A Complex Systems Theory of Technological Change: A Case Study Involving a Morphometrics Analysis of Stone Age Flake Debitage from the Horn of Africa

Material Information

A Complex Systems Theory of Technological Change: A Case Study Involving a Morphometrics Analysis of Stone Age Flake Debitage from the Horn of Africa
Copyright Date:


Subjects / Keywords:
Archaeology ( jstor )
Aspect ratio ( jstor )
Complex systems ( jstor )
Excavations ( jstor )
History of technology ( jstor )
Ideal type ( jstor )
Stone age ( jstor )
Stone tools ( jstor )
Technological change ( jstor )
Trucks ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Erich Christopher Fisher. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
71315356 ( OCLC )


This item has the following downloads:

Full Text




Copyright 2005 by Erich Christopher Fisher


This document is dedicated to my family a nd friends for their continued support. Thank you all.


iv ACKNOWLEDGMENTS This thesis is the product of massive support from many wonderful individuals. Most pertinent to this thesis is the support and guidance of my supervisory committee. Steve Brandt introduced me to Ethiopia, and during each trip helped me develop the ideas that eventually became the chapters within this thesis. He has provided unwavering support for my myriad ideas and research opport unities. He has also become a wonderful friend and colleague. Susan Gillespie has opened my eyes to avenues of knowledge I never dreamt existed through he r voluminous knowledge of ar chaeological theory. Her critical comments about my research and ideas have pushed me to achieve more than I knew possible for myself. She is a thoughtful and caring person whom I am proud to know. Michael Binford has expanded my focus beyond the narrow confines of archaeology and has provided invaluable comm ents on my research. I would also like to thank my committee as a whole for enduri ng my successes and failings throughout this process with extreme patience and understanding. I consider each of these individuals true friends and I am indelibly honored that I was fortunate enough to work with each of them. I want to thank each of my committ ee members very much, and I hope that I will never let any of them down. I also wish to acknowledge the unending s upport of my family and friends. This thesis and degree would never have been po ssible without my parents. Words alone cannot describe how indebted I am to th em for their unending care, devotion, and concern, even on those rare occasions when I know they have absolutely no interest in


v what I am talking about. My brother has added that little bit of excitement in my life that makes life worth living and I hope he has found his future as I have mine. I am also equally indebted to my Grandmother a nd Grandfather Bunnell and Grandmother and Grandfather Fisher. Not only di d they bring me to my first archaeological site but their enthusiasm and encouragement over the years ha ve fueled my ambitions in life. And to my friends, my life would have been so bor ing without them. Drew, Mike, Josh, Luke, and Walker keep my boring personality in check There is no better friend than Sarah. And Jim, life is all about the peaches and tuna fish. I would also like to thank my many friends within the Department of Anthropol ogy who have given me reason to enjoy life and debate all those inco nsequential little detail s. Thank you everyone.


vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT....................................................................................................................... xi CHAPTER 1 TECHNOLOGY, STONE TOOLS AND THE HORN OF AFRICA..........................1 Introduction................................................................................................................... 1 Impetus for this Research.............................................................................................3 Problems Associated with the Study of Technological Change...................................6 Resolving the Problems of Prehistoric T echnological Change in the Archaeology of the Horn of Africa..............................................................................................10 Conclusion..................................................................................................................11 2 COMPLEX SYSTEMS THEORY.............................................................................13 Introduction.................................................................................................................13 Complex Systems Theory, Chaos Theory, and Non-linear Theory: Relationships and Characteristics.................................................................................................13 Characteristics of Complex Systems Theory......................................................15 Context within Technological Systems...............................................................17 Technological Change within Complex Systems................................................19 Ideal Types..........................................................................................................20 Intentional and Unintentional Changes...............................................................25 The Technological Life Cycle....................................................................................30 The Trajectory of Long Term Technological Systems........................................31 Growth and Complexity of a Technological System..........................................32 Peak.....................................................................................................................34 Decline.................................................................................................................34 Conclusion..................................................................................................................35


vii 3 PUTTING A SHAPE ON TECHNOLOGY...............................................................39 Morphometrics............................................................................................................39 Standardization....................................................................................................41 Measures of Shape and Dimension.....................................................................44 Flakes...................................................................................................................47 Conclusion..................................................................................................................49 4 THE GILGEL GIBE AND GOGOSHIIS QABE MORPHOMETRICS ANALYSES...............................................................................................................54 Introduction.................................................................................................................54 Liben Bore..................................................................................................................55 Morphometrics Analysis.....................................................................................57 Understanding the Liben Bore Data....................................................................60 Gogoshiis Qabe...........................................................................................................65 Morphometrics Analysis.....................................................................................65 Understanding the Gogoshiis Qabe Data............................................................69 Conclusion..................................................................................................................72 5 CONCLUSIONS........................................................................................................88 APPENDIX A AN EXPERIMENT ON TECHNOLOGI CAL PROCESS USING STONE TOOLS.......................................................................................................................95 B MATHEMATICAL FORMULAE...........................................................................103 BIOGRAPHICAL SKETCH...........................................................................................113


viii LIST OF TABLES Table page 4-1 T-test results of the inter-period differen ces in the mean values for formfactor, roundness, aspect ratio, elongati on, and X-feret and Y-feret...................................74 4-2 Averaged standard deviation of the formfactor, aspect ratio, and roundness values organized per excavation level time period..................................................74 4-3 T-test results of the inter-period differen ces in the mean values for formfactor, roundness, aspect ratio, elongati on, and X-feret and Y-feret...................................74


ix LIST OF FIGURES Figure page 2-1 The technological life cycle of thr ee different technological systems.....................38 3-1 Measurement of flake “length” varies based on the methods used..........................50 3-2 The steps necessary to conduct a mor phometrics analysis using computerassisted shape software.............................................................................................51 3-3 X-Feret and Y-Feret represent measure th e widest points of an object dependent upon an object’s orientation to the X and Y axis.....................................................52 3-4 In contrast to X and Y-Fe ret (figure 2-4), Figure 3-5 illustrates how the Image Processing Toolkit represents “lengt h” and “breadth” of an object.........................52 3-5 How different morphometrics statistics measure different shapes of objects..........53 4-1 Morphometric formfactor values of the Li ben Bore flakes organized by the level of excavation............................................................................................................75 4-2 Morphometric roundness values of the Libe n Bore flakes organized by the level of excavation............................................................................................................76 4-3 Morphometric aspect ratio values of th e Liben Bore flakes organized by the level of excavation...................................................................................................77 4-4 Morphometric elongation values of the Li ben Bore flakes organized by the level of excavation............................................................................................................78 4-5 Morphometric X-Feret and Y-Feret ratio values of the Liben Bore flakes organized by the level of excavation........................................................................79 4-6 Irregularity (formfactor) divided by the roundness of the Liben Bore flakes organized by the level of excavation........................................................................80 4-7 A hierarchical cluster analysis using the mean and standard deviation values per excavation level for formfactor, roundness, and aspect ratio from Liben Bore.......81 4-8 Morphometric formfactor values of the Gogoshiis Qabe flakes organized by the level of excavation...................................................................................................82


x 4-9 Morphometric roundness values of the Gogos hiis Qabe flakes organized by the level of excavation...................................................................................................83 4-10 Morphometric elongation values of the G ogoshiis Qabe flakes organized by the level of excavation...................................................................................................84 4-11 Morphometric aspect ratio values of the Gogoshiis Qabe flakes organized by the level of excavation...................................................................................................85 4-12 Morphometric X-Feret and Y-Feret ratio values of the Gogoshiis Qabe flakes organized by the level of excavation........................................................................86 4-13 A hierarchical cluster analysis usi ng the standard deviation values per excavation level for formfactor, roundne ss, elongation, and aspect ratio from Gogoshiis Qabe........................................................................................................87


xi Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts A COMPLEX SYSTEMS THEORY OF TECHNOLOGICAL CHANGE: A CASE STUDY INVOLVING A MORPHOMETRICS ANALYSIS OF STONE AGE FLAKE DEBITAGE FROM THE HORN OF AFRICA By Erich Christopher Fisher May 2005 Chair: Steven Brandt Major Department: Anthropology This thesis provides theoretical, archaeo logical and experimental archaeological evidence to support a complex systems framew ork for identifying and explaining lithic technological change within the Stone Age of the Horn of Africa. Complex systems theory is used to provide an interpretati on of the process of technological change by describing the proximal effects of individua ls on larger, longer-term technological systems via multi-scalar spatial and temporal dynamics, temporal continuity, recognition of material and non-materi al archaeological culture, and the intentional and unintentional actions of technol ogical producers. Technological change is proposed here to be found in the modification of “ideal t ypes,” which describe how a technology is produced and used by members of a technological system. To support a complex systems theory approach to technological ch ange, this thesis presents the analysis of two Stone Age sites from the Horn of Africa. Each of the two Stone Age sites was analyzed using com puter-assisted morphometrics—the study of


xii shape—to identify the changes in flake shape through time. Changes to flake shape are correlated with modifications to the ideal type of the technology and are taken to represent technological ch ange through time. The results of the morphometrics analysis suggest that this method is a useful research tool. In particular, the morphometrics analyses are ab le to identify technological changes and technological variations within the lithic technological systems at both Stone Age study sites. Using complex systems theory to interpret these re sults, it appears that there were several independent technological ch anges at each site th at may correlate to changes in the “ideal types” of these technological systems.


1 CHAPTER 1 TECHNOLOGY, STONE TOOLS AND THE HORN OF AFRICA Introduction Stone Age African archaeology research has held a long-standing relationship with the cultural historical met hod since Wayland applied culture historical ideas to his “pluvial sequence” in the early twentieth century (Robertshaw 1993:20, 79). Wayland believed that the pluvial method was a way to examine the relationship between African and European cultures by recording the loca lized pluvial episodes at an archaeological site. Each African pluvial data was compared against Europe an glacial sequences because it was believed that this would provide a re lative dating method for African archaeology (for example, see Wayland 1930). Louis Leakey also used a culture histor ic method through his application of the pluvial method. Leakey’s use of the pluvial method is now referred to as the “organic model” because of his gene ral conception of cultures as organic forms (Robertshaw 1993:80). In particular, Leakey defined culture s based on highly specif ic artifact types, disregarded functional explanations, consider ed change to be exogenous, believed that variability was a product of temporal differences, and na rrowly associated different human types with specific cu ltures (Robertshaw 1993:81). Perhaps the most prominent applicati on of culture histor icism in African archaeology is the current rendition of the three-age se quence developed by A.J.H. Goodwin and Clarence van Riet Lowe (1929). The development of their sequence was contemporaneous with Wayla nd’s pluvial sequence but instead of relying on extra-


2 African data sources, such as European glacial sequences, Goodwin and LoweÂ’s classification was based primarily on local ly recovered (South African) technological material culture (McBrear ty and Brooks 2000:456). However, the use of culture history today presents two major problems to contemporary theories of technological cha nge within African Stone Age archaeology: interpretation and continuity. First, cultu re history does provide a useful methodology to organize large bodies of data by using trait lists to create large scale t ypological classes. This is especially useful in the Horn of Africa where the archaeological record of stone tool use is temporally greater than anywhere else in the world and may span more than 2.5 million years (cf. Semaw 2000). And when compounded by an equally great spatial area, the culture historical approach presents a useful system to manage an incredible diverse (spatio-temporal) body of lithic archaeologi cal data in the Horn of Africa (i.e., the three-age sequence developed by Goodwin and Lowe 1929). On the other hand, the culture historical method also limits the interpretations of data and may also distort actual trends within a dataset. When an archaeological site is studied, a culture historical approach creates massive poten tial for the misinterpretation of archaeological data because it relies on over-general ized, large-scale chronostratigraphic trait lists. McBrearty (1988) notes that the assignment of connotation-laden chrono stratigraphic periods (i.e., Early Stone Age, Middle Stone Age, and Late Stone Age) to artifacts based on t echno-typological characte ristics is dangerous and even misleading. Poorly studied data ma y be identified into a culture historical typology that, by definition, has to have a s ubstantial amount of referential meaning attached to each typological class. As a resu lt, these data are direc tly associated with the


3 particular meanings of the t ypological classification regard less if the typology is truly accurate or not (cf. McBrearty 1988). Second, culture history presents a nondynamic and discontinuous conception of time and change because the use of essentialism restricts any idea of change to a series of distinct periods or phases wit hout any leeway to consider co ntinuity through time. The essentialism of culture histor y requires that the traits us ed to create typologies are considered real and unique, a nd are manifest in artifacts and representative of only a single “cultural” or “technological” type or time period. Therefore, each trait must fall into only one category. Due to this very pr oblem, the Third Pan-African Congress (Clark 1957) introduced the use of “transitional pe riods” because certain Stone Age industries (i.e., Sangoan) showed features characteristic of two different pe riods and thus proved difficult to place into one specific cla ssification (McBrearty and Brooks 2000). Therefore, in order to model technological change as a continuous process through time, and be able to measure these changes accurately, the preexisting application of culture historical essentia lism and methodology to study the African Stone Age must be re-worked. Before this model can be de veloped, however, certain problems within contemporary theories of technological change must also be addressed. Impetus for this Research The impetus for this thesis is a reac tion to perceived deficiencies in the predominant application of culture histori cal methodology in current African Stone Age archaeological research and existing theories of technological change. This thesis focuses on the application of complex system s theory (CST), which is a non-reductionist theory to explain multi-scalar interaction with in historically contingent systems (Bentley and Maschner 2003) to interpret the process of technological change as it is observed in


4 the prehistoric archaeological record in the Horn of Africa. In particular, CST is used to interpret the process of technological change identified using computer assisted shape analysis (morphometrics) to identify lithic technological change through time with specific focus on the Middle and Late Stone Age transition within Ethiopia and Somalia. Complex systems theory is rooted in ch aos and non-linear theo ry and has several core principles (Bentley and Maschner 2003; see also McGlade and Van der Leeuw 1997; Van der Leeuw and Torrence 1989). Thes e principles include the reliance on an open systems model recognizing the free flow of energy into and out of the system including the constant creation of diversity, non-equilibrium, and continuous systemic fluctuations operating at different scales, mu lti-scalar interaction including agentive and extra-agentive processes and influences short and long term systemic dynamics, historical contingency, and intentional and non-intentiona l technological changes. Although genres of CST have been applie d within archaeology (cf. Bentley and Maschner 2003; McGlade and Van der Leeuw 1997; Roux 2003; Van der Leeuw and Torrence 1989), the theory itself was not developed for the social sciences. As a result I have associated new ideas with base principles of CST in order to account for the actions and ideas of people within a theory of t echnological change. These ideas focus on my assumption of an essentialist metaphysic underl ying the process of technological change as interpreted using CST. The adopti on of an essentialist position may seem contradictory to a reaction against culture hist orical research. However, the essentialist position maintained within culture history research creates strict limitations on interpreting change and vari ation because the normative characteristics underlying


5 typologies are knowable and unchanging throughout the entire spatial or temporal extent of the typological class or period under question. I assume that there are underlying essential ideas that direct our actions in creating technological products, but I reje ct the ability to know the ex act ideal characteristics of an object now or in the past. This idea does condone a Cartesian mind-body dichotomy, but only to differentiate between the ideas dire cting actions and the actions themselves. Therefore, by rejecting the abil ity to ever observe or define the essential qualities of an object I hope to measure instead the vari ation of many objects around a hypothesized ideal type thereby creating a bridge between thoughts and actions. In addition, I emphasize five primary theo retical and methodological aspects of CST within this thesis. First, emphasis is placed on multi-scalar spatial and temporal influences in order to account for the different scales of lithic technological change. Second, emphasis is also given to multi-scala r individual and group le vel influences in order to account for the creati on of lithic technological chan ge at all scales within a technological system. These first two attri butes are specifically addressed at moving beyond seamless and linear sequences of time as well as the general inability of other theoretical approaches to operate within different scales of research. Third, technological change within a lithic te chnological system is concei ved as a continuous process operating at all scales of the system. By conceptualizing tech nological change as a continuous process it resists the synchrony of the culture historical approach. Furthermore, if technological change is recognized as a multi-scalar process this equalizes the effects of both small and large scale systemic processes and introduces the capability that both endogenous and exogenous influences can affect the technological


6 system. Fourth, the visible effects of prehisto ric behaviors (i.e., mate rial culture) must be explored, but the fifth, and final, characteris tic must be the equally important reliance on the materially invisible cognitive and symbolic correlates of Stone Age lithic material culture. These final two characteristics seek to move beyond the blinding reliance on materialism inherent in prehistoric Stone Age archaeology, specifically studies of prehistoric technological change. The followi ng chapters, therefore, are an intellectual journey as I play out this th eoretical position in an attempt to measure the variation in flake morphology assumed to be directly rela ted by one or more ideas that underlay the actions of individuals in the past that created the artifacts. Problems Associated with the Study of Technological Change “Change” is defined by the Oxford English Dictionary (2004) as “to become different, undergo alteration, alter, vary.” The ontology of this definition suggests that “change” is the mechanism for movement betw een multiple states of being within the object under observation. The epistemological quality of “change,” however, is much more important to the study of archaeol ogy because it provides the contrast and comparison between the multiple states of being within the study object. The epistemological recognition of “change” compels us, as contemporarily situated researchers creating and interpreting archaeo logical data in the present (cf. Hodder 1991:30, 1999:72; Thomas 2000:4; Tilley 1991:115) to differentiate between the “current” state of the world and the interpretati on of the state of the world as it existed at some point in the past. The study of “change”, though, is much too gene ric. Therefore, this thesis focuses upon the phenomena of “change” as it is a pplied to the study of “technology”. The American Heritage Dictionary (2000) provides a robust and delimited definition of


7 “technology” stating th at technology is “the body of knowledge available to a society that is of use in fashioning implements, practici ng manual arts and skills, and extracting or collecting materials.” This definition recogni zes material products and their non-material correlates, and it does not differentiate for west ern industrialism. From this definition of “technology”, and that given above for “cha nge”, the study of “tec hnological change” can be “variations and alterations in the body of knowledge availabl e to a society that is of use in fashioning implements, practicing ma nual arts and skills, and extracting or collecting materials” (cf. American Heritage Dictionary 2000) However, this definition of technological change stil l lacks recognition of technolog ical knowledge and skills for the symbolic and visual arts such as religi ous and other graphic re presentation including body decoration. Therefore, within this thes is “technological chang e” is defined as the study of the “variations and a lterations in the body of knowle dge available to a society that is of use in fashioning implements, pr acticing manual, symbolic and visual arts and skills, and extracting or collecting materials” (cf. American Heritage Dictionary 2000). Although the study of technological change in archaeology is not new, there is a renewed interest in the mechanisms of th e process of change within technological systems (for example, Bentley and Maschne r 2003; McGlade and Van der Leeuw 1997; Roux 2003; Van der Leeuw and Torrence 1989). A technological system describes the dynamic and open interrelations of the material and behavioral components that serve to classify knowledge and techniques for the pr oduction of a specific technology. This definition of “technological systems” incor porates specific elements utilized by French cultural technologists (i.e., tec hnological systems as a netw ork of chanes opratoires practiced by a group) (Lemonnier 1992; Roux 2003) as well as other researchers who


8 employ a complex systems approach (i.e., open systems model and non-equilibrium) (Bentley and Maschner 2003; Van de r Leeuw and Torrence 1989). The principle problem with current, and pa st, interests in technological systems is an over-reliance on using discrete events to ex plain changes over time in a technological system without adequately describing the ac tual processes involve d (cf. Van der Leeuw 1989:3). According to McGlad e (1999:141), the lack of stu dy into the processes of technological change is because time was c onsidered to be self-evident. The linear theories of the twentieth century objectified tim e, quantified it, and structured it in a way to provide seamless historical narrativ es (Bailey 1983; Binford 1981; Fabian 1983; McGlade 1999; Raper 2001). When an attemp t was made to provide an explanation for technological change, the answer comm only was a rather mundane and cursory description of the (mostly e xogenous) factors involved in technological change, and not the actual process, with the inevitable outcome being something “better” than before. For example, a common contemporary explanation of technological change is that it is the product of fitness enhancing or goal-oriented so lutions either at the society or individual levels (see Fitzhugh 2001; Kim 2001; Mokyr 200 0; Pfaffenberger 1992). But, these ideas do little to explain the actual processes involved in the change itself and merely describe the ability to increase ener gy efficiency or productivity. Another noteworthy issue w ithin technology studies is the scalar difference between the myriad ideas of technological change. Evolutionary and evolutionary ecological approaches take a broad view of technology with specific focus upon the origins of a technology rather than look at technology system s as a whole over time (see; Fitzhugh 2001; Mokyr 2000; O’Brian and Lymen 2000:ch.2; Ziman 2000a).


9 Functionalist approaches al so view technology in a holistic fashion, but focus upon technological change more as a disequilibrium effect to particular environmental influences or stress rather than a continuous process of change through time (see Binford 1965; Kuhn and Stiner 1998:152; Pfaffe nberger 1992:429-492, 508; Watson et al. 1971:74). As a result of scalar differen ces within these ap proaches, prior and contemporary research into technological sy stems is nearly incompatible. And yet without each scale of research together, th e study of technology a nd technological change would seem wholly incomplete. A final issue within contemporary studies of technological change is the lack of discussion concerning unintentiona l technological changes. Th e archaeological literature that specifically addresses ideas of tec hnological change consistently lacks any substantial discussion of the effects or pro cess of unintentional tec hnological changes in lieu of a reliance on intentionality to create technological change (for example see Lemonnier 1992, 1993; McGlade and Van de r Leeuw 1997; Roux 2003; Schiffer 2001; Schiffer and Skibo 1987; Torrence and van der Leeuw 1989; Pfaffenberger 1992; Ziman 2000b). A research agenda that combines intentional and uninten tional technological changes, however, would provide a more robus t conception of the nature of technological change. Such an agenda would incorporat e the punctuated even ts of directed and purposeful action (i.e., intentiona l change) in association with the constant acquisition of unanticipated techniques, tool characteristics, tool types, or technological knowledge in general acquired during the directed and purpos eful events of inte ntional technological change (i.e., unintentional change).


10 Resolving the Problems of Prehistoric T echnological Change in the Archaeology of the Horn of Africa A solution to these deficiencies in culture history and theories of technological change would be a theory of technological change that accounts for dynamic processes rather than just origins or di screet periods and also integrat es the small-scale events of individuals (cf. Lemonnier 1992) with large-scale events operating over hundreds, if not thousands of years or more (Allen 19 97:40; cf. Braudel 1980:27; Roux 2003:12). Although there have been previous attempts to move between different scales of research (cf. Braudel 1980) these attempts have been more descri ptive account s of the processes involved rather than operationalized methods, thus leaving the user still unable to dynamically move between multiple scales of research. In order to overcome the limitations of contemporary, mostly culture hist orical, archaeology in the Horn of Africa and emplace it within a more diverse fram ework, a theoretical and methodological approach must be employed that has the cap abilities to: 1) mane uver between, and link logically, individual and group level dynamics, 2) maneuver between, and link logically, multi-scalar spatial and temporal extents, 3) rely on a model for continuous and dynamic changes, and 4) employ both the visible effect s of prehistoric behaviors (i.e., material culture) as well as their materially invisi ble cognitive and symbolic correlates. These ideas still do not resolv e the initial dilemma that the spatio-temporal expanse in African archaeology is too gr eat and the number of researchers too small to afford each and every archaeological site the requisite attention it deserves. Therefore, a fifth criterion must be a methodology that brings control to the archaeological record using particular data to identify the temporal extent of a site, and its part icular characteristics therein, without relying initially, or solely, on generalized trait lists. The explicit purpose


11 of this criterion should only address the expedi ent initial summation of a site, or region, as a means for more accurately proceeding with more detailed methods of research in the future and not simply as a standalone t ool for archaeological data collecting. The benefits these recommendations provi de to prehistoric African archaeology include the abilities to 1) quickly identify cu ltural and temporal characteristics of a site for excavation, heritage protection, mana gement, or salvage without relying on generalized trait lists, 2) st udy regional/large scale trends while concurrently being able to study 3) small scale trends such as indivi dual or group actions in cluding technological variations, 4) mediate between large and small scale processes within a continuous dynamic framework of change and not as a di scontinuous series of time periods, and 5) utilize material culture but also recogni ze and begin to interpret its non-material behavioral, symbolic, and cogniti ve correlates. Effectively, this method would facilitate a more holistic research agenda for the docum entation of archaeological sites to enable archaeologists to comprehend and manage be tter the prehistoric record and employ a more expansive array of theore tical ideas and particular arch aeological methodologies. And as a further endeavor, the application of CST, in conjunction with the addition of an essentialist metaphysic, may provide a more robust interpretive schema with which to infer the development of modern human beha vior in the late Pleistocene or early Holocene. This, however, is outside the re alm of discussion here and will only be touched upon briefly in th e concluding chapter. Conclusion This thesis will use CST in association w ith an essentialist metaphysic to explore the process of lithic technological change duri ng the Middle to Late Stone Age transition within Ethiopia and Somalia. Changes in lithic technology will be identified using


12 morphometrics. The implementation of this particular theory and method should provide a more robust conception of technological ch ange that is multi-scalar, continuous, and links ideas within peopleÂ’s heads with material ized actions. In turn, this will provide an avenue of research around perc eived deficiencies within cultu re history and other theories of technological change. The thesis is structured as follows: Ch apter 2 provides a succinct synopsis of complex systems, chaos, and non-linear theori es and further discu sses their application here within a theory of t echnological change. Chapter 3 in troduces computer assisted morphometry and discusses the methodology and techniques used to perform a morphometrics analysis. Chapter 4 begins with a discussion of the Gilgel Gibe morphometrics lithic analysis and draws upon the theoretical ideas discussed in chapter two. It also introduces the Gogoshiis Qabe morphometrics analysis as a comparative study with the Gilgel Gibe analysis to assess the morphometrics method. In the conclusion of this thesis I summarize my ideas and provide direction for future research.


13 CHAPTER 2 COMPLEX SYSTEMS THEORY Introduction Complex systems theory (CST) is not a pplied by mainstream archaeology. The reason for this may be due to an overall misr epresentation that the application of this theory requires substantial high level mathemati cal knowledge. In fact it is easy to reach such an idea as many discussions using a CST, and its related con cepts non-linearity and chaos, in archaeology contain page after page of mathematical jargon not decipherable to the non-mathematically specialized archaeologist. But, this is a fallacy as the concepts of CST can be usefully applied to interpret archaeological pr oblems without the need for complex mathematical modeling. Therefore, th e focus of this chapte r is to discuss the history of CST, its relationship with non-lin ear and chaos theory, and describe how the concepts of CST will be applied to understa nd technological change within this thesis. This chapter will also present the concepts of ideal types and unintentional technological change in association with the basic fr amework of a complex systems theory. Complex Systems Theory, Chaos Theory, and Non-linear Theory: Relationships and Characteristics Complex systems theory is grounded in ch aos and non-linear theo ry (Bentley and Maschner 2003; see also McGlade and Van der Leeuw 1997; Van der Leeuw and Torrence 1989). The foundation of chaos theory can be traced as far back as the late nineteenth century to James Clerk MaxwellÂ’s research on l ong-term unpredictability and sensitivity to initial conditions (Williams 1997:17). During the early twentieth century


14 researchers developed other ke y components of chaos theory including the concept of entropy. Much of the current interest in ch aos can be traced to a 1963 paper written by Edward Lorenz; however, it was the advent of affordable, high-power computers in the 1970s that brought about the wide-spread applic ation of chaos theory to the physical and social sciences (Williams 1997:18). Chaos is a theoretical idea that seeks to explain changes over time within longterm, naturally occurring systems. In particul ar, it is designed funda mentally to interpret temporal changes strictly within determ inistic, complex, nonlinear dynamic systems. Characteristics of chaos include: 1) determ inism (mathematical laws underlying systemic processes), 2) sensitivity to initial conditions (two slightly different initi al inputs can create two vastly different results), 3) emergence (ability of the system to create new, more complex levels of order over time) and self-organization (ability of the system to create order from irregularit y without external influences ), 4) dynamics (changes through time), and 5) non-linearity (definitions adapted from Williams 1997). According to Williams (1997:14), the benefits of using chaos theory include 1) the ability to identify randomness within a system and explore systemic determinism (see also Bentley and Maschner 2003:2) 2) greater accuracy for s hort-term predictions, and 3) the ability to identify time-limits for relia ble predictions. However, chaos theory does lack the ability to reveal pa rticular details of any underlyi ng physical laws in nature (Williams 1997:15). Non-linearity is a component of chaos th eory that describes systemic dynamics— movement and readjustment—through dispr oportional changes betw een variables and reactions within a complex system. A non-linea r equation does not pl ot a straight line on


15 a graph or take on the proportional, linear equation form y = mx + b where x and y are variables and m and b are coefficients. Nonlinearity is useful for describing systemic fluctuations (Allen 1989:272; Allen 1997:42; McGlade 1999; see also Torrence and Van der Leeuw 1989:8; Van der Leeuw and McGlade 1997:334) and non-equilibrium, including the historical depe ndency of these fluctuations (McGlade and Van der Leeuw 1997:2), long-term systemic unpredictability via unaccountable or changing systemic variables (McGlade 1999: 151), and the capacity of the system to change (Allen 1997:40). Complex systems theory, also known as “comp lexity theory”, is a recent term used to describe multi-scalar intra-systemic intera ctions whose future trajectory is dependent upon its history (Bentley and Maschner 2003) via the interactions of “p articular-like units or ‘agents’ ” (Williams 1997:234). In partic ular, CST focuses on the emergence of new levels of order (self-organization) within systems that exhibi t non-linear and chaotic characteristics (Williams 1997:234). Characteristics of Complex Systems Theory Williams (1997:449 emphases original) define s complexity as “a type of dynamical behavior in which many i ndependent agents continuall y interact in novel ways, spontaneously organizing and reorganizing themselves into larger and more complicated patterns over time.” Based on this definiti on, a complex system relies on five basic qualities: 1. An open systems model to describe the influx of new matter and energy into the system (i.e., births of chil dren, creation of new artifact s) (Bentley 2003:9; Bentley and Maschner 2003:2). 2. Non-linearity to describe dynamics (move ment) of the system via the continuous fluctuation of systemic energy (non-equi librium) (Bentley and Maschner 2003:2). In association with an open systems model, long-term systemic prediction is unable to be made (Allen 1989:272; Allen 1997: 42; McGlade 1999; Torrence and Van der Leeuw 1989:8). As a result, the focus cha nges to predicting the capacity to change


16 within the system through sensitivity dependence on initial conditions thereby introducing the ability for creativity, innovation, and multiple results (Allen 1997:40). 3. Determinism (governance by underlying laws). However, because of non-linear systemic dynamics and an open systems model, the history of the system, and the laws that regulate it, may be impossible to physically identify or predict (Bentley and Maschner 2003:2; Williams 1997:15) 4. Sensitivity to initial conditions (slight altera tions in initial conditions creates two or more vastly different trajectories) (W illiams 1997:466). Sensitivity to initial conditions is related to the concepts of the “critical path network” that Allen (1989:249) describes as “a scheduling technique for portr aying a group of interrelated steps which make up a whol e process” and “contingency”, which describes how similar events at different ti mes can trigger vastly different reactions within a complex system because the current setting could not sustain the appropriate chain of events (Bentley a nd Maschner 2003:2; see also McGlade and Van der Leeuw 1997:2) 5. Emergence (ability of the system to create new, more complex levels of order over time) through self-organiza tion (ability of the system to create order from irregularity without exte rnal influences) (McGlade 1999; Van der Leeuw and McGlade 1997:334; Williams 1997:234) The most important aspect of CST, howeve r, is the capability to maneuver between numerous scales of reference when the theo ry is implemented (cf. McGlade 2003:116). Multi-scalar analysis is vitally important to this approach because CST views systems as highly complex, interlinked events occurringat any possible scale (B entley and Maschner 2003; Williams 1997). And true to its non-linear and chaotic roots, CST posits that the most microscopic event can gene rate disproportionately greate r, or different, macroscopic events (Bentley and Maschner 2003:5). In particular, Bentley a nd Maschner (2004:5) note that one of the goals of CST research is to “discover how movements at a small scale translate into emergent phenomena at a larger scale or at least what emergent phenomena can be expected.” The scale of analysis within CST is variable. CST attempts to bridge both reductionist and constructi onist methodologies (Bentle y and Maschner 2003:1) by


17 observing how small scale events can create di sproportionately larger macro-scale events within complex dynamical systems (Allen 1997: 40; Bentley and Maschner 2003:5). In particular, Williams (1997:234) emphasizes a soci ally situated (micro) scale of analysis by suggesting that complexity research focuse s on the interaction of “particle-like agents” including the “hierarchical progression in the evolution of rules and structures.” In this thesis, the use of the term “ag ents” does connote individuals but refers more generally to human and non-human agents of change including people, ideas, and objects. Context within Technological Systems Complex systems theorists describe c ontext through “sensiti vity to initial conditions” and “contingency”. “Sensitivity to initial conditions” specifically refers to the reliance of the syst em upon historical events such that a miniscule action in the past can translate into disproportionately larger events in the future actions of the system (Williams 1997). “Sensitivity to initial conditions ” also refers to the particular trajectory of a complex system due to its unique su ite of initial conditions (ibid). “Contingency” on the other hand, describes how similar events at different times can trigger vastly different reactions with in a complex system (Bentley and Maschner 2003:2; McGlade and Van der Leeuw 1997:2). “C ontingency” also refers to the concept of the “critical path network” that Allen (1989:249) describes as “a scheduling technique for portraying a group of in terrelated steps which make up a whole process.” Relying strictly on the complex system s theory terminology described above, “context” refers specifically to 1) historical actions, 2) contemporary setting, and 3) sequences of events, both historical and curr ent, within a complex system. However, these concepts only describe the mechanical setting and processes of the system and do


18 little to discern the socially situated nature, or m eaning, of a complex technological system that is being investigated here. A more meaningful approach to study pa st context within complex technological systems can be found within the various themes of interpretive archaeo logy. The specific epistemology of interpretive archaeol ogy can provide explicit boundaries on what meaning can or cannot be discerned from the archaeological record, especially considering the meaning drawn from lithic de bitage relating to underlying ideals. Accordingly, these phenomenologically in clined ideas place a great deal of responsibility upon the contempor ary interpreter, and their actions, to understand past social processes because it is assumed that on ly through the interpreterÂ’s present and past subjective experiences are possible any iden tifications of meaning (cf. Hodder 1991:30; Hodder 1999:72; Knapp 1996:143; Shanks and Hodder 1995:5; Thomas 2000:4; Tilley 1991:115; Tilley 1993:3,7). As such, these positions are less concerned with the explanation of events than with creating an understanding of the event, as a product of contemporary subjective knowledge (cf. Whitley 1998:13). Although not explicitly discussed throughout this thesis, the recognition of an underlying essentialist metaphysic guidi ng a technological system ultimately is concerned with meaning. The assumption of ideals guiding technol ogical systems entails the assumption that these ideals connote specif ic and subjective soci al meanings to the individuals who employ them. However, it is beyond the scope of this thesis to address these issues in great detail. Suffice to say, th e conception of ideals (i.e., ideal types) here is assumed to never be directly identifiabl e or materialized by t hose in the past who employ them or those of us in the present w ho wish to study them. I assume instead an


19 underlying essentialist metaphysic guiding th e production of technological objects but deny the ability to know and utilize these ideals to classify these products. As a result, I hope to measure the variation of many t echnological objects around a hypothesized ideal type thereby creating a bridge between t houghts and actions, the te chnological society and the individual. Ultimately, we may be able to hypothesize individual and social meaning of these variations around certain id eal types. For now, however, I simply assume that the creation and maintenance of ideal types by a technological community connotes some sense of shared meaning be tween individuals about how to produce and use the technology, which in turn creates homogeneity within a technological system. Thus, the employment of interpreti ve archaeological epistemology may significantly enhance the concep t of “context” within comp lex technological systems by moving beyond a description of mechanical sett ings and processes and elaborating on the meaning of the essentialized properties of technological syst ems and technological products. An interpretive archaeological epistemology facilitates the contemporary creation, and elaboration, of a past technological system through contemporary social actions and knowledge. As such I assume th at by relying on socially situated actions, historical and contemporary sequences of technological events can be more fully described in terms of the human and non-hum an agents creating th e action, the complex interrelations between agents within the technol ogical system, and how these agents recursively create, and are influenced, by the la rger technological syst em as a whole. Technological Change within Complex Systems Within a complex systems approach, tec hnological change is described through individual and group level behavior (A llen 1997; McGlade 1999; Roux 2003; Spratt 1989; Van der Leeuw 1997:34) adapting to partic ular environmental influences resulting


20 in non-equilibrium systemic processes and s ubsequent re-organiza tion of the system (Allen 1989:273; McGlade 1999:150; Roux 2003: 6; Torrence and Van der Leeuw 1989:7-8; Van der Leeuw and McGlade 1997:339). For example: It is the existence of processes such as reproduction, cooperation and competition at the interface of individua l and community levels which can, under specific conditions of amplification, generate unstable and potential ly transformative behaviour. (McGlade 1999:150) Adaptability and change come from the in terplay of internal variability, system structure, and environmenta l conditions. (Allen 1989:273) The non-linear coupling of a relatively stru ctured, slow environmental dynamic and a more rapid and stochastic human one ge nerates bifurcation behaviour. (Van der Leeuw and McGlade 1997:339) This thesis, however, app lies an essentialist position advocating ideal types to describe technological change. This position su ggests that ideas direct the actions of active agents who manipulate their availabl e resources and knowledge within a preconceived (but not necessarily achieved) prediction of his/he r consequences in order to gauge the value in retaining or discarding technological elemen ts for future use. As a result changes in the ideal type affect ch anges in the technological system, but these changes are instituted only through the di scursive and contextualized choices of technological agents. Ideal Types A “type” is nothing more than a “general character or structure held in common by a number of people or things considered as a group or class” (American Heritage Dictionary 2000). In particul ar, ideal types are mental repr esentations an individual or society has as to the appropriate form a t echnological product shoul d have for a specific function. This section discusses the use of ideal type within th is thesis and other research


21 and suggests two reasons why the use of ideal types within technology studies is important. The use of types is essen tialist and normative. However, the primary difference between the uses of ideal types in this thes is compared to prior research is assumed inability of past and current individuals to ever identify the ideal type or produce it perfectly in material form. This is differe nt from culture historical research where normative values are used by a researcher to classify and compare one culture against another through the use of trait lists (Trigger 1989). The use of types in culture historical research suggests that any person is able to know, and materially express, their shared normative values. In contrast, the employment of ideal types here is employed strictly as a heuristic to describe general similarities in thinking among memb ers of a technological system, and it is the variation around ideal types with which we can infer change. Many other theories allow for mental or cognitive contributions to technology, but references to ideal types within technol ogy studies seem to be highly variable. Lemonnier (1993:3-4) suggests th at there are underlying ment al processes directing our actions. Roux (2003), following Lemonnier and others, suggests that the compromise between formal and ideal properties of artif acts advocated by the behaviorist approach (cf. Schiffer and Skibo 1997; Skibo and Schi ffer 2001) creates an over-simplified linear sequence of change and unrealistic duality be tween technology and society. In addition, through their recognit ion that choices can be used to improve an artifactÂ’s performance characteristics and that those creating a technological product ha ve a conception of optimum levels of performance, Schiffer and Skibo (1987), imply the existence of ideas that direct actions, but not an explicit c onception of ideal, shared form types:


22 ideally then, the tinkering artisan tries out different technical choices, attempting to optimize an artifact’s activity relevant pe rformance characteristics. In practice, however, many performance characteristics fa ll short of optimal levels because of their complex causal relationships with t echnical choices and formal properties. (Schiffer and Skibo 1987:599) Others recognize ideal types but misrepresent its use. For example, Rolland and Dibble (1990:483) note that any of the retouched tools found in Middle Paleolithic assemblages represent wornout, discarded objects rather th an intentional end products . in this case it would not be true that the lithic types repres ent deliberately shaped objects reflecting normative values Rolland and Dibble are absolutely correct to state that the original “normative value” may not be preserved in the final morphology of the tool. However, contra Rolland and Dibble (1990:483), from a heuristi c point of view there are normative ideal types in the final morphology of a lithic too l; they are just different from the original “normative value” of the tool. An idealist basis for technology requires that the ideas must always precede actions. Even later m odification events conform to some “ideal” characterization. This section has, up until this point, concen trated on a discussion of the infrequent and implicit use of ideal types w ithin other research. In spit e of the lack of use within contemporary technology studies, and a co mmon opinion throughout the discipline of archaeology that essentialist ideas are outdated and largely rejected, I still intend to argue for two benefits that an essentialist con ception of ideal types can offer a theory of technological change. First, the ideal t ypes privileges the mental conceptions of a technology held within the minds of the technological agents Second, the ideal type is the foci of the technological system creating similarity within its materialized effects (artifacts).


23 First, my use of ideal types only pr ivileges the mental conceptions of a technological form held within the minds of the technological agents. The materialized effects of a technology (i.e., artifacts) are simply the by-products of the process that attempts to recreate—but never attain—the “ideal tool” materially based on the mental conceptions of the “ideal type.” Technol ogical change, therefore, is found within variations of materialized tool form around an ideal type as well as a lterations of the ideal type itself held within the minds of the tec hnological agents and s econdarily manifest in the materialized by-products of that technol ogical system. Precisely for this reason I hope to move beyond prior conceptions, and use, of essentialist id eas and ally it more closely with a materialist metaphysic that a llows for measurable variation of change through time. Second, the ideal type becomes the foci of a technological system allowing for similarity within its materialized effects (artifacts). Using CST terminology, the ideal type can be likened to the convergence of a system towards an “attr actor” (cf. Williams 1997:447). However, the state of systemic equi librium also associated with the concept of an attractor can never be achieved using my concept of ideal type s because the type is assumed to never be achievable. This presumption creates a constant and dynamic locomotion of change within the technologi cal system. The drive of innumerable technological agents to invent the “ideal to ol” creates a technological system that is constantly in the processes of non-li near and seemingly chaotic action. This action, what I see as a gravitati onal movement around the ideal type, is perturbation. Perturbation is a “displacement in a trajectory or any difference between two neighboring trajectories or observations at any given time” (Williams 1997:169). It


24 is a means for describing the movements of a technological system over time though selforganizing fluctuations (c f. McGlade 1999; Torrence a nd Van der Leeuw 1989:8; Van der Leeuw and McGlade 1997:334) that are esse ntial to the survival of complex and nonlinear systems (Van der Leeuw and McGlade 1997:338). Perturbation is directly associated with the sensitivity to initial conditions (cf. Williams 1997:466), contingency (cf. Bentle y and Maschner 2003:2; see also, McGlade and Van der Leeuw 1997:2) and critical path (cf. Allen 1989:249) of a technological system. Allen (1989:269) notes “because of fl uctuations the real system is always, in fact, probing the stabil ity of the particular situation and, dependi ng on which fluctuation occurs at a critical moment, the system will move to one or another of the stable behaviors which are possible.” However, perturbation by itself is an incomplete concept to describe how technological changes occur because pert urbation only describes movement in a trajectory. The origin of tec hnological change is always with the technological producer. In particular, within my model technologi cal changes result from the recognition and decision to implement intentional and uni ntentional changes th rough a technological producer’s available knowledge base and concep tion of the ideal type for that particular technological system. The next section in troduces the concepts of intentional and unintentional changes. Partic ular attention is focused on unintentional changes as the primary catalyst for technological changes. Ideal types are a critical co mponent of this thesis’ inte rpretation and application of CST as it pertains to technological systems. Th e ideal type is the fo ci of a technological system providing continuity in form and func tion of what is produced and it is the locus


25 of technological change because as the idea l type changes so too does the technological system as a whole. But, the ideal type is just the focus of technological change. The actual means by which technological systems ch ange is more precisely described through the processes of intentional and unintentiona l changes incurred th rough the process of technological production and the infl uence of technological agents. Intentional and Unintentional Changes The predominance of intentional technol ogical change within archaeologicallybased research restricts the ab ility to conceive of technolog ical changes in any other way (however, see McGlade 1999:152; Schiffer and Skibo 1987:597; Torrence and Van der Leeuw 1989:10). Equally important, however, is the idea of uninten tional technological change. This section describes the differe nces between intenti onal and unintentional technological changes and advocat es a research agenda that utilizes both concepts of technological change. In a ddition, Appendix A presents the results of an experimental archaeological project designed to investigate the process of intentional and unintentional technological changes. A research agenda that combines intenti onal and unintentional changes is useful for two reasons. First, a combined research ag enda incorporates the punctuated events of directed and purposeful action (intentional change) in association with the acquisition of unanticipated techniques, tool characteristics, tool types, or technological knowledge in general acquired during the directed and pur poseful events of technological change (unintentional change). Second, intentiona l technological change facilitates an open systems model of complex systems because unintentional changes are hypothesized to occur constantly as a source of diversity but are only r ecognized infrequently. In addition, the concepts of sensitivity to init ial conditions, critical path, and contingency


26 describe the incorporation of unintentional changes into a technological system. The adoption of unintentional change s requires first the recogniti on of the variation based on the contingency of the situat ion and critical path of th e person’s knowledge base, and second the emergence of a qualitatively m odified technological system through selforganization in order to incorporate the new t echnique, tool characteristic, tool type, or technological knowledge into the pre-exis ting conception of the ideal type. The concept of “intentional change” desc ribes the active and directed process of technological creation and i nvention. This position, wh ich I call “intentionalist,” maintains that technological ch anges occur as the result of the purposeful and conscious influence of directed individuals (for example see Fairtlough 2000; Lemonnier 1993; Martin 2000:99; Schiffer 2001b; Schiffer and Skibo 1987:599), larger technological systems (for example see Pfaffenberger 1992; Roux 2003), or even creativity (Boden 1998; Hodder 1998; Kuhn and Stiner 1998). In particular, the inte ntionalist position relies on the direct reproduction of the forms and techniques that structure the creation of technological products (cf. Roux 2003:5; Schiffer and Skibo 1987:597; Ziman 2000:5a). Examples of the intentiona list position are found in H odder (1998:62 emphasis added) who notes that creativity is “a ssociated with the more active process of problem solving, imagination, and invention” and Schi ffer (2001b:217) who succinctly sums up the intentionalist position by suggesting simply “in the invent ion process, people create…” On the other hand, unintentional technologica l change refers specifically to the decision to implement unanticipated techniqu es, tool characteris tics, or tool types achieved during the intentional production of technological pr oducts that alter the ideal type through its design strategy and use. Th e basis of unintentional technological change


27 is associated directly with the inability to create a perfect one-to-one relationship between the mental conception and physical producti on of a technological product (cf. Ziman 2000:7a). The imperfect reproduction of a si milar material product, or production of a new idea, is a result of the inherent qualities in raw material and the influences of various social, ideological, and physical contexts of both the technological product and producer. As a result, the process of technological produ ction can be more accurately described as “production-in-kind” whereby the technol ogical agent assesse s any identifiable unintended variations based on the current ideal type and reorganizes his or her knowledge to accept or reject these variations. Underlying the concept of unintentional te chnological change is the idea that unintentional changes occur constantly duri ng every act of tec hnological production or modification. However, a technological agen t may not identify any constantly occurring unintentional changes until some later date. The ability to identify, and assess the potential of, unintentional cha nges is directly linked to th e socially situated current context of the technological agents including the critical path (cf. Allen 1989:249) and contingency (Bentley and Maschner 2003:2) of their technological knowledge. The critical path of the knowledge available to a technologi cal agent is essential for identifying potential in the unintended varia tion. A technological agent will discard an unintended variation if they have no extant knowledge to identify any usefulness of a technological variation into an existing tec hnology or aid in the development of a new technology. A preliminary experimental archaeology project studying the ability of nine volunteers with no knowledge of stone tool production or use to make and use stone tools


28 underlines the duality of inte ntional and unintentional techno logical change and the role knowledge and context serves to identify and implement these changes (Appendix A). The results of this experiment provide verbal and visual evidence that each group did rely on pre-conceived ideas to direct their actio ns. Furthermore, during the course of intentional, directed actions to make a specific stone tool, unintentional variations were observed to occur frequently and when identi fied and adopted had th e potential to change the ideal conception of a certain tool for the group. Finally, direct observation of each of nine participants as they a dopted unintentional technological changes suggests that this process is controlled by the individualÂ’s knowle dge critical path to see potential within the variation as well as how the variation can be applied within the contingency of current actions. The contingency of a technological system concerns the social and historical situation and knowledge base of a technol ogical system and it is crucial to both intentional and uninten tional technological changes (Ben tley and Maschner 2003:2; see also McGlade and Van der Leeuw 1997:2). In particular, continge ncy is a tool to describe how the context of a technological system must be capable to sustain sufficiently an unintentional technological change because the incorporation of unintended variations can initiate rapid and punctuated cha nges within a technological system. A prime example of contingency within a technological system is the Fairbanks Morse (FM) H-20-44 Trainmaster diesel locomo tive. According to the foreman of the machine shop at FM during the time the Trainmaster was unveiled (Kenneth Bunnell, 2004) and Ingles (1996), the Trainmaster wa s introduced on the market in 1953 and was considered to be decades ahead of its time in terms of technological efficiency. The


29 Trainmaster was hailed as the single greatest improvement in diesel engines since World War II. The 2400 HP Trainmaster was the first diesel locomotive to have six individual traction motors operating each of its six axles, and it was also the first locomotive to have an engine with aluminum bushings. But, the feature that has had th e greatest impact on the railroading industry was the automation built into the control of the locomotive. This automation meant that one person could operate the engine instead of the standard three persons on other types of locomotives. From the TrainmasterÂ’s technological s uperiority versus co ntemporary diesel locomotives, it seems logical that the Trainm aster would have been readily accepted by the railroading industry. Certainly, there woul d be more than just one of these engines remaining today. However, efficiency a nd superior technological qualities are meaningless unless the social context is receptive to the technology. Soon after introducing the Trainmaster, FM entered into negotiations with the New York Central (NYC) railroad for the purchase of 100 Tr ainmaster units. However, the NYC was concurrently also in contract negotiations with the railroad ing unions. The unions, fearful of the three-quarters job loss the Trainmaste r would bring with it, offered to accept the NYC contract proposals so long as the NYC di d not accept the FM Trainmaster contract. As an unintentional consequence of the effici ency of the Trainmaster, the NYC did not buy any new Trainmasters. Three years later, the NYC filed for bankruptcy from having to pay high personnel salaries and the chairman of the NYC railroad committed suicide. In addition to this debacle of contingenc y, Fairbanks-Morse, a once grandiose company among the railroading industry, was effectiv ely cut out because they had hoped the Trainmaster would save them from impending financial ruin. The railroading industry


30 eventually implemented many of the technolog ical changes forecasted in the Trainmaster once the context was receptive to these ch anges thereby transforming the technology of railroading in the process. Furthermore, the intentional developm ent of the Trainmaster had specific unintentional consequences as a result of the contingency of the then current social context. It was not how the Trainmaster cha nged FM or the railroading industry directly, but how the Trainmaster was unable to be ma intained that was contingently important; the Trainmaster was just too far ahead of its time for its social and technological context and this caused radical unin tentional change throughout the railroading community. A technological system must qualitativel y reorganize itself to accommodate the changes incurred following the identificati on and adoption of both intentional and unintentional technological change s. In particular, the ideal type of a technology must be reconsidered and modified in order to institute the tec hnological change. Often, the qualitative shift in the ideal type is no more than a reconsideration of the creation and use of a particular technological product since ideal types are no mo re than a concept held in the minds of technological agents. As fo r the Trainmaster example above, there were several scalar qualitative shifts in the enorm ously large railroading technological system. At a proximate scale, Fairbanks-Morse stopped producing railroad locomotives and focused on engines for marine or other appl ications. On a much larger scale, the American railroading system saw the potenti al, and problems, of the Trainmaster and eventually developed engines with greater personnel and mechanical efficiency. The Technological Life Cycle No discussion of technological change is complete without due consideration for the macroscopic, long term changes affecting technological systems. In particular, when


31 discussing technological changes occurring ove r a long period of time there must be some discussion as to how one particular tec hnology ends and another begins in its stead. Up until now, this discussion has focused on the explanation of proximate technological events and processes. This section has one primary objective. By focusing on the long term development and eventual end of tec hnological systems as a whole entity, this section presents a model to discuss the or igin, development, peak, and decline of a complex technological system over the long term. The Trajectory of Long Term Technological Systems Due to the sensitivity to initial conditions of any technological system (cf. Williams 1997:466), no two technological systems will follow the same trajectory. Though some technological systems may appear more an alogous than others, the underlying chaotic principles of any complex technological system dictates that the s lightest alterations in initial conditions of any complex system create vastly different systemic trajectories. Add to this that the critical path of knowledge (cf. Alle n 1989:249) and contingency (cf. Bentley and Maschner 2003:2) of any technol ogical system creates a unique set of influences and each technological system then, by definition, is exclusive. All technological systems, however, do follow a very similar, though much generalized, pathway from initial conception, growth, peak, decline, and abandonment. This model describes the crucial relati onship between two sequential technological systems, one system succeeding the other. Furthermore, the application of complex system theory to this model provides ample ability to maneuver between the large scale, long-term processes of a technological system and the small scale, short-term dynamics between a technological agent a nd a technological system.


32 Ideally, the generalized, long-term trajecto ry of a technological system is best represented by a logistic curve. Logistic curves are capable of describing the origin and development of a complex technological sy stem over time, including the limitation of systemic growth and eventual systemic declin e. The natural growth pattern for a logistic system is through the S-shaped logistic curve. According to Modis (2003), the beginning of any log curve is exponential. This s uggests that two users of a technology would pass on the idea to two other users, and they would pass on the technology to two other users each, and so on and so forth. Exponential systems, however, require an equally exponential amount of resources to sustain the rapid growth of the system. Therefore, exponentia l systems cannot last indefinitely because of the limitations imposed by the availability of resources. Once a technological system reaches its maximum cap acity of resources, the growth rates begin to slow down and eventually stabilize. McGlade and McGlade ( 1989:283) show that the adoption of a technology in a society follows a logistic curve in response to the initial adopters of a system, those that adopt after some delay, and the “laggards” adopting the technology after much delay. The underlyi ng mechanism to their model is diffusion (ibid). Modis (2003) also di scusses the potential of logi stic curves in technological systems although he relies upon competition as bei ng the underlying factor in the model. Growth and Complexity of a Technological System In figure 2-1, ‘A’ is the genesis of t echnology ‘1’. The rise in the curve (B) represents the exponentially based rate of adoption for a technology by technological users. In the past, technol ogical adoption has been percei ved to be a product of the effectiveness of a technol ogy (Torrence and Van der Leeu w 1989:10), the representation between technology and society (Pfaffenbe rger 1992; Roux 2003 following Lemonnier


33 1989, 1993), extra-technological factors (Sch iffer and Skibo 1987), and the perceived future benefits, costs, and risks of ad opting the technology (Kim 2001). Technological adoption is seen here as a product of each of these fact ors because these variables establish the natural and soci al limitations with which to gauge any new technology. Following adoption of a technology, the increas e in technological users expands the capability for variation within the technological system. The expansion of the possible variation that can occur during production likewise in creases the possibility of introducing greater complexity in to the technological system. Complexity is a contested and speculative subject. Modi s (2003:29) notes that comple xity cannot be quantified. Instead he argues that the relative amount of complexity can be calculated by determining the importance of an event as directly proportion al to the complexity it introduces into the system, and also by the length of the stasis preceding the next event. According to Modis’ (2003) formula, a more important event would then introduce more complexity into the system, and also have a longer peri od of stasis following th e event, than a less important event. However, Modis’ definition of technological complexity would be very difficult to apply to the African Stone Age because neit her the important even t nor the resulting stasis may be archaeologically visible. Therefore, instead of trying to quantify “complexity” in this thesis ( sensu Modis 2003), I suggest that it is more appropriate to discuss the “approximation of the degree of co mplexity” as proportional to the frequency of technological users sharing a similar idea l type for employing, creating, and modifying a technology. As the number of users within a technologi cal system increases, the possibility for more variation within the sy stem is likewise increas ed, thereby introducing


34 the likelihood for greater systemic and t echnological complexity. This idea of complexity does assume a direct correlati on between quantity of individuals within a technological system and the quantity of vari ation within that system, but with future research this premise could be tested archaeologically. Peak The peak in the technological life cycle (Figure 2-1 label C) is a direct result of the introduction of a competing technology, and like the approximation of the degree of complexity, the peak of the technological life cy cle also refers to the frequency of use of that technology. The assumption of this m odel is that this peak is brought on by a competing technology that draws away users of the current technological system holding other variables such as populati on or material resour ce availability consta nt. This limits the amount of resources and energy available to the current technological system because these resources are being redirected towards the competing system. A viable competing technology must have significant qualities that make it more appealing to the technological users, and also ha ve the ability to operate within the users’ social contexts. The new technological system must also possess enough similarity to the existing technological system to be positively recogni zed and received by a society. The example given earlier in this discussion of the Fairba nks-Morse corporations showed exactly this concept; the Trainmaster was too far ahead of its time, therefore, it was ill-received and greatly affected the railroadi ng industry at that time. Decline If two technological systems are in dire ct competition then the adoption of a new technological system causes the decline of th e existing technological system (Figure 2-1 label D). The use of the term “decline” here does not suggest the immediate eradication


35 of the former technological system. The adoption and decline between two competing technological systems is simply a shift in the predominant applic ation of resources, knowledge, and use from one system to another. As this shift occurs, the former technological system becomes less and less ut ilized. The loss of users limits the amount of variation (i.e., the approximation of the degr ee of complexity) within the technological system thereby restricting the frequency of future intenti onal and unintentional systemic changes. Ideal types are assumed here to structure the process of d ecline. While in decline, the innumerable variations and knowledge attr ibuted into the technological system will be lost (i.e., unable to be maintained) until all that remains is only enough knowledge to accomplish the task set forth by the ideal type. The removal of technological attributes during the decline of the technolog ical life cycle is simply ju st another qualitative change to the ideal type and nothing more. In esse nce, the ideal type has changed yet again while the system is in decline, and it is in no way comparable in terms of quality, quantity, or character to prior and future states of the ideal type because of the contingency and critical path of the syst emÂ’s history, and ever-changing contemporary influences on the system which may seek to propel the system in new and unpredictable directions. Conclusion This chapter has introduced several new con cepts into CST and its application to a theory of technological change. First, I have introduced an underlying essentialist metaphysic of ideal types controlling the pr ocess of technological change. Second, I redefined context using interpretive arch aeological hermeneutic and phenomenological theories in order to allow for the experiential and relativist recogniti on of meaning in past


36 technological systems. I assume that ideal types imply a shared sense of meaning about how to produce and use a technol ogical product and that this definition of context brings together thoughts and actions, the technological society and th e individual. Third, I have introduced the conception of unintentional technological change drawing from the assumed inability of a technological producer to ever identify and create a perfect copy of the preconceived mental conception (i.e., ideal type) and physical production of a technological product. Fourth, I have introdu ced a model of long-term growth, peak, and decline of technological sy stems including the creation of complexity within a technological system. Therefore, using CST, as I have modified it here, I believe that human actions and internal and external influences can be appropriately modeled through the device of a system. In particular, this modified Co mplex Systems Approach is aptly suited for modeling technological systems because it faci litates dynamic events that can abruptly alter the trajectory of a tec hnological system while also accounting for the contemporary and historical context of the technological system. Furthermore, the multi-scalar capability of CST easily allows for descriptions of technological influences and change on both micro and macroscopic levels. By far the most advantageous quality of CST, as it is applied within this thes is, is its ability to incorpor ate both material culture and nonmaterial cognitive ideas and behaviors of i ndividuals to explain the changes within a technological system. Thus, I believe that the application of CS T within studies of technological change has the potential to provide significant improvements in the way technological change is identified and described. The following chapters move beyond a purely theoretical


37 discussion of CST within technological change and provide a methodology for operationalizing this theoretical approach within archaeological and ethnoarchaeological research. To do this, chapter three introduces the application of com puter assisted shape analysis (morphometrics) to analyze mor phological changes in archaeological flake debitage. Finally, chapter four presents the results of a morphometrics analysis of two Stone Age sites from the Horn of Africa and app lies CST, as described in this chapter, to interpret the results.


38 Figure 2-1 The technological life cycle of three different technological systems. Technology 1 is a complete life cycle representing the genesis, adoption, stasis, and decline of any technologi cal system. Technology 2 represents a technological system in decline and T echnology 3 represents the genesis of a technological system competing with Technology 1 and 2 for technological users and other resources.


39 CHAPTER 3 PUTTING A SHAPE ON TECHNOLOGY Until now, this thesis has focused primarily on the theoretical background for a complex systems approach to technological change. The task is now to show how a complex technological system, relying on the assumed cognitive ideal types posited in the theory, can be identified in the real world. This chap ter seeks to find a method to answer the following questions: 1) how can we observe technological change in the archaeological record using computer-assisted shape analysis and interpret these changes using a complex systems approach, 2) how can we identify the variations around cognitively-based ideal types through an anal ysis of form in the archaeological record, and 3) how can we observe the change s of ideal types through time? Morphometrics In my model I propose that the process of technological change is measured in terms of the attempts to replicate the ideal conception of a particular tool based on a specific function and context. However, because of deterring factors found in the somatic actions of the technological agents properties of the raw materials, and extrasomatic influences such as the spatio-temporal environment, the ideal type can actually never be materialized. Thus, accordi ng to this model technological change is found in the shifts of the id eal type over time as evid enced by the always imperfect materialization of the ideal type. Therefore, if the ideal type consists of a configuration of specific fundamental formal properties of a t echnological product, as I have suggested, then it is plausible that the morphometry of the materi alized technolog ical products


40 should provide an avenue to study the vari ations around an ideal type through time and the change from one ideal type to another. The pursuit of this theoretical model further requires methods that can identify the relativ e shapes of the ideal types and map their changes over time. Morphometrics is the study of the shape, or more general morphology, of an object. The analysis of shape has been a research topi c in anthropology for a great deal of time. In particular, the use of shape to classify stone tools has existed since the early 20th century. Black and Weer (1936) presented a lith ic typology that iden tifies the basic form of the object by classifying it either as “rec tanguloid”, “trianguloid”, or “circuloid.” Currently, many lithic analyses commonly categorize artifacts based on their crosssection, profile, and plan-view. However, the problem with such terminology, and application methodology, is the inherent su bjectivity and ambiguity found in the distinctions between one shap e type and another (cf. Gero and Mazullo 1984:317; Rovner 1995). What is “sub-rectangular,” and how is it different from “rectangular” or “square,” for example? The problem lies in the ordinal nature of this system. Shape is not based on an objective and pre-defined reference poin t. Rather shape is defined based on the degree of difference from another simila r shape (Gero and Mazullo 1984:317). For example, rectangular is more elongated than “sub-rectangular,” wh ich is more elongated than “square,” “sub-circular,” or “circular1.” Although the classes of shapes such as “square,” “rectangular,” or “circular” may be used with explicit definition, the gradations between these classes are ambiguous. Furthermor e, shapes of actual artifacts rarely, if 1 Here, square refers to an object with four equidistant planes connected through rights angles. ‘Rectangular’ refers to an object with two pairs of equidistant planes connected through right triangles whereby one pair of planes is longer than the other. ‘Circle’ refers to a plane cu rve equidistant from a fixed point.


41 ever, conform to easily classifiable shapes such as “square” or “circle.” As a result this system of morphometry does not provide an adequate method to differentiate objectively between categories of shapes either qualitatively or quantitatively. There is also currently no well-defined standard of shape or measurement in the analysis of stone tools; ten di fferent projects can have ten different systems to describe shape and record measurements. Quantitative systems of measurement, which might be perceived as objective and stra ight-forward, actually contai n a great deal of ambiguity and subjectivity. Take the measurement of flake “length,” for example. Andrefsky (1998: ch. 5, 85-109) devotes an entire ch apter on measuring stone tools, including several pages describing solely how to measur e flakes. According to him (ibid: 97), one method to measure a flake is by orientation. In this method th e recorded length of a flake can actually be shorter than the “actual” length of the flake if that flake is oriented so that the line perpendicular to the striking platform intersects a lateral margin before reaching the distal end (Figure 3-1a). Andrefsky ( 1998:98) also mentions two other methods to measure flake length including length perpendi cular to striking platform width (Figure 31b) and length as the maximum distance from the proximal to dorsal ends (Figure 3-1c). Rovner (1995) also argues that mathematica lly the greatest length of a flake is not parallel to the axis of the fl ake but is, in fact, diagonal to the proximal and distal edges (Figure 3-1d). Standardization There has been scant research designed to transcend the lack of standardization and subjectivity found in lithic shape analysis and dimensions measurement. Gero and Mazullo (1984) discuss the use of a Fourier test to objectify the shape of a flake, while Dibble and Chase (1981) employ a parallel scal e to measure ten points on an artifact.


42 Yet, in spite of solutions such as these, c ontemporary lithic analysis still has not adopted a less-subjective and standard ized methodology to record the shape and dimensions of stone artifacts. Therefore, to move beyond this perceived deficiency, I have employed computer-assisted imagery analysis as a met hod to provide an object ive measure for lithic archaeological research. Computer-assisted imagery analysis uses computers to measure the shapes of objects using digitized raster images taken of the objects. Any object can be digitized, and simple digitization can be achieved us ing a digital camera, scanner, or video recorder. Digital images are recorded in a pi xel-based Raster format. In this format each pixel in the image has a particular value corresponding to its color displayed on the screen. The Raster format is significant because it introduces the possibility of mathematically analyzing and manipulating the pixels in an image. There is a primary drawback to this method, however. Raster images are dependent on the pixel resolution of the image. Therefore diagonal lines have to proceed over and down around the borders of specific pixels. If the resolution of an image is set too low, then the pixelation of the object could potentia lly alter an analysis. The computer assisted imagery package used in this research was the Image Processing Toolkit version 5 (IPTK) devel oped by Reindeer Graphics, Inc. As shown in figure 3-2, after digitizing an image, the ne cessary next step transforms this image into binary raster format. Color images can contain millions of colors and thus an equally great amount of values for its pixels. Progr ams such as the IPTK have the ability to isolate each pixel value, but th is does not identify the actual object in the image. In essence, the program cannot sense the forest from the trees. Th erefore, the user must first


43 select each object intended for analysis with in an image (Figure 3-2a). Within Adobe Photoshop, objects within an image are select ed using the Magic Wand tool. This tool selects a range of pixels with similar digital number values ba sed on the input pixel value. The conversion to binary format isolates ea ch selected feature within an image and reclassifies the value of pixels contained within its area as '1 '. All other pi xel values are reclassified as '0'. The end result of a bina ry transformation is a white image with a series of black silhouettes of the objec ts that were selected (Figur e 3-2b). Although this process eliminates all other color information embedde d within the image, by reclassifying pixels as either '1' or '0' (i.e., “on” or “off”), it also provides the ability to accurately employ mathematical calculations of shape and dime nsion on the silhouetted objects within the image. Morphometrics analysis has several primar y benefits. First, hundreds of artifacts can be digitized in an expedient manner pr oviding more time for an analysis or the collection of more data in the same amount of time. Second, morphometrics increases the efficiency of data collection because more than one object can be represented within a single image (for example see Figure 3-2a). Personal experience has shown that up to sixty flakes can be included within one digi tal image without varia tions in morphometric statistical results within four significant digits. Howeve r, as noted above the pixel resolution of the image is a significant constr aint in the ability to record many objects in one image. Although pixel resolution can be in creased, this also increases the size of the computer file. Therefore, the researcher mu st first decide on the relation between pixel resolution and image size versus recorded ar ea and number of objects. Third, the entire analysis process from image procurement, binary conversion, and tabular data output


44 rarely takes longer than a few mi nutes. For example, as will be discusses in chapter four, in preparation for this thesis research, I was able to analyze morphometrically over 1800 flakes in two weeks’ time. By far the most advantageous aspe ct in employing computer assisted morphometrics is the ability to ensure a meas ure of objectivity and standardization within lithic morphometry. Computer-assisted image analysis programs such as the IPTK come with a pre-set suite of shape and dime nsion functions based on contemporary morphometrics techniques. The only subjec tivity within computer assisted shape analysis is found in the relationship between the real world shape of the object and its pixelized raster image. The techniques us ed to measure the shape and dimension of objects quantitatively provided in morphom etrics programs are not subjective. Computer-assisted shape analysis is object ive because the program does not analyze shapes as “shapes” per se rather it simply qua ntifies the patterns of pi xels with a value of ‘1’. Measures of Shape and Dimension The standardization and limited subjectivity of computer assisted shape analysis is linked to the methods the program employs to measure the selected objects. Among the many measures of shape and dimension that the IPTK records are area, perimeter, “xferet,” “y-feret,” length, and width. Unlike the ambiguity di scussed earlier for measuring the length and breadth of an object, compute r-assisted analysis work s around this issue by relying on only two distinct and predefined measures of length and breadth. The first measure of length and breadth, for example, is called “X-Feret” and “Y-Feret.” These are measures of the widest points of an obj ect on the x and y axes respectively. These measures are dependent upon the orientation of the object relative to the x and y axes


45 (Figure 3-3). Values for “X-Feret” and “Y-Feret ” are recorded in pixe ls. In contrast, the measure for “Length” records the distance of the longest arc betw een two points within an object (maximum caliper distance) whereas “Breadth” measures the minimum caliper distance (Figure 3-4) (Russ 2002; Rovner 1995). Both “Length” and “Breadth” operate independent of object orientation. The IPTK software outputs a text file of over 20 different calculations of shape and dimension. Not all of these values were used in this research. Therefore, this discussion will be limited strictly to those measures used to analyze lithic flakes. In particular, I will discuss the use of “formfactor,” “roundne ss,” “aspect ratio,” and “elongation” in conjunction with a separate measure that r ecords the ratio of th e “formfactor” and “roundness” values of an object. Formfactor, roundness, aspect ratio, and el ongation are dimensionless measures. This means that they record similar va lues depending on the shape of an object irregardless of size. The mathematical formulas for these measures are shown in Appendix B. Formfactor measures the change s in the perimeter of an object irrespective of the elongation of the obj ect (Russ 2002; Rovner 1995). Es sentially, the formfactor measures the edge roughness of an object. Roundness, on the other hand, measures how round an object is independent of the object’s perimeter roughness (ibid). Used together, the formfactor and roundness indices provide a very descriptive analysis of the shape of an object. Both of these measures are cal ibrated against a perfect circle which has a value of ‘1’. Within both m easures, greater values indicat e the degree of departure from a perfect circle. Figure 3-5a shows several “generic” shapes including a circle, square, rectangle, and triangle and their corres ponding formfactor, roundness, elongation, and


46 formfactor divided by roundness values for co mparison. Figure 3-5b shows these same morphometrics measures for more amorphous sh apes. In particular figures 3-5a and 35b illustrate the differential change in form factor and roundness values as the perimeter (irregularity) and roundness (e longation) diverge more and mo re from a perfect circle. The quantification of shape is certainly a benefit for objectivity and standardization in analysis, but it is also a pr oblem as well. If I told anot her person that an object is “subcircular” then that other person might have a rough indication of what shape I am discussing albeit most likely with slight subj ective differences from what I think is “subcircular.” However, if I were to tell th is same person that a specific object had a formfactor and roundness value of ‘0.89’ this most likely would not connote that I have actually just described a very similar “s ub-circular” shape because we do not use numbers to classify shapes in everyday activ ities. The ability to use this quantitative shape analysis, therefore, hinges on a firm understanding of how the measures operate because the values themselves have little relevance to everyday activities. The other two default measures of sh ape introduced earlier are elongation and aspect ratio. Elongation measur es the skeleton length of an object divided by the mean fiber width (Russ 2002; Rovner 1995). Skeleton length is the length of the centerline of the object. The skeleton is created by removi ng all pixels in an object except those that make up the midline. Mean fiber width measur es the mean distance from all pixels in the object skeleton. Aspect ratio is essentially the same measure as cephalic index and it measures the length of an object divided by its breadth (ibid). Figures 3-5a and 3-5b illustrate the properties of elongation using si milar shapes as formfactor and roundness.


47 Finally, the Formfactor divided by Roundness was used to provide a relative index of the irregularity of an object versus its roundness. A perfect circle still has a value of ‘1’. However, as the perimeter increases relative to the elongation (greater irregularity in the shape of the perimeter) the morphometry values will be lower. In contrast, an object with a low perimeter irregularity and a high el ongation will have a larger number (Figure 3-5a and 3-5b). Flakes The whole purpose of employing a morphometric s approach in this research is to identify the relative patterns of the shape of stone tools over time and interpret the process of these patterns using a Complex Syst em Theory and cognitive ideal types. At first it seems logical to analyze the actu al tools that embody di rectly the thoughts and actions of technological producers. Such shap ed tools theoretically reference the ideal type in their very shape; that is, their shap e reflects the ideas pe ople held about how to produce the tool as best they can. Other researchers have also focused on the shape analysis of formal tools as well (Rovner 1995). However, it is widely known that the morphology of formal tools is modified over time through use, repa ir, and modification (Frison 1968; Jelinek 1976; Rolland and Dibble 1990). If it could be reasonably ascertained that a formally designed tool was not modified thr oughout its use life then this would be an appropriate lin e of investigation. However, while formal tools approach a direct one-plus representation of the ideal t ype of the technological agent, the debitage produced to make these formal tools should in fact be an “inverse representation" of the ideal type. To phrase this another way, stone tool produc tion debitage reflects those characteristics that are invers ely associated with the ideal type. Essentially, this method is akin to analyzing Michela ngelo’s David by only studying the shapes of the bits of rock


48 that he chipped off the main block. Predom inantly (because there are always mistakes and unintended variations), the debitage pr oduced during the sculp ting of stone (i.e., flaking) was removed because the producer c onsciously seeks to remove characteristics incompatible with their ideal end type conception. During the analysis of the data for this re search, I have relied on whole flakes as the primary morphological data source. I believe that flakes are the by-products of an intentional and predetermined process by the toolmaker and when analyzed together they represent the generalized underlyi ng intentions of the technol ogical system referenced to create that assemblage of artifacts. Primar y, secondary, and tertiary flakes, as well as stage of manufacture to produce flake debitage may prove morphometrically identifiable with future research. However, this study gr oup encompasses all of these groups into just one category—flakes—as a representation of the overall inverse ideal type. The incorporation of a statistically significant sample size of fl akes should then represent an average ideal type conception for the range of tools used during one specific time period. Changes in the shapes of the flakes should then correspond with changes in the conceptions of the ideal tool forms, intenti onally or unintentionally. For example, if a lithic technological system is modified to produce more elongated tool forms than were produced prior, the flake debitage should s how a related altered mo rphology to facilitate these changes. If the changes in the concep tions of tool forms represent a gradual and continuous process of technological change as I have suggested previously then the application of morphometric s to a study of flakes should show trends of gradual technological change within the archaeological record.


49 Conclusion Computer-assisted shape anal ysis is a unique technique to study the morphometrics of objects. It provides standa rdization within an analysis and a means to analyze large datasets relatively quickly. Mo rphometrics is also relatively inexpensive. By applying a morphometrics analysis to study flake debitage and explaining this data using a Complex Systems approach, the next chapter will expl ore how this method and theory combination usefully identifies technological change s within two sets of data.


50 Figure 3-1 Measurement of flake “length” vari es based on the methods used. In Figure 3-1A, flake length is the inte rsection with a lateral margin of the flake prior to reaching the distal edge of the flake. Figure 3-1B represents flake length perpendicular to striking platform wi dth. Figure 3-1C represents how the flake shape of Figure 1A also can be measured as the maximum length from the proximal to dorsal flake ends irregard less of lateral margins. Figure 3-1D, however, shows that mathematically th e greatest length of any rectangular flake is actually diagonal to the pr oximal and distal flake edges.


51 AFlake SelectionBConversion to Binary FormatCOutput table of Morphometrics values Figure 3-2 The steps necessary to conduct a morphometrics analysis using computerassisted shape software. Flakes must first be positioned within a photograph and then selected using a graphics pr ogram. Next, the selected flakes are converted into binary format. Fina lly, after running the morphometrics software the program creates a ta ble of morphometr ics values.


52 Figure 3-3 X-Feret and Y-Feret represent measure the widest points of an object dependent upon an object’s orient ation to the X and Y axis. Figure 3-4 In contrast to X a nd Y-Feret (figure 2-4), Figure 3-5 illustrates how the Image Processing Toolkit represents “length” and “breadth” of an object. Here, length represents the distance of the l ongest arc within an object (maximum caliper distance) whereas the breadth measures the minimum caliper distance of an object.


53 Figure 3-5 How different morphometrics statis tics measure different shapes of objects. Figure A illustrates how the formfact or, roundness, elongation and formfactor divided by roundness statistics char acterize common shapes. Figure B represents the same morphometrics statistics applied to more amorphous shapes. Figure B adapted from Rovner 1995. 1.011 0.776 0.582 FF / RR 21.819 12.236 9.909 Elongation 0.169 0.389 0.631 Roundness 0.171 0.302 0.368 Formfactor 1.043 1.428 1.221 1.583 0.942 FF / RR 5.103 2.334 1.009 2.294 1.007 Elongation 0.570 0.464 0.646 0.485 0.986 Roundness 0.595 0.662 0.789 0.767 0.929 Formfactor A B


54 CHAPTER 4 THE GILGEL GIBE AND GOGOSHIIS QABE MORPHOMETRICS ANALYSES Introduction Computer-assisted morphometrics analysis was presented in the preceding chapter as a means for identifying technological chan ges within the archaeo logical record using flake debitage. Morphometrics analyzes the shapes of object, such as flakes, and can provide a clear indication as to how sh apes of objects change through time. Flake debitage is especially useful to study within a morphometrics analysis for three reasons. First, flakes are frequently found at archaeo logical sites, many times being the most prevalent artifact type. Second, whole and unmodified flakes, unlike shaped tools, do not change morphology during their use-life. There may or may not be a correlation between flake debitage shape and stage of manufacture for the tool being produced, between primary, secondary, tertia ry, and retouch flakes, and even raw material. However, this analysis assumes th at grouping all whole, unmodified flakes into one category, and measuring the total variati on of flake shape within this generalized group, will provide an overall measure of the fl ake shape per assemblage. Third, in spite of being unable to determine reliably the exact stage of manufacture of a shaped tool, an analysis of flakes provides an alternative window into the technological system. Instead of looking directly at final tool forms, anal yzing the debitage removed to create such tools provides an inverse por trait of the ideas and intentions within the minds of prehistoric technological producers using a redu ctionist process. As noted in chapter four, a morphometrics analysis of flakes is akin to studying Mich elangeloÂ’s David by


55 studying only the debris removed from the final shaped form; the key element—the sculpture itself—can not be clearly discerned, but by studying what was removed to make that sculpture, we can acquire some id ea of what it may have represented. Therefore, the focus of this chapter is th e application of morphometrics analysis on flakes from the two Stone Age sites of Li ben Bore in southwestern Ethiopia and Gogoshiis Qabe, southern Somalia. Drawi ng upon CST, particular emphasis is given towards identifying long term technological ch ange within each dataset and describing the perturbative actions of the lithic tec hnological system around the assumed existence of ideal types. At the conclusion of each section, each dataset is compared against current culture historical sequences within each area in order to discern more particular technological characteristics within the da ta and also provide some measure of unification with currently empl oyed chronological methods. Liben Bore Between 1999 and 2002, archaeologists from the Gilgel Gibe Archaeological Project (GGAP) conducted a series of emerge ncy surveys and site excavations in the Gilgel Gibe River Basin of Southwestern Ethiopia in conjuncti on with the Ethiopian government’s plan to construct a major hydr oelectric dam and reservoir (Brandt 2000; Kinahan 2004). GGAP identified and record ed many archaeologica l sites, including GG35, an open air site situated 30 meters west of a small river that feeds into the Gilgel Gibe River, but unfortunately directly in th e path of a new road to be built around the edge of the reservoir. First identified by a lithic and ceramic surface scatter, two 1m2 test units were excavated at Liben Bore, followe d by four 1x2 m units taken down in 10 cm spits. These excavations exposed over tw o meters of stone and ceramic artifacts embedded within silty clay deposits ther eby revealing the deepest archaeological


56 sequence in southern and western Ethiopia and the second deepest sequence in all of Ethiopia. Since no faunal remains and only very sm all amounts of charcoal from the uppermost levels were recovered from the homogeneous silty clay deposits, radiocarbon dating of the sequence has not been possibl e. Furthermore, the homogenous clay deposits of Liben Bore have not as yet lent themselves to OSL dating. Typologically, occupation of the site appear s to span more than 30,000 year s from the late Middle Stone Age/early Later Stone Age through the Iron Age, and may span the Pleistocene/Holocene transition as there appears to be no evidence of a depositional hiatus or disconformity. Excavations revealed three major litho-stratigraphic units from surface to bedrock: LSU1 occurs from the surface to 20cm below surface (surface to excavation level 1) and is a very dark brown clay encomp assing Later Stone Age, Neolithic or Iron Age lithics and ceramics. LSU2 occurs from about 20cm to 60cm be low surface (excavation level 1 to level 3). LSU2 is a reddish brown clay contai ning Later Stone Age/Ne olithic lithics and ceramics. Lithics increase in quantity within this unit. LSU3 occurs from 60cm to 200cm below surf ace (excavation level 4 to level 18). This unit is a dark reddish brown cla y. At approximately 100cm below surface pottery disappears and only lithics remain. At this time new elements in lithics appear including higher freque ncies of blades and unifacial and bifacial points. The analysis of the Liben Bore artifacts was undertaken at the National Museum in Addis Ababa, Ethiopia. A. Negash (2004) re cently completed an attribute analysis of the Liben Bore shaped and unshaped tools and cores, while B. Kimura (2003) has analyzed the Liben Bore ceramics. Due to time constraints, I focused largely upon the morphometric and techno-typol ogical analysis of flakes from two excavation units (S14W1 and S23W1).


57 Prior to GGAP, very little was known a bout the Stone Age archaeology of the Gilgel Gibe area, or for that matter Southweste rn Ethiopia. It was this precise lack of knowledge for this large area that provided the impetus to search for new and more expedient methodologies (i.e., morphometr ics) that might help resolve the inadequacies of culture historical typologies. Morphometrics Analysis Morphometric analysis of the flakes included a series of measurements such as formfactor, roundness, elongation, aspect ratio, and X-Feret and Y-Feret. Before any descriptive statistics are ge nerated from the data, the da ta are further represented graphically by plotting the excavation levels of the entire dataset on the X-axis and the particular morphometric value on the Y-axis (Fi gure 4-1 to 4-6a and b) Visual analysis of these graphs is an essential first step in th e analysis of this data because it provides the analyst the opportunity to observe how the data changes and note any possible patterns within the dataset. Within this analysis initial pattern recogn ition was based on the changes in the mean morphometrics value between excavation leve ls, changes in the standard deviations of morphometrics values between excavation levels, and the changes in the extreme upper and lower morphometr ics values between excavation levels. Particular emphasis was also given to rapid changes (i.e., between one or two contiguous excavation levels) in the vari ance of morphometrics values between excavation levels. There are two initial observations of the Liben Bore morphometrics data based only on the visual analysis of the data. First, th e flakes represented w ithin the morphometrics graphs appear to change shap e through time. Since there appears to be no evidence of disconformities in the depositional sequence, the Liben Bore flakes appear to change shape gradually. Second, there appear to be three different patterns observable within the


58 Liben Bore sequence. These patterns correla te with excavation levels 1 to excavation level 8 (pattern 1), excavati on level 9 to excavation level 14 (pattern 2), and excavation level 15 to excavation level 26 (pattern 3). Of the total flakes within the dataset, pattern one contains 11.55%, pattern tw o contains 49.31%, and pattern three contains 39.14%. The trends within and between patterns s uggest that pattern th ree, the oldest and lowermost pattern of the dataset, contains flakes that are di stinctly ovate and lack much elongation (Figure 4-4) or edge irregularity (Figure 4-1) than the rest of the dataset. By comparison, in pattern two flake shape is much more elongated and has greater edge irregularity. In pattern one the flake shape has transitioned back to a less elongated form than in period two but does not appear as ova te as seen in period three. Because each morphometrics measure is independent of ot her morphometrics measures, the patterns noted visually within each mo rphometrics measure (n= 6) of the dataset are suggestive of several lines of independent evidence of specific patterns within the data Therefore, the inter-pattern differences a nd trends within the dataset were further identified using quantitative stat istical analysis including analys is of descriptive statistics, t-test, and cluster analysis. In order to quantify each pattern, the mean, minimum, maximum and standard deviation of each mo rphometrics measures per excavation level was first calculated. These data were then used to calculate the average standard deviation, minimum, maximum and mean values per pattern range within the data. The results of this analysis su ggest that pattern three has a mean aspect ratio of 1.5384, a mean formfactor of 0.6895 and a mean r oundness of 0.6104. The aspect ratio value suggests a slightly elongated form, the roundness value also suggests a slightly elongated


59 shape but also fairly ovate, and the formfactor indicates a low level of edge irregularity (refer to Figures 4-1, 4-2, and 4-3 comparison). The values for pattern two, however, contrast markedly with pattern three values. Pattern two has a greater mean aspect ra tio value (1.760) indicating more elongation, a lower formfactor value (0.655) indicating mo re edge irregularity, and a lower roundness value (0.551) indicating these flakes are less round than flakes in pattern three. In particular, the aspect ratio (Figure 4-3) in pattern two is greater than pattern three by a value of 0.222. This suggests that the flakes in pattern two are much more elongated than flakes in pattern three. When this value di fference is compared to the mean aspect ratio of flakes in pattern one (1.613) it indicates that the flakes in pattern two are the most elongated of the entire dataset. Pattern one is relatively similar to patter n three in terms of having a more ovate flake shape with decreased edge irregularity. In particular, the mean aspect ratio value (1.613) for flakes in pattern one indicates a mo re elongated form than the flakes in pattern three (1.538) though by no means as elongated as flakes in pattern two (1.760). A t-test was further conducted on this data to determine the statistical difference between the inter-pattern mean values within the dataset (Table 4-1). According to this test, there is no statistical di fference between the mean values for pattern one or pattern two, there is a small difference between mean va lues in pattern one and pattern three, and the mean values for pattern two are statistic ally different than each of the other two patterns. The alpha value for this test was set at 0.05. A cluster analysis of the Li ben Bore data using the st andard deviation and mean values per excavation level fo r the roundness, aspect ratio, a nd formfactor measures also


60 supports the differentiation of period two fr om periods one and three (Figure 4-7). According to these results, most of the excavation levels identified earlier into pattern two (excavation level 9 to level 14) as well as levels 6, 8, 17, and 18 cl uster together at a rescaled cluster case distance of 25. The re maining excavation levels cluster together into another main group with small sub-clusters at differing scaled cl uster distances. The significance of these results wi ll be discussed below. Understanding the Liben Bore Data Although it is still unknown when certain t echnological changes occurred precisely at Liben Bore, the morphometrics analysis of the Liben Bore flakes suggests evidence for technological change within the 26 excavation levels of units S14W1 and S23W1 at this site. The Liben Bore dataset represents thr ee morphologically different technological patterns during the times of occ upation at this site. Theore tically speaking, and ignoring other potential sources of vari ation, the analysis of the morphometrics data suggests that pattern three (excavation levels 26 to level 15) is characterized by an ideal type emphasizing the production of ovate flakes with little edge irregularity or tools requiring the removal of ovate flakes for proper tool production. The flakes in pattern two (excavation levels 14 to level 9) indicate a much different ideal type. Here, the technological system emphasi zes the production of more el ongated flake forms, which may be indicative of more elong ated tool forms. In patter n one, however, the ideal type changes yet again, this time back to a more ova te shape reminiscent of pattern three. Drawing upon the theoretical di scussion of the technological life cycle presented in chapter two, the Liben Bore da taset could represent the peak and initial decline of the lithic technological system within the Gilgel Gibe region of Ethiopia. If this is an


61 accurate assumption, pattern three would repres ent a still-developing lithic technological system while pattern two represents the peak of the lithic technologica l system within this area. The end of pattern two (level 9) or th e beginning of pattern one (level 8) may further represent the introduction of a compe ting technological system as perceived from the conception of complexity, which reflect s the number of user s of a technological system and emphasizes the ability for more intentional and unintentional variations within a technological system for greater technological variab ility. Technological variability is identified within the Libe n Bore dataset by calc ulating the standard deviation of each morphometrics statistic per excavation level. Excavation levels with less flake shape variability are assumed to correlate with less technological complexity, and should exhibit a lower standard deviati on than excavation leve ls with greater s technological complexity. Once the standard deviation values of each morphometrics statistic are calculated and grouped into the three patterns, the mean of these values is taken and then used to compare against each of the other two patterns. In comparison against the mean standard deviation for the formfactor, aspect ratio, a nd roundness of pattern one and pattern three, pattern two has a much greater standard deviation than either the other two patterns (table 4-2). If standard deviation is an accurate re presentation of the degree of perturbation due to the complexity of a technological system, th en it can be concluded that pattern two has possibly the greatest technologica l variability and t echnological complex ity of the entire dataset.


62 Furthermore, a separate technological system, in the form of ceramics, may have directly or indirectly influenced the local lithic technological syst em. Analysis of the ceramics recovered at Liben Bore suggests that ceramics are first introduced to this site around level 8. Negash (2004) has suggested that the introduction and adoption of ceramics most likely also entailed the restru cturing of food procurement strategies and group mobility (Negash 2004). Therefore, the sh ift in procurement strategies and group mobility may have inadvertently enacted changes to the predominant application of certain technological resources including knowledge, technologi cal producers, and users from the lithic technological system. This does not deny that a lithic technological system still occurred during and after the a doption of ceramics. However, based on the model of complexity advocated within this thesis, the lithic tec hnological system would exhibit less variability following the adopti on of a competing tec hnological system and would be in a state of decline because the shift in subsistence strategies and mobility would re-direct the applica tion of knowledge and resour ces to produce and maintain certain technologies. The possible loss of technological producers wi thin pattern one also would create a technological system incapable of mainta ining existing technologi cal variability or creating new variability. Unnecessary variat ions were discarded leaving in place only those variations necessary for the fundamental application of lithic technology. As a result, the remaining vestige of the lithic technological syst em within trend one assumed the appearance of a less complex and less variab le technological syst em. Due to similar levels of complexity, trend one and trend th ree should appear similar. The t-test and cluster analysis results of the morphometri cs values do suggest that pattern one and


63 pattern two are statistically similar to one a nother and cluster together when compared to pattern two. However, any such similariti es between pattern one and pattern three are purely superficial as the essential character —that is the ideal type—in pattern one is qualitatively different than the ideal type in pattern three; pattern one merely represents the continuation of qualitative change to th e ideal type and technological system via the technological life cycle. Thus far, the morphometrics an alysis has been able to identify three patterns within the Liben Bore dataset, statistically differentiate these patterns, and identify the morphology of the flakes within each pattern that are indicative of the ideal types. Furthermore, this analysis has suggested how to measure the perturbation around ideal types and the technological ch anges between each pattern vi a alterations to underlying ideal types. Given the limitati ons and restrictions of the mo del, this analysis has been able to identify technological change th rough time based only on a shape analysis of flakes. Morphometrics analysis, however, is not a standalone method. If only a morphometrics analysis on flakes was conducted for a site, this leaves out vast bodies of data including shape tool analyses. Concu rrent with the Liben Bore morphometrics analysis, which analyzed 1,810 flakes, this au thor also conducted a techno-typological analysis of the shaped tools (n > 1,900) r ecovered from excavation unit S14W1 at Liben Bore. This analysis was followed with a si milar analysis (n = 2,555) by A. Negash who analyzed Liben Bore units S23W1 and S23E1 in conjunction with the prior analysis results from S14W1.


64 From these analyses, a much more coherent picture can be made of the Liben Bore data. As a whole, the dataset is scraper-dom inated in all three excavation units with a large quantity of points also at S14W1 (n = 19) (cf. Negash 2004). Negash (2004) subdivides the Liben Bore data into two distinct time periods based on the introduction of pottery in level 8 and the more apparent ex pedient nature of the lithic assemblage between level 8 and level 1. The expedient natu re of the lithic assemblage between level 8 and level 1 could indicate a more sede ntary lifestyle as a result of changed technological, economic, and associated mobility patterns. Furthermore, Negash (2004) recognizes that the earliest levels in the Libe n Bore sequence, which contain the Levallois technique, may in fact also repres ent another separate time period. Based on the very similar results from the morphometrics an d techno-typological analyses, it seems very likely that at least one, if not more technological shifts did occur during the time of occupation at Liben Bo re. Regardless, both analyses identify a technological shift in level 8 possibly due to the introducti on of ceramics thereby issuing more substance to the claim that level 8 represents the introduction of a competing technological system (i.e., pottery) and subs equent decline of th e lithic technological system. In fact, the expediency of shaped tools that Negash note s between level 8 and level 1 can be interpreted to reflect changing ideal types enacted by a shift in the predominant application of technological know ledge, resources and us ers. According to the model, these changes created less tec hnological variation, loss of prior specific technological characteristics, and resulted in the subsequent appearance of a lithic technological system lacking any significan t technological complex ity at this time.


65 Gogoshiis Qabe Gogoshiis Qabe (GQ) is a rockshelter lo cated in the inter-ri verine Buur Heybe region of southern Somalia (Bra ndt 1988). The site was first excavated by P. Graziosi in 1935. Further excavations by the Buur Ecologi cal and Archaeological Project (BEAP) in 1983 and 1985 found the burials of fourteen indivi duals providing the ea rliest evidence of mortuary practice in the Horn of Africa. In total, the BEAP project excavated thirty-two contiguous 1m2 units in 5cm levels (ibid). The primary goal of the Gogoshiis Qabe mo rphometrics analysis was to test the reliability of the morphometrics approach b ecause of an overall concern that the Liben Bore analysis could have been misleading due to previous knowledge of the site. Since the Liben Bore morphometric analysis wa s conducted simultaneously with a technotypological lithics analysis, in sp ite of the persistent lack of information about the site in particular, this nonetheless allowed for dire ct questioning of the reliability of the morphometrics method itself. In particular, the question at hand was which analysis led the other; the results of th e morphometrics method might onl y have been reached through a priori conclusions and observations based on the techno-typologica l lithics analysis. Therefore, the Gogoshiis Qabe morphometric s analysis was conducted without any prior knowledge of the technological characteristics or prior conclusions reached of the GQ assemblage, temporal periods, or precise spatia l distribution of the materials. For this reason, more detailed background data for G ogoshiis Qabe will be given following the discussion of the morphometrics analysis. Morphometrics Analysis The GQ dataset has been very difficult to expl ain in spite of being relatively easy to interpret. The GQ dataset is composed of flakes spanning thirty-six 5cm excavation


66 levels. Unfortunately, at the time of this analysis several ex cavation levels were unavailable for analysis because their st orage bags had disintegrated allowing the artifacts to mix together (e.g., excavation level 17 and level 18). Based on the formfactor and roundness values, the GQ data set does not exhibit any mark ed differences indicative of technological changes when graphed against their excava tion levels (Figures 4-8 and 4-9). There are some notable constrictions in the range of formfactor and roundness values around level 7 and level 27 but this more likely represents poor sample size and not actual restriction of technological variation. However, a different picture emerges wh en the elongation values are graphed by their excavation levels (Figure 4-10). This graph clearly shows a distinction between the elongation of the flakes from level 1 to leve l 16 and the flakes in level 19 to level 36. According to this graph, the flakes from leve l 19 to level 36 have a much greater range of elongation, if not more elongated as a whole. When the aspect ratio is similarly gra phed (Figure 4-11), the trend noted in the elongation graph again disappears. Though elongation and aspect ratio are both dimensionless measures of the elongation of an object, they operate differently. Aspect ratio measures the length of the object agains t its breadth. In part icular, the “length” measured in the aspect ratio is independent of orientation and is simply a measure of the longest arc length within an obj ect. Width conversely is th e measure perpendicular to the length. Therefore, if an object has a greater width than leng th (e.g., a side-struck flake) the aspect ratio will be high because the lengt h is actually measuring the widest point of the flake, that which is often referred to as “breadth.”


67 Elongation, on the other hand, measures the skeleton length of the object against the mean fiber width. Skeleton length is dete rmined by removing all pixels in an object except those that make up the midline. Mean fiber width measures the mean distance from all pixels in the object skeleton. Theref ore, when a flake is “triangular” in shape—a shape very common to both the Liben Bore and Gogoshiis Qabe datasets—the skeleton length actually splits and creates a main ar m with two branches similar to a “peace” symbol. As a result, triangular and even ova te shaped flakes can have high elongation values whereas the aspect ratio value might suggest co mpletely otherwise. In order to circumvent this problem, the X-feret and Y-feret values were divided similar to the aspect ratio. X-feret and Y-fere t measures are useful in a situation such as this because they do not just measure the longest arc length within an object (i.e., “length”). Rather X-feret and Y-feret are m easures of the two widest points on the X and Y axes and are thus sensitive to orientation. Since all flakes within the Gogoshiis Qabe (and Liben Bore) analysis were oriented platform down, the X-Y Feret ratio should indicate if the flakes from level 19 to level 36 are actually more elongated than the flakes from level 1 to level 16. When plotted against the excavation levels the X-Y feret ratio values indicate a pattern very similar to the as pect ratio suggesting that the elongation of flakes between levels 1 to 16 and levels 19 to 36 are not greatly different (Figure 4-12). The flakes between levels 19 to level 36 are actually slightly less elongated than flakes from level 1 to level 16 by an aspect ratio value difference of 0.128. The distribution of values within the el ongation graph, however, still suggests that there are underlying patterns within this data that have not been recognized. According


68 to the elongation values, the flakes between le vels 19 to level 36 are more triangular, or at least have a slightly greater width versus length, thereby creati ng a greater skeleton length and larger elongation value. Though the elongation measure may not be a good measure of actual elongation per se, this meas ure is still an accurate indication of the shape of these flakes. As a result, when the range and standard deviation of flake elongation values are compared between level 1 to level 16 and level 19 to 36 the results are striking. According to these results, the range of elongation values between level 19 to level 36 (n = 26.399) is over twice as great as the range of elongation values between level 1 to level 16 (n = 13.867). This ratio is further supported by averaging the mean range of elongation values per excavation level. Furthe rmore, the standard deviation of elongation values between level 19 to leve l 36 is 69% greater (n = 2.338) than the standard deviation between level 1 to level 16 (n = 1.620). This su ggests that there is greater variability in the overall shape of flakes between level 19 to level 36 in spite of the slightly greater roundness and less irregularity of shape co mpared from level 1 to level 16. A t-test conducted on this data also shows that there is a statistical difference at a 0.05 confidence level between the in ter-pattern mean values with in the dataset in four out of five morphometrics measures (Table 4-3). According to this test, there is a statistical difference between the mean values between excavation level 1 to level 16 (pattern 1) and excavation level 19 to level 36 (pattern 2). Furthermore, the results of a cluster analysis on the GQ data using the standard deviation values per excavation level for the formfactor, roundness, elongation, and aspect ratio group the majority of levels from pattern two independent of pattern one at a


69 rescaled cluster case distance of 4 (Figure 4-13). Fourteen out of seventeen excavation levels categorized morphometr ically within patte rn two do in fact group within cluster two whereas eight out of thir teen levels classified mor phometrically within group one also group within cluster one. Only eight out of thirty morpho metrically analyzed excavation levels do not share similar morpho metrics and cluster analysis results. Finally, the shape of flak es between levels 19 to 36 suggests an ideal type qualitatively different from th e ideal type underlying the lith ic technological system in level 1 to level 16. Due to similarities in the formfactor, roundness and aspect ratios between these two time periods, the ideal type s must not have been greatly different. However, there must have been a great enough change to the lithic technological system around level 19 to create more solidarity within the shape of flakes between level 1 to level 16. Understanding the Gogoshiis Qabe Data According to Brandt (1988), the Gogoshiis Qabe assemblage actually represents the transition from the Eibian LSA to the Bardaal e LSA-Neolithic industry. This shift occurs with the environmental change from cool, hyperarid conditions of the Pleistocene to warm and humid Holocene conditions (Brandt 1988) Brandt specifically identifies this transition at Level 19. The similar identification of two patterns separated at a similar excavation level shows that the application of morphometrics was again able to de termine technological change within an archae ological site using only fl ake debitage even though morphometrics on the Gogoshiis Qabe data wa s not as successful as the Liben Bore analysis. For instance, technological changes within the Liben Bore dataset were much


70 more pronounced than those identified at Gogoshiis Qabe. This can be accepted especially if technological changes ar e more gradual rather than punctuated. Thus, the discussion of Gogoshiis Qabe is significant for two reasons. First, the GQ analysis once again shows the validity of the morphometrics method for identifying technological change within the archaeological record. Second, the unique nature of the GQ data bring to bear certai n deficiencies or assumptions within the morphometrics method that warrant explanation. Specificall y, a deficiency of the morphometrics method rests on an underlying assumption that the archaeological deposits within a morphometrics analysis represent a con tinuous deposition of archaeological and geological materials over time. This assump tion was noted in the Liben Bore analysis, but warrants further discussion. Gradual and punctuated technological change s can be obfuscated or created within a morphometrics dataset by differing rates of archaeological and sedimentary deposition. It is important to note here th at the application of this ty pe of morphometrics analysis implies an archaeological sequence with a continuous rate of sedimentary and archaeological deposition over time with few, if any, abrupt changes in the rates of deposition. Keep in mind that this assumpti on does not preclude that differential rates of cultural or sedimentary deposition do occur and can distort the distribution of a dataset. It merely points out that abr upt changes in the sedimentary deposition at a site should be clearly identifiable whereby a period of rapid sedimentary deposition should show an equalized morphometrics distribution with few, if any abrupt changes. On the other hand a period of slow sedi mentary deposition should show a very distinct demarcation between the morphometrics values of artifacts. This demarcation


71 corresponds to the much coarser temporal resolution, and collection of archaeological materials, represented in the period with slow sedimentary deposition versus other periods with more refined temporal re solution of archaeological materials. At Liben Bore, there is no evident di stinction in changes to the rate of archaeological or geologic deposi tion at that site. Due to the consistently fluctuating morphometrics values throughout the entire Libe n Bore sequence, this suggests that there was continuous sedimentary and archaeological deposition over the entirety of the Liben Bore sequence creating uniformity in the frequency of archaeological materials and temporal resolution. Gogoshiis Qabe suggests otherwise. Although archaeological materials were recovered in excavation level 17 and level 18, the abrupt nature of this division as shown in the statistical analyses between pattern one and pattern two, whic h correlates with a major sedimentary and environmental shift as noted by Brandt (1988), suggests the possibility for either slow sedimentary de positions or dis-continuous archaeological deposition during this restricted period of time. If there is a gap in the deposition of archaeological materials then one can only expect to see significant technological changes between the two patterns (i.e., leve l 1 to level 16 and le vel 19 to level 36) separated by a large sp an of time and significant changes to the environment. However, as the shape of flakes remain similar in many respects throughout the entire Gogoshiis Qabe sequence, this suggests other, as yet unknown, factors homogenizing the morphology of flakes within each lithic te chnological systems operating in this area through time.


72 Conclusion This chapter has been able to show that the morphometrics method is capable of identifying technological changes and tec hnological variations within the lithic technological systems at Gilgel Gibe and Gogoshiis Qabe. It is hypothesized that these shifts correlate to equivalent changes in the ideal types within the technological systems. At Liben Bore, computer assisted morphom etry has been able to identify three statistically different morphologi cal patterns within the datase t indicative of technological change whereas at Gogoshiis Qabe, the re sults were similar and identified two statistically different patterns within th e dataset correlated with environmental and sedimentary changes. Thus not only does computer assisted mor phometrics work when applied to flakes, but it is also fast. The analys is of 1,810 flakes for the Gilgel Gibe analysis took less than three weeks while the an alysis of 1,809 flakes for the G ogoshiis Qabe analysis took less than two weeks. Therefore, a morphometr ics analysis could theoretically operate concurrent with an excavation with minor extr a expenditure of time. This would afford a significant advantage towards understanding the nature of a site in the process of excavation versus knowing these difference afte r excavation has finished and the artifacts are being analyzed in a laborat ory. Furthermore, the applica tion of morphometrics is also capable of providing significant insight into the character of archaeological deposits in previously studied, or unstudie d, archaeological sites and even regions. The point I must stress here is that morphometrics is by no means a standalone method, but in conjunction with CST, the application of this theo retical and methodologi cal package provides significant advantages towards understanding St one Age archaeological sites in a timely, cost-effective manner. In so doing, archaeologist s may be able to better tackle the spatial


73 and temporal immensity of Stone Age archaeol ogy within the Horn of Africa and impart a much more firm foothold towards understa nding the complex sequence of archaeology within this region.


74 Table 4-1 T-test results of the inter-period di fferences in the mean values for formfactor, roundness, aspect ratio, elongation, and X-fe ret and Y-feret. Each trend was compared against the other two. The result s of this test suggest that there is no statistical difference between the mean values for trend 1 and trend 2, there is a small difference between mean va lues in trend 1 and trend 3, and the mean values for trend 2 are statistically different than each of the other two trends. Bold values indi cate statistical difference. Roundness Elongation Aspect Ratio Formfactor X-Feret / YFeret T1/T2 0.188 0.509 0.101 0.398 0.261 T2/T3 1.54E-05 0.520 9.75E-05 0.005 0.003 T1/T3 0.036 0.949 0.051 0.120 0.207 Alpha Value = 0.05 Null Hypothesis: No Statistically Sign ificant Difference in mean value. Table 4-2 Averaged standard deviation of the formfactor, aspect ratio, and roundness values organized per excavation level time period. Roundness Formfactor Aspect Ratio Period 1 0.112 0.060 0.387 Period 2 0.134 0.088 0.530 Period 3 0.111 0.067 0.335 Table 4-3 T-test results of the inter-period di fferences in the mean values for formfactor, roundness, aspect ratio, elongation, and X-fere t and Y-feret. The results of the test show that trend one (excavation level 1 to level 16) is statistically different from trend two (excavation leve l 19 to level 36) in four out of the five morphometric measure results. T1/T2 XFeret and Y-Feret 0.001 Formfactor 0.022 Roundness 0.002 Aspect Ratio 0.002 Elongation 0.423 Alpha Value = 0.05 Null Hypothesis: No Statistically Si gnificant Difference in mean value


75 Figure 4-1 Morphometric formfactor values of the Liben Bore flakes organized by the level of excavation. According to these graphs, the flake shape at Liben Bore became more irregular between level 16 and level 8. Lower levels of excavation correspond with greater excavation level numbers. The top graph is a box-and-whisker plot representing th e mean, standard deviation, first and third quartiles of the formfactor data Circles in this graph correspond to outliers whereas asterisks represent extreme values. The lower graph represents the raw formfactor data poin ts used to create the box-and-whisker graph.


76 Figure 4-2 Morphometric roundness values of the Liben Bore flakes organized by the level of excavation. According to these graphs, the flake shape at Liben Bore became less round between level 15 and le vel 8. Lower levels of excavation correspond with greater excavation leve l numbers. The top graph is a boxand-whisker plot representing the mea n, standard deviation, first and third quartiles of the roundness data. Circles in this graph correspond to outliers whereas asterisks represent extreme valu es. The lower graph represents the raw roundness data points used to create the box-and-whisker graph.


77 Figure 4-3 Morphometric aspect ratio values of the Liben Bore flakes organized by the level of excavation. According to thes e graphs, flake shape between level 15 and level 8 became more elongated at Libe n Bore. Lower levels of excavation correspond with greater excavation leve l numbers. The top graph is a boxand-whisker plot representing the mea n, standard deviation, first and third quartiles of the aspect ratio data. Circ les in this graph correspond to outliers whereas asterisks represent extreme valu es. The lower graph represents the raw aspect ratio data points used to create the box-and-whisker graph.


78 Figure 4-4 Morphometric elongation values of the Liben Bore flakes organized by the level of excavation. According to thes e graphs, flake shape between level 15 and level 8 became more elongated at Libe n Bore. Lower levels of excavation correspond with greater excavation leve l numbers. The top graph is a boxand-whisker plot representing the mea n, standard deviation, first and third quartiles of the elongation data. Circles in this graph correspond to outliers whereas asterisks represent extreme valu es. The lower graph represents the raw elongation data points used to create the box-and-whisker graph.


79 Figure 4-5 Morphometric X-Feret and Y-Feret ratio values of the Liben Bore flakes organized by the level of excavation. According to these graphs, flake shape between level 15 and level 8 became more elongated at Liben Bore. Lower levels of excavation correspond with gr eater excavation level numbers. The top graph is a box-and-whis ker plot representing the mean, standard deviation, first and third quartiles of the X-Feret and Y-Feret ratio data. Circles in this graph correspond to outliers whereas aste risks represent extreme values. The lower graph represents the raw X-Feret and Y-Feret ratio data points used to create the box-and-whisker graph.


80 Figure 4-6 Irregularity (formfactor) divided by the roundness of the Liben Bore flakes organized by the level of excavation. These graphs suggest a trend towards more elongated and irregular flakes betw een level 15 to level 8. Lower levels of excavation correspond with greate r excavation level numbers. The top graph is a box-and-whisker plot repr esenting the mean, standard deviation, first and third quartiles of the Formfa ctor divided by Roundness ratio data. Circles in this graph correspond to outliers whereas as terisks represent extreme values. The lower graph repr esents the raw Formfactor divided by Roundness ratio data points used to create the box-and-whisker graph.


81 Figure 4-7 A hierarchical cluster analysis us ing the mean and standard deviation values per excavation level for formfactor, r oundness, and aspect ratio from Liben Bore. The graph shows that excavati on levels identified into pattern two (excavation level 9 to level 14) mos tly group independently of excavation levels identified into patt ern one and pattern three. In this graph “pattern” refers to the three identified patterns discussed in the text.


82 Figure 4-8 Morphometric formfactor values of the Gogoshiis Qabe flakes organized by the level of excavation. According to these graphs, flake irregularity fluctuated minimally at Gogoshiis Qabe. Lower levels of excavation correspond with greater excavation leve l numbers. The top graph is a boxand-whisker plot representing the mea n, standard deviation, first and third quartiles of the formfactor data. Circle s in this graph correspond to outliers whereas asterisks represent extreme valu es. The lower graph represents the raw formfactor data points used to create the box-and-whisker graph.


83 Figure 4-9 Morphometric roundness values of th e Gogoshiis Qabe fl akes organized by the level of excavation. According to these graphs, flake roundness fluctuated minimally at Gogoshiis Qabe. Lower levels of excavation correspond with greater excavation level numbers. Th e top graph is a box-and-whisker plot representing the mean, standard devia tion, first and third quartiles of the roundness data. Circles in this graph correspond to outliers whereas asterisks represent extreme values. The lower graph represents the raw roundness data points used to create the box-and-whisker graph.


84 Figure 4-10 Morphometric elongation values of the Gogoshiis Qabe flakes organized by the level of excavation. According to these graphs, flake elongation appears to be greater between level 36 to leve l 19 at Gogoshiis Qabe. Lower levels of excavation correspond with greate r excavation level numbers. The top graph is a box-and-whisker plot repr esenting the mean, standard deviation, first and third quartiles of the elonga tion data. Circles in this graph correspond to outliers whereas asterisk s represent extreme values. The lower graph represents th e raw elongation data points used to create the boxand-whisker graph.


85 Figure 4-11 Morphometric aspect ratio values of the Gogoshiis Qabe flakes organized by the level of excavation. These graphs contradict the trend for greater elongation between level 36 to level 19 as shown in figure 15. Here, flake length appears to remain relatively co nstant throughout th e entire Gogoshiis Qabe sequence. Lower levels of excavation correspond with greater excavation level numbers. The top graph is a box-and-whisker plot representing the mean, standard devia tion, first and third quartiles of the aspect ratio data. Circles in this graph correspond to outliers whereas asterisks represent extreme values. The lower graph represents the raw aspect ratio data points used to create the box-and-whisker graph.


86 Figure 4-12 Morphometric X-Feret and Y-Fere t ratio values of the Gogoshiis Qabe flakes organized by the level of exca vation. These graphs support figure 16 by also showing that the elongation of fl akes remained fairly constant over time at Gogoshiis Qabe. Lower leve ls of excavation correspond with greater excavation level numbers. Th e top graph is a box-and-whisker plot representing the mean, standard devia tion, first and third quartiles of the XFeret and Y-Feret ratio data. Circle s in this graph correspond to outliers whereas asterisks represent extreme valu es. The lower graph represents the raw X-Feret and Y-Feret ratio data po ints used to create the box-andwhisker graph.


87 Figure 4-13 A hierarchical cluster analysis using the standard deviation values per excavation level for formfactor, roundne ss, elongation, and aspect ratio from Gogoshiis Qabe. The graph shows that excavation levels identified into pattern one (excavation level 1 to level 16) group independently of excavation levels identified into patte rn two (excavation level 19 to level 36).

PAGE 100

88 CHAPTER 5 CONCLUSIONS This thesis focused on the application of Complex systems theory (CST) to interpret the process of te chnological change as it is observed in the prehistoric archaeological record of the Horn of Africa. The impetus for this thesis was based in reaction to perceived deficiencies in the pr edominant application of culture historical methodology in current African Stone Age arch aeological research and existing theories of technological change. The recognized deficiencies in culture history include lim itations to interpret data within the culture historical framew ork and the non-dynamic and discontinuous conception of time presented by culture hist orical typologies and essentialism. Deficiencies noted within contemporary studie s of technological change include an overreliance on using discrete events to explain changes over time in a technological system without adequately describing the actual pr ocesses involved (cf. Van der Leeuw 1989:3), scalar differences between the varied ideas of technological change, and a lack of discussion concerning unintentiona l technological changes. As a result, I have relied on eight primary theoretical aspects of CST within this thesis to derive specific e xpectation regarding the identi fication and interpretation of technological change in the African Stone Age archaeological record. These eight theoretical aspects include an opens sy stems model, non-linearity, determinism, sensitivity to initial conditions, critical path networks, contingency, emergence, and selforganization. The expectations I have derive d from CST within this thesis include an

PAGE 101

89 overall conception of continuous technologica l change with periods of punctuated changes, multi-scalar dynamics including th e ability to theorize on the existence of shared social-technological id eas (ideal types) and how th ese relate with in long term technological systems, and the ability to inte grate theoretically material culture and nonmaterial thoughts and actions of past people. I have appended several other theoretical ideas onto the existing complex systems model in order to more precisely account fo r the actions and ideas of people within a theory of technological change. First, I ha ve assumed an essentialist position advocating a mind-body dichotomy between the ideas in people Â’s heads and their subsequent actions. In contrast to culture histor ical essentialism I assume that there are underlying essential ideas that direct our actions in creating technological products but I reject the ability to know the exact ideal characteristi cs of an object now or in the past. By rejecting the ability to ever observe or define the essent ial qualities of an obj ect I hope to distance myself from culture historical essentialism by attempting to measure instead the variation of many objects around a hypothesi zed, but unattainable, ideal type thereby creating a bridge between thoughts and actions. In this way I hope also to find common ground between diametrically opposed essentialist and materialist theoretical positions. Second, I have assumed that technologi cal changes occur intentionally and unintentionally. In my theorization of CST in chapter 2 and my application of certain aspect of CST in chapter 4 I ignore many other internal and external systemic factors that may cause technological change and instead argu e that technological ch ange is located in the changes to cognitively-base d ideal types. More precis ely, I argue that technological change occurs due to a combination of inten tional and unintentional events that can serve

PAGE 102

90 to alter the trajectory of a technological system through the knowledge to identify their potential and successfully incorporate them into a technological system. This idea incorporates the punctuated ev ents of directed and purposef ul action (i.e., intentional change) in association with the constant acquisition of una nticipated techniques, tool characteristics, tool types, or technological knowledge in general acquired during the directed and purposeful events of intenti onal technological change (i.e., unintentional change). In summation of this thesis, I feel confiden t to conclude with two points. First, a morphometrics method using computer assisted shape analysis of flake debitage can be used to identify technological change in the archaeological record. Second, CST, as modified within this thesis to explore on certain parameters and explanation, can be applied usefully to the archaeological record to interpret the pr ocess of technological change. The application of computer assisted morphometry to anal yze lithic debitage in the archaeological record is still in its infancy, and the results of my morphometrics analysis at Liben Bore and Gogoshiis Qabe pose many new directions for future research. For instance, is there an observable difference in the flake morphometry between primary, secondary, or tertiary flakes in the archaeo logical record? How might style affect the adoption of unanticipated varia tions thereby affecting the ideal type? Does raw material affect the morphology of flake shape in sp ite of shared prop erties of chonchoidal fracture? Could a morphometric s analysis even be used to identify style in the form of flakes within the archaeological record?

PAGE 103

91 Within this thesis I have had to assume th at many of these questions will not have a direct influence on technological change. Ho wever, this does not be little the recognition of their possible significance within a theory of technological change, nor does it affect my conclusion about morphometrics in general. In this thesis I have merely set out in part to demonstrate that com puter assisted shape analysis can or cannot be used to identify technological change through time by analyzing how flake shape varies within an archaeological sequence. As an initial study, I feel c onfident to conclude that morphometrics does indeed id entify changes through time. However, identifying technological change through time was only half of this thesisÂ’ objective. The identified changes had to be interpreted as we ll in order to make them useful to archaeology. Th e application of CST, as I advocate it here, can provide a useful explanation of the process of technol ogical change identified in the archaeological record. Using expectation from a complex systems model I have been able to offer a minimal explanation of how the technologica l systems at both Liben Bore and Gogoshiis Qabe changed through time without any evid ence that facilitates why the changes occurred in the first place. Furthermore the complex systems model br ought together materialized actions and cognitive ideas and situated them within a multi-scale continuous and dynamic conception of time and change. And, it is here that I think CST may have the most to offer archaeological research in the future. When associated with a conception of ideal types, CST provides a useful model to pos it a relationship between ideas and actions, individuals and groups, and shor ter periods of time within mu ch greater spans of time. Using a complex systems model we do not have to restrain ourselve s to typologies that

PAGE 104

92 imply synchronicity. Instead we can begin to explore the many levels of variation operating continuously through time within a complex system that serve to create technological, social, and even behavior change s within people. In th e short section that follows I intend to posit how CST, as it is applied here, may be able to provide an alternative way to explain the development of modern human behavior, a direction for future research I intend to pursue. The epistemological implications of th e modern human origins debate include Eurocentrism, materialism and culture-historica l essentialism, but herald a much greater impact on the discipline itself by re-defining key concepts such as “modern behavior” (cf. Henshilwood et al. 2001; Henshilwood a nd Marean 2003; McBrearty and Brooks 2000:534; Wadley 2001). Current tr ends within this debate seek to move beyond simple normative and direct-historical trait-list approaches reminiscent of the culture-historical roots of European and African archaeology to wards a more social approach advocating intangible, socially-produced behavior as we ll as material culture (for example see Deacon and Deacon 1999:101-102; Henshilwood and Marean 2003:635; McBrearty and Brooks 2000:491-492; Mitchell 2002:104-105). As a result, the application of CST as a tool for the identification of cognitive id eal types and intentional and unintentional changes might be able to provide salient advantages towards answering key questions about the development of technology a nd behavior within this debate. First, recognition of both intentional and unintentiona l changes and how these changes are adopted into a tec hnological system through the cr itical path and contingency to transform the technological system can render obsolete the necessity of a strict dichotomy between “punctuated” and “gradualis t” models within the debate (cf. Brooks

PAGE 105

93 et al. 1995; Henshilwood et al. 2001; Hens hilwood and Marean 2003:630; Klein 1999; Kusimba 2003:117; McBrearty and Br ooks 2000:454; Milo 1998:99; Shea 2003; Thompson et al. 2004; Yellen et al. 1995; Milo 1998). Instead a complex systems model assumes that both processes must work togeth er to propel the traject ory of technological systems, and the behavior, actions, and know ledge of technological agents in new and unforeseen directions. Second, a complex theory model may be able to describe more accurately the interaction between material culture and behavior within this debate, thereby disqualifying certain current traits recognized as modern human behavior and possibly even moving beyond a culture historical trait list approach. The r eason for this is found in the ideal type. The ideal type, and a study of the variation around th ese ideal types, is a link between the material record and inta ngible behavior. The transition from a technological system with little or no fo rmal object definition to a technology with clearly pre-defined ideas that influence the overall forms of objects over vast geographic areas and time periods suggests the adopti on of specific ideologi cal principles. Precisely, these principles ma nipulate behavior by eleva ting the form of an object to that of “symbol” thereby transcending re gional specialization. If symbolization and standardization of tool forms does in fact he rald in part the founda tion of modern human behavior as others have suggested (c f. Deacon and Deacon 1999:101-102; Henshilwood and Marean 2003; Klein 1999:512; McBrearty and Br ooks 2000:491-492; Mitchell 2002:104-105; Wadley 2001) then the advent of the Acheulian “handaxe” may be the first tangible evidence for the developm ent of modern human behavior.

PAGE 106

94 This is just one example of possible futu re research for CST but is by no means the limits of this research. The applicab ility of both CST and morphometrics for archaeological research is much more diverse. Complex systems theory already has been used be other researchers to explain social change within archaeol ogical contexts (cf. Bentley and Maschner 2003; Van der Leeu w and McGlade 1997). Yet the complex systems model is still cutting-edge theorization of systems theory, chaos, and nonlinearity. As the literature on these diverse c oncepts increases, this should in turn open up many new avenues of theorizing how people behaved and acted in the past. However, computer assisted morphometry might have more practicality to mainstream archaeology. The expediency, reliability, cost-effectiveness, and standardization of morphometrics can play a significant role in future lithic analysis methodology. Refinement of mo rphometrics techniques can le nd this method very useful for identifying the character of archaeologica l deposits during an excavation and even using this method as a tool to correlate the morphological trends at one site with one or more other sites. Thus, morphometrics can b ecome just the method essential to Horn of African archaeology that provides a foothold on understanding the sp atial and temporal enormity of Stone Age assemblages within this area. Only time will tell though. Regardless, I intend to pursue both CST and co mputer assisted mor phometry in the hopes of advancing archaeological knowle dge and practice in order to contribute to the story of humanity.

PAGE 107

95 APPENDIX A AN EXPERIMENT ON TECHNOLOGICAL PROCESS USING STONE TOOLS Introduction A practical experiment was conducted dur ing the Fall of 2003 to investigate the process of intentional and uni ntentional technological change within an experimental archaeological setting as interpreted using a CST. This experiment enlisted nine volunteers with no prior experi ence making stone tools. The lack of lithic tool production experience was essential to ensure a contemporary de novo lithic technology with little influence from prior archaeological knowledge as to how stone tools have been made prior and a relatively even beginning point to compare each of the three groups against. This project was unde rtaken to observe whether: 1. Technological change occurs primarily th rough unintentional va riations incurred through intentional t echnological production. 2. Intentional and goal-or iented developments follow a pr e-planned route of research and technology while uninte ntional variations may al ter the trajectory of technological research, development a nd implementation in new and unforeseen ways. 3. Experiential knowledge plays a vital role in th e identification and implementation of unintentional variation with in a technological system. In particular, agents with low experiential knowledge of a technology w ill not grasp the fu ll significance of certain unintentional variations until a substantia l experiential knowledge base (i.e., critical path) is achieved. These three project goals were tested by observing the actions of each group member during each task and noting how thei r behavior changed, or did not change, between tasks and series. Each task was de signed to be cumulative thereby allowing the knowledge acquired in task one to also be appl ied to task two. However, each task was

PAGE 108

96 also distinctly different thereby forcing gr oup members to create new solutions for each task. The ability to recognize intentional and unintentional variations and note their influence on a groupÂ’s developed technologica l system hinged on my greater experience producing, using, and studying stone tools. Through my knowledge, I was able to identify the accidental production of tool form s or the accidental use of tool production techniques based on my prior experience with similar situations and examples from the archaeological record, and obser ve if the group member identified the usefulness of the unintentional variation or not. For example, if a group member accidentally produced a flake during a task requiring them to cut th rough a piece of leather, I observed if the group member identified this poten tially useful tool and how th ey incorporated it within their technological system. In later tasks I was able to observe how group members further utilized flakes, for example, in term s of their ideal tool types and actual tool production and if the group members had suffici ent knowledge to create more flakes or were unable to sustai n the tool type. The observations and interpretations made within this project rest on several underlying assumptions. These assumptions are: 1. Participants prior knowledge will not affect the results of their ability to create and sustain a lithic technology because their l ack of knowledge to produce stone tools. However, prior knowledge may limit the c hoices perceived available to each group (i.e., to chop though a piece of wood you need an axe-like tool). 2. Ideas direct actions. Shared ideas create sim ilarity within tool forms and tool use. 3. Technological change results primaril y from intentional and unintentional variations recognized during the process of technologica l production. The adoption of these variations changes the ideal t ype and technological system in general. 4. Technological change may or may not occu r through other proce sses but this is inconsequential to the effect of in tentional and unintentional changes.

PAGE 109

97 5. Modern-day experimental technological production should replicate comparable real-world archaeological technological production for both intentional and unintentional technol ogical change. The experiment consisted of each group i ndependently conducting a series of 4 tasks that were repeated twice in sequence (Series 1 and Series 2). Each task was designed so that every group could apply the kno wledge gained from the prior task to the next task. The specific tasks fo r each group included the following: 1. Develop an implement to cut through a section of leather. 2. Develop an implement to chop through a 2 inch thick diameter oak branch. 3. Develop an implement to inscribe fine, parallel lines onto a piece of wood. 4. Develop of an implement to bor e a hole in a section of wood. The rules for the experiment were simple and required that each group initially fill out a questionnaire asking what they intende d to produce, why they wanted to produce a particular design and how they conceived the tool manufactur ing process. Furthermore, each group was also required to submit a scale drawing of the intended final outcome of the tool(s) with a brief explanation of its particular hypot hesized morphology and function. This information was thought to be useful to compare a gr oupÂ’s intentions with the final tools produced. However, the drawi ngs of a groupÂ’s ideal tool were frequently much too generalized to be much use in comparison with the act ual tool(s) produced. Furthermore, the tool(s) produ ced during the task were fr equently modified, thereby creating inconsistency in the ability to compar e ideal tool form drawings against the endproducts. As a result, I found that my direct conversations with group members about their ideal tool and how they intend to make the tool were much more informative and useful for identifying thei r ideal types and how thes e change through time though intentional and uninten tional variations.

PAGE 110

98 The methodology was straightforward a nd consisted of giving each group a collection of variably-sized qua rtzite and chert cobbles. While the participants were told the common name of each rock, caution was ta ken not to use descriptive or functional terminology such as “core”, “hammerstone” or “flake” and thus bi as the participant’s conceptions of the “proper” functions of th e supplied materials. In addition to the quartzite and chert stones, ot her raw materials available to each group included a 26cm elk antler baton, a 15cm white tail deer antler tine and a 28cm x 18cm piece of leather. Series 1 Task 1: Cutting a Piece of leather The first task was to cut a large piece of leather using only the stone tools provided. All three groups decided to rely initially only upon naturally sharp edges of the chert instead of attempting to secondarily manufact ure a sharp edge. Frequently, a large rock was used as an anvil to suppor t a cutting tool from above. Alternately, a large rock was also situated on the ground a nd the leather simply rubbed over an upward facing sharp edge. Most importantly, one group eventually began to experiment with flaking chert after experimenting with an uni ntentionally produced flake. But while this group seemed to grasp the importance of flaking raw materi als to produce sharp edges, they could not duplicate the sharp flakes through their random and haphazard smashing two stones together. Therefore, they reverted back to a simple hammer and a nvil cutting technique. It was only toward the end of this task that the same group found a sharp flake unintentionally produced duri ng the hammer and anvil technique and swiftly and successfully scored the leather with the distal end of the flake. While two groups still did not recognize and utilize a flake tool technique, the recognition and utilization by the third gr oup demonstrated that the unintentional

PAGE 111

99 production of tool variations can alter the c onception of the ideal tool and how they go about accomplishing a task. Task 2: Chopping through a Branch The next task was designed to build on experiential knowledge gained from the previous task and required each group to c hop through a 5cm (2 inch) thick branch of oak. All three groups clearly no ted their influence with the im age of an axe as the pattern of choice because of its sharp bifacial blade, high mobility and the mechanical force it afforded to chop through wood. A second gr oup began to recognize the properties and value of flakes during this task. However, like the group in task one, this second group lacked the knowledge to c onsistently produce flakes. Flakes were still a novelty produced only by accident. Task 3: Inscribe a Series of Lines into Wood This task required each group to score a series of crosshatched and zigzag lines into a piece of wood. As in past experiments, a ll groups initially chose the simplest solution by utilizing readily available material of natu rally pointed stones. This process quickly failed to produce desired results so all groups eventually reverted to lessons learned in previous tasks and sought a dditional stones for flaking shar p edges and points. It was during this process of attempted flake pr oduction that one groupÂ’s quartz nodule they used as a hammerstone shattered. The qua rtz shatter produced pr edominantly non-flake angular debris with numerous sharp corners and edges. This gr oup quickly discovered that the angular debris from the quartz boulde r provided an excellent tool for inscribing the wood and shifted their focus from usi ng and producing flakes to using non-flaked angular debris initially made unintentionally.

PAGE 112

100 Up until this point I have emphasized th e usefulness and unintentionally produced nature of flake materials. However, the example from task three clearly shows that flakes are not the only useful unintended variation but just another vari ation nonetheless. Interestingly, in this task, the most succe ssful final tool was unintentionally produced while trying to produce flakes. What followe d was the important discovery that rough, angular and pointed pieces of quartz could also be successfu lly utilized for the required task at hand instead of expending time and en ergy on the precise production of flakes. Task 4: Drill a Hole through a Piece of Wood The requirement for this task was to dr ill a hole of any diameter through a 0.5cm piece of wood. Similar to past experiments, the participants were equally influenced by their knowledge of modern drilling technology and immedi ately set out to produce a drill-like stone tool that fit the hand well and could be appl ied in a spinning motion. Two groups began experimenting with flaking chert nodules in attempts to produce thin, drilllike stone tools representative of their prec onceived ideal tool form. The first group accidentally produced their flake and created a successful tool from it whereas the second group, which had up until now not utilized flak e technology, began to experiment with creating and using flakes. The third group uns uccessfully completed this task because their hammering method split the wood in two. Most groups began this task with a clear-cut ideal tool form in mind and quickly set about its manufacture. It was only after repeat ed unsuccessful attempts to make the tool that the focus was then shifted to utilizing the resulting flake debitage instead. While the manufacture of these ultimately successful flak es was originally intentional, their final utility was not initially recognized until later. It was again observed that unintentional

PAGE 113

101 variations and usage of manufactured pr oducts could be successfully applied to accomplish the task at hand. Summary of Series 1 At the conclusion of Series 1, the te chnological systems of all groups had progressed significantly in bot h the design and implementatio n of stone tools. Each group had developed a flake-based technology th at allowed the rudimentary manufacture and modification of tool edges. While func tionally-driven, goal-ori ented tasks were the motivation for each group, it was equally clear that the tool designs of each group also reflected the intention of the task and th e prior knowledge of the group members. Through direct observation and conversati on with participants during each task, experiential knowledge seemed to provide each group the ability to ultimately recognize and utilize unintended outcomes during inte ntional production that they otherwise may have missed earlier. It was observed that the frequency to recognize useful unintended variations increased as group members acquired more direct knowledge of stone tools. This in turn enabled them to identify and e ither implement or rej ect coincidental and unintended variations. Series 2 Series 2 consisted of identical tasks from the previous series and was designed to allow direct observation of the importance of experiential knowl edge and how it may influence the recognition and ultimate application of unintended outcomes. All groups initially expressed confidence in using stone tools to complete the required tasks. Sharp flake production was the dominant objective for cutting the leather and wood although one group still a ttempted to reproduce an unu sual triangular-shaped tool with a hook that was unintentiona lly created in a previous tas k. After considerable effort

PAGE 114

102 and numerous tool failures from breakage, this group eventually discarded the hooked stone tool idea and reverted back to simp le flake technology to successfully accomplish the cutting tasks. It was observed that the id entification of an earlier unintended variation (the triangular flake) clearl y influenced the intentional production of the technology and facilitated rapid change of th e ideal type conception for th is group even if the produced tool was eventually proven unsuccessful. This unstable and perturbative process ultimately led to the emergent development of a similar tool form that might not have been pursued had the initial c onditions been slightly different Similar albeit less original variants were also utilized in the drilling and scoring tasks as all groups simply relied upon their past experience to reproduce stone implements found acceptable in the previous series. Conclusion The primary objective of this experiment was to investigate whether technological change through unintentional va riations can be directly observed and documented. In addition, this project observed the influence of increasing experien tial knowledge in the identification and implementation of uninten tional variations w ithin a technological system. Each group consistently relied on a pre-pla nned ideal tool type to accomplish each task but the actual manufactur ed product was rarely similar in design. The reasons for this possibly relate to the low manufacturing sk ill level of the partic ipants combined with their inability to recognize a nd utilize unintended tool vari ants to better accomplish the tasks. As the knowledge base of the tec hnological producers increas ed, the ability to recognize unintended outcomes also increase d thereby altering the overall trajectory of the technological system.

PAGE 115

103 APPENDIX B MATHEMATICAL FORMULAE Formfactor 4 A / P2 Roundness 4A / L2 Elongation skeleton length / mean fiber width Aspect Ratio L / B Formfactor Divided by Roundness 2AL2 / AP2 Where: A = Area B = Breadth L = Length P = Perimeter

PAGE 116

104 LIST OF REFERENCES Allen, Peter M. 1989 Modeling Innovation and Change. In What’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), p. 258280. London: Unwin and Hyman. Allen, Peter M. 1997 Models of Creativity: Towards a New Science of History. In Time, Process, and Structured Tr ansformation in Archaeology J. McGlade and S.E. Van der Leeuw (eds), p. 35-55. London: Routledge. Andrefsky, Willam Jr. 1998 Lithics: Macroscopic Approaches to Analysis Cambridge Manuals in Archaeology Cambridge: Cambridge University Press. Bailey, G. N. 1983 Concepts of Time in Quaternary Prehistory. Annual Review of Anthropology 12:165-192. Bentley, R. Alexander 2003 An Introduction to Complex Systems. In Complex Systems and Archaeology: Empirical and Th eoretical Applications, R. Alexander Bentley and Herbert D. G. Maschner (eds), p. 9-23. Foundations of Archaeological Inquiry. Salt Lake City: University of Utah Press. Bentley, R. Alexander and Herbert D. G. Maschner 2003 Complex Systems and Archaeology : Empirical and Theoretical Applications Foundations of Archaeological Inquiry. Salt Lake City: University of Utah Press. Binford, Lewis R. 1965 Archaeological Systematics and the Study of Culture Process. American Antiquity 31:203-210. Binford, Lewis 1981 Behavioral Archaeology and the “Pompeii Premise.” Journal of Anthropological Research 7:195-208. Black, Glenn A. and Paul Weer 1936 A Proposed Terminology for Shape Classifications of Artifacts. American Antiquity 1(4):280-294.

PAGE 117

105 Boden, Margaret A. 1998 What is Creativity? In Creativity in Human Evolution and Prehistory Steven Mithen (ed), p. 22-59. New York: Routledge. Brandt, Steven A. 1988 Early Holocene Mortuary Practices and Hunter-Gatherer Adaptations in Southern Somalia. World Archaeology 20(1):40-56. Brandt, Steven A. 2000 Emergency Archaeological Fieldwork a nd Capacity Building at the Gilgel Gibe Hydroelectric Project, Deneba Southern Ethiopia. Unpublished Report to the Ethiopian Tourism Commission and Cent er for Research and Conservation of the Cultural He ritage, Addis Ababa, Ethiopia. Braudel, Fernand 1980 History and the Social Scie nces: The Longue Duree. In On History p. 2554. Chicago: University of Chicago Press. Brooks, Alison S., David M. Helgren, Jon S. Cramer, Alan Frnklin, William Hornyak, Jody M. Keating, Richard G. Klein, William Rink, Henry Schwarcz, J.N. Leith Smith, Kathlyn Stewart, Nancy Todd, Jacques Verniers, John Yellen 1995 Dating and Context of three Middle Stone Age sites with bone points in the Upper Semliki Valley, Zaire. Science 268:548-553. Clark, J. Desmond and Sonia Cole (Eds.) 1957 Third Pan-African Congress on Prehistory, Livingstone 1955 London: Chatto and Windus. Deacon, H. J. and J. Deacon 1999 Human Beginnings in South Africa Walnut Creek: Alta Mira. Dibble, Harold L. and Phillip G. Chase 1981 A New Method for Describing a nd Analyzing Artifact Shape. American Antiquity 46(1):178-187. Fabian, Johannes 1983 Time and the Emerging Other. In Time and the Other: How Anthropology Makes Its Object p. 1-35. New York: Columbia University Press. Fairtlough, Gerard 2000 The Organization of Innovative Enterprises. In Technological Innovation as an Evolutionary Process John Ziman (ed), p. 267-277. Cambridge: Cambridge University Press.

PAGE 118

106 Fitzhugh, Ben 2001 Risk and Invention in Huma n Technological Evolution. J ournal of Anthropological Archaeology 20: 125-167. Frison, George C. 1968 A Functional Analysis of Ce rtain Chipped Stone Tools. American Antiquity 33(2):149-155. Gero, Joan and Jim Mazullo 1984 Analysis of Artifact Shape Using Fourier Series in Closed Form. Journal of Field Archaeology 11(3):315-322. Goodwin, A. J. H. & Van Riet Lowe, C. 1929 The Stone Age cultures of South Africa. Annals of the South African Museum 27:1-289. Henshilwood, C.S., J.C. Sealy, R. Yates, K. Cruz -Uribe, P. Goldberg, F. E. Grine, R. G. Klein, C. Poggenpoel, K. van Niekerk, I. Watts 2001 Blombos Cave, Southern Cape, Sout h Africa: Preliminary Report on the 1992-1999 Excavation of the Middle Stone Age Levels. Journal of Archaeological Science 28:421-448. Henshilwood, Christopher S. and Curtis W. Marean 2003 The Origin of Modern Human Behavior : Critique of the Models and Their Test Implications. Current Anthropology 44(5):627-651. Hodder, Ian 1991 Interpretive Archaeo logy and Its Role. American Antiquity 56(1):7-18. Hodder, Ian 1998 Creative Thought: A Long-Term Perspective. In Creativity in Human Evolution and Prehistory Steven Mithen (ed), p. 61-77. New York: Routledge. Hodder, Ian 1999 Towards a Reflexive Methodology. In The Archaeological Process: An Introduction p. 80-104. Oxford: Blackwell. Ingles, David 1996 Landmark Locomotives: Too Big, Too Early, Too Different. Trains Magazine, January. Jelinek, A. J. 1976 Form, Function, and Style in Lithic Analysis. In Culture Change and Continuity: Essays in Honor of James Bennett Griffin, ed. C. E. Cleland, p. 19-33. New York: Academic.

PAGE 119

107 Kim, Jangsuk 2001 Elite Strategies in the Spread of Technological Innovation: The Spread of Iron in the Bronze Age Societies of Denmark and Southern Korea. Journal of Anthropological Archaeology 20: 442-478. Kinahan, John 2004 Draft Final Report on the Gilgel Gi be Archaeological Project (Reservoir Component). Unpublished Report to the Authority for Research and Conservation of the Cultural He ritage, Addis Ababa, Ethiopia. Klein, Richard 1999 The Human Career: Human Biological and Cultural Origins 2nd ed. Chicago: University of Chicago Press. Knapp, A. Bernard 1996 Archaeology Without Gravity: Postmodernism and the Past. Journal of Archaeological Theory and Method 3(2):127-158. Kuhn, Steven L. and Mary C. Stiner 1998 Middle Paleolithic ‘Creativity’ : Reflection on an Oxymoron? In Creativity in Human Evolution and Prehistory Steven Mithen (ed), p. 143-164. New York: Routledge. Kusimba, Sibel B. 2003 African Foragers: Environmen t, Technology, Interactions Walnut Creek: Alta Mira. Lemonnier, Pierre 1989 Towards an Anthropology of Technology. Man 24:526:527. Lemonnier, Pierre 1992 Elements for An Anthropology of Technology Anthropological Papers No. 88. Ann Arbor: Museum of anth ropology, University of Michigan. Lemonnier, Pierre (ed) 1993 Introduction. In Technological Choices: Trans formation in Material Cultures Since the Neolithic Pierre Lemonnier (ed), p. 1-35. New York: Routledge. Martin, Gerry 2000 Stasis in Complex Artifacts. In Technological Innovation as an Evolutionary Process John Ziman (ed), p. 90-99. Cambridge: Cambridge University Press.

PAGE 120

108 McBrearty, Sally 1988 The Sangoan-Lupemban and Middle St one Age Sequence at the Muguruk Site, Western Kenya. World Archaeology 19(3):388-420. McBrearty, Sally and Alison S. Brooks 2000 The Revolution That Wasn’t: A New Interpretation of the Origin of Modern Human Behavior. Journal of Human Evolution 39:453-563. McGlade, James 1999 Archaeology and the Evolution of Cultural Landscapes: Towards an Interdisciplinary Research Agenda. In The Archaeology of Landscape Peter Ucko and Robert Layton (eds ), p. 458-482. New York: Routledge. McGlade, James 2003 The Map is Not the Territory : Complexity, Complication, and Representation. In Complex Systems and Archaeology: Empirical and Theoretical Applications, R. Alexander Bentley and Herbert D. G. Maschner (eds), 111-119. Foundations of Archaeological Inquiry. Salt Lake City: University of Utah Press. McGlade, James and Jacqueline M. McGlade 1989 Modelling the Innovative Compone nt of Social Change. In What’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), 281-296. London: Unwin and Hyman. McGlade, James and Sander E. Van der Leeuw 1997 Introduction: Archaeology and Non-Li near Dynamics—New Approaches to Long-Term Change. In Time, Process, and Structured Transformation in Archaeology J. McGlade and S.E. Van der Leeuw (eds), 1-29. London: Routledge. Milo, Richard G. 1998 Evidence for hominid predation at Kl asies River Mouth, South Africa, and its implication for the behavior of early modern humans. Journal of Archaeological Science 25:99-133. Mitchell, Peter, Royden Yates, John E. Parkington 2002 At the Transition: The Archaeology of the Pleistocene-Holocene Boundary in Southern Africa. In Humans at the End of the Ice Age: the Archaeology of the Pleistocene-Holocene Transition ed by Lawrence Guy Strauss, Berit Valentin Eriksen, J on M. Erlandson, David R. Yesner. Plenum Press: New York. Modis, Theodore 2003 The Limits of Complexity, p. 26-32. The Futurist May-June.

PAGE 121

109 Mokyr, Joel 2000 Evolutionary Phenomena in Technological Change. In Technological Innovation as an Evolutionary Process John Ziman (ed), p. 52-65. Cambridge: Cambridge University Press. Negash, Agazi 2004 Unpublished Report on the Lithic Anal ysis at Liben Bore, Ethiopia. University of California, Berkeley. OÂ’Brian, Michael J. and R. Lee Lyman 2000 A pplying Evolutionary Archaeol ogy: A Systematic Approach New York: Kluwer. Pfaffenberger, Bryan 1992 Social Anthropology of Technology. Annual Review of Anthropology 21:491-516. Raper, Jonathan 2001 Multidimensional Geographi c Information Science New York: Taylor & Francis. Robertshaw, Peter 1993 The Development of Archaeology in East Africa. In A History of African Archaeology edited by P. T. Robertshaw, p. 320-331. New York: Heineman. Rolland, Nicholas and Harold L. Dibble 1990 A New Synthesis of Middle Paleolithic Variability. American Antiquity 55(3):480-499. Roux, Valentine 2003 A Dynamic Systems Framework for Studying Technological Change: Application to the the Emergence of the PotterÂ’s Wheel in the Southern Levant. Journal of Archaeological Method and Theory 10(1): 1-30. Rovner, Irwin 1995 Complex Measurements Made Easy: Morphometric Analysis of Artefacts Using Expert Vision Systems. In Computer Applications and Quantitative Methods in Archaeology J. Wilcock and K. Lockyear (eds), BAR International Series 598. Russ, John C. 2002 T he Image Processing Handbook Boca Raton: CRC Press.

PAGE 122

110 Schiffer, Michael Brian 2001a Toward an Anthropology of Technology. In Anthropological Perspectives on Technology Michael Brian Schiffer (e d), p. 1-13. Amerind Foundation New World Studies Series, No. 5. Albuquerque: University of New Mexico Press. 2001b The Explanation of Long-Term Technological Change. In Anthropological Perspectives on Technology Michael Brian Schiffer (ed), p. 215-232. Amerind Foundation New Wo rld Studies Series, No. 5. Albuquerque: University of New Mexico Press. Schiffer, Michael B. and James M. Skibo 1987 Theory and Experiment in th e Study of Technological Change. Current Anthropology 28(5): 595-622. Schiffer, Michael B. and James M. Skibo 1997 The Explanation of Artifact Variability. American Antiquity 62(1): 27-50. Semaw, Sileshi 2000 The World’s Oldest Stone Artef acts from Gona, Ethiopia: Their Implications for Understanding Ston e Technology and Patterns of Human Evolution Between 2.6 – 1.5 Million Years Ago. Journal of Archaeological Science 27:1197-1214. Shanks, Michael and Ian Hodder 1995 Processual, Postprocessual, and Interpretive Archaeologies. In Interpreting Archaeology Ian Hodder, Michael Shanks, Alexandra Alexandri, Victor Buchli, John Carm en, Jonathan Last, and Gavin Lucas. p. 3-29. London: Routledge. Shea, John J. 2003 Neandertals, Competition, and the Orig in of Modern Human Behavior in the Levant. Evolutionary Anthropology 12:173-187. Skibo, James M. and Michael B. Schiffer 2001 “Understanding Artifact Variabil ity and Change: A Behavioral Framework. In Anthropological Pers pectives on Technology Michael Brian Schiffer (ed), p. 139-150. Albuque rque: University of New Mexico Press. Spratt, D.A. 1989 Innovation Theory Made Plain. In What’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), p. 245257. London: Unwin and Hyman.

PAGE 123

111 “Technology” 2000 Def. American Heritage Dictionary 4th ed. “Technology” 2004 Def. Oxford English Dictionary 3rd ed. Thomas, Julian 2000 Introduction: The Polarities of Post-Processual Archaeology. In Interpretive Archaeology: A Reader p. 1-18. London: Leichester University Press. Thompson, Jessica C., John R.F. Bower, Erich C. Fisher, Audax Z. P. Mabulla, Curtis W. Marean, Kathy Stewart, and Carl F. Vondra 2004 “Loiyangalani: Behavioral and Ta phonomic Aspects of a Middle Stone Age site in the Serengeti Plain, Ta nzania.” Paper presented at the 2004 Paleoanthropology Society Me etings, Montreal, Canada. Tilley, Christopher 1991 An Hermeneutics of Meaning. In Material Culture and Text: The Art of Ambiguity p. 114-148. London: Routledge. Tilley, Christopher 1993 Introduction: Interpretation and a Poetics of the Past. In Interpretative Archaeology Christopher Tilley (ed), p. 1-27. Oxford: Berg. Torrence, Robin and Sander Van der Leeuw (eds) 1989 Introduction: What’s New About Innovation. In What’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), p. 1-15. London: Unwin and Hyman. Trigger, Bruce G. 1989 A History of Archaeological Thought Cambridge: Cambridge University Press. Van der Leeuw, Sander 1989 Risk, Perception, Innovation. In What’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), p. 300325. London: Unwin and Hyman. Van der Leeuw, Sander E. and James McGlade 1997 Structural Change and Bifurcation in Urban Evolution: A Non-Linear Dynamical Perspective. In Time, Process, and Structured Transformation in Archaeology J. McGlade and S.E. Van der Leeuw (eds), p. 331-369. London: Routledge.

PAGE 124

112 Van der Leeuw, S.E. and R. Torrence 1989 Introduction: WhatÂ’s New About Innovation. In WhatÂ’s New? A Closer Look at the Process of Innovation S.E. Van der Leeuw and R. Torrence (eds), p. 1-15. London: Unwin and Hyman. Wadley, Lyn 2001 What is Cultural Modernity? A General View and a South Africa Perspective from Rose Cottage Cave. Cambridge Archaeological Journal 11(2):201-221. Watson, Patty Jo, Steven A. LeBlanc, and Charles L. Redman 1971 The Normative View of Culture and the Systems Theory Approach. In Explanation in Archeology: An Explicitly Scientific Approach p. 61-87. New York: Columbia University Press. Wayland, E. J. 1930 Pleistocene Pluvial Periods in Uganda. The Journal of the Royal Anthropological Institute of Great Britain and Ireland. 60: 467-475. Whitley, David 1998 New Approaches to Old Problems: Ar chaeology in Search of an Ever Elusive Past. In Reader in Archaeological Th eory: Post-Processual and Cognitive Approaches David S. Whitley (ed), p. 1-30. New York: Routledge. Williams, Garnett P. 1997 Chaos Theory Tamed Washington D.C.: Joseph Henry Press. Yellen, John, Alison Brooks, Els Cornelis sen, Michael Mehlman, Kathlyn Stewart 1995 A Middle Stone Age Worked Bone Industry from Katanda, Upper Semliki Valey, Zaire. Science 268:553-556. Ziman, John 2000a Evolutionary Models for Technological Change. In Technological Innovation as an Evolutionary Process John Ziman (ed), p. 3-12. Cambridge: Cambridge University Press. 2000b Selectionism and Complexity. In Technological Innovation as an Evolutionary Process John Ziman (ed), p. 41-51. Cambridge: Cambridge University Press.

PAGE 125

113 BIOGRAPHICAL SKETCH Erich Fisher grew up in Houston, Texas. He is a graduate of Texas State University, San Marcos (formerly Southwest Texas State University), where he majored in anthropology and cartography/GIS. He is currently working in Ethiopia, Tanzania, and South Africa on various Stone Age archaeol ogical projects and 3D GIS applications within archaeology.