977.801 P4952 a PHASE II EXCAVATION AND EVALUATION OF EIGHT SITES AT FORT LEONARD WOOD PULASKI Co MISSOURI by Paul P. Kreisa, Gregory R. Walz, Brian Adams, Kevin P. McGowan and Jacqueline M. McDowell with contributions by Cynthia Balek and R. Eric Hollinger Public Service Archaeology Program Research Report No. 24 ~M U. OF I. LIBRARY AT URBANA/ CHAMPAIGN BQCKSTACKS PHASE II EXCAVATION AND EVALUATION OF EIGHT SITES AT FORT LEONARD WOOD, PULASKI COUNTY, MISSOURI For Submission To: U.S. Army Construction Engineering Research Laboratory Champaign, Illinois 61826-9005 Contract No. DACA88-94-D-0008 Delivery Order No. 0006 By: Paul P. Kreisa Gregory R. Walz Brian Adams Kevin P. McGowan Jacqueline M. McDowell with contributions by: Cynthia Balek R. Eric Hollinger Dr. Thomas J. Riley, Principal Investigator Public Service Archaeology Program Department of Anthropology University of Illinois at Urbana-Champaign 109 Davenport Hall 607 South Mathews Avenue Urbana, Illinois 61801 April 1996 Digitized by the Internet Archive in 2013 http://archive.org/details/phaseiiexcavatio24krei PH°l 5" ^ ABSTRACT From August 1995 through November 1995 the Public Service Archaeology Program of the University of Illinois at Urbana-Champaign conducted Phase II evaluations at eight prehistoric sites on the Fort Leonard Wood Military Reservation, Missouri, for the United States Army Construction Engineering Research Laboratory (USACERL). The investigations were designed to provide archaeological inventory and management recommendations for sites previously recommended for Phase II National Register of Historic Places (NRHP) eligibility evaluation. Six of the investigated sites are recommended for listing in the NRHP because of the integrity of cultural deposits and their demonstrated ability to yield scientifically significant information. TABLE OF CONTENTS Chapter Page 1. INTRODUCTION 1 2. ENVIRONMENTAL SETTING 4 Geology and Physiography 4 Soils 6 Climate and Climatic Change 8 Floral and Faunal Communities 9 Cultural Study Units 10 3. CULTURAL OVERVIEW AND REVIEW OF PREVIOUS WORK 11 Cultural Overview 11 Previous Investigations 18 4. RESEARCH GOALS 23 Chronology and Culture History 23 Settlement Pattern Change 24 Prehistoric Subsistence 24 Effects of Environmental Conditions on Human Populations 25 Summary 26 5. METHODOLOGY 27 Field Methods 27 Laboratory Methods 29 Curation 32 6. RESULTS OF INVESTIGATIONS 33 23PU58 33 23PU452 53 23PU594 63 23PU421 71 23PU424 92 23PU468 106 23PU481 114 23PU485 126 7. RESEARCH RESULTS 134 Refinement of the Prehistoric Local Chronology 134 Settlement Pattern Change 139 Subsistence Patterns 144 Climatic or Environmental Change 149 Conclusions 151 TABLE OF CONTENTS CONCLUDED Chapter Page 8. SUMMARY OF EVALUATIONS AND RECOMMENDATIONS 152 Summary of Investigations 152 Summary of NRHP Recommendations 158 Summary of Research Potential of Sites 160 Summary of Recommendations for Further Work 160 REFERENCES CITED 162 APPENDIX A. LIST OF ARTIFACTS 182 APPENDIX B. Geomorphology of Five Sites at Fort Leonard Wood (Cynthia Balek) 211 Introduction 212 Geomorphic Setting 212 Soil-Geomorphic Study 216 Results of Investigations: Upland Sites 219 Results of Investigations: Terrace Site 225 Upland Soils and Depth of Artifact Recovery 230 References Cited 231 APPENDIX C. SITE LOCATION MAPS Bound Separately APPENDIX D. ARCHAEOLOGICAL SURVEY OF MISSOURI SITE FORMS Bound Separately APPENDIX E. SUPPORTING FIELD AND LABORATORY NOTES AND DOCUMENTATION Bound Separately APPENDIX F. PHOTOGRAPHIC DOCUMENTATION Bound Separately in LIST OF FIGURES Figure Page 1. Cultural Resource Zones at Fort Leonard Wood 3 2. Physiographic Provinces of Missouri and the Location of Fort Leonard Wood 5 3. Chronological Sequence of the Fort Leonard Wood Area 12 4. Site Plan and Topographic Map of 23PU58 34 5. Density Distribution of Surface Collected Materials at 23PU58 36 6. North Wall Profile from Test Unit 1, 23PU58 38 7. Wall Profiles of Test Units 2 and 3 from 23PU58 39 8. Diagnostic Projectile Points from the Surface and Test Unit 1 at 23PU58: a) Motley; b, e) Rice Side Notched; c-d) Etley or Ledbetter Cluster; f) Bottleneck Stemmed .... 45 9. Diagnostic Projectile Points from Test Unit 2 Stratum 1 at 23PU58: a) Pickwick Stemmed; b) Saratoga Expanding Stem; c) Lowe Flared Base; d) Rice Side Notched; e-k) Scallorn Cluster 46 10. Diagnostic Projectile Points from Test Unit 2 Strata 2-4 at 23PU58: a) Ledbetter Cluster; b) Rice Side Notched; c) Wadlow Cluster; d) Possible Lowe Cluster Rice Side Notched 47 11. Site Plan and Topographic Map of 23PU452 54 12. Wall Profiles of Test Unit 1, 2 and 3 from 23PU452 56 13. Frequency of Artifacts by Level Within Test Units at 23PU452 60 14. Site Plan and Topographic Map of 23PU594 64 15. Wall Profiles of Test Units 1 and 2 from 23PU594 66 16. Early Archaic Hidden Valley Projectile Point from 23PU594 71 17. Site Plan and Topographic Map of 23PU421 72 18. Density Distribution of Surface Collected Materials at 23PU421 74 19. West Wall Profile and Plan View of Test Unit 1 at 20 cm Below Surface 75 20. Wall Profiles from Test Units 3 and 4 at 23PU421 77 21. Diagnostic Projectile Points from 23PU421: a) Afton Corner Notched; b) Wadlow Cluster; c) Dickson Cluster; d) Big Creek; e) Jakie Stemmed; f) Triangular g) Karnak Stemmed; h-p) Scallorn 80 22. Selected Meramec Cordmarked rimsherds from 23PU421 84 23. Site Plan and Topographic Map of 23PU424 93 24. Wall Profiles of Test Units 1, 2 and 3 from 23PU424 95 25. Diagnostic Projectile Points from 23PU424: a) Etley Cluster; b) Unidentified 103 26. Site Plan and Topographic Map of 23PU468 107 27. Wall Profiles from 23PU468 109 28. Site Plan and Topographic Map of 23PU481 115 29. Wall Profiles from of Test Units 1, 2 and 3 from 23PU481 117 30. Wall Profiles of Test Units 4, 5 and 6 from 23PU481 119 31. Diagnostic Artifacts from 23PU481: a-b) Large Side Notched Cluster; c) Unidentified Side Notched Cluster; d) Jakie Stemmed; e) Wadlow 122 32. Site Plan and Topographic Map of 23PU485 127 33. Wall Profiles of Test Units 1, 2 and 3 from 23PU485 129 IV LIST OF FIGURES CONCLUDED Figure Page 34. Distribution of Radiocarbon Dates from Fort Leonard Wood Sites 136 35. Artifact Density at Archaic Period and Late Woodland Subperiod Sites at Fort Leonard Wood 143 36. Nutshell Density at Late Archaic and Late Woodland Subperiod Sites at Fort Leonard Wood 147 37. Faunal Assemblage Composition of Late Woodland Subperiod Sites at Fort Leonard Wood 150 38. Particle Size Distribution of Soils at 23PU481, Test Units 2 and 3 220 39. Particle Size Distribution of Soils at 23PU481, Test Units 2 and 3 222 40. Particle Size Distribution of Soils at 23PU468, Test Unit 3 224 41. Particle Size Distribution of Soils at 23PU594, Test Unit 1 226 42. Particle Size Distribution of Soils at 23PU424, Test Units 1, 2 and 3 228 43. Particle Size Distribution of Soils at 23PU424, Test Units 1, 2 and 3 229 LIST OF TABLES Table Page 1. Summary of Physical and Location Attributes of Tested Sites at Fort Leonard Wood 2 2. Distribution of Historic Artifacts from 23PU58 42 3. Distribution of Prehistoric Ceramics at 23PU58 43 4. Lithic Artifacts from 23PU58 44 5. Composition of the Faunal Assemblage from 23PU58 49 6. Faunal Remains from 23PU58 50 7. Distribution of Artifacts from 23PU452 58 8. Distribution of Artifacts at 23PU594 68 9. Distribution of Lithic Artifacts at 23PU421 79 10. Distribution of Ceramics at 23PU421 82 11. Number of Identified Specimens from 23PU421 84 12. Human Remains from 23PU421 86 13. Archaeobotanical Remains from 23PU421 87 14. Distribution of Artifacts at 23PU424 98 15. Distribution of Artifacts from Shovel Tests at 23PU424 99 16. Distribution of Historic Artifacts at 23PU424 100 17. Distribution of Prehistoric Artifacts by Test Unit at 23PU424 101 18. Distribution of Artifacts at 23PU468 Ill 19. Distribution of Artifacts at 23PU481 121 20. Distribution of Artifact Categories at 23PU485 131 21. Radiocarbon Dates from Fort Leonard Wood Sites 135 22. Selected Metric Attribute Averages of Scallorn/Sequoyah Points 138 23. Selected Data from Fort Leonard Wood Archaic Period Sites 141 24. Selected Data from Fort Leonard Wood Woodland Period Sites 142 25. Density of Nutshell at Fort Leonard Wood Sites 146 26. Composition of Late Woodland Faunal Assemblages by Percent 149 27. Summary of NRHP Evaluation Criteria and Final Evaluations 159 28. Summary of Research Issues Pertinent to NRHP-Eligible Sites 161 VI ACKNOWLEDGMENTS Projects of the scope and size of the Phase II testing of the eight sites described herein are completed only with the help of numerous individuals. The initiation of this project was aided by Dr. Richard Edging of Fort Leonard Wood and Dr. Michael Hargrave of USACERL. At the University of Illinois, Dr. Thomas Riley served as Principal Investigator for this project, and Dr. Janet Dixon Keller provided support as department head. All of these individuals provided the administrative support so necessary for conducting projects of this nature. I also would like to thank Ms. Janice Pankey of the Department of Anthropology for her help with project accounting. The completion of the fieldwork was aided by a number of individuals. Dr. Richard Edging and Mr. Curt Rankin provided site location maps and aided in transporting equipment to 23PU58. Ms. Suzanna Doggett of USACERL acted as a liason between PSAP, Fort Leonard Wood, and USACERL. The archaeological field team included Dr. Brian Adams, Dr. Cynthia Balek, Todd Brenningmeyer, Collen Clark, Pennie Copley, Stacy Craft, Alisa DeMarco, Elizabeth Garibay, David Hixson, Vincent McMillan, and Gregory Walz. Once fieldwork was completed, many individuals on the staff at PSAP conducted the laboratory tasks associated with this project. These included Dr. Brian Adams, Dr. Cynthia Balek, Todd Brenningmeyer, Pennie Copley, Stacy Craft, Alisa DeMarco, Elizabeth Garibay, Eric Hollinger, Jacqueline McDowell, Vincent McMillan, Gregory Walz, and Carl Wendt. Analytical specialists assisting this project were Dr. Brian Adams, Gregory Walz and Vincent McMillan (lithic analysts), Dr. Cynthia Balek (geomorphologist), myself (faunal analyst), and Gregory Walz and Eric Hollinger (paleobotanical analysts). Finally, report production was aided by a number of individuals: Susan Brannock-Gaul and Paula Luesse produced the computer and line drawings included in this report and Jacqueline McDowell edited the manuscript. I thank all of these individuals for their help with this project. They should be pleased that their efforts have aided in furthering the scientific, cultural and historical understanding of the study region. I hope that the content and contributions of this report have made their efforts worthwhile. P.P.K. February 1996 vn CHAPTER 1. INTRODUCTION In March 1995 the United States Army Construction Engineering Research Laboratory (USACERL) contracted with the University of Illinois at Urbana-Champaign to conduct Phase II evaluations of eight prehistoric sites at Fort Leonard Wood Military Reservation in Pulaski County, Missouri. The project was conducted by personnel from the University of Illinois Public Service Archaeology Program, with fieldwork taking place between August and November 1995. This report details the results of the investigations, evaluates each site in terms of its potential eligibility for inclusion in the National Register of Historic Places (NRHP), and, based on the data obtained from these sites, evaluates certain aspects of prehistoric settlement and subsistence in the region. As part of the effort to comply with federal regulations regarding historic preservation, specifically the National Historic Preservation Act of 1966, as amended, and Army Regulation 420-40, Phase II NRHP evaluations have been recommended for numerous sites at Fort Leonard Wood. Large-scale Phase I cultural-resource inventories have been conducted within the military reservation since 1981, resulting in identification of more than 300 historic and prehistoric sites. Additional evaluation of sites to establish their eligibility for inclusion in the NRHP (Phase II evaluation) has been recommended for over 70 sites. Until recently (e.g., Ahler, Kreisa, McDowell, and McGowan 1995; Kreisa 1995), most archaeological research has been directed toward site inventory rather than Phase II evaluation. The investigations reported here represent a continuing program to conduct Phase II evaluations on the military reservation, and constitute an important aspect of cultural resource management at Fort Leonard Wood. The sites investigated are located in two of the four cultural resource zones previously delineated at Fort Leonard Wood (Table 1 and Figure 1). Phase II evaluations targeted a variety of open-air sites and caves/rockshelters using similar investigation methods. The investigations were designed to provide archaeological inventory and management recommendations for the eight sites previously recommended for Phase II evaluation of their eligibility for nomination to the NRHP. Of the eight sites investigated, six are recommended for listing in the NRHP because of the integrity of cultural deposits and their demonstrated ability to address locally significant research issues. This project also was formulated to contribute to the fulfillment of several basic research goals identified in the Historic Preservation Plan for Fort Leonard Wood (Harland Bartholomew and Associates 1992). Of critical importance was collection of data relevant to prehistoric chronology building, subsistence and settlement patterns, and paleoenvironmental research questions. Radiocarbon and flotation samples were collected and analyzed for each of the eight sites excavated in order to address these research-oriented goals. Two of the sites investigated are caves/rockshelters because of the higher potential for preservation of botanical and faunal remains. Five of the sites are located on upland ridges, while the remaining site is situated on a floodplain terrace. One of the sites located in the uplands is adjacent to a spring and may represent specialized use of a microenvironment. The excavated sites were chosen to represent a variety of drainage, landform and site attributes. The sample also includes a broad range of site types, site sizes and ages. Table 1 summarizes the general characteristics of the eight tested sites. Three sites are located in the Big Piney resource zone, Table 1. Summary of Physical and Location Attributes of Tested Sites at Fort Leonard Wood. Site Number Site Type Resource Zone Size (m 2 ) Landscape Position 23PU58 Cave Big Piney 200 Bluff Slope 23PU452 Open Air Big Piney 12,000 Ridge Crest 23PU594 Open Air Big Piney 5,600 Ridge Crest 23PU421 Rockshelter Upper Roubidoux 200 Bluff Slope 23PU424 Open Air Upper Roubidoux 4,000 Terrace 23PU468 Open Air Upper Roubidoux 1,500 Uplands 23PU481 Open Air Upper Roubidoux 20,000 Ridge Crest 23PU485 Open Air Upper Roubidoux 375 Ridge Crest while the remaining five sites are in the Upper Roubidoux resource zone. Open-air sites are located in a variety of landscape positions, and both large and small sites are represented. The archaeological investigations used standard excavation, data recording and analysis techniques employed throughout the Midwest. The field and laboratory methods are detailed in Chapter 5, and include controlled surface collections (if surface visibility permitted); postholing at close intervals to define site stratigraphy, artifact concentrations and site limits; and hand excavation of multiple 1-x-l-m or l-x-2-m test units. All cultural material and records associated with this project have been submitted to USACERL. A listing of the cultural materials can be found in Appendix A. Figures indicating precise site locations are distributed separately because of the sensitive nature of site location data. The remainder of this report provides environmental and cultural overviews of the Fort Leonard Wood area, details the research goals and field and laboratory methods used during this project, and describes and interprets the results of the investigations at the eight sites. Chapter 2 is a brief description of the regional environmental setting. Chapter 3 provides a summary of the regional prehistoric chronology and an overview of previous work performed on and near Fort Leonard Wood. Chapter 4 provides the underlying research orientation and details specific research questions that are addressed in Chapter 7. Chapter 5 outlines the field and laboratory methods and procedures. Chapter 6 presents the results of the investigations, including site descriptions, descriptions of the investigations conducted and their results, description and analyses of the artifact assemblages, interpretation of the results, and recommendations for further work. Chapter 7 addresses the results of the project in terms of the research questions outlined in Chapter 4. Chapter 8 summarizes the results of the project and provides a summary of NRHP recommendations for each site. References Cited are followed by appendices, which include an artifact inventory (Appendix A), a discussion of the geomorphology of sites investigated (Appendix B), Archaeological Survey of Missouri (ASM) Site Forms (Appendix C), site location maps (Appendix D), copies of all field notes, maps and laboratory analyses, grouped by site (Appendix E), and photographs (Appendix F). Appendices C through F are bound separately as requested in the USACERL Scope of Work since information contained in them is considered confidential. Pulaski Co. and Ft. L Wood N 23PU58 23PU594 l-JL 23PU468 — IH 23PU481 23PU485 r KEY Cultural Resource Zones: - Lower Roubidoux - Upper Roubidoux - BigPiney V//X Cantonment Area Interior Uplands « Ft. L. Wood Reservation Boundary 1 5km Figure 1. Cultural Resource Zones at Fort Leonard Wood. CHAPTER 2. ENVIRONMENTAL SETTING Fort Leonard Wood, Pulaski County, Missouri, is located within the Salem Plateau Section of the Ozark Plateau region (Figure 2). The unique rolling landscape is characterized by upland ridge plateaus separated by deeply entrenched stream valleys. Caves, rockshelters, solution cavities, and sinkholes are numerous and add to the unique character of the region. All such features were significant factors in the prehistoric settlement of the Fort Leonard Wood area. At present, much of the area is covered by an oak-hickory forest with occasional junipers. This section presents a more detailed overview of the environment of the Fort Leonard Wood area, including geology and physiography, soils, climate, and floral and faunal communities. Geology and Physiography Fort Leonard Wood is located in the Salem Plateau section of the Ozark Plateau region (Whitefield 1989) of the Interior Highlands Physiographic Province (Madole et al. 1991). The unglaciated Ozark Plateaus overlie a broad asymmetrical cratonic dome; subaerial erosion has exposed its Precambrian core in the St. Francois Mountains (Thornbury 1965). According to Allen et al. (1975), Paleozoic tectonic features are a reflection of Precambrian tectonic elements in that deformation initially occurred during the Precambrian eon. Erosion of the uplifted Precambrian surface was followed by subsidence during the early Paleozoic era. Tectonic activity throughout the Paleozoic resulted in the formation of arches, basins and faults. Toward the end of the Paleozoic, major uplift, which may be related to the Appalachian Orogeny (Thornbury 1965), occurred along the Ozark Arch. Final uplift, followed by tectonic stability, occurred during the Mesozoic Cretaceous period for most of the Ozark region but extended into the Tertiary and Quaternary in the St. Francois Mountains. Post-Cretaceous erosion of the gently dipping Paleozoic sedimentary rocks surrounding the Ozark Arch has produced a series of escarpments. The Salem Plateau is bordered on the west and east, respectively, by the Burlington (Eureka Springs) and Crystal escarpments (Marbut 1896). The Eureka Springs Escarpment marks the boundary between younger Mississippian rocks of the Springfield Plateau to the west and the older Ordovician-aged rocks of the Salem Plateau to the east (Thornbury 1965). Three major rock formations outcrop in the Fort Leonard Wood area (Whitefield 1989). The oldest formation, early Ordovician Gasconade Dolomite, is exposed in about 12 percent of the installation (Albertson et al. 1995) and is typically a light brownish gray, cherty dolomite that is massive in its upper part and thin to medium-bedded in its lower part (Allen et al. 1975); beds of chert can be more than 1.5 m thick (Whitefield 1989). The Gasconade Dolomite is overlain by cherty dolomite and sandstone of the Roubidoux Formation. The Roubidoux Formation is exposed in about 64 percent of the base and forms the broad upland in the northern part of the base (Albertson et al. 1995). The dolomite is brown to brownish red, fine to medium crystalline rock that is locally sandy (Whitefield 1989) and can contain up to 50 percent chert (Albertson et al. 1995). The Roubidoux sandstone is brown to red in color and composed of fine to medium sand-sized quartz grains; it is locally cherty with the chert being sandy and oolitic to banded and porcelaneous (Albertson et al. 1995). Jefferson City Dolomite is the youngest Ordovician rock, and it outcrops in 17 percent of the base (Albertson et al. 1995). It is a cherty gray to brown (Whitefield 1989), medium-bedded, fine to medium crystalline i r ISON 1 J" l GRUNDY i DAVIESS N0$Jj[$vest Pfaine; Region i | ilYWSTON j L |NN J ; CALDWELL ! _T ~j_ ! ( CHARITON ; j CARROLL RAY -/ BUCHANAN J CLINTON j RROLL "*-> JMNMLPHJ RALLS .<• I MONROE : rtheastN Prairie Region —r' LAFAYETTE : JACKSON j I l BOONE LINCOLN JOHNSON pfnis /OPER f\ ) CALLAWAY >| j-^ oQ , : ,,, il1LFS /-' wA^r-i 5; '■' ur -V y /) ST LOUIS (/ sUl OSAGE 19 FRANKLIN / NHY ( I ~ _J r J" } benAn i r -M -4 / \iA A^ MILLER i J*"**"" 1 **^ \rPJl" r~1 CAMDEN , ~— , 1 .- I CRAWFORD' U^^f7i~|Oza5riigtad|Regi^ 1 J PULASKI ; JEFFEKON I -I. CEDAR BARTON | r~ ' H,NGT0N S <\enJvievc>— . J i ,„..™.^^^ perry > I DADE j DALLAS j uatDE \ FRANCOIS \ IRON ' — r— / MADISON I JASPER 1 "*V i ! IAWRENI? V i GREENE_ - J&*£!! slER ' WRtGHT ! HJEYNOLDS 1 i $ ~] CHRISTIAN ! DOUGLAS I i. I ! L L j j So^hj^t ^an^Re^i^ ( • l SCOTT VSV Southeast^ STODDARD f C?.SSBS«»< 5 OREGON j fliPxV \ / lER )RiKerineXj V / \ NEW n / 1 '~\ ) MADRID iV S j — r ^ _) j PEMISCOT^ 'Region* UNKLIN | r / DUNKLIN f Figure 2. Physiographic Provinces of Missouri and the Location of Fort Leonard Wood (Adapted from Chapman 1975). dolomite that is locally argillaceous with lenses of orthoquartzite, shale and conglomerate (Allen et al. 1975). Although it overlies a structural lowland, the Salem Plateau is a topographic upland (Marbut 1896). Local relief on the upland surface is commonly less than 30 m but can be up to 150 m along major streams (Thornbury 1965). Present-day drainage in the Ozarks probably was initiated no later than the Tertiary (Allen et al. 1975; Thornbury 1965). The Big Piney River and Roubidoux Creek, which are north-flowing tributaries of the Gasconade River, itself a north-flowing tributary of the Missouri River, display a radial drainage pattern reflective of the underlying domal structure (Marbut 1896). A Tertiary age for valley incision implies that the uplands overlooking Roubidoux Creek and Big Piney River are also pre-Pleistocene in age (Madole et al. 1991). Upland divides on the Salem Plateau are commonly wide and flat or gently undulating in areas away from major river valleys but can be almost absent near the larger streams (Marbut 1896). According to Krusekopf (1958) broad, rolling upland divides characterize surfaces that are relatively gravel-free. Although nearly all of the Ozark soils on the uplands possess some amount of chert gravels, either throughout the entire profile or in the subsoil and substrate, the distribution of surficial chert gravels is thought by Krusekopf (1942) to be a function of microclimate. South-facing slopes tend to have stony soils whereas north-facing slopes do not. Krusekopf (1942) asserts that the drier conditions associated with the south-facing slopes decrease the rate of chert decomposition (hydrolysis), resulting in an increase in the number of stones at the surface. In contrast, the wetter conditions on the north-facing slopes cause greater weathering of the chert, resulting in fewer (to no) rock fragments at the surface and the development of a silica-rich hardpan (indurated horizon) below the B horizon. Areas of karst topography occur most frequently in the Gasconade (Albertson et al. 1995) and Roubidoux (Albertson et al. 1995; Allen et al. 1975) formations. Sinkholes (Allen et al. 1975) and rock shelters (Paul Albertson, personal communication, 1996) are typical of the Roubidoux Formation whereas caves and springs are common in the Gasconade Dolomite (Albertson et al. 1995). The Roubidoux dolomites and sandstones are relatively permeable rocks, and the residuum formed in them also tends to be highly permeable. Streams in these areas tend to be diverted to subsurface conduits, resulting in low surface runoff (Williams and Maxwell 1975) and in reduced potential for erosion (Thornbury 1965). More specifically, Fort Leonard Wood occupies part of the broad dissected uplands between the Big Piney River and Roubidoux Creek (Figure 1), which form parts of the eastern and western boundaries of the installation respectively. Both streams flow northward and are tributaries of the Gasconade River, which is part of the Missouri River drainage. Most of the installation is within the upland interfluve between the Big Piney River and Roubidoux Creek. This zone is not as deeply dissected as the areas along the major watercourses and has distinctive geologic, soil and biotic characteristics. Soils Soils on the Ozark Plateau range in age from Tertiary to Holocene. Tertiary residuum, which is preserved throughout the Salem Plateau, can exceed 45 m in thickness (Albertson et al. 1995; Krusekopf 1958); it is thickest on level divides and moderate slopes and thinnest on steep slopes (Madole et al. 1991). The residuum on the Ozark Plateau is typically a reddish, stony (cherty) clay (Williams and Maxwell 1975) similar to the red clay soil, or terra rossa, that is characteristic of landscapes underlain by carbonate rocks in other parts of the world. Although terra rossa soils traditionally have been associated with mediterranean-type climates, they are not restricted to such zones since they are known to occur in the midcontinental United States (Frolking et al. 1983; Olson et al. 1980), in the Middle East (Barshad et al. 1956), in eastern Europe (Ciric and Senic 1985; Glazovskaya and Parfenova 1974) and in western Europe (Sevink and Verstraten 1979; Verstraten and Sevink 1979). Terra rossa soils are characteristically high in kaolinite clays whereas the underlying unweathered limestone or dolomite is not (Barshad et al. 1956); Ozark carbonate rocks are high in illite (Scrivner 6 1975). In spite of the mineralogical differences (which is attributed to neoformation and/or alteration of clay minerals due to weathering) (Madole et al. 1991), terra rossa soils are thought to have formed in place in the underlying carbonate rock. According to Duchaufour (1977), terra rossa soils form by decalcification of indurated carbonate rocks under alternating wet and dry seasons. During the wet season, iron in the bedrock parent material is released upon weathering (decarbonation) but precipitates out in the weathered profile (where it binds to clay-sized particles) during the dry season. Duchaufour found that residuum formed in iron-poor rocks will not rubify (redden). Rubification, according to Glazovskaya and Parfenova (1974), does not occur because of the presence of primary red-colored minerals, but instead is the result of biochemical weathering. The iron oxides and clays that make up the terra rossa are predominantly insoluble residue left after the carbonates have been removed, although some insoluble minerals may have been added by eolian deposition (Ciric and Senic 1985). The preservation of primary bedrock structures in the residuum suggests an in situ origin and volume- for-volume replacement of dissolved carbonate rock by clays (Frolking et al. 1983; Madole et al. 1991; Williams and Maxwell 1975). It is generally accepted that formation of terra rossa on limestone and dolomite is a slow process. Therefore, the soil is considered to be fairly old where thick. Ciric and Senic (1985) estimate that 1.5- 2.5 million years of weathering is required to produce a 20-cm thick terra rossa developed in a limestone with .3 percent insoluble material. Thickness of residuum in the Ozarks varies depending upon parent material properties. Generally, thicker residuum (averaging 4.5 m) is developed in the Gasconade and Roubidoux dolomites, which have a higher percentage of insolubles (in the form of quartz sand and chert) than the Jefferson City Dolomite, where the residuum averages only 2.4 m thick (Madole et al. 1991). Much thinner residuum (less than one meter thick) is developed in sandstone where the amount of carbonates is low and the amount of insolubles high (Madole et al. 1991). Residuum formed in easily weathered Jefferson City dolomite is typically a gravelly red clay with low plasticity; in thicker beds that are more resistant, the residuum is a relatively stone-free, thin plastic clay of low permeability (Williams and Maxwell 1975). The low permeability of these soils inhibits karst formation and streams tend to maintain a surface flow (Williams and Maxwell 1975). Residuum derived from the permeable Roubidoux Formation sandstones and dolomites are relatively thick and gravelly (gravel content often exceeds 50 percent) (Williams and Maxwell 1975) and highly permeable (Allen et al. 1975). The high permeability of these soils promotes development of karst topography wherein there is a loss of surface water to groundwater flow (Williams and Maxwell 1975). Residual soils formed in the Gasconade Dolomite are commonly a gravelly, yellow brown plastic clay or a gravelly, red clay with low plasticity (Allen et al. 1975). Tertiary residuum in the Ozark Plateaus is buried beneath variable thicknesses of Pleistocene loess. Loesses in Missouri were derived from the Missouri and Mississippi river valleys, and their thicknesses decrease with increased distance from the source valley. Consequently, loess ranges from about 30 m thick along the rivers to less than one meter thick in southern Missouri (Davis 1973; Madole et al. 1991). Illinoian Loveland Loess is preserved along the Missouri River (Bayne et al. 1971a), but it is rarely preserved on the uplands in south-central Missouri (Allen et al. 1975; Madole et al. 1991). In Laclede County, immediately west of Fort Leonard Wood, less than one meter of Loveland Loess is preserved in a sink, and none is preserved on the surrounding uplands, suggesting a period of widespread erosion between the Illinoian and late Wisconsinan (Allen et al. 1975). However, as Bayne et al. (1971b) point out, the inability to identify Loveland Loess on the uplands may be due to its complete pedogenic assimilation into the Tertiary-Sangamon soil and not to erosion. At least 4.9 m of middle Wisconsinan Roxana Silt may occur along the Missouri River (Bayne et al. 1971a, 1971b). Late Wisconsinan Peoria Loess is the most widespread and recognizeable loess in the area (Allen et al. 1975; Krusekopf 1958; Madole et al. 1991; Scrivner 1975). At Fort Leonard Wood, 45-75 cm of loess is found on the more stable parts of the uplands (Albertson et al. 1995). Johnson et al. (1981) found over 2 m of Peoria Loess locally preserved on the flat uplands near the Pomme de Terre river valley, northwest of Fort Leonard Wood, and Allen et al. (1975) note less than 1.2 m on uplands in Laclede County. Peoria Loess at Fort Leonard Wood is differentiated from the underlying residuum by its less weathered appearance, its high silt content (greater than 75 percent) (Balek, unpublished data) and a general lack of particles greater than 2 mm in diameter. Peoria Loess deposition occurred between 25,000 and 13,000 years ago in southwestern Illinois (McKay 1979) while in the Osage Plains physiographic section (i.e., eastern Kansas, central Oklahoma and west-central Texas) Peoria Loess deposition ceased about 10,500 to 10,000 years ago (Madole et al. 1991). A Holocene (post-9,000 years ago) loess, the Bignell Loess, occurs in parts of the Osage Plains along the Platte and Missouri Rivers (Bayne et al. 1971b), but none has been reported to occur in the Fort Leonard Wood area. Where the Tertiary residuum is completely buried by relatively thick Peoria Loess, the surface soil is developed in the loess and is essentially Holocene in age (i.e. , post- 13, 000 years). Where thin Peoria Loess has been incorporated pedogenically into the residuum such that it is not recognizable as a distinct loessial unit, the surface soil may be Tertiary through Holocene in age and, hence, can be considered a relict soil. Relict soils, as defined by Nettleton et al. (1989:59), are "ancient, pedogenic soils that have persisted on land surfaces of Pleistocene or greater age (and) may have been thickened by gradual deposition after the land surface stabilized, but never thickened by deposition so rapid or great that the soil is now a buried paleosol". Wolf (1989) identifies four soil associations within Fort Leonard Wood that also correlate closely with major physiographic zones. In the uplands between the Big Piney River and Roubidoux Creek is the Lebanon-Plato association. The Lebanon-Plato association consists of silty soils formed in loess deposits that have variable drainage characteristics. The dissected upland interfluve and the summits and shoulders of larger ridges near major streams are covered by Viraton-Clarksville-Doniphan association soils. These soils are characterized as deep, well-drained and silty to cherty. They formed in thin loess deposits overlying the cherty Jefferson City /Cotter dolomite residuum. Steep side valleys and bluffs in the major streams are covered by the Clarksville-Gepp association. These soils are thinner, well- to excessively drained and cherty to very cherty. Major stream valleys and some minor valleys are covered by Nolin-Huntington-Kickapoo association soils. These soils are deep, nearly level to gently sloping silty and loamy soils on floodplains and adjacent terraces. These soil associations are further divided into several soil series (Wolf 1989), which exhibit distinctive and consistent texture, structure and drainage characteristics. The soil series for each site is discussed in the individual site discussions. Climate and Climatic Change The climate of the Fort Leonard Wood area is midcontinental, with warm summers and cool winters (Wolf 1989). Average annual rainfall is about 100 cm, distributed relatively evenly throughout the year. The driest months tend to be during the late fall through early winter, while the wettest are during spring and early summer, although rainfall tends to be rather evenly distributed (Wolf 1989:84, Table 1). Research has indicated that this pattern has not been stable through the Holocene, with the 8 early Holocene climate being cooler and wetter than present. This climate regime then shifted during the middle Holocene to warmer and drier conditions (Bryson et al. 1968; Deevey and Flint 1957; Wendland 1978). This period, known as the Hypsithermal Interval, took place between 8,500 and 5,000 years ago. The effect of this climate change on human populations remains poorly understood and somewhat controversial at present. Minimally, by 5,000 years ago, the climate and vegetation patterns characteristic of the Fort Leonard Wood area began to stabilize in their modern form. Floral and Faunal Communities Flora Braun (1950) has classified the flora in the Fort Leonard Wood area as part of the Southern Division of the Oak Hickory Forest. This division is characterized by southern oak species with pine present locally. Within the oak-hickory forest species composition differs with slope aspect and orientation, drainage, and soil characteristics. Cedar is often present on the most xerophytic slopes. The oak-hickory forest is interrupted by occasional prairie openings. But within this classification, the Ozark Highlands also exhibit a wide diversity of plant communities corresponding with topographic, geologic and hydrologic variability within the region (Steyermark 1963). Additional distinctive plant communities are found on rolling uplands, poorly drained uplands, steep slopes, bottomland terraces, floodplains, near springs and sinkholes, and in ravine bottoms (Harland Bartholomew and Associates 1992:9). Recently, the U.S. Forest Service identified over 40 plant communities found in the adjacent Mark Twain National Forest (Miller 1981). The present vegetation differs from the prehistoric and early historic vegetation patterns. Schoolcraft (1853) traveled across the Ozarks in 1818 and noted that large tracts in the Ozarks were either unforested or had stunted tree vegetation. This vegetation pattern may have been either culturally promoted by Native Americans setting fires to improve hunting conditions (Chapman 1946) or due to the development of nearly impermeable fragipan soils that inhibit root growth on flat and poorly drained uplands (Rafferty 1980). In other areas, large tracts of pine forest covered the well-drained uplands, prairies were found on flat uplands, oak-hickory stands were present in high elevations, and cane thickets were abundant in bottomland settings (Harland Bartholomew and Associates 1992). A detailed list of species present at Fort Leonard Wood has been presented by Niquette et al. (1983). The present vegetation is dominated by oak forests on uplands and side slopes, with white, post, black, and blackjack oaks most common. The valleys support a greater variety of trees, with sycamore, ash, cottonwood, sugar maple, walnut, butternut, hackberry, red oak, willow, and pecan present. Fauna Niquette et al. (1983) provide a detailed listing of fauna present in the Fort Leonard Wood area. The faunal species in the Fort Leonard Wood area include those commonly encountered in the North American midcontinent. The most common large mammal in the region is the white-tailed deer, and a variety of medium-sized mammals (raccoon, squirrel, mink, muskrat, beaver, red fox, gray fox, skunk, opossum, cottontail rabbit, and coyote) are also present (Harland Bartholomew and Associates 1992:10). Jones and Birney (1988) identify most of the mammalian species present at Fort Leonard Wood as being associated with the Eastern and Prairie/Grassland zoogeographic units. Terrestrial bird species that were also of potential economic importance include wild turkey, quail and prairie chicken. As Fort Leonard Wood is not within a major migratory route, seasonal fluctuations of ducks, geese and swans are minimal (Bellrose 1976). Lastly, the rivers and streams of the Fort Leonard Wood area have been noted to contain numerous sunfish, catfish, sucker and gar species (Pflieger 1975). Cultural Study Units Based on the physiographic characteristics of the Fort Leonard Wood region, the base has been incorporated into a number of different cultural study unit models. The philosophy behind the creation of the cultural study unit models is that particular characteristics of the landscape, including geology, hydrology, soils, flora, fauna, and climate, affect the pattern of human occupation within a region. Chapman (1948a, 1975, 1980) used the natural divisions of Missouri (see Figure 2) to define environmentally based units with which to compare the development of cultural traditions. Chapman (1975) included the Fort Leonard Wood area in the Ozark Highland archaeological-physiographic region, with Pulaski County divided between the Lower Osage and Gasconade localities. Elaborating on the concept of combined cultural and natural areas within Missouri, Weston and Weichman (1987) employ drainages as the basis for cultural divisions. In this model, Fort Leonard Wood is in the Gasconade Study Unit. At a finer scale, Edging (1992) has defined a series of cultural resource zones for Fort Leonard Wood. This model was originally designed as a cultural resource management tool, although it is also useful in providing an initial basis of comparison regarding human use of the base. Drainage, physiographic and soil series characteristics divide Fort Leonard Wood into five broad zones. The Cantonment zone is located on the northern end of the base and consists of uplands between the Big Piney River and Roubidoux Creek. This zone includes military facilities and is viewed as having low potential for containing intact prehistoric cultural resources because of construction activities. The Interior Uplands resource zone covers the remainder of the uplands south of the cantonment. The potential for prehistoric and historic sites is generally low because of distance to permanent water sources. The Lower Roubidoux resource zone includes the Roubidoux Creek valley and adjacent terraces and uplands in the northwest corner of the installation. The Upper Roubidoux resource zone consists of an area similar to the Lower Roubidoux zone but in the southwest portion of the installation. Finally, the Big Piney resource zone includes the Big Piney River valley and adjacent terraces and bluffs in the northeast corner of the installation. Prehistoric site density is highest in these latter three zones. 10 CHAPTER 3. CULTURAL OVERVIEW AND REVIEW OF PREVIOUS WORK This section presents a general outline of prehistoric and historic cultural development in the southern Midwest area and Ozark region and is based upon earlier reports (Ahler, Kreisa, McDowell, and McGowan 1995; Ahler and McDowell 1993; Kreisa 1995) and regional archaeological syntheses, especially Chapman (1975, 1980), Douthit et al. (1979), Wright (1987), the prehistoric overview presented in Harland Bartholomew and Associates (1992), and an historical overview of Fort Leonard Wood (Smith 1993). Information specific to the Gasconade drainage has been derived from syntheses by McMillan (1965) and Reeder (1988). This overview provides an interpretive framework for evaluating archaeological resources at Fort Leonard Wood. Cultural Overview The cultural-historical sequence for Missouri is divided into six major periods (Figure 3), a number of which are subdivided into early, middle or late subperiods (Chapman 1975, 1980). Missouri prehistory appears to have experienced a number of developments similar to those identified in other areas of eastern North America, including population increase, focalization on locally abundant and seasonal foodstuffs, the eventual adoption of cultivated plant foods, and increasing social and political complexity. Expressions of these developments, though, differ across the state as it encompasses a number of different regions, including the Midwest, the Great Plains, the Southeast, and the Ozarks. Paleoindian Period (14,000 to 10,500 years ago) It was during the Paleoindian period that the earliest occupation of the New World took place, when people migrating from northern Asia crossed an exposed land mass that is now covered by the Bering Strait. This period is well-documented in portions of the United States where a series of distinct lanceolate hafted bifaces is associated with regional traditions. These lanceolate styles have long, narrow flakes removed from the base, forming a characteristic channel or flute to facilitate hafting onto bone or wood handles. Many of the more refined specimens, made from high-quality nonlocal chert, indicate a high degree of mobility and participation in exchange networks. Paleoindian groups were organized as small, highly mobile groups integrated into egalitarian bands. Settlement systems were apparently based on residential mobility. Subsistence was generalized, exploiting locally available megafauna and a variety of smaller terrestrial mammal species as well. Patterns of plant use for Paleoindian groups is poorly documented. No major Paleoindian sites are reported for the Gasconade drainage, and no sites of this time period are reported from Fort Leonard Wood or the adjacent Houston-Rolla District of the Mark Twain National Forest. A few isolated surface finds of Paleoindian points are reported in Chapman (1975) for the Gasconade drainage, though none are from Pulaski County. Dalton Period (10,500 to 9,800 years ago) The transition from late Pleistocene to Holocene environment brought about extinctions of megafauna across North America and the development of modern biotic regimes. Archaeologically, 11 YEARS B.P. PERIOD SUB PERIOD HORIZON - PHASE "■"" '■■■••-■'■•'■ ■ , . . ■ . . ■ ,-..... ■:... , ■....,: .. , ■.,.,■■... ■;-,■,.- ,-.,-:. , -..,; ,..: . .. HISTORIC 500 - MISSISSIPPIAN Late Maramec Springs Phase mnn 1UUU Late Early Maramec Springs Phase 1,600 - - WOODLAND Middle Spring Creek Complex 2,500 - 1 (~\f\T\ Early J,V\J\J Sedalia Phase - Late 5,000 - or James River Complex 8,000 - ARCHAIC Middle Early Tick Creek Complex 10,000 PALEO INDIAN Late Dalton Complex Middle T? nnn Early Figure 3. Chronological Sequence of the Fort Leonard Wood Area. 12 this transition is associated with the Dalton culture (Goodyear 1982). This manifestation was originally defined in northern Arkansas and southern Missouri (Goodyear 1974; Morse 1973; Morse and Goodyear 1973; Price and Krakker 1975) and is characterized by a chipped-stone tool assemblage that includes the distinctive lanceolate, unfluted Dalton projectile point and its variants, chipped-stone adzes and spurred end scrapers. Dalton period settlement patterns and systems have been examined in detail in the southern Ozarks and Missouri Bootheel region (Morse 1975, 1977; Schiffer 1975). Settlement includes a variety of site types, consisting of base camps occupied for long periods of time, resource extraction camps, smaller generalized residential camps, and special-purpose cemetery sites (Goodyear 1974), suggesting a logistically organized system oriented toward the exploitation of seasonally abundant aquatic resources by larger population aggregates. Several important Dalton components have been located in Missouri, including stratified deposits at Rodger's Shelter (Kay 1980), Graham Cave (Klippel 1971; Logan 1952) and Arnold-Research Cave (Shippee 1966). Two Dalton sites have been reported from Fort Leonard Wood; 23PU190 (Niquette et al. 1983) and 23PU494 (Ahler and McDowell 1993) are located in the Interior Uplands resource zone. Phase II investigations at Sadie's Cave (23PU235) yielded evidence of a Dalton component at that site as well (Ahler, Kreisa, McDowell, and McGowan 1995). Early Archaic Period (9,800 to 7,000 years ago) More common are Early Archaic period sites characterized by a variety of lanceolate (Rice Lanceolate), contracting-stemmed (Hidden Valley), straight-stemmed (Hardin), side-notched (Graham Cave), corner-notched (Thebes, St. Charles, Kirk, Jakie stemmed) and bifurcate-base (Rice Lobed and LeCroy) hafted bifaces. The Ozark highland region contains a number of sites, mostly caves/rockshelters with significant Early Archaic components. These include Jakie Shelter, the Rice site, Standlee Shelter (Table Rock Reservoir), Rodger's Shelter (Pomme de Terre drainage), and Tick Creek Cave (Gasconade drainage). The suite of hafted bifaces listed above compares favorably with many of the hafted bifaces described by Roberts (1965) and McMillan (1965) in their original description of the Early Archaic Tick Creek Complex. However, Chapman (1975) points out that Middle and Late Archaic materials represented by later point types (Stone Square Stemmed and large sided-notched points) also may have been included in the original Tick Creek Complex assemblage, and Reeder (1988: 185) suggests that the complex should be either redefined or abandoned. Based on work in other portions of the Midwest, it is likely that the temporal and functional variability represented by the Tick Creek Complex can be separated into more restricted phases and components through excavation of stratified sites. Early Archaic period sites are usually small sites with relatively few artifacts, suggesting that populations were composed of small, highly mobile residential groups organized into egalitarian bands. Local populations may have coalesced periodically into larger population aggregates to take advantage of seasonally abundant resources, but these were probably episodic events of short duration. Several sites at Fort Leonard Wood and the nearby Houston-Rolla Ranger District have yielded Early Archaic points. These include 23PU229 (an upland lithic scatter), 23PU304 (a floodplain lithic scatter), 23PH231 (a multicomponent upland lithic scatter), and 23PU210 (a disturbed stratified cave in the Upper Roubidoux resource zone). At the Kofahl Tract on the Big Piney River upstream from Fort Leonard Wood, four sites yielded Early Archaic points (Fraser et al. 1981). Seven Early Archaic sites were identified in the Big Piney and Upper Roubidoux resource zones through recent surveys by 13 the University of Illinois (Ahler and McDowell 1993; McGowan 1995). Phase II testing has recently been conducted at one of these sites, 23PU482 (Kreisa 1995). Middle Archaic Period (7,000 to 5,000 years ago) In the Midwest this cultural period is marked by a shift in settlement toward major river valley margins and increasing use of aquatic resources by larger population aggregates (Ahler 1984; Brown and Vierra 1983; Jefferies and Butler 1982; Styles 1986). Hafted bifaces characteristic of this period include large and small side-notched points (Godar, Matanzas, Raddatz, and Big Sandy), small corner- notched points (Jakie stemmed) in the early part of the period, medium to large corner- notched/expanding stem points (Big Creek, Saratoga cluster, Table Rock Stemmed) late in the period, and large straight-stem points (Stone Square Stemmed, Smith/Eva and Karnak). No specific Middle Archaic phase or assemblage complex has been defined for the Gasconade drainage, although McMillan (1965) describes several points characteristic of a Middle to Late Archaic complex. New tool types such as the fully grooved axe and ground-stone celt were added to the technological assemblage during this period. Ozark-region sites with major Middle Archaic occupations include Rodger's Shelter, Jakie Shelter, the Rice site, Standlee Shelter, and possibly Tick Creek Cave. Based on the recovery of only moderate numbers of Middle Archaic artifacts from sites in the Gasconade drainage (Tick Creek Cave, Goat Bluff Cave and Miller Cave [McMillan 1965; Roberts 1965]), Chapman (1975) proposes that the Gasconade drainage was used mainly for hunting during the Middle Archaic by populations with base camps located outside the drainage. This conclusion may be an artifact of sampling, since Middle Archaic sites apparently are fairly numerous at Fort Leonard Wood. At least 23 sites have yielded Middle Archaic materials (Ahler and McDowell 1993; Markman and Baumann 1993; Moffat et al. 1989; Niquette 1984; Niquette et al. 1983). Of these, Phase II investigations have been conducted at 23PU251, 23PU368, 23PU235, and 23PU457 (Ahler, Kreisa, McDowell, and McGowan 1995; Ahler, Kreisa, Theler et al. 1995; Kreisa 1995). Late Archaic Period (5,000 to 3,000 years ago) The Late Archaic period in the Ozarks is marked by continuation in the manufacture of many projectile point styles in use at the end of the Middle Archaic period. While large side-notched types are not part of the assemblage, the medium and large corner-notched/expanding stem points apparently continue into the first half of the Late Archaic period. New hafted biface types were introduced, including a variety of corner-notched (e.g., Afton), stemmed (Burkett, Etley) and lanceolate (Sedalia, Wadlow) types. Distinctive tools include the triangular, unifacial Clear Fork gouge, possibly used as a woodworking tool, and the rectanguloid, bifacial Sedalia Digger, possibly used for grubbing or digging. A greater variety of ground-stone tools, including 3/4-groove axes, celts, pestles, manos, bannerstones, plummets, is also present in Late Archaic assemblages. Many of these tools are associated with plant processing. Increased reliance on plants is supported by recovery of some of the earliest domesticated squash and gourd remains in the eastern United States from Late Archaic contexts at Phillips Spring (Chomko 1978; Kay et al. 1980). Chapman (1975, 1980) indicates that the entire Late Archaic adaptive strategy persisted into the Woodland period in the Ozark region. Unfortunately, many of the projectile point types listed above have been found in poorly dated or mixed deposits containing both Woodland and Late Archaic cultural 14 materials. The absence of well-dated stratigraphic contexts poses a serious drawback to basic cultural historical interpretations in the Ozarks. Two local Late Archaic cultural manifestations have been defined for areas adjacent to Fort Leonard Wood. The James River complex is based on assemblages from Table Rock Reservoir and includes Afton, Smith Basal Notched, Stone Square Stemmed and Table Rock point types (Chapman 1960). The Sedalia Complex is centered in the lower Missouri and Osage drainages and includes Etley Stemmed and Sedalia Lanceolate point types (Chapman 1975). A review of previous reports indicates that no Sedalia points, Sedalia Diggers or Etley points have been recovered from Fort Leonard Wood. The Fort Leonard Wood Late Archaic sites lack the diagnostics associated with the Sedalia complex, while the base's geographic and physiographic position make it less likely to be associated with the James River complex. Late Archaic components are common on the installation, with at least 20 sites having components assigned to this period (Ahler and McDowell 1993; Markman and Baumann 1993; Moffat et al. 1989; Niquette 1984; Niquette et al. 1983). Phase II investigations have been conducted at 23PU492, 23PU249, 23PU235, and 23PU457 (Ahler, Kreisa, McDowell, and McGowan 1995; Ahler, Kreisa, Theler et al. 1995; Kreisa 1995). Early Woodland Period (3,000 to 2,500 years ago) Traditionally, the beginning of the Woodland period is marked by the appearance of pottery in archaeological assemblages. In many areas, subsistence, settlement and social organization remain essentially unchanged from Late Archaic patterns (see Chapman 1980; Farnsworth and Emerson 1986). Distinctive projectile point styles are few in number, and a local expression of Early Woodland adaptation has not been defined. In addition, pottery types indicative of the Early Woodland period have not been identified in the region. Chapman (1980) maintains that Late Archaic adaptations, including many projectile point styles, were maintained in the Ozarks well into the Early and Middle Woodland periods. Niquette et al. (1983) and Niquette (1984) list nine sites assigned to the Early Woodland period. However, these assignments are based on recovery of Gary and Langtry points, which have a long history of manufacture that began in the terminal Archaic (3,500 years ago) and continued at least through Middle Woodland times (1,500 years ago). It may be more accurate to assign sites with these sites to a generalized Woodland temporal affiliation than to a more specific time period. Middle Woodland Period (2,500 to 1,600 years ago) As with the Early Woodland period, a local Middle Woodland manifestation has yet to be been defined for the Ozark region. If the distinctive Middle Woodland Hopewell ceramic styles are absent, local Middle Woodland expressions are difficult to identify. Other artifacts that have been used as markers of the Middle Woodland period include Snyders Corner-Notched projectile points and small lamellar blades (Montet- White 1968), neither of which is common in the Gasconade drainage. The lack of evidence for participation in the panregional Hopewell network has been interpreted by Chapman (1980) and McMillan (1965) as evidence for absence of permanent Middle Woodland settlement in the Gasconade drainage, or even abandonment of the region. Based on work at the Feeler site, Reeder (1982, 1988) defined the Middle Woodland Spring Creek complex and proposed temporal boundaries of A.D. l^KX). This alternative interpretation of Middle Woodland adaptation indicates the 15 presence of a resident population with an essentially aceramic technology identified by Kings and Snyders corner-notched projectile points (Reeder 1988). Thin, grit-tempered pottery similar to Middle Woodland utilitarian wares from other parts of the Midwest has been identified in low frequencies at a few sites in the Gasconade drainage, including at Fort Leonard Wood (23PU152 and 23PU210 [Niquette et al. 1983] and 23PU265). Other sites on base have been assigned to the Middle Woodland period based on identification of Snyders projectile points or lamellar blades. It is unclear whether other Middle Woodland cultural patterns common in the Midwest, such as intensive use of native cultigens, two-level settlement hierarchy or social differentiation in mortuary practices, are expressed in any local Middle Woodland period manifestation. Phase II investigations have been conducted at 23PU265, 23PU173, 23PU291, and 23PU235, all of which have Middle Woodland or Middle to Late Woodland components (Ahler, Kreisa, McDowell, and McGowan 1995; Ahler, Kreisa, Theler et al. 1995). Late Woodland Period (1,600 to 1,100 years ago) Late Woodland culture is well-expressed in the Gasconade drainage. Locally, Late Woodland sites are associated with the presence of cordmarked or plain limestone/dolomite-tempered ceramics. Diagnostic Late Woodland lithic artifacts include Kings Corner-Notched and Rice Side-Notched in the early part of the period and a variety of small arrow points (Crisp Ovate, Sequoyah, Scallorn, and Hayes, among others) in the later part of the period. In addition, there appears to be some degree of social differentiation in mortuary programs, with some individuals are interred rock cairns located on upland prominences and others in caves and rockshelters. The local Late Woodland manifestation is the Maramec Springs Focus (Marshall 1958, 1965), which has been subsequently divided into early and late Maramec Springs phases (Reeder 1988). Maramec Springs phase settlement patterns include extensive villages in both valley and upland ridge crest settings, smaller hamlets and extractive camps, and extensive use of caves and rockshelters, probably for specialized activities. The Late Woodland period is the best documented period at Fort Leonard Wood and in the Gasconade drainage in general. Maramec Springs occupations (either early or late) have been identified at over 35 sites on Fort Leonard Wood (Ahler and McDowell 1993; Markman and Baumann 1993; Moffat et al. 1989; Niquette 1984; Niquette et al. 1983). Phase II investigations have been conducted at 23PU426, 23PU492, 23PU172, 23PU248, 23PU249, and 23PU265 (Ahler, Kreisa, McDowell, and McGowan 1995; Ahler, Kreisa, Theler et al. 1995; Kreisa 1995). Mississippian Period (1,100 to 300 years ago) The Mississippian period generally is used to denote both the segment of time following the Late Woodland Period and a particular cultural tradition, known as Mississippian. Culturally, in the major river valleys of the southern Midwest and much of the Southeast, this period is marked by technological changes such as shell-tempered pottery and use of the bow and arrow and changes in social integration and complexity identified with hierarchical settlement systems, large town-and-mound complexes, elaborate elite burial ceremonialism, and inferred differential access to subsistence resources. The Gasconade drainage apparently does not contain a well-developed Mississippian cultural expression. Instead, the Maramec Springs Late Woodland cultural focus continues, with the addition of a few elements indicative of interaction with Mississippian peoples or peripheral participation in Mississippian 16 ceremonial/ideological/exchange systems, such as occasional shell-tempered pottery, rare incising and punctation on ceramic vessels, and occasional recovery of exotic marine shell artifacts. No Mississippian local manifestation has been clearly defined for the Fort Leonard Wood area. While Mississippian influences are apparently represented in some of the late Maramec Springs phase ceramic assemblages, the degree of local interaction and participation in the larger Mississippian cultural system has not been defined. Historic Period (300 years ago to present) Smith (1993) provides an excellent summary and developmental context for the historic period in the Ozarks and at Fort Leonard Wood in particular. The following summary is extracted from his report; additional information is found in McGrath and Ray (1987). In the eighteenth century, Pulaski County and the Fort Leonard Wood area was inhabited by the Osage Native American tribe, who used the area mainly for hunting rather than permanent habitation. In the late eighteenth and early nineteenth centuries the area also came to be populated by small numbers of Kickapoo, Delaware, Shawnee, and Cherokee, who had been driven westward by expanding Euroamerican and African- American settlement. All remnants of these tribes had been removed from Missouri by about 1830. The first non-Native American settlers in the region were French explorers and lead miners who began traveling through the Ozarks as early as 1719. French settlement in the Ozarks was sparse and its effect on the landscape and archaeological record is minimal. American settlement of the area began around 1800, with the occupation of the area by scattered pioneer hunters, subsistence farmers and lumbermen. Most settlements were located in the larger stream valleys. Gradually, more people arrived in the area, but the rugged topography kept population levels low. Most pioneers in the early nineteenth century bypassed the Ozarks in favor of the more abundant and fertile farm territories along major river valleys such as the Missouri, White and Arkansas rivers. Pulaski County was organized in 1833, and increasing numbers of permanent settlers arrived from eastern states. The settlement pattern appears to have been one of detached residences or farmsteads and hamlets scattered along rivers and larger creeks. Actual farming was done in small floodplain plots, with house locations on valley margin footslopes or side slopes. Lumbering, hunting and subsistence farming remained the major occupations. By 1860, population levels were rising, road systems had been constructed, and railroads were helping to develop the lumber industry in the area. Civil War actions in the county resulted in abandonment of many rural farmsteads and general economic decline. Repopulation of the area began again after the Civil War, but the landscape and farmsteads had been affected adversely by abandonment. Fueled by construction of a railroad through the county in 1867, population increased greatly and economic concerns began to focus on rural industries, especially lumbering. Railroads also brought material goods into the area, promoting a change from self-sufficient farming/hunting/lumbering households to more consumer-oriented households linked to, and more dependent on, outside markets and manufactories. Cross-tie production was the major industry for rural upland settlers without access to good river bottom farmland. This occupation was the main rural industry in the county from the 1870s through World War I. 17 The economic and cultural prosperity of the area reached its peak around 1910. Soon after that, overcutting of lumber resources and soil erosion brought economic decline and depopulation to southern Pulaski County. The area was also affected strongly by the Great Depression. The largely self- sufficient farmers and hunters that dominated the Ozark population of the nineteenth century had given way to larger consumer and producer communities with increasingly intimate ties to regional and national economies. Population peaked around 1910 and remained stable until about 1940. Several small crossroads communities located within the boundaries of Fort Leonard Wood had been established in the nineteenth century and continued to flourish during this period. The largest and most economically important was Bloodland; other trading communities included Big Piney, Cookville, Bailey, Moab, Wharton, Wildwood, and Tribune. The 1930s witnessed the increasing emergence of the federal government as an economic and social factor in Pulaski County. Various projects sponsored by the U.S. Department of Agriculture and the Civilian Conservation Corps (CCC) directly affected economics and population in the Fort Leonard Wood area. The CCC had established numerous camps within the Mark Twain National Forest and the present Fort Leonard Wood boundary. In 1940, the U.S. Army announced the planned purchase of 65,000 acres in southern Pulaski County. This created an immediate massive influx of laborers and support personnel into the area. As many as 30,775 workers camped within a 50-mile radius of the base (Mayes 1941, in Smith 1993). The economic and social transformation of the area has had a lasting effect on local communities. The resident populations of Bloodland and other small communities located within the installation boundaries were displaced, and the former hamlets were razed. Small communities outside the base (Waynesville, St. Robert) witnessed unprecedented economic and population growth. After World War II, population in the area and military use of the base fluctuated greatly until it was given permanent status in 1965. Previous Investigations The following summary has been drawn mainly from earlier reports (Ahler and McDowell 1993; Ahler, Kreisa, McDowell, and McGowan 1995; Baumann and Markman 1993; Kreisa 1995; McGowan 1995; Markman 1993; Markman and Baumann 1993; Moffat et al. 1989; Niquette 1984; Niquette et al. 1983) and a dissertation (Reeder 1988) of work conducted at Fort Leonard Wood and in adjacent areas. Synopses of these previous investigations are presented below. Previous Work in the Gasconade Drainage The first reference to archaeological sites in the region around Fort Leonard Wood was made early in the nineteenth century by Beck (1975[1823]), who described the remains of a prehistoric village in Pulaski County. Later, in a history of Missouri, Conant (1879) noted the presence of prehistoric remains in caves, rockshelters and mounds in the Gasconade area. This work included the first report on an excavation in this area, a cave site containing human burials. The Goodspeed Publishing Company (1974[1889]) later described mounds also located in Pulaski County. Both Beck and the Goodspeed Publishing Company made reference to stone walls, and the latter noted that some had burials beneath them. Later discussions of mounds, rock cairns, villages, campsites, and caves and rockshelters were made by Bushnell (1904) and Houck (1908). 18 The first scientific investigations in central Missouri were made by Fowke in the early twentieth century (Fowke 1922, 1928). Several of the sites he visited are located on Fort Leonard Wood; namely, Kerr Cave, the Davis caves and Miller Cave. The latter produced artifacts assignable to the Early Archaic through Late Woodland periods, and possibly into the Mississippian period. Fowke, unfortunately, was interested mainly in lithic artifacts and hoped to demonstrate a relationship between early American cultures and the European Paleolithic, ultimately ignoring most other types of artifacts. Later, Fenenga (1938) reported on the ceramics from six caves, including four in Pulaski County. He developed a ceramic classification from the excavated materials, finding two ceramic types: one grit- tempered and one shell-tempered. These were associated with the Woodland and Mississippian cultures, respectively. By the 1940s, little work had been done in the region, and Chapman (1946, 1947, 1948a, 1948b), who was developing a statewide cultural sequence, was faced with a general lack of data. There was little evidence of Paleoindian or Archaic occupation, but he was able to identify a "Late Woodland Highland Aspect" for the area. Beginning in the 1950s, some of the first investigations driven by historic preservation concerns were carried out in the region. Shippee began a survey of the proposed Richland Reservoir in 1957 (Anonymous 1957; McMillan 1965), and Marshall (1963, 1965, 1966) worked on the Meramec River and near the headwaters of the Bourbeuse and Dry Fork rivers. In 1961 McMillan (1963, 1965) conducted a large-area survey of the Gasconade River and its major tributaries. As a result, 160 sites were identified, and test excavations were made at several of them. These sites included villages or campsites, caves/rockshelters, and cairns. It was noted that villages and camps were usually located at the base of bluffs, on second terraces, or near the mouths of hollows and springs. Cairns were most often found on bluff tops over large streams, and major villages were often at the mouth of large tributaries. In addition to these surveys, investigations at individual sites also were made. Descriptions of these can be found in site reports by Delling (1962a, 1962b), McMillan (1962), Price (1964a, 1964b), and Roberts (1965). Roberts, an amateur archaeologist from Waynesville, excavated at Tick Creek Cave, and his work resulted in the first major study of faunal remains from this region (Parmalee 1965). By this time, more documentation of Archaic and Late Woodland sites had been gathered, but still little was known about Middle Woodland or Mississippian occupation of the area. Such a bias was to continue in the 1970s, when the implementation of federal legislation on historic preservation led to an increase in archaeological investigations. In 1975 investigations began along a 20-km corridor in association with a highway project on Route 63. These included the first studies of the smaller tributaries and uplands of the central Gasconade drainage (McGrath 1977). Four of the open-air sites located in this survey were tested by Southwest Missouri State University between 1975 and 1977 (Cooley and Fuller 1977). In 1978 personnel from the University of Missouri at Columbia tested an additional eight sites (Reeder and Oman 1979). Smaller surveys also were conducted, such as those by Butler (1976), Cooley and Fuller (1975, 1976), and Turner and Helm (1979). Additionally, amateurs examined various caves and rockshelters (Jolly and Roberts 1974a, 1974b; Wessel 1974) and expanded some of the Route 63 excavations (Roberts 1978). These investigations also formed the basis of Reeder's (1988) dissertation in which he argued 19 that inhabitants of the Gasconade River basin were not isolated geographically from other groups and that the culture history of the area fits the general midwestern cultural pattern. Large-scale excavations also were conducted at two Late Woodland sites: the Feeler site, 23MS12 (Reeder 1982, 1988), and the Kimberlin site, 23CR301 (Geier 1975). The former project included the first paleoethnobotanical analysis for this section of the drainage (Voigt 1982) as well as the first soil- geomorphic study (Johnson 1982). Previous Work in the Fort Leonard Wood Area In the 1980s a series of major surveys was undertaken in the Mark Twain National Forest. Surveys were conducted outside the major valleys and, for the most part, identified small lithic scatters without diagnostic artifacts (Fraser et al. 1981; Klinger and Cande 1985; Perttula et al. 1982; Purrington 1985). The American Resources Group surveyed the 243-ha Kofahl Tract in 1981 (Fraser et al. 1981). Seventeen prehistoric sites, both open-air camps and lithic scatters, were identified. The sites are located on floodplain levees, terraces, alluvial fans, terminal ridge spurs, bluff crests, upland ridge crests, and hollow bottoms. Personnel from Southwest Missouri State University (Purrington 1985) surveyed an additional 1,215 ha and identified 15 limited activity areas, 12 "field camps," 6 cave/rockshelter sites, and 1 stone cairn. Most of these sites are located near large streams, but some are also present in the uplands. Purrington (1985) suggests that prehistoric site density was highest in the Rolla-Houston area of the forest and that this density is indicative of the real site distribution pattern and not a result of sample bias. Perttula et al. (1982) also conducted investigations in a series of tracts and found two small prehistoric sites, 23PU194 and 23PU195. Purrington (1983) also located two sites, 23PU198 and 23PU199, in another survey for the National Forest. Klinger and Cande (1985) surveyed 632 ha and located 10 small lithic scatters on ridge tops. The 1970s also saw the first scientific investigations at Fort Leonard Wood. Garrison (1976) undertook a survey of a 16-ha tract for an air-to-ground weapons range, but no prehistoric sites were located. Purrington and Turner (1981) carried out a survey of a 34-ha landfill tract in an upland area. Two prehistoric lithic scatters, 23PU167 and 23PU168, were identified. Two surveys also were conducted by Major John Hargis (ca. 1980sa, ca. 1980sb) in 1980 and 1981. The first was located on Roubidoux Creek and the second in the Interior Uplands zone. He found 46 prehistoric sites, of which 13 are in the interior uplands. Due to changes in the boundaries of the military reservation, only 42 of these sites are currently located on the installation. The sites include cave/rockshelters, rock cairns, and villages or campsites. In 1981 and 1982, Environment Consultants, Inc. examined scattered tracts across the installation, totaling 2,024 ha (Niquette et al. 1983). Identified and partially evaluated were 53 prehistoric and historic sites. The prehistoric sites include 15 caves or rockshelters, 25 open sites (including two with Dalton points), and 6 rock cairns. The historic sites are three late nineteenth/early twentieth century farms, a church from the same time period, and two military facilities (a bridge and generator). Twenty-five prehistoric sites and one historic site were thought to be eligible for listing in the National Register of Historic Places (NRHP). Another 68 prehistoric sites were identified but not evaluated, and no formal report on these sites was published. Based on review of the Archaeological Survey of 20 Missouri (ASM) forms for these sites, the Late Woodland period is best represented, but Middle and Late Archaic sites are also identified. Niquette (1984) conducted a survey of an additional 1,035 ha in which 22 prehistoric and 11 historic sites were identified. Again, Late Woodland sites predominated, although Late Archaic sites were also well-represented. In the same year, Niquette (1985) also investigated 243 ha of timber-sale tracts on steeply sloped terrain. This was considered to be an area with low potential for containing archaeological sites and, not surprisingly, only three isolated finds were recovered. In 1988 the American Resources Group studied 46 separate survey areas in Pulaski County, totaling 974 ha (Moffat et al. 1989). They identified 33 archaeological sites, eight of which had been previously recorded. Twenty-seven of the sites are prehistoric, and they are mainly open camps and lithic scatters. Also recorded were five rockshelters, a rock cairn, and a cave site. The six historic sites are late nineteenth/early twentieth century residences or farmsteads. Phase II evaluation was recommended for 25 sites. Also identified were 22 isolated finds. The survey results indicate a higher site density on Fort Leonard Wood than in the Mark Twain National Forest as a whole. In 1990 and 1991, American Resources Group conducted further Phase I investigations of selected timber sales and military training areas on Fort Leonard Wood (McNerney 1992; McNerney and Neal 1992). One prehistoric and eight historic sites were documented during survey of 17 timber-sale parcels totaling 684 ha. Two military ordnance training sites and four historic farmsteads were recommended for additional work (McNerney 1992). An additional 705 ha was surveyed in two timber-sale parcels and three training areas located in the Interior Uplands and Big Piney resource zones, respectively. Eight prehistoric isolated finds, one historic isolated find, nine isolated historic features, one historic farmstead, and two prehistoric sites were located. One prehistoric site and the historic farmstead were recommended for additional Phase II investigation (McNerney and Neal 1992). In 1992, Markman and Associates conducted Phase I inventory of two large contiguous tracts of 972 ha and 1,943 ha. The larger survey documented 25 historic sites, 20 prehistoric sites and 24 prehistoric isolated finds. Phase II test excavation evaluation was recommended for four prehistoric open sites, and avoidance and preservation was recommended for two cairns and four rockshelters. Phase II evaluation was recommended for four historic sites, and one cemetery was recommended for avoidance and preservation (Markman and Baumann 1993). In the smaller survey, 12 historic farmsteads or residences were documented, and two sites were recommended for Phase II NRHP evaluation. No prehistoric sites were recorded (Baumann and Markman 1993). More recently, the University of Illinois conducted two Phase I survey projects at Fort Leonard Wood. The first project consisted of an inventory of 783 ha in scattered tracts. Forty-three new sites and 21 previously reported sites were documented (60 prehistoric, three historic, and one with both historic and prehistoric components). Phase II evaluation was recommended for 43 prehistoric sites, one historic site, and the single site with mixed components. Four cairns are recommended for listing in the NRHP, and Phase II evaluation was recommended for 20 cave/rockshelter sites and 21 open habitation sites (Ahler and McDowell 1993). The second survey of 1,613 ha was completed in 1995 (McGowan 1995). Areas surveyed include upland tracts adjacent to the Big Piney River, west of Musgrave Hollow, and encompassing Hurd Hollow. A total of 57 prehistoric and historic sites and isolates was located, of which 15 were recommended as potentially eligible for listing in the NRHP. 21 Until recently Phase II investigations at Fort Leonard Wood have been few in number. The investigations conducted by Niquette et al. (1983) included test excavations at all reported sites. These excavations were usually limited to excavation of a single test unit, cleaning looter pit profiles and documenting the exposed stratigraphy, or occasional excavation of deep backhoe trenches. These activities were considered to constitute formal Phase II evaluation to assess NRHP eligibility criteria, and NRHP nomination forms were completed for the 14 sites included in the proposed Roubidoux Creek Archaeological District. Another Phase II investigation was conducted in 1992 by Markman (1993). These activities were limited to test excavations and damage assessment of Miller Cave (23PU2). Although the cave has been looted repeatedly, the excavations demonstrated that intact Early Archaic deposits were present and that the site should be nominated to the NRHP. The Miller Cave evaluation was the first Phase II work done at Fort Leonard Wood since the 1982-1983 work conducted by Niquette et al. (1983). Phase II excavations were conducted at 15 sites during 1992 and 1993 (Ahler, Kreisa, McDowell, and McGowan 1995). The project was designed to sample different time periods, resource zones and site types. Sites investigated include both open-air and rockshelters in the Interior Uplands, Big Piney and Roubidoux resource zones. Most importantly, this project produced a large suite of radiocarbon dates and paleoethnobotanical data. Based on the excavations, 10 of the 15 sites were recommended as eligible for the National Register of Historic Places. During 1994 expanded fieldwork at 23PU235, first investigated as part of the 1992-93 Phase II testing project, was conducted as a Legacy project (Ahler, Kreisa, Theler et al. 1995). While additional work also was conducted at 23PU2 and 23PU255, most of the research effort centered on Sadie's Cave (23PU235). Work included the excavation of a number of test units, systematic coring of the cave area to identify intact deposits, and chemical and physical analyses of soil samples. The research identified intact Middle Archaic through Late Woodland period deposits. In addition, numerous charcoal samples were assayed and detailed faunal, floral and lithic analyses were conducted. Most recently, NRHP evaluations were conducted at seven sites during 1994 and 1995 (Kreisa 1995). Similar to the previous Phase II testing project, sites included a number of different time periods, resource zones and site types. Included were open-air sites and rockshelters in the Big Piney and Upper Roubidoux resource zones. Geomorphological investigations were conducted as part of this project at terrace, upland and rockshelter sites. Based on the excavations, five of seven sites were recommended as eligible for the NRHP. 22 CHAPTER 4. RESEARCH GOALS In conjunction with, and as an essential part of, determining National Register of Historic Places eligibility, this chapter defines four research goals that are designed to evaluate the scientific potential of the eight sites investigated. Under Criterion D, a crucial part of determining NRHP site eligibility is an ability to contribute to the scientific understanding of aspects of prehistory or history. Upon delineation of appropriate research goals and acquisition of data during fieldwork, evaluation can proceed based on the ability of data collected to address the defined research goals. The research goals selected for this project are modifications of the research domains discussed in the Historic Preservation Plan for Fort Leonard Wood (Harland Bartholomew and Associates 1992) and are a continuation of research efforts undertaken by previous investigators (Ahler, Kreisa, McDowell, and McGowan 1995; Kreisa 1995; Niquette 1982, 1984). The research goals discussed below are not intended as a comprehensive listing of the research potential of cultural resources at Fort Leonard Wood. It also should be noted that the research goals discussed below are broad topics that cannot be addressed through a single project. However, it was hoped that data pertinent to these research goals would be collected and could begin to contribute toward their understanding. The research goals selected for investigation consist of the refinement of local chronological and culture historical sequences, the investigation of settlement pattern change, the documentation of prehistoric subsistence strategies, and the collection of environmental data to identify potential effects of climatic change on human populations. Each of these topics is discussed below. Chronology and Culture History Without a coherent local chronology that is linked to regional cultural/historical developments, other research goals cannot be meaningfully addressed. This project collected chronological data through two methods— radiocarbon assays and typological comparisons. Both of these methods are necessary for the development and refinement of local culture/historical sequences and integration of these local sequences with regional cultural developments. General age and cultural affiliation was determined through typological comparisons of all temporally diagnostic projectile points and ceramic sherds to previously defined types within the region. The provenience and age or cultural affiliation of each diagnostic artifact was recorded. These data provide general chronological control for most sites and stratigraphic components. Radiocarbon assay of carbonized plant remains is used to provide precise age ranges for proveniences. To date, 33 radiocarbon samples have been assayed from various Fort Leonard Wood sites. Of this total, 17 are from Sadie's Cave (23PU235). There are 11 Late Woodland, four Middle Woodland, six Late Archaic, four Middle Archaic, and six Early Archaic dates. Two assays are modern. The earliest date of ca. 7520 B.C. is from Miller Cave (23PU2) while the latest date is ca. A.D. 1420 from 23PU249. Clearly, additional dates from sites would balance any biases that may be inherent in the Sadie's Cave radiocarbon series. This research goal was implemented by attempts to recover charcoal for radiocarbon assay, to identify and describe projectile point or ceramic types tied to specific date ranges, to describe local 23 material culture assemblages related to poorly known time periods that are associated with specific date ranges, and to differentiate local phases or subphases within the current local prehistoric chronology, based on ceramic typology, lithic typology and assemblage variation, and associated with specific date ranges. Settlement Pattern Change Throughout the Midwest there is evidence for a shift in settlement toward increased sedentism and a reliance on cultigens through time. Given that the Fort Leonard Wood landscape is highly variable in terms of resource distribution and abundance, it is predicted that changes in settlement patterns and site functions can be documented for specific time periods. The cultural resource zones discussed in Chapter 2 were constructed to maximize physiographic and environmental differences. Therefore, one prediction is that differences in human use of the defined cultural resource zones, potentially in terms of the intensity of site use and the types of sites present. It is hypothesized that sites in the Interior Uplands resource zone should be characterized by low intensity or duration of use, while sites in the Big Piney and Upper Roubidoux resource zones should show more intensive or longer duration of use regardless of time period. This prediction is based on greater abundance of subsistence resources in the cultural resource zones associated with large streams. An exception to this may occur during time periods characterized by high group mobility, such as the Dalton horizon and Early Archaic period. During those periods, prior to the Hypsithermal Interval, resource locations may have demonstrated less clustered patterning. Density and diversity of artifacts and subsistence remains and site size may be similar between the Interior Uplands and Big Piney and the two Roubidoux cultural resource zones during these periods, whereas, after the Middle Archaic period and the subsequent clustering of subsistence resources in and along river valleys, the use of these resource zones should increasingly differ. This trend should increase with the addition of native and tropical cultigens to the subsistence base during the Woodland subperiods. This research goal will be implemented through a comparison of artifact density and diversity and subsistence remains between sites in different resource zones, with time period held constant. Interpreted site functions also will be compared in this manner. Prehistoric Subsistence During the last three decades, archaeologists have increasingly focused on the reconstruction of prehistoric subsistence strategies. One goal of the current project was to collect subsistence remains. Until recently, little systematic investigation of subsistence has been undertaken in the Fort Leonard Wood area. Notable exceptions are the analyses of recovered bone, shell, human skeletal, and botanical remains presented in Niquette et al. (1983) and Niquette (1984) and the analysis of bone (Whitehead and Kelly 1993) and botanical remains (Wright 1993) from recent test excavations in. Miller Cave (Markman 1993). More recently, detailed analyses of botanical and faunal assemblages from various Fort Leonard Wood sites have been presented in Ahler, Kreisa, McDowell, and McGowan (1995); Ahler, Kreisa, Theler et al. (1995); and Kreisa (1995). Several general subsistence trends evident throughout the Midwest are expected to be documented at Fort Leonard Wood as well. These include the increasing use of aquatic resources through the 24 Archaic period, the introduction of squash and gourd in the Late Archaic period, the cultivation of oily- and starchy-seed cultigens during the Woodland period, and the introduction of maize in Late Woodland times and its eventual dominance after A.D. 1100. A goal of this project was to collect subsistence data that could determine whether these trends are evident at Fort Leonard Wood. Given that bone and shell preservation is good at most caves/rockshelters sites at Fort Leonard Wood, two such sites were selected for investigation. In addition, flotation samples were taken from undisturbed deposits at each of the excavated sites. It has been noted that cave/rockshelter sites produce abundant bone, shell and carbonized plant remains while this is often not true at open sites because of poor preservation conditions. Effects of Environmental Conditions on Human Populations Climate change, especially during the middle Holocene Hypsithermal Interval, has been a long-term research issue at Fort Leonard Wood (see Ahler, Kreisa, McDowell, and McGowan 1995; Ahler, Kreisa, Theler et al. 1995). This research issue centers on the effects of climate change on local vegetation, fauna and hydrology. Archaeologists assume that changes in climate produces change in the environment— the plant/animal composition, rates of sedimentation, and hydrological characteristics of streams, among others. Humans adapt to these changes, and this adaptation is mirrored in site location, site types, artifact assemblages, subsistence remains, and the geomorphology of soils at sites. This research goal is designed to contribute to basic data concerning potential environmental change and human responses to that change. For example, the possible effects of the Hypsithermal Interval should be evident in the site artifact, floral and faunal assemblages. This could include evidence from geomorphology for increased side-slope erosion and/or increased deposition rates in valley settings during the Middle Archaic period, resulting from vegetation cover deterioration on uplands and dry slopes; an increase in the abundance or ubiquity of acorn nuts in Middle Archaic botanical assemblages, resulting from the spread of oak forests and barrens; and an increase in deer and turkey remains in Middle Archaic faunal assemblages associated with the increased oak-forest cover and oak-mast production. If the climatic changes discussed above characterize the Fort Leonard Wood area, changes in local resource abundance and distribution could have affected human settlement. Earlier studies in the Truman Reservoir area (McMillan and Klippel 1981; Wood and McMillan 1976) have identified several changes in settlement systems resulting from the Hypsithermal climatic interval. This climatic change should result in several changes in prehistoric settlement. First, Middle Archaic components can be expected to be in closer proximity to major stream valleys due to the desiccation of the uplands and smaller tributary drainages. Second, upland areas away from permanent streams would be either little used during the Middle Archaic period or contain sites focused on extraction of locally abundant resources such as deer, oak mast or turkeys. Third, human population would be increasingly concentrated in and near major stream valleys. These trends are expected to produce evidence of increasing site size, artifact density and artifact diversity for Middle Archaic components and should indicate a shift in settlement systems toward more logistical organization with repetitive and structured use of the landscape and resources. To identify the effects of environmental change, geomorphological analysis of site soils will be conducted, subsistence remains analyzed for changes in taxa that are subject to narrow environmental conditions, and artifact assemblages analyzed to identify shifts in site function. 25 Summary The general research goals outlined above cover a wide range of topics and can be refined to greater levels of specificity. Linking all together are comparisons based on physiographic and temporal variability. Site selection, then, when incorporating both physiographic and temporal variability ■, can be used to construct robust bases of data with which to address these research goals. However, it is expected that these research goals cannot be addressed fully through the investigation of this single suite of sites at one locality, but can continue to be addressed in future excavations while adding Fort Leonard Wood data to a growing regional body of information. Successive projects can aid in the refinement of settlement and subsistence models. 26 CHAPTER 5. METHODOLOGY The Phase II archaeological testing project undertaken by the Public Service Archaeology Program of the University of Illinois at Urbana-Champaign at Fort Leonard Wood consisted of excavation and analysis of eight prehistoric archaeological sites to determine their eligibility for inclusion in the NRHP and to recover material pertinent to other areas of significant regional research issues. This section describes the excavation, laboratory processing and analysis methods used in this project. Sites were selected for investigation based on their indeterminate status for NRHP eligibility and hence their potential to yield data on prehistoric cultures in the Fort Leonard Wood area. A series of Phase I cultural resource investigations performed on the installation prior to this work (see Chapter 3) had documented over 400 sites. Of this total, 123 sites have been recommended for Phase II NRHP evaluation. Only sites with prehistoric components were considered here, further reducing the number considered for investigation to 96. In selecting sites for NRHP evaluation, an effort was made to select sites from different cultural resource zones. Variety was considered to be essential, and sites were selected from two of the five cultural resource zones (Big Piney and Upper Roubidoux zones). Other factors that were considered included representation of sites in a variety of physiographic/topographic settings— bluff crest, bluff slope and stream terrace. Effort was made to select both large and small sites, open-air sites and rockshelter sites, and to have representation of several time periods. Finally, logistical considerations such as military restrictions were considered. Using these criteria, the eight sites listed in Table 1 were selected for investigation. Field Methods The field methods used on both open habitation sites and rockshelter sites are similar. However, because of the differences in site size, condition and potential for recovery of organic remains from intact, stratified deposits, field methods for these two types of sites differed in some respects. The general field methods employed are discussed below, with methods differentiated by site type when appropriate. In general, three tasks were undertaken at each site. First, the site was relocated and mapped. Second, a posthole grid was excavated or a controlled surface collection was conducted. Third, several test units were excavated. A detailed discussion of each of the three tasks is presented below. The initial task was to relocate and record a preliminary description of the site. Relocation of the site was based on previous documentary evidence including maps, photographs or written descriptions in site reports or Archaeological Survey of Missouri (ASM) site forms. Once relocated, photographic documentation of the site along with a description of the site prior to excavation was made. Information recorded included a general site description, its condition, and its surrounding environment, including the present vegetation, topographic position, elevation, and evidence of disturbances. Upon completion of these preliminary tasks, a detailed map of the site was begun, using either a total station or a transit. Each site map details the surrounding topography, surface features, site boundaries, grid orientation, and locations where posthole tests and test units were excavated. 27 Second, either a controlled surface collection was made or a grid of postholes was excavated. These tasks provide information on artifact distributions, artifact density, the cultural and natural stratigraphy of the site area, and are important for determining accurate site boundaries. Posthole tests were excavated at open-air sites, since all had less than 30 percent surface visibility. At the two cave/rockshelter sites, leaf litter and recent debris were cleared from the floor, and a controlled surface collection was conducted. Third, test units were hand-excavated at each site. Test units were placed within artifact concentrations or were distributed across the area to examine representative sections of the site. The number and size of the test units varied, depending on site size and goals of the excavations. All sediment was screened through 1/4-inch (6.4-mm) mesh hardware cloth. All excavations were continued to culturally sterile sediments. Test unit documentation included unit summary and level forms, with plan maps drawn to 1:10 or 1:20 scale prepared for each level and scale drawings of two vertical wall profiles recorded for each unit. Black-and-white photographs and color slides document each wall profile. Photographs were taken to document individual excavation levels. Cultural and natural strata were identified, drawn and described. Colors were described using the Munsell soil color chart (Munsell 1975). Features were encountered at one of the sites investigated. Initially, plan maps of the features were drawn. Features were then cross-sectioned and a vertical profile drawn. Flotation samples were collected from at least one-half of each feature. Pollen, soil/sediment and radiocarbon samples were collected when appropriate. Black-and-white photographs and color slides document each feature in plan and the cross-section profile. Soils were described based on USD A standards and the local soil survey (Wolf 1989). Upland soils at Fort Leonard Wood are predominantly alfisols and ultisols formed under natural forest vegetation (Wolf 1989). Alfisols are less weathered than ultisols, although both soil profiles are morphologically characterized by A-E-Bt horizon sequences. The A horizon is a mineral horizon formed at the surface or below an O (organic) horizon and is characterized by an accumulation of humus (decomposed organic matter). Organic matter derived from leaf litter is mixed into the upper 10 cm or so of mineral material by earthworms and other fauna, producing a dark-colored (gray to black depending upon the amount of organic carbon present) A horizon. Ingestion and excretion of the humus-rich mineral material by earthworms and other soil organisms gives rise to a granular or crumb structure. Where anthropically disturbed, the surface horizon (Ap horizon) may become mixed with the underlying horizons such that an Ap-Bt horizon sequence is produced; Ap horizons commonly have abrupt lower boundaries. E horizons are mineral horizons characterized by minimal organic matter content and by loss of silicate clays, iron and aluminum oxides resulting in a concentration of resistant silt and sand-sized (quartz) minerals. Consequently, E horizons are lighter-colored and coarser-textured than A and B horizons. E horizons commonly exhibit platy structure but can possess granular structure if biologic activity is pronounced. Platy structure is thought to result from a combination of processes, namely expansion and contraction during freeze-thaw and wetting-drying cycles and from the parallel orientation of planar silt-sized particles. 28 Bt horizons are zones of silicate clay accumulation; they are more dense and have a higher bulk density than A and E horizons. Desiccation of Bt horizons produces prismatic or blocky structure. The surfaces of the blocky peds are often coated with illuvial clays, organic matter and/or silt-sized quartz particles. Where well-drained, the Bt horizon is reddish brown or yellow brown in color; in more poorly drained environments it is blue, green or gray, and a highly mottled Bt horizon denotes a fluctuating water table. Whereas most soil horizons form in situ due to vertical biomechanical and biochemical processes sediments in caves and rock shelters derive from erosion of upland soil material. Red cave clays and overlying gray brown silt loam material are mineralogically similar to upland Tertiary-Pleistocene residuum and late Wisconsinan Peoria Loess respectively (Madole et al. 1991). Vertical variations in cave and rock shelter sediments are geogenic (accretionary) not pedogenic in origin and reflect changes in source material properties and/or in erosion-deposition regimes; they are not the result of pedogenic leaching and eluviation-illuviation processes that produce horizons in the upland soils. Laboratory Methods All recovered materials were transported to the laboratory facilities at the University of Illinois at Urbana-Champaign where they were washed, labeled, inventoried, analyzed, and packaged for curation. Inventory forms document artifact types, counts and weights for each provenience. All of the sites investigated here were selected because of the potential of their prehistoric components. Historic artifacts were recovered in small numbers from only a few of these sites and are described individually. All prehistoric material was counted and weighed, with the data entered on a preliminary inventory form (Appendix E), where artifacts are divided into major material classes (lithic, ceramic, bone, plant, etc.). Fire-cracked rock (Taggert 1981; Zurel 1979, 1982) also was segregated as a major material class. The most common material class of prehistoric artifacts is lithic remains. Ceramic, faunal (bone and shell) and botanical remains also were recovered, though in fewer numbers than the lithic artifacts. Lithic, ceramic, bone, shell, and botanical remains were subjected to detailed analyses. Lithic artifacts were divided into three broad categories: tools, manufacturing debris and miscellaneous lithic material. Tools include those made from both chipping techniques (e.g. , projectile points and scrapers) and grinding and pecking techniques (e.g. , celts and axes). Tools can be of formal manufacture, such as the examples listed above, or of incidental manufacture (e.g., hammerstones and pitted stones). Manufacturing debris usually is composed of the waste materials (e.g. , cores, flakes and block shatter) generated from the production of chipped-stone tools. Miscellaneous lithic material includes fire-cracked rock, limestone, unmodified and tested chert chunks and nodules, and ocher. Fire-cracked rock is unintentionally produced debris that results from alternating processes of heating and cooling (Taggert 1981; Zurel 1981). Unmodified or minimally flaked chunks and nodules of chert represent raw material collected and reserved for tool production. Artifacts from each of these categories were recovered. Definitions of the tool and debris categories represented in the artifact assemblage are provided below. Debitage (flaking debris) categories comprise a majority of all the chipped-stone remains. These categories include block shatter, broken flakes and whole flakes. The whole flake classification was used for items characterized by the presence of a bulb of percussion on the ventral surface and a 29 striking platform. The whole flakes were further divided into primary, secondary and tertiary flake types based on the amount of visible cortex present: > 50 percent, < 50 percent and > percent, and percent respectively. Secondary characteristics also were assessed. Primary flakes tend to have a pronounced bulb of percussion, secondary flakes have a less pronounced bulb, and tertiary flakes are generally smaller than the other two flake types and often have a reduced or no bulb of percussion. Broken flakes are debris items that lack a platform or bulb of percussion, or are too small to place accurately within the whole flake category. Block shatter has irregular shapes that lack flake and core characteristics. Bifacial thinning flakes have a distinct lip on their bulb of percussion, have an angled striking platform and have distinctive negative flake scars on their dorsal surface. Related to debitage are cores, the parent stones from which flakes are removed. Formally flaked stone tools initially were divided into unifacial and bifacial categories. Unifaces show evidence for retouch only on one surface. Bifaces demonstrate retouch on both their dorsal and ventral surfaces. When possible, each tool is assigned to a more detailed morphological-functional use category. Unifaces are most commonly classified as scrapers, with the particular type determined by the placement of the edge modification. Bifaces can be placed into a number of distinct categories. Among these are such items as projectile points, drills, knives, scrapers, and thick and thin bifaces. The most recognizable of the chipped-stone tools are projectile points. Projectile points are symmetrically thinned bifaces that show evidence of hafting. These items have been examined in detail for comparison with projectile point types known from the Midwest and are particularly important for the placement of sites within a cultural and temporal context (see Bell 1958, 1960; Chapman 1975, 1980; Justice 1987; Kay 1980; McMillan 1965; May 1982; Morrow 1984; Perino 1968, 1971). The other tool types are largely descriptive in nature. Perforators are typically small, narrow, often bifacial tools. Knives are larger, thin bifaces with a low edge angle to facilitate cutting while scrapers have a higher edge angle to facilitate scraping. Thick and thin bifaces are not finished tools but stages in tool manufacture. A thick biface is one that has been modified, is not a finished implement, and is in need of further modification. Typically, the thick biface can be modified into a number of different tool types (Bradley 1975). Thin bifaces are the result of further modification of thick bifaces. They also are not finished implements, but their morphology indicates that they can be further modified into only a single tool category (Bradley 1975). Thin and thick bifaces were differentiated based on flake morphology. In analyzing the chipped-stone tools and lithic debris, both core-reduction and bipolar models were followed (Collins 1975; see also Bradley 1975; Hayden 1980). Collins (1975) defines five stages of chipped-stone manufacture and use for the core-reduction model. These stages consist of acquisition of raw materials, core preparation-initial reduction, primary trimming, secondary trimming, and use- maintenance-modification. Each of these categories, called activity sets (except for raw material acquisition), is associated with waste by-products and objects that are further used or modified. Core preparation-initial reduction is a stage in which the core is shaped and flakes are detached. Suitable flakes may be retained and further used with the core being discarded, or both can be. retained for additional modification. End products of this stage are primary flakes, block shatter, discarded cores, and thick bifaces. The next stage, primary trimming, is used to shape the object. Flakes can be retouched into usable tools, or thick bifaces can be flaked into a thin biface. These activities result in the production of secondary flakes, retouched flakes, thin bifaces, and items broken during manufacture. Following primary trimming is the secondary trimming of thin bifaces. This stage produces tertiary flakes, finished tools and items broken during processing. Finally, the tools are used, maintained and 30 perhaps modified. Bifacial thinning flakes are the most important waste by-product of tool maintenance activities, although they also could be produced while thinning thick bifaces. Following this model, the following considerations have been made in the analysis of lithics. Cores, primary flakes and block shatter are classified as evidence of initial-stage reduction activities. Secondary flakes, tertiary flakes, and thick and thin bifaces evidence later-stage reduction activities. Bifacial thinning flakes are indicative of tool-maintenance activities. Since broken flakes can be produced by a number of prehistoric and modern processes, they were not considered when characterizing the lithic tool production activities at the site. Less common at Fort Leonard Wood is the use of a bipolar technique. In this technique, small cobbles are generally not well-suited for use in the direct hammer or core reduction technique described above, although a bipolar technique can be used to manipulate these items. When using a bipolar technique, the cobble is placed on an anvil and struck. This action yields bipolar debris and, eventually, a spent core. The flakes can either be discarded, used as-is, or further modified into tools. The bipolar technique also produces pitting in anvil stones due to the striking force used. The other class of lithic artifacts, ground-stone tools, consists of pecked and ground items generally made from metamorphic or igneous rock. Included in this category are items that are intentionally formed, such as celts and axes, and unintentionally formed, such as hammerstones, grinding stones and pitted stones. Intentionally formed artifacts consist of items that were modified for a specific use. Unintentionally formed items have areas of pitting, battering, or smoothing that was caused through use. Definitions of the individual artifact categories are based on those used by other researchers in the Midwest (e.g., Brose 1970; McElrath 1986; McGimsey and Conner 1985, among others). Additional data also was gathered from prehistoric ceramics. Data collected for each rim or body sherd includes temper type, temper density and size, surface treatment, cord twist, and decoration, where appropriate, thickness, and vessel portion. For rim sherds, additional collected data includes vessel form, rim shape, lip shape and treatment, following standard definitions for these attributes (Rice 1987; Shepard 1965). Vessel profile definitions and terms of Ozuk (1990) were used in an attempt to standardize terminology. When possible, each sherd is assigned to a ceramic type for chronological purposes that follow previously described Maramec types defined in central Missouri (e.g., Marshall 1958; McMillan 1965). Species and elements represented in the faunal assemblage were identified using the comparative collections of the Zooarchaeology Laboratory at the University of Illinois at Urbana-Champaign and the Illinois State Museum. All material was identified to the most exclusive taxonomic level. Number of Identified Specimens (NISP) and Minimum Number of Individuals (MNI) are presented for each taxon below the level of family. The derivation of MNI is based on the single most frequently paired element occurring in the sample (Grayson 1984). MNI for each taxon was calculated separately for each unit of analysis. Except for 23PU58, the assemblages include material from the 2-mm flotation fraction in addition to the general excavation level material. Flotation samples were processed in the laboratory using a Flote-Tech II flotation machine. Light and heavy fractions were separated for analyses. Lithics and ceramic artifacts from the 2-mm size grade were analyzed along with the screened samples from excavation units and features. The primary focus of the flotation fraction analyses was recovery and identification of botanical remains. 31 Both heavy and light fractions yielded botanical remains, and these were supplemented by samples of hand-picked charcoal collected during excavation. Carbonized plant remains were extracted from both flotation fractions, which were combined and sieved through a 2-mm geologic screen. All fragments >2.0 mm were segregated into wood charcoal, nutshell and seeds. All nutshell and seeds and a sample of 20 pieces of wood charcoal from each sample were identified to the most exclusive possible taxon. The smaller fraction (<2.0 mm) was scanned for identifiable seeds and nutshell fragments under lOx to 30x magnification. Taxonomic identification was effected through comparison to specimens illustrated in standard botanical identification texts (Core et al. 1979; Martin and Barkley 1961; Montgomery 1977) and to the comparative collections at the University of Illinois. After analysis of all materials was completed, updated ASM site forms were compiled for all sites. The completed ASM forms are included in Appendix C. Because site location data are confidential, this appendix is bound and distributed separately. Curation All cultural material recovered during this project and all documents relating to the fieldwork and laboratory analysis of these materials are the property of the federal government. University of Illinois personnel have compiled lists of the artifacts recovered from each site (Appendix A). In addition, copies of all photographs, analysis forms and field forms pertaining to these sites are included in Appendices E and F. Qualified researchers interested in access to these collections should contact USACERL. 32 CHAPTER 6. RESULTS OF INVESTIGATIONS This chapter presents the results of the Phase II test excavations conducted at eight sites in Fort Leonard Wood. The sites are grouped by cultural resource zone— Big Piney River sites are presented first, followed by those located within the Upper Roubidoux Creek zone. For each site a short description, including physiographic location and current condition, is presented. Following this are sections on previous research, field investigation results, the analysis of the artifact assemblage, and an interpretation of these data. Finally, based on the field and laboratory results, an interpretation of the site's research potential and an NRHP recommendation is presented. Included in this last section are potential avenues of future research that could be undertaken at those sites recommended as eligible for listing in the NRHP. Material from specific provenience units is detailed in the material inventory in Appendix A. 23PU58 (Freeman Cave) Site 23PU58 is located within the Big Piney cultural resource zone, on a steep bluff face approximately 50 m above the north bank of the Big Piney River. The site lies at approximately 304 m asl on the Ramsey Peninsula, a large upland ridge formed by a bend in the Big Piney River. Archaeological deposits cover an area of approximately 240 m 2 . At the time of the investigations, two atmospheric monitoring stations had been set up in the cave, one at the mouth, the other in the small back (north) chamber. The data from these stations will be used in a study of bat behavior. Looting, military training activities and previous archaeological investigations have impacted up to 50 percent of the site area, although the excavations identified intact deposits below many of the disturbed areas. Site Description 23PU58 consists of a large main chamber that opens to the south, with maximum dimensions of approximately 22 m east- west and 18 m north-south, and a small chamber at the back of the cave (north), measuring approximately 4-x-4 m (Figure 4). The main chamber, which narrows from the mouth (south) to the rear (north), is separated from the small chamber by a "step" or ledge of exposed bedrock. The cave floor slopes upward from south to north, or from the mouth of the cave to the rear chamber. The cave ceiling is approximately 6 m high at the entrance and 2 m high in the rear. An active drainage is present in the back of the cave along the east wall, although very little water was observed during the Phase II investigations. Water was not flowing but dripped down the cave floor, forming a series of small connected puddles. This drainage originates in the rear chamber and follows the east wall, where it disappears approximately 10 m from the cave entrance. Several large blocks of roof fall are located in the west half of the main chamber. Outside the cave drip line is a very narrow (2-3 m wide) terrace. South of the terrace is a steep slope that descends to the Big Piney River. Initial inspection of the cave indicated that there has been much disturbance in the front of the main chamber, extending from the mouth to about 7 m north, which covers an area of approximately 80 m 2 , or about 30 percent of the interior surface area. Here there is a large crater-like depression containing recent refuse. The slope in front of the cave is littered with prehistoric lithics, shell and ceramics, most likely looter's spoil. Prehistoric material was observed throughout the cave floor. As discussed below, 33 Figure 4. Site Plan and Topographic Map of 23PU58. 34 limited scientific excavations were conducted in the front of the cave by Fowke (1922), who also reports that pot hunters had been active in the cave during the early twentieth century. Although the site is not easily accessible, the modern refuse on the cave floor indicates that it is frequented by the general public. It also appears that in the past, perhaps during World War II or the Korean conflict, military training activities took place in the cave. Previous Investigations Freeman Cave was first described by Fowke (1922), who states that the site also was known as Ramsey Cave. Fowke conducted limited archeological excavations and indicated that the deposits were generally thin and little archeological material was present. While the precise location of Fowke's excavations was not recorded, he indicated that ".. .several holes were dug at intervals between the front and the rear... very little was found in them" (Fowke 1922:82). Artifacts recovered consisted of two bone perforators, a worked mussel shell fragment, nine "rough" chert artifacts, and a "few" pot sherds. Fowke also reports that prior to his work, skeletal remains of two individuals had been discovered in the east side of the cave and partially removed (but later replaced) by a local looter. The data from Test Unit 1, to be discussed below, indicate that looting continues to the present time. Results of Investigations Investigations at 23PU58 were conducted during November 1995 after a nesting colony of endangered migratory gray bats had vacated the cave. Included in the activities were a controlled surface collection, the excavation of three test units, systematic coring of the cave to identify areas of intact deposits, and the compilation of a topographic map and site plan. Surface Reconnaissance, Subsurface Probes and Mapping. The first task undertaken was a controlled surface collection. A 5-x-5-m grid was established, and all cultural material in each grid square was collected. Results of the surface collection indicate that cultural debris is densest at the center of the cave mouth near the terrace and at a point approximately 8 m north of the mouth (Figure 5). These points correspond to the outer edges of the large area of disturbance. Based on the results of the surface collection, it was decided to locate 1-x-l-m test excavation units at the cave mouth (Test Unit 1), in the main chamber (Test Unit 2) and at the rear of the main chamber (Test Unit 3). Although few artifacts were found in the rear of the main chamber, a test unit was placed there to investigate stratigraphy. Following the surface collection, the cave was mapped (Figure 4). Finally, subsurface coring was performed using an Oakfield corer in an attempt to determine the presence and depth of intact archaeological deposits in the cave (Figure 4). Cores were placed at 5-m intervals on north-south and east-west lines in the cave, yielding profiles from a total of 14 locations. The maximum depth reached was 23 cm, although most (n=12) probes were less than 20 cm deep. In general, the deepest deposits appear to be located in the front of the cave, from the mouth to a point about 8 m north, where depths between 19 and 23 cm were reached. This crosses the area which appears to be heavily disturbed but also includes the vicinity of Test Unit 2. These probes suggest that most of the remaining cave deposits are shallow (less than 20 cm), although thicker intact deposits appear to be preserved in the east part of the main chamber in the vicinity of Test Unit 2. Test Unit 1. This 1-x-l-m unit is located near the east wall on the terrace outside of the drip line of the cave (Figure 4). Several artifacts, including ceramics, were found on the surface during the 35 Artifact Density Per Unit Figure 5. Density Distribution of Surface Collected Materials at 23PU58. 36 controlled surface collection, and vegetation growth suggests that this area has not been disturbed recently. Since the terrace is very narrow, it is assumed that this area has not been disturbed because it represents the only access to the site from the east. Test Unit 1 was excavated to 1.65 m below ground surface in fifteen arbitrary 10-cm levels. Two strata were identified in Test Unit 1 (Figure 6). Stratum 1, from surface to between 110 and 125 cm below surface, is looter's backfill, consisting of recent and prehistoric cultural material in a loose, dry deposit. Stratum 1 is a dark brown (7.5YR3/2) to grayish brown (10YR5/2) to brown (10YR5/3) to pale brown (10YR5/3) loose to very friable, silt to silty sand to sandy silt with alternating bands of small pieces of dolomite debris. Stratum 2, 35 to 40 cm thick, is an undisturbed prehistoric deposit. Stratum 2 is a dark brown (7.5YR3/2) clayey silt with weak fine and very fine subangular blocky structure. This stratum contained several large dolomite blocks, which eventually prevented further excavation. Only prehistoric cultural material was recovered from this stratum, although density of this material was low. Recovered material consists of shell, bone, lithic debitage, and charcoal. Test Unit 2. This 1-x-l-m unit was placed along the east side of the main chamber, approximately 8 m from the cave entrance, and north and east of the heavily disturbed area (Figure 4). It was excavated in 10-cm levels to 80 cm below ground surface. Four strata were identified in Test Unit 2 (Figure 7). Stratum 1, between 11 and 32 cm thick, is a grayish brown (10YR5/2) silty loam. The sediment was dry, loose and very friable, and contained a mix of both historic and prehistoric cultural material. This stratum most likely represents a disturbed deposit consisting of both looter's spoil and Fowke's backdirt. Stratum 1 rests on a compact layer of small pieces of dolomite debris, which may represent an old buried cave floor. Beneath this layer is Stratum 2, a very dark brown (10YR2/2) silty loam with dolomite fragments that increases in clay content with depth. This stratum, between 5 and 15 cm thick, contained only prehistoric artifacts. Stratum 2 is separated from Stratum 3 by a clear boundary. Stratum 3, 10-16 cm thick, is very dark brown (10YR2/2) clayey silt loam with numerous dolomite and chert fragments. Large dolomite blocks were exposed in the northwest corner of the test unit in this level and appear to represent weathered pieces derived from the cave wall. This stratum produced only prehistoric artifacts, including ceramics, lithics, shell, bone, and charcoal. Finally, Stratum 4, from 20 to 40 cm thick, consists of a dark yellowish brown (10YR3/4) clay with numerous dolomite and chert fragments. The dolomite fragments increase in size with depth and extend into the western half of the unit, and are interpreted to represent cave wall degeneration and redeposition. Due to difficulties in removing the limestone blocks, excavation was suspended. Although solid bedrock was not exposed, the size and distribution of the dolomite blocks, together with the high clay content of the deposits, suggest that the bedrock floor was near. Attempts to determine the thickness of the remaining deposits below the base of the test unit failed due the high density of gravel debris. Stratum 4 produced only prehistoric cultural material, including ceramics, lithics, shell, bone, and charcoal. Two radiocarbon samples were submitted for assay from Test Unit 2. The first sample (BETA- 88675), from Level 6, consisted of 7.9 g of hand-collected bark and wood charcoal and carbonized hickory nutshell. This level, part of Stratum 2, is associated with Late Woodland ceramics. An assay of 1300±90 B.P. was obtained, yielding a calibrated date at two sigma of A.D. 600 (690) 960. The second sample (BETA-88676), from Level 7, also a part of Stratum 2, is associated with Late Woodland ceramics and an unidentified terminal Middle Woodland to Late Woodland projectile point. 37 Test Unit 1 North Profile Rock Fragments Stratum 1 Stratum 1 •-•- Gravel Lens Rock cm 50 Figure 6. North Wall Profile from Test Unit 1, 23PU58. 38 Test Unit 2 North Profile Test Unit 3 North Profile O Flake O Shell • Rock ■ 1 Limestone Layer I ^ Z > Dense Gravel and Clay cm 50 Figure 7. Wall Profiles of Test Units 2 and 3 from 23PU58. 39 This sample consists of 5.4 g of hand-collected oak and beech wood charcoal and carbonized hickory nutshell. Because of its size, an extended count was performed on this sample. It yielded an assay of 1510±80 B.P. Calibrated, the sample dates at two sigma to A.D. 400 (560, 590) 670. Thus, the two dates are stratigraphically correct and are in general agreement with the artifact assemblage. The assays indicate that intact deposits in Test Unit 2 date to the early portion of the Late Woodland Maramec Springs phase, or about A.D. 400 through 1000. Test Unit 3. Test Unit 3 was located in the rear of the main chamber, just to the south of the bedrock "step" which separates the main chamber from the rear chamber and at a point where the east and west cave walls almost converge (Figure 4). This unit, which measures 1-x-l m, was excavated to 25 cm below surface, at which point a rocky, clayey, culturally sterile deposit was reached. Cultural material recovered from this unit consists of prehistoric and recent debris from the initial 2-5 cm of deposits. This unit is very close to the small drainage which runs along the east wall of the cave, and it is likely that little prehistoric occupation or deposition occurred in this area. Two strata were identified in Test Unit 3 (Figure 7). Stratum 1, the initial 5-7 cm of deposits, consists of grayish brown (10YR5/2) silty clay with gravel and both prehistoric and historic cultural debris. Stratum 2, minimally 5-18 cm thick, is a culturally sterile subsoil deposit of strong brown (7.5YR 5/6) clay with gravel. Discussion. Both intact and disturbed deposits are present at 23PU58. Soil characteristics and the presence of historic artifacts were used to define disturbed strata. Looting, archaeological excavations by Fowke (1922) and military use of the cave have disturbed the upper 110 cm of Test Unit 1, the initial 30 cm of Test Unit 2 and all of Test Unit 3. The disturbed strata of test units 1 and 2 contain prehistoric artifacts dating to the Late Archaic and Late Woodland periods. This mixing of temporally distinct components appears to be due to the deposition of backfill from looter's pits and the excavations conducted by Fowke. Below the redeposited fill are intact deposits. The presence of a Late Archaic projectile point dates levels 12-15 of Test Unit 1. Terminal Middle Woodland to Late Woodland projectile points and Late Woodland ceramics, in conjunction with two radiocarbon assays, date the deposits between 30-80 cm in Test Unit 2. The radiocarbon assays and artifact analysis suggest these deposits date to the early portion of the Late Woodland Maramec Springs phase. It is possible that a late Maramec Springs phase component is present at 23PU58 as well, based on the shell-tempered ceramics and Scallorn projectile points found in the disturbed deposits of test units 1 and 2. It is likely, though, that these deposits have been impacted during the historic use of the site. Artifact Assemblage A total of 3,767 artifacts was recovered during the controlled surface collection and the excavation of the three test units. This material includes historic glass, metal, plastic and ceramics, and prehistoric lithics and ceramics. In addition, floral and faunal remains were recovered from the controlled surface collection, general excavation levels and flotation samples. Lithics from the general excavation levels were analyzed while ceramics from the general levels and flotation samples were analyzed. Each assemblage is discussed below and is analyzed by strata. Test Unit 2 yielded the largest artifact assemblage (n= 1,798; 48 percent), followed by Test Unit 1 (n= 1,654; 44 percent). Relatively few artifacts were recovered from Test Unit 3 (n=89; 2 percent) or the controlled surface collection (n=226; 6 percent). This trend is apparent when the total artifact count per unit is standardized to 1 m 3 of fill. Material is over twice as dense in Test Unit 2 (2,248) as Test Unit 1 (1,002), and five times as dense as Test Unit 3 (445). This suggests that the Test Unit 3 area, near the rear of the cave, was only incidentally occupied while the area in which Test Unit 2 was placed is the main area of 40 occupation of the site. The lower figure for Test Unit 1 is probably due to both its location outside the drip line of the cave and the fact that most of the deposits consist of redeposited fill. Historic Artifacts. A total of 48 historic artifacts was recovered from the surface and three test units (Table 2). Aside from the surface collected material, historic artifacts were found in the upper 1 10 cm of Test Unit 1, the initial 30 cm of Test Unit 2, and the initial 10 cm level in Test Unit 3. The presence of the historic artifacts indicates a widespread historic disturbance of prehistoric deposits at 23PU58 and aids in the definition of intact deposits at the site. Most of the material consists of bottle glass and metal. The bottle glass (n=28, 58 percent) is composed of clear and green soft drink beverage- container fragments. Two other pieces of glass from Test Unit 1 may represent lamp glass fragments. Metal is also fairly common (n=ll, 23 percent). This class includes rusted wire-cut nails, a rusted wire-cut nail embedded in mortar, soft drink cans, military ration cans, wire, and an electric light base, perhaps from a flashlight. Other historic artifacts recovered during investigations include a metal soft drink bottle cap, a plastic spoon, a plastic cigar mouth piece, a shotgun shell casing, a coal cinder, and a piece of Bristol-slipped stoneware. Clearly, all of the historic artifacts recovered from 23PU58 are recent, probably representing a mid- to late- twentieth century date of manufacture. Their deposition is likely due to two separate historic uses of the site. One would appear to be the use of the cave during military training, while the other is no doubt associated with either Fowke's 1920s excavations or the extensive looting evidenced at the site. Prehistoric Ceramics. A total of 266 prehistoric ceramics was recovered from the three test units and the surface collection at 23PU58 (Table 3). Ceramics were found in the disturbed stratum of Test Unit 1, from disturbed and undisturbed strata in Test Unit 2, and from the disturbed stratum in Test Unit 3. No prehistoric ceramics were found in Stratum 2 of Test Unit 1. Test Unit 1 contained the greatest number of ceramics (n=120, 45 percent), followed by Test Unit 2 (n=99, 37 percent). Most of the ceramics were recovered either from the surface or looter disturbed deposits (n=223, 84 percent), with relatively few from intact deposits. Aside from those sherds either too small or eroded to be analyzed, six ceramic types were identified in the 23PU58 assemblage (Table 3). Most common are Maramec Cordmarked, comprising 61 percent of the assemblage. An additional four percent are smoothed-over-cordmarked. Less common are Maramec Plain sherds (21 percent). Two small dolomite-tempered sherds have incised lines. These four Maramec ceramic types tend to have a fairly fine dolomite temper, although occasionally larger inclusions are present. One Maramec Plain sherd has a combination of dolomite and grit as tempering agents. Single examples of shell-tempered plain and grit-tempered plain sherds are also present in the assemblage. In addition, a single piece of dolomite-tempered baked clay was recovered from Stratum 2 of Test Unit 2. This piece may represent a baked clay object, figurine fragment or vessel adorno. Cordage twist data was also collected from 125 of the 166 Maramec Cordmarked sherds. S-twists dominate the assemblage (n=91, 73 percent) with fewer Z-twists present (n=34, 27 percent). Additionally, S-twists are numerically superior in each of the separate strata. A total of 13 rim fragments also are present in the assemblage. Nine of these are Maramec Cordmarked jars and four are Maramec Plain jar rims. Of the nine Maramec Cordmarked jar rims, four profiles are inslanting to vertical-incurvate, two are inslanting, one is outslanting-incurvate, and 41 Table 2. Distribution of Historic Artifacts from 23PU58. Artifact Class Surface Test Unit 1 Test Unit 2 Test Unit 3 Total Stoneware 1 1 Bottle Glass 7 6 12 3 28 Unid. Glass 2 2 Metal 6 5 11 Cinder 1 1 Plastic 3 1 4 Unidentified 1 1 Total 8 17 19 4 48 two are unidentifiable. Four of the lips are rounded, three are outslanted and one is pointed. In addition, one of the rounded lips is cordmarked. A profile could be determined for only one of the Maramec Plain jar rims, an inslanted form. Two of the rims have pointed lips, one is rounded and one is outslanted. Temporally, the ceramics are in agreement with both the radiocarbon assays discussed above and the projectile points recovered. Aside from ceramics in the disturbed strata, all are dolomite tempered with a high frequency of cordmarking. Such an assemblage is rather typical of the early Maramec Springs phase assemblage (Reeder 1988). The presence of shell-tempered sherds and incised ceramics in the upper, disturbed levels, may indicate the presence of a late Maramec Springs phase component at the site as well. Lithic Assemblage. Lithics are the single largest artifact assemblage from 23PU58 (Table 4). The assemblage consists of numerous flake categories, bifaces— including temporally diagnostic projectile points— unifaces, and several ground-stone tools. Most of the lithics were recovered from test units 1 and 2. When standardized to 1 m 3 , Test Unit 2 has over two times the density of lithics as Test Unit 1 (2143 to 958), with fewer in Test Unit 3 (240). The surface collection consists of 97 chert artifacts and one ground-stone tool, a grinding stone (Table 4). Two of the chert artifacts are temporally diagnostic: a Late Archaic Motley point (Chapman 1975; Justice 1987) and a Terminal Middle Woodland to Late Woodland Rice Side Notched point (Chapman 1975; Justice 1987) (Figure 8, a-b). The remainder of the lithic material consists of block shatter and flake types indicative of various stages of lithic reduction. Given the mixing of deposits due to looting and previous archaeological investigations at the site, no meaningful statements can be made regarding lithic manufacturing sequences based on the surface collection. 42 Table 3. Distribution of Prehistoric Ceramics at 23PU58. TU 1 TU2 TU2 TU 2 TU 2 TU3 Type Str. 1 Str. 1 Str. 2 Str. 3 Str. . 4 Str. 1 Surface Total Too Small 12 8 3 5 1 5 34 Maramec Plain 26 6 4 3 3 3 11 56 Maramec Smoothed- 6 2 1 1 1 11 Maramec Cordmarked 74 40 13 7 2 12 13 161 Maramec Incised 2 2 Shell-Tempered Plain 1 1 Grit-Tempered Plain 1 1 Total 120 56 21 16 6 22 25 266 Test Unit 1 yielded a large lithic assemblage, although most of the material was derived from the disturbed Stratum 1 levels (Table 4). Diagnostic projectile points derived from Stratum 1 consist of two Late Archaic Etley or Ledbetter Cluster projectile points (Justice 1989) (Figure 8, c-d), a Rice Side Notched point dating to the terminal Middle Woodland or Late Woodland periods (Chapman 1980) (Figure 8, e), and a single distal portion of a possible Scallorn projectile point (Bray 1956; Chapman 1980). The remaining retouched artifact assemblage consists of thin biface fragments (n=ll), end scraper fragments (n=4) and a single burin. Six cores were recovered, including two plano-convex, two multidirectional and two fragments. Debitage consists of 1,352 pieces, representing all flake categories, but since this stratum is composed of mixed deposits, no meaningful conclusions can be inferred regarding lithic reduction sequences. A small sandstone grinding-stone fragment also was recovered from this stratum. Test Unit 1 Stratum 2 has a small assemblage of 202 lithic artifacts (Table 4). One diagnostic artifact, a Bottleneck Stemmed point (Chapman 1975) (Figure 8, f), was found in this stratum. No other retouched artifacts were recovered. The lack of cores and paucity of early-stage manufacturing debris suggest that tool production was not an important activity at this time. Bifacial thinning flakes are relatively abundant, especially when compared to strata in Test Unit 2, indicating that biface maintenance may have been an important component of the Late Archaic activities at 23PU58. Four stratigraphic units are present in Test Unit 2 (Table 4). Stratum 1 contains a mixture of prehistoric and historic artifacts and is interpreted as backdirt from looting and possibly Fowke's archeological investigations. A total of 704 lithic artifacts was found in Stratum 1, including eleven temporally diagnostic projectile points. Late Archaic types are represented by one Pickwick Stemmed 43 oo in P cu co CN £ o ,*H 1-H '•§ 03 H o H ■o c y 3 c O o O OO 1) o CS <4-C c 3 o In Oh oocNCN-rJ-r-OcOONO OnO\OOoocoO\nono m cn f- co ^- *— i ^f ^H CO ON o CO i-H ^ in cn -H O CN in O on CN CO CN ON in co in o\ NO t-H CN ^h i— i in O ^ © © ON ©no©i-h^h©©^hQn CNN©^Ht^inNOcOi-Hi— i ^oo©ONNOO\CNNO ~HinCN^Tj-rtCN^ ON CO i— I no r- t-- ^h