Lithic variability and the cultural elaboration of Upper Pleistocene North China.

Item Type text; Dissertation-Reproduction (electronic)

Authors Miller-Antonio, Sari.

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 07/10/2021 04:01:47

Link to Item http://hdl.handle.net/10150/185777 INFORMATION TO USERS

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

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

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

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

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

V·M·I University Microfilms International A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor. M148106-1346 USA 313/761-4700 800/521-0600

Order Number 9220'108

Lithic variability and the cultural elaboration of Upper Pleistocene North China

Miller-Antonio, Sari, Ph.D. The University of Arizona, 1992

D·M·I 300 N. Zeeb Rd Ann Arbor, MI 48106

LITHIC VARIABILITY AND THE CULTURAL ELABORATION OF UPPER PLEISTOCENE NORTH CHINA

by Sari Miller-Antonio

A Dissertation Submitted to the Faculty of the DEPARTMENT OF ANTHROPOLOGY In Partial Fulfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY In the Graduate College THE UNIVERSITY OF ARIZONA

1 9 9 2

- . ---_._------2 THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have

read the dissertation prepared by Sari Miller-Antonio entitled _L_I_THI__ C_V_:ARI __ AB_I_L_I_TY_' _AND __ THE__ aJU__ rruRAL ___ ELAB

OF UPPER PLEIS'l'OC.mE NORrH CHINA

and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy

12/20/91 Date 12/20/91 Date 12/20/91 Date

Date

Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

12/20/91 sen Date 3

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library. Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

------4 ACKNOWLEDGMENTS The opportunity to study China's extraordinary prehistoric record has deepened both my understanding of and respect for that country's rich cultural heritage. I would like to express my thanks to the Chinese Academy of Sciences, especially the Institute of Vertebrate Paleontology and Paleoanthropology (IVPP) and its former director Zhang Miman, f,or facilitating my dissertation research. This work would not have been possible without the generosity and logistical support of Professor Huang Weiwen who graciously shared with me his office space, his time and his vast knowledge of the Chinese Paleolithic. I am also indebted to Professors Jia Lanpo and Zhang Senshui for their advice and assistance. My colleaques on the third floor of IVPP became my extended family during the summer field seasons since 1987. I thank them collectively. They have greatly enriched my visits to China. Many thanks to the IVPP artists for providing the outstanding stone and bone artifact illustrations. I extend my gratitude to my dissertation committee. Dr. John W. Olsen's impressive reputation and pioneering work in China made it possible for me to undertake this project. As my committee chair, he offered me his expertise and quidance. As my friend, his belief in my abilities helped me through many difficult times. It is often true that the toughest criticism is, in the end, the most productive. I thank Dr. Arthur J. Jelinek for his constructive criticism tempered by his wisdom and humor. At the start of my graduate work in Anthropology, Dr. William A. Longacre played a pivotal role in encouraging me to pursue this degree. I will always be grateful for his support and kindness. Pursuing this degree would have seemed less worthwhile, were it not for the continuous encouragement of family and friends. I thank Susan Bierwirth, Donald Whelan, Irving Miller, Debbie Fogelman, Barb Roth and Laura Fulginiti for their support. My deepest thanks to my husband, Mark Antonio, for his endless patience, reassuring disposition and for cheerfully sharing the sometimes overwhelming demands of our family. 5

For my daughter, Lili 6 TABLE OF CONTENTS LIST OF ILLUSTRATIONS ••••••••••••••••••••••••••••••••• 9

LIST OF TABLES ••...•..•..•••..••..•••...... •.•.•••. 11

ABSTRACT •••••••••••••••••••••••••••••••••••.••.•••••• 13 1. RESEARCH OBJECTIVES AND INTERPRETIVE MODELS IN THE CHINESE PALEOLITHIC ••••••••••••••••••.••••••••••••..• 15 Introduction and Research Objectives ••••••••••••••••• 15 A Current Model of Paleolithic Traditions in China ••• l? Historical Foundations •••••••••••••••••••••••••• l? Small Tool vs. Microlithic Traditions ••••••••••• 19 Typology ...... 24 Large Tool Tradition ••••••••••••.•••••••••••..•• 25 The Question of Style and Stone Artifacts •••••••••••• 2? Composition and Methodology of this Study •••••••••••• 31 Chronological Terms ...... 34 Summary ...... •...•...... ••...... 35 2. PREHISTORIC TECHNOLOGIES .•••.•••.••••...•••..••.••••. 39

Li thic Technol 09Y ...... 40 Raw Material for Lithics in China ••••••••••••••• 40 African Technology ...... 43 The Organization of Lithic Technology ••••••••••• 46 Lithic Reduction Strategies and Projectile Technology ...... 51 Bone Technology ••••••.•...••••••.••.•.•••...•.••....• 55 Bone and Antler as Raw Material for flaked Implements ...... 56 Taphonomic Considerations ••••••••••••••••••••••• 59 Wood Technology ...... 61 Summary and Conclusions .••••••••••••••••••••••••••••• 64 3. THE QUATERNARY PALEOENVIRONMENT OF NORTH CHINA ••••••• 68

Introduction ...... 68 Mega-geomorphic Regions and the Origins of Aridity in North China ••••••••••••••••••••••••••••••• 69 Arid Zones Defined ••••••••••••••••••••.••••••••• 69 The Formation and Expansion of North China's Deserts ...... 71 The Formation of the Loess Plateau •••••••••••••• ?3 The Marine Oxygen Isotope Record and Pleistocene Climate ...... 76

------_._------7 Table of Contents (continued) The Loess-Paleosol Sequence Correlated with the Oxygen Isotope Curve •••••••••••••••••••••••• 77 The Pollen Record and Pleistocene Climate •••••••••••• 79 Indicator Species and the Comparative Approach ...... 81 North China Pleistocene Pollen Assemblages and Paleoclimate...... • ••• 82 Pleistocene vertebrate Paleoecology •••••••••••••••••• 83 Animal Fossils in the Loess ••••••••••••••••••••• 85 Artiodactyla ...... 86 Perissodactyla...... •••••• • ••••• 87 Summary ...... 89 4. THE SALAWUSU SITE, INNER MONGOLIA ••••••••••• • ••• 95 Introduction ...... • •.• 95 A History of Archaeological Investigations at Salawusu ...... • ••• • 96 Chronology ...... 96 Artifacts from Pre-l980 Collections: Early Interpretations ...... 99 The Human Fossils From Salawusu •••••••••••••••• 103 The Salawusu Fauna: Chronological Impl ications ...... •••• 106 The stratigraphy of the Salawusu site ••••.• .•.• 110 The Salawusu Stratigraphy: Chronological Implications ...... • ••. 113 Pollen and Faunal Correlations with Paleocl imate ...... • .•• 116 The 1980 Stone Artifact Assemblage ••••••• • ••• 117 Cores and Flakes .•••••••••••••••••••••••••••••• 118 The Reconstruction of Flaking Technology: Flake Form and Size, Termination, Striking Platform, and Cortex...... • ••• 121 Tools...... 123 Scrapers •••••••••••• •••• 123 Circular Scrapers ••• • ••• 124 Borers ...... ••• 125 Gravers and Notches. • ••••• . ..•• 125 Bone Tools at Salawusu •••••••••••••• • ••• 126 The ,Absence of Bone Flakes •••••••••• . ••• 127 Flaking Patterns ••••••••••••••• • ••• 128 Polished Bone Implements .••••.• ...... 129 The Use of Antler...... • ••••• • ••• 130 Discussion ...... • ••• 130 Summary and Conclusions •••••••••••••••••• • ••• 134

------8 Table of Contents (continued)

5. THE SHIYU SITE, SHANXI PROVINCE ••••••• .•...... 159 Introduction ...... 159 The 1963 Excavation of Locality 63661 ••••••••••••••• 160 Site Stratigraphy •••••••••••••••••••••••••••••• 161 Faunal Remains, Paleoenvironment and Chronology ...... 162 The Cultural Remains and Designation of the Small Tool Tradition ••••••• .165 The Shiyu Lithic Analysis ••••••••••••••••••• .167 Typoloqy ...... 168 Artifact Classes •••••••••••••••••••• .168 Flakes...... 168 Bipolar Flakes ••••••••••••••••• • •••• 169 Blades ...... 169 Flake-Blades •••••••••••••••••• .170 Points ...... 170 Side Scrapers ••••••••••••.••• • •• 170 Transverse Scrapers •••••••••••••...... 171 Notches .....••.•.•...•....•...... 171 Discuss ion ...... 172 Artifact Dimensions and Standardization of Form ••...••••.•..••••••••.••••••.•....•••..• 172 Tool Typology and Functional Interpretations •.. 174 Raw Material Use ••••••••••••••••••••••••••••.•• 175 Flaking Techniques and Reduction Sequence •••••• 176 Retouch and Edge Damage...... • •• 178 Comparative Analysis Between the sites.. • .• 180 Summary and Conclusions...... 183

6. CONCLUSIONS •••..•••••..••.•••••••••••••.••..• • .211 Chronoloqy, Typology and Hominid Tool Makers. ..212 Inter-site Variability as a Reflection of Hominid Activities ...... ••• 216 Technological Antecedents ••••••••• • •• 219 APPENDIX A - LIST OF VARIABLES •••• • •• 222

REFERENCES •••••••••••••••••••••••••••••••••••••••••• 225 9 LIST OF ILLUSTRATIONS Figure Page

1. Archaeological sites mentioned in the text ••••••••••• 37 2. A current model of paleolithic traditions in North China ...... 38

3. A hypothetical assemblage formation model •••••••••••• 66 4. Summary diagram of the interaction of prehistoric technologies ...... 67 s. The distribution of loess deposits in China •••••••••• 91 6. Loess section from Lujiaowan, North Xinjiang ••••••••• 92 7. Loess/paleosol correlation with 6180 chronology •••••• 93 8. Summary pollen diagram by vegetation type ••••••.••••. 94 9. Small tools from the 1923 Salawusu excavations ••.•.• 136 10. Small tools from the 1923 Salawusu excavations •••••• 137 11. The Paleolithic industry at Shuidonggou •.••••••••••• 138 12. The Salawusu stratigraphic Section (A and B) •••••••• 143 13. Flake measurements and attributes ••••••••••••••••••• 144

14. Flake area/flake thickness plot for Salawusu flakes ...... 149

15. Flake area/platform area plot for Salawusu flakes ...... 150

16. Platform area/flake thickness plot for Salawusu flakes ...... 151 17. Length of Salawusu flakes and tools ••••••••••••••••• 152

18. Width of Salawusu flakes and tools •••••••••••••••••• 153 19. Thickness of Salawusu flakes and tools •••••••••••••• 154

20. Tools recovered from Salawusu in 1980 ••••••••••••••• 155

------10 List of Illustrations (continued) 21. Tools recovered from Salawusu in 1980 ••••••••••••••• 156

22. Worked bone from S,alawusu •••••••••••••••••••.••••••• 157

23. Assemblage formation model •••••••••••••••••••••••••• 158

24. The Shiyu stratigraphic section ••••••.•••••••••••••• 185 25. Flake length/width ratios for Shiyu flakes •••••••••• 186 26. Flake-blades from Shiyu ••••••••••••••••••••••••••••• 187

27. Points from Shiyu •••••••••••••••••••••••••••..•••••• 188 28. Blades from Shiyu ••••••••••••••••••••••••••••••••••• 189

29. Cores, scrapers and notches from Shiyu ••••••.••••••• 190 30. Length of blades, points and flake-blades .•••.•••••• 198

31. width of blades, points and flake-blades .•...•..••.. 199 32. Thickness of blades, points and flake-blades .•.••.•. 200 33. Length of Shiyu flakes and tools ••••••...•••.•...••• 201

34. width of Shiyu flakes and tools ••••••••••••••.•••••• 202

35. Thickness of Shiyu flakes and tools •••••••••••.••••• 203

36. Length/width plot for Salawusu and Shiyu flakes ••••• 204

37. Length/width plot for Salawusu and Shiyu tools •••••• 205

38. Length/thickness plot for Salawusu and Shiyu flakes ...... 206

39. Length/thickness plot for Salawusu and Shiyu tools ...... 207

40. Flake area/platform area plot for Shiyu flakes •••••• 208

41. Flake area/flake thickness plot for Shiyu flakes ...• 209

42. Platform area/flake thickness plot for Shiyu flakes ...... 210 11 LIST OF TABLES Table Page Sa1awusu Data Tables (1 - 23) 1. The human fossils recovered from Sa1awusu ••••••••••• 139

2. The Sa1awusu fauna. .141

3. Artifact class ••••• .145

4. Tools ...... 145

5. Raw material relative frequencies. .145

6. Tool blank •.•••••••.•• .145

7. Termination of flakes. .145

8. Flake platform type ••• .146

9. Flake shape ••••••••••• .146

10. Bulb of percussion on flakes. .146 ll. Amount of cortex on flakes. .146

12. utilization on flakes ••.••• .146

13. Intensity of utilization on flakes. .147

14. Location of retouch on scrapers •••• .147

15. Ratio of retouch to circumference of scraper. .147

16. Scraper and borer retouch morphology •••••••• .147

17. Frequency of flaking patterns on bone tools. .147

18. Flake length. .148

19. Flake width •• .148

20. Flake thickness. .148 2l. Tool length...... 148 22. Tool width •• .148 12 List of Tables (continued)

23 • Tool thickness ...... 148 Shiyu Data Tables (24 - 43) 24. Artifact class frequencies •••••••••••••••••••••••••• 191

25. Artifact length ...... 191

26. Artifact width ...... 191

27. Artifact thickness •••••••••••••••••••••.•••••••••••• 192 28. striking platform width and thickness •.•••.•.•.••••• 192

29. Mean length of retouch on edges •••••.••...•.•••••... 192 30. Mean width of notch ••••••••••••••••••••••••••....••• 192 31. Completeness of artifact ••••••••••••••••••••••.••••• 193 32. Raw material use by artifact class ..•....•...•...... 193

33. Raw material use for flakes vs. tools ••••••••.•••••• 193 34. Interior flake surface •••••••••••••••••••••••••••••• 194

35. Striking platform type by artifact class •••••••••.•• 194 36. % of prepared or modified platforms ••••••.•••••••••. 194 37. Number of exterior flake scars •••••••••••••••••••••• 195

38. Pattern of exterior flake removal ••••••••••••••••••• 195

39. % of cortex ...... 195 40. Retr)uch location by artifact class •••••••••••••••••• 196

41. Edge damage location ••••••••••••••••••••••••••.••••• 196

42. Shiyu flakes ...... 197

43. Shiyu tools ...... 197 13 ABSTRACT The study of Paleolithic archaeology in China has flourished over the last two decades, providing a unique perspective on the development of human culture and emergence of modern Hom,o. This dissertation is a comparative study of two Upper Pleistocene sites in North China: the site of Salawusu, in Inner Mongolia and the Shiyu site in northern Shanxi province. Lithic assemblages from both sites have been assigned to the broad 9ategory of "small tool tradition" as defined by contemporary Paleolithic archaeologists in China. This study addresses the nature of the variability within this tool tradition as it relates to reduction strategies, the organization of prehistoric technology and the role that resource type and availability played in this organization. A critical discussion of the radiocarbon dates for these sites and a correlation of the stratigraphy with the Oxygen Isotope chronology and paleoenvironmental evidence, suggests that these sites are separated by a much greater temporal span than their radiocarbon dates have indicated. The interpretation of the Salawusu section proposed in this study correlates the cultural materials recovered in the

1980 excavations with cS 180 Stage 5 while the Shiyu remains

18 are correlated with cS 0 stage 2.

------14 The lithic analysis documents quantitative and qualitative differences between these artifact assemblages which also call into question the closeness of their relationship. The Shiyu lithics are a homogeneous collection dominated by elongated flakes, blades, points and implements fashioned on these. The Sa1awusu assemblage is less standardized and shows no trend toward flake elongation or blade technology. 15

CllAP'l'BR 1 RESEARCH OBJECTIVES AND INTERPRETIVE MODELS IN THE CHINESE PALEOLITHIC

Introduction and Re.earch Objective.

The first excavations, conducted in the 1920's, at the now famous cave site of Zhoukoudian, southwest of Beijing, uncovered the Middle Pleistocene remains of Homo erectus and an associated material culture (Pei 1931~ Weidenreich 1935~ Wu and Lin 1983). This caught the attention of scholars worldwide and established the antiquity of the Hominidae in China. until then, Eugene Dubois's 1891 recovery of a Homo erectus calvarium and femur along the Solo River near the village of Trinil in Java had been the easternmost discovery of this hominid. In the decades that followed, and especially in the last 15 years, the study of Paleolithic remains in China has flourished. The archaeological record from this vast region of the Asian continent provides a unique perspective on the development of human culture and the emergence of modern Homo. This work is a comparative study of two Upper Pleistocene sites in North China: the site of Salawusu, 16 located in the Ordos region of Inner Mongolia and the Shiyu site in northern Shanxi province (See Figure 1). Lithic assemblages from both sites have been assigned to the broad category of "small tool tradition" as defined by contemporary Paleolithic archaeologists in China (Jia et al. 1972; Jia and Huang 1985). This definition is presented and discussed in detail later in this chapter. The research focus of this study addresses the nature of the variability within this tool tradition. This lithic analysis offers a quantitative test of the similarities and differences of the stone assemblages recovered from these two sites. One objective is to determine whether patterned attributes of lithics, often inferred to be stylistic, reflect ethnically distinct paleocultural groups. This study proposes that the variability within these North Chinese small tool assemblages is closely tied to a prehistoric cultural system with a technological component using resources of stone, bone and wood. This technological interaction was especially critical in marginal arid environments where lithic raw materials were scarce or of poor quality. While this study interprets archaeological data from a comparatively small geographic region in relation to the 17 entire Chinese landmass, it is hoped that it will make the prehistory of East Asia more accessible and stimulate the continued exchange of scholarly cooperative research projects.

A CUrrent Hodel of Paleolithic Traditions in China

Historical Foundations

For the past twenty years, China's foremost Paleolithic archaeologists have been concerned with sorting the typological variability in Pleistocene lithic assemblages (Jia et a1. 1972; Jia and Wei 1976; Huang 1985; Zhang 1985; Qiu and Li 1978). This section presents the current, most popular model of Paleolithic traditions in China. The stone artifact assemblages of the Upper Pleistocene in western Europe were the standard of reference for the analyses of patterning in lithic assemblages worldwide through the mid-twentieth century. Perceptive archaeologists in China during this period realized that characterizing the Asian material in this manner was counterproductive (Jia et a1. 1972; Chang 1986). Clearly, the Paleolithic assemblages that they were excavating and studying were typologically distinct from those of Europe. 18 Their unique character was due in part to the nature of the raw material. Sources of fine-grained cherts, flint and obsidian are rare in China. Quartz and quartzite pebbles are abundant. still, given that there are a finite number of ways of flaking any stone, the distinctivness of the Asian lithics could not be explained on the basis of lithic raw material differences alone. In the 1970's, Chinese archaeologists reclassified Paleolithic sites into complexes which shared certain typological traits (Jia et al. 1972). Initially, two traditions were identified and assemblages were placed into one of these two series. The Kehe-Dingcun series was characterized by heavy triangular points and large chopper-chopping tools made on flakes. The Zhoukoudian Locality 1-Shiyu series was typified by small tools made on irregular flakes exhibiting fine retouch (Jia et al. 1972). As a result of this reorganization, Chinese archaeologists concluded that there existed a continuum of technological features manifest even as early as the Middle Pleistocene in the tool assemblage associated with Homo erectus at Zhoukoudian Locality 1 that could be traced through lithic assemblages into the late Upper Pleistocene and culminated in the true microlithic industry of Neolithic North China {Jia et al. 1972; Gai 19

1985). Figure 2 illustrates the multilinear model that exists at present. The authors of this model have stated (Jia and Huang 1985: 262) that their hypothesis was meant to stimulate research and discussion. Indeed it has, and it forms one of the foundations upon which this study is based.

Small Tool vs. Microlithic Traditions

In this study, a "tradition" implies a demonstrable continuity of culturally meaningful activities as reflected in prehistoric technologies. It suggests a cultural transmission system which may differentiate social entities. The use of "small tool" tradition in this study should not be confused with a much younger microblade and burin tradition (ca. 2500 BC) known as the "Arctic Small Tool" tradition and identified from northwestern Alaska, Arctic Canada and Greenland (Dumond 1978). The differences between small tool assemblages and microlithic assemblages can best be explained by clarifying what a microlithic industry is. In recent years, the characterization of microlithic industries has been delimited by Chinese archaeologists studying the problem as 20 it applies to North China (Gai 1985). A microlithic industry is one which may consist of a combination of microblades, scrapers, and small projectile points with flat or concave bases and a variety of microcores from which the microblades are produced. Microblades are flakes of stone, typically about 2 mm thick, with a triangular or trapezoidal cross section. The length of microblades is variable, and ranges between 20 and 60 Mm. They are usually less than 10 mm in width (Gai 1985). Microblade cores are often classified on the basis of their shape. Descriptive terms such as, wedge-shaped, conical or cylindrical are used. The relative proportions of tool types in microlithic assemblages varies. For example, in some of the surface localities recently dis,covered on ancient river terraces in southern Xinjiang, the assemblages consist almost exclusively of microblades with microblade cores being next in abundance (Huang et ale 1988; Olsen et ale 1989). However, in one locality a collection of core preparation flakes was found in a wide scatter accompanied by very few microblades or cores. It is possible to interpret these differences in assemblage composition as a reflection of site function. However, this distribution bias could be a function of complex site formation processes including

------21 differential weathering, tectonic activity, and even local weather conditions during the field season. A characterization of the small tool tradition is not quite as easily delimited. The range of variability in tool types and in tool size is greater than in micro1ithic assemblages. In addition, the temporal span through which these assemblages extend is much larger. An integral part of Jia and Huang's (1985) model is the implication that there is an evolutionary relationship between the small tool tradition and the later micro1ithic industries. Several sites, informally designated as "type" sites, were initially used to characterize the small tool tradition. The following discussion of these sites is intended to accentuate the variability within this tool tradition. The extensive lithic assemblage of tools made on small flakes from the site of Zhoukoudian Locality 1 has been cited in support of the antiquity of the roots of micro1ithic technology in North China (Zhang 1985, 1989). It also contains large chopper-chopping tools, exceeding 100 mm in length, and stone spheroids exhibiting a broad size range. The long temporal sequence of stone tools found here in association with Homo erectus beginning at about 500,000 BP exhibits technological trends through time that suggest 22 an increasing ability to manipulate the available stone resources. These trends include a reduction in tool size, the increasing use of bipolar flaking and more regularized retouch. By about 200,000 BP, both the quantity and quality of flakes indicate a considerable degree of skilled tool manufacture. It is in this stratigraphic context that small irregular flakes and bipolar flakes which are longer than they are wide, are recovered along with a variety of scrapers, points, shouldered-awls, and gravers (Zhang 1985). This early manifestation of knapping skill is particularly impressive in light of the fact that the dominant raw material is quartz, an intractable stone to work. Ironically, it may have been the abundance of a relatively poor quality raw material that stimulated the development of more proficient knapping techniques. Several other "type" assemblages have contributed to the characterization of the small tool tradition. Xiaochangliang and Donggutuo, two sites located in the Nihewan basin of northern Hebei province, are particularly interesting for several reasons. The stone artifacts come from a sand layer more than 40 m below the present ground level and are associated with mammalian fauna suggesting a Lower Pleistocene antiquity for these localities (Huang 1985). A paleomagnetic date of 1,000,000 BP has also been

------23 determined (Wei 1985; You et a1. 1980). The 1ithics are fashioned almost exclusively on small chert pebbles from a nearby gravel bed. There are a great variety of scrapers, with fine retouch on one or two edges as well as endscrapers, notches, points, borers, small chopping tools and burins. Simple direct percussion is the dominant flaking and retouch technique. Bipolar flaking is also evident here (Jia et ale 1982). Again, there are a few large chopper-chopping tools as well as pieces of flaked fossilized bone (Wei 1985). The stone artifacts recovered from Xujiayao, Shanxi were found in association with fossils of early Homo sapiens and an early Upper Pleistocene fauna (Wu 1980). The assemblage consists primarily of small irregular flakes fashioned into scrapers, gravers and denticulated points. More than 1,000 stone spheroids ranging in size from over 100 mm to less than 50 mm in diameter have been recovered from this site. Discoidal, prismatic and funnel-shaped cores thought to be antecedants to microblade cores have also been found here. This interesting collection of artifacts contains a wide variety of raw material including vein quartz, chert, agate and marble (Jia and Wei 1976; Jia et a1. 1979). 24 Typology

In part, the difficulty in defining the small tool tradition is due to the lack of a universally accepted descriptive typology specifically tailored to the Chinese Paleolithic. One of the positive aspects of such a typology is to enhance communication among researchers and facilitate comparative studies. Interpretive studies of Chinese Paleolithic remains have, to an extent, been hampered by this deficiency. On the other hand, there is less tendency to place an assemblage into one of the traditonal Paleolithic subdivisions as they have been defined for other geographic regions. Assigning Chinese assemblages to inappropriate chronological divisions would certainly bias subsequent interpretation. This study borrows from two typological systems for the express purpose of promoting a clear representation to the reader of the distinct morphology of the artifacts from the sites of Shiyu and Salawusu. It is not £he place or purpose of this dissertation to create a new typological system. Broad tool classes such as scraper, point and notch as defined by Bordes (1961a) will be used throughout this study. By way of supplemental description, each site chapter contains definitions of artifact types as proposed 25 by Akazawa et ale (1980) which are most applicable to the North Chinese small tool flake industries. Patterns of use and retouch on both complete and broken specimens are particularly important to the interpretation of both assemblages and are discussed within each site chapter.

Large Tool Tradition

The large tool dominated tradition, characterized on the basis of the 1ithics from the Rehe and Dingcun sites located in shanxi province, exhibit slightly more homogeneity than the small tool assemblages. Several localities at Rehe have yielded an assortment of spheroids and large choppers exceeding 100 rom in length and 500 g in weight. Most tools are unifacia1 but the pointed specimens are bifacia11y flaked trihedral pick-like implements (Zhang 1985; Huang 1987). Dingcun, a series of 10 localities along the east bank of the Fen River, has yielded over 2,000 artifacts mostly fashioned on large hornfels flakes. The heavy trihedral points and pick-like implements from this assemblage are considered to be characteristic early Upper Pleistocene tools of the Fen River valley region (Qiu and Li 1978). In several recent 26 publications, Chinese scholars have emphasized that the biface as a tool type is indeed a component of the Paleolithic in both North and South China (Huang 1987; Huang et ale 1990; Ceng 1983; Huang and Qi 1987). An important question for future research will be to determine how much natural sorting of tools has taken place in these assemblages dominated by larger implements. The site of Shuidonggou is located in the eastern Ningxia Hui Autonomous Region. Jia and Huang's model places the Shuidonggou lithic assemblage in a medial position (see Figure 2) because of its large and small tool components. The larger tools consist of elongated points and blades resembling the Upper Paleolithic blade industries of the western Old World (Jia et ale 1964). The small tools are dominated by finely retouched end scrapers, notches, points, small blades and blade cores with faceted striking platforms. Radiocarbon dates of 26.230 ± 800 years BP and 17,250 ± 210 years BP have been determined for this site (Ningxia Hui Museum 1987). The model suggests that this type of mixed assemblage finds its precursors in both tool traditions and that later microlithic industries could have developed directly from this technology. The chacterization of both traditions is based on identifiable attributes in lithic artifacts. Some 27 researchers (Wobst 1977; Close 1978; Weissner 1983; Sackett 1973) have suggested that these attributes, when inferred to be stylistic, imply the possibility of identifying paleocultural groups.

The Question of Style and Stone Artifacts

Arriving at an explanation for the observed variability in synchronic lithic assemblages is a common thread running through studies of culture change in the Paleolithic. The typological/technological approach, inherited from the 1950's and 60's studies of the Mousterian in the Perigord region of southwestern France, has dominated interpretations of cultural development through the Paleolithic (Laville et al. 1980; see Jelinek [1988], for a concise review of this subject). This approach, generating heated debates on the roles of form and function in lithic assemblages, has focused on the nature of lithic variability in the Middle Paleolithic (Bordes 1961b; Binford and Binford 1966; Bordes and Bordes 1970; Binford 1973). It stems from the extrapolation of the clear stylistic patterns of the Upper Paleolithic back into earlier periods, without justification. In larger perspective however, it has served as the

-~------28 stimulus for addressing the general question of stylistic variability as a reflection of paleocultural differentiation through prehistory (Sackett 1973, 1977, 1982, 1986; Jelinek 1976). The idea that types of stone tools in patterned combinations may be diagnostic as ethnic markers has been questionably applied to assemblages as early as the Acheu1ean of East Africa (Kleindienst 1961; Leakey 1971, 1975; Isaac 1977). Similarly, patterning in other artifact classes has also been investigated to promote an understanding of the social mechanisms involved in transmitting a range of craft traditions (Deetz 1965; Longacre 1964: David 1973). It is necessary, initially, to define some basic ideas relevant to this discussion of style. First, this study accepts a definition of culture as the means by which people solve adaptive problems imposed upon them by the social and physical environment (after White 1949 and Flannery 1972).

Archaeological tool ass~mb1ages as material remains reflecting the technological solution of adaptive problems in a cultural system, in turn reflect human behavioral patterns (after Binford 1962). These behavioral patterns are controlled by decision making processes which influence stylistic variability. Sackett (1977) implies this when he states that style is the result of choices among functional 29 alternatives. Applying this idea to East African assemblages, Isaac (1977: 212) suggested that "a variety of sharp stones could fulfill equally well the basic requirements of nonagricultural humans," and as a result we might expect a fair degree of variation in both artifact morphology and assemblage composition. What role, then, does resource availability and quality have in creating the patterning of variation that is archaeo10gica11y visible? In a study of stylistic variation in Epipa1eo1ithic capsian industries in North Africa, Sheppard (1987) outlines a model of hunter-gatherer social interaction as related to resource abundance and predictability. It is useful to briefly examine his model in application to this study. His work proposes that under conditions of low resource predictability and abundance, social groups are highly mobile and style appears heterogeneous within sites, resulting in a low level of regional similarity without spatial discontinuity. In regions with high resource predictability and abundance, social group mobility decreases and style becomes homogeneous within site and within region. Wobst (1977) similarly suggests that under conditions of increased sedentarism, style might signal ethnic boundary maintenance. In late Upper Pleistocene North China and the study 30 area this dissertation is concerned with, lithic resources are not only scarce but they are of relatively poor quality. The quality of the raw material has a profound effect on the morphological variability of artifacts because the number of attributes free to vary is seriously constrained by material limitations. Further, as Clark (1989) cautions, the lithics recovered in Upper Paleolithic contexts are often only the durable parts of multicomponent tools originally mounted in non-preservable hafts, handles or foreshafts of wood or bone. It is likely that these perishable parts of hunting weapons or implements might have contained some stylistic features meant to signal information about personal and social identity (after Wiessner 1983). Close's (1978) concept of style as a microtradition learned within restricted spatial, social and temporal context and transmitted from one generation to the next is compatible with the way the broader concept of tradition is used earlier in this chapter. Most important however, is her recognition of the technological causes of variability. In this study, variations in patterns of retouch are characteristic within each site. It is argued that these patterns are most constrained by the size and lithology of the raw material. They reflect an intensity of use and reuse and are not governed by the same decision making 31 processes involved in stylistic choices (sensu Wiessner and Sackett) because restrictions imposed by the raw materials take precedence. It is important to keep in mind the relationship that the remains of lithic assemblages (accessable to the archaeologist) may have had to the full cultural system in which they were manufactured, as Jelinek (1977) has suggested. This study addresses this issue and maintains that it is the integrated system of prehistoric technologies of stone, bone and wood that was critical to the cultural elaboration of late Upper Pleistocene North China. composition and Methodoloqy of this study

The sites of Salawusu and Shiyu are suited to a comparative study for the reasons discussed below. Radiocarbon dates for these sites have been determined and both fall within the Upper Pleistocene. Salawusu is dated at 35,340 ± 1900 BP (Huang and Wei 1981). The Shiyu materials have yielded two radiocarbon dates; 28,945 ± 1370 and 28,135 ± 1330 BP (Chen and Olsen 1990; Aigner 1981). All these radiocarbon dates are on animal bone and their acknowledged limitations are discussed within the framework of the site stratigraphy.

. - .. ---.. ' -- 32 There are faunal and lithic components at both of these sites that can be compared, and it is also possible to reconstruct the paleoecological setting. Both sites have yielded human fossils in association with the archaeological remains. The Sa1awusu fossils have been interpreted as "possessing some advanc,ed characteristics of modern humans on the one hand while retaining some archaic characters on the other hand" (Dong et a1. 1981: 1193). The hominid fossil recovered at Shiyu is an occipital fragment identified as Homo sapiens (Jia et ale 1972). The interpretation of the stratigraphy and chronology of the Salawusu site is particularly critical to the interpretation of materials from the 1980 excavation presented in this study. The lithic assemblages are morphologically distinct at both sites with characteristic patterns of utilization and retouch. Technologically, there are some similarities as well as important differences. Both are flake industries where small irregular flakes produced mostly by simple direct percussion dominate. However, other flaking techniques such as bipolar flaking and particular retouch patterns are important to the interpretation of these sites as well. The emphasis of the lithic analyses is on identifying reduction strategies and stages in the 33 manufacturing sequence. The intensive and extensive use of raw materials is documented. The interpretive emphasis is on a reconstruction of the organization of the prehistoric technologies and the role that resource type and availability played in this organization. The numbers of specimens (both lithic and faunal) which constitute the archaeological record at these sites is limited; therefore, the utility in structuring this study to incorporate additional interdisciplinary data facilitates the interpretation within a paleoecological framework. The following chapter examines the physical properties of the raw materials that were available for fashioning prehistoric tool kits. Reduction strategies, projectile technology and the organizational aspects of prehistoric technologies are also discussed in Chapter 2. The Quaternary paleoenvironment of North China is the subject of Chapter 3. Regional geomorphology, the pollen record and biostratigraphy are discussed and the loess­ paleosol sequence is correlated with the Oxygen Isotope chronology. Studies of the Salawusu and Shiyu sites constitute Chapters 4 and 5, repectively. Each site chapter contains a history of excavation, site stratigraphy and the lithic analysis; data and interpretations. Chapter 5 also includes

------_.. -_. 34 a comparative analysis of the two lithic assemblages. The final chapter offers the conclusions of this research and prospects for further studies in Chinese Paleolithic archaeology.

ChroDoloqica1 Terms

This study will employ geochronological terms (Lower, Middle and Upper Pleist,ocene) for the categorization of archaeological assemblages in North China in order to avoid the problematical use of European-derived concepts such as Early, Middle and Late Paleolithic. In the decades that have passed since Movius' (1944, 1949) pioneering attempts to arrive at a large-scale synthesis of East Asian prehistory, it has become clear that the application of traditional Paleolithic subdivisions in China may be misleading (Clark and Yi 1983; Pope 1988, 1989; Watanabe 1985). For example, labelling every early Upper Pleistocene industry in China as "Middle Paleolithic" does little to enhance our understanding of these assemblages since in most cases there is little or no technological congruence between industries of this age in China and the western Old World to warrant such a blanket statement (Olsen and Miller-Antonio, in prep.) 35 The chronological divisions adopted for this work consider the Lower Pleistocene to include the period from roughly two million years ago to the Matuyama [R] chron/Brunhes [N] chron paleomagnetic boundary at approximately 730,000 years BP. The Middle Pleistocene is defined as that period from the beginning of the Brunhes chron to the Oxygen isotopic stage 5 - stage 6 boundary at 128,000 ± 3,000 years. Finally, the Upper Pleistocene includes the most recent glacial cycle of the Ice Age from roughly 128,000 to 10,000 years before present (Oxygen Isotope stages 5 - 2).

Summary

Quaternary global climatic fluctuations have been the focus of a growing number of multidisciplinary studies over the last two decades (Whyte 1984; Martin and Klein 1984; Briggs and waters 1977). The evidence suggests a continuous pattern of glacial/interglacial cycles that dominated the Pleistocene climatic regime (Lowe and Walker 1984). Even the tropics, long thought to have been unaffected by Quaternary climate change, experienced environmental oscillations as evidenced by remnant sand dunes in the Pleistocene stratigraphy of these regions (Thomas 1989;

--_._------36 Roberts 1984). Ability to adapt to environmental stress by Pleistocene hunter-gatherers must have been a prime criterion for survival. Fluctuating climatic regimes may necessitate human adaptability but, as Conkey (1987: 66) points out, "environmental stress doesn't operate on an organizational vacuum ••• it is the state of a given organizational system and its past history that provide the direction (of culture change)." Because of the demands that environmental stresses imposed on the social systems of mobile hunter-gatherers in late upper Pleistocene North China, it is proposed that directional culture change through time may be viewed more as "cultural flexibility" rather than as a unidirectional increase in complexity. This flexibility is in part reflected in the organization of prehistoric technologies and this organization is in turn linked to raw material type and availability. The interactIon of prehistoric technological systems using stone, bone and wood as raw material was critical to the cultural elaboration of late Upper Pleistocene North China. 37

o

1. Salawusu, Inner Mongolia 2. Shiyu, Shanxi 3. Dingcun, Shanxi 4. Rehe, Shanxi 5. Xujiayao, Shanxi 6. Xiachuan, Shanxi 7. Xueguan, Shanxi 8. Shuidonggou, Ningxia 9. Donggutuo, Hebei 10. Xiaochangliang, Hebei 11. xiaonanhai, Henan 12. Zhoukoudian, Locality 1 and 15

Figure 1. Archaeological sites mentioned in the text. 38

HOLOCENE . Neolithic/Microlithic Polished 10015 E'mookou Hulouliang PLEISTOCENE Xiaananhai (11,000-9000) (13,000-9000) I Xiachuan I ( 23,000-16,000) Shiyu I . 1".2,00) Sh7\"00

Sjara-osso-gol (Salawusu) I \ 150,30,000) -"// \'~?9~":a.oOO)

small tool Xujiayaa _...... /large tool ' tl.' on (ca.l00,000) tradl. tradition SanmenKla

2"~"'~ ",,,n., 1 1? 4j"XX» 170~000\Oo,ooo) ",~ =.,)

Dongguluo / ( ca. 1,000,000) ' ...... ? .... Xlhoudu (ca.l,aOO,OOO)

Figure 2. A current model for the development of Paleolithic traditions in North China (from Jia and Huang 1985: 262). Dates are in years B.P.

------.. - . 39 Chapter 2 PREHISTORIC TECHNOLOGIES

The technological nature of the North Chinese Paleolithic must address the issue of raw material use and constraints. Stone, because of its superior preservability, is overrepresented in the archaeological record as an indicator of the technological aspects of prehistoric human behavioral systems. Animal bone is also an important component of the archaeological assemblage at the sites of Shiyu and Salawusu although it is subject to diagenetic processes that complicate its interpretation in archaeological context. Several scholars, including Boriskovsky (1966), Hutterer (1977), Lian (1981), Pope (1988,1989) and Clark and Schick (1988), have suggested that bamboo and hardwoods may have served as important resources upon which non-lithic aspects of technologies were based. These versatile materials could have furnished not only flaked tools but also cooking and food storage containers, spears, traps and rope (Olsen and Miller-Antonio, in prep). Some authors (e.g. Hutterer 1985; Pope 1989) suggest that the objective of the Asian chopper/chopping tool tradition (with an antiquity of at least 1,000,000 years) was principally to 40 produce stone flakes used to manufacture and maintain non-lithic tool kits. The physical properties of these raw materials and the organization of these technologies are explored in this chapter and a summary of the hypothetical interaction of prehistoric technologies based on the use of available raw materials in Upper Pleistocene North China is presented.

Lithic Technology

Raw Material for Lithics in North China

The knapping process, as with all skills requiring hand and eye coordination, is a series of decisions. Beginning with selection of the initial piece of raw material, the flaking process proceeds essentially with a decision at every strike based on the outcome of the previous step and the form of the implement desired. The physical properties of the raw material naturally constrain these steps because of their effect on the range of possible artifact forms (Toth 1985). Problems that may arise from imperfections in the raw material must be accomodated or the implement may be abandoned in its incomplete or broken state. Rocks with dense bomogeneous fine-grained structure 41 such as basalts, flints and cryptocrystalline cherts behave in a predictable and relatively consistent way when they are fractured by the skilled knapper (Purdy 1973). When a sharp blade was required, the chosen material might be obsidian which breaks with conchoidal fractures leaving extremely thin keen cutting edges. These edges may be so thin that it is necessary to thicken them with retouch to increase their durability (Inskeep 1988). A chopping tool, subject to repeated impact force might preferentially be made on tougher, more granular rocks. There are differences relating to the size of available stone nodules and distance from the source area in the lithic raw material at each site in this study. These are discussed in detail in the chapters that present the sites. The pervasive scarcity of ideal stone and the lithology of available material however, is similar through the study area. There are several varieties of quartzite, vein quartz and quartz pebbles. Chert and limestone are also present in limited quantities. The flaking properties of these materials are discussed below. Quartz, a silcon dioxide mineral, may occur in small pebbles or massive chunks large enough to fashion handaxes. However, it tends to fracture along structural planes that make it difficult for the craftsman to control for the size

-- - _.- ... __ ...._------42 and shape of flakes detached in the knapping process (Toth 1985). On coarse-grained quartzite flakes, it is often difficult to see the features of conchoidal fractures that are associated with flint. Ripples, which are pronounced on flaked vitreous material such as obsidian, are not clear on coarse-grained quartzite. This makes it difficult to determine the direction of flake removal. Bulbs of percussion are often shallow, making flake scars difficult to distinguish especially in specimens which have been abraded or weathered in any way. Chert, a cryptocrystalline form of quartz, occurs in many varieties and as such can have a wide range of flaking properties. It can occur as large tabular blocks with flaking properties very similar to flint. Other forms may occur as nodules or small pebbles and flaking properties are constrained by their small size. Some types of chert are considered to be amorphous. These forms flake isotropically (having the same tensile properties in all directions) making it a high quality desirable raw material because it fractures so predictably (Purdy 1973). Lithic analysts and current day knappers have suggested that quartz and coarse-grained quartzites are intractable and undesirable tool materials (e.g. Maloney et ale 1988).

------43 However, there is archaeological evidence from several geographic regions that toolmakers with access to inferior materials still made adequate tools (MacRae and Moloney 1988; Clark 1954; Leakey 1971; Toth 1985; Villa 1983). An examination of the nature of some African assemblages with particular emphasis on raw material is useful for an understanding of this variable in tool production as it applies to North China, since there are many similarities in raw material use and the nature of lithic artifacts between these two regions.

African Technology

Sandstones, clays and gravels are widespread throughout African Cretaceous deposits. Limestones from Tertiary layers in the area of the Nile Valley and the Sahara were an important source of chert in nodular or tabular form and chalcedony, quartz and indurated shale were locally important to stone age industries around Zambia and Zimbabwe (Inskeep 1988; Clark 1970). Singer and Wymer (1982) report that quartzite is the dominant raw material in the Middle Stone Age industries at Klasies River Mouth in South Africa and Clark's (1954) monograph on the prehistory of the Horn of Africa describes 44 artifacts fashioned on basalt, chert, silcrete, sandstone, shale, crystal quartz, limestone and quartzite. Many of these rock types, particularly volcanics and quartzite, were available as lithic raw materials in the form of cobbles in river and stream beds. The technique of reducing these cobbles to finished implements results in morphologically similar artifacts from very diverse regions of the world. Jones (1982) compared implements fashioned on quartzite cobbles from Berinsfield, Oxfordshire to Early stone Age implements from Olduvai Gorge. He found that the size and form of these cobbles exerts an overriding influence on the morphology of the implements. The technique of detaching very large flakes from massive quartzite or igneous boulders was another method of working intractable stone to produce large flake blanks. These flakes are well represented at both olorgesailie and Olduvai Gorge and were subject to a minimum of secondary retouch to form bifacial implements (Isaac 1977; Leakey 1971). This suggests that the massive nature of African raw materials underlies the presence of bifaces and cleavers. While large unifacial and bifacial implements such as handaxes, picks, cleavers and choppers are the hallmark of the Acheulean, scrapers and other small flake tools are numerically the most important category at olorgesailie

.-.~~-~-.. - -_. -~~ .. -.------45 (Isaac 1977). Most of these tools are made of basalt and exhibit relatively little standardization. Some scholars (Clark and Schick 1988; Movius 1949) have suggested that the North Chinese Paleolithic is characterized by a similar casual nature . In an experimental study designed to identify some of the causal factors of variability in Oldowan lithic industries, Toth (1985) suggested that the availability of larger pieces of quartz and quartzite at Olduvai allows for the presence of spheroids and subspheroids at that site, while the absence of this artifact type in the Oldowan assemblage at Koobi Fora is attributed to a raw material difference. Presumably, the basalt which is the the principal raw material at Koobi Fora is not suitable for fashioning spheroids. Toth attributed most of the variation in the Oldowan to "waste" forms or end products that result from a reduction sequence designed to produce sharp flakes. Mary Leakey (1971) reported on the recovery of flaked mammalian bone from Beds I and II at Olduvai. She suggested that the massive limb bones of large animals such as elephant, giraffe or Libytherium were employed as a SUbstitute for stone in tool-making. Similar usage of long bones of Coelodonta antiquitatis at Salawusu will be discussed.

------46 The use of an alternate raw material to produce sharp flakes that could be utilized in the same way as stone flakes but that was more susceptible to diagenetic processes can profoundly affect assemblage composition. There is a great chronological disparity between the African Oldowan and the late Upper Pleistocene assemblages of North China and two different species of hominids are responsible for each of these industries. Nevertheless, this discussion is meant to illustrate that raw materials impose characteristic limitations on artifact forms and hence influence assemblage variability. A correlate of this is that the nature of the raw material places constraints on the organization of lithic technology within a given cultural system.

The Organization of Lithic Technology

Paleolithic archaeologists have long recognized that the variables and attributes of lithic artifacts may assume technical, stylistic and/or functional meaning (Jelinek 1976). These three components are organized by a set of principles providing the decision-making mechanism for selecting specific patterns of manufacture, use, maintenance and recycling. stylistic/functional studies were emphasized 47 in mid-century lithic analyses but for the past 15 years, the emphasis has been on the technological component. Diverse lithic studies concentrating on the technical aspects of assemblage formation have ranged from reduction strategies and debitage analysis (Sullivan and Rozen 1985; Dibble 1987, 1988) to tool design parameters (Shott 1989: Bleed 1986) to contributions from ethnoarchaeological studies designed to link sUbsistence strategies and settlement patterns with stone tool manufacture, maintenance and discard (Binford 1977: Pokotylo and Hanks 1989). When Binford (1977, 1979) suggested that lithic technologies may be organized along a continuum ranging from curated to expedient, it became apparent that the notion of curation was imbued with a complexity all its own because of the range of behavioral implications it carries. For example, if curation implies a degree or kind of foresight applied to tool production and conservation, what is its relation to resource availability and can it be used to distinguish between Middle Paleolithic and Upper Paleolithic industries? Marks' (1988) Negev study suggests that curation of chipped stone tools was not an important part of either period's technological system because his study area has abundant flint sources. He maintains that curated tools are 48 more apt to be important and identifiable in the technological system where flint sources are lacking or widely separated. He further links the behavior with resource availability by suggesting a "curational flexibility" in flint-rich areas that would favor the curation of bone tools, wooden hafts and spear shafts rather than chipped stone (Marks 1988: 284) regardless of chronology. Identifying lithic attributes assignable as indicators of curative behavior is also problematical. Toth (1985) maintains that the hominids responsible for the Oldowan industries exhibited simple curation behavior by transport of raw material around the landscape for future use. Curation as defined by Binford (1977,1979) implies a complex range of behaviors. He states that curated technologies consist of tools manufactured in anticipation of use, that are useful in a variety of tasks, maintained through a number of uses, transported from location to location and recycled to other tasks when no longer functional for their original purpose. He concludes that curation should produce assemblages that are technologically sophisticated and formally distinct while expediency should produce assemblages that are technologically simpler and formally less patterned as a result of tool manufacture in 49 immediate response to a task at hand. In a more quantitative approach, Shott (1989: 24) states that curation is "the realized utility of a tool and curation rate can be measured as the ratio of realized to potential utility." There is no agreement among researchers on what aspects of a full prehistoric cultural system constrain the lithic technology most. Binford suggests that settlement organization places the most fundamental constraints on the lithic technology. Bamforth (1986) argues instead that raw material availability is the primary factor in determining the technological characteristics of tool kits. Torrence (1983) suggests that time limitations on the activities for which tools are used is the constraining factor. The assemblages at Salawusu and Shiyu show intense use, maintenance and recycling of lithic material, however none of the above models adequately apply to these curated technologies. What they do not account for is the combination of expedient and curative characteristics that can be present in assemblages where two or more technologies based on different raw materials with varying availability, are equally important to sUbsistence strategies. They do not consider the technological interdependence that must logically exist. A stone point meant to function as a

------~~~- =~----~~---- - 50 projectile is useless (for this task at least) without a wooden or bone haft. The lithic industry associated with a bone technology dominated by the production of simple bone flakes may exhibit the curative characteristics of recycling and maintenance due to raw material scarcity and/or inferior flaking properties. In direct conflict with Binford's (1979) conclusion however, it will not necessarily appear technologically sophisticated or formally distinct when viewed on its own. Finally, from a purely technical view, every lithic assemblage may be considered "curated" if characterized in terms of a lithic reduction sequence where tools go through several stages of modification and use (cf. Jelinek's (1976) "Frison effect"; Dibble 1987). Changes in artifact morphology naturally accompany this reduction process at eac:h stage, but the degree of change is variable. Maintenance, discard and replacement occur at different rates depending on raw material availability. The contemporary use of "curation" implies a degree of foresight which results in the transport of useful material around the prehistoric landscape. Conservative flaking technology or intensity of utilization may be manifestations of this behavior interpreted from the archaeological record. 51 Any attempt to explain the variability or understand the organization of technology in the late Upper Pleistocene small tool assemblages of North China should ideally consider the contribution to the whole tool kit from bone and wood based technologies as well as stone and the reduction strategies applied to these materials. Unfortunately, the archaeological record is a reflection of only a very small part of the full prehistoric cultural system.

Lithic Reduction strategies and Projectile Technology

In a recent article, offering a new synthesis of variability in the French Middle Paleolithic, researchers suggest that raw material constraints and the intensity of artifact reduction constitute "more basic and observable factors of variability than function and style." (Rolland and Dibble 1990: 480). In assemblages where lithic raw material is scarce, it is expected that the degree of reduction intensity would be greater than in assemblages where raw material is plentiful. In situations of raw material stress, what kinds of reduction strategies would best conserve resources and how would these strategies be reflected in archaeological assemblages? 52 Bipolar flaking is a useful technique for working small materials and poor quality stone although if offers little control over the morphology of the flake produced. It is a technique in which a core is supported vertically on an anvil and struck at the opposite end with a hammer so that flakes are simultaneously removed from both ends of the core (Clark 1954). Flakes detached in this manner show various degrees of crushing at opposite ends and they usually have a straight interior surface. Studies which emphasize the utility of debitage analysis in reconstructing manufacturing stages have provided some insight on assemblage formation, group mobility and site occupation (Sullivan and Rozen 1985: Kelly 1988). As a result of reduction experiments focusing on debitage variability, Magne (1989) suggests several variables he finds most useful in determining early vs. late manufacturing stages. Among these are platform presence or absence: dorsal scar count: platform scar count: cortex presence or absence and presence of bipolar reduction techniques. Magne (1989) presents a hypothetical model on assemblage formation employing debitage/tool ratio and the proportion of late stage debitage (Figure 3). He tested this model by plotting the data from seven Paleo indian assemblages.

------.- -_. ------53 The 1980 Salawusu lithic assemblage consists of a high percentage of debitage that was analyzed by an analogous approach, i.e. a choice of several variables to determine the proportions of early and late stage debitage in the assemblage. This percentage was then plotted against the debitage/tool ratio (see Figure 3) to illustrate different technological strategies. Debitage from the Shiyu assemblage was not available to analyze in the same way. These studies which emphasize the importance of debitage as an analytical class are relatively recent and it is expected that their impact on lithic analyses will increase. Paleolithic archaeologists in China today are also well aware of the need to recover debitage in order to accurately assess the full range of manufacturing stages represented in a lithic assemblage (Huang, pers. commun.). Intensely reduced assemblages imply a high proportion of multiple-use and reused tools. This makes typological assignments particularly difficult and can lead to a misrepresentation of artifact classes and their relative proportions in an assemblage. A study by Odell (1988) addressing prehistoric practices through lithic analysis concluded that the projectile point type was only one of several kinds of lithic artifacts employed in antiquity to tip projectiles. 54 This study has implications for the casual flake and blade dominated industries of North China and in particular, the Shiyu lithic assemblage. Both artifact classes of flakes and blades from his Illinois Valley data set (spanning the late Archaic, Woodland and Mississippian periods) appear to have been modified to produce a point at one end and a blunting or notching of the edge to receive fastening devices. To assess impact damage from projectile use, several experimental studies were devised (Odell and Cowan 1986; Odell 1988; Bergman and Newcomer 1983» and impact fractures were compared between the archaeological and experimental pieces. Odell concluded that retouched flakes and debitage of a variety of shapes and sizes were numerically much more significant as functional projectile points in his data than are the actual morphologically "typical" projectile points. While the results of these and related studies have been questioned on the basis of capability to distinguish projectile impact fracture from other percussive activity in archaeological context (Jelinek, pers. commun.; Shea 1988, 1990; Holdaway 1989), it is likely that several materials and morphological designs may be equally suitable to perform the same task or that these effects may result from causes other than use as projectile points.

.- .. ---.-- ._- 55 An examination of flaked bone technology and bone as raw material follows.

Bone Technology

upper Paleolithic humans throughout Eurasia had developed an impressive bone, antler and ivory technology with intricately carved harpoons, spear points and awls, finely crafted bone needles and animal figurines. These rich bone industries were accompanied by lithic industries of tools such as burins, specifically suited to working these materials (Gowlett 1984). Bone, antler and ivory , especially from mammoth and reindeer, were also important raw materials for tools as well as for art objects and architecture in the Late Paleolithic sites of Mezin and Mezhirich in the Ukranian Republic (Gladikh et al. 1984; Soffer 1985) and at Mal'ta in Soviet Siberia (Chard 1974). These sites date to about 15,000 B.P. and the inhabitants were clearly skilled at fashioning bone into utilitarian as well as ornamental objects. Two important cave sites in northeast China, xiaogushan in Liaoning province and Zhoukoudian Upper Cave near Beijing, have yielded evidence of skilled bone and antler

-~------56 technology. Both of these localities date to appoximately 18,000 B.P. and the assemblages contain barbed antler harpoons, bone awls and fine bone needles with eyes (Chen and Olsen 1990). These sophisticated bone industries suggest a familiarity with the physical properties of bone, teeth and antler. Several scholars concerned with research questions involving the arrival of early humans in the New World and the technology that accompanied this colonization process have suggested that bone technology may have been an important component of Beringian and Paleoindian tool kits

(Guthrie 1983, 1990~ Bonnichsen 1979~ Morlan 1980; Frison and Zeimans 1980; West 1981). The type of bone industry these researchers have investigated is not necessarily analogous to the sophisticated Magdelenian of Upper Paleolithic Europe. Rather, it is an industry meant to produce bone flakes and core implements based on flaking techniques such as simple direct percussion (hard and soft hammer) similar to those employed in working stone resources.

Bone and Antler as Raw Material for Flaked Implements

The pioneering work of Breuil (1939) at Zhoukoudian 57 Locality 1 illustrates the antiquity of flaked bone technology. Additional reference to this type of industry, while not common in the literature, can be found in several other volumes such as Semenov (1964) and Freeman (1978). Semenov (1964: 147) suggests that the diaphysis of a long bone which had been split longitudinally could be worked by resting the splintered bone (in this case, mammoth leg bone) on a hard surface and directing blows along the lateral edges from the outside inwards. The bone assemblage from Cueva Morin, a Middle Paleolithic site in Cantabrian spain, consists of 298 bone artifacts made mostly on irregular long bone fragments by flake removal exhibiting conchoidal flake scars similar to those produced by the stone knapping process (Freeman 1978). From these reports it is clear that the long bones of large mammals are the overwhelming choice for this type of flaking technology. The reason for this has to do with the structure of bone itself. Non-fossilized bone consists of two fractions, organic and inorganic. Hydroxyapatite crystals, a brittle mineral, constitutes the inorganic phase and a fibrous protein, collagen, composed of long-chain amino acids, forms the organic phase (Runia 1987). As a composite, these materials form a strong matrix giving bone its supportive qualities. 58 other structural features such as osteon size and number, cement lines and the presence of Haversian canals affect the stiffness of bone (Katz 1980). The mineral content of bone is about 10% higher than antler which accounts for its greater rigidity and higher density. Long bone shafts (diaphyses) have a thick layer of compact, dense cortical bone. This quality makes it easier to fracture and chip. The resilient properties of antler make it less subject to fracture and it is less able to hold a sharp edge (Guthrie 1983). These respective properties of bone and antler have interesting functional implications for assemblages where both raw materials are available for utilization. Bone flakes and flaked core implements would be suitable for cutting and scraping tasks while antler is more useful fashioned into projectiles, where some degree of flexibility or resistance to breakage is desirable. Guthrie (1983) has shown, through a comparative experimental study, that antler does in fact make a superior projectile when compared to those he constructed of wood and long bone. This does not exclude the use of long bone for fashioning projectiles, however. Frison and Zeimens (1980) have reported the recovery of bone projectiles made on bison femur from a Folsom site in Wyoming. 59 While both bone and antler are more easily worked when wet, the supple quality of antler is dramatically increased by soaking and the working shape it is given while wet will be retained when dry (Semenov 1964; Bonnichsen 1979). Experiments by Morlan (1980) and Bonnichsen (1979) have shown that green bone is more suitable for flaking and that fossilized bone will rarely exhibit conchoidal fracturing, which makes it difficult to distinguish humanly flaked fossil specimens from those altered by other agencies.

Taphonomic Considerations

After deposition, bones may be altered by many agents which produce not only flake scars but marks that are difficult to distinguish from humanly produced cut marks (Behrensmeyer and Hill 1980; Shipman 1986). Recent reassessments of the faunal assemblage at Zhoukoudian Locality 1 have emphasized the problematic issues in the interpretation of hominid activity and the accumulation of faunal remains at archaeological sites (see Binford and Ho 1985; Binford and stone 1986, and especially the comments by other authors that follow these articles). Some of the most important natural modifying agents as they apply to this study will be discussed here.

----_... _- _._. --.-~.~~-~=.-~.. _. -~.-~-~~- .. ~--.. _-- 60 The rapid movement of bone and stone by water in a river or stream has been shown to detach flakes along the borders of fragmentary bones. The pattern of this flaking is usually one of small scars on opposite margins of the bone as if it were rotating around a central axis (Morlan 1980) . The chipping and splintering of bones by carnivores is a critical factor in the interpretation of many faunal assemblages. Trampling by large animals can cause spiral fracturing and flake scars on bone fragments (Myers et al. 1980). Large flakes can be detached from the edge of a broken bone when a carnivore hooks its canines or carnassial teeth over the edge and uses the opposite jaw to pry against the edge. The teeth leave other contact marks on the bone sometimes characterized by rotational scarring or punctures. In Binford's (1981: 55) opinion, the bone artifacts from Cueva Morin mentioned previously are the result of this kind of carnivore activity. Freeman (1983) agrees that some of the bone accumulation and modifications at Cueva Morin are the result of carnivore activity, however, he maintains that humans were the principle bone accumulators. He maintains that the types of animal body parts (cranial and dental), the large size of the fragments and the selection of bones from particular species are not suggestive of hyena bone 61 accumulations. Additional supportive evidence of his interpretation of the specimens as bone artifacts is contextual. Much of the bone is in association with other evidence of human activities such as stone tools and piles of stones. weathering processes cause exfoliation of bone cortex and the formation of splinters which might be confused with humanly produced fragments. Wind blown sand and water abrasion can create highly polished surfaces which might be misinterpreted as artifact utilization (Miller 1973). The criteria applied to the classification of flaked bone artifacts at Salawusu were conservative. Any bones with evidence of carnivore activity were excluded, even if at a distance from the carnivore marks, they exhibited flake scars that might fit the classification of cultural fracture. In agreement with Morlan's (1980) conclusion, it is apparent that the orientation of flaking is an important criterion for distinguishing these specimens as artifacts.

Wood Technology

Wood does not preserve under most circumstances and its use is not well documented in the Pleistocene archaeological record but many prehistorians have acknowledged its probable 62 place as a fundamental tool in hominid evolution and suggest that it may pre-date the use of stone as a resource for tool making. The oldest wooden implement to date was recovered from peat deposits at Clacton, England. It is a spear-like specimen with a tapered point which seems to have been , trimmed by a stone implement (Gowlett 1984). Clark (1970) has reported wooden artifacts from Kalambo Falls in Africa. These specimens, preserved in a waterlogged environment which inhibits the action of destructive bacteria, include sharpened sticks and a wooden club. Clark suggests that one successful method of working wood was a charring and scraping process. Very hard wood can also function as a soft hammer for percussion flaking. Pope (1989), following an idea proposed by Boriskovsky (1966), correlates the geographic distribution of bamboo with that of the Asian chopper/chopping tool tradition and suggests that this giant grass was in fact the most important technological resource ~or early hominids. The physical properties of the stalk make it possible to fashion cooking and storage containers, heavy and light projectile points, traps, spears and living structures. Its rapid growth rate makes it a predictable resource and the seeds and shoots of many species are edible. It also possesses a

------63 degree of flexibility making it possible to bend into a bow shape. Other versatile resources such as liana, rattan and teak are also found in tropical Asian forests. Northern Chinese deciduous forests could also have furnished a variety of useful resources for tool making. Guthrie (1983) eva.luated the performance of birch as a projectile with respect to bone and antler and found the wooden point had no cutting ability but it was held tightly by the dilated skin ope.ning of the wound. Although it blunted easily, it could be quickly resharpened. Guthrie also mentions ethnographic sources which illustrate the use of wooden arrows for bird and other small game hunting. In terms of flaking properties, wood is a variable and often difficult medium. It is porous, inhomogeneous and anisotropic (Wilson and White 1986). It's fibrous structure gives it a greater elasticity than the crystalline matrix of rock. The vibrational energy that travels through a material when it is struck must overcome this elastic capacity in order to detach a flake (Bodig and Jayne 1982). Only the most compact, brittle and dense-grained woods are likely to exhibit flaking properties. Harrison (1978:44) suggested that hardwoods and bone were both extensively used to "supplement and elaborate simple stone-cutting tools" through the Paleolithic of 64 Island Southeast Asia. He reported the recovery of a bifacially flaked ironwood handaxe from Borneo (Kalimantan) Cop. cit.). Unfortunately, details on the provenance and age of this artifact remain unpublished.

Summary and Conclusions

A summary diagram of ideas discussed in this chapter which relate to the interaction of prehistoric industries that utilized stone, bone and wood resources is presented in Figure 4. A cultural system constrained to make efficient use of stone resources and expedient use of bone could develop an industry characterized by the use of irregular pointed flakes and blades to tip projectiles. This is one model which may apply to the cultural elaboration of late Upper Pleistocene North China and data from the study area are presented in the next two chapters in support of this. Basic flaking techniques applied to variably available resources results in archaeological assemblages with specific characteristics reflecting the organization of the technology. The degree to which these different raw materials are utilized through time may account for some of the observed

------_._-_._-- 65 variability in these assemblages. Viewed individually, these industries will not necessarily exhibit the traditional features of sophisticated Upper Paleolithic Western Old World assemblages. However, technological interaction exploiting a variety of material resources can be a culturally innovative process.

------_ ... _--_._-- . _._.- - ._- - 66

High

TOOUBLANK MANUFACTURE TOOL MAINTENANCE HIGH EXPORT RATE LOW DISCARD RATE HIGH CONSERVATION

o a:ra ~ REOCCU~EDsrrES ~ I------i HIGH REUSEJSCAVENGING RATE 1--______~ :0 oQ)

TOOUBLANK MANUFACTURE TOOL MAINTENANCE HIGH REJECTION RATE HIGH DISCARD RATE LOW CONSERVATION RATE

SITUATIONAL SITUATIONAL REPAIR REPAIR RAW MATERIAL RAW MATERIAL AVAILABLE SCARCE

Low~ ____~ ______~ ______~ ____~_ Low %Late Stage Debitage High

Figure 3 - A hypothetical formation model employing debitage/tool ratio and late stage debitage proportion, based on reduction experiments focused on debitage variability (from Magne 1989: 20) •

-_. ------67 UPPER PLEISTOCENE Raw Material characteristics: STONE variable flaking predictable plentiful quality and size of nodule Technological I characteristils: curated curated expedient

l.n. t enSl.veI . opportunisticI use and use reuse

smallI artl.facts . consistent variable halfted flakes size size flakes, and blades as shafts polished bone points and points

TERMINAL PLEISTOCENE Epipaleolithic innovations: developed bows microlithic

projectile Technology (bow and arrow)

Figure 4 - A hypothetical summary diagram of the interaction of prehistoric technologies through time based on potentially available raw material.

------~~~~-~ -~------~~~--~-""-~ _. _. ~~-~ ~~~~~.------~ - 68

CHAPTER 3 THE QUATERNARY PALEOENVIRONMENT OF NORTH CHINA

Intro4uction

Chinese scientists have established an overall climatic trend toward desiccation in North China which has accelerated since the Pliocene (Zhao and Xing 1984; Zhang Lansheng 1988). They also acknowledge the pattern of Pleistocene climatic fluctuations and alternating combinations of cold-warm and dry-moist stages. The question is, how fine-grained can the resolution be in reconstructing the Pleistocene paleoenvironment within stratigraphic and chronological bounds. Published data suggest that climatic changes from desert to vegetated land and vice versa are very rapid, spanning no more than a few hundred years. For example, in the modern West Sahara, dune fields became active no more than 300-400 years ago and Portuguese explorers still observed summer rains at 24° N. Today these rains hardly reach 19° N (Sarnthein 1978). Many relatively "short" (in geological time) desert phases could easily remain unrevealed in the terrestrial geological record. However, a study of lake sediments from Qinghai Lake, situated in an 69 intermontane basin at the northeastern corner of the Qinghai-Tibet Plateau, suggests that sedimentation patterns respond with higher resolution to climate change than vegetation (Kelts et al. 1989). This chapter examines appropriate methods of reconstructing paleoenvironment for Pleistocene North China by integrating the faunal and pollen evidence with Oxygen Isotope chronology and climatic change. It begins with a discussion of the regional geomorphology of that area of North China encompassing both Salawusu and Shiyu sites.

Mega-geomorphic Regions and the origins of Aridity in North China

Arid Zones Defined

There are several ways to define the degree of aridity of a region. The designation "Im" (indices of moisture availability) was used in the 1950's in reports for UNESCO. Later in the 1970's, mean annual rainfall was added to the classification of arid zones (see Thomas 1989a: 2, for a brief explanation of this subject). Zhao and Xing (1984: 250) report that a committee on Natural Regionalization composed of scientists from the 70 Academia Sinica chose the "aridity index" which represents the ratio between potential evaporation and precipitation for delimiting natural regions in China. As a result, the arid lands of North China have been classified into 3 natural zones: 1. Semi-arid lands have an aridity index of 1.5 - 2.0 and occur mainly in the steppe zone of eastern Inner Mongolia and the eastern slopes of the Da Hinggan Ling. 2. Peripheral areas of arid lands have an aridity index of 2.0 - 4.0 occur in the semi-desert zone of Ningxia and Central Inner Mongolia. 3. Arid lands have an aridity index of greater than 4.0 and are situated in the desert zones of northwest China, including the Tarim, Junggar and Qaidam Basins, the region between Xinjiang and Gansu and Helan Shan. China's deserts are classified into one of two types: the shamo, which includes sandy deserts (sand seas) and mobile dunes and the gobi, including stony and gravel paved desert surfaces. The shamo and gobi of China have their origin in geologic, atmospheric and biological processes operating since the late Cretaceous and early Tertiary (Zhao and Xing 1984).

------71 The Formation and Expansion of North China's Deserts

North China's deserts are the result of two regional factors: their distance from the ocean, preventing the penetration of rain bearing winds into the center of continents and mountain barriers which enhance this continentality, creating a rain shadow effect. Landforms such as the Qinghai-Tibet, Mongolian, and Loess Plateaus, by virtue of their vast spatial extent are designated mega-geomorphic units (Gardiner and Scoging 1983). The evolution of these landforms is important to an understanding of the paleoenvironment of North China and the Quaternary human occupation of the region. At the close of the Mesozoic and until the early Tertiary, the vast Chinese landmass reflected a tectonically stable environment of low hills and peneplains interspersed with deep inland depositional basins (Liu and Ding 1984; Wang 1984; Xu 1984). The Mongolian peneplain covered the region which later becomes the Mongolian Plateau. The southwest margin of the Tarim Basin in the far northwest was a relict area of the ancient Tethys Sea which had once submerged much of the continent. This quiet period came to an end by the Late Tertiary with the Himalayan Tectonic Orogeny which began the uplift of the Qinghai-Tibet Plateau

------_..... -_.'. --_.. - .. - --- _. - 72 and surrounding areas and forced the extinction of the Asian Tethys Sea (Zhao 1988). The characteristic monsoon climate of East Asia with cold, dry winters and warm, moist summers became established as the Qinghai-Tibet Plateau rose to an average elevation of 3000 m by the early Pleistocene. other mountain ranges including the Kunlun, Tian Shan, Qilian and Altay were tectonically active as well. continuous uplift and faulting resulted in alternating series of high plateaus and mountains and low graben basins, a pattern which prevails in North China today. These basins, isolated by lofty mountain ranges exceeding elevations of 5000 m, became increasingly desiccated. Their floors were covered with thick beds of sands and gravels derived from the weathering of surrounding plateau and mountain regions. These sediments were a rich source of material for shamo and gobi desert formation. The Mongolian Plateau including the Ordos region reached its average elevation of 1000 m with the uplift and warping of the Mongolian peneplain. Both shamo and gobi deserts developed over this plateau by the processes of fluvial and aeolian erosion of quaternary sediments (Zhao and Xing 1984) . In arid regions, the action of rivers as an erosive and 73 depositional agent is often underestimated (Reid and Frostick 1989). Gravels are carried and deposited on gobi surfaces by rivers and floods. Aeolian processes deflate and erode these gravels, transporting the smaller-grained sands away and leaving a pavement of closely packed stones over a layer of fine sediment (Breed et a1. 1989). These aeolian sands become the material of the shamo, dunes and sand seas (ergs). Finer grained silts travel even further and are responsible for forming another mega-geomorphic feature of North China, the Loess Plateau.

The Formation of the Loess Plateau

Aeolian silts, called loess when they accumulate on the land surface, are initially transported by wind. Dust storms carry large volumes of silt-sized material over distances of thousands of kilometers and at heights of 4 km or more (Watson 1989a). Goudie (1983) proposes that the thick loess deposits of eastern Asia derive from expanded Quaternary arid zones situated to the northwest of the Loess Plateau. Therefore they are desert loess deposits as opposed to glacial rock flour. The lack of loess deposits on the fringes of most desert regions is due to two factors. First, desert-derived

------_._---_._-_ .. 74 silt is frequently transported far beyond the immediate desert margins and is consequently well-represented in ocean sediment cores. Second, in vast desert regions, when silt­ falls occur, the silt may become incorporated into pre­ existing sediments such as the fine sediment layer just under the gobi pavement (Yair 1987). During the Pleistocene, strong prevailing northwestern winds across the Ordos Plateau carried fine sands and silt, depositing this material to the southeast and forming a vast Plateau with sediments over 300 m thick in some places (Derbyshire 1983). Most of the 317,600 sq. km. of the Loess Plateau is situated north of 34° N in the southern half of the great bend of the Huang He (See Figure 5). It is bounded by mountains on three sides: the Taihang Shan on the east, the Qinling Shan in the south and the Qilian Shan and Qinghai-Nanshan on the west. From southeast to northwest, the granularity of the sediments goes from fine to coarse (Derbyshire 1983). The middle reaches of the Huang He preserve a complete sequence of loess deposition from Early to Upper Pleistocene. The classic loess sequence distinguishes four loess layers. The Early Pleistocene is represented by the Wucheng loess. The Middle Pleistocene is divided into the lower Lishi loess and the upper Lishi loess (Wang et ale 75 1988). For this study we are most ,concerned with the deposition of the Malan loess during the Upper Pleistocene because these are the sedimentary layers which are associated with the Salawusu and Shiyu cultural material and fossils. The Malan loess period is equal to cycle II of the deep-sea core record, Oxygen Isotope stages 5-2, dating from 128,000 to 10,000 years ago (Liu and Ding 1984). The Malan loess may be divided into the upper "LI" Malan loess and the

lower "LI-2" Malan loess. A greyish-brown paleosol (SL), radiocarbon dated to 24,000 ± 1400 BP is intercalated between these two stratigraphic units at the Niuquanzi section in northern Xinjiang (Wen and Zheng 1988). This date is consistent with an age of 23,000 BP as reported by An and Lu (1984). This simple division of loess/paleosol stratigraphy for the Malan loess is deeply entrenched in the literature but it is being intensively studied by both Chinese and Western scientist. This research emphasis was clearly in evidence at the XIII INQUA Congress in Beijing in August 1991. Numerous Congress participants suggested that the loess/paleosol stratigraphy, as it reflects climatic fluctuations through the Upper Pleistocene, is much more complex than the classic sequence indicates (International union for Quaternary Research, XIII International Congress Abstracts 1991).

..;.:.;--...:::-~~-.---.------.=-;;:-;:-:.-- .----.------" - 76 Correlations of the loess-paleosol sequence with the deep-sea core oxygen-isotope record has enabled scientists to make some interpretations of Pleistocene climate change in North China (Liu et ale 1985; Wang et ale 1988; Wen and Zheng 1988). These interpretations are discussed next.

The Marine Oxygen Isotope Record and Pleistocene Climate

Oxygen isotope ratios (180 : 160 ) of well-preserved marine calcareous fossils are indicative of the temperature of ancient ocean waters. This approach is based on the fact

18 that the difference in 180 : 160 ratios (expressed as 6 0) between calcium carbonate and the water from which it precipitates is a function of temperature (Anderson 1990). The 6180 value increases as the water temperature decreases.

Shackelton and Opdyke (1973) concluded that Pleistocene glacial cycles were the primary cause of changes in oceanic isotopic composition. The relatively warm interglacials, assigned as odd-numbered stages on the oxygen isotope curve, have lower 6180 values, while the even-numbered glacial stages have higher 6180 values. Also, oceans are enriched in

180 during glacial epochs relative to nonglacial epochs because H2160 is preferentially stored in polar icecaps and continental icesheets (Anderson 1990). The variations in 77

the ratio of 180 : 160 provide an index of ice volume and therefore, sea level fluctuations.

The Loess-Paleosol Sequence Correlated with the Oxyqen Isotope Curve

Imbrie et al. (1984) present a revised Middle-Late Pleistocene 0180 chronology which correlates oscillations in five 0180 records from pelagic foraminifera over the past 780,000 years. This chronology is used here. It sets the Brunhes-Matuyama boundary at 734,000 ± 5,000 BP and the stage 5 - stage 6 boundary at 128,000 ± 3,000 BP. Paleosols intercalated in the Pleistocene loess provide a record of paleoclimatic fluctuations. The Middle and Early Pleistocene loess contain 10-20 paleosol beds distributed extensively throughout the Loess Plateau. These are designated as red-brown cinnamon soils (Wang et al. 1988). While loess is the product of pedogenesis under cold-dry conditions, the cinnamon soils are formed in relatively warmer and wetter periods under forest-steppe conditions. So, the alternation of loess-paleosol layers reflect fluctuating climatic conditions of dry to humid and cold to warm within glacial and interglacial stages. It is important to realize however, that the differences in

------78 paleosols depend heavily on local and regional variations in pedogenesis as well. There are several well-preserved loess/paleosol sequences in the region around the middle reaches of the Huang He. One such sequence is as follows. At the top of the Upper Pleistocene Malan loess there is a greyish-black paleosol, So (called black loam in the area around the middle reaches of the Huang He). Black loam forms under steppe and forest-steppe conditions when the climate is slightly moist. The greyish-brown paleosol (SL, as in the Niuquanzi section) associated with Ll and Ll-2 Malan loess layers is present only in some stratigraphic sections, mostly in the south-eastern Loess Plateau. Cinnamon soils are considered well-developed compared to greyish-brown or black paleosols and they often contain claey layers (Wang et al. 1988). The fact that the Holocene and late Upper Pleistocene loess layers have less extensive and less developed paleosols (one or two of the black loam or greyish-brown type) as compared to the Middle and Early Pleistocene loess (10-20 paleosol beds of the cinnamon soil type) supports two conclusions. First, that the degree of development of paleosols increases with age and second, that the overall climatic trend in North China has been toward increasing 79 aridity. Examining a "typical" loess-paleosol sequence covering Middle to Upper Pleistocene stratigraphy as in Figure 6, the following correlations between paleosol development, oxygen isotope stages and climatic conditions may be made [Figure

7, (Wen and Zheng 1988)]. The pedogenic period of black loam (So) can be correlated with the warm and moist 0180

Stage 1; Ll and LI-2 with cool and dry glacial Sages 2 and 4 repectively; greyish-brown paleosol SL with the slightly warmer and moister Stage 3; cinnamon paleosol SI with wetter and warmer Stage 5 and ~, S2, LJ, S3, and Lt with Stages 6 through 10 respectively.

The Pollen Record and Pleistocene Climate

Pleistocene climatic fluctuations are also reflected in the pollen record. Pollen grains are formed in the anthers of seed-producing plants (angiosperms and gymnosperms). Lower plants such as ferns and mosses produce spores. Both pollen and spores are extremely small (25 - 80 ~m) and are dispersed in large numbers into sediments for plant propagation. They are able to fossilize because the outer layer or exine of the grain is composed of a resistant waxy coating called sporopollenin which functions to protect the 80 grain from desiccation or microbial attack (Lowe and Walker 1984). The shape and size of the grain form the basis for pollen and spore identification. It is most important to keep in mind some limitations of palynological analyses. First, pollen data are not necessarily synchronic with geologic data. Davis (1976) has shown that plant migration occurs at varied rates for different species. Therefore, the. absence of pollen for some taxa does not necessarily mean that unfavorable growth conditions existed at that point in time. Second, not all plants produce the same amount of pollen. Those that produce less will therefore be under-represented in the fossil record. Third, it is necessary to know something about the depositional history and the mechanisms involved in transport of the pollen. Pollen analysis from lacustrine environments is particularly problematical due to input from streams and groundwater. Disturbance of sediments on the lake floor either by currents or burrowing organisms can also complicate the fossil record because pollen may be remobilised and incorporated into later sediments. Aeolian pollen can be transported great distances and this may bias interpretation of pollen assemblages particularly in environments where local pollen production is low (Birks and Birks 1980). 81 Finally, there is the overriding question of whether plant communities which characterized the glacial and interglacials of the Quaternary have modern analogues which can be extrapolated back into the past (See Birks and Birks 1980:1-33 for a discussion of the application of Uniformitarianism to paleoecological reconstruction).

Indicator Species and the Comparative Approach

Most pollen studies of North Chinese Pleistocene vegetation use a combination of these two interpretive techniques. The first is based on the identification of taxa with a well-defined narrow ecological tolerance which may be used as indicator species. The second compares fossil pollen spectra with modern spectra from known vegetation types. Birks et ale (1975) report the application of a range of numerical methods including a principal components analysis and minimum-variance, to regional data from the western Canadian Interior. The results suggest that pollen assemblages correspond closely in area to the vegetation types. They were able to separate the following environments on the basis of percentage of total pollen: dwarf-shrub tundra, forest-tundra, conifer forest, mixed 82 forest, deciduous forest, and grassland. Figure 8 shows a summary pollen diagram for these main vegetation types.

North China Pleistocene Pollen Assemblages and Paleoclimate

A wide variety of sporo-pollens has been recovered from the northern China loess/paleosol deposits (Wang et al. 1988). Palynological studies show that the loess layers are dominated by pollen from shrub and herbaceous plants with only minor amounts of woody plant pollens such as Pinus, Betula, Quercus, and Ulmus present. Artemisia, Chenopodiaceae and Gramineae in large amounts indicate a steppe/grassland vegetation during the time of loess deposition. The paleosol layers on the other hand, contain abundant shrub and herbaceous pollen as well as increased amounts of pollen from woody plants and thermophilic broadleaf trees. This suite of vegetation indicates deposition of the paleosols under the slightly warmer moister conditions of a mixed forest-prairie environment (Wen and Zheng 1988). In some loess/paleosol sections which preserve a long temporal sequence of deposition through the Quaternary, such as those from Shaanxi and Qinghai provinces, an overall trend toward desiccation is indicated from the bottom to the 83 top of the section (Zhang Lansheng 1988: Jiang 1988). For example, Wang et a1. (1988) report finding BbY§ and Broussonetia in the Early and Middle Pleistocene Liujiapo and Luochuan sections. These taxa indicate a warmer humid climates. Zhou (1984) has suggested a sequence of alternating warm and cool periods through the Quaternary of North China which can be correlated with vegetation. The warmer wetter periods are characterized by broad-leaf and mixed conifer forest as indicated by Pinus, Quercus, Ulmus, and Betula pollen. The succession of vegetation through the cool periods is dependent on available moisture. The presence of Picea, Abies, and Pinus indicate wet and cold coniferous forest conditions while Artemesia, Chenopodiaceae and Gramineae pollen indicate a more arid cool steppe environment.

Pleistocene vertebrate paleoecology

The geological, palynological and oxygen-isotope records indicate that the Pleistocene was characterized by climatic fluctuations. Regional vegetation in North China alternated between steppe/grassland during cool arid periods and coniferous and mixed temperate forests during moister 84 warmer phases. Animals had to either migrate or adapt to changing conditions. Warm-blooded mammals, able to maintain a constant body temperature, are tolerant of a wide temperature range and it is generally the overall biotype which is more important to their survival (stuart 1982). For example, many herbivores needed open steppe/grassland to accomodate large herds but they could occupy this kind of habitat in a range of temperatures. However, the succession of a closed forest biotype changed the character of the fauna. Most paleoecological studies of Pleistocene organisms employ comparative modern animal ecology to make paleoenvironmental inferences (Kurten 1968; stuart 1982; Boyle 1990). These studies acknowledge a serious limitation with this approach. That is, that the ecological preference of an individual species may have changed during the evolutionary process. It is therefore important to study an assemblage of organisms. This section will discuss the paleoecology of selected taxa which are most useful to the interpretation of the Upper Pleistocene paleoenvironment of North China in the region of the Salawusu and Shiyu sites. 85 Animal Fossils in the Loess

Wang et a1. (1988) reports that 90% of the animal fossils recovered from strata in the center of the Loess Plateau are of the order Rodentia. As Kurten (1968:191-192) predicted, they are furnishing better evidence on the details of past environments than any other mammals because of the narrow adaptive ranges of most species of the order. Zhang Yongzhu (1988) has identified eight species of the genus Myospalax whose Pleistocene distributions coincide precisely with the main area of loess deposition. These blind, burrowing rodents avoid wooded or forest habitats, favoring instead open grassland to semi-arid steppe. The rodent, Dipus sagitta, which inhabits dune habitats, extended its range through the Pleistocene from Central Asia to the northeast China plain, attesting to the environmental degradation of this region. The distribution of the souslik, Citellus, a Holarctic steppe element, also coincides roughly with the Loess Plateau and the North China plain. By contrast, both extant and Pleistocene species of the hamster, cricetulus, occur as far east as the coastal region and south to the lower reaches of Chang Jiang (Yangtze River), exploiting a range of habitats from steppe to 86 temperate forests but avoiding extreme aridity (stuart 1982).

Artiodactyla

The North China Pleistocene faunal assemblages are rich in representatives of this order. Some are more useful than others as paleoenvironmental indicators. The red deer, Cervus elaphus is a highly adaptable species and an eclectic feeder, occupying ecological niches from temperate and coniferous forest to treeless steppe and semi-desert. A snow cover of >40-50 cm is the major limiting factor to its geographic distribution (Jia et al. 1972). The giant deer, Megaloceros, may have been more specialized in its habitat. Jia et al. (1972) have suggested that its large and outspread elaborate antlers would make movement in dense forests difficult. Other researchers maintain that interglacial forests were sparse and did not restict the range of Megaloceros (stuart 1982). Also, if there was a closed canopy, the spacing of mature trees would not inhibit the movement of large antlered species (Jelinek, pers. commun.). These animals also exploited open environments with rich herbaceous vegetation. The gazelle (Gazella spp.) of the North China 87 Pleistocene exploited semi-arid grassy plains as did the antelope, Procapra. By contrast, the water buffalo, Bubalus wansjocki, preferred a wetter environment and the wild cattle, Bos primigenius were confined to dense overgrown grasslands and open woodlands. It is possible that Bubalus was a migratory and seasonal occupant of Pleistocene North China, but this is not the case for the much more sedentary Bos primigenius (Kurten 1968). The presence of many grazing forms could be indicative of a riparian woodland vegetation on the valley floor and sides. The wild boar, Sus scrofa, has also been recovered from North China Upper Pleistocene faunal assemblages. This is a form that needs forest. This animal is omnivorous and requires a large forage intake of acorns, nuts, fruits, tubers, small rodents, grubs and carrion (Boyle 1990). Since they rooted much of their food out of the ground, extreme cold and snow cover limited their geographic range. They are associated with interglacial deciduous, especially broad-leaf forest (Kurten 1968).

Perissodactyla

The Pleistocene woolly rhinoceros, Coelodonta antiguitatis, was slightly smaller than extant rhinos and 88 relatively long-legged. It had a long, thick, soft pelt and its upper lip was wide, much like the lip of the white rhino, Ceratotherium, of Africa (Guthrie 1990). Enamel patterns on the cheek teeth were very complex and the teeth were high crowned indicating a grazing adaptation. The low carriage of the head and neck also support a grass-eating specialization. Woolly rhinoceros has been associated with cool loess/steppe conditions as in Upper Pleistocene Siberia (Chard 1974), but has also been recovered from completely different environments. Remains of woolly rhinoceros and pollen from the interglacial deposits at Cueva del Toll, Spain indicate a dry, mild climate with extensive grassland and some broad-leaf forest (Kurten 1968). Pleistocene fossil Equids are Ubiquitous in North China. According to the historic distribution of Eguus przewalskyi, they favored cool, grassy environments such as that of the Central Asian steppe and because of their hooves, were intolerant of deep snow and soft, marshy ground (Boyle 1990). Their diet is one of high fiber and low protein and while they are primarily grazers, they can survive on a wide variety of plant foods including coarse, low quality grasses and shrubs (Artemisia, Juniperus). As a herd animal adapted to running at great speeds, an open environment was favored. The species has also been found in 89 association with interglacial temperate forest, an environment which probably offered shelter and alternative forage. The kulan, Eguus hemionus, had a wide distribution in northern Asia, also inhabiting steppe regions preferably near rivers and lakes (Kurten 1968).

Summary

The continuous uplift of the Qinghai-Tibet Plateau through the Pleistocene accentuated the desiccation of North China and caused changes in the distribution of land vertebrates. Fluctuating climatic regimes of cold-dry, cold-wet, warmer-dry, and warmer-moister are reflected in the loess/paleosol strata which in turn can be correlated with the Pleistocene 6180 chronology. Current research on these correlations suggest a more complex regime of climatic fluctuations indicating many more glacial cycles through the Pleistocene than have traditionally been assigned. Pollen recovered from these layers also supports an alternating succession of vegetation which ranged from arid, cool steppe grasslands to coniferous forest to temperate forest in warmer, moister periods. The fauna recovered from Salawusu and Shiyu is 90 discussed within each site chapter and correlations with the pollen record, site stratigraphy and chronology are presented. 91

\~;;;[.\~: Loe.. deposits G A N S U Province name

~!J(~~.i\' .. ' :"':j;?

o Kilometres 1000 , n n , , ,

Figure 5. The distribution of thick loess deposits in China (Derbyshire 1988: 217) 92

SII" 211' , LII ! " u~ ~.-<:~ ~ L, ~ I J , 5, I I ' I I: , , , /1 I , I ,I ~rtl I 'I iI III I ~10m , 1 I; . . I~ I !: , ' ' I AT H I " I II Ii /..:to .•... O.·o·o·.·.·.O.O .•. · ••••.••.•.. . . .: .;y:, s A5. XA P II i I' , ! L, Ii i II ': II ii ! I II ~ I Ii Ii II I II . IIII , I A1 ,y..)0,t ,Y . . " '( L. .~~Irlfll ...... • .•.• . .• .• . ••• • 81 1\",,"'12 ~3 IT!:!:!::l4 [ll]]5 • • • [][ID6 1·.·· .. 17

1. surface soil 2. black loam 3. paleosol 4. Holocene loess 5. Malan loess 6. Lishi loess 7. sandy gravel

Figure 6 - Loess section from Lujiaowan, Shawan County, North Xinjiang (from Wen and Zheng 1988:205) 93

11e ..c" J! V 28 238 'J ~ >, "t j u ";s Ollllcn i50loJll! C "'J g :s! rc':llrd "'''Ill.' 2. Il ~ 1 c OJ 8 .~ ~ " '7 c· 1>1.. ~ I! ~ C "" "" 2.11 - --...... -_.-......

/ / / / / / I / / / /-==----.-----'~,._t / 1/ //

1. loess 2. black loam 3. greyish brown paleosol 4. cinnamon paleosol

Figure 7 - Loess/paleosol stratigraphic correlation of paleoclimate and oxygen isotope chronology. Loess sequence from North Xinjiang. (from Wen and Zheng 1988: 212) 94

I.....-.. Pictl. - """-- - - Pinus .:.. S,ru/. "------~ --- OUlfCUS - Ulmus - Frllxinus I. Cory/us I...... A/nus - "" Gramineae 1-~ - - Chenopodium·type '-- ~ Ambrosill-type ~ i-.- Arrtlmisia ...... - Cyperaceae ""'-- - Ericaceae h--. i--r-1 ...,.-, I--r-T-o I--r--r-.

Percentages 01 total pollen

Figure 8. Summary pollen diagram of mean pollen percentages for each vegetation type. Data from western interior Canada. (from Birks and Birks 1980: 248)

------_...... '-'~---"'---- __ 95

CHAPTER .. THE SALAWUSU SITE, INNER MONGOLIA

Introduction

Thick Pleistocene loess deposits covering a vast expanse of North China gave this region the descriptive term "Yellow Earth," popular in the first few decades of this century (Teilhard 1926). It was from these deposits that Chinese Paleolithic remains were recovered by European and American scientific expeditions of the 1920's. The site of Salawusu (also known as Sjara-osso-go1) was one of the earliest Paleolithic localities identified in North China. It's long history of archaeological investigation is discussed in this chapter. Salawusu is located in the southeastern part of the arid plateau enclosed by the great bend of the Huang He (Yellow River). This vast plateau, known as the Ordos, is classified today as a semi-desert zone with both shamo and gobi desert environments. This chapter presents an analysis and interpretation of the data from the most recent excavations at Salawusu in 1980. The basic unit of analysis for this study is the assemblage, which is defined here as a contextually 96 associated group of artifacts of one raw material type. Therefore both the bone artifact and lithic artifact assemblages form the archaeological database for this study. This definition differs from the conventional use of the term as defined by Bordes (1972:158): "Assemblage: All the artifacts found in a given layer at a site." The specimens included in this study are housed at the Institute of Vertebrate Paleontology and Paleoanthropology in Beijing.

A History of Archaeological IDvestigatioDs at Salawusu

Chronology

In 1922, Father Emile Licent, who at that time, was director of the Huang He - Bai He Museum at Tianjin, southeast of Beijing, received a message from a MOhgolian native named Wansjock which prompted his initial investigation of the locality. During this brief survey, Licent recovered a variety of mammalian fossils from the Pleistocene strata as well as a hominid femur from the debris at the bottom of a cliff (Boule et al. 1928). Recognizing the potential of this region for a rich Paleolithic record, Licent organized a larger scale 97 investigation of the Ordos in 1923. The Institut de Paleontologie Humaine sent Teilhard de Chardin to join Licent and they embarked on what was to be an extremely successful series of excavations which demonstrated a new level of antiquity of human occupation in North China. The team first discovered and excavated the site of Shuidonggou, located about 150 miles west of Salawusu in Ningxia Hui Autonomouus Region (refer to Chapter 1 for a brief discussion of this site). They then excavated the locality of Shaojiagouwan at Salawusu. The cultural remains from this locality included stone, bone and antler artifacts and traces of fire use. These were in association with an abundant mammalian fauna that has come to occupy a position of prime importance in the comparative faunal dating of Paleolithic localities throughout North China (Wu and Olsen 1985) . A preliminary report on the Ordos sites was published by Licent and Teilhard in 1925. Subsequently, Henri Breuil studied the cultural materials from the Ordos expeditions and this resulted in a comprehensive monograph by Boule et al. (1928) which presented the stratigraphy, paleontology and archaeology of these sites as well as one of the earliest synthetic interpretations of the Paleolithic of North China. This synthesis was based on the comparison of

.. - .. -----. --- 98 stone artifact assemblages from Shuidonggou, Sa1awusu and sites further to the southeast on the Loess Plateau. A second Paleolithic locality at Sa1awusu, Fanjiagouwan, was identified in 1956 by Wang Yuping of the Inner Mongolian Museum. Three investigations were made at this locality from 1956-1960 which yielded stone artifacts and a human femur and parietal (Wang 1963; Wu 1958). The Institute of Vertebrate Paleontology and Paleoanthropology (IVPP), a research branch of the Chinese Academy of Sciences, sent a field team led by Professor Pei Wenzhong to Salawusu in the period from 1963 to 1964. These excavations resulted in the definitive identification and interpretation of the Salawusu fauna (Pei and Li 1964; Qi 1975). Work continued at Salawusu in 1978-79 by a field team from the Lanzhou Desert Institute. Clarification of the stratigraphy of the Main Bank Deposit and valuable paleoclimatic data resulted from a variety of sediment analyses and pollen studies (Yuan 1978). The most recent excavations at Sa1awusu were in 1980 near Fanjiagouwan. They were a cooperative project by the Lanzhou Desert Institute and the Institute of Vertebrate Paleontology and Pa1eoanthropo1ogy (Huang and Wei 1981). This field research resulted in the recovery of 11 human 99 fossil fragments, stone and bone artifacts. These lithics and bone artifacts are the specimens analyzed for this study.

Artifacts from Pre-1980 Collections; Early Interpretations

The Shaojiagouwan locality excavated in 1923 yielded 200 stone artifacts of extremely small size (Figures 9 and 10). Breuil recognized the difficulty in classifying the Salawusu lithics within the European typological system he was familiar with (Boule et ale 1928; 123). He attributed part of this difficulty to the small size of the specimens. Another aspect that hindered their classification was the fact that many specimens appeared to be multi-use tools, or fragmentary and reutilized. For example, he identifies several specimens as "cores which have been transformed into scrapers; small chunks, pebbles or small thick flakes transformed into tools by retouch into micro-endscrapers and micro-sidescrapers," (op. cit.; 123, 126). Recognizing the reuse, of broken specimens, he describes "a distal fragment of a flint blade fashioned into a small endscraper by semi­ circular retouch around the tip." (op. cit.; 127). The intensity of utilization of these lithics is clear from Breuil's descriptions of small flakes and flake 100 fragments retouched on most available edges. Frequently, the same specimen exhibits notched retouch to form a serated or denticulated edge along one side while the other edges may be contiuously or alternately retouched. Breuil's use of the term "microlithic" in reference to this assemblage should not be interpreted in the strict technological terms that apply to microlithic industries in North China today (see definition in Chapter 1). He is referring mostly to the overall size of these artifacts. In one case however, that is his classification of a few specimens as micro­ burins, he does imply a microlithic technique. This classification is controversial and called into question by

Koz~owski (1971). The bone artifacts recovered during the 1923 excavations at Shaojiagouwan were of two types: long bone diaphyses which were identified as artifacts because of polished utilized tips and diaphyses with flaked ends similar to the retouch on the lithics. Antler from several species of Cervids was recovered from shaojiagouwan as well. Some of the naturally shed Cervus meqaloceros antler showed evidence of having been utilized as hammers. The major interpretive question that the French scientists were trying to resolve after these initial

---~ -_ .. __._------101 excavations in the Ordos, was the chronological position of Salawusu in relation to Shuidonggou. Recall that the Shuidonggou assemblage, which falls along the mid-line of Jia and Huang's (1985) model of tool traditions (see Figure 2), was a mixed assemblage of large and small artifacts. It contains a variety of side and endscrapers as well as triangular points similar to the European Mousterian point. There are also numerous blades and burins (Figure 11). Breuil suggested that this assemblage resembled a "highly evolved Mousterian and an incipient Aurignacian, or a combination of both" (Boule et al. 1928:121). On typological evidence, based mostly on the small size of the artifacts, Breuil suggested the Salawusu assemblage post-dated Shuidonggou. However, the stratigraphy suggested the opposite conclusion. with the new material from the 1980 excavations and a clarification of the stratigraphy it is possible to offer a new interpretive framework for this aspect of the North Chinese Paleolithic. The artifacts collected by Teilhard and Licent during the Ordos expeditions are housed at the Institut de Paleontologie Humaine in Paris. The most recent comprehensive .study of these materials was published by

Janusz Koz~owski in 1971. His was a significant contribution to the interpretation of Chinese Paleolithic 102 assemblages because it addressed the issue of the existence of an "Ordos Culture," a term first proposed by Teilhard (1941), based on distinctive regional lithic traditions in Upper Pleistocene North China.

Koz~owski felt that the Salawusu occupation preceeded Shuidonggou and was more closely aligned both chronologically and typologically to Zhoukoudian, locality 15 (Locality 15 is located 70 m south of locality 1, the "Peking Man site"). His analysis stressed the technological similarities between the Salawusu lithics and those at Zhoukoudian, locality 15. For example, both sites exhibited the production of flakes from discoid cores, but the use of very small pebbles as raw material at Salawusu influenced the size of these flakes. The Salawusu flakes were designated "a spicchio" flakes (short sectional flakes from a small thick discoid core, resembling cloves of garlic separated from a head [Koz~owski 1971: 73]). The larger available raw material at locality 15 allowed for larger discoid cores and hence the production of bigger flakes. Both sites produced small points with unifacial retouch, notched and denticulated tools, and a variety of scrapers. Qiu Zhonglang (1985) has observed that shallow bulbs of percussion on the flakes from Zhoukoudian, locality 103 15 suggest a flaking technique of simple direct percussion with a soft hammer of bone or wood. Such use of cervid antlers at Salawusu was noted previously (Boule et ale 1928). The synchrony of these two sites was further supported by the biostratigraphy. The animal species represented at Zhoukoudian, locality 15 were represented in the Salawusu fauna as well. The antiquity of these sites was suggested as early Upper Pleistocene (Koz~owski 1971: 72).

Koz~owski's conclusion was that Salawusu represented an early stage of human occupation in the Ordos region and that it was unrelated to a later occupation represented in the various levels at Shuidonggou and other sites in the Gobi to the north. He proposed the use of the term "Ordos Culture" for the blade and scraper assemblage at Shuidonggou and corresponding Gobi sites but not the Salawusu assemblage. He considered the "Ordos Culture" as unequivocally Upper Paleolithic.

The Human Fosssils from Salawusu

The potential for the discovery of hominid fossils at Salawusu had been recognized after Licent made his initial survey of the locality in 1922 and recovered a hominid femur

------~~- ----.---~=-,------.------.- . 104 from debris at the bottom of a cliff. Since it was not recovered in situ, its antiquity was uncertain. The only distinguishing feature of the diaphysis was a particularly well-defined area for muscle attachment (on the linea aspera?) [Boule et al. 1928: 17] which might be suggestive of a robust individual. When the first collections of vertebrate fossils from the 1923 Shaojiagouwan excavations were brought back from the field to the laboratory for study, a heavily fossilized human upper incisor was identified. This specimen was recovered approximately 500 m from a hearth at the boundary of the Pleistocene sediments and recent river terrace gravels (op. cit.: 3). It exhibits shovel-shaped morphology on its lingual surface. This tooth was studied and named "the Ordos tooth" by Dr. Davidson Black, then director of the Department of Anatomy at Peking Union Medical College (Licent et al. 1926). Two additional human fossils, a fragmentary parietal and femur, were recovered during the 1956-1960 work at Salawusu. These specimens were described by Wu Rukang (1958). Excavations during 1978-1980 yielded an additional 15 human fossil specimens (Huang and Wei 1981). Table 1 presents a list of all the hominid remains recovered to date from this site.

------_.... _- . 105 In 1981, Dong et ale published a report on the human fossils recovered from Salawusu in 1978 and 1980. Their conclusion was that these specimens exhibited "some advanced characters of modern humans ••• while retaining some archaic characters .•• they may belong to archaic Homo sapiens or early Homo sapiens sapiens." (Dong et ale 1981: 1193). Since the initial recovery of the Ordos tooth, the stratigraphic and chronological interpretation of the hominid remains has been controversial. Conflicting information in the literature makes it difficult to resolve the taxonomic position of these specimens. Some of this confusion may result from the fact that several of the cranial specimens, especially the more complete ones found in situ in the lower Salawusu formation, appear to be from adolescent individuals. This would make it difficult to assess the occurrence of "primitive" features that might not be fully expressed in the skeletal material of young individuals. Several of the more complete specimens are post-cranial bones which have fewer diagnostic attributes useful for a phylogenetic determination than cranial bones. Another problem in assigning a subspecies designation is an acknowledged limitation of paleoanthropological studies: that is, how to interpret the range of human variation in a small sample size such as the

------106 fossil record for archaic Homo sapiens in China and worldwide. The archaeological data from the 1980 excavations presented in this dissertation considered along with the stratigraphy of the Salawusu site may be interpreted in such a way as to shed new light on this problem.

The Salawusu Fauna: Chronological Implications

The faunal record from Salawusu is a rich collection of fossil vertebrates which has been studied extensively. Because of this, it has become the type collection of North Chinese Upper Pleistocene fauna (Han and Xu 1985). The monograph by Boule et al. (1928) identified 32 vertebrate species and described these specimens in detail. since then, 13 additional species have been identified. Table 2 provides a complete list of the Salawusu fauna. This faunal suite has played a prominent role in establishing a relative chronology for the Ordos region and much of North China in general. The percentage of extinct mammals in a faunal assemblage is important in arriving at a relative chronological range when comparing faunal collections from various sites. Approximately 30% of the mammalian fauna in this 107 collection represent extinct taxa including Elephas CPalaeoloxodon) namadicus (considered the type fossil for the Upper Pleistocene in North China), Coelodonta antiguitatis, Megaloceros ordosianus, Spiroceros kiaktensis, Bos primigenius, Bubalus wansjocki, Crocuta ultima, and Struthiolithus (Han and Xu 1985). Boule et a1. (1928) concluded that this assemblage was early Upper Pleistocene. However, a report published in 1975 by Qi Guochin interprets the Salawusu fauna as late Upper Pleistocene based on fossils recovered in 1963 and 1964 by an IVPP field team. Complete skeletons of wooly rhinoceros and tiger were found in a distinct horizon of the lower Salawusu formation and Qi (1975) suggests that these animals lived near the river and perhaps became mired in a muddy environment. A radiocarbon date of 35,340 ± 1900 BP has been obtained on burned bone excavated in 1980 near the Fanjiagouwan locality. Single radiocarbon dates, however, must be viewed with great caution, especially dates on problematical material such as burned bone. Problems of contamination by older or younger carbon are often encountered where bone is used for dating. Carbon exchange takes place readily after death in the inorganic phase of the bone matrix. Bone diagenesis as well as the process of 108 fossilization, where the primary hydroxyapatite is transformed into a less soluble carbonate-fluorapatite, both introduce potential sources of contamination (Prevot and Lucas 1990). A comparison with the fauna from locality 100 at Dingcun, Xiangfen County, Shanxi suggests that Dingcun is an earlier assemblage because of the higher percentage of extinct taxa there. Twenty-seven mammalian species have been identified from Dingcun and over half are extinct (Jia et al. 1972). The Dingcun fauna contains two Middle Pleistocene taxa, Palaeoloxodon tokunagi and Dicerorhinas kirchbergensis (Yuan 1978). Relative chronology based on faunal comparisons must consider several factors. Differential preservation of bone and a consideration of Pleistocene climatic fluctuations affecting the persistence of animal species northward and southward, are important variables when constructing relative chronologies based on faunal remains. In composition and percentage of extinct species, the Shiyu fauna is thought to be closely related to the Salawusu fauna (Yuan 1978). It contains approximately 40% extinct taxa including Felis tigris, Coelodonta antiquitatis, Megaloceros ordosianus, Bubalus wansjocki, and Bos primigenius. There is less variety of fauna at Shiyu than 109 at Salawusu. The radiocarbon dates for this site place it within the late Upper Pleistocene very cold and dry stage 2 glacial. The sparse representation of animal species at Shiyu may be the result of a less favorable paleo-habitat in North China at this time or it may be a consequence of diagenetic factors affecting site formation. The composition of two other faunal assemblages have been compared to Salawusu. The Upper Cave assemblage at Zhoukoudian has yield a large faunal collection assigned to the late Upper Pleistocene. Approximately 12% of the fauna is represented by extinct taxa, including Crocuta ultima, Ursus spelaeus, Elephas, and Struthiolithus (Yuan 1978; Jia et ale 1972). A radiocarbon date of 18,865 ± 420 BP was obtained on deer bone from the cave's lower fissure and is consistent with the preconceived ideas about the cultural material and human fossils recovered here (Wu and Zhang 1985) . The cave deposits of Xiaonanhai near Anyang in northern Henan on the western edge of the North China Plain have yielded 18 mammalian species of which 4 (22%) are extinct. This fauna is associated with a large collection of stone artifacts which are mostly small tools made on chert flakes. A range of radiocarbon dates between 24,000 BP (on charcoal) and 13,000 BP (on Bubalus bone) has been generated for this

--- ._._------110 collection of fauna and cultural materials (Aigner 1981; An 1965). These dates suggest that the Shiyu fauna may be more closely aligned with these sites while the Salawusu fauna is of greater antiquity.

The stratigraphy of the Salawusu Site

In regional perspective, the Salawusu site is located just at the boundary of two mega-geomorphic regions, on the southeastern edge of the Ordos Plateau and the northwest boundary of the Loess Plateau. The environment today is considered a peripheral arid area of steppe and desert­ steppe (Zhao 1988). The site is named after the Salawusu River which originates in the Loess Plateau to the south and flows north in a sinuous path through the flat southeastern Ordos. It cuts deeply into unconsolidated sediments forming furrowed river bends called gouwan by the local inhabitants. Most localities along the Salawusu River are named with this term as a suffix. The river is greater than 20 m in width and its muddy color is a result of the large quantities of sand and clay which it carries. It has cut a valley between 60 - 80 m in depth and 300 - 1500 m in width. At the locality of 111 Dixiaogouwan, two banks have exposures with complete quaternary sequences (Yuan 1978). This sequence is called the Main Bank Deposit of the Salawusu Formation. The bedrock, a red Cretaceous sandstone, is also exposed here. Six river terraces have been identified measuring 8 - 12 m, 15 - 20 m, 25 - 28 m, 30 - 35 m, 40 - 45 m and 52 - 57 m respectively above the present river level. Dynastic cultural relics associated with the Song and Yuan Dynasties have been recovered from within the first and second terrace sediments suggesting that they were recent deposits between 1,000 and 2,000 years ago (Huang and Wei 1981; Dong et al. 1981). During this period, there was frequent tectonic activity, active downcutting by the river and snake-like river meanders developed (Zheng 1988). Twelve stratigraphic layers are associated with the Quaternary sediments (Figure 12A). These are divided into three stages. The uppermost layers, 11 and 12, represent post-glacial Holocene deposits and are called the Dagouwan formation. They are lake basin-bog deposits of fine clay sediments interpreted as an environmentally warm, moist period in the early Holocene. Uplift of the Ordos Plateau continued through the Holocene and by the top of layer 12, the lake-bog facies disappears and the present desert-steppe environment was formed (Yuan 1978). 112 The middle layers, 6 - 10, are 25 m thick and compose the Upper Salawusu formation (Huang and Wei 1981). These are primarily fluvial sediments of fine-grained sand accumulated by the river during a generally cold, dry period. Layers 7 and 9 are sediments of thin, horizontally bedded silty clay. They are thought to represent two short pluvials within this generally dry period and the depositional environment of these two layers was a bog or marsh (Yuan 1978). The bottom layers, 1 - 5, are 40 m thick and comprise the Lower Salawusu formation. They are a fluvo-lacustrine facies preserving evidence of Pleistocene climatic fluctuations which may be correlated with the tectonic history of the Ordos Plateau. The lower part of this sequence (layers 1, 2 and 3) was deposited during a time when much of the southeastern Ordos was sUbmerged. In the middle of the Plateau, loose Tertiary residuum was easily eroded and carried by rivers. Small lakes and vegetation were plentiful and the climate was mild. The lakes accumulated large volumes of fine sandy sediments. The stone artifacts, animal and human fossils are associated with these sediments. Iron-bearing reddish puple clay deposits, indicating an oxidizing environment, are found at the top of layer 5. By this time, the small lakes formed

------113 one large shallow one and the climate became colder and more arid.

The Salawusu stratigraphy: Chronological Implications

There have been many stratigraphic studies on Upper Pleistocene fluvio-lacustrine deposits and there are many interpretive views on how the sequence should be subdivided (Huang and Wei 1981; Dong et al. 1981; Qi 1975; Yuan 1978; You and Xu 1981). A recent refinement of the Salawusu section, meant to differentiate depositional characteristics of the layers and to explain their sedimentary facies was reported by Zheng (1988, Figure 12B). The layers differentiated by Zheng (1988) and Yuan (1978) correlate in the following way: Zheng's Layers Yuan's Layers Formation layers 35-37 layers 11-12 Dagouwan layers 30-34 layers 6-10 Upper Salawusu layers 1-29 layers 1-5 Lower Salawusu

The mammalian fossils, human fossils and stone tools from the 1980 excavations were recovered from Zheng's layers 6-8. Radiocarbon dating of organic remains in Zhenq's layers 4 and 5 show them to be beyond the dateable limits 114 for this technique and he reports a TL (thermoluminescence) date of 174,000 ± 14,000 BP for the sediments in layer 9 (Zheng 1988:237). TL dating of sediments is a technique which has recently been applied to the mineral constituents of sedimentary deposits to obtain the time of deposition (Berger and Huntley 1980). Thermoluminescence has been used with great success in dating pottery but its utility for that purpose is based on the principle that the heating of minerals during the ceramic firing process removes the TL signal derived during their previous geological history (Aitken 1974). For sediments (unless they are volcanically derived), no heating occurs at the time of deposition. This fundamental difference has critical repercussions regarding the proper calibration of TL dates. Wintle and Huntley (1982) have published a comprehensive review on the problems and complexities involved in producing and interpreting TL dates of sediments and the interested reader is referred to their paper. Their conclusions suggest that at least a ±10% loss of accuracy exists for absolute dates. Futher, they stress the necessity in reporting the result of four critical tests (plateau test, anomalous fading test, alpha­ effectiveness value, test for non-linearity at low doses) to support the validity of any reported absolute date.

------115 In light of the above discussion, the following interpretation of the Salawusu section is proposed. The iron-bearing reddish purple clay deposits indicating strong weathering at the top of Yuan's layer 5 (Zheng's layer 29) may be interpreted as having been deposited during the cold wet phase preceding the 0180 stage 4 glaciation. The colder temperatures of this relatively short stage are not obvious in this section and may be represented by an erosional hiatus between the Upper and Lower Salawusu Formations. The somewhat ameliorated climatic 0180 stage 3 may be reflected in the Upper Salawusu Formation. The depositional environment of these layers (Yuan's 6-10 and Zheng's 30-34) was mainly lake swamp. The sediments ar~ clays containing a molluscan fauna, Radix lagolix, interbedded with fine fluvial yellow sands. The very cold and dry 6180 stage 2 could be represented by another erosional hiatus separating the Dagouwan Formation at the top of the section. If this interpretation is accepted, it implies that the Lower Salawusu Formation is between 75,000 - 130,000 years old and the Upper Salawusu is 30,000 - 55,000 years old. 116 Pollen and Faunal Correlations with Paleoclimate

During the period of deposition of the Lower Salawusu Formation, a relatively stable lacustrine environment was dominant. Zheng (1988) reports that the pollen recovered is mostly Artemisia sp. and Chenopodiaceae, indicating relatively warm, semi-a.rid grasslands. Quercus sp. pollen is also present in small amounts indicating some deciduous forest, perhaps a sparse interglacial open woodland environment. The fauna recovered from the Lower Salawusu Formation (see Table 2) are well suited to this vegetation and to the interglacial temperature range. Of the order Rodentia, Microtus sp. feeds largely herbaceous vegetation, particularly grasses. Cricetulus is commonly found in grassland along river or lake margins. Lepus sp. generally inhabit open or sparsely wooded country and eat grasses and other herbs (stuart 1982). All of the species of Artiodactyla and Perissodactyla recovered from these layers could thrive in grassland and sparsely wooded environments where grazing animals could find plentiful food and the woodland would afford some protection from predators such as the hyaena, crocuta. The occurrence of the ostrich, Struthiolithus sp., and the 117 camel, Camelus knoblochi suggest a relatively warm, dry grassy plain. Both these animals can adapt to cooler temperatures and the ostrich is often found in association with woolly rhinoceros (Jia et ale 1972). The water buffalo, Bubalus, has been found in numerous interglacial strata and confines itself to muddy shores of rivers or lakes (Kurten 1968).

The 1980 stone Artifact Assemb1aqe

The 1980 excavations near Fanjiagouwan were a cooperative investigation by a field team from the IVPP working with a group fr'om the Desert Research Institute in Lanzhou. They recovered 11 specimens of human fossils, 167 stone artifacts and 344 fragments of animal bone. In the Salawusu lithic assemblage, as in many other Paleolithic assemblages, there is a range of variation within the attributes of the artifact categories so that the entire assemblage may be viewed as a technological continuum. This is especially apparent when reduction strategies are conservative as a result of small and scarce raw material. The nearest source of raw material consistent with the size and lithology of the Salwusu lithics is a distance of 118

40 km northwest of the site where Middle Pleistocene pebble deposits are found in the highlands (Huang and Wei 1981). At Salawusu there are no pebble deposits but only accumulations of fine sand, silt and clay. The analysis of the stone and animal bone follows.

Cores and Flakes

These two classes of artifacts provide a great deal of information about an assemblage in terms of sophistication of knapping techniques. As Isaac (1978) has noted, technological patterns evident in cores usually complement those observed in the flakes of the same assemblages. The cores and flakes recovered in 1980 from Salawusu give little indication of elaborate standardization of form or preparation of the core.

Cores:

Cores represent only 6% of the stone artifacts at Fanjiagouwan. They are all cores with un faceted platforms made on small chunks of stone with no cortical surface remaining. This type of core may be used for the production of ordinary or wide flakes (definitions follow). 119 While cores are not a numerically significant part of the assemblage (N=10), they show some interesting technological features. They are almost all made on chert and 60% of these specimens exhibit at least one retouched edge. Because of this type of utilization, they may be considered part of the tool assemblage at this site, indicating an intensity of use and conservation of raw material. Almost all the cores have more than one striking platform and at least 3 flake scars, with 10 flake scars being the maximum.

Flakes:

Flakes are numerically the most significant stone artifact at this site. They comprise 66.5% of the assemblage and can be classified into the following types (after Akazawa et a1. 1980). ordinary flakes: A flake which is longer (based on percussion axis length) than, but less than twice its width (measured at midpoint). In this assemblage, 36% of the flakes are classified as ordinary flakes. 120 Wide flakes: A flake that is wider than it is long. This is the most common type of flake in this assemblage with 64% of the flakes classified as wide flakes.

The following metric variables were recorded on flakes. Refer to Figure 13 for an illustration of these variables.

1. Percussion axis length

2. Width at approximate midpoint (using percussion axis

length) 3. striking platform width

4. striking platform thickness 5. Flake thickness at midpoint of width

6. Weight (g)

7. Flaking angle: angle between platform and bulbar surface

In addition, the following descriptive attributes were recorded:

1. Raw material type

2. The type of distal termination: feather, hinge or step

3. Platform type: un faceted (plain), dihedral, point,

recessed

4. Flake morphology: parallel sides, convergent sides, fan (divergent) sides, scale (convergent triangular)

5. Bulb of percussion: salient or diffuse 121

6. cortex amount

The distributions within these categories are summarized in Tables 5 - 11.

The Reconstruction of Flaking Technologv: Flake form and size. termination. striking platform. and cortex

It is apparent from Tables 18 - 20, that the Salawusu flakes are very small and mostly unretouched flakes. Only 11% of the flakes exhibit edge scars which might be interpreted as utilization and these are all larger, thicker flakes (only one utilized flake is less than 100 mm long and 3.5 mm thick). It must be kept in mind that it is often not possible to distinguish utilization from natural edge damage. When the total sample of flakes is plotted by flake area (lxw) over thickness (Figure 14), it is clear that they represent a continuous unimodal group. Figures 17, 18 and 19 illustrate the relative frequencies of flakes within intervals for each variable (length, width and thickness) independently. Platform areas, calculated for flakes whose striking platform width and thickness could be measured with some 122 degree of accuracy, suggest that these flakes have small platforms in relation to flake size (Figure 15). Recent experimental lithic analysis suggests that this is a distinguishing feature of flakes produced by soft hammer percussion (Hayden and Hutchings 1989). These researchers found that two distinct populations of flakes were apparent. The "typical" billet flake with uniformly small platforms and the more unusual type with variable platform sizes. contrary to Hayden and Hutchings (1989) results, a cluster analysis of the Salawusu population of flakes on the basis of platform area and flake size does not show two distinct populations although Figure 15 does suggest a bimodality for platform size independent of flake area. There does not appear t,o be a relationship between platform area and flake thickness for these flakes (Figure 16). Salawusu however, is much more limited in terms of the flaking properties of the raw material and blank size then experimental population analyzed by Hayden and Huthchings (op. cit.) Almost all the striking platforms are un faceted (plain). Crushed platforms, suggesting bipolar flaking, are rare. The flake morphology is predominantly square (equidimensional flakes with parallel sides) or convergent 123 (greatest width at the platform). This is consistent with the predominance of wide flakes and the lack of elongation. There is no blade or microblade technology evident in these flakes despite their small size and this point is clear from the lack of core preparation as well.

Tools

Most (69%) of the tools at Salawusu are fashioned on chert flakes. When these flake tools are graphed by length, width and thickness in comparison to the unretouched flakes, two distinct distributions are observed (Figures 17, 18 and 19) . Four tool classes may be distinguished at Salawusu: scrapers (side and circular), borers, notches and gravers. These are illustrated in Figures 20 and 21).

Scrapers

Almost half of the tools in this assemblage may be classified as scrapers with side scrapers being most numerous. These tools are recognizable by the presence of a beveled margin, asymetric to the horizontal plane of the piece (termed "planoclinal" by Isaac 1977: 147). These side 124 scrapers have an average length of 18 mm. There is a low degree of standardization among this tool class. The retouch of the working edge is mostly irregular, approaching a denticulate morphology. Considering the small size of these tools, the retouch is steep and invasive and the cutting edge angle is generally between 51-75° and sometimes as steep as 86°. This suggests an intense use and resharpening of the tool. The retouch is sometimes continuous along one edge of the tool, sometimes intermittent and can occur on an exterior or interior surface. There is no consistent overall retouch pattern. More than half of the side scrapers are retouched in a ratio of less than 50% of their circumference. While this seems contradictory to an intense reuse pattern, it may be explained by viewing these specimens as the exhausted stage of a tool that is too small to retouch further.

Circular Scrapers

These specimens are fashioned on wide flakes or broken flake remnants. They are almost circular in outline with the working edge made along whole or mostly whole margins of the blank. The angle of the working edge is steep to almost perpendicular and the retouch style is fine, small scale- 125 shaped flake scars. The average length and width of these specimens is 14.2 rom and 14.6 rom respectively. Their thickness is variable (between 4 and 8 rom). Their small size and steep working edge angle suggests that these are end products in a resharpening and reshaping sequence conserving the better quality raw material (chert).

This category is the second most numerous tool class in the Salawusu lithic assemblage after scrapers. Borers are pointed tools made on relatively thick flakes or chunks. The bore point averages 3.1 rom in length and is formed by lateral retouch of both edges of the piece. This tool class has an average length of 19 rom and an average thickness of 7 mm. This suggests that a borer with a broken point might be retouched into a scraper.

Gravers and Notches

Both of these tool types are very rare in this collection of artifacts. Koz~owski (1971: 73), on studying the 1923 Salawusu stone tools, maintains that this inventory

------"_. " 126 "lacks typical burins." The few controversial specimens from this collection that have a questionable burin facet might best be referred to as gravers consistent with similar implements from Zhoukoudian locality 15 and Xujiayao, Shanxi (Qiu 1985). Both these sites are dominated by small tool types as well and are dated on the basis of their faunal assemblages as early Upper Pleistocene.

Bone Tools at Salawusu

There are three categories of tools identifiable among the bone fragments of the 1980 excavations at Salawusu. For descriptive purposes, this study calls these flaked bone cores, flaked bone with a scraper edge and polished bone implements. The use of the term "scraper" here is meant to accentuate the similarity in morphology between the working edge of lithic scrapers and these bone tools. A total of 344 bone fragments were recovered. Of these, 12% are identifiable implements, falling into one of the three categories above. The determination of bone fragments as artifacts for this study was conservative. Any fragment exhibiting tooth marks (rodent gnawing or large carnivore tooth marks), even if these occurred at a distance

-----_._--- 127 from percussion flake scars, was not included in the total artifact count. Only specimens with the flaking patterns defined below were included so that it may be said with some confidence that the coarse flaking on the bone cores and scrapers was caused by repeated adjacent percussion blows. Most of the accumulated and recovered bone fragments are mid-shaft long bone fragments of small artiodactyls (antelope, gazelle) or larger herbivores (woolly rhinoceros, cattle). They are therefore listed to the closest taxonomic category allowed, i.e. Cervidae, gen. et sp. indet.

The Absence of Bone Flakes

Before proceeding further with the bone tool analysis, the absence of bone flakes in this assemblage must be addressed. Only two flake specimens could be identified: one measuring 16 rom in length, 9 rom in width and 3 rom in thickness and the other 46 rom in length, 30 mm wide and 8 mm thick. The size of these flakes are consistent with flake scars measured on the smaller mid-shaft long bones of gazelle or antelope and the larger flake scars on the more massive rhinoceros or camel bones, respectively. The absence of bone flakes may be due to a combination of factors. Thinner bone flakes removed from the cortex 128 would be more susceptible to diagenetic processes than the more massive limb bone fragments which served as cores and scrapers. Lambert et al. (1985) has shown that decay rates are inversely proportional to bone size. Thinner bone flakes would also be subject to breakage and discard and these small fragments vulnerable to salt weathering processes. It is also possible that the smaller bone flakes might not have been recovered by the excavation techniques.

Flaking Patterns

The orientation of flaking is an important criterion for classifying the bone cores and tools because it has an influence on both the size and shape of flakes detached and the working edge produced. The flaked bone artifacts may be separated into the following categories: 1. Bone cores with broad flakes removed from the exterior cortical surface in a direction parallel to the long axis of the bone. 1a. Flakes removed from only one end of the fragment lb. Flakes removed from both ends of the fragment 2. Bone cores with flakes removed from a lateral edge of the fragmentary bone shaft perpendicular to the long axis of the bone. None of the specimens were flaked on both lateral 129 edges. 3. Bone cores exhibiting a combination of both flaking orientations: one or both ends and a lateral edge. 4. Bone tools with scraper edge(s) formed by: 4a. Small flake scars, trimmed on one end 4b. Small flake scars, trimmed on both ends This pattern of flaking was confined to the termination of fragmentary shafts. None of the specimens exhibited this type of flaking on a lateral edge. Table 23 lists the frequencies of each flaking pattern. Figure 22 illustrates these artifacts. Pattern la, 1b and 2 were exclusively found on the more massive limb bone fragments of the larger herbivores, woolly rhinoceros, Camelus, or Bovidae, gen. et sp. indet. Pattern 3 was found on both larger herbivore long bones as above and on smaller limb bones of artiodactyls and perissodactyls, Equidae, gen. et sp. indet. Patterns 4a and 4b were found exclusively on the fragmentary mid-shafts of long bones of gazelle, antelope and deer.

Polished Bone Implements

The polished implements are all diaphyseal fragments on smaller long bones such as radiae and ulnae of gazelle or 130 antelope. The fragmentary ends show evidence of utilization producing a polished tip.

The Use of Antler

Antler from several artiodactyl species including Megaloceros ordosianus, Cervus elephus, and Gazella was recoverd at this site. Several of the cervid antlers had cranial fragments adhering and therefore were not naturally shed. The naturally roughened burr of one of the large shed Megaloceros antlers had been worn smooth by continual pounding as in a flaking wand or hammer. This massive antler would make a more durable and effective hammer implement than antler from smaller species. A durable hammering implement of this nature would be a valuable tool for breaking and flaking long bone and stone in an assemblage where large pieces of lithic raw material which could serve as hammerstones were rare.

Discussion

The lack of cortex on the specimens from Salawusu and the scarcity of cores suggest that the initial reduction of 131 raw material was done elsewhere. The small size and steep working edges suggest that the implements recovered were the exhausted remnants of tools that had been conserved through a series of reduction stages. Evidence of the use of bipolar flaking has been recovered from Chinese Paleolithic assemblages associated with Homo erectus at Zhoukoudian locality 1 (Zhang 1985). This technique is found throughout Paleolithic assemblages in North China (Wu and Olsen 1985) and researchers have suggested that it is particularly useful for working small raw material (Clark 1954; Binford 1979). Bipolar flaking has also been cited as one of the indications of a curative technological strategy (Magne 1989). It is interesting that despite the small size of the Salawusu lithics, there is very little evidence of bipolar technique. Flakes with crushed platforms are rare and there are no cores which exhibit the characteristic crushing on both ends. The reason for this may be that either 1) the technique was not used or 2) the final end-products of this technique are unrecognizable. It is possible that the final stage of reduction involved a bipolar technique to shatter a small core or implement into pieces with a "one-use" edge or point. These shattered bipolar remnants might not have been recovered by the excavation techniques. 132 The use of soft hammer percussion at Sa1awusu has long been a controversial question since the earliest excavations at the locality. In particular, heavy suitable hammerstones for hard hammer percussion flaking would have been extremely rare in the Middle Pleistocene pebble deposits utilized as a lithic raw material source by the population at Salawusu. This problem may have been solved in part by the use of shed cervid antler and perhaps wooden hammers as well. Lipped platforms, a feature usually attributed to flakes produced by soft hammer, are very rare in the Salawusu stone assemblage. Only five specimens have this feature. This feature may be dependent in part, on the flaking properties of the raw material. From the diversity of the Salawusu faunal record we know mammals were plentiful there. Chinese scientists have suggested that a wide variety of animals lived near the Salawusu River and at times became mired in boggy ground (Qi 1975). This raises the question of how much of the bone assemblage at the site is a direct result of human activity. Observations of modern African savannas suggest that under conditions of normal animal mortality, a very low concentration of bone results on the landscape. Taxonomic and ecologic diversity is low, there is little mixing of bones from different animals, and vertebrate and other axial 133 skeletal parts tend to remain near the death site while limb bones tend to be removed (Hill 1979; Behrensmeyer and Boaz 1980). At Olduvai, by contrast, animal bones were densely concentrated. There is a high degree of mixing of different carcasses, high taxonomic and ecologic diversity and a disproportionately small frequency of axial skeletal parts relative to limb bones (Potts 1988). The dense concentration of bone at Salawusu is almost exclusively limb bone fragments, only 2% are axial skeletal fragments, meaty upper limb bones are well represented. There is a high degree of mixing of different species and high taxonomic and ecologic diversity. The overwhelming majority of the bones recovered from the 1980 excavations belonged to herbivores. Carnivore remains are not well represented at the site, although a complete tiger skeleton was recovered at a nearby locality (Qi 1975). The high percentage of unretouched small flakes and small tools with steeply bevelled edges do not support a hunting scenario for the site use. Researchers have suggested that this type of implement is more suitable for whittling wooden objects or paring staves (Clark 1954). The abundance of herbivores relative to carnivores at this locality must have made it a favorable place for hominids to scavenge carcasses for meat, marrow and raw material on

...... _ _-_ ------" ... ------..---- 134 which to fashion an expedient bone tool kit to supplement their poorly supplied lithic tool kit. Magne (1989) presents an experimental model to relate lithic reduction strategies to regional assemblage diversity in curated technologies (see Chapter 2). He tests his model using data from seven Paleoindian assemblages. Viewing Salawusu as a curated assemblage, or at least one where lithic raw material is highly conserved, it can be placed into Magne's proposed model. If we separate the Salawusu flake assemblage into early and late stage debitage on the basis of flake size and plot the debitage/tool ratio (1.4:1) over the percentage of late stage debitage (66%) using Magne's model (1989) in Figure 23, we find that this assemblage falls into the upper left corner of the "reoccupied sites - high reuse/scavenging rate" box. Similarly, the low diversity of tool types and the high percentage of late stage debitage causes this assemblage to fall within the "special use sites, situational emergency camps" box. These are reasonable scenarios given the archaeological data from this site. summary and Conclusions

The correlation of the Salawusu stratigraphy with the 135

&180 chronology proposed here suggests a much greater antiquity for this site than does the radiocarbon date on burned bone. Problems with the C-14 dating of this material have been discussed. The TL date of 174,000 ± 14,000 BP reported by Zheng (1988) for his layer 9 may be too old, again an inherent problem with the dating technique. If the Lower Salawusu Formation is greater than 75,000 years old as the correlation with the &180 chronology supports, than this site cannot be considered as belonging to a late Upper Pleistocene small tool tradition. Rather, it should be viewed as adding a further dimension of variability to the inhomogeneity of the early Upper Pleistocene North China assemblages. The lack of flake elongation, blade technology and bipolar flaking and the presence of an expedient bone tool assemblage are distinguishing features of this site. The nature of the lithic tool assemblage suggests a behavioral trend toward conservative use of stone resources. A comparison with the Shiyu lithic assemblage which follows accentuates the technological and typological discontinuity ,between these assemblages. 136

Figure 9. Small tools from the 1923 excavations at Salawusu (Sjara-osso-gol) (from Boule et ale 1928: 123) (actual size) 137

. . J~' ,...... ~ !r,;~ .' @~; ~.;

Figure 10. Small tools from the 1923 excavations at Salawusu (Sjara-osso-gol). (actual size) (from Boule et al. 1 '928: 126) -::-- .138 . -

(12 . I ~

~ -:-:-...~..... ~.: ~. '~....~ ...... ", . - . ~ . \ -"

,~' .# ~ _~: f r I...- • ''""1:..::. Figure 11. The Paleolithic industry from Shuidonggou (from Boule et ale 1928: 110,113) (actual size) 139

Table 1 - Human POBBi1B Recovered at Sa1awuBu Specimen Condition S'tratigraphic Layer Date Frontal almost complete Lower Salawusu Fm. 1978 adolescent Frontal almost complete discontinuity 1978 adolescent Frontal mid-upper fragment Lower Salawusu Fm. 1978 young adult Parietal right posterior frag. alluvial sediments 1980 young adult Parietal right mid-upper frag. alluvial sediments 1980 adolescent Parietal right upper frag. alluvial sediments 1956 adult Occipital large middle frag. Lower Salawusu Fm. 1980 adult Occipital mid fragment alluvial sediments 1980 child Incisor upper left alluvial sediments 1923 adult Mandible right frag. Lower Salawusu Fm. 1978 child Mandible right frag., with alluvial sediments 1980 third molar, adult Vertebra first thoracic alluvial sediments 1980 almost complete Scapula right, adult Lower Salawusu Fm. 1980 almost complete Scapula right, adult alluvial sediments 1980 almost complete Humerus left, adult alluvial sediments 1980 almost complete 140

Table 1 (continued)

Humerus left, adult alluvial sediments 1922 Femur right, adult alluvial sediments 1978 almost complete

Femur right, adult alluvial sediments 1980 no distal end

Femur distal end discontinuity 1922 adult

Femur right frag. discontinuity 1922 adult

Femur left, distal half alluvial sediments 1956 adult

Tibia right complete Lower Salawusu Fm. 1978 adult

Fibula left, adult alluvial sediments 1980 proximal fragment

(from Huang and Wei 1981: 29) 141

Table 2 - The Salawusu Fauna Order: INSECTIVORA Erinaceus sp. Hedgehog scaptochirus moschata Mole Chiroptere, gen. indet. Bat RODENTIA Lepus sp. Jack rabbit Ochotona sp. Pika citellus mongolicus Souslik Allactaga cf. annulatus Jerboa Dipus sowerbyi Kangaroo rat Meriones meridianus Tamarisk gerbil Myospalax fontanieri Mole-rat Eothenomys sp. Vole Alticola cf. strachyi Mountain vole Microtus cf. ratticeps Root vole Microtus sp. Vole Cricetulus cf. griseus Hamster CARNIVORA Canis lupus Wolf Felis tigris Tiger Crocuta ultima Hyaena Meles taxus Badger PROBOSCIDEA Palaeoloxodon cf. namadicus Elephant PERISSODACTYLA Eguus cf. przewalskyi Horse Eguus hemionus Kulan Coelodonta antiguitatis Woolly rhinoceros ARTIODACTYLA Sus scrofa Pig Camel us knoblochi Camel Cervus elaphus Red deer Megaloceros ordosianus Giant deer Cervus mongoliae Deer Procapra picticaudata Antelope Gazella przewalskyi Gazelle spirocerus kiakhtensis Bovid Ovis ammon Sheep Bubalus wansjocki Water buffalo Bos primigenius Cattle 142 Table 2 (continued) AVES Buteo cf. ferox Hawk Vultur monachus Vulture Passereaux sp. Sparrow Perdix cf. perdix Grey partridge Coturnix sp. Quail Syrrhaptes paradoxus Sandgrouse Echassier Podiceps auritus Horned grebe Anas boschas Duck (teal) Tadorna tadorna Sheld-duck Struthiolithus sp. Ostrich 143

l~ ------0 -=-..:: .:.:': f-...... -=---..:: -=: .•. 10 .7 10 - -- 30 40

10

10

•...... : ,".:,' 70

10

10

100 m Figure 12A The Salawusu section (after Yuan 1978:224) o

5

10

15

20

25 I .r:. 30 C. k.:~·i.'·~:·~···~··~···5·:t··19 c!: :...... ;' ...... ~ !! .::;;:~'J;:11fr.ls 35 . ;". ~~:r .. ,":~ .... ~ 14 .~.'P;' •• ~.:r:. '::,,13 .~.::~.~:.~ ~.7:.12 40 :', : .,:.: ~~. .'.~; 11 ... ~~.!-~-.~ '5:' :=~~:: 10 45

50

55

------~60 Figure 12B The Salawusu section (after Zheng 1988:236) 144

1. Percussion axis length 2. width at approximate midpoint 3. striking platform width 4. Striking platform thickness s. Flake thickness at midpoint of width 6. Flaking angle: angle between platform and bulbar surface

Flake Form Q o convergent divergent convergent parallel triangular

striking Platform Types ., ~ <:::p I I ~ r tt ~\ ~ (\ un faceted dihedral point recessed

Figure 13 - Flake measurements and attributes

._.. __._------.... - ..---- .. ------_... _- -.- 145

DATA TABLES FOR THE 1980 SALAWUSU ARTIFACTS

TABLE 3 - ARTIFACT CLASS

FLAKES CORES TOOLS N 111 10 46 167 66.47 5.99 27.54 100%

TABLE 4 - TOOLS (excluding reutilized cores) SIDE ROUND BORER NOTCH GRAVER N SCRAPER SCRAPER 22 4 14 3 3 46 47.83 2.40 8.38 1.80 1.80 100%

TABLE 5 - RAW MATERIAL RELATIVE FREQUENCIES

QUARTZITE QUARTZ CHERT FLAKES 36.84 15.79 47.37 CORES 0.00 20.00 80.00 TOOLS 17.39 23.91 58.70

TABLE 6 - TOOL BLANK

FLAKE CORE FRAGMENT PEBBLE ND N 36 6 8 1 1 52 69.23 11.54 15.38 1.92 1.92 100%

TABLE 7 - TERMINATION OF FLAKES

FEATHER STEP HINGE N 90 15 6 111 81. 08 13.51 5.41 100%

------. 146 TABLE 8 - FLARE PLATFORM TYPE

PLAIN DIHEDRAL POINT N 102 7 2 111 91.89 6.31 1.80 100%

2 TABLE 9 - FLARE SHAPE (for flakes with area >90.00 mm )

PARALLEL CONVERGENT FAN DIVERGENT NO N SIDES SIDES 17 9 7 2 3 38 44.74 23.68 18.42 5.26 7.89 100%

2 TABLE 10 - BULB OF PERCUSSION (flakes with area >90.00 mm )

SALIENT DIFFUSE ABSENT N 27 10 1 38 71.05 26.32 2.63 100%

TABLE 11 - AMOUNT OF CORTEX ON FLARES* (with flake area >90.00 mm2)

o 1-25% 26-50% 51-75% 76-99% 100% N* 35 1 o 1 1 0 38 92.11 2.63 o 2.63 2.63 0 100% *note: it was difficult to distinguish cortex on the very small debitage but they (73) all appeared to be non-cortical flakes.

TABLE 12 - UTILIZATION ON FLARES

ABSENT PRESENT N 99 12 111 89.19 10.81 100% 147 TABLE 13 - INTENSITY OF UTILIZATION ON FLAKES

ABSENT ONE EDGE TWO EDGES N 99 12 o 111 89.19 10.81 o 100%

TABLE 14 - LOCATION OF RETOUCH ON SCRAPERS ONE EDGE TWO EDGES N 16 6 22 72.73 27.27 100%

INTERIOR EXTERIOR ALTERNATE N 9 9 4 22 40.91 40.91 18.18 100%

TABLE 15 - RATIO OF RETOUCH TO CIRCUMFERENCE OF SCRAPER

<50% 50% >50% N 14 1 7 22 63.64 4.55 31.82 100%

TABLE 16 - SCRAPER AND .BORER RETOUCH MORPHOLOGY IRREGULAR DENTICULATE STRAIGHT ND ON 25 9 2 4 40 62.50 22.50 5.00 10.00 100%

TABLE 17 - FREQUENCY OF FLAKING PATTERNS ON BONE CORES AND TOOLS

1a 1b 2 3 4a 4b polished N 5 2 3 6 14 3 7 40 12.50 5.00 7.50 15.00 35.00 7.50 17.50 100% 148 TABLE 18 - FLAKE LENGTH

MIN MAX MEO MEAN SO N 4.40 25.00 9.00 10.08 4.19 111

TABLE 19 - FLAKE WIDTH

MIN MAX MEO MEAN SO N 3.50 . 25.00 10.20 10.88 4.01 111

TABLE 20 - FLAKE THICKNESS

MIN MAX MEO MEAN SO N 1.00 9.60 3.00 3.61 1.80 111

TABLE 21 - TOOL LENGTH

MIN MAX MEO MEAN SO N 7.50 55.00 16.80 19.11 7.51 52

TABLE 22 - TOOL WIDTH

MIN MAX MEO MEAN SO N 6.30 35.00 11.65 13.45 5.06 52

TABLE 23 - TOOL THICKNESS

MIN MAX MEO MEAN SO N 2.70 14.50 6.50 7.20 3.21 52 ....t'Zj ~ '1 FLAKE AREA / FLAKE THICKNESS (I) .... (SALAWUSU SITE) ,j:o. 700 r------I ...... Hlt'Zj PI PI 600 r:~ A rn • PI

~ 500 ~ ...... ••...•.••••••.••••••••••••'> •••••••••••••••••••••••••• ·······················································1 ~ ....HI ~ A PI E r: 's400 U) A ....~ a:: A o « 9 ~ 300 (I) A A rn ~ rn LL A • t ttl.... 200 .A ...... ~ ...... o ...... ········i··············· .. rt HI ..... A. ~A A A A A.. A o 100 A '1 t t til l~t'j~~~*·· .::1. PI.... o I~------~------~------~------~------~------~ ~rn o 2 4 6 8 10 12 s:: FLAKE THICKNESS (mm) .... ,j:o. \0 150

o ON' .... .sE US 0«a: co~ a: ft 5 ~: ~a.. ~:

~

o C\I

o o o o o 8 8 8 o ,.... co v C") .... (zww) V3t:1V 3)I\fl.::l

Figure 15 - Flake area/platform area plot for Salawusu flakes.

------151

r---~----~----~--~----~------~--~ ~

(J) co (J) w ~ z ... o~ i

~a.

~ww) V3t:lV ~t:lO.::Il.Vld

Figure 16 - Platform area/flake thickness plot for Salawusu flakes. 152

~ 0 ~ Q ~ E i z tn W ~ ~ ~ ~ L&. W~ :;) III tn c~ :I: ~ ~ u.. Z !tn W ...I • U. 0 i5 z "~

S3E>\flN30~3d

Figure 17 - Length of Salawusu flakes and tools. 153

o... o S3E>V1N30}:l3d

Figure 18 - width of Sa.lawusu flakes and tools. 154

!) 0 ~ Q 'E Z E C z 0 ~ ~ ~ ::5 D: U. W ~ ::) In ~ ::) :::::::.. • en en ~ ~ w ~ ~ ~ u. 0 '0 -J: • 0 I- 0 LLI ~ (.) :z:- ~

S3E>V1N30~3d

Figure 19 - Thickness of Salawusu flakes and tools. 155

o tem L.'_.L----'- ....'

Figure 20: Tools recovered from Salawusu in 1980. Side Scrapers (A,B); gravers (C,D); circular scraper (E); flake core (F); Notch (G) 156

o 3Cm 1.... _...1.-_.1.----1'

L..---I__ J ICm

C

Figure 21: Tools recovered from Salawusu in 1980. Side scrapers (A,B); Borer (C) 157

polished

--- -

Ib

o 3Cm 1.... _"""-_"'--...... J'

Figure 22 - Worked bone recovered from Salawusu in 1980. (refer to text for description of flaking patterns by number) 158 Hlah

TOOIJlLANIC MANUFACTURE TOOL MAINTENANCE HIGH EXPORT RATE LOW DISCARD RATE HIGH CONSERVAtION

J • Selewusu B REOCCUPIED IITaI ~ 1------1 HIGH REUSEilCAYENGlNO RATE 1-_____ f

TOOUBLANK MANUFACTURE TOOL MAINTENANCE HIGH REJECTION RATE HIGH DISCARD RATE LOW CONSERVATION RATE

SITUAnOHAl SlTUAnOHAl REPAIR REPAIR RAW MATERIAL RAW MATERIAL AYAILABlE SCARCE

Low~ ______~ ______~ ______~ ____~~ Low %Late Stage Oebitage High

Assemblage formation model employIng debIt.age/t.ool ratio and lat.e st.age deb1t.age proporUon. (from Magne, 1989: 20)

RIp ...... IlEllDENCES ...... i ~ ------,------I IlAHUFACTURINQ lITES

• Sale\lLlSu

~~------~------~ %Late Stage LIthIc aa.e.blase tonaat.ioD .odel ..,lo,iO& div.rait, alope and proportion ot lat.e at.ase debIt.ace.

Figure 23 - Assemblage formation model (after Magne 1989: 20, 28) 159

CHAPTBR 5 THE SHIYU SITE, SHANXI PROVINCE Introduction

Forty years after the discovery of the Salawusu site, another important North China Paleolithic locality was uncovered about 450 km east of the boundary between the Ordos and the Loess Plateaux. Locality 63661, the Shiyu site, is located in Suoxian county (39°24'11"N, 122°21' 05"E) , northern Shanxi province (Jia et al. 1972). The Shiyu artifacts and associated faunal remains were clearly within the context of Upper Pleistocene loess deposits. The diminuitive size of the lithics and the dominance of tools made on flakes seemed to ally this assemblage closely with the cultural remains from Salawusu. While this site was excavated in 1963, the preliminary report of the excavation was delayed for nine years, eclipsed by the political publications which dominated the Chinese presses during the tumultuous early years of the Cultural Revolution. Pogroms by the Red Guard, directed toward the destruction of China's monumental dynastic period archaeological remains were a tragic by-product of this period's social upheaval and mass hysteria. Fortuitously, prehistoric relics were largely ignored as targets, and on a

------_... _-_. -- ... --.--.-.-~~~~= 160 positive note, this period saw the first applications of chronometric dating techniques employed to interpret Chinese Paleolithic assemblages (Olsen 1987). Shiyu was one of the initial assemblages for which radiocarbon dates were obtained. This chapter discusses the stratigraphy and chronology of the Shiyu site and presents an analysis of the Shiyu stone artifacts. This lithic analysis includes all the specimens housed and curated by the Institute of Vertebrate Paleontology and Paleoanthropology (IVPP) in Beijing. The site excavation, initial interpretations of the cultural remains, site stratigraphy, the faunal remains and their paleoenvironmental implications are discussed in the following section as backround to the lithic analysis. An interpretation of this lithic analysis and a comparison with the Salawusu assemblage concludes Chapter 5.

The 1963 Excavation of Locality 63661

In the early summer of 1963, a survey team from the Xi'an Mining Institute recognized animal fossil remains eroding from sediments near Shiyu Village in northern Shanxi province. In response to this initial discovery, Wang zeyi and You Yuzhu of the IVPP began a survey of this area and 161 subsequently discovered. Locality 63661. Two additional IVPP scientists, Jia Lanpo and Gai Pei, joined the excavation team and their work resulted in the recovery of this rich faunal and lithic collection. site stratigraphy

Shiyu village is located in the eastern foothills of Mt. Heituo at the mouth of the Sangan River. To the east of the village is a broad alluvial plain. Locality 63661 is situated at the confluence of two smaller rivers, Shiyu He and Xiaoquangou He. The sediments of sand, gravel and silt are typical of Loess Plateau stream deposits. The archaeological materials were recovered from the second terrace, 25-30 m above the river bed. The thick, loosely consolidated loess layers are easily eroded. Consequently, the deposit has been isolated as a small hill

2 with a total area of 1000 m • The excavation comprised 70 m3 (Jia 1980). The Upper Pleistocene stratigraphy is separated into four units (Figure 24). These are described from top to bottom as follows: unit 4 is 18m thick and composed of unconsolidated gray-yellow fine grained powdery sand with some occasional concentrations of small gravel. unit 3 is 162 8.9 m thick and composed of fine and coarse interbedded gray and white sands. Some faunal remains were recovered from the bottom of this layer and they appear to have been compressed. These specimens are mostly mammalian teeth and small bits of bone (Jia et al. 1972). The human fossil occipital fragment, most of the fauna and all of the stone artifacts were recovered from unit 2. The total thickness of this unit is 0.9-1.5 m. The sediments are mostly sand with some clay and there is a layer of ash approximately 2.5 cm thick. Some of the bone is burned. Quartzite cobbles are also found in this unit. The bottom unit (0.5-1 m thick) is composed of small gravel and sand and angular pieces (2-5 cm) of limestone, sandstone and quartzite. A layer of fine-grained sand separates Unit 1 and Unit 2. A Permian coal-series underlies the deposit (Jia 1980). This is a potential contaminant source of "old" carbon which could invalidate a C-14 determination.

Faunal Remains. Paleoenvironment and Chronology

A fragmentary human occipital was recovered from unit 2 at Shiyu. This specimen has been assigned as fully modern Homo sapiens. However, Jia (1980: 47) explains the

~---~- ~------~ 163 following two characteristics of this fragment as primitive traits: 1) The external occipital protruberance is higher than the internal occipital protruberance and 2) The cerebellar fossae are much shorter and narrower than the cerebral fossae. The faunal remains recovered from the second and third layers of locality 63661 include the following animals: ostrich, struthiolithus sp.; tiger, Felis (Panthera) tigris; hyaena, Crocuta sp.; red deer, Cervus elephus; giant deer, Megaloceros ordosianus; Tibetan gazelle, Gazella przewalskyi; goitered gazelle, Gazella subgutturosa; woolly rhinoceros, Coelodonta antiguitatis; horse, Eguus przewalskyii Asiatic wild ass, Eguus hemionusi and the bovids, Bubalus wansjocki and Bos primigenius. 40% of these are extinct species (Jia et ale 1972). Equids are a disproportionate number of the faunal remains. The bones are highly fragmentary but teeth are plentiful and at least 200 individual animals have been counted (Jia 1980). Over 400 deciduous equid teeth have also been recovered here. The high percentage of ungulates at this locality suggests a paleoenvironment of dry, cool, grassy steppe. Megaloceros, with its giant rack of antlers was presumably not a dense forest inhabitant, therefore the largest 164 vegetation was probably shrub or sparse woodland. The winter range of Cervus elephus today is restricted to areas with 20-30 cm of snow cover and hyaena and ostrich can both adapt to cooler temperatures. Bubalu~ may have been a seasonal inhabitant in a paleoenvironment with a large temperature difference between summer and winter (Jia et ale 1972). In 1977, two radiocarbon dates, 28,945 ± 1370 and 28,135 ± 1330, were generated for the Shiyu assemblage. Both dates are on Bubalus bone (ZK-109-01,02 Institute of Archaeology, Chinese Academy of social Sciences 1977). Some of the problems involved with radiocarbon dates determined on bone have already been discussed in Chapter 4 and the presence of underlying Permian coal bed at this site compounds these problems. An attempt to correlate the Shiyu stratigraphy with the

6180 record lends some credibility to the radiocarbon dates above. There is a lack of paleosol development and the sandy sediments suggest deposition during a glacial period, of which stage 2 is the most likely candidate given the late Upper Pleistocene faunal representatives. 165 The cultural Remains and Designation of the Small Tool Tradition

The 1972 publication by Professors Jia Lanpo, Gai Pei and You Yuzhu initially describing the Shiyu stone artifacts was a milestone in the history of Paleolithic archaeology in china. It was one of the earliest publications which acknowledged the importance of interpreting lithic variability within North Chinese stone tool assemblages. The stone artifacts from Locality 63661 were classified into the following types (Jia et ale 1972: 48-52): amorphous cores, flakes, blades, four types of points (separated on the basis of morphology and retouch pattern), single and double side scrapers, end scrapers and burins. Evidence of bipolar flaking was also apparent in the form of bipolar flakes and cores. Some "one of a kind" artifacts were also described as follows: a wedge-shaped or Gobi core, like those commonly found in terminal Pleistocene and Neolithic microlithic sites across North China: a small crescent­ shaped knife made of chalcedony with a tanged base opposite the finely flaked working edge and an arrowhead fashioned on a long thin flint flake with fine retouch on one edge of the tip and a tapered base. These last two artifacts are similar to those described by Alimen (1957) from the North 166 African Aterian assemblages. A broken ornament made on graphite which had been smoothed on one side and pierced at the center and a bone point fashioned from a long bone fragment were also recovered. Jia and Huang (pers. commun.) attribute some of the fine retouch to indirect or pressure flaking and Jia (1980) has proposed that bow and arrow technology for hunting horse was in use at this locality. Professors Jia, Gai and You's interpretation of the Shiyu lithics within the framework of North Chinese assemblages prompted them to write: At least two traditions are apparent in the development of China's Paleolithic cultures. One of them may be termed the Kehe-Dingcun series and is characterized by large chopper­ chopping tools made on flakes and heavy triangular points ••• The second tradition is referred to as the Zhoukoudian Locality 1- Shiyu series. This tradition was typified by the use of small, irregular flakes in the production of microlithic artifacts which exhibit meticulous retouch ••• The Zhoukoudian Locality l-Shiyu series is well represented in Middle and Upper Pleistocene contexts throughout North China and is the forerunner of the microlithic Mesolithic in the area (Jia et ale 1972: 54-55).

The following Shiyu lithic analysis and its comparison with the Salawusu 1980 assemblage is meant to contribute toward refining, quantifying and characterizing the variability within North Chinese Paleolithic assemblages. It tests the usefulness of the dichotomous small tool/large 167 tool model as it is currently used to interpret Chinese cultural remains in Upper Pleistocene stratigraphic context.

The Shiyu Lithic Analysis

Data for the following set of variables were collected for the Shiyu stone artifacts. (See Appendix A for a complete description of the coded variables) 1. Condition of specimen (whole or fragment) 2. Percussion axis length (mm) 3. width at approximate midpoint (mm) 4. Thickness at approximate midpoint (mm) 5. striking platform width 6. striking platform thickness 7. Artifact type 8. Retouch location 9. Appearance of the bulb of percussion on interior surface of the flake 10. Length of the line of retouch on the lateral edge(s) 11. Type of striking platform 12. Platform preparation 13. Raw material 14. Edge damage location 15. Percentage of cortex present 168 16. Number of exterior flake scars 17. Pattern of flake removal on exterior surface

Typology

In several aspects, this typology differs from that of Jia et ale 1972. For the purpose of suggesting a reduction sequence and to facilitate a comparison with Salawusu, the flakes are separated into ordinary, wide and elongate flakes on the basis of the ratio of flake length to flake width. The specimens which were classified as bur ins in the early report are here classified as distal fragments of broken points. Denticulation is viewed as a retouch attribute (after Isaac 1977: 154) rather than distinguishing a tool class of "denticulates" as Jia and his colleagues have done. The blades are separated into flake-blades and blades on the basis of their symmetry in cross-section.

Artifact Classes

Flakes: The Shiyu un retouched flakes may be divided into three types on the basis of the ratio of flake percussion axis length to flake width. These divisions are as follows: Wide flakes - Those flakes with a length/width ratio of 169 <1.00. 29.5% of the whole unretouched flakes are wide flakes. Ordinary flakes - Those flakes with a length/width ratio of 1.00 - 1.39. 31.4% of the whole unretouched flakes are ordinary flakes. Elongate flakes - Those flakes with a length/width ratio of >1.39. 39.1% of the whole flakes are elongate flakes. Figure 25 presents the histograms of distribution of this index in support of this tripartite classification.

Bipolar Flakes: Flakes which have been produced by the bipolar flaking technique and exhibit characteristic crushing on both ends.

Blades: The Shiyu blades have the following characteristics: Their percussion axis length is more than twice the width. They have a regular shape and parallel lateral margins. In cross-section these artifacts exhibit an equilateral triangle or a quadrilateral with two equal sides depending on the number of exterior ridges. They may be retouched. The distal end of blades may be rectangular, convex or pointed.

------~..... ------~~~-~-=~- ..------_. _.- 170 Flake-Blades: These artifacts have the following characteristics: Their percussion axis length is more than twice the width as in blades but in cross-section these exhibit an asymetrical triangle or quadrilateral due to their irregular exterior ridges. Their non-parallel lateral margins give them a less regular shape than blades. They may be retouched. The distal end of flake-blades may be naturally pointed or rounded.

Points: Points are fashioned on wide or elongated flakes, blades or flake-blades. They have a point which has been shaped by retouch. This retouch may consist of small flake scars down one or both sides on the exterior, interior or alternate surface to taper the point. There is sometimes evidence of notching or roughening at the base and the striking platform may be trimmed. The pattern of flake scars on the exterior surface is most often convergent or opposing (bidirectional).

Side Scrapers: These artifacts are made on thick chunks or ordinary flakes. The lateral retouch may be single or double occurring on the exterior or interior surface. The retouch ranges from small fine shallow flake scars to semi 171 abrupt retouch and clear even denticulation.

Transverse Scrapers: These tools are all fashioned on wide rectangular flakes with denticulated retouch on the working edge opposite the platform of the flake.

Notches: These artifacts are scarce in the Shiyu assemblage. With the exception of two specimens made on complete flakes, they are all made on thick flake or chunk fragments. They are all of the clactonion type, one large notch made with a single blow of the hammerstone.

Examples of these artifact types are illustrated in Figures 26 - 29 and Table 24 presents their frequencies within this assemblage. Tables 25 - 30 present the descriptive statistics on the metric variables for each artifact class. Tables 31 - 41 present the results of the non-metric variables.

~ .. ~-~ --~-- -- ~------172 Discussion

Artifact Dimensions and Standardization of Form

An overall characterization of this lithic assemblage is one of unretouched flakes and tools made on elongated flakes. The trend toward elongation of flakes suggests that ·core preparation was an integral aspect of the flaking technology. While unretouched flakes are the most numerous artifacts, flake-blades, blades and points occur in significant numbers in this assemblage. The flake-blades have comparatively wide ranges in length and width and a moderate standard deviation for both mean length and mean width. In contrast, the narrow ranges of length and width for blades and their low standard deviation for mean length and width reflect their regularity of form. The histograms presented in Figures 30 - 32 illustrate these relationships. Points are variable in length but fairly standardized in width and thickness. Flake-blades, points and blades taken as a group show less variability in length, width and thickness than the other tool classes of transverse, side scrapers and notches. The mean striking platform width and thickness is very

______•••• '0" .,_._~~~, ,_.____ ••• 173 similar for flakes, flake-blades and blades as a group while points, transverse scrapers and side scrapers tend to have similar platform dimensions (Table 28). The mean dimensions of bipolar flakes are very similar to those of blades suggesting that the bipolar flaking technique was not necessarily used to flake the smallest raw material as a final reduction stage. It may have been a quick, relatively easy way of getting several sharp flakes at once and of utilizing an intractable raw material such as quartz. Table 29 shows that the length of the retouched edge or edges is relatively consistent for all tools, suggesting some standardization in their production or of available working edge. The breakage patterns for these artifacts are particularly interesting with the highest percentage of fragments occurring among the flake-blades, blades and points (Table 31). As discussed in Chapter 2, Odell (1988) has suggested that pointed flakes and blades were halfted and utilized as projectiles and were a quantitatively important aspect of Archaic hunting tool kits. It is possible that some of the elongated flakes, flake-blades and blades at Shiyu were half ted as expedient points when more carefully fashioned points were unavailable. Another 174 feature of these particular artifact types that lends credibility to this interpretation is the proportion of these specimens with roughening or notching near the proximal end of the flake. 44.9% of the whole and proximal fragments of flake-blades and 34.3% of the whole and proximal fragments of blades exhibited this type of modification.

Tool Typology and Functional Interpretations

One of the problems inherent in assigning typological categories to specific artifacts is that it biases their functional interpretation. I have used the classification categories of flake-blades, blades and points for much of the Shiyu assemblage and I have suggested their possible function as hafted projectiles. clearly, there is a range of variation in artifact form within these categories and there is no absolute certainty about this interpretation. Another functional interpretation for these artifact types must be considered. The alternate retouch pattern found on a significant number of the pointed specimens in this assemblage may be the result of their utilization as borers (Jelinek, pers. comm.). This interpretation would not be inconsistent with 175 their modification for hafting. Several of the specimens I have classified as flake-blades and blades fit within Bordes category of typical or atypical backed knives (Bordes Type List for the Upper Paleolithic, #36 and #37). Taken together, borers (percoirs) and backed knives could be interpreted as part of a hide-working tool kit. The contextual association of these artifacts at Shiyu is not supportive of other domestic activities. shiyu is more convincing as a horse-kill site. However, there are many variables involved in site formation and preservation and much of the Shiyu site remains unexcavated (Jia 1980). Future investigations emphasizing taphonomic processes might clarify the function of the stone assemblage at Shiyu.

Raw Material Use

Lithic raw material was readily available at this locality in the eastern foothills of Mt. Heituo. Numerous small streams and rivers carried cobbles of quartzite, limestone, chert and sandstone. Representatives of these raw materials were recovered from the site strata. Fine­ grained quartzite, chert and limestone were the most frequently utilized raw materials (Table 32). For the purpose of this analysis two groupings of 176 artifacts were done. Raw material was correlated with artifact class by typological category. Then, specimens were separated into the broader categories of flakes or tools on the basis of the presence or absence of retouch (Table 33). The percentages of tools made on chert and limestone are only slightly higher than flakes. Flint, a high quality and relatively scarce raw material, was used twice as much for tools and most often for points which required more control over shaping the artifact.

Flaking Techniques and Reduction Sequence

Impact features on the interior flake surface and the appearance of the bulb of percussion suggest that both hard hammer and soft hammer percussion may have been aspects of the flaking technology at Shiyu. Table 34 shows that over half of the whole specimens or proximal fragments exhibit a crushing or fracturing of the bulbar area and point of percussion. This is thought to be characteristic of flakes produced by hard hammer percussion (Cotterell and Kaminga 1987). In contrast, where this damage was not present the bulb was generally diffuse, a feature often attributed to soft hammer or billet flakes (Crabtree 1973). Lipping, also a feature usually attributed to billet flakes is not present

-----_. '- ,_ .. ----'".-~. '-.="~- 177 in any of these specimens. In addition, the Shiyu flakes do not exhibit pronounced rippling on the interior surface. These last two features however, should not be used as a hard hammer/soft hammer distinction since they are dependent on the flaking properties of the raw material and probably the type of soft hammer used (ie. wood vs. bone vs. "soft" stone like sandstone). While most of the striking platforms are un faceted , the percentage of dihedral platforms is highest among the blades (Table 35). cortical platforms, indicating an early stage of reduction, are highest among the side scrapers and among wide and ordinary flakes on which scrapers were mos·t likely made (70% of the flakes with cortical platforms are wide or ordinary). Less than 5% of the platforms have been trimmed, but these all occur on points. Approximately 43% of the artifacts exhibit platform preparation or modification in the form of flake scars on the exterior proximal edge to recess the platform. Half of the blades and over 75% of the points have this feature (Table 36). These high percentages are consistent with core preparation for the production of blades and elongated flakes. In the case of points, the modification may have served to taper the base to facilitate halfting. For blade manufacture, recessing the platform (a feature of punctiform 178 platforms) offers more control over the shape of flakes produced (Jelinek, pers. commun.). The number of exterior flake scars is also consistent with the pattern of blade core reduction. The average number of flake scars for all the artifact types is 3, with the exception of transverse scrapers which average 4. Flake-blades and blades tend to have 2, 3 or 4 exterior flake scars while flakes and points are more variable (Table

37) • There is very little cortex present on these specimens (Table 39) and the 16% with some trace of cortex are mostly flakes, scrapers and points with cortical platforms as discussed above. The initial step in the reduction sequence is thus not well represented here. The larger wide and ordinary flakes and flake-blades, some retouched into scrapers and points, may represent an early stage in the reduction sequence. Elongated flakes and blades would represent a later stage of reduction.

Retouch and Edge Damage

Almost half (48%) of the Shiyu artifacts exhibit some retouch. The type of retouch ranges from small fine flake scars to more abrupt intrusive scars to the clear, fairly

------.------179 even denticulated edge on transverse scrapers. However, the small fine retouch is not of the consistently parallel type which would indicate pressure flaking. Characteristic small flat indentations with negative bulbar areas are not apparent. There is no strong evidence at this site for heat treatment of chert and flint, which would make them more brittle and easier to pressure flake. Only one flint specimen out of the entire assemblage shows some evidence of possible heating in the form of pot-lid fractures. The patterns of retouch on specific artifacts shown in Table 40 and the percentages of artifact types present, argues against this assemblage being mostly debris from a stone workshop, as Aigner (1981:227) suggests in her brief reference to this collection. Points are the most intensely retouched tools and the pattern of alternate retouch is most often found on these implements. Side scrapers tend to be retouched on only one exterior or interior edge. Flake­ blades and blades are moderately retouched. Side-scrapers are mostly fashioned by retouch on a single interior or exterior edge by semi-abrupt flake scars. Table 41 reports the occurrence of edge damage in the form of small flake scars or abrasion. This flaking does not look like intentional retouch. The "broken tip" category refers to the loss of just the very end of the

------_..... ---. _.. -.-., .. ~.~------~-._--- --. ------. 180 specimen. It is impossible to distinguish the cause of this edge damage, an attempt to do so would be speculative. The patterns are reported here to make the data accessible for future studies. comparative Analysis Between the sites

Histograms of flake and tool distributions within length, width and thick.ness intervals for the Salawusu flakes and tools have been presented in Figures 17, 18 and 19. They clearly show two distinct populations of artifacts within this assemblage. Comparable histograms for the Shiyu flakes and tools are presented in Figures 33, 34 and 35. They illustrate that the Shiyu stone assemblage must, by contrast, be considered a homogeneous population on the basis of these three variables. There is very little variation in artifact length and thickness and even less variation in width between the Shiyu flakes and tools. This reflects the elongated flake/blade dominance of this assemblage. The descriptive statistics for the Shiyu flakes and tools are presented in Tables 42 and 43. The scattergrams in Figures 36, 37, 38 and 39 are comparative plots of length, width and thickness for the flakes and tools from 181 each site. They illustrate the size differences between the two artifact collections. There is very little overlap in each of these combined plots between site assemblages. The dimensional differences are especially dramatic for the flakes from each site. It is true that raw material size and availability may account for most of the difference in dimensions of artifacts between the two sites. However, there are clearly differences in technology and patterns of reduction and retouch as well. There are no cortical or prepared (recessed) platforms on the Salawusu lithics and there is no blade technology in that assemblage. By contrast, Shiyu has a high percentage (43%) of prepared platforms and is dominated by elongated flakes and blades. Early (but not the initial) reduction stages are represented by cortical platforms at Shiyu. The early stages of the reduction sequence are not represented at.Salawusu. The flaking technique at Salawusu may have been predominantly soft hammer. Antler billets have been recovered there and very few specimens (8, in the entire assemblage) show any impact damage on the bulbar area interior flake surface. 58% of the Shiyu artifacts show this impact damage associated with hard hammer percussion

... - .. _----_ .. _-_.- -_ ... _-- --- . 182 and there were more available suitable lithic hammerstones in the environs of locality 63661. Figures 40, 41 and 42 are scatter plots of flake area, platform area and flake thickness for the Shiyu flakes. These may be compared to similar plots for the Salawusu flakes in Figures 14, 15 and 16. The Shiyu flakes do not exhibit a bimodal distribution for platform size independent of flake area as do the Salawusu flakes. They are a continuous unimodal distribution with variable platform size. This may represent the predominance of flakes produced by hard hammer percussion at Shiyu. There are differences in raw material utilization between the two sites as well. The better lithic raw material at Salawusu seems to have been utilized for tool production while at Shiyu there is not much difference in raw material usage between flake and tool production. There are no intentionally shaped points at Salawusu, while at Shiyu 17% of the artifacts are classified as points or pointed implements. These exhibit patterned, often alternate, retouch and notching, roughening or trimming on the proximal end. Denticulated retouch is found on transverse scrapers, side scrapers, flake-blades, points and blades at Shiyu. There is less patterning to the retouch on Salawusu tools •

... .. _-_ .._--- . __._------183 summary and conclusions

Since the preliminary report on the Shiyu site, every reference to these artifacts has closely aligned this lithic collection with the Salawusu stone assemblage. This analysis shows that while both these assemblages are composed of relatively small flakes and flaked implements, they are significantly different in many respects and should not be so closely aligned. Overall size differences of the artifacts between sites may be due largely to the ready availability of raw material at Shiyu but, there are clear differences in flaking techniques and core reduction as well. There is more evidence for soft hammer percussion at Salawusu and a heavier reliance on bone as raw material for fashioning a tool kit. At Shiyu, an incipient blade industry is reflected in the evidence of platform preparation and the percentages of elongated flake-blade artifacts. There is no flake elongation at Salawusu and no blades among the Salawusu lithics. Shaped points also distinguish the Shiyu lithic assemblage. Both the Salawusu and Shiyu lithic assemblages have been cited as having technological roots to the microlithic industries that were widespread across China by the terminal

------184 Pleistocene (Gai 1985, 1991; Jia and Huang 1985; Chen 1984). This idea is discussed in the following chapter.

. .. _------185

" :'I:.... I',:. I,' .' ':(.1'

.' aD ", " ", " ," ", " ...... " .,' " ~,

" Shfyu River

Xfaoquengou hat Entrence

0.lIJ',:::.::":" "f' Powdery sand 1:::,:.: .. .1 Fine sand ~ Coarse sand ~ I . I . 1,1,1 Sand with some clay ITITIIJ.1.1. r.:-:7:1 Small gravel and sand l:..!.:..!.:J .. , ~.: ...... : Permian coal series [!] Stone artifacts [I] Mammalian fossils

Figure 24 - The Shiyu s.tratigraphic section (after Jia et al. 1972). ....~ IQ C 11 (I) LENGTH TO WIDTH RATIOS FOR SHIYU FLAKES 20 l\) U1

~ I-' PI ~ (I) 15 1--...... Mc>Sa ...... I-' ::s(I) IQ rt m <...... : Do Z~ 10 J-·················································vvvt··· ...... rt ::r W ~ ~ w rt a...... o (I! 51-···························k)M········ HIo 11 {Jl ::r.... ~ HI o !OfX! VVYJ !OcOC1 I-' PI .40-.59 .~.99 1.20-1.39 1.60-1.79 2.0-2.19 2.40+ ~ (I) FlAKE LENGTH/FLAKE WIDTH QNTERVALS IN mm) (I! ~WlDE FLAKES • ORDINARY FLAKES D ELONGATED FLAKES .... 0) 0'1 187

,, 2cm ,~' /--:'/.

Figure 26 - Flake-blades from Shiyu.

------_._-- 188

2cm

Figure 27 - Points from Shiyu.

------189

~/~ , ' , ' , , I , I~ I I

2cm

Figure 28 - Blades from. Shiyu.

~- -~-~--~~~-~-~ ------~.~------190

2cm 7

Figure 29 - Cores, scrapers and notches from Shiyu. 191

Data Tables ~or the Shiyu Lithic ADalysis TABLE 24 - ARTIFACT CLASS FEQUENCIES

CLASS FREQUENCY PERCENT Flakes 132 28.7 Flake-Blades 86 18.7 Bipolar Flakes 34 7.4 Points 79 17.2 Blades 52 11.3 Side Scrapers 32 7.0 Transverse Scrapers 18 3.9 Notches 10 2.2 Cores 11 2.4 Chunk Fragments 11 1.3 Total N 460 100.0 TABLES 25 - 30: Descriptive statistics of Metric Variables for Each Artifact Class (N = number of specimens in which the variable is measurable) TABLE 25 - ARTIFACT LENGTH (mm)

CLASS N MIN MAX MEDIAN MEAN SO Flakes 103 11.51 43.54 24.26 24.77 6.7'7 Flake-Blades 29 23.51 79.58 34.05 36.88 11. 72 Bipolar Flks. 34 17.64 56.46 26.64 27.87 8.35 Points 29 12.85 47.01 32.41 31.83 8.73 Blades 12 21.65 44.31 34.21 34.05 6.98 Side Scrapers 29 14.95 58.51 29.04 30.99 11.11 Tran.Scrapers 18 13.24 47.73 21.68 22.62 7.34 Notches 3 18.51 45.45 21.80 28.59 14.70 Cores 11 17.14 61.22 35.46 33.99 13.62

TABLE 26 - ARTIFACT WIDTH (mm)

CLASS N MIN MAX MEDIAN MEAN SO

Flakes 126 8.85 52.15 21.21 21.50 6.77 Flake-Blades 84 5.22 42.49 17.38 18.19 5.75 Bipolar Flks. 34 8.68 31.86 13.74 14.28 4.26 Points 72 6.32 29.13 14.11 17.12 4.81 Blades 52 8.69 27.08 15.37 15.59 3.67 Side Scrapers 32 11.58 40.32 25.95 26.36 7.35 Tran.scrapers 18 11.68 52.15 28.89 29.72 8.66 Notches 6 8.86 34.64 27.04 23.48 10.47 Cores 11 16.07 34.98 23.86 23.75 5.97 192 TABLE 27 - ARTIFACT THICKNESS (mm)

CLASS N MIN MAX MEDIAN MEAN SO

Flakes 131 1.54 17.19 5.36 5.73 2.64 Flake-Blades 85 1.90 10.45 4.56 4.87 1. 72 Bipolar Flks. 34 2.27 11.33 4.79 5.04 1.97 Points 79 3.04 10.09 5.31 5.55 1.60 Blades 52 2.23 12.48 4.77 5.28 2.09 Side Scrapers 32 2.78 13.51 6.83 7.72 3.14 Tran.Scrapers 18 4.00 10.87 6.81 7.10 2.05 Notches 9 2.21 15.46 5.02 7.38 4.47 Cores 11 6.27 27.46 10.74 12.88 5.91

TABLE 28 - STRIKING PLATFORM WIDTH AND THICKNESS (nun)

CLASS N MEAN MEDIAN SO MEAN MEDIAN SO WIDTH THICKNESS Flakes 111 11.99 11.47 6.39 4.07 3.55 2.35 Flake-Blades 48 11.43 10.85 5.54 4.04 3.99 1. 77 Blades 35 11.90 11.35 4.62 4.12 3.78 1.61 Points 23 17.33 18.26 7.56 5.83 6.16 2.29 Side Scrapers 26 18.73 20.34 10.41 5.76 5.41 2.69 Tran.Scrapers 12 20.24 19.96 6.43 6.28 6.29 2.17 Notches 2 13.77 6.74 6.51 6.60

TABLE 29 - MEAN LENGTH OF RETOUCH ON LATERAL EDGE(S) (nun)

CLASS N EDGE 1 SO N EDGE 2 SO

Flake-Blades 29 22.01 7.81 10 21.56 8.46 Points 69 22.21 8.23 44 19.53 6.60 Blades 20 17.85 4.25 6 19.97 4.88 Side Scrapers 31 24.20 8.11 7 24.30 7.80 Tran.Scrapers 18 24.06 5.57

TABLE 30 - MEAN WIDTH OF NOTCH

N MEAN WIDTH SO

10 14.48 6.61 193 TABLE 31 - COMPLETENESS OF ARTIFACT (Freq. of identifiable proximal or distal fragments)

WHOLE PROXIMAL DISTAL CLASS N % N % N % Flakes 103 78.0 8 6.1 16 12.1 Flake-Blades 28 35.6 21 24.4 29 33.7 Bipolar Flakes 34 100.0 Points 29 36.7 4 5.1 43 54.4 Blades 12 23.1 23 44.2 10 19.2 Side scrapers 26 81.3 4 3.1 Tran.Scrapers 13 72.2 1 5.6 Notch 2 20.0 1 10.0

TABLE 32 - RAW MATERIAL USE BY ARTIFACT CLASS (%)

A B C D E F G, H Flakes 3.8 49.2 3.0 28.8 9.9 2.3 2.3 0.7 Flake-Blades 4.7 51.2 9.3 23.2 10.5 0 0 1.1 Bipolar Flakes 8.8 17.7 5.9 26.5 8.8 0 29.4 2.9 Points 1.3 36.7 10.1 27.8 21.5 0 1.3 1.3 Blades 5.8 36.5 9.6 23.1 23.1 0 0 1.9 Side Scrapers 9.4 40.6 3.1 37.5 3.1 0 0 6.3 Tran.Scrapers 0 33.3 0 50.0 16.7 0 0 0 Notches 0 50.0 0 30.0 20.0 0 0 0 Cores 0 36.4 9.1 9.1 45.4 0 0 0 A = Coarse-grained Quartzite B = Fine-grained Quartzite C = Flint D = Chert E = Indurated Limestone F = Sandstone G = Quartz H = Chalcedony TABLE 33 - RAW MATERIAL USE FOR FLAKES VS. TOOLS (where flakes = all un retouched specimens and tools = those specimens with retouch) N FLAKES(%) N TOOLS(%) RAW MATERIAL B 111 59.4 76 40.6 C 9 32.1 19 67.9 D 61 46.6 70 53.4 E 28 46.7 32 53.3

~-- -- .-~------194 TABLE 34 - INTERIOR FLAKE SURFACE (appearance of bulb of percussion)

N %

Impact damage 149 57.5 Salient 3 1.2 Diffuse 103 39.8 Double 4 1.5 Total N 259

TABLE 35 - STRIKING PLATFORM TYPE BY ARTIFACT CLASS (% of whole specimens or proximal fragments)

Unfaceted Dihedral Cortical Trimmed

Flakes 67.6 10.8 21.6 0 Flake-Blades 72.3 17.1 10.6 0 Points 36.4 12.1 15.1 36.4 Blades 67.6 20.6 11.8 0 Side Scrapers 71.4 4.8 23.8 0 Tran.Scrapers 83.4 8.3 8.3 0 Total 65.3 12.7 17.4 4.6

TABLE 36 - % OF PREPARED OR MODIFIED PLATFORMS BY ARTIFACT CLASS

N %

Flakes 47 42.3 Flake-Blades 21 43.8 Points 16 76.2 Blades 17 50.0 Side Scrapers 4 15.4 Tran.Scrapers 4 33.3 Total 109 42.6% of total whole specimens or proximal fragments

.... - .. ----.. _--- 195 TABLE 37 - NUMBER OF EXTERIOR FLAKE SCARS BY ARTIFACT CLASS (Frequency)

number of scars: 1 2 3 4 5 6 7 8 Flakes 29 27 26 20 2 2 0 1 Flake-Blades 4 31 20 14 0 1 1 0 Bipolar Flakes 1 6 6 6 1 0 0 0 Points 11 16 35 9 1 0 0 0 Blades 2 16 19 9 2 0 0 0 Side scrapers 3 5 13 5 1 0 0 0 Tran.Scrapers 2 2 2 5 2 0 0 0 Notches 1 0 2 1 0 0 0 0

TABLE 38 - PATTERN OF EXTERIOR FLAKE REMOVAL BY ARTIFACT CLASS IN %

Simple Opposing Parallel Convgt. Radial Flakes 48.3 16.4 6.6 6.6 22.1 Flake-Blades 15.5 35.2 18.3 22.5 8.5 Bipolar Flakes 26.3 36.8 15.8 10.5 10.5 Points 16.7 18.1 2.8 47.2 15.3 Blades 4.2 14.6 66.7 8.3 6.2 Side Scrapers 25.0 17.9 7.1 7.1 42.9 Tran.Scrapers 30.8 15.4 7.7 0 46.1 Notches 50.0 25.0 0 0 25.0

TABLE 39 - % OF CORTEX

FREQUENCY % No Cortex 362 78.6 Trace-25% 73 15.9 26-50% 18 3.9 51-75% 4 .9 76-100% 3 .7

~.-.------196 TABLE 40 - RETOUCH LOCATION BY ARTIFACT CLASS (%) (total N=220; 48% of total count)

Flake-Blades Points Blades SideScprs. N % N % N % N %

One exterior 11 21.6 19 37.3 6 11.8 16 29.4 edge

Both exterior 6 20.7 16 55.2 3 10.3 4 13.8 edges

One interior 7 29.2 3 12.5 7 29.2 7 29.2 edge

Both interior 4 22.2 10 55.6 2 11.1 2 11.1 edges

Alternate 5 22.7 14 63.6 2 9.1 1 4.5

Shaped tip only 2 10.0 15 75.0 2 10.0 0

TABLE 41 - EDGE DAMAGE LOCATION (total N=224)

Interior Exterior Broken Tip N % N % N %

Flakes 31 38.3 35 46.7 10 14.7 Flake-Blades 12 14.8 13 17.3 26 38.2 Points 11 13.6 7 9.3 10 14.7 Blades 8 9.9 12 16.0 18 26.5 Bipolar Flakes 6 7.4 4 5.3 2 2.9 Side Scrapers 5 6.2 2 2.7 1 1.5 Tran.Scrapers 3 3.7 1 1.3 0 Notches 5 5.2 1 1.3 1 1.5 197 TABLE 42 - SHIYU FLAKES

MIN MAX MEAN MEDIAN SO N

LENGTH 11.51 79.48 27.18 26.11 9.26 152

WIDTH 5.22 50.52 19.14 18.03 7.07 227

THICKNESS 1.54 17.19 5.44 4.94 2.42 232

PLATWDTH 1.93 33.17 12.03 11.26 6.19 156 PLATTHCK 0.98 12.73 4.07 3.78 2.13 155

TABLE 43 - SHIYU TOOLS

MIN MAX MEAN MEDIAN SO N

LENGTH 12.85 58.51 29.86 28.40 9.58 106

WIDTH 6.32 52.15 20.00 18.73 7.56 197

THICKNESS 2.21 15.51 7.76 5.30 7.76 211 PLATWDTH 2.64 40.37 15.39 14.17 7.82 95

PLATTHCK 1.14 11.84 5.19 5.04 2.44 95

------_._--- .....~ \Qs:: 11 (I) LENGTH OF BLADES, POINTS AND FLAKE-BLADES w (SHIYU SITE) 0 35

~ ::s 30 \Q rt ::t 0 en 25 HI W D' C} I-' PI Slo (I) ~20 ...rn Z 'tJ ~ 15 .....0 ::s a: rt W rn Q.. 10 PI ::s Slo HI 5 I-' PI ~ (I) I 0 D' I-' 5-9.9 15-19.9 25-29.9 35-39.9 45-49.9 PI Slo (I) LENGTH (INTERVALS IN mm) rn • BLADES • POINTS I2B FLAKE-BLADES .... \0 Q) ....t'Zj "l ~ 11 WIDTH OF BLADES, POINTS AND FLAKE-BLADES (D (SHIYU SITE) W I-' 50

....~ Q, ::rrt 40 o HI en w ....tr PI Q, ~30 (D en ~ ... Z ItS W ....o 020 ::s a: rt W en n. ::sPI Q, 10 ....HI PI ~ (D I ....tr o PI 5-9.9 15-19.9 25-29.9 35-39.9 45-49.9 Q, (D .OJ WIDTH (INTERVALS IN mm) • BLADES • POINTS LiJ FLAKE-BLADES I-' \0 \0 200

~ ..... I .....

~ E CD CIJ 05 E w d> 0 Z 5 CD I ~ ~ 5 CD ~ u. r-.:I a: W Il2d " ~ Z CIJ e l-z en CD ~ to en Ii> W ~ •CIJ w U 0 - 5 J: CD CD ~ ~ •

...... I

o o ~ ~ @ ..... S3E>V1N30~3d

Figure 32 - Thickness of blades, points and flake-blades.

---_.. _-- 201

Figure 33 - Length of Shiyu flakes and tools.

------_.. , t'Zj.... \Q ~ WIDTH OF SHIYU FLAKES AND TOOLS 11 CD 35 w ~ 30 ....~ Do rt ::r en 25 0 W HI (!l en ::r.... ~20 ~ Z W HI.... o 15 S» ~ a: CD m W 0.. 10 ::sS» Do rt 0 5 ....0 .m o 5-9.9 15-19.9 25-29.9 35-35.9 45-49.9 55-59.9 WIDTH (INTERVALS IN mm)

• FLAKES D TOOLS

to.) o to.) ....~ ~ 11 THICKNESS OF SHIVU FLAKES AND TOOLS (I) 30 w U1 I 25 :f.... o 9 CJ) (I) m W20 rn (!) o HI en ~ :::r Z 15 .... W ~ o HI a: ..... W 10 PI a... ~ m PI ::s 5 fl, rto .....o rn o . 1-1.9 3-3.9 5-5.9 7-7.9 9-9.9 11-11.9 13-13.9 15+ THICKNESS (INTERVALS IN mm)

• FLAKES D TOOLS

f\) o w 204

0 I.l) U)w ~ ~ LL :J 0 > 0 "It -J: U) 0 0 Z o 0 E >- J: -I (J) J-- 0 (J) ~ w -4 a: (J) 0 => LL e(~ .... (J) 0 ..J D- O ....J: or- -0 ~ ~ ....J: 0 0 0 0 0 0 Zg co co "It C\I "W..- ..J (WW) H1DN3l

Figure 36 - Length/width plot for Salawusu and Shiyu flakes. 205

0 UJ (0 ..J 0 0 I- 0 :J 0 > If) -:r: UJ c Z 0 0 3: 0 >- E :E ~OO °I--- (J) :3 C')~ ~O 0 (J) LL o 00 ~O 0 e ; o~q ~, 0 ~ I- C\J c( '~I ~~~~ (J) 0 o ~O :0 ..J . . ~ ~ 0. . ° 0 • ~~~~~ :r: @: : ~~ I- o~ ~ o : ~O ~ 0 C T"" , : 0 0: ~ ~ - . . '~ 3: <31~ ----:r: I- 0 Z 0 0 0 0 0 0 0 0 ,.... If) C") or- "W (0 v C\J ..J (ww) H.lE>N31

Figure 37 - Length/width plot for Salawusu and Shiyu tools. 206

0 C\I

::) > -::J: tn 0 C Z ct 10 ::) "'"" tn ::) 0

rn UJ ~ 00 -E ~ 0 E ~O ::::> ~ -Cf) >- 0 0 oCf) :i: a: en "'""W rn OW ... Z u.~ ::.::: rn ... () ~ ... ::5 :f 5 gu...... ~ ~ ... C. ::::>

en ~

Z 0 "w 0 0 0 0 ..J 8 co g 'I:t C\I "'"" (ww) H1E>N31

Figure 38 _ Length/thickness plot for Salawusu and Shiyu flakes. ....I'%j I.Q LENGTH/THICKNESS PLOT FOR SALAWUSU AND SHIYU ~ Ii (D TOOLS W \0 70 r1"tot o ID 60 o ::s o ..... I.Q A rtJr1".::r ...... r1" 50 ::r 00 0 0 .... o CO COo 0 0 o -E ::s::0;­ .§.40 00 ...... 8 0 0 OJ (D '0:1 888t6 0 0 .0 ...... rtJ ::r: rtJ t- 'tS..... ~30 0063,9 0 0<8 0,"'" 000 •• 0 W 0 o w ~ r1" ...J t)o t a .o()~o· A ~ o A 000 ~ t 8 0 o 20 o 0 CO Ii t ~ 0 03 <9 • "VA A t A AQ.. A d-51 A 0 {Jl PI A AJ' OA A A ..... 10 PI A A ~ rtJ ~ o ~I------~------~------~------~------~------~------~ PI 2 4 6 8 10 12 14 16 [ THICKNESS (mm) {Jl ::r.... SALAWUSU TOOLS SHIYU TOOLS .It. 0 ~ l\) o -..J ....I'Jj ~ FLAKE AREA/PLATFORM AREA (I) (SHIYU SITE) ~ 3,~,~------~

....I'Jj PI ~ 2,500 (I) • PI t1 (I) 01 ...... 't:S.... f2,~ 01 E rt HI - o «W I. a: 1,500 ~ « • 01 t1 (I) • • 01 !:! t • • • • • 't:S.... ~ 1,000 • o . ... • • • rt 4· .~ ..... HI ... ~. • o _~If&. ..A. ..4· ...... •• • t1 500 ~).~ .~ • fIl . ::t ~a. •• • .... • II> ••• • '< s:: o ~I ______~ ______~ ______-L ______~ ______--J ....HI o 100 200 300 400 500 01 ~ PLATFORM AREA (mm2) (I) en t\) . o CD 209

It) T""

.-.. E ...... E Cf) Cf)w ... oZ T""D .. I ...... I- ... .. ~ ...... :5 ...... LL ...... ~'" ... -, ... .. "'-II "'-II .. 4...... ~ ...... It) ... tf..... ~ ... ..~· ...... 1 ... ~ ... .f.c .. ~ ...... 14 ......

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 It) 0 It) 0 It) ~- N N T"" T"" (c: ww) V'3l::lV' 3)l'v'l:J

Figure 41 - Flake area/flake thickness plot for Shiyu flakes. 210

r------~ C\I0

tn tn W Z o~ :I:- E I- ...... E W Cf) ~ Cf)w o~ :3~ T"""() LLfi) ... I ...... ~ f- --~ ....

5c......

o o o 0 o o o o o 0 o 1.0 v M C\I T""" ~WW) V3l:1V V\Il:10::l1 Vld

Figure 42 - Platform area/flake thickness plot for Shiyu flakes

------211

CllAP'l'BR , CONCLUSIONS

The archaeological assemblages from Salawusu and Shiyu have provided valuable information on Upper Pleistocene human cultural development. These data have also served to highlight some important topics in Asian Paleolithic archaeology, including the nature and degree of lithic variability and the difficulties involved in arriving at a chronology for these assemblages. The historical foundation for this study was established by reviewing the most popular current model of cultural traditions used to explain the diversity in northern Chinese Paleolithic assemblages. This model regards the Salawusu and Shiyu assemblages as representative of a "small tool tradition" which is thought to be antecedent to the terminal Pleistocene microlithic industries of North China. However, this study has documented quantitative and qualitative differences between these lithic assemblages which call into question the closeness of their relationship. scatterplots have shown that there is a significant difference in overall artifact size between the two sites. Interassemblage variability is greater for the Salawusu

------212 assemblage where tools and flakes separate into two distinct populations on the basis of artifact dimensions and raw material utilization. The Shiyu lithics are a more homogeneous group with no apparent separation of flakes and tools on the basis of artifact dimensions. This represents the dominance, in the Shiyu assemblage, of elongated flakes and blades and tools fashioned on them. Technological attributes, such as prepared or modified platforms, patterned retouch on specific artifact types and the use of hard hammer versus soft hammer percussion, also accentuate the differences between these two assemblages. How might these analytical differences be explained and what do they mean in terms of the behavior of early humans in Asia?

Chronology, Typology and Hominid Tool Makers

On the basis of the radiocarbon dates reported for both sites, the oldest potential date for Shiyu subtracted from the youngest for Salawusu, could bring these sites as close in age as 3,125 years. This is unlikely however, given the acknowledged difficulties involved in radiocarbon dating of bone and the geological evidence as well. The two dates for Shiyu show no significant difference

------~.-~... - .. -.. -.-~--~~=== • 213 (28,945 ± 1370 BP and 28,135 ± 1330 BP) and the site stratigraphy correlates well with 6180 stage 2. However, only one radiocarbon date (35,340 ± 1900 BP) has been determined for Salawusu. This date has been called into question by Zheng (1988) who found that the organic remains

in the same layer as the C-14 dated mammalian bone were

beyond the limits of radiocarbon dating. Zheng's (1988) TL date on the layer above the cultural material recovered in

1980 was 174,000 ± 14,000 BP. The interpretation of the Salawusu section proposed in

this study correlates the 1980 lithic and bone artifacts from the Lower Salawusu Formation with 6180 stage 5. These

materials could thus be 75,000 - 130,000 years old, a conservative estimate in light of the above information. In this context, if we examine the general features of the

lithic and bone artifacts from the 1980 excavations and the human fossils at Salawusu, some of the inter-site differences are illuminated.

Salawusu has yielded 23 hominid fossil fragments (see Table 1) although the provenance is unclear for many of these specimens. The cranial fragments, which usually contain the most useful features for a taxonomic assignment, appear to be from adolescent individuals, which makes it difficult to assess the degree to which anatomical features 214 might be regarded as primitive or archaic in these specimens. Shiyu has yielded only one fragmentary occipital and this has been assigned to anatomically modern HQmQ sapiens. There is enough uncertainty in the dating of Salawusu and in assigning a fully modern status to the Ordos fossil hominids that it is possible that the Salawusu assemblage might have been produced by archaic Homo sapiens in the early Upper Pleistocene. Variables such as raw material availability, paleoenvironment and preservation of cultural remains dramatically affect archaeological assemblage composition. Because of the differential control of these variables, we cannot "translate" the biological organism (hominid) directly into technological terms. r have previously stated that there is very little technological congruence between early Upper Pleistocene industries in China and the western Old World. There is also very little homogeneity among early Upper Pleistocene archaeological assemblages in northern China, which makes it impossible to define the "Middle Paleolithic" as a distinct archaeological entity (see Qiu 1985). The Salawusu lithics show less standardization in form and less patterned retouch than the Shiyu lithics. No platform preparation or evidence of blade technology is 215 present. The Sa1awusu 1ithics are very small, but smaller does not necessarily mean that they are technologically advanced. The size and scarcity of raw material and the distance from a lithic source were most likely the dominant factors controlling tool size. The same flaking technology that was applied to stone at Sa1awusu was applied to the more accessible resource of animal bone. However, only expedient bone tools were found, not the finely crafted bone needles or ornamental worked bone of the late Upper Pleistocene. There is growing archaeological and osteological evidence from Europe and the Levant that archaic populations of Homo sapiens coexisted with anatomically modern populations of Homo sapiens sapiens (stringer et a1. 1984; McBrearty 1990; Trinkhaus and Smith 1985). However, a recent study integrating the chronological, cultural and environmental evidence for Middle Paleolithic hominids in Southwest Asia illustrates the questions that remain in the interpretation of the course of human development at this time in this relatively well-known region (Jelinek 1990). Most of the published interpretations of the Chinese fossil hominid evidence for the course of human evolution favor a model of in situ evolution from Homo erectus to anatomically modern Homo sapiens sapiens (Wo1poff et a1. 216 1984; Wolpoff 1985; Wu 1961; wu and Wu 1985). Paleolithic studies in Asia would be well served by examining the cultural and biological evidence from a more integrated approach which is especially critical of chronology and the geological evidence. Differences in raw material accessibility, task variability, different.ial preservation and natural sorting of artifacts are in part, responsible for the inhomogeneity of early Upper Pleistocene archaeological assemblages in North China.

Inter-site variabillity as a Reflection of Hominid Activities

The nearest presently known source of lithic raw material is 40 km from Salawusu in upland Middle Pleistocene pebble deposits. stone accessibility was therefore not the determinant for this locality. The full reduction sequence in lithic manufacture is not in evidence at Salawusu; there are very few cortical flakes and very few cores. The stone tools most likely represent implements which could not be utilized or sharpened further. The intensity of lithic raw material use is supported by the fact that some implements have been retouched on a broken edge. It appears that the initial reduction of the raw material was done elsewhere and 217 that the larger flakes and tools may have been further conserved or exported from the site. Animal bone, on the other hand, is plentiful at Salawusu. The faunal remains recovered from the 1980 excavations, as well as from nearby localities, indicate that the paleoenvironment was capable of supporting a variety of mammalian species. The Salawusu hominids may have utilized animal bone for expedient tools and antler as billets. While there are marks on some of the long bone fragments that may represent butchering, these have not been studied in detail and carnivore activity cannot be ruled out. There are no lithic implements (i.e. spear points) recovered from the 1980 excavations to support the interpretation of hunting activities at this site and there are no elongated flakes or blades which might have been hafted as spear points. The Shiyu lithics are a relatively homogeneous assemblage, at least based on artifact dimensions. There is clear evidence of platform preparation which accompanies the trend of flake elongation and blade production. Points with tips shaped by secondary retouch are a significant part of this assemblage. Roughening or notching near the base of 45% of the flake-blades and 34% of the blades has been interpreted as potential modification to facilitate hafting. 218 The mean thickness in mm of the Shiyu flakes, f1ake­ blades, blades and points are 5.73, 4.87, 5.28 and 5.55 respectively. These have been compared with m~an thicknesses of an ethnographic sample of arrowheads and spear points reported by Thomas (1978). Thomas's sample of arrowheads had a mean thickness of 4.0 mm while the spear points averaged 4.9 Mm. They have also been compared with an experimentally produced sample of unretouched flakes utilized as arrowheads and spear points by Odell and Cowen (1986). Their arrowheads averaged 5.5 mm in thickness and the spear points, 7.3 Mm. The Shiyu artifacts fall within Thomas's range as spear points but within Odell and Cowen's range for unretouched flakes utilized as arrowheads. However, this does not rule out the possibility that these were neither spear points nor arrowheads but pointed implements, hafted and utilized as borers. More information on the context and depositional history of these implements, perhaps from future excavations, might resolve this issue. In the preliminary Shiyu site report and in subsequent references to the site, the quantity of horse teeth and bone has been interpreted as representing the prey of the hominids who occupied the locality (Jia et ale 1972: Jia 1980). The question of horse hunting by the inhabitants of

------219 Shiyu remains controversial. This conflicting evidence does not help resolve the question of hunting technology at Shiyu but it does offer a direction for additional research applicable to other Northern Chinese late Upper Pleistocene lithic assemblages with reported projectiles such as Xiaonanhai, Xiachuan and Xueguan (Chen and Wang 1989; Jia and Huang 1985). Since the quantities of stone artifacts recovered are a function of the excavation strategy, it is difficult to say anything about the duration or permanence of occupation of these sites from the lithic analysis. At both localities, almost all of the cultural materials come from one layer. This may represent an ephemeral occupation by small mobile groups. There is no evidence to suggest repeated occupation of either site over a long temporal span.

Technological Antecedents

The value of lithic analysis of these materials is that it provides quantitative means for describing these assemblages which can then be used in comparative studies. It offers a parsimonius way to approach the developmental origins of certain technologies by quantifying lithic attributes within an assemblage rather than relying on the 220 presence of one or two specimens to draw conclusions about technological trends. Microlithic industries are distributed across North China by the end of the Pleistocene and there has been much debate over the roots of this technoloqy (Gai 1985; Chen and Wang 1989). Most recently, Professor Gai (1991) has suggested a series of four technological conditions which must be satisfied before an assemblage may be considered a true typo-technological antecedent to microblade production. These are 1) small sized artifacts 2) the presence of bifacially retouched tools 3) evidence of core preparation and 4) end-products of production such as bipolar flakes, burin spalls or small elongated flakes. This may be a reasonable list of technological features for immediate predecessors of terminal Pleistocene industries, however it is important to realize that this technological development was a gradual process. These features did not all appear at once. According to the results of this analysis, the Salawusu artifacts fulfill only the first of these criteria and raw material accessibility plays a critical role in this. The Shiyu assemblage does have evidence of platform preparation and production of elongated flakes, but the retouch is confined to the edges of tools as well as shaping or 221 resharpeninq the tips of points. If we recoqnize that the basic element of microlithic industries is the microblade, the presence of blades in siqnificant quantities at Shiyu may well reflect an initial essential first step on the way to those later industries. As blade production became prevalent, it miqht have been applied to different raw materials of decreasinq size until the refinement of the microblade core and microblade production was attained. The presence of an industry with flake-blades or backed knives and pointed implements in early-mid stage 2 in North China shows interesting parallels to western Eurasia. This may reflect anatomically modern Homo sapiens similar solutions to coping with rigorous conditions by manufacturing better hide clothinq and shelter.

------.--- 222 Appendix A

LIST OF VARIABLES

ION - artifact number

OEG - degree of specimen present

1. complete flake or tool 2. proximal fragment 3. distal fragment 4. mid fragment 5. chunk fragment 6. split, mid-line vertically

AL - percussion axis length in mm

AW - width at approximate midpoint in mm

AT - thickness at approximate midpoint in mm

PW - striking platform width

PT - striking platform thickness

ART - artifact type

1. flake 2. flake-blade 3. bipolar flake 4. point 6. blade 7. core 8. side scraper 9. circular scraper 10. transverse scraper 11. notch 14. borer 15. graver

RET - retouch location

1. one exterior edge 2. both exterior edges 3. one interior edge 4. both interior edges 223 5. alternate retouch 6. shaped tip with shallow retouch 7. retouch on distal exterior edge 8. retouch on distal interior edge 9. retouch on all exterior edges 10. retouch on all interior edges 11. no retouch

BUB - appearance of bulb of percussion on interior surface of flake

1. impact damage 2. salient 3. diffuse 4. double

LR 1 - Length of the line of retouch on one edge (mm)

LR 2 - Length of the line of retouch on second edge (mm)

PLAT - type of striking platform

1. un faceted 2. dihedral 3. cortical 4. point 5. removed/trimmed

PREP - preparation of the platform

1. yes 2. no

RAW - raw material type

1. quartzite (coarse) 2. quartzite (fine) 3. flint 4. chert 5. limestone (indurated) 6. sandstone 8. quartz 9. chalcedony 224 EDG - edge damage 1. on interior edge 2. on exterior edge 3. none 4. broken tip or small scars on tip 5. damage on interior and exterior edge

CRTX - percentage of cortex present

1- 0 2. trace-25% 3. 26-50% 4. 51-75% 5. 76-100%

FS - number of exterior flake scars (0 - 8)

FD - pattern of flake removal on exterior surface 3. simple (same platform) 4. opposed scars 5. transverse scars 6. parallel scars (same platform) 7. convergent scars (same platform) 8. sub-radial and radial scar pattern

Note: The raw data are available on disk from the author by request. 225 .eferenoe. Aiqner, J. 1981 Archaeological Remains in Pleistocene China. Forschunqen zur Allqemeinen und Verqleichrende Archaoloqie, Band 1. Verlaq C.H. Beck, Munich. Aitken, M.J. 1974 Physics and Archaeology, Second Edition. Oxford University Press, Oxford. Akazawa, T., S. Oda, and I. Yamanaka 1980 The Japanese Palaeolithic: A Techno-Typological Study. Rippu Shobo Pub. Co. Ltd., Tokyo. Alimen, H. 1957 The prehistory of Africa. Translated by A.H. Brodrick. Hutchinson, London. An Zhimin 1965 Trail Excavations of the Palaeolithic Cave at Hsiao-nan-hai in Anyang, Honan. Acta Archaeologica Sinica 1:1-28.

An Zhisheng and Lu Yanzhou 1984 Climo-stratigraphic Division in the Late Pleistocene Malan Period in North China. Kexue Tongbao 4:228-230. Anderson, T.F. 1990 Temperature from Oxygen Isotope Ratios. In Paleobiology, edited by D.E.G. Briqgs and P.R. Crowther, pp.403-406. Blackwell Scientific Publications, Oxford. Bamforth, D.B. 1986 Technological Efficiency and Tool Curation. American Antiquity 51:38-50. Behrensmeyer, A.K. and D. Dechant Boaz 1980 The Recent Bon.es of Amboseli National Park, Kenya in Relation to East African Paleoecology. In Fossils in the Making, edited by A.K. Behrensmeyer and A.P. Hill, pp.72-92. University of Chicago Press, Chicago. Behrensmeyer, A.K. and A.P. Hill (editors) 1980 Fossils in the Making. University of Chicago Press, Chicago. 226 Berger, G.W. and D.J. Huntley 1980 Thermoluminescence Dating of Terrigenous Sediments. Proceedings of the Second seminar on Thermoluminescence Dating. PACT, Oxford. Bergman, C.A. and M.H. Newcomer 1983 Flint Arrowhead Breakage: Examples from Rsar Akil, Lebanon. Journal of Field Archaeology 10:238-243. Binford, L.R. 1962 Archaeology as Anthropology. American Antiquity 28: 217-225. 1973 Interassemblage Variability: The Mousterian and the "Functional" Argument. In The Explanation of Culture Change, edited by C. Renfrew, pp.227-254. Duckworth, London. 1977 Forty-Seven Trips: A Case Study in the Character of Archaeological Formation Processes. In stone Tools as Cultural Markers: Change. Evolution and Complexity, edited by R. Wright, pp.24-36. Australian Institute of Aboriginal studies, Canberra. 1979 Organization and Formation Processes: Looking a Curated Technologies. Journal of Anthropological Research 35:255-273. 1981 Bones: Ancient Men and Modern Myth. Academic Press, New York. Binford, L.R. and S.R. Binford 1966 A Preliminary Analysis of Functional Variability in the Mousterian of Levallois Facies. American Anthropologist 68:238-295. Binford, L.R. and Chuan Run Ho 1985 Taphonomy at a Distance: Zhoukoudian, "The Cave Home of Beijing Man." Current Anthropology 26:413-442. Binford, L.R. and N. stone 1986 Zhoukoudian: A Closer Look. Current Anthropology 27(5):453-475. Birks, H.J.B. and H.H. Birks 1980 Quaternary Palaeoecology. Edward Arnold, London. 227 Birks, H.J.B., T. Webb and A.A. Berti 1975 Numerical Analysis of Surface Samples from Central Canada: A Comparison of Methods. Review of Palaeobotony and Palynology 20:133-169. Bleed, P. 1986 The Optimal Design of Hunting Weapons: Maintainability of Reliability. American Antiquity 51: 737-747. Bodig, J. and B.A. Jayne 1982 Mechanics of WOod and Wood Composites. Van Nostrand Reinhold Company, New York. Bonnichsen, Robson 1979 Pleistocene Bone Technology in the Beringian Refugium. National Museum of Man Mercury Series Paper, Archaeological Survey of Canada Paper No. 89. National Museums of Canada, Ottawa. Bordes, F. 1961a Typologie du Paleolithigue Ancien et Moyen. Imprimeries Delmas, Bordeaux. 1961b Mousterian Cultures in France. Science 134:803-810. 1962 A Tale of Two Caves. Harper and ROw, New York. Bordes, F. and D. de Sonneville-Bordes 1970 The Significance of Variability in Paleolithic Assemblages. World Archaeology 2:61-73. Boriskovsky, P.I. 1966 Pervobytnoe Proshloe V'etnama. Nauka, Moscow. 1978 Some Problems of the Paleolithic of South and Southeast Asia. In Early Paleolithic in South and East Asia, edited by F. Ikawa-Smith, pp.87-95. Mouton Publishers, The Hague. Boule, M., H. Breuil, E. Licent and P. Teilhard 1928 Le Paleolithigye de la Chine. Archives de l'Institute de Paleontologie Humaine, Memoire 4. Masson et Cie, Paris. Boyle, K.V. 1990 Upper Palaeolithic Faunas from South-West France. BAR International Series 557, British Archaeological Reports, Oxford. 228 Breed, C., J.F. McCauley and M. Whitney 1989 Wind Erosion Forms. In Arid Zone Geomorphology, edited by o. Thomas, pp.284-307. Belhaven Press, London. Breui1, H. 1939 Bone and Antler Industry of the Choukoutien Sinanthropus Site. Pa1eontologica Sinica, n.s., 06(117). Briggs, O.J. and R.S. waters (editors) 1977 Studies in Quaternary Geomorphology. Proceedings of the VIth British-Polish Seminar sponsored by the Institute of British Geographers. University Press, Cambridge. Ceng Xiangwang 1983 Guangxi Baise Oiqu Xin Faxian de Jiushiqi. Shigian yanjiu 2:81-88. Chard, C. 1974 Northeast Asia in Prehistory. The University of Wisconsin Press, Madison. Chang, R.C. 1986 The Archaeology of Ancient China, Fourth Edition. Yale University Press, New Haven. Chen Chun 1984 The Microlithic in China. Journal of Anthropological Archaeology 3:79-115 Chen Chun and J.W. Olsen 1990 China at the Last Glacial Maximum. In The World at 18,000 B.P., volume 1, edited by o. Soffer and C. Gamble, pp.276-295, Unwin Hyman, London. Chen Chun and Wang Xiang Qian 1989 Upper Paleolithic Microblade Industries in North China and Their Relationships with Northeast Asia and North America. Arctic AnthrQPology 26(2):127-156. Clark, G.A. 1989 Romancing the Stones: Biases, Style and Lithics at LaRiera. In Alternatiye ApprQaches to Lithic Analysis, edited by D.O. Henry and G.H. Odell, pp.27-50. Archaeological Papers of the American Anthropological Association Series, Number 1. 229 Clark, G.A. and S. Yi 1983 Observations on the Lower Paleolithic of Northeast Asia. current Anthropology 24:181-202. Clark, J.D. 1954 The Prehistoric CUltures of the Horn of Africa. University Press, Cambridge. 1970 The prehistory of Africa. Praeger Publishers, Inc., New York. Clark, J.D. and K. Schick 1988 Context and content: Impressions of Palaeolithic Sites and Assemblages in the People's Republic of China. Journal of Human Evolution 17:439-448. Close, A. 1978 The Identification of Style in Lithic Artifacts. World Archaeology 10: 223-237. Conkey, M.W. 1987 Interpretive Problems in Hunter-Gatherer Regional Studies: Some Thoughts on the European Upper Paleolithic. In The Pleistocene Old World. Regional Perspectives, edited by o. Soffer, pp.63-77. Plenum Press, New York. Cotterell, B. and J. Kaminga 1987 The Formation of Flakes. American Antiquity 52:675-708. Crabtree, D. 1973 Comments on Lithic Technology and Experimental Archaeology. In Lithic Technology: Making and Using Stone Tools, edited by E. Swanson, pp.105-114. Mouton Publishers, The Hague. David, N. 1973 On Upper Paleolithic Society, Ecology and Technological Change: The Noaillan Case. In The Explanation of Cultural Change: Models in Prehistory, edited by C. Renfrew, pp.277-303. Duckworth, London. Davis, M.B. 1976 Pleistocene Biogeography of Temperate Deciduous Forests. Geosciences and Man 13:13-26. 230 Deetz, J. 1965 The Dynamics of stylistic Change in Ankara Ceramics. University of Illinois Press, Urbana. Derbyshire, E. 1983 On the Morphology, Sediments and Origin of the Loess Plateau. In ~-Geomorphology, edited by R. Gardner and H. Scoging, pp.172-194. Clarendon Press, Oxford. 1988 Granulometry and Fabric of Quaternary Silts from Eastern Asia. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol.I, edited by P. Whyte et Al., pp.215-245. Centre of Asian Studies, University of Hong Kong. Dibble, H.L. 1987 Reduction Sequences in the Manufacture of Mousterian Implements of France. In The Pleistocene Old World. Regional Perspectives, edited by o. Soffer, pp.33-45. Plenum Press, New York. 1988 Typological Aspects of Reduction and Intensity of utilization of Lithic Resources in the French Mousterian. In The Upper Pleistocene Prehistory of Western Eurasia, edited by H. Dibble and A. Montet­ White, pp.181-198. The University Museum of the University of Pennsylvania, Philadelphia. Dong Guangrong, Gao Shangyu and Li Baosheng 1981 New Materials of Human Fossils from the Ordos. Academia Sinica, Institute of Desert Research, Lanzhou, pp.1192-1194. Dumond, D.E. 1978 Alaska and the Northwest Coast. In Ancient Native Americans, edited by J.D. Jennings, pp.43-93. W.H. Freeman, San Francisco. Flannery, K.V. 1972 The Cultural Evolution of civilization. Annual Annual Review of Ecology and Systematics 3:339-428. Freeman, L.G. 1978 Mousterian Worked Bone from Cueva Morin (Santander, Spain): A Preliminary Description. In Views of the Past, edited by L.G. Freeman, pp.29-52. Mouton Publishers, The Hague. 231 1983 More on the Mousterian: Flaked Bone from Cueva Morin. current Anthropology 24(3):366-376. Frison, G.C. and G.M. Zeimens 1980 Bone projectile Points: An Addition to the Folsom Cultural Complex. American Antiquity 45:231-237. Gai Pei 1985 Microlithic Industries in China. In Palaeoanthropology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.225-241. Academic Press, Orlando. 1991 Microblade Tradition Around the Northern Pacific Rim: A Chinese Perspective. Paper Presented at the XIII INQUA Congress, August 1991. Beijing, China. Gamble, C. 1983 Caves and Fauna from Last Glacial Europe. In Animals and Archaeology, vol. I, edited by J. Clutton­ Brock and C. Grigson, pp.163-172. BAR International Series 163, British Archaeological Reports, Oxford. Gardner, R. and H. Scoging (editors) 1983 Mega-Geomorphology. Clarendon Press, Oxford. Gladikh, M.I., N.L. Kornietz and o. Soffer 1984 Mammoth-Bone Dwellings on the Russian Plain. Scientific American 251(5):164-175. Goudie, A.S. 1983 Environmental Change, 2nd edition. University Press, Oxford. Guthrie, R.D. 1983 Osseous projectile Points: Biological Considerations Affecting Raw Material Selection and Design Among Paleolithic and Paleoindian Peoples. In Animals and Archaeology, vol I., edited by J. Clutton-Brock and C. Grigson, pp.273-294. BAR International Series 163, British Archaeological Reports, Oxford. 1990 Frozen Fauna of the Mammoth Steppe. University of Chicago Press, Chicago. Gowlett, J. 1984 Ascent to Civilization. Alfred A. Knopf, Inc., New York.

------.. - ..-----. --- 232 Han Defeng and Xu Chunhua 1985 Pleistocene Mammalian Fauna of China. In Palaeoanthropology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.267-286. Academic Press, Orlando. Harrison, T. 1978 Present status and Problems for Paleolithic Studies in Borneo and Adjacent Islands. In Early Paleolithic in South and East Asia, edited by F. Ikawa-Smith, pp.37- 58. Mouton Publishers, The Hague. Hayden, B. and W.K. Hutchings 1989 Whither the Billet Flake. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin, pp.235-250. BAR International Series 528, British Archaeological Reports, Oxford. Henry, D., C.V. Haynes and B. Bradley 1976 Quantitative variations in Flaked stone Debitage. Plains Anthropologist 21(7):57-61. Hill, A. 1979 Disarticulation and Scattering of Mammal Skeletons. Paleobiology 5:261-274. Holdaway, S. 1989 Were There Hafted Projectile Points in the Mousterian? Journal of Field Archaeology 16:79-86. Huang Weiwen 1985 On the Stone Industry of Xiaochangliang. Renleixue Xuebao 4(4):301-307. 1987 Bifaces in China. Renleixue Xuebao 6(1):61-68. 1989 The Late Paleolithic of China. In Zhongguo Yuangu Renlei, edited by Wu Rukang et al., pp.220-244. Science Press, Beijing. Huang Wei wen, Leng Jian, Yuan Xiaofeng and Xie Guangmao 1990 Advanced Opinions on the Stratigraphy and Chronology of the Baise Stone Industry. Renleixue Xuebao 9(2):105-112. Huang Weiwen, J.W. Olsen, R.W. Reeves, S. Miller-Antonio and Lei Jiaqiang 1988 Xinjiang Talimu Pendi Nanyuan Xin Faxian de Shiqi. Renleixue Xuebao 7(4):294-301. 233 Huang Weiwen and Qi Guochin 1987 Preliminary Observation of Liangshan Paleolithic site. Renleixue xuebao 6(3):236-244. Huang Weiwen and Wei Qi 1981 The Salawusu Ordos Man and His Culture. In E'erdousi Wenwu Kaoqu Wenii, edited by The yikezhao League Cultural Relics Work Station, pp.24-32. Head Office for Inner Mongolian Agriculture and Animal Husbandry, Hohhot. Hutterer, K. 1977 Reinterpreting the Southeast Asian Paleolithic. In Sunda and Sahul: Prehistoric Studies in Southeast Asia. Melanesia. and Australia, edited by J. Allen, J. Golson and R. Jones, pp.31-71. Academic Press, London. 1985 The Pleistocene Archaeology of Southeast Asia in Regional Context. Modern Quaternary Research in Southeast Asia 9:1-25. Imbrie, J., J.D. Hays, D.G. Martinson, A. McIntyre, A.C Mix, J.J. Morley, N.G. Pisias, W.L. Prell and N.J. Shackelton 1984 The Oribital Theory of Pleistocene Climate: Support from a Revised Chronology of the Marine &180 Record. In Milankovitch and Climate, edited by A. Berger, J. Imbrie, J. Hays, G. Kukla and B. Saltzman, pp.269-305. D. Reidel Publishing Company, Dordrecht. Inskeep, R.R. 1988 Africa: Everything but Flint. In Non-Flint Stone Tools and the Palaeolithic Occupation of Britian, edited by R.J. MacRae and N. Moloney, pp.235-242. BAR British Series 189, British Archaeological Reports, Oxford. Isaac, G. Ll. 1977 Olorgesailie. University of Chicago Press, Chicago. International Union for Quaternary Research 1991 XIII International Congress Abstracts, August 2-9, Beijing, China. IVPP (editors) 1980 Atlas of Primitive Man in China. Science Press, Beijing. 234 Jelinek, A.J. 1976 Form, Function and Style in Lithic Analysis. In CUlture Change and Continuity. edited by C.E. Cleland, pp.19-33. Academic Press, New York. 1977 The Lower Paleolithic: CUrrent Evidence and Interpretations. Annual Review of Anthropology 6: 11-32. 1988 Technology, Typology and CUlture in the Middle Paleolithic. In Upper Pleistocene Prehistory of western Eurasia, edited by H. Dibble and A. Montet-White, pp.199-212. The university Museum of the University of pennsylvania, Philadelphia. 1990 Problems in the Chronology of the Middle Paleolithic and the First Appearance of Early Modern Homo sapiens in Southwest Asia. Paper prepared for The Evolution and Dispersal of Modern Humans in Asia, November 14-17, 1990. Tokyo. Jia Lanpo 1980 Early Man in China. Foreign Language Press, Beijing. Jia Lanpo, Gai Pei and Li Yanxian 1964 New Materials from the Shuidonggou Paleolithic site. vertebrata PalAsiatica 8(1):75-83. Jia Lanpo, Gai Pei and You Yuzhu 1972 Report of Excavation in Shiyu, Shanxi - A Paleolithic site. Acta Archaeologica Sinica 1:39-60. Jia Lanpo and Huang Weiwen 1985 The Late Palaeolithic of China. In Palaeoanthropology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.211-224. Academic Press, Orlando. Jia Lanpo and Wei Qi 1976 A Palaeolithic Site at Hsu-chia-yao in Yangkao County, Shansi Province. Acta Archaeologica Sinica 2: 97-114. Jia Lanpo, Wei Qi and Chen Chun 1982 The Lower Palaeolithic of China. Kokogaku Z~sshi 206: 36-41. 235 Jia Lanpo, Wei Qi and Li Chaorong 1979 Report on the Excavation of Hsuchiayao Man site in 1976. Vertebrata PalAsiatica 17(4):277-293. Jiang Dexin 1988 Quaternary Palynofloras and Paleoclimate of Qaidam Basin, Qinghai, China. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol.I, edited by P. Whyte et al., pp.571-578. Centre of Asian Studies, University of Hong Kong. Jones, P. 1982 The Berinsfield Quartzite Palaeoliths and their significance. In Palaeolithic Artifacts from Berinsfie1d, Oxfordshire, by R.J. MacRae. Oxoniensia 47:8-11. Katz, J.L. 1980 The Structure and Biomechanics of Bone. In The Mechanical Properties of Biological Materials. symposium of the society for Experimental Biology, 34. pp.137-168. Cambridge University Press, Cambridge. Kelly, R.L. 1988 The Three Sides of a Biface. American Antiquity 53: 717-734. Kelts, H., Chen Kezao, G. Lister, Yu Junqing, Gao Zhanghong, F. Niessen and G. Bonani 1989 Geological Fingerprints of Climate History: A Cooperative Study of Qinghai Lake, China. Eco10gae Geo1. He1v. 82:167-182 Kleindienst, M.R. 1961 variability within the Late Acheulian Assemblage in East Africa. South African Archaeological Bulletin 16: 35-52. Koz):owski, J.K. 1971 The Problem of the So-Called Ordos Culture in the Light of the Palaeolithic Finds from Northern China and Southern Mongolia. Folia ouaternaria 39:63-97. Kurten, B. 1968 Pleistocene Mammals of Europe. Weidenfe1d and Nicolson, London. 236 Laboratory of the Institute of Archaeology, Chinese Academy of Social Science 1977 Report on Carbon-14 Dates (4). Xaogu 3:200-204. Lambert, J., S. Simpson, C. Szpunar and J. Buikstra 1985 Bone Diagenesis and Dietary Analysis. Journal of Human Evolution 14:477-482. Laville, H., J.P. Rigaud and J. Sackett 1980 Rock Shelters of the Perigord: Geological Stratigraphy and Archaeological Succession. Academic Press, New York. Leakey, M.D. 1971 01duvai Gorge. Vol. 3. Excayations in Beds I and II. 1960-1963. cambridge University Press, Cambridge. 1975 Cultural Patterns in the Olduvai Sequence. In After the Australopithecines. edited by K.W. Butzer and G.Ll. Isaac, pp.477-493. Mouton Publishers, The Hague. Lian Zhaomei 1981 Tainanxian cai1iaoxi de Ren1ei Huashi. Guoli Taiwan Daxue Kaoqu Ren1eixue Kan 42:53-74. Licent, E.P., P. Tei1hard de Chardin and D. Black 1926 On a Presumable Pleistocene Human Tooth from the Sjara-osso-go1 Deposits. Bulletin of the Geological Society of China 5:285-290. Liu Dongsheng and Ding Meng1in 1984 The Characteristics and Evolution of the Palaeoenvironment of China Since the Late Tertiary. In The Evolution of the East Asian Environment, vol. I, edited by R.O. Whyte, pp.11-40. Centre of Asian Studies, University of Hong Kong. Liu Dongsheng et ale 1985 Loess and the Environment. Ocean Press, Beijing. Longacre, W.A. 1964 Archaeology as Anthropology: A Case Study. Science 144:1454-1455. Lowe, J.J. and M.J.C. Walker 1984 Reconstructing Quaternary Environments. Longman Inc., New York.

~ ~~---~-.------.._------.-----.---.-~ - -_. - 237 Magne, M.P.R. 1989 Lithic Reduction stages and Assemblage Formation Processes. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin. pp.15-32. BAR International Series 528, British Archaeological Reports, Oxford. MacRae, R.J. and N. Moloney (editors) 1988 Non-Flint Stone Tools and the Palaeolithic occupation of Britian. BAR British Series 189, British Archaeological Reports, Oxford. Maloney, N., C. Bergman, M. Newcomer and F. Wenban-smith 1988 Experimental Replications of Bifacial Implements using Bunter Quartzite Pebbles. In Non-Flint stone Tools and the Palaeolithic Occupation of Britian, edited by R.J. MacRae and N. Maloney, pp.25-48. BAR British Series 189, British Archaeological Reports, Oxford. Marks, A.E. 1988 The Curation of Stone Tools During the Upper Pleistocene: A View from the Central Negev, Israel. In Upper Pleistocene prehistory of western Eurasia, edited by H.L. Dibble and A. Montet-White, pp.275-286. The University Museum of the University of Pennsylvania, Philadelphia. Martin, P.S. and R.G. Klein (editors) 1984 Quaternary Extinctions. The University of Arizona Press, Tucson. Miller, G.J. 1973 A study of Cuts, Grooves and Other Marks on Recent and Fossil Bone: Weathering Cracks, Fractures, Splinters and Other Similar Natural Phenomena. In Lithic Technology, edited by E. Swanson, pp.211-228. Mouton Publishers, The Hague. Morlan, R.E. 1980 Taphonomy and Archaeology in the Upper Pleistocene of the Northern Yukon Territory: A Glimpse of the peopling of the New World. National Museum of Man Mercury Series, Archaeological Survey of Canada Paper No. 94. National Museums of Canada, Ottawa.

------238 Movius, H.L. 1944 Early Man and Pleistocene stratigraphy in Southern and Eastern Asia. Papers of the Peabody Museum of American Archaeology and Ethnology 19(3). Harvard University. 1949 Lower Paleolithic Archaeology in Southern Asia and the Far East. In Studies in Physical Anthropology 1. Early Man in the Far East, pp.17-82. Wister Institute Press, Philadelphia. Myers, T.P., M.R. Voorhies and R.G. Corner 1980 Spiral Fractures and Bone Pseudotools at Paleontological sites. American Antiquity 45(3): 483-502. Ningxia Hui Museum 1987 Preliminary Report on the Excavation of Shuidonggou Site in 1980. Odell, G.H. 1988 Addressing Prehistoric Hunting Practices Through stone Tool Analysis. American Anthropologist 90: 335-355. Odell, G.H. and F. Cowen 1986 Experiments with Spears and Arrows on Animal Targets. Journal of Field Archaeology 13:195-212. Olsen, J.W. 1987 The Practice of Archaeology in China Today. Antiquity 61:282-290. Olsen, J.W., R. Reeves, Huang Weiwen, S. Miller-Antonio and Lei Jiaqiang 1989 The University of Arizona--Chinese Academy of Sciences Joint Investigation of the Prehistory of Southern Xinjiang: Results of the 1986 and 1987 Campaigns. Central and Inner Asian Studies 3:57-82. Olsen, J.W. and s. Miller-Antonio in prep. The Palaeolithic in Southern China. Solicited article for a volume on the Palaeolithic in Southeast Asia, Cambridge University. 239 Pei Wenzhong 1931 Notice of the Discovery of Quartz and Other stone in the Lower Pleistocene Hominid-Bearing Sediments of the Choukoutien Cave Deposit. Bulletin of the Geological Society of China 11(2):109-146. Pei Wenzhong and Li Youheng 1964 Tentative Opinions on the Sjara-osso-gol Series. vertebrata Pal Asiatica 8(2):99-119. Pokotylo, D.L. and C.C. Hanks 1989 Variability in CUrated Lithic Technologies: An Ethnoarchaeological Case Study from the Mackenzie Basin, Northwest Territories, Canada. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Maudlin, pp.49-66. BAR International Series 528, British Archaeological Reports, Oxford. Pope, G.G. 1988 Current Issues in Far Eastern Paleoanthropology. In The Paleoenvironment of East Asia from the Mid­ Tertiary, vol.II, edited by P. Whyte et al., pp.1097- 1123. Centre of Asian Studies, University of Hong Kong. 1989 Bamboo and Human Evolution. Natural History 10: 49-57. Potts, R. 1988 Early Hominid Activity at Olduvai. Aldine de Gruyter, New York. Prevot, L. and J. Lucas 1990 Phosphate. In Paleobiology, edited by D. Briggs and P. Crowther, pp.256-257. Blackwell Scientific Publications, Oxford. Purdy, B. 1973 Fractures for the Archaeologist. In Lithic Technology: Making and Using stone Tools, edited by E. Swanson, pp.133-144. Mouton Publishers, The Hague. Qi Guochin 1975 Quaternary Mammalian Fossils from Salawusu River District, Nei Mongol. Vertebrata Pal Asiatica 13 (4) : 239-249. 240 Qui Zhonglang 1985 The Middle Palaeolithic of China. In Palaeoanthrgpology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.187-210. Academic Press, Orlando. Qui Zhonglang and Li Yanxian 1978 Chinese Palaeo,lithic Archaeology in the Past 26 Years. In Gurenlei Lunwenii (Collected Papers of Palaeoanthropology), edited by the IVPP, CAS, pp.43-66. Science Press, Beijing. Reid, I. and L.E. Frostick 1989 Channel Form, Flows and Sediments in Deserts. In Arid Zone Geomorphology, edited by D. Thomas, pp.117- 135. Belhaven Press, London. Roberts, N. 1984 Pleistocene Environments in Time and Space. In Hominid Evolution and Community Ecology, edited by R. Foley. Academic Press, London. Rolland, N. and H.L. Dibble 1990 A New Synthesis of Middle Palaeolithic Variability. American Antiquity 55(3):480-499. Runia, L.T. 1987 The Chemical Analysis of Prehistoric Bones. BAR International Series 363, British Archaeological Reports, Oxford. Sackett, J. 1973 Style, Function and Artifact variability in Paleolithic Assemblages. In The Explanation of Culture Change, edited by C. Renfrew, pp. 317-325. Duckworth, London. 1977 The Meaning of Style in Archaeology: A General Model. American Antiquity 42(3):369-380. 1982 Approaches to style in Lithic Archaeology. Journal of Anthropological Archaeology 1(1):59-112. 1986 Style, Function and Assemblage variability: A Reply to Binford. American Antiquity 51:628-634. Sarnthein, M. 1978 Sand Deserts During Glacial Maximum and Climatic optimum. Nature 272:43-46. 241 Semenov, S.A. 1964 Prehistoric Technology. Cory, Adams and MacKay, London. Shackelton, N.J. and N.D. Opdyke 1973 Oxygen Isotope and Paleomagnetic stratigraphy of Equatorial Pacific Core V28-238: o~gen Isotope Temperature and Ice Volumes on a 10 year and 106 Year Scale. Quaternary Research 3:39-55. Shea, J.J. 1988 Spear Points from the Middle Paleolithic of the Levant. Journal of Field Archaeology 15:441-450. 1990 A Further Note on Mousterian Spear Points. Journal of .Fie1d Archaeology 17:111-114. Sheppard, P.J. 1987 The Capsian of North Africa: Stylistic Vari.ation in Stone Tool Assemblages. BAR International Series 353, British Archaeological Reports, Oxford. Shipman, P. 1986 scavenging or Hunting in Early Hominids: Theoretical Framework and Tests. American Anthropologist 88(1):27-43. Shott, M. 1989 On Tool-Class Use Lives and the Formation of Archaeological Assemblages. American Antiquity 54:9-30. Singer, R. and J. Wymer 1982 The Middle Stone Age at Klasies River Mouth in South Africa. The University of Chicago Press, Chicago. Soffer, O. 1985 The Upper Paleolithic of the Central Russian Plain. Academic Press, Orlando. stringer, C.B., J.J. Hub1in and B. Vandermeersch 1984 The Origin of Anatomically Modern Humans in western Europe. In The Origin of Modern Humans, edited by F. smith and F. Spencer, pp.51-136. Alan R. Liss, New York. Stuart, A.J. 1982 Pleistocene Vertebrates in the British Isles. Longman, London.

------242 Sullivan, A.P. and K.C. Rozen 1985 Debitage Analysis and Archaeological Interpretation. American Antiquity 50:755-779. Teilhard de Chardin, P. 1926 Fossil Man in China and Mongolia. Natural History 26(3) :238-245. 1941 Early Man in China. Institut de Geo-Biologie Publication 7, pp.1-99. Peking. Thomas, D.S.G.(editor) 1989 Arid Zone Geomcrphology. Belhaven Press, London. Torrence, R. 1983 Time Budgeting and Hunter-Gatherer Technology. In Hunter-Gatherer Economy in Prehistory, edited by G. Bailey, pp.11-22. Cambridge University Press, New York. Toth, N. 1985 The Oldowan Reassessed: A Close Look at Early stone Artifacts. Journal of Archaeological Science 12:101- 120. Villa, P. 1983 Terra Amata and the Middle Pleistocene Archaeological Record of Southern France. University of California Press, Berkeley. Wang Kelu, Ding Menglin, yin Zhaomin and Liu Huimin 1988 The Loess in the Middle Reaches of the Yellow River of China and its palaeogeographic Environment. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol.I, edited by P. Whyte et al., pp.433-444. Centre of Asian Studies, University of Hong Kong. Wang Xianzeng 1984 The Palaeoenvironment of China from the Tertiary. In The Evolution of the East Asian Environment, vol. II, edited by R.O. Whyte, pp.472-482. Centre of Asian studies, University of Hong Kong. Wang Yuping 1957 A Preliminary Report on Archaeological Investigations in the Salawusu Valley, Yikezhao League, Wenwu Cankao Ziliao 4:22-25.

------243 Wang Yuping and J.W. Olsen 1988 Aspects of the Inner Mongolian Palaeolithic. In Palaeoanthropology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.243-258. Academic Press, Orlando. Watanabe, H. 1985 The Chopper-Chopping Tool Complex of Eastern Asia: An Ethnoarchaeological-Ecological Reexamination. Journal of Anthropological Archaeology 4:1-18. Watson, A. 1989a Windflow Characteristics and Aeolian Entrainement. In Arid Zone Geomorphology, edited by D. Thomas, pp.209-231. Belhaven Press, London. 1989b Desert Crusts and Rock Varnish. In Arid Zone Geomorphology, edited by D. Thomas, pp.25-55. Belhaven Press, London. Wei Qi 1985 Palaeoliths from the Lower Pleistocene of the Nihewan Beds in the Donggutuo site. Renleixue Xuebao 4(4):290-300. Weidenreich, F. 1935 The Sinanthropus Population of Choukoutien. Bulletin of the Geological Society of China 14:427-461. Wen Qizhong and Zheng Honghan 1988 Paleoclimatic Records in the Loess Sections of North Xinjiang, China. In Loess, edited by D.N. Eden and R.J. Furkert, pp.203-213. A.A. Balkema, Rotterdam. Wenban-Smith, F.F. 1989 The Use of Canonical Variables for Determination of Biface Manufacturing Technology at Boxgrove Lower Paleolithic Site and the Behavioural Implications of Technology. Journal of Archaeological Science 16(1):17-26. west, F.H. 1981 The Archaeology of Beringia. Columbia University Press, New York. White, L.A. 1949 The Science of Culture. Ferrar, Straus and Giroux, New York. 244 White, J.P. and J.F. O'Connell 1982 A Prehistory of Australia. New Guinea and Sahul. Academic Press, Sydney. Whyte, R.O. 1984 The Eyolution of the East Asian Enyironment, two volumes. Centre of Asian Studies, University of Hong Kong. Wiessner, P. 1983 Style and Social Information in Kalahari San projectile Points. American Antiquity 48:253-276. Wilson, K. and D.J.B. White 1986 The Anatomy of Wood: Its Diversity and Variability. Stobart and Son Ltd., London. Wintle, A.G. and D.J. Huntley 1982 Thermoluminescence Dating of Sediments. Quaternary Science Reviews 1:31-53. Wobst, M. 1977 stylistic Behavior and Information Exchange. In Papers for the Director, edited by C. Cleland, pp.317- 342. University of Michigan Museum of Anthropology, Anthropological Papers No. 61, Ann Arbor. Wolpoff, M.H., Wu Xinzhi and A.G. Thorne 1984 Modern Homo sapiens: A General Theory of Hominid Evolution Involving the Fossil Evidence from East Asia. In The Origin of Modern Humans, edited by F. Smith and F. Spencer, pp.411-484. Alan R. Liss, New York. Wu Maolin 1980 Human Fossils Discovered at Xujiayao Site in 1977. Vertebrata PalAsiatica 18(3):229-238. Wu Rukang 1958 Fossil Human Parietal Bone and Femur from Ordos, Inner Mongolia. Vertebrata PalAsiatica 2(4):208-212. Wu Rukang and Lin Shenglong 1983 Peking Man. Scientific American 248(6):78-86. Wu Rukang and J.W. Olsen (editors) 1985 Palaeoanthropo,logy and Palaeolithic Archaeology in the People's Republic of China. Academic Press, Orlando. 245 Wu Xinzhi and Zheng Zhenbiao 1985 Homo sapiens Remains from Late Palaeolithic and Neolithic China. In Palaeoanthrgpology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.107-133. Academic Press, Orlando. Xu Ren 1984 Changes of Vegetation in China since the Late Tertiary. In The Evolution of the East Asian Environment, vol.II, edited by R.O. Whyte, pp.426-432. Centre of Asian studies, University of Hong Kong. Yair, A. 1987 Environmental Effects of Loess Penetration Into the Northern Negev Desert. Journal of Arid Environments 13: 9-24. You Yuzhu, Tang Yingjun and Li Yi 1980 New Discovery of Palaeoliths in the Nihewan Formation. Quaternaria Sinica 5(1):1-13. Yuan Baoyin 1978 Sedimentary Environment and Stratigraphic Subdivisions of the Sjara-osso-gol Formation. ACTA Geologica Sinica 3:220-234. Zhang Lansheng 1988 The Trend Toward Dryness in North and western China Since the Mid-Pleistocene. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol.I, edited by P. Whyte et al., pp.445-452. Centre of Asian Studies, University of Hong Kong. Zhang Senshui 1985 The Early Palaeolithic of China. In Palaeoanthropology and Palaeolithic Archaeology in the People's Republic of China, edited by Wu Rukang and J.W. Olsen, pp.147-186. Academic Press, Orlando. Zhang Yongzu 1988 Preliminary Analysis of the Quaternary Zoogeography of China Based on the Distributional Phenomena Among Land Vertebrates. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol. II, edited by P. Whyte et Al., pp.883-896. 246 Zhao songqiao 1988 A Preliminary study on the Evolution of the Physical Regionalization of China Since the Early Tertiary. In The Palaeoenvironment of East Asia from the Mid-Tertiary, vol.I, edited by P. Whyte et al., pp.393-400. Centre of Asian Studies, University of Hong Kong. Zhao Songqiao and Xing Jiaming 1984 Origin and Development of the Shamo (sand deserts) and the ~ (stony deserts) of China. In The Evolution of the East Asian Environment, vol.I, pp.230-251. Centre of Asian Studies, University of Hong Kong. Zheng Honghan 1988 Late Pleistocene Aeolian Loess and Fluvio­ lacustrine Deposits in North China. In Loess, edited by D.N. Eden and R.J. Furkert, pp.233-245. A.A. Balkema, London. Zhou Kunshu 1984 Quaternary Succession of Vegetation and Climatic Evolution in North China. Scientia Geologica Sinica 2:165-172.