Biomechanical Strategies during Oldowan and Acheulean Stone Tool Production
by Erin Marie S. Williams
B.A. in Anthropology, 2000, Grinnell College M.A. in Anthropology, 2007, The George Washington University M.Phil in Hominid Paleobiology, 2008, The George Washington University
A dissertation submitted to
The Faculty of The Columbian College of Arts and Sciences of The George Washington University in partial fulfillment of the requirements for the degree of Doctor of Philosophy
May 15, 2011
Dissertation directed by
Alison S. Brooks Professor of Anthropology and Brian G. Richmond Associate Professor of Anthropology
The Columbian College of Arts and Sciences of The George Washington University certifies that Erin Marie S. Williams has passed the Final Examination for the degree of
Doctor of Philosophy as of February 28, 2011. This is the final and approved form of the dissertation.
Biomechanical Strategies during Oldowan and Acheulean Stone Tool Production
Erin Marie S. Williams
Dissertation Research Committee:
Alison S. Brooks, Professor of Anthropology, Dissertation Co-Director
Brian G. Richmond, Associate Professor of Anthropology, Dissertation
Co-Director
Peter W. Lucas, Professor of Anthropology, Committee Member
Adam D. Gordon, Assistant Professor of Anthropology, University at Albany—SUNY , Committee Member
ii
Dedication
This dissertation is dedicated to three sets of educators:
To my teachers and professors, starting with Geneva Ballard and Anne Miller in the nursery school at church, for my formal education,
To Bella-Boo, Charlie-Choo, John-J, Greggy-Goo and ‘Ria-Roo Sprenkel and C-Bear,
Anna-Banana, Katie-Did, and ‘Tuffer Gass, for my informal education, and
To my mom and dad, Sara Lou and David Williams, for teaching me most of what resides in between the two.
iii
Acknowledgements
I would like to begin by thanking my primary graduate school advisor, Dr. Alison
S. Brooks. Alison’s Paleolithic Archaeology class was the first class I took in graduate school, and her early assistance and support gave me the confidence to apply to the
Hominid Paleobiology Doctoral Program. To me, Alison is the consummate scientist and role model. She cares a great deal for all of her students and treats everyone in the scientific community with dignity and respect. Alison has walked me through a number of joyful and trying experiences while I have been at GWU, and I do not think it is unwarranted to say that I would have not made it this far, or perhaps even into the program, without Alison’s guidance.
I would also like to thank Dr. Brian G. Richmond, who may have been surprised when he realized that I’d snuck in and stuck myself among his students, as well. Brian is one of the best teachers I have ever had the privilege to study with and I know that he has exercised enormous patient teaching me about biomechanics, anatomy, and physics during my time at GWU, for which I am very grateful. I would also like to thank Brian for always cushioning his critiques with at least one positive statement before getting to the “issues,” and for talking me back from the cliff’s edge when I thought that I had erased NPR’s FTP site and possibly taken down their website, as well.
Dr. Peter W. Lucas served as the third adviser in my advisory trifecta. Peter helped me look at things from unique perspectives and challenged me to investigate the underlying causes governing why systems work as they do. He has consistently been a source of
iv
encouragement and support and he is the only academic I know that gets as excited about experiments as my father does about life in general. I was sad to see Peter pick up and move to another continent and I will miss his sly jokes and watching him doodle on his tablet.
Dr. Adam D. Gordon has also acted as one of my main advisors since the collection of my pilot data. Adam has an amazing ability to take the most difficult concepts and slice them up into manageable pieces that make sense to me. He also has a mysterious way of guiding students to craft better experiments without dictating protocol from on high, which helped me learn far more than I would have otherwise. I am grateful for the enormous amount of the time he dedicated towards my education and dissertation. I also appreciate
Adam’s good humor and his continual reminders that THIS IS FUN!
The four people listed above formed the core of researchers that helped guide me through the dissertation process. They all read multiple drafts of grants and manuscripts, listened to dozens of podium and poster presentations, and let me find the right answers on my own time. I have been lucky to learn from them and their help has meant a great deal to me.
I would also like to thank Dr. Daniel Schmitt, who served as one of my external committee members. Dan is another teacher that manages to make difficult concepts less intimidating. He has been supportive of me and my research from the early stages, and his excitement made me more excited. Dan has been willing to discuss my questions and to loan me rather expensive equipment on what has turned out to be a long term basis. I am grateful for all that Dan has done for me.
v
Dr. Dietrich Stout also served as an external committee member, and helped usher me through the final stages of the process. I am grateful for Dietrich’s thoughtful comments and discussions of my research.
I am also grateful to Dr. Bernard Wood, first and foremost for all of his hard work over the years as the Director of the Hominid Paleobiology Doctoral Program. Bernard is not one to mince words or to leave Ts uncrossed, which means that manuscripts that pass across his desk come out cleaner (notwithstanding the abundance of red ink) and presentations that he previews paint a coherent picture and are free from spelling and grammatical errors. I appreciate the time he takes with all of us making sure that we are
“stage ready.” I am particularly grateful for his help in securing the postdoc I received and for his help at the end of my dissertation process.
I am endlessly grateful to the faculty and students of The Center for the Advanced
Study of Hominid Paleobiology and I am proud to be counted as one of their peers. The people that make up our department support one another personally and professionally in a variety of ways, and they made the long hours far more enjoyable than one may imagine they would be. I would like to thank our faculty and postdocs for all of the instruction they gave me and for making graduate school comfortable and fun, including: Dr. Robin
Bernstein, Dr. Shannon McFarlin, Dr. Chet Sherwood, Dr. Muhammad Spoc(ter), and Dr.
Erin Vogel. I would particularly like to thank Chet and Erin for their helpful discussions regarding cognition and statistics, respectively. I would also like to thank all of my fellow
Hom/Pal students, past and present, for creating an atmosphere free from competition and full of mutual support and respect, including: Jen Baker, Iowa Bauernfeind, Kallista Bernal,
Serena Bianchi, soon-to-be Dr. J-9 Chalk, Habiba Chirchir, Dr. Piz-aul Constantino,
vi
double-Dr. Rui Diogo, Du-Du Du, Dr. T Faith, soon-to-be Dr. Felicia Gomez, Dr. Deej
Green, Kevin Hatala, Dr. Amanda Henry, Dr. Griff, Dr. Lisa Nevell, Liz Renner, Kestopher
Schroer, Cheryl Stimpson, Dr. Robin Teague, Dr. Matt Skinner and Andrew Zipkin.
Thanks are due in particular to Amy, Kevin, and Tyler for their assistance in all things related to cognition (Amy) and statistical help (Kevin and Tyler), respectively. I am grateful to Deej, J-9, Amanda, and X-tyna for their help in avoiding holiday moments, and acknowledge that many sticky situations were avoided with the help of their proof reading and editing. J, you are clever and have lovely brown eyes.
Before entering the PhD program, I was a Masters student in the Anthropology department at GWU. A number of people made that experience easier and more enjoyable and I thank them for their help and support during that time. They include: Dr. Kitty Allen,
Dr. Robin Bernstein, Dr. Alison Brooks, Dr. Richard Grinker, and Dr. Stephen Lubkemann.
I would particularly like to thank Alison and Robin for their constant support and encouragement and for writing the references that helped me get into the PhD program.
Dr. Rick Potts and Jenny Clark at the Smithsonian National Museum of Natural History were also extremely kind and supportive while I worked towards my MA, and I appreciate all of their help, from the job (and air mattress) they gave me to their tolerance of my odd working hours.
Our program functions as well as it does in large part because we have a lot of support from the larger GWU community. Within the Anthropology department I am grateful to the entire administrative staff, particularly Jonathan Higman, Savannah
Fetterolf, Goby Mann, and Amanda Warner. Hom/Pal students also receive an additional helping hand from the CASHP Administrator, and I feel lucky to have received a great deal
vii
of help from the four that worked in this position while I was a student in the program:,
Kayla Jarvis, Gaby-Baby Mallozzi, Phillip Williams, and Sophie Thibodeau.
In the larger GWU community, I am grateful to Iva Beatty and Geri Rypkema for all of the administrative help they have given me over the years. I am also extremely grateful to Dr. Tara Wallace and the GW chapter of the Edward A. Bouchet Society
Graduate Honors Society for their support and for the funding which allowed me to complete my dissertation.
Beyond GWU, I have been supported and cared for by a large community of friends and family. They have provided friendship and support for many years now, and in general they make life more joyful. I am extremely grateful for my friends from Greenhills High
School, Grinnell College (even Jeff Brumfield), Memorial Christian Church, Ann Arbor, the GPF, NPR (particularly Jessica Goldstein, Chris Joyce, and Alison Richards) and my new community of friends that live just off of the aptly named One Wild Place on Portland
Street.
I was very fortunate to attend Grinnell College for my undergraduate education, where I majored in Anthropology. Grinnell’s Anthropology faculty made learning fun and exciting and they turned me on to the discipline that has come to mean so much to me. I hope one day I can become the type of teachers and mentors that they all are and I thank them for their early guidance. They include: Dr. Jon Andelson, Dr. Vicki Bentley-Condit,
Dr. Doug Caulkins, Dr. Katya Gibel Mevorach, and Sondi Burnell. I would particularly like to thank Dr. Kathy Kamp and Dr. John Whittaker. Together they have housed me, taught me, fed me, and encouraged over the last 10+ years, and their support continues to mean a great deal to me.
viii
I am also grateful to the organizations which funded my studies and research, including the Wenner-Gren Foundation’s Dissertation Fieldwork Grant (#7995), the
National Science Foundation’s Doctoral Dissertation Improvement Grant (# BCS-0903652) and Integrative Graduate Education and Research Traineeship (IGERT # DGE 9987590 and # DGE 0801634), The George Washington University’s chapter of Sigma Xi’s Grant in
Aid of Research, The George Washington University’s Research Enhancement Fund, and
The George Washington University’s Selective Excellence Fund.
Lastly, I would like to thank my large and loud family: Mom, Dad, Michael, the
Williams (in Chicago and Omaha), the Sprenkels (in Kansas and Pittsburgh), the Stovers, the Grahams, and the Broujous families.
Acknowledgments by chapter:
Chapter 2:
I wish to thank the knappers that flew in from around North America to participate in this study: Dr. Michael Bisson, Dr. Harold Dibble, Dr. Bruce Huckell, Dr. Grant McCall, Dr.
Dennis Sandgathe, and Merritt Sanders. Maria Pasquale and Susan Diekrager provided training and significant assistance with the Pliance system, for which I am grateful. I wish to thank Dr. Peter Lucas, who provided thoughtful comments regarding force and pressure on this chapter and chapters 4 and 5. I also wish to thank Craig Ratzat at Neolithics.com for his careful selection of raw materials for this project and all that followed. I am also grateful to Dr. Erin Vogel for her help with statistical analyses and suggestions.
Chapter 3.
ix
I am grateful to the students and volunteers that participated in the pilot studies I conducted, even those who mistook “knapping” for “napping” and thought they had signed up for an afternoon snooze, including: Dr. Jonah Choiniere, Matt Bukowski, Dr. Tyler
Faith, Nic Fourie, Dr. Nicole Griffin, Dr. Amanda Henry, Michael Frank, Nick Lonergan, and Dr. Christian Tryon. I wish to thank Dr. Can Kirmizibayrak for training me to use the
Vicon equipment and his invaluable help throughout my early data collection process. I am grateful to my co-authors, Dr. Brian Richmond and Dr. Adam Gordon, for helping me throughout this early project (and throughout all of the projects which comprise this dissertation) and helping me learn how to put together a manuscript for publication. I also wish to thank Dr. Alison Brooks for her helpful comments and insights on this project and throughout my dissertation.
Chapter 4
I wish to thank the knappers that took part in the experiments described in this chapter and in chapter 5: Dr. Michael Bisson, Dr. Harold Dibble, Dr. Grant McCall, Merritt Sanders,
Dr. Dennis Sandgathe, Steve Schwortz, Dr. John Shea, and Dr. Joanne Tactikos. I am very appreciative of the large chunk of time they all took out of their schedules in order to fly to
DC and participate in my dissertation experiments.
Chapter 5
The same eight knappers that participated in the experiments described in chapter 4 also took part in the experiments described in chapter 5, and again I am grateful for their participation. I am also grateful to Dr. Herzl Chai and Dr. Brian Lawn for all of the discussions we had regarding fracture mechanics and their help in applying their work to stone tool manufacture. I wish to thank Dr. Mark Reeves for his help and guidance in
x
figuring out how to calculate strike force, and for allowing me to practice a portion of my defense on his unwitting physics class. Dr. Dave Braun and Wesley Flear helped with the lithic analysis, for which I am grateful and I look forward to future collaborations as we wade through the massive experimental collection together. Lastly, I would like to thank
Dr. Chet Sherwood and Amy Bauernfeind for their help with matters relating to cognition.
xi
Abstract of Dissertation
Biomechanical Strategies during Oldowan and Acheulean Stone Tool Production
Multiple hominin species used and/or produced stone tools (e.g., Australopithecus afarensis, Paranthropus robustus, Homo habilis), yet evidence suggests that only later
Homo (i.e., H. erectus sensu lato) intensified and developed the behavior. This difference has been attributed to later Homo’s ability to execute efficient tool production, to the exclusion of earlier hominin species. However, we lacked the data on upper limb motions needed to evaluate the biomechanical context of stone tool production. With this in mind, the goal of this dissertation was to investigate the kinematic strategies used by modern humans in the production of Early Stone Age stone tools in order to test the primary hypothesis that modern humans’ upper limb condition contributes to efficiency and accuracy during stone tool production.
My collaborators and I used high-speed 3-D motion capture technology and a high- speed manual pressure sensor system to capture some of the only quantitative data on knapping kinematics, and the only quantitative data on manual pressure distribution during stone tool production presently available. The data and conclusions produced during this dissertation document the upper limb motions employed during Oldowan and Acheulean stone tool production. In doing so, my collaborators and I have 1) provided evidence
xii
against hypotheses directly linking the derived pollical condition to stone tool manufacture;
2) demonstrated that knappers employ a common kinematic strategy that has proven to be energetically efficient in a variety of contemporary activities; 3) support the hypothesis that modern humans exploit the upper ranges of their wrist extension ranges during knapping and in doing so achieve greater accuracy and efficiency; and 4) provided evidence that large-scale motion sequences (e.g., sequence of force application) rather than small scale motion sequences (e.g., sequence of joint motions) contribute to greater right hemisphere activity during Acheulean handaxe manufacture compared with Oldowan flake production.
This dissertation and the data collected in its course represent another step towards understanding the manner in which modern humans produce stone tools and the relationship of our upper limb anatomy to this developmentally significant behavior.
xiii
Table of Contents
Dedication iii
Acknowledgements iv
Abstract of Dissertation xii
List of Figures xvi
List of Tables xviii
Chapter 1: Introduction 1 The Oldowan and Acheulean stone tool industries 2 Early hominin upper limb anatomy and stone tool production 12 Dissertation goals and hypotheses 20
Chapter 2: Manual pressure distribution during Oldowan stone tool production 23 Abstract 23 Introduction 24 Methods 27 Results 32 Discussion 33
Chapter 3: Upper limb kinematics and the role of the wrist during stone tool production 56 Abstract 56 Introduction 58 Methods 63 Sample 63 Raw materials 63 Motion capture 63 Data analysis 65 Results 68 Upper limb motion patterns 68 Wrist extension/flexion and radial/ulnar deviation patterns 70 Work production 71 Discussion 71
xiv
Conclusion 76
Chapter 4: Achieving accuracy and efficiency during Oldowan stone tool production 89 Abstract 89 Introduction 91 The proximal-to-distal joint sequence and muscular efficiency 92 Strike accuracy 94 Methods 95 Sample 95 Motion capture 96 Kinematics and lithic analysis 98 Results 101 The knapping swing 101 Joint angles 102 Joint motion initiations 102 Peak velocity 104 Timing of peak joint angular velocity 104 Braced joint angular velocity and knapping accuracy 105 Discussion 105 Muscular and interactive torque 108 Peak joint velocity 109 Contributions of wrist extension 110 Conclusion 113
Chapter 5: Acheulean and Oldowan knapping strategies 132 Abstract 132 Introduction 134 Upper limb kinematics 138 Methods 139 Sample 139 Motion capture 141 Kinematics analysis 142 Lithic analysis 146 Results 147 Standard swings 147 Trimming swings 150 Acheulean v. Oldowan knapping swings 152 Discussion 154 Conclusion 161
Chapter 6: Conclusion 186
Literature cited 193
xv
List of Figures
Chapter 2 figures 45 Figure 2.1, Oldowan bifacial chopper produced by Subject E 45 Figure 2.2, Relationship between angular acceleration (m/s2) at the wrist and peak normal force 46 Figure 2.3a, Peak normal force (N) 47 Figure 2.3b, Peak pressure (kPa) 48 Figure 2.3c, Normal force (N) at strike 49 Figure 2.3d, Pressure (kPa) at strike 50 Figure 2.4a, Impulse 51 Figure 2.4b, Pressure-time integral 52 Figure 2.5a&b, Peak normal force (N) and pressure (kPa) during up-swing 53 Figure 2.5c&d, Peak normal force (N) and pressure (kPa) during pre-strike down-swing 54 Figure 2.5, e&f, Peak normal force (N) and pressure (kPa) during post-strike down- swing 55
Chapter 3 figures 84 Figure 3.1, Placement of reflective markers on subjects’ dominant hand 84 Figure 3.2, Vertical position and vertical velocity of the olecranon process, radial styloid process, and second metacarpal head through a typical knapping cycle 85 Figure 3.3, Lateral and dorsal views of the forearm with a depicture of extension/flexion and radial/ulnar angles 86 Figure 3.4, Model of the forearm with the radial styloid process, the second metacarpal head, and their respective paths during a typical knapping cycle 87 Figure 3.5, Total extension-flexion and radial-ulnar excursions 88
Chapter 4 figures 125 Figure 4.1, Oldowan bifacial choppers produced under unbraced knapping circumstances 125 Figure 4.2, Points used to calculate angles at the elbow joint 126 Figure 4.3, Wrist angles through one knapping swing, moving through the dart thrower’s arc 127 Figure 4.4, 3-D model of the upper limb through one knapping swing 128 Figure 4.5, Change in wrist motion and vertical position of the RSP and MC II head at the end of up-swing through down-swing 129 Figure 4.6, Peak wrist flexion velocity and peak elbow extension velocity 130
xvi
Figure 4.7, Elbow angle and angular velocity, and wrist angle and angular velocity through the knapping cycle 131
Chapter 5 figures 182 Figure 5.1, Experimental handaxes produced by Subjects A, B, C, and D 182 Figure 5.2, Wrist angles through one swing, moving through the dart-thrower’s arc 183 Figure 5.3, Oldowan and Acheulean strike forces 184 Figure 5.4, Strike force derived from lithic and kinematic data 185
xvii
List of Tables
Chapter 2 tables 38 Table 2.1, Cumulative peak force (strike) in relation to peak angular acceleration 38 Table 2.2, Peak normal force (N) and pressure (kPa) 39 Table 2.3, Normal force (N) and pressure (kPa) at strike 40 Table 2.4, Impulse and kPa-time integral 41 Table 2.5, Peak normal force and pressure during up-swing 42 Table, 2.6, Peak normal force and pressure during pre-strike down-swing 43 Table, 2.7, Peak normal force and pressure during post-strike down-swing 44
Chapter 3 tables 78 Table 3.1, Wrist joint limits of Pan, Gorilla, Pongo, and Homo 78 Table 3.2, Timing of peak linear velocities (m/s) in the upper limb relative to the transition from up-swing to down-swing and strike 79 Table 3.3, Peak linear velocities (m/s) in the upper limb during down-swing 80 Table 3.4, Percent of total wrist excursion employed during knapping 81 Table 3.5, Timing of peak extension, angular velocity (m/s) and angular acceleration (m/s2) at the wrist 82 Table 3.6, Work production (J) at the second metacarpal head and the radial styloid process 83
Chapter 4 tables 115 Table 4.1, Muscular-induced maximum joint angles 115 Table 4.2, Strike accuracy 116 Table 4.3, Peak knapping angles (⁰) and total excursion ranges 117 Table 4.4, Correlation between wrist extension/radial deviation and flexion/ulnar deviation 118 Table 4.5, Initiation of elbow extension and wrist flexion relative to strike 119 Table 4.6, Peak segment endpoint linear velocities 120 Table 4.7, Peak angular velocity at the elbow and wrist 121 Table 4.8, Timing of peak linear velocities 122 Table 4.9, Timing of peak angular velocities relative to strike 123 Table 4.10, Flake production rate distribution between unbraced and braced knapping conditions 124
Chapter 5 tables 163
xviii
Table 5.1, Upper limb masses (kg) and lengths (mm) 163 Table 5.2, Maximum handaxe length, width, and thickness (mm) 164 Table 5.3, Standard and trimming knapping swing count and associated flake count 165 Table 5.4, Timing of segment endpoint transition to down-swing motion direction 166 Table 5.5, Timing of initiation of elbow extension and wrist flexion relative to strike 167 Table 5.6, Correlation between wrist extension/radial deviation and flexion/ulnar deviation 168 Table 5.7, Peak knapping angles during standard and trimming Acheulean swings 169 Table 5.8, Post-strike wrist extension 172 Table 5.9, Peak linear velocities (mm/s) at segment endpoints 173 Table 5.10, Peak angular velocity (⁰/s) at the elbow and wrist 174 Table 5.11, Comparison between peak angular velocities (⁰/s) at the elbow and wrist between knapping positions 175 Table 5.12, Timing of peak angular velocities (⁰/s) at the elbow and wrist relative to strike 176 Table 5.13, Timing of the initiation of elbow extension and wrist flexion relative to the transition to down-swing 177 Table 5.14, Timing of peak angular velocity (⁰/s) at the wrist and elbow relative to the transition from up-swing to down-swing 178 Table 5.15, Timing of standard Oldowan and Acheulean joint initiations at the elbow and wrist 179 Table 5.16, Peak angular velocities (⁰/s) at the elbow and wrist during standard Oldowan and Acheulean knapping swings 180 Table 5.17, Comparison of the timing of peak angular velocity (⁰/s) at the wrist and elbow relative to the transition to down-swing 181
xix
xx
Chapter 1: Introduction
No other species on the planet uses technology to the same extent as modern
humans. From processing our foods to cleaning those foods from our teeth; from
scribbling equations on a piece of paper to using mathematics to send people into space;
technology is an integral component of countless human activities across the world.
Despite its ubiquity in modern human cultures, our relationship with technology has rather humble beginnings. More than 2.6 million years ago our early human ancestors began creating sharp edged implements by knocking stones together, sometimes removing no more than 3 flakes from a core (Kimura, 1999). Thus begins technology in the archaeological record.
The simplicity of early stone tools belies their importance to early humans’ success. Stone tool behaviors (e.g., tool production and use) are widely regarded as a significant adaptation in the evolution of our species. These behaviors provided early hominins with significant advantages over contemporary competitors, such as increased access to high-quality foods and defensive implements (Schick and Toth, 1993; Plummer,
2004; Braun et al., 2010; McPherron et al., 2010). These advantages, in turn, may have accelerated a series of further adaptations that culminated in the emergence of our own genus, Homo (Aiello and Wheeler, 1995).
Numerous researchers have hypothesized that stone tool production was a major selective pressure inducing changes in upper limb anatomy from the primitive, ape-like condition to the derived condition of anatomically modern humans (AMH) (Napier,
1962a; Susman, 1994; Marzke, 1997; Tocheri et al., 2008). Yet virtually no data on hand
1
and upper limb biomechanics during stone tool production are available for evaluating
these hypotheses. This dissertation is an investigation into the upper limb’s
biomechanical strategy used during stone tool production, undertaken to test hypotheses
relating hand and arm structure and function to stone tool production. It examines the upper limb motions used during the production of Oldowan and Acheulean stone tools through the lenses of functional anatomy, archaeology, and fracture mechanics to test the primary hypothesis that the modern upper limb anatomy found in later Homo plays a key
role in efficient stone tool production.
The remainder of this Introduction focuses on two main topics: 1) Oldowan and
Acheulean tools and their context and 2) the upper limb anatomy of Oldowan- and
Acheulean-era early hominins as it relates to stone tool manufacture. Next, the chapters that make up this dissertation are briefly described, as well as the specific hypotheses that will be tested in each chapter.
The Oldowan and Acheulean stone tool industries
The Oldowan and Acheulean stone tool industries make up the two primary industries of the Early Stone Age [2.6 – 0.3 mya (McBrearty and Brooks, 2000; Semaw,
2000)]. Although it is possible that hominins used and/or made stone implements prior to the known dates of the Oldowan (Panger et al., 2002), and recently discovered 3.4 million
year old cut marked bones from Dikika, Ethiopia may provide further support for this
hypothesis (McPherron et al., 2010, but see Dominguez-Rodrigo et al., 2010 for a
contrary interpretation of the specimens) , pre-2.6 mya tools have not yet been
discovered.
2
The Oldowan Industrial Complex is named for Tanzania’s Olduvai Gorge, where
large cores, chopping tools, and flakes were excavated by Mary and Louis Leakey starting in the mid 1930s (Leakey, 1971). The earliest Oldowan assemblages, dated to ~
2.6 mya, occur in Ethiopia’s Awash Valley in the Gona River drainage area of the Hadar
Region (Semaw et al., 1997). The industry continued throughout East and Southern
Africa until 1.7 – 1.6 mya when the Developed Oldowan and Acheulean industries
became dominant (see Klein, 2009 for a thorough review of Oldowan characteristics and
chronology). Homo habilis [2.3 – 1.6 mya (Kimbel et al., 1997)] is widely regarded as having made Oldowan tools (Leakey et al., 1964). However Paranthropus boisei [2.3 –
1.4 mya (McHenry and Coffing, 2000)] and P. robustus [1.9 – 1.4 mya (McHenry and
Coffing, 2000)] were also contemporaneous with the Oldowan, and fossils from both species have been found in the same stratigraphic layers as Oldowan tools, making it difficult to rule out their contribution to Early Stone Age assemblages (Leakey et al.,
1964; Brain, 1988).
Though conceptions of the character of the Oldowan industry vary, with varying levels of complexity and cognitive ability attributed to assemblages and their makers
(Wynn, 1981; Roche et al., 1999; Semaw, 2000), its main tenets are commonly agreed upon. Adherence to a final flake or core form does not appear to have been a goal of
Oldowan tool-makers; artifacts are informal and generally lack standardization (Klein,
2000; Wynn, 2002). Despite this, broadly defined tool forms are consistently present in
Oldowan assemblages. Manuports (transported but non-modified stones), hammerstones
(stones used in percussive activities, mainly to remove flakes from other stone pieces),
and altered cores and flakes make up the classic Oldowan tool kit (Leakey, 1971; Isaac,
3
1977). Mary Leakey developed one of two typologies commonly employed to describe the flake and core forms seen in Oldowan assemblages, categorizing them as scrapers, choppers, proto-bifaces, bifaces, discoids and spheroids (Leakey, 1971). Alternatively, cores and flakes may be categorized simply as whole, fragmented, retouched, or other non-functional terms to avoid the undue imposition of function onto form (Isaac, 1974).
Although Oldowan tools are frequently named and categorized according to function
(e.g., scrapers and choppers), experimental studies have demonstrated that these subjective categorizations are largely unwarranted (Toth and Schick, 2009). Instead, much of the variability that is found in Oldowan assemblages exists simply as an artifact of flake manufacturing processes (Toth, 1985).
The Oldowan is generally viewed as an opportunistic tool strategy in terms of reduction processes and raw material procurement. It is regarded as mainly static in terms of tool form and production complexity throughout the course of its history (Isaac and Harris, 1997; Kimura, 2002; Semaw et al., 2009). However, some researchers oppose the stasis hypothesis on the basis of assemblage variability, the diverse species of probable Oldowan tool-makers, and curation and production variability (Roche et al.,
1999; Delagnes and Roche, 2005). Still more researchers have demonstrated that a number of assemblages show evidence of the selective preference for higher quality raw materials. For instance, hominins at Gona, Ethiopia preferentially selected fine grained materials over those with a coarser grain (Semaw et al., 2003), the makers of the assemblages in the Koobi Fora Formation in Kenya avoided materials with internal flaws
(Toth, 1982) and hominins at Kanjera South, Kenya showed a preference for more durable materials (Braun et al., 2009).
4
The primary production techniques utilized by Oldowan hominins were hard-
hammer percussion, bi-polar reduction, the anvil technique, and a simple throwing
method. Of the four, hard-hammer percussion was the most commonly relied upon
(Schick and Toth, 1993; Toth and Schick, 2009). During hard-hammer percussion a smaller, rounded stone is held in the hammer hand (i.e., dominant hand) and struck against a core to produce either a sharp edge on the core itself or sharp edged flakes that come off of the core. Trimming, in which small flakes are removed from the edge of a larger flake or core in order to shape the core or to set up a platform for the removal of a flake, was not as frequently applied until the Developed Oldowan or the Acheulean
(Clark, 1994).
Regardless of whether one adheres to the stasis hypothesis or believes that tool practices changed during the Oldowan, it is clear that the Oldowan marks a dramatic shift in hominin behavior which resulted in the earliest record of human material culture.
Stone tools recovered from Oldowan sites demonstrate that the makers held at least a
basic understanding of rather complex fracture mechanics. Mechanically, successful
flake production requires the application of a force sufficient to induce material failure in
the contact region and propagate the resulting crack through the material without causing
shattering or crushing. The nature of crack propagation, flake formation, and flake
morphology are governed by aspects of the nodule itself and the manner in which it is
struck (Dibble and Whittaker, 1981; Cotterell et al., 1985; Cotterell and Kamminga,
1987; Dibble and Pelcin, 1995; Dibble, 1997; Pelcin, 1997b; Pelcin, 1997a; Andrefsky,
1998). Oldowan-era hominins were able to exploit platform angles and platform depths
and apply an appropriate level of force to the core in order to remove a series of flakes; a
5
feat which novice knappers can attest should not be underestimated. Though perhaps
unable to attain a specific flake size or shape, Oldowan hominins clearly understood the principles necessary to produce flaked implements.
It is highly likely that stone tool manufacture and/or use occurred to some degree
before 2.6 mya, particularly in light of stone tool use among extant chimpanzees, our
closest living relatives (Panger et al., 2002). Chimpanzees across Africa exhibit a variety
of tool behaviors, including the use of hammerstones to crack open nuts, and tool cultures differ between the populations (Whiten et al., 1999). Additionally, evidence of chimpanzee stone tool use dates to at least 4,300 years ago (Mercader et al., 2007). This date may well be extended in light of the emergence of a new sub-discipline focusing on primate archaeology. Given the existence of stone tool behaviors in both humans and chimpanzees, it is likely that the last common ancestor (LCA) of the two also participated in stone tool related behaviors (Haslam et al., 2009). However, there is no doubt that the
Oldowan at a minimum indicates the intensification of a series of novel behaviors in early humans’ adaptive regime, including the use of direct hard hammer percussion to fashion stones into tools, the long-distance transport of raw materials, and the inclusion of meat into the hominin diet (Toth and Schick, 2009). Stone tools became so prevalent and so significant to hominin subsistence strategies that it is reasonable to hypothesize that early humans would have shown morphological adaptations towards stone tool behaviors
(Napier, 1962a; Napier, 1962b; Susman and Creel, 1979; Susman, 1994; Marzke, 1997;
Richmond and Strait, 2000; Tocheri et al., 2007). Yet data on the biomechanical context of stone tool production has largely been lacking, making it difficult to evaluate the numerous hypotheses that have been erected. Using manual pressure pads and 3-D
6
motion capture technology, this dissertation research generated such data to investigate whether some of the modern anatomical features identified as adaptations towards stone tool manufacture in fact contribute to efficiency and/or accuracy during stone tool production.
The Oldowan was followed by the Acheulean Industrial Tradition, which spanned
1.9 – 0.3 mya (Asfaw et al., 1992; Deino and McBrearty, 2002). The Acheulean is found beyond the researches of the Oldowan in Europe, India, and arguably in East Asia, as
well (Klein, 1999; Yamei et al., 2000). The Acheulean is recognized as the industry of
Homo erectus sensu lato [1.9-0.5 mya (Swisher et al., 1994; Swisher et al., 1996)] and H.
heidelbergensis [600 – 250 kya (Clark et al., 1994; Parés et al., 2000)], and it marks the
dominance of the biface in archaeological assemblages. Acheulean-era bifaces include
picks, cleavers, and handaxes. All of these forms are implements commonly made on
large flakes, 30 cm or more in length. The tools themselves were frequently between 10
cm to 17 cm long (Ambrose, 2001). The ability to produce flakes of this size is regarded
as one of the primary characteristics distinguishing the Acheulean from the Oldowan
(Isaac, 1975; Klein, 1999).
The handaxe, a bifacial implement which has been flaked over all, or most, of the
surface on both sides, is the iconic tool of the Acheulean. The reduction technique used
to create handaxes produces a sharp edge around the entire perimeter of the tool.
Handaxes are generally classified in shape as ovates, teardrops, triangular, or sub-
triangular (Clark, 1954; Leakey, 1971; Clark, 2001). During the early Acheulean
handaxes tended to be crude and thick, distinguished from preceding Oldowan tools
mainly by the extent of flaking across both faces of the implement. However, by the end
7 of the tradition Acheulean handaxes were remarkably thin and flat and they had reached a high degree of sophistication and symmetry (Delson et al., 2000; Klein, 2000). At
Olduvai, Acheulean sites were further distinguished from Oldowan sites by a 40% minimum representation of bifaces among all of the tool types (Leakey, 1971).
In addition to the manufacture of large flakes for the production of bifaces, the
Acheulean, particularly the later stage, also ushered in the use of soft hammer percussion, platform preparation, and prepared core techniques such as Levallois and Kombewa
(Klein, 2000). As their name implies, soft hammers are implements that deform much more readily than the hammerstones commonly used in Oldowan stone tool production because of the materials from which they are made Platform preparation is the intentional shaping of the striking surface in order to facilitate removal of long and wide or thin flakes (Schick and Toth, 1993; Whittaker, 1994). Prepared core knapping techniques were used to produce a flake of a specific size and shape. In attaining this goal, a series of smaller flakes are systematically removed from the perimeter and both surfaces of a bifacial core. The final flake, removed after a striking platform is carefully prepared at one end, generally encompasses much or all of one surface of the core (Van
Peer, 1992).
Whereas Oldowan tools are frequently regarded as largely arbitrary in form,
Acheulean bifaces are thought to reflect preplanning and the use of high-level operational skills in order to obtain an end product of a specific size and shape (Wynn, 1979; Toth and Schick, 1993; Ambrose, 2001; de la Torre et al., 2003; Pelegrin, 2005). This, in combination with the increased cranial capacity and generally more modern body form of
H. erectus sensu lato (Wood and Collard, 1999a; Wood and Richmond, 2000) has lead
8 many researchers to assign greater cognitive abilities to Acheulean-era hominins. Others, however, view the assignment of increased cognitive capacity to Acheulean hominins on the basis of standardization and symmetry as premature (Silverman, 2002; Simão, 2002;
McNabb et al., 2004). Although it is difficult to determine whether Acheulean implements reflect a greater cognitive capacity per se, from a production point of view there are clear differences between the two industries.
Oldowan tools are frequently described as “crude,” “blocky,” or “irregular,”
(Leakey, 1971; Klein, 2000), whereas the Acheulean, particularly the later Acheulean, is described as “complex,” “standardized,” and “finely flaked,” (Clark, 1966; Wynn, 1979;
Ambrose, 2001; Noll and Petraglia, 2003). These descriptions seem warranted when one considers that bifacially flaked Oldowan implements frequently have no more than eight to eleven removals (Kimura, 1999; Roche et al., 1999), while Acheulean bifaces pass through three to four production stages, use hard and soft hammer percussion, and that it is not uncommon to remove 50 or more flakes in the production of a single tool (and this says nothing of the smaller flakes that are removed in the trimming and shaping process)
(Newcomer, 1971; Whittaker, 1994). Despite these clear differences, it remains difficult to determine the impetus for the Oldowan – Acheulean transition, particularly given the co-occurrence of the two industries at sites such as Olduvai Bed II and throughout the
Middle Awash (Leakey, 1971; Schick and Clark, 2003). Researchers have attributed the transition to invading populations of hominins, differences in activities, and raw material availability, among other reasons (Clark, 1978; Kimura, 1999; Schick and Clark, 2003).
Yet these explanations are somewhat unsatisfying because they fail to address the underlying reason for the transition. For example, if the shift is evidence of invading
9
hominin populations, did one population possess refined motor skills compared to the
other? If the two industries represent different activities, what were those activities and
how was one implement more appropriate than another for the task at hand?
Research recently conducted by Stout and Chaminade (2007) and Stout and
colleagues (2008) has begun elucidating some of the underlying causes behind the
transition. These researchers demonstrated that among modern human knappers the right
hemisphere of the brain was significantly more active during Acheulean tool production
compared with Oldowan tool production, including the right ventral premotor cortex, the right inferior prefrontal cortex, and the right supramarginal gyrus of the inferior parietal lobule. The authors attributed this greater right hemispheric activity to greater sensorimotor control, task-set switching, the inhibition of action and the regulation of complex action and action sequences. In the making of a bifacial handaxe, the success of
subsequent removals is determined by the patterns of flake scars created from the
preceding removals. This means that a misplaced flake removal has the ability to hinder
or arrest further reduction, potentially resulting in a failed attempt to produce a handaxe.
Consequently, the location, size, and shape of each flake must fit into the overall plan for
the entire tool and the knapper has to adjust his or her knapping strategy as the
morphology of the core changes.
Noting previous hypotheses linking language and tool use, Stout and colleagues
(2008) argued that their finding lent further support to these connections based on the right hemisphere’s contribution to the execution of hierarchical behavioral sequences and inhibition. The authors concluded that language skills and complex tool-making co-
10
evolved, and that their concurrent development reinforced the underlying processes of
both behaviors.
Faisal and colleagues (2010) compared hand postures and hand joint sequences in
the left hand between Oldowan and Acheulean tool manufacture, to determine whether
the increased right hemisphere activity simply reflected greater motor complexity in the
left hand. They found that no differences existed in the complexity of hand postures, and
concluded instead that greater right hemisphere activity during Acheulean stone tool
manufacture must reflect greater demands for cognitive control of action, task-set
switching, and action inhibition.
The control and inhibition of action sequences may refer to large scale actions,
such as alternating high force and low force hammerstone strikes, or small scale actions,
such as the sequence of joint activation in the course of executing a single knapping
swing. If Acheulean reduction sequences call for greater control of action and the
regulation of more complex action sequences on a large or small scale, these differences
may be evident in the upper limb kinematics. To investigate this issue, kinematics of the
dominant arm during Acheulean and Oldowan reduction sequences were captured and their actions and action sequences were compared using 3-D motion capture technology.
3-D motion capture records both large scale and more refined upper limb motions, making this technique particularly appropriate to investigate the issue of motion sequences and complexity.
11
Early hominin upper limb anatomy and stone tool production
Since the recovery of the OH 7 (H. habilis) hand from Olduvai Gorge, we have known that stone tool manufacture did not require a modern hand (Napier, 1962b). There are now multiple lines of evidence demonstrating that the primitive upper limb anatomy was sufficient for stone tool production of Oldowan-like flakes and cores (Napier, 1962b;
Leakey et al., 1964; Marzke, 1997; Tocheri et al., 2003). For example, Kanzi, a male bonobo chimpanzee, has been making stone tools for more than two decades. The upper limb anatomy of the LCA is commonly hypothesized as closely approximating the condition of extant chimpanzees, making Kanzi an appropriate living analog for the tool- making capabilities of the LCA (Tocheri et al., 2008: and references within). Although the flakes and cores that Kanzi has produced are distinct from Oldowan tools, and his preferred style of knapping is quite different than that of modern humans, there is no doubt that Kanzi is capable of manufacturing stone implements (Toth et al., 1993; Schick et al., 1999). Similarly, Homo florensiensis [~ 74 – 14 kya (Brown et al., 2004)] made
Oldowan-like tools despite retaining a primitive wrist morphology (Morwood et al.,
2004; Tocheri et al., 2007).
However, there is a large gulf between mere ability and behavioral efficiency
(Marzke and Marzke, 2000), particularly in the context of competition among species.
This dissertation investigates that gulf, by testing whether aspects of the modern upper limb condition contribute to accuracy and efficiency during stone tool production.
12
Manual pressure distribution during stone tool manufacture
We currently lack fossil evidence of the LCA between chimpanzees and humans1.
However, as stated above, the inferred condition of the LCA’s hand and wrist is quite similar to those of a modern chimpanzee. The LCA is hypothesized to have had a high finger to thumb length ratio, curved manual phalanges with strong flexor muscle markings, narrow apical tufts, and a gracile pollical metacarpal (Tocheri et al., 2008: and references within). It is likely that the LCA’s carpals were oriented similar to chimpanzees’ to better withstand the compressive loads experienced during locomotion.
Such loads are transmitted through MCs 2-4 to the carpal region and up into the forearm
(Tocheri et al., 2005; Wunderlich and Jungers, 2009). In contrast, modern humans’ carpals are oriented such that they are better able to withstand transverse loads that are transmitted across the MC I-trapezium joint through the scaphoid, such as occur during strong pinching activities (Momose et al., 1999). High compressive loads are also thought to occur across the first carpometacarpal joint during stone tool production and the modern configuration has been cited as an evolutionary method for withstanding wear in this region. The ability to resist loads along the pollex is viewed as so integral to a species’ ability to sustain stone tool behaviors that Susman (1994) proposed that the thumb’s structure was selected specifically to resist loads sustained during strong repetitive percussive activities such as stone tool manufacture. He further proposed that a single feature describing the thumb’s robusticity (the ratio of metacarpal I head width to metacarpal I length) is sufficient to detect human-like tool making abilities in fossil hominins.
1 In fact, despite a rich hominin fossil record, the chimpanzee fossil record consists of only three teeth from the Kapthurin formation in Kenya. McBrearty S, and Jablonski NG (2005) First fossil chimpanzee. Nature 437:105-8. 13
Two groups of researchers have indirectly investigated Susman’s (1994)
hypothesis by looking at pollical musculature recruitment during hard hammer
percussion2. Hamrick and colleagues (1998) reported high levels of muscle activity in
the flexor pollicis longus (FLP), the largest thumb muscle, during stone tool production
and use, indicating its important role in stabilizing the thumb and load resistance.
However, Marzke and colleagues (1998) captured data on a wider range of hand and
forearm muscles and reported varied FLP recruitment during their tool replication
experiments, with the lowest recruitment levels in the most experienced knapper. Noting
that the FLP is particularly susceptible to fatigue during strong pinching activities, the
authors hypothesized that Oldowan tool makers would have avoided those grips requiring
strong FLP activity to circumvent this problem. Surprisingly, Marzke and colleagues
(1998) reported that three of the five most heavily and consistently recruited muscles
were not directly related to thumb movement or stability, calling into question whether
the thumb does indeed experience larger forces compared to the other digits. However,
neither study directly tested whether the thumb is subjected to greater loads compared
with the other digits of the hand during stone tool manufacture.
Analyses of the carpal joint surfaces of fossil hominins have yielded further
surprising results regarding which hominins were anatomically committed to stone tool
production. H. habilis, the “handy man,” has long been regarded as a tool maker, and
was once touted as the very first (Napier, 1962b; Leakey et al., 1964; 2003). However,
2 Hamrick and Inouye (1995) also tested Susman’s (1994) hypothesis by expanding the hominoid sample set to include gorillas. In doing so they found that the range of pollical metacarpal head breaths recorded among gorillas overlaps those recorded for modern humans. Additionally, the ratio of pollical metacarpal head breadth to length recorded in mountain gorillas overlaps that of modern humans more than that of chimpanzees. The authors concluded that because all taxa possessing a wide pollical metacarpal head relative to length do not used a precision grip or make stone tools, the feature is not a reliable test for tool use in the fossil record. Hamrick MW, and Inouye SE (1995) Thumbs, tools, and early humans. Science 268:586-7; author reply 9. 14
Tocheri and colleagues (2003) demonstrated that the shape and orientation of the trapezium of A.L. 333-80 [Australopithecus afarensis, 4.18-3.0 mya (Kappleman et al.,
1996; Wood and Richmond, 2000)] is more similar to that of modern humans, while the trapezium of O.H. 7 is more similar to gorillas. The authors concluded that A. afarensis would have been capable of executing modern human-like pad-to-side and three-jaw chuck grips, and with greater force than chimpanzees. H. habilis, on the other hand, would have been required to knap in a manner distinct from modern humans. Yet neither species would have been able to resist pollical joint wear—which is assumed to arise from hard hammer percussion—as well as H. sapiens (Tocheri et al., 2007).
Despite studies indirectly testing whether the thumb is subject to high pressures during stone tool production and others that evaluated its ability to withstand this, the pressure distribution across the hand during such production is still unclear. This significant gap in our knowledge of tool production parameters has made it difficult to evaluate hypotheses relating pollical metacarpal and carpal shape and orientation to stone tool production. This dissertation addresses this issue by capturing manual pressure data during the production of Oldowan-like chopping tools. Pressure and normal force data were collected from six experienced stone tool producers at 200 Hz in order to test the hypothesis that during the production of Oldowan stone tools the thumb is subjected to significantly greater normal force and/or pressure compared to other regions of the hand, particularly along the remaining four digits.
15
The contributions of wrist extension to stone tool production
Cut marked bones recovered from Dikika, Ethiopia have been loosely associated with A. afarensis, making this early hominin the first that is in any way potentially associated with stone tool behaviors (McPherron et al., 2010). However, the stone tools themselves are currently lacking and the validity of the cut marks have been questioned
(Dominguez-Rodrigo et al., 2010). Whether the lack of associated tools is merely absence of evidence as opposed to evidence of absence is not yet clear. Researchers do, however, believe that if A. afarensis participated in stone tool behaviors, the species’ mixture of primitive and derived upper limb features would have restricted their knapping motions and rendered them perhaps less efficient compared to AMHs (Marzke, 1997).
In addition to a carpal orientation that is in some ways more modern than that of
H. habilis, A. afarensis also exhibited a more modern finger to thumb length ratio
(Marzke, 1983; Latimer, 1991; Alba et al., 2003). Many of the other features of the wrist and hand, however, remained primitive in form. A. afarensis retained curved proximal phalanges with strong flexor sheath markings (Bush et al., 1982; Stern and Susman,
1983; Susman et al., 1984; Haile-Selassie, 2001), narrow apical tufts (Bush et al., 1982), a gracile metacarpal I (Smith, 2000), and a projection on the dorsal distal radius which is also seen in extant Pan and Gorilla, as well as A. anamensis (Richmond and Strait, 2000;
Richmond et al., 2001). This combination of features indicates that A. afarensis lacked anatomical components that have been hypothesized to contribute to effective stone tool manufacture (Marzke, 1997; Tocheri et al., 2003). However, due to the nature of the primitive condition (e.g., short fingers compared to thumb length), it is difficult to mimic and model the effects of the primitive anatomy on stone tool production (Susman, 1998).
16
Consequently, it had not been possible to say with any certainty how A. afarensis would
have knapped. This difficulty is heightened by the variable representation of primitive
and derived features—i.e., unique anatomy—in each extinct hominin taxa. Thus,
although it appeared likely that A. afarensis would have manufactured stone tools in a
less efficient manner compared with modern humans, this hypothesis had not been
directly tested.
Oldowan-era hominins also showed a mixture of primitive and derived upper limb
features. However, it is difficult to determine the species designation of many of these
fossils due to the co-occurrence of multiple species within a given site. For example,
cranial and postcranial elements that may belong to either P. boisei or H. habilis have been recovered from Olduvai Gorge, Tanzania, although the OH 7 hand is generally regarded as H. habilis [(Napier, 1962b; Wood and Richmond, 2000) though see Moyà-
Sola and colleagues (2008) for an alternative classification]. Similarly, postcranial elements from Swartkrans, South Africa may belong to either P. robustus or H. cf. erectus. Susman (1988) argues that because 95% of the craniodental remains in Member
1 have been assigned to P. robustus, it is likely that the postcranial remains also belong to the same species. However the possibility that the fossils are actually Homo remains
(Trinkaus and Long, 1990).
Assuming the fossils have been correctly classified, features that may facilitate
stone tool behaviors that have been recorded among them include the broadening of
apical tufts in presumed H. habilis, P. robustus, and A. africanus specimens (Napier,
1962b; Ricklan, 1990; Smith, 2000), a reduction in curvature in the proximal phalanges,
and a more robust pollical metacarpal in P. robustus (Susman, 1988; Susman, 1994;
17
Tocheri et al., 2008), and a modern finger to thumb length ratio (Green and Gordon,
2007), a more modern carpal shape and orientation (Tocheri et al., 2003), and a
stabilizing styloid process at the MC III base (Ricklan, 1987) in A. africanus [3.5-2.4 mya
(Clarke and Tobias, 1995)]. Specimens of A. africanus and those presumed to belong to
P. robustus also exhibit flatter distal radii that is similar to the modern condition (Grine and Susman, 1991; Richmond and Strait, 2000). However, all of these species also retain a number of primitive features, including curved proximal phalanges in H. habilis and A. africanus (Susman and Creel, 1979; Ricklan, 1987), a primitive carpal arrangement in H. habilis (Tocheri et al., 2003), and an ape-like pollical breadth to length ratio in A. africanus (Green and Gordon, 2007). The few upper limb elements that have been described for the newly erected species A. sediba (Malapa, South Africa) are also primitive. Berger and colleagues (2010) report that sediba had long, curved phalanges with strong muscle markings for the flexor digitorum superficialis muscle.
As stated above, the functional significance of the majority of these features during stone tool manufacture is difficult to test directly. For instance, we cannot directly test the effects of a high finger to thumb length ratio or narrow apical tufts on one’s ability and efficiency in manufacturing stone tools. This difficulty is exacerbated by the fact that different species exhibit different combinations of primitive and derived features
(Susman, 1998). With Susman’s (1998) concern in mind, one of the goals of this dissertation was to investigate the upper limb kinematics of modern human knappers, as well as changes in their knapping kinematics across skill level and lithic tradition. The role of the wrist, and specifically the use of extension and flexion, is highlighted throughout the dissertation for three reasons. First, mobility in wrist extension to allow
18
greater rnages of wrist movements has been hypothesized to contribute to efficient stone
tool manufacture (Marzke, 1971; Ambrose, 2001; Richmond et al., 2001). Second,
among the fossils that currently make up the fossil record, the dorsal distal radial
projection differs greatly among taxa; A. anamensis and A. afarensis exhibited the
African ape-like condition, while P. robustus, A. africanus, H. neanderthalensis, and H.
sapiens did not (Grine and Susman, 1991; Richmond and Strait, 2000). And third, wrist extension may be non-invasively limited by using a simple extension-limiting wrist restraint, thereby mimicking what may have been the ancestral condition. Through this method it becomes possible to isolate wrist extension and examine its significance to aspects of stone tool production, such as accuracy and efficiency.
Following P. robustus, manual elements and carpals become scarce in the fossil record until H. neanderthalensis [300-30 kya (Arsuaga et al., 1997; Wood and Richmond,
2000)]. Those which exist have either yet to be made public or are of a fragmentary nature and consequently difficult to analyze. For example, upper limb elements from
Area 1A of the Koobi Fora Formation in Kenya (specimen KNM-ER 47000 date to 1.52
mya) resemble Pliocene Australopithecus (Richmond et al., 2011) and have been
cautiously assigned to P. boisei (Richmond et al., 2009). However, a full comparative
analysis on the fossils is currently unavailable. Additionally, one complete capitate and
11 fragmentary hand bones have been recovered from Gran Dolina, Spain which have
been assigned to H. antecessor (Lorenzo et al., 1999). Together the elements exhibit features which are derived with respect to Australopithecus and primitive with respect to modern humans. However their fragmentary nature makes a detailed analysis currently difficult. Manual elements and carpals are nearly absent from H. ergaster and H.
19 erectus, with the exception of a metacarpal from WT 15000 (Walker and Leakey, 1993), a lunate from Zhoukoudian (Weidenreich, 1941), and two distal phalanges from Dmanisi
(Lordkipanidze et al., 2007). Some of the hand elements from Swartkrans may belong to
H. erectus, however their species attribution remains unclear (Susman, 1988; Susman,
1989; Trinkaus and Long, 1990).
The lack of securely attributable H. erectus sensu lato hand and wrist fossils is particularly unfortunate given the association between H. erectus and the Acheulean
Industrial Tradition. Though clearly distinct and primitive with respect to H. sapiens, cranial dental and postcranial material overwhelmingly indicates that H. erectus sensu lato practiced an adaptive strategy with more similarities to H. sapiens than any previous hominin (Wood and Collard, 1999b; Wood and Collard, 1999a). It is regrettable that we cannot state with any degree of certainty the morphology of H. erectus sensu lato hands.
By the time manual elements are again abundant in the fossil record, with H. neanderthalensis, the upper limb was highly derived (Wood and Richmond, 2000;
Tocheri et al., 2008) and both species of Homo manufactured complex composite Middle and Upper Paleolithic stone tools, and arguably symbolic ornaments (d'Errico, 1998; d'Errico, 2003; Henshilwood et al., 2004).
Dissertation goals and hypotheses
The goal of this dissertation is to investigate the biomechanical strategies used by modern humans in the production of Early Stone Age stone tools in order to test the primary hypothesis that aspects of the modern upper limb condition contribute to
20
efficiency and accuracy during stone tool production. Specific sub-hypotheses will be
tested in four manuscripts presented as dissertation chapters as follows:
Chapter 2: Chapter 2 describes the investigation into the distribution of pressure and normal force across the hand during the production of Oldowan-like choppers. This experiments was undertaken to test the hypothesis that during stone tool production the thumb is subject to significantly greater pressure and/or normal force compared to other regions of the hand, particularly along the remaining four digits. Manual impulse and pressure-time integrals are also analyzed.
Chapter 3: Chapter 3 describes the knapping kinematics of four amateur knappers during simple flake production. Three sub-hypotheses were tested:
1) Across subjects, upper limb kinematics during stone tool production are grossly similar, but vary in the timing of specific events.
2) Upper limb joint kinematics will occur in a proximal-to-distal joint sequence (PDJS).
3) Wrist motions will significantly influence efficiency and accuracy.
Chapter 4: Chapter 4 describes of the knapping kinematics of eight experienced subjects during the production of Oldowan bifacial chopping implements. These data are compared to the knapping kinematics of the same eight subjects when their ability to extend at the wrist was limited to ~30⁰ – 35⁰ . The knapping swing is described in detail and the following sub-hypotheses are tested:
1) Skilled knappers will initiate joint motions in a proximal-to-distal sequence and utilize a full proximal-to-distal joint sequence in terms of peak linear segment endpoint velocities, peak joint angular velocities, and the onset of peak angular velocities.
21
2) Limiting wrist extension will result in significantly lower angular velocities at the wrist and in turn lower strike accelerations and forces.
3) Limiting wrist extension will significantly decrease target accuracy during stone tool production.
Chapter 5: Chapter 5 describes the kinematic strategy utilized by experienced knappers during the production of Acheulean bifacial handaxes. This strategy is compared to the kinematic strategy employed during the production of Oldowan bifacial choppers to test the hypothesis that Acheulean motion sequences exhibit greater complexity in upper limb action sequences.
Chapter 6: Chapter 6 is a summary of the four studies conducted in the course of this dissertation and a discussion of the implications of the findings in regard to stone tool production and upper limb adaptations as they related to stone tool production.
22
Chapter 2: Manual pressure distribution during Oldowan stone tool production
ABSTRACT
Modern humans possesses a highly derived thumb that is robust and long relative to the other digits, with enhanced pollical musculature compared to extant apes. Researchers have hypothesized that this anatomy was initially selected in early Homo in part to withstand high forces acting on the thumb during stone tool production’s hard hammer percussion; however, data were lacking on loads experienced during stone tool production and their distribution across the hand.
Here we report the first quantitative data on manual pressures (kPa) and normal forces (N) acting on the hand during Oldowan stone tool production, captured at 200 Hz.
Data were collected from six experienced subjects replicating Oldowan stone tools, the earliest tools in the archaeological record. Our data do not support hypotheses asserting that the thumb experiences greater pressure compared to other regions of the hand. Peak pressure, normal force, impulse, and the pressure/time integral were significantly greater on the 2nd and/or 3rd digits compared to the 1st in every subject. Our findings call into
question hypotheses linking modern human thumb anatomy specifically to load resistance
during stone tool production.
23
INTRODUCTION
Humankind’s ability to manipulate technology is commonly cited as an essential
component of what defines us as Homo sapiens. Consequently, the origin of our
relationship with technology is a subject of intense interest across a range of disciplines.
Stone tool manufacture, the earliest form of technology in the archaeological record
(Semaw, 2000), is regarded as a principal selective pressure that may have acted on some
aspects of early human development, such as cognition (Holloway Jr., 1969; Stout et al.,
2008; Faisal et al., 2010) and hand and wrist anatomy (Napier, 1965; Marzke, 1997;
Tocheri et al., 2007). In regard to the latter, anatomically modern H. sapiens possess a
number of derived features which enhance our ability to forcefully grasp and manipulate
small objects in a single hand, two important components of stone tool production
(Marzke and Shackley, 1986). These include short, relatively straight phalanges, broad
apical tufts on the distal phalanges, and carpometacarpal and metacarpolphalangeal shape
and articular orientation that allow simultaneous flexion and rotation of the 2nd and 5th
metacarpals, among others (Napier, 1962b; Lewis, 1977; Susman, 1979; Susman, 1994;
Panger et al., 2002). The thumb, in particular, has been directly linked to stone tool production and is hypothesized to have been selected to withstand the high repetitive
percussive forces experienced during stone tool production (Marzke, 1992; Susman,
1994; Hamrick et al., 1998; Tocheri et al., 2007). However, the data on manual pressure
during stone tool production that is needed to evaluate such hypotheses has not been
available until now. Here we present the first quantitative data on the manual pressures
and normal forces experienced during stone tool production, and their distribution across
24
the hand. Data were collected from six stone tool makers (i.e., knappers) experienced in
the replication of Oldowan stone tool industries, using a high speed manual pressure
sensor system and 3-D motion capture technology (both recording at 200 Hz). With these data we test the hypothesis that during the production of Oldowan stone tools, the
thumb is subject to significantly greater pressure and/or normal force compared with
other regions of the hand, particularly along the remaining four digits.
A number of features of the modern human thumb contribute to stability at the
first carpometacarpal joint and to our ability to forcefully oppose the thumb with the
finger digits. Both are considered important during stone tool manufacture. For
example, the modern human thumb is robust in build, particularly at the metacarpal base
and head, which helps reduce increased joint stress due to our enhanced thumb
musculature (Susman, 1994). We have a relatively flat first carpometacarpal joint
(Guthrie, 1991; Marzke et al., 2010) and the highest mean thumb length to index finger
length ratio of all extant primates (Napier, 1993). Together these assist in true thumb-to- finger opposition, necessary for precision handling. Our carpals are also oriented in a manner better able to withstand forces directed into the thumb—as has been hypothesized to occur during stone tool manufacture—compared with the primitive condition and that seen in extant African apes (Tocheri et al., 2003; Tocheri et al., 2007).
H. sapiens also possess a unique representation of pollical muscles, including two ventrally derived pollical muscles and one dorsally derived pollical muscle which cross the first metacarpophalangeal joint. Nearly all other primates lack these three muscles— the first volar interosseous of Henle, a true flexor pollicis longus (FPL) muscle, and an extensor pollicis brevis (Susman, 1994; Tocheri et al., 2008; Diogo and Abdala, 2010;
25
Diogo et al., 2010). The first muscle is a small intrinsic hand muscle associated with fine
manual motor control and more recently with grasping ability (e.g., holding a
hammerstone)(Marzke et al., 1998). The last muscle is an extrinsic hand muscle which
inserts into the first phalanx in humans to assist extensor pollicis brevis and abductor
pollicis brevis with thumb extension. The FPL is a large extrinsic thumb muscle,
comprising approximately 22% of the thumb’s total muscular physiological cross-
sectional area (Marzke, 1997). In humans the muscle belly of FPL is separate from that
of flexor digitorum profundus, and thus constitutes a separate and far stronger muscle
than the extrinsic pollical flexor in the former species, unlike the condition found in other
great apes (Straus, 1942).
FPL is widely hypothesized to be significant during stone tool manufacture
(Susman, 1988; Marzke, 1992; Susman, 1994; Hamrick et al., 1998; Wood and
Richmond, 2000). Indeed, just the presence of a large insertion site for the FLP is nearly sufficient to bestow tool-making capabilities on fossil hominins [(Napier, 1962b;
Susman, 1988; Ricklan, 1990; Susman, 1994) however see Marzke and colleagues (1998) and Tocheri and colleagues (2008) for an alternative interpretation of the volar
depression].
FLP’s hypothesized contribution to stone tool production is the primary role it
plays during strong pinching activities, allowing for forceful opposition of the thumb
against the other digits, which act in concert to stabilize objects held between the thumb
and finges. Researchers argue that this strong precision pinching ability helps ensure a
firm grip on the hammerstone and help resist hammerstone displacement that may occur
26 due to strong upward reaction forces at strike (Susman, 1994; Marzke, 1997; Hamrick et al., 1998; Susman, 1998).
METHODS
Normal force (N) and pressure (kPa) data were captured from six subjects
(Subjects A-F) during the production of Oldowan tools (n = 2 tools per subject, n = 148 swings). Associated kinematic data were captured from five of the six subjects (Subjects
A-E, n = 98 swings). Informed consent was obtained from each subject prior to experimentation. Data were captured in The George Washington University’s Motion
Capture and Analysis laboratory in Washington, DC.
Five males (Subjects A-C, E, and F) and one female (Subject D) participated in the study. All subjects were professional archaeologists or archaeological graduate students familiar with Oldowan tool typologies and proficient in the production of
Oldowan tools. All subjects were healthy adults free from muscular and/or osteological arm, forearm, and hand conditions which may have compromised their data. Subjects B-
F were right-hand dominant, Subject A was left-hand dominant.
All reduction sequences were conducted on nodules of cortex-free raw English flint (material toughness ≈ 1 Kc). Subjects were given two fist-sized flint “cobbles”, which were knobs of flint that been removed from larger nodules, and were requested to produce Oldowan bifacial chopper from each cobble with no more than 15 flake removals per chopper (Figure 2.1). Each subject was allowed to select a hammerstone of his or her choosing from among a group of hammerstone weighing < 0.75 kg each. The
27
sole stipulation placed on hammerstone selection was the knapper’s ability to wield the
hammerstone with digits I-III during stone tool production. Most employed a 3-jaw chuck grip or a variant of that grip. Previous knapping experiments have demonstrated that the 3-jaw chuck would have been particularly effective in Oldowan stone tool production, and modern hand morphology has been related specifically to this grip
(Marzke and Shackley, 1986; Marzke, 1992; Marzke, 1997). All knapping occurred with subjects seated in a wooden chair (seat height = 48.26 cm). Subjects all naturally assumed similar knapping positions, with the core nodule resting against one leg (i.e., the left leg for right-handed subjects, the right left for left-handed subjects) and the hammerstone held in their dominant hand. Subjects naturally employed a standard knapping swing (Williams et al., 2010) to remove flakes, swinging with their dominant hand and arm across their midline to the core, which rested on their contralateral leg. Up to three leather pads (5 mm thick) were laid across subjects’ core-leg to protect it from injury.
Dynamic normal force and pressure data were captured using a Novel Pliance pressure sensor system. The Pliance system is comprised of five sensor strips (100 x 10 mm2), each connected via a sensor cable to a pliance-x electronic analyser. Each sensor strip contains ten sensor elements (each 10 x 10 mm2, thickness: < 1.2 mm) arranged in a
single row and embedded between two thin sheets of rubber coating (thickness: 0.1 – 0.6
mm). A 1 cm extension of the rubber coating (i.e., the lead) extends off of the distal end
of each sensor strip. Sensors strips were connected to the palmar sides of subjects’ 1st -
3rd digits. The strips were held in place by wrapping the lead around the distal phalanx from the palmar side to dorsal side of the digit and taping the lead onto the dorsal side of
28 the digit. Care was taken to avoid placing any tape on the section of the strips containing sensor elements. Finger condoms were placed over each digit to further secure the sensor strip. The sensor strips traveled across the palm and wrist to connect to the sensor cable just proximal to the wrist. The trailing end of each sensor strip was held in place at the wrist using another small strip of Velcro. Within each strip, the sensor elements stopped mid palm while the sensor-free strip continued across the entire palm and wrist.
Proximal to the wrist, sensor cables were connected to the Pliance-x electronic analyser.
The Pliance-x electronic analyser was placed either in a utility belt worn by each subject or on a small table next to the subject. Both positions rendered it out of the knapping swing path.
The five subjects from whom kinematic data were collected also wore a tight, fingerless glove (rayon/cotton/rubber blend) fitted with five 10 mm diameter reflective markers to capture corresponding kinematics data. Each glove had separate holes for the thumb and index fingers, and a single large hole for the remaining digits. Reflective markers were placed at the following landmarks on the hand: the radial and ulnar styloid processes (RSP and USP, respectively), and metacarpal (MC) heads I, II and V.
Additional markers were placed on the olecranon process (OP) and the point of the shoulder (SH). Marker bases were taped to the subjects using double sided tape to further prevent their displacement.
The Vicon kinematics motion capture system operates using high speed infrared cameras that digitally record the position of the reflective markers afixed to each subject.
Up to eight cameras simultaneously recorded each subject’s motion from various angles to provide multiple views of the same movements. During data collection subjects and
29 markers move across a calibrated space allowing extraction of each landmarks’ coordinates, from which is it possible to derive landmark velocities and accelerations, joint angles, and joint angular velocities and accelerations. Coordinate data were compiled to create a 3-D model of the relevant anatomical region. A digital camera (60
Hz) was also used to capture digital images of the reduction sequence to verify subject behavior.
After each subject was fitted with the sensor strips and glove, the Pliance system was zeroed to factor out the pressure exerted by the attachment apparatus. Subjects then progressed through the production of two Oldowan bifacial choppers. Each removed simple flakes from the flint nodule by striking the hammerstone, held in a 3-jaw chuck, against the nodule as described above. Normal force, pressure, and kinematic data were captured at 200 Hz. Recording was paused after the production of each flake and each flake was retrieved.
Associated kinematic data was captured for Subjects A-D (n = 130 swings).
Those swings with missing data were removed from the sample set, resulting in 98 sets of associated normal force/pressure and kinematics data. Swings were sectioned into up- swing, pre-strike down-swing, and post-strike down-swing. The position of the radial styloid process was used as a proxy for the wrist’s position. Up-swing was defined as encompassing the instant of the wrists’ lowest vertical position immediately prior to the initiation of the wrists’ vertical ascent through the instant of the wrists’ highest vertical position immediately prior to beginning a downward trajectory. Pre-strike down-swing was defined as encompassing the instant immediately after the wrist reached its highest vertical position through the instant immediately prior to strike. Post-strike down-swing
30
was defined as encompassing the instant immediately after strike through the wrists’
lowest post-strike vertical position.
The coupling of kinematic and force data verified that strike occurred at the
instant of cumulative peak normal force across all three digits. Cumulative peak force, or
strike, is nearly coincident with peak angular acceleration at the wrist (extension to
flexion), occurring within 2 frames or 0.01 seconds of each other (Table 2.1, Figure 2.2).
The association between peak grip force and peak angular acceleration has previously
been reported by Werremeyer and Cole (1997). The present conclusion was strengthened
by findings that peak angular acceleration is nearly coincident with strike (occurring
0.032 seconds prior to strike when recorded at 50 Hz) (Williams et al., 2010).
Captured normal force and pressure data were analyzed on a per-swing basis for each subject. Peak normal forces and pressures acting on each digit were recorded for each swing, as well as forces and pressures occurring at strike. Peak pressures reported for each digit may occur anywhere along that digit, while peak normal force reports the total normal force acting across the entire digit. Peak normal forces and pressures occurring along an individual digit constitute a series of moments distinct from the instant of cumulative peak force across all digits (i.e., strike). Impulse and pressure-time integrals were also calculated for each digit during each swing. A nonparametric
Kruskal-Wallis was used to test for differences among the digits. P-values were determined using a post hoc pair-wise Mann-Whittney U test and treated with a standard
Bonferroni correction to determine significance: Mann-Whitney pairwise P-values were multiplied by the number of number of pairwise comparisons made and determined significant if they were < 0.05 (Zarr, 1996).
31
RESULTS
Across all six subjects, peak normal forces during knapping were consistently
significantly greater on the 2nd and/or 3rd digits compared to the 1st digit (p < 0.04, Table
2.2, Figure 2.3a). Peak pressures occurring anywhere along each digit at any point in the
knapping swing were similarly distributed; peak pressures were consistently significantly greater on the 2nd and/or 3rd digits compared to the 1st (p < 0.04, Table 2.2, Figure 2.3b).
Mean peak pressures acting on the 1st digit were greater compared to the means of the 2nd and/or 3rd digits in only one subject (Subject A), however this difference was insignificant.
Manual normal force distribution at strike was similar to the distribution of peak
normal force and pressure; normal forces acting on the 2nd and/or 3rd digits were
significantly greater than those acting on the 1st digit across all six subjects (p ≤ 0.03,
Table 2.3, Figure 2.3c). Pressures acting anywhere along a given digit at strike were
significantly greater on the 2nd and 3rd digits compared to the 1st digit in all six subjects (p
≤ 0.001, Table 2.3, Figure 2.3d). Mean normal force and pressure at strike were never greater, significantly or otherwise, on the 1st compared to either the 2nd or 3rd digits.
Impulse (i.e., force applied over time) and the pressure-time integral consider not only the load applied, but the duration that load is experienced. Results for both impulse and pressure-time integrals were significantly greater on the 2nd and/or 3rd digits
compared to the 1st for five of six subjects (Subjects A-C and E-F, p < 0.03 and p <
0.0001, Table 2.4, Figure 2.4a and 2.4b). Differences between the 1st and 2nd and 1st and
3rd digits were insignificant in Subject D.
32
Using corresponding kinematic data, normal forces and pressures were separated
into up-swing, pre-strike down-swing, and post-strike down-swing for five subjects (A-
E). During up-swing two of five subjects displayed significant differences between peak
normal forces (2nd and 3rd significantly greater than 1st, Subjects B and E). All other
differences were insignificant between the 1st and 2nd and 1st and 3rd (Table 2.5, Figure
2.5a). Peak pressures were significantly greater on both the 2nd and 3rd compared to the
1st in three of five subjects (B, C, and E) and significantly greater on the 3rd compared to
the 1st in Subject D (Table 2.5, Figure 2.5b). Subject A did not exhibit significant differences in pressure. During pre-strike and post-strike down-swing, peak normal forces were significantly greater on the 2nd and/or 3rd digits compared to the 1st digit in all
subjects (p < 0.02, Tables 2.6 and 2.7, Figure 2.5c, 2.5e). Peak pressures during the pre-
strike and post-strike phases were significantly greater on both the 2nd and 3rd digits
compared to the 1st in all subjects (p < 0.015 and p < 0.035, respectively, Tables 2.6 and
2.7, Figure 2.5d, 2.5f).
DISCUSSION
Results from the present study do not support the hypothesis that the thumb is subject to significantly greater forces and/or pressures compared to the other digits during
the production of Oldowan stone tools. Peak and strike normal forces and pressures, impulse, and pressure-time integrals were consistently significantly greater on the 2nd and/or 3rd digits compared to the 1st digit in all analyses. Normal force and pressure
acting on the thumb were never significantly greater compared to either of the other
33 digits, and were insignificantly greater in only Subject A in the analysis of peak normal forces and pressures (Table 2.2, Figure 2.3a, 2.3b). Corresponding kinematic data demonstrate that differences in normal force and pressure distribution across the three digits are not consistently present during up-swing. However, these differences are established prior to strike during the pre-strike phase of down-swing and continue to strike and through the post-strike down-swing phase (Tables 2.5 – 2.7, Figure 2.5).
Expectations regarding high forces acting on the thumb during stone tool production may lie in a misinterpretation of the hand’s orientation to the hammerstone and of the knapping swing. In regard to stone tool manufacture, the precision grip is traditionally described as a grip that secures the hammerstone between the volar aspect of the pollical distal phalanx and the volar pads of one or more of the remaining digits
(Napier, 1956; Napier, 1962a; Marzke, 1997). The palm is not employed in this grip. In depicting precision grips around hammerstones, Napier (Napier, 1956; Napier, 1965) displayed the thumb as strongly abducted, flexed, and rotated towards the 3rd and 4th digits so that it lay in full opposition to the fingers. If knappers were to position their hand around the hammerstone in this manner, they would also have to execute knapping swings with the forearm positioned midway between pronation and supination—similar to position used when hammering with a traditional hammer—in order to avoid crushing the thumb between the core and the hammerstone at strike. Thus, this implies that the knapping swing is similar to a hammering swing, with an emphasis on radial and ulnar deviation at the wrist (Leventhal et al., 2010) such that force is directed up into the pollical distal phalanx.
34
If one were to knap under these conditions, the thumb would likely experience
large loads as it acts to stabilize the hammerstone throughout down-swing, particularly at
strike. Further, FLP would be heavily recruited, as the pollical distal phalanx flexes to
resist hyperextension while acting to hold the hammerstone in place (Johnson and
Forrest, 1970; Hamrick et al., 1998).
In discussions of the knapping abilities of fossil hominins, multiple researchers have invoked a traditional hammering swing (Ricklan, 1987; Marzke et al., 1992).
Others, however, have regarded the two swings as constituting distinctly different motion
patterns (Hamrick et al., 1998). The results of our knapping kinematics study upheld this distinction (Williams et al., 2010). We found that the knapping swing relies primarily on wrist extension and flexion, with the forearm pronated and the palm facing the nodule, rather than on radial and ulnar deviation. This pattern of wrist motion and forearm orientation allows the knapper to execute a wrist snap immediately prior to strike and to aim from his or her most distal joint (i.e., the wrist), thereby increasing work production and potentially improving accuracy (Bernstein, 1967; Anderson and Sidaway, 1994; Hore et al., 1996; Chowdhary and Challis, 1999). The emphasis on wrist extension and flexion necessitates that the palm of the hand is oriented towards the nodule. This placed the hammerstone directly beneath the 2nd and 3rd metacarpal heads so that throughout swing and at strike force is directed up into the 2nd and 3rd metacarpal heads. These findings
have previously been qualitatively reported by Marzke and Shackley (Marzke and
Shackley, 1986), and support their hypothesis that the hand is not oriented in a traditional
precision grip in full opposition as described by Napier (Napier, 1956; Napier, 1965).
35
Results from the present study and our previous kinematics study (Williams et al.,
2010) suggest that knappers employ the thumb as a buttress against the side of the
hammerstone in a modified precision grip (Marzke and Shackley, 1986; Marzke, 1997),
rather than using a traditional, fully-opposed precision grip. When used as a buttress, and
in combination with the knapping swing (Williams et al., 2010), the thumb is necessarily
abducted, extended and rotated laterally, rather than abducted, flexed and rotated medially in true opposition. If the thumb were flexed and rotated medially, the knapper would either be required to employ a knapping swing that does not rely on extension/flexion and accordingly does not offer the accuracy and work production advantages previously reported, or risk catching his or her thumb between the
hammerstone and the nodule as the wrist quickly flexes through strike. EMG data of muscle recruitment during stone tool production indirectly supports this reinterpretation
of the thumb’s role during knapping (Marzke et al., 1998). Marzke and colleagues
(1998) found that FLP was not consistently strongly recruited in the dominant hand, suggesting that strong pinching activities were not occurring. Activity in the intrinsic thumb and index finger muscles did, however, display a consistent tendency to peak during the swing.
Results from the present study do not support the traditional interpretation of the
thumb’s role during stone tool production. When the entire swing was considered,
normal forces and pressures acting on the hand were consistently significantly greater at
the 2nd and/or 3rd digits compared to the 1st digit (Table 2.2, Figure 2.3a 2.3b). Normal
forces and pressures at strike were also significantly greater on the 2nd and/or 3rd digits compared to the 1st digits (Table 2.3, Figure 2.3c, 2.3d). The same was true for impulse
36
and pressure-time integrals—the 2nd and/or 3rd digits were significantly greater than those acting on the 1st digit across all subjects except Subject D for whom there was not a
significant difference between the 1st and 2nd digits nor a difference between the 1st and
3rd digits (Table 2.4, Figure 2.4). Normal force and pressure differentiation did not consistently occur during up-swing, however the distribution pattern described above was established prior to strike during pre-strike down-swing and maintained through the end
of post-strike down-swing (Tables 2.5 -2. 7, Figure 2.5). These findings directly support
Marzke and Schackley (1986) hypothesis that the thumb is used as a buttressing agent
against the side of the hammerstone during stone tool production, rather than as the
primary stabilizing fulcrum against the hammerstone while oriented in true opposition.
Further, the findings highlight the roles the index and middle fingers play in stabilizing
the hammerstone during down-swing and at strike.
These results call into question hypotheses directly linking modern human thumb
anatomy specifically to the resistance of high loads experienced during stone tool
production. It now appears that the thumb is unlikely to have played the role traditionally
assigned to it during the manufacture of the earliest stone tools, however this does not
mean that the thumb is not an integral component of stone tool behaviors. Preliminary
results on pressure distribution during stone tool use demonstrate that the thumb
experiences relatively higher loads compared with the other digits when gripping flakes
with a two-jaw pad-to-side grip (Marzke, 1997), as you would hold a key. These
preliminary results suggest that the derived human thumb anatomy may be selected to
withstand high loads during stone tool use, specifically those grips requiring a two-jaw
pad-to-side grip, rather than stone tool production.
37
Table 2.1. Cumulative peak force (strike) in relation to peak angular acceleration
Subject seconds ± CI A 0.008 ± 0.0096 B 0.002 ± 0.0009 C -0.0005 ± 0.0016 D -0.001 ± 0.0016 E 0.0005 ± 0.001 Negative results indicate peak force occurred before peak angular acceleration. Positive results indicate peak force occurred after peak angular acceleration.
38
Table 2.2. Peak normal force (N) and pressure (kPa)
Peak normal N Peak kPa
Subject A, n = 24 Subject A, n = 24 1st 2nd 3rd 1st 2nd 3rd Mean 27.652 26.87 28.859 Mean 191.09 168.478 156.63 1st - - - 1st - - - 2nd 0.860 - - 2nd 0.089 - - 3rd 0.036 0.000 - 3rd 0.031 1 - Subject B, n = 36 Subject B, n = 36 1st 2nd 3rd 1st 2nd 3rd Mean 45.046 59.239 51.398 Mean 166.25 222.841 213.977 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.000 - - 3rd 0.055 0.004 - 3rd 0.107 0.228 - Subject C, n = 28 Subject C, n = 28 1st 2nd 3rd 1st 2nd 3rd Mean 28.313 46.031 47.609 Mean 93.281 173.594 213.594 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.004 0.160 - 3rd 0.000 1.000 - Subject D, n = 13 Subject D, n = 13 1st 2nd 3rd 1st 2nd 3rd Mean 20.519 39.058 21.539 Mean 60.385 190.192 114.808 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.000 - - 3rd 1 0.001 - 3rd 0.001 0.005 - Subject E, n = 29 Subject E, n = 29 1st 2nd 3rd 1st 2nd 3rd Mean 3.917 50.286 18.583 Mean 15.119 145.357 57.976 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.000 - 3rd 0.000 0.000 - Subject F, n = 18 Subject F, n = 18 1st 2nd 3rd 1st 2nd 3rd Mean 5.885 34.442 16.231 Mean 27.308 110.962 39.808 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.000 - 3rd 0.018 0.000 - Mean peak normal force and pressure occurring along each digit at any point during the swing and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
39
Table 2.3. Normal force (N) and pressure (kPa) at strike
Peak normal N Peak kPa
Subject A, n = 24 Subject A, n = 24 1st 2nd 3rd 1st 2nd 3rd Mean 20.531 23.344 25.885 Mean 116.15 147.19 135.21 1st - - - 1st - - - 2nd 0.104 - - 2nd 0.001 - - 3rd 0.004 0.000 - 3rd 0.000 1 - Subject B, n = 36 Subject B, n = 36 1st 2nd 3rd 1st 2nd 3rd Mean 44.319 57.979 48.750 Mean 115.208 216.806 206.597 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.030 0.004 - 3rd 0.001 0.228 - Subject C, n = 28 Subject C, n = 28 1st 2nd 3rd 1st 2nd 3rd Mean 23.063 47.750 37.786 Mean 70.446 161.786 149.464 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.004 0.160 - 3rd 0.000 1.000 - Subject D, n = 13 Subject D, n = 13 1st 2nd 3rd 1st 2nd 3rd Mean 18.673 38.904 20.327 Mean 39.808 184.615 94.423 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.000 - - 3rd 1 0.001 - 3rd 0.001 0.005 - Subject E, n = 29 Subject E, n = 29 Mean 2.638 49.724 16.638 Mean 10.517 133.534 48.707 1st 2nd 3rd 1st 2nd 3rd 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.000 - 3rd 0.000 0.000 - Subject F, n = 18 Subject F, n = 18 1st 2nd 3rd 1st 2nd 3rd Mean 2.611 29.486 15.667 Mean 7.778 87.500 33.889 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.000 - 3rd 0.000 0.000 -
Mean strike normal force and pressure occurring along each digit and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
40
Table 2.4. Impulse and kPa-time integral
Impulse kPa/time integral
Subject A, n = 24 Subject A, n = 24 1st 2nd 3rd 1st 2nd 3rd Mean 7.605 9.435 9.268 Mean 43.267 67.512 61.093 1st - - - 1st - - - 2nd 0.010 - - 2nd 0.000 - - 3rd 0.027 1.000 - 3rd 0.000 1.000 - Subject B, n = 36 Subject B, n = 36 1st 2nd 3rd 1st 2nd 3rd Mean 17.061 21.194 17.340 Mean 47.408 86.863 60.738 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 1 0.000 - 3rd 0.306 0.000 - Subject C, n = 28 Subject C, n = 28 1st 2nd 3rd 1st 2nd 3rd Mean 13.933 12.354 17.540 Mean 44.357 53.310 78.151 1st - - - 1st - - - 2nd 0.070 - - 2nd 0.352 - - 3rd 0.006 0.000 - 3rd 0.000 0.000 - Subject D, n = 13 Subject D, n = 13 1st 2nd 3rd 1st 2nd 3rd Mean 12.919 11.214 12.534 Mean 12.919 11.214 12.534 1st - - - 1st - - - 2nd 0.655 - - 2nd 0.655 - - 3rd 1 1.000 - 3rd 1 1.000 - Subject E, n = 29 Subject E, n = 29 1st 2nd 3rd 1st 2nd 3rd Mean 0.584 16.284 5.075 Mean 3.031 47.867 18.650 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.000 - 3rd 0.000 0.000 - Subject F, n = 18 Subject F, n = 18 Mean 1.876 9.106 8.103 Mean 6.322 27.948 15.592 1st 2nd 3rd 1st 2nd 3rd 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.000 0.375 - 3rd 0.000 0.000 - Mean impulse and kPa/time integrals and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
41
Table 2.5. Peak normal force and pressure during up-swing Peak normal N Peak kPa Subject A 1st 2nd 3rd 1st 2nd 3rd Mean 11.075 10.35 10.388 Mean 80.625 74.625 68 1st - - - 1st - - - 2nd 1 - - 2nd 0.444 - - 3rd 0.718 1 - 3rd 0.64 1 - Subject B 1st 2nd 3rd 1st 2nd 3rd Mean 18.264 27.564 21.4 Mean 51.786 108.643 86.071 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.004 0.000 - 3rd 0.000 0.02 - Subject C 1st 2nd 3rd 1st 2nd 3rd Mean 15.988 16.9 18.263 Mean 54.375 71.875 81.875 1st - - - 1st - - - 2nd 1 - - 2nd 1 - - 3rd 0.655 1 - 3rd 0.004 0.25 - Subject D 1st 2nd 3rd 1st 2nd 3rd Mean 14.5 15.577 15.212 Mean 35.385 75.385 82.5 1st - - - 1st - - - 2nd 1 - - 2nd 0.031 - - 3rd 1 1 - 3rd 0.000 1 - Subject E 1st 2nd 3rd 1st 2nd 3rd Mean 0.1 11.6 4.625 Mean 1 27.5 21.75 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.001 - - 3rd 0.001 0.006 - 3rd 0.001 1 -
Mean peak normal force and pressure separated into knapping swing phases and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
42
Table 2.6. Peak normal force and pressure during pre-strike down-swing Peak normal N Peak kPa Subject A 1st 2nd 3rd 1st 2nd 3rd Mean 19.738 23.813 26.713 Mean 114.875 149.375 140.625 1st - - - 1st - - - 2nd 0.243 - - 2nd 0.009 - - 3rd 0.000 0.686 - 3rd 0.001 1 - Subject B 1st 2nd 3rd 1st 2nd 3rd Mean 41.214 57.643 47.586 Mean 145.286 212.357 186.143 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.001 0.004 - 3rd 0.003 0.023 - Subject C 1st 2nd 3rd 1st 2nd 3rd Mean 22.088 48.875 34.638 Mean 70.750 163.375 131.75 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.016 0.033 - 3rd 0.012 0.82 - Subject D 1st 2nd 3rd 1st 2nd 3rd Mean 18.731 32.039 20.039 Mean 48.077 148.462 106.538 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.000 - - 3rd 1 0.008 - 3rd 0.000 0.082 - Subject E 1st 2nd 3rd 1st 2nd 3rd Mean 2 54.675 17.9 Mean 11.75 147.5 60.25 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.001 - - 3rd 0.001 0.001 - 3rd 0.001 0.001 -
Mean peak normal force and pressure separated into knapping swing phases and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
43
Table 2.7. Peak normal force and pressure during post-strike down-swing Peak normal N Peak kPa Subject A 1st 2nd 3rd 1st 2nd 3rd Mean 21.813 22.838 25.825 Mean 123.75 146.75 146.875 1st - - - 1st - - - 2nd 0.702 - - 2nd 0.007 - - 3rd 0.014 1 - 3rd 0.001 1 - Subject B 1st 2nd 3rd 1st 2nd 3rd Mean 44.243 55.457 44.221 Mean 159.571 219.286 189.857 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.693 0.000 - 3rd 0.033 0.040 - Subject C 1st 2nd 3rd 1st 2nd 3rd Mean 23.188 47.7 36.025 Mean 73.875 163.375 138.625 1st - - - 1st - - - 2nd 0.000 - - 2nd 0.000 - - 3rd 0.015 0.088 - 3rd 0.014 1 - Subject D 1st 2nd 3rd 1st 2nd 3rd Mean 18.904 35.077 20.442 Mean 57.115 170.385 110.192 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.000 - - 3rd 1 0.003 - 3rd 0.003 0.036 - Subject E 1st 2nd 3rd 1st 2nd 3rd Mean 1.400 53.7 18.75 Mean 7.5 140.75 54.75 1st - - - 1st - - - 2nd 0.001 - - 2nd 0.001 - - 3rd 0.001 0.001 - 3rd 0.001 0.002 -
Mean peak normal force and pressure separated into knapping swing phases and p-values from post-hoc pairwise Mann-Whittney U tests. Bolded results indicate that the 2nd or 3rd digit is significantly greater than 1st.
44
A B
Figure 2.1: Oldowan bifacial chopper produced by Subject E. The bifacial edge is shown from 2 sides.
45
6000 Force (N) Angular acceleration (m/s2) 1100 4000 1000
2000 900 ) 2 800
(m/s 0
wrist
700 (N)
‐2000 force 600 acceleration,
‐4000 400 Normal Angular 300 ‐6000
200
‐8000 100
‐10000 0 0 0.05 0.1 0.15 0.2 0.25 Seconds Figure 2.2: Relationship between angular acceleration (m/s2) at the wrist (blue) and peak normal force (red) averaged over 24 swings for Subject A. The black line indicates strike.
46
Figure 2.3, a 2nd significantly greater than 1st 108 3rd significantly greater than 1st 96
84
72
60 (N)
48 force
Peak 36
24
12
0 A BC D EF Subjects
Figure 2.3: Peak normal force and pressure occurring at any time during the knapping swing (a and b, respectively) and normal force and pressure occurring at strike (c and d) acting on digits I, II, and III.
47
Figure 2.3, b 2nd significantly greater 450 than 1st 3rd significantly greater st 400 than 1
350
300 (kPa)
250 Pressure 200 Peak 150
100
50
0 A B CD E F Subjects
Figure 2.3: Peak normal force and pressure occurring at any time during the knapping swing (a and b, respectively) and normal force and pressure occurring at strike (c and d) acting on digits I, II, and III.
48
Figure 2.3, c 2nd significantly greater than 1st 108 3rd significantly greater than 1st 96
84
72 (N)
force
60
Strike 48
36
24
12
0 A B CD E F Subjects
Figure 2.3: Peak normal force and pressure occurring at any time during the knapping swing (a and b, respectively) and normal force and pressure occurring at strike (c and d) acting on digits I, II, and III.
49
Figure 2.3, d 2nd significantly greater than 1st 320 3rd significantly greater than 1st
280
240 (kPa) 200
pressure 160
Strike 120
80
40
0 A B CD E F Subjects Figure 2.3: Peak normal force and pressure occurring at any time during the knapping swing (a and b, respectively) and normal force and pressure occurring at strike (c and d) acting on digits I, II, and III.
50
2nd significantly greater than 1st Figure 2.4,a 3rd significantly greater than 1st 32
28
24
20 (N)
16 Impulse
12
8
4
0 A B CD E F Subjects
Figure 2.4: Impulse (a) and pressure-time integral (b) acting on digits I, II, and III through the entirety of the knapping swing.
51
Figure 2.4,b 2nd significantly greater than 1st 3rd significantly st 112 greater than 1
96
80 integral
64
kPa/time 48
32
16
0 A B CD E F Subjects Figure 2.4: Impulse (a) and pressure-time integral (b) acting on digits I, II, and III through the entirety of the knapping swing.
52
2nd significantly Figure 2.5, a&b greater than 1st 3rd significantly a b greater than 1st 45 225 40 200 35 175 30 swing
‐ 150 swing up ‐ 25
up 125
(kPa)
(N), 20 100 force 15 75 pressure
Peak 10
Peak 50
5 25
0 0 A B C D E A B C D E
Figure 2.5: Peak normal force and pressure during each phase of the knapping swing, acting on digits I, II, and III during up-swing. a: peak normal force during up-swing; b: peak pressure during up-swing; c: peak normal force during pre-strike down-swing; d: peak pressure during pre-strike down-swing; e: peak normal force during post-strike down-swing; f: peak pressure during post-strike down-swing.
53
2nd significantly greater than 1st 3rd significantly greater than 1st Figure 2.5, c&d c d 360 108
swing 320 ‐ swing
‐ 96 280 down
down 84
strike 240 72 ‐ strike ‐ pre
pre 200 60 (kPa), (N),
160 48 force
A B CD E 120 pressure
36 normal
Peak 80 24 Peak 12 40
0 0 A B C D E A B C D E
Figure 2.5: Peak normal force and pressure during each phase of the knapping swing, acting on digits I, II, and III during up-swing. a: peak normal force during up-swing; b: peak pressure during up-swing; c: peak normal force during pre-strike down-swing; d: peak pressure during pre-strike down-swing; e: peak normal force during post-strike down-swing; f: peak pressure during post-strike down-swing.
54
2nd significantly Figure 2.5, e&f greater than 1st 3rd significantly e f greater than 1st
320 90 swing swing
‐ 280
80 ‐
down 70 down
240
strike 60 strike ‐ ‐ 200 post post
50 160 (N), 40 (kPa),
120 force 30
pressure 80
normal 20
Peak 40 Peak 10
0 0 A B C D E A B CD E Subjects
Figure 2.5: Peak normal force and pressure during each phase of the knapping swing, acting on digits I, II, and III during up-swing. a: peak normal force during up-swing; b: peak pressure during up-swing; c: peak normal force during pre-strike down-swing; d: peak pressure during pre-strike down-swing; e: peak normal force during post-strike down-swing; f: peak pressure during post-strike down-swing.
55
Chapter 3: Upper limb kinematics and the role of the wrist during stone tool production
ABSTRACT
Past studies have hypothesized that aspects of hominin upper limb morphology are linked
to the ability to produce stone tools. However, we lack the data on upper limb motions
needed to evaluate the biomechanical context of stone tool production. This study seeks
to better understand the biomechanics of stone tool-making by investigating upper limb
joint kinematics, focusing on the role of the wrist joint, during simple flake production.
We test the hypotheses, based on studies of other upper limb activities (e.g., throwing),
that upper limb movements will occur in a proximal-to-distal sequence, culminating in
rapid wrist flexion just prior to strike. Data were captured from four amateur knappers
during simple flake production using a VICON motion analysis system (50 Hz).
Results show that subjects utilized a proximal-to-distal joint sequence and disassociated the shoulder joint from the elbow and wrist joints, suggesting a shared strategy employed in other contexts (e.g., throwing) to increase target accuracy. The knapping strategy included moving the wrist into peak extension (subject peak grand mean = 47.3°) at the beginning of the down-swing phase which facilitated rapid wrist flexion and accelerated the hammerstone towards the nodule. This sequence resulted in the production of significantly more mechanical work, and by extension greater strike forces, than would otherwise be produced. Together these results represent a strategy for increasing knapping efficiency in Homo sapiens and point to aspects of skeletal anatomy
56 that might be examined to assess potential knapping ability and efficiency in fossil hominin taxa.
57
INTRODUCTION
Modern humans are the most eurytopic primate species the planet has ever hosted,
capable of inhabiting regions that our closest living relatives, the African apes, would
find inhospitable and even hostile to their survival. Our success has been attributed in
part to our elaborate relationship with technology, of which stone tools represent the
earliest evidence in the archaeological record (Stiner and Kuhn, 1992; Schick and Toth,
1993; Foley, 1995; Semaw et al., 1997; Wood and Brooks, 1999; Wrangham et al., 1999;
Ambrose, 2001; Lutz and Qiang, 2002; Wood and Strait, 2004; Wrangham, 2007).
Consequently, the ability to produce and use tools is recognized as a key adaptation in hominin evolution.
Hominin hands and wrists have undergone numerous alterations over the course of human evolution, many of which occurred soon after the origin of early stone tool
technologies (Tocheri et al., 2008). While major gaps remain, the fossil record
documents changes including broader apical tufts of the fingertips, a more robust thumb,
and a rearrangement of carpal and radiocarpal anatomy. Due in part to the temporal
proximity of these events, researchers have hypothesized that stone tool production was a
major selective pressure inducing some of these changes in upper limb anatomy (Napier,
1962b; Marzke and Shackley, 1986; Susman, 1988; Susman, 1994; Marzke and Marzke,
2000; Ambrose, 2001; Richmond et al., 2001; Panger et al., 2002; Tocheri et al., 2008).
With few exceptions (Marzke et al., 1998; Biryukova et al., 2005), we currently
lack the quantitative data on upper limb kinematics necessary for evaluating hypotheses
on the mechanical context of stone tool production. The current project was undertaken
58
to begin rectifying this issue by investigating upper limb kinematics during the
production of simple flakes. We examined upper limb motion patterns associated with
stone tool production as a means of testing basic assumptions underlying hypotheses
about the functional demands and selective pressures that may have been acting on the
upper limb during early hominin evolution. Susman (1998) noted that because experiments using modern humans cannot account for the mosaic combinations of primitive and derived features in fossil hominin upper limbs, their relevance to understanding the tool behaviors of early hominins is limited. While this critique may apply to some studies, it is not applicable here. The goal of the present study was to examine upper limb kinematics of knappers and the role of the wrist during stone tool production as currently practiced, enabling a more informed evaluation of some of the functional hypotheses linking the derived upper limb and wrist conditions to stone tool production. We concur with Lauder's (1995) and Marzke and Marzke and Marzke’s (2000) argument that functional data obtained through direct observation are necessary for evaluating functional hypotheses.
Compared with our closest living relatives, the African apes, human wrist anatomy differs in many respects (see Richmond et al, 2001; Tocheri et al, 2008 and references therein). Chimpanzees and gorillas possess a suite of traits related to knuckle- walking, which are thought to stabilize the wrist transversely and improve the resistance of compressive stresses experienced during the support phase of knuckle-walking (Tuttle,
1970; Jenkins and Fleagle, 1975; Richmond and Strait, 2000; Richmond et al., 2001;
Kivell and Schmitt, 2009). A component of this suite is a distally-projecting dorsal ridge
of the distal radius (Tuttle, 1967; Richmond and Strait, 2000). During the support phase,
59
as the wrist extends the dorsal portion of the distal radius reaches a close-packed position
with maximal articular congruence with the lunate and scaphoid that, in conjunction with
palmar ligaments, maintains a stable joint and prevents the wrist from further extension
(Tuttle, 1967; Jenkins and Fleagle, 1975; Richmond and Strait, 2000). In this manner,
the dorsal ridge of the radius plays a key role in minimizing the degree of maximum wrist
extension and contributing to a stable support column during knuckle-walking.
Empirically, average ranges of maximum wrist extension are greater in modern humans
[x = 70o (Almquist, 2001)] than in Gorilla [x = 58o (Tuttle, 1969)] and Pan [x = 34o
(Tuttle, 1967; Tuttle, 1969; Jenkins and Fleagle, 1975; Richmond, 2006)] (Table 3.1).
An African ape-like projecting dorsal ridge is present in Australopithecus anamensis and Au. afarensis, suggesting a lower range of wrist extension compared with later hominins (Richmond and Strait, 2000). Lovejoy et al. (2009) argue that the carpal
anatomy of Ardipithecus ramidus indicates that the wrist was capable of greater degrees
of extension than that observed in African apes. However, their analysis unfortunately
does not demonstrate a link between midcarpal morphology and ranges of motion in
extension, and lacks sufficient comparisons with primate taxa other than African apes and
humans to permit clear functional interpretations about wrist mobility. Although data
were not presented, Lovejoy et al (2009; see also White et al. 1994) imply that the distal
radius of Ar. ramidus has an African ape-like distally-projecting dorsal ridge. Therefore,
the functional significance of the Ar. ramidus wrist remains open to question and requires
further analysis. Furthermore, Ar. ramidus probably post-dates the Pan-Homo last
common ancestor (LCA) by over 1 million years (White et al., 2009). Without direct
evidence of the LCA, wrist morphology in the LCA will remain a matter of debate.
60
What can be concluded with certainty is that over the course of human evolution
the morphology of the wrist has undergone significant modifications, including a change
in the distal radius from an African ape-like morphology in Au. anamensis and Au.
afarensis to a relatively flat, modern human-like distal radius in later hominins
(Richmond and Strait, 2000; Richmond et al., 2001). Researchers have hypothesized that
such anatomical changes in wrist anatomy may have offered later hominins, including
Homo, increased wrist mobility that facilitated a variety of behaviors including throwing
and stone tool production (Marzke, 1971; Richmond and Strait, 2000; Ambrose, 2001;
Richmond et al., 2001). However, this hypothesis rests on the assumption that wrist mobility, particularly in extension, plays an important role in stone tool production.
In order to examine the role of the wrist in stone tool-making, this study tests several hypotheses. Previous research on upper limb activities such as writing, throwing, and piano playing has demonstrated that subjects maintained general kinematic uniformity (e.g., sequence of kinematic events) within a given task. Specific patterns
(e.g., kinematic values), however, varied among subjects and across competency levels
(Newell and Van Emmerik, 1989; Hore et al., 1996; Fleisig et al., 1999; Minetti et al.,
2007). We predict that during stone tool production knappers will similarly demonstrate
consistent gross upper limb motion patterns.
Biomechanics research on activities such as pitching (Debicki et al., 2004), dart throwing (McDonald et al., 1989; Jeansonne, 2003), hammering (Cote et al., 2005), and soccer kicking (Putnam, 1991) show that the consistent gross movement pattern of the
limb occurs in a proximal-to-distal sequence. This pattern allows for greater accuracy because movements of proximal joints have larger effects at the end of the limb than do
61
movements of distal joints (Hore et al., 1996). Therefore, the distal joints are thought to
act late in order to refine the final position before object release or strike. Distal joint
kinematics (e.g., position and velocity) have a demonstrated effect on accuracy in upper
limb activities such as pitching a baseball and striking a target (Southard, 1989; Hore et
al., 1996; Hirashima et al., 2007). In this study, we test the hypothesis that during
knapping upper limb movements will similarly occur in a proximal-to-distal sequence,
culminating in peak wrist extension followed by rapid wrist flexion just prior to strike.
Finally we predict, based on previous upper limb kinematics research, that wrist
motion will significantly influence knapping mechanical efficiency and strike accuracy.
In their EMG study of muscle recruitment during stone tool production, Marzke et al.
(1998) reported that all subjects maximally recruited their flexor carpi ulnaris (FCU)
muscles (the only forearm muscle monitored whose chief action is motion at the wrist
and/or midcarpal joint) during down-swing. FCU activity and wrist flexion may help
knappers reach higher joint velocities, thereby producing more mechanical work and greater strike forces than could be achieved with a rigid wrist (Bunn, 1955; Putnam,
1991). Here, we test the hypothesis that during knapping wrist movements significantly
increase mechanical work. We also test the hypothesis that wrist extension in particular
plays an important role in producing this increased mechanical work.
62
METHODS
Sample
Data were captured from four knappers; three males (Subjects A-C), and one
female (Subject D). Two of the subjects periodically participated in knapping (i.e., fewer
than five times per year, Subjects A and C), and two had limited-to-no prior knapping
experience (Subjects B and D). All subjects were healthy, right hand dominant adults
free from muscular and/or osteological conditions that may have compromised their
motion patterns.
Raw materials
Experiments were conducted in cortex-free raw Texas flint (material toughness =
1 Kc, unpublished data from Herzl Chai). The mass and dimensions of the nodules were
initially similar (n = 8, mean mass = 5.85 kg; stdev = 1.22; mean length = 40.5 cm, stdev
= 2.76; mean circumference = 34.1 cm, stdev = 6.65). Flint was obtained from
Neolithics.com. A single quartzite hammerstone was used for all flake production (0.765 kg).
Motion capture
Kinematics data were captured using the VICON motion capture analysis system
in The George Washington University’s Motion Capture and Analysis lab. The VICON
system uses high speed cameras to digitally record reflective markers applied to subjects as they move across a calibrated space, allowing extraction of motion data such as
63 landmark coordinates, joint angles, joint (e.g., wrist) angular velocity and acceleration, and segment (e.g., forearm) velocity and acceleration. Multiple cameras are linked and simultaneously record the subject’s motion from various angles, thereby providing multiple views of the same movement. Six to eight infrared cameras and one digital video recorder were used to capture each subject’s knapping motions, recorded at 50 Hz.
Each subjects’ dominant hand (i.e., hammer hand) was fitted with a tight, fingerless glove (rayon/cotton/rubber blend) with separate holes for the thumb and index finger, and a single large hole for the three remaining digits. Six 10-mm diameter reflective markers were affixed to the glove and arm at the following landmarks: the olecranon process (OP), the radial and ulnar styloid processes (RSP and USP, respectively), and metacarpal (MC) heads I, II and V (Figure 3.1). Each marker consisted of a base with a short rod projecting upward and a reflective globe. Globes can be screwed and unscrewed from the rods to secure or unsecure them from the bases.
Markers were affixed to the appropriate position on the glove by pushing the rod through the glove from the inside and screwing the globe back onto the rod such that the glove was sandwiched between the base and globe of each marker. Marker bases were taped to the subjects using double sided tape to further prevent their displacement. Markers had initially been taped directly to subjects without the glove; however their frequent displacement upon strike necessitated the alternative method described above.
Data capture occurred in two phases for each subject. Phase 1 consisted of recording subjects’ excursions at the wrist (extension, flexion, radial deviation, and ulnar deviation). Data were captured from a “neutral” position in which subjects held their dominant arm flexed at the shoulder joint with the arm, forearm, and palm parallel to the
64
floor, fully pronated with the palm facing the floor and fingers fully extended. Subjects sequentially moved their wrist in each direction to their natural excursion maximum and held the position for five seconds before releasing, returning to the neutral position, and proceeding to the next direction. We note that the wrist may be capable of greater ranges of movement than these natural excursion maxima due to accelerations during rapid movements or if subjected to an external force. However, these provide baseline excursion maxima based on each subject's voluntary muscle activity.
Phases 2 consisted of data capture during flake production. Subjects removed simple flakes from the nodule without regard to flake dimensions or mass by forcefully striking the hammerstone against the nodule, which subjects balanced on their left leg.
Recording was halted after the successful production of each flake and the flake was
retrieved and labeled to allow later association of each flake with its appropriate knapping trial and landmark coordinate data. Data for each subject were captured at two or more separate knapping sessions to prevent fatigue of the upper limb.
Data analysis
Captured coordinate data were sectioned into individual swings. Those swings with relevant missing data were removed from the sample set, resulting in 66 flake production knapping cycle swings (Table 3.1). When subjects did not flex or deviate in the ulnar direction past their neutral position reported flexion and ulnar deviation means represent degrees above the neutral position (i.e., extension or radial deviation). The position of the radial styloid process was used as a proxy for the wrist’s position to demarcate swing initiation and termination, vertical and lateral excursions, and the
65
transition from up-swing to down-swing. Swing initiation was demarcated by the lowest
vertical position of the wrist immediately prior to the initiation of the wrist’s vertical
ascent during up-swing. Swing termination was demarcated by the lowest vertical
position of the wrist immediately following the termination of the wrist’s vertical decent
during down-swing. The transition from up-swing to down-swing (TR) was demarcated by the highest vertical position of the wrist during each swing (Figure 3.2). Motion at the shoulder joint was estimated by tracking directional changes in the horizontal plane at the
OP in relation to its starting position (e.g., the transition from flexion to extension during the knapping cycle).
Wrist extension/flexion and radial/ulnar deviation angles were calculated using captured coordinate data from the OP, RSP, USP, and the MC II head, and evaluated through the course of each swing. Using the midpoint between the RSP and USP as the angle vertex (Point A), the following procedure was used to derive extension/flexion angles (ΦDV [dorsal/ventral], Figure 3.3a) and radial/ulnar deviation angles (ΦML
[medial/lateral], Figure 3.3b). Two planes were established: 1) Forearm Plane ML, representing 0° of extension/flexion or the neutral plane, defined by the RSP, USP and
OP, defined as Point B, and 2) Forearm Plane DV, the plane perpendicular to Forearm
Plane ML which runs along Line AB, representing 0° radial/ulnar deviation. The MC II
3 head , defined as Point C, was projected onto Forearm Plane DV (Point CDV) for
3 The ulnar deviation angles reported here, based on the MCII head, are likely to be systematically lower than those reported by studies that used the MC III head to calculate radial and ulnar deviation Youm Y, McMurthy RY, Flatt AE, and Gillespie TE (1978) Kinematics of the wrist. I. An experimental study of radial-ulnar deviation and flexion-extension. J Bone Joint Surg Am 60:423-31 Moritomo H, Goto A, Sato Y, Sugamoto K, Murase T, and Yoshikawa H (2003) The triquetrum-hamate joint: an anatomic and in vivo three-dimensional kinematic study. J Hand Surg Am 28:797-805 Murgia A, Kyberd PJ, Chappell PH, and Light CM (2004) Marker placement to describe the wrist movements during activities of daily living in cyclical tasks. Clin Biomech (Bristol, Avon) 19:248-54., but ranges of motion would be comparable. 66
calculation of ΦDV or onto Forearm Plane ML (Point CML) for calculation of ΦML.
Extension/flexion angles (ΦDV) were calculated as the angle supplementary to BACDV.
Radial/ulnar deviation angles (ΦML) were calculated as the angle supplementary to
BACML. All angles were calculated relative to each subject’s neutral position as recorded during Phase 1. For statistical analyses, all wrist excursion angles were measured in degrees and converted to radians to standardize measurements.
Vertical velocity (v, m/s) and acceleration (a, m/s2) were calculated through each
swing at time (t) and position (x) for each landmark using captured coordinate data
according to:
X(t+1) - X(t-1) v (t) = (Eq. 1) (t+1) - (t-1)
V - V a = (t +1) (t -1) (Eq. 2) (t) (t+1) - (t-1)
Angular velocity and angular acceleration at the wrist were calculated through the course
of each swing using ΦDV. These angles replaced x in Eq. 1. Wrist angles, velocity,
acceleration, angular velocity, and angular acceleration were derived using the R
statistical program language, versions 2.5 and 2.7 (Ihaka and Gentleman, 1996).
Total work per swing (Wtotal, measured in Joules) was defined as the sum work for the up-swing and down-swing phases for each swing:
Wtotal = Wup + Wdown (Eq. 3).
The coordinate position of the MC II head was used as a proxy for the hammerstone
position for all work production calculations, unless otherwise stated. Up-swing work
production was calculated according to:
Wup = FΔdup (Eq. 4), where
67
F = (9.8 m/s2)(m);
m = hammerstone mass (kg); and
Δdup = (maximum hammerstone vertical position – hammerstone vertical position at
swing initiation).
Down-swing work production was defined as the difference between maximum kinetic
energy of the hammerstone and maximum potential energy of the hammerstone:
1 2 2 Wdown = ( /2mv ) – (9.8m/s mΔddown) (Eq. 5), where
m = hammerstone mass (kg);
v = maximum hammerstone velocity (m/s); and
Δddown = (maximum hammerstone vertical position – hammerstone vertical position at
strike).
In order to maintain the error rate across multiple (k) comparisons, a modified
Bonferroni adjustment method was employed to determine significance (alpha: 0.05) for
all analyses: the k p-values were ordered and the smallest p-value compared was
compared to 0.05/k; if that was found to be significant, then the next smallest p-value was
compared to 0.05/(k-1), etc. (Holm, 1979).
RESULTS
Upper limb motion patterns
Flake production knapping cycles were divisible into two phases with distinct, consistent sets of motion: an up-swing and a down-swing phase. Up-swing was
characterized by upward limb motion, flexion of the shoulder and elbow joints, and
68
increasing wrist extension. Down-swing was characterized by downward limb motion,
extension at the shoulder joint, continued elbow flexion and wrist extension through peak
wrist extension, followed by rapid elbow extension and wrist flexion (Figure 3.4).
Following the termination of down-swing, motion patterns varied widely by subject and trial. Knapping was often halted between swings to adjust the core or hammerstone.
Subjects rarely maintained a fluid knapping rhythm, in which they proceeded directly into the next up-swing following down-swing termination.
Comparison of the timing and magnitude of peak linear velocities of limb segment endpoints was used to evaluate joint coordination through down-swing—the knapping phase that is responsible for positioning the hammerstone to strike the nodule.
All subjects began down-swing with velocity increasing in the negative direction (i.e., downwards) at the OP, RSP, and MC II head (Figure 3.2). The OP reached peak linear velocity first, then reduced velocity. The RSP and the MCII head continued to increase velocity inferiorly, until their velocities peaked just prior to strike. Post strike, linear velocities of all segment endpoints quickly decreased inferiorly, approaching zero velocity (i.e., all segment endpoints decelerated while continuing to travel downwards).
The temporal onset of upper limb joint peak linear velocities showed a partial proximal- to-distal relationship, with the OP peaking significantly before the RSP and the MCII head (p < 0.003 and p < 0.001, respectively). All temporal differences between the RSP and MCII head were insignificant (Table 3.2). Peak linear velocity at each segment endpoint proceeded in a complete proximal–to–distal fashion, with velocity significantly increasing from the OP to RSP (p < 0.0001), and from the RSP to MC II head (p < 0.05) for each subject (Table 3.3).
69
Wrist extension/flexion and radial/ulnar deviation patterns
Subjects’ peak extension range encompassed 30.6° - 70.1° (total Subject low and high values, respectively). Subjects’ peak knapping extension grand mean was 47.3°
(Table 3.1). Measured relative to their respective neutral positions, Subjects C and D utilized 58.7% and 56.4% of their muscular-induced extension range, respectively, while
Subjects A and B utilized 82.5% and 98.4%, respectively (Table 3.4). Negative values represent the extent to which subjects failed to flex past their neutral position. Subjects A and B passed their passive muscular-induced extension maxima (as recorded in Phase 1) in 17% and 30.8% of their trials, respectively. All subjects did not approach their passive muscular-induced flexion maxima, flexing on average 12.82° – 34.14° below their knapping extension peak and all failed to flex beyond their respective neutral positions except Subject C in 20% of his trials.
Radial deviation was emphasized over ulnar deviation. All subjects employed at least 35% of their radial deviation range, but only one (Subject B) deviated in the ulnar direction past his neutral position (Tables 3.1 and 3.4). All subjects employed significantly more absolute motion in the dorsal-ventral plane compared to the radial- ulnar plane (Bonferroni adjusted p < 0.02, Figure 3.5). However, each subject employed similar percentages of their respective total available range of motion in the dorsal- ventral and radial-ulnar planes (Table 3.4).
The timing of peak extension, angular velocity and angular acceleration were temporally constrained across all subjects in relation to both the transition to down-swing and to strike (Table 3.5).
70
Work production
Knapping work production was calculated at the MCII head as described above to generate a work production baseline. To isolate the contribution to work production gained through increased velocity at the MCII head, work production was calculated a second time by substituting velocity and vertical displacement at the MCII head with the
RSP. All subjects produced significantly more work employing the greater velocities achieved at the MC II head during both phases of flake production (p ≤ 0.006, Table 3.6).
DISCUSSION
The results of this study demonstrate that the wrist plays an important role in simple stone tool production, and supports the hypotheses set out in the introduction.
First, all knappers displayed broadly similar upper limb motion patterns. This was evident in the temporal consistency of significant events within each subject (i.e., segment endpoint velocity [Table 3.2]) and among all subjects (i.e., peak wrist extension, angular velocity, and angular acceleration [Table 3.5]) during the down-swing phase.
These similarities were expected given the task uniformity and subjects’ prior exposure to knapping demonstrations and theory.
The second hypothesis that during down-swing upper limb movements will occur in a proximal-to-distal sequence culminating in rapid wrist flexion just prior to strike was also supported. The shoulder, elbow, and wrist joints moved in a coordinated fashion through the up-swing phase. However, subjects replaced a rigid upper limb with mobile joints that act synergistically—a transition from simple to complex motion patterns
71
(Newell and Van Emmerik, 1989)—following the transition to down-swing. A mobile upper limb offers two advantages over maintenance of a rigid limb. One, subjects are
able to utilize a proximal-to-distal joint sequence in which the most proximal joint begins
forward motion before the distal joints, and the proximal joint reaches peak linear
velocity and begins to slow down prior to the distal joints reaching their respective peak
linear velocities (Figure 3.2; Tables 3.2) (Putnam, 1991). This motion sequence can
result in a velocity “summation effect” at the most distal joint due to torque interactions
among the preceding joints such that the distal joint experiences greater velocities than
could otherwise be achieved (Bunn, 1955; Southard, 1989; Putnam, 1991). The results of
this summation effect are evident in the knapping kinematics reported here; segment
endpoints’ linear velocities significantly increased from the OP to RSP and from the RSP
to MC II head during flake production knapping cycles (Table 3.3).
The second advantage of a mobile upper limb is the accuracy increase afforded by disassociating motion of the distal joint from the proximal joints (Bernstein, 1967;
Arutyunyan et al., 1968; Newell and Van Emmerik, 1989; Southard, 1989; Anderson and
Sidaway, 1994; Hore et al., 1996). Bernstein (1967) hypothesized that additional degrees
of freedom are liberated through joint disassociation, thus increasing the potential to
coordinate a greater number of degrees of freedom and increase motor control. However,
Hore et al. (1996) cautioned that the magnitude of the effects of joint kinematics on strike
accuracy increases in a proximal-to-distal fashion. Therefore, the wrist joint plays a
greater role than more proximal joints in determining strike accuracy by virtue of its
distal position. Joint disassociation was evident in the timing and magnitude of segment
endpoints’ peak linear velocities (Tables 3.2 and3. 3). Temporal disassociation between
72
the RSP and MC II head was not evident, which may be due to subjects’ relative inexperience with this behavior, resulting in reduced capability to achieve joint disassociation (Bernstein, 1967). Given its utility in other accuracy-seeking activities,
such as pistol shooting, writing, throwing, and kicking, we suggest that joint
disassociation is naturally similarly employed during stone tool production to increase
strike accuracy (Bernstein, 1967; Arutyunyan et al., 1968; Newell and Van Emmerik,
1989; Southard, 1989; Anderson and Sidaway, 1994; Hore et al., 1996). Ongoing kinematic studies are being conducted with skilled knappers to directly measure strike accuracy as it relates to joint disassociation.
The consistent proximal-to-distal sequence supports the hypothesis that the wrist
plays an important role in stone tool production, through greater mechanical work and
likely greater accuracy. The wrist reaches peak extension 0.05 seconds after the
transition to down-swing, setting up the wrist for flexion prior to strike (Table 3.5). The
degree of wrist extension (individual means of 36 °– 66o for subjects in this study) used is
likely a function of the manner in which the moment arms and tensile strengths of the
forearm’s flexor muscles change with the degree of wrist extension/flexion. The
relationship between the moment arms of two of the forearm’s flexors (FCU and flexor
carpi radialis) and wrist posture approximate positive second order polynomial
relationships (Pigeon et al., 1996), meaning that their mechanical advantage increases as
wrist extension increases. Thus, by utilizing greater degrees of wrist extension knappers are able to more fully exploit the forearm’s flexors.
The rapid release of peak wrist extension (i.e., burst of wrist flexion), illustrated by angular velocities and accelerations recorded at the wrist, peaked at 0.019 seconds and
73
0.032 seconds prior to strike, respectively (Table 3.5). In their EMG study of muscle
recruitment during stone tool production, Marzke et al. (1998) reported that FCU muscle
activity in the hammer hand peaked immediately prior to strike, inducing a wrist “flick” which further accelerated the hammerstone towards the nodule. The peak angular velocities and angular accelerations recorded here are likely due in part to strong FCU
recruitment following peak wrist extension.
The relationship between muscles’ mechanical advantage and the position of the
wrist also appears to influence the degree of flexion subjects utilized. All subjects
employed greater than 50% of their respective extension range, but subjects flexed
minimally out of peak extension, and typically avoided wrist flexion past their neutral
position (Tables 3.1 and 3.4). Maintaining the wrist in an extended position may reflect a
strategy to maintain hammerstone control against the strong reaction forces produced at
strike. According to Pigeon et al., (1996), the digital flexors are weaker when the wrist is
held in a flexed position compared to an extended position, which may render the
hammerstone more susceptible to displacement when the wrist is strongly flexed. By
avoiding excessive wrist flexion the knapper may be better able to maintain a tighter grip
on the hammerstone. The avoidance of strong flexion, particularly past Forearm Plane
ML, has also been reported in pitching activities (Debicki et al., 2004), during which
flexion-inducing interactive torques produced by the more proximal upper limb segments
and experienced at the distal joints were dampened by heightened forearm extensor
activity. The dampening effect on the part of the extensors allowed pitchers to better
control wrist flexion and ball release, thereby increasing target accuracy.
74
The results of this study also support the final hypothesis that rapid wrist flexion from an extended position significantly contributes to the production of mechanical work