west and the Caney Creek flood plain to the east, north, and south (Fig. 3-19). The terrace covers an area approximately 80,000 m square and is about 8 m above the level of the flood plain. The flood plain is 400 m wide in the site vicinity. Although the terrace is relatively level, the periphery slopes about 1 m vertically for each 10 horizontal meter. Localized erosion, accentuated by recent agricultural practices, has created several small, meter-high knolls on the terrace (Fig. 3-20). Most prehistoric occupation is located atop these knolls. Formation of the terrace occurred during the Pleistocene and since this time the terrace has been a relatively stable surface with little or no soil build-up (Corwin 1978). Since the terrace is not inundated by Caney Creek overflow, the only potential aggradational source is the uplands to the west. However , the distance to the site from this landform

THE SITES

27 classes and combinations of classes were produced , the maps of lithics and ceramics were found most useful for subsequent testing at the site (Figs. 4-10 and 4-9). Eight artifact clusters were defmed based on the maps (Fig. 3-21). Six of the clusters correspond to middens observed on the surface of the site and in subsequent backhoe pro fLies. The clusters which correspond to middens are WE64/69A through WD64/69E and WD64/69G. Seven of the clusters WD64/69A through WD64/ 69G, consist primarily of ceramics and represent Caddoan occupations. The remaining cluster, designated WD64/ 69H, is a result of the distribution of lithics. The ceramic clusters are scattered generally on slight rises across the site. An exception to this is WE64/690 and its associated midden, located on the slope of a swale. Whether this cluster was originally situated on a rise and erosion has created the present topography , or whether the midden was intentionally placed on the slope could not be fully determined. Some erosion from the top of the deposit has taken place, but the amount of soil removed does not seem sufficient to account for the present elevation of the midden. The midden probably was created upon the slope. Backhoe trenches were positioned across the site to obtain proflles through the SYMAP clusters and to look for subsurface features. All backhoe trenches, with the exception of a trench which bisected cluster WD64/69F, were connected to provide a continuous proflle and to locate features unassociated with the clusters. The distance between WE64/69F and the other clusters was too great to connect them with a backhoe trench. The backhoe trenches through WD64/69A, WD64/69B, WD64/69D and WD64/69 E were extended beyond the cluster limits by several meters to look for house patterns. At the DeShazo Site in east central Texas, house patterns were found around a central midden (D. A. Story, personal communication), and a similar pattern was thought possible at the Killebrew Site. All trenches were proflled along one face and in certain instances, particularly when an anomaly was located, along both faces. The northern or western walls were proflled to provide optimal sun exposure. The final phase of work at the site was the excavation of test units. Consideration shifted at this stage to obtaining a sample of artifacts from each of the clusters. The profile information verified the integrity of the clusters by demonstrating that each represented a subsurface as well as a surface locus of cultural debris. A sample of artifacts from each cluster was felt to be important for comparisons and interpretation of cluster function. Two by two m test units were used exclusively for this purpose. Since the site surface has been relatively stable (i.e., nonaggradational; Corwin 1978) and since soil churning from biotic activity has moved artifacts within the soil, the test units were dug without vertical provenience. This lack of vertical control is further justified considering the shallow subsurface deposit under the clusters, which in most instances is less than 30 centimeters.

.

Figure 3-19.

Killebrew Site (X41WD64/69) with Respect to Local Topography (10 ft. contour intervals).

is great enough so that appreciable colluviation probably has not taken place. The soil profile at the site supports the interpretation of long-term stability for the terrace. The A horizon consists of a dark brown, organically stained, silty loam. Below this is the B horizon, made up of yellow to brown silts and clays. Thicknesses of the A and B horizons vary considerably from area to area throughout the site. Excluding areas where erosion has altered the proflle, the A horizon is 20 to 40 em thick and the B horizon is 30 to SO em thick. For the past several years the site has been in pasture, but prior to this, according to local inhabitants, it was under cultivation. Plowing, with the resultant increase in erosional susceptibility of the soil, has probably been a major factor in exposing the B horizon along the periphery of the terrace.

RESEARCH METHODOLOGY The first step in testing the Killebrew Site was the excavation of nine I x 1 m test units across the terrace. The site was initially plowed, but while waiting for sufficient rainfall to expose artifacts on the surface, two suspected middens (i.e., dark stains observed on the site after plowing) were tested. After sufficient rainfall, the site was systematically surface collected in 5 x 5 m units. Collection of the site required 14.25 mandays. · Although several different SYMAPs for Qifferent artifact

28


A("eo",~:,I

'SJ

V I~OE•

CAlVEY

CREEK

FLOOD PLAIIV IrA

I DOE N t1

• I

A

f£A1 URE 4 !Rue.._ Conc e n i (OI•O"~l • I FE AT+JIH .!

l otecftl>l

/ /

•

I

G TURf

sq m}

TEST

4 sq m

•

W!OO(N

BACKHOE

UNITS

!tl•·••'"'

M! 00€N f

TRENCHES

/

y /

/

/

/

/ \t AG NORH

/

// \

,..-, I

( e I

\

FE~TUFIE 2 I ROlli CH ll'lfOl•O,I

' _,..POSS , 8L£ I

" OOEN t

FEATURE INSERT

\\

rS

\

L imits

t

M AG. NORT H

0 15 l""5iiwe

30

meters

/

50 em

/Figure 3-20.

Killebrew Site (X41 WD64/ 69).

contour

interval

~

THE SITES

29

---o-

MAG . NORTH

0

es;;;10

20

30 I

me lara

~ ~WD64/69D

Su rface

Co llect• on

l tmtf l

l WD64/69F

Fi~re

3-21.

Artifact Cluster Locations at the Killebrew Site.

FEATURES Midden A

•

This midden (part of artifact cluster WD64/69A) is comparatively small, measuring 50 sq m in area, and is located adjacent to Midden B (Fig. 3-20). A 2 x 2 m test unit (8% of the total midden area) was excavated in this feature, and this unit plus the backhoe trenches represent a 28% sample of the midden . Maximum depth of the midden , which occurs near the center, is 40 centimeters. Phosphate levels range from 0 to 3 based on a scale devised by Eidt (1973), and pH is 6.0, or slightly acid. Darkest soil color occurs at the center and is IOYR3/4.

.. Midden B Midden B (part of artifact cluster WD64/69B), covering a 250 sq m area, is located immediately to the southwest of Midden A (Fig. 3-20). In fact, the two stains blend together on their tangent peripheries. Backhoe and test unit excavation removed 12% of Midden B, with hand excavations alone accounting for two percent. Maximum depth is 45 em, and phosphate values range from 2 to 4. The midden is slightly acid with a pH value of 6.0, and the darkest midden color is dark brown (1 OYR3/3) in the center. Midden B shows similarities to Midden A in SYMAP

results. The map for ceramics shows an elongated cluster over both middens, while the map for aU prehistoric artifacts combined tends to show an elongated cluster with two peaks. Each peak corresponds to one of the midden centers. The SYMAP for clay daub also shows two peaks, each corresponding to the middens.

Midden C Midden C (part of artifact cluster WD64/69C) is located 20 m south of Middens A and B (Fig. 3-20). It covers an area of approximately 450 square meters. The 2 x 2 m test units were excavated in Midden C because the SYMAP artifact clusters did not exactly correspond to the midden stain observed in backhoe trenches. One unit was placed in the center of the artifact cluster representation on the SYMAP and one was placed in the darkest part of the midden stain. Two percent of the midden was excavated by hand and a total of 5% of the midden, including the backhoe trenches, was excavated. Darkest midden color was a dark yellowish brown (10YR3/4) which contrasted with the surrounding yellowish brown (lOYRS/6) soil. Phosphate readings in the midden varied from 1 to 3 and pH was 6.0 , or slightly acid. Depth of the midden was 35 em near the center.

PREHISTORIC SETTLEMENT PATTERNS AT LAKE FORK RESER VOIR

30 Midden D

This locality (part of artifact cluster WD64/69D) is situated in the central portion of the site, 35 m from Middens, A, B, and C (Fig. 3-20). Midden Dis unique in comparison to others at the site in that it is situated on the slope of a swalc. As discussed earlier, the midden apparently was formed on the slope. The deposit covers an area of 180 m and a total of 18% of the midden was excavated, including the backhoe trenches. The test units account for 5% of the midden area. The midden is located above a clayey soil and is about 20 em thick. Midden color is dark yellowish brown (lOYR3/4). Phosphate levels range from 1 to 5, the latter ncar the center of the midden. Midden E

Midden E (part of artifact cluster WD64/69D) is unusual. in that it is located farther from the terrace edge and presumably is farther from a water source than any of the other middens (Fig. 3-20). The deposit, which is located atop a small rise, is 115 sq min area. Twenty-four percent of the midden was removed through backhoe and hand excavation, with 7% removed by hand excavation alone. Midden color was dark brown (10YR3/3) near the center grading to yellowish brown (l0YR5/6) outs1de the midden. Phosphate levels range from l to 3 with a pH range of 6.0 to 7.2 (from slightly acid to neutral).

observed in the backhoe trench and partially uncovered by the 2 x 2 m unit. The midden measured 11 m in diameter (based on the observable stain) and 16% of the 95 sq m area was excavated by backhoe or by hand. The 1 x I m and the 2 x 2 m test units were screened through 1/4 inch mesh: the 1.5 x 2 m unit was not screened. Midden depth was difficult to determine, but based on the artifact densities seemed to be about 40 centimeters. Feature I

This feature is a hearth located in the approximate center of Midden E (Fig. 3-22). In vertical proflle it consists of two parts. The first is a very dark grayish brown (JOYR3/2) layer (the gray color due to higl1 ash content) which overlies a reddish yellow clay (7.5YR6/6). The grayish brown layer begins 10 to 15 em below the surface and extends downward another 8 em in the center; this layer lenses out along the periphery. The reddish yellow clay layer begins immediately below the grayish brown layer and extends 5 em deeper in the center, also lensing out along the periphery. In horit.ontal plan, the hearth IS oval and measures 32 em by 42 centimeters. The reddish yellow color of the clay is the result of heat. The entire grayish brown layer was collected for flotation recovery. The pH of the grayish brown level was 7 .2, or neutral.

Midden F

At best, this is a possible midden (part of artifact cluster WD64/69F) since only a very slight soil stain was observed in the field (Fig. 3-20). The observed color may in fact be the result of natural processes. Midden F was detected by the SYMAP for ceramics (Fig. 4-1 0), but only by a very low artifact density. The low surface density is confusing since subsurface testing found a relatively higher density. No appreciable soil aggradation seems to have taken place on this part of the site, eliminating this as a cause of the low surface dcnsi ty. The majority of artifacts simply were located below the plow zone, in contrast to the other artifact clusters. It may be that bioturbidic activity has been more prevalent in this area, resulting in movement of more artifacts to lower levels. However, no definitive evidence exists in support of this hypothesis. Midden F was examined by a single backhoe trench, which bisected it on an east-west line, and by three test units. The first test unit was a 1 x 1 m pit placed prior to surface collection to investigate the midden stain observed on the site after plowing. A 2 x 2 m unit was dug later to obtain a larger artifact sample. A frnal unit, measuring 1.1 x 2 m, was excavated to uncover the remainder of Feature 2

brown cloy t>OYR$/il

UNEXCAVATED

ol:~·',o

// /

10

Figure 3-22.

Hearth (Feature 1) at the Killebrew Site.

Feature 2

This feature consists of a rock concentration found in the center of M1dden F (Fig. 3-23). The rock, most of which shows evidence of burning, consisted of ferruginous sandstone and minor amounts of petrified wood lying on a common horizontal plane about 20 em below the ground. Midden artifact debris (including sherds, lithic debris, tools, and fire cracked rock) was found around the rock concentration, above the feature, and generally throughout the midden. The feature position relative to the midden suggests that the rock had been placed on a level floor. Feature

THE SITES

31 diameter. The rocks seem to have been burned, and unlike Feature 2, they form a roughly circular ring. No associated charcoal or other artifactual remains were encountered.

--'~.

!\.'

Figure 3-23.

Subsurface Rock Concentration (Feature 2) at the Killebrew Site.

2 measures about 2 m x 1.2 m, compensating for a portion taken out by a backhoe trench, and individual rocks measure between 2 to 20 em in diameter. The distribution of rocks was not uniform within the limits of the concentration and no patterns could be seen in the distribution. No charcoal or other evidence indicating that the feature had been used as a hearth were noted. Feature 3 A hearth similar in appearance to Feature 1 was uncovered in a backhoe trench in the approximate center of Midden B and was labeled Feature 3. The backhoe had removed the majority of thls hearth (a profile was not drawn due to disturbance) and what remains consisted of an orange clay layer intermixed with several large pieces of charcoal. Feature 3 extends from 30 to 55 em below the surface and is 30 em in diameter. A radiocarbon date obtained on the charcoal gave an age of 190 ± 50 (TX-3045). This date is too old for the Caddoan midden above the hearth, so apparently the charcoal is from an earlier Archaic Period occupation. However, the majority of the artifact deposit from Archaic occupation is along the bluff edge, about 30 m east of the hearth.

Feature 4 This feature (Fig. 3-24) is another rock concentration found in the approximate center of Midden B near Feature 3. It consists of ferruginous sandstone lying on a common horizontal plane; no examples varied from this plane. The depth to the top of the rocks is 40 to 50 em, and the horizontal extent of the feature, estimated from the remaining portion untouched by the backhoe, is 1 m in

Figure 3-24.

Subsurface Rock Concentration (Feature 4) at the Killebrew Site.

OSBORN SITE (X41WD16) The Osborn Site is situated upon a moderately dissected remnant terrace and is located 60 m south of the present channel of Lake Fork Creek. The Lake Fork Creek flood plain surrounds the terrace on the northern, eastern, and western sides, and uplands adjoin on the southern side (Fig. 3-25). The terrace drops steeply in elevation on the north, east, and west, while sloping gently upward on the south. The terrace surface is approximately 43 ,200 sq min area, with the majority of prehistoric artifacts found on two knolls along the northern periphery. The terrace was arbitrarily divided into two areas, A and B; Area A was the focus of work in 1975 (Bruseth and others 1977) and 1976 and Area B was the focus in 1978 (Fig. 3-26). The A horizon on the terrace is a dark to light brown silty loam, ranging from 30 to 60 em in thickness, although it is thinner in eroded areas on the northern margin of the terrace. The B horizon averages 30 em in thickness and consists of yellow clay. The Osborn Site was planted in pasture when located in 1975. Area A was extensively damaged in 1976 by Reservoir-associated bulldozer activity. In 1978, the site vegetation was Johnson grass, scattered persimmon bushes, and a few oak trees along the northern terrace edge.


32

Figure 3-25.

Osborn Site (X41 WDI6) with Respect to Local Topography (I 0 ft. contour intervals).

RESEARCH METHODOLOGY Portions of the Osborn Site had been systematically surface collected during the 1975 field season. Most of Area A was plowed while Area B was partially plowed (Bruseth and others 1977:75). Two dark brown midden stains, Midden A in Area B and Midden B in Area A (Fig. 3-26), were observed on the surface at this time. The subsurface content of each midden was tested in the 197 6 field season. A 1 x I m test unit was placed in the approximate center of Midden A, while four 1 x 1 m test units were placed in and around Midden B. Test pits were excavated in arbitrary 5 em levels with the plow zone removed as one level. All soil was screened through 1/4 inch mesh. Due to the destruction of Area A by bulldozing, 1978 work at Osborn concentrated on Area B. To initiate the 1978 work, approximately 4 ha of the terrace surface including all of Area B. were plowed. Artifacts were surface collected in 5 x 5 m units, with 57% of the plowed area collected. SYMAPs were computed on the surface artifacts from both the 1975 surface collections and the 1978 surface collections, and these maps indicated several minor and three major artifact clusters on the terrace. The major clusters include one in Area A (WD16B) and two in Area B (WD16A and WD16C) (Fig. 3-27). Cultural material was dispersed over Area B along the center of the terrace knoll

upslope from the eroded terrace edge. Artifact cluster WDl 6A, associated with Midden A. has a high density of ceramic artifacts. WD16B, associated with Midden B, also has high ceramic surface density . WD16C is too poorly investigated, due to its destruction before final testing, to be assigned to a cultural period. Before backhoe trenching was conducted, WD16A was tested with five 1 x 1 m excavation units and WDI6D was tested with five 1 x I m excavation units and WD 16C was evaluate the significance of the cultural deposit in order to avoid unnecessary backhoe damage. All hand excavated units were dry-screened through 1/4 inch mesh with vertical control of 10 em levels. Backhoe trench locations were placed primarily in and around WDI6A, since WD16B and WD16C had been damaged in 1976. Eleven backhoe trenches, 0.8 min width and of varying lengths, were placed across the Osborn Site, bisecting the approximate center of each SYMAP-defined artifact cluster. The lengths of individual trenches were dependent upon the size of the artifact clusters, and certain trenches were extended to investigate l.:>w surface density areas adjacent to the clusters. No cultural features were located in these areas. Backhoe trenches were excavated into the underlying B horizon to a depth of 50 to 150 centimeters. After initial trenching and hand testing, exposed cultural features were excavated. At the conclusion of excavation, a 34 m x 18 m area m and around Midden A was stripped by a bulldozer to a depth of 60 em (the top of the B horizon) to search for additional cultural features below the midden. A road grader was next used to smooth the area, providing a clean surface conducive to locating cultural features (King 1977). Several features were located and are reported below.

Midden A Midden A (part of WD16A) is a distinct black silty loam deposit (Fig. 3-26). The midden is 64 sq m in area, and is located on the crest of the knoll in Area B. It is likely that part of the knoll elevation is the result of midden accumulation. Midden A is 50 em thick at the center and gradually thins out along the peripheries. Chemical analysis of soil samples taken along a backhoe trench through the midden shows increased quantities of phosphate in the midden . Excavated units were placed across Midden A to sample intra-midden variability in artifact. ecofact, and feature contents. Six 1 x I m. one 3 x 3 m, and four .25 x .25 m units were excavated. One of the 1976 excavations units was also located in this midden. In all, 11 sq m of the midden were hand excavated (I 7.2% of the deposit), and 6.2 sq m were excavated by backhoe trenches.

Midden B Midden 8 was destroyed by a Reservoir contractor be-

t

..

,,JO%to ~20%. >20% to ~5%, >45% to ~1 00%. SYMAPs were run at the Bradfield Computer Center at Southern Methodist University Since the SYMAP program has a limitation of 1000 data points, several partial maps were produced from sites and spliced together. In cases where it was necessary to run more than one map, contig· uous borders were overlapped by one or more rows of col· lection units to provide continuity between sections for the overall map. One map per artifact class was computed, as well as additional maps for combinations of artifact classes and a final map for all artifact classes combined. The maps were produced in approximately 4 hours of computer terminal work for each site. Actual computing time averaged 30 seconds per map. Key punching varied from site to site but generally took about 2 to 6 hours. COMPARISON OF THE SYMAPS WITH SUBSURFACE DATA

POTENTIAL PROBLEMS A major assumption underlying the use of surface col· leeted artifacts to define horitontal components is the belief that the surface and subsurface distributions correspond to the same cultural phenomena. Although this assumption has been tested at a few sites (e.g., Binford and others 1970; Redman and Watson 1970; Brockington 1976), gencralitations cannot be extended to all sites. In fact, it is doubtful that generalizations about the surface-to-subsurface correspondence of artifacts will ever be possible since this is directly dependent on the nature of individual sites, something which obviously varies from site to site. The major variable controlling artifact surface-to-subsurface correspondence is soil genesis. If for instance, a site was reoc· cupicd over time and soil genesis was constant through

time, a collection of surface artifacts would not necessanly reflect the subsurface artifact distribution. On the other hand, if no soil build-up occurred between occupations, then there would be no subsurface artifact deposit, and the surface artifacts would represent the entire populatton . Since at least before human occupation, all cxarnim•d sites in the Reservoir have possessed nonaggrading surfaces (Durler 1976; Corwin 1978), thus presenting the situation where artifacts were deposited on a stable surface. However, these sites arc also affected by an additional variable that influences vertical artifact distribution. This variable is bioturbation, or soil churning due to biotic activity (Wood and Johnson 1978; Brown 1974). The five Lake Fork sues considered in this chapter all have considerable deposits of subsurface artifacts, despite their nonaggrading surfac.:cs. The artifacts were deposited on the surface during site oc· cupation and became vertically displaced through bioturba· tion (Durler 1976). A common type of bioturbation obscrv· able at the sites is gopher burrowings, and in deep sandy deposits "pockets" of artifacts, collected and deposited by these animals, can often be found. The effect of bioturbation upon the artifactual depostts can be best seen in the sotl profiles from the sites. The typical profile consists of a sandy loam A hori10n , a clayey B horizon, and C horiwn. The thickness of the A hodton varies from a few cenhmcters, in areas where erosion has removed overlying soil, to over a meter in undisturbed areas. The B horizon is formed by clays leaching from the A horiwn and collecting at a common depth below the sur· face (Durler 1976). Invariably, where the A hori1.0n extends deeply, artifacts are also encountered deeply. In many cases artifacts arc found directly on the clay B horiwn. Given that this B horizon formed through leaching and was never an exposed surface, these artifacts must have migrated downward from the surface. In addition to simple vertical displacement of artifacts, the concept of bioturbation brings into play othrr variables. Do all artifacts filter at the same rate? This is difficult to answer precisely, but all Archaic deposits at the sites had surface manifestations. suggesting that all artifacts either had not flltered downward or that some had been churned back to the surface. Thus, a period of at least 2500 years is not sufficient to completely move all artifacts to the B horiwn. At the other extreme. evidence from an historic site in Hugo Reservoir shows artifacts can move vertically as much as 76 em in 145 years (Rohrbaugh 1972). Although this evidence shows variability in artifact Hltra· tion, what is here termed an "age effect" docs seem to operate. That is, artifacts that were deposited on the sur· face during time period A. for instance, have had more time for vertical displacement than artifacts deposited on the surface during period B. The results of the age effect can be seen in Fig. 4-1, which displays the vertical distribution of expanding stem dart points, contracting stern dart points,

57

EVALUATION OF mE SYMAPARTIFACT DISTRIBUTIONS

-------------~ -------------~~~=-----~-~-~-::~

0-6 7-9

-., Q)

.s::. (.) c: Q)

-

---

---;-.,.- ceramics

10-12 13-15 16-18 19:.21

(.)

0

,_

;::,

en ~

0 Q,l

CD

-

.s::.

Q.

Q)

Cl

(N=50)

-Contracting Stem Tradition (N=88)

22-24 25-27 28-30 31-33 34-36

-Expanding Stem Tradition (N=86)

37-39 40-42 43-45 46-48 49-51 52-54 0

4

Figure 4-1.

8

12

16

20

24

28%

Vertical Distributions of Artifacts at the Yarbrough Site.

and pottery sherds from the Yarbrough Site, 20 km southwest of the Reservoir. This site has been selected as an example because of the large sample of materials attributable to three cultural periods: Middle Archaic (expanding stem dart points), Late Archaic (contracting stem dart points), and Caddoan (sherds). The Middle Archaic expanding stem dart points show the deepest penetration, the Late Archaic contracting stem dart points show the next deepest penetration, and Caddoan sherds display the shallowest penetration. However, considerable overlap among the three artifact classes is seen.lt is suggested that the distribution of these three classes represents the effect of bioturbation upon cultural assemblages deposited at different points in time on a stable surface. If so, the "age effect" can have profound influence upon surface collection results, with earlier deposits underrepresented with respect to later de· posits. In addition to the vertical dislocation of artifacts, biotur· bation also affects horizontal provenience. However, the process does not have the destructive effect on horizontal provenience that it does upon the vertical integrity of artifact deposits. The reason for this is the result of scale differ· ences between the horizontal and vertical patterning of artifacts. Vertical patterning seldom involves more than a meter or two, while horizontal patterning often is spread over a much greater area. A 50 em dislocation, for instance,

is far more destructive to vertical integrity than to horizontal integrity. Thus, despite horizontal disturbance, larger activity loci such as trash middens and household debris concentrations are still defmable. At the same time, smaller activity areas such as single chipping localities or cooking locations may be too thoroughly mixed with surrounding debris for recognition.

CORRELATION WITH SUBSURFACE ARTIFACTS The first criterion for evaluating the relationship between surface artifact density and subsurface artifact and feature distributions is surface-to-subsurface artifact correlation. The surface sample from a site was used as a planning tool for subsequent site testing and a main concern was how closely the surface sample matched the subsurface sample. Three sites have been selected for consideration of this problem: Grimes, Osborn, and Taddlock. Although surface collections and artifact density maps were also made for the Killebrew and Hines Sites, the three selected sites adequately reflect the range of surface to subsurface results obtained and, for brevity in presentation, are used as examples. Two artifact classes, ceramics and lithics, have been selected for the surface-to-subsurface comparisons. These classes were selected because artifact clusters on the maps


58

generally fell into two groups corresponding to these classes. The lithic clusters were identified with Archaic occupations while the ceramic clusters were linked to Caddoan occupations. At the Grimes Site, ceramic and lithic artifact surface/ subsurface correlations (Figs. 4-2 and 4-3) show agreement at only a gross level. Subsurface test unit densities are shown in Table 4-l. Highest ceramic densities are from unit 9 which is within a contour level C and lowest sursurface densities are from units 1, 2 and 3 which are located outside contours. Units 4, 5, and 6 show moderate densities although located on the periphery of ceramic clusters. Units 7 and 8 are located further from a ceramic cluster and simply do not correlate well with the surface distributions.

Table 4-1 Subsurface Ceramic and Lithic Artifact Densities from Test Units at the Grimes Site. Unit l

Ceramic Density 1

0 8

2 3

2 48

4 5

41

Lithic Density 10

15 7 11

20 28

6

29

7

25

8 9

35

16 31

56

12

1Average

Density per 10 em X 2 m X 2 m level.

The lithic SYMAP (Fig. 4-3) from the Grimes Site indicates poor surface-to-subsurface correlation. Only at the grossest level, looking at units farthest from the clusters, does there appear to be a correlation. Units 1, 2, 3, and 4 have lowest subsurface densities and lie farthest from surface clusters. The remaining units, with the exception of unit 9, are located near major clusters and have higher densities. Unit 9 has a low density and is located within a major cluster. Overall, the SYMAPs from the Grimes Site show a poor surface-to-subsurface artifact correlation. One explanation for poor correlation is recent farming practices used at the site. Prior to fieldwork, the site had been plowed for watermelon cultivation. This type of plowing is extremely destructive to archaeological deposits in that deep swales are made in the field by repeatedly plowing the same paths. Dirt from these areas is thrown onto adjacent paths, forming a small-scale ridge and swale topography. This type of plowing extends the plow zone several centimeters below

the normal levels in the swales and may have caused considerable horizontal movement of artifacts at the Grimes Site. Another, perhaps more likely, explanation is that the deep sand A horizon at the site allowed artifacts to be bioturbated further down than at other sites. All other sites had a shallower depth to a clayey B horizon which seems to have acted as a barrier to much bioturbation activity. The Osborn Site SYMAPs for ceramics and lithics are shown in Figs. 4-4 and 4-5 respectively. Table 4-2 gives the subsurface test unit densities. The ceramic map shows a good surface-to-subsurface correlation, with subsurface densities increasing progressively towards the cluster centers. Two exceptions to this are units 1 and 7, which are each located in small surface clusters but have low densities. The failure of these two units to reflect the SYMAP results may relate to the small size of the surface clusters. These small clusters are the result of a single unit with a higher number of artifacts in comparison to surrounding units. However, the addition of only a few artifacts-i.e., within the range of sampling error-created the SYMAP cluster and may not actually reflect the true surface distribution.

Table 4-2 Subsurface Ceramic and Lithic Artifact Densities from Test Units at the Osborn Site Unit 1 2

Ceramic Density 1 17

Lithic Density 109

35 35

22

3 4

8

11

5 6 7 8

14

28

1

25

1

19

76

54 37

37

Average Density per 10 em X 2m X 2m level.

Figure 4-5, the lithic SYMAP for the Osborn Site, shows a less precise, but nonetheless a general correlation. The lower density units ( 1,4, and 8) are generally located in areas of lower surface density, although unit 4 is located in a level B contour. The higher density units (2,6, and 11) are situated closest to a level C contour, the highest surface contour. The final surface-to-subsurface comparisons are from the Taddlock Site (Figs. 4-6 and 4-7). Subsurface artifact densities are shown in Table 4-3. A good correlation is seen in both the ceramic and lithic maps. The lowest density units are located either outside artifact clusters or on the margin of a level A contour, and the high density units are located within or on the margins of level C contours. The only exceptions to this are units 6 and 7 from the lithic SYMAP.

EVALUATION OF THE SYMAP ARTIFACT DISTRIBUTIONS

59

Q) "uo

0

(] 0

I

MAG NOR T H

0

10

20

...,.,.

EQC

SYMAP Contour Intervals: A 2-4 B 5- 10 c 11-23

Figure 4-2.

•

30

'

TfS T UN rTS

Comparison of Ceramic SYMAP and Subsurface Artifact Densities for Test Units at the Grimes Site.

~ Ul

I

',___f ') \

,~

'--'

I

p

MAG NORTH 10 c:-=:

0

SYMAP Contour Intervals: A 5- 11 B 12-24 c 25-53

Figure 4-3.

2:~

30 2

I'!\ I I I "

•

TCST U N ITS

Comparison of Lithic SYMAP and Subsurface Artifact Densities for Test Units at the Grimes Site.


60

0

0

• TEST UNITS

0

MAG.

o 10 l""""'!iiii

NORTH

20

SYMAP Contour Intervals: Western Clusters

30

meters

A B

0 Figure 4-4 .

c

Eastern Clusters A B

c

4-8 9-19 20-42

Comparison of Ceramic SYMAP and Subsurface Artifact Densities in Test Units at the Osborn Site.

0

•

MAG.

0

0

10

~

TEST UNITS

NORTH

20

meters

Figure 4-5 .

I 3 4-6 7 15

30 I

SYMAP Contour Intervals: Western Clusters A 6-12 B 13-23 c 29·61

Eastern Clusters A 5-10 B 11-22 c 23-48

Comparison of Lithic SYMAP and Subsurface Artifact Densities in Test Units at the Osborn Site.

EVALUATION OF THE SYMAP ARTIFACT DISTRIBUTIONS

I

IIU G

0

61

VII I

N O fl fM

..,.'

0 0 > ~

'

•

,....

T fS T UNfT$

~~\

SYMAP Contour Intervals: 5- 11 A B 12-24 c 25-53

(j)

u•o•

j

I

n~. (V~I ~V2 I .1

•

'

'0 , :\.....

n ••.(Jru~ ~ ~~ "-u . u•

I

Figure 4-6.

L

~ut

.

'>) uo

ILl-(]'•::

I'·

~

J

'- .

Comparison of Ceramic SYMAP and Subsurface Artifact Densities for Test Units at the Taddlock Site.

I

Ill>

.I

\ •"

I("

9

0

'-O•n-•

tu '

~'

""'

, .~.

'

• TIS T UNITS

~e,:

~

SYMAP Contour Intervals:

2-5 6- 11 12-25

A B

c

0

;

"-j

')

\..

_.., ·r:~~

!~ l')j.

()

I

\"...

?~

.._0

.·:re later col· lapsed into a single polished category since inter-observer bias and differenrial weatht'nng of sherds (O'Brien 1972 :4) made the subdivisiOns virtually meaningless. The suggesuon that surface preparation might be a useful temporal indi· cator came from examination of amateur collections, which showed polishing more frequent in Late Caddoan ceramic components. Another variable selected for its potential temporal char· acter was lip profile, or the cross-section of the terminal margin of the vessel orifice (Brown 1971 :20) Seven catego· ries were defmed (see Table 5-S). The major impetus for selection of this variable came from established ceramic types, which indicate that LateCaddoan types tend to have rolled lips and early types tend to have rounded or flat· tened lips (Suhm and Jelks 1962). Again, it was felt that detailed analysis of lip profiles among the AC might indi·

71

CERAMIC ANALYSIS cate finer temporal changes than presently embodied in the established types. Temper was also selected for examination with four attributes recognized : calcareous, noncalcareous (or grog), both calcareous and noncalcareous, other (including sand, lignite, and unidentifiable), and none. Because of insufficient time, visual inspection of sherd temper was the dominant method, with microscopic and chemical analysis undertaken only on selected sherd samples to determine the frequency of bone temper relative to shell temper. Therefore , the temper data must be viewed as "apparent temper" results, indicating that the categories are not unequivocal. Primary justification for selection of temper as a temporal variable carne from the analysis of ceramics from the 1976 Lake Fork field season (Bruseth, Perttula, and Skiles 1980), which suggested that calcareous tempering agents decreased in frequency through time. Although the 1976 data were few and inconclusive, they were sufficient to give impetus to more detailed examination of the 1978 ceramic data. Moreover, other researchers have noticed trends in temper of Caddoan sherds (Webb and others 1969:3-4; Skinner and others 1969:64-67; Rohrbaugh 1973:144-145). Decoration type, the final temporal variable, refers to design elements observed on the decorated sherds. The term "element" is used in this study to refer to a basic, irreducible part of a design (Shepard 1956:266). Such decoration as parallel horizontal lines or punctations within an incised triangle are examples of "elements" observed on the Lake Fork ceramics. A number of elements was recorded for the Lake Fork ceramics. Decoration type was selected as a variable for the temporal dimension because ceramic design has long been a major chronological indicator in Caddoan archaeology (e.g., Suhm, Krieger, and Jelks 1954; Suhm and Jelks 1962; Krieger 1946; Webb 1959).

Recording Procedure Recording of the attributes was carried out by visual examination of each sherd. All attributes were observed on decorated rims, while progressively fewer attributes were observed on decorated body, plain rim, and plain body sherds. Observations were recorded on coding forms for key-punching and subsequent computer analysis. Computer analysis was undertaken through the aid of various programs maintained by the Bradfield Computer Center at Southern Methodist University.

RESULTS Presentation of the ceramic analysis results is divided into two parts corresponding to the functional and temporal problem domains.

Functional Attributes The method outlined for recognizing functional similarities among the AC was the indirect examination of vessel form and size through a combination of vessel form identification and orifice diameter. The sherd location data allow a general picture of the shape and size of a vessel to determined through comparison of the frequencies of rim sherds to body sherds. Table 5-1 presents the relative frequencies of rim and body sherds from the AC. Generally , sample sizes are sufficient, with the exception of cluster WD16D, which has a sample size of four.

Table 5-l Sherd Location and Vessel Form by Artifact Cluster Artifact Cluster WD16A WD16B WD16C WD16D WD39A WD39B WD39C WD39D WD39E WD64/ 69A WD64/ 69B WD64/69C WD64/69D WD64/69E WD64/69F WD64/69G WD64/ 69H WD83A WD838 WD87A WD87B WD87C WD95A WD95B WD95C WD95D WD95E RA20 41WD109

Sherd Location (%) Rim Body 7.3 92.7 6.3 93.7 5.4 94.6 100 7.8 92.2 8.0 92.0 8.6 91.4 6.9 93.4 6.4 93.6 6.6 93.4 4.8 95.2 4.8 95.2 2.6 97.4 6.1 93.9 2.2 97.8 11.5 88.5 6.6 93.4 5.0 95.0 4.5 95.5 4.7 95.9 3.0 97.0 3.5 96.5 10.7 98.3 7.4 92.6 9.7 90.3 7.6 92.4 19.6 80.4 4.2 95.8 5.9 94.1

Totals 1,613 221 37 4 15,476 2,130 788 102 109 61 372 418 485 1,107 269 52 91 101 289 1,220 101 57 503 190 144 381 102 24 2,581


72

The following AC show low rim-to-body sherd ratios as defined by the graph: WD63/69B, WD64/69C, WD64/69D, WD64/69F, WD83A, WD83B, WD87A, WD87B, and WD87C. As outlined earlier, this low rim-to-body sherd ratio indicates a dominance of vessels with small orifices relative to capacity, or a high frequency of bottles. However, the vessel form data rule out bottles as the cause of the low ratio, since some of the abow AC have no bottles, while the highest incidences of bottles occur at AC with high rim-to-body ratios. The sherd location results indicate considerable variability among the AC. To better understand this variability, the rim sherd percentages were graphed, with the results shown in Fig. 5-l. It is necessary to graph only the rim sherd figures, since the body sherd percentages are comple· mentary. The graph is bimodal, with the first peak at four and five percent and the second peak between six and seven percent. These peaks serve to separate the AC into categories of low and high rim-to-body sherd ratios. Orifice diameter, the other functional attribute utilized to aid vessel form and size comparisons, offers little help in understanding the sherd location data. Theoretically, orifice diameters were measured to determine if a particular rimto-body sherd ratio was due to large or small orifice bowls. The results, presented in Table 5-2, show that too few rim sherds are measurable to make a contribution. Only one AC has a sample size over 17 (WD39A), and only four have sample sizes over eleven.

Table 5-2 Orifice Diameters by Artifact Clusters Artifact Cluster1 WD16A WD16B WD39A WD39B WD39C WD64/69A WD64/69B WD64/69C WD64/69D WD64/69E WD64/69F WD64/69H WD87A WD87B WD87C WD95A WD95B WD95C WD95E 41WD109

Orifice Diameter mean s.d. range 10.7 2-23 5.8 8.0 13.5 7.1 11.4 6.2 12.5 3.5 14.0 8.0 5.6 2- 13 9.6 3.2 5- 13 11.0 11.9 4.4 8.5 9.2 11.5 3.5 9- 14 12.7 13.0 2 - 50 14.0 8.0 10.8 4.4 3- 18 4.0 1.4 3-5 10.6 6.7 3- 19 9.7 5.7 5- 16 3.2 24.1 7- 10

Totals 21 1 67 8 2 1

3 3 1 17 2 2 12

1 1 17 2

5 3 13

1The following AC are omitted from the table since no measurable rim sherds were recovered: RA20, WD16C, WD16D, WD64/69G, WD83A, WD83B, WD95D, WD39D, IDd WD39E.

10

..

!

~

u

.. ~

z

8 7 6

u

J!

D

9

5

3

2 I

0~~~~~~~~~~~-r-r~~~~ 2

3

4

~

8

7

8

9

10 I I

12

13

14

~~

16

17 18

19 20

"• R1m Sherds

Figure 5-1.

Graph of Artifact Cluster Frequency by Percentage of Rim Sherds.

The cultural reason for the low rim-to-body sherd ratio is difficult to determine. Based on the presence of motifs identifiable with established ceramic types (Suhrn and Jelks 1962), five of the AC are Late Caddoan components and four are Early Caddoan components. Considering that only six LateCaddoan components occur in the total AC sample, there may be temporal significance to the low rim·to-body ratio AC. However, the four other AC areEarlyCaddo, thus suggesting that more than time is involved. Since WD87 A possessed a series of three overlapping house patterns, it is possible that these clusters represent sedentary sites. This is further supported by WD64/69B, where, although no clear house patterns were uncovered, a number of postholes was present. However, WD87C is a predominantly Archaic scatter with some ceramics present and probably does not represent a sedentary occupation. The following AC have high rim-to-body sherd ratios (frequencies of rims of 6% or greater), baaed again on the graph in Fig. 5-1: WD61A, WD61B, WD39A, WD39B,

CERAMIC ANALYSIS WD39C, WD39E, WD64/69A, WD64/69E, WD64/69G, WD64/69H, WD95, WD95C, WD95D, WD95E, and 41WD109. As with the low rim-to-body sherd AC, the presence or absence of bottles can influence the body sherd proportions. However, since the bottles are scarce among these AC, the ratios must be the result of vessels with large orifices relative to capacities. The only common denominator among the high rim-tobody sherd AC is, with the exception of WD64/69A, that they are all EarlyCaddoan. This chronological placement is based on element and motif similarities to types established by Suhm and Jelks (1962). Combined into this class of AC are clusters that must be the result of sedentary occupations (such as WD39A, WD39B, and 41 WD109), and others that are not likely to have been sedentary AC (such as WD95E). Clusters WD35A and WD39B are a well-defined midden and trash dump pair, and have a combined artifact assemblage that argues for a sedentary occupation. Cluster WD95E, on the other hand, represents a large lithic cluster with a small ceramic constituent; no midden or other indication of permanent habitation is evident at this cluster. Thus, these AC fail to show a common functional orientation. In conclusion, the functional equivalency of the AC cannot be firmly established. Several AC are similar in the ratios of rim sherds to body sherds, but the specific cause of this similarity is unknown. Temporal differences seem to correlate to a degree with the groupings of AC, and it may be that vessel form and size changes with time are the source of the rim-to-body sherd ratios (cf., Brown 1971 :25). Carinated bowls, a vessel form with a relatively low orifice-to-capacity ratio, are more common in late Caddoan sites (Krieger 1946), and this vessel type may be partially responsible for the results. This argument suggests that within the early or late Caddoan periods, vessel form and size reflected uniformity in function, and that between the early and late Caddoan periods, functional change took place.

Temporal Attributes Decoration method data, results of the first temporal variable, are shown in Table 5-3. Although sample sizes are small, temporal significance can be seen in the frequencies of various decoration methods. Late Caddoan AC, designated as such on the basis of design element similarity to established ceramic types (Suhm and Jelks 1962), tend to have higher percentages of engraving and applique, while EarlyCaddoan AC tend to have higher percentages of punctations and incisions. To examine this temporal relationship in more detail, a manual a bun dance matrix seriation (Ford 1962:42; Marquardt 1978:261) was undertaken for the decorative methods engraving, incision, and punctation. The remaining

73 categories-neck banding, applique, brushing, and otherwere eliminated, since not enough AC contained examples of these methods. The results of the seriation are shown in Fig. 5-2 (only AC with sample sizes greater than or equal to 20 were included to minimize sampling error). The basis for the seriation is the increase in engraving with the concomitant decrease in punctation over time. The seriation of early Caddoan sites correlates well with radiocarbon dates from the sites; the dates are listed alongside the seriation in Fig. 5-2. The AC with the oldest radiocarbon determination is WD16A, which dates at A.D. 760±50. Seriated next to WD16A are WD16B, RA20, WD16B, WD39C, and WD39F. These AC all have higher proportions of punctations, with engraving distinctly low in frequency. The other AC with radiocarbon determinations are WD39A, with dates of A.D. 940±60, A.D. 930±40, and A.D. 1000±50; WD87 A, with a date of A.D. 1040±50; and 41WD109, with five dates of A.D. 940±80, 1000±50, 1230±80, 1260±60, and 1260±70. The dates from WD39A and WD87 A indicate that they could be virtually contemporaneous to as much as 157 years apart based on the sigma values. The 5 dates from 41WD109 suggest two occupations at the site, one between A.D. 860 to 1050, and another between A.D. 1150 to 1330. Since the ceramics from the site could not be sorted by provenience according to these two occupations, placement of the site on the seriation represents an average of these two occupations. No independent dates are available for any of the late Caddoan AC. However, a date of A.D. 1470±80 (TX 3473) from the Steck Site, a LateCaddoan site located a few kilometers outside the Reservoir, offers a clue to the temporal position of these AC. The Steck ceramic motifs are virtually identical to the Reservoir ceramics. According to the seriation, WD64/69 is the earliest, with WD83B following, and WD64/69 the latest. Considering the relative similarity among these AC, and taking into consideration the small sample sizes used for the seriation, it is difficult to judge the merit of this ordering. The second variable that was examined from a temporal perspective was surface preparation (see Table 5-4). Again AC with sample sizes below 20 were eliminated from discussion. The results of this attribute analysis are disappointing for two reasons. First, only 10 to 20% of the sherds show evidence of surface preparation, resulting in low sample sizes. Second, differential weathering of sherd surfaces introduces bias into the frequencies of surface preparation types. This bias is best seen in the frequency of polishing, which varies from a high of 31.4% at WD39 A to nonexistent at several AC. WD87 A is a good example. Although WD87A dates close to WD39A, it contains a much smaller percentage of polished sherds and a much greater percentage of sherds with no surface treatment. The

74


Table 5-3 Decoration Method by Artifact Cluster Artifact Cluster WD16A WD16B WD16C WD39A WD39B WD39C WD39D WD39E WD64/69A WD64/69B WD64/69C WD64/69D WD64/69E WD64/69F WD64/69G WD64/69H WD83A WD83B WD87A WD87B WD87C WD95A WD95B WD95C WD95D WD95E RA20 41WD109

Engraving 8.3 15.0 100.0 18.6 17.4 3.0 40.0 66.7 42.1 35.1 53.2 53.2 19.4 28.6 55.6 47.8 53.3 26.4 38.4 33.3 11.7 50.0 30.4 42.5 23.8 43.6

Incision 46.1 45.0

Decoration Method (%) Punctation Neck Band. Applique 35.0 0.6 30.0 5.0

Brushing

Others 10.0 5.0

Totals 180 20

.01 .4

4.2 9.3 8.9

1,950 271 101 14 5 6 53 39 33 61 19 7 9 15 23 50 10 3 71 18 21 44 21 6 296

I

53.9 54.6 63.4 85.7 40.0 17.6 22.6 24.1 64.6 14.3 11.1 13.1 13.3 44.9 46.1 66.7 69.4 16.7 47.8 27.6 47.5 66.7 25.7

17.5 16.4 23.8 14.3 20.0 16.7 12.3 25.0 2.1 19.3 16.1

.1 1.0

10.5 10.0 2.1

14.3 13.1 20.0 24.1 15.4 9.4 11.1 8.7 21.3 19.0 16.7 18.6

deposit at cluster WD87 A had been extensively eroded, with no primary context materials observed when the site was examined. Cluster WD39A, in contrast, was a wellpreserved midden with good faunal preservation. It is argued that the differences between the two AC are not a reflection of time, but a reflection of differential preservation. O'Brien (1972:4) has noted the same phenomenon on ceramics from Cahokia, where sherds from village areas were more weathered than sherds from mounds. Unless compensation can be made for this bias, differential frequencies of sherd surface preparation are of little value for chronological refinement. The degree to which the slipping category is also affected by differential weathering is unknown. In any case, slipping is present in such low frequencies that time control with this variable alone is impossible. Generally, slipping

2.4 1.6

3.4

16.7 14.0 2.5 14.9 1.7

7.0 5.0 1.7

14.3 22.2 17.4 13.3 1.1

4.3

2.4 5.6 4.3 6.6 4.8 2.7

4.7 5.6 4.3

4.1

3.4

4.8 16.7 2.0

seems to occur in low frequency among AC where sample sizes are large (Table 54). The next variable examined for potential temporal significance was lip proft.le (see Table S-5). The results add little to what was previously known based on established ceramic types (Suhm and Jelks 1962), i.e., rolled lips are more common in LateCaddoan sites, and flat and round lips are more common in Early Caddoan sites. In an attempt to develop a finer control, the lip profile data were seriated by hand. The results are inconclusive; no single best seriation could be found. Instead, a number of poor fits were produced, and apparently the lip profile data do not lend themselves to temporal arrangements. The poor results are partially due to low frequencies in some attributes and the complete absence of certain profJ.le types at some AC. The last variable examined as a potential tool for devel-

CERAMIC ANALYSIS Cl 4 DAT E

75

In cisio n

-

760~50

W 039C

-

W0 9 5 A

-

W0398

-

940~60

W D39A

980~40

1000!50 WD9 5C W0 8 7A W OI09

Punctation

-

wo 168 W016A

Engrav i ng

-

-

-

-

1040!50 9 4 0t80

to 1260t70 •

W0 9 5E WD6 4/69C W0 9 50

-

W064/69E

-

W064/69B

-

W0 8 38

-

-

WD64/ 69 D

• ADDI T IONAL

Figure 5-2.

. :.

OATES:

1000t50, 12 3 0t.80, 1260t60.

Seriation of Artifact Clusters by Selected Decoration Methods.

oping finer time control from the project AC was temper. Previous ceramics analyses from the Reservoir had shown that the utilization of calcareous temper seemed to die out during the Late Caddoan (Bruseth and Huckabay 1980). Table 5-6 presents the results of temper observations on all rim and decorated sherds from the AC. As this table shows, many of the AC sample sizes are small. In an effort to increase sample sizes, the decorated and rim sherds were combined with temper results for plain sherds. Since temper on the plain sherds was recorded by fewer attributes (because of limited analysis time), the decorated and rim results had to be condensed into three attributes : calcareous, noncalcareous, and none. The combined results for all sherds are shown in Table 5-7. Sample sizes are generally larger, with only two AC (WD16D and WD39F) having Jess

than 23 sherds. Frequencies are generally high in the calcareous and noncalcareous categories, but are lower in the no temper category. Analysis of calcareous-tempered sherds by hydrochloric acid (I :3 strength) failed to indicate shell temper in samples of sherds. The samples consisted of about 10% of AC total sherds, and shell temper, if present in appreciable quantities, should have been detected. Contrary to temper findings elsewhere (e.g., Burton 1970), shell temper does not seem to have been used during Reservoir prehistory. A hand seriation was undertaken on the temper results (Fig. 5-3). The AC seriate well, largely due to the fact that only two categories are being seriated. Nevertheless, four of the six late Caddoan sites are located at the upper part of the figure and the majority of early Caddo sites below. This

76

PREHISTORIC SETTLEMENT PATTERNS AT LAKE FORK RESERVOIR Table 5-4 Surface Preparation for Rim and Decorated Sherds by Artifact Class

Artifact Cluster WD16A WD16B WD16C WD39A WD39B WD39C WD39D WD39E WD64/69A WD64/69B WD64/69C WD64/69D WD64/69E WD64/69F WD64/69G WD64/69H WD83A WD83B WD87A WD87B WD87C WD95A WD95B WD95C WD95D WD95E RA20 41WD109

Slipping 3.4 7.1

Surface Preparation (%) Slipping/ Polishing polishing 0.4 8.9 3.6 3.6

3.9 6.7 3.5

31.4 18.7 9.1

18.2 20.0 1.5

10.0 12.1

.9

1.9 3.5

17.1 4.4

9.1 10.5 3.4 2.7 7.1

5.3 12.1 .9

2.8

2.8

3.7 2.9 2.7

2.9 2.7

1.3

5.2

.9 3.7

supports the suggestion of 1976 ceramic analysis that calcareous tempering is popular during the Early Caddo and declined in usage during theLateCaddo. Closer examination of the ordering, however, shows a few problems. For example, WD87A has been dated to A.D. 1040±50 and lies three AC lower ("older") in the seriation than WD16A, which has been dated to A.D. 760±50. Other variables besides time are probably involved in the temper seriation. The fmal attribute selected for examination under the temporal problem domain was decoration type manifested by design elements on sherds. During attribute examination of the ceramics, each distinctive element was drawn and coded for computer analysis; this resulted in the definition

1.3

None

Totals

87.3 85.7 100 62.8 71.1 87.3 100 81.8 70.0 86.4 100 82.9 94.8 100 100 90.9 84.2 84.8 96.4 92.9 100 93.5 100 92.6 94.1 94.6 100 93.4

237 28 3 2,411 342 142 20 11 10 66

55 47 115 36 100 11 19 33 110 14 6 107 25 27 98 37 8 381

of 149 separate elements. The first step in analysis of these elements was to lump similar designs together. This resulted in 22 element groups each of which represents elements that vary slightly (e.g., orientation, placement, or size) but maintain essential element integrity. The next step in the analysis process was to combine element groupings at a higher level into type categories. The type categories are an effort to relate the element groups to established ceramic types as presented in Suhm and Jelks (1962). It is at this level that the ceramic analysis becomes comparable to other analyses that have used a typological approach. At the same time, the element groupings are also retained to allow the reader to decide if a particular ele-

77

CERAMIC ANALYSIS

Table 5-5 Lip Profile by Artifact Cluster Lip Profile

•

Artifact Cluster WDI6A WD16B WD16C WD39A WD39B WD39C WD39D WD39E WD64/ 69A WD64/69B WD64/69C WD64/69D WD64/ 69E WD64/ 69F WD64/69G WD64/69H WD83A WD83B WD87A WD87B WD87C WD95A WD95A WD95C WD95D WD95E RA20 41WD109

n 28.8 21.4 50.0 13.3 15.0 38.5 27.3

5.6 7.7 32.2 10.0

14.3 20.4 33.3 50.0 5.0 18.2 7.7 7.1 14.3 20.0 16.3

n 34.2 57.1 45.6 45.5 32.3 27.3 50.0 75.0 36.8 50.0 46.2 59.3 80.0 60.0 40.0 40.0 50.0 69.4 66.7 50.0 78.3 72.7 84.6 92.9 61.9 60.0 76.4

"' 8.1

3.9 3.0 1.5

(%)

(\

r1

rt

15.3 14.3 50.0 10.7 8.4 6.2 9.1

0.2

9.9

7.1 7.8 6.2

18.2 18.6 13.9 36.4 50.0 25.0 57.9 22.2 46.2 1.7

5.3 11.1 1.7

5.1

n 0.2 7.1 .005 1.8 1.5

5.6

Totals Ill 14 2 1,202 167 65 II 6 4 19 18 13

59

10.0 40.0 20.0

10 5

2.0

2.0

3.3 9.1

1.7

3.3

8.3 7.7

4.8

4.8

9.5

4.8 20.0 5.7

ment group was justifiably lumped into a type category. The reader should be cautioned at this point that the type ascriptions are based only on design elements, without regard for temper, vessel form, and other attributes commonly included in typological classification. The results of the design element analysis are presented in Table S-8. Twenty of the element groups were combined into six type categories; four groups were unique enough not to be included in the categories. The first element category is termed Canton Incised (Fig. S-4), based on the presence of elements included in the type description (Suhm and Jelks 1962: 23). Four etcmen t groups make up this category:

.8

5

40.0 60.0 35.7 6.1

1) Parallel diagonal incised a,b).

line~

1.7

5 14 49 3 4 60 II 13 28 21

5 .8

uound the rim (Fig. 5-4

2) Diagonal incised lines that also have punctations be· tween parallel incised lines or in triangles fonned by diagonal lines set at opposing angles (Fig. 5-4 c,d,c). 3) Crosshatched incised lines which only vary in spacing and angle (Fig. 5-4 f,g). 4) MisceUaneous crosshatched incised elements that differ other than in the slant or spacing of lines; elements include crosshatched designs with the addition of punctations between lines or where the cross-hatching is located aiOund only part of the rim (Fig. 5-4 h,i).

123

PREHISTORIC SE1TLEMENT PA1TERNS AT LAKE FORK RESERVOIR

78

Table 5-6 Temper for Rim and Decorated Sherds by Artifact Cluster Artifact Cluster WD16A WD16B WD16C WD39A WD39B WD39C WD39D WD39E WD64/69A WD64/69B WD64/69C WD64/69D WD64/69E WD64/69F WD64/69G WD64/69H WD83A WD83B WD87A WD87B WD87C WD95A WD95B WD95C WD95D WD95E RA20 41WD109 1

Calcareous 1

Noncalcareous

14.8 10.7

44.7 39.3 33.3 60.7 57.0 52.1 70.0 45.5 70.0 86.4 74.5 89.4 .9 2.8 10.0 81.8 89.5 78.8 75.5 80.0 66.7 83.2 64.0 77.8 83.8 48.6 25.0 69.3

10.3 11.7 11.3 5.0 9.1 10.0 3.6 8.7

9.1 6.1 .9

.9 3.7 1.5 2.7 25.0 3.2

Temper(%) Calcareous Noncalcareous 20.3 17.9 33.3 14.6 16.1 26.8 5.0 36.4 10.0 1.5 10.9 4.3 53.5 63.9 60.0

Other

None

Totals

0.1

18.9 32.1 33.3 7 .I 9.1 4.2 10.0 10.0 10.6 10.9 6.4 22.4 13.9 10.0 9.1 5.3 6.1 10.9 6.7 33.3 7.5 8.0 11.1 7.4 5.4

237 28 3 2,410 342 142 20 11 10 66 55 47 116 7 10 11 19 66 110 15 6 107 25 27 68 37

27.3

381

7.7 5.6 5.6 10.0 9.1 1.5

13.8 19.5 20.0

9.1 12.7 6.7 8.4 24.0 7.4 7.4 43.2 37.5

4.0

12.5 .3

5

See text for discussion of bone versus shell in calcareous temper.

The second element category, Davis/East Incised, implies that a precise type name could not be given. Elements of this category show similarity to both Davis Incised and East Incised; however, as the distinction between these types is based largely upon vessel form, temper, wall thickness, and appendages, no type designations have been made (Suhm and Jelks 1962). The two element groups that comprise this category are parallel-horizontal incised lines (Fig. 5-5 a,b,c, d) and parallel-horizontal incised lines with punctations (Figs. 5-5 e,f). Both groups consist of parallel-horizontal lines encircling the vessel and located on the rim. The second group differs in that it includes both blunt-end implement and fmgemail punctations located between the parallel lines.

The fmal incised element group, miscellaneous incised, has not been given a category assignment. As the name suggests, this group consists of various incised elements that do not conform to an established type. The next group, punctated, also lacks a type category designation. It consists of all punctate elements without associated incisions or appliques. The punctations show various rim arrangements, including horizontal rows, diagonal rows, and geometric patterns, as well as haphazard placements. The c-ategory Sanders Engraved (Suhm and Jelks 1962: 137) consists of seven element groups (Fig. 5-6) that are variations on the type Sanders Engraved. These groups are:

79

CERAMIC ANALYSIS

Table 5-7 Temper for All Sherds by Artifact Cluster Artifact Cluster WD16A WD16B WD16C WD16D WD39A WD39B WD39C WD39D WD39E WD64/69A WD64/ 69B WD64/ 69C WD64/69D WD64/69E WD64/69F WD64/69G WD64/ 69H WD83A WD83B WD87A WD87B WD87C WD95A WD95B WD95C WD95D WD95E RA20 41WD109

Calcareous 28.9 39.8 5.4 19.2 22.9 31.3 .6 30.6 4.9 2.4 15.0 14.0 30.6 32.8 9.6 23. 1 1.9 2.3 11.4 7.0 12.3 7.3 14.3 8.9 11.2 28.3 69.6 3.5

Temper(%) Noncalcareous 65.1 54.8 89.2 100 79.5 75 .3 67.3 49.1 67.6 93.4 95.8 83.6 85.4 67.2 65.6 88.5 75 .8 97.2 97.0 57.1 92.0 84.2 91.0 84.7 82.6 87.5 69.6 30.4 92.5

1) Crosshatched engraved Jines along the rim panel (Fig.

5-6 a,b). 2) Diagonal engraved lines along the rim panel; similar to

1) except that lines are not crosshatched (Fig. 5-6 c,d). 3) Concentric semicircle engraved Jines nested in sets.

Sets occur in various positions and combinations along rim panel (Fig. 5-6 e).

None 6.0 5.4 5.4 1.3 1.8 1.4 50.3 1.9 1.6 1.9 1.4 .6 2.2 1.7 1.9 1.1 .9 .7 31.5 1.0 3.5 1.7 1.1

8.5 1.3 2. 1 4.0

Totals 1,597 221 37 4 15,476 2,130 788 102 109 61 372 418 485 1,107 269 52 91 101 239 1,220 100 57 503 190 144 381 92 23 2,584

7) Engraved triangles at various places along rim. Some examples have triangles pendant from lip (Fig. 5-6 j). Others have triangles among diagonal engraved lines (Fig. 5-6 k) and still others have triangles pendant from horizontal engraved lines.

The next element category is Ripley Engraved (Suhm and Jelks 1962:127), consisting of three element groups (Fig. S-7).

4) Curved engraved lines diagonal to lip or rim panel;

similar to 2) except lines are curved. Often curved lines emanate from a horizontal engraved line (Fig. 5-6

f). 5) Horizontal and vertical engraved lines intersecting at

right angles; no consistent patterns observed (Fig. 5-6 g).

6) Horizontal engraved lines parallel to rim (Fig. 5-6 h,i).

1) Sun symbols, radiating "sun rays" of straight lines or triangles within a circle (Fig. 5-7 a, b). 2) Elements possessing equal arm crosses with the arms

comprised of flared ends contained within a circle (Fig. 5-7 c). 3) Longitudinal diagonal lines and excised parabolae:

catch-all for various parts of the Ripley scroll motif

es

Table 5-8

~

Design Element Frequencies by Artifact Cluster

c

~

~

Type Category1

~

Canton Incised

~

Davis/East Incised

~~

I

~

Sanders Engraved

..--'---1

Element

""""' """

-o

C ategory

Q::;

g§ ~

-.; ~ ·~ ~

-"

~~

6h =' cb ~ o:l 0. ·g ·-....... ..~ .~0 o - o ::.: u .s "' ·-

~

~

~

~ ;:;z 0

t; ~

~

~

WD16A 8 WD 168 3 WD39A 183 WD39B 24 WD39C ll WD39D WD64/69A WD64/69B WD64/69C WD64/69D WD64/69E l WD64/69F l WD83A WD83B WD87A 4 WD87B WD95A 9 WD95C 1 WD95D WD95E RA20 41 WD 109 4 1

0

00

~ ·u

_

~ ~

·u

"' g

20 1 64 22 8 l 2

17 4 388 42 10

-o

-o

oil

~

u

e.s .s ~ ·5 ~ ~ -o ~ "iico .s co -9~ o ..c:0 = o o 'tlc:: 0~ ~ "' ~ ~ c:: "'

c::-oc::g..c:-ou~

Cl..;

~

-o ~

Q .~ "' -o

~

t;:

~

r--1---, "0

~ V'l

~

McKinney La Rue/Nash Plain Neck Banded

I

I

~

~

Ripley Engraved

;,;;:: (;j

~ ..c:

1 4 1

o:l

P..

11 2 23 6 4 1

o:l ;:l

p., p.,

2

·-

;:l

~

P..

4

29 2 226 26 13 2

12 3

2 2

..!.

~.

u

~

E

~

c::

"'

c::

~w-o

~ ·~ ~ ~

o

;:J

Q

...;

.. Q

< sg~-o -~ ~

c:: w

·o

c:: 0 ;:l c: ..0 Q() 0 .. c:: u ·o u ::t w

1

5

2

79 9 l

147 4 1

..

u

Q()

o:l

7

o.,)

G,)

3

4 1

•

~

~

~ c:

c::

~ 0

sg e»

.N

o:l

1:

F-