Thinking beyond the Tool: Archaeological computing and the interpretive process 9781407309279, 9781407339108

Citation preview

BAR S2344 2012 CHRYSANTHI, MURRIETA FLORES & PAPADOPOULOS (Eds) THINKING BEYOND THE TOOL

B A R

THINKING BEYOND TOOL Archaeological computing and the interpretive process

Edited by

Angeliki Chrysanthi Patricia Murrieta Flores Constantinos Papadopoulos Commentary by Jeremy Huggett

BAR International Series 2344 2012

Thinking beyond the Tool Archaeological computing and the interpretive process

Edited by

Angeliki Chrysanthi Patricia Murrieta Flores Constantinos Papadopoulos Commentary by

Jeremy Huggett

BAR International Series 2344 2012

ISBN 9781407309279 paperback ISBN 9781407339108 e-format DOI https://doi.org/10.30861/9781407309279 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

About the Editors* Angeliki Chrysanthi is a PhD candidate at the University of Southampton. Central to her research is the examination of methodologies for exploring visitor movement and the new potential for onsite presentation introduced by the ‘augmented space paradigm’. She has a background in Archaeology (BA), Preservation of Monuments and Sites (MSc) and has worked in several archaeological and heritage site preservation projects in Greece and, more recently, at the Akropolis Restoration Service. Her growing interest in studies such as mobile augmented reality, HumanComputer interaction and visual cognition has unfolded a series of new research directions and collaborations. She is part-time research assistant on PATINA research project (funded by RCUK Digital Economy programme) and senior member staff of D.E.P.A.S. Mycenae fieldwork project in Greece (The Archaeological Society at Athens in collaboration with the Archaeology Department of Dickinson College, Carlisle).

Patricia Murrieta Flores holds a PhD in Archaeology by the University of Southampton. She specialises in spatial technologies, GIS, High Performance Computing, and other applications of technologies to archaeological research. Her main interests include Landscape Archaeology, computational approaches to past human movement, the structuration of megalithic and symbolic landscapes of prehistoric Europe, and Zoque-Mayan Archaeology. She studied a BA in Archaeology at the National School of Anthropology and History of México and obtained an MSc in Archaeological Computing from the University of Southampton. She is field supervisor and scientific advisor of several archaeological projects in Spain and México. She has an extended record of publications and is co-director of projects in the Balearic Islands, Tarragona (Spain), and Chiapas (México). She is currently working to set in place a cross-institutional collaboration for the creation of a ‘Centre for the study of past human mobility through computational modelling’.

Constantinos Papadopoulos is a PhD candidate at the University of Southampton employing formal and informal analytical tools and computer graphics to critically evaluate the ways that modern methodological tools, and especially three-dimensional visualisations, can enhance archaeological interpretation. He has studied Archaeology and History of Art (BA) and Archaeological Computing (MSc), and he is interested in the theory of visualisation, perception and interpretation in prehistoric archaeology, the processes of recording in archaeological fieldwork, as well as the importance of light in the understanding of ancient built spaces. He has published his work in peer-reviewed journals, edited volumes, conference proceedings and a monograph and has produced archaeological documentaries which have been presented in internationally acclaimed festivals. He works as a supervising archaeologist, IT manager and 3D visualisation specialist in the Zominthos Project in Crete (The Archaeological Society at Athens), and in Koutroulou Magoula Archaeology and Archaeological Ethnography Project in Thessaly, Greece (Ephoreia of Palaeoanthropology and Speleology of Southern Greece in collaboration with the University of Southampton). * Editors are listed in alphabetical order i

 

Advisory Committee and Referees*

Vasko Démou - PhD candidate, University of Southampton, UK Graeme Earl - Senior Lecturer, Archaeological Computing Research Group, University of Southampton, UK César González-Pérez - Staff Scientist, Institute of Heritage Sciences (Incipit), Spanish National Research Council (CSIC), Spain Susanne Hakenbeck - McDonald Institute Anniversary Research Fellow, University of Cambridge, UK Jeremy Huggett - Senior Lecturer, Archaeology, University of Glasgow, UK Gary Lock - Emeritus Professor, Archaeology, University of Oxford, UK Eleftheria Paliou - Post-doctoral Researcher, Marie Curie Fellow, Radio-Past Project, University of Évora, Portugal Cesar Parcero-Oubiña - Staff Scientist, Institute of Heritage Sciences (Incipit), Spanish National Research Council (CSIC), Spain Silvia Polla - Junior Professor, The Institute of Classical Archaeology, Free University of Berlin Laia Pujol-Tost - Independent Researcher, Project Officer at the Acropolis Museum, Athens, Greece Apostolos Sarris - Research Director, Laboratory of Geophysical-Satellite Remote Sensing & Archaeo-environment, FORTH/IMS, Crete, Greece Christopher Witmore - Associate Professor, Department of Classical and Modern Languages and Literatures, Texas Tech University, USA The editors of the volume also acted as referees for the papers related to their fields of interest and expertise.

* Names are listed in alphabetical order

ii

Contributors* Tom Brughmans Archaeological Computing Research Group, University of Southampton

Jeremy Huggett Department of Archaeology, University of Glasgow

Paul Cripps Wessex Archaeology

Karen Jeneson Thermenmuseum, Heerlen, The Netherlands

Rosa Cuesta Universidad Rovira i Virgili (URV), Catalan Institute for Classical Archaeology (ICAC), Archaeological and Archaeometric Research Team, University of Barcelona (ERAUB)

Markos Katsianis Aristotle University Thessaloniki, Greece

Andrew Dufton Joukowsky Institute for Archaeology and the Ancient World, Brown University

Elaine Massung University of Bristol

Gary Lock School of Archaeology, University of Oxford

Hannah Pethen School of Archaeology, Classics and Egyptology, University of Liverpool

Corisande Fenwick Department of Anthropology, Stanford University

Eva Subias Universidad Rovira i Virgili (URV), Catalan Institute for Classical Archaeology (ICAC), Archaeological and Archaeometric Research Team, University of Barcelona (ERAUB)

Ignacio Fiz Universidad Rovira i Virgili (URV), Catalan Institute for Classical Archaeology (ICAC), Archaeological and Archaeometric Research Team, University of Barcelona (ERAUB) Tom Frankland Archaeological Computing Research Group, University of Southampton,

Philip Verhagen Research Institute for the heritage and history of the Cultural Landscape and Urban Environment (CLUE), Faculty of Arts, VU University, Amsterdam, The Netherlands

César González-Pérez Institute of Heritage Sciences (Incipit), Spanish National Research Council (CSIC), Spain

Alice E. Watterson Glasgow School of Art, Digital Design Studio Mu-Chun Wu School of Archaeology, University of Oxford

* Names are listed in alphabetical order iii



Acknowledgments This book would not have been possible without the support of colleagues and friends. The editors wish to express their sincere thanks to the participants of our session Thinking beyond the Tool at the 32nd annual meeting of the Theoretical Archaeology Group at Bristol: Tom Brughmans, Andrew Dufton, Corisande Fenwick, Tom Frankland, Eleonora Gandolfi, Karen Jeneson, James Miles, Hanna Pethen, Andrew Taylor, Ertu Unver, Philip Verhagen and Alice Watterson. As well as to several anonymous participants who contributed to the roundtable discussion. We owe our deepest gratitude to Jeremy Huggett for writing the commentary and facilitating this publication with his remarks. We would also like to thank the members of the advisory and reviewing committee for their comments and support: Vasko Démou, Graeme Earl, César González-Pérez, Susanne Hakenbeck, Jeremy Huggett, Gary Lock, Eleftheria Paliou, Cesar Parcero-Oubiña, Silvia Polla, Laia Pujol-Tost, Apostolos Sarris and Christopher Witmore. We are also grateful to Javier Pereda for designing the cover.

iv

Table of Contents Introduction: Archaeological Computing: Towards Prosthesis or Amputation?.......................... 7-13 Angeliki Chrysanthi, Patricia Murrieta Flores, Constantinos Papadopoulos 1. The Value and Application of Creative Media to the Process of Reconstruction and Interpretation ....................................................................................................................................................... 14-23 Alice Watterson 2. A CG Artist’s Impression: Depicting Virtual Reconstructions Using Non-photoreal-istic Rendering Techniques................................................................................................................... 24-39 Tom Frankland 3. Little by Little, One Travels Far ............................................................................................... 40-50 Paul Cripps 4. Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations .............. 51-71 Markos Katsianis 5. Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions .... 72-87 Cesar Gonzalez-Perez 6. The Spatial Construct of Social Relations: Human Interaction and Modelling Agency ....... 88-102 Mu-Chun Wu and Gary Lock 7. The Old and the New in Egyptian Archaeology: Towards a Methodology for Interpreting GIS Data Using Textual Evidence .................................................................................................. 103-122 Hannah Pethen 8. A Roman Puzzle. Trying to Find the Via Belgica with GIS ................................................ 123-130 Philip Verhagen and Karen Jeneson 9. Deconstructing and Reconstructing The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS ................................................................................... 131-154 Jose Ignacio Fiz Fernández, Eva Subias, Rosa Cuesta 10. Beyond the Grave: Developing new tools for Medieval Cemetery Analysis at Villamagna, Italy ................................................................................................................................................... 155-167 Andrew Dufton and Corisande Fenwick 11. Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites 168-190 Elaine Massung 12. Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology ... 191-203 Tom Brughmans Commentary: What Lies Beneath: Lifting the Lid on Archaeological Computing ................ 204-214 Jeremy Huggett 

v

Introduction Archaeological Computing: Towards Prosthesis or Amputation? Angeliki Chrysanthi, Patricia Murrieta Flores, Constantinos Papadopoulos Archaeological Computing Research Group, University of Southampton ________________________________________________________________________________ Keywords: Archaeological Computing; Interpretation; Prosthesis; Amputation.

________________________________________________________________________________ This is also true in the case of traditional archaeological research, where the usual point of departure, both in conferences and academic journals, has been the presentation of methodologies and case studies. Identifying and responding to a diachronic (and increasing) demand to support technological developments in our field with more rigorous theoretical grounding, led us to organise this session at TAG. It is worth noting that although TAG is not a typical archaeological computing venue, this was not the first time it successfully accommodated such a topic (e.g. Session , Southampton 2008, workshop on Cyber-archaeology, Stanford 2009 (Forte 2010), CASPAR Session, Bristol 2010). Furthermore, the publication of our book coincides with the celebration of the 40th year of the Annual Conference on Computer Applications and Quantitative Methods in Archaeology (CAA) where, for the first time in its history, there was a call for sessions on ‘Theoretical Approaches & Context of Archaeological Computing’.

1. Origins and scope of the Book The idea of putting together this book was inspired by the session ‘Thinking beyond the Tool: Archaeological Computing and the Interpretive Process’, which was held at the Theoretical Archaeology Group (TAG) conference in Bristol (17-19 December 2010). The session, as well as the regular format of paper presentations, included a round table discussion at the end of the session, to provide a debate forum for the participants, and encourage the development of the dialogue which emerged from the various presentations. This format not only facilitated the discussion on a better theorised approach to computer applications in archaeology, but also allowed delegates with diverse backgrounds to elaborate on common concerns from different perspectives. The overarching theme of the session, which revolved around how the various computational tools affect the ways we practice archaeology and interpret and disseminate aspects of the past, generated a series of stimulating debates, some of which we will attempt to highlight during the course of this introduction.

The scope of the book aims to build on previous approaches to archaeological method and theory, and place emphasis on the fact that archaeological computing, with its wide range of applications and methodologies, has gradually become central to most archaeological practice. Yet archaeological computing is still seen as a set of tools, which provide a methodological exactitude to overcome problems and constraints. This mentality marginalises and separates it from the rest of archaeological practices. However, the diverse range of approaches presented in the book’s 12 chapters, make it apparent that computational approaches are not mere media, but processes of data recording, management, analysis and presentation and any attempt to distinguish archaeology and archaeological computing invalidates our efforts to make archaeology richer. The notion of interpretation as a process is the foundation stone of our book, as we postulate that computer applications in archaeology are not only related to end products, they are also integral elements in the process of cognition.

Although there are several fora that focus on computer applications in cultural heritage (to name a few, the Annual Conference on Computer Applications and Quantitative Methods in Archaeology; the Conference on Cultural Heritage and New Technologies; the International Congress of Archaeology, Computer Graphics, Heritage and Innovation; Arqueologica 2.0; and the International Symposium on Virtual Reality, Archaeology and Cultural Heritage), as well as several magazines and academic journals, (such as the Archeologia e Calcolatori, Internet Archaeology, Archaeomatica, Journal on Computing and Cultural Heritage, and the International Journal of Heritage in the Digital Era), most of them are driven by an increasing technological fetishism (Huggett 2004). The general trend has been to lay emphasis on the technological aspects of these approaches, and the significance of developing technical solutions to approach cultural heritage data. Several try to keep a balance between the theoretically informed papers and the applications of technologies. 7



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process the 1980s onwards, archaeological research started widely adopting the notion of interpretation as a subjective process in archaeological practice, while trying to stay as detached as possible from the objective study of the past. Since then, however, the meaning of the word wavers between two different concepts; firstly, as an explanation and understanding, and secondly as a process that takes place throughout an archaeological research project. A Google Scholar search on ‘interpretation and archaeology’ yields about 340,000 results. These seem to be arranged around two large clusters: i) case studies examining how different approaches enhance the interpretive process, and ii) studies on the nature of archaeological interpretation and its links to archaeological theory. The former significantly outweighs the latter. It is also important, to make a clear distinction of the term, as it is used in the quintessential Ename Charter (Icomos 2007, 3), where interpretation ‘refers to the full range of potential activities intended to heighten public awareness and enhance understanding of cultural heritage sites’. In that sense, archaeological computing has always played a key role in the interpretive process by employing a range of rich media for public dissemination.

2. Interpretation in archaeological computing: are we on the same page? The notions of interpretation and interpretive process are inextricably linked with archaeological practice. Interpretation, starting ‘at the trowel’s edge’ (Hodder 1997), came to the forefront in the late 70s with the postprocessual school of thought, which emphasised the subjectivity in all stages of knowledge production, and gradually became synonymous with interpretive archaeologies (Shanks and Hodder 1995, 5). Accordingly, interpretation does not only mean a personal explanation or meaning, but also embraces agency, which contains cultural meaning, intentions and purposes (Hodder and Hutson 2003). These variables, since they are highly dependent on the social and cultural context of the present, make interpretations flexible, and therefore neither a single interpretation can be provided, nor could multiple interpreters give the same interpretation. In this context, post-processualism encouraged experimentation with multiple and alternative readings of the datasets, which may not necessarily lead to one acceptable truth (Shanks and Hodder 1995, 6-11). The gradual shift from the scientism and positivism of processual archaeology (quantification and statistics in archaeological computing) to the structuralism and human agency of post-processual thinking, influenced the way that the interpretive nature of archaeology was confronted. In both cases, it was the methodologies employed to transform data into theory, and reach a desirable outcome, that changed the nature of archaeological interpretation (Lock 2009, 76).

This book aspires to fit between these categories, as we believe that archaeological computing not only constitutes the tools that enhance archaeological interpretations, but also provides an active interpretive process, from the stage of conceptualising research questions through modelling and the end products. Apart from the case studies presented in the 12 chapters of the book, the authors have in most cases incorporated a theoretically informed discussion, in which, they either raise issues of wider significance in archaeological computing, or discuss the decision making process and the formulation of concepts that finally led to the interpretations. However, especially in the realm of archaeological computing, the criticism is that the interpretive process may entertain greedy reductionism (Dennet 1995, 82).

However, in the realm of archaeological computing, we came to realise that there is still a lack of consensus regarding the content of the term ‘interpretation’. This was not only derived from the round table discussion, but also from the content of the papers and the reviewing process. Is interpretation synonymous with theory? Does theory include the process of interpretation? When does interpretation take place in the workflow? And finally, is interpretation a process or a product? The word interpretation has been used in different contexts, with a slightly different meaning in each case, though always emphasising that interpretation involves a distinctive personal opinion and explanation. In archaeology, it has the same meaning, with the process for reaching an interpretation being emphasised, since it encompasses arguments, hypotheses and understandings. These distinctive elements flag interpretation as a process rather than a product. However, the meaning of the word interpretation is not always so profound. According to Ascher (1961), archaeologists have four main tasks: ‘First there is the formulation and refinement of concepts; second, data gathering and processing; third, the interpretation of the data, and finally, synthesis.’ Although this phrase defines interpretation as an essential step in archaeological practice, it takes place at the end, or near the end of a project. Considering the epistemological context of this saying, i.e. New Archaeology, it is reasonable that ‘objectivity’ and scientific authority precedes interpretation. However, in

3. Towards prosthesis Reductionism is used to describe the understanding of complex phenomena by simplifying their components, and examining their individual elements by incorporating them into a broader whole, a more general reality (Jones 2000, 13-16). Reductionism has been the basis of most scientific fields and is an unavoidable step (Johnson 2010, 169) in the development of processes and explanations. Reductionist approaches (cf. holism by Smuts 1926, holistic archaeology by Marcus and Flannery 1994, 55), which were central to the scientism of processual archaeology, are necessary to explain phenomena which in other ways may have been difficult to appreciate. However, quite often, these approaches fail to consider the complexity of the datasets and the sociocultural variables that existed in the past. Due to the nature of the archaeological record, there is a general consensus regarding the impossibility of evaluating all 8

Chrysanthi, Murrieta Flores, Papadopoulos: Archaeological Computing: Towards Prosthesis or Amputation? would have been impossible with the use of conventional methods. They actually function as prosthetics of our body and mind constructing strands of research, knowledge and perception.

the variables that might have existed in archaeological reasoning. Senses, perception and identity for example, or any other non-quantifiable human factors, are variables which cannot be approached by established conventional means.

3.1. Computational tools as prosthetics On the contrary, although computer applications in archaeology follow the same reductionist strands as any decision making process in traditional archaeological practice, they have often been criticised for a tendency for over-reductionism, described by Dennett (1995, 82) in the context of cognitive science as greedy reductionism. The term mainly refers to the misrepresentations and misinterpretations because of scientists’ eagerness to explain the unreachable. In the case of archaeology, this may be synonymous to bad archaeologists and practices. However, in the context of archaeological computing, this becomes more complicated. For instance, computational methods have lately provided the means to simulate and analyse illumination in ancient structures (Papadopoulos & Earl in press). By highlighting the paucity of means provided by traditional archaeology, we have demonstrated an eagerness to exploit these tools to their full extend. In this process, however, it is not only human agency that is responsible for bad archaeology; we often underestimate the complexities of both the archaeological record and the tools we use, resulting in unwarranted conceptual and methodological leaps. It should also be noted that these advanced computer-based methodological approaches - as with most scientific and theoretical applications in archaeology - were developed and firstly applied within other disciplines. As a result, many variables, crucial for understanding, interpreting and presenting the past, are deficient or completely missing, as they cannot be modelled and/or simulated with the tools we use. For example, complex social phenomena cannot be dealt with, since at the core of computing processes quantification is a prerequisite for any analysis. In several cases this may also mean that since we rarely have adequate knowledge to repurpose the tools used, we are often forced to revisit our research questions and interrupt our research practice or reasoning.

The term prosthesis, from the Greek word ʌȡȩıșİıȚȢ, meaning addition, is used in the discourses of anthropology, media theory and computing, to refer to any artefact or device which functions as an additional part of the human body; as an instrumental extension of human performance. In essence, a tool is an artefact that functions as a mediator between us, and the environment in which we operate (Ingold 1993). Used as such, artefacts have diachronically played a key role in human innovations, from hominid stone-tools to current mobile and wearable technologies. The notion of prosthesis is also interrelated to the phenomenon of ‘technological embodiment’, the situation where machines and/or technologies assume the organic functions of the body (Balsamo 1995). The concept of cyborg, a hybrid humanmachine entity, is useful here, as it is being used in cyborg theory to demonstrate that boundaries between human and machine, among other traditional binary oppositions, are blurred in our contemporary and technologically equipped society (Haraway 1991). McLuhan (1994, 7) in his seminal book Understanding Media: The Extensions of Man, stresses the physicality of media that function as ‘any extension of ourselves’, and pinpoints that electric technology could be seen as an extension of our central nervous system. The notion of prosthetic technologies as extensions assumes a synergy between the physical and mental aspects of this humanmachine interaction. To fully comprehend how computer aided methodologies act as an extension of our research processes, we need to look beyond the tool itself, and shift our attention to the notion of the ‘awareness of the task’. This is better understood in Heidegger’s classic example of operating a hammer, where after repeated and skilful handling of the tool, awareness is gradually shifting from the tactile sense of holding the wooden shaft, to the corporeal motion of hammering, until awareness is focused on the task itself (Heidegger 1962, 70/99). Taking this example in the archaeological scenario of fieldwork, we may recall holding a trowel and experiencing a similar transition of awareness, while excavating archaeological layers. A synchronous task would be to keep mental notes about the texture and composition of the soil and the finds themselves, before recording our observations in the diary and taking levels with a total station. These tools form part of an ‘equipmental totality’, which functions ‘in a network of pragmatic relations assigning tools to contexts, to tasks, to goals, and to the ultimate underlying point of what we are doing’ (Carman 2003, 851). Computational tools and media, whether or not they were built for archaeology, are at our disposal to use within a network of other methodological approaches, all participating in a higher level of inquisition and synthesis, hence their increasing presence in

The spur of multidisciplinary collaborations in archaeological research and the development of purpose built computing tools were motivated by realising their limitations. In this context, reflexion and self-criticism were also encouraged. Realising that reductionism is ingrained in archaeological computing, we may have found its Achilles’ heel, leading both these practices and their practitioners to the fringes of archaeology. Considering archaeological computing as a set of tools (as a matter of meaning and not terminology) in an attempt to defend our academic and scientific integrity, we abstract by definition their essential role, reducing them to an auxiliary asset. However, the purpose of this book is to look through and think beyond any obstacles posed by the media used. The emphasis is laid upon the capabilities of these methods to work not only as physical, but also as mental, extensions of our work. As such, they boost an everlasting process of experimentation, testing and discussion in ways that 9



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process interpretive devices, as they distract visitor’s gaze from the actual exhibits. By valorising hearing, they diminish the rest of the perceptual mechanisms and eventually, they interrupt the exploratory nature of the visit.

interpretation. In Heidegger’s words ‘Dealings with equipment subordinate themselves to the manifold assignments of the ‘in order to’ (Heidegger 1962, 69/98). The same dominance of ‘task’ and, ‘in order to’, rather than tool awareness occurs when excavating with a trowel, writing up our reasoning with a pen or a keyboard, and interacting with a dataset on particular software.

Ocularcentrism: an amputation of archaeology? The concept of amputation, as a corollary of the prosthetic nature of computational tools, could be better understood when linked to the way that visualisations have formed archaeology (and archaeological computing) and transformed the way of comprehending the past. Computer applications, in the same way as archaeology (Moser 2001, Smiles and Moser 2005, Thomas 2008), have been long haunted by fierce ocularcentrism. The production of visual material during the processing, analysis and presentation of data is a necessary step in order to evaluate and follow the interpretive process developed via and within the methodological means we use. The main problem - if one can sense a problem in that - is that the tools employed in our work analyse and present the data mainly by using different forms of visualisation. Firstly, we have to transform real world data into information that can be articulated by the tools. Therefore, the qualities of the real world are flattened in to two-dimensional senseless visuals (Tufte 1990) by a wide range of methods traditionally used in archaeology. Our input is largely based on graphical interfaces, and consequently vision dominates this stage as well. Arguably, tactility plays a significant role, as the visual interface is manipulated by the use of devices (e.g. mouse, keyboard, digitising tablet etc.), which require interaction by touch. However, our familiarity with these devices has moved this process to the ‘awareness of the task’, moving beyond the conscious interaction with them. Still, this subconscious physical interaction influences (consciously or not) our thinking about the task (cf. the adoption of IPads in Pompeii Archaeological Research Project). However, vision remains dominant. The output as well, whether it is a photorealistic rendering, an animation, a least cost path analysis in GIS, an Entity-Relation model in a database or an XML Schema, are also visual, projected on screens or printed on paper. Although this ocularcentrism may sound reasonable, since Western culture privileges vision over a multisensory perception and understanding of the world (Classen 1993; Jay 1996; MacGregor 1999), there is the underlying danger of the neglect of fundamental elements, formed by different sensoria, that mould our knowledge about the past. Consequently, have we as researchers come to terms with a form of self-amputation in the process of expanding and amplifying our understandings of the past?

However, Heidegger refers also to the term ‘readiness-tohand’ (Zuhandenheit), arguing that it requires familiarity with the object in use in order to sustain awareness to the task or the ultimate goal for which it is being used. The example of the hammer perhaps led many scholars to mistake the Heideggerian ‘readiness-to-hand’ with the Gibsonian ‘affordance’, where an object affords the agent to perform a certain action (Gibson 1977). In the hammer example, the tool affords grasping and hammering, and this does not require any previous knowledge of using the tool. Nevertheless, the inexperienced user cannot use it as the skilful carpenter does, and thus, it is more likely that he cannot transit to the stage of task awareness. Perhaps, it is more appropriate to parallel Heidegger’s term in the example of a brass musician, who during the warm up stage, is aware of his face contractions or the sense of the metal mouthpiece on his lips, whereas during performance - an immersed stage of being - the instrument becomes one with his body and mind. Appropriating Heidegger’s phenomenological observations, we could argue that a prerequisite to successfully manipulate computational tools and immerse ourselves in the interpretive process is to have attained a certain level of mastery. This notion applies to all stages of scholarship and can hardly constitute a criticism to archaeological computing. In the last decade computing analyses, for instance through GIS, are carried out by archaeologists with greater awareness of the actual mathematical structures and assumptions implemented behind each software. Nevertheless, the influence and impact they have in the interpretive process are rarely discussed or included in publications and reports. It could be said, then, that we should be able (and keen) to discuss how is that computing approaches shape archaeological interpretation, and how these extensions influence the reconstitution of our discipline. 3.2. The implications of prosthesis Prosthesis, as a situation where technological tools become human extensions, is interwoven with the concept of amputation. Each time the human body and mind uses an extension in order to amplify a certain sense or activity, a synchronous numbness of a different area of our perception occurs (McLuhan 1994, 46). This is widely known in physiology as ‘autoamputation’. To exemplify this we will use a well-known case from heritage studies; visitors who have used audio guides during their visit to a museum or a heritage site reported to have experienced irritation and limited perception of the visited environment, a situation which also entails safety issues as Massung mentions in her paper (chapter 11). This is also a typical criticism of the use of such

Attempting to overcome the problem The realisation of these limitations in archaeology have led to phenomenological, experiental (Tilley 1994, 2004, 2008; Ingold 2000, 2007), corporeal (Hamilakis, Pluciennik and Tarlow 2002) and kinaesthetic/multisensory approaches (Cummings 2002; Rainbird 2002; Hamilakis 2002, 2011; Skeates 2008, 10

Chrysanthi, Murrieta Flores, Papadopoulos: Archaeological Computing: Towards Prosthesis or Amputation? archaeological practice and the computational methods applied to archaeological datasets. However, this attempt encountered scepticism, as well as conceptual and practical issues. By considering senses in the context of archaeology, it was thought that a major drawback of archaeological reasoning and interpretation could be overcome; however, the framework of this attempt, mainly derived from the westernised perception of sensory experience, was still problematic. Our research and explanations are dominated by a problematic concept, according to which our five senses are autonomous agents which form our perception of a given environment. In other words, each sense is independently contributing to our experience. Sensory scientists challenge the notion that only our five senses form the experience of the physical world. They argue that intermediate or additional senses, that work subconsciously, as well as the evocation of memories, transform our unilateral experience into a synaesthetic journey. By way of analogy, we could argue that senses are (extensions of) ourselves, the same way that computational tools are becoming prosthetics of ourselves. The dubious notion of five distinctive senses still stands between us and our interpretations, in the same way that the tools used in our research are still considered distinct entities, existing in a toolbox, rather than on our desk or in our mind, as physical and mental extensions of ourselves respectively. However, there is also a great difference; as archaeologists, we have not reached the point where we will consider tools synonymous to our mental and physical existence, the same way that we think (or do not think) about our senses. Senses are immanent and are developed since the time of our birth. In contrast, the tools, and consequently the knowledge and skills needed to manipulate and master them, are acquired. This brings in mind Freud’s words when he talks about science and technology in his Das Unbehagen in der Kultur (1961, 38-39):

2010). In addition, a series of developments in technologies, programming languages and devices indicate a premise for a more holistic experience of the past. Such approaches led to the design of computergenerated environments, which could simulate human presence and interaction in the real or in an imaginary world, involving hearing and often touch, and creating a sense of immersiveness (Reilly 1991, Gillings 1999). Virtual Reality technologies were mainly facilitating the public aspect of cultural heritage, while archaeological research also tried to benefit from this trend (see for example Gaitatzes et al. 2000, Goodrick and Gillings 2000, Ch’ng et al. 2005). Lately, the attempts to emulate immersive environments have moved to the development of devices which can simulate high-fidelity multisensory environments. A characteristic example is the virtual reality helmet called Virtual Cocoon (Chalmers et al. 2009, Chalmers and Zányi 2009), which is currently under development by leading universities in the UK, and promises a ‘real virtuality’ experience. Further developments in the field, in the ‘continuum of virtuality’ (Milgram et al. 1994) but on the opposite side of virtual reality, such as augmented reality applications (Noh et al. 2009), attempt to create or disseminate interpretation in mixed reality environments, where presumably more of our senses will be able to interact with the information. In fact, recent trends in human-computer interaction and ambient intelligence are pointing towards more subtle ways of linking technologies to human performance. In these attempts, although technologies are physically present, maintain a discreet role -and in the ultimate scenario are even transparent- in assisting human agency. Finally, the advent of Web 2.0 enabled on-line interactivity, information sharing and the creation of multi-participatory platforms (e.g. Second Life) and contributed to the formation of a cyber-archaeology (Forte 2010). All these attempts possess great potential towards promoting a synergy of human-machine interaction, as it was envisaged by futurists from the 80s and onwards. However, crucial practical, methodological, and theoretical issues still pose certain limitations to its realisation; and this is not only because human senses cannot yet be sufficiently simulated in computergenerated environments.

‘Man has, as it were, become a kind of prosthetic God. When he puts on all his auxiliary organs he is truly magnificent; but those organs have not grown on to him and they still give him trouble at times…Future ages will bring with them new and probably unimaginably great advances in this field of civilization and will increase man’s likeness to God still more’

4. Putting the pieces together So what is the relationship among theories of senses, ocularcentrism, amputation, prosthesis and archaeological computing? We will demonstrate this by providing an overview and raising an analogy. The discussion on senses came to the forefront after the realisation that interpretations based only on visual data and visual mechanisms for capturing, processing and analyses have certain limitations. Arguably, in certain cases visuals can activate a series of perceptive mechanisms (for example cognitive scientists argue that static visuals can evoke the presence of movement - Gibbs 2005, 55). Nevertheless, as vision remains dominant, we are subconsciously led to a self-amputation. In this context, theoretical discussions and applications related to the multisensory perception of environments were introduced, revolutionising

5. Conclusion This paper, in an attempt to introduce the underlying concept of the book, presents only some preliminary thoughts on the way archaeological computing is confronted. Touching upon the ideas of prosthesis and amputation, we contributed to the discussions regarding computational approaches, not as media, but as thinking processes. We will continue acquiring knowledge and skills to dominate the tools and to come to terms with them as extensions of ourselves. Theorising and problematising our interaction with them is a crucial step in the process of learning and acquiring skills. Bringing new concepts and ideas from within archaeology and 11



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process other disciplines, communicating the processes to different audiences, and adopting diverse approaches, are key factors towards ‘thinking beyond the tool’.

Forte, M. (ed.) 2010. Cyber-archaeology. British Archaeological Reports International Series 2177. Oxford, BAR Publishing.

Acknowledgements

Freud, S. 1961. Civilization and its discontents [Translated and edited by Strachey, J.] (first edition 1930). New York, W. W. Norton & Company Inc.

The concept of prosthesis in the context of archaeological fieldwork was mentioned by Graeme Earl in a meeting of the Personal Architectonics Through INteractions with Artefacts (PATINA) research project. The idea was related to the use of tools, as archaeologists’ physical extensions. We would like to thank George Thomopoulos for our discussion on his self-reflective observations on the matter of prosthesis and musical performance. We are grateful to Vasko Démou, Jeremy Huggett and Emilie Sibbesson for their comments and suggestions in earlier versions of this paper.

Gaitatzes, A., Christopoulos, D., Voulgari, A., Roussou, M. 2000. Hellenic cultural heritage through immersive virtual archaeology, in Proceedings of the 6th international conference on virtual systems & multimedia (VSMM 2000). 57-64. Gifu, Japan. Gibbs, R. W. Jr. 2005. Embodiment and cognitive science. New York, Cambridge University Press. Gibson, J. J. 1977. The theory of affordances, in R. Shaw, R. and Bransford, J. (eds). Perceiving, acting, and knowing: toward an ecological psychology. 67-82. Hillsdale, NJ, Lawrence Erlbaum.

References Ascher, R. 1961. Analogy in archaeological interpretation. Southwestern Journal of Anthropology 17, 317-325.

Gillings, M. 1999. Engaging place: a framework for the integration and realisation of virtual-reality approaches in archaeology, in Dingwall, L., Exon, S., Gaffney, V., Laflin, S., VanLeusen, M. (eds) Archaeology in the age of the internet. Proceedings of computer applications and quantitative methods in archaeology conference (CAA 1997). Oxford, British Archaeological Reports International Series 750.

Balsamo, A. 1995. Forms of technological embodiment: reading the body in contemporary culture. Body & Society 1 (3-4), 215-237. Carman, T. 2003. Husserl and Heidegger, in Bunnin, N. and Tsui-James, E. P. (eds) The Blackwell companion to philosophy 2nd edition. Oxford, Blackwell Publishers Ltd.

Goodrick, G. and Gillings, M. 2000 Construct, simulations and hyperreal worlds: the role of virtual reality (VR) in archaeological research, in Lock, G and Brown, K. (eds) On the theory and practice of archaeological computing. Oxford University Committee for Archaeology monograph 51. Oxford, Oxbow Books.

Ch’ng, E., Stone, R. J., Arvanitis, T. N. 2005. A virtual reality archaeological framework for the investigation and interpretation of ancient landscapes, in Proceedings of the 9th IASTED international conference - Internet and multimedia systems and applications. 527-532. Grindelwald, Switzerland.

Hamilakis, Y. 2011. Archaeologies of the senses, in Insoll, T. (ed.) The Oxford handbook of the archaeology of ritual and religion. 208-225. Oxford, Oxford University Press.

Chalmers, A. and Zányi, E. 2009. Real virtuality: emerging technology for virtually recreating reality. Becta Leading Next generation learning.

Hamilakis, Y., Pluciennik, M. and Tarlow, S. (eds) 2002. Thinking through the body: archaeologies of corporeality. New York, Kluwer/Plenum.

Chalmers, A., Howard, D. and Moir, C. 2009. Real virtuality: a step change from virtual reality, in Spencer, S. N. (ed.) Proceedings of the 2009 spring conference on computer graphics (SCCG 2009). 9-16. New York, USA, Association for Computing Machinery.

Haraway, D. J. 1991. Simians, cyborgs and women: The reinvention of nature. New York, Routledge.

Classen, C. 1993. Worlds of sense: exploring the senses in history and across cultures. London, Routledge.

Heidegger, M. 1962. Being and time (Transl. by Macquarrie, J. and Robinson, E). New York, Harper and Row.

Cummings, V. 2002. Experiencing texture and transformation in the British Neolithic, Oxford Journal of Archaeology 21 (3), 249–61.

Hodder, I. and Hutson, S. 2003. Reading the past: current approaches to interpretation in archaeology (third edition). Cambridge, Cambridge University Press.

Dennett, D. C. 1995. Darwin’s dangerous idea: evolution and the meanings of life. New York, Touchstone.

Huggett, J. 2004. Archaeology and the new technological fetishism, Archeologia e Calcolatori 15. 81-92 .

12

Chrysanthi, Murrieta Flores, Papadopoulos: Archaeological Computing: Towards Prosthesis or Amputation? Icomos 2007. The ICOMOS charter for the interpretation and presentation of cultural heritage sites. Available at: http://www.enamecharter.org/downloads/ICOMOS_Inter pretation_Charter_EN_10 -04-07.pdf.

education system design and development (Edutainment 2009). 50 -61. Berlin, Heidelberg, Springer-Verlag. Papadopoulos, C. and Earl, G. in press. Formal threedimensional computational analyses of archaeological spaces, in Paliou, E., Lieberwirth, U., Polla, S. (eds) Spatial analysis in past built spaces – workshop (Berlin, 1-2 April 2010). Topoi, Excellence Cluster.

Ingold, T. 2000. Stop, look and listen! Vision, hearing and human movement, in Ingold, T. (ed.) The perception of the environment: essays in livelihood, dwelling and skill. 243–87. London, Routledge. Ingold, T. 2007. Materials against Archaeological Dialogues 14 (1), 1–16.

Rainbird, P. 2002. Making sense of petrogryphs: the sound of rock art, in David, B. and Wilson, M. (eds), Inscribed landscapes: marking and making place. 93-103. Honolulu, University of Hawaii Press.

materiality,

Ingold, T.1993. ToolǦuse, sociality and intelligence, in Gibson, K. and Ingold, T. (eds) Tools, language and cognition in human evolution. 429-45. Cambridge, Cambridge University Press.

Rainbird, P. 2002. The past as oral history: towards an archaeology of the senses, in Hamilakis, Y., Pluciennik, M. and Tarlow, S. (eds). Thinking through the body: archaeologies of corporeality. 121-136. New York, Kluwer/Plenum.

Jay, M. 1996. Vision in context: reflections and refractions, in Brennan, T. and Jay, M. (eds) Vision in context: historical and contemporary perspectives on sight. London, Routledge.

Reilly, P. 1991. Towards a virtual archaeology, in Lockyear, K. and Rahtz, S. (eds) Proceedings of the 18th computer applications and quantitative methods in archaeology conference (CAA1990), Southampton, UK. Oxford, BAR Publishing. British Archaeological Reports International Series 565.

Johnson, M. 2010. Archaeological theory: an introduction (second edition). UK, Blackwell Publishing. Lock, G. 2009. Archaeological computing then and now: theory and practice, intentions and tensions. Archeologia e Calcolatori 20, 75-84.

Shanks, and Hodder, I. 1995. Processual, postprocessual and interpretive archaeologies, in Hodder, I. et al. (eds) Interpreting archaeology: finding meaning in the past. 3-29. UK, Routledge.

MacGregor, C. 1999. Making sense of the past in the present: a sensory analysis of carved stone balls. World Archaeology 31 (2). 258-271.

Skeates, R. 2008. Making sense of the Maltese temple period: an archaeology of sensory experience and perception. Time and Mind 1 (2), 207–38.

Marcus, J., Flannery, K. V. 1994. Ancient Zapotec ritual and religion: an application of the direct historical approach, in Renfrew, C. and Zubrow, E. B. W. (eds) The ancient mind: elements of cognitive archaeology. 55-74. Cambridge, Cambridge University Press.

Skeates, R. 2010. An archaeology of the senses: prehistoric Malta. Oxford, Oxford University Press. Smiles, S. and Moser, S. (eds) 2005. Envisioning the Past. Archaeology and the image. UK, Blackwell Publishing Ltd.

McLuhan, M. 1994. Understanding media: the extensions of man (first edition 1964). USA, The MIT Press.

Smuts, J. C. 1926. Holism and evolution. London, MacMillan and CO. Limited.

Milgram, P., Takemura, H., Utsumi, A., Kishino, F. 1994. Augmented reality: a class of displays on the realityvirtuality continuum, in Proceedings of telemanipulator and telepresence technologies. 282-292. SPIE 2351.

Thomas, J. 2008 On the ocularcentrism of archaeology, in Thomas, J. and Jorge, V. O. (eds) Archaeology and the politics of vision in a post-modern context. 1-12. UK, Cambridge Scholars Publishers.

Moser, S. 2001. Archaeological representation: the visual conventions for constructing knowledge about the past, in Hodder, I. (ed.) Archaeological theory today. 262-283. Cambridge, Polity.

Tilley, C. 1994. A phenomenology of landscape. Oxford, Berg.

Noh, Z., Sunar, M. S., Pan, Z. 2009. A review on augmented reality for virtual heritage, in Chang, M., Kuo, R., Kinshuk, Chen, G-D. and Hirose, M. (eds) Proceedings of the 4th international conference on elearning and games: learning by playing. Game-based

Tilley, C. 2004. The materiality of stone. Oxford, Berg. Tilley, C. 2008. Body and image. Walnut Creek, Left Coast Press. Tufte, E. R. 1990. Envisioning Information. Chesire, CT, USA, Graphics Press.

13



1 The Value and Application of Creative Media to the Process of Archaeological Reconstruction and Interpretation Alice E. Watterson Glasgow School of Art, Digital Design Studio

____________________________________________________________________ Abstract For many years illustration (digital or otherwise) and the process of reconstruction have been considered as the ‘final step’ or accumulation of archaeological practice rather than an integral part of the interpretation process (Sorrell 1981). There has been much of disillusionment with digital reconstruction in archaeology, and, certainly, there is still a resistance to the use of 3D modelling as part of the research process. Morgan (2009) has noted that building virtual models of archaeological sites has been perceived as a legitimate mode of representing the past but that the models produced are usually the end products of a process in which archaeologists have relatively little involvement. Riedel and Bauer (2008) express concern that such 3D models are used only to show what they term ‘pretty pictures’, with their full technical capacity reduced to that of mere presentation tools. Building upon my own experience as an archaeologist, illustrator, and digital artist, this paper will investigate the potential and application of digital media to the process of archaeological reconstruction and interpretation. In order to investigate how the act of creating an archaeological reconstruction and the consequential output function as interpretive processes, this paper will consider a case study from a digital reconstruction and animation of the prehistoric ceremonial complex at Forteviot in Perthshire, Scotland. Keywords: Digital Reconstruction; 3D Animation; 3D Modelling; Virtual Archaeology; Neolithic; Forteviot; SERF Project. ________________________________________________________________________________________________ which are most constructive to the interpretation of a given site.

1. Introduction Using a case study from the digital reconstruction of a prehistoric timber monument at Forteviot in Scotland, this paper will investigate how the act of assembling an archaeological reconstruction and the consequential visual output function as interpretive processes. Following initial data collection in the field, there are two important stages of interpretative potential to consider when producing a reconstruction. Firstly there is a process of ongoing interpretation as the artist engages with the archaeological evidence from the site during the creation of their visualisation. Subsequently interpretive involvement takes place following the consumption of this image by an audience. Here experts and non-experts alike may interpret the archaeologists’ visualisation in a variety of ways. Consequently, this paper is split into two sections; interpretation during creation and interpretation following consumption.

Initially, the following case study began with the intention of gaining the majority of analysis from the outcome of the reconstruction. Instead, as the project progressed it became apparent that the real body of analysis was conducted from the methodology during the course of the modelling itself. The 3D model became a dynamic and flexible virtual environment within which it was possible to theorise upon certain elements of the evidence and visually work through problems.

2. The Site The site is located in Perthshire, Scotland, to the south of the modern village of Forteviot (Figure 1). A late Neolithic timber palisade some 270 meters across forms one of the earliest and biggest constructions in what was to become one of the largest prehistoric ceremonial landscapes in the British Isles. From aerial photography we see that a number of smaller cropmarks have also been identified. These represent a series of later, smaller timber circles and henges (Figure 2).

Although it is true that many archaeological reconstructions upon completion offer little more than a vague sense of ‘it may have looked like this in the past’, it is the process of reconstruction itself, as well asthe questions raised and decisions made during this time,

14



Watterson: The Value and Application of Creative Media to the Process of Reconstruction and Interpretation

Figure 2 Annotated map showing the location of the SERF Excavations and the main features transcribed from the aerial photograph (Brophy and Noble 2011b, 70. Transcription reproduced with permission of the University of Glasgow). Traditional reconstructions of timber circles and enclosures convey a static impression of these monuments, depicting them as newly constructed, pristine ‘telegraph poles’ (pers. comm. K. Brophy 2009), and avoiding any deeper questions of their dynamic and evolving life-path. The site would have changed dramatically in the decades following its construction. We have evidence from excavation that the posts were not maintained but instead left to rot away in the postholes naturally, evidently with little human interference. As the timber posts disintegrated back into the soil, additions were made to the site, indicating that the area remained a hub of ritual activity during and after the decline of the palisaded enclosure. This means that the timber posts may have still been present in various stages of decay during later phases of activity on the site, as further henges and enclosures were constructed.

Figure 1 These maps show the location of Forteviot in Scotland. The monument can be found in the fields to the south of the modern village. Top: (http://www.antiquity. ac.uk/projgall/driscoll323/ reproduced with permission of the University of Glasgow); Bottom: Google Earth view. The SERF (Strathearn Environs and Royal Forteviot) project aims to investigate the long-term development of the landscape and theorise upon the reasons behind the continued use and later development of a Pictish power centre on this site. In order to do this, understanding the significance of the prehistoric complex is key (Brophy and Noble 2007, 2008, 2009 and 2011a).

The model and animation produced aimed to investigate the establishment and deterioration of the timber enclosure at Forteviot, visually mapping its lifetime within the context of the wider ceremonial complex using 3D modelling and animation. The excavations have sampled different areas and periods amongst the monument complex over the past 4 years, and visualising the whole site over time brought a lot of the work together, offering an overview of the development of this unique prehistoric landscape over a period of a few hundred years.

As Noble (2006, 53) explains, it is difficult to imagine what it would have been like to live in a wooded environment, but for the people of Neolithic Scotland life would have been encompassed by forest, This is something that is rarely considered in the literature. 

15

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process construction being about creation and can only fulfil its intended function when completed.

3. Interpretation during creation As Winterbottom and Long (2006, 1356) note in their work in the Kilmartin Glen in Western Scotland, virtual reality provides a framework within which to explore visually rich representations of past environments for site interpretation, and allows the possibilities between the extremes of different scenarios to be explored, with the potential for a radical change in our interpretation. 3.1. Monument aesthetics As with any excavation of a specific monument type, the work is approached with a certain amount of presumption over what will be found. However, as Noble and Brophy (2007, 3-4) realised during their excavations at the Forteviot enclosure, the avenue postholes revealed that the timbers were far more irregular and organic than previously thought. They suggest that, unlike the reconstructions of the Meldon Bridge site (see Burgess 1976, 177), for example, the timber posts at Forteviot may have retained some of their branches and leaves (Noble and Brophy 2007, 19) (Figure 3). Although only a fraction of this massive monument has been excavated, the excavations revealed a great deal about the construction and destruction of the timber posts and allowed speculation on the nature of the monument as a whole. In beginning to consider the monument aesthetics and possibilities for any decoration on the posts, it was important to remain sceptical, as obviously no wood carvings from the Neolithic in Scotland survive today. Bocquet and Noël (1985, 1) stress that objects produced during the Neolithic would have been technically highly sophisticated, and properties of different types of wood would have undoubtedly been well known. With this in mind, a decision was taken to visualise the possibilities for decoration of the monument (Figure 3). The motifs chosen for decoration of the timber posts were clearly thought out based upon similar art from the late Neolithic in the Strathearn region. A recurrent motif in Neolithic art in northern Scotland is spiral or concentric rings patterning, and when we look to fragments of Grooved Ware pottery from the same era we see combing techniques, triangular motifs and zigzagging.

Figure 3 Top: Some of the painting on the posts, influenced by the predominant Neolithic styles of art from the region. Bottom: A view of the palisaded enclosure from a distance, note that some of the timbers have retained their branches. We should, he argues, consider the possibility that the actual use of a monument lay in its building. In the animation, the painting and carving on the timbers are synonymous with the ‘completed’ monument. Given that no direct evidence survives to suggest that the timbers were decorated, the possibilities for their visualisation and the sequence in which they were modified are infinite. For example, it is entirely possible that if the timbers were decorated it was not a process which was seen as being synonymous with the erection of the posts. Instead it was a process which occurred following construction or which was continued throughout the life of the monument by different groups of people – perhaps the posts were re-painted during certain ‘ritual events’, or at important times of the year? Initially, only the posts close to the avenue entrance to the north had been decorated in the model. This decision was partly influenced by the impression given by the aerial photograph revealing focussed later prehistoric activity around the avenue. Furthermore, all of the excavations prior to the creation of this visualisation were

It is the prerogative of the illustrator to always maintain a level of scepticism towards their own work and to cast a critical eye over the images they produce. There is the worry that, as with the wider monument itself, too much emphasis has been placed on the idea of a ‘finished artwork’. McFadyen (2006, 93) emphasises the importance of considering the construction process in detail when investigating any monument, claiming that an approach which considers only the finished monument is short-sighted. She likens this approach to ‘seeing the end of a film and only ever talking about the end-point’, i.e. the ‘completed’ monument. Similarly, Richards (2004, 73) believes that the Neolithic view of architecture may be in reverse of our own, in that we are accustomed to

16

Watterson: The Value and Application of Creative Media to the Process of Reconstruction and Interpretation concentrated around the avenue area, giving a bias to the evidence available. However, if we consider the possibilities for the construction phases of the monument, then it becomes apparent that this peripheral impression of the ‘outer limits’ of the monument could be redundant. For example, if the construction was overseen by a central authority, but with individual input from separate groups or families (pers. comm. L. Clancy August 2010). Then it is likely that there would have been competition between the families to make their posts look the most impressive, and therefore all of the posts throughout the monument would have been given similar attention.

notion of ‘social time’ which is harder to interpret, investigating developments in the demise of a monument and interruptions to the broader pattern of change, considering the individuals’ experience of a place (Bradley 1998, 87-88). As Bradley (1998) himself notes, it is rare that the archaeologist can work with periods of less than fifty years – for example, the burning of a tree within the palisaded enclosure at Forteviot may happen in a day, but we can only date the event to the nearest century. Given this intense focus of activity at Forteviot over such a vast period of time, issues of chronological sequencing and so forth are often difficult to grasp, and as such a visualisation of the site which deals with some of these issues has been very valuable in consolidating the data we have and improving the archaeologists’ engagement with the site. In order to facilitate the investigation of a more social timescale, characters were included in the model, and sequences of animation of their activity planned out. In moving from a static model to a dynamic, populated animation, it was possible to engage with a narrower dayto-day timescale and begin to incorporate the little everyday events and routines which can define a society.

During these early stages of the reconstruction process the value of visualisation for interpretation was beginning to be realised. As the reconstruction evolved, the images produced along the way became a dialogue of the interpretive process in the sense of mapping decisions made and conclusions reached. 3.2. Modelling on a social timescale In light of the concerns raised through the initial aesthetic modelling process, it became apparent that, in order to begin to understand the sequencing and use of the palisaded enclosure, a series of static phasing models simply would not suffice. Similarly, an animation which spans the few hundred years in the life of this monument could end up being very broad-brush and distanced from any significant consideration of the lives of the people who constructed the monument, and the timescale on which they operated day to day. The storyboard for the Forteviot animation needed to find a comfortable balance between elements which were taken directly from evidence onsite and a creative narrative which would allow the animation to flow from scene to scene. Using evidence from excavations and deductions made from other similar sites, a timeline of events were put together to form a visual narrative. Visualising the site in this manner reflected the research aims of the SERF project which focuses on the region in terms of its significance as a power centre over a number of centuries, and reinforces the concept that this was not a static site.

3.3. Storyboarding the animation The animation itself was split into five different scenes which all represented a period in the life of this monument (Figure 5): A) The animation begins with a Mesolithic/Early Neolithic flint knapping scene which represents the earliest known activity on the site and implies that this locality was important prior to any monuments being built. By expanding the period of interest to include the early flint knapping activity on the site, a whole new dimension to the establishment of the importance of the locale could be explored. The scene was inspired by Cummings’ (2002) theories on the establishment of important symbolic sites and ideas in the Mesolithic and Early Neolithic which led to the construction of the first monuments. Often, we assume that the construction of monumental architecture and ‘ritualised deposits’ mark the beginnings of a spiritual connection to a place, but indeed it is often the case that the more routine day to day activities and rituals are the key to understanding the establishment of an important site.

As Laycock et al (2008) argue, investigation of the enclosure in this way is important because the life of a monument is far more dynamic than its initial intended period of use. As Brophy and Noble (2009, 26) note, in Scotland (particularly in the east), a cycle of deposition, destruction, repetition, ceremony, death and burial seems to be a recurring theme for specific places in prehistory. For example, at Yeavering in the North East of England Frodsham and O‘Brien (2005, 96-97) believe that cultural memory, tradition, and association are key to understanding the site, with each new era making reference to the last in order to acquire legitimacy.

B) This scene depicts the aftermath of a storm collapse which has opened a clearing in the forest; we see a character investigating the damage as one man raises his hands in disbelief. Here, the potential ritual significance of natural clearings within woodland was considered, justified by the super-natural power of the events which created them. We have evidence of tree throws at least in the area which has been excavated around the avenue, indicating mature tree fall. Is it possible that the ritual significance of the site at Forteviot began with a natural event which opened up the canopy of trees? As Brown (2000, 51) notes, it is difficult to assess the extent of damage possible in a prehistoric forest, but based on a

Time itself is an interesting concept and it is easy to map Braudel‘s (1969) concept of ‘geographical time’ which operates at the scale of environmental change and is closest to our archaeological concepts of chronology and monument life times. However, it is the more complex 17

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process survey of the modern forest in the Chiltern Hills, southeast England following a storm in 1987, it appears that clearances upwards of 80m are not uncommon.

and consequently this process raised more questions than it answered. There is far more to the reconstruction of this monument than simply the structure itself. In order to theorise upon the possible stages of construction and interaction with the site it was not possible to simply ignore the social structure of things. Often when monument construction is discussed in archaeology the issues of scale and labour required are fixated upon (e.g. Gibson 1998 and 2002). However, little considered are the more routine day to day tasks associated with a project of this size. For example, if such a large group of people are engaged in monument construction, are they living near the monument? If so, who tends these temporary camps (Figure 4) and provides sustenance by hunting and gathering? How do these people relax and interact with one another? Although these are only small details, they serve the purpose of highlighting that monumental construction is a multi-faceted process and involves more than simply the building of a structure.

C) The next scene is perhaps the most complex as it depicts the construction of the monument itself. The social complexity of the Neolithic society is something which was beginning to be suggested here as we enter the scene through a peripheral campsite with activity occurring in the distance. Through the eyes of a character within the scene we see organised groups of people transporting worked posts towards the monument, one character rubs paint onto a standing post with his hands, while another group push a post into place. D) Given the ambiguity regarding the ‘purpose’ of these monuments (Gibson 2002) in this scene it was important to maintain a somewhat passive stance in terms of the storyline of the animation. This scene let the model tell the story as opposed to using the characters to provide narrative through their activity. We see the monument from the reflexive perspective of a character wandering around the peripheries of the enclosure, some hazy nondescript activity is visible within.

We will never know for certain how these monuments were decorated or how they were used by the societies that built them, but in visualising some of the possibilities we place ourselves in a virtual environment where it is possible to experiment with the options available to us. Having worked with a variety of creative media, in the author’s experience, 3D modelling has always been felt to be the most productive media to work in when approaching research questions and the interpretation of an archaeological site. However, when it comes to the consumption of an image by an audience, the issues change and it can be argued that other forms of media provide greater potential for an audience’s interpretation.

E) We cut to between 150 and 200 years into the life of the palisaded enclosure, the posts are beginning to deteriorate, any painting has long since faded. Many of the posts are leaning, we can see that some have fallen over, others have rotted down, and some appear charred from deliberate burning. Although the forest is more open than it previously was, areas of the monument have become overgrown, ‘reclaimed’ by nature. New monuments are beginning to be built, and we see a character walking towards the henge and timber circle to the west of the remains of the palisade. Because of the nature of storyboarding, elements of Neolithic life, which were not immediately obvious for consideration from the evidence alone, were brought to light. Having worked previously with hand-drawn 2D illustration for reconstructions of this site, the 3D medium proved to be far more complex in comparison. For example, in the pen and ink illustrative work in the vein of Alan Sorrell’s work (James 1997, 26) it was possible to use a strategically placed tree or puff of smoke from a campfire to hide those details which were unclear in the archaeological record. In the 3D software, however, hiding uncertainty is far more problematic. The whole model was visible, and as a result, the author was forced to give more time to those areas of ambiguity. For example, construction of the monument was an important phase represented in the animation and it was during this part of the reconstructive process that the value of visualising this monument was realised. In deciding how to represent the construction process, questions began to be asked of the social structure of the prehistoric society,

Figure 4 3D Max and Vue Xstream Model, A still from the animation, walking through the campsite towards the activity during the construction phase of the Late Neolithic palisaded enclosure at Forteviot (27002600BC). 

18

Watterson: The Value and Application of Creative Media to the Process of Reconstruction and Interpretation

A

B

C

D

E Figure 5 Storyboard of stills from an animation investigating the establishment of the ceremonial complex at Forteviot (2010).

19

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process character, it does begin to place the observer into the model in some physical sense and gives a far better appreciation of the scale of the reconstructed scene. In one sense, we could talk about how the Forteviot animation is effective in giving us a sense of what this experience could have physically been like, but to what extent is this recreating the past given that this scene is a fully simulated experience which was created in order to evoke such emotions?

4. Interpretation following consumption In terms of consumption and public outreach, with the Forteviot reconstruction a major objective was to investigate the potential of a more immersive experience of the site, in order to promote greater engagement with the evidence presented. In terms of the success of creating an engaging experience with this reconstruction, a conscious effort was made throughout to place the viewer in the scene, as opposed to the vantage point of the more traditional ‘fly through’ model, where a detached observer views the virtual environment from a vantage point that no inhabitant would have ever experienced. Reconstructions such as this have the effect of falsifying the experience for the audience by creating an unnatural viewpoint, and are perhaps more suited to architecturally based reconstructions. In this instance, it was decided that a more personally inclusive approach was needed.

Ch’ng’s argument is compelling, but seems somewhat unobtainable given that it implies a level of engagement and embodiment which facilitates an understanding of the world from the perspective of an inhabitant of the past. In this sense, the best we can hope to achieve through the embodiment of a virtual character is a sense of physicality in a reconstructed environment, gaining a feeling for the overall scale of the monument and the activity on a purely aesthetic level. Ingold (2000) warns that anything more would risk a projection of our modern understanding of the world on to the Neolithic. We must always be wary of projecting our own preconceptions onto the past, but in using creative methods as a means to begin the thought process of interpretation we can create environments which are grounded in theory and hard evidence, but which do not lose the human element an audience can relate to.

Earl and Wheatley (2002, 6) discuss the theories of psychologist James J Gibson (1950, 1966, 1979) who argued for the entire ambient optical array as the source of human perception, as opposed to static retinal images. Furthermore, he refused to consider the visual (or other) senses in isolation from what he termed the sensory system, arguing that our perception of the world starts with a mobile observer with two eyes mounted on a mobile head. The method conceived was the attachment of a virtual camera to the head of a character in each scene. In this way, the bouncing motion of walking is conveyed in the animation which gives a better impression of movement through the space, thus facilitating and enhancing an experience which proved to be far more engaging than conventional ‘flyover’ approaches (Figure 6).

The process of creating an animation at first seemed straightforward, though the very concept of storyboarding an archaeological reconstruction raised issues in itself. The storyboard created issues from an analytical perspective, given that every aspect of the animation has been pre-defined by the author. In influencing the narrative of the reconstruction by using the medium of an animation, have we inadvertently weakened the representation by reducing the ability of the audience to produce their own narratives, and, as such, engage with the archaeology in greater depth in the same way as a hand drawn still image might? Given that the model was built in different stages and the animation put together from a series of various scripted scenes, from an illustrator’s perspective the reconstructive process in the digital medium felt far more accommodating and fluid throughout the methodology stages. However, from the perspective of the audience is it the case that ‘less is more’ with archaeological reconstruction, and would the facility for their own interpretations have been more encouraged from a traditional hand-drawn, still image which some would argue encourages the imagination?  The idea that different forms of media can evoke a varied level of engagement, both for the archaeological illustrator and the audience is intriguing. This statement relies heavily on what is expected from an archaeological reconstruction. For instance, if we are to agree with James (1997, 23), then the mechanics of projection are irrelevant to a reconstruction if we accept its purpose simply as an image which conveys information.

Figure 6 3D Max and Vue Xstream Model, 3D character model of a Neolithic man and a view of the monument from eye level. The animation of the Forteviot ceremonial complex allows the audience to embody one of the characters and see the reconstructed scene through their eyes much like the ‘first person’ perspective favoured in video games. Although this method is not ‘fully immersive’ as the viewer has no control over the movement of the

When considering our responsibilities of reconstruction as archaeologists, we must always consider our audience. 20



Watterson: The Value and Application of Creative Media to the Process of Reconstruction and Interpretation Hermon and Nikodem (2008, 140) note, there are remarkably few archaeological papers which present new results and revelations achieved while using 3D media as a research tool, instead their worth seems to be emphasised in terms of their communicative potential.

As Miller and Richards (1995) note, there are multiple problems which occur when differentiating between what is fact and based on evidence, next to what has been embellished for the purpose of storytelling and narrative flow. These issues become more problematic for the nonspecialist, who will often trust what is represented as the truth, not just one academics’ opinion and interpretation of the evidence. Turkle (2009, 7-8) voices concern that computer visualisations make themselves ‘easy to love and difficult to doubt’ by translating the archaeologists’ theories on the evidence into compelling virtual objects and scenes which are highly engaging for an audience. In this sense she believes that the audience is vulnerable and at risk of being so taken in by the graphics that they forget their scepticism. Smiles and Moser (2005, 1) insist that it would be naive to imply that the media we use have no subconscious influence over the way we view a reconstructed scene, and there is no doubt that an audience will perceive varying media differently. Undoubtedly representation is never innocent. Parallel to these issues of how much freedom we allow for an audience’s interpretation, ultimately there is the desire on the part of the archaeologist to educate their audience. In this sense, we need to find a balance between our archaeological theories on the use of a site and the extent to which we allow for the independence of an audiences’ imagination, providing at least a minimum basis of interpretation in order to avoid imagination leading to a reinforcement of preconceived ideas about the past.

Although this paper argues for virtual reality to be used as an interpretive scientific tool in archaeology, Tost (2008) advocates interaction as the solution. The issue here is that this idea of ‘interaction’ implies a finished model. In order to evaluate the research potential of archaeological reconstruction it is the creation process that holds the most interpretive potential. In this sense, we need to shift the focus away from this idea of a ‘finished’ model and, instead, begin to see the 3D software as a visual dialog of the interpretation of the site or artefact in question as it inevitably evolves over time, an approach advocated by Garner (2008, 16-17). In this way, it will be possible to investigate the value of this creative media for interpretation in terms of its role as a dynamic and flexible virtual environment for problem solving and the development of ideas and research questions.

6. Conclusion Archaeologists as a whole are not currently exploiting the interpretative values of creative technologies to their full potential, and reconstructions more often than not are tagged onto the end of a piece of fieldwork or research as an afterthought (James 1997, 27) with their interpretative potential overlooked. Riedel and Bauer (2008, 141) express concern that such 3D models are used only to show what they term ‘pretty pictures’, with their full technical capacity reduced to that of mere presentation tools. Archaeological excavations and surveys continue to produce ever growing datasets through both traditional recording techniques and remote sensing. As a consequence of our expanding and increasingly complex methods for collecting this data it is becoming difficult to make sense of everything as a whole. Van Dam et al (2002, 535) have advocated the use of immersive virtual reality as a potent tool which allows researchers to ‘see’ patterns, trends and anomalies in their data in a way which is far superior to conventional 3D desktop displays. This paper suggests, therefore, that digital reconstruction has the potential to provide a solution to some of these issues.

5. Discussion There has been a lot of disillusionment with digital reconstruction in archaeology in recent years, and, certainly, there is still a resistance to the use of 3D modelling as part of the research process. A number of articles have been written in the last decade critiquing digital approaches to reconstructing cultural heritage. Gillings (2005) argues that archaeologists often reconstruct simply for the sake of it, without a researchdriven question in mind. Morgan (2009, 468) elaborates, noting that building virtual models of archaeological sites has been perceived as a legitimate mode of representing the past, but that the models produced are usually the end products of a process in which archaeologists have relatively little involvement. Certainly, it can be argued that the majority of archaeological reconstructions offer little research potential on completion (Goodrick and Earl 2004), and serve more as a case-study for the technologies, presenting little interpretive archaeological use (Earl and Wheatley 2002, 5). Consequently, at present, the number of archaeological projects using virtual reality technologies for approaching specific research questions, as opposed to the presentation of those questions, is relatively small (Goodrick and Earl 2004).

Acknowledgements Work discussed in this paper completed as part of the Virtual Pasts MSc course at the University of Southampton. Particular thanks to my supervisor Graeme Earl for his advice and support throughout the writing of the dissertation on which this paper is based, to my present supervisor Jeremy Huggett for theoretical guidance and to Kenny Brophy, Gordon Noble and Dene Wright of the SERF Project for their enthusiasm and suggestions throughout the reconstruction process and for encouraging me to become an illustrator back in 2007

Tost (2008, 105) agrees that there is a void in the application of virtual reality in archaeology, wherein the technology is utilised as a purely descriptive tool, with any interactivity with the data beyond navigation within a ‘hyper-realistic’ reconstruction largely ignored. As 21



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process during my first field-school with the University of Glasgow.

Contemporary themes in archaeological computing. 515. Oxford, Oxbow.

References

Frodsham, P. and O‘Brien, C. 2005. Yeavering: people, power and place. Tempus.

Bocquet, A. and Noël, M. 1985. The Neolithic or Wood Age. Endeavour, New Series 9. Pergamon Press

Garner, S. 2008. Writing on drawing. Bristol, Intellect.

Bradley, R. 1998. The significance of monuments: on the shaping of human experience in Neolithic and Bronze Age Europe. London, Routledge.

Gibson, A. 1998. Hindwell and the Neolithic palisaded enclosures of Britain and Ireland, in Gibson, A. and Simpson, O. (eds) Prehistoric ritual and religion, 67–79. Sutton Publishing Ltd.

Braudel, F. 1969. Êcrits sur l’histoire. Paris, Flammarion.

Gibson, A. 2002 The Later Neolithic palisaded enclosures of the United Kingdom. Behind wooden walls: Neolithic palisaded enclosures in Europe. 5-23. British Archaeological Reports International Series 1013. Oxford, BAR Publishing.

Brophy, K. and Noble, G. 2007. Forteviot, Perthshire: excavations at the Entrance Avenue of the Neolithic Palisaded Enclosure. Unpublished Interim Report and Data Structure Report.

Gibson, J. J. 1950. The perception of the visual world. Westport CT, Greenwood Press.

Brophy, K. and Noble, G. 2008. Forteviot, Perthshire: excavations of a henge monument and timber circle. Unpublished Interim Report and Data Structure Report.

Gibson, J. J. 1966. The senses considered as perceptual systems. Boston MA, Houghton Mifflin.

Brophy, K. and Noble, G. 2009. Forteviot, Perthshire: excavations of a henge and cist burial. Unpublished Interim Report and Data Structure Report.

Gibson, J. J. 1979. The ecological approach to visual perception. Hillsdale NJ, Lawrence Erlbaum Associates.

Brophy, K. and Noble, G. 2011a. Ritual and remembrance at a prehistoric ceremonial complex in central Scotland: excavations at Forteviot, Perth and Kinross. Antiquity 85, 787-804.

Gillings, M. 2005. The real, the virtually real, and the hyperreal: the role of VR in archaeology, in Smiles, S. and Moser, S. (eds) Envisioning the past: archaeology and the image, 223-239. Oxford, Blackwell.

Brophy, K. and Noble, G. 2011b. Big enclosures: the Later Neolithic palisaded enclosures of Scotland in their Northwestern European context. European Journal of Archaeology 14, 60-87.

Goodrick, G. and Earl, G. 2004. A manufactured past: virtual reality in archaeology. Internet Archaeology 15, Available at: http://intarch.ac.uk/journal/issue15/earl_toc. html.

Brown, A. G. 2000. Floodplain vegetation history: clearings as potential ritual spaces? Plants in Neolithic Britain and beyond, Neolithic Studies Group Seminar Papers 5, 49-62. Oxford, Oxbow.

Hermon, S. and Nikodem, J. 2008. 3D modelling as a scientific research tool in archaeology. Layers of perception: proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA2007), CD-Rom.

Burgess, C. 1976. Meldon Bridge: a Neolithic defended promontory complex near Peebles, in Burgess, C. and Micket, R. (eds) Settlement and economy in the third and second millenia B.C. 151-180. British Archaeological Reports International Series 33. Oxford, BAR Publishing.

Ingold, T. 2000. The perception of the environment: essays on livelihood, dwelling and skill. London, Routledge.

Ch’ng, E. 2009. Experiential archaeology: is virtual time travel possible? Journal of Cultural Heritage 10, 458470.

James, S. 1997. Drawing inferences: visual reconstructions in theory and practice, in Molyneaux, B. L. (ed.) The cultural life of images: visual representation in archaeology, 22-48. London, Routledge.

Cummings, V. 2002. All cultural things: actual and conceptual monuments in the Neolithic of western Britain, in Scarre, C. (ed.) Monuments and landscape in Atlantic Europe: perception and society during the Neolithic and Early Bronze Age. 107-121. London, Routledge.

Laycock, R. G., Drinkwater, D. and Day, A. M. 2008. Exploring cultural heritage sites through space and time. Journal on Computing and Cultural Heritage, 1, (2), Article 11, Available at: http://jocch.acm.org. McFadyen, L. 2006. Material culture as architecture. Approaching prehistoric and protohistoric architectures of Europe from a dwelling perspective, Journal of Iberian Archaeology 8, 91-99.

Earl, G. and Wheatley, D. 2002. Virtual reconstruction and the interpretative process: a case-study from Avebury, in Wheatley, D., Earl, G. and Poppy, S. (eds)

22

Watterson: The Value and Application of Creative Media to the Process of Reconstruction and Interpretation Miller, P. and Richards, J. 1995. The good, the bad, and the downright misleading: archaeological adoption of computer visualisation, in Computer applications and quantitative methods in archaeology 1994 (CAA1994). 19-22. British Archaeological Reports International Series 600. Oxford, BAR Publishing.

Smiles, S. and Moser, S. 2005. Envisioning the past: archaeology and the image. Oxford, Blackwell. Sorrell, M. 1981. Alan Sorrell: reconstructing the past. London, Book Club Associates. The Strathearn environs and Royal Forteviot project Available at: http://www.antiquity.ac.uk/projgall/driscoll 323

Morgan, C. 2009. (Re)building Çatalhöyük: changing virtual reality in archaeology. Archaeologies: Journal of World Archaeological Congress, 5 (3), 468-487.

Tost, L. P. 2008. Does virtual archaeology exist?, in Layers of perception: proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA2007), 101107.

Noble, G. 2006. Tree architecture: building monuments from the forest. Approaching prehistoric and protohistoric architectures of Europe from a dwelling perspective, Journal of Iberian Archaeology 8, 53-67.

Turkle, S. 2009. Simulation and its discontents. Cambridge, MIT Press.

Richards, C. 2004. Labouring with monuments: constructing the dolmen at Carreg Samson, south-west Wales, in Cummings, V. and Fowler, C. (eds), The Neolithic of the Irish Sea: materiality and traditions of practice, 72-80. Oxford, Oxbow.

Van Dam, A., Laidlaw, D. H. And Simpson, R. M. 2002. Experiments in immersive virtual reality for scientific visualization. Computers and Graphics 26, 535-555.

Riedel, A. and Bauer, T. 2008. Pretty and useful? Three dimensional computer models as a working tool for documentation and investigation in building archaeology. Layers of perception: proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA 2007).

Winterbottom, S. J. and Long, D. 2006. From abstract digital models to rich virtual environments: landscape contexts in Kilmartin Glen, Scotland. Journal of Archaeological Science 33, 1356-1367.

23



2 A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering Techniques Tom Frankland Archaeological Computing Research Group, University of Southampton

____________________________________________________________________ Abstract Archaeologists have been creating virtual reconstructions for over thirty years. Yet, despite numerous concerns with depicting these in a ‘photorealistic’ style, virtual reconstructions depicted in alternative styles are generally a rarity. This paper attempts to evaluate the potential benefits of presenting virtual reconstructions using ‘non-photorealistic rendering’ (NPR): a relatively new discipline in computer graphics that aims to depict 3D graphics in artistic and expressive styles. The study described here examined how using non-photorealistic rendering techniques might overcome the various problems that have become associated with using photorealistic styles of depiction. The study also highlighted the potential benefits that non-photorealistic graphics could offer archaeologists creating virtual reconstructions. Recent multi-disciplinary research in computer graphics and psychology has suggested that nonphotorealistic rendering techniques can influence a viewer’s psychological response to an illustration. For example, using NPR techniques can encourage conversation, influence a viewer’s judgements, and direct their gaze. These findings were compared to the results of an online survey that assessed how viewers’ responded to archaeological reconstructions depicted in both photorealistic and non-photorealistic styles. The results suggested that choice of style clearly influences the way viewers respond to virtual reconstructions, and indicated that archaeologists should be considering the impact style has on their reconstructions, especially interpretive reconstructions where creating a strong aesthetic or a sense of engagement are not essential requirements. Keywords: Computer Graphics; Reconstruction; Illustration; Artistic Rendering; Non-photorealistic Rendering. ________________________________________________________________________________________________ confirming the reliability of the largely anecdotal evidence that supports them.

1. Introduction Archaeologists have been using computers to create reconstructions of archaeological sites and artefacts since the beginning of the 1980s (Miller and Richards 1994, 19). However, virtual reconstructions have also been subject to a variety of criticism from archaeologists. Frequently this criticism relates to the prevalent visual style of these reconstructions (Miller and Richards 1994; Eiteljorg 2000), referred to as rendering 3D graphics ‘photorealistically’.

The study described here therefore aimed to establish whether the use of photorealistic rendering could affect the way a viewer interprets an archaeological reconstruction. In order to assess this, it was necessary to provide reconstructions of alternative visual styles for comparison. This was achieved using a relatively new branch of computer graphics known as non-photorealistic rendering (Strothotte and Schlechtweg 2002, 2), which offers a variety of methods and techniques for depicting 3D graphics in artistic and expressive styles (Gooch and Gooch 2001, 2).

Photorealistic rendering aims to depict 3D graphics in a manner that is indistinguishable from a photograph (Gooch and Gooch 2001, 1; Elhelw et al. 2008, 1-2). This level of visual realism is highly desirable to the primary industries that use computer graphics software, such as the film, TV, and computer games industries, as there is a clear market for it. However, for many archaeologists this visual style is problematic. For example, many archaeologists feel that rendering 3D graphics photorealistically removes the visual cues that indicate that the reconstruction is an interpretation, increasing its authority (Eiteljorg 2000). However, it is suggested that archaeologists have been more focused on discussing possible solutions to these concerns (Bateman 2000) than

As the techniques for photorealistic rendering have become increasingly sophisticated, and the level of perceptual realism attainable has improved dramatically, interest in non-photorealistic graphics has also increased, especially with researchers who are concerned with how the content of their illustrations is communicated to the viewer (Gooch and Gooch 2001, 2). Consequently, this study also aimed to identify whether there were any significant benefits to depicting archaeological reconstructions using non-photorealistic rendering techniques. 24

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering Thomas 2009, 9). Typically photorealistic renders take a long time to generate, and are therefore only usually presented as a single illustration or animation. In contrast, Swogger suggests that the best solution to overcoming the problem of ‘monumentality’ is to adopt an approach based on ‘fluidity’ and ‘volume’ (Swogger 2000, 149)

The study took two approaches to addressing these issues. The first was to review a variety of literature from numerous disciplines, including archaeology, computer science, and psychology. This literature is described below in the background to the study. The second approach was to create a number of virtual reconstructions depicted in both photorealistic and nonphotorealistic styles, which were subsequently used to assess viewer’s responses to these styles.

Attempting to depict human figures photorealistically is also problematic, due to the issues associated with the ‘uncanny valley’. Mori, a Japanese roboticist, observed that as robots were built with increasingly human-like features, the familiarity a viewer experiences with the robot decreases (Mori 1970). A similar phenomenon has been suggested to occur with highly realistic virtual depictions of human figures. There are many reasons why archaeologists may choose to depict human figures in reconstructions; these figures can provide an easy way for a viewer to interpret scale and can demonstrate how objects or spaces in a scene may have functioned. Murgatroyd (2008) suggests that the best way for archaeologists to avoid the ‘uncanny valley’ is to ‘scale back their attempt to simulate ancient humanity, settling for an easier, more abstract approach’.

2. Background 2.1. Virtual reconstructions The first virtual reconstructions were created in the 1980s (Miller and Richards 1994, 19). They included the Roman Temple Precinct at Bath and the Roman legionary bathhouse at Caerleon in Wales (Woodwark 1991). The main purpose of these early models was to create accessible reconstructions for the public (Miller and Richards 1994, 19). Since these reconstructions, archaeologists and technical specialists have attempted to reconstruct numerous cultural heritage sites and monuments using 3D graphics.

One of the main reasons that photorealistic reconstructions have become so popular is undoubtedly their aesthetic appeal. Especially with early virtual reconstructions such as those presented by Forte and Siliotti (1997) and Novitski (1998), the desire to create an attractive illustration often outweighs the archaeological motivations for creating virtual reconstructions (Miller and Richards 1994, 19). Miller and Richards (1994, 20) also suggested that photorealism’s aesthetic appeal has resulted in a large number of virtual reconstructions based of Classical or Romanesque architecture, which provides a romantic, attractive subject for computer depiction. This is problematic because as there is generally a good understanding of Classical or Romanesque architectural principles, few archaeological insights are gained through reconstruction.

Computer graphic modelling, which uses 3D graphics as a visual language to represent a theoretical model (Barceló 2001, 221), is the predominant way by which virtual reconstructions are created by archaeologists. 3D graphics can also be animated, and many virtual reconstructions will often include an animated camera that ‘flies-through’ the three-dimensional model. Reconstructions are frequently disseminated to the public at cultural heritage attractions and via TV (Addison 2001, 350; Earl et al. 2005). There is also a growing interest in using computer game technologies for the dissemination of archaeological 3D models (Addison 2001, 350). 2.2. The problems with photorealism

Another issue is that most archaeologists are well accustomed to artistic forms of illustration (Roussou and Drettakis 2003), and they use a number of accepted conventions and ‘abstract-graphical symbols’ to both create and read these illustrations (Strothotte and Schlechtweg 2002, 26). With some authors suggesting that in the near future three-dimensional models will form the majority of visual graphics used in archaeology (Cignoni and Scopigno 2008), this tradition and understanding of the semantics of archaeological illustration could be lost if archaeologists depict these photorealistically.

As suggested in the introduction, archaeologists have numerous concerns with rendering 3D graphics photorealistically. One of the most common concerns is that photorealism increases the authority of a reconstruction; the viewer might interpret the visual realism as a definitive statement or the ‘historical truth’ (Miller and Richards 1994; Lewin et al. 1997, 639; Eiteljorg 2000; Addison 2001, 349; Roussou and Drettakis 2003; Döllner and Buchholz 2005, 1; Zubrow 2006, 24; Cripps 2006, 150; Arnold 2008, 10; Thomas 2009, 9). 3D graphics created for photorealism tend to compound this problem, as a complete and detailed 3D model is usually required (Novitski 1998, 9).

It is also worth discussing one increasingly active area of research in archaeology, which focuses on creating highfidelity, hyper-realistic reconstructions of archaeological sites. These reconstructions typically use laser scanners or photogrammetry to create extremely accurate 3D models of an archaeological site and/or attempt to authentically illuminate the environment. Supporters of this approach, such as Devlin and Chalmers (2001, 43),

Archaeologists are also concerned by the ‘monumentality’ of virtual reconstructions. ‘Monumentality’ refers to the significant influence representations can have on the way archaeologists think about the past, and is particularly problematic in regards to the influence out-dated theories and concepts can retain due to being visually depicted (Swogger 2000, 148; 25



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process are critical of the way most visualisations of cultural heritage are produced, stating that they owe more to an artist's imagination than to an interpretation based on accurate simulation. Other archaeologists, such as Dawson and Levy (2005), believe that accurate reconstructions could allow a viewer to ‘experience’ the past. Similarly, Happa et al. (2010) directly equate the accuracy of a photorealistic rendering with its scientific value: ‘for the photorealistic visualisation to be of scientific value, it is necessary to recreate illumination of cultural heritage reconstructions to be both physically and historically accurate as possible’. However, some archaeologists are critical of this approach, such as Gillings (2005, 233), who suggests that attempts to reach an ever greater level of ‘verisimilitude and technological sophistication’ are based upon the false assumption that ‘vision equates directly with perception’.

photorealistic visual effects from the advertising, media and entertainment industries. Another reason empirical evidence is required is that some archaeologists disagree over the concerns described here, such as Lewin et al. (1997, 637) who argue that in their experience people treat photorealistic illustrations with just the same degree of scepticism as impressionistic illustrations.

2.3. The London Charter

One advantage of abstraction is that it makes annotation easier. Döllner and Buchholz (2005, 1) suggest that it is particularly difficult to augment a photorealistic scene because of its visual complexity. This also means that photorealistic graphics are often unsuitable to use with mobile devices, as their complexity means that when they are scaled-down they are less comprehensible than a scaled-down non-photorealistic rendering. Finally, it has been suggested that a viewer can experience a greater sense of engagement with purposefully abstracted illustrations.

2.5. The benefits of abstraction Archaeologists are already well aware of the benefits of abstraction. An archaeological illustrator will purposely simplify and generalise the characteristics of an object, based on his or her archaeological knowledge. This creates an illustration that focuses the viewer's attention on the details deemed archaeologically significant (Steiner 2005).

The use of style in regards to reconstruction is also referred to in the London Charter, a set of guidelines for creating computer-generated visualisations of cultural heritage (Beacham et al. 2006; Denard 2009). Paragraph 2.3 of the charter states that: ‘While it is recognised that, particularly in innovative or complex activities, it may not always be possible to determine, a priori, the most appropriate method, the choice of computer-generated visualisation method (e.g. more or less photo-realistic, impressionistic or schematic; representation of hypotheses or of the available evidence; dynamic or static) or the decision to develop a new method, should be based on an evaluation of the likely success of each approach in addressing each aim.’

Currently, Roussou and Drettakis (2003) have presented one of the only in depth examinations into the potential of non-photorealistic rendering for archaeology. Their case study focused on reconstructions of the Tholos monument from Argos, for which there are multiple archaeological hypotheses. Non-photorealistic rendering was used to render an animated walkthrough of the reconstruction, using a paper grain and watercolour effect. Although the authors identified several benefits associated with presenting archaeological graphics nonphotorealistically, such as the ability to isolate areas of interest and to suggest uncertainty, they suggest that further work is required to determine when photorealism and non-photorealism are suitable presentation methods.

Beacham et al. (2006, 174) state that this paragraph was written to ensure ‘the full range of visualisation options would be considered and that no single approach (for example, 'photo-realism' or 'real-time navigation') would be considered a 'default' expectation’. However, whilst this was written to address the use of photorealism or any other style as a ‘default’ expectation, it provides no guidance as to how one might evaluate which is the most appropriate style. Considering the lack of empirical studies which suggest whether a ‘less-photorealistic’ or ‘impressionistic’ rendering would be of value to cultural heritage practitioners, it is difficult to see how one could accurately evaluate their ‘likely success’.

2.6. Empirical studies A number of studies looked specifically at the effects non-photorealistic rendering has on a viewer. Many of these studies, and especially the most recent studies conducted in psychology, suggest that non-photorealistic rendering can create implicit meaning in an illustration that could be of benefit to archaeologists.

2.4. The lack of empirical evidence As stated in the introduction, although archaeologists have raised these concerns, they remain for the most part unsupported by empirical evidence. For example, Eiteljorg (2000) bases his evidence for the increased authority of photorealistic illustrations on his own personal experience and the public’s reaction to old movies and news film. This is especially problematic as the out-dated evidence Eiteljorg uses is clearly irrelevant to a contemporary society subject to a daily barrage of

One of the first to observe the effect that differing styles of depiction could have on the viewer of an illustration was Wong (1992, 84), who suggested that rougher, sketched forms were sometimes more appropriate for graphic design than visuals which appear ‘complete’. Wong found that viewers would be tempted to comment on the appearance of an illustration if the representation they were viewing appeared ‘complete’, whereas rough 26

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering opposed to low-level processes such as those involved with perceiving the realism of an image (Duke et al. 2003, 367). Halper et al. (2003) were similarly interested in the relationship between style and affective responses. One significant finding by Halper et al. was that line style could have an impact on the viewer’s social judgements, such as whether an object is considered safe or dangerous (Halper et al. 2003, 70). The authors suggested that designers could use these results to create intent-driven illustrations (Halper et al. 2003, 76).

or sketchy ‘incomplete’ drawings helped to avoid discussion about these low-level issues, and instead focused the discussion on other more important issues such as functionality. Wong also found that purposely leaving out some of the detail from an illustration encouraged viewers to consider alternatives to the design (1992, 84). Strothotte et al. (1994, 456) observed that architects wishing to depict a building at an early stage of the design process would often trace over a virtual model to produce a hand drawn illustration. The reason for this was that the architects felt these illustrations were more representative of an early stage of the design than a ‘finalised’ photorealistic illustration.

Santella and DeCarlo (2004) conducted a psychological study using eye tracking as way of determining whether photographs or non-photorealistic illustrations could generate more visual interest. Their results showed that the more detailed regions of partially abstracted images attracted a viewer’s gaze the most (Santella and DeCarlo 2004, 76). However, the authors also concluded that ‘meaningful’ abstraction was vital to achieving this (Santella and DeCarlo 2004, 77).

Schumann et al. (Schumann et al. 1996, 36) conducted the first true empirical study of non-photorealistic rendering, in an attempt to verify Strothotte et al.’s (1994) observations. The authors introduced Peeck’s classification scheme (Willows and Houghton 1987) as a way of analysing the psychological effects of an illustration. This split the responses to an illustration into three groups: x

Cognitive – the understandability, clarity or spatiality of the illustration.

x

Affective – an emotional response, such as how interesting or imaginative an illustration is.

x

Motivational – the extent to which the viewer is encouraged to engage with the design process.

Isenberg et al. (2006) conducted a recent study that examined how participants assessed hand-drawn pen-andink illustrations in comparison to computer-generated non-photorealistic renderings of the same objects. The results of the study showed that people could easily tell the difference between the two types of illustrations, but that they considered the computer-generated illustrations as accurate and useful (Isenberg et al. 2006, 123). In summary, these studies suggest that the choice of style used for an illustration can affect far more than its aesthetic appearance. Non-photorealistically rendered illustrations can guide conversation (Wong 1992), suggest incompleteness (Strothotte et al. 1994), encourage a viewer to respond cognitively or affectively (Schumann et al. 1996), influence judgements (Duke et al. 2003; Nick Halper et al. 2003)or direct a viewer’s gaze (Santella and DeCarlo 2004). Furthermore, Isenberg et al. (2006) suggest that viewers regard nonphotorealistically illustrations as accurate and useful as traditional hand drawings.

Their study surveyed 54 architects, who compared sketches generated by a computer with the typical photorealistic output. The authors found that the sketch style illustrations were better for conveying the early stages of a design and were better at creating affective and motivational responses in a viewer. In contrast, the photorealistic illustrations created a stronger cognitive response (Schumann et al. 1996, 38).

3. The case study and illustrations

The authors suggested that the sketch style illustrations required increased ‘cognitive effort’ to decipher the sketched forms. They suggest that this is why the sketch style performed poorly in the cognitive group, and that the resulting ambiguity meant they rated higher in the affective and motivational groups (Schumann et al. 1996, 40). The authors also referred to Gombrich’s (1960, 191) idea of a viewer projecting their anticipations into the areas of empty space of a picture. The authors suggest that the sketches provided more ‘projection space’ than the more ‘complete’ shaded illustrations, which is why a sketch style generates more overall discussion (Schumann et al. 1996, 40).

The second approach to the study used virtual reconstructions based on an archaeological case study to assess a viewer’s response to different styles of depiction. The archaeological site chosen for the study was crannog I at Milton Loch, near Dumfries in Scotland (as described by Piggott 1953). 3.1. The archaeological case study Crannogs are artificial or adapted natural islands, which date from the Later Bronze Age to the Early Historic period. They are often interpreted as defended settlements, occupied by farming communities who exploited the local resources on the shoreline to grow crops and keep animals. Crannogs are only found in Scotland and Ireland, with a single exception in Wales (Armit 2005).

Research by Duke et al. (2003) continued to explore the affective qualities of non-photorealistically rendered illustrations. The authors demonstrated that style could convey meaning and influence judgment in relation to the more complex higher-level cognitive processes, as 27



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process details only, detailed, and detailed with representations of human figures) and style (hand-drawn sketch, watercolour, and photorealistic).

Crannog I at Milton Loch, Kirkcudbrightshire was excavated by C.M. Piggott in 1952 (Piggott 1953). The excavation revealed the remains of a large, wooden crannog complete with wooden gangway running from the shoreline and a large number of loose stones to the rear, which were interpreted as breakwaters for a harbour (Piggott 1953, 143). The internal structure of the crannog was suggested to have been subdivided into smaller rooms, and the roof a rush thatch supported on rafters and purlins (Piggott 1953, 139).

4. The survey 4.1. Background It was decided, based on the results of several pilot studies, that the survey would ask the respondents both closed and open questions. This was because many of the responses to the illustrations in these pilot studies were very subjective and therefore did not fit a more traditional closed survey model. Because of the subjectivity and personal nature of these responses, it was also felt that the survey should reach as varied an audience as possible. For this reason, the survey was hosted online and promoted by word of mouth, email, and through social media sites.

Piggott’s interpretation of Milton Loch has had a significant influence on subsequent reconstructions of crannogs (Morrison 1985, 48). The reconstructions created for this study did not deviate largely from this interpretation. However, the internal layout of the crannog, which Piggott proposed as being square walled, has been changed to circular, an interpretation supported by the archaeological evidence and by parallels to other crannog sites that have since been excavated or recorded in Scotland.

Ultimately, a broad audience did respond, with 103 respondents completing the survey over a two-week period. The majority of these respondents were from the UK, however primarily due to the use of social media, nine of these respondents were from international locations. The use of social media sites is considered largely responsible for the good response to the survey, although a possible downside is that this meant the majority of respondents were friends, colleagues or contacts of the author. Although it is acknowledged that this may have biased the results to some degree, the benefits of being able to analyse trends in the survey on a wider scale was considered more valuable.

3.2. The illustrations The model of the crannog was created using Google SketchUp. The main reason for choosing Google SketchUp was that with the addition of third-party plugins it was capable of rendering this model in both photorealistic and non-photorealistic styles. The V-Ray plugin for Google SketchUp was used to render the photorealistic reconstructions. Some of the non-photorealistic reconstructions were also enhanced using Piranesi. Piranesi is a ‘three-dimensional painting program’, which enables a user to paint 3D models in order to create non-photorealistic illustrations (Richens 2007). Using Piranesi it is possible to simulate a variety of artistic media including watercolour painting, which was the effect chosen for this study. Although it is possible to use Piranesi to create very convincing artistic illustrations, the reconstructions produced for the study used the programs default settings in order to standardise the effect across several reconstructions.

The respondents were balanced in terms of gender and were evenly matched in terms of their experience working in the cultural heritage sector (47% worked in this field, 53% were members of the public) (Figure 1.). Age was unfortunately biased towards a younger audience (respondents aged 18-35 made up 70% of the total surveyed) which was a result of promoting the survey to staff and students at two universities. A variety of ages were however represented in the survey data and contrasts were made between them.

Nine reconstruction illustrations were created in all (Figure 2), which varied in level of detail (structural

  Figure 1 Breakdown of the respondents to the survey. 28

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering

         

Illustration 1

Illustration 2

Illustration 3

Illustration 4

Illustration 5

Illustration 6

Illustration 7

Illustration 8

       

Illustration 9 Figure 2 The illustrations created for the survey.

29

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process 4.2. Section 1

4.3. Section 2

The first section in the survey asked the respondents to rate on a scale of one to five how well a variety of adjectives described three reconstruction illustrations. The only variation between these reconstructions was the style of depiction. The adjectives chosen were similar to those used in the survey by Schumann et al. (1996). Despite being simplified for a non-specialist audience, the adjectives used in the survey were chosen to distinguish between cognitive responses (understandable, clear) and affective responses (believable, interesting, lively, imaginative, original). Motivational responses were not assessed; as the survey was aimed at a non-specialist audience it was felt they would generally view rather than create archaeological reconstructions.

The second section in the survey attempted to assess whether style affected the respondents interpretation of the underlying archaeological evidence each reconstruction was based on. This section was again based on three illustrations of differing styles. The question in the survey stated that each of these illustrations were based on a differing amount of archaeological data; however the structural features depicted in each illustration were chosen because the excavation report suggested that there was only minimal archaeological evidence to support each of the interpretations. It was accepted as inevitable that the respondents would provide biased answers to this section; for example, respondents with an archaeological background are likely to have based their response to this section on their own knowledge and experience. It was felt that there was no obvious solution to this problem, as it would be unrealistic to expect a respondent to not base their decision on their own knowledge, and this was taken into account when considering the results.

The votes for each of the adjectives were averaged, and are shown in Figure 3. This figure contrasts the responses of the cultural heritage specialists and nonspecialists. The sketched style illustration scored the lowest of the three illustrations for all the adjectives rated. The highest average scores were for ‘believable’, ‘clear’ and ‘understandable’, suggesting sketches create a strong cognitive response. The average scores for the adjectives describing the watercolour style were higher than for the sketched style; however, the photorealistic rendering scored the highest out of all the illustrations, and averaged over four for five of the seven adjectives.

The responses to the questions (Figure 4) were again split into those from the cultural heritage specialists and non-specialists. Non-specialist respondents were evenly divided between all three possible answers for the sketched style illustration. In contrast, the large majority of those with experience working in the cultural heritage sector felt there was an equal balance between archaeological data and imagination. This was also by far the most popular response in regards to watercolour style illustration, with 50% of all respondents choosing this response. Interestingly this was also the only illustration that was rated higher by the cultural heritage specialists than the non-specialists as being based on a large amount of archaeological evidence. The responses from the public to the photorealistic illustration are surprisingly almost identical to their responses to the sketched style illustration. In contrast, the cultural heritage specialists responded to this illustration with scepticism, with 50% suggesting it was based on a small amount of archaeological evidence.

These responses appear to suggest that the photorealistic style produced strong cognitive and affective responses in the viewer. It is therefore suggested that this style would be most suitable for archaeological illustrations that are created for a variety of purposes. In contrast, the ratings for the sketch style illustration have the most variation between the cognitive and affective effects on a viewer, suggesting that using a sketch style might be a good way to communicate spatial information or to make a complex illustration more understandable. One of the most surprising results from the first section was the difference between the responses to the survey conducted by Schumann et al. (1996) and to this survey. The results of Schumann et al.’s survey found that sketch style illustrations were rated as being more interesting, lively, imaginative, creative, individual (Schumann et al. 1996, 38), whereas this survey found that sketches were rated lowest in regard to all of these adjectives.

The responses to this section clearly suggest that there is a large difference of opinion between specialists and non-specialists. Above all, there is an obvious similarity between the way the public responded to the sketch and photo-real illustrations, suggesting that style may not make a large amount of difference to the authority of the illustrations. Equally, the public might be unaware of the differing levels of data and imagination that form the basis of an archaeological reconstruction.

One possible explanation for this is the difference in respondents. Architectural students completed the survey by Schumann et al.; it is therefore likely that they were familiar with the benefits of sketches and probably used sketching themselves on a frequent basis. Another possible explanation is the difference in time between the two surveys; a contemporary audience is likely to have completely different expectations of computer graphics to an audience from 1996.

The exception to this is the watercolour style illustration, which the public voted for as being balanced in respect to imagination and archaeological data.

30



Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering

Illustration 1

Illustration 2

Figure 3 The response to section 1 of the survey. 31



Illustration 3

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Illustration 4

Illustration 5

Figure 4 The response to the second section of the survey. 32

Illustration 6

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering The response to the second question (Figure 5.) differed somewhat, as the sketch style illustration received a larger amount of votes compared to the watercolour style illustration. However, again the photorealistic illustration received the largest amount of votes. Many respondents explained that they voted for this illustration because it showed more detail and colour and material information, which a number of respondents felt would be important to building a physical reconstruction. The ‘technical’ aspect of this style was significant as well; several of the respondents felt that this meant the design could have been tested prior to construction. Several of the respondents suggested that they found it difficult to choose an illustration, or that none of the illustrations were suitable, and instead suggested a scaled plan would be most appropriate.

Therefore, it could tentatively be suggested that if an archaeologist wanted to imply that a reconstruction is based equally on imagination and data, an expressive style of rendering might be more suited to this. Finally, those with cultural heritage experience were highly sceptical of the evidence the photorealistic rendering was based on. This might be because roof structures are unlikely to survive archaeologically, but may also be because cultural heritage specialists are aware of the problems associated with photorealism and are wary of its use. 4.4. Section 3 The following section of the survey was based on a final three illustrations, which not only had a high-level of detail (showing the interior of the crannog via a cutaway) but also featured representations of human figures. As such they were the most ‘complete’ illustrations and are the most akin to the type of reconstruction illustrations used for display and publication. Therefore, the questions in this section related to this subject.

4.5. Section 4 The final question in the survey asked respondents to select their favourite illustration and to explain why. The openness of this question provided respondents with an opportunity to expand on their answers to the previous questions. The response to this question is shown in Figure 6.

The first question in this section asked respondents to select which illustration they felt would be the most appropriate for a visitor centre that was being built close to the site and to explain their choice. The second question asked respondents to select which illustration would be the most appropriate for creating a scale model or reconstruction of the site and to explain their choice again.

The combined response to this question suggested that illustration 9 was the most popular with 46% of the overall vote. However, this changes significantly when contrasting the responses from cultural heritage specialists and non-specialists. Those with cultural heritage experience were almost evenly split (40% and 44%) between a preference for the watercolour style illustration and the photorealistic illustration, whereas the non-specialists clearly preferred the photorealistic illustration. This preference for the watercolour style illustration was again largely based on the belief it best represented the uncertainty of the reconstruction. There was also a smaller but significant number of votes for illustration 6, which received 10% of the entire vote.

The reason for asking these questions was to try and understand whether respondents identified the purpose of an illustration with a particular style of depiction. For example, respondents may be likely to select a visually attractive or interesting illustration for use in a visitor centre (affective response), and an illustration which presents spatial information in a clear and understandable manner for the second (cognitive response). The response to the first question (Figure 5.) indicated that the sketch style illustration was by far the least desirable for display in a visitor centre with 2% of the total vote. 29% of respondents selected the watercolour style illustration as the most suitable; the respondents who chose this option felt that this choice best presented the uncertainty in the archaeological data. The clear favourite for this purpose was the photorealistic rendering, which received 69% of the total vote.). The response to this question suggests that most respondents feel photorealism is the style most suited to public presentation. The main attraction to the photorealistic illustration for the public was its aesthetic appeal; many of the respondents referred to liking ‘realism’ or simply ‘3D’. On the other hand, one of the main reasons the cultural heritage specialists voted for the photorealistic illustration was that they felt it was more suitable for the public. Many referred to the use of visual effects in contemporary culture, and to the importance of fulfilling the public’s expectations.

These responses suggest that there is an overwhelming preference for illustrations that depict a large amount of detail (illustrations 8 and 9). These illustrations also included representations of human figures, which were popular with the respondents as a way to judge scale. Some of the respondents also felt that these illustrations (8 and 9) best represented the social aspects of the site, which in terms of archaeological dissemination is clearly important. Therefore, visual detail and representations of human figures seem to play important roles in creating an engaging illustration. Many of the respondents who chose illustration 8 as their favourite explained that their reason for doing so was that it best represented the hypothetical nature of the reconstruction. This might suggest therefore that the nonspecialist respondents who voted for illustration 9 are likely unfamiliar with the problems of using photorealism, or that presenting the illustration in a photorealistic style outweighs these concerns.

33



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Illustration 7

Illustration 8

Figure 5 The response to section 3 of the survey.

34



Illustration 9

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering

Illustration 1

Illustration 2

Illustration 3

Illustration 4

Illustration 5

Illustration 6

Illustration 7

Illustration 8

Illustration 9

Figure 6 The response to section 6 of the survey.

35



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process As illustrations 8 and 9 were the respondents’ clear favourites, the other illustrations generally received a low amount of votes. The exception to this was illustration 6, which received approximately 10% of the total vote. The main reason respondents liked this illustration was the perspective, which rather than being from overhead, was positioned at approximately eye-level. This clearly produced a more dramatic illustration and enhanced the aesthetic appeal, but for many of the respondents it was the greater understanding of scale they appreciated from having this viewpoint.

style suggests that using another style to solve the problems associated with it might be counter-productive.

The age of respondent was also compared with the response to this question (Figure 7.). This revealed that younger respondents (aged 18-35) voted almost equally for the watercolour style illustration and photorealistic illustration (39% and 42% respectively), whereas respondents aged 36+ had a clear preference for the photorealistic rendering (55%). This appears to contradict some of the beliefs of the cultural heritage specialists, who suggest that photorealism is more appropriate for a younger audience. Perhaps it is worth considering whether this argument has now come full circle, and whether younger audiences are so used to seeing photorealistic graphics that artistic presentation is now a novel medium.

Clearly non-photorealistic techniques could offer a number of benefits to archaeologists. It is apparent that non-photorealism is well suited to promoting discussion and emphasising important areas of a reconstruction through purposeful abstraction. Therefore, it is suggested that non-photorealism is more suited to depicting reconstructions intended for interpretation rather than dissemination. This may come as no surprise to the many archaeologists who have used rough sketches as a quick way to communicate their thoughts and ideas to their colleagues for years. In contrast, it is suggested that photorealism is more appropriate when the emphasis is to create an aesthetically pleasing and engaging reconstruction, primarily for the public. The response to section three of the survey also highlighted that photorealism is suited to depicting a high level of detail, which may be relevant when displaying or printing reconstructions at a large scale.

The second aim of the study was to establish whether non-photorealistic styles could offer archaeologists any benefits over photorealistic styles of depiction. As the majority of the respondents to the survey were enthusiastic about the use of photorealistic graphics to depict reconstructions, the answer to this question appears to relate directly to the intended purpose of the reconstructions.

5. Discussion In the background to the study a variety of problems were identified with photorealism. However, it was observed that the evidence for these concerns was often based on anecdotal evidence. Therefore, one of the primary purposes of the survey was to provide empirical data that could be contrasted to this anecdotal evidence.

5.1. Limitations of the study As discussed in the participants’ background section, age was unfortunately biased towards a younger audience, a result of promoting the survey to staff and students at two universities. Statistical methods, which were not applied to the data acquired in this study, could potentially have been used here; however, as the age of respondents was not considered particularly important to fulfilling the broad aims of the survey, no attempt was made to address this. Interestingly, the results of the survey highlighted the importance of considering age, for example, several of the cultural heritage professionals considered photorealism a more ‘suitable’ style for younger audiences, and this would definitely be worthy of further study.

The broad range of responses to the second section of the survey suggests that the non-specialists were generally unsure of the underlying data the reconstructions were based on. However, several of the non-specialists who voted for the watercolour illustration as their favourite in the last question explained that they had chosen this as felt this best represented the hypothetical nature of the reconstruction, indicating that the public have some awareness of the problems associated with photorealistic illustrations. In contrast, the cultural heritage specialists had a greater awareness of the problems associated with photorealism and were more likely to be sceptical of the evidence underlying the reconstructions.

Another limitation to the study was that each of the illustrations compared in section 2 depicted different views of the crannog. As previously discussed, the inherent difficulty of attempting to understand how the respondents reacted to the underlying data of each reconstruction made the use of different illustrations a necessity for the questions in this section of the survey, but this ultimately had an effect further on when respondents were asked to choose their favourite illustration. Due to the nature of the study, the amount of time and effort that could go into creating multiple illustrations was finite, and, therefore, the last question (respondents’ favourite illustration) was based on the illustrations from the previous sections, and it is

Despite the problems associated with photorealism, based on the responses to section 1, 3 and 4, the survey showed that this style of depiction was particularly popular with both the cultural heritage specialists and non-specialists. The response to these questions suggests that the concerns archaeologists have with photorealism may be outweighed by the desire to produce an aesthetically pleasing or engaging illustration, as nearly three-quarters of all the respondents would use a photorealistic illustration to display the crannog to the public in a visitor centre. Ultimately, the popularity of photorealism as a

36



Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering acquisition of quantitative data, such as through the use of eye tracking technologies.

acknowledged that this may have biased the responses to this question.

There is a clear case for using non-photorealistic rendering for archaeological illustrations, and it is felt that this style is most suited to depicting interpretive reconstructions where creating aesthetic appeal or a sense of engagement are not essential. However, further work is required to verify these benefits.

Acknowledgements I would like to thank Tim Taylor for his consistent support throughout the study, and John Hodgson and Graeme Earl for their comments and feedback. Thanks also to Amy Jeffrey for her patience and encouragement. Finally, a huge thank you to everyone, who filled out the survey, without whom, this study would not have been possible.

References Addison, A. C. 2001. Virtual heritage, in Proceedings of the 2001 conference on virtual reality, archeology, and cultural heritage - VAST ’01. 343-380. New York, USA, ACM Press

Figure 7 Respondents favourite illustration by age.

6. Conclusion

Armit, I. 2005. Celtic Scotland, London, Batsford.

The study had two main aims, to determine if archaeologists’ concerns with photorealism were justified by providing empirical evidence, and to examine the potential of using non-photorealism for archaeologists. The first aim was primarily fulfilled via the survey, which indicated that whilst some individuals were aware of the problems with photorealism, the responses to section 2 of the survey suggested that the public are generally uncertain about the inferences made in reconstructions. In contrast, the cultural heritage specialists were perhaps overly sceptical of photorealism, although it is difficult to be certain in view of the difficulties posed by asking this question.

Arnold, D. 2008. Editorial for inaugural issue of JOCCH: Pasteur s Quadrant: cultural heritage as inspiration for basic research in computer science. Journal on Computing and Cultural Heritage (JOCCH) 1 (1), 1-13. Barceló, J. A. 2001. Virtual Reality for archaeological explanation: beyond “picturesque” reconstruction. Archeologia e Calcolatori 12, 221-244. Bateman, J. 2000. Immediate realities: an anthropology of computer visualisation in archaeology. Internet Archaeology. Available at: http://intarch.ac.uk/journal/ issue8/bateman_index.html.

The second aim was to assess whether non-photorealism could meet the needs of archaeologists. Whilst this study suggests there are numerous benefits to depicting virtual reconstructions using non-photorealistic styles, it is clear that the purpose of an illustration should be the determining factor behind stylistic choice. The survey demonstrated that there was enthusiasm for photorealistic graphics, and that it would be counter-productive to present reconstructions aimed at a public audience in a different style. It was also highlighted that nonphotorealism appears to be highly beneficial to the way that illustrations are interpreted and understood, and therefore it seems that there would be benefits to using non-photorealistic graphics for interpretive graphics, with other archaeologists the primary audience.

Beacham, R., Denard, H. and Niccolucci, F. 2006. An introduction to the London charter, in Proceedings of evolution of Information Communication Technology in Cultural Heritage. Cignoni, P. and Scopigno, R. 2008. Sampled 3D models for CH applications. Journal on Computing and Cultural Heritage 1 (1), 1-23. Cripps, P. 2006. Spatial technologies in archaeology in the twenty-first century, in Greengrass, M. and Hughes, L. (eds) The virtual representation of the past. 148-156. Farnham, Ashgate.

It is hoped that the empirical data obtained through the study might provide a starting point for further research and discussion. It is suggested that the qualitative data provided via the survey would be complemented by the

Dawson, P.and Levy R. 2005. Using computer modelling and virtual reality to explore the ideological dimensions of Thule whalebone architecture in Arctic Canada. 37



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Available at: http://intarch.ac.uk/journal/issue18/dawson _index.html

international symposium on non-photorealistic animation and rendering. 115-126. Annecy, France, ACM.

Denard, H. 2009. The London charter for the computerbased visualisation of cultural heritage (version 2.1). Available at: http://www.londoncharter.org/.

Lewin, J., Ehrhardt, M. and Gross, M. D. 1997. Not just another pretty face: images and arguments in an anthropology web site, in CAAD futures 1997 Digital Proceedings. 635-654. Munich, Germany, Klewer.

Devlin, K. and Chalmers, A. 2001. Realistic visualisation of the Pompeii frescoes, in Proceedings of the 1st international conference on Computer graphics, virtual reality and visualisation. 43-48. Cape Town, South Africa, ACM.

Miller, P. and Richards, J. 1994. The good, the bad and the downright misleading: archaeological adoption of computer visualization, in Huggett, J. and Ryan, N. (eds) CAA94 Proceedings of the 22nd CAA conference, 19-22. Oxford, BAR Publishing.

Duke, D.J. et al. 2003. Rendering and affect. Computer Graphics Forum 22 (3), 359-368.

Mori, M. 1970. Bukimi no tani [the uncanny valley]. Energy 7 (4), 33–35.

Döllner, J. and Buchholz, H. 2005. Non-photorealism in 3D geovirtual environments, in Auto Carto proceedings. Las Vegas, Nevada.

Morrison, I. A. 1985. Landscape with lake dwellings: the crannogs of Scotland, Edinburgh, Edinburgh University Press.

Earl, G.P. 2005. Video killed engaging VR? Computer visualizations on the TV screen, in, Smiles, S. & Moser, S. (eds) Envisioning the past: archaeology and the image. 204-222. Oxford, Blackwell.

Murgatroyd, P. 2008. Appropriate levels of detail in 3-D visualisation: the house of the surgeon, Pompeii. Available at: http://intarch.ac.uk/journal/issue23/murga troyd_index.html

Eiteljorg, H. 2000. The compelling computer image: a double-edged sword. Internet Archaeology. Available at: http://intarch.ac.uk/journal/issue8/eiteljorg_index.html.

Novitski, B. J. 1998. Rendering real and imagined buildings, Gloucester, Rockport.

Elhelw, M. et al. 2008. A gaze-based study for investigating the perception of visual realism in simulated scenes. ACM Transactions on Applied Perception 5 (1), 1-20.

Piggott, C. 1953. Milton Loch Crannog, I. A native house of the 2nd century AD in Kirkcudbrightshire. Proceedings of the Society of Antiquaries of Scotland, 87, 134-152.

Forte, M. and Siliotti, A. 1997. Virtual archaeology: recreating ancient worlds, Harry N. Abrams.

Richens, R. 2007. Beyond photorealism. Available at: http://www.piranesi.co.uk/downloads/literature/beyond_p hotorealism_white_paper.pdf

Gillings, M. 2005. The real, the virtually real, and the hyperreal: the role of VR in archaeology, in Smiles, S. and Moser, S. (eds) Envisioning the past: archaeology and the image. 223-239, Oxford, Blackwell.

Roussou, M. and Drettakis, G. 2003. Photorealism and non-photorealism in virtual heritage representation, in Chalmers, A. Arnold, D. and Niccolucci, F. (eds) Proceedings of the first eurographics workshop on graphics and cultural heritage.

Gombrich, E. H. 1960. Art and illusion: a study in the psychology of pictorial representation. Oxford, Phaidon. Gooch, B. and Gooch, A. 2001. Non-photorealistic rendering. Natick, A. K. Peters.

Santella, A. and DeCarlo, D. 2004. Visual interest and NPR: an evaluation and manifesto, in Proceedings of the 4th international symposium on non-photorealistic animation and rendering. 71-150. Annecy, France, ACM.

Halper, Nick et al. 2003. Towards an understanding of the psychology of non-photorealistic rendering. Computational Visualistics, Media Informatics, and Virtual Communities 11, 67-78.

Schumann, J. et al. 1996. Assessing the effect of nonphotorealistic rendered images in CAD, in CHI ’96. 3541. Vancouver, British Columbia, Canada, ACM.

Happa, J. et al. 2010. Illuminating the past: state of the art. Virtual Reality 14 (3), 155-182.

Steiner, M., 2005. Approaches to archaeological illustration: a handbook. York, Council for British Archaeology.

Isenberg, T. et al. 2006. Non-photorealistic rendering in context: an observational study, in Proceedings of the 4th

38

Frankland: A CG Artist's Impression: Depicting Virtual Reconstructions Using Non-photorealistic Rendering Willows, D. M. and Houghton, H. A. 1987. The psychology of illustration, New York, Springer-Verlag.

Strothotte, T. et al. 1994. How to render frames and influence people. Computer Graphics Forum 13 (3), 455466.

Wong, Y. Y. 1992. Rough and ready prototypes: lessons from graphic design, in CHI ’92 posters and short talks of the 1992 SIGCHI conference on human factors in computing systems. 83-84. Monterey, California, ACM.

Strothotte, T. and Schlechtweg, S. 2002. Nonphotorealistic computer graphics: modeling, rendering, and animation, San Francisco, Morgan Kaufmann.

Woodwark, J. 1991. Reconstructing history with computer graphics. IEEE Computer Graphics and Applications 11 (1), 18-20.

Swogger, J. 2000. Image and interpretation: the tyranny of representation?, in Hodder, I. (ed.) Towards reflexive method in archaeology: the example at Çatalhöyük. 143152.Cambridge, McDonald Institute for Archaeological Research.

Zubrow, E. B. W. 2006. Digital archaeology: a historical context, in Evans, T. L. and Daly, P. (eds) Digital archaeology: bridging method and theory. 10-31. Abingdon, Routledge.

Thomas, G. 2009. Images of the past: illustrating and imagining archaeology. EuroREA 6/2009, 8-12.

39



3 Little by Little, One Travels Far Paul Cripps Wessex Archaeology

____________________________________________________________________ Abstract This paper will look across a range of recent developments in the specific field of archaeological spatial technologies in order to show how archaeological theory and technological practice can be successfully related; a reflexive process whereby these two strands working together can provide more meaningful interpretation(s) and, at the same time, help improve the underpinning technologies through extreme use cases, archaeological exemplars being typically at the more complex end of the spectrum. Using examples from recent and ongoing projects, including cultural resource management, archaeological investigation, and academic research projects, this paper aims to show how many small steps have indeed brought us a long way with regard to spatial technologies in archaeology. Keywords: GIS; Laser-scanning; Ontology; CIDOC-CRM; Metric Survey; Linked Data; Data Model; Archaeological Theory; Spatial Technology; API; Standards. ________________________________________________________________________________________________ positively contributing to the development and critical evaluation of archaeological theories, particularly by making the links between data and resultant theory more explicit through the use of technology.

1. Introduction It is through dogged and determined small steps that larger successes arise. Many such small steps have been taken over the last few decades in various technologies applied within archaeological contexts, and we have indeed come far as a result. One not insubstantial result is the presence of this very session with its resulting publication emerging from a Theoretical Archaeology Group conference, a session which demonstrates the intrinsic links between doing and thinking, practice and methodological and theoretical discourse.

2. Overview Do not meddle in the affairs of wizards, for they are subtle and quick to anger. (Tolkien, J. R. R. 1954. The Fellowship of the Ring) Cutting edge technology is always the preserve of a few skilled technologists, wizards to some who see what they do as some kind of magic. A key aspect of the nature of technology is the way in which it is adopted, absorbed and made available to others through time, changing from the preserve of the few to the tools of the masses: Magic quickly becomes mundane. But whilst it is new and magical, a barrier exists between those with knowledge and those without; those without are not always able to engage. It is important to meddle in the affairs of wizards, as it is important for wizards to explain their magic, since, contrary to Tolkien’s claim, wizards are not generally quick to anger although they may well be subtle.

For too long, there has been a gap and lack of engaging discourse between archaeological computing specialists and archaeological theorists. This has been well reported on and is not the thrust of this paper but rather background to it. Today’s abstractions, models, visualisations, development platforms and tools provide richer ground for discussion and critical thinking regarding the archaeological process. The use of technology, particularly spatial technologies, the focus of this paper, is arguably the norm across a range of activities. Archaeological computing specialists are in a central position here to bridge the perceived divide between the technological and the theoretical.

It is possible to argue that archaeology as a discipline has drawn on and been quick to adopt new technologies and techniques from other disciplines. For example, the very fundamentals of stratigraphy, superposition and relative chronology have their origins in the geological sciences whilst notions of classification and taxonomy draw heavily on the biological sciences. In many cases, rightly or wrongly, this assimilation can be seen as being neutral (for example survey techniques to capture spatial data or the use of spatial databases), whilst at other times, such

As such, we are in an excellent position to examine, evaluate and drive the discourse forward as the initial session abstract sought to. Looking at some select interrelated areas relating to the use of spatial technologies in archaeology, it is possible to argue that small technological steps in a range of disciplines are 40

Cripps: Little by Little, One Travels Far assimilation is problematised (for example GIS based analysis of landscapes, the use of structured data and formal language to describe data structures).

3. Geometry To begin, let us start with spatial technologies used for data capture and data management. I would include here the use of survey instruments and techniques such as photogrammetry, CAD, and GIS, but would, at this point, exclude any form of spatial analysis from this discussion; that aspect will be investigated subsequently. Applications of these instruments and techniques are often considered to be unproblematic from a theoretical perspective. However, more recent developments have brought to the fore rather fundamental theoretical issues about data capture and data management which had largely remained implicit previously.

The nature of and relationship between technology and theoretical discourse has been a recurring theme at conferences such as the Theoretical Archaeology Group and Computer Applications in Archaeology in many ways characterised by bipolarisation with limited engagement. It is not the aim of this paper to go over this ground again; instead I would refer to published works on this topic (e.g. Cripps et al. 2006). However, this background is very relevant as to why sessions such as this are vital to fostering engagement between archaeologists who necessarily approach the same or similar problems but from different angles, but are often separated by a technological and/or theoretical divide.

Selective modes of capturing data such as the use of Total Stations or GNSS, where the archaeological surveyor makes careful choices in the field about which points to measure, have become widespread in the academic, commercial, and the public sectors. Used for all manners of activities from cultural resource management to prospection to (analytical) recording, the ways in which we use such technologies to capture data have become commonplace (Figure 1).

Increasingly, the realm of spatial technologies, to which this paper refers, is becoming much broader and integrated with other related technologies currently making the transition from the realm of pure computer science to the study of archaeology. Their practitioners look for real world examples and case studies, and archaeological computing specialists look to take advantage of these new developments in exactly the same way archaeologists have always adopted new methods, approaches, and technologies from other disciplines, sometimes with great success, other times less so.

Furthermore, as the processes have become democratised and open to more archaeologists, the process of data gathering has increasingly become a more mechanised activity. This is countered only by the inherent decision making required by selective data capture processes; thus, some specialist interpretive ability must still be present.

Crucially, spatial technologies are providing meaningful thinking spaces (after Gillings and Goodrick 1996) and there is a growing range of standards and support documents to help with the design and construction of such spaces (e.g. RCHME 1995; ADS 2011; CIDOC ASWG 1995; EH 2007).

Figure 1 A traditional drawn elevation showing individual stones in the castle wall but produced directly from a point cloud. An example of traditional output produced using the latest technology.

41



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process If we are talking about buildings and structures, then this other information may include modern and/or historic photographs, documentary sources and, crucially, the richer information captured on site. Putting all this into a common 3D coordinate system typically using CAD, we have a reflexive, interpretive analytical platform capable of not only visualising but responding to and informing our interpretations. This is now a typical approach for large and complex structures for a number of archaeological units including Wessex Archaeology.

As such, whilst arguably in many cases recording is simply about capturing data points, the process having been to a large extent divorced from any explicit theoretical basis about why particular points are selected, it must also be appreciated that the (implicit) decision making process is still going on. The general level of skill associated with making appropriate decisions amongst those undertaking such work has increased, with more archaeologists now being aware of good recording practices when using such technologies. Thus, the choice of how to record becomes internalised without an explicit external frame of reference.

This is not, however, a fundamental change in the philosophy of data capture, as it has always been the case that an archaeological surveyor makes educated and informed decisions about which measurements to capture; rather the change is in the way in which we go about making these decisions, at which point in the process and with which evidence to hand.

The recent shift from highly selective field data capture techniques (such as TS or GNSS or photogrammetry) towards more mass capture techniques is changing the way in which we think about data capture and subsequent manipulation and use (Cripps 2004; Cripps 2008; Goskar et al. 2003). Rather than being selective in the field to capture carefully selected spatial data points representative of the subject to the specification required, fieldwork now often comprises indiscriminate data capture using instruments such as laser scanners resulting in massive spatial datasets which contain far more information about the subject (Figure 2). Important contextual information can also be gathered for little extra investment (Figure 3). That is not to say however that the interpretive process has been eliminated and that what we are doing is creating some kind of facsimile of reality, despite what some curators and commissioners of work would argue. The shift in technology has simply resulted in a shift in associated processes and the way in which we, as skilled archaeologists, form the archaeological record. Interpretation necessarily changes from being an action undertaken at point of data capture to an activity which can take place post survey, back in the office with a wealth of other pertinent information to hand.

With respect to landscapes, such data (notably airborne LiDAR, but also terrestrial laser scan data) can form the basis for an involved, analytical approach to archaeological remains using the battery of tools in the GIS toolbox. In addition to standard GIS derived products such as hillshades, slope and/or aspect models, we can process and visualise the data using techniques such as Reflectance Transformation Imaging (RTI), a real world technique developed for examining physical objects, which archaeologists have demonstrated that works eminently well for looking at digital objects including Digital Surface Models of landscapes in a virtual environment (Goskar et al. 2010). Of course, we can, as some would advocate, reduce this rich data to a series of images akin to aerial photographs and subject them to exactly the same analytical processes (e.g. Bewley et al. 2005), but, at the same time, by working with this data as 3D data and using novel workspaces as thinking spaces, it is possible to do far more

Figure 2 A 3D metric survey of a castle comprising a point cloud containing 1.5 billion individual measured points. An example of indirect data capture resulting in decisions usually made in the field being deferred to the office postfieldwork. 42



Cripps: Little by Little, One Travels Far models are able to be more complex and we can incorporate models of uncertainty (e.g. Nackaerts et al. 1997; Fisher 1991, 1992), we can include a whole range of variables (e.g. Kvamme 2006) or even run multiple complex models for comparison and qualification (e.g. Llobera 2003). Archaeological spatial technologies can support diverse forms of investigation and analysis, including those focussing on the more ephemeral postprocessual ideas of experience and being (e.g. Cripps 2007; Llobera 2003) (Figure 4).

So, the way in which we capture and manipulate spatial data has been changed for the better. Richer virtual working environments allow us to critically engage with our data more directly than ever before. Combined with improvements in the speed and ease with which such datasets can be captured or otherwise obtained and manipulated allows for more intellectual freedom when it comes to a reflexive process of investigation, analysis, visualisation, and dissemination; it is easier to try out ideas, to investigate and explore unfettered by technological complexities. This is not to say that we are in any way creating ‘better’ proxies for reality or that our virtual thinking spaces populated with digital data replace or supersede what has gone before; the use of innovative technologies simply provide alternative ways of recording and interacting with the archaeological resource, a progression, as it were, incorporating and building on the myriad of tools and technologies archaeologists have assimilated to date.

It is indeed true that these more complex models are still models, reductionist abstractions of reality, if you will. But, as complex spatial analysis becomes more accessible and more widely used within archaeology, it becomes possible to use such technologies in the same way as we use word-processors, spreadsheets or other technologies, i.e. to rapidly assess, appraise and then investigate hypotheses. Specialists are, as always, best placed to work at the cutting edge with the latest, newly adopted spatial technologies, but at the same time, non-specialists can undertake analyses that were until recently, the preserve of the specialist. This transfer and propagation of understanding and knowledge relating to spatial analysis fosters better appreciation and understanding of the usefulness of available technologies in the furtherance of theoretical discourse, and moves on from simply focussing on positive or negative reviews and critiques of scientific methods.

Wessex Archaeology now routinely uses technologies like these described as part of digital workflows for all purposes (Isaksen et al. 2008). Formulation of timelines for past coastal change, monitoring and assessment of impact as part of ongoing monitoring, recording and interpretation of historic buildings, investigation of historic drainage and agricultural patterns; all these varied types of work have benefitted from the use of highly detailed spatial datasets in environments where theoretical constructs can be investigated and more rigorously assessed in order to formulate more explicitly data driven archaeological interpretations. Such datadriven assessments facilitate a more robust link between our archaeological observations and records and the interpretations we buildupon them, and the assertions we then make. Our interpretive processes build new layers of information over the (richer) spatial data we collect in the field.

4. Text Spatial technology is not restricted to pure geometry however: Equally important as the points, lines and polygons are the related properties, including various forms of numeric and textual data. The use of numeric properties is typically limited to summary information, representing counts, populations and the like. The use of textual data falls into two main classes: the use of categorical values to support classification and the use of ‘free text’ for all manner of purposes both sensible and suitable, but also often poorly defined and counterproductive.

Following on from this, with regard to spatial analysis, developments in available computing power have allowed archaeologists to really get to grips with some of the more problematic theoretical problems. Our GIS

Figure 3 A 3D laser scan survey of a castle placed in context using a topographic laser scan of its environs. An example of indirect data capture resulting in decisions usually made in the field being deferred to the office post-fieldwork. 43



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 4 A 3D reconstruction of the Beckhampton Avenue at Avebury, used to investigate patterns of spatial relationships associated with the monument. An example of how technologies such as 3D models and GIS can be used to investigate and rigorously examine archaeological theories regarding the phenomenology of landscape. Archaeologists typically rely upon classification and typology as part of the interpretive process, closely allied with notions of narrative and storytelling. One of the bigger steps taken here in recent years is the adoption of GIS as a means of facilitating this, almost to the point where the very application of GIS for managing spatial data is seen as something akin to a panacea. As with the adoption of textual databases through the 1980’s, the late 1990’s and the first decade of the new millennium has seen the proliferation of the term GIS, often bandied around with little knowledge or understanding of what it is, what it can do or the raft of spatial theory which underpins it. The very statement that GIS will be applied is often seen as a ‘good thing’ and this has led to significant cases of misuse, misapplication, and overheightened user expectations.

GIS as a means of interacting with archaeological data is, however, often highly productive. Whilst the Cartesian view of the world is often critiqued (e.g. Thomas 1999; 2001), it is undoubtedly true that map-based interfaces onto the archaeological record allow archaeologists to gain a better understanding of their records; this is even true where the underpinning data model is inadequate or even flawed from a purist GIS point of view, what matters is the way in which a map based interface can provide a form of thinking space for non-specialist users. It is certainly the case that poor implementation can lead to poor interpretation, and there is the issue of unquestioning belief in what the computer says that leads to uncritical viewpoints. Equally, simple, crude and sometimes flawed data models have in turn led archaeologists to question what it is that they are seeing on the map, preferring not to trust everything presented to them, but simply use the map based interface as a means of interrogating the resource, a shortcut as it were to finding the primary, paper-based records; potentially also as the base for manually creating derived analogue spatial records. The GIS database functions here as (spatial) index rather than as repository of data to be formally analysed. This is GIS as a form of thinking space or simple facilitating tool rather than analytical platform,

As with any database technology, data needs to be structured in an appropriate way. Poor data structure can not only hinder use of the data but can actually lead to misleading results. The modelling of reality to infer a suitable attribute data structure is a process which really does need to be carefully thought through. If not, the data structure can bias or even determine any analytical outputs based upon it.

44

Cripps: Little by Little, One Travels Far one which, despite the apparent limitations, is still capable of bringing significant benefits.

Heritage which described archaeological data and processes in terms of the CIDOC CRM (Cripps et al. 2004). This was followed by subsequent work building on this model through the STAR project by a team from English Heritage and the University of Glamorgan. This has demonstrated the potential for using the ontology to work with a number of diverse datasets. More recently,

5 Ontology and Semantics There is nothing like looking, if you want to find something. You certainly usually find something, if you look, but it is not always quite the something you were after. (Tolkien, J. R. R. 1937. The Hobbit, or There and Back Again). Moving away from spatial technologies which purely deal with geometric data and textual information, another strand of research, again relating to technologies born in the discipline of computer science, has made significant advances over the past few years: the ontology. Ontologies as philosophical constructs are well known, world views, if you will, used as the basis for singular or shared understanding, discourse, and critique. Ontologies in the computer science world represent models of understanding in a similar way, but use formal, structured language capable of being used by machines as well as being human readable.

Figure 5 The top level entities from the CIDOC CRM core ontology, each of which has been specialised using class inheritance to produce entities representing archaeological recording units. An example of a core ontology used in the cultural heritage domain.

As such, archaeological ontologies have enormous potential for a variety of tasks. A major function being to facilitate resource discovery through semantic mediation of search tools. They can also be used for the integration of data from heterogeneous sources, disambiguating contradictory evidence, description of inference chains, and visualisation of data amongst other purposes.

the same team, through the STELLAR project, produced a number of tools to allow pretty much any archaeological excavation dataset to be transformed into a CIDOC CRM compliant RDF triple store (Binding et al. 2008; Binding et al. 2010). Furthermore, the ontology is being used as a base for the application of Natural Language Processing (NLP) techniques to extract structured data from archaeological reports and grey literature (Vlachidis et al. 2010). The production of semantically robust incarnations of any heritage data is the first step towards an archaeological semantic web of data.

Archaeological data is, by its very nature, fragmentary and incomplete. This is compounded by archaeological recording strategies which are often semantically unclear, merging concepts, such as culture and material, and at the same time using a variety of (often competing and contradictory) typologies, taxonomies, and classification schemata based on this partial knowledge and interpretations of it. Furthermore, there has been, and arguably still is, a tendency to publish more of the narrative and summary information as opposed to raw data. It is into this scenario that ontologies and semantic tools can be deployed to help unpick the complexity of archaeological discourse.

Once in this form, it is possible to use tools to visualise and analyse the data, e.g. jnet, a graphing tool (Ryan 2001). Dynamic graphs can be produced to represent and edit stratigraphic and physical relationships. Logic-based routines can be deployed to check for cyclic errors in stratigraphic relationships. As methods for interacting with ontologically-mapped datasets develop, the potential is there for directly positing complex research questions and accessing relevant data for more detailed analysis. Such a technological base, being robust and explicit with respect to our understanding of archaeological data, and being flexible enough to allow for modification to reflect changes in our understanding promises to make the process of archaeological interpretation more closely linked to the data or evidence from which assertions are made; the very process of archaeological interpretation becoming more explicit and interrogable, the process itself becoming part of the archaeological data archive.

The underlying theoretical basis to the use of domain ontologies is well understood and published widely. Their application within archaeology is similarly becoming more widely acknowledged, particularly the use of the CIDOC Conceptual Reference Model (Doerr 2003; Doerr et al. 2003). The primary role of the CRM is to serve as a core ontology for the cultural heritage domain, to act as the basis for mediation of cultural heritage information and thereby to provide the semantic ‘glue’ needed to transform today’s disparate, localised information sources into a coherent and valuable global resource (Figure 5). There is now a growing number of projects working with this core ontology. One such case involved an initial modelling/mapping project undertaken by English

So our archaeological semantic web comprises raw and interpreted data from a wide range of sources, it includes 45



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process analytical systems. The ontologies in such cases not only provide the semantic glue to bind heterogeneous archaeological datasets together, they can provide the framework though which automated reasoning can be applied, changing the ways in which analysis of archaeological data can be undertaken.

the higher level inferences and assertions built upon this data and the resultant narratives which ultimately form outputs from the research process. The disparate sources of data structures and typologies can be laid bare, exposed and examined. Furthermore, impacts of changes made in one place can be traced through the semantic web, for example to see how the chronology of a site is affected by changes to the pottery sequences used to provide the primary dating evidence.

Of course, there are potentially as many ontologies as there are archaeologists and the proliferation of ontologies is problematic. To achieve integration and interoperability of resources, we need some kind of commonality, an integrative framework. Importantly, this needs to be semantically clear and unambiguous. To this end, one approach centres on the use of core ontologies to govern the production and describe the nature of more detailed domain specific ontologies, and also allow horizontal integration between domain ontologies. It is only through discourse and agreement on the fundamental building blocks of archaeological domain ontologies that such approaches become useful; yes, it is often much easier to develop a simple stand-alone ontology for a particular use case, but that is to ignore the bigger picture and we end up back at square one, with disparate conceptual models and no little or no semantic linkage between them. This relates closely to the growing area of Linked Data (Berners-Lee 2006) and what that means for archaeological resources.

Importantly, spatial data can also be represented in ontological frameworks; archaeological spatial data is only part of a much bigger world of digital data. Within the broader computer science community, particularly the GIS community, ontologies are being used to describe and manipulate not only spatial data, but temporal data, together with spatiotemporal data (e.g. Stock 1997; Binding 2010). By allying archaeological domain ontologies with related ontologies and an integrating core ontology, we can retain the detail of our current (disparate and potentially contradictory) archaeological context records, finds classifications and records, environmental data, remote sensing data, and other information. There is also potential for drawing on ontologies from related disciplines so as to draw on relevant expertise, for example domain ontologies relating to geology (e.g. Zhong et al 2005) to cover site formation processes. By concentrating on modelling archaeological resources using emerging ontological standards such as the CIDOC-CRM, which in turn are capable of being related to associated spatio-temporal and other ontologies, the way is paved for truly semantic interoperability and advanced retrieval, visualisation and

The drive for Linked Data and the growing desire to see datasets opened up, warts and all, and exposed in ways that allow for clients, both human users and automated systems, to work with the data has both significant advantages but also pitfalls (Bizer et al 2009).

Figure 6 Data from Wiltshire Council HER overlain over aerial photography using Google Earth with the online HER record linked to each marker. An example of textual and spatial information combined to inform a user.

46

Cripps: Little by Little, One Travels Far Such approaches are being driven by the raft of new web technologies, including mapping systems, that are capable of being bolted together to form platforms capable of a broad variety of tasks via Advanced Programming Interfaces (APIs). Critically, these are not systems implemented or published by the owners of individual datasets. Rather they are community-driven systems, powered by a growing number of individuals with interest and some skills in web technologies (Figure 7; a landscape tour deployed using Keyhole Markup Language which draws data from Wiltshire Council HER using compatible software, in this case Google Earth). Interface and data are truly separated, joined only by a plethora of users who consume and share, use and reuse data as they see fit. Importantly, the skill level required is diminishing with the appearance of intermediary platforms, which mediate between the underlying technology and the general public. So it becomes possible to rapidly create an online mapping application without having to write a single line of code, simply by using an abstraction layer to interface with online databases and write dynamic HTML or even, at its simplest, by means of a GUI on a website such as Google Maps; the code is still being written but this is abstracted from the user perspective providing more open access to non-IT specialists.

ways that are fundamentally flawed, depending entirely on a variety of variables, notably the intentions of the originator, with obvious problems for downstream consumers of such data. Having said this, the combination of Linked Data with a robust ontological basis will provide great potential for archaeological research. Being able to visualise and analyse archaeological information from a range of resources simultaneously with semantic differences mediated automatically and made explicit is a significant step forward in and of itself. Combine this with mapbased views and spatial and temporal analysis tools and the potential for archaeological research is significant. Importantly, this thinking space is no longer the preserve of an individual technologically adept researcher. Rather the process of interpretation and forming understanding can be externalised and shared more easily. Indeed, the explicit nature of inference through such processes also facilitates thinking about thinking; the very nature of archaeological theory can be tied to underlying data, challenged and expressed in a myriad of ways, true multivocality. Furthermore, the potential for explicit nature of processes applied to data resulting in derived data and assertion allows methodologies to not only be rigorously exposed to analysis, but potentially shared more openly, to be critiqued and assessed and collaboratively worked upon and developed, even taken out into the field for further investigation (Figure 8 - the landscape tour from Figure 7 running on a mobile phone at the location indicated on the map).

So Linked Data allows users to draw on online resources, combine them using a variety of online tools, and produce new content from them. The notion of the mashup is key to this and a growing number of archaeologists and digital humanities specialists are able to make use of data made available in suitable forms. As the number of datasets published in this way grows, so does the potential for use. Interested in the relationship between particular site types and particular objects? Then use a web mapping application to view the relevant Linked Data sources over an Ordnance Survey map (provided through the OpenSpace API). Such approaches have obvious potential for resource discovery but, more than that, the potential is there for more detailed spatial analysis. Of course, there is the ubiquitous limitation of quality of source data, but archaeologists perhaps more so than practitioners of other disciplines have considerable experience in dealing with and understanding the notion of partial, incomplete or even inaccurate data, as outlined above. This last point relates directly to a point made earlier and which has often been missed in the rush for Linked Data, relating specifically to the notions of data quality, meaning and usefulness: semantic interoperability. A Linked Data approach does not have to be semantically unambiguous. It is simply a means of expressing data using arbitrarily defined structures capable of being parsed and processed in particular ways. The way in which one user implements concepts and/or relationships between concepts can be very different from another. Simply representing data as XML or RDF says nothing about the inherent meaning of the data other than that imposed by the originator. As such, Linked Data can not only be limited, but can comprise data expressed in such

Figure 7 Displaying the same HER data on a Google Map whilst stood at the very position of the archaeological site displayed, as indicated by the onboard GPS within the smart phone. An example of how data can be taken out into the field for ground truthing, enhancement and other analysis.

47



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process such as the Ordnance Survey and communities such as the Open Source groups including the Open Geospatial Consortium all providing valuable platforms for archaeologists to work with. Archaeological theory becomes data driven, more exposed and open to others whilst at the same time, data is available in more accessible ways on which to base theories and make assertions.

6. Conclusion The wide world is all about you; you can fence yourselves in, but you cannot forever fence it out. (Tolkien, J. R. R. 1954. The Fellowship of the Ring) To conclude, it is the assertion of this paper that many small steps in a number of disciplines including archaeology and computer science are incrementally moving towards the possibility of much tighter integration between archaeological theory and archaeological data through a more explicit process of research. Structured spatiotemporal resources combined with a range of tools and technologies capable of supporting data driven research and theoretical development has the potential to improve the ways in which archaeological narratives and understandings of the past are pieced together. Data and outputs such as narratives and theories are becoming more explicitly linked. This is happening on small scale, local developments where spatial technologies such as GIS and databases are used as thinking space and reporting tools, activities which have become very much the norm in a wide range of archaeological institutions from academic departments to contracting units. This is also happening on a large scale with developing and formalised national and international standards and frameworks for heritage data combined with the might of commercial organisations such as Google and Microsoft, agencies

Despite this, there is still critique and discussion about the theoretical implications of such developments, problematising the use of such technologies per se rather than engaging with their use and integration within digital workflows and the very real discourse regarding this. The use of technologies as described here is part of 21st century archaeological processes, the debate should move on to how we use these technologies and the impact they may have. Whilst some platforms and technologies are seen as neutral (e.g. the use of mobile devices and in some cases use of GIS and databases), many are not, and these are not necessarily groundbreaking or digital; many of the same issues apply to the use of spatial technologies as relate to the production of text or images. The very idea of structured data, the use of typology and classification, archaeological visualisation and associated technologies such as VR, GIS, and databases in general is still not universally accepted and only slowly moving forward into more fertile territory through publications such as this book.

Figure 8 The world of Linked Data; datasets published as of September 2010 (Source:http://en.wikipedia.org/ wiki/Linked_Data). This shows how many datasets are being published as Linked Data, into which cultural heritage data can be linked.

48

Cripps: Little by Little, One Travels Far Bizer, C., Heath, T. and Berners-Lee, T. 2009. Linked data - the story so far. International Journal on Semantic Web and Information Systems (IJSWIS).

However, time moves on and so does the discourse. The rate of change in the world of information technology and the increasing proficiency of archaeologists with respect to technological advances, particularly the growth in digital humanities specialists, is driving change forward; the obvious feedback loop here resulting in an ever accelerating rate of change. And so, what was only dreamt of a few years ago by some and seen by others at that time as being reductionist and with a closer relationship to the so-called new archaeology rather than more theoretically driven and reflexive strands of archaeological thinking, is slowly maturing. Archaeological computing is, as evidenced by this very session at TAG, engaged with theoretical archaeology, the barriers between the two becoming less distinct than they were in the first decade of the 21st century. Through tighter integration, it is possible to develop further our theoretical models and narratives whilst at the same time improve the tools and technologies with which we can interact and produce the archaeological record and academic discourse.

Cripps, P. 2004. Moving into the third dimension: threedimensional data capture and manipulation. Archaeology Computing Newsletter 61, 24-26. Cripps, P., Greenhalgh, A., Fellows, D., May, K., and Robinson, D. 2004. Ontological modelling of the work of the Centre for Archaeology. CIDOC Technical Paper Cripps, P., Earl, G. and Wheatley, D. 2006. A dwelling place in bits, in Jorge, VO et al (eds) Approaching ‘prehistoric architectures’ of Western Europe from a ‘dwelling perspective.’ Journal of Iberian Archaeology 8, special issue. Adecap, Porto. Cripps, P. 2007. Pathways, perception and the development of place; computational approaches to movement and perception of landscape in prehistory, in Figueiredo, A. and Leite Velho, G. (eds) 2007. The world is in your eyes – Proceedings of the XXXIII computer applications in archaeology conference.

Acknowledgements The work discussed and described here is very much the property of the referenced authors and is included here solely for the purposes of providing exemplars of relevant works. The direction is entirely my own opinion. With thanks to those authors with whom I have discussed these matters and to the attendees of TAG and CAA UK conferences where I have presented some of these ideas for discussion previously. The title and quotes forming subheadings are all taken from the work of J.R.R. Tolkien, a scholar who had an uncanny ability to produce intriguing and pithy statements.

Cripps, P. 2008. Spatial technologies in archaeology in the twenty-first century, in Greengrass, M. and Hughes, L. (eds) The virtual representation of the past. Ashgate Publishing. Doerr, M. 2003. The CIDOC CRM – An ontological approach to semantic interoperability of metadata. AI Magazine. 24 (3). Doerr, M., Hunter, J., Lagoze, C. 2003. Towards a core ontology for information integration. Journal of Digital Information. 4 (1).

References Berners-Lee, T. 2006. Linked data. International Journal on Semantic Web and Information Systems. 4(2).

Fisher, P. F. 1991. First experiments in viewshed uncertainty: The accuracy of the viewshed area. Photogrammetric Engineering and Remote Sensing 57 (10). 1321-7.

Bewley, R. H., Crutchley, S. P. and Shell, C. 2005. New light on an ancient landscape: LiDAR survey in the Stonehenge world heritage site. Antiquity 79, 636–47.

Fisher, P. F. 1992. First experiments in viewshed uncertainty: simulating fuzzy viewsheds. Photogrammetric Engineering and Remote Sensing 58 (3). 345-52.

Binding, C., Tudhope, D., May, K. 2008. Semantic interoperability in archaeological datasets: data mapping and extraction via the CIDOC CRM, in Proceedings (ECDL 2008) 12th European conference on research and advanced technology for digital libraries, Aarhus, 280– 290. Lecture Notes in Computer Science, 5173, Berlin, Springer.

Gillings, M. and Goodrick, G. T. 1996. Sensuous and reflexive GIS: exploring visualisation and VRML. Internet Archaeology 1.

Binding, C. 2010. Implementing archaeological time periods using CIDOC CRM and SKOS, in Aroyo, L. et al. (eds) Proceedings 7th extended semantic web conference, Heraklion, ESWC 2010, Part I, Lecture Notes in Computer Science, 6088, 273–287, Berlin Heidelberg, Springer-Verlag.

Goskar, T., Carty, A., Cripps, P., Brayne, C. and Vickers, D. 2003. The Stonehenge laser show. British Archaeology. 73. Goskar, T., and Earl, G. 2010. Polynomial texture mapping for archaeologists. British Archaeology. 111.

49



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Goskar, T. and Cripps, P. Interactive landscape relighting. Remote Sensing and Photogrammetry Society Archaeology Special Interest Group Newsletter, Spring 2011. 3-6.

Thomas, J. 2001. Archaeologies of place and landscape, in Hodder, I. (ed.) Archaeological theory today. Cambridge and Oxford, Polity Press. Vlachidis, A., Binding, C., May, K., Tudhope, D. 2010. Excavating grey literature: a case study on the rich indexing of archaeological documents via natural language processing techniques and knowledge based resources. ASLIB Proceedings journal. 62 (4, 5).

Isaksen, L., Goskar, T., and Cripps, P. 2008. Making people believe. British Archaeology. 100. Kvamme, K. 2006. Integrating multidimensional geophysical data. Archaeological Prospection 13. 57-72.

Zhong, J., McGuinness, D. L., Antonellini, M., Aydin, A. 2005. Ontology for structural geology. American Geophysical Union, Fall Meeting (AGU2005). Eos Trans. AGU 86(52), Fall Meet. Suppl., Abstract IN43A0323.

Llobera, M. 2003. Extending GIS-based visual analysis: the concept of visualscapes. International Journal of Geographical Information Science 17 (1). 25-48. Nackaerts, K., Govers, G. and Loots, L. 1997. The use of Monte-Carlo techniques for the estimation of visibility, in Dingwall, L., Exon, S., Gaffney, V., Laflin, S. and Van Leusen, M. (eds) Computer applications and quantitative methods in archaeology, CAA 1997: Archaeology in the age of the internet. British Archaeological reports International Series. Oxford, BAR Publishing.

STELLAR project. Available at: http://hypermedia.resea rch.glam.ac.uk/kos/stellar/ STAR project. Available at: http://hypermedia.research. glam.ac.uk/ kos/star/ CIDOC Archaeological Sites Working Group 1995. Draft international core data standard for archaeological sites and monuments. International Council of Museums.

Ryan, N. 2001. jnet, a successor to gnet, in Borner, W. (ed.), Archaologie und Computer, Workshop 6. Wiener Stadtarchaologie, Forschungsgesellschaft November 2001. Stock, O. (ed.) 1997. Spatial and temporal reasoning. Dordrecht/Boston/London, Kluwer Academic Press.

Royal Commission on the Historical Monuments of England and English Heritage 1995. Thesaurus of monument types: a standard for use in archaeological and architectural records. Swindon, RCHME.

Thomas, J. 1999. Understanding the Neolithic. London, Routledge.

Archaeology Data Service. Guides to good practice (G2GP). http://guides.archaeologydataservice.ac.uk/ English Heritage 2007. MIDAS Heritage.

50

4 Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations Markos Katsianis Aristotle University Thessaloniki, Greece Abstract The article examines the main conceptual and practical issues involved in the use of digital technology to facilitate the documentation and interpretation of intra-site archaeological research. Emphasis is placed on the need for theoretically informed Geographic Information Science (GISc) models that mediate effectively excavation data representation and exploratory analysis in support of the archaeological knowledge production process. Towards this end, the application of GIS technology in excavation documentation involves some careful thinking with respect to archaeological ontology, 3D visualization and temporal reasoning. The observations reached from this exercise are then used to consider some practical applications of GIS in modern excavation. Relevant examples are taken from an excavation information system developed within the framework of the archaeological project of Paliambela Kolindros, Greece. Keywords: GIS; Excavation; Ontology; 3D Visualisation; Temporal Reasoning; Conceptual Representation Model; Pyramid Model; Paliambela Kolindros. _______________________________________________________________________________________________ archaeological theory and digital technology and reflect upon the limitations and advantages that digital media bring to archaeological practice (Beck & Seif 2006).

1. Introduction In the last few decades archaeology has accomplished significant steps in expanding the amount of information extracted from material remains through novel recovery techniques and the application of digital technologies. However, despite the fact that a modern archaeological excavation extracts a significant amount of information through meticulous recording processes, extensive sampling protocols and innovative artefact processing techniques, there has been limited success in developing methods that enhance interpretative reasoning in the subsequent process of generating new archaeological ‘knowledge’.

Building on the insights gained during the development process of a digital excavation documentation system for the ongoing archaeological project in the prehistoric site of Paliambela Kolindros in Greece, I will try and highlight a number of theoretical issues that impact upon the use of GIS in intra-site research with respect to facilitating archaeological interpretive reasoning. The key point is to understand what we actually want from digital technology. Towards this end, a close look at archaeological knowledge production practices with respect to the notions of ontology, visualization and temporal reasoning can allow the creation of more expressive conceptual models that are closer to our cognitive perspective and can pave the road to advanced implementations.

Digital technologies have contributed significantly to the better management of excavation archives, but not in the collation, analysis and verification of information. Intrasite applications, other than management tools, should facilitate interpretive reasoning, ongoing evaluation of the excavation process and reflexive excavation strategies (Lucas 2001a). In practice, this is related to the provision of digital functionality that enhances field recording, detailed logging, dynamic management, realistic visualization and inductive analytical treatment of excavated data, allowing effective and critical control over the course of the investigation.

2. Excavation as knowledge production Excavation forms a core research process that provides archaeology with its object of study, archaeological material. Archaeological reasoning is based upon the recognition that archaeological artefacts are carriers of information allowing us to formulate ideas about past processes. As Hourmouziades (1995, 249) notes, ‘archaeological artefacts... through a proper scientific approach can obtain meaning and lead to sound interpretations; they can reveal their essence; because...they are correlated with events, suggest

To advance these aims, we need to understand the implications of digital technology and promote novel ways of integrating computer applications within established archaeological practices. In this respect, it is crucial to comprehend the relationship between 51



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 In a similar manner, archaeological interpretation is based on the interaction of researchers with scientific instruments and means of documentation as part of a chain of reasoning that results in multiple data transformations. Witmore (2005) understands documentation resources as media, highlighting their active role within knowledge production processes. He defines media as ‘the modes of articulation through which knowledge is mobilized, manifested, and materialized’. All physical objects that are used as carriers of archaeological information, such as diaries, drawings, photographs, find catalogues, and more recently computers, can be considered as documentation media.

functions, demonstrate uses, constitute social data, help in the transformation of empirical observation to scientific knowledge, identify and explain processes’. In this respect, an excavation can be understood as a scientific procedure that retrieves and observes material remains related to past human activity with the goal of producing knowledge about the past in the present. In essence, every archaeological excavation is an observation and recording process that not only constitutes a rich and complex information archive, but employs this archive to unravel past processes. The emphasis on excavation recording methods and systematic documentation can be explained by the fact that the excavation is considered an ‘unrepeatable experiment’ that destroys its object of study (Barker 1982, 12).

Recently, attention has been given to this point in the context of Symmetrical Archaeology, a new epistemological approach to social practice in archaeological research that introduces the basic ideas of Latour in the field of archaeology (Olsen 2007; Shanks 2007b; Webmoor 2007; Witmore 2007). Rather than thinking in terms of subjects and objects, both categories are understood as actors who are mutually constituting as part of the scientific process. Each actor, i.e. archaeologists, documentation resources and archaeological data cannot be defined independently. Data only exist in the context of the media on which they are registered, while every medium has particular properties that affect the way recording is carried out by the archaeologists (Webmoor 2007). This interaction among archaeologists, documentation media and information are, in the words of Shanks (2007a), ‘a mediation process’.

This claim has been criticised as implying the inherent weakness of documentation to replace the materiality of the real world (Witmore 2005; after Andrews et al. 2000, 527). Rather than thinking of the excavation process as an inevitable observation technique that results in the deterioration of archaeological material, different conceptualisations of it, as transformation (Lucas 2001b) or as creation of a metaphor (Merlo 2004) have been proposed, implying the enhancement of archaeological material through its transcription into a research dataset. In this framework, the excavation record acts as the primary medium for subsequent archaeological interpretation, providing at the same time a necessary means for the repeated interaction with an otherwise historically finite excavation event (Lucas 2001b). Archaeologists interact with the archive contents across multiple levels, in which field observations are constantly processed leading to interpretations of increasing scale (Hodder 1999). In this sense, the documentation of the archaeological process is not related solely to the results of fieldwork, but also to the integration of all interpretations developed during the study of the excavation archive.

This is where the relationship between archaeology and information science lies. The employment of available technology to collect, organize and study data with the aim of producing new knowledge is the reason for the existence of information systems. The mutual constitution of researchers, media and data forms an information system, ‘a socio-technical discipline involving both human and non-human entities’ (Tatnall 2003, 266). Research conducted around information systems is strongly related to the analysis of the contribution of actors, human and otherwise, in the complex process of knowledge production (Tatnall 2003).

For Witmore (2004) the successive transformation of the excavation archive across multiple fields, away from the excavation site, produces a chain of successive and mutual references that should warrant the iteration of any interpretation to its supportive material evidence. The ability to refer back to each interpretative step between the excavation and its publication is an important way of promoting greater credibility in archaeological research.

An information system consists of the material resources that allow the existence of the system, the data that are the product of an observation process, the operations that facilitate data management and processing, and the users of the system (Stefanakis 2003, 21-22). In an excavation information system, material resources include recording forms, drawings, photographs, Harris Matrix diagrams etc. The data are recorded archaeological observations, while operations are shaped by archaeological theories and methods of observation. Finally, the users are archaeologists. In the case of a digital or hybrid excavation information system things get complicated, as the use of computers begets the alteration of the basic information management technology. Therefore, it is

These observations have been influenced by ideas expressed by Latour (1999), who emphasises the impact of wider social, political, economic and technological conditions on scientific knowledge production processes. He observes that in practice the validity of an interpretation is usually black boxed, in the sense that it is based on the results of research conducted via scientific instruments and complex methodologies that cannot be verified or evaluated easily by the outsider or the uninitiated. 52

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations and technical developments, such as photography and cinema (Shanks 2007a). As indicated by Brittain and Clack (2007, 14), old technologies are not simply replaced by new ones. They are combinations, mixing those elements that prove to have useful applications. At the same time, they maintain, improve or displace other components along with their associated knowledge production practices.

crucial to reflect upon the impact of digital technology on the rest of the actors in the system.

3. Digital information systems in excavation From as early as the 1960s, digital technology has gathered an audience convinced of its potential to contribute to every aspect of archaeological research. However, its application has been, and continues to be, a slow process compared to other scientific disciplines (Huggett 2000, 19). Most difficulties are associated with the operating backgrounds of new digital media, which are different to that of their analogue equivalents.

New digital media offer significant opportunities, not only because they are effective in the management of information, but also because they allow investigation of a variety of new modes of representation and uses of information, enhancing knowledge production strategies (Llobera 2010). However, it should be noted that new media can only offer opportunities for improvement rather than solutions to the methodological problems of excavation. Thus, a digital information system may never construct knowledge by itself in ways envisaged by many researchers, e.g. plotting all evidence and running complicated algorithms that display meaningful objective patterns (see Carver 2005, 20-21). It can, nevertheless, mediate the systematic or spontaneous synthesis of archaeological knowledge.

The advent of digital technology has changed both the materiality of the medium and the way information is stored (Manovich 2001). In digital media, information is maintained in a dynamic state through successive codifications, calculations and transformations as a result of the numerical representation located at the foundation of any graphic, sound or verbal display in the computer. Digital information is transformed into a signal stored in much less physical space, reproduced instantaneously, transmitted directly, and detected automatically not only by humans, but also by machines. Computers can handle text, image, sound and animation in the same way. Consequently, the volume of data is increasing and expanding available formats. At the same time, information integration, management, processing and dissemination is greatly enhanced.

To achieve this, digital media have to be smoothly integrated within established practices that support archaeological knowledge production. Lock (1995, 13) points out that ‘for archaeological computing to become an integral part of the archaeological process rather than just a set of tools which can be used at appropriate times, there has to be a fundamental link between the computing and the underlying theoretical stance’. In this context, computer application development efforts can contribute decisively to reviewing the theoretical, methodological and practical conventions by which we survey the past.

The famous dictum of Marshall McLuhan (1964) ‘the medium is the message’ emphasizes the fact that human activity is more influenced by the media used rather than the actual information they convey. Any new form of media creates different messages requiring thereby the introduction of new modes of experiencing their understanding. In this context, each new medium affects communication practices and information management, as well as human perception.

4. Knowledge production excavation using GIS

in

The observations made can be further elaborated with respect to the use of GIS in excavation practice. Until their advent, the representation of excavation space was achieved through maps and plans at various scales, which provided a means of organizing and presenting archaeological data, but could not support quantitative analysis or easy integration between maps. GIS effectively integrated databases, graphical user interfaces and spatial analysis tools, offering flexibility in combining and exploring archaeological data (Wheatley and Gillings 2002, 1-18).

The employment of digital technology in excavation research is often said to distance our interpretation from reality, thus leading to the interpretation of a digital copy rather than the actual excavation event. Without doubt, fieldwork, like any research process, contributes to the generation of a primary body of observations. As recorded information, however, any observation must be used to inform or adjust subsequent interpretations during fieldwork and after its completion (see for example Hodder 1999, 30-65). In this sense, the excavation archive preserves its autonomy and comprises the background, as well as the basic tool for any future interpretive effort.

Kvamme (1999, 154) singles out the representational capacities of GIS as their most powerful quality, noting that ‘information may be assembled according to its position in space and displayed in the form of maps, allowing ready visualization of structure and relationships in complex regional data sets and creating something of a revolution in the transmission of information’. The importance of visual data examination is associated with the advanced abilities of the human brain to distinguish

In this context, digital media do not need be considered as another black box. As indicated by Cripps et al. (2006), the forms of representation used in digital media do not differ from those of analogue media. Many theorists point out that the former, rather than breaking the chain of development in the representation of reality, extend forms of representation that have their roots in the Renaissance 53



practices

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 program reinforce a false impression of objectivity in the sense that their results are the product of a calculation procedure and not a process of interpretation. They point out that GIS displays are virtual models of reality and include several conventions, inaccuracies and errors caused either by the need for complex digital data structures or by the digitization process. It becomes immediately obvious that GIS includes additional constraints, which may have implications for how to organize data and analytical perspectives.

visual patterns in two or three dimensions. Through an informal and almost intuitive analysis of a cartographic display, useful observations can be directly extracted in a manner similar to interpretive reasoning (Lock and Harris 1992). The integration of excavation data with cartographic display has shifted the focus of archaeological analysis to the digital map rather than the excavation diary. It is now possible to investigate patterns inherent in an archaeological dataset in an inductive manner using a series of techniques grouped under the more general term, Exploratory Data Analysis. These encourage an early stage of data investigation for possible patterns that can be subsequently evaluated by standard confirmatory spatial statistic techniques (Tukey 1977).

Yet, as indicated by Cripps et al. (2006, 26-28), the use of GIS provides tools for the study of real world phenomena in a simulation environment, leading to interpretations that are equally reliable to analogue or other digital technologies. Even if they affect the final interpretation, GIS mediate the conduct of complex analyses and, in this sense, are a vital component of research reasoning. Moreover, interpretation is not merely the end-product of a GIS analysis, but is actually composed by the researcher who incorporates the results of the analysis into a wider hermeneutic body (Rajala 2004, §5.3). In this sense, Conolly and Lake (2006, 10) identify the successful application of GIS in archaeological research in an understanding of its inherent theoretical limitations and technical operating parameters. Towards that end, recent studies attempt to explore ways for extending the analytical capacities of GIS in order to fit post-processual reasoning (i.e. Llobera 2006).

However, these capacities have not been utilized to their full extent. Existing intra-site GIS exhibit some weaknesses that prevent their effective and widespread application in archaeological research. Overall, different trends are not only due to the off-the-shelf technological solutions offered by current software or to staff skills, but are also relevant to the research methodology used in each archaeological project. A major problem relates to the quality of the excavation record (Wheatley and Gillings 2002, 235-236). Emphasis on the spatial aspects of archaeological data leads to the intensification of field survey as well as the adoption of specific, often complex, digitization methodologies. Without these, problems arise concerning the consistency and accuracy of spatial registration rendering spatial visualization and data analysis ineffective.

As far as usage of GIS in intra-site research is concerned, there is enough room for the successful integration of both theory and technology. Towards this end, González Pérez (2002) suggests that application development procedures should be context aware, validating both the theory of the application domain and the technology outlined in the computing domain. After all, most inefficiencies relate either to the methodologies we use to approach the excavation process or the choice of the available technological solutions. Uncritical views of archaeological fieldwork practices and lack of application development strategies contribute to restrictive models of reality, which do not serve archaeological reasoning (Hadzilakos and Stoumbou 1996).

Several researchers also argue that there is a gap between archaeological theory and GIS techniques, which demonstrate a difficulty in integrating the latest archaeological theoretical trends in a productive manner. There is a general feeling that GIS are more suited to processual approaches and quantifiable data, one reason being the fact that GIS development largely took place within scientific disciplines that favour positivist approaches (Gidlow 2000). In this context, GIS applications have been criticized as fostering the objectification and simplification of complex archaeological reality. All their aspects, including spatial representation models and statistical processing techniques, affect the choice of analytical approaches, as well as the kind of data that can be entered (Barceló and Pallares 1996). Similarly, the mathematical and geometric background of spatial analytic procedures in GIS includes restrictions on data interaction (Goodchild 1996). As a result, GIS can lead to bias through convenient but unwanted adjustment of the questions submitted and the data collected (Harris and Lock 1995, 357).

5. Conceptual modelling The implementation of an information system involves the selection of elements of interest to the application and their transcription into a digital data structure that can serve the intended functions. In this respect, the functionality of an information system is directly related to the effectiveness of the conceptual representation of a wider Universe of Discourse (UoD). Any representation involves the division of reality into individual components and the clarification of their relationships, the choice of appropriate representation types, the advancement of practical and logical operations that support the transition from reality to representation and vice versa, as well as the clear definition of the

In a similar style, many researchers (Bateman 2000; Johnson 2002; Reilly and Rahtz 1992) argue that cartographic representations contained in any GIS 54

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations to

the development of functional a GIS-led archaeological information system.

Conceptual modelling breaks reality down into a complex set of discrete and mutually associated entities. An entity can be defined as ‘any unit or object that exists physically or conceptually’ (Stefanakis 2003, 4). Each entity can be described by a set of properties that can be divided into three dimensions: the thematic or descriptive, the spatial and the temporal (Hadzilakos and Stoumbou 1996, 22). The creation of models that incorporate the previous dimensions of archaeological data is a key element to a successful implementation of GIS in excavation practice.

Based on the analysis of different modes of acquisition, representation and use of spatial and temporal knowledge, Peuqeut proposed a relative simple methodology for attaining knowledge about complex geographic realities (1994). The logic of the method distinguishes between spatial, temporal and thematic questions that determine the characteristics of each entity in real world. In particular, it employs a trinity of questions ‘Where?’, ‘When?’ and ‘What?’. The question ‘Where?’ refers to the location of an entity, the question ‘When?’ is related to the ‘history’ of the entity, while ‘What?’ focuses on the definition of the entity through its thematic properties. Based on this distinction a review of current research in the fields of archaeological theory, information technology and cartographic representation is presented below.

objectives behind this attempt, with respect information communication (van der Scans 1978).

The basis of any modelling process is conceptual thinking defined as ‘a system of concepts and categories which divide up the corresponding universe of discourse into objects, processes and relations in different sorts of ways’ (Smith and Mark 2001, 593). Digital representation of reality involves two models: a) an abstract model of reality, i.e. an abstract description of a human perception of reality and b) a data model, i.e. the digital transcription of human perception by applying standard concepts (Stefanakis 2003, 212-213).

5.1. ‘What?’ or ‘excavation ontology’ The process of defining the building blocks of the excavation universe requires a reflexive review of our current research practices on the basis of how we define our observables in the present in order to arrive at valid conclusions about the past. Ramenofky and Steffen (1997) point out that the establishment of observation units is a necessary step for archaeologists to organize their research subject and formulate a methodology for the exploration of significant variations in a phenomenon, i.e. the ‘archaeological site’. They maintain that, in essence, every excavation is a process of subdividing a complex stratigraphic matrix into separate entities or units of observation.

Obviously, the abstract model is a simplification in the sense that it cannot describe reality completely and therefore the conceptualization created is different from reality itself. The reasons for this is the subjectivity of perception of the observer, the deliberate omission of minor details of reality, the technical imperfections of data collection, and the inherent weakness of the concepts of a model to accommodate all aspects of reality (Stefanakis 2003, 57-58). However, its usefulness lies in achieving a clearer understanding of a complex reality via a simplified description. The data model refers to the systematic, consistent and standardized description of the structure of a database, which is composed of data types, the correlations between the data and the restrictions applicable to the foregoing.

Adams and Adams (1991) discuss the definition of the entities that form observation units within a taxonomy or a typology. The segmentation of reality into entities and their organization into categories based on common characteristics leads to the creation of types that are interconnected within a broader model that describes some aspects of reality. A type may be a component of broader conceptual classifications, which together form a typological hierarchy. For example, entities such as a grinder, a spindle whorl etc., denote distinct categories and also comprise observation units that can be integrated within the more general class tools, which in turn forms one of the categories belonging to the generic type of artefacts or finds. Madsen (1999, 129-130) summarizes the practical effect of this procedure in the gradual reduction of the range of features that characterize a type in every level of specialization.

Even prior the design of a digital information system, an abstract model of reality exists, which is identical with our mental conceptualization of that reality, at least as it pertains to the intended scope of our application. When designing an information system, a conceptual model is an important first step. The success of such a conceptual model lies in its expressiveness in terms of procedures and data. The intertwining of the previous theoretical arguments into a functional GIS requires the use of a general conceptual model of spatio-temporal data. In recent years more and more attention has been given to the development of spatio-temporal conceptual data models of a general nature, which provide guidelines for implementing multi-dimensional (x, y, z, t[ime]) systems. In archaeology, an early model was advanced by ArroyoBishop and Lantada Zarzosa (1995) in the context of their Archeo-DATA system, which emphasized the unification of Object, Space and Time as a fundamental condition for

Each taxonomic distinction of this kind requires semantic integrity, based on the existence of clearly defined boundaries between types and therefore in mutually excluded conceptual definitions (Adams and Adams 1991, 76-77). As noticed by Ramenofsky and Steffen (1997, 5-6), the definition of a type results either from the classification of a set of parameters in empirical entities or is made up arbitrarily, according to the research scope. 55



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 simply the subjectivity of the observer (Madsen 1999, 126-127). Their effect is evident when comparing similar data from different archaeological projects.

Problems arise as to whether a type is determined by a set of variables and to what extent they objectively characterise its definition (the 'typological debate' started as early as the 1940s; see Spaulding 1953; Ford 1954).

Overall, the identification of the entities involved in an archaeological excavation is a fairly complex intellectual effort affected by many factors. Among these is existing scientific knowledge, the theoretical background of the researcher, the research objectives, the measurement tools, as well as the particularities of the archaeological material. All affect both the research means used and the interpretive outcomes reached. As Adams (1988, 52) points out: ‘typologies, like other archaeological procedures, are not ends in themselves but means to an end, and the ends must justify the means’.

Kotsakis (1983) notes that the definition of variables and the relationship between them depends on the research purposes. He believes that in the framework of a preestablished theoretical model it is possible to evaluate the degree according to which some variables are more important for the definition of an entity than others. In this respect, the correspondence of the derived categories with reality is based on cognitive theory, not on the existence of an inherent pattern within the attested data variables. Hence, archaeology can support multiple classifications or multiple worldviews depending on the research scope, accepting the inevitable problems of data correlation.

Kotsakis (1983) in his analysis of painted pottery from the Middle Neolithic site of Sesklo in Thessaly (Greece), describes his methodology for constructing a typological reference system beginning with a theoretical proposal that explains the scope of his investigation. He then sets standard rules for determining diversity and applies these rules to an observed sample of pottery. Despite the new concepts he proposes, he also emphasizes use of a common vocabulary, abstract concepts and coherent classifications, as well as the need to acknowledge preexisting classifications. As opposed to extensive typologies, this approach allows for the comparison of data between different studies. The same pattern can be observed to apply to the overall fragmentation of the excavation universe into observation units.

Further observations can be made in relation to partonomy, i.e. the division of reality into entities that are components of broader groupings (e.g. a context as part of a group layer). In effect, partonomic structures in an excavation define relationships between two conceptually identical entities described by basically the same characteristics. Their difference lies in the level of interpretation. Interpretive reasoning essentially is a move from the part to a whole. The ultimate purpose of the excavation process is to create an overall idea for a whole site through the gradual association of individual observations. Most digital documentation systems have difficulty in managing such relationships that involve the creation of new hermeneutic categories that did not exist during excavation, but arise later in study as a result of the interpretation process (e.g. a stratigraphic layer or a set of buildings).

The creation of a classification system following this logic is followed by the sorting of a population within the defined categories. This process frequently requires arbitrary decisions, since, in practice, a real world object may display characteristics related with multiple categories (Adams 1988). Often fuzzy logic approaches are proposed as a solution (Niccolucci et al. 2001). However, Rajala (2004, §3.6) remarks that ‘the problem with fuzzy logic is that it tries to quantify, often arbitrary, subjective decisions. It looks scientific but is a product of qualitative evaluation’. Such quantitative approaches seem to correspond to the numerical foundation of digital media, but avoid addressing problems inherent in uncertainty formalisation, such as subjectivity or individual aptitude for sorting decisions. These problems in the definition of object types and in the attribution of archaeological data can be addressed in practice through the inclusion of statements of confidence and self-critical information (Chadwick 2003, 110). Moreover, interpersonal consistency in the use of terminology (Adams 1988) and the adoption of standards that ensure the required semantic correctness of any definition can further reduce the problem (Rajala 2004, §3.6).

The classification of the excavation universe into discrete units of observation is part of the problem of archaeological information management. The most common problem of digital applications in archaeology arises when new entities that cannot be integrated into predefined categories or new properties that break current classifications are identified. In such cases, major adjustments of the data structure are often required. A useful suggestion is to avoid inclusive descriptive strategies that seek to integrate an extensive account of reality, and rather employ classifications that are aware of the research purposes of an application, maintain high levels of abstraction and carefully select attributes that focus on the description of the relevant aspects of reality (Madsen 1999). The complexity of data modelling increases dramatically when the question of attribute definitions and their respective values is added to the equation. There are differences in terminology (e.g. awl or needle), in measurement units (e.g. metres or feet), in terms of quantitative or qualitative assessments (e.g. few, some, many or low, medium, high). These discrepancies are a result of the polysemy and ambiguity of natural language, the differences in archaeological recording methods or

As the above discussion should make clear, archaeological data modelling problems are not an inevitable outcome of the inherent complexity of the archaeological record. Rather they are due to the differences in perspective through which archaeologists approach the excavation material. Primary observation 56

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations Conceptual Reference Model (CIDOC-CRM, ISO 21127: 2006) in the case of Centre for Archaeology’s (CfA) Revelation project. Its use contributed to a better understanding the archaeological research practices of the organisation through their mapping in a detailed conceptual data framework. The resulting data model has been employed further in the context of the STAR and STELLAR projects as an umbrella conceptual framework for the semantic linkage and cross searching of archaeological datasets with diverse data structures (Tudhope et al. 2011).

data and all subsequent information reflect the goals of the researcher and the broader context of the archaeological research in the present. The distinction of data in observation units is mainly a conceptual abstraction in the mind of the archaeologist who decides what characteristics of reality relate to the scope of the study. In this context, the solution to the complexity of archaeological information should not be sought in the implementation of increasingly complex and detailed data models that seek to fully describe the characteristics of an objective phenomenon, but to the better conceptualization of the excavation universe and the correlation of different abstract models through standard conceptual descriptions. This approach makes the problem one of excavation ontology.

In a similar way, the notions of the CIDOC-CRM were employed to provide a semantic account of the excavation domain of Paliambela Kolindros during the development of the application data model (Katsianis et al. 2008; Katsianis 2009). An example is provided in Figure 1. The semantic description of the terms contained in the data model revealed the logic inherent in the excavation methodology of the case study site. Its usage proved particularly useful in distinguishing between interpretive events in the present (i.e. interpreting a set of features as a house) and events that are related to past human activity (i.e. the actual construction of these features). In this respect, the modelling process, besides verifying the consistency of the conceptual description of the application data, helped to clarify confusing views on the definition of basic concepts of the excavation process that existed between different members of the excavation team.

An ontology can be defined as ‘a logical theory which gives an explicit, partial account of a conceptualization’ (Guarino and Giaretta 1995, 4), which ‘serves as a concretization mechanism that defines the vocabulary of the knowledge base’ (Tomai and Kavouras 2002). Ontologies are not categorizations of real phenomena, but can be understood as socially constructed definitions. This means that they do not deal with the objective nature of a system of concepts, but have as a starting point the concepts themselves, as shaped by human perception in each case. Because of the subjectivity of human perception, there can be many equally plausible ontologies (Smith and Mark 2001, 594). In archaeology, as in many other scientific fields, different practices and attitudes to research methodologies may lead to the perception of different elements as information or data, resulting in a variety of conceptual classifications.

5.2.‘Where?’ or ‘spatial representation’ The management of the spatial characteristics of a geographic entity requires a set of geometric structures that facilitate graphical display. However, these structures in digital cartography do not comprise a single spatial representation model. Rather, there are two distinct methods in widespread use, i.e. discrete and continuous, that correspond to vector and raster or grid models respectively. They differ in how they store and visualize the spatial location of real world entities (Conolly and Lake 2006, 24). The graphical output of any spatial entity could follow either of these representation models. This creates a problem for the selection of an appropriate spatial model for graphical representation of a geographical entity in a GIS. As reported by Daly and Lock (1999, 290), the choice of the appropriate spatial model affects all subsequent analysis.

Ontologies do not impose a specific vocabulary. Rather, they act as thesauri or concise dictionaries, allowing researchers to be more precise in their semantic definitions by adapting their vocabulary to the respective concepts contained in an ontology. In this way, research freedom is not directly affected, while the attempt to link archaeological knowledge is transferred to a more abstract level. According to Zhang et al. (2002, 214) an effective ontology should be explicit, (i.e. presenting a clear definition of the types and constraints used), formal, (i.e. machine readable) and shared (i.e. describing generally accepted cognitive categories). The integration of established ontologies in the process of conceptual modelling can relocate the excavation practice in a more structured analytical region. Thus, it is possible to create more expressive and semantically correct data models, which when combined with advanced software implementation solutions can exploit the structure of thematic information and link it to the spatial characteristics of a GIS.

Vectors are mainly used to display distinct entities which have clear boundaries and can be described. In archaeological research anything that can be displayed in a traditional map or a line drawing can be digitized in vector form. Every single vector object can be identified uniquely and linked to records in a table that contain the descriptive characteristics of the distinct geographical entity that it represents. In this sense, it is understood that the inherent problem of a vector model is the need to determine the exact boundaries of an archaeological object. This can be misleading, as in many cases such

Cripps et al. (2004) provide a sound approach to information modelling using the Committee for Documentation of the International Council of Museums57



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 clearly identifiable boundaries do not exist or can be argued over (Conolly and Lake 2006, 24-30). Ǽ73: information object (drawing) Ǽ38: image (picture)

P67: is referred to by

Ǽ62: string (comment…) Ǽ55: type (pendant)

P3: has note

P2: has type

Ǽ57: material Ǽ3: P2: has type Ǽ55: type (stone) Condition (intact) state Ǽ53: place P45: (stratigraphic layer) P53: has consists current or of former P122: borders location with P44: Ǽ53: place (excavation unit) has condition Ǽ73: information object P53: (GIS object) has current or former location

Ǽ42: identifier P1: is P67: is referred to by (SF2345) identified P2: has type by Ǽ19: physical P87: is Ǽ52: time-span object identified Ǽ55: type (12/08/06) (spatially) (point) P140: was attributed P4: has P87: identifies by time span (spatially) P14: carried Ǽ13: attribute Ǽ39: actor Ǽ47: spatial coordinates out by assignment (excavator) (x:3247, y:27456, z:60.57)

Figure 1 An example of semantic modelling using the CIDOC-CRM notions.

This discussion is particularly relevant with respect to 3D space. The limitations of archaeological recording in 2D are widely acknowledged (Conolly and Lake 2006, 38). Archaeologists often express their desire to develop the means to represent the archaeological site in both the horizontal and vertical axes, as this novelty is believed to play a significant role towards the better understanding of the complex stratigraphic relationships. Although the management of spatial data in existing commercial 2D GIS is effective, the goal of threedimensional functionality is a complex issue, since there are persistent limitations in the geometric construction of 3D objects, 3D spatial data management (e.g. 3D topology), and 3D spatial analysis. The complexity of these issues in combination with the diversity of approaches in different scientific fields has hindered the development of an integrated cross-disciplinary method. This situation suggests that spatial representation forms are strongly correlated with particular conceptual issues in each discipline (Zlatanova et al. 2002). In this respect, the effectiveness of the 3D analytical capabilities of GIS in archaeology depends largely on the chosen model of spatial representation. The problem lies mainly in the visualization of volumetric data, e.g. the archaeological deposits. The approaches proposed so far in the display of volumetric data in GIS are extremely cumbersome and overly simplifying (Stoter and Zlatanova 2003).

The representation of phenomena or spatial objects with more vague boundaries is facilitated using the raster model, where space is considered as a continuous field and every position is determined by a value related to one or more attributes. The structural component of every raster is the pixel which can have variable sizes, be regular or irregular in shape and refer to two or more dimensions. Pixels do not overlap and occupy the entire space of a mapped region. In archaeology, rasters are used primarily for displaying the distribution of spatial phenomena, such as altitude, or the dispersion of objects in space (Conolly and Lake 2006, 24-30).

In recent years, however, 3D representation of excavation deposits in GIS is becoming an important research question (Barceló et al. 2003; Barcelȩ and Vicente 2004; Harris and Lock 1996; Zhukovsky 2002). The solutions proposed refer to both the vector and raster model (Abdul-Rahman and Pilouk 2008; Raper 2000, 145-158). Cattani et al. (2004) and more recently Losier et al. (2007) provided a detailed discussion on methods of manufacturing three-dimensional excavation deposits (Figure 2).

Couclelis (1992) has transferred this issue to the realm of spatial cognitive understanding stating that vector and raster representations reflect the dual perception of space as a set of discrete entities or as a continuous field made of locations with varying properties. In the first case, space is broken down into discrete elements which are interpreted as objects of interest. The second describes phenomena unfolding within a space-time continuum. Because both views are based on cognitive mechanisms by which people perceive reality, the choice of the correct way to visualize a phenomenon of a GIS depends heavily on the intended function of the information system (Peuquet 2002, 269-270).

Many researchers promote Volume Representations using voxels as the most suitable format for representing volumetric entities or continuous phenomena, such as the stratification of an excavation site (Bezzi et al. 2006; Cattani et al. 2004; Lieberwirth 2008). In general, 3D grids present the same characteristics as their 2D equivalent. Value differences can be shown with colour gradients or other visual variables, such as transparency and contrast. They support topological and distance queries as well as 3D map algebra. However, they still present limitations in multiple attribute relations.

Most applications include both approaches, since specific aspects of reality have a direct correlation with both perceptions of space. The development of a rich GIS application with respect to spatial perception requires conceptual models informed by the scope of the application. This parameter implies that the configuration of the basic spatial building blocks of an application results primarily from the thematic dimension of the data, i.e. the way reality is broken down into distinct entities rather than the limitations of each spatial representation model (Peuquet 2002, 269-270).

A different approach to the representation of discrete entities uses Boundary representation, or B-reps. There is a variety of b-reps including point clouds, wireframes, surface and volumetric models. Their differences lie in the effectiveness of the representation by incorporating vertices, edges, faces, and volume respectively. Value differences can be displayed using colour, texture or shading and further visual variables such as transparency 58

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations or contrast. They are very efficient in the visualization of complex and irregular objects and effective in their linkage with multiple attributes through an appropriate geo-relational system (Apel 2004).

distribution in space, in relation to deposition or use patterns. The preference of archaeologists for representing excavation deposits using 3D grids is related to the treatment of archaeological layers as a continuous phenomenon of deposit accumulation that evolves over time and the availability of tools that support voxel generation and representation. However, this idea is contradicted by the fact that excavation deconstructs this phenomenon in the present in individual observations units (see also Cripps and May 2010; Llobera 2010). In the context of excavation, the purpose of GIS implementation is the linkage of the information collected for an observation unit, either a deposit or a find, with the spatial interface of this unit. The reduction of individual observation units to a phenomenon comes as a result of the interpretation process, through the study of the relationships between them. Therefore, the discrete model is better suited for the graphical representation of excavation data, whereas the continuous model can be used in final stages, to link the results of archaeological research procedures to events that refer to the past.

Figure 2 3D representation of excavation deposits using (a) a surface b-rep model and (b) a 3D grid. In both cases, shape accuracy is directly related to the accuracy of field measurements. The characteristics of each type of spatial representation show that in 3D space both models offer applicable approaches to human spatial perception (verifying the observations made by Couclelis 1992 with respect to 2D space). Moreover, both approaches are equally feasible, but with individual differences in the functionality they support. Therefore, the choice of the best approach for the purposes of an excavation application is related to issues of spatial perception.

Following the above reasoning in the case of Paliambela Kolindros the basic excavation spatial entities that have to be represented in 3D were defined as shown in Figure 3. Each entity acts as the recovery framework of the thematic and temporal information captured in the field and stored in the database. This way, the spatial reference of any recorded thematic or temporal information can be achieved.

Frank notes (2003) that human cognitive perception is effective because it manages through an amorphous pattern of sensory stimuli to separate out independent entities that present an interest to the perceiver. According to Gibson (1986), the identification of affordances in the environment is subtle and involves the division of reality into conceptually distinct entities. This is achieved through experience and the gradual classification of reality through linguistic and cognitive structures.

5.3. ‘When?’ or ‘temporal reasoning’ From the mid 20th century onwards, many geographers have attempted to describe dynamic phenomena using innovative cartographic representations. The work of Hägerstrand (1967) in mapping human motion via a single spatiotemporal mapping framework was especially influential and is often referred to as Time Geography. This initiative was followed by several studies attempting to portray dynamic phenomena in analogue maps using different applied examples. Bertin (1967) organized the possible methods of depicting dynamic events in three categories: a) using a range of maps pertaining to successive temporal moments, b) indicating the temporal path of a phenomenon and c) using visual variables, such as colour, texture, etc., to distinguish between objects with different temporal attributes in a static map.

During the excavation event, interpretation ‘at the trowel’s edge’ (Hodder 1999) suggests that the interaction of an archaeologist with the terrestrial matrix requires the decomposition of excavation space into objects of interest (or observation units). These objects of interest are formulated in the present on the basis of conceptual classifications established by the research scope and the individual experience of each archaeologist. Their association with episodes of the past takes place in latter stages of archaeological study.

The advent of digital cartography led to new ways of mapping dynamic phenomena that aimed to create functioning temporal GIS (TGIS) applications. Langran (1992) presented one of the first documented attempts to link spatial and temporal data using either map snapshots or temporal zones. Different methods used animation techniques (Harrower 2004; Koussoulakou and Kraak 1992) and other innovative approaches, such as the spacetime cube (Gatalsky et al. 2004; Kraak and Koussoulakou

For example, the definition of an excavation context is primarily a cognitive act in the present. Whether this context actually corresponds to a single episode of deposition in the past has to be decided upon through the correlation of all available evidence. In other words, during excavation the material objects must be identified as distinct entities and segregated by the surrounding matrix. Only then it is possible to understand their

59

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 rather than trying to specify an arbitrary temporal scale (i.e. real dates) to which all evidence should be attributed with a certain level of certainty using probabilistic methods, archaeologists first assess the chronology of available evidence and then decide on the temporal resolution to which they can base all subsequent interpretations. In this respect, we should also focus on developing qualitative ways of temporal reasoning that help adjust the level of resolution in temporal assessments through the temporal correlation of available evidence.

2005) or the use of voxels (Crema in press) and taxels (Huisman et al. 2009) models.

The incorporation of time into an archaeological GIS is directly associated with the interpretative use of temporal reasoning. Although temporal reasoning has only recently been given a theoretical focus in archaeology (Gosden 1994), already many researchers have stressed the fact that the linear, objective and irreversible notions of time should be replaced by more flexible approaches that incorporate the concepts of different time-scales (Bailey 1983) and the prospect of non-linear trajectories (McGlade 1999). These approaches highlight the idea of the multi-temporality of the archaeological material. Olivier (2004, 206) notes that the urban environment is a palimpsest, i.e. a collage composed of the traces left by different occupation episodes, each with different duration and different effects on the final deposit formation. This idea bears many resemblances to a snapshot view of the world comprising entities with different chronology.

Figure 3 Different types of excavation entities and their respective visualization: a) excavation units as 3D closed b-rep objects, b) stratigraphic units as 3D surfaces (breps), c) drawings as 3D lines, d) features as 3D surfaces (b-reps), e) finds as 3D points and f) samples as 3D points. Despite the progress in real time data management systems, such as monitoring vehicle movement, most GIS applications associated with historical data remain deficient in their organization of temporal information and their display of temporal change (Beser de Deus and Ferreira da Silva 2005). These persistent limitations relate to differences in approach to the concept of time in different disciplines, so it is not possible to provide a generalized model that applies in all cases (Frank 1998).

In addition, it is implied that every archaeological object has a unique biography that begins from the time of construction and runs across multiple contexts depending on its role as a utilitarian or symbolic object, as part of a deposition or as a single find. In each case, different correlations that modify the meaning, the usage, or even the material of this find (e.g. secondary use) are involved. Holtorf (2002) extends this debate by arguing that all characteristics of an archaeological object, including its age, are the result of interpretations made in the present.

Archaeology is a promising field for TGIS applications, with the potential to create alternative representations of time. Originally, Castleford (1992) provided a thorough review of research in multidimensional GIS to be followed by Daly and Lock (1999). They note that, contrary to the majority of TGIS applications concerned with dynamic spatiotemporal phenomena evolving in the present using precise temporal measurements, archaeology requires a different treatment of the temporal dimension. Current archaeological approaches to time are difficult to implement in a GIS, because time comprises secondary information derived through the study of excavation data and is characterised by uncertainty. Recent studies (Green 2009) have made significant contribution using probabilistic modelling for recording and analysing time-series data such as radiocarbon dates and dendrochronology readings. Such attempts open new paths in the analysis of temporal data patterns.

In an excavation things can get more complicated, as the development of techniques that focus on the microstratigraphic level have showed that what can be seen as a homogeneous deposit in the field, may actually contain multiple depositional episodes at the micro-level (Barham and MacPhail 1995; Lucas 2001a, 152-159). Schiffer (1987, 266) has suggested that a single depositional event can result in the creation of multiple depositional layers, while a single layer can be associated with multiple depositional events. These observations suggest that every excavation action defines a spatial and physical entity that takes its temporal characteristics from its relationships with other equally demarcated entities. In this sense, the excavation process does not define physical entities that correspond to individual events in the past, but creates heterogeneous groups that refer to past temporal intervals. The accuracy involved in determining this interval (i.e. whether a deposit covers a range of, say, 700 or 100 years) depends

However, the incorporation of time into an archaeological excavation GIS is primarily associated with the interpretative use of temporal reasoning. In practice, 60

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations different methods of temporal reasoning. To support effective temporal reasoning, an excavation database must incorporate multiple temporal values for each archaeological object and include operations that handle chronological sorting and temporal querying.

on the degree of temporal granularity that characterizes these groups. The potential gradients result from the individual temporality of each and every content of the deposit as well as the chronological correlation between them (Lucas 2001a, 156-162). However, temporal relationships among the contents of a deposit are related to the biography of each object. In this sense, the use of absolute dating as evidence for the chronology of an entire deposit, can lead to erroneous estimates, as it is related to the construction of the object and not its use or its deposition. If the remaining contents support a later dating, it is probable that the find was used for a longer period of time until it ended up in the same deposit as the rest. This aspect reinforces the idea that every excavation object is defined as an entity in space and time in relation to a particular context of reference. These theoretical considerations have to be properly conceptualised, if they are to be integrated in a GIS. For example, the difference between relative and absolute dating lies essentially at the scale of measurement. Ramenofsky (1997) argues that absolute dating coincides with the interval scale (3500 BC, 4500 bp, 12/05/2008), while relative dating is equated to the ordinal scale of measurement (older, newer). Frank (1998) in an analysis of different temporal types associated with specific temporal conceptualizations depending on the scope of research, notes that in the case of interval time two concepts can be identified that are related to the ideas of continuity and discretization (Figure 4).

Figure 4 Interval-scale types of archaeological time and their effect on temporal reasoning. MN stands for Middle Neolithic and LNI for Late Neolithic I.

The first considers time as continuous in size and evolving linearly. In between two events another can always be inserted. This time corresponds to temporal measurement in absolute terms using small units, such as dates. This time is typical of the time of an excavation. The second considers time as a discrete dimension containing variable resolutions. Time is shaped through successive temporal units that subdivide it. In this approach, the correspondence of two events with the same temporal unit renders them contemporary. This time is typical of archaeological chronological classification in periods, sub-periods and phases (Lock and Harris 1997, 4), as well as absolute dating, implying a margin of error and thus a temporal interval (Ramenofsky 1997).

Figure 5 Ordinal archaeological time on the basis of stratigraphic information. The succession of deposits (e.g. ǹ and Ǻ) from below suggests transition to the more recent ( (Is Later Than) Layer Y

Cultural Archaeological chronological time framework

interval

Late Neolithic

Excavation Site phase time chronological framework

interval

Phase IV

interval

6730 bc +/150 years

Database time Creation time Stratigraphic time

On the whole, time is translated into an attribute of every archaeological entity (i.e. a feature) defined during excavation and every interpretive entity formed during study (i.e. a house). In other words, the excavation universe is made up of separate 3D entities with temporal characteristics of finite duration that operate in alternative contexts of reference. In this way of thinking, archaeological entities are static in the sense that they do not incorporate spatial or thematic variation across time. To illustrate temporal variation between entities an emphasis is placed on their spatial, taxonomic and part-whole relationships (Harris and Lock 1996, 291-2; Daly and Lock 1999). Depending on the location of an entity (e.g. a ceramic bowl) and its relation to neighbouring features (e.g. a hearth) or those that surround it (e.g. an archaeological layer), a series of thematic (e.g. food preparation) and spatial correlations (top, bottom, side, NW, insects) are created that can be traced back to temporal relations (later, earlier or simultaneous) (Lock and Harris 1997, 4).

Absolute time

Relative temporal distinction

Absolute chronology

Figure 6 Temporal categories recorded as timestamps. All or some of these temporal paths apply to different excavation objects depending on the interpretive objectives.

The utilization of the third dimension for the display of excavation data excludes many of the proposed temporal visualization approaches, such as the space-time cube. The representations that can be used in a 3D environment are confined to the display of snapshots and the use of animation. Both do not match the multi-linear nature of archaeological time, as they refer to successive states organized in a single timeline.

This framework is provided by the Pyramid model that builds on the spatio-temporal trinity of When?, Where? and What? questions discussed earlier (§5). According to this model, an entity (e.g. an excavation unit) can be represented as a discrete object in space (3D geometry), refers to one or more temporal intervals (e.g. it was excavated on 06/15/1908 and was originally formed during the Middle Neolithic) and is characterized by observable variables (e.g. colour, moisture, texture). In the first level, the overall characteristics are primary observations. In the context of interpretative reasoning the same entity can form part of broader groupings (e.g. Layer 15 of phase III) or participate in some taxa (e.g. observation unit).

Therefore, the model of a temporally unified static or atemporal map can be considered as the most appropriate for exploratory temporal data analysis in a cartographic environment, since it allows the inclusion of objects with similar or even seemingly unrelated temporal characteristics in complex representations. In such an

The pyramid conceptual representational framework is a complex theoretical model to implement (see Mennis et al. 2000), but it is more enlightening as to the way human mental representations of geographical reality are shaped. In the case of excavation research, it conforms to archaeological interpretive reasoning that determines the 62

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations definition of an archaeological entity through its thematic, spatial and temporal parameters. Its use can facilitate a formalized description of the excavation universe and support the composition of new information during post-excavation study. Its potential functionality is manifested with respect to stratigraphic analysis, as it allows the modelling of a hierarchical aggregation structure that reflects the stratigraphic grouping process in successive stages. The creation of a stratigraphic group is in essence new information reached through the interpretation of thematic, spatial, temporal and temporal variables among the available excavation data. The new entity (e.g. the stratigraphic group) is the sum of all components (i.e. excavation units) as well as their thematic, spatial and temporal attributes. The same time its creation corresponds with the inheritance of its characteristics and relationships by its components. The organization of stratigraphic reasoning through an aggregation process allows the re-examination of excavation work and provides a mechanism for the graduated re-assemblage of the fragmented archaeological evidence. In addition, the model organizes the excavation interpretative process by connecting every archaeological entity to the event that produces it, allowing for the separate storage of interpretive remarks and the ability of the researchers to revisit their interpretation, without affecting the primary data stored during fieldwork.

Figure 7 The entity of an excavation unit according to the pyramid model using UML notation.

7. Representing and exploring excavation data across space and time

The implementation of this model in GIS favours objectoriented software development methods, such as those offered by the Unified Modelling Language - UML (see Dennis et al. 2002). These allow the analysis of a chosen application domain to be followed by the incremental and iterative building of the conceptual and the implementation data model through a series of diagrammatic depictions (Figure 7). UML notation provides increased expressiveness in the representation of the excavation universe, allows for the communication of the data modelling process to the project’s participants, and facilitates the identification of patterns within the data model. Its use from the early stages of a project facilitates object-oriented data modelling procedures resulting to conceptual models that are adaptable to a range of particular software solutions.

Access to all excavation data is provided by the native 3D display environment of ESRI’s ArcScene (see also Katsianis et al. 2008). At this functional level, the focus can be shifted to facilitating navigation in 3D space and visual exploration of archaeological data. The spatial interface of each excavation entity can be represented in 3D using standard cartographic solutions in the form of point, line, surface or volumetric theme layers. The representation of each theme can be limited to the objects that follow specific spatial, temporal and thematic constraints set by the users using definition queries, e.g. display the excavation units of trench 5 that belong to layer XIV. Emphasis should be placed on helping the user navigate in 3D space by providing ways of understanding his/her viewpoint, the relative size of features and objects, as well as their orientation. This oriented context can be provided by additional tools such as a complementary 2D interface or 3D cartographic legend items, i.e. north arrow, 3D scale. After all, there is nothing more frustrating than showing a full 3D display of the contents of a trench to its excavator only to be rewarded with the request to switch the viewpoint to vertical, so that he/she can orientate him/herself better!

This pyramid-style representation framework formed the conceptual basis for the advancement of the excavation data model for Paliambela Kolindros (see Katsianis et al. 2008, fig. 3, 4 and 5). The model was complemented by the CIDOC-CRM semantic description of every component through the use of stereotypes and tagged values. Its use revealed a number of inconsistencies in the conceptual definition of entities, mainly through the identification of missing events, and allowed timely corrections. Gradually, the conceptual data model was finalised and implemented as a georelational data schema in an ArcGIS geodatabase.

To optimize visual output and enhance visual discrimination of the displayed spatial objects, particular retinal variables can be utilized for each theme (Bertin 63



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 of Paliambela Kolindros within a custom cartographic environment developed specifically for archaeological use (Tsipidis 2009).

1967). Using standard classification methods, the spatial objects can be used to explore all types of information contained in the database (Figure 8). Additionally, most GIS provide advanced capacities for visualization including advanced symbologies, 3D labels, etc. Realism in representation can be greatly enhanced by draping linear drawings and stratigraphic section faces with raster orthophoto images (Figure 9).

Overall, the combination of all components of an excavation record into a single information system contributes to better management, access and collation of information. In particular, the representation of thematic, spatial and temporal queries, allows us to correlate finds with their respective excavation contexts, facilitating the reunification of data that were recorded and stored separately after initial recovery. The combination of querying capabilities, mapping techniques and symbologies that enhance the presentation of interpretive results extends the potential contribution of the computer system to all stages of research.

3D Excavation spatial entities

Colour

Point

Line

Surface

Volume

N

N

N

N

N

NO

Texture Size

OI

I

8. Conclusions

Shape

N

N

Tone

O

GIS gradually demonstrate their potential for intra-site archaeological research, since they can provide interactive environments that facilitate inductive reasoning in a similar manner that archaeological interpretation is envisioned to take place from a highly reflexive, post-processual perspective.

OI

Figure 8 The effectiveness of retinal variables for the display of relationships among the modelled excavation entities. Relationship types are denoted in each cell (N for nominal, O for ordinal and I for interval scale relationships). Inappropriate usage is denoted by crossed cells (after Tsipidis 2009).

The crucial factor for the constructive use of digital technology is to avoid the creation of yet another ‘black box’ that increases our distance from reality and undermines the interpretive significance of the excavation moment. If the employment of digital technology remains consistent with archaeological theory and excavation methodology, it can truly be a means to expand our range of knowledge production strategies. In this sense, GIS technology does not break the flow of archaeological reasoning. On the contrary, it can enhance it, if properly employed. The proper way to do so is directly related to the referral capacities in between the multiple transformations of archaeological data from the field to the computer screen.

On-screen selection facilitates attribute viewing, including all of stratigraphic relationships in which the object participates. Complex queries using thematic, temporal and spatial data attributes can be formulated resulting in the selection of the corresponding objects within the map frames (Figure 10). Stratigraphic analysis can take advantage of 3D display and collate excavation units with actual recorded stratigraphic sections, so as to discover discrepancies and possible mistakes that can be taken into account during the formulation of a final stratigraphic sequence. In this sense, the users are not restricted to set investigation procedures, but they can interactively examine all available data, review, find mistakes, create alternative stratigraphic sequences and test various hypotheses (Figure 11).

This issue has been addressed in three ways in the case study outlined above. First, the techniques used for the digital transcription of reality (i.e. the breaking-up of the excavation universe into well defined units of observation), the discrete model of spatial display and the use of multiple temporal variables are directly related to an individual’s perception and mental organization of archaeological information in the process of excavation.

Of course, a number of issues require special attention in terms of 3D GIS functionality. These can be summarized in stratigraphic analytical tools that make use of 3D topology as well as 3D spatial statistics, two fields that only recently have begun to develop (Anderson and Burke 2008). However, at present there is the possibility to create custom tools that take advantage of the data model capacities and 3D visualization. Among these are tools that implement stratigraphic aggregation, 3D distance queries, 3D custom section generation within excavation space and chronological querying procedures using interactive graphs (Katsianis et al. 2008). Most of these tools have been implemented for the case study site

Secondly, the semantic description of the terms contained in the data model makes it easy to link archaeological information to the individual interpretation processes that have produced it, allowing the mapping of all intermediate interpretative steps within the data structure. Thirdly, the distinction between primary and secondary interpretations allows for an ongoing review of the latter whenever new data appear, defining part-whole relationships in an efficient and flexible manner.

64

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations

Figure 9 Navigation and orientation in 3D using an overview map and viewing angle. 3D cartographic elements, such as 3D extent, scale, north arrow and 3D labels facilitate map reading in 3D space.

Figure 10 Visual exploration of excavation data. The example illustrates the difference in texture values between the interior and exterior of a Neolithic house (wall foundations shown in yellow with a degree of transparency) suggesting that it collapsed mostly inwards. Further possibilities for the examination of find spatial patterning employ onscreen selection and object identification.

65

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process The goal of functional 3D GIS environments is beginning to take shape and a growing number of archaeological excavation projects incorporate some form of GIS data management and adopt recording methodologies that target 3D representation. In the case study outlined above the implementation of an expressive spatio-temporal representation data model for GIS has made possible the full use of current computational possibilities in excavation recording, information collation, exploratory data analysis in 3D, and data-led interpretation that render the excavation archive dynamic and open to continuous re-evaluation.

account established cognitive practices within archaeological research, future work in GIS can play an important part in archaeological theory building. In any case, the integration of digital technology into excavation documentation should evolve around the intended cognitive outcome of the excavation process and the overall methodology that governs it. The creation of excavation data archives is not an end in itself. On the contrary, the success of an archaeological project depends on the methods it utilizes to translate information into effective knowledge, and therefore, as noted by Hourmouziadis (1999) into a ‘social good’ in the present. In this sense, the effective introduction of new technologies in excavation is directly related to their role in reassessing and improving the ways we interact with reality to create knowledge about the past.

Besides the practical advantages of GIS usage in intrasite research, every attempt to employ digital technology in the archaeological process provides an opportunity to reflect on our own theoretical constructs and fieldwork practices. As long as research moves towards new applications through representation models that take into

aa 

c

b b 

d

Figure 11 Visual comparison of excavated contexts with actual stratigraphy as recorded in the trench section. a) excavated contexts, b) pottery chronology, c) texture values, d) aggregated contexts after stratigraphic analysis.

Acknowledgements and his valuable comments on previous versions of this paper. Finally, I would like to thank the two anonymous referees of the present edition for their useful remarks.

The work presented here is the outcome of a research project funded by the EU and the Greek Ministry of Development (Operational Programme ‘Competitiveness’, Priority Axis 8: Human resources, Measure 8.3: Human resources in Research and Technology). I am thoroughly indebted to my friend and colleague Spyros Tsipidis for his contribution in the formulation and implementation of these ideas during our PhD research. I would also like to thank Kostas Kotsakis, Alexandra Kousoulakou and Yiannis Manolopoulos for their guidance in archaeological theory, geovisualization and spatiotemporal data structures. Finally, a great thanks to Andy Bevan for his active participation in the project

References Abdul-Rahman, A. and Pilouk, M. 2008. Spatial data modelling for 3D GIS. Berlin, Springer-Verlag. Adams, W. Y. 1988. Archaeological classification: theory versus practice. Antiquity 62, 40-56.

66

 

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations Adams, W. Y. and Adams, E. Archaeological typology and practical Cambridge, Cambridge University Press.

Beck, A. and Seif, A. 2006. Computer-aided reflexivity and data management in archaeology, in Campagna, M. (ed.) GIS for sustainable development, 367-83. Boca Raton, CRC Press.

1991. reality.

Allen, J. F. 1983. Maintaining knowledge about temporal intervals. Communications of the ACM 26 (1), 832-43.

Bertin, J. 1967. Sémiologie graphique. Paris, Editions Gauthier-Villars.

Anderson, K. L. and Burke, A. 2008. Refining the definition of cultural levels at Karabi Tamchin: a quantitative approach to vertical intra-site spatial analysis. Journal of Archaeological Science 35 (8), 2274– 85.

Beser de Deus, L. A. and Ferreira da Silva, L. F. C. 2005. Conceptual basis for time usage in GIS, in XXII international cartographic conference (ICC2005), A Coruña, Spain.

Andrews, G., Barrett, J. C. and Lewis, J. S. C. 2000. Interpretation not record: the practice of archaeology. Antiquity 74, 525-30.

Bezzi, A., Bezzi, L., Francisci, D. and Gietl, R. 2006. L’utilizzo di voxel in campo archeologico. Geomatic workbooks 6.

Apel, M. 2004. A 3D geoscience information system framework, Unpublished PhD thesis, Freiberg University.

Brittain, M. and Clack, T. 2007. Introduction: archaeology and the media, in Brittain, M. and Clack, T. (eds) Media and archaeology, 11-65. Walnut Creek, Calif, Left Coast Press.

Arroyo-Bishop, D. and Lantada Zarzosa, M. T. 1995. To be or not to be: will an object-space-time GIS/AIS become a scientific reality or end up an archaeological entity?, in Lock G. R. and Stancic, Z. (eds) Archaeology and geographic information systems: A European perspective, 43-53. London, Taylor and Francis.

Carver, G. 2005. Archaeological Information Systems (AIS): adapting GIS to archaeological contexts, in International congress of cultural heritage and new technologies - Workshop 9 "Archäologie und Computer". Wien, Phoibos Verlag, (CD-ROM).

Bailey, G. N. 1983. Concepts of time in quaternary prehistory. Annual review of Anthropology 12, 165-92.

Castleford, J. 1992. Archaeology, GIS and the time dimension: an overview, in Lock, G. and Moffett, J. (eds) computer applications and quantitative methods in archaeology 1991, 95-106. Oxford, BAR Publishing.

Barceló, J. A., De Castro, O. Travet, D. and Vicente, O. 2003. A 3D model of an archaeological excavation, in M. Doerr, M. and Sarris, A. (eds) The digital heritage of archaeology. Proceedings of the 30th computer applications and quantitative methods in archaeology conference, Heraklion, Crete, April 2002, 85-7. Athens, Hellenic Ministry of Culture.

Cattani, M., Fiorini, A. and Rondelli, B. 2004. Computer applications for a reconstruction of archaeological stratigraphy as a predictive model in urban and territorial contexts, in Ausserer, K. F. Börner, W. Goriany, M. and Karlhuber-Vöckl, L. (eds) Enter the past. The e-way into the four dimensions of cultural heritage. CAA 2003, computer applications and quantitative methods in archaeology, 299-303. Oxford, BAR Publishing.

Barceló, J. A. and Pallares, M. 1996. A critique of GIS in archaeology. From visual seduction to spatial analysis. Archeologia e Calcolatori 6, 313-326. Barcelȩ, J. A. and Vicente, O. 2004. Some problems in archaeological excavation 3D modelling, in Ausserer, K. F. Börner, W. Goriany, M. and Karlhuber-Vöckl, L. (eds) Enter the past. The e-way into the four dimensions of cultural heritage. CAA 2003, computer applications and quantitative methods in archaeology, 400-4. British Archaeological Reports International Series. Oxford, BAR Publishing.

Chadwick, A. 2003. Post-processualism, professionalization and archaeological methodologies. Towards reflective and radical practice. Archaeological Dialogues 10 (1), 97-117.

Barham, A. and MacPhail, R. (eds) 1995. Archaeological sediments and soils: analysis, interpretation and management. London, Left Coast Press.

Couclelis, H. 1992. People manipulate objects (but cultivate fields): beyond the raster-vector debate in GIS, in Frank, A. U., Campari, I. and Formentini, U. (eds) Theories and methods of spatio-temporal reasoning in geographic space, 65-77. Berlin, Springer-Verlag.

Conolly, J. and Lake, M. 2006. Geographical information systems in archaeology. Cambridge, Cambridge University Press.

Barker, P. 1982. Techniques of archaeological excavation. London, Batchford.

Crema, E. R. in press. Aoristic approaches and voxel models for spatial analysis, in Jerem, E., RedĘ, F. and Szeverényi, V. (eds) On the road to reconstructing the past, Proceedings of the 36th annual conference on computer applications and quantitative methods in

Bateman, J. 2000. Immediate realities: an anthropology of computer visualisation in archaeology. Internet Archaeology 8.

67



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 archaeology. CAA2008. Budapest, 2-6/4/2008. Budapest, Archaeolingua.

Green, C., 2008. Winding Dali’s clock: the construction of a fuzzy temporal-GIS for archaeology, Unpublished PhD thesis, School of Archaeology and Ancient History. Leicester, University of Leicester.

Cripps, P., Earl, G. and Wheatley, D. 2006. A dwelling place in bits. Journal of Iberian Archaeology 8, 25-39. Cripps, P. and May, K. 2010. To OO or not to OO? Revelations from ontological modelling of an archaeological information system, in Niccolucci, F. and Hermon, S. (eds) Beyond the artefact - Digital interpretation of the past. Computer applications and quantitative methods in archaeology (CAA). Proceedings of the 32nd international conference, Prato, Italy, 13-17/4/2004, 57-61. Budapest, BAR Publishing. Daly, P. and Lock, G. 1999. Timing is everything: commentary on managing temporal variables in geographic information systems, in Barceló, J. A., Briz, I. and Vila, A. (eds) New techniques for old times CAA98. Computer applications and quantitative methods in archaeology, 287-93. Oxford, BAR Publishing. Dennis, A., Wixom, B. H. and Tegarden, D. 2002. Systems analysis and design: an object-oriented approach with UML. New York, Wiley. DiBiase, D., MacEachren, A. Krygier, J. B. and Reeves, C. 1992. Animation and the role of map design in scientific visualization. Cartography and geographic information science 19 (4), 201-14. Ford, J. A. 1954. Comments on A. C. Spaulding, statistical techniques for the discovery of artifact types. American Antiquity 19, 390-1. Frank, A. 1998. Different types of "Times" in GIS, in Egenhofer, M. and Golledge, R. G. (eds) Spatial and temporal reasoning in geographic information systems, 40-62. Oxford, Oxford University Press. Frank, A. 2003. Ontology for spatio-temporal databases, in Sellis, T., Koubarakis, M., Frank, A., Grumbach, S., Gueting, R. H., Jensen, C., Lorentzos, N., Manolopoulos, Y., Nardelli, E., Pernici, B., Schek, H. J., Scholl, M., Theodoulidis, B. and Tryfona, N. (eds) Spatiotemporal databases: the chorochronos approach, 7-63. Berlin, Springer Verlag. Gatalsky, P., Andrienko, N. and Andrienko, G. 2004. Interactive analysis of event data using space-time cube, in Proceedings of international conference on information visualisation (IV’04), London, 145-52. Gibson, J. J. 1986. The ecological approach to visual perception. Hillsdale (NJ), Lawrence Erlbaum Associates. Gidlow, J. 2000. Archaeological computing and disciplinary theory, in Lock, G. and Brown, K. (eds) On the theory and practice of archaeological computing, 23-30. Oxford, Oxford University Committee for Archaeology.

68

González Pérez, C. A. 2002. Sistemas de información para arqueología: teoría, metodoloxía y technoloxías. Oxford, BAR Publishing. Goodchild, M. F. 1996. Geographic information systems and spatial analysis in the social sciences, in Aldenderfer, M. S and Maschner, H. D. G. (eds) Anthropology, space, and geographic information systems, 241-50. Oxford, Oxford University Press. Gosden, C. 1994. Social being and time. Oxford, Blackwell. Guarino, N. and Giaretta, P. 1995. Ontologies and knowledge bases: towards a terminological clarification, in Mars, N. (ed.) Towards very large knowledge bases, 25-32. Amsterdam, IOS Press. Hadzilakos, T., and Stoumbou, P. M. 1996. Conceptual data modelling for prehistoric excavation documentation, in Kamermans, H. and Fennema, K. (eds) Interfacing the past: computer applications and quantitative methods in archaeology 1995, 21-30. Leiden, University of Leiden. Hägerstrand, T. 1967. Innovation diffusion as a spatial process. Chicago, University of Chicago Press. Harris, E. C. 1979. Principles of archaeological stratigraphy. London and New York, Academic Press. Harris, T. M. and Lock, G. 1995. Towards an evaluation of GIS in European archaeology: the past, present and future of theory and applications, in Lock, G. and Stancic, Z. (eds) Archaeology and geographical information systems: a European perspective, 349-65. London, Taylor and Francis. Harris, T. M. and Lock, G. 1996. Multi-dimensional GIS: exploratory approaches to spatial and temporal relationships within archaeological stratigraphy, in Kamermans, H. and Fennema, K. (eds) Interfacing the past: computer applications and quantitative methods in Archaeology 1995, 307-16. Leiden, University of Leiden. Harrower, M. 2004. A look at the history and future of animated maps. Cartographica 39 (3), 33-42. Hodder, I. 1999. The archaeological process: an introduction. Oxford, Blackwell. Holtorf, C. 2002. Notes on the life history of a pot sherd. Journal of Material Culture 7, 49-71. Hourmouziades, G. 1995. Peri tou arhaiologikou mnemeiou genika, in Analogies, 249-60. Thessaloniki, Vanias (in greek).

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations Hourmouziades, G. 1999. Logia Thessaloniki, Skopelos (in greek).

apo

choma.

Latour, B. 1999. Pandora's hope: essays on the reality of science studies. Cambridge (Mass.), Harvard University Press.

Huggett, J. 2000. Computers and archaeological culture change, in Lock, G. and Brown, K. (eds) On the theory and practice of archaeological computing, 5-22. Oxford, Oxford University Committee for Archaeology.

Lieberwirth, U. 2008. Voxel-based 3D GIS - modelling and analysing archaeological stratigraphy, in Frischer, B. and Dakouri-Hild, A. (eds) Beyond illustration - 2D and 3D digital technologies as tools for discovery in archaeology, 78-86. Oxford, BAR Publishing.

Huisman, O., Feliciano Santiago, I. T., Kraak, M.-J. and Retsios, B. 2009. Developing a geovisual analytics environment for investigating archaeological events: extending the space-time cube. Cartography and Geographic Information Science 36 (3), 225–36.

Llobera, M. 2006. The nature of everyday experience: examples on the study of visual space, in Unwin, D. and Fisher, P. (eds) Re-presenting GIS. 127-134. London, John Wiley & Sons.

Johnson, I. 2002. Contextualising archaeological information through interactive maps. Internet Archaeology 12.

Llobera, M. 2010. Archaeological visualization: towards an archaeological information science (AISc). Journal of Archaeological Method and Theory, 18 (3), 193-223.

Kähler Holst, M. 2001. Formalizing fact and fiction in four dimensions: a relational description of temporal structures in settlements, in Stancic, Z. and Veljanovski, T. (eds) Computing archaeology for understanding the past CAA2000. Computer applications and quantitative methods in archaeology, 159-64. Oxford, BAR Publishing.

Lock, G. 1995. Archaeological computing, archaeological theory, and moves toward contextualism, in Huggett, J. and Ryan, N. (eds.) Computer applications and quantitative methods in archaeology 1994, 13-8. Oxford, BAR Publishing.

Kähler Holst, M. 2004. Complicated relations and blind dating: formal analysis of relative chronological structures, in Buck, C. E. and Millard, A. R. (eds) Tools for constructing chronologies: crossing disciplinary boundaries, 129-48. New York, Springer.

Lock, G. and Harris, T. M. 1992. Visualizing spatial: the importance of geographic information systems, in Reilly, P. and Rahtz, S. P. Q. (eds) Archaeology and the information age, 81-96. London, Routledge.

Katsianis, M. 2009. Anaskafiki methodologia ke sxediasmos systimatos diaheirisis arhaiologikon tekmirion, Unpublished PhD Thesis, Aristotle University Thessaloniki (in greek).

Lock, G. and Harris, T. M. 1997. Analysing change through time within a cultural landscape: conceptual and functional limitations of a GIS approach, in Sinclair, P. and Wandibba, S. (eds) Urban origins in eastern Africa. Uppsala

Katsianis, M., Tsipidis, S., Kotsakis, K. and Koussoulakou, A. 2008. A 3D digital workflow for archaeological intra-site research using GIS. Journal of Archaeological Science 35 (3), 655-67.

Losier, L.-M., Pouliot, J. and Fortin, M. 2007. 3D geometrical modeling of excavation units at the archaeological site of Tell ‘Acharneh (Syria). Journal of Archaeological Science, 34, 272-88.

Kotsakis, K. 1983. Kerameiki tehnologia ke kerameiki diaforopoiisi. Unpublished PhD thesis, Aristotle University Thessaloniki (in greek).

Lucas, G., 2001a. Critical approaches to fieldwork: contemporary and historical archaeological practice. London, Routledge.

Koussoulakou, A. and Kraak, M. J. 1992. Spatiotemporal maps and cartographic communication. The Cartographic Journal 29, 101-8.

Lucas, G., 2001b. Destruction and the rhetoric of excavation. Norwegian Archaeological Review 30 (1), 35-46.

Kraak, M. J. and Koussoulakou, A. 2005. A visualization environment for the space-time cube, in Fisher, P. (ed.) Developments in spatial data handling: proceedings of the 11th international symposium on spatial data handling, Leicester, UK (August 2004), 184-200. Leicester, Springer.

MacEachren, A. 1995. How maps work: representation, visualization, and design. New York, Guildford Press.

Kvamme, K. L. 1999. Recent directions and developments in geographical information systems. Journal of Archaeological Research 7 (2), 153-201.

Manovich, L. 2001. The language of new media. Cambridge (Mass), MIT press.

Madsen, T. 1999. Coping with complexity. Towards a formalised methodology of contextual archaeology. Archeologia e Calcolatori 10, 125-44.

McGlade, J. 1999. The times of history: archaeology, narrative and non-linear causality, in Murray, T. (ed.) Time and Archaeology, 139-62. London, Routledge.

Langran, G. 1992. Time in geographic information systems. London, Taylor and Francis.

69



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

 the

Shanks, M. 2007b. Symmetrical archaeology. World Archaeology 39 (4), 589 - 96.

Mennis, J., Peuquet, D. J. and Qian, L. 2000. A conceptual framework for incorporating cognitive principles into geographical database representation. International Journal of Geographical Information Science 14 (6), 501-20.

Smith, B. and Mark, D. M. 2001. Geographical categories: an ontological investigation. International Journal of Geographical Information Science 15 (7), 591-612.

McLuhan, M., 1964. Understanding media: extensions of Man. Cambridge 9Mass, MIT press.

Spaulding, A. C. 1953. Statistical techniques for the discovery of artifact types. American Antiquity 18, 30513.

Niccolucci, F., D'Andrea, A. and Crescioli, M. 2001. Archaeological applications of fuzzy databases, in Stancic, Z. and Veljanovski, T. (eds) Computing archaeology for understanding the past CAA2000. Computer applications and quantitative methods in archaeology, 107-16. Oxford, BAR Publishing.

Stefanakis, E. 2003. Vaseis geografikon dedomenon ke systimata geografikon pliroforion. Athens, Papasotiriou (in greek).

Olivier, L. 2004. The past of the present. Archaeological memory and time. Archaeological Dialogues 10 (2), 20413.

Stoter, J. and Zlatanova, S. 2003. 3D GIS, where are we standing. IPRS joint workshop on spatial, temporal and multi-multi-dimensional data modeling and analysis.

Olsen, B. 2007. Keeping things at arm's length: a genealogy of asymmetry. World Archaeology 39 (4), 579 - 588.

Tatnall, A. 2003. Actor-network theory as a sociotechnical approach to information systems research, in Clarke, S. Coakes, E., Hunter G. M. and Wenn Hershey, A. (eds) Socio-technical and human cognition elements of information systems, 266-83. PA, IGI Publishing.

Peuquet, D. J. 1994. It's about time: a conceptual framework for the representation of temporal dynamics in geographic information systems. Annals of the Association of American Geographers 84 (3), 441-61.

Tomai, E. and Kavouras, M. 2002. Development of geographic ontologies. 2nd panhellenic conference of the Hellenic GIS Society Athens, 28/2-1/3, 2002.

Peuquet, D. J. 2002. Representations of space and time. London and New York, Guildford Press.

Tsipidis, S. 2009. Geo-optikopoiisi chorochronikon arhaiologikon dedomenon, Unpublished PhD thesis, Aristotle University Thessaloniki, (in greek).

Rajala, U. 2004. Sense and sensibility - reflections on the epistemology and ontology of GIS studies. Internet Archaeology 16.

Tudhope, D., May, K., Binding, C. and Vlachidis, A. 2011. Connecting archaeological data and grey literature via semantic cross search. Internet Archaeology, 30. Available at: http://intarch.ac.uk/journal/issue30/tudhope _index.html.

Ramenofsky, A. F. 1997. The illusion of time, in Ramenofsky, A. F. and Steffen, A. (eds) Unit issues in archaeology: measuring time, space, and material, 7484. Utah, University of Utah Press.

Tukey, J. W. 1977. Exploratory data analysis. Philippines, Addison-Wesley Publishing Company.

Ramenofsky, A. F. and Steffen, A. 1997. Units as tools of measurement, in Ramenofsky, A. F. and Steffen, A. (eds) Unit issues in archaeology: measuring time, space, and material, 3-17. Utah, University of Utah Press.

van der Scans, R. 1978. In Vakjes stoppen - over het waarnemen en analysere van der werkelijkheid. Technische Hogeschool Delft, Afdeling der Geodesie.

Raper, J. 2000. Multidimensional GIS: extending GIS in space and time. London, Taylor and Francis.

Webmoor, T. 2007. What about 'one more turn after the social' in archaeological reasoning? Taking things seriously. World Archaeology 39 (4), 563 - 78.

Reilly, P. and Rahtz, S. P. Q. 1992. Introduction: archaeology and the information age, in Reilly, P. and Rahtz, S. P. Q. (eds) Archaeology and the information age, 1-27. London and New York, Routledge.

Wheatley, D. and Gillings, M. 2002. Spatial technology and archaeology. The archaeological applications of GIS. London and New York, Taylor and Francis.

Schiffer, M. B. 1987. Formation processes of the archaeological record. Albuquerque (NM), University of New Mexico Press.

Witmore, C. L. 2004. On multiple fields. Between the material world and media: two cases from the Peloponnesus, Greece. Archaeological Dialogues 11 (2), 133–64.

Shanks, M. 2007a. Digital media, agile design and the politics of archaeological authorship, in Clack, T. and Brittain, M. (eds.) Media and archaeology, 273-289. Walnut Creek (CA), Left Coast Press.

Witmore, C. L. 2005. Multiple fields approaches in the Mediterranean: revisiting the argolid exploration project. Unpublished PhD Thesis, Stanford University. 70

Katsianis: Conceptual and Practical Issues in the Use of GIS for Archaeological Excavations Zhukovsky, M. 2002. Handling digital 3-D record of archaeological excavation data, in Burenhult, G. (ed.) Archaeological informatics: pushing the envelope CAA2001. Computer applications and quantitative methods in archaeology, 431-8. Oxford, BAR Publishing.

Witmore, C. L. 2007. Symmetrical archaeology: excerpts of a manifesto. World Archaeology 39 (4), 546 - 62. Zhang, C., Cao, C., Gu, F. and Si, J. 2002. A domainspecific formal ontology for archaeological knowledge sharing and reusing, in Karagiannis, D. and Reimer, U. (eds) Practical aspects of knowledge management. Proceedings of the 4th international conference, PAKM 2002, 213-5. Berlin, Springer.

Zlatanova, S., Rahman, A. A. and Pilouk, M. 2002. Trends in 3D GIS development. Journal of Geospatial Engineering. 4 (2).

71



5 Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions César González-Pérez Institute of Heritage Sciences (Incipit), Spanish National Research Council (CSIC) ________________________________________________________________________________ Abstract The description of an archaeological entity is a process that often begins with the classification of the entity into one of a set of available categories; once the category is known, the description process itself may begin. This approach, albeit powerful and intuitive, is prone to causing category bias, since the category system at use has probably been created under the influence of ontic, epistemic and pragmatic conditions that are different to those of the current situation. Category bias, in turn, produces alterations in the information being generated that are very difficult to correct and, sometimes, even to detect. In this work we propose a category-less approach to the description of archaeological entities that avoids classification as an a priori mechanism, using instead the entity’s properties as units of description. Under this approach, category bias is avoided, information is equally obtained, and categorisation can still be employed on the data as an a posteriori mechanism to explore information patterns. Keywords: Conceptual Modelling; Data Modelling; Category Bias; Cognitive Bias; Archaeological Information. ____________________________________________________________________________________________

1. Introduction

today commonplace as tools of the archaeological practice.

Most genuine archaeological work involves some sort of information recording, processing and preservation. The material culture that we study, the activities that we carry out in order to recover it (such as surveys or excavations), as well as the interpretive processes that we perform on said material culture, are sources of large amounts of information that must be recorded, processed and preserved. Deciding how information is so managed is not a trivial matter, and the importance of the role that these decisions play within the archaeological discipline is exemplified by the publication of best practice guides e.g. Archaeology Data Service 2011 and the funding of efforts such as Digital Research Infrastructure for the Arts and Humanities (DARIAH www.dariah.eu) or Project Bamboo (www.projectbamboo.org).

However, and despite the broad and demanding use that archaeology has made of information technologies, it has failed to challenge certain aspects of them. In this respect, archaeology has behaved as an obedient user, taking what technology had to offer and being oblivious to the fact that the state of the art might be improved. This is the case of conceptual modelling, which underpins information recording, processing and preservation, and on which this chapter focuses. The following sections discuss this issue in depth. Section 2 presents an analysis of how information is usually recorded, processed and preserved in archaeology. Section 3 introduces the concept of category bias and explains why it is an inherent problem to most (if not all) approaches to information management in archaeology. Then, Section 4 suggests an approach to solve the problem, based on typeless information modelling, and Section 5 builds on this to propose a particular solution named value cluster modelling. After this, Section 6 presents some tools that have been developed to experiment with this solution and advances some preliminary observations that we have made in this regard. Section 7 discusses some advantages and limitations of the proposed approach in comparison to conventional methods, and describes some work that follows similar directions. Finally, Section 8 offers some conclusions and directions for future research.

Regarding this, archaeology has been using information technologies as an ancillary field for a few decades already: in 2012, the Annual Conference on Computer Applications and Quantitative Methods in Archaeology will be held for the fortieth time; specialised postgraduate courses such as the MSc in Archaeological Computing at the University of Southampton are nowadays common; and even some specific approaches to information recording and processing have been agreed upon and standardised, such as the CIDOC Conceptual Reference Model (ISO 2006). Specific technologies such as Geographical Information Systems and databases are

72

 

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions archaeology, in our experience, share a common methodological trait: ‘there is a predefined set of thing types, each of them characterised by a particular set of properties, to which reality is expected to conform’. Let us examine what this means.

2. The way we describe At Incipit, where I work, there is a large shelf cabinet near the common area where fieldwork forms are stored. These are paper, single-page forms designed to be taken into the field and filled in there in order to capture the information that is necessary to describe whatever archaeological entities are found or created during surveys, excavations or other activities. There are 34 different form models, corresponding to 34 different types of things that we know we are likely to find or create during archaeological work. These include the usual kinds of archaeological evidence such as Feature, Artefact, Stratigraphic Unit, Site and Zone; they also include kinds of things that are not uncovered, but created during archaeological work, such as Photography, Video, Diary Entry, or Sample. And they also include things that are elaborated later, once the information has been interpreted and transformed, such as Impact Assessment, Heritage Assessment or Report.



There is a predefined set of thing types. This means that, over time and thanks to the accumulation of experience and knowledge in archaeological practice, we have been able to develop a consensus about what kinds of things are likely to be found or created within the realm of our work. And we have fossilised this knowledge as types of things, thus capturing what we believe is invariant. For example, we at Incipit have a predefined thing type called Site because we know that we will find sites very often.



Each thing type is characterised by a particular set of properties. This means that we choose to describe the variability of individual things of a given type through specific properties, and these properties vary from type to type. For example, sites have different names, dimensions and cultural ascriptions, yet all of them fall within what we call ‘sites’. Hence Name, Dimension and Cultural Ascription are properties that become useful to describe individual sites.



Reality is expected to conform to thing types and their properties. This means that the conceptual construct composed by the thing types and the properties that describe them is used as a descriptive model, i.e. we expect to be able to describe the reality that we face by applying (or ‘enacting’) this model on it. For example, as soon as we find something during a survey that matches our definition of ‘site’, we expect it to be described through the properties associated to the Site thing type.

Each form contains fields that are supposed to be filled in with relevant data. For example, the Site form contains the following fields:       



Code. This is generated in the field as the site is recorded, and is composed by a sequence of letters and numbers. Name. This is a descriptive name for the site. Context. This describes the environment of the site. Dimensions. This describes the physical dimensions of the site. Description. This contains a free description of any relevant details of the site. Type. This assigns a specific type to the site, taken from a list of predefined options, such as Hillfort or Tumulus. Cultural ascription. This assigns a chronocultural period to the site, taken from a list of predefined options, such as Late Neolithic or 20th Century. Sketch. This contains a hand-drawn sketch of the site.

Archaeology manuals usually explain that archaeological data pertains to artefacts, features, ecofacts or sites (or a similar classification), and then go on to subdivide each of these large types into more concrete ones. These are, indeed, predefined thing types, and the characterisations that are often made of them refer to their properties. Some examples of this are (Renfrew and Bahn 2004, 5356) and (Burke and Smith 2004, 340-355).

Information can be also recorded at Incipit using a computer-based system, named SIA+ (Sistema de Información Arqueológica; Archaeological Information System) (Gonzalez-Perez 2002, section VI); this system mirrors the structure of types of things and fields that is inherent to paper forms, being fully compatible with them. Thus, using SIA+ may be faster, and not as cumbersome as using paper-based forms, but the information is recorded and entered following exactly the same protocols.

The subdivision of types into subtypes that we mention in the previous paragraph alludes to an important aspect of thing types: they are hierarchical. That is to say, there can be types within types, the outer ones semantically subsuming the ones inside. For example, at Incipit we have a thing type for Artefact that refers to any artefact that may be found during a survey or excavation, regardless of its particular kind or material; in addition, there are additional thing types for Pottery Artefact,

2.1. Common approach to description We are aware that other organisations work in different ways, and this situation may not be too similar to your case. However, most organisations that work in 73

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  Metal Artefact and Lithic Artefact, which incorporate specific properties that are often recorded for artefacts made of these materials. Similarly, thing types such as Site and Zone can be seen as specialised versions of a more generic type called Place.

2.2. The description workflow The approach outlined so far is both natural and extremely powerful. It is natural because it fits perfectly with the mechanisms of categorisation that operate in our mind (Lakoff 1990, 5-6). Seeing the world in terms of predefined categories and assuming that it will conform to them can only be expected. However, we need to admit that the thing types that we describe above are classical categories (Lakoff 1990, 161) and, as such, any approach based on them will be necessarily rooted in an essentialist view of the world. We discuss the implications of this in Section 3

In addition to being linked through subsumption relationships, thing types may also be linked to one another via other kinds of relationships; continuing with the examples in the previous paragraph, the thing type Artefact can be said to be related to the Site thing type via an ‘appears in’ relationship. Relationships link thing types together into complex systems, in which each thing type becomes interdependent on many others, since both its definition and the properties that characterise it rely on such a network of relationships. In our example, and also in accordance to the reality at Incipit, we do not include a Site Name property for the Artefact thing type, because we assume that artefacts are always linked to a site, and it is the Site thing type which has a Name property.

2.3. Category bias On the other hand, the outlined approach is powerful because it allows a very efficient way of working ( Figure 1).

At this point, some may argue that defending the existence of universal predefined thing types, or even simply predefined thing types, implies an extremely reductionist and impoverished approach to the richness and complexity of the archaeological record. The point, however, is not that people agree on the definition of such predefined thing types or what their properties are; this is not necessary and, in fact, rarely the case. The point is that predefined thing types do exist, although different people and different organisations define them differently, and attach different properties to them. In other words, what we have observed is not an agreement on what an X is, or what properties Xs have; but an agreement that archaeology is practiced by considering the concept of X (however it is defined) as an a priori; where X is any relevant word such as ‘site’, ‘artefact’, ‘feature’, etc. In the words of (Korfhage 1997, 8), ‘all data have some organization, if only the organization imposed by the process of collecting and storing the data’.

Figure 1 Typical workflow when using a description approach based on classical categories. Whenever an entity is to be described, a classification process is carried out in order to assign it to a category. Once the category is known, the properties to be described are automatically given, since they are inherent to the category. The description process follows, which, taking said properties as a starting point, yields an information object as a result. Let us consider the following example. During an archaeological excavation, something is unearthed. According to this workflow, the first step that takes place is that of classification, i.e. determining what category this thing belongs to. Is it an artefact, a feature or something else? By using our experience we can immediately (and possibly unconsciously) determine that it is an artefact. Now, is it a lithic artefact, a pottery shard, or some other kind of find? Let us assume that it is a pottery shard. Now that a specific category (or thing type) has been determined, the properties to describe are automatically given; according to the fieldwork form that we usually employ at Incipit, pottery shards need recording of properties such as code, brief description, whether it is decorated, fragment shape, vessel part, and vessel type. Once the properties to be described have been determined, the description process itself may start for each of them; in this example, we would need to create a code for the pottery shard, write a brief description, determine whether it is decorated or not, and choose a fragment shape, a vessel part and a vessel type from lists of predefined options. The outcome of this, namely, the set of specific values that we generate, is an information object Figure 2).

It is also necessary to note that not all categories are equal. At one end of the spectrum, some categories are shared by practically everyone across organisations, nationalities and even theoretical inclinations; these categories often have very abstract definitions. For example, when presented with a pottery shard, very few archaeologists would hesitate when asked ‘Is this an archaeological artefact?’. The Artefact category is an example of what we call base categories. At the opposite end of the spectrum we find interpretive categories, which are heavily based on very specific interpretations made by one or few individuals and are, therefore, unlikely to be shared by many. For example, categories that make up typological systems such as ‘Classical Attic Tombstone’ or ‘Toralla Pot’ are usually highly interpretive. Between the two extremes, categories vary along a wide range of ‘interpretiveness’; for example, many archaeologists would agree on what constitutes a site and what does not, but some would see multiple sites where others see only one. 74

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions cognitive device that led palaeontologist Charles Doolittle Walcott to classify a wide number of fossil specimens from the Burgess Shale in Canada in the wrong taxa during the 1910s and 1920s. Walcott followed a description mechanism similar to that depicted in Figure 1, with the fossil specimens being the entities to be described, and the then-recognised taxa playing the role of the available categories. Thus, Walcott classified the fossils in the available categories, ‘shoehorning’ them into place with mixed success. Evidently, the existence of a clear, well-established taxonomic system was useful to Walcott because it gave him a reference frame for his work; at the same time, it led him into what today we judge as a number of classification errors. Why?

Figure 2 An information object (plus its values) next to the category from which it has been created. It is easy to find parallels between the overall approach that we have described so far and the mainstream approaches to information modelling in the software engineering or information technologies fields. The concepts of category and information object that we have just introduced match nicely those of class and object of the object-oriented paradigm (Meyer 1997), or even those of entity set and entity of the entity-relationship paradigm (Chen and Pin-Shan 1976). They can even be pushed to map to those of table and row in the relational database model (Codd 1970). Similarly, the fact that thing types are hierarchically linked through subsumption relationships is nicely captured by the object-oriented concept of generalisation/specialisation (Rumbaugh et al. 1991, 38-42), and the more general kinds of relationships that link them together in complex networks correspond to the object-oriented concept of association (Rumbaugh et al. 1991, 27-31).

First and foremost, and as we explained in Section 2 The way we describe, a category system reflects what we believe is invariant, and is constructed through the accumulation of experience and knowledge. This means that any given category system can only be applied to entities of the same nature as those from which it was generated. Walcott was fortunate (or unfortunate) enough to find fossils that were extremely different to what paleontolgy had ever found so far, and hence customary taxonomy was not applicable. Using the established taxonomy of the early 20th century on the Burgess Shale fossil record was akin to trying to describe Australian Indigenous sites using categories defined from Atlantic European megalithic landscapes. Secondly, the properties attached to each category are defined so that they capture what we believe are relevant characteristics of the associated categories. For example, at Incipit we have included a Name property for the Site category because we think that most sites should have a name, and recording this name is important. But important for what? It is clear that the purpose of the description plays a crucial role in determining what properties must be attached to each category. For example, a property for the Distance to the Nearest Water Source may be appropriate for the Site category if the purpose of the description is to carry out research about the use of the territory, but probably superfluous if the purpose is cataloguing the site for a tourist guide. In other words, properties are not natural features of the categories, but subjective attributions that we construct according to our needs, experience, and limitations.

In addition, the relationship between the information object obtained as final outcome of the information recording process and the category assigned to the entity being described can be mapped to one of instantiation (Rumbaugh et al. 1991, 69-70); the information object can be said to be an instance-of the category (Figure 3).

Figure 3 The information object generated by the description process ( Figure 1) is an instance of the category to which the entity was assigned.

Most people would accept that categories and properties only reflect our experience, rather than a fixed, objective underlying world. However, once a category system is in place, we often use it as a descriptive model (see Section 2 The way we describe), and we actually behave as if the underlying world were composed of classical categories with natural properties that are necessary and sufficient to define them. This worldview is often named essentialism (Lakoff 1990, 160-162), and even when we do not share it, we are biased towards it because of its mere existence. Evidently, the consequences of essentialism are more serious when arising from interpretive categories than from base categories.

Being natural, powerful, and well-grounded in information technology theory, the category-based approach contains one major flaw: given its reliance on the figure of classical categories, it inherits all of their weaknesses. The next section explores in depth the particular issue of category bias.

3. Category bias In ‘Wonderful Life’ (Gould 2000), Stephen Jay Gould introduces the concept of the Walcott’s shoehorn, a 75

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  3.1. Characteristics of category systems

free decision: too much variability, and the category becomes too broad and vague; too little variability and it becomes constraining and useless to describe many entities.

The issues described above can be summarised as pertaining to three specific aspects. 





The coverage, partitioning and variability balance of a particular category system, as we have shown, are not predefined. The factors that influence them and ultimate determine the outcome can be described as follows.

Coverage. The categories in a category system cover the fraction of reality that we have had experience with, but do not contemplate other areas of reality. Similarly, the properties in any given category address the characteristics that we have considered as relevant, leaving out the rest. For example, a category system may contain many categories that reflect, in utmost detail, the types of burial sites that are often found in Atlantic Europe, but contain only one category corresponding (from a very abstract viewpoint) to artefacts of all kinds. Partitioning. The categories in a category system ‘cut up’ reality in a particular way, chunking it into discrete units that may be more or less applicable to different situations. For example, a category system may define a Site category as including both the archaeological and geographical aspects of the site; another may define Site as including only the archaeological aspects, moving the geographical details to a separate category named Place. Similarly, the properties of any given category express the characteristics of the category in a particular manner, and at a certain level of abstraction. For example, the Site category may have a Dimensions property where the site’s dimensions are supposed to be described at length; or, alternatively, it may have separate properties for Length and Width, where one is supposed to enter specific measurements in metres or some other unit. Variability balance. As we explained in Section 2 The way we describe, categories express the invariants that we observe as common across entities, whereas properties express the variables that change from one entity to the next. Thus, the decision of whether to make a trait part of the category definition or a property in the category is simply a matter of deciding whether that trait varies across individual members of that category or not. For example, a particular category system may state that all pottery artefacts are part of a pottery vessel (or a whole pottery vessel itself), and therefore properties such as Vessel Type and Vessel Part make sense within the Pottery Shard category. Another category system may acknowledge that some pottery shards may be fragments of bricks, tiles or other non-vessel things, making what could have been an invariant (and part of the category definition) into a variable, and thus a property such as Object Type. Picking the right degree of variability in a category is not a consequence-



Scenario. This refers to the contextual setting where the category system is created. For example, a category system for the archaeology of Galicia (North-West Spain) will more likely contain categories that catpure the archaeological record of Galicia rather than the archaeological record of Australia.



Background. This refers to the previous experience, knowledge and disciplinar approach of the person or persons creating the category system. Particular backgrounds tend to impose particular worldviews; for example, a category system created by an archaeologist will probably differ from one created by an anthropologist or an art historian. Even within the same discipline, two people or teams with different past experiences or degrees of expertise often create category systems that differ significantly. This is related to the concept of user-fit dimension of representations in (Peterson 1996, 10).



Purpose. This refers to the goal for which the category system is created. As we said at the beginning of this Section, the purpose of the description plays an important role in determining what properties must be attached to each category; also, deciding what categories should be included and which left out is also influenced by the purpose; this corresponds to what (Peterson 1996, 9) calls the task-fit dimension of representations.



Limitations. In addition to the previous factors, there are often practical constraints in the development process of the category system that limit the scope and depth of the outcome. For example, it is not uncommon that some of the categories must be compatible with an existing recording or processing tool such as a database or information system; in cases like these, the categories that are created and the properties that are attached to them are bounded by pragmatic (rather than ontic or epistemic) reasons.

Figure 4 summarises the interplay of category system characteristics and the factors that influence them. The conclusion so far is that any given category system is just one out of many potential solutions to the same problem. This resonates with the old adage that ‘all 76

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions models are wrong; some models are useful’, attributed to George Box. In other words, a category system is just one specific category system, the coverage, partitioning, and variability balance of which are strongly influenced by the setting where it was created, the experiential and disciplinary background of its creators, the particular goals in their minds at that moment, and the practical limitations that they found.

are two major families of cognitive processes at work: intuitive and reflective. Intuitive cognitive processes are automatic, effortless, associative, rapid, and often rely on skills. Reflective processes, on the contrary, are controlled, effortful, deductive, slow, self-aware, and often rely on the application of explicit rules. Also according to Kahneman and Frederick ibid.), intuitive processes quickly propose answers to judgment problems as they arise, and reflective processes monitors the quality of these proposals, endorsing, correcting or overriding them as necessary. The judgment that we ultimately express is called an intuitive one if it matches what the intuitive processes initially suggested. In this regard, and assuming an ill-suited category system, category bias occurs because: 

To start with, the category system at use comprises a set of formal or semi-formal definitions (of categories and properties) that serve as rules to our reflective cognitive processes, allowing us to carry out the necessary controlled, deductive judgments; this means that, even with relatively unfamiliar category systems, we are able to classify and describe entities in a biased manner.



As exposure to the category system increases and proficiency is acquired, our ability to make judgments migrate from the reflective to the intuitive realm (Kahneman and Frederick, 2005, 51); this means that, over time, we start making category-biased decisions in an intuitive, automatic and effortless way.

Figure 4 Ontic and epistemic reasons (scenario, background and purpose), together with pragmatic limitations, determine the characteristics of a category system, namely its coverage, partitioning and variability balance. 3.2. Introducing cognitive bias We said in Section 2 The way we describe that a category system can be ‘enacted’ or applied in order to describe the reality that we observe in relation to the categories and properties defined by it. However, and according to the previous paragraph, any category system is just one out of many potential options. When we use it, we are implicitly absorbing the consequences of the particular influencers that underpin it, i.e. we are espousing the setting where it was created, the experiential and disciplinar background of its creators, the particular goals in their minds at that moment, and the practical limitations that they found. By doing this, we are implicitly, and often inadvertently, adopting a particular set of ontic, epistemic and pragmatic influencers of category system characteristics without stopping to critically assess their suitability. Once this has happened, the classification process (Figure 3) that assigns a category to a particular entity is bound to employ a category that inherits these issues; in turn, the resulting information object will exhibit them as well. Thus, we define category bias as the deviation in classification and description that occurs when assigning an entity to a category of a system that is ill-suited to the task at hand. This means that, when category bias occurs, the resulting classification and description of the entity is different to what it would have been if the category system in use had been created taking into account the scenario, background, purpose, and limitations of the particular description situation at hand. This difference will be much larger for interpretive categories than for base categories, since, by definition, base categories are intersubjectively shared by most people, whereas interpretive ones are not.

If no measures are taken to fight category bias, problems are bound to arise. The most evident kind of problem to be detected is that of situational mismatch. As we explained above, the scenario, background, purpose and limitation influencers strongly determine the outcome of any particular category system. When a category system is applied in a context with different situational parameters, i.e. different scenario, background, purpose, and limitations, then a mismatch occurs between the situations where the category system was created and is being enacted. Another kind of problem, which is much subtler but potentially more threatening to the integrity of the information, is that of objectivism. We explained in Section 2 The way we describe that categories as most often used are, in fact, classical categories; they are defined by an invariant, or statement of what characteristics of things are necessary and sufficient to determine whether any given thing belongs to the category or not (Lakoff and Johnson 2003, 186-188). This, together with other associated principles that govern classical categorisation, makes up a convenient way to formalise the notion of category and, in fact, all mainstream information technology approaches have adopted it, including object-orientation and entityrelationship. However, reality is often more complex than

Category bias is a specific kind of cognitive bias. According to Kahneman and Frederick (2005, 51), there 77

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  this; as Lakoff has shown, radial (Lakoff 1990, ch. 6) and other kinds of categories cannot be defined by invariants, and there is no set of characteristics that all of their members share. In fact, a radial category is one that is not defined by a common set of characteristics that are shared by all its members, but one that emerges by accretion of its members starting from one individual, each additional one being linked to the previous by some shared characteristic. In a radial category, the characteristic that is shared by two members may vary from a pair to the next; in a classical category, all members share exactly the same characteristics. Lakoff uses the example of the category ‘mother’ (Lakoff 1990, 91), which includes adoptive mothers, birth mothers, foster mothers, stepmothers, surrogate mothers, etc. There is no common characteristic invariantly shared by all of them.

we explore an alternative avenue, namely, the total avoidance of categories as a descriptive device.

Similarly, it can be argued that concepts such as that of ‘archaeological site’ are better conceptualised as radial rather than classical categories. We seem to have a very clear, intuitive idea of what an archaeological site is, but providing an explicit and complete list of characteristics that are necessary and sufficient to signify membership to this category proves to be an extremely difficult task. Therefore, a classical category named Site is almost certainly ill-suited in whatever situation, since it would be imposing an artificial set of common characteristics on all of its members.

Figure 5 Information description workflow taken from Figure 3 with the category concept removed.

4.1. A category-less description workflow Is it possible to describe an entity in the absence of a category system? Figure 5 shows the typical information description workflow (taken from Figure 3) with the category concept removed.

The resulting approach (Figure 6) comprises, first of all, a property determination process, which establishes what the relevant properties to be considered are for the entity to be described. Once this has been done, the description process follows, which takes said properties as a starting point to yield an information object as a result.

How can category bias be avoided? We propose an approach that is as drastic as simple: by avoiding categories altogether. We are aware that other alternatives (such as developing a formalisation of radial, generative and graded categories) are worth exploring as well but we leave them for future research. The next section describes the overall approach to typeless (or category-less) information modelling, which establishes the foundation for non-biased archaeological descriptions.

Figure 6 Information description workflow with no category concept (resulting from Figure 5). For example, let us imagine that something is unearthed during an archaeological excavation. According to this new workflow, the first step that would take place is that of property determination. We would not try to decide what the entity to be described is, but what relevant properties it has. For example, we may decide that we want to give it a code, write a brief description of it, record its location using 3D coordinates, and also record the material of which it is made. Once the properties to be described have been determined, the description process itself starts for each of them; in our example, we would need to create a code for the entity, write a brief description, capture its 3D coordinates, and choose the material it is made of from a list of predefined options. The outcome of this, namely, the set of specific values that we generate, is an information object (Figure 7).

4. Typeless information modelling It could be argued that category bias can be mitigated by using highly flexible categories that adapt to whatever needs are presented by the application context. Following this approach, categories would not be rigid structures, but elastic guidelines that could be adjusted or ‘customised’ to whatever reality they are enacted in. However, a category like this would fail to help us organise reality around us in an effective way, since it would admit an undefined range of things into it, becoming liable to the ‘Texas sharpshooter fallacy’ (Wikipedia contributors, 2011). In other words, by removing strictness we are also removing the power of categories to arrange things into intersubjective, wellknown formations. In addition, flexible categories are still categories, i.e. they still cause category bias, albeit, arguably, a lesser amount of it. Exploring whether a reduction in category bias is worth a considerable forfeit in categorisation power may merit separate research efforts (see 8 Conclusions), but is out of scope here. Here

Note that, since there is no classification process in place anymore, the instance-of relationship that was present in the case of the category-based approach is now lost. In other words, information objects created under a category-less approach are not instances of anything in particular; they stand as self-contained aggregates of values.

78

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions anything with values for at least two coordinates, plus Material and Cultural Ascription properties’, and observe how the software tool recomputes what information objects are members of the Artefact category. There are two major advantages of a posteriori categories as compared to a priori (i.e. conventional) categories. First of all, a posteriori categories are exploratory, volatile and temporary; we define them in terms of the conditions that information objects must meet, and information automatically acquires the shape of the category on the fly. This is to say, membership of the category is not fixed ontologically, as in conventional, a priori categories, but emerges as we explore possibilities.

Figure 7 An information object (plus its values) created within a category-less information description workflow (cf. Figure 2). At this stage, we can freely admit that the workflow shown in Figure 6 is a simplification of the process that most often takes place in practice. According to our informal experience, the determination of properties and the description of their values do not happen as two separate, sequential activities; rather, properties are determined and data about the entity is absorbed within a complex and tangled mix of cognitive processes that, ultimately, yield the information object (Figure 8).

Secondly, multiple category systems can be active at the same time; we do not need to choose between mutually exclusive ways to organise the information. This is especially relevant in the context of teamwork on a shared set of data, where each individual may be using the same information under different epistemic presuppositions. In this regard, an a posteriori category system is akin to a view on the information that gives us a particular perspective without altering the nature or structure of said information. 4.3. Typeless information modelling If modelling means giving shape, then information modelling alludes to the process of shaping information so that it is useful for our ends. As we explained in previous sections, all the mainstream modelling approaches in information technologies are heavily category-based; our proposal, in this respect, deviates from the norm significantly. How can shape be given to information once the major shaping device, i.e. categories, has been removed?

Figure 8 Property determination and data intake occur simultaneously, feeding back into each other (cf. Figure 6). 4.2. A posteriori categorisation The absence of a category system does not mean that information cannot be organised and arranged; it only means that information is not organised and arranged a priori. But nothing prevents us from exploring the patterns in the information after the fact, and using them to overlay specific arrangements onto the data a posteriori, once the information objects have been created.

In information technologies, the concept of category is often formalised as that of type. Roughly speaking, a type is a formal definition of the common ‘look and feel’ of a similar set of data. Like classical categories, types are defined through an invariant, which establishes the characteristics that all members of the type must possess; and types may also introduce a series of features, which are similar to the variable properties that we described for categories, and which describe how members of the category can be different to one another.

Under a category-less description approach, information is composed of discrete information objects, each one comprising an aggregate of values (Figure 7). A posteriori categorisation works by applying declarative category definitions on such information objects, and dynamically obtaining the category’s extension. For example, let us imagine that a few information objects similar to the one show in Figure 7 have been created. Each would probably have different values of similar but different properties. We could declare an a posteriori category like this: ‘an Artefact is anything with values for the Coord X, Coord Y, Coord Z and Material properties’. A software tool could easily compute which information objects comply with this declaration, and consider them as members of the just introduced Artefact category as long as its declaration is kept active. We could easily redefine the statement, for example, ‘an Artefact is

A category-less approach to archaeological description, therefore, must be based in a typeless infrastructure from a computing point of view. The following section introduces a proposal for a specific modelling language that, accordingly, avoids types. This language allows the shaping (i.e. modelling) of information in the absence of types.

5. The valster modelling language In information technologies, a modelling language is an artificial language that can be used to express information 79

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  following a well-known set of rules. Modelling languages usually include a small lexicon (tens or at most hundreds of ‘words’) and a formal or semi-formal syntax specification, often conveyed as a metamodel (GonzalezPerez and Henderson-Sellers 2008, 18). By using a modelling language, different people can express information using a shared, well understood vehicle that minimises ambiguity and enhances communication clarity and effectiveness. This section introduces the valster modelling language, which implements a typeless approach to information modelling. It is expressed as a collection of UML 1.4.2 (ISO/IEC, 2005) class diagrams and accompanying text descriptions.

would probably hold data such as ‘flint’, ‘clay’, ‘birch wood’, etc.; it is possible (and recommended) to use an enumerated type for this property rather than an openended Text base type. Figure 9 shows a diagram illustrating the concepts described so far.

5.1. The valster modelling language metamodel Information objects created under a category-less approach are simple aggregates of values (Figure 7); for this reason, we propose the name value cluster, or valster for short, to refer to each of such information objects. A value cluster (or valster), therefore, is an aggregate of one or more individual values, where each value refers to a specific property and contains a specific piece of data for that property. Valsters do not exist in isolation. For example, a valster that represents an arrow head found during an archaeological excavation must be linked in some way to the valster that represents the excavation itself. This is accomplished in two ways. First of all, related valsters can be put into a model, i.e. an informational abstraction that represents a complex portion of reality. For example, all the valsters representing things unearthed during the excavation, plus the valster for the excavation itself, plus any other valster that we may regard as being related to this activity, can be put into a single model and managed as a whole. Secondly, valsters can hold pointers to other valsters, via values of the reference data type; this is explained in the following paragraphs.

Figure 9 Metamodel of the valster modelling language. Note that, in addition to classes for the concepts introduced above, the metamodel also contains an InformationStore class. An information store is a generalpurpose container for models and properties. As we will see in 7 Discussion, information stores play an important role as semantic repositories of properties that can be used across a number of models.

Each value in a valster, as we said, refers to a specific property. Each property has a name and is of a certain type. The type of the property determines what kind of data its values may contain; for example, values of a property of type Number may only contain numeric data, but not text or dates. Acceptable base data types are Boolean (only ‘yes’ or ‘no’ data), Number, Time (any point in time, at any resolution), Text or Data (any other data).

The metamodel also incorporates Category and CategorySystem classes; these implement a posteriori categories as described in Section 4 Typeless information modelling. The Category class represents an a posteriori category, whereas the CategorySystem class represents a collection of a posteriori categories that work together as a system. For example, a researcher exploring patterns to organise the information gathered during an archaeological survey may create a ‘material culture’ category system containing specific categories such as ‘artefact’, ‘clay artefact’, ‘stone artefact’, etc.

An additional base data type, Reference, exists in order to allow valsters to hold pointers to each other. A value of a property of type Reference can hold a pointer to another valster in the same model, rather than a conventional piece of data.

5.2. Using the language

In addition, enumerated types exist in order to allow for custom lists of predefined items; this is useful for properties that take values within a well defined range rather than across a whole type domain. For example, values of the the property Material in the example at the beginning of Section 4 Typeless information modelling

Let us imagine that an excavation campaign is taking place at a site named Chan da Rocha, and that a number of artefacts are recovered during the works. The first step that is necessary in order to capture information about the excavation and the associated artefacts is an information 80

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions store plus some properties. We can anticipate that this particular excavation campaign at Chan da Rocha will result in a new model being added to the information store; however, since properties are not particular to a specific model, but external to them, it is a good practice to define them with reuse in mind. If the information store already contains properties that have been defined (and hopefully used) by other models, probably pertaining to previous projects, then we may use them; alternatively, or in addition to this, we can add new properties to the information store before we attempt to put any information in, so that the necessary infrastructure will be in place as soon as it is needed. In any case, let us assume that the information store contains the properties shown in Figure 10. Code

Text

Start

Moment in time when the entity begins occurring.

Name

End

Moment in time when the entity finishes occurring.

Description

Is Decorated

Indicates whether the entity has been artificially decorated.

Map Ref

Material

Text

A conventional reference to a well‐ known map or chart where the entity appears.

Provenience

Reference

The physical location where the entity was originally found.

enum Material

The material of which the entity is made.

Cultural Origin

Excavation at Chan da Rocha

Map Ref

IDG92 1520 F8

Start Description

11 September 1997 Archaeological dig at Chan da  Rocha, 1997 campaign. Late Iron Age

Code

AVR971202V16

Material Description Is Decorated

Clay Rim fragment, well preserved.  Broke in two at dig. Yes

Cultural Origin

Late Iron Age

Provenience

17

25

enum Culture

The culture that created or gave origin to the entity.

Figure 10 Properties contained in the information store. For each property, the name, type name and description are given (cf. metamodel in Figure 9).

Code

AVR971203S01

Material Description Is Decorated

Granite Circular mill stone fragment,  about one third of circle. No

Provenience

17

Figure 11 Three valsters in the model. For each one, an id is given at the top, followed by a list of values; each value is depicted as a pair consisting of the name of the associated property plus the data content (cf. metamodel in Figure 9). Arrows have been added to illustrate how reference-typed values work.

Properties Material and Cultural Origin are specified as of ‘enum Material’ and ‘enum Culture’, respectively, so the enumerated types Material and Culture should also be introduced in the information store, and their enumerated items determined; of course, it is possible to alter the list of enumerated items later.

Note that, in contrast to classical category-based systems, we are not bound to use a predefined set of properties to describe any particular entity. In the example, two artefacts (a clay fragment and a mill stone fragment) are described by very similar sets of properties, but not exactly the same ones. It is to be expected that very similar entities described in very similar contexts (including very similar subjectivities) will tend to use very similar sets of properties, but the category-less approach imposes nothing in this regard.

Once the properties are in place, a model can be created in the store, and valsters added to it. As we explained in Section 4 Typeless information modelling, a valster represents an entity that is described under a categoryless approach. For the sake of this example, let us assume that a valster will be created to describe the excavation campaign, and additional valsters will be added to describe each artefact that is unearthed during the dig. Figure 11 shows a diagrammatic representation of three sample valsters; the one on top, labelled with the id 17, represents the excavation campaign, and contains values for properties Code, Name, Map Ref, Start, Description and Cultural Origin. The other two valsters, labelled with ids 24 and 25, represent artefacts recovered during the dig, and contain similar lists of values; both have been

After the information has been entered, category systems can be used to overlay a posteriori categorisation views on the information without changing its structure, as we explained in Section 4, Typeless information modelling. In order to do this, a category system must be created in the information store, and categories added to it. Let us assume that we wish to look at the Iron Age material 81

 

AVR97S01

Name

24

Boolean

A brief narration of the most relevant characteristics of the entity.

Code

Cultural Origin

Time

The proper name of the entity, which denotes it unambiguously and distinguishes it from others.

Text

17

Time

A unique identifier that is assigned to the entity for the sake of reference.

Text

described through properties Code, Material, Description, Is Decorated, and Provenience. One of them, in addition, has a value for property Cultural Origin. Note that values corresponding to the Provenience property, being of type Reference, hold the id of the referred-to valster as data. This is interpreted in Figure 11 as, for example, ‘the provenience of the entity represented by valster 24 is described by the valster with id 17’.

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  culture recovered from the excavation in our example; we could introduce the two categories shown in Figure 12.

Figure 13 is a screenshot showing the properties described in the example used in the previous section (Figure 10).

Figure 12 Two a posteriori categories in the information store. For each one, a name is given, followed by a formal declaration and a description (cf. metamodel in Figure 9).

Figure 13 List of properties in an information store as shown by the demonstrator tool (cf. Figure 10).

Categories are formally defined through a declaration that uses property names, values and a set of simple operators to construct expressions that a software tool can evaluate. For example, the Iron Age category in Figure 12 is declared as ‘Cultural Origin = Iron Age OR Cultural Origin = Early Iron Age OR Cultural Origin = Late Iron Age’; this expression will hold for any valster that has a value for the Cultural Origin property that matches one of the given options. Similarly, the Iron Age Material Culture category is declared as ‘IS Iron Age AND HAS Material’; this means that a valster will be considered a member of this category if it is a member of the previous one and, in addition, has a value (with any data content) for the Material property.

Note that the type names used by the tool do not exactly match those defined by the valster modelling language. This is because the tool was developed before the valster modelling language metamodel stabilised into the final form presented here; still, there is a clear and unambiguous mapping: ‘String’ means ‘Text’, ‘DateTime’ means ‘Time’, and ‘Ref’ means ‘Reference’. Also, since the tool does not support enumerated types, properties such as Material and Cultural Origin, which were defined as of enumerated types in the previous section, are defined here as of type ‘String’ for the sake of this example. This tool is a demonstrator for experimental purposes, and does not aim to provide a complete, fully-functional user interface. A tool that aimed to support archaeologists doing actual work would need a more powerful set of features to manage properties, such as a search feature to find properties by name and/or description, and a semantic network feature to assist in finding the best fitting property for any given situation; this is an idea that is further explored in section 7. Discussion.

Given the extremely simple declarations of these categories and the reduced number of valsters in our example, it can be easily verified that, under this category system, valsters 17 and 24 in Figure 11 would be considered members of the Iron Age category, and only valster 24 would be considered a member of the Iron Age Material Culture category. The power and flexibility of the valster modelling language in general, and of a posteriori categorisation in particular, becomes prominent when complex category systems are overlaid onto models with hundreds or more valsters. To do this, software tools are necessary. The following section presents some efforts in this direction.

Figure 14 is a screenshot showing the valsters in the same example of the previous section (Figure 11). Each valster is shown as a node in a tree labelled with the valster id, under which the valster’s values are listed. Data contents are shown alongside, using some conventions. For example, reference-typed values, such as Provenience in valsters 24 and 25, show the id of the valster it refers to. Double-clicking on the value jumps to the referred-to valster, allowing for easy navigation.

6. Tools for valster modelling A simple demonstrator software tool has been implemented to experiment with the theoretical approach proposed here. The tool supports properties and valsters, which can be interactively created and modified on screen. The information so created can be saved to disk for later retrieval. The tool does not support a posteriori categories or enumerated types yet. The tool was developed by the author using the Microsoft .NET environment.

Through the user interface shown above, new valsters can be easily added, and values can similarly be added, modified or deleted. Figure 15 is a screenshot showing the process of entering data for a new value for the Description property of a valster. Again, this user interface would not be enough for a production-level tool that aimed to address the needs of actual archaeological work. There are multiple options to 82

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions display valsters and the relationships amongst them, and the best way to implement the interaction between the user and the tool in this regard is still an area of active research.

matter of implementing a few abstract principles. However, the design of the valster modelling language entailed some important decisions, even after having the overall approach to category-less information modelling in place. Most of these decisions are related to the issue of how to optimise the use of semantics within the workflow. First of all, the concepts of information store and model were introduced in the language so that valsters could be grouped at two levels: closely related valsters compose a model, and related models, in turn, compose an information store. If we conceive a valster as an information object that represents an entity in the ‘real world’, then a model is akin to a database in a conventional information system; like table records exist within a database, valsters exist in models; and like related records hold references to each other within the database, related valsters hold references to each other with the model (Figure 16).

Figure 14 List of valsters in a model as shown by the demonstrator tool (cf. Figure 11).

Figure 16 The hierarchy of information stores, models and valsters, and a database analogy. No equivalent to information stores exists in the database world. In our approach, however, there is an additional level of grouping, which collects multiple models together into an information store. The models that are put into an information store are semantically related, i.e. they must share something. In fact, types, properties and category systems are the three aspects that establish the sematic similarities across all models that reside in the same information store (Figure 9). To start with, all models share the same set of types; this is especially relevant with regard to enumerated types. For example, we can create a Material enumerated type in the information store, and all the models in it would share it. Secondly, all models share the same set of properties. For example, if we introduce a property such as Weight, of type Number, and description ‘The weight of the entity in grams’, all models would share this. Finally, a posteriori category systems are also shared by all the models in an information store in a similar manner. Sharing common types, properties, and category systems means that models in an information store share semantics to a certain extent. The information content of each model will be different, but the semantic context where that

Figure 15 Adding a new value to a valster.

7. Discussion So far we have presented an approach for category-less information description in archaeology. Before this approach is adopted or developed further, some additional considerations should be taken into account, and some works by other authors should also be studied. 7.1. Additional considerations The transition from the overall approach to category-less information modelling described in Section 4, to the specific design of the valster modelling language described in Section 5, may look like a simple step and a 83

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

  information lives is the same: the same types, the same categories from which valsters can be created, and the same categories that can be overlayed on the information.

biased, should not we criticise our approach for being property-biased? In a way, yes. Figure 17 is a replica of Figure 4 where the original ‘category system’ has been replaced by ‘property set’. This illustrates that the characteristics that apply to a category system (coverage, partitioning, and variability balance) also apply to the set of properties that are defined as part of an information store; and that the ontic, epistemic, and pragmatic influencers that affect the outcome of these characteristics for category systems also play a role for property sets.

Arguably, shared semantics can help homogenise models of related segments of reality. For example, let us consider a private company carrying out a series of archaeological excavations over time. It is conceivable that some standardised approach, common to all excavations, exists in the organisation. This approach could be modelled in terms of enumerated types, properties and category systems, and put into an information store; then, a model would be created in this information store for each particular excavation that the company carries out, all of them sharing the same semantic context. If, at some point, the company performs a research project, or some other endeavour that does not share the same semantic context as the previous projects, the corresponding model can be placed in a separate information store with its own semantic context. We believe that this way of modelling the semantic context of an organisation is superior to other approaches. Traditionally, organisational and domain knowledge has been modelled in terms of ontologies (Green and Rosemann 2005; ISO 2006) or similar constructs, which are heavily category-based, focussing on the entities that the organisation works with. Standardising categories, and pursuing reuse at the category level, is much more difficult than doing it at the property level, because categories are cognitively more complex than properties, and thus it is harder to define them in unambiguous ways, communicate them effectively, and apply them unchanged to new scenarios. Standardising properties, to the contrary and according to our experience, requires less effort to find good definitions, communicate their meaning, and apply them to new situations. In addition, categories are often defined in ontologies as part of an interconnected mesh, reflecting what can be observed in reality; properties, conversely, are independent from each other: they are defined as self-contained, unrelated ‘atoms’ of significance with no dependency relationships. This means that a property can be easily altered or removed altogether from an information store without triggering a chain of resultant changes to other properties. In this regard, maintenance of the standardised information is arguably much simpler with our approach than with a category-based one.

coverage Scenario Background Purpose Limitations

Property Set

partitioning

variability balance

Figure 17 Ontic and epistemic reasons (scenario, background and purpose), together with pragmatic limitations, determine the characteristics of the property set in an information store, namely its coverage, partitioning and variability balance (cf. Figure 4). In other words: any given property set that we might define is just one out of many potential solutions to the same problem, and, like in the case of category systems, adopting it means that we are possibly applying it to a situation that does not fit its appropriate usage scenario. However, this is not as big a problem as in the case of category systems, for two reasons:

7.2. Criticism Despite the advantages of a property-based approach, we must admit that properties present an important limitation. Properties are, after all, formalisations of relevant characteristics of entities in the world; the fact that they are formalisations means that they are cognitively constructed. This is obvious, but we need to bring this to attention in order to make the following criticism: the cognitive processes that we employ in order to construct properties are subject to bias as much as those that we may use to construct categories. So, if we have criticised conventional approaches as category-



As we said in the previous section, properties are cognitively much simpler than categories, and are defined as independent, self-contained ‘atoms’ of significance rather than interconnected nodes in a mesh. This means that the cognitive bias that a wrong property may bring about is necessarily smaller than the bias imposed by the wrong category.



Properties qualify a thing (e.g. ‘this thing is of Roman origin’), whereas categories establish what a thing is (e.g. ‘this thing is an amphora’). This means that categories determine the structure of the information, whereas properties add details to whatever structure is in place. A wrong property may add incorrect details, but a wrong category creates an incorrect structure, which is a much deeper problem.

Since we must use our cognition to describe things, we must admit that some risk of bias is inherent to the information creation process. Placing this risk at the level of properties rather than categories seems to yield a better balance between descriptive power and potential bias.

84

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions 7.3. Related work

(McKeehan and Rhodes 1996), created by Apple to develop software for the Newton devices, offers the necessary capabilities to create and maintain soups, frames and slots, and contains special logic that treats the ‘_class’ slots in the special way that they are intended to. The reasons why Apple decided to implement data storage using this hybrid approach rather than a conventional file system seems to be related to ,data flexibility, interoperability and extension (McKeehan and Rhodes 1996).

Within archaeology, some works have been carried out to explore category-less or category-weak description approaches. (Cobas Fernández 2003) suggests an approach to the description of decoration on material culture that consists of two steps: primary description and classification. Primary description is described as a process consisting of the deconstruction of the object under study into small relevant units or attributes; the subsequent process of classification is described as a an ensuing arrangement of the information obtained into a coherent structure so that patterns can be determined (Cobas Fernández 2003, 24). It is interesting that the author explicitly puts classification after description, and that classification is seen as a pattern-searching mechanism rather than an information-structuring one. Furthermore, the author describes her approach as being bottom-up rather than top-down, since it builds from the simplest up to the more complex, rather than starting with predefined categories in order to determine what to keep and what to discard. We cannot agree more.

Taking tagged data storage and Newton soups as inspiration, (Gonzalez-Perez, 2002, section V.2.3.2) introduced the concept of sub-structured information (‘información subestructurada’ in the Spanish source), i.e. information that is very weakly structured. In contrast to the previous two approaches, which are motivated by computer-oriented reasons, sub-structured information is introduced in the context of representational systems as a way to minimise metainformation redundancy (GonzalezPerez 2002, 145-147). Although the issue of category bias is not explicitly mentioned, the motivation for this approach is based on a trend to reduce categorisation as much as possible.

From an information technology perspective, typeless approaches to information modelling do not seem to be very popular in academia. If at all, they have been explored in the context of programming languages and data storage. Vineyard (1997) suggests tagged data storage, which uses tags to store typeless data in a database. In this approach, ‘objects’ are composed of ‘tags’, each tag having a name plus a piece of data. Objects are equivalent to valsters in our approach, and tags are equivalent to values, with the exception that there is no specification of a data type. The motivation behind the tagged data storage is not mitigating category bias, but, according to the author, increased power and flexibility when writing software that stores and retrieves the objects.

Finally, the Existential Programming approach of (Wallace, 2007) shows some interesting parallels with the approach presented here. The author claims that this approach ‘means to embrace the notion that existence precedes essence, i.e. develop data models, object models, programming frameworks, etc. without imposing a single entity-relationship model, object-oriented class hierarchy, ontology, etc.’. Also, the author’s statement that ‘When objects are independent of a single ‘‘strong data type’’, they can integrate multiple ‘‘strongly typed’’ data models’ resonates with our notion of a posteriori categorisation as an overlay onto the data. However, there seems to be no academic discussion of Existential Programming as far as we know, and how these vague declarations are supposed to be implemented remains to be explored.

In the mid 1990s, Apple launched the Newton range of handheld computers. The Newton did not employ a conventional file system to store files as most personal computer systems do nowadays; instead, it used an approach based on information ‘soups’. A soup is a database where ‘frames’ are stored, each frame consisting of a collection of ‘slots’; in turn, a slot consists of a name plus a piece of data. Again, the mapping is clear: soups are equivalent to models in our approach, frames to valsters, and slots to values. And, again, no data type is specified for the data itself.

8. Conclusions The process of describing an archaeological entity is almost always assumed to start by classifying that entity into one of several available categories. This assumption has established the basis for classical category-based approaches to information modelling, which often produce category bias, which, in turn, causes problems of two kinds. On the one hand, there is a situational mismatch, i.e. the category system is being applied to a situation with potential incompatible parameters (namely, scenario, background, purpose, and limitations). On the other hand, the consequences of objectivism become relevant, and especially serious in the case of interpretive categories, since a mismatched category implies a forced accommodation of reality into an ill-fitting mould. The ultimate result of this is that, when category bias occurs, the information that we generate as part of the archaeological description process is different to what it

In addition to regular slots, a frame in this approach can contain a special slot identified by the name ‘_class’, which points to another frame that is considered its class. The word ‘class’ here means, in fact, an implementation of the concept of category. The ‘_class’ slot is completely optional, and thus this approach is a truly hybrid that implements the notion of a priori categorisation on top of a category-less approach. In this regard, it is up to each individual using the Newton data storage system and creating a frame to assign it to a class (i.e. category) or not. The NewtonScript programming language 85

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process INCITE Programme) by the Xunta de Galicia, Spain. The author would like to thank Isabel Cobas for her comments to a draft of this work.

would have been in absence of the bias. This is a serious problem. In order to escape category bias, we propose an approach based on avoiding classical categories altogether as an a priori classification mechanism. This means that archaeological entities are described in terms of their observed properties, rather than in terms of what kind of thing they are. Categories can still be used as a posteriori instruments to explore the emergent patterns in the information, rather than to create structure in it.

References Archaeology Data Service 2011. Archaeology Data Service / Digital Antiquity guides to good practice. Available at: http://guides.archaeologydataservice.ac.uk/ Burke, H. and Smith, C. 2004. The archaeologist's field handbook. Allen and Unwin.

The environment where information description takes place can be standardised for a given realm (such as an organisation or a project) as a set of enumerated types, a set of properties and a set of a posteriori category systems; these establish a shared semantic context that is easier to reuse and apply than one created from a conventional, ontology-based approach.

Cobas Fernández, I. 2003. Formas de representar, mirar e imaginar: metodología para el estudio de la decoración geométrica en la prehistoria reciente, in Tortosa, T. and Santos Velasco, J. A. (eds) Arqueología e iconografía: indagar en las imágenes. 14-40. L'Erma di Bretschneider. Codd, E. F. 1970. A relational model of data for large shared data banks. Communications of the ACM 13, 377387.

A few lines of research need to be developed in the future in order to complete this work. A major one involves the exploration of alternative options to the category-less approach that could mitigate or avoid category bias. We said that the simplest way to avoid category bias is to avoid categories altogether but we are aware that there might exist other options. For example, formalising nonclassical categories such as radial, generative, and graded categories, among others, into an information modelling language, is worth exploring. As far as we know, there have been no research efforts whatsoever in this direction.

Chen, P. and Pin-Shan, P. 1976 The Entity-Relationship Model - toward a unified view of data. ACM Transactions on Database Systems 1, 9-36. Gonzalez-Perez, C. 2002. Sistemas de información para arqueología: teoría, metodología y tecnologías. British Archaeological Reports International Series 1015. Oxford, BAR Publishing. Gonzalez-Perez, C. and Henderson-Sellers, B. 2008. Metamodelling for software engineering, Chichester, UK, Wiley.

Another research problem that needs to be addressed is the issue of user interaction with typeless information models. Since a category-less approach, by definition, is less structured than a conventional, category-based one, the on-screen visualisation of information presents additional challenges: the information is harder to organise, there are fewer visual clues that aid the user in navigating the data, and, in general, it is easier to ‘get lost’. Richer user interfaces that exploit a posteriori categorisation of valsters constitute a promising avenue of research.

Gould, S. J. 2000. Wonderful life: The Burgess Shale and the nature of history, Vintage. Green, P. and Rosemann, M. (eds) 2005. Business systems analysis with ontologies, Idea Group. ISO 2006. Information and documentation - A reference ontology for the interchange of cultural heritage information. ISO 21127:2006.

Finally, the crucial role that the semantic context plays in a category-less approach means that any software tool that manages typeless information must offer an excellent degree of support for semantics. Although this is a relatively mature field, most recent work in semantics assumes category-based underpinnings, and (tacitly or explicitly) puts the emphasis on entities rather than properties. Enhancing semantic support at the property level is one more research line that will help complement this work.

ISO/IEC 2005. Unified modeling language (UML) version 1.4.2. ISO/IEC 19501: 2005. Kahneman, D. and Frederick, S. 2005. Representativeness revisited: attribute substitution in intuitive judgment, in Gilovich, T., Griffin, D. and Kahneman, D. (eds) Heuristics and biases: the psychology of intuitive judgment. 49-81. Cambridge University Press.

Acknowledgments

Korfhage, R. R. 1997. Information storage and retrieval. Wiley.

This work has been partially funded within Project MIRFOL (research grant number 09SEC002606PR of the

Lakoff, G. 1990. Women, fire, and dangerous things. University of Chicago Press. 86

González-Pérez: Typeless Information Modelling to Avoid Category Bias in Archaeological Descriptions Lakoff, G. and Johnson, M. 2003. Metaphors we live by. The University of Chicago Press.

Rumbaugh, J., Blaha, M., Premerlani, W., Eddy, F. and Lorensen, W. 1991. Object-oriented modeling and design. Englewood Cliffs, NJ, Prentice-Hall. Vineyard, J. 1997. Tagged data storage, in Dr. Dobb's Journal. Available at: http://drdobbs.com/database/ 184410275.

McKeehan, J. and Rhodes, N. 1996. Programming for the Newton using windows. Morgan Kaufmann. Meyer, B. 1997. Object-oriented software construction. NJ, Prentice-Hall, Upper Saddle River.

Wallace, B. 2007. Introduction to existential programming blog. Existential Programming. Available at: http://existentialprogramming.blogspot.com/2007/09 /introduction-to-existential-programming.html

Peterson, D. (ed.) 1996 Forms of representation. Exeter, UK, Intellect, Renfrew, C. and Bahn, P. 2004 Archaeology: theories, methods and practice. Thames and Hudson.

Wikipedia contributors 2011. Texas sharpshooter fallacy. Available at: http://en.wikipedia.org/w/index.php?title= Texas_sharpshooter_fallacy

87

 

6 The Spatial Construct of Social Relations: Human Interaction and Modelling Agency Mu-Chun Wu and Gary Lock School of Archaeology, University of Oxford Abstract The relationship between the spatial and the social has always been of major interest in spatial analysis. Recent research by social theorists has shown that not only is space a social construct, it is also a constituent of social relations. In terms of a spatial analysis in archaeology, most research has been focused on how social structure is reflected in spatial configuration, and how spatial layout supports and consolidates social order. However, the spatial construction of social relations is rarely discussed. This research argues that interpersonal relationships are not entirely based on social identities, and social relations should also be investigated, regardless of their hierarchical status, but through intimate human interaction. By applying Ingold’s ‘wayfaring’ theory, this research models human agency from a ‘meshworked’ perspective, and demonstrates how social relations are influenced by agents walking around a settlement. We suggest that, by taking a meshwork approach, researchers can examine the agents’ social relations in more detail and gain better control over their relational attributes. Furthermore, this bottom-up approach allows the examination of the allocated social relations as opposed to the delegated social identity, and benefits from understanding internal transformations. Keywords: Space Syntax; GIS; Social Relations; Agency; Meshwork; Movement; Taiwan; Saqacengalj. _______________________________________________________________________________________________

1. Introduction

However, many spatial technologies are tasked to provide an explanation of how and why certain artefacts,

Archaeologists have long been aware of the importance of the spatial component within the archaeological record. In fact, much, if not all, of the data archaeologists recover are spatial in nature, or have an important spatial component. Researchers have always been interested in spatial patterns and their implications for material objects (Lock 2009, 2010). The classic example is the widelyapplied ‘settlement pattern’ approach, which has long held an abiding position in terms of spatial analysis. This involves the mapping of sites according to their functional and hierarchical places within the settlement system, and has became widely influential (Meskell 2004, 12).

structures or sites are located and ordered; rarely do they serve to investigate how people interact with the existing spatial structure. Human interactions and social relations are usually discussed through material interpretation although research from various social theorists (Bourdieu 1985, 1989; Giddens 1985; Gregory 1985; Ingold 2011; Massey 1985; Pandya 1990; Saunders 1985; Sayer 1985; Soja 1985; Urry 1985) has shown that space is not only a social construct, but also a constituent of social relations. There is, therefore, no reason why spatial technologies cannot be applied to examine the formation and restructuring of human interaction and social relations. Here we propose that there are three levels of relationship between spatiality and social relations: (1) the ideational governing structure which produces spatial layout; (2) the physical spatial layout which reproduces and maintains the same structure which produced it; and (3) the actual social relationships which are constrained by the spatial layout. These three levels are not entities with separate existences; rather, they are involved in a continuous ongoing entangled relationship. By applying Ingold’s recent theory of ‘wayfaring’, this paper will explore the possibility of using ‘spatial technologies’ as a tool to model ‘meshworked’ social relationships from spatial relations in order to reveal the spatial construct of social relations.

Numerous spatial statistics and their spatial technology applications support spatial analysis. Technologies such as nearest neighbour analysis, network analysis, movement analysis, activity area analysis, and visibility analysis are adopted and applied at different scales. These approaches enable archaeologists to compare spatial patterns and establish meaningful results from what seem to be random distributions. Spatial technologies have also informed us of past human beings’ understanding of landscape and spatial conception, as well as their perception of the environment in which they dwell.

88

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency and furthermore, suggests that cultural complexes are defined by the clustering and association of material remains and artefacts. Once such complexes are recognised, it is then possible to correlate spatial features into patterns which represent former social institutions. To Rouse, a settlement is merely the temporal locale where a particular community carried out its various activity components, but is not a component in itself. The value of a settlement pattern investigation is only significant in the fourth procedural step of interpretation when the previous three steps have already been established.

2. Settlement pattern, spatiality and social structure ‘Settlement pattern’ as a focused methodological concept of archaeological research was brought to the forefront by the Virú Valley Project conducted by Gordon Willey in 1946. He regarded an archaeological settlement pattern as evidence of the relationship between human groups and the natural environment, and contended that the way human groups adapt themselves to nearby natural resources can be revealed by studying these patterns. Although Willey had a strong ecological anthropological background, he also observed that many other factors of a social and cultural nature were reflected in the archaeological record and therefore, he refused to treat these as merely being a reflection of general patterns of ecological adaptation (Trigger 1996, 376). In consequence, with respect to other social and cultural factors, Willey used the settlement pattern to approach the function and structure of specific past societies.

In the continuing debate of settlement pattern as an ‘approach’, Willey concluded that the investigation of a settlement pattern did not, and could not, in itself comprise a self-contained approach to prehistory (Willey 1968, 208). He agreed with Rouse that the concept of an archaeological settlement is not on a primary level of investigation, rather, it is important on the secondary level of investigation, that of interpretation. However, Willey also raised the problematic issue of the difference between cultural and social by questioning that, if archaeologists treat settlement patterns as a cultural trait, why cannot settlement serve as a device in the definition of a cultural complex or cultural phase (Willey 1968, 213)? That is, when settlement patterns are classified into types (such as building structures, spatial arrangements, and modes of dwelling), they should be able to suggest cultural distinctions just like other material traits.

Following Willey’s research, Chang stated that this approach had overlooked the importance of the ‘whole community’ and the ‘entire settlement’, and was confined by concepts such as types, styles, and attributes of ‘artefacts’. He argued that archaeological types range in multiple scales from the single attributes of artefacts to entire cultural traditions and, therefore, using the unitary social group to describe and characterise a prehistoric people can also be considered to be an alternative to a range of various scales (Chang 1968, 2). On this premise, Chang proposed a pattern investigation of three categories: households, communities, and the aggregate of communities (Chang 1958). By examining a settlement using these three scales, archaeologists are able to start to understand the social structure, and furthermore, social organisation, in the local and wider region. This particular methodological framework proposed by Chang is focused on the acquisition and interpretation of archaeological data, by way of defining, characterising, and relating prehistoric social groups so that social ‘structure’ is its main goal, and reconstruction, or ‘articulation’ in Chang’s terms, is the means to achieving that goal. Although the ‘structure’ inevitably falls short of the ideal of past reality, ‘articulation’ can be conceived and pursued as a scientific process, and can be improved upon constantly (Chang 1968, 2). The significance of using settlement patterns to reconstruct prehistory is seen to be the result of various ‘organisations’ or ‘institutions’ which are reflected in those patterns.

The main reason for the settlement approach paradox lies in the obscure relationship between the cultural and the social. Rouse argues that temporal prehistory is composed of different cultures across the landscape, whereas Chang breaks down the temporal landscape into separate regional social community areas. The former sees prehistoric people as being bound by culture, which is reflected in artefacts, and the latter sees them as being bound by social structures, which are reflected by regional distribution patterns. Willey also noted this different perspective and stated that his difference with Chang was not ‘over the inevitability of a close relationship between archaeological field methods and cultural-historical interpretation, but over what I (Willey) believe to be the necessity of taking certain basic methodological steps prior to the settlement patternprehistoric society interpretation step’ (Willey 1968, 211). However, Willey did not specify the kind of information which needs to be established in order to be considered appropriate for a ‘second level’ of investigation, i.e. interpretation. Following his statement that the culturalhistorical aspect is not a factor of his disagreement with Chang, the logical sense would be to examine the attributes of artefacts regardless of their cultural category, and reconstruct their structural order and relationship. In this case, there still remains the question of why the settlement pattern, referring to its house structure, building material, distribution pattern, location selection, etc., cannot be seen to be a type of ‘artefact’ like other

2.1. The settlement pattern approach However, during the early development of settlement pattern studies, many researchers questioned the logistics of this approach. For instance, Rouse argues that there are four logical steps in reconstructing prehistory: the recovery, classification, reconstruction, and interpretation of the remains (Rouse 1968). He argues that it is impossible to define a meaningful prehistoric settlement without the help of a defined cultural complex 89

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process material remains to ‘approach’ prehistory. To answer this question, a further examination of space and society is needed.

data alone. This spatial perspective is not aimed to resolve all the broad cultural issues, but to answer questions of a social-relational nature. The beginnings of social archaeology once again highlight the difference between a cultural and a social perspective of prehistory although this is not to deny the significance of material culture studies. Material remains are still at the very heart of identifying social status, network flow, identity, subsistence, etc., and any distinction merely implies different interests in prehistory, some focussing on activity context, and others on human relations. The relationship between the two is complimentary rather than contradictory.

2.2. Spatiality and social structure Some Geographers, and this is by no means restricted to geographers, assume that there is a spatial cause behind every spatial pattern that characterises the spatial structure and, henceforth, can explain its demise (Massey 1985, 11). The argument about whether ‘purely spatial processes’ exist, is similar to, but not necessarily the same as, the ‘settlement pattern approach’ of the 1970s. It was formerly argued that only social processes were operating over space and, therefore, the spatial was only the result of social constructs. However, as Sayer pointed out later, although it is common to separate space and substance, and speak as if they are independently existing objects, it is actually the deceptive structure of the human language which separates the three inseparable aspects of human behaviour: the spatial, temporal, and process (Sayer 1985, 52). Soja also suggested that spatial structures and relations are the ‘material form’ of social structure and relations (Soja 1985, 94). So, that only the structural ‘idea’ of human behaviour and interaction exists together with the ‘physical appearance’ of this idea. The ‘idea’ is in the mind but the ‘physical appearance’ is in the combined body of space, time and artefact. The realisation that social life is materially constituted in its spatiality is the keystone of the later interpretation of spatiality. Furthermore, space is not simply a social construct, but it also limits and constrains human social interaction. The one-way relationship of ‘social structure’ equals ‘spatial distribution’ ignores the crucial impact of spatially organised opportunities and limitations of social interaction. Gregory has argued that social relations are not merely expressed within a spatial structure, but they are, in fact, also constituted through that very same structure (Gregory 1978, 120). As a result, the analysis of the spatial structure itself is not secondary to the analysis of the social structure but the relationship between the two is actually a recursive one; one cannot be theorised without considering the other. Therefore, spatial order should be understood as both the outcome and the medium of social action and relationship with the spatial structuring of social life defining how social actions and relations are materially constituted and made concrete.

2.3. Spatial pattern and space syntax In applying Renfrew’s social perspective, the issue is how to extract a meaningful pattern from space itself and how a spatial pattern provides significant meaning. According to Renfrew the landscape is broken down into territories that represent individual social units. The spatial patterning and relations of these territories can then be utilised to investigate their social relationships. In the study of architectural space, there is a very similar approach to revealing spatial meaning, that of Space Syntax. Proposed by Hillier and Hanson (1984), space syntax theory provides archaeologists with a wider view of spatial relationships by arguing that space is not just a neutral background, but a fundamental component of human activity. Furthermore, it is proposed that, by breaking space down into individual units, these can be analysed using formal methods, a premise in alignment with Renfrew’s earlier proposal. The spatial units approach has since been adopted by many scholars and later evolved into methods such as ‘access graphing’, ‘axial mapping’, and ‘convex mapping’, which have impacted on different aspects of spatial studies (Batty 2004; Foster 1989; Hillier 1984; Turner 2004). In terms of settlement pattern analysis, space syntax has been widely applied by archaeologists to investigate household relations, road networks, spatial movements and spatial patterns in general. Among all the space syntax analytical methods, access analysis concentrates on the differentiation and integration of space, enabling the investigation of the interconnection of certain spatial components, and highlights the location of high-level integration. With the use of reconstructed road networks, the basic network connection can then produce an access graph diagram for further analysis. This includes an investigation of household spatial relationships, integrations and separations and can help to identify distinctive areas and divisions which may indicate the presence of individual settlement segments. Furthermore, the division results can be examined either individually or in their entirety in order to highlight the characteristics and relationships between the segments. Applying an access analysis aids the possibility of identifying separate social groupings in

Despite Willey’s objection towards the settlement pattern as an approach in itself, later archaeologists accepted it as a possible perspective (Hodder 2004; Patterson 2004; Renfrew 1984; Trigger 1968, 2004). In his book, ‘Approaches to Social Archaeology’, Renfrew stated that, by adopting a spatial approach, and breaking down the landscape into discrete social units each of which occupies a definite territory, archaeologists can start to investigate social relations (Renfrew 1984, 10). He acknowledges the limitations of this approach in that it only uses one of a number of possible pieces of information, i.e. space, but nevertheless argues, although rather simplistically, that it is possible to begin to investigate questions of dominance and power from these 90

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency the settlement and provides insights into their social structure.

particular space. However, these can only be meaningful to those who are familiar with these symbolic representations, and such presentation can, in fact, be more difficult to understand than the original site plan. Nonetheless, space syntax is not mainly about how visual graphs can represent reality in detail, although the alternative perception offered does provide unique insights into understanding space; it is more concerned with statistical analysis and pattern-identification, and the social structures and relationships that these can help us understand. Therefore, there are many ways in which an access graph or a convex map can be improved, either by adding functional symbols, separate floor heights or assigning distance gaps. These extra attributes assist in looking at specific correlations and distributions, although the major issue is still the lack of human representation.

Following Hillier and Hanson, space syntax and access analysis have been broadly applied to studying settlements in archaeology. For example, Mark Grahame has used an access analysis as a framework to investigate the spatial structure of Roman houses in Pompeii, showing that the central courtyard, which provides the main access route leading to closed rooms, served as a main defence mechanism for inhabitants against outsiders (Grahame 2000). Looking at historic Zuni architecture, Ferguson showed how convex mapping could provide information that was instrumental in understanding the spatial layout and division of social activities (Ferguson 1996). However, in addition to the quantitative technique and pattern-identifying mechanism space syntax has to offer, it can also serve as a ‘tool to think with’. Marion Cutting (2003, 3) has categorised the analytical process into four stages: the identification of convex spaces; the insertion of the links joining those spaces; the representation of the linked spaces visually through a justified access graph; and the calculation of the numerical relationship (usually expressed in terms of integration values) between those spaces. The first two steps in this process are usually the most difficult to define in terms of analysing settlement layout. A single convex space is defined as an identified space within a layout plan, likely to be a somewhat controversial process as, unlike architectural buildings where space syntax originated, archaeologists are usually faced with settlement plans involving open spaces. It is difficult, if not impossible, to recreate a spatial division if it is not apparent in the archaeological record, not to mention the problem of different individual perceptions of spatial divisions. However, the process of forming a definition stimulates the redefinition of space itself and its function, and forces rethinking the relationship between human behaviour and spatial structure. It is this ‘tool to think with’ aspect that can be of most use to prehistorians.

Space syntax analysis has always involved the depopularisation of space as, when defining a single meaningful space, it is difficult to know how the furnishings and human activity originally took place, since they have disappeared from the archaeological record. This aspect of Hillier and Hanson’s work has received most criticism because it treats space as a totally independent discourse in itself. The behaviour and decision-making of human activity is removed from the construct, leaving only the spatial distribution behind, an assumption which has a major impact on the social interpretation and the use of recognised structures. That is, how do we, as archaeologists, link back the spatial pattern to human activity and its social implication? Originally space syntax emphasised the examination of ‘control over space’ (Hillier 1984), a purely spatial relationship point of view that leaves very little room to take function and use of the space into account. Hillier and Hanson were most concerned with the movement, encounter, and direction of access between outsiders and inhabitants. The idea of inhabitants and outsiders aimed to emphasise access and control over the space in which they dwell rather than also including the symbolic or practical function of space. As Fairclough rightly put it, ‘the reason for access, and the frequency and nature of movements and encounters, tend not to be considered in detail, and other methods are required to contextualise and explain’ (Fairgclough 1992, 350). Space syntax provides a way of identifying spatial distinctions and layout patterns, categorising spatial divisions and possible structural order. However, in terms of social relations and human interaction, there is no direct link between pattern and social structure. Space syntax is a single perspective on space, a useful analytical tool, and also a ‘tool to think with’, but it is not a ‘theory of space’.

Despite the spatial definition problem, there are some other limitations and shortages regarding the original access analysis and convex mapping; the most apparent issue is that both are ‘adjusted graphs’. This results in the spatial location and connection being misplaced along with the loss of many other spatial attributes, such as location, size, distance, function, etc. Many researchers have managed to work around this issue by devising special maps, for example in his Bolton Castle planning diagram, Faulkner (1963) assigned alphabetic symbols to each space in order to represent different spatial functions, and, in addition, the layout of the diagram is in accordance with horizontal spatial accessibility and vertical floor difference. In his planning diagram of Edlingham Castle, Fairclough (1992) further assigned different kinds of lines to represent different means of accessibility, and different shapes of convex to represent their functions or locations. To a certain extent, these later graphs do assist in clarifying the articulation of that

3. Community and human interaction Spatial structures and relations are, then, the material form of social structure and relationships and the identification of a settlement pattern should reflect its social structure and the various institutions of social 91

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process interaction (Willey 1968). The issue is therefore, how do we repopulate the de-humanised spatial pattern?

social relations. One major issue common among archaeological definitions of social institutions, like the community, is that they often end up as a method for recognition rather than informing analytical theory. This is mainly due to the fact that defined social institutions are, by nature, constrained by the limits of the material record. The emphasis on recognition over analysis allows concepts like ‘community’ to remain unproblematic, because the object studied arises directly from the data (Canuto and Yaegor 2000, 5). Since material remains are regarded primarily as being products or by-products of human behaviour in social and cultural contexts, they are, by their nature, something more than just systematic patterning data. Settlement archaeology should characterise itself as the archaeology of humans in settlements more than the archaeology of the settlement in its physical sense. In addition, Canuto and Yaegor have pointed out that pattern identification approaches frequently ignore issues of social creation, manipulation, and meaning which have become increasingly important as archaeologists inject concepts of agency, practice, structuration, and interaction into models of the past (Canuto and Yaegor 2000, 5). One of the shortcomings of structuralist and post-structuralist approaches is that the role of social agents is often overlooked. The growing interest in agency and interaction highlights the appropriation of social institutions through constant daily practices which shape and reshape social relations. Consequently, there is a growing interest in the utilisation of the form and function of a settlement pattern to ‘reflect’ social institutions so that community is conceived as a dynamic socially-constituted institution which relies on human agency for its creation and continuing existence. The social interest has shifted from the archaeology of ‘settlement patterns’ toward the archaeology of ‘communities’.

On the basis that a settlement pattern is defined as a study of social relationships using archaeological evidence, the identification of the social group is crucial. Here we refer to the ‘scale’ of the social group rather than the ‘cultural background’ of the people, as Chang (1958) rightly pointed out, cultures can fluctuate, whereas social groups are clear-cut. The first step in studying a settlement pattern is to delimit local social groups rather than identifying archaeological regions and areas by timespacing material traits, since cultural traits are meaningless if they are not described in their social context. Material and cultural traits are supplementary tools useful to interpret discovered social relationships and contexts. In order to identify basic social groupings, both Chang and Trigger have drawn attention to the concept of ‘community’ and considered it universally manifest in the archaeological settlement record. 3.1. The archaeology of community ‘Community’ to both Chang and Trigger means ‘a group of people sharing the same interests in the land in which they dwell’. Since this group of people live side-by-side with each other, they would naturally form a distinctive single society regardless of their political, cultural and religious backgrounds. In general, therefore, a community corresponds to a single settlement, and can be identified as an archaeological component. The composition and social implications of this group are usually investigated from three different perspectives: individual structures, the settlement, and the relationship between settlements (Chang 1958; Trigger 1967). Although different terms are used, such as ‘house’, ‘community’ or ‘aggregate’, the underlying principle remains the same and highlights a settlement pattern focus as being an analysis of individual ‘house structures’, a ‘community’ consisting of multiple houses, and a wider ‘region’ involving multiple communities. In order to reveal the social relationships implied in these spatial patterns and structures, Trigger pointed out that the layout of communities is strongly influenced by family and kinship organisation in nonindustrialised societies, and there is a high degree of correlation between community plans and village social life (Trigger 1968, 62).

Traditionally, archaeologists have defined community as being a residential unit such as a ‘site’ or a ‘settlement’ although anthropologist George Murdock’s definition of community as ‘the maximal group of persons who normally reside together in face-to-face association’ is now widely accepted. It is not hard to see from this that the human community is essentially produced by two things: (1) shared residence or space, and (2) human interaction based on face-to-face association. George Homans has argued that community is a necessity due to the material interdependency of people requiring human interaction which, whether consciously or unconsciously, produces group sentiments and identity (Isbell 2000).

Based on this, the correspondence between specific community patterns and certain social forms, such as bands, chiefdom, and states, has been attempted. Chang (1958) categorised communities into ‘Siberian’ types and ‘Eskimo’ types based on present groups, kinship lineage system, and social composition, in order to identify and correlate their settlement layout pattern. Trigger and Renfrew have both discussed the possible social implications by scrutinising particular spatial structures (Renfrew 1984; Trigger 1968), however, they have acknowledged that a direct link between certain spatial structures and specific social institutions does not exist, suggesting that mere spatial patterning and structure is not enough to understand intimate human interaction and

In localised regions, such as a small settlement, community identity is formed by its members’ daily routines such as subsistence reproduction from the nearby environment and understanding survival within the place in which they dwell. Their common pragmatic concerns produce a shared behavioural reaction to adapt to the physical and social world, similar to Bourdieu’s ‘habitus’ (1977). This structures their perception and guides their ‘practices’ in the world, and eventually consolidates a sense of ‘being in common’, which leads to the existence of a community and mental identity. It is this practice 92

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency theory perspective that has stressed ‘human interaction’ most strongly and positioned agency at the centre of creating and recreating society. Instead of seeing community as a place where social interaction and reproduction takes place, the agent-orientated approach views social institutions as being socially constituted.

dominance of transformation.

and

the

internal

social

Research into community is aimed at investigating human interaction and social relations and the first step is to identify the units of analysis. Since the focus is social in nature, these should be a type of social group, or be relevant to the notion of its social identity. As Renfrew stated (1984, 39), a social group cannot be analysed if one cannot recognise it, for it is the recognition of the group that governs group behaviour and provides archaeologists with the path to investigate social relations. This being the case, the social group ought to be seen, if possible in the archaeological record, although a number of social groups and identities which fall into the category of ‘imagined community’ are quite difficult, if not impossible, to identify, thus greatly increasing the analytical challenge.

However, this perspective of ‘community’ presents additional issues for archaeological analysis. To begin with, community is a social, rather than an archaeological concept which cannot be excavated or identified, even though the inhabitants of an area can be inferred to be a community. This issue is not hard to overcome, and appropriate spatial patternings have been widely discussed (see below). The most problematic issue, however, is that this interactive perspective on community unveils the potential of non-spatially-based social network groups. For this approach the core of the community is an idea of membership and identity which is not necessarily rooted in a fixed location or space. Furthermore, now that the concept does not exclude other types of social groupings, the issue of multiple identities, such as those based on ethnicity, gender, lineage, or faction becomes important.

Furthermore, we would argue, Isbell’s rejection of the ‘natural community’ is an error reflecting those of the 1960s and 1970s because this still provides an alternative spatial perspective which is significant in itself. The natural community should be regarded as being a ‘neighbourhood identity’ and retain a position in the ‘imagined community’. The logic is the same as for the previous social and cultural argument and although this does not answer all the questions, it still serves as a meaningful approach among many others. Accepting that social identities, such as kinship and ethnicity etc., are hard to distinguish among archaeological evidence, space and ‘neighbourhood identity’ presents itself as a more fixed and stable factor of human relations.

Isbell has proposed that the archaeology of ‘community’ faces a crossroads (2000, 263). On one hand is the ‘natural community’, a real and bounded entity based on a spatial pattern, and on the other the ‘imagined community’, constructed through discourses and processes. The advantage of adopting an ‘imagined community’ perspective is that it populates the past with agents who act as interested individuals to examine social fractions. Furthermore, it promotes an interest in contingent change within an individual ‘culture’ rather than the universal cultural evolution based on a comparison of settlement patterns. Based on this, Isbell suggests that archaeologists should abandon the ‘natural community’ concept and adopt the ‘imagined community’ (Isbell 2000, 263) although here we partially disagree with this.

3.2. The community of neighbourhood and modelling agency In terms of the relationship between spatiality and social relations, we propose three different levels: (1) the ideational governing structure that produced the spatial layout; (2) the spatial layout that reproduces and maintains the very same structure in which it was made; and (3) the social relationship which is constrained by the spatial layout. The first two levels are relatively straightforward as they are essentially the same as the early ‘settlement pattern’ approach aimed to reveal. However, as previously stated, the ordered structure revealed is merely the stratigraphy of status and does not illustrate the kind of social relations and human interaction which actually took place. Therefore, the third level introduces individual agents into the space in order to model and investigate human behaviour.

This community concept shift from a spatial to an ideological base has resulted in 'community' becoming a redundant alternative for the term 'social'. It is also gradually becoming a collective concept including 'kinship', 'economic unit', 'ethnicity', 'wealth division' etc; community has become a substitute term for the previously established term 'social'. This is not to deny the contribution and potential of the 'imagined community' which has stimulated a further investigation of the obscure concept of 'social' and 'societal', and stressed the significance of a different 'identity'. More importantly, an individual agent’s interaction within a community cannot be overlooked so that the community serves as a crucible where multiple and potentially incompatible or antagonistic identities, such as factions, lineages, genders, and ethnicities interact, competing with, or complementing, one another (Canuto and Yaeger 2000, 7). As Isbell correctly suggested (Isbell 2000, 249), it benefits archaeologists to investigate further the

Mehrer has shown (Mehrer 200, 50) that the dominance of power is epistemologically different between top-down and bottom-up perspectives. The source of social power is divided into allocated and delegated, whereas local elites gain delegated social dominance from a superior authority, people also receive allocated social power and resources from commoner neighbours. Our proposed third level can highlight this structural difference and allows for the possibility of a contingent change within 93

 

power

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process the social group. The community concept focuses the contrast between individuals and local and regional scales, a potentially fruitful approach in understanding social transformation.

the ABM in the British Museum (Turner 2004, 2007), where agents move around based on their line-of-sight in the building. The goal of ABM is to simulate human interaction within a computer program and effectively retrieve behavioural and social encounter information for reference and interpretation. However, the successful application of ABM not only involves extensive knowledge of computer programming, but most importantly, a reasonable and detailed understanding of how humans would behave in different circumstances based on various kinds of social status, profession, gender, etc. ABM remains an under-developed technology, an approach which needs to be explored and researched more extensively, and here we prefer to retain the term and approach of ‘modelling agency’.

A spatial construction/settlement pattern does not just ‘reflect’ the kind of ‘social organisation’ which may have occurred, but also potentially shows how people interacted with each other. On the one hand, the spatial layout is a social construct reflecting how society was structured whilst on the other, due to the fixed layout of the spatial structure, people dwelling within it are constrained by it. Their interaction, therefore, can be modelled effectively based on individual spatial behaviour using a modelling agency. The issue of agency has become important when discussing how individuals perceive and react toward a wider construct, such as that of culture and social entity. Anthropology and sociology have shown how individuals influence and interpret the world in which they dwell. The ‘practice’ and ‘habitus’ theory proposed by Bourdieu has been influential in the switch from studying the collective consciousness to investigating the perception and behaviour of a ‘native’ view (Bourdieu 1977). Archaeology has followed, Christopher Tilley and phenomenology for example have helped the move away from typology, structuralism, and technology, to thinking about the agency of perception, senses, and individual behaviour (Tilley 1994).

Our main objective is to explore agency modelling in a social-relational context. Most existing modelling of agency using spatial technologies has centred on the senses and perception of the world and on modelling individual movement across a given environment. The challenge, therefore, is how to model an individual’s relationships with a collective society from a social perspective. To address this we start with Bourdieu’s ‘practice’ and ‘habitus’ theory and how daily practices such as movement around domestic space, the discarding of refuse, and the construction of an oven, all have social weight. People gain their gender identity and social status, and form relationships by behaving and performing in a certain way; it is through one’s ‘action’ that one receives social recognition. The first step in recognising one’s social identity is, therefore, to identify its bodily practice and behaviour which is complicated because, although these practices can be observed through ethnography, they are lost in the archaeological record. Despite traces of practices being left in the assemblage of material culture, we cannot be sure how they relate to specific situations. Here, therefore, we take a similar, but alternative, view of modelling agency in a social context.

Such approaches have become widespread within landscape archaeology with an increasing interest in scrutinising how human senses perceive the surrounding environment, how they make sense of the information they perceive, and, in turn, how they behave and react to this information. This kind of agency-based modelling often relies on software applications like Geographic Information Systems (GIS), spatial analysis tools to model visibility, soundscape, and their relations with the immediate surroundings using spatial statistics and attributes (Conolly 2006; Fitzjohn 2007; Lake 2007; Llobera 2003; Llobera 2007; Lock 1995, 2000; Wheatley 2000, 2002). Furthermore, researchers are also starting to model how the perception of surroundings affects human behaviour, such as moving across a landscape (Batty 1999; Castle 2006; Llobera 2000, 2005; Turner 2007). The work of Llobera (Llobera 2000, 2005, 2007) is important here for establishing the methodological ground rules for GIS-based movement and visibility. He also provides a well-argued theoretical framework based on the social theory of inhabiting a landscape. These various kinds of modelling studies have demonstrated and highlighted the importance of agency in understanding the wider social-cultural construct.

In his recent book Being Alive, Ingold proposes a new way of thinking about social relationships (Ingold 2011) by introducing the idea that people perceive the world and form knowledge by embodying themselves in a ‘meshwork’ of reality. He argues that, in real life, people perceive not from a fixed point, but along a continuous itinerary of movement (Ingold 2011, 46) Thus, perception is a function of movement and what we perceive must depend on the way we move. Henceforth, locomotion, not cognition, must be the starting point for the study of perception, and it is through locomotion that everything surrounding us is woven into life, and lives are woven into everything surrounding us. As opposed to traditional ‘network theory’, where everything is connected through points and nodes, he proposes a spider-web like ‘meshwork theory’, where:

‘Agent-Based Modelling (ABM)’ is now a familiar approach in which computer-generated agents are programmed to behave and react to certain circumstances in a predictable way. By releasing multiple agents into a given environment, they can then be monitored, and their social interaction and activity recorded. One example is

‘the web is not an entity. That is to say, it is not a closedin, self-contained object that is set over against other objects with which it may then be juxtaposed or conjoined. It is rather a bundle or tissue of strands, tightly 94

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency drawn together here but trailing loose ends there, which tangle with other strands from other bundles…It is as though my body were found through knotting together threads of life that run out through my many legs into the web and thence to the wider environment’ (Ingold 2011, 91)

4. Case Study: Saqacengalj The settlement of Saqacengalj is located in Mudan County, Pindong District, at the southern tip of Taiwan (Figure 1). It is an abandoned settlement of the Kau-Shi people (originally called Kus-Kus) who belonged to the Paiwan ethnic group, one of the Taiwanese indigenous groups. The earliest document which mentions the KusKus was written in 1647-1656 AD by the Dutch, providing historical evidence that the Kau-Shi people have existed in some form for at least 350 years (Chen 2008). The slate-stone settlement was constructed using a dry-stone walling technique, and consists of 81 house structures situated on a gentle slope (Figure 2).

The meshwork is everything happening around us and the very reality in which we dwell. Thus, rather than viewing things happening around us as different entities in a linked network relationship with each other, we see everything as being interwoven or ‘meshed’ together. The way to understand this mesh is to walk through it and experience it or, in his words, to experience ‘wayfaring’. He argues that to perceive the environment is ‘not to look back on the things to be found in it, or to discern their congealed shapes and layouts, but to join with them in the material flows and movements contributing to their – and our – ongoing formation’ (Ingold 2011, 88). As to what we experience in life, things do not so much exist as occur. Lying at a confluence of actions and responses, they are identified not by their intrinsic attributes or their definition nametags, but by the memories they conjure up. Human behaviour is not the result of agency which is distributed through the network, but rather emerges from the interplay of forces which are conducted along the lines of the meshwork (Ingold 2011, 92). The world is, therefore, not classified into nametags nor is it entitynetworked, but meshworked. The way to understand the world around us is by walking through it, acting and reacting to upcoming encounters, and forming a memory of understanding and knowledge about what has happened. The similarity between Bourdieu and Ingold lies in the essence of ‘practicing’ or ‘acting’ any upcoming event and it is this behaviour that provides individuals with the way to perceive and understand the world. However, the same behavioural process does yield different implications. For Bourdieu, the process of practice is to define and redefine one’s self-identity, which is a fixed entity within the cosmological order. In contrast, the process of wayfaring is to understand one’s relative relationship with the rest of the world. For Ingold, there is no fixed entity since identity is a fluid concept, forever changing.

Figure 18 The location of Saqacengalj, southern Taiwan. The Kau-Shi oral history narrates how their ancestors migrated from the ocean to the Mudan region, how they selected Saqacenglj as their first settlement, and how internal conflict and resulting suicidal events made the place a socially and cosmologically defined ‘forbidden land’ which they were forced to abandon and then move to a new locale (Chen 2008). Based on their oral history, the Kau-Shi people moved and established alternative settlements six times before they settled in their current location. Oral tradition also suggests that their oldest and earliest abandoned settlement, Saqacengalj, was the first settlement established by their ancestors when they immigrated into the region about 500 to 600 years ago. Since then, they have abandoned their settlements and migrated five times to different locations before finally moving to and building their current village 50 years ago. Although there are plenty of ethnographic documents relating to the Paiwan ethnic group, most research has been focused on kinship or material culture studies and interest in settlement patterns or landscape archaeology

As previously stated, we aim to understand the spatial implications of social relations through agency modelling. The first two of the three levels of relationship outlined above between spatiality and social relations are better understood from Bourdieu’s practice theory, since they are concerned with social structure and the maintenance of the structural order. The third level, the actual social relationship which is constrained by the spatial layout, is more appropriately addressed by applying Ingold’s wayfaring theory here. Our case-study is an abandoned aboriginal settlement, Saqacengalj, southern Taiwan, in which we highlight the concept of meshwork in social relations, and how it can benefit archaeologists in understanding spatial layout and social relationships. 95

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process have been neglected. Despite this, ethnographic research still provides crucial insights into Paiwan social relations and how they reflect physical structures. We also know from ethnographical and historical sources that the Paiwan were traditionally a ranked society and practiced primogeniture.

roads; and (4) stairways. The household area was a place for private family indoor interactions; open spaces were where major social activities took place, such as group gatherings and festivals; the major roads were where daily social interaction such as occasional encounters and day-to-day chats happened; and stairways were a narrow passageway where social interaction rarely occurred. By applying these four types of activity areas, the settlement was divided into around 150 convexes (Figure 3).

4.1. Networked social relations According to a previous study of Saqacengalj, the settlement was divided into several distinctive areas or neighbourhoods (Wu 2010), each of which may have held a group of people of different social status within the hierarchy. This suggestion is mainly based on the classification of house size and the presence of large houses in every division or area. The results correspond with Chang’s segmented village characteristics in which each segment is arranged to surround special buildings, and all the segments further constitute a larger community. As one of the earliest Paiwan settlements recorded, Saqacengalj is most suitable for establishing an understanding of the fundamental settlement pattern of Paiwan culture, it serves as a cultural prototype uninfluenced by foreign cultural contact and provides a path towards understanding Paiwan’s original form.

 

 

Figure 20 Convex diagram of Saqacengalj. Square= household, Diamond= open space, Eclipse= road and stairway.

 

Since exploring social relations was the objective, each household was treated as a single social unit so as to investigate its relationship with others. It is argued that the settlement plan was designed to have social meaning and this, therefore, was used to investigate the relationship between each household. Wu also argued that the social separation between households was not defined by physical distance in such a small site, as this was not significant or apparent to a single agent, but alternatively it was measured by the number of social activity areas one must pass to go from one house to another, in other words the step depth. As a result, each household had a relationship with others based on the number of convexes crossed when moving around the settlement. Using step depth as a key variable, the settlement was divided into five segments, or neighbourhoods: A, B, C1, C2, and C3 (Figure 2).

Figure 19 Site plan of Saqacengalj showing individual houses and the three areas discussed in the test. In order to understand the configuration of the settlement layout, Wu took a space syntax approach by breaking the space down into individual convexes (Wu 2010). Since the main objective was to investigate social relations, each convex was defined by the kind of social activity it was able to hold, this ‘activity-orientated approach’ categorised the settlement into four different activity areas: (1) household areas; (2) open spaces; (3) major 96

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency Combined with the statistical analysis of house size and the size of different open spaces for social activities, Wu argued that there was a close relationship between the settlement’s spatial pattern and the social structure of the Paiwan people (Figure 4).

acquired from wayfaring. It is the relationship between the agent and the outside world that makes the meshwork meaningful. Three key features are fundamental to modelling a social meshwork: (1) a neutral agent without any attachment to a fixed structural identity; (2) a pathway which the agent can move along; and (3) a subject that is in an encountering relationship with the agent. Just like Ingold’s example of the spider’s web the agent is the spider itself, the pathway is the spider’s web, which it can move along, and the encountering subject is the prey that is caught in the web. Although this modelling may look similar to the network modelling by Wu described above, there is a fundamental difference. The pathway of the agent’s movement lays down the conditions for the possibility of the agent interacting with other households although they are not in themselves lines of interaction. The pathways can be seen as relations, although they are not relations between, but along. In Wu’s previous research, the social relation was described as the direct step depth relationship between one household and another, and just like social status, the step depth between the two was a fixed relationship. Whereas, in the modelling of wayfaring, the interaction between two households is dependent on how the agent moves around the site and may take a more distant route and decrease the possibility of interaction, or take a closer route to increase the chance of an encounter. In so doing, the relationship is not fixed between any two households, but relies on how the agent moves along the pathways in the settlement. Here we will demonstrate the difference between social relationships within a network and a meshwork by modelling wayfaring in areas A and B of Saqacengalj.

Wu’s research on Saqacengalj is a typical networked approach to social relations, where each household and open space was a social entity and all were connected in a relationship of social structure. Based on house structure attributes and spatial function, each household had a fixed social status in the wider social structure, just as each space had a fixed functional ranking in the socio-political construct. The social network relation was described through social identity within a structural composition. Therefore, Wu further discussed how social structure shapes spatial layout and, in turn, how the spatial layout supports and consolidates the cosmological order. Yet, this is but the first two levels of the relationship between spatiality and social relations discussed above, namely how social structure shapes the spatial configuration and how the spatial layout supports and consolidates the social order. It cannot answer the question raised by the third level: the actual social relationships which are constrained by the spatial layout. Social structure is a concept confined within the human mind; it is how humans think the world is structured and ordered, and what it should be. However, real social relations are often far from this ideal as people do not necessarily have closer relationships with their kin or someone who is of the same social status, and they are often good friends with someone from a different background or not related by kinship. Traditional network analysis is too constrained for all structural possibilities, and it is therefore unable to explore social relations beyond the rigid ideal social order. In consequence, the network approach is unlikely to answer what Mehrer (Mehrer 2000, 51) calls the ‘allocated power’ from the bottom-up and, in turn, it is difficult to address the issue of internal transformation raised by Renfrew (Renfrew 1984) as opposed to the exterior transformation.

  Figure 21 Correlation between social structure and spatial order (after Wu 2010). Figure 5 Pathway mesh of Area A and B.

4.2. Meshworked social relations

Initially we need to rebuild the three key features as described above. The first and third are quite straightforward as they are the households identified in the site plan. The second is the meshed pathway in which agents move around the site, the relationship of one agent with all the rest, and here we use ‘Least-Cost Path (LCP)’ analysis to reconstruct the pathways from one household to another and create a meshed web-like pathway system

Unlike a network, which breaks social relations down into individual social status entities, wayfaring is focused on the agent moving along pathways to experience and understand its relationship with the outside world. That said, in terms of modelling meshworked social relationships, the first thing is to discard the definition of social identity as the meshwork relationship is based purely on the agent’s movement and its knowledge 97

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process (Figure 5). The pathways based on one single household are then accumulated together to highlight the pathway density of its wayfaring (Figure 6). The contrast between Figures 5 and 6 highlights how an agent’s movements influence the perception and understanding of the same pathway system. Whereas Figure 5 shows all possible pathways between all houses as a single line, Figure 6 shows all possible pathways weighted according to the number of times used.

valuable insights into understanding network flow, it has different social implications. A common mistake concerning axial mapping is the misunderstanding of the ‘integration rate’ of an axis as being the probability of movement. The integration rate highlights the integration of a specific axis within a network and reflects the way in which a central pathway is situated; it mainly functions as a reference to interpret political control over others. The integration rate does not reflect how people move around the site, nor does it imply a higher or lower possibility of movement (Figure 9). Also, it could be argued that the result of the meshwork perspective is no different from that of the network perspective because the outcome still produces a fixed relational reference between two households and, therefore, the result still suggests a fixed social status. This is also not the case because, as mentioned earlier, the relationship between two households is dependent on wayfaring, that is, it depends on how agents move along the pathways. As an example, we can take an agent starting from house 0S4 and compare its relationship with household 00, the fluidity of this is highlighted by the difference between the agent visiting area A compared to visiting area B (Figures 10 and 11). In Figure 10, the number in each convex space shows how many times the agent has passed when visiting area A and, likewise, in Figure 11, when visiting area B, 6 for the former and only 4 for the latter showing that the social relationship of the two households is not fixed but, rather, fluid and subject to how one moves across the site. For example, a villager would gain one specific version of understanding when he/she walks around the village, but when specifically visiting the headman’s house he/she becomes more familiar and bonds with the headman and everything along the way.

  Figure 6 Path thickness highlighting times used. Thick is high; thin is low. The high-density pathway increases the possibility of encountering its nearby households thus suggesting more social interaction, whereas the low-density pathways suggest less social activities. In addition, counting the number of times the agent passes through a single convex can provide a reference to examining the relationship between the agent and nearby households (Figure 7). To go one step further, the meshwork of all the settlement’s inhabitants can be accumulated to show the comprehensive social relations of areas A and B (Figure 8).

Figure 8 Accumulative meshwork of all agents in Area A and B. This kind of wayfaring modelling highlights the advantages of the meshwork approach. Once a relational connection, such as kinship, between the two households is identified in the archaeological record or suggested by other data, the path between the two can be further weighted to distinguish their relationship from one another. The network and meshwork approaches to

Figure 7 Wayfaring of a single agent starting from 0S4 shown in numbers. From a networking perspective, some would argue that a similar kind of analysis is already possible within space syntax using axial mapping. Although this does provide 98

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency modelling are not entirely opposed, and their relationship is complimentary rather than contradictory. In addition, wayfaring can also explore relationships between specific subjects. For example, if a villager has a relationship with his/her hunter friend, his/her cousin would not be involved.

in discussing the possibility of internal transformation when compared with later settlements.

Likewise, when an agent visits a friend in area A, his/her relative in area C would have nothing to do with it, but the people living along the path would, because the agent is passing by them. In this case, the meshwork approach allows the examination of how mutually-related people stand in terms of the relationship between the two friends.

  Figure 11 Wayfaring of an agent startinf from 0S4 moving to Area B. (Times passed household 00 is 4.)

5. Conclusion

In this paper, we argue that space can still serve as an instrumental viewpoint in understanding prehistoric social relationships. By applying concepts of human agency, we can further understand human interaction within a community, and investigate how people perceived and reacted to their dwellings.

Figure 9 Axial integration of Area A and B. Thick is high; thin is low.

This research proposes that there are three levels of relationship between spatiality and social relations: (1) the ideational governing structure which produces the spatial layout; (2) the spatial layout which reproduces and maintains the same structure within which it was made; and (3) the actual social relationships which are constrained by the spatial layout. From the case study of Saqacengalj we have shown that, whereas the first two are relationships taken from a network perspective, the latter is better understood by a meshwork approach. Networked and meshworked approaches should not be in opposition since they are complementary, rather than contradictory. By applying both approaches, we can start to discuss how delegated social power interacts with the allocated power of the inhabitants. While natural selection and external forces may favour some cultural forms over others this does not induce or stimulate adaptive culture change. From this bottom-up meshwork approach, we can start to discuss internal transformation and understand how culture gradually transforms itself through human action and interaction. Adopting meshwork and wayfaring theory also introduces a possibility to bridge the gaps that exist in between each diachronic phase within a settlement and the observed differences.

 

  Figure 10 Wayfaring of an agent starting from 0S4 moving to Area A. (Times passed household 00 is 6.)

 

Again, if archaeological evidence suggests that a mutually-related person is living somewhere distant, this person can be cost-weighted in to that specific relationship. This enables an evaluation of how a certain household is influential within a neighbourhood or the entire community. Furthermore, such results can provide an insight into how social power is allocated among the inhabitants as opposed to the delegated power passed down through the social structure, an insight that can help

This work is ongoing and aims to discuss the social transformation of the Kau-Shi Paiwan. Several possible research directions have been identified, firstly the relationship between a space and the social activity it carries with the social implications and the significance 99

 

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process of the times-passed value of each individual convex offering in depth examination. The second focus is on the divisions of the area within the site, further exploring the correlation between area division and neighbourhood. Questions such as how social relations in an area consolidate and form a local identity similar to that of a neighbourhood will be raised and examined. Finally, there is the interaction between a socially structured/networked identity and spatially structured/meshworked social relationships. By investigating the similarities and differences between the two, we can further discuss political dominance and social control, both from the delegated identity and the allocated social relations. This last stage could be the most significant and influential in interpreting social transformation from an internal perspective.

Chen, M. L. 2008. Settlement patterns at Saqacengalj, a slate house settlement in southern Taiwan. Asian Perspectives 47, 210-241. Conolly, J. and Lake, M. 2006. Geographical information systems in archaeology, Cambridge, Cambridge University Press. Cutting, M. 2003. The use of spatial analysis to study prehistoric settlement architecture. Oxford Journal of Archaeology 22, 1-21. Fairclough, G. 1992. Meaningful constructions - spatial and functional analysis of medieval buildings. Antiquity 66, 348-366. Faulkner, P. A. 1963. Castle planning in the 14th century. Archaeological Journal 120, 215-235.

Acknowledgements

Ferguson, T. J. 1996. Historic Zuni architecture and society: an archaeological application of space syntax, Tucson, University of Arizona Press

We would like to thank Prof. Chris Gosden for his indepth socio-cultural insights. Special thanks to Mr. John Pouncett for his technological support. In addition, we are very grateful to Dr. Maa-Ling Chen for providing the valuable data from Kau-Shi Paiwan.

Fitzjohn, M. 2007. Viewing places: GIS applications for examining the perception of space in the mountains of Sicily. World Archaeology 39, 36-50.

References

Foster, S. M. 1989. Analysis of spatial patterns in buildings (access analysis) as an insight into social structure: examples from the Scottish Atlantic Iron Age. Antiquity 63, 40-50.

Batty, M. 2004. A new theory of space syntax, Available at: www.casa.ucl.ac.uk/working_papers/paper75.pdf Batty, M., Jiang, B. 1999. Multi-agent simulation: new approaches to exploring space-time dynamics within GIS. Available at: www.casa.ucl.ac.uk/working_papers/paper 10.pdf

Giddens, A. 1985. Time, space and regionalisation, in: Gregory, D., Urry, J. (eds) Social relations and spatial structures, 265-295. Houdmills, Macmillan Publishers LTD.

Bourdieu, P. 1977. Outline of a theory of practice. Cambridge, Cambridge University Press.

Grahame, M. 2000. Reading space: social interaction and identity in the houses of Roman Pompeii: a syntactical approach to the analysis and interpretation of built space. British Archaeological Reports International Series 886. Oxford, BAR Publishing.

Bourdieu, P. 1985. The social space and the genesis of groups. Theory and Society 14, 723-744. Bourdieu, P. 1989. Social space and symbolic power. Sociological Theory 7, 14-25.

Gregory, D. 1978. Ideology, science and human geography. London, Hutchinson.

Canuto, M. A., Yaeger, J. (eds) 2000. The archaeology of communities: a new world perspective. London, Routledge.

Gregory, D. and Urry, J. 1985. Introduction, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 1-8. Houdmills, Macmillan Publishers LTD.

Castle, C. J. E., Crooks, A. T. 2006. Principles and concepts of agent-based modelling for developing geospatial simulations. Available at: www.casa.ucl.ac.uk/ working_papers/paper110.pdf

Hillier, B., Hanson, J. 1984. The social logic of space, Cambridge, Cambridge University Press. Hodder, I. 2004. The 'Social' in archaeology theory: an historical and contemporary perspective, in Meskell, L., Prucel, R. W. (eds) A companion to social archaeology 23-42. Oxford, Blackwells Publishers LTD.

Chang, K. C. 1958. Study of the neolithic social grouping: examples from the new world. American Anthropologist 60, 298-334. Chang, K. C. (ed.) 1968. Settlement archaeology, Palo Alto National Press Books.

Ingold, T. 2011. Being alive: essays on movement, knowledge and description. London, Routledge.

100

Wu and Lock: The Spatial Construct of Social Relations: Human Interaction and Modelling Agency Isbell, W. H. 2000. What should we be studying: the 'imagined community' and the 'natural community', in Canuto, M. A., Yaeger, J. (eds) The archaeology of communities: A new world perspective, 243-266. London, Routledge.

companion to social archaeology. 66-81. Oxford, Blackwell Publishers LTD. Renfrew, C. (ed.) 1984. Approaches to social archaeology, Cambridge, Massachusetts, Harvard University Press.

Lake, M. 2007. Viewing space. World Archaeology 39, 13.

Rouse, I. 1968. Prehistory, typology, and the study of society, in Chang, K. C. (ed.) Settlement archaeology, 1030. Palo Alto, National Press Books.

Llobera, M. 2000. Understanding movement: a pilot model towards the sociology of movement, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies, 65-84. Amsterdam, IOS Press.

Saunders, P. 1985. Space, city and urban sociology, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 67-89. Houdmills, Macmillan Publishers LTD.

Llobera, M. 2003. Extending GIS-based visual analysis: the concept of visualscapes. International Journal of Geographical Information Science 17, 25-48.

Sayer, A. 1985. The difference that space make, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 49-66. Houdmills, Macmillan Publishers LTD.

Llobera, M. 2005. The Nature of everyday experience: examples from the study of visual space, in Fisher, P. (ed.) Re-presenting GIS, 171-194. Chichester, John Wiley and Sons Ltd.

Soja, E. W. 1985. The Spatiality of social life: towards a tranformative retheorisation, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 90-127. Houdmills, Macmillan Publishers LTD.

Llobera, M. 2007. Reconstructing visual landscapes. World Archaeology, 39, 51-69.

Tilley, C. 1994. A phenomenology of landscape: places, paths and monuments, Oxford, Berg Publishers.

Lock, G. (ed.) 2000. Beyond the map: archaeology and spatial technologies, Amsterdam, IOS Press.

Trigger, B. G. 1967. Settlement archaeology. Its goals and promise. American Antiquity 32, 149-160.

Lock, G. 2009. Human activity in a spatial context, in Cunliffe, B., Gosden, C., Joyce, R. A. (eds) The Oxford Headbook of archaeology. Oxford, Oxford University Press, 169-88.

Trigger, B. G. 1968. The determinants of settlement patterns, in Chang, K. C. (ed.) Settlement archaeology, 53-78. Palo Alto, National Press Books.

Lock, G. 2010. Representations of space and place in the humanities, in Bodenhamer, D. J., Corrigan, J., Harris, T. M. (eds) The spatial humanities: GIS and the future of humanities scholarshi Bloomington, Indiana University Press, 89-108.

Trigger, B. G. 1996. A History of archaeological thought, New York, Cambridge University Press. Trigger, B. G. 2004. Cross-cultural comparison and archaeological theory, in Meskell, L., Prucel, R. W. (eds) A companion to Social Archaeology, 43-65. Oxford, Blackwells Publishers LTD.

Lock, G. and Stancic, Z. (eds) 1995. Archaeology and geographical information systems, London, Taylor and Francis.

Turner, A. 2004. Depthmap 4: a researcher's handbook. London, Bartlett School of Graduate Studies, UCL.

Mehrer, M., 2000. Heterarchy and hierarchy: the community plan as institution in Cahokia's polity, in Canuto, M. A., Yaeger, J. (eds) The archaeology of communities: a new world perspective, 44-57. London, Routledge.

Turner, A. 2007. UCL Depthmap 7: from isovist analysis to generic spatial network analysis. Available at: citeseerx.ist.psu.edu/viewdoc/download?doi =10.1.1.103. 4400andrep=rep1andtype=pdf

Massey, D. 1985. New directions in space, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 9-19. Houdmills, Macmillan Publishers LTD.

Urry, J. 1985. Social relations, space and time, in Gregory, D., Urry, J. (eds) Social relations and spatial structures, 20-48. Houdmills, Macmillan Publishers LTD.

Meskell, L., Prucel, R. W. (eds) 2004. A companion to social archaeology, Oxford, Blackwell Publishers LTD.

Wheatley, D., Gillings, G. 2000. Vision, perception and GIS: developing enriched approaches to the study of archaeological visibility, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies, 1-27. Amsterdam, IOS Press.

Pandya, V. 1990. Movement and space: Andamanese cartography. American Anthropologist 17, 775-797. Patterson, T. C. 2004. Social archaeology and marxist social thought, in Meskell, L., Prucel, R. W. (eds) A 101

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Wheatley, D., Gillings, M. 2002. Spatial technologies and archaeology: the archaeological applications of GIS, London, Taylor and Francis. Willey, G. R. 1968. Settlement archaeology: an appraisal, in Chang, K.C. (ed.) Settlement archaeology, 208-226. Palo Alto, National Press Books. Wu, M. C. 2010. Spatial integration and consistency: exploring settlement patterns in Saqacengalj, Taiwan. Unpublished MSc thesis, University of Southampton.

102

7 The old and the new in Egyptian archaeology: towards a methodology for interpreting GIS data using textual evidence Hannah Pethen School of Archaeology, Classics and Egyptology, University of Liverpool ________________________________________________________________________________________________ Abstract This paper proposes an approach to prevent GIS methodology from dominating research, and to assist the resulting project in addressing the question of the ancient meaning behind the archaeological remains. GIS is used as an investigative tool in conjunction with traditional Egyptology and theoretical approaches in order to interpret ritual features constructed within the landscape of Egyptian mining sites. By engaging with the Egyptian textual material it is possible to move towards an understanding of the perceptual framework by which Egyptians interpreted their landscape. This will provide the basis for the development and interpretation of GIS-based investigations into the visual experience of the sacred and numinous in landscapes around Egyptian mining sites. Keywords: Archaeology; Egyptian; GIS; Landscape; Minerals; Mining; Mythology; Phenomenology; Religion; Ritual; Text. ________________________________________________________________________________________________ of landscape. How textual, and other material, is to be integrated into the method is of great importance in ensuring it is theoretically sound. Although some work is being undertaken on the relationship between literature and landscape (www.lancs.ac.uk/mappingthelakes.htm), ancient textual material has not generally been used to provide insights into the ancient experience of landscape. One aim of this research is, therefore, to assess how far textual material can provide a context for the interpretation of visual experience.

1. Introduction This paper proposes a methodology for the integration of geographic information systems (GIS) and phenomenological approaches in the analysis and interpretation of ritual landscapes at ancient Egyptian mines. The great wealth of Egyptian textual material concerning religion and ritual provides considerable information on the Egyptian attitude to landscape, and permits the development of hypotheses about the Egyptian experience of landscape. These hypotheses can then be investigated through GIS-based investigation of the visual properties of certain ritual remains. It is hoped that the results of this investigation will provide insights into how far Egyptian attitudes to landscape, as revealed in textual material, are reflected in physical remains constructed for ritual purposes. This should provide insights into the nature and use of ritual features, some of which are difficult to interpret using traditional methods. This paper recognises that such an approach is not straightforward. The relationship between GIS and postprocessual concepts associated with experience and phenomenology has often been antagonistic. By some, they have been seen as almost mutually exclusive (Brück 2005; Chadwick 2004, 21; Cummings 2008; Tilley 2004b; Thomas 2004, 198–201). The development of a methodology which is inclusive and acceptable of both sides of this debate is therefore difficult.

As this is an initial paper during the development of a methodology, I do not intend to consider here the technicalities of cumulative viewshed analysis, its advantages, flaws, and results. Rather, I will use the insights gained during the development of a research project to consider how GIS may be usefully integrated into a research project, while avoiding the risks of technological determinism and the danger that the project will ultimately be so focussed upon the GIS-method as to be irrelevant to the original field of research. The focus is therefore upon the theoretical background and the rich research context provided by the Egyptian textual and archaeological material, and how these aspects will allow the GIS research to be integrated into a hermeneutic spiral, which aims to achieve a better understanding of Egyptian religious concepts and visual experience as expressed in features constructed at Egyptian mining sites.

The Egyptian context may provide the means to integrate such ideas, but it also adds further complexity. The method relies upon the presence of textual material to provide information on ancient attitudes and experience 103



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Phenomenological accounts often acknowledge that there may have been superficial change, but dismiss any suggestion that such differences could affect the validity of their interpretations. Instead, they may appeal to the idea that the underlying topography or the ‘bones of the land’ have remained the same (Tilley 2008, 268). This may or may not be true, but, even where the underlying topography remains the same, the appearance of the surface of the landscape can still make a dramatic impact on how it is experienced.

2. Theoretical background The increasing use of GIS in archaeology has coincided with the development of post-processual approaches to landscape archaeology. These include phenomenology, which prioritizes the experience of the individual (Bender, Hamilton and Tilley 2007; Brück 2005; Ingold 2005; Thomas 2001; Tilley 2010). Such approaches emphasise that the understanding of a place is derived from one’s sensory perception of it, in a reflexive and dynamic dialogue, whereby the identity of the individual and the identity ascribed to the landscape are each constructed and modified by the other (Tilley 1994; Thomas 2001; 2004; Watson 2001). This approach is derived from the philosophy of Heidegger (1962) and others (such as Merleau-Ponty 1996), for whom individual personal experience, mediated through the physical body, is preeminent.

GIS researchers have sought to engage with postprocessual debate concerning experience and meaning. The juxtaposition of GIS and phenomenology has often been framed around visibility and visual experience, and many have seen GIS technology as a means of solving some of the problems associated with the interpretation of visual experience in a landscape context (Gillings 2009; Hamilton et al. 2006; Lake et al. 1998; Llobera 2003; 2006; 2007a; Sturt 2006; Wheatley 1995). GIS can be used to systematically analyse the visual properties of an entire landscape, including archaeological remains and natural features. If necessary, the landscape can be modified within the GIS to reflect information on past topography, hydrology, and vegetation (Geary and Chapman 2006; Llobera 2007b). The results can also be output in a format suitable for statistical testing (Kvamme 1990; Wheatley 1995). The results of such a process are therefore systematic, repeatable, and easily demonstrable to a reader or audience. It is therefore argued that GIS has much to offer phenomenology.

Phenomenology has been of great value in providing a critique of the underlying philosophies of processual landscape archaeology (Shanks and Tilley 1992) and offering new perspectives on ritual landscapes. It has been used as a method of investigating landscape, particularly prehistoric ritual landscapes in northern Europe (Bender, Hamilton and Tilley 1997; Cummings 2002a; Thomas 1993; 2001). Central to phenomenological methods of investigation, is the physical engagement of the archaeologist with the archaeological remains in their landscape. It is argued that this allows archaeologists to experience the landscape for themselves and gain insight into the possible meaning of that landscape for those who constructed the archaeological remains within it (Tilley 1994; 2010).

At the same time, GIS researchers have turned to phenomenology as a means of interrogating the meaning behind the data they analyse. It is recognised that current GIS lacks the means to integrate meaning into the analysis of visual perception. This is a key criticism of the use of GIS to investigate visual experience, and one that has been made by phenomenological archaeologists (Chadwick 2004, 21; Gidlow 2000, 28). If the interpretation of landscape is viewed (as by Ingold 1992, 74; Tilley 1994, 2) as embedded purely within the physical sensory experience of it, GIS cannot replicate the complex emotional and intellectual responses generated by that physical experience. Alternatively, if it is believed (as by Jones 1998, 9) that sensory perception is mediated through the ‘perceptual framework’ of the individual’s culture and memory (Johnston 1998; Jones 2002, 35; Thomas 2004, 185), GIS cannot encode the complex and dynamic perceptual framework of a past culture (Chadwick 2004), even assuming some knowledge of that perceptual framework is available to the archaeologist.

Despite prompting new interpretations of prehistoric remains, such approaches have been rightly criticised for a number of problems. Phenomenological archaeologists often fail to demonstrate that the visual, or other sensory, relationships they identify represent the genuine choices of the ancient population (Fleming 2005; Renfrew 1994; Tarlow 2000, 719; Watson 1990). There is often no evidence to show that these relationships are not accidental (DeBoer 2004; Lawson et al. 1998, 119–20) or the result of post-depositional processes (Fleming 1999). In terms of visibility and visual experience, human beings have a wide field of view within which to identify significant or numinous landscape features. Unless it is possible to show that a group consistently chose to construct sites with reference to specific types of feature or features, it is difficult to be sure that the visual relationships identified by the phenomenologist were genuinely of significance to the ancient population (Brück 2005, 53).

At most, GIS can provide information on what visual sensory experiences would have been available to those present at a specific location within the landscape. Through further analysis, it may reveal if different archaeological features were constructed with reference to the same sensory experiences. With the application of statistical testing, this information may confirm if relationships between archaeological remains and certain

It may also be difficult for the phenomenologist to show that the landscape they are engaging with is equivalent to that experienced by an ancient culture, and has not changed dramatically during the intervening period (Criado Boado and Villoch Vasquez 2000). 104

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence In addition to a general dislike of its philosophical origins, there are a number of specific criticisms of GIS. It has been suggested that use of computers and the apparently scientific nature of the process, could promote uncritical acceptance of the results (Miller and Richards 1995, 20; Shanks and Tilley 1992, 56–7). However, robust criticism of viewshed analysis from various sources, including phenomenological archaeologists (Brück 2005; Chadwick 2004; Webster 1999) and GIS researchers (Gillings and Goodrick 1996; Wheatley 1995, 180–2; Witcher 1999), suggests that there is sufficient scrutiny and criticism, at least amongst the academic community.

visual properties of the landscape are statistically significant, and are therefore likely to have been deliberately chosen or constructed by those people who created the archaeological features. GIS cannot explain why those visual properties were chosen or exploited, or what they meant. The attempt to synthesise GIS and phenomenology is, in some cases, an attempt to overcome the inability of GIS technology to address the meaning behind the data. This synthesis is often implicit in GIS projects. In such cases, researchers reference the concepts of phenomenology before suggesting that their GIS methods can provide a useful tool to augment phenomenological analysis (e.g. Llobera 2001). Although well-meant, such approaches are often badly received by phenomenologists, who see GIS-based analyses of visual experience as a resurgence of discredited processual approaches, reliant upon a dehumanising view of landscape (Gidlow 2000), associated with an incipient environmental determinism and false scientific objectivity (Brück 2005; Cummings 2008; Thomas 2004, 198–201).

GIS research has also been accused of being technologically deterministic (Brück 2005; Chadwick 2004; Lake and Woodman 2003; Thomas 2004, 198–201; Wheatley and Gillings 2000). It is true that there has been stress on the technological and the methodological in many publications of GIS-based research, but I would suggest that this is only natural. Those engaged in analysis using a new and continually developing technology will be anxious to demonstrate its capabilities, and eager to show how new advances may be applied to their work. The development and application of new methods is desirable and there is no reason why these methods cannot be undertaken within a specific research context to answer questions of genuine archaeological significance.

This distaste for GIS rests on the argument that all methods are so wreathed in the philosophies and ideologies within which they were developed, that these philosophical foundations intrinsically affect how the results of those methods are perceived (Gidlow 2000, 27). Thus, for post-processual archaeologists, GIS is inherently associated with Cartesian measurement, scientific thought, and objectification of the landscape. This is because it is derived from maps created using Cartesian measurements and developed in the context of western philosophy, which emphasises measurement, quantification and control rather than engagement and experience (Thomas 2001, 168–70; 2004, 198–201). Therefore, it is argued, any GIS analysis would result in an interpretation that is inherently dispassionate and dehumanised, the very antithesis of phenomenology (Chadwick 2004, 21; Tilley 2004b).

I would suggest that rather than the process being inherently technologically deterministic, it is the style of research report which unwittingly conveys this impression. For example, where GIS is used to investigate archaeological research questions, these may be described as ‘case studies’ or treated as such (for example Lake et al. 1998; Llobera 2001; Wheatley 1995). This may suggest to phenomenologists, who are predisposed to distrust GIS-based research, that the technological advance is more important than the research questions it is used to answer. Similarly, readers of GIS research are often encouraged to consider how a particular technology or method may be applied to their field of research (for example Lake et al. 1998, 27; Llobera 2001, 2005; Llobera 2006, 161; Llobera 2007a; 2007b; Wheatley 1995, 171). This could be interpreted as evidence that new technological opportunities are driving research rather than being applied appropriately to archaeological issues.

These criticisms make a valid point. It is important to be aware of the origins of the methods that are used, but the exclusion of any technology which relies upon processual or scientific principles seems a somewhat extreme and restrictive position. Surely, such attitudes can only result in further polarisation between phenomenologists and GIS researchers (Gidlow 2000). The use of Cartesian measurement and quantification cannot be entirely avoided, as it is necessary for the basics of archaeological site location, recording, and investigation (Hamilton et al. 2006, 37). As the work of Hamilton et al. (2006) and others (such as Criado Boado and Villoch Vasquez 2000) demonstrates, GIS may originate in particular philosophical systems, but its use need not imply total acceptance of those systems to the exclusion of all others. Rather, it is suggested that the best defence against the propagation of potentially dehumanising processual concepts is greater integration of theoretical concerns into GIS projects.

A significant factor in the general distrust and dislike felt by some phenomenologists towards GIS is that researchers using GIS can focus more on what relationships are present in the visible landscape than on why those relationships are present (Gillings 2009, 339). In one particularly obvious example, Llobera comments that ‘only the faintest of archaeological implications will be sought here, given space constraints and the intention to focus on methodology (Llobera 2006, 151)’. While there is nothing inherently wrong with a focus on method in one publication, such phrases could appear to post-

105



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process for a future GIS that can analyse the ancient meaning of the landscape, as well as its visual attributes, current GIS can be used with archaeological theory and other evidence to improve understanding. In such a scenario, GIS becomes a method for analysing data and investigating patterns. Archaeological theory and other sources of evidence, which will vary depending on the area of research, provide a context for the development of the GIS analysis and consideration of the results.

processual archaeologists as evidence of the incipient technological determinism in GIS research. Even when greater consideration is given to the meaning behind the data, the scientific style can sometimes make those explanations appear almost as an afterthought (for example Lake et al. 1998; Llobera 2001; Wheatley 1995), though this is surely not the intention of the researcher. Such stylistic attributes have only served to antagonise phenomenological archaeologists, who are naturally highly interested in meaning and ideologically predisposed to be suspicious of the scientific approach that has been adopted. This is not to claim that such an approach should be abandoned, nor should it diminish the value of methodological and technological contributions, or be taken to imply that genuinely significant archaeological research cannot be undertaken using new technologies (for example Gillings 2009; Lake et al. 1998; Wheatley 1995). Rather, it should be taken as evidence that the division between GIS and phenomenology is often based on appearance rather than reality. To resolve this there may be a need for GIS researchers to engage more directly with theoretical issues (see also Gillings 2009), and for phenomenologists to look beyond the style and see the underlying intent of GIS researchers to contribute useful methods, which will assist in answering significant archaeological questions.

Gillings (2009) has made an effort to explicitly integrate phenomenology and GIS and address some of the stylistic issues which have affected the reception of GIS by phenomenologists. In his research into the megalithic remains on Alderney, Gillings aims to show that GIS is an appropriate tool for investigating those visual relationships, first suggested by phenomenology. By demonstrating his interest in the meaning behind the data and grounding his analyses in the research context, he also aims to develop a more meaningful dialogue between GIS researchers and phenomenologists. It seems likely that this approach would find acceptance amongst those who have sought a more systematic phenomenological methodology (see Criado Boado and Villoch Vasquez 2000; Hamilton et al. 2006). The integration of multiple theoretical and methodological approaches within a single research project is increasingly being considered by phenomenological and post-processual archaeologists. Many are advocating a hermeneutic and dialectical approach as a means of integrating different sources of evidence, methodologies and theories into the interpretative process (Hodder 2003, 195–205; Jones 2002, 8; Lock 2003, 1–13; Shanks and Tilley 1992, 103–115; Thomas 2004, 235–43; Tilley 2004a, 224). Such approaches allow a role for GIS, but by locating it in a wider theoretical and evidential framework seek to limit the impact of processual philosophy and technological determinism.

This is particularly important because the apparent lack of interest in the meaning of landscape, for which GIS researchers are often criticised, is due to the same limitations of GIS which originally attracted those researchers to phenomenology. As GIS cannot integrate human perceptual frameworks or the intellectual and emotional experience of landscape into an analysis, GIS researchers are often reticent to venture into interpretations of the meaning of landscape. Yet this reticence does not mean that GIS researchers are disinterested in the meaning behind their data, and many have shown a genuine interest in methods of integrating the intangibles of society, culture, community and memory into GIS analysis (Curry 1998; Lake et al. 1998; Llobera 2003, 34; 2006, 166; Lock 2003, 175–6). Unfortunately technological development is often seen as the solution to this problem. Many researchers have expressed the hope that future GIS will allow for experience and social meaning to be fully incorporated into analysis. This future may indeed eventuate, and is likely to generate much innovative technological development and archaeological research. Sadly, at present, an overt yearning for a technological solution has only contributed to the impression of technological determinism, resulting in further polarisation between phenomenologists and GIS researchers (Chadwick 2004, 21; Gillings 2009, 341).

This hermeneutic approach has also proven necessary because it has been recognised that phenomenology possesses an inherent problem with meaning and experience. If it is acknowledged that current GIS cannot incorporate the individual’s perceptual framework, emotional, intellectual or mnemonic response into the analysis of visual experience, it must equally be recognised that a modern archaeologist, with a different perceptual framework, is also likely to respond in a different way to a Neolithic or ancient Egyptian individual (Brück 2005). How far then, can a modern archaeologist truly perceive the landscape as an ancient individual would have done? If the experience, and therefore the interpretation, of the modern archaeologist is based upon a different perceptual framework from those held by ancient people, can it be considered a valid method of interpreting ancient meaning?

Too great a focus on technological development may also obscure opportunities for investigating visual experience using GIS in combination with other theoretical approaches or sources of evidence. Rather than waiting

Both GIS and phenomenology therefore require the addition of further material to explore visual experience fully and grapple with how ancient landscapes were perceived. GIS cannot include the emotional, intellectual 106

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence The use of textual material is not without its own particular problems and issues. Texts, particularly religious and ritual examples, are typically products of elite culture and can promote an idealising and unifying view of that culture (Baines 1997), which may not be true in practice. The purpose of Egyptian religion was to secure order, through a combination of divine and human governance (Assmann 1989; Kemp 2006, 49; Shafer 2005). Texts contained the details of the rituals enacted to maintain that order, and, through the metaphysical power ascribed to writing, could contribute to the maintenance of stability simply by virtue of their very existence (Hornung 1992, 54). It was, therefore, advisable that texts with a religious or ritual content should be unifying and positive, and they are likely to exclude the marginal and contentious aspects of society. As a result, they may not represent the perceptual framework of non-Egyptians, non-elite, and non-literate persons.

and spiritual response of a person physically experiencing a landscape, while phenomenology can struggle to show that the emotional, intellectual, and spiritual responses it generates are genuinely applicable to ancient populations. Many post-processual archaeologists (for example Hodder 2000, 92; Thomas 2004, 241; Tilley 2000, 425) have argued that the archaeological remains themselves place limitations upon the interpretations phenomenologists can derive from them and that, when archaeological remains are considered within the landscape, ‘we cannot describe them in any way we like (Tilley 2004a, 219)’. This is certainly true in the general sense, but even when considered within their landscape, archaeological remains are often capable of very different interpretations. This is particularly true when those interpretations are concerned with emotion, spirituality, and other intangibles, and the societies which created them have left relatively little evidence to constrain our interpretations.

If only textual material was used to understand the Egyptian perception of landscape, this elitist and unifying trend could lead to interpretations based only on the perceptual framework of the elite, which may not necessarily be applicable to the rest of society (Baines 1991, 124; Wendrich 2010, 1).

It must therefore follow that the more archaeological evidence, the greater the constraint upon interpretation. Where archaeological evidence is not only plentiful, but includes textual references to those very intangibles we seek to understand, it can provide a more meaningful context for interpretation, even if a complete and accurate understanding will never be possible (Hodder 2000, 93; Thomas 2004, 242 – 47; Tilley 2004a, 223).

Just as with phenomenology and GIS, texts cannot reveal past visual experience if used in isolation. They neither describe experience directly, nor can they provide evidence of a complete perceptual framework for all society. It is therefore necessary to integrate textual and archaeological material with GIS analysis that is grounded in the physical evidence of the landscape and the archaeological features present at particular sites. By analysing the visual properties of ritual archaeological remains within a specific landscape, significant visual relationships may be identified that either confirm or confound expectations derived from textual material. The results could then reveal whether those involved in constructing archaeological features were operating within the perceptual framework of the elite, or not.

This is not to suggest that textual or other archaeological evidence can replace the complex nuances of individual human experience and response, any more than GIS analysis can replicate the experience of being within a landscape. In the Egyptian context, texts do not generally describe experiences directly. Rather they are derivative expressions of past experience, refined and reduced for a particular, often religious, purpose. As such, they are not the description of an experience, but the interpreted conclusions of that experience. Despite this, such texts do contain evidence of the Egyptian perceptual framework of theology, symbolism, and cosmology through which experience was interpreted. What is more, that perceptual framework is likely to have been derived from the intellectual, mnemonic, emotional, and spiritual responses to particular past experiences. The glimpses of the Egyptian perceptual framework preserved in their texts are thus both evidence of past responses to experience and constructs which would have reproduced similar responses in subsequent generations of Egyptians. The use of ancient texts is therefore likely to result in a better understanding of the ancient perceptual framework. A better understanding of that perceptual framework is, in turn, likely to result in a better series of research questions and a better context for the interpretation of results derived from GIS or phenomenological approaches. Textual evidence cannot therefore replace experience, but it can provide a greater constraint to our interpretation of the data, and thereby move us towards a better understanding of it.

Another problem with the use of Egyptian religious texts is that they do not generally allow much room for individual expression of religious or other attitudes. The principal categories of religious text include those which have been preserved as funerary texts, such as the Coffin Texts or Book of the Dead (Faulkner 2004; 2010); those which were inscribed on temple walls (Aufrère, 2001; Baines 1997; Bell 2005; Hornung 1992, 125); and those religious elements included in a large body of inscriptions within the landscape, on stelae, tomb, and chapel walls (for example Gardiner 1955; Seyfried 1982; Simpson 1974). The latter type of texts contain slightly more scope for individualism, but are often formulaic, particularly where they concern religion and ritual (see Blumenthal 1977; Eichler 1994). The use of this material cannot therefore provide an individual dimension, but only some limited insight into the perceptual framework across the group which produced it.

107



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process The question of individual experience, is one with which phenomenology has also grappled. The emphasis on the universality of the human body in phenomenology has been seen as a denial of uniqueness of each person and their individual experience (Brück 1998; Meskell 1996, 6). This remains a valid concern, but cannot be easily addressed, except where individuals have left very personal accounts of their experiences, including diaries or descriptive letters. Such documents are not present amongst the extant Egyptian evidence pertaining to desert landscapes.

stelae; rock-cut shrines (Castel and Soukiassian 1985a); artefacts (Pinch 1993); and a variety of more ephemeral features built of small stones (Engelbach 1933; 1939; Petrie and Currelly 1906, 64; Shaw 2010, 68). These ephemeral features occur at a number of sites from the Sinai to the south-western desert and from the Red Sea Coast to the Nile Valley. In addition to those which have been identified, it is likely that similar features remain to be recorded at other sites. The presentation of these different features varies. At Gebel el-Asr the enclosures were directly associated with cairns, while at Gebel Zeit and Serabit el-Khadim they were not. Small cairns associated with stone alignments have been recorded at Hatnub (Figure 4), but, unlike the other sites, no textual material or artefacts were associated with them (Shaw 2010, 97–107).

Similarly, GIS only provides limited access to the individuality of physical experience. GIS analyses can be varied according to physical characteristics, such as the height of the viewer, but cannot incorporate the different perceptual frameworks derived from social class, gender or ethnicity. It is not, therefore, possible to address questions of individual experience within this study. At most, it is possible to consider how far the different choices of visual attributes at some ritual sites may reflect whether the individuals involved in creating them were operating within the perceptual framework known from elite texts. It is recognised that this is a poor substitute for the variety of individual experience but it is suggested that with the available evidence, little more can be achieved at this time.

However there are variations in how different features were presented. For example, at Gebel el-Asr the enclosures were directly associated with cairns, while at Gebel Zeit and Serabit el-Khadim they were not. Small cairns associated with stone alignments have been recorded at Hatnub (Figure 4), but unlike the other sites, no textual material or artefacts were associated with them (Shaw 2010, 97–107).

In this paper, the integration of GIS and phenomenology with each other, and with other sources of evidence, is seen as a very positive development in the theoretical and methodological debate. The furtherance of this integration and the development of methods within which it can occur, is a significant goal of my research into visual relationships and experience at Egyptian ritual sites. To this end, it is hoped that the combined use of GIS, phenomenology, and other archaeological evidence will ensure the research is aimed at improving understanding of the meaning behind the visual attributes of the landscapes under investigation. The use of evidence from the Egyptian context will assist in this and will ensure that the theories and methods involved are embedded within current research questions, and are not merely included for their own sake. The use of GIS and archaeological theory will assist in ensuring the research is able to address questions which are not accessible using textual and archaeological evidence alone. This integration is intended to result in a more satisfactory, but still incomplete, interpretation of the ancient meaning behind the archaeological remains.

Timna Serabit el-Khadim Gebel el-Zeit

* * * *Hatnub

Gebel elͲAsr

*

*Wadi el-Hudi

3. Religion and mining in the desert

Figure 1 Ancient Egyptian mining and quarrying sites in the text

There is a number of ancient Egyptian mining and quarrying sites scattered across the deserts of Egypt (Figure 1) and originally located to take advantage of outcrops of suitable rock and precious stones. It is clear that many of these sites included a ritual area. The most elaborate is the Middle Kingdom (c. 2055–1650 BC) temple at Serabit el-Khadim (Valbelle and Bonnet 1996), but other sites include inscriptions, either rock cut or on

It is likely that some of these remains may be of a functional nature, rather than purely ritual. Some of the cairns, for example, may have functioned as route markers or could have been used to identify particular locations (see Engelbach 1939). Many of these mining sites are quite extensive, and it is possible that some cairns were created to ensure those moving around the

108

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence It is therefore likely that some proportion of these ephemeral features were associated with ritual or religious activity at these mining sites. How these features were used individually, why they were positioned so widely across the landscape, and how they related to more obviously ritual or religious structures is uncertain. Where features may have had a practical use, it is not even possible to determine which features were of ritual significance, except in cases where they are associated with obviously religious inscriptions or artefacts.

site did not become disorientated and get lost in the desert. However, the texts and artefacts associated with some features indicate that cairns and other ephemeral features could also fulfil a ritual function. At Gebel el-Asr, excavations in the 1930s recorded a group of eight cairns grouped together on adjoining ridges (Figure 3). These cairns had a flat eastern side facing onto a small court outlined in pieces of rock. In the area of the cairns the expedition found stelae (from which the area acquired the name ‘Stelae Ridge’) and ‘votive’ offerings in the form of hawks (or falcons), offering tables and small pyramids (Engelbach 1933, 68; 1939, 372 – 378 and pl. LIV; Shaw 2003, 453). The textual material recorded the names of Twelfth Dynasty Pharaohs and the expedition leaders who served them. The texts contained references to certain mythological and religious concepts associated with mining sites (see below). Taken together, this evidence suggested that these features were deliberately constructed in that location for a specific ritual purpose or purposes (Darnell and Manassa Forthcoming b; Engelbach 1939; Pethen 2006).

Where possible, recognisably Egyptian artefacts, inscriptions, and religious structures from mining sites have been investigated using traditional interpretative methods (Andrén 1998, 37–38; Wendrich 2010, 1). There has been analysis of artefacts associated with the worship of Hathor at the mines of Timna, Gebel Zeit and Serabit el-Khadim as part of a wider study into the worship of the goddess (Pinch 1993).

Figure 2 Vertically set orthostats in an alignment at Hatnub (Photo courtesy of Ian Shaw) Other ephemeral features, including the enclosures at Serabit el-Khadim and Gebel Zeit, have been found to contain ritual artefacts and religious texts (Castel and Soukiassian 1989; Valbelle and Bonnet 1996, 70). At Hatnub, no ritual artefacts or religious texts were found near the ephemeral features, although textual material was present in other areas. While some of the Hatnub cairns may have been of a functional nature, no practical function could be ascribed to the small ‘shrines’ (Figure 4), the orthostats (Figure 2), or the petroglyphs. The ‘shrines’ particularly defy explanation. They comprise a short approach outlined in stones, with a small cairn or niche constructed of local stones at the end of the approach. These niches were too small to be used for shelter, showed no evidence of the burning which might be expected if they had functioned as ovens or kilns, and were also too small to make effective storage spaces for more than a few token items (Shaw 2010, 101). A religious or ritual use therefore seems probable for these structures.

Figure 3 Sketch plan of the cairns at Stelae Ridge, with adjacent courts outlined in small stones on their eastern sides (Modified from Engelbach 1939, pl. LIV)

109



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Studies and records have been made of the inscriptions left by the ancient Egyptian expeditions at a variety of sites (for example Castel and Soukiassian 1985a; 1985b; Couyat and Montet 1912; Darnell and Manassa Forthcoming a; Forthcoming b; Gardiner et al. 1955; Sadek 1980; Seyfried 1982). At Gebel el-Asr, investigations of the textual (Darnell and Manassa forthcoming a) and archaeological evidence (Engelbach 1933; 1939), concluded that the cairns of Stelae Ridge functioned as chapels, commemorating the members of the mining expeditions, and emphasising the divine right of the Pharaoh to the mineral wealth (Darnell and Manassa, forthcoming b; Pethen 2006). At Hatnub a consideration of the archaeological remains included the evidence from nearby inscriptions and similar sites (Shaw 2010), and a recent investigation at Serabit el-Khadim encompassed the Egyptian temple, archaeological remains, and textual evidence (Valbelle and Bonnet 1996).

This problem is symptomatic of the elite nature of the evidence, which is typically used to investigate Egyptian religion. More general studies of personal religion, popular religion, piety and magic (Baines 1987; 1990; Kemp 1995; Sadek 1988), have also concentrated on elite sources and have therefore had limited ability to determine how far the elite concepts and observances they describe were common to the whole of society. This concentration upon elite material has made the investigation and interpretation of ephemeral or atypical features more difficult. The predominance of certain typical types of religious material, much of which is of an elite nature, fostered a perception that Egyptian ritual sites are ‘self-evident’ (Kemp 2006, 113). The presence of religious iconography, textual references and ‘formal’ temple architecture has traditionally been, and often still is, the basis upon which sites are identified as ‘religious’ or ‘cultic’ (see for example Darnell et al. 2002; Derchain 1971; Edel 1996; Haarlem 2003; Pinch 1993; Vörös 1998). As ephemeral features, like those within the mining landscapes, often lack the inscriptional, artefactual and architectural evidence which investigations into religion and ritual are believed to require, they were often not identified as ritual sites. Even where ephemeral ritual sites were identified, interpretations were limited and such remains were often ascribed to foreign influence or activity. Thus, in describing groups of enclosures, standing stones, and cairns at Serabit el-Khadim, Petrie assumes the ‘origin of this custom cannot be attributed to the ancient Egyptians’ (Petrie and Currelly 1906, 64). It is true that similar remains are present in the Levant, but such features also occur within Egyptian religious tradition. With improved excavation techniques, predynastic (Neolithic) standing stones and circles have been recorded at Nabta Playa in the eastern Sahara (Wendorf et al. 1996; 2001; 2002); at the dynastic period quarry site of Hatnub in Middle Egypt (Shaw 2010; 97–108); and elsewhere in the Nile valley (Dreyer 1986; Friedman 1996; 1999; Haarlem 2003).

Figure 4 ‘Shrine’ S2 from Hatnub (Photo courtesy of Ian Shaw) These investigations have revealed some interesting elements of religious and ritual activities, but much of the material which has been interpreted is related to the culture of the elite and literate officials and priests. It is possible that some of those who commissioned stelae and inscriptions were not themselves literate, but by commissioning those inscriptions they reveal that they subscribed to the conventions of literate society, even if they could not practise it themselves. Similarly some of those recorded in the inscriptions as subsidiary members of expeditions may not have been literate, and may not have subscribed to the ideals encapsulated in the texts. Unfortunately, it is not possible to know whether this was the case, without knowing whether the inclusion of one’s name in an inscription was widely felt to be spiritually or ritually advantageous and, therefore, something to be desired. It is entirely possible that elite officials, who commissioned the stelae, considered it beneficial for their more lowly associates to be included, even if those associates did not subscribe to the same norms of elite culture and were therefore indifferent to their inclusion. As a result, although highly informative, such texts cannot reveal with any certainty the ritual activities or religious beliefs of non-elite sections of society (Baines 1987, 84).

The interpretation of archaeological remains is often still based on traditional Egyptology (e.g. Baines 1997; Castel and Soukiassian 1985a; Friedman 1999; Taher 2010; Valbelle and Bonnet 1996); and interest in archaeological theory has been limited, (although see Bloxam 2006; Kemp 2006; Meskell 1996; 2004 for more theoretical approaches to Egyptian archaeology). As a result, those sites without associated texts, religious iconography or artefacts remain relatively difficult to interpret (e.g. Darnell 2002, 121; Petrie and Currelly 1906, 63–70; Shaw 2010, 97–107; Wendorf et al. 1996). Investigation of religion and ritual in ‘off site’ or ‘nonsite’ contexts (see for example. Alcock et al. 1994; Kealhofer 1999; Knapp and Ashmore 1999; van Dommelem et al. 2005) has also been limited. Where ‘ritual landscapes’ have been considered, investigation has often focussed on the ‘sacred landscape’ of a particular site or larger area, in terms of its relationship with mythological landscapes (for example Friedman 110

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence ancient Egyptians experienced, interpreted, and operated within their landscape, to construct and emphasise its numinous elements. Although the underlying analysis will be undertaken in a GIS, the foregoing theoretical discussion has underlined the importance of explicit engagement with the theoretical issues of phenomenology and the contextual evidence, in order to avoid an overemphasis on the environment, and ensure that human experience remains at the heart of the research.

1996; McCarthy 2007; Tait 2003); processional landscapes (for example Eaton 2007); or with particular landscape features, believed to resemble significant religious symbols (for example Aldred 1976; Chittick 1957; Donohue 1992; Kendall 1988). These difficulties have meant that little investigation of atypical or ephemeral ritual remains has been undertaken. For example, the temple at Serabit el-Khadim with its typically Egyptian architectural components, inscriptions and artefacts has been extensively studied. However, the surrounding landscape contains a variety of smaller stone enclosures, rock-cut shrines, and inscriptions cut into the entrances of the mines. At least one of the stone enclosures contained a stela of the early Middle Kingdom pharaoh Senusret I (Valbelle and Bonnet 1996, 70). There are also a number of alignments of orthostats. The individual stelae and inscriptions have been recorded (Gardiner et al. 1955), but there has been limited investigation of the layout of the site and the relationship of the inscriptions, orthostats, and enclosures to the temple, the landscape, and each other.

4. Visual perception and landscape The emphasis on sight and visual experience in both GIS and phenomenology (although see Cummings 2002b; Houston and Taube 2000; MacGregor 1999; Tilley 2008; Watson 2001; Watson and Keatings 1999 for phenomenological studies associated with other senses) has been criticised as a function of the primacy of vision in Western thought (Brück 2005, 50; Cummings and Whittle 2003). This is a counterpart to the Western understanding of landscape as something external to be viewed (Thomas 2004, 199). It is linked to the visual appropriation of landscape, encapsulated in Cartesian mapping, which seeks to divide up and control a complex environment by representing it visually in two dimensions (Thomas 1993, 168–70).

Similarly, the stelae from Gebel el-Asr have been translated (Darnell and Manassa, forthcoming) and the archaeological evidence for the purpose and meaning of the Stelae Ridge chapels has been analysed (Pethen 2006). A synthetic study (Bloxam 2006) has also been undertaken, which considered the role of the deities recorded at the site, in relation to the organisation and power structure associated with mineral extraction. However, Engelbach indicates that the eight cairns of Stelae Ridge were only one of several groups of such features. Some of the groups of cairns produced stelae and artefacts like those at Stelae Ridge, while others did not (Engelbach 1933, 68–9). The different types of cairns may, therefore, have had divergent purposes. Further investigation of the relationships between the different types of cairns and the wider landscape may provide additional information about their purpose and origin.

In view of this criticism and since the proposed research will entail investigation of visual properties, it is appropriate to determine how meaningful visual experience was to the ancient culture which created the ephemeral features under consideration. There is considerable evidence that visual perception was significant in ancient Egypt, in terms of the processes of myth-making and religious discourse and their relationship to the visible landscape. This is perhaps not surprising given that much Egyptian myth and ritual, and particularly that which is accessible to us, was associated with the literate elite. The nature of Egyptian scripts, which blurred the boundary between text and image (Baines 2007, Hornung 1992, 17–36), encouraged those who read them to see symbols and hieroglyphs in the landscape around them (Wilkinson 1994). Many ritual constructs, spells and religious conventions were associated with punning and word play, which were rendered peculiarly visual by the nature of the script (Hornung 1982; Pinch 2006, 68; Wilkinson 1994). As a result of this highly visual language, the process of theological explanation and discourse (Assmann 2001) was undertaken in part through the medium of the visual symbol. These symbols made ideal motifs for the inscription of theological concepts on temple and tomb walls (Assmann 1995; Rundle Clark 1959, Wilkinson 1994). The reinterpretation of these motifs as part of ongoing theological discourse, reified visual symbolism.

The ritual remains present at such sites are, therefore, an ideal subject for further investigation using theoretical and methodological approaches derived from European and North American archaeology, namely phenomenology and GIS. The remains are located within, and spread across, a landscape. Some proportion of them is of undoubted ritual purpose, and they are directly or indirectly associated with textual and artefactual material that has already been analysed and published. The question of how such remains relate to their landscape and each other is of great interest, but cannot be addressed without the addition of new theoretical and methodological approaches to landscape. Furthermore, the considerable archaeological, textual, and architectural evidence provides a highly detailed context within which to locate such analysis.

The importance of visual perception in the development and reification of Egyptian ideas concerning the supernatural is also evident in the significance attached to natural objects which resembled known religious

I suggest that by undertaking an investigation of the ephemeral features using GIS analysis in the context of archaeological evidence and theory, it may be possible to engage in a new hermeneutic dialogue regarding how 111



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process symbols. An Egyptian text describes how the King found a highly significant stone, celebrated for its shape, which resembled the symbol for a ‘divine hawk’ (Wilkinson 1994, 16). Similarly, there is some evidence that certain landscape features were chosen as foci for ritual or religious sites because of their resemblance to symbols associated with deities or mythic structures (for examples see Aldred 1976; Arnold, D. 2003, 179; Donohue 1992; Edwards 1993, 279; Reeves and Wilkinson 1996, 17; Strudwick and Strudwick, 1999, 97).

Egyptian texts indicate that the landscape was perceived as imbued with divine power and spiritual presence (Hornung 1982). The gods were felt to be present within the landscape, both specifically and generally. As in many cultures, certain specific landscape features were strongly associated with a certain god or goddess and were clearly felt to be numinous in that context. The peak of El Qurn above the Valley of the Kings, for example, was considered sacred to the serpent goddess Meretseger (Dodson 2000, 30; Reeves and Wilkinson 1996, 17).

Egyptian iconography and language were dynamic and reflexive, so the recurrence of real features which resembled religious symbols would have confirmed and enhanced religious understanding. This would have established the significance of visual perception in the identification of the numinous.

Egyptian texts also reveal that the gods were felt to be immanent within the landscape more generally. In the mythology of Heliopolis, the god Shu and the goddess Tefnut represent air and moisture respectively, while their son Geb and daughter Nut are associated with the earth and the sky (Wilkinson 2003, 18). Although their precise associations and attributes vary, the relationship between these divinities and the natural world appears in most periods of Egyptian history. As Hornung (1982, 68) points out, this is not to say that the landscape was necessarily thought to be divine itself. The names of the gods of earth and sky, for example, are not the same as the names of the elements with which they are associated. Thus the name of the earth god is Geb (gb), but the earth is ta; the name of the sky goddess is Nut (nt), but the sky is pt. The relationship between the sky and the sky goddess, or the earth and the earth god is thus a deeply complex one.

It is of course logical that information from all five senses would have contributed to the experience of a place and any numinous or spiritual interpretation of it. It is also true that if only the visual dimension of human experience is considered, the resulting interpretation is equally partial (Brück 2005, 50). Nonetheless, there is considerable evidence that the visual appearance of a structure, landscape or object was of great significance in ancient Egyptian perception. There is clear evidence that the resemblance between natural objects, landscape features and religious symbols was both recognised and ritualised. Although it is understood that a focus on visibility will result in only a partial interpretation of experience, it is nonetheless considered appropriate to investigate how visual perception of the surrounding landscape affected the construction of ephemeral ritual features at Egyptian mining sites.

The complex spiritual significance of the landscape is revealed most clearly in its relationship with the Egyptian temple. The temple as a complete structure had to represent the cosmos, from the encircling desert hills to the verdant Nile swamps. The re-creation of the cosmos, as the home of the deity, enhanced the interaction of the deity with the real universe. So temple columns mimic papyrus and lotus plants to re-create the verdant swamps of the Nile Valley; the solid bulk of the temple pylons recall the distant desert hills; and the dark rooms within the temple form man-made caves (Aufrère 2001; Hornung 1992; Mostafa 1989; Phillips 2002, 260–263; Saleh 1969; Wilkinson 1994, 28–9).

5 The Egyptian perceptual framework and the mythology of minerals It was earlier suggested that both GIS and phenomenology share a similar problem with regard to the interpretation of social meaning and the visual experience of landscape. GIS cannot currently encode social meaning, perceptual frameworks, and individual experience into its analyses (Chadwick 2004, 21; Gidlow 2000, 28). Similarly, the visual experiences of the modern phenomenologist may reflect a very different perceptual framework from those of the ancient Egyptian (Brück 2005). In order to address these complementary issues it is necessary to include Egyptian textual, artefactual, archaeological, and architectural evidence within the hermeneutic dialogue to provide as detailed and nuanced a context for interpretation as possible. Just as the interpretation of archaeological remains within their landscape provides limits and boundaries for interpretation in northern European phenomenology (Hodder 2000, 92; Thomas 2004, 241; Tilley 2000, 425; 2004a, 219), so this evidence will constrain the research by providing an insight into the intangibles of religion and spirituality which formed one element of ancient perceptual frameworks.

The role of the temple as desert ‘Mountain’, as opposed to the Nile valley, was also particularly significant. The desert landscape was numinous by virtue of its hostility and otherness in relation to the hospitable valley. It formed the horizon familiar to most Egyptians as the point where the earth met the sky, and this world met the otherworld. The temple was therefore also described as a ‘horizon’, as it was a place where the material world met the supernatural (Hornung 1992, 115). The re-creation of components of the desert landscape within the structure of temple was therefore conducive to the temple’s function as a locus for interaction with the divine. To enhance that interaction the temple was deliberately constructed as a repository of fabulous minerals, just like the desert. In the temple, such minerals were used to create the divine statues, furnishings and utensils, which were imbued with a divine power by virtue of the 112

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence materials from which they had been created. This power was required by the divinities within the temples to reproduce the mysteries of creation and ensure cosmological stability (Aufrère 1991, 315).

(Assmann 1995; Hornung 1982), religion (Baines 2007; Hornung 1992), mythology (Assmann 2001), personal piety (Baines 1987; Kemp 1995; Sadek 1988), magic (Pinch 1993; 2006), and how such aspects are interwoven together. Since such material is mostly derived from an elite context, the extent to which such analyses reflect the attitudes of other sections of society is often unknowable (Baines 1990, 84). As ritual features at desert mining sites may lack textual or artefactual material, they are often difficult to interpret using evidence from elite contexts elsewhere. There is no way to tell if such evidence is widely applicable, or only relevant to those elite sections of Egyptian society which may or may not have constructed these features.

The numinous power of precious minerals was conceived of as derived from the supernatural origin of the minerals themselves, which were believed to be emanations of the divine, created within the numinous landscape of the desert horizon at the junction of natural and supernatural existence. In texts from the Late (c. 747–332 BC) and Ptolemaic Periods (305–30 BC) of Egyptian history, this relationship is described in mythic terms as the return of a wandering goddess, known as the ‘Distant One’, who restores spiritual cohesion to the divine world and peace to the material world (Aufrère 1997). The repetition of this myth on temple walls, and the equivalent scenes of the real mineral wealth brought into the temple, ensured by sympathetic magic the continuing return of the Distant One and resulting stability on earth. In earlier periods, the textual material is less explicit, but nonetheless confirms that mineral wealth was perceived to have a divine origin. That such beliefs were common across Egypt is revealed by references from Serabit elKhadim in the north-east (Valbelle 1997), to Wadi elHudi in the south (Sadek 1980, 81 no. 143 and pl. XXIII). At Serabit el-Khadim, turquoise is referred to as the ‘garment of Geb (Inscription 110. Gardiner et al. 1955; Valbelle and Bonnet 1996, 123)’. At Wadi el-Hudi inscription 143 confirms that ‘Geb has decreed for him (i.e. pharaoh) his secret treasures (Sadek 1980, 84)’ and refers to the myth of the Distant One when it says ‘the Eye (another name for the Distant One) brings for him from what is in her’. Elsewhere references to the mythic basis for mineral extraction are more oblique. The goddess Hathor regularly appears at mining sites as ‘Mistress of’ the various minerals extracted from them. Thus at Serabit elKhadim Hathor is described as Mistress of Turquoise (ýerny et al. 1955, 41; Seyfried 1981, 237–238; Valbelle and Bonnet 1996, 37); at Wadi el-Hudi she is Mistress of Amethyst (Fakhry 1952, inscription 24; Seyfried 1981, 137); at Gebel Zeit, Mistress of Galena (Castel and Soukiassian 1985, 291); and at Gebel el-Asr, Mistress of Carnelian (Figure 5). Despite their differing titles, the various forms of the goddess are depicted with the same appearance and iconography as forms of Hathor from elsewhere in the Nile valley. The regular relationship between the mining districts and Hathor is perhaps a reference to the myth of the Distant One, who could be associated with Hathor as well as other goddesses. As the myth encapsulated the idea of the mineral as a form of the goddess (Aufrère 1997), it is perhaps not surprising that the goddess is described as ‘Mistress of’ the relevant minerals.

Figure 5 Hathor ‘Mistress of Carnelian’ shown on a stela from Gebel el-Asr (Photo courtesy of Ian Shaw) In order to overcome this problem, I propose to combine textual and archaeological evidence with GIS analysis and archaeological theory. The textual and artefactual material will be used to move towards an understanding of the elite Egyptian perceptual framework concerning the desert landscape. The issues involved in this have been discussed above. Essentially, the religious concepts detailed in such material can be understood as a reflection of those things which had previously been identified as numinous on the basis of past sensory experience. At the same time, the recording of such concepts reified the power of the experiences which had given rise to them,

Previously, this type of mainly elite textual material has been assessed through the traditional processes of textcriticism, artefactual and architectural study. This has provided much information on Egyptian cosmology 113



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process ground levels (Chapman 2006, 85), depending on whether skeletal and geoenvironmental evidence indicate there is merit in such changes. In other environments it may also be necessary to allow for thick vegetation (Geary and Chapman 2006; Llobera 2007b), but in the desert this is likely to pose a minimal problem.

and encouraged future generations to experience similar features as numinous. The resulting perceptual framework can be glimpsed in the textual and artefactual material which recorded and generated it. I recognise that any understanding of ancient perceptual frameworks which is based on textual and archaeological material, is likely to be partial at best. It is not possible to access the full complexity of human experience from the available evidence, which can only represent a limited element of the fully embodied experience of landscape. Furthermore, due to the elite origin and the essentially conservative nature of Egyptian religious material (Baines 1997), any understanding gained will reflect only the homogenised view of the group rather than the nuances of individual intelligence and spirituality. This group will represent the literate Egyptian elite, as the perceptual framework of other social classes, ethnicities, genders, and illiterate groups cannot be accessed through this elite material.

Perhaps the most significant factor in the choice of GIS is that it can produce a systematic analysis of the landscape (Gillings 2009; Lake et al. 1998; Wheatley 1995), which is amenable to statistical testing (Kvamme 1990; Wheatley 1995). Although not always favoured by phenomenologists (Brück 2005; Cummings 2008; Thomas 2004, 198–201), this is considered necessary to determine and demonstrate whether the results of the work are consistent. Without such evidence it will be difficult to move beyond unsupported inference, for which phenomenology has previously been criticised (Brück 2005, 53; Criado Boado and Villoch Vasquez 2000; DeBoer 2004; Fleming 1999; 2005; Lawson et al. 1998, 119–20; Renfrew 1994; Tarlow 2000, 719; Watson 1990), and which has contributed to a polarisation of debate in areas where phenomenology has been extensively employed.

Nonetheless, any insight into the cosmological foundations of landscape perception is to be welcomed. It is necessary to make use of as much evidence as possible in seeking a new interpretation of the ephemeral ritual features. In order to achieve a fuller understanding of such features and how they were experienced within their landscape, the available Egyptian archaeological and textual evidence must be included in the hermeneutic dialogue together with other theories and techniques. Within this dialogue, the Egyptian archaeological and textual evidence will provide a perspective, albeit a limited one, on the perceptual framework of those who may have operated within and constructed the ephemeral, and other features, at Egyptian mining sites.

It is recognised that the use of GIS to investigate visual experience and perception is not straightforward. GIS is derived from processual methods of measurement and quantification (Gidlow 2000; Thomas 2001, 168–70; 2004, 198–201), and the need for systematic analysis and statistical testing is also a product of such approaches. In some cases over-emphasis on such methods has led to research which merely quantifies, but does not explain. This could resurrect methods that are dehumanising and dispassionate, obscuring human experience rather than exploring it (Chadwick 2004, 21; Tilley 2004b). I would agree that Cartesian quantification and measurement cannot be avoided in archaeology (Criado Boado and Villoch Vasquez 2000; Hamilton et al. 2006, 37), and suggest that only an explicit consideration of archaeological theory and contextual evidence (Gillings 2009) can ensure that the processual nature of the analysis does not overtake the essentially human and post-processual aims behind it. It is to this end, that the theoretical background and Egyptian contextual evidence have been considered in such detail.

6. The role of GIS GIS has been chosen as an appropriate method of investigating the visual properties of these landscapes for several reasons. Consideration of Egyptian perceptual frameworks and religious experience has suggested that the appearance of things was a significant element in their interpretation as numinous in many contexts. Although any interpretation made is likely to be partial, it is nonetheless appropriate to consider visibility and GIS is particularly appropriate for the investigation of visual attributes.

It is also important to consider that the construction of a landscape model in the GIS is a form of interpretation. The landscape recreated in the GIS is not ‘real’, but involves a level of interpretation of data derived from the original landscape (Goodrick and Gillings 2000, 43 discuss this with regard to Virtual Reality technology, but it is equally applicable to GIS). This is not unique to GIS, but is common to any archaeological technique (Shanks and Tilley 1992). Any method of visualising or depicting a landscape is likely to be interpretative to some extent. Even those methods attempted by phenomenological archaeologists, such as fan-shaped collages (Shanks 1992), photo montages (Bradley 1998), and 360Û diagrams (Cummings 2002), require some form of interpretation in terms of what to include and how it is

It is not currently possible to develop an intimate relationship with these landscapes. Unlike northern European phenomenologists (such as Tilley 2004, 185), we cannot build an understanding of these inhospitable sites over a long period. It may not be possible to visit some of these sites personally, while visits to others may comprise a brief expedition to obtain the necessary data. This is not ideal, but cannot be avoided given current circumstances. The use of GIS also provides an opportunity to change certain variables, such as the height of the viewer and 114

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence (Brück 2005; Chadwick 2004, 21; Gidlow 2000, 28). The very glimpses of the Egyptian perceptual framework, which are evident in textual and other material, cannot be integrated into the GIS in any way. It therefore only provides a means of assessing the visual properties of the landscape. It cannot recreate the experiences such visual properties might have prompted in an ancient Egyptian individual. The GIS cannot therefore answer directly the question of how the landscape was experienced or perceived. Rather, it can answer very specific questions about what is visible from where, and with what frequency and significance. The results must then be interpreted in the light of the archaeological material discussed above, to suggest how a certain experience and perception of the landscape may have resulted in the data which is present.

depicted. Such interpretations may be implicit, and therefore hidden, and could easily affect understanding of the meaning of that landscape. Conversely, the level of interpretation undertaken by GIS to construct a digital terrain model and undertake viewshed analysis is limited to physical interpolation and line of sight algorithms. It cannot include any implicit interpretation of the meaning behind the data. Indeed, the inability of GIS to encode, either implicitly or explicitly, any meaning or interpretation has already been noted and is widely recognised by both GIS researchers (for example Gillings 2009; Llobera 2001) and phenomenologists (such as Chadwick 2004, 21; Gidlow 2000, 28). Furthermore, in many cases the extent and accuracy of the physical interpretation can be quantified statistically (Chapman 79–81; Conolly and Lake 2006, 103–7) and empirically, if one has access to the landscape concerned. The ‘unreality’ of the GIS process must therefore be considered, but is not a reason to avoid its use.

The resulting interpretation is likely to be an incomplete understanding, since our knowledge of the perceptual framework of the ancient Egyptians is limited and the visual analysis undertaken in the GIS will be limited to certain aspects of visual experience. Nonetheless, it is hoped that the hermeneutic approach will provide a better, if incomplete, insight into the visual experience of Egyptian mining landscapes and the nature of the ritual remains constructed therein.

GIS has also been characterised by a preoccupation with technology and method, which has appeared technologically or environmentally deterministic (Brück 2005; Chadwick 2004; Lake and Woodman 2003; Thomas 2004, 198–201; Wheatley and Gillings 2000). I have argued above, that the apparent technological determinism is often an illusion derived from the stylistic attributes of research papers, a natural enthusiasm for the technology, and, paradoxically, an awareness of its limitations. Nonetheless, this illusion would be less deceptive if GIS research could be integrated more explicitly with theoretical approaches (Gillings 2009; Hamilton et al. 2006). To this end, and to address some of the concerns considered earlier, the use of GIS in visual experience has been explicitly considered here in light of theoretical approaches and criticisms.

7. The hermeneutic spiral The theoretical, archaeological, and historical background described above will form the foundation for the development of those research questions, which will be addressed through GIS analysis. Those glimpses of the Egyptian perceptual framework, revealed in the archaeological and textual evidence from the mining sites and elsewhere, will provide the context for the research and prompt the formulation of certain questions. These questions will be addressed in the current project through the use of GIS cumulative viewshed analysis.

Furthermore, although the investigation includes GIS, it does not propose a new method or process. Instead it uses cumulative viewshed analysis, an existing approach that has been used in a variety of studies over a number of years, to investigate different levels of intervisibility (e.g. Wheatley 1995; Gillings 2009; Lake et al. 1998). The research is not, therefore, predicated upon the development of new and exciting technological advances, but rather upon the appropriate application of an existing tool to a situation where the archaeological and historical context (outlined above) suggests it will be helpful. Although the development of new methods need not preclude genuinely valuable research, in this case new techniques are not necessary. It is not therefore necessary to focus extensively on the technique. Instead it is hoped that by focussing on the integration of GIS analysis with archaeological evidence and theories, the research will remain embedded in the hermeneutic dialogue and the different components of the project, theoretical, technical, and contextual, will be held in balance.

The theoretical background has identified limitations and considerations in the use of both the contextual evidence and GIS. In view of these limitations, the research cannot seek to replicate the ancient experience of landscape, but rather will aim to move towards a better understanding of what that experience might have been, the perceptual framework within which it was understood, and the responses it generated in terms of the construction and use of ritual features. The following questions are those which have so far been identified from the textual and archaeological evidence. They provide an initial focus to the research which is based on the context and not on the technological capabilities of the GIS programme. Additional research questions may arise during the course of the investigation:

This integration of GIS within a hermeneutic spiral is necessary not only for theoretical reasons, but also because GIS lacks the capacity to include human emotion, intelligence, and memory into an analysis

Are any particularly prominent outcrops or hills visible from the ritual features and are these consistently visible from ritual features at different locations across the site? 115



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process located to prevent their visibility from certain locations. At Gebel el-Asr, for example, it is noticeable that the Stelae Ridge cairns had their courts on the eastern side, while most of the mines were to the west. It would be interesting to determine if this location screened the mines from the view of those operating in the courts.

Phenomenology has emphasised that areas of high ground may appear significant or numinous when viewed from elsewhere (see for example Bender et al. 1997, 155; Cummings 2002a; Cummings et al. 2002; Tilley 1994). This is also attested in the Egyptian tradition, which attaches a general significance to the ‘mountain’ (Aufrère 2001; Hornung 1992, 115). In some cases, specific mountains have also been associated with a particular deity or considered particularly numinous (Dodson 2000, 30; Donohue 1992). Within the desert, mountains are amongst the most obvious topographic features, and may often be the defining feature of a locality or region. It is, therefore, reasonable to question whether the ritual features of the mining sites provided strategic views of any outcrops of hills. If particular landscape forms were consistently visible from a large number of individual features, this might indicate a particular significance for those landforms. This could, in turn, indicate that those landforms were experienced as particularly numinous and, when ritual features were constructed, such landforms were deliberately referenced.

The converse of hiddenness is prominence. While access to Egyptian temples was restricted, they were designed to be prominent features of the urban or rural landscape by virtue of their large size and stone construction. Prominence, particularly topographic prominence, has also been a subject of interest to phenomenology (Cummings 2002a, 134; Cummings et al. 1997; Tilley 1994) and of GIS research (Gillings 2009; Llobera 1996; 2001; Wheatley 1995). It has been noted that a number of the ritual features at mining sites are associated with higher ground. These include the temple at Serabit elKhadim, the cairns at Stelae Ridge and Hatnub, the shrines at Hatnub, and certain alignments of orthostats. However, such features may not be associated with the highest ground, or the most prominent location. This may suggest that some interplay of prominence and another factor determined the location of these features.

Do any prominent outcrops or hills dominate the view when approaching ritual features?

What is the visual relationship, if any, between the larger formal remains and the more ephemeral features?

In addition to the significance of certain specific hills, phenomenology has also identified examples where sites were constructed to allow views over certain areas of land or sea (Cummings and Whittle 2003; Scarre 2002). At desert mining sites, the construction of some features delimits the direction of approach. They might have short approaches delineated in stones as at Hatnub (Shaw 2010, 63); or possess courtyards dictating where activities should take place, like those at Gebel el-Asr (Engelbach 1939). At Serabit el-Khadim, the layout of the temple provided for limited routes into and away from the sanctuary (Valbelle and Bonnet 1996). The influence of these defined approaches upon the visual experience of the person moving towards, operating within and leaving the features is of considerable interest and has implications for both the use of the features and the perception of the landscape. The textual and artefactual evidence indicates that some ritual features operated as miniature shrines or chapels, where royal power and the divine origins of the pharaoh were emphasised to reiterate his rights to the minerals (Pethen 2006). In the Nile valley, the approaches to chapels and temples are delineated to reflect their significance. A similar significance might be anticipated for the approaches to desert shrines, and may be related to particular landscape features or views.

As most of the sites feature both ephemeral features and more typical inscriptional, architectural and archaeological evidence, there is a question as to how the larger features, such as the temple at Serabit el-Khadim or the Stelae Ridge cairns, related to more ephemeral or subsidiary features. The relationship between different types of remains has also proven of interest in phenomenological studies (see for example Bender et al. 1997), but the addition of textual and archaeological evidence provides more possibilities for interpretation in the Egyptian context. The elite nature of the textual evidence has already been acknowledged. Indeed it has been identified as a potential problem as it does not allow any insight into the perceptual framework of non-elite Egyptians. These texts, and other evidence for the presence of elite perceptual frameworks, are mostly associated with larger, more formal and typically Egyptian types of evidence, such as the temple at Serabit el Khadim. By contrast, certain ephemeral features are not associated with any textual or artefactual material that might indicate an elite origin. If such features are associated with very different visual references to the large temples and elite remains, this might suggest they were constructed and used by individuals with a different perceptual framework from the elite Egyptians. This could suggest different perceptual frameworks amongst those present at the sites, and may even reflect the presence of other classes or ethnicities.

Where can the ritual features be seen from? Phenomenological studies have also suggested that restriction and control of visual experience was intended at certain monuments (for example Watson 2001). Egyptian religion also emphasised hiddenness and exclusivity. The interior parts of the temple were accessible only to the Pharaoh and the priests, his representatives (Baines 1990; Wilkinson 2003, 63). There is therefore a question of whether ritual features were

Are any visual relationships consistent across many different sites? In terms of prehistoric British monuments, phenomenologists and GIS researchers have identified 116

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence particular patterns of references amongst relatively discrete geographic groups (Cummings 2002a; Tilley 1994; Wheatley 1995). The sites under consideration here cover all of Egypt, from the north-east to the south-west of the country. If the ritual features present at these sites consistently reference certain types of landform, this might suggest that certain elements of Egyptian religion were as homogenous across the country as has often been assumed.

8. Conclusion This paper has considered how both GIS and theoretical approaches rooted in phenomenology can remain too focussed on their method or philosophy and so fail to address the flaws inherent in their individual methods or the ultimate question of the meaning behind the archaeological remains. In common with others (e.g. Hodder 2003, 195–205; Lock 2003, 1–13; Thomas 2004, 235–43; Tilley 2004a), I suggest that it is necessary to take a more hermeneutic approach, in which different theories and methods can be embedded in a reflexive discussion which takes into consideration various sources of evidence.

It should also be noted that there are various different groups of features within the study, including orthostats, shrines, enclosures and cairns. It would be interesting to determine if visual attributes are also consistent within a group or between groups.

In terms of the investigation of Egyptian ritual landscapes, the wealth of textual material provides an important context in which to embed both phenomenological theories and GIS-based analyses. This textual material is used to move towards a limited understanding of the Egyptian perceptual framework which, together with relevant theoretical concerns, formsthe basis for subsequent investigation. In view of the large area of landscape concerned and the visual emphasis in Egyptian myth and religion GIS provides a useful analytical tool, because it offers a means of systematically assessing visual relationships within the landscape. The GIS analysis seeks to answer specific research questions intended to elucidate how far the visual properties of ritual features at the mining sites reflect the perceptual framework revealed by the textual and other evidence. The emphasis is on GIS as a tool for the interrogation of data with regard to specific questions derived from a combination of archaeological theory and Egyptian textual and archaeological evidence.

These questions will be investigated using cumulative viewshed analysis based on the process undertaken by Lake et al. (1998) and Gillings (2009). The analysis will be undertaken using digital terrain models created using satellite images. The intention is to improve the quality of the models in close proximity to the ritual features by undertaking on-site surveys. Once the results of the GIS analysis have been obtained it will be necessary to interpret these in the light of the elite Egyptian perceptual framework to assess how far the visual attributes of the ephemeral features reflect that framework. Thus the results of the GIS analysis will provide additional evidence to refine understanding of the choices made in the construction of ritual sites in the desert. These choices should reflect the visual experience the builders’ of such features intended to create, and so reveal those elements of the landscape they understood and experienced as visually significant. How far such choices can be explained using the perceptual framework glimpsed in elite texts and archaeological remains is of great significance for understanding the penetration of elite religious concepts within society, and how these were expressed in a constrained and inhospitable environment.

Acknowledgements This research paper was originally prepared for the Theoretical Archaeology Group Conference in December 2010. It forms part of an on-going PhD research project undertaken at the University of Liverpool. I would like to thank Constantinos Papadopoulos for organising the publication of the proceedings of the Thinking beyond the Tool session of the Theoretical Archaeology Group Conference and for his help and advice regarding submission deadlines and submission format. I would also like to thank my supervisor Dr Ian Shaw for his encouragement and permission to use some of his photographs and Colleen Manassa for sight of an early copy of her forthcoming article and translations of the inscriptions from Gebel el-Asr, both written with John Darnell.

The results of the GIS analysis will, therefore, be incorporated back into the hermeneutic spiral and considered in relation to the other evidence (textual and archaeological) and theories pertaining to the experience of the landscape. This is intended to ensure that the results remain of interest and significance to those who study this area and are able to contribute meaningfully to the debate about the ancient experience of landscape. The hermeneutic approach is necessary because neither the Egyptian evidence and, archaeological theory nor GIS analysis is able to address the issues fully. Rather, the three strands must be integrated together and balanced against each other in order to develop a better understanding of the role of visual experience in the construction of ritual sites at Egyptian mines, and contribute to the understanding of Egyptian religion and visual experience in general.

References Aldred, C. A. 1976. The horizon of the Aten. Journal of Egyptian Archaeology 62, 184. Alcock, S. E. Cherry, J. F. and Davis, J. L. 1994. Intensive survey, agricultural practice and the classical 117



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process landscape of Greece, in Morris, I. (ed.), Classical Greece: ancient histories and modern archaeologies, 137–70. Cambridge, Cambridge University Press.

Bloxam, E. 2006. Miners and mistresses: Middle Kingdom mining on the margins. Journal of Social Archaeology 6, 277–303.

Andrén, A. 1998. Between artifacts and texts: historical archaeology in global perspective. New York, Plenum Press.

Blumenthal, E. 1977. Die textgattung expeditionsbericht in Ägypten, in Assmann, J., Feucht, E. and Grieshammer, R. (eds) Fragen and die Altägyptische literatur: studien zum gedenken an eberhard otto, 86–118, Wiesbaden.

Anthes, R. 1928. Die Felsinschriften von Hatnub. Leipzig, J. C. Heinrichs.

Bradley, R. 1998. The significance of monuments. London, Routledge.

Arnold, D. 1974. Der Tempel des Königs Mentuhotep von Deir el-Bahri I: architektur und deutung. Cairo, Deutsches Archäologischen Institut, Abteilung Kairo 8.

Brophy. K. 2001. A void. A biography of phenomenology in archaeology. Paper presented to the 23rd annual conference of the theoretical group, University College Dublin. archaeology December 2001.

Assmann, J. 1995. Egyptian solar religion in the New Kingdom. Trans. Anthony Alcock. London and New York, Kegan Paul International.

Brück, J. 1998. In the footsteps of the ancestors. A review of Tilley’s ‘A phenomenology of landscape. Places, paths and monuments’. Archaeological review from Cambridge 15, 23–36.

Assmann, J. 2001. The search for God in ancient Egypt. Trans. David Lorton. Ithaca and London, Cornell University Press. Aufrère, S. 1991. L’univers minéral dans la Pensée Égyptienne. Cairo, Institut Français d’Archéologie Orientale du Caire.

Brück, J. 2005. Review essay: experiencing the past? The development of a phenomenological archaeology in British prehistory. Archaeological Dialogues 12 (1), 45– 72.

Aufrère, S. 1997. L’univers minéral dans la Pensée Égyptienne; essai de synthèse et perspectives (Autour de l’univers minéral X). Archéo – Nil. 7, 113–144.

Castel, C. and Soukiassian, G. 1985a. Dépôt de Stèles dans le Sanctuaire du Nouvel Empire au Gebel Zeit. Bulletin de l’Institut Francais d’Archéologie Orientale 85, 285–293.

Aufrère, S. 2001. The Egyptian temple – substitute for the mineral universe, in: Davies, W. V. (ed.), Colour and painting in Ancient Egypt, 158–163. London, British Museum Press.

Castel, C. and Soukiassian, G. 1985b. Les Mines de Galene du Gebel Zeit, in Schoske, S. (ed.) Akten des vierten internationalen ägyptologen congresses, München 1985, 161–170. Hamburg, Helmut Buske Verlag.

Baines, J. 1987. Practical religion and piety. Journal of Egyptian Archaeology 73, 79–98.

Castel, C. and Soukiassian, G. 1989. Gebel Zeit I: Les Mines de Galène. Cairo, Fouilles de l’Institut Francais d’Archeologie Orientale du Caire 35.

Baines, J. 1990. Restricted knowledge, hierarchy and decorum: modern perceptions and ancient institutions. Journal of the American Research Centre in Egypt 27, 1– 23.

Chadwick, A. 2004. ‘Geographies of sentience’ – an introduction to space, place and time, in Chadwick, A. (ed.) Stories from the landscape. Archaeologies of inhabition, 1–31 British Archaeological Reports, International Series 1238. Oxford, BAR Publishing.

Baines, J. 1997. Temples as symbols, guarantors and participants in Egyptian civilisation, in Quirke. S (ed.) The temple in ancient Egypt, 216–241. London, British Museum Press.

Chapman, H. 2006. Landscape archaeology and GIS. Tempus.

Baines, J. 2007. Visual and written culture in ancient Egypt. Oxford, Oxford University Press

Chittick, H. N. 1957. An inscription on Gebel Barkal. Journal of Egyptian Archaeology. 43, 42–4.

Bell, L. 1997. The New Kingdom divine temple: the example of Luxor, in Shafer, B. E. (ed.) Temples of Ancient Egypt, 127–184. Cornell University Press.

Conolly, J. and Lake, M. 2006. Geographical information systems in archaeology. Cambridge, Cambridge University Press

Bender, B. Hamilton, S. and Tilley, C. 1997. Leskernick. stone worlds; alternative narratives; nested landscapes. Proceedings of the Prehistoric Society 63, 147–178.

Couyat and Montet. 1912. Les inscriptions du Ouâdi Hammâmât. Cairo, Memoires publies par les Membres de l’Institut Francais d’Archéologie Orientale 34.

Bender, B. Hamilton, S. and Tilley, C. 2007. Stone worlds: narrative and reflexivity in landscape archaeology. London, Publications of the Institute of Archaeology, UCL.

Criado Boado, F. and Villoch Vasquez, B. 2000. Monumentalising landscape. From present perception to 118

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence Edel, E. 1996. Studien zu den relieffragmenten aus dem Taltempel des Königs Snofru, in Manuelian, P. (ed.) Studies in honour of William Kelly Simpson, 199–208 Boston, Museum of Fine Arts.

the past meaning of Galician megalithism (north-west Iberian Peninsula) European Journal of Archaeology 3, 188–216. Cummings, V. 2002a. Between mountains and sea. A reconsideration of the Neolithic monuments of southwest Scotland. Proceedings of the Prehistoric Society 68, 125–46.

Edwards, I. E. S. 1993. The pyramids of Egypt. Revised edition. London, Penguin Books. Eichler, E. 1994. Zur kultischen bedeutung von expeditionsinschriften, in Bryan, B. M. and Lorton, D. (eds) Essays in egyptology in honour of Hans Goedicke. 69 – 80. San Antonio, Texas, Van Siden.

Cummings, V. 2002b. Experiencing texture and transformation in the British Neolithic. Oxford Journal of Archaeology 21, 249–61.

Engelbach, R. 1933. The quarries of the Western Nubian Desert: a preliminary report. Annales du Service des Antiquités d’Égypte 33, 65–74.

Cummings, V. 2008. Virtual reality, visual envelopes and characterising landscape, in David, B. and Thomas, J. (eds) Handbook of landscape archaeolog, 285–90. California.

Engelbach, R. 1939. The quarries of the Western Nubian Desert and the ancient road to Tushka. Annales du Service des Antiquités d’Égypte 39, 369–390.

Cummings, V. Jones, A. and Watson, A. 2002. Divided places. Phenomenology and asymmetry in the monuments of the Black Mountains, southeast Wales. Cambridge Archaeological Journal 12, 57–70.

Fakhry, A. 1942. The Egyptian Deserts: Bahria Oasis I. Cairo, Service des Antiquités d’Égypte.

Cummings, V. and Whittle, A. 2003. Tombs with a view. Landscape, monuments and trees. Antiquity 77, 255–66

Faulkner, R.O. 2004. The ancient Egyptian coffin texts. Revised edition. Oxford, Aris and Phillips.

Darnell, J. Darnell, D. Friedman, R. and Hendrickx, S. 2002. Theban desert road survey in the Egyptian Western Desert I: rock inscriptions at Gebel Tjauti and Wadi ElHôl. Chicago, University of Chicago Oriental Institute Publications 119.

Faulkner, R.O. 2010. The ancient Egyptian book of the dead. Revised Edition. London, British Museum Press. Fleming, A. 1999. Phenomenology and the megaliths of Wales. A dreaming too far? Oxford Journal of Archaeology 18, 119–125.

Darnell, J. and Manassa, C. forthcoming a. Yale Toshka desert survey I: inscriptions of the “Chephren Quarries” in the Nubian Western Desert. Yale.

Fleming, A. 2005. Megaliths and post-modernism: the case of Wales. Antiquity 79, 921–32.

Darnell J. and Manassa, C. forthcoming b. A trust-worthy seal-bearer on a mission: The monuments of Sabastet from the Khephren Diorite quarries, in Middle Kingdom studies in memory of Detlef Franke.

Friedman, R. 1996. The ceremonial centre at Hierakonpolis locality HK29A, in Spencer, J. (ed.) Aspects of Early Egypt, 16–35. London, British Museum Press.

DeBoer, W. 2004. Comment on Tilley (2004). Cambridge Archaeological Journal 14, 199–201.

Friedman, R. 1999. Pots, pebbles and petroglyphs Part II: 1996 excavations at Hierakonpolis locality HK64, in Leahy, A. and Tait, J. (eds) Studies in ancient Egypt in honour of H. S. Smith, 101–108. London, Egypt Exploration Society Occasional Publication 13.

Derchain, P. 1971. Elkab I. Brussels Donohue, V. A. 1992. The goddess of the Theban mountain. Antiquity 66, 871–885. Dodson, A. 2000. After the pyramids. The valley of the Kings and beyond. London, Rubicon Press

Gardiner, A. H. ýerny, J. and Peet, T. E. 1955. The inscriptions of Sinai II. Second edition. Egypt Exploration Society, London.

Dreyer, G. 1986. Elephantine VIII: Der Tempel der Satet. Mainz, Deutsches Archäologisches Institut Abteilung Kairo 39.

Geary, G. and Chapman, H. 2006. Digital gardening, in Daly, P. and Evans, T. (eds) Digital archaeology. 171– 90. London.

Eaton, K. 2007. Memorial temples in the sacred landscape of Nineteenth Dynasty Abydos: an overview of processional routes and equipment, in Hawass, Z. and Richards, J. (eds) The archaeology and art of ancient Egypt: essays in honour of David B. O’Connor, 231–250. Cairo, Supreme Council of Antiquities Annales du Service des Antiquités d’Égypte Cahier 36.

Gillings, M. 2009. Visual affordance, landscapes, and the megaliths of Alderney. Oxford Journal of Archaeology 28(4), 335–356. Goodrick, G. and Gillings, M. 2000. Constructs, simulations and hyperreal worlds. The role of Virtual Reality (VR) in archaeological research, in Lock, G. and 119



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Brown, K. (eds) On the theory and practice of archaeological computing, 41–58. Oxford.

landscape: contemporary perspectives, 1–30. Oxford, Blackwell.

Haarlem van, W. M. 2003. The excavations at tell Ibrahim Awad (Sharqiya Province) seasons 1995 – 2000, in Hawass, Z. and Pinch – Brock, L. (eds) Egyptology at the dawn of the 21st century: proceedings of the 8th international congress of Egyptology I, 536–540. Cairo and New York, American University in Cairo Press.

Kvamme, K. L. 1990. One sample tests in regional archaeological analysis: new possibilities through computer technology. American Antiquity 55 (2), 367–81. Lake, M. and Woodman, P. E. 2003. Visibility studies in archaeology: a review and case study. Environment and planning B: planning and design 30, 389–707.

Hamilton, S. Whitehouse, R. Brown, K. Combes, P. Herring, E. and Seager-Thomas, M. 2006. Phenomenology in practice: towards a methodology for a ‘subjective’ approach. European Journal of Archaeology 9(1), 31–71.

Lake, M. Woodman, P. E. and Mithen, S. 1998. Tailoring GIS software for archaeological applications: an example concerning viewshed analysis. Journal of Archaeological Science 25, 27–38.

Hodder, I. 2000. Symbolism, meaning and context, in Thomas, J. (ed.) Interpretative archaeology; a reader. 86 – 96. London and New York, Leicester University Press.

Lawson, G. Scarre, C. Cross, I. and Hills, C. 1998. Mounds, megaliths, music and mind. Some acoustical properties and purposes of archaeological spaces. Archaeological review from Cambridge 15, 111–134.

Hodder, I. 2003. Reading the past: current approaches to interpretation in archaeology, 3rd edition. Cambridge, Cambridge University Press

Llobera, M. 2001. Building past perception with GIS: understanding topographic prominence. Journal of Geographic Information Science 17 (1), 25–48.

Hornung, E. 1982. Conceptions of God in ancient Egypt. Trans. J. Baines. New York, Cornell University Press. Hornung, E. 1992. Idea into image. Basle.

Llobera, M. 2003. Extending GIS-based visual analysis: the concept of the visualscape. International Journal of Geographic Information Science 17 (1), 25–48.

Houston, S. and Taube, K. 2000. An archaeology of the senses. Perception and cultural expression in ancient Mesoamerica. Cambridge Archaeological Journal 10:261–94.

Llobera, M. 2006. What you see is what you get? Visualscapes, visual genesis and hierarchy, in Daly, P. and Evans, T. (eds) Digital archaeology, 148–167. London.

Ingold, T. 2005. Comments on Christopher Tilley, ‘The materiality of stone: explorations in landscape Phenomenology’. Norwegian Archaeological Review 38, 122–129.

Llobera, M. 2007a. Reconstructing visual landscapes. World Archaeology 39(1), 51–69. Llobera, M. 2007b. Modelling visibility through vegetation. International Journal of Geographic Information Science. 21(7), 799–810.

Jones, A. 2002. Archaeological theory and scientific practice. Cambridge.

Lock, G. (ed.) 2000. Beyond the map: archaeology and spatial technologies. Amsterdam.

Kealhofer, L. 1999. Creating social identity in the landscape: Tidewater, Virginia, 1600–1750, in Ashmore, W. and Knapp, A. B. (eds) Archaeologies of landscape: contemporary perspectives, 58–82. Oxford, Blackwell.

Lock, G. 2003. Using computers in archaeology. London, Routledge. Lock, G. and Stanþiþ, Z. (eds) 1995. Archaeology and geographical information systems. London.

Kemp, B. 1995. How religious were the ancient Egyptians? Cambridge Archaeological Journal 5(1): 25– 54.

MacGregor, G. 1999. Making sense of the past in the present. A sensory analysis of carved stone balls. World Archaeology 31:258–71.

Kemp, B. J. 2006. Ancient Egypt: anatomy of a civilisation. 2nd Edition. London and New York, Routledge.

McCarthy, H. L. 2007. The Beit el-Wali Temple of Ramesses II: A cosmological interpretation, in Hawass, Z. and Richards, J. (eds) The archaeology and art of ancient Egypt: essays in honour of David B. O’Connor, 127–145. Cairo, Supreme Council of Antiquities Annales du Service des Antiquités d’Égypte Cahier 36.

Kendall, T. 1988. The cliff shrine of Tarharqa and the ‘Uraeus’ on Gebel Barkal, in Cherif, A. (ed.) Fifth international congress of egyptology: abstracts of papers 153–4. Cairo, Egyptian Antiquities Organisation. Knap, A. B. and Ashmore, W. 1999. Archaeological landscapes: constructed, conceptualised, ideational, in Ashmore, W. and Knapp, A. B. (eds) Archaeologies of

Meskell, L. 1996. The somatisation of archaeology. Institutions, discourses, corporeality. Norwegian Archaeological Review 29, 1–16. 120

Pethen: The Old and the New in Egyptian Archaeology: Interpreting GIS Data Using Textual Evidence Shaw, I. 2003. New fieldwork at Gebel el-Asr: ‘Chephren’s Diorite quarries’, in Hawass, Z. and Pinch – Brock, L. (eds) Egyptology at the dawn of the 21st century: proceedings of the 8th international congress of egyptology I, 448–454. Cairo and New York, American University in Cairo Press.

Meskell, L. 2004. Object worlds in ancient Egypt. Berg, Oxford and New York. Mostafa, D. M. 1989. Reflexions sur la function cosmographique due temple Égyptien. Varia Aegyptica 5, 103–118. Pethen, H. 2006. Royal power, personal gain and the mythology of mineral extraction at Stelae Ridge, Gebel el-Asr, Lower Nubia. Unpublished MA Dissertation, University of Liverpool.

Shaw, I. 2010. Hatnub: quarrying travertine in ancient Egypt. London, Egypt Exploration Society. Simpson, W. K. 1974. The terrace of the Great God at Abydos. Publications of the Pennsylvania–Yale expedition to Egypt 5, New Haven and Philadephia.

Petrie, W. M. F. and Currelly C. T. 1906. Researches in Sinai. London, Egypt Exploration Society.

Strudwick, N. and Strudwick, H. 1999. Thebes in Egypt. London, British Museum Press.

Phillips, J. P. 2002. The columns of Egypt. Manchester, Peartree Publishing.

Sturt, F. 2006. Local knowledge is required. A rhythmanalytical approach to the Late Mesolithic and Early Neolithic of the East Anglian Fenland, UK. Journal of Maritime Archaeology 1, 119 – 39.

Pinch, G. 1993. Votive offerings to Hathor. Oxford, Griffith Institute. Pinch, G. 2006. Magic in ancient Egypt. London, British Museum Press.

Tarlow, S. 2000. Emotion in archaeology. Current Anthropology 41, 713–746.

Reeves, N. and Wilkinson, R. H. 1996. The complete Valley of the Kings. London, Thames and Hudson.

Taher, A. W. 2010. A special report on the lost temple of Sety I at Wadi Kanayis. Ancient Egypt Magazine 11(2), 17–19.

Renfrew, C. 1994. Towards a cognitive archaeology, in Renfrew, C. and Zubrow, E. B. W. (eds) The Ancient mind, elements of cognitive archaeology, 3–12. Cambridge, Cambridge University Press.

Tait, J. 2003. The ‘Book of the Fayum’: Mystery in a known landscape, in O’Connor, D. and Quirke, S. (eds) Mysterious lands, 183–202. London, UCL Press Encounters with ancient Egypt Series.

Rundle Clark, R.T. 1959. Myth and symbol in ancient Egypt. London, Thames and Hudson.

Thomas, J. 1995. The politics of vision and the archaeologies of landscape, in Bender, B. (ed.) Landscape, politics and perspectives, 48–49. Providence RI and Oxford, Berg.

Sadek, A. I. 1980. The amethyst mining inscriptions of the Wadi el-Hudi. Warminster. Sadek, A. I. 1988. Popular religion and piety in Egypt during the New Kingdom. Hildesheimer Ägyptologische Beiträge 27.

Thomas, J. 2001. Archaeologies of place and landscape, in Hodder, I. (ed.) Archaeological theory today, 165–186. London, Blackwell.

Saleh, A. 1969. The so-called ‘primeval hill’ and other related elevations in ancient Egypt. Mitteilungen des Deutsches Archäologischen Instituts, Abteilung Kairo 25, 110–120.

Thomas, J. 2004. Archaeology and modernity. London and New York, Routledge. Tilley, C. 1990. Reading material culture: structuralism, hermeneutics and post–structuralism. Oxford, Blackwell.

Scarre, C. 2002. Monuments and the landscape in Atlantic Europe. London

Tilley, C. 1994. A phenomenology of landscape. Oxford, Berg.

Seyfried, K. J. 1982. Beiträge zu den expeditionen des mittleren reiches in die Ost-Wüste. Hildesheim, Hildesheimer Ägyptologische Beiträge 15.

Tilley, C. 1999. Metaphor and material culture. Oxford, Blackwell.

Shafer, B. E. 1997. Temples, priests and rituals: an overview, in Shafer, B. E. P. (ed.) Temples of ancient Egypt, 1–29. Cornell University Press.

Tilley, C. 2000. Interpreting material culture, in Thomas, J. (ed.) Interpretative archaeology; a reader. 418 – 428. London and New York, Leicester University Press.

Shanks, M. 1991. Experiencing the past. London, Routledge.

Tilley, C. 2004a. The materiality of stone. Explorations in landscape phenomenology. Oxford.

Shanks, M. and Tilley, C. 1992. Re–constructing archaeology 2nd Edition. Cambridge, Polity Press.

121



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Tilley, C. 2004b. Round barrows and dykes as landscape metaphors. Cambridge Archaeological Journal 14, 185– 203.

Watson, A. and Keating, D. 1999. Architecture and sound: an acoustic analysis of megalith monuments in prehistoric Britain. Antiquity 73, 325–36.

Tilley, C. 2008. Body and image: explorations in landscape phenomenology 2. Walnut Creek California, Left Coast Press.

Webster, D. S. 1999. The concept of affordance and GIS: a note on Llobera (1996). Antiquity 73, 915–917. Wendorf, F. Schild, R. and Zedeno, N 1996. A late Neolithic megalith complex in the Eastern Sahara: a preliminary report, in Krzyzaniak, L. (ed.) Interregional contacts in the Later Prehistory of Northeastern Africa, 125–132. Poznan, Poznan Archaeological Museum.

Tilley, C. 2010. Phenomenological approaches to landscape archaeology, in David, B. and Thomas, J. (eds) Handbook of landscape archaeology, 271–276. Walnut Creek, California, Left Coast Press Valbelle, D. 1998. Les Dieux Égyptiens et la Rayauté au Sinaï, in Valbelle, D. and Bonnet, C. (eds) Le Sinai durant l’Antiquité et le Moyen Age: 4000 ans d’histoire pour un désert, 50–55. Paris, Editions Errance.

Wendorf, F. Schild, R. and Associates. 2001. Holocene settlement of the Egyptian Sahara. Volume 1. Wendorf, F. Schild, R. and Associates. 2002. Holocene settlement of the Egyptian Sahara. Volume 2.

Valbelle, D. and Bonnet, C. 1996. Le sanctuaire d’Hathor maîtresse de la turquoise. Paris, Picard Editeur.

Wendrich, W. 2010. Egyptian Archaeology: From text to context, in Wendrich, W. (ed.). Egyptian archaeology, 1– 14. Wiley-Blackwell.

van Dommelen, P., Gerritsen, F. and Knapp, A. B. 2005. Common place. Archaeologies of community and landscape, in Attema, P., Nijboer, A. and Zifferero, A. (eds) Communities and settlements from the Neolithic to the Early Medieval Period, 55–63. Papers in Italian Archaeology 6: British Archaeological Reports, International Series 1452 (1). Oxford, BAR Publishing.

Wheatley, D. 1995. Cumulative viewshed analysis: a GIS-based method for investigating intervisibility and its archaeological application, in Lock, G. and Stanþiþ, Z. (eds) 1995. Archaeology and geographical information systems, 171–86. London.

Vörös, G. and Pudleiner, R. 1997. Preliminary report of the excavations at Thoth Hill, Thebes: The temple of Montuhotep Sankhara (Season 1995–1996). Mitteilungen des Deutsches Archäologischen Instituts, Abteilung Kairo 53, 283–287.

Wheatley, D. and Gillings, M. 2000. Vision, perception and GIS: developing enriched approaches to the study of archaeological visibility, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies, 1–27. Amsterdam.

Watson, R. A. 1990. ‘Ozymandias King of Kings’: Postprocessual radical archaeology as critique. American Anthropology 92, 613–629.

Wilkinson, R. H. 1994. Symbol and magic in Egyptian art. London, Thames and Hudson.

Watson, A. 2001. Composing Archaeology 33: 296–314.

Avebury.

Wilkinson, R. H. 2003. The complete Gods and Goddesses of ancient Egypt. London, Thames and Hudson..

World

122

 

8 A Roman Puzzle. Trying to Find the Via Belgica with GIS Philip Verhagen1, Karen Jeneson2 1

Research Institute for the heritage and history of the Cultural Landscape and Urban Environment (CLUE), Faculty of Arts, VU University, Amsterdam, The Netherlands 2 Thermenmuseum, Heerlen, The Netherlands

  Abstract In this paper we address the issue of using least cost path (LCP) modelling for a practical case study: the prediction of a 7 km stretch of Roman road (the Via Belgica) in the Dutch province of Limburg. Despite extensive archaeological research, the nature of the evidence is such that it does not allow us to project the route with sufficient confidence. LCP modelling can then be helpful to develop possible scenarios, departing from the available evidence and general assumptions about Roman road building. Using these scenarios, we managed to come up with a few plausible routes that we hope to test in the near future. Developing the scenarios made us think harder about the nature of Roman road building strategies and the interpretation of the available evidence. However, we also had to conclude that the available tools and theories are not very well suited for the kind of models that we would like to produce. Keywords: Least Cost Paths; Via Belgica; Roman Roads; Openness. ________________________________________________________________________________________________

1. Introduction: the Via Belgica

(Coriovallum) and Jülich (Juliacum) (Figure 1). In this paper we will focus on the Dutch stretch of this route, that is running through the southern part of the province of Limburg, from the Dutch-Belgian border near Maastricht through the valley of the Geul river on to Heerlen, and from there to the Dutch-German border at Rimburg, over a total distance of approximately 40 km (Figure 2). This stretch has been the subject of a substantial research project carried out in 2003 (Demey and Roymans 2004). To establish the exact position of the road all historically documented archaeological finds were analyzed, and fieldwork (augering and trial trenching) was carried out to find evidence for the existence of the road in various locations. Despite this effort, the published reconstruction of the road remains speculative in many places. A particularly hard nut to crack is the 7 km stretch between the villages of Valkenburg-aan-de-Geul and Voerendaal, for which various alternatives have been suggested. This is all the more surprising since in many places, especially in the German Rhineland, the road runs a very straight course and should therefore not be too difficult to reconstruct. The fact that the landscape in Dutch Limburg varies strongly from that in the Rhineland, due to a strong relief, could be a contributing factor.

  The Roman road between Boulogne-sur-Mer (France) and Cologne (Germany) is nowadays generally referred to as the Via Belgica. However, it was never known by this name in Roman times. The term was coined in the early 20th century by archaeologists who tried to reconstruct the Roman road system in the province of Germania Inferior. From the Peutinger Table and various other sources (such as milestones found at several locations along the way) it is clear that a major Roman road existed between the naval port of Boulogne-sur-Mer (Gesoriacum) and the frontier city of Cologne (Colonia Claudia Ara Agrippinensis). After the conquest of the region around 50 BC and its subsequent usurpation into the Roman empire, the familiar formula of towns and roads was implemented here. The Via Belgica was constructed under August and initially its main purpose was to connect the new territories in the North to the rest of the empire. In the second century AD its role as a transportation axis between the Rhine and Meuse and beyond became even more important, as it led through rich loess soils where large quantities of cereals were produced for both the military and civilian markets. It passed through various towns that were all founded under August, including Tongres (Atuatuca Tungrorum), Maastricht (Traiectum ad Mosam), Heerlen

123

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 1 Approximate route of the Via Belgica from Boulogne-sur-Mer to Cologne. which only the 8 m wide talus can be easily recognized in the field when digging trenches or test pits. Since the main survey methods used in the region are augering and field walking, it may not come as a surprise that the chances of actually finding it in a standard reconnaissance survey are not very high. This is particularly the case in Dutch Limburg, where the marked relief in combination with the loess soils there can lead to serious soil erosion, so that the road could either be covered in colluvium, or be washed away. In addition to this some parts of the road are known to have been used as a gravel quarry as recently as the early 20th century.

Figure 2 Approximate route of the Dutch stretch of the Via Belgica Figure 3 Cross section of the Via Belgica in the vicinity of Voerendaal.

2. Roman road building: what do we really know?

Concerning the criteria used by the Romans to decide where to put the road, we have to go with some general guidelines that can be deduced from the fact that over the whole Roman Empire we often encounter straight roads, pointing to a preference for speedy connections with relatively low construction and maintenance costs. However, many exceptions to this rule are observed in places where the landscape offered challenges to the Roman engineers that had to be overcome by either taking a different course, or by choosing alternative solutions like bridges, switchbacks, dikes or even tunnels. The study area does not seem to offer extreme difficulties to Roman engineering skills, even with the marked relief

As far as can be judged from the archaeological evidence, the Via Belgica is not the archetypical Roman road paved with slabs of stone. In Dutch Limburg, in those places where it has been found, it manifests itself as a layer of gravel up to 1.5 m thick and approximately 8 m wide. The gravel was used as paving material and the talus (or agger) is somewhat convex to allow water to drain more easily from the road (Figure 3). On each side, a zone of approximately 8 m wide is found that is bounded by a ditch. No paving is found here, but it is supposed that these zones were used for transport as well, though presumably not for ‘official’ purposes. This implies that the total width of the road is approximately 25 m, of 124



Verhagen and Jeneson: A Roman Puzzle. Trying to Find the Via Belgica with GIS indicated earlier. However, Demey and Roymans (2004, 37) conclude:

past routes. Modelling studies on issues of movement and transport in archaeology have mostly focused on the effect of slope on walking speed and energy expenditure (Bell and Lock 2000; Llobera and Sluckin 2007; Herzog 2010). So, despite the utility and popularity of LCP models to explore the potential for movement through the landscape, very little work has been done that has resulted in a predictive success (but see Becker and Altschul 2008). No case studies have taken into account the effect of terrain costs on wheeled transport, basically because of a lack of empirical data that could be used for this.

‘The way in which it is built shows a fundamental knowledge of the terrain. During road construction, a creative compromise was sought and found between following the straight line and evading difficult parts, like a marshy hollow near Voerendaal.’ Which parts of the landscape constitute the difficult bits is, however, largely a matter of conjecture. The attested route of the road suggests that wet areas were avoided. Apart from the marshy hollow mentioned near Voerendaal, the crossing of the river Worm at Rimburg at the current Dutch-German border is located at an eccentric northerly location, presumably because the river valley was too wide and wet to cross comfortably further south. Apart from that, it seems that Roman roads almost never take slopes over 15%, and in many cases not even over 8%. The main reason for this may have been the challenges that steep slopes offer to wheeled transport. The Roman army relied on supply trains with wagons, and these are obviously difficult to get moving again once they get stuck. A typical Roman army cart is the carpentum, drawn by 2 mules with a supposed maximum load of 500 kg (Roth 1998, 208-212). According to the figures given by Raepsaet (2002, 23), such a cart cannot move up a slope of more than 9% on a well-paved road when loaded. While the equation given by Raepsaet1 is probably not very realistic, it gives us some idea of the limitations of using wheeled transport at the time.

Furthermore, it has repeatedly been stressed (Llobera 2000; Bellavia 2006; Zakšek et al. 2008; Lock and Pouncett 2010; Murrieta Flores 2010) that the choice for a particular route is not only dependent on whether it constitutes the shortest connection, but will also be influenced by other factors. Especially for military routes like the Via Belgica it seems probable that visibility may have played an important role. Locations that are more vulnerable to ambush may have been avoided, and strategically placed watchtowers may have been employed to supervise stretches of road. This was acknowledged by local amateur archaeologist Harry van Aken, who published a number of possible reconstructions of the Via Belgica between Valkenburg and Voerendaal (Figure 4). Van Aken stated in an interview with Dagblad De Limburger on 7 Feb 2009: ‘Vicar Crutzen of Klimmen and Ransdaal wants to investigate whether the 11th century tower of the St. Remigius church was a Roman watchtower. This is plausible, since this is a strategic viewpoint. From this spot at an elevation of 140 m you can see the whole Eastern Mining Area. The Goudsberg near Valkenburg had such a tower as well. Its foundations were found. From there you can oversee the whole Geul valley.’

3. A case for least cost path modelling? Given the uncertainties regarding the way in which Roman engineers chose routes through difficult terrain, it seems logical to apply least cost path (LCP) models to try to find the most plausible ones. Least cost paths can be used to find the optimal connection between two or more locations based on both the distance between these points and the effort that is needed to cover this distance. On steeper slopes and wet ground, the time needed to cross the terrain increases and it may therefore take less time and energy to take a detour over easier ground. The basis of the calculations is a raster surface that specifies the costs of traversing a single grid cell. From this, the cumulative costs of accessing a specific location from anywhere in the landscape can be calculated. This principle has been applied to archaeological case studies on numerous occasions (e.g. Bell and Lock 2000; Bell et al. 2002; Howey 2007; Fiz and Orengo 2008). However, since the evidence for (pre-)historic roads and routes is usually very limited, it is in most cases hard to judge whether these models are actually very good at predicting

In their 2004 report, Demey and Roymans also indicated the route across the Goudsberg as being the most probable, due to the presence of the watchtower. However, as is often the case in archaeology, the situation is less clear-cut than that. The available archaeological evidence for this hypothesis is in fact fairly limited. The only place where the road has actually been found is near Voerendaal, and we can assume that it passed through Valkenburg because of a substantial number of burials located in close proximity to eachother west of the town, indicating a small necropolis that would have developed alongside the road at this location in the Geul valley. The existence of a Roman watchtower at the Klimmen church is completely unconfirmed, and the interpretation of the tower on the Goudsberg is not certain either.

 1

T = kP + Pi where T = traction force needed for movement P = weight of loaded cart in kg k = rolling coefficient i = slope in m/m The rolling coefficient k is composed of a pavement friction factor and an axle friction factor

125



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 4 Suggested routes for the Via Belgica between Valkenburg-aan-de-Geul and Voerendaal. G = Goudsberg, K = Klimmen church . The structure that was excavated there in 2002 visibility on the suitability of the road for military (Bazelmans et al. 2004) is of Late Roman origin (310purposes. Conceptually this is rather straightforward: 360 AD) and may very well have been a so-called various models should be constructed to connect the start burgus, a civilian construction in the form of a tower and target location in which a varying importance will be where the population could flee to in the case of enemy attributed to slope, wetness and visibility. In the raids. While it is logical to assume that such a tower execution of these models we are however confronted would be built close to the existing road, this could not be with several problems. established with certainty; and furthermore, it does not prove that the route was originally chosen because of the First of all, as already mentioned, almost all published potential to place a watchtower at this specific location. studies in archaeology on LCP models and movement have used walking speed as the parameter that determines We are basically dealing with a fairly common situation the suitability of a route. Several authors have pointed out in archaeology here: scant and disparate data sets, that the available equations specifying the effects of slope interpretations of the available data that highly depend on on walking speed will result in different outcomes of the expert judgement, and a lack of formal definition of the LCP models (Gietl et al. 2008; Herzog and Posluschny problem at hand. As such, it forms a typical case where 2008). Furthermore, alternative equations for wheeled quantitative modelling can be helpful because it will transport seem to be absent. The abovementioned force us to think about these issues from a more equation provided by Raepsaet (2002) seems to be formalized point of view, and experiment with the unsuited. While it provides figures for determining the variables involved to see how they influence optimal traction force needed to get a cart with a certain amount route finding. Both the heuristic element in this (learning of load moving on different surfaces and slopes, in by experimenting with the different options) and the practice it is severely limiting the slopes that can be spatial visualisation of the model results are the most negotiated with the type of carts that the Roman army is important contributions that GIS-based modelling can supposed to have used. When introducing this equation offer to archaeological research questions. It is therefore into a LCP model, it tends to produce large detours. For all the more surprising to see that LCP modelling in GIS this reason we have not used this equation, and have is actually not very well adapted to executing this settled for the most popular walking speed equation that approach. was originally published by Tobler (1993).

W

4. Modelling issues

Where W = walking speed in km/h e = the base of natural logarithms s = slope in m/m

As indicated above, a model trying to predict the location of the Roman road between Valkenburg and Voerendaal should take into account at least the physical conditions limiting transport, as well the potential influence of 126



6e 3.5|s  0.05|

Verhagen and Jeneson: A Roman Puzzle. Trying to Find the Via Belgica with GIS isotropic and anisotropic cost distance calculations is however considerable, even in a not very mountainous area like South Limburg (Figure 5).

A comparison of Tobler’s equation with other hiking functions shows differences, but, from a practical point of view, they are not that big. The major factor determining whether a different route will be chosen by the LCP algorithm is the difference in cost values encountered. Applying Tobler’s walking speed-based hiking equation in the study area will give ranges between 1.3 km/h and 5.0 km/h, a difference of 3.7 km/h. However, the change in cost between cells usually shows much less variety; the mean range of speed values in a 3x3 cell neighbourhood is 0.7 km/h, with a standard deviation of 0.37. It is only by substantially modifying this change in cost between cells that different routes will be created, especially since the different walking equations all follow more or less the same pattern of speed reduction with increasing slope (see also Herzog 2010).

These problems only become more complex when trying to incorporate measures of visibility into the cost surface. Very little consensus exists about what constitutes a good indicator for terrain visibility. Viewshed analysis, much like LCP modelling, has resulted in a number of archaeological publications (Wheatley and Gillings 2000; Llobera 2003; Llobera 2007; Lock and Pouncett 2010) pointing to the importance of including the scale of visibility into the analysis. The total viewshed, while a seemingly popular measure of overall visibility of a location, is in practice still a complex parameter to calculate, especially when trying to include multiple viewing distances. For reasons of practical performance, we decided not to use total viewsheds for our assessment of visibility, but a measure known as openness (Yokoyama et al. 2002), which measures the angles of zenith and nadir within a selected neighbourhood. It gives a readily interpretable image of the degree in which a location is sheltered or open to vision, and as such is a quicker alternative to total viewshed calculations. However, a comparison between this method and other options such as the skyview factor (Watson and Johnson 1987; Kokalj et al. 2010) has never been performed, let alone that we can be sure that it actually captures what we want to model, i.e. the potential that a position in the landscape offers for visual control.

A second issue is the accuracy of the digital elevation data that can be used to create the cost surfaces based on slope. While there is a nation-wide elevation coverage of the Netherlands obtained through LiDAR (Actueel Hoogtebestand Nederland or AHN), the derived product that is most commonly used is a 5x5 m filtered DEM that has most of the vegetation removed, but leaves most standing buildings and a considerable amount of other man-made structures, like railways and highways. In this particular case, the highway connecting Maastricht to Heerlen runs just north of Valkenburg in the direction of Voerendaal and crosses the possible location of the Roman road just to the southwest of this village. A test using the filtered AHN and Tobler’s hiking function resulted in a LCP that moved up the highway talus and followed it for much of the stretch. The only other easily available product is the Aster DEM, which has a coarser resolution of approximately 35x35 m and a reduced vertical accuracy with elevations in meters, not in centimers. Nevertheless, this DEM is better suited for our purpose since the influence of man-made structures on the cost surfaces and LCPs is much reduced. It gives a better approximation of the relief in Roman times, but at the same time cannot make claims to extreme accuracy. A third issue, also identified by various authors, is the use of different cost distance algorithms for path finding (Conolly and Lake 2006; Herzog and Posluschny 2008). Most GIS packages use a version of Dijkstra’s algorithm to calculate the cumulative costs. However, in most cases these costs are isotropic: it does not matter whether one moves in one direction or another, since the cost is dependent on the slope value in each raster cell. Since slope values are always averages of slopes in 8 directions, this means that slope maps will almost never give the true value of slope in the direction of movement. Even though several authors have warned for this and even provided alternatives (Conolly and Lake 2006; Zakšek et al. 2008), most GIS packages deal with this in a highly unsatisfactory way. Only the Path Distance module in ArcGIS seems to be able to calculate movement costs directly from a DEM and combine these with hiking functions and other cost specifications. Unfortunately, the documentation provided with the software is not specific on how this is achieved. The difference between using

Figure 5 Least cost paths based on Aster DEM using Tobler’s hiking function. The black path was created using isotropic costs, the white path using anisotropic costs

5. Results Despite the limitations of the available data set and modelling procedures, the modelling resulted in some quite interesting results. First of all, as already shown in figure 5, it makes a difference whether we assume isotropic or anisotropic costs. Both options produce fairly straight lines between Valkenburg and Voerendaal. The anisotropic route however takes advantage of the terrain by slowly moving uphill and following the plateau ridge, 127



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process whereas the isotropic path tries to follow the valleys before crossing the plateau at a relatively low elevation.

feasible option for the Romans. The only possible bottleneck on this stretch is the slope just to the west of the Goudsberg that may have been too steep for wheeled transport.

All this changes drastically when we let the LCP depend solely on the openness factor (Figure 6). It creates a route that consequently takes the ridges and avoids all valleys. However, this result could only be achieved by first calculating the mean openness value within a 3-cell circular neighbourhood and then normalizing (contrast stretching) the values obtained. In this way, a pattern of openness ‘cost’ values was created that would force the LCP to follow the ridge. Subsequently introducing slope into the equation did not produce very different results. This again poses serious questions about what methods are the most suitable to model visibility, and how to combine the results with other cost parameters. The route eventually modelled largely coincides with the preferred option suggested by Demey and Roymans and van Aken, passing close by the Klimmen church and the Goudsberg watchtower. This in itself indicates that van Aken’s ideas on the position of the watchtowers are correct: both the church and the Goudsberg are positioned in locations with good visibility. However, if we look more closely at the viewsheds from both locations (Figure 7), it seems that they are hardly intervisible. If we assume that the watchtowers were 10 m high (and allowing for some inaccuracy of the DEM) it might just be possible that each tower controlled about half of the possible route over the plateau, but it is highly improbable that they could see each other very well. As the main purpose of watchtowers was that the people manning them could communicate with each other, by means of signalling with torches for example, the fact that the location of these two supposed towers prohibits such communication surely weakens the argument in favour of them. Another important question is the efficiency of each modelled route, both in distance and in speed. A comparison of modelled route distances and speeds to a straight line results in the following figures:

Figure 6 Least cost path based on openness factor. The modelled route passes close by the Klimmen church (K) and the Goudsberg tower (G).

Route

Length

Walking Time

straight line

6700 m

1h 20’

Figure 7 Viewshed from Klimmen (K) and Goudsberg (G), based on Aster DEM and an assumed height of 10 m for Roman watchtowers. Dark grey= Goudsberg viewshed, light grey= Klimmen viewshed, medium grey = overlap of viewsheds.

isotropic slope

6840 m

1h 24’

6. Thinking beyond the tool?

anisotropic slope

7440 m

1h 30’

openness

7730 m

1h 33’

While working on this case study, we have tried to maintain a pragmatic approach to the issue of finding plausible routes of the Roman road section between Valkenburg and Voerendaal. In contrast to the existing theories, we have tried to be specific about the factors involved, using LCP models to find the optimal connections based on slope and visibility. However, in the process we were forced to think more about the tool than beyond it. Several issues in LCP modelling are not solved from a practical point of view. The software is a major limiting factor in this. Comparison of LCPs and the manipulation of the different variables involved should be much easier than it is now, and algorithms should be

The difference between the shortest and longest variant is 13% (890 m). In terms of walking speed it is 11.5% (9 minutes). It will of course be very difficult to judge what would be the financial consequences of construction and maintenance of a detour to the Roman administration. However, the difference does not seem to be prohibitive, and we can conclude that a route following the most visible locations in the landscape may have been a 128



Verhagen and Jeneson: A Roman Puzzle. Trying to Find the Via Belgica with GIS suggest that archaeologists should stop theorizing about fundamental notions on how people moved in the past, we do feel that more energy could and should be invested in developing software tools that will allow us to more easily compare different theoretical perspectives.

open and more versatile. In theory, building different models with different weights attributed to the various factors should not be that difficult. In practice, it takes a number of convoluted steps to get to this point. Since ArcGIS is the only package that will more or less accurately deal with anisotropic costs from DEMs, we were also stuck with all the other aspects of using the accompanying interface. For example, if we want to know the actual costs of each LCP instead of its length, this is not automatically produced by ArcGIS. We also have to be aware that LCP models only provide a single solution. We cannot specify what will be the second best, third best or n-best path on the basis of a single cost surface.

We hope to use the models developed to guide field testing using ground penetrating radar in the near future. Ultimately, ground truthing is necessary to establish the validity of the modelling results, something which is all too often neglected in predictive modelling exercises. After all, GIS-based modelling is only one of the options available to try to predict archaeological remains. It can only be considered successful if we can actually prove that it did a better job than an approach based on expert judgement alone.

Theoretical concerns had to be confronted as well: the modelling of movement speeds for (pre-) historical wheeled transport is an area where no previous work is done, and it would seem fairly difficult to take this much further without a form of experimental archaeology. An additional problem is found in the modelling of visibility and its possible impact on movement. For pragmatic reasons we have used the openness parameter for this and with some manipulation succeeded in creating a visually prominent route that intuitively seems to make sense. However, this was more based on trial and error than on a good understanding of the interplay of the factors involved.



Acknowledgments The writing of this paper was partly made possible by a VENI research grant from NWO (the Netherlands Organization for Scientific Research) awarded to the first author.

References Bazelmans, J., Bakels, C. and Kocken, M. 2004. De Romeinse wachtpost op de Goudsberg. Een verslag van de opgraving Valkenburg aan de Geul 2002. Historische en heemkundige studies in en rond het Geuldal (Jaarboek 2004), 61-86.

Having said this, we also see clear positive sides to using a GIS-based approach for trying to predict the location of the Via Belgica. First of all, we were able to shed some light on the plausibility of the hypotheses of Demey and Roymans and van Aken on the possible location of the route. While the latter concluded that the road should have connected the two watchtowers, in fact the route suggested by him is favouring good visibility all along the stretch. Interestingly, the suggested watchtower locations may not have been in the best position to control the road. If they really were watchtowers, there may have been a third one in between. Secondly, thanks to the modelling we were able to compare options, not just in length, but also in walking speed. A route favouring good visibility is longer and slower than a route taking the flattest areas, but it is not prohibitive.

Becker, K. M. and Altschul, J. H. 2008. Path finding. The archaeology of trails and trail systems, in Altschul, J. H. and Rankin, A. G. (eds), Fragile patterns. The archaeology of Western Papaguería, 420-446. Tucson, SRI Press. Bell, T. and Lock, G. 2000. Topographic and cultural influences on walking the Ridgeway in later prehistoric times, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies, 85-100. Amsterdam, IOS Press/Ohmsha. Bell, T., Wilson, A. and Wickham, A. 2002. Tracking the Samnites: landscape and communication routes in the Sangro Valley, Italy. American Journal of Archaeology 106, 169-186.

Within the constraints of the current case study - which had to be carried out on a very limited budget - it was however impossible to seriously tackle the fundamental theoretical issues concerning GIS-based modelling of (pre-) historic movement. Most of these issues have been, and still are, the subject of long-standing debate in archaeological (computing) literature, without a real consensus having emerged on the best way to deal with them. And those studies that have reported successes have not yet resulted in the development of specific software tools that are available to a wider archaeological community. In a sense, we are still lacking the appropriate ‘spatial language’ and accompanying toolboxes to adequately approach questions like where the Romans put their roads. And while we don’t want to

Bellavia, G. (2006). Predicting communication routes, in Haldon, J. F. (ed.), General issues in the study of Medieval logistics, 184-198. Leiden, Brill. Conolly, J. and Lake, M. 2006. Geographical information systems in archaeology. Cambridge, Cambridge University Press. Demey, D. and Roymans, J. 2004. De Romeinse weg van Boulogne-sur-Mer naar Keulen. Provincie Limburg. Een archeologisch onderzoek. Historische Studies Geuldal 14, 7-60. 129



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Lock, G. and Pouncett, J. 2010. Walking the ridgeway revisited: the methodological and theoretical implications of scale dependency for the derivation of slope and calculation of least-cost pathways, in Frischer, B., Webb Crawford, J. and Koller, D. (eds), Making history interactive. Computer applications and quantitative methods in archaeology (CAA2009). Proceedings of the 37th international conference, Williamsburg, Virginia, United States of America, March 22-26, 2009, 191-202. Oxford, BAR Publishing.

Fiz, I. and Orengo, H. 2008. Simulating communication routes in Mediterranean alluvial plains, in Posluschny, A., Lambers, K. and Herzog, I. (eds), Layers of perception. Proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA 2007), Berlin, Germany, April 2–6, 2007, 316-321. Bonn, Dr. Rudolf Habelt GmbH. Gietl, R., Doneus, M. and Fera, M. 2008. Cost distance analysis in an alpine environment: comparison of different cost surface modules, in Posluschny, A., Lambers, K. and Herzog, I. (eds), Layers of perception. proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA 2007), Berlin, Germany, April 2–6, 2007, 342 and CD-ROM. Bonn, Dr. Rudolf Habelt GmbH.

Murrieta Flores, P. A. 2010. Traveling in a prehistoric landscape: exploring the influences that shape human movement, in Frischer, B., Webb Crawford, J. and Koller, D. (eds), Making history interactive. Computer applications and quantitative methods in archaeology (CAA 2009). Proceedings of the 37th international conference, Williamsburg, Virginia, United States of America, March 22-26, 2009, 249-267. Oxford, BAR Publishing.

Herzog, I. and Posluschny, A. 2008. Tilt–slopedependent least cost path calculations revisited. 36th international conference on computer applications and quantitative methods in archaeology CAA2008, Budapest, 2-6 April.

Raepsaet, G. 2002. Attelages et techniques de transport dans le monde gréco-romain. Brussels, Timperman. Roth, J. P. 1998. The logistics of the Roman army at war (264 B.C. - A.D. 235). Leiden, Brill.

Herzog, I. 2010. Theory and practice of cost functions. 38th annual conference on computer applications and quantitative methods in archaeology CAA2010, Granada, 6-10 April.

Tobler, W. 1993. Three presentations on geographical analysis and modeling. Santa Barbara, University of California.

Howey, M. C. L. 2007. Using multi-criteria cost surface analysis to explore past regional landscapes: a case study of ritual activity and social interaction in Michigan, AD 1200-1600. Journal of Archaeological Science 34, 18301846.

Watson, I. D. and Johnson, G. T. 1987. Graphical estimation of sky viewfactors in urban environments. Journal of Climatology 7, 193-197. Wheatley, D., and Gillings, M. 2000. Vision, perception and GIS: developing enriched approaches to the study of archaeological visibility, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies, 1-27. Amsterdam, IOS Press/Ohmsha.

Kokalj, Ž., Zakšek, K. and Oštir, K. 2010. Archaeological Application of an Advanced Visualisation Technique Based on Diffuse Illumination, in Reuter, R. (ed.), Proceedings of the 30th EARSeL symposium: Remote Sensing for Science, Education and Culture, Paris, 31 May – 3 June, 113-120. Hannover, EARSeL. Available at: http://www.conferences.earsel.org/system/uploads/ass et/file/58/earsel-symposium-2010_2-09.pdf.

Yokoyama, R., Shirasawa, M. and Pike, R. J. 2002. Visualizing topography by openness: a new application of image processing to digital elevation models. Photogrammetric Engineeering and Remote Sensing 68, 257-265.

Llobera, M. 2000. Understanding movement: a pilot model towards the sociology of movement, in Lock, G. (ed.) Beyond the Map: archaeology and spatial technologies, 65-84. Amsterdam, IOS Press/Ohmsha.

Zakšek, K., Fovet, E, Nuninger, L. and Pobodnikar, T. 2008. Path modelling and settlement pattern, in Posluschny, A., Lambers, K. and Herzog, I. (eds), Layers of perception. Proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA 2007), Berlin, Germany, April 2–6, 2007, 309-315. Bonn, Dr. Rudolf Habelt GmbH.

Llobera, M 2003. Extending GIS-based visual analysis: the concept of visualscapes. International Journal of Geographical Information Science 17, 25-48. Llobera, M 2007. Reconstructing visual landscapes. World Archaeology 39, 51-69. Llobera, M. and Sluckin, T.M. 2007. Zigzagging: theoretical insights on climbing strategies. Journal of Theoretical Biology 249, 206-217.

130



9 Deconstructing and Reconstructing the Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS Ignacio Fiz, Eva Subias, Rosa Cuesta Universidad Rovira i Virgili (URV), Catalan Institute for Classical Archaeology (ICAC), Archaeological and Archaeometric Research Team, University of Barcelona (ERAUB) ________________________________________________________________________________ Abstract In this study, we present the work done within the project ‘The organization of space in Greco-Roman Egypt’, carried out with the support of the Plan I + D + I (HAR2008 -01 623) and the additional grant HAR2010-10 368-E, both of which were provided by the Spanish Ministry of Science and Technology. Our study involved the collection and analysis of data on archaeological sites and traces on the landscape. All this information was recorded using cartography and satellite images in order to reconstruct the ancient nomos of the Greco-Roman city of Oxyrhynchus (elBahnasa, Al Minya). Our work is the result of reading and analysis of texts written in the nineteenth and twentieth centuries on management of the flooding of the Nile and the reforms in the irrigation systems in the Middle Nile Valley. By amalgamating this information we were able to reconstruct the Egyptian landscape as it was before the arrival of Napoleon Bonaparte's expedition in 1799. We also use these results to discuss the application of remote sensing in archaeology. Keywords: GIS; Remote Sensing; Description of Egypt; Oxyrhynchus; CORONA; ASTER; LANDSAT; PCA; TTC.

________________________________________________________________________________ irrigation or drainage channels, and determine the organisation and structure of the anthropised landscape of the present and future. They are types of landscape which, become ‘morphogenic’ after they are created, due to their tendency to become widespread especially among the subsequent forms constructed with the passing of time (Levau 2000, 559-560). This morphological approach will enable us to recognise and contextualise them. However, this approach must be ‘historicised’ if it is not merely to be an exercise in detecting and modulating the traces (Orejas et al. 2002, 290)

1. Introduction The study of the landscape of ancient Oxyrhynchus inevitably involves a process of ascertaining and consideration of the workings and management of the flooding of the Nile. This is understandable when we consider that Egypt has always depended on the seasonal and cyclical flooding of the Nile in all its facets. Its effects, apart from those of a natural cause, had a cultural, visual, social, economic, and religious impact on the entire country, creating a unique society that was different from other Mediterranean cultures. Control and management of the waters has been essential in ensuring food supplies and social order from the Predynastic era until the present day.

The nineteenth century is therefore important to us due to the major changes and projects that took place, first under the French engineers directed by Mehmet Ali and then under the British protectorate. All these factors contributed to the major transformation and modernisation of the way the Nile's water was managed, in terms of both seasonal flooding and perennial irrigation. In the twentieth century, the construction of the Aswan Dam in the mid-1960s led to the final transformation and the end of a natural, social and cultural way of understanding and conceiving of the Nile.

An understanding of the how the procedures, techniques and methods for managing and controlling flooding developed, especially during the nineteenth century when the country modernised the way it irrigated its land, is a prerequisite for any application of other landscape analysis techniques. This assumption, which is the cornerstone for this project, is based on the tradition begun by the Besançon School in the 1970s. According to this school of thought, a process of understanding and analysis of major anthropic transformations of the natural landscape is necessary. These processes, which are related to the ways in which territories are organised and exploited, leave their imprints in the form of paths, land divisions, dikes and

The most spectacular example of this change is nineteenth-century Cairo. In August every year, after the announcement of the annual flooding by the Nilometer on Rhoda Island, the flooding began by breaking the dike that held the waters of one of the channels that crossed the city: the Khalig al-Masri (Linant de Bellefonds 1872, 39; Norden 1755, 44). The floodwaters entered the city, 131



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process turning some of its streets and squares into canals and lagoons. The French artists and engravers that accompanied Napoleon's expedition to Egypt (Figure 1), who must have been influenced by the ‘vedute’ pictorial genre as painted by Canaletto, Guardi and Carlevarijs, depicted a city of Cairo affected by flooding which was completely adapted to the urban environment. However, the changes made to the Nile's flooding system in the nineteenth and the twentieth centuries drastically changed this phenomenon, banishing it not only from the landscape, but also very probably from social memory.

transformation, management, and improvement of the irrigation system by means of flooding or the perennial supply of water. The sources studied from the nineteenth century were first, the series of works about Egypt published between 1809 and 1829 entitled Description de l'Égypte, ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française. This major work provided an encyclopaedic and scientific overview of ancient and modern Egypt. It was a collective work by a number of scientists, researchers, artists, and technicians, known as the ‘Savants,’ who accompanied Napoleon Bonaparte's expedition between 1798 and 1801. The second source, which was also very important, was written by Linant de Bellefonds (1872) who was the Minister of Public Works and a member of Muhammad Ali's private council. His book, published in 1872, is a memoir of his experiences dating back to 1817. The text, which at some points is merely a narrative, goes into considerable depth at other points, presenting reports, budgets, and essential reference information on the irrigation system before the major changes, to which the author was a first hand witness and one of the parties involved, took place during the nineteenth century.

Figure 1 ‘Vedute’ of the named Birket-el-Fyl square, in Cairo, during the flood. (Martin 1822, 80, plate 39) Using this knowledge of the flooding and management of the waters of the Nile, we worked with an archaeological map of the area of the project, and recorded the archaeological sites and traces of water management (channels, dikes, etc.). These were recovered using map interpretation, photointerpretation, and remote sensing techniques, using first modern and historical maps, and then the medium and high resolution satellite images available.

J. Barois (1887), the first Secretary to the Minister of Public Works, gives us essential information on the hydraulic management of irrigation in the Middle Valley of the Nile in 1887, almost 14 years after the construction of the Ibrahimiya Canal. This canal provided perennial irrigation to the eastern half of the Middle Valley and had therefore completely changed the water distribution channels and drainage system. The study was subsequently reprinted in 1904, with the details on the flooding by basins completed and updated.

We can use these traces and texts, which tell us about major irrigation projects, to reconstruct the landscape as it was prior to the nineteenth century. By ascertaining the major nineteenth-century projects that reformed or changed the landscape, we can identify the features that existed beforehand. Taking this as our starting point, and by reading primary sources, we attempted to interpret and reconstruct the landscape as it was in the ancient world, in anticipation of confirmation on the ground and corroboration of data using geomorphological and paleoenvironmental techniques.

Finally, Sir William Willcocks published the book ‘Egyptian Irrigation’ just two years after Barois, and it was reprinted in 1913 with James Ireland Craig (Willcocks, Craig 1913). This reprint is of major interest as it lists the changes and projects that took place between the end of the nineteenth century and beginning of the twentieth century. Willcocks began working in the Egyptian Public Works Department in 1883 onwards, and the projects in which he was involved included the supervision of the construction of the Aswan and Assiut dams. His work, divided in two volumes, is an exhaustive review of the defining features of the geographical, geological, and natural features of the Nile and its irrigation system. His work also defines, describes and classifies the dams, canals, dikes, and regulators and all the other items necessary for management of hydraulic engineering in Egypt, and how they were built.

2. From the eighteenth to the twentieth century: transformation and continuity in the management of irrigation in Middle Egypt. 2.1 The sources The regressive study of the irrigation system in Egypt was based on an exhaustive review of a series of publications dating from the eighteenth, nineteenth, and early twentieth centuries. It should be noted that some of the nineteenth-century works were written by authors who were directly involved in the study, analysis,

2.2. Flooding in the eighteenth century A review of some texts dating from before Bonaparte's expedition in 1799 gives only limited evidence of how the flooding of the Nile took place and the country's 132



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS ‘Quand il est cru a son point, and qu’il repandu ses eaux sur la surface de la terre; c’est alors qu’on pense de les retenir durant quelque tems, afin de que les terres ayent le loissir de s’abbreuver sufissanment. Pour cet effet, on practique les digues, apellées Giffer, qui empêchent que l’eau ne s’ecoule, and l’arretent auntant que le tems qu’on le juge à propos. Enfin quand la terre est assez arrosé, on coupe le Giffer, pour faciliter l’ecoulament des eaux(...)’.

spectacular seasonal transformation. It should be noted that these impressions, which are sometimes written as travel memoirs, are part of the spirit of discovery of the Grand Tour, which had previously been limited to Italy. Some of these publications therefore take the form of expeditions or journeys, and contain a mixture of geographical, ethnographic, naturalistic, and archaeological descriptions, as well as mere anecdotes. Agriculture and flooding are therefore described in general terms, without entering into any detail, and the works are therefore not of a specialist nature. A meticulous description of flood management mechanics is only provided in a few cases.

This Danish author enables us to imagine how the flooding of the Middle Nile Valley was undertaken. During the flooding, excess water from the Nile was channelled into canals, which flooded the land that they surrounded after they overflowed. The dikes held back the water and stored it long enough for its supply to enhance the fertility of the harvest.

Among the more interesting works is that of the impressions of Benoït de Maillet (1740, 87), who was Consul General in Egypt between 1692 and 1708. His perspective, which was that of a traveller, mentions the flooding from the point of view of a spectacle:

Jean Baptiste D’Anville, the great eighteenth-century geographer, presented a vision of Egypt in 1766 which was subsequently rejected by the Savants. According to the French author, the Middle Nile Valley had three major watercourses: the Nile, the Bahr-Yusef, and the Bahr Baten. The latter, which he mistakenly identified as the Lake Moeris mentioned by classical sources, ran parallel between the Bahr-Yusef and the Nile. The Bahr Baten, which came from the Nile somewhere between Minia and Samalut, continued until Benisuef. A series of canals ran across the valley between the Nile and the Bahr. D’Anville (1766, 154-156) makes the assumption based on classical sources that perhaps the Baten was not a natural phenomenon and there had been a series of dikes to hold back its waters. However D’Anville does not describe the phenomenon of flooding and when he observes the Baten he apparently does not understand its seasonal nature. His preoccupation with identifying and reconstructing the territory of ancient Egypt based on source materials prevented him when considering the Baten as a natural phenomenon and analysing it its anthropic management, as Norden had suggested.

‘La vue de l'Egipte dans les terms d'inundation est sans contredit un spectacle le plus charmans du monde. C'est alors que du haut des montagnes on découvre une vaste mer, d'ou s'elevent des villes et bourgardes sans nombre qui n'ont de communication entr'elles que par des chausées élevées a ce dessein. Les eaux quelquefois sont si abondantes, qu'elles inondent les chausées mêmes. Alors la communication se fait en bateaux and ce n'est pas un mediocre agrément de voir, tout le pais couvert des maisons flotants.’. This text, published in 1735, basically presents a situation very similar to that described by Herodotus two thousand years previously (Hdt, II, 97): ‘When the Nile comes over the land, the cities are alone and are seen rising above the water, resembling more nearly than anything else the island in the Aegean Sea; for the rest of Egypt becomes a sea and the cities alone rise above water. Accordingly, whenever this happens, they pass by water not now by the channels of the river but over the midst of the plain.’

Finally, in 1789, Claude-Etienne de Savary briefly described the effect of the flooding and tells us about canals as a means of distributing the floodwaters, and about the dikes that held back the waters (Savary 1789, 198-199).

It is obvious that the flooding created the same sense of astonishment despite the centuries that had elapsed. De Maillet (1740, 88) tells little more, except for the canals' role as an essential part of the flooding process:

Over the eighteenth century the various authors who mention the flooding of the Nile did not, therefore, go into exhaustive detail about the mechanics of the canals and dikes. They did know that the canals distributed floodwater in the Valley, while the dikes were used to contain the water and as a means of communication between towns and villages. Some of them justified the changes and transformations that took place over the subsequent century in terms of neglect and lack of maintenance of the canals and dikes, and the serious danger that this entailed for the country's future (Norden 1755, 62; Martin 1813, 199).

‘C'est par là que l'Egypte approfite les acroissements du Nil; c'est par là que ses eaux son repandues dans tout cette contrée, and i portent la fertilitée’. Louis Frederic Nörden, a Danish ship's captain, who explored Egypt and Sudan between 1737 and 1738, gives us a more specific view of the flooding of the Nile and its management and use (1755, 61-62): ‘Ces moyens consistent en des Digues and en des Calichs, ou canaux, que l’on coupe, on creuse dans les endroits, òu le bord de le Nil est bas. On les conduït jusqu’a les montagnes, au travers des provinces entieres; de sorte que, quand le Nil croît, les eaux entrent dans les Calichs qui les introduisent au dedans du pais, à proportion de la hauteur du Fleuve(...)’. 133



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process respectively as d’inoundation.

2.3. Management of flooding in the early nineteenth century

irrigation

basins

or

bassins

In the early nineteenth century, the mechanics of this system consisted of the basins being flooded one after the other, so that when one of them was full and the water had been held back long enough for the sediment to be deposited, the dike was then breached. The water was allowed to pass along the canal that had flooded the first basin to drown the fields in the next one, until it was once again held back by another transverse dike.

Dikes and basin flooding All the nineteenth-century authors agreed that the control of flooding by means of dikes and canals had begun earlier in Egypt’s history (Barois 1887, 8-9). However, nothing remained of these hydraulic structures, according to Barois, as neglect and disuse had destroyed any traces of their existence. However, Linant maintained that some ancient features of hydraulic engineering could still be seen on the landscape. This was true of the Kocheïche dike, located 6 km to the south of the modern town of Maidum, or the stone structures on both sides of the Nile between the cataracts of Aswan and Wadi Halfa in Nubia. Linant's experiences on his travels, when he explored the country for years before becoming a civil servant for the Egyptian government, confirms this.

These dikes, and especially the older ones, have a winding layout due to the fact that the breaches in the circulation points of the flow meant that the walls had to be repaired by moving them backwards or forwards from their original position. As a result, the dikes ceased to be arranged in a straight line as time passed (Barois 1904, 265-266). The waters were not only controlled by breaching the dikes. In some dikes, brick bridges had been built with arches 3 metres long, with each pillar acting as a regulator. These enabled the water to circulate after it had been held back for a sufficient length of time in the previous basin.

The situation in Egypt prior to hydrographic management in the early nineteenth century was something that Linant, Barois, and Willcocks mention vaguely and without going into detail. Taking into account that there are no exhaustive details from the eighteenth century, and in some cases they are even confusing, only the Description by the Savants provides us with enough data for a reconstruction. The team of scientists that accompanied Napoleon's expedition to Egypt described how the system worked in the following terms (Girard 1823, 496-497):

The same work, when describing the hydrography of the province of Benisuef, lists three types of dikes: large, medium and small. The large dikes crossed the valley from one side to another, with the largest being the dike of Oukechechy or Kocheicha (Martin 1813, 198-199) and the other dikes being:

‘Ces canaux sont diriges dans le haut Egypte plus ou moins obliquement, vers les deux chaines de montagnes qui ordent la vallée: parvenus a leur pied, ils se prolongent parallèment au desert; mais des diques transversales en interrompen le cours, de sorte que ses eaux interropes par ces diques s'élevent contre elles et submergen une partie des terres qu'elles enferment.’.

‘(...)Behabchyn, Safanyeh, Saftrachin, el-Noueyreh, Choubak, Ehoueh, Badahal ou el Chantour, Samalout, Menbaâl et Bardanoah.’ This list is very valuable and useful as it is the first time that a source provides a list of dikes prior to the nineteenth-century reforms. A comparison with other lists for the basins in operation during the nineteenth century has enabled us to date some of them as modern. The same classification distinguishes the medium-sized dikes as those originating in the Nile, and the large dikes, which ended in towns built on mounds.

‘(...)Quand cette submersion atteinte sa plus grande hauteur on coupe la dique qui soutenait les eaux; elles s'ecoulent alors au-delà de cette digue, en suivant le même canal, qui se prolongue lui-même sur la limite du desert, jusq'a un second barrage qui, arretant de noveau les eaux, les oblige de se gonfler, et de se repandre, sur une partie de l'espace renferme entre deux digues tranversales consécutives.’.

The flood canals

‘On coupe la seconde digué comme on avait coupé la premiere; les eaux descendent de la même maniere contre una troissieme ; qui produït à son tour la submersion d'une certaine étendue du terrain; et ainsi de suite (...)’.

Part of the seasonal flooding of the Middle Valley was due to the flooding of the Bahr-Youssef. This natural canal follows a very winding course, with a width of between 50 and 60 metres and a maximum depth of 6 to 8 metres. With the flooding of the water, the high water level is confined by dikes and by the mountains defining the valley at all points to the west. Its waters reach the Fayoum, but it also acts as a feeder for the flooding of basins and as an overflow channel for their drainage when the waters fall (Linant de Bellefonds 1873, 14).

In other words, the irrigation system consisted of a series of natural basins, separated by dikes, which were flooded by canals during the flooding season. Full advantage was therefore taken of the gradient of the Nile Valley until its mouth, in order to facilitate a uniform distribution of water loaded with sediment. As well as holding back the waters, the dikes, which were built with reinforced earth, also acted as a communications system during flooding periods (Girard 1823, 497; 1813, 352-353). These flood basins were known by the English and French

However, this natural course was not the only one involved in the water's flooding. The Bahr Baten, which was mentioned by D’Anville some years before, was studied by scientists on the French expedition. The Arab term Baten was a generic name applied to: 134



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS earlier canal in this area in his memoirs in the nineteenth century, we must assume that it already existed at that time and was one of the preexisting Fyad Baten canals that crossed several provinces.

‘(…)presque tous les canaux qui parcourent l’interieur des terres dans direction du sud au nord’ (Martin 1813, 206). However, the expression Fyad Baten was used to refer to large canals. The largest of these, according to the Description, was the source of some confusion with Lake Moeris by Granger, Sicard and D’Anville. The French scientists on Bonaparte's expedition say that the Middle Valley was crisscrossed by large canals running parallel to the Nile, which easily flooded the fields when the waters rose. Another Fyad Baten in the Middle Valley is described as follows: ‘(...)dont l’origine sur le Nil est entre le village de NazletAbou-Esné et celui de Qalousaneh. Il passe au pied du village du Matâtyeh, ou il se divide en deux branches, dont l’une a l’est devient petit Bathen, et se perd, à deux lieus de là, dans les terres d’Abou Girgeh; l’autre, al’oest comunique pendat l’inundation avec le Bahr-youssef, au village d’El-Houeh: mais il n’a plus de trois lieues de longueur’ (Martin 1813,206).

Figure 2 Identified traces of the fyad Bathen described by Martin 1813,206

We have been able to identify this Fyad Baten as one of the restored traces by studying the cartography and remote sensing. This canal follows a similar route to the one described by the Frenchmen (Figure 2). However, the savants do not all agree with each other about the use of this term. Jomard (1809, 104, note 2) says that

Another plate in the Description (Martin 1822, 14, Plate 6) depicting the hydrography of the Middle Valley shows us a plan that places this Baten between Hermopolis and Minieh. The authors produced a whole series of sections of various stretches of the valley which included the Nile, the Bahr-Youssef and the Bahr-Baten. Unfortunately, the work only focused on the area between Samalout and Minieh, which is beyond the scope of our study.

‘On dit un bâtin et plusieurs batin [el-bâtin, el-baouâten] ; ce mot Arabe, qui signifie intérieur, est parfaitement bien appliqué aux bas-fonds dont Je parle, puisqu'ils forment la partie la plus basse et la plus intérieure du pays : ils conservent de l'eau presque toute l'année, et ils offrent, par endroits, l'aspect d'un canal continu.’

Almost sixty years later, Linant de Bellefonds (1873, 3) gives us another perspective on how the flooding of the Nile worked before the nineteenth-century reforms: ‘Lorsque les eaux coulaient librement dans les plaines, celles ci se trouvaient ravines partout par les eaux des crues, et plusieurs de ses ravines conservaient pendant les etiages una certain quantite d’eau courante; aujourd’hui meme il existe beaucoup de ces ravines ou cours d’eau naturales dans toute l’Egypte.’

They were therefore not natural canals, but instead lowlevel locations which retained the water throughout the year after the flooding took place. The explanation for this phenomenon was the unusual layout of the Middle Valley. It was defined by two sloping plains, one from the Nile, and the other from the Bahr-Youssef. Where they came together, they formed a basin or lower-level area which allowed water to be retained for a long period of time and which is why it was called Bahr Baten (Martin 1813, 201), i.e. Inner River. In some cases, these low-lying beds ended up becoming natural canals due to overflows from the Nile. Bahr Baten was therefore a generic name, although if it was necessary to distinguish a large canal, the term Fyad Baten was applied.

‘En descendant du sud vers le nord, on recontrait de ces cours d’eau considerables pendant les crues, et a peu a près dans le même état qu’ils étaient avant que des travaux fussent venus régulariser les débordaments du fleuve; mais aujourd’hui ils sont en grand partie maîtrises et utilisés pour les arrosages.’ In other words, the flooding consisted of allowing the water that overflowed from the Nile to be channelled naturally along natural streams or watercourses and the valley was flooded from those. Some courses retained part of the water throughout the year, while the rest stayed dry. Linant therefore stressed the natural character of these canals as a consequence of the retention and regulation of water by means of dikes. Once they were opened or if they originated in the regulators, built, the waters were channelled seasonally by natural means (Linant de Bellefonds 1873, 4):

In another volume of the Description Jomard (1818, 260) once again describes the phenomenon of the Bahr Baten. According to the text, it flowed irregularly from near the ruins of Hermopolis to further south of Minieh, and was known as Tera’t el Ghouetat and Tera’t el-Sebakh. Both Linant de Bellefond and one of the cartography plates in the Description depicted this canal in great detail (Figure 3). Taking into account that the first major canal to be built in the in Middle Nile Valley was the Ibrahimiya, which was constructed in 1873, and Linant mentions no 135



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process ‘Plusieurs autres ravines ou cours d’eau naturels existent dans les plaines de la Haute-Egipte et surtout dans la Moyenne, mais ils sont de bien moindre importance que les deux precedents (Sohagieh y Bahr Youssef). Souvent ils prennent leur origine en aval d’un des nombreux deversoirs pratiques dans le diques des bassins d’inondation; les eaux, qui echapent par ces déversoirs ravinent la plaine et forment alors des bas fonds que l’on nomme Bathen , ce qui veut dire lieu bas.’

to be flooded, at which point the canal disappeared. Some crossed entire provinces and channelled water by means of diversions that were used for floods. Linant made a distinction between these canals and those known as sefi which supplied perennial irrigation, and which he attributed directly to Mohammad Ali's government, as there were no references to the contrary. However, by doing so Linant shows us that the Nili were canals that predated Mohammad Ali's government. Linant's description matches the Fyad Baten or the Bahr Baten in the Description. 2.4. Subsequent transformations The Ibrahimiya Canal was opened in 1873, beginning perennial irrigation in the Middle Valley and leading to the introduction of sugar cane (Barois 1887, 28).The head of this 268-km long canal is at Assiut, and it waters the higher land near the Nile, from which it takes its water. This canal, which runs parallel to the river, was built using some of the old stretches of canal (Barois 1887, 40), in the same way as many of the sefi canals built during the reign of Muhammad Ali were built using stretches of the old Nili canals. Its discharge provides water during the summer, together with that of the BahrYussef. The latter's head was changed to receive the discharge of the Ibrahimya. During flooding, the Bahr Youssef was used to feed the basins that covered the area between the dikes located in Derout and the dike of Kocheicha, in the north of the province of Beni-Souef. In the second half of the nineteenth century, the Bahr-Youssef itself acted as a feeder for a series of 16 basins that covered a distance of 200 km. The basins flooded by the Bahr-Youssef and the land irrigated by the Ibrahimiya were separated by a large longitudinal dike called mohit (Barois 1904,94), which ran from Derout until the end of the canal. This is very interesting, as it gives us details of a landscape adapted to the two different systems for water management and cultivation which were separated from each other by a line set at a right angle.

Figure 3 The Bahr Bathen: 1) Jacotin 1826, 44, plate 14. 2) Linant de Bellefonds 1854. 3) 60's CORONA image, detail of the same area. Linant de Bellefonds also uses the term Bathen to refer not to the canals, the Fyâd Bathen in the Description, but instead to the riverbed where the waters accumulated. The water from one of the flood basins was channelled and distributed by means of natural watercourses, forming natural lagoons at the lowest points above sea level. However, one can assume that these watercourses or streams were part of a system of canals of anthropic origin which had become natural due to the lack of maintenance mentioned by various eighteenth century authors.

In 1897, work also began on the conversion of the basins system in Upper Egypt to perennial irrigation. In 1904 Barois reprinted his work and says that both systems operated separately but that there were some exceptions to the rule. First, there was the low-level land in the flood basins which could extract water from the water table using a water pump, a sakia or shaduf (1904, 51-52). Crops could therefore be irrigated before the flooding took place. This system was called a qedi. The canals used for the flooding, the nili mentioned by Linant that were only used when waters rose, were also used to channel the water extracted using pumps or waterwheels located near the Nile and the Bahr-Yusef (Barois, 1904, 97-99).

However, these natural courses were not the only ones, as according to Linant (1873, 19-20) there were canals called Nili, which were only used during the Nile floods. They were dug at four metres below the water circulation level and their depth decreased until they reached the land

Willcocks says that ideal model for a basin would be one with perennial irrigation from the subsoil enabling the 136



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS As for cartography produced prior to the construction of Aswan Dam, we digitised and georeferenced the topographical plans (1:50K, 1:100K, 1:200K) produced by the Survey Department of Egypt between 1906 and 1939, when the country was a British protectorate. The precision and detail of these maps was essential in the detection of the structures used in the country's hydrographic management.

cultivation of sugar cane for one or two years, which would then be flooded for one year (Willcocks, Craig, 1913, 347) Finally, in 1894, a technical committee of the Ministry of Public Works produced a report to ascertain which of the locations in the Wadi proposed by the government engineers in Cairo (including Willcocks) would be chosen for the construction of a large dam (Willcocks, Craig, 1913, 681). This committee's unanimous conclusion was that Aswan was the only possible location for a reservoir that could store enough water from the flooding and which could supply a sufficient volume of flow to the Nile Valley during the summer. After the plan was approved, work began in 1898 and the Aswan Low Dam was completed in late 1902. It was nearly overtopped in 1946, and it became apparent that a larger project was necessary. This project was not planned until 1954, as part of the shift towards Egyptian nationalism under Nasser. Work did not begin until 1960, the first planned dam was completed in 1964, and the project was finished in 1976, when the reservoir reached its full capacity, providing all of Egypt with perennial irrigation. It appears that the flood basin system was still in operation in 1964, and that the dikes were no longer used after that date, and the flooding that was a feature of the Egyptian landscape until that point passed into memory.

In both cases, the use of these sources facilitated the location of place names of settlements mentioned in documents from the nineteenth century, despite the variation they have undergone over the past two hundred years. All the features recognised using this medium were added to the SIG, creating the appropriate vector layers: cities, current place name, hydraulic infrastructures, courses of the Nile and Bahr-Youssef, etc. The nineteenth-century cartography used was based mainly on the Carta Hidrographique de la Moyenne Egipte by Linant de Bellefonds, produced in 1854 and revised in 1883. This plan was very useful in the reconstruction of the Middle Valley during the nineteenth century due not only to the quality with which it was produced over a large area, but also because it includes very useful place names and details such as the location and structure of dikes and regulators, and the location of several planned construction projects in 1854.

3. Methodology

A problem for which we did not find a solution was involved in the 51 1:100K scale plans that accompany the Description de l’Egypte. The problems and errors in its location of settlements and the location and paths of dikes and canals (highlighted by Gomà) were sufficient grounds not to use them as a main source of work or as support cartography in the research. We only occasionally used some of the plates in the work by the Savants for the purposes of illustration of the discussion on the Bathen.

The sections above show that the process of transformation and change in the Middle Valley was a complex one. The large hydraulic infrastructures, irrigation and drainage channels built during the nineteenth century overlaid and concealed the systems dating from before the 1799 expedition. We therefore know how the flood basins structured by dikes operated, the various definitions applied to the term Baten, the difference between a sefi and a nili canal and we know that the valley was transformed in various stages over more than one hundred years, with each change leaving its imprint on the land. In view of these assumptions, we applied methods and techniques, which we used to reconstruct the various landscapes of the Middle Valley in the nineteenth century, as well as an initial consideration of the area around the Oxyrhynchus site.

Satellite images and DEM Remote sensing was the second means of study, due to the spatial information it provides and which is not subject to conceptual filtering by cartographers. Its application in archaeology can be dated back to the aerial photographs of Stonehenge taken by Sharpe on the Salisbury plain in 1906. In subsequent years, the various wars in the international arena, the length of combat fronts and the need to ascertain military and industrial logistics on the opposing side during the Second World War led to large-scale aerial photography. Archaeologists were recruited to the secret services to help interpret them, and at the end of the conflict they used these resources and knowledge to publish their observations during the 1950s. The Cold War and the Space Race led to satellites being placed in orbit that were able to generate global coverage of the planet, and these coincided with the first publications on multispectral treatment of applied images in archaeology in the early 1970s. The spatial resolution

3.1. Data sources Cartography Our work, based on the premises and methods of an archaeomorphological study above all required a regression analysis and synthesis of the modern and historical cartography of the territory studied. Using this criteria, we compiled the modern topographical maps of the province of Minieh and Beni Suef (scale 1:50K) produced by the Egyptian Survey Authority.

137



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process orthorectification of images requires digital elevation models (DEM). In these cases, it was necessary to use public servers such as CGIAR-CSI, which provides a 90 m/pixel DEM, STRM-90 which has global coverage and the recent ASTER GDEM product ERSDAC with resolutions of 30 m/pixel.

of satellite images and multispectral capacity is now the latest major technique that archaeology that has been used to locate smaller scale structures (Parcak 2009, 1339). Over the years, the publications on remote sensing applied to the Egyptian landscape have focused above all on the areas of the delta, the Middle Nile Valley and the Sinai peninsula (Parcak 2007; Parcak, Mumford 2002), the reconstruction of the course of the Nile at Karnak and Memphis (Hillier et al. 2007; Bunbury et al. 2008; Lutley, Bunbury 2008); the formation of the island of Edfu during the New Kingdom (Bunbury et al. 2009), and the detection of a Canopic channel (Stanley, Jorstad 2006).

Problems with GDEM Using Aster GDEM showed us a series of anomalies that are not visible on high-resolution images or on Google Earth. We initially thought these anomalies were old dikes or large deposits of water. However, the ERSDAC GDEM website contains a recent update which provides information on anomalies in the product similar to those we saw, and emphasises its ‘experimental’ nature, which therefore forced us to discard the hypothesis.

During our project, we carried out a meticulous search and selection of images from various satellites, and endeavoured to include photographs from various times of the year. This latter consideration was in order to increase the probability of detection of traces that probably appear during specific seasons.

Generating our own DEM These problems forced us to seek other solutions. One of these was to create our own DEM using ASTER 3N and 3B bands. These two bands, captured at the same time but with a stereoscopic timelag, use photogrammetry techniques to create stereo images and to extract a DEM with a resolution of 15 m/pixel. Other projects have used 3N and 3B bands to generate a DEM of the area near the city of Zagora, in southern Morocco (Lee et al. 2008).

Most of the images were obtained from the USGS Earth Explorer server available on the Internet. In specific terms, the images were captured by the LANDSAT and CORONA sensors. The ASTER 1B images were purchased from the company ERSDAC (Japan). The CORONA images, which were declassified in 1995 by the Clinton administration, are one of the first series of images on a satellite with global coverage obtained during the Cold War in the period between 1960 and 1975. The KH 4B series provides the best results due to its high quality and the fact that the images were taken at resolutions of around 2 metres/pixel. Their global coverage, especially in Asian and African regions, makes them an essential benchmark in any landscape study (Parcak 2009, 52-57).

Another source used to create the DEM was the images from the CORONA sensor as it simultaneously captures two stereoscopic images (Dashora et al. 2007). Other projects have applied this technique to generate high resolution digital elevation models (Galiatsos et al. 2008). Database and GIS We created a database containing the sites included in the study by Farouk Gomaà, Renate Müller-Wollermann and Wolfgang Schenkel (1991). This database was imported in ArcGis 9.3 with the basic cartography and the satellite images. By superimposing all this information, we created layers matching the traces of canals or dikes we found in the observation and analysis of both the cartography and the images.

In the case that concerns us here, the major point of interest is that it is an extraordinary source of information about the Middle Nile Valley between 1960 and 1976, the period during which the Aswan Dam was constructed. The images show features and traces in the basin flooding system. The landscape captured was still unchanged by the urban and territorial development of the 1980s and 1990s. Its use combined with more modern images from other sensors show the major transformations that have taken place over the last thirty years.

3.2. Spectrum analysis techniques Thanks to the software ERDAS Imagine 2010, it was possible to apply various spectrum analysis techniques to the series of selected images. The different spatial resolutions provided by each type of sensor meant that it was possible to use various approaches depending on the working scale we were adopting.

However, it was necessary to acquire Quickbird (5-022009) and Worldview 2 (23-06-2010) scenes with resolutions of 0.6 and 0.5 m/pixels respectively for a more detailed study of the area closest to the city of Oxyrhynchus. The multispectral features of WV2 (8 bands) were used, applying various types of analysis which will be discussed in the sections below. We also used these images to georeference the Corona images taken in the Oxyrhynchus area.

Combination of bands The multispectral images were composed of bands capturing various regions in the electromagnetic spectrum. Each of these bands acts on natural or geological features in a different way. It is initially only

Furthermore, analysis of the topography of the Middle Nile Valley, and some operations such as 138



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS linear combination of the original bands, enhancing the features of interest in each scene. The difference compared to PCA is that TTC shows components with specific physical characteristics; in other words, regardless of the type of image that is being analysed.

possible to combine three bands in one image, highlighting differences and contrasts such as those resulting from different vegetation growth. For example, it is possible to use LANDSAT images to combine the 43-2 (IR-R-G) bands to produce a new image in which the IR band highlights cases in which there is a greater or lesser density of vegetation, related to greater wetness (Mumford, Parcak 2002; Altaweel 2005)

Kauth and Thomas (1976) created this technique as part of the LACIE project (Large Area Crop Inventory Experiment) developed by NASA and the US department of agriculture (USDA) in the 1970s. The basic variation components in a Landsat MSS image were analysed and their physical characteristics described from the perspective of monitoring crops. The authors of the study made a distinction between three components in the series of images: one called brightness, which is the pondered total of the four original Landsat bands; another called greenness, related to vegetation growth; a third known as yellowness which was used to describe the decline in the vitality of vegetation, and, finally, a fourth, nonsuch, with no apparent meaning. They began to be more widely disseminated in the 1980s, and were applied to other sensors, particularly the LANDSAT TM and the AVHRR (Crist and Cicone, 1984b; Cicone and Metzer 1984). This application led to the detection of a fifth component related to wetness.

Normalized Difference Vegetation Index (NDVI) It is also possible to use an image to work with its constituent bands so the result only highlights specific characteristics recorded within their spectrum. For example, this is the case with the Normalized Difference Vegetation Index (NDVI), which is associated with the health of the vegetation and therefore the existence of walls, trenches or canals beneath sediment. The presence of these structures implies a higher or lower level and decomposition of organic components that retain humidity to a greater or lesser extent, and therefore highlight their existence by means of changes in the density of vegetation (Parcak 2009, 92-94). NDVI is therefore an analytical technique that compares the IR and R bands from multispectral images. We used this technique with images obtained from medium and high resolution multispectral sensors: Landsat, Aster, and WV2. The problem arises when it is applied to areas in which there is a high level of desertification or the presence of dunes. In these circumstances, which apply in the area we studied, we found that it was better to apply another type of analysis, such as PCA or Tasseled Ca

The technique was also applied to high spatial resolution multispectral sensors such as IKONOS, with the territory of Metapontum selected for its application (Horne 2003). Fusion of images One of the problems we found when working with multispectral images was that some of the bands of which they are composed are captured at a different spatial resolution, thereby hindering the combination and treatment of the bands as a whole. For example, it was impossible to combine the SWIR and TIR bands in an ASTER scene with the UNIR bands because the former had resolutions of 60 and 90 m/pixel compared to 30 m/pixel in the latter. In another case, if we applied a combination or PCA to the multispectral bands in a WV2 scene, the resulting resolution would be 1.8 m/pixel instead of 0.5 m/pixel in the panchromatic image.

By using NDVI, we located some canals that had silted up through disuse or which were replaced in the nineteenth century by deep canals designed for permanent irrigation (Merola et al., 2006; Rowlands, Sarris, 2007; Masini, Lasaponara, 2007) Principal Component Analysis (PCA) One way of eliminating the redundancy in the electromagnetic spectrum captured by a scene from a multispectral sensor is to reduce the bands to a series of new components that contain much of the original information. This method, known as Principal Component Analysis (PCA), enables the identification of both the features recorded on most bands and those that are specific to a group of them. The synthesis capacity of PCA makes it a very useful technique for filtering the images as a step prior to other multistation analyses. It is this capacity which is useful in the selection of the most significant information from each period or season studied (Chuvieco, 1996). Its application in archaeology has led to the detection of sites (Stafford et al 1992), buried walls (Garbuzow 2003) and archaeological anomalies on images from the IKONOS sensor.

To solve this problem, we used fusion tools in order to combine a high resolution image (such as WV-2 Panchromatic 0.5 m/pixel) with the other multispectral bands at a lower resolution (1.8 m/pixel). One of the techniques applied was Ehlers Fusion, designed and implemented by Manfred Ehlers at the University of Osnabrück (Ehlers, 2008). However, version 9.3 of ArcGIS includes a feature for online fusion of a multispectral image with a high resolution panchromatic image (Pan-Sharpening), thereby avoiding the need to create large new files.

Tasseled Cap (TTC)

3.3. The Oxyrhynchus landscape: working scales

Tasseled Cap (TTC) is a transformation applied to multispectral images which obtains new bands from the

We used several strategies to study the Oxyrhynchus landscape, applying different working scales depending on the various sources used. 139



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process We considered that before undertaking an archaemorphological study based on traces of division, paths, smaller dikes, and canals, we needed a regression which reconstructed the flood basin system and the changes that took place due to the construction of knowledge irrigation channels such as the Ibrahimiya canal.

In practical terms, the application of TTC to this area has shown that the best results were obtained with a combination of the components 2-3-4 (greenness, yellowness, and wetness) and the components 3-4-6 (yellowness, wetness, and the unidentified component 6).

The study of the basins system in the Middle Nile Valley first required a large scale approach, using written and surveyed nineteenth century sources. We used this approach to reconstruct the various phases and transformations that changed this region in the arrival of Bonaparte's expedition in 1799 and the end of the nineteenth century. The area studied (Figure 4, label 1) is the valley between Samalout and Maghagha, with an approximate total area of 1,304 km2. Using a smaller scale and in the absence of detailed historical cartography for Banhasa and Sandafa el Fahr, the two settlements that have emerged near Oxyrhynchus, we worked on the landscape directly by analysing the satellite images. In this second approach, we organised the study based on three approaches or open windows on the project area. The first (Figure 4, label 2) begins one kilometre to the south of Oxyrhynchus and covers an area of 80 km2 between Bahr-Youssef and the foothills of the Libyan mountains. The second (Figure 4, label 3), a smaller area of 5.25 Km2, is located to the west of the site, and our intention was to analyse the immediate physical area surrounding the Greco-Roman city. Finally, the third area on which we used this approach was the site itself (Figure 4, label 4).

Figure 5 PCA applied on diferent Landsat images. These detected marks and the synthesis and elimination of the inherent redundancy between the various bands in a single image were then superimposed on CORONA, QuickBird and WV-2 high resolution scenes. PCA and TTC techniques were also applied the latter case in order to analyse them in greater detail.

4. Results 4.1 The large-scale approach: The Middle Valley in 1800 The reconstruction of the basin system as it was in 1800 is more interesting. In the absence of documentation about the Mamluk period, we found sufficient items in the Description de l’Egipte to give us some idea of the hydraulic organisation of the Middle Valley. Unfortunately, the data were not compiled directly from the cartography in the Description, which, as mentioned above, is impossible to use due to its errors in the location and representation of dikes and canals. Furthermore, the list of the dikes in the text is not based on any geographical order enabling direct identification. However, a combination of Linant's 1855 plan, which is more precise, and the list of the dikes by the authors of the Description helped to reconstruct the landscape as it was in 1800. There are indeed phonetic variations between the place names used in the two sources and this means that the location of some of the dikes may be open to argument. However, we found no others in Linant's cartography that fit the description better. The list of large dikes includes those that existed in 1800 and which could have structured the flood basins system in the Middle Valley. The Savants' assumption is very clear - a large dike is one that crosses the Middle Valley transversally from the Nile to the Bahr-Youssef. On the 1855 plan, Linant labels dikes that meet that condition, but which were not listed in the Description. These dikes can therefore be included in Mehmet Bey's remodelling

Figure 4 Project area. In all three areas, we first applied TTC and PCA techniques to the images captured by various multispectral sensors (Landsat 4-5 TM, MSS) between 1984 and 1999. The result (Figure 5) showed various traces of paths, wadis, canals, and unidentified features which we analysed using a diachronic process, eliminating those that did not exist prior to 1984 the first date for which scenes are available. At the same time, we recorded the degree of transformation of the rural area near the modern town of Bahnasa. 140

 

Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS programme that took place in the Middle Valley between 1820 and 1855.

are clearly shown next to an important system of regulators.

A comparison of the two sources shows the following place name equivalents:

Safanyeh is another dike named by the Savants (Figure 6, label 5). Linant draws a dike near a settlement labelled as Safaniya which is today the town of Safania. This dike still follows the same route between the Nile and the Bahr-Youssef.

Description de l’Egypte

Linant de Bellefonds

Actual

Behabchyn

Babechine

Haud Bahbasin

Safanyeh

Saffannie

Safaniya

Saftrachin

Safi Rachine

Saft Rachine

el-Noueyreh

Nonewere?

Haud anNuwaira?

There are then three dikes Linant draws on his plan but which do not match any of those mentioned by the French expedition, and must therefore date from after 1800. The first still follows a significant part of its course to the south of the modern town of Muzura, the second must be located near El Fashn, and the third, which we were unable to identify, must be in Ifaqhs. Continuing northwards, we find the sixth dike in the Description (Figure 6, label 6) for which the Savants gave two possible names for reference: Chantour and Badahal. On Linant's plan, the first place there has the same phonetic and is a town today known as Ash Shantour. There is greater variation in the second place name, as Linant labels it as Bedal and we can identify it as the modern town of Badah.

Nuera? Choubak,

Chobak

¿?

Ehoueh,

Elloué

Helwa?

Badahal, Chantour

Bedal, Chantour

Badah, AshShantour

Samalout

Samalout

Samlut

Menbaâl

Minbal

Minbal

Bardanouâh

Bardanou

Bardanuha

The seventh dike (Figure 6, label 7), called Saftrachine or Saft Rachine, retains almost the same name today. To the north is the eighth dike (Figure 6, label 8), Choubak, in a settlement that Linant calls Chobuk. The ninth dike (Figure 6, label 9) is the one that could cause the greatest problems due to its phonetic variation. The Description calls it Noueyreh, while Linant labels a settlement with the name of Nonewere, located between the settlements of Dindil (today Dandil), Bouche Cora (Bush), Ennasi el Medina (Ihanasya el Medina) and Beni Souef. Monitoring of this dike of Noueyreh by remote sensing provided us with data that could date it from prior to 1800. To the south west of Ihanasya el Medina, at the western end of the dike, is the tell of Herakleopolis Magna. The idea that the city had a large dike for holding back the waters that also acted as the main communication system for travellers on foot during the floods is thought-provoking.

Based on the traces found and the previous studies Faruk Gomà, we followed a north-south order for the location of the eleven dikes in the Description. First, we identified the dike known as Samalout in the Savants’ list (Figure 6, label 1). On Linant's cartography, a little further to the north of the old city of DehirSamalout is the dike which we identified as the Samalout named in the Description. The second and third dikes (Figure 6, label 2, 3) are easier to identify as they are either match operating flood basins on Barois' list or place names that still exist: Bardanoah or Berdanou according to Linant (today Bardanuha), and Menbaâl, today the village of Minbâl. On his plan, Linant draws two dikes near these settlements.

It is followed (Figure 6, label 10) by the dike of Behabchyn or Babebchine. Its location is confirmed by the name that Barois (1887, 42, table) uses for the flood basin located immediately to the south: Bahahshin or Bahahehin. The modern place name is Haud Bahbasin.

The fourth dike in the Description is called Ehoueh (Figure 6, label 4). In our opinion, it matches the settlement labelled by Linant as Elloue, which could be what is known today as Hilwa or Helwa. Linant also shows a large transverse dike a short distance from the town.

Finally (Figure 6, label 11), there is the dike of Cocheicha or Oukchechy, about which there is no argument, as it is mentioned by all the sources as being ancient (Barois 1904, 267). Linant includes it on his map and labels it Cocheicha. It is located approximately 6 km to the south of the town of Maidum, running from the Nile between the modern towns of Al Maslub and Bani Ghunaym to the desert at Kom Abou Radi.

The traces detected match two dikes located between Garnus and Tambidi. We found no place names that we were able to assign to any of the dikes listed in the Description. However, on Linant's plan these two dikes 141



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process required for the basins to fill suggest that the system was changed in order to share them with other dikes.

There is a significant gap on our list. The scientists who wrote the Descripton do not include the dike of Garnus on their list. It is surprising that, despite the dike of Garnus meeting the requirement of running from the Nile to the Bahr-Youssef, it is not mentioned in the Description, but is shown on Linant's plan. Based on the criteria of the Savants, we must assume that at least one of them was built between 1820 and 1855.

Figure 7 1800's Bassin inundation system on middle Valley The observation of Barois (1904, 60) is interesting in this respect, as he says that the last basin in a chain is always the longest and widest. The reason for this was that if an unanticipated disproportionate flooding took place or mistakes were made in the emptying manoeuvres, the solution was to create successive breaches in the chain of dikes, thereby relieving the overflow. The last basin bore the excess water as its area was the largest. However, a smaller basin meant that the dikes were subject to less tension and stress from the water stored. They also solve the problem caused by sharp gradients, thereby making the water easy and economical to distribute. According to this argument, the smaller basins of Bardanoah, Menbâal, and Ehoueh, and the larger basin defined by Safaniya most probably by Chantour would form a chain of basins fed by the Bahr-Youssef, the Nile and the various Fyad Baten.

Figure 6 Dikes identified on middle valley. This line of reasoning does not concur with the study by Wolfgang Schenkel (1994) which identified this dike using a text dating from the Roman era. Given their configuration, these dikes came together at a point where the modern town of Tambidi is located, which could suggest an error in appraisal by the Savants, who, it must be remembered, classified the dikes originating in the Nile that were linked to another large dike or with a settlement located on a mound as medium-sized. As a result, taking into account Schenkel's hypothesis, they may have assumed the two dikes to have been three medium-sized dikes. However, Butzer (1976, 12) says that the dikes built prior to the nineteenth century had been used to make the alluvial plains into natural flood basins, which were longer than the nineteenth-century subdivision of the basins. Butzer's opinion is of particular importance given the large area covered by the two continuous basins separated by the dikes of Elloue, Safaniya, and Badahal if our hypothesis is correct. However, assuming that these really were the basins that were operating, the problems arising from the long time

The intermediate nature of the area between these two large basins separated by the dike of Safaniya is interesting. Rough calculations suggest that the distances between the dikes of Elloue and Safaniya and the latter and Chantour are around 26/27 and 27/30 Km respectively. These data could suggest that this was a rational structuring of the space prior to the nineteenth century reorganisation. In this previous design, the 142



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS study, Butzer (1959, 78-79) also analysed the geological sequence of the Middle Valley, identifying two layers of aeolian sand deposited on the Greco-Roman levels. The first, which he called the Upper Younger Dunes, began to be deposited in the early eighteenth century on top of alluvial soils dating from the fifteenth and seventeenth centuries. The second, the Lower Younger Dunes, was formed between 300 and 1450 with a break between 800 and 1200 caused by the accumulation of alluvial soils due to the high levels of flooding in the Nile during this period.

Safaniya dike would have been an important feature in the definition of spaces as it is almost equidistant from the dikes of Elloue and Chantour. Based on this analysis, we agree with Schenkel (1994, 29) who believed that the dike of Safaniya (the seventh dike on our list) dated from before 1800, and he identified it as the Seper-merou in Wilbour's Ramesside papyrus. Using these data, we constructed a final plan (Figure 7) showing the flood basin organisation of the Middle Nile Valley in 1800. In the layout of these basins, one is of particular interest due to its size. We have called it Safanieh as it is the oldest dike and due to the tendency to use the name of the northern enclosing dike for the basin as a whole. On this basis, we included the traces of the canals identified by map interpretation and remote sensing. We saw above how one of them matched the Fyad Baten in the Description, and we assume that the other trace also does so. The westernmost enabled the floodwaters to be distributed in the basins of Samalout, Membaâl, Bardanoah, Ehoueh, and Safanieh, finally overflowing into the Bahr-Youssef; while the second, described by the Savants, enabled the flood waters of the Nile to be distributed directly into the basins of Bardanoah, Ehoueh, and Safanieh, with its water discharged into the latter as the water level fell. Finally, we have shown the old course of the BahrYoussef on the plan, which the Savants mistakenly thought was the real course of the canal. 4.2. A smaller scale study of the Oxyrhynchus landscape: a study in remote sensing The area located to the south of Oxyrhynchus The outstanding geographical feature in this framework begins at a distance of 4.5 kilometres to the south of Oxyrhynchus, between the settlements of Dayr asSanquiriyyha and Hilwah, where the Bahr Youssef forms a large meander (Figures 8 and 10). The composition of bands with Landsat images, NDVI analysis and CORONA images shows that the west bank of the meander is crisscrossed by a series of canals similar to the Nili canals described by Linant. Some are today silted up as they no longer channel and distribute irrigation in the area during flooding. Others were channelled as part of the change that took place in the 1960s with the construction of the Aswan Dam.

Figure 8 Flooding channels (nili) near the meander at south of Oxyrrinchus. 1) CORONA image 2) LANDSAT 4-3-2 bands 3) NDVI analysis. Furthermore, the lateral dike that accompanies the Bahr Youssef is aligned east-west to the north of the meander (Figure 11, label 7). When the Bahr completes its meander, returning to its N-S alignment, the dike no longer accompanies it, but instead continues in a W-E direction until it reaches this series of longitudinal dunes. The enclosure of a basin is thereby completed. At the same level, at the confluence of Bahr-Youssef with one of the flood canals, there is the regulator of El-Miheigra (Figure 11, label 8). After the flood waters had been held back for a sufficient period of time, the water was drained directly into the Bahr-Youssef by opening the regulator. Significantly, from this point onwards there are no large accompanying dikes in the stretch adjacent to the towns of Sandafa and el-Bahnasa.

To the east of the meander, the topography shows a slight longitudinal elevation (Figure 11, label 10) which is a large sandbank or dune, the width of which varies between 300 and 500 metres in a NE-SW direction. According to Butzer (1959,75-76), these dunes are aeolian deposits limited to the western side of the valley, which extend longitudinal and on an irregular basis over the 128 kilometres between Gebel Deshasha (Biba) and the monastery of Deir al-Miharraq (Qusiya). In the same 143



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process part of the reforms that took place in the second half of the nineteenth century for improving the existing flood basins system on the left bank of the Bahr-Youssef, which was operating when Linant de Bellefonds published his plan in 1854. However, if we use Butzer's study as our starting point, which dates the formation of the series of longitudinal dunes from 1700 onwards, this suggests that the basin was not constructed prior to that date.

This series of data suggests that there was a flood basin that linked the Bahr Youssef meander, the possible Nili canals and the western elevation mentioned above, covering an area of 15.9 km2. This flood basin does not match any of the main basins named by Barois or Willcocks. However, as mentioned above, the two authors discuss the flood basin system to the west of the Bahr Youssef without giving them any names.

Other marks on the landscape.

A little further to the west, between this longitudinal elevation and the foothills of the Libyan mountains, there is a series of depressions, with minimum elevations similar to those of this flood basin. When the Corona and Aster images and modern maps are analysed, these depressions present an area with wetland vegetation (Figures 9 and 11).

As mentioned above, the traces detected in the Landsat scenes were superimposed onto high resolution scenes from Quickbird, Crown and WV2. Initially, we found a structure with a function that we were unable to identify, but which could be a possible dike, a channel or a path (Figure 10, label 1 and 12, label 1). This is longitudinal, running SW-NE. We thought that it was these used when the CORONA scene was captured, as a series of breaks or discontinuities in some of the traces is visible. However, when superimposed on the topography, they may also be due to irregularities in the terrain.

Careful observation of the series of images shows a series of canals (Figure 10, label 6) running across the longitudinal elevation, creating a network that provided the series depressions with a water supply. The form of these canals appears to suggest that they are not Nili canals, but instead canals dug at some depth. We are therefore unable to say that it was used as an area for flooding and that it was used to make these lands available for cultivation, as is apparent in more recent images. The supply of water with mud would enable this land to be made fertile.

Figure 12.4 shows a detailed view of a WV2 scene taken in 2010. The remains of the same structure are visible, and it is now very dilapidated and ruined due to the recent construction of a motorway a short distance away. The structure appears to consist of a slope with a road on top, delimited by two possible walls. Its width at its highest point is approximately 3 metres.

Further north, between Banhassa and flood plain of the meander of the Bahr-Youssef, some CORONA images show a longitudinal trace (Figure 11, label 5 and 12, label 6) distinguished by an off-white colour which sets it apart from the surrounding area. Due to its width, we believe that it could be a possible paleochannel of the BahrYoussef. Furthermore, the particular shape of the crop plots in this area defines the course that this canal would have followed (Figure 12, label 7).

Two transversal alignments in the structure are also visible (Figure 10, label 2, 12, label 3, 4). Each one of these in turn consists of two possible walls that are approximately 10 metres wide. They are both aligned SE/NW, and funnel out when they come into contact with the longitudinal structure. The TTC analysis also vaguely reveals a series of linear forms (Figure 12, label 5) with an alignment similar to that of this longitudinal structure. This alignment is different to that of the divisions of the modern fields of crops.

Of particular interest is the width of this paleochannel, which is similar to that of the Bahr-Youssef. To the west of the detail, we can see how a trace emerges to the north in the form of an incision that could be a canal (Figure 11, label 4). However, the image in this area is not clear enough, and more modern scenes from Quickbird and WV2 provide no more information due to the profound changes that have taken place in this area.

Further to the north, we found other traces (Figure 11, label 3) of a possible road documented in the 1908 cartography, linking Bahnasa with another transversal road running along the edge of the Nile Valley on the western side and located further to the west.

These wetlands were perhaps planned in order to relieve surplus flooding. As a result, they would have supplied the depressed areas adjacent to the desert. As well as the presence of trace of the paleochannel, the El-Miheigra regulator, and the lack of large accompanying dikes in the space between the enclosure dike of the meander basin until Benhesue and Sandafa to the north reinforces the idea that the changes to this basin were also aimed at preventing flooding.

The area to the west of Oxyrhynchus In this second study area, we also applied TTC and PCA analysis techniques to Landsat images. Both showed the system of wadis and the network of paths that converge on Oxyrhynchus. The wadis are of particular interest due to component 4 of the TTC analysis, which has a tonality that may be due to higher levels of wetness and therefore higher values of that component.

It is impossible for us to date when this basin was built, as we do not have textual or cartographic information that is specific enough, or a field study. It is possible that it is 144



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS The urban context: some key traces

The confluence of the wadis at Oxyrhynchus could be a good reason to undertake studies on possible occupancy during the predynastic period. This work would involve a geomorphological analysis and examination of the area on the ground, based on the studies carried out in Nag-elQarmila in Upper Egypt (Gatto et al. 2009). These authors believe that the structure of the wadis was used for irrigation during this period. During this period, the flooding of the Nile reached the height of the wadis which were used as reservoirs or reserves of water for irrigation during the rest of the year, after the water subsided.

Combining 8-3-2 World View multispectral Image bands and applying TTC and PCA functions to the Oxyrhynchus archaeological area provides two key pieces of evidence, among others, located elsewhere in the site. The analyses applied to the area in the immediate vicinity of the theatre (Figure 17) showed a quadrangular trace and a N/S alignment located to the north of the site which runs towards it. The combination of the 8-3-2 bands shows a greater tonality of the colour red in the first, indicating better health among the vegetation and therefore an area with high wetness levels. In the TTC analysis of the wetness component, there is again a difference that shapes the quadrangular form.

In this area we first found two marks on the landscape. The first (Figure 13, label 1) is a wadi in the Landsat analyses. It follows an E-W path, continuing beyond the modern quarry until it reaches the area around elBahnasa. The second (Figure 13, label 2) is a path which in the 1980s connected the quarry with a transversal road that ran along the edge of this part of the valley, and was abandoned after the construction of a new road from elBahnasa. We believe that this old path initially runs along a wadi and continues after the transversal road as its traces are still visible, despite being concealed by a field of crops (Figure 13, label 2b). We also found two longitudinal signs (Figure 13, label 2d, 2d’) at an angle of 60º to the wadi from the fields of crops located a little further to the north. Other longitudinal marks can also be seen further to the east (Figure 13, label 2e, 2e’). The four are all connected by a perpendicular fifth with a SW-NE alignment which runs towards the site of Oxyrhynchus.

The second piece of evidence (Figure 18) is an alignment with an approximate width and length of 16 and 258 metres respectively. This would reveal the possible existence of a canal entering the urban area. The tonality of the 8-3-2 combination of the image and component 3 of the PCA shows that there may be a trench or canal that has been silted up, which could contain more humus than the surrounding area, thereby fostering the growth of vegetation.

5. Discussion According to Rippon (2008, 3), the term 'historic landscape analysis' is used to cover a range of approaches focused on resolving the issue of how today's landscape was constructed. This means it is necessary to include an extensive and varied range of materials to ascertain the process or processes that led to changes in the landscape.

A little further to the north (Figure 14), we found another possible wadi aligned towards the site. TTC analysis of the Landsat images shows that it was used starting from the transversal road, and is still used in the same manner today. The same TTC, but applied to the WV2 image, shows in the combination of the 4/3/2 bands the width of the wadi that can be seen by the lighter shades (Figure 14, label 1) and which appears to be divided into two parts one in a straight line towards the north of the site (14.1a) and a second makes a sharp turn to the NW-SE towards el-Bahnasa (Figure 14, label 1b) This analysis is confirmed by verifying the data shown by the TTC with the Corona images (Figure 15).

Based on this assumption, GIS and ICTs are ideal tools given their capacity to store geospatial information and the ease of superimposing and checking various sources from various origins (maps, DEM, remote sensing images, etc.). These varied sources of information can be deconstructed, reconstructed and synthesised to create new information about the landscape. However, in the way we worked with the information, the GIS on its own does not produce any new knowledge as it would do if a Viewshed, a prominence analysis, a flooding simulation or Least Cost Route was applied. In this specific case, it is not a positivist tool which provides quantitative results based on data entry (Chapman 2009, 129). On the contrary, it more closely resembles qualitative confirmation of the digital information with the knowledge acquired on the Middle Valley.

Finally, these wadis converge towards the site of Oxyrhynchus. We can see some degree of relationship between the orientation of the wadis when they reach the outskirts of Oxyrhynchus, and some longitudinal outcrops with a darker colour than the predominant colour of the land (Figure 16). These outcrops, which are generally aligned in a SW-NE direction and appear to be modulated, are located to the NW of the Oxyrhynchus archaeological area, and do not follow the orientation of the urban insulas suggested by the various images analysed.

145



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process An essential consideration is that the research must be 'past oriented' (Bloemers 2002), i.e. we need to understand the processes of change that take place in the landscape. The nature of GIS as a means of storing multitemporal information contributes to regression analysis. However, it is only possible to reinterpret the information based on the researcher's prior knowledge of the ancient landscape. The restoration of the landscape as

it was in 1800 was a process which involved various sources of information. The superposition of historical maps was useful for collecting place names that have undergone a great deal of variation in only 200 years. These enabled us to locate the eleven dikes listed in the Description. These eleven dikes had previously been identified using the CORONA images. However, we first had to ascertain the shape of a dike.

Figure 9Wetland vegetation.

Figure 10 South area. 146



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS

Figure 11 Landscape features identified.

Figure 12 West of Oxyrrinchus archaeological area.

147



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 14 West of Oxyrrinchus archaeological area.

Figure 13 West of Oxyrrinchus archaeological area.

Figure 15 West of Oxyrrinchus archaeological area.

Figure 16 West of Oxyrrinchus archaeological area.

148

 

Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS Under these conditions, applying photogrammetry functions, it is possible to extract a high resolution DEM. If a Quickbird image has a resolution of 0.5 m/pixel, the resolution of the DEM will therefore be similar. It will obviously be necessary to set control points by DGPS, but the number of points necessary will be much lower. In these cases, it is a question of assessing the costs of these high resolution images and comparing them with the time required for the high resolution topographical surveys carried out on the ground. We found a lower resolution (5 metres/pixel) in the CORONA images. This is a solution that reduces costs in the acquisition of the product, but requires a longer processing time. The images were captured with panoramic analogue cameras. This generates very distorted scenes without any georeferencing. It is therefore necessary to pre-process them, in order to subsequently extract the high resolution DEM (5 metres/pixel). In these cases, image analysis and treatment programmes such as ERDAS 2010 and ENVI 4.7 2010 are the tools that we have to use. We are still in a trial phase with the CORONA images, and have generated a DEM at 5 m/pixel, but only in a small working area. In the future we will generate total coverage of our project at high resolution.

The same occurred in the identification of canals. In our case, we found a semi-silted canal between Matay and Abou Girge. If we do not know how the flooding behaves, or that in eighteenth and nineteenth century it was channelled through the Fyad Baten and the Nili, it is impossible to generate new knowledge. Based on this perspective, the GIS is the ICT medium that provides new knowledge, but it is not directly involved in the Thesis-Antithesis-Synthesis sequence at any point. From the other perspective, the observation of a semi-silted up canal without any relationship to other larger, deeper, longer canals running in a straight line, may lead us to contextualise and identify the point at which they were planned or their possible natural source. A second factor should be taken into account. Some of the traces are taken from images or plans dated from periods in which the landscape had not yet undergone the major transformations of the nineteenth and twentieth centuries. These features may therefore not be recognised and directly analysed on the ground today, as the surrounding area was completely transformed. In these cases, selection and analysis in the field is only possible for those that remain in today's landscape. In this case, GIS can provide the solution as its geospatial database contains a diachronic series of images.

The experimental nature of the GDEM product was also noteworthy. According to a validation report written by the institutions participating in the METI/ERSDAC project, NASA/LPDAAC, USGS/EROS (2009, 22-23) two types of anomalies were found in the product. The first takes the form of a series of longitudinal elevated shapes (pits) and the second are elevated circular forms with a depression in the centre (bumps). We saw these anomalies in the work area at the beginning of our project. We were unaware of the existence of the report mentioned above, and interpreted the pits as dikes or canals that were not recognisable in the satellite images. In the second case, the error was more serious as we found the text of an Oxyrhynchus papyrus, the 44.3167(BL 9), which specifically mentioned the use of circular constructions for water storage. These large constructions were allowed to flood during the flooding season. When the water reached the top of the construction the sluices were closed to prevent it from leaving the construction when the water levels fell. As mentioned above, the report emphasised the experimental nature of the product GDEM (METI, NASA, USGS 2009, 27) and as a result we gave up studying shapes in the landscape relief using GDEM.

We believe that the identification of forms and their morphological interpretation is an essential part of our work. As mentioned above, this is because they can tell us about how a landscape was structured at specific points in time, and the nature of the major changes that took place. The data collection methodology used is vital in this respect. Chapman (2009, 91-98) mentions two methodological approaches: the contour survey and the hachure survey. The first, the contour survey, is an objective method as it requires no interpretation when a trench or wall is observed. Some functions of GIS enable the surface to be created in order to visualise and identify topographical irregularities in the area. The use of GPS and the creation of high resolution topographies are a major breakthrough in this area (see for example Sanjuan et al. 2009, 163168). However, in the application of our project, the DEM we used were generated and made freely available on the Internet by various international institutions. However, their low resolution, with a minimum of 30 metres per pixel, prevents an analysis of this type, as it requires data to be acquired on the ground, which is in some cases difficult to obtain. First, the work scales make such an approach impossible to undertake. Our project studies the Middle Nile Valley and we would therefore have to undertake a survey gathering data for the entire valley, which would involve costs in resources and time that are impossible to meet for a team. There are also administrative (Parcak 2009, 186-187) and logistical difficulties hindering such an approach. Other solutions need to be found, such as the acquisition of highresolution stereo pair Quickbird or WorldView scenes.

Other solutions such as LIDAR are applicable in European or American contexts, but are currently impossible in the Middle Valley due to a number of issues which are beyond the scope of this article. Our work was therefore based on cartography and satellite images, and was a qualitative and interpretative study. GIS is part of this approach, and is almost a part of the mechanism for processing the information, and contributes to the deconstruction and reconstruction of new knowledge. We have seen how a flood basin which used the system of side dikes in the Bahr Youssef and the 149



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process of the NDVI index filters the traces that structure the landscape (river, canals, paths, etc.) and therefore helps to identify large divisions quickly.

sandbank located on the western side of the valley was located to the south of Oxyrhynchus. However, there was no direct evidence for the existence of such a structure. It is not shown on any of the images or plans, either in writing or visually. This system was deduced from various indicators (such as the side dikes and the network of flood canals). However, we did not know of any possible western enclosure.

On a local scale, we considered acquiring high resolution scenes, from either Quickbird (0.5 m/pixel) or WorldView 2 (0.6 m/pixel). Our work area is not covered by cartography or orthophotography produced by the appropriate Egyptian institution available for consultation.

By determining the latter, we confirmed the need for precision topography. The modern and older cartography does not provide sufficient information. In this case, the capture of DEMs such as ASTER GDEM and STRM-90 and the creation of our own DEM based on ASTER scenes gave us a detailed topography of the work area. In other words, the resources provided by remote sensing contributed new information that led to continuity in the analysis. The DEM study confirmed that this flood basin was enclosed by the longitudinal elevation of the sandbank identified by Butzer and located on the western side of the Nile Valley.

As mentioned above, remote sensing analysis of the Oxyrhynchus archaeological area provides two details for consideration. The first is a quadrangular structure highlighted by the combination of the NIR band and the TTC wetness component. This shows a structure for accumulating and using water, or as they are generally constructions made of adobe, a series of demolished walls belonging to a large building. Adobe is high in humus content and facilitates the development of vegetation and the accumulation of wetness. We know from the topographical reconstruction by Krüger (1990) based on the Oxyrhynchus papyri that there were large public baths near the theatre. Let us assume that the structure is one of the infrastructures of those public baths.

It is at this point where prior knowledge could lead to error if it is not complemented with other data. Initially, direct viewing of satellite images, and NDVI analysis showed that there was indeed this longitudinal trace of sand which we did not initially identify as a system of dunes. We mistakenly assumed that it could be an old course of the Bahr-Youssef, now part of the desert. However, it was not until we superimposed the DEM and saw the positive increase in elevation that we identified it as the sandbank. Reading Butzer's study enabled us to confirm its aeolian origin, its chronology, and that of the nineteenth-century flood basin. It also gave us very interesting data about the workings of a secondary flood basin, used for local irrigation by one or several communities.

The second structure had a longitudinal shape and was aligned towards the theatre. The trace, detected by the NIR-1 component of WV2 and TTC analysis, also shows a structure dug to a depth that facilitates the growth of vegetation, which is unusual in an area in which vegetation is sparse. In this case we assumed that it was infrastructure constructed to enable the circulation of water. This would control the entry and supply of surplus water caused by the flooding of the Bahr-Youssef at an urban level. Both cases show how the data from the analysis of images provide information for the reconstruction of the urban topography of Oxyrhynchus. In view of the problem with resolving or validating the reconstruction of Kruger, remote sensing analysis provides evidence for where to apply an archaeological examination which would confirm it.

Another issue to take into account is work scales. High and medium resolution satellite images can be adapted to various approaches for landscape analysis that may be local or regional, and these substituted the missing detailed cartography, one of the main problems with this project.

As regards the study of the landscape at a regional level, no work has been done on the ground to date. Our work has strictly speaking been a laboratory study. In theoretical terms, it runs counter to the approach and experimentation with the landscape established by Tilly (1994). We are lacking confirmation on the ground for the canals and dikes, stratigraphic drilling, geomorphological studies, and paleoenvironmental restoration. They will all be new GIS layers.

On a regional scale, analysis of the flood basins followed a top-down process (Rippon 2008, 26). This consists of identifying the large divisions that fit a consistent description. In this case, the large divisions are determined by the flood basins and dikes. Various systems of land division were structured around them, and changed annually after the flooding since the Pharaonic era. The remote sensing analysis techniques applied (PCA and TTC) showed traces of paths, wadis, canals, and other features. These techniques enabled some information to be highlighted, and other redundant information to be suppressed. However, they cover the regional work scale when medium resolution images are applied. The solution involving working with Landsat and Aster images is therefore the most appropriate. For example, application

6. Conclusions Regression analysis of the sources and cartography enabled us to reconstruct the dikes that were operational when the French expeditionary forces arrived in 1799. Our study also clarified how the flooding was channelled and distributed by the Fyad Baten and the Nili canals 150



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS before being channelled or added to sections of other new canals such as the Ibrahimiya as part of the programme of nineteenth century transformations. We have seen how these can still be found on cartographic plans and satellite images despite the years that have passed since the construction of the Aswan Dam.

However, we must also consider the possibility of finding ourselves in a similar situation to that identified by Lutley and Bunbury (2008) further to the north. According to the study undertaken by their team, these canals are really the result of the Nile's migration across the valley's entire width. These variations have been calculated for the Memphis area and are in the region of 9 km for every thousand years. When the seasonal flooding occurs, these paleochannels or abandoned courses regained their volume and naturally fed the constructed flood basins. In the area that concerns us here, we should also add the variability in the course of the Bahr-Youssef. The latter possibility is reinforced by the fact that the two sloping plains from the Bahr-Youssef and the Nile formed basins or low-lying beds - Linant's Bathen. This layout would encourage migration of the two courses, leading to the formation of abandoned courses or paleochannels. Future geomorphological studies will have to resolve these questions and establish chronologies that enable the landscape to be reconstructed in greater detail.

Figure 17 Theatre area.

However, the Muslim invasion and the Mamluk government occurred between the two points in time. This period is too long for there to have been no changes and modifications to the preceding system. If the eighteenth and nineteenth century sources talk about the decline and neglect of the infrastructures, then they must have been maintained at least until that point in time. The restoration of the basins system in 1800 allows us to define two large basins that were interlinked and separated by the Safaniya dike, which was identified by Schenkel as the Seper-merou mentioned in the Wilbour Ramesside papyrus. It was possibly a very large system of basins, which were perhaps subdivided by means of smaller dikes. These basins would have been defined by the enclosure dikes located to the north and to the south (Ehoueh and Chantour) which were constructed almost equidistant from the Safaniya dike. Our analysis on a larger scale enabled us to determine that continuity and intensity in archaeomorphological studies to re-establish the systems of division prior to the arrival of French expedition must focus on the eastern banks of the Middle Valley, and above all on the mohit dike created after the construction of the Ibrahimiya Canal. This does not mean that the western bank should be ignored, but the main changes and alterations took place there, and as a result the introduction of sugar cane and cotton accompanied by perennial irrigation must have changed the pre-existing systems of division.

Figure 18 Archaeological area. 1) Wordview 2 NIR2/4/3 bands. 2) CORONA. Considering the magnitude of the transformations that took place during the nineteenth century (perennial irrigation, partition of existing basins, changes in the flooding, irrigation channel and drainage system) one might assume that the reconstruction of the flood basins that took place in 1800 was similar to the system that existed in the Greco-Roman period. During the flooding, these basins were fed by a network of traces from the Bahr Baten, or the low-lying beds that we found. It remains to be ascertained whether these inner canals or rivers are of natural or anthropic origin, although Jomard stated that he had not found any remains of hydraulic infrastructures (Jomard 1809, 105, note 1)

Meanwhile, our work based on remote sensing in the Oxyrhynchus area has enabled us to reconstruct the system of wadi which was the cornerstone of the GrecoRoman population structure and we have shown the presence of a possible displaced paleochannel to the west of the Bahr-Youssef.

151

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process imagery: archaeological applications from the Near East, Antiquity 82, 732-749.

The routes of nili canals on the western bank of the Bahr and low level points and enclosure dikes show that a flood basin was built to the south of el-Bahnasa in around the mid-nineteenth century, which enhanced an area affected by the sands and dunes of aeolian origin that had accumulated since the eighteenth century.

Chapman, H. 2009. Landcape archaeology and GIS, Gloucestershire.

Detected traces of the canals and dikes nili have helped us also to reconstruct the irrigation system during the Greco-Roman period (Subias, Fiz, Cuesta in press)

Cicone, R. and Metzler, M. 1984. Comparison of landsat MSS, Nimbus-7 CZCS, and NOAA-7 AVHRR features for land use analysis, Remote Sensing of Enviroment 14, 257-265.

Finally, TTC and PCA analysis provided us with information on the hydraulic features and structures in the Oxyrhynchus archaeological area. These are explained in Subias (in press).

Crist, E. P., Cicone, R. C. 1984b. Application of the tesseled cap concept to simulated thematic mapper data, Photogrametric Engineering of Remote Sensing 50, 343352.

Remote sensing remains a technique that is based on distance, taking advantage of the extensive spatial coverage of the sensors used. However, this distance cannot be the only method that is used. This study, which was carried out on various working scales, will not be complete until the routes located have been confirmed on the ground, and accompanied by exploration in the territory. This work must be complemented by geomorphological and paleoenvironmental perspectives that complete the overview of the landscape of the Oxyrhynchus area.

Crist, E. P., Cicone, R. C. 1984c. A physically-based transformation of thematic mapper data the TM tasseled cap, IEEE transactions on geoscience and remote sensing GE22, 256-263. Dashora, A, Lohani, B., Malik, J. 2007. A repository of earth resource information CORONA satellite programme, Current Science 92 (7), 926-932. Ehlers, M. 2008. Multi-image fusion in Remote Sensing: spatial enhancement vs. spectral characteristics preservation, in ISVC '08 proceedings of the 4th international symposium on advances in visual computing, Part II, 75-84.

References Altaweel, M. 2005. The use of ASTER satellite imagery in archaeological contexts, Archaeological Prospection 12, 151–166.

Galiatsatos, N., Donoghue, D. N. M., Philip, G. 2008. High resolution elevation data derived from stereoscopic CORONA imagery with minimal ground control: an approach using IKONOS and SRTM data, Photogrammetric Engineering and Remote Sensing 74 (9), 1093-1106.

Bloemers, J. H. F 2002. Past and future-oriented archaeology: protecting and developing the archaeological landscape in the Netherlands, in Fairclough. G, and Rippon. S (eds) Europe's cultural landscape: archaeologist and the management of change. Brussels, Europae Archaeologiae Consilium, 89-96.

Gatto et al. 2009. Landscape reconstrucion of the Predynastic site at Nag-el-Qarmila (Upper Egypt), Sahara Journal 20, 63-69.

Bunbury, J. M, Graham, A, Hunter, M. A 2008. Stratigraphic landscape analysis: charting the Holocene movements of the Nile at Karnak through ancient Egyptian time. Geoarchaeology: An International Journal, 23 (3), 351–373.

Gomaà, F., Müller-Wollermann, R., Schenkel, W. 1991. Mittelägypten zwischen Samalut und dem Gabal Abu Sir. Beiträge zur historichen topographie des Pharaonischen Zeit, Wiesbaden.

Bunbury, J. M, Graham, A, Strutt, K. D 2009. Kom elFarahy: a New Kingdom island in an evolving Edfu floodplain, British Museum Studies in Ancient Egypt and Sudan 14, 1–23.

Goossens, R., De Wulf, A., Bourgeois, J., Gheyle, W., Whillems, T. 2006. Satellite imagery and archaeology: the example of CORONA in the Altai Mountain. Journal of Archaeological Science 33, 745-755.

Butzer, K. 1959. Some recent geological deposits in the Egyptian Nile Valley, The Geographical Journal, 125 (1), 75-79.

Hillier, J. K., Bunbury, J. M., Graham, A. 2007. Monuments on a migrating Nile, Journal of Archaeological Science 34, 1011-1015.

Butzer, K. 1976. Early hidraulic civilization in Egypt. Chicago.

Horne, J., 2003. A tasseled cap transformation for IKONOS images, in Proceedings of the ASPRS: 2003 annual conference and technology exhibition. Kauth, R. J., Thomas, G. S. 1976. The tasseled cap. A graphic description of the spectral-temporal development

Casana, J., Cothren, J. 2008. Stereo analysis, DEM extraction and orthorectification of CORONA satellite

152



Fiz, Subias, Cuesta: The Landscape of Oxyrhynchus Using Textual Sources, Cartography, Remote Sensing and GIS Parcak, S. 2009. Satellite archaeology, New York.

of agricultural crops as seen by Landsat, in Proceedings of the symposium on machine processing of remotely sensed data, Purdue, 41-51.

sensing

for

Rippon, S. 2008. Historic landcape analysis. Deciphering the countryside, York.

Krüger, J. 1990. Oxyrhynchus in der Kaiserzeit. Studien zur topographie und literaturrezeption, Frankfurt am Main/New York.

Rowlands, A. Sarris, A. 2007. Detection of exposed and subsurface archaeological remains using multi-sensor remote sensing, Journal of Archaeological Science 34 795-803.

Lasaponara, R., Masini, N. 2007. Improving satellite quickbird-based identification of landscape archaeological features through principal component analysis and tasseled cap, in Georgopoulos A. (ed.), AntiCIPAting the future of the cultural past. Proceedings of the transformation in XXI CIPA symposium-Athens, Greece, 1 October - 6 October 2007.

San Juan, L. Wheatley, D. Murrieta, P. F., Marquez, J. 2009. Los SIG y el análisis espacial en arqueología. aplicaciones en la prehistoria reciente del sur de España, in Nieto, X. and Cau, M. A. (eds), Arqueología Naútica Mediterrànea, Monografies del CASC 8, Girona, 163180.

Lee, J., Hans, S., Lee, S., Parl, J. 2008. Correcting DEM extracted from Aster stereo Images by combining cartographic DEM. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. XXXVII. B1, Beijing, 829-834.

Shenkel, W. 1994. Les systèmes d’irrigation dans l’Égipte ancienne et leur genèse, Archéo-Nil, Mai-Juin, 27-35. Subias, E. in press. Oxyrhynchos: metrópolis y paisaje, in The Graeco-roman space of the city in Egypt: image and reality, International Workshop held at Tarragona, Tarragona.

Levau, P. 2000. Le paysage aux époques historiques. Un document archéologique, Annales HSS 3, 555-582. Lutley, K., Bunbury, J. M. 2008. The Nile on the move. Egyptian Archaeology, 3-5.

Subias, E., Fiz, I., Cuesta, R. in press. Elementos del paisaje del nomo oxirrinquita, in The Graeco-roman space of the city in Egypt: image and reality, International Workshop held at Tarragona, Tarragona.

Masini, N., Lasaponara, R. 2007. Investigating the spectral capability of QuickBird data to detect archaeological remains buried under vegetated and not vegetated areas, Journal of Cultural Heritage 8, 53-60.

Stanley, J. D., Jorstad, T. F. 2006. Short contribution: buried canopic channel identified near Egypt’s Nile Delta coast with Radar (SRTM) Imagery, Geoarchaeology: An International Journal 21 (5), 503–514.

Merola, P., Allegrini, A., Fuglietta, D., Sampieri, S. 2006. Using vegetation indices to study archaeological areas, British Archaeological Reports International Series 1568, 303-310, Oxford, BAR Publishing.

Stafford, J. D., Leigh, D., Asch, D. 1992. Prehistoric settlement and landscape change on alluvial fans in the Upper Mississipi River valley, Geoarchaeology 7 (4), 287-314.

METI/ERSDAC, NASA/LPDAAC, USGS/EROS 2009. ASTER Global DEM validation. Summary Report Available at: http://www.ersdac.or.jp/GDEM/E/image/ ASTERGDEM_ValidationSummaryReport_Ver1.pdf.

Tilley, C. 1994. A phenomenology of landscape. Oxford.

Muller-Wollerman, R. 1993. Ein deich in Oxyrrinquites, Zeitschrift für papyrologie und epigraphik 96, 123-124.

XVIII - XX century sources Barois, J. 1887. Irrigation in Egypt. Washington.

Mumford, G., Parcak, S. 2002. Satellite image analysis and archaeological fieldwork in El-Markha Plain (South Sinai), Antiquity, 766, 953-4.

Barois, J. 1904. Les irrigations en Egypte. Paris D’ anville 1766. Memoires sur L’Egypte ancienne et moderne suivis d’une description du Golfe Arabique ou de la mer Rouge. Paris.

Palet, J. M. 1997. Estudi territorial del Pla de Barcelona: estructuració i evolución del territorio entre l’època iberoromana i l’altmedieval. Segles II-I aC – X-XI dC, in Estudis i memòries d’Arqueologia de Barcelona, I. Barcelona.

De Maillet 1740. Description de l'Egipte. La Haya. Girard, M. S. 1813. Observations sur la vallée d'Egypte et sur l'exahussement séculaire du sol qui la recouvre, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 9 : Histoire Naturalle, 343-404.

Parcak, S. 2007. Satellite remote sensing methods for monitoring archaeological tells in the Middle East, Journal of Field Archaeology 32, 65-81.

153



remote

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process depuis la plus haute antiquité jusqu'à nos jours: accompagné d'un atlas renfermant neuf planches grand in-folio imprimées en couleur, Paris.

Girard, M. S. 1823. Memoire sur l'Agriculture, l'industrie et le commerce de l'Egypte, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 6 : L’Etat Moderne, 491-700.

Linant de Bellefonds, L. M. A. 1882. Carte hidrographique de la Moyenne Egypte. Gravée au depot de la Guerre en 1854 d’apres les travaux de M. Linant de Bellefonds. Revue et completée en 1882 pour les chemins de Fer. Echelle 1/250000, Paris.

Jacotin, 1826. Carte topographique de l’Égypte et de plusieurs parties limitrophes, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 6 : Atlas Geographique.

Martin 1813. Description hydrographique des provinces de Beny-Soueyf et du Fayoum, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 6 : Ètat Moderne II, 194-228.

Jomard, E. 1809. Memoires sur le Lac Moeris compare au Lac de Fayoum, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 2: Antiquités Memoires I, 84-114.

Martin 1822. Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 7: État Moderne I.

Jomard, E. 1818. Description des Antiquités de la Heptanomide, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 2: Antiquités Descriptions II, 219-294.

Norden, L. F. 1755. Voyage d’Egipte et de Nubie. Copenhague. Pockocke, R. 1972. Voyages de Richard Pockocke. En orient, dans l’Egypte, l’Arabie, la Palestine, la Grèce, la Trace, etc., IV, Paris.

Le Père J. M. 1809. Mèmoire sur la communication des Indes a la Mediterranée par la Mar Rouge et l'Isthme de Soueys, in Description de l'Egypte ou Recueil des observations et des recherches qui ont été faites en Égypte pendant l'expédition de l'armée française, 5 : État moderne I, 21-186.

Savary, M. 1787. Letters on Egypt, London. Willcocks, W., Craig, I. 1913. Egyptian irrigation, New York.

Linant de Bellefonds, L. M. A. 1872. Memoires sur les principaux travaux d'utilité publiqueé éxecutés en Egypte

154



10 Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy Andrew Dufton1 and Corisande Fenwick2 1

Joukowsky Institute for Archaeology and the Ancient World Brown University 2

Department of Anthropology, Stanford University

Abstract There is much new and exciting theoretical research in funerary archaeology, revealing the potential of archaeological evidence for understanding the responses, attitudes and practices surrounding burial. Yet in practice, there is often a disjunction between archaeological and anthropological data, theories about burial, field methodologies, and postexcavation analysis. Web-based technologies offer a means to integrate a variety of data types and disseminate this information to facilitate a finer-grained analysis of burial and funerary practices. We examine the implications of these new techniques through a detailed case-study of the excavations of the medieval cemetery at Villamagna, Italy (20062010). At Villamagna, we used an integrated, reflexive and multivocal approach to manage and record anthropological, osteological, archaeological, and topographical data. Collating plans, images, finds, and specialist data enables a more holistic understanding of individuals from the funerary record by combining evidence of identity as signaled after death with evidence for identity during life. The Villamagna case study illustrates the potential that new technologies, specifically web applications, have for the integration of data from excavations and the production of theoretically nuanced analyses. Keywords: Funerary Archaeology; Methodology; Medieval Archaeology; Italy. ___________________________________________________________________________________________ that are time-consuming and difficult to excavate and analyse. This paper argues that the peculiar complexities of medieval cemetery stratigraphy have created a methodological quandary for medieval archaeologists about how to record and analyse the diverse types of information produced during a cemetery excavation.

1. Introduction The last two decades have produced a spate of increasingly sophisticated social research in funerary archaeology (e.g. Morris 1992; Parker Pearson 1999; Tarlow 1999; Williams 2006). This work underscores the huge potential of archaeological evidence for understanding the responses, attitudes and practices surrounding burial, and the ways in which mortuary practices can construct and express social identities. Scholars of the medieval period have equally emphasised that the medieval dead were a distinct social group who continued to have a significant presence among the living (Binski 1996; Geary 1994). The funerary sphere was a key arena of political, religious and social debate in the middle ages (Ariès 1974). Individuation, commemoration, social status and corporate identity were closely associated and, for much of the medieval period, the grave and funerary rituals played a key role in promoting these concepts.

We also argue that this methodological quandary has a methodological solution. New web-based technologies offer a more effective means to record and integrate data from excavations and laboratory analyses. Paired with existing spatial technologies they also provide a tool for conducting complex spatial and chronological analyses, thus facilitating theoretically informed micro and macro scale analyses of burial and funerary practices. This paper illustrates the potential of these tools for the problemoriented excavation and theoretically informed analysis of medieval cemeteries through a detailed case-study of the cemetery excavations, and how they unfolded, at Villamagna, Italy.

2. The state of the problem

Yet, arguably this research has had limited impact when it comes to excavation and the publication of medieval cemeteries in Italy. Modern scholarly assumptions about the nature of medieval burial practices are perhaps in part responsible for this state of affairs. However, there is also a pragmatic explanation: medieval Christian cemeteries typically consist of deeply stratified, intercutting burials

In a recent book on the medieval monastic cemetery in Britain, Roberta Gilchrist and Barney Sloane (2005, 1) argued that progress in interpreting medieval burial has been hampered by three key factors: modern assumptions about medieval burial practice; the absence of broader

155



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process theoretical perspectives on burials; and methodological limitations encountered in dealing with complex evidence. The main purpose of this discussion is to address the latter of these factors. Through the demonstration of some of the methodological problems faced by archaeologists dealing with the complexities of medieval burial practice in the field, we suggest that it is these practical difficulties that have been the critical limiting factor to more theoretically informed analyses of medieval funerary evidence.

The failure to recognise the potential of funerary data may stem in part from the relative isolation of medieval archaeologists from contemporary theoretical debates in funerary archaeology until recently. Several scholars have argued that the lack of engagement with broader theories of death and burial on the part of medieval archaeologists has resulted in the elevation of description over interpretation, and the bolstering of historical paradigms (e.g. Gilchrist and Sloane 2005; Williams 2003, 228). This stands in stark contrast with archaeological studies of death and burial in prehistory and historical periods (e.g. Parker Pearson 1999; Morris 1992; Tarlow 1999).

Traditionally, the study of medieval burial practices focused on the artifact-rich cemeteries of the early medieval period (5th – 7th centuries). By contrast, burials in succeeding centuries were typically thought to be undifferentiated in terms of status, gender and age, a development largely attributed to the conversion to Christianity (see Lucy and Reynolds 2002; Gilchrist and Sloane 2005). Such opinions about the undifferentiated nature of the majority of medieval burials are no longer widely held, but have resulted in an imbalance in research efforts.

Recently, however, medieval archaeologists have been drawing on broader theoretical perspectives from anthropology, history, and prehistoric archaeology. Medieval graves are now seen not as static, self-contained entities, but as a complex set of interactions between the living and dead. Scholars have used funerary evidence to examine questions about social memory, gender, ethnic identity, consumption practices, social structure, and ideology (e.g. Barbiera 2005; Effros 2003; Halsall 2003; Gilchrist 1994; Gilchrist and Sloane 2005; Lucy and Reynolds 2002; Williams 2006). These more theoretical approaches tend to derive from analyses of national or regional patterns. Whilst this work has been very productive, the detail of individual cemeteries and their micro-social relevance often becomes lost in these comparative studies. Another weakness of this research, shared with funerary archaeology more generally, is the relative lack of attention paid to the biological individuals in question, and a failure to integrate anthropological data (Gowlands and Knüsel 2006, ix). Yet skeletal evidence can shed light on questions of social identity, particularly the social construction of age and gender (Sofaer 2006), as well as lifestyle, diet, culture, environment and population movement (Larsen 1997).

In Italy, although a large number of medieval cemeteries have been excavated, scholarly attention has focussed on Goth or Lombard burials, where grave goods are plentiful, and little has been published about other types of medieval burial in the same or later periods (e.g. Brogiolo and Cantino Wataghin 1998). Sauro Gelichi (1997, 157) has argued that this grave-good centred approach ‘has marginalized the study of cemeteries which postdate the 7th century and, at the same time, has funneled into a single stream research on the centuries of the barbarian migrations’.1 Certainly, very little has been published about burials from the numerous non-Goth/ Lombard cemeteries exposed during the course of excavations of churches or monasteries. The location of graves is rarely marked on published plans, and they are often barely mentioned in reports aside from important finds in their graves. Even in the case of the well-studied Gothic and Lombard burials, scholars have only recently begun to pay attention to the form and structure of the tomb, the arrangement and treatment of the body, and cemetery topography. Moreover, anthropological and palaeopathological studies are rarely integrated into the synthetic discussions of cemeteries (see Fornaciari and Mallegni 1981; Ginatempo 1988 for still applicable critiques). These biases have resulted in a dearth of information about medieval cemeteries in Italy. Despite the huge wealth of information we have about churches, we know almost nothing about medieval mortuary practices, the location of cemeteries in relation to churches and monasteries, the division between spaces for lay-people and clergy/ religious orders, etc. (Gelichi 1997).

These theoretical developments, however, have only had limited impact on excavation practices and the publication of cemetery excavations. This is not a problem unique to medieval archaeology, though it is aggravated by the complexities of medieval cemeteries and the difficulties that archaeologists face in excavating multiple inter-cutting burials with no clear limits. There are some very real barriers to the problem-oriented and theoretically informed analysis of funerary remains. Whilst the last few decades have seen the standardisation of archaeological recording methodologies, including for human remains, these changes have also created a significant gap between data and interpretation in site practices. Restricting the contributions of the excavator to a series of prompted, pro-forma fields designed to facilitate efficient, quality recording often also has the unintended effect of denying the interpretive voice of the excavator and presenting a univocal account of the excavation process (Berggren and Hodder 2003; Lucas 2001). This criticism is particularly valid in the singlecontext recording system, initially conceived by the Museum of London's Archaeological Site Manual but also advocated for Italian recording by Andrea Carandini in Storie dalla terra (Spence 1993, Carandini 1991).

 1

‘ha confinato ai margini lo studio dei cimiteri posteriori al VII secolo, e, nel contempo, ha convogliato in maniera unidirezoniale le ricerche anche per i secoli delle migrazioni barbariche’. 156



Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy problems at the Villamagna project – both through interpretation at the trowel’s edge and as we now write up the cemetery for final publication.

Although the pro-forma sheets adopted in both singlecontext systems include a prompt for the interpretation of the excavator, this information is usually included at the time of initial excavation and is based almost exclusively on stratigraphic characteristics. There is little provision for amending or improving this interpretation after specialist consideration of associated finds, or as the excavation progresses. Furthermore, the limited space on paper records can constrain attempts at multivocality. For example, the addition of new or differing interpretations usually results in the deletion of the initial text, and thus the loss of the excavator’s original interpretation of a context. Finally, the sheer bulk of documentation generated in most excavations often means that the majority of primary records for excavated contexts are not revisited in post-excavation analysis. Standardisation in methodology has also come with an increasing compartmentalization of specialisms on many projects, with archaeologists tending to excavate human remains in the field and physical anthropologists or osteoarchaeologists analysing them in the laboratory, often after the excavation has been completed. Excavators therefore tend to work in a relative vacuum, distant from the anthropologists, who can provide information which might be useful in the ongoing process of excavation and interpretation. Equally, anthropologists often work in a relative vacuum without sufficient knowledge of contextual relationships or mortuary practices (Hodder 1997).

Figure 1 The location of Villamagna in the wider landscape.

3. Villamagna: project background

When it comes to publication, excavation monographs of cemeteries usually focus on chronology and typology and rarely consider in detail social questions. This trend is not restricted to medieval archaeology, but is characteristic of cemetery publications more generally. As Gilchrist and Sloane (2005, 13) argue, excavation reports are beset by some unnecessary problems. Most burials are not integrated within the structural sequence. It is usually difficult to work out what artefactual evidence may have been recovered from the graves, leading to omission or mis-dating. Different sexes or age categories or the distribution of finds are rarely displayed clearly. Frequently there is no plan with the burial/ skeleton/ grave cut numbers on it. Anthropological studies of the human remains and catalogues of the grave goods are usually confined to specialist catalogues at the end of the report, and rarely integrated into the interpretation. As Henri Duday complains, ‘If one evaluates the respective place of the diverse elements that make up a burial as a function of the number of written lines an author devotes to them in a publication, one often has the unfortunate impression that the deceased had been placed as an offering to a ceramic vessel or to a flint projectile point, rather than the other way around’ (Duday 2006, xx).

The site of Villamagna is in southern Lazio and lies just south of Anagni, some 40 miles south of Rome, Italy (Figure 1). The site was known as Villa Magna in antiquity. The emperor Hadrian probably built the villa in the second decade of the 2nd century AD. A famous pair of letters from Marcus Aurelius to his tutor, Fronto, describe two days spent there with his adopted father hunting, watching the pressing of the grapes and finally dining with the emperor in the wine-pressing room (Fronto, Epistolae 4.5). A document of the 10th century refers to the presence of a monastery called S. Pietro in Villamagna (Flascassovitti 1994), and the medieval church still stands among the ruins of the villa. The first systematic excavations at the site were conducted between 2006-2010 by an international team under the direction of Elizabeth Fentress, and under the auspices of the British School at Rome, the University of Pennsylvania and the Archaeological Superintendency of Lazio. The first author of this paper was in charge of the digital aspects of the entire project, while the second author was responsible for the excavations of the medieval cemetery. The project excavated part of the 2nd-3rd century imperial villa, including the winery and a bath house; a barracks complex dating from the late 3rd through the 6th century; the medieval church and cemetery dating from the 5th to 15th centuries; and some huts from the medieval village dating from the 9th to 13th centuries. Further information about the project can be found on the project website (www.villa-magna.org). Interim reports have been published in Fasti Online Documents and Research (Fentress et al. 2006; 2007;

There are reasons for this lack of integration in field practice and cemetery publications: lack of conversation between specialists, time constraints between excavation and publication and, in particular, the absence of suitable tools to aid interpretation. In the remainder of this paper, we will outline the ways in which open source and popular commercial programs have helped address these 157



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process 2008; 2009; 2010) and in Fentress and Maiuro forthcoming. A final excavation monograph is currently in preparation.

‘…responds to what is found and tries to use a wide range of techniques to make sense of what is found. It tries to surround the excavator in greater knowledge. It is more rigorous because starting assumptions and things taken for granted are always being reevaluated and open to critique, especially as different partners in the process bring their different perspectives’ (Berggren and Hodder 2003, 431).

The medieval excavations, directed by Caroline Goodson (Birkbeck College, University of London), focused on the area around the still-standing church and revealed a complex history of building and cemetery use after the abandonment of the villa (Figure 2). In the 6th century, a church with a narthex was built in the middle of the paved courtyard of the former imperial villa on the foundations of an earlier apsed building in brick, possibly an earlier church. In the 10th century, a monastery was built to the north of the church, while a village lay some 400m to the south, near the remains of the old winery. The cemetery expanded and burials in communal brick and tile tombs have been found in front of the church and in the monastery. In the 12th to 13th centuries, the monastery cloisters were built, along with a bell tower and a new monumental porch. An ossuary pit was built, presumably to house the bones disturbed by this construction. The area in front of the church was given over to a massive cemetery containing some 350 simple earthen graves, a few burials were found in the cloisters. At the end of the 13th century the monastery was suppressed, the monks excommunicated, and fortifications were built around the church. Burials were initially confined to the area outside the fortifications and to a chapel inside the church, though some perhaps later burials have been found inside the fortified area.2

A reflexive archaeology therefore speaks directly to some of the issues specific to cemetery excavation raised earlier in the paper: it implies less of a separation between data and interpretation; more integration through excavation and post-excavation; and better incorporation of different sources of information. Although there has been much discussion of the need for reflexive archaeology in excavation methodology (see Andrews et al. 2000; Bradley 2003; Chadwick 2003; Hodder 1997; Lucas 2001), to date there is a dearth of recording tools designed to help facilitate this reflexivity. As a result, technological or practical limitations have too often prevented effective practice. At Villamagna we used an integrated database and GIS system known as the Archaeological Recording Kit, or ARK (ark.lparchaeology.com). ARK is an open source, web-based system for the collection and dissemination of archaeological data. In conceptual terms, ARK serves as a loosely structured development framework allowing the recording of archaeological or non-archaeological data. In technical terms, the database relies on a PHP web interface serving a MySQL database. Javascript, Mapserver and OpenLayers code is also used in the integration of images and spatial data, respectively (Eve and Hunt 2008).3 Using a hybrid of the Entity-Attribute-Value (EAV) data model and a more traditional relational database, the system does not rely on a single concept of archaeological recording such as a context, locus, bucket or stratigraphic unit. Instead, individual data fragments such as strings of text, dates, images or actions are grouped under a primary record identifier and displayed in a web-based user interface (Figure 3). This fragmentary approach to archaeological data can also be extended to include geospatial information such as plans, point locations, or topo-surfaces. By relying on a fragmentary EAV data model, ARK also allows the inclusion of multiple fragments of the same type to a single record. This facilitates a degree of multivocality in the excavation record not always possible in pro-forma recording systems, with a variety of team members adding their voices to emerging interpretations in ARK without replacing the preliminary ideas of the excavator.

Figure 2 Plan showing the key features of the medieval monastery.

4. At the trowel’s edge As outlined above, one of the factors constraining a theoretically informed analysis and interpretation of medieval cemeteries is the recording systems employed in the field. The complexity of burial practice during this period necessitates a holistic and reflexive understanding of the excavated remains not only during the analysis, but also in the field throughout the excavation process. A reflexive archaeology:

 3

Similar database systems are being developed by the Tracing Networks Project at the University of Leicester (http://www.tracingnetworks.ac.uk/content/web/collabora tive_system.jsp) and by the Centre for Digital Archaeology at Berkeley for the Last House on the Hill Project (http://www.archaeovault.org/lhoth/), but are still at beta-stage.

 2

It should be noted that dating is still provisional and that a suite of radiocarbon dates is currently being processed by the Oxford Radiocarbon Accelerator Unit, Oxford. 158



Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy The implementation of ARK at Villamagna (http://www.villa-magna.org/ark), while following concepts and recording data common to archaeological excavation, was therefore designed to meet specific recording needs unique to the project and site. Excavations were conducted following standard techniques of open area excavation and using the single context recording system. The site's primary record forms were entered into ARK on a daily basis throughout fieldwork. Site photographs were also added regularly and linked within the system to the relevant stratigraphic units. Specialist data was added as processing was completed, including a basic inventory of ceramics and associated spot-dating and catalogues of objects, coins, architectural elements, and glass. These disparate types of records were collated within ARK and presented together, facilitating a wholesale understanding of the ongoing excavation as well as individual stratigraphic units. Acknowledging the importance of the spatial component of the archaeological record in understanding the site, the creation and processing of site plans was also fully integrated with other on-site recording. Hand-drawn plans were scanned, georeferenced and digitised concurrently throughout the excavation. This digital spatial record was then augmented with targeted survey by total station of selected structural stratigraphic units or other significant features. Digital plans of features were held in a dedicated geodatabase linked dynamically to records in ARK.

ARK to take into account new information. The ability to include multiple interpretations, with associated metadata, within the data structure also offered a means of tracking these emerging interpretations across excavation seasons. Having all the excavation data in a unified and easily accessible database facilitated a more reflexive interpretive process than may have otherwise been possible in a research excavation with limited time and resources. In the second season of excavation, it became increasingly apparent that the cemetery area in front of the church was much larger and more complex than initially assumed (Figure 4). On average over 100 skeletons were removed during each subsequent field season, eventually totalling 1457 contexts and 395 burials in trench BI alone, an area only 14.20 metres by 12 metres excavated to a maximum depth of 1.5 metres (Figure 5). One might be tempted to dismiss the complexity of the Villamagna cemetery as an extreme case, but it is wholly typical of medieval cemeteries, and this, we argue, is partly responsible for our poor understanding of their development. As such, the cemetery became an ideal opportunity for both applying reflexive archaeological techniques in the field with the aid of an integrated digital record and conducting the theoretically informed and problem oriented excavation of a medieval cemetery in Italy. Building on our understanding of Christian medieval burial practices, and the settlement history of the site, we formulated five overlapping research agendas which we hoped the excavations would provide information about: 1. The changing role of the cemetery within the medieval landscape and its relationship with other structures. 2. Social identity (gender, age and status). 3. Burial rites (how the corpse was prepared, grave types, and body arrangement). 4. Cemetery organization and management (rows and zoning, grave visibility, grave disturbances). 5. Demography (disease, mortality, stature, diet etc.). Our research questions shaped our research design and our field methods. We wanted to be able to integrate information about the individual (age, sex, health, diet, corpse arrangement) with information about the grave (type, orientation, associated finds), as well as to record the stratigraphic and spatial location of each burial. A recording system therefore had to be developed which would enable the integration of archaeological and anthropological methodologies in the field and laboratory.

Figure 3 A screenshot of integrated HRU records held within the ARK database. Rather than replace the traditional hand-drawn and written recording methods ingrained in archaeological excavation practice, the ARK database was designed as a tool to complement these basic methods. The integration of the drawn and written paper records with digital photography and specialist data created a combined digital record able to inform emerging interpretations of the archaeology of the site. The database was hosted online on the project website, and could be accessed and edited by specialists in the off-season. To promote a reflexive approach to field practice, site supervisors were encouraged to revisit the data from past field seasons and to update interpretations and stratigraphic relationships in

We felt that the field recording system that we initially employed for human remains, similar to that of the Museum of London system, did not record all the grave information available in the field. Physical anthropological methodologies have evolved markedly in recent years, and scholars are becoming increasingly interested in the taphonomic processes by which skeletons and their graves decompose. The 159



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process ‘anthropologie de terrain’ approach pioneered by Henri Duday and his students uses taphonomic analysis to reconstruct funerary practices performed for the dead (e.g. Duday et al. 1990). This approach is commonly used in France and Italy, though in recent years a number of Anglophone scholars have drawn attention to its potential (e.g. Nilsson 1997; Willis and Tyles 2009). Analysis allows a detailed reconstruction of mortuary practices, including whether the deposit was a primary or secondary burial, or in the case of multiple interments whether the bodies were deposited simultaneously or in sequence. It can establish whether the body was wrapped in a shroud or placed in a coffin or under a cover, where offerings, ornaments and wrappings/ clothing were initially placed on the body, and whether other organic elements such as pillows were present in the grave at the time of burial. It also can reveal if the body was manipulated in any way before, during and/or after the process of decomposition (Duday 2009).

context sheet by an archaeologist. For this reason, from the second year, we established a training programme at Villamagna to provide archaeologists and anthropologists with the necessary skills to do both. The decision to use the ‘anthropologie de terrain’ method meant that the in-field documentation had to be revised over the duration of the project. In 2006, a standard Human Remains Unit Sheet was used (Figure 6), which recorded only basic information about the skeleton, and no taphonomic data. In the second and third years, we introduced an Italian anthropological recording sheet as well as a slightly revised Human Remains Unit Sheet, causing unnecessary duplication of data and translation issues. In later years, a bilingual recording sheet that integrated the archaeological and anthropological information was used (Figure 7).

Figure 4 Evidence of an emerging complexity of burial practice.

Figure 6 The original Human Remains Unit pro-forma recording. The resulting Human Remains Unit (HRU) context sheet recorded general information (Trench, HRU number, plans), archaeological data (stratigraphic information, grave typology, location, orientation, type of deposition), associated finds (type, location in respect to the body) as well as anthropological data (age, sex and anatomic connection) and taphonomy (corpse arrangement and post-burial activities). Excavation was viewed as an intellectual as well as a technical process. In the field, those digging had to think through the taphonomic processes by which graves and corpses become part of the archaeological landscape, and to interpret how the body had been prepared for burial. Interpretation therefore was a routine part of primary recording.

Figure 5 The whole picture: a plan of all skeletons recorded in the main medieval cemetery area. The ‘anthropologie de terrain’ approach uses in-field observations to provide detailed information about human burial practices, particularly those regarding preparation of the corpse, and interment. It also has the potential to bridge gaps between physical anthropologists and field archaeologists and create a more nuanced dialogue about a particular cemetery. In practice, however, divisions are often reinforced in the field: anthropological and taphonomic data tends to be recorded on a separate sheet by an anthropologist, and stratigraphic data recorded on a 160



Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy However, the implementation of a reflexive field practice at Villamagna, and indeed any attempts at changing recording practice mid-excavation, are not without consequences. The decision to amend our recording methodology in the cemetery has some implications for the integrity of our data set. Fundamentally, in-field taphonomic analysis was not conducted on the 28 skeletons excavated in 2006. Additionally some of these skeletons were not digitised to the standards later adopted. Fortunately, however, some key indicators of taphonomic processes can be successfully identified from hi-resolution photographs of excavated skeletons (see Nilsson 1997; Willis and Tyles 2009), and it has been possible to ascertain the types of graves that these individuals were buried in.

mandate of the ARK system, the project was able to create a broader digital tool to help analyse the social aspects of medieval burial.

5. From site to synthesis A reflexive archaeology involves an ongoing to-and-fro between initial hypotheses and data, with room both for evolving interpretations to shape data collection and for collected data to influence the interpretive process. It is important that this interaction is not limited purely to the time spent on site. As archaeological material collected on site is segmented for further study by a series of disconnected, often international specialists there is a risk that practical limitations put an end to reflexive practices. The integrated online ARK system served as an integral hub for the project in moving from site excavation to the specialist analysis, synthesis and publication both of the medieval cemetery and the wider site as a whole.

In addition to the above changes, from the beginning of the second excavation season skeletons were planned digitally from photographs, targets and Electronic Distance Measure (EDM) plots saving much time in the field (Figure 8). This spatial information was then paired with the HRU record, site photographs and summary specialist anthropological metrics (age, sex, pathology) in ARK.4 Daily digitising and data-entering rotas meant that ARK was kept up-to-date and allowed us to contextualise burials as they were excavated. It was thus possible to build up a picture of chronological and spatial patterns that were emerging in the cemetery.

Following standard practice, we gathered related contexts into groups of related and contemporaneous activity. Thus three contexts (grave cut, skeleton, backfill) become a single group (burial). As individual contexts can possess different attributes: A grave cut can have shape and size and can contain multiple burials. A skeleton can have position, age, sex, pathology etc.

In this way, our tool did not simply encourage and enable reflexive practice in the field, it also forced us to confront the question of how best to analyse the mass of data produced in a medieval cemetery excavation. Initially, we intended to use ARK primarily as a database: a place to store and manage data in an accessible format, and a supplement to the primary drawn and photographic record of our excavation. As the number of skeletons multiplied at an alarming rate, we began to consider the potential of our tool for not simply managing data, but for interpreting it and analysing the social aspects of cemetery development, burial practice and the lives of past individuals.

A grave lining can have extent, construction material. A grave fill can have associated finds etc. A grave wall can have orientation, extent, construction material. The group acts as a means of linking together all of these different attributes and permits their close analysis as constituent components of a burial. The database provided a means of creating tomb groups through linking contexts and was developed to allow analysis of burials through a consideration of the various traits these constituent parts may possess.

We wanted to use the excavation report not simply to publish the excavated data, but as an opportunity to analyse data in relation to our five research themes. This had to be done using the tools already in place, thus mitigating the costs and time-issues that make this type of analysis prohibitive for most projects excavating complex medieval cemeteries. By moving beyond the original

Complex medieval cemeteries pose a number of specific problems in post-excavation analysis, as the usual tools of archaeological interpretation are often not that helpful. For example, the density of medieval cemetery use means it is often difficult to accurately determine stratigraphy and the limits of graves. Similarly, ceramics are usually residual and thus provide poor indicators of chronological shifts. For these reasons, it is essential to integrate vertical (i.e. chronological) with horizontal (i.e. spatial) patterning to answer questions of a social nature.

 4

Study of the osteological material was undertaken by a team led by Dr. Francesca Candilio of the University of Rome, with the participation of graduate students in anthropology from the University of Pennsylvania. They have established the age, sex, measurements and major pathologies of each individual. A further project has involved the stable isotope analysis of the remains as part of the doctoral research of Erika Nitsch at Oxford University. The results of these studies have been integrated into ARK.

Our understanding of the history of the founding of the monastery and later church made it reasonable to suppose that we might detect archaeological differences in the ways that burial was conducted over time as the church’s function shifted.

161



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 7 An amended Human Remains Unit including both archaeological and anthropological data.

Figure 8 An example of the integrated workflow for skeleton digitisation. For example, were these changes represented in how bodies were prepared for burial, the construction of the grave, and the performance of commemorative rites after death? The integration of demographic data was essential to evaluate the pattern of the social groups and identities of people buried at Villamagna. What was the relationship between the cemetery and the other parts of the built landscape? Could we detect the existence of ‘lay’ and ‘monastic’ zones in the monastic period? Could we identify differences in social status in different burial areas? What about the health of individuals buried in different contexts?

User-friendly tools commonly used by research and commercial archaeologists alike provide an opportunity to answer these questions. Most projects will use some form of database to record contextual information. They will also use a GIS to record spatially the (x, y, z) location of individual contexts, including skeletons. Basic information about burials – age, sex, grave type, body position, grave goods, pathologies – can be extrapolated from this in the form of a comma-separated values file and imported into any standard GIS package. This information can then be linked to a context – whether a grave fill or skeleton – and a series of queries can be run to discover vertical and horizontal patterning in the archaeological record. 162



Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy To facilitate the interpretation of the material at Villamagna, elevation data recorded by EDM during cemetery excavation was combined with the HRU record and spatial information held within ARK. This information was then manipulated within ArcGIS to create a series of 2.5-dimensional images highlighting possible spatial patterning of graves. Given the integrated nature of the excavation record, all elements of the burials can be considered alone or in combination: the presence/absence of grave goods; stratigraphic relationships and phasing; grave typology; and anthropological or osteological classifications.

A quick consideration of this distribution shows a concentration of burials directly facing the church entrance, and a certain predominance of male burials in this area. In comparison, the burials of children are seemingly relegated to the periphery of the cemetery to the north and south of the entrance. This is particularly evident in a series of infant burials in the far southern area. Expanding on this basic distribution, a more complex question was posed: what is the pattern between the sex distribution and the presence/absence of possible status objects like rings and buckles? To consider this problem a plan was created highlighting the sex distribution of only those burials with recorded grave goods (Figure 10). As above, the most striking observation is the location of the majority of these burials in the area immediately in front of the church entrance. Graves associated with finds during this phase are also predominantly male or child burials, with only three recorded female burials containing grave goods. In combination with the sex distribution discussed above, an early hypothesis can be formulated suggesting a preference for burial in front of the church for the burials of males and high status individuals.

As a first attempt at making use of this inclusive record a very simple query was asked: is there any pattern to the distribution of male/ female and child burials? The massive cemetery in front of the church, in use from the 10th to 13th centuries, served as an appropriate target for analysis as the intercutting earthen burials typical of this phase present some of the greatest challenges to further interpretation. The resulting plan highlights the distribution of male, female, child and indeterminate earthen burials from the main cemetery area (The sex categories employed by the project are Male, Male?, Indeterminate, Female?, Female. Child/ juvenile burials cannot be sexed, and so are assigned to a different category) (Figure 9).

Figure 9 A plan of the sex distribution of the earthen cemetery.

163

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process

Figure 10 A plan of both grave sex and the distribution of grave goods within earthen burials.

Figure 11 An interpretive snapshot using recorded elevations to highlight possible phases of cemetery development. The height of the buildings is exaggerated so as to position the burials

164

 

Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy web-based technologies have for the integration of data from excavations and the production of theoretically nuanced analyses. Tools like these, which are not difficult to implement or use, can therefore offer huge potential for field projects to address broader social and theoretical questions in excavation publications.

Focussing on the most heavily populated and possibly preferential status area of the earthen cemetery, immediately to the front of the church entrance, the burials from this period were grouped according to a series of standard deviations of recorded elevation (Figure 11). By manipulating and grouping the elevation distribution of skeletons originally recorded in only two dimensions, and displaying this information in a 2.5 dimensional figure, a pattern of possibly discrete grave alignments with matching depths begins to emerge. These distinct phases can thus be incorporated both in the stratigraphic analysis of the earthen cemetery and in the nascent interpretation of burial practice at the site.

Acknowledgements This paper derives from a much larger collaborative project at Villamagna, and we would like to thank Elizabeth Fentress and Caroline Goodson for graciously allowing us to discuss the cemetery before final publication, and for their comments on earlier drafts. We are very grateful to Francesca Candilio for providing us with the anthropological data, and for helping develop the cemetery methodology. Corisande Fenwick would like to thank the cemetery team - Claudia Asper, Luciano Bruni, Samantha Cox, Karen Heslin, Erica Nitsch - and all those who participated in the ‘skellie rota’ for forcing her to be reflexive in the field. Andrew Dufton would like to thank all those that helped survey and digitise the skeletons as part of the GIS rota, particularly Karen Heslin.

These examples illustrate only the early stages of interpretation of the medieval cemetery and by no means represent a final narrative of burial practice at Villamagna. In keeping with the reflexive method adapted throughout the project, the results of further study on the osteological material and grave goods will be integrated into emerging interpretations throughout post-excavation, resulting in evolving hypotheses and analysis. The primary point of this paper, rather than attempting a quantitative analysis of the material so often endorsed by GIS and database applications, is to highlight the potential for such tools as an effective aid to a more qualitative discussion and to bridge the gap between data management and the interpretive process so common in archaeological projects.

References Andrews, G. A., Barrett J. C. Lewis J. S. C. 2000. Interpretation not record: the practice of archaeology. Antiquity 74, 525-530.

As an additional benefit to using web-enabled technologies throughout this process, the extensive digital record of the project also serves as a valuable component of the online dissemination of the excavation results. The data will continue to be hosted online after the completion of the project for future reference. The online companion to the excavation publication also incorporates aspects of the digital record. Through the process of transclusion, in this case inclusion of live data from ARK within elements of the online publication, the interaction between data and written text remains dynamic and bidirectional. Such an approach uses the digital record as an essential link between the primary record and the final text and establishes the excavation data of Villamagna as a valuable and accessible tool for future research.

Ariès, P. 1974 (translated by M. Ranum). Western attitudes towards death: from the Middle Ages to the present. Baltimore, The John Hopkins University Press. Barbiera, I. 2005. Changing lands in changing memories. migration and identity during the Lombard invasions. Florence, Edizioni all’Insegna del Giglio. Binski, 1996. Medieval death: ritual and representation. London, British Museum Press. Berggren, A., and Hodder, I. 2003. Social practice, method, and some problems of field archaeology. American Antiquity 68, 3, 421-35. Bradley, R. 2003. Seeing things: perception, experience and the constraints of excavation. Journal of Social Archaeology 3 (2), 151-168.

6. Conclusion

Brogiolo, G. P, and Cantino Wataghin, G. (eds) 1998 Sepolture tra IV e VIII secolo, atti del 7° seminario sul tardo antico e l’alto Medioevo in Italia centro settentrionale (Gardone Riviera, 24-26 ottobre 1996), Mantova.

To conclude, we hope that we have demonstrated the need for an integrated approach to medieval cemeteries, and the potential that commonly used and user-friendly tools offer for recording, visualizing and interpreting data. At Villamagna, we used an integrated, reflexive and multivocal approach to manage and record anthropological, osteological, archaeological and topographical data. This has enabled us to create a more holistic understanding of individuals from the funerary record by combining evidence of identity as signalled after death in mortuary practices with evidence for identity during life from osteological analysis. The Villamagna case study illustrates the potential that new

Carandini, A. 1991. Storie dalla terra. Manuale di scavo archaeologico. Torino, Einaudi. Chadwick, A. 2003. Post-processualism, professionalization and archaeological methodologies. Towards reflective and radical practice. Archaeological Dialogues 10 (1), 97-117.

165



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Gerrard, C. 2003. Medieval archaeology: understanding traditions and contemporary approaches. London, Routledge.

Duday, H., Courtaud, Crubézy, E., Sellier, P. and Tillier, A. M. 1990. L’anthropologie “de terrain”: reconnaissance et interpretation des gestes funéraires. Bulletins et Mémoires de la Société d’Anthropologie de Paris n.s. 34.

Giantempo, M. 1988 Corpi e uomini tra scienza e storia: studi di osteo-archeologia umana per l’Italia medievale. Archeologia Medievale 15, 7-64.

Duday, H. 2006 (translated by C. J. Knu sel). L’archéothanatologie ou l’archéologie de la mort (archaeothanatology or the archaeology of death), in R. L. Gowland and C. J. Knu sel (eds.) Social archaeology of funerary remains, 30 - 56. Oxford, Oxbow Books.

Gilchrist, R. and Sloane, B. 2005. Requiem: The Medieval monastic cemetery in Britain. London, Museum of London Archaeological Services. Halsall, G. 2003. Burial writes: graves, texts and time in Early Merovingian Northern Gaul, in J. Jarnut and M. Wemhoff (eds) Erinnerungskultur im bestattungsritual, 61–74. Munich, Wilhelm Fink.

Duday, H. 2009. The archaeology of the dead: lectures in archaethanatology. Oxford, Oxbow Books. Effros, B. 2003. Merovingian mortuary archaeology and the making of the early Middle Ages. Berkeley, University of California Press.

Hodder, I. 1997. Always momentary, fluid and flexible: towards methodology. Antiquity 71, 691-700.

Eve, S. and Hunt, G. 2008. ARK: A developmental framework for archaeological recording, in A. Posluschnya, K. Lambers and I. Herzong (eds). Layers of perception. Proceedings of the 35th international conference on computer applications and quantitative methods in archaeology (CAA), Berlin, Germany, April 2–6, 2007. Bonn, Rudolf Habelt.

Hodder, I. 2003. Archaeological reflexivity and the ‘local’ voice. Anthropological Quarterly 76 (1), 55-69. Knüsel, C. 2010. Bioarchaeology: a synthetic approach/bioarchéologie: une approche synthétique. Bulletins et mémoires de la société d'anthropologie de Paris, 22, 1-2, 62-73.

Fentress, E., Goodson, C., Hay, S., Kuttner, A. and Maiuro, M. 2006. Excavations at Villa Magna 2006. FOLD&R 68. Available at: www.fastionline.org/docs/ FOLDER-it-2006-68.pdf.

Larsen, C. S. 1997. Bioarchaeology: interpreting behavior from the human skeleton. Cambridge, Cambridge University Press Lucas, G. 2001. Critical approaches to fieldwork. Contemporary and historical archaeological practice. London, Routledge

Fentress, E., Fenwick, C., Goodson, C., Kuttner, A., and Maiuro, M. 2007. Excavations at Villa Magna 2007. FOLD&R 96. Available at: www.fastionline.org/docs/ FOLDER-it-2007-96.pdf.

Lucy, S. 2002. Burial practice in early medieval eastern England: constructing local identities, deconstructing ethnicity, in Lucy, S. and Reynolds, A. (eds) Burial in Early Medieval England and Wales, 72-87. (Society for Medieval Archaeology Monograph 17). London, Maney.

Booms, D., Candilio, F., DiMiceli, A., Fentress, E., Fenwick, C., Goodson, C., McNamee, M., Privitera, S., Ricciardi, R. 2008. Excavations at Villa Magna 2008. FOLD&R 126. Available at: www.fastionline.org/docs/ FOLDER-it-2008-126.pdf.

Lucy, S. and Reynolds, A. 2002 Burial in Early Medieval England and Wales: past, present and future, in Lucy, S. and Reynolds, A. (eds), Burial in Early Medieval England and Wales, 1-23. (Society for Medieval Archaeology Monograph 17). London, Maney.

Fentress, E., Goodson, C., and Maiuro, M. 2009. Excavations at Villa Magna 2009. FOLD&R 169 Available at: www.fastionline.org/docs/FOLDER-it-2009 -169.pdf.

Morris, I. 1992. Death-ritual and social structure in classical antiquity. Cambridge, Cambridge University Press.

Fentress, E., Goodson, C. and Maiuro, M. 2010. Excavations at Villa Magna 2010. FOLD&R 204. Available at: www.fastionline.org/docs/FOLDER-it-2010 -204.pdf.

Nilsson, L. 1998. Dynamic Cadavers: a fieldanthropological analysis of the Skateholm II burials. Lund Archaeological Review 4, 5-17.

Fentress, E. and Maiuro, M. 2011. Villa Magna near Anagnia: the emperor, his winery, and the wine of Signia Journal of Roman Archaeology 24. 333-369.

Parker Pearson, M. 1999. The archaeology of death and burial. Stroud, Sutton Publishing.

Flascassovitti, C. 1994. Le pergamene del Monastero di S. Pietro di Villamagna (976-1237). Galatina.

Sofaer, J. The body as material culture. A theoretical osteoarchaeology. Cambridge, Cambridge University Press.

Geary, J. 1994. Living with the dead in the Middle Ages. Ithaca, Cornell University Press.

Spence, C. 1993 (ed.). Archaeological site manual, 3rd Edition. London, Museum of London.

Gelichi, S. 1997. Introduzione all'archeologia medievale. Rome, La Nuova Italia Scientifica.

Tarlow, S. 1999. Bereavement and commemoration: an archaeology of mortality. Oxford, Blackwells.

166



Dufton & Fenwick: Beyond the Grave: Developing New Tools for Medieval Cemetery Analysis at Villamagna, Italy Williams, H. 2003. Remembering and forgetting the Medieval dead, in Williams, H. (ed.) Archaeologies of remembrance. Death and memory in past societies, 227254. New York, Kluwer/Plenum.

Williams, H. 2006. Death and memory in Early Medieval Britain. Cambridge, Cambridge University Press. Willis, A. and Tyles, N. 2009. Field anthropology: application to burial contexts in Southeast Asia. Journal of Archaeological Science 36, (2) 547-554.

167



11 Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites Elaine Massung University of Bristol Abstract Interpretation at archaeological and heritage sites can take many forms: signs and labels, guidebooks, leaflets, audio guides, tour guides, and touchscreen kiosks. One of the most recent additions to these methods has been the utilisation of the visitor's own device, such as a mobile phone or MP3 player, and thereby providing interpretation through downloadable files (e.g. podcasts). Adding a further layer to this form of interpretation is the employment of locationsensing technology (e.g. GPS, Wi-Fi, RFID tags, etc.) to create a ‘mediascape’, a virtual world of sights and sounds that is triggered automatically and displayed on a PDA or smartphone. Although this offers enormous potential for sharing information about the past, prior research has typically concentrated on the capabilities of the technology, rather than exploring whether visitors want to access interpretation in such a manner, and in turn, how to create a positive experience. As a result, an investigation was undertaken to examine how this novel technology can be used to its full potential. Through an analysis of the quantitative and qualitative data obtained in the visitor evaluations, a framework of factors termed the Seven Cs—Context, Clarity, Choice, Control(s), Cost, Contact, and Content—has been developed to assist in the creation of location-based interpretation. Keywords: Location-based; Interpretation; Mediascape; Mobile Phone; Smartphone; Apps. ________________________________________________________________________________________________ The technology operates through the use of a contextaware device, such as a mobile phone enabled with a Global Positioning System (GPS) receiver or the infrared sensor on an audio guide, and automatically provides content when the user visits a pre-programmed area of interest. By knowing the user’s location, site specific content can be developed that offers far greater freedom and flexibility than hitherto seen in heritage interpretation. However, the intent of this research was not to endorse a specific method of utilising this technology, but rather to explore the opportunities and limitations offered by location-based media.

1. Introduction Through interpretation – understanding; through understanding – appreciation; through appreciation – protection. (Tilden 1977, 38) An anonymous National Park Service official penned these words over fifty years ago, and his sentiments remain as pertinent today as they did then. Although the importance of providing high-quality interpretation has not diminished, the methods used to deliver such content have evolved over time, from tour guides and guidebooks to audio wands and computer kiosks, and the ubiquitous nature of technology is likely to lead to a paradigm shift within the discipline of heritage interpretation.

A number of sites have already trialled location-based technology as a method of providing interpretation: the gardens of Ashton Court and Stourhead served as the testing grounds for a proprietary GPS system; the BBC television programme Coast established visitor trails in coastal towns by using Quick Response (QR) codes; and the World Heritage Site of Blaenavon investigated the use of GPS-enabled personal digital assistants (PDAs). However, none of these methods have become permanent.

The rapidly developing field of location-based media is one such method that has the potential to usher in a new era in the presentation of interpretive content. By allowing the public to access this information anywhere, at any time, it is hoped the technology will make the past more accessible and enable a more nuanced view of the historical palimpsest that underlies most sites, whilst also providing visitors with a far greater degree of personalisation than available at present.

Such a situation illustrates that although the tools necessary to make this a viable technology are in existence, there is still a missing component: do visitors 168



Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites background of the technology utilised; (4) the survey methodology employed; (5) an overview of results from evaluations at the Roman Baths; (6) design of the prototype guides; (7) results from the Clifton Suspension Bridge; and (8) the Seven Cs framework that developed as a result of audience testing.

themselves want to use such technology? If so, what features are necessary to provide users with an enjoyable experience that will enable location-based interpretation to move beyond an experimental novelty? In order to answer these questions, a two-prong approach was taken, consisting of visitor evaluations and the creation of a prototype system. The first of these evaluations was carried out at the Roman Baths Museum to explore how visitors rate various aspects of existing audio guides; this is a traditional form of interpretation that is typically emulated in the delivery of content through location-based means.

2. Background: Interpretation Common dichotomies underlie the creation of interpretation, and this section begins by examining several such aspects that form the foundation of modern interpretive theory. There is no one method of interpreting the past that is ‘best’, but rather each facet can aid in guiding a visitor’s understanding of the past.

The results obtained from this study were then utilised in the development of content and the design of the prototype guides used at the Clifton Suspension Bridge; the resulting data was then compared along a spectrum of interpretive methods to determine whether any visitor preference is evident.

Approaches to interpretation: Positivistic versus constructionist Tim Copeland (2004, 2006) identifies two methods in which interpretation can be presented: (1) positivistic, in which visitors are provided with ‘facts’ and a description of the site as it was in the past; and (2) constructivist, which seeks to be challenging, allowing visitors to make their own meanings. The former is employed at the Roman Baths and presented in the prototype guides at the Suspension Bridge; this is more of a didactic approach and provides visitors with one way of interpreting the available evidence (Pearce 1993, 239). The latter, however, seeks to provide visitors with primary sources whenever possible and allows them to draw their own conclusions, i.e. visitors interpret the site in their own way, with the method of interpretation instead serving as a device to question visitors, encourage thought, and foster discussion (Copeland 2006).

Research aims As mentioned above, the overarching aims of this research were to determine whether visitors to archaeological and heritage sites have an interest in accessing interpretation using location-based technology, and what aspects of delivering this content must be taken into account if it is to be successful. The practicalities of using location-based media to present archaeological interpretation were investigated through a combination of quantitative and qualitative methods, allowing the author to seek answers to the following questions: x

How do visitors interact with both the device and their environment?

x

How much interpretation should be provided, e.g. what is the ideal length of audio or text segments?

x

The use of a display screen should enrich the experience and not distract from the artefacts or site on display; how should images, video clips, and maps best be utilised?

x

What factors are necessary to create a positive user experience?

Although it is the constructivist approach–’selfinterpretation’–that Copeland favours for interpretation at archaeological and heritage sites (2006, 83), he identifies that it ‘might engender resistance from more traditional visitors’ (Copeland 2006, 94). Indeed, comments regarding a desire for ‘just the facts’ were seen during the study at the Roman Baths, in opposition to the opinionbased Bryson tour. The key similarity between these approaches is that it is the interpretative designer who selects which method to follow. Likewise, most methods of interpretation rely on only one such approach in order to present the past, yet through the development of digital delivery systems it will be possible to provide interpretation in the manner that best suits the visitor, to give him or her the choice as to which method is preferred. Over two decades ago, N. James (1986, 47) hinted at these possibilities by commenting, ‘there could be as many interpretations of history—in effect, as many pasts—as there are social interests.’

Bellotti et al. (2002, 37) cut to the heart of the matter, concluding that the design of future guides must join interpretation with ‘an understanding of human psychology when developing mobile multimedia tools.’ According to O’Hara and Kindberg (2008, 256), it is very rare for visitors themselves to be queried as to their preferences. This research sought to rectify this through the use of audience evaluations and observations on working prototypes. This paper is divided into multiple sections, each examining a different aspect of research: (2) the factors of interpretation considered within this study; (3) a 169



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process The champion of constructionist interpretation, Tim Copeland, favours this latter method of presenting the past, claiming (2006, 92): ‘If the linear route is dispensed with, and the ‘big concepts’ used as a structure, then it should be possible for visitors to explore more widely.’ From a purely practical standpoint, this encouragement for exploration takes advantage of decades worth of research into the ways visitors behave in museums (e.g. Robinson 1928; Melton 1935, 1972), and which has shown that visitors often impose their own route upon an exhibition or site, despite the best efforts of interpretive planners (Falk and Dierking 1992, 69; Bitgood 2006).

Approaches to interpretation: Education versus entertainment In Presenting the Past (2003, 10), Larry Zimmermann makes the grand claim that, ‘For most of the public, however, the primary use of archaeology is for entertainment.’ Freeman Tilden (1977, 29) adds that ‘. . . we cannot forget that people are with us mainly seeking enjoyment, not instruction.’ Taking the opposite tack, Richard Prentice (1991, 298) claims that, ‘These enhanced interpretative provisions are primarily designed to inform, and sometimes to entertain, visitors.’

However, location-based media does away with the necessity of following either the ‘compartmentalised’ or the ‘big concepts’ method, and instead ties interpretation into the very location a visitor is standing, creating a more immediate, more intimate view of the past.

Regardless of the author’s belief in the purpose of interpretation, statements such as these seem to imply that education and entertainment are mutually exclusive, that visitors seek out only one or the other when visiting a museum or heritage site. Yet this all or nothing approach seems outdated, especially in light of the development of ‘edutainment’, a term that crept into the lexicon in 1983 with regards to the convergence of education and entertainment in the sector of software development and game-based learning (Oxford English Dictionary; Addis 2005, 730), and which has since permeated other fields, e.g. archaeologist and television presenter Mark Horton (pers. comm., September 2010) relates how the BBC has made ‘edutainment’ a driving force behind their programming.

Approaches to interpretation: multivocality A recent buzzword to appear in the field of heritage interpretation, multivocality seeks to give voice to those who are otherwise silent in the interpretation of the archaeological record: women, children, the poor, the disabled (Hodder 2008, 196). The greater storage capacity available in digital interpretation allows visitors to easily access such stories, and there has been a trend in interpretive design to have these alternative narratives available. For example, the children’s audio tour at the Roman Baths attempts to achieve this through the use of characters, and the ss Great Britain uses separate audio tours to provide different perspectives about the past, e.g. relating the experience of a First Class passenger versus that of a person in Third Class steerage. Multiple time periods can also be provided for; after all, every epoch and culture makes its own material impact on the landscape (Jones 1993, 18).

Much of it boils down to semantics; a study by Barbara Birney (1986; as cited in Falk and Dierking 1992, 14) found that the public tends to associate the idea of ‘learning’ with that which occurs in a formal setting, such as a classroom. But when the definition is broadened to include ‘the results of curiosity’ and the ‘urge to explore’, learning becomes a major motivation for visiting a museum. James Arnold (1993, 229), writing about the industrial heritage of the site of New Lanark, Scotland, seeks to unite these two disparate approaches: ‘Our visitors come to us to have an enjoyable day out, and if we can educate them a little, as well as confirming their enjoyment, we feel we have been enormously successful...’

3. Background: Technology Interpretation should occur where the site best illustrates the point being made (Arnold 1993, 226), and therefore, strictly speaking, location-based media is not new: traditional audio tours, such as those at the Roman Baths discussed in Section 5, rely on knowing a user’s location within a site in order to impart interpretation. Even notech methods such as signs are placed where visitors can best benefit from them (as Humphries (2006, 72) describes, typically with the use of a ‘you are here’ indicator to provide orientation), and leaflets commonly employ maps with highlighted areas of interest to guide the visitor.

Approaches to interpretation: Linear versus thematic In telling the story of the past, the very nature of archaeology lends itself to a linear narrative: first this happened, then that, illustrated with a straightforward timeline. According to Morris (1988, 72), ‘Devised in the context of museum display, this way of compartmentalising the past has proved difficult to avoid...’ However, in recent years archaeologists have begun to move towards a more holistic approach, examining the relationship between individual elements and the landscape itself (Bath 2006, 163), with such an approach allowing for different ways of providing interpretation, typically through the use of themes (e.g. Hill 1993) or concepts.

However, in these instances the method of interpretation is passive, relying on the user to enter a location code in the case of audio guides, or ‘access’ the interpretation through reading. New technology allows the device to play an active role in the part of interpretation, not only automatically sensing the user’s present location, but also recording where the user has been previously in order to 170



Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites from the Tourist Information Office, and in addition to providing interpretation about the historic sites of the city centre, visitors were also able to communicate directly with the office if they had queries. Since then, the development of location-based guides for interpretive purposes has proceeded at a rapid rate. Reinhard Oppermann and Marcus Specht (2000) outline the development of Hippie, which takes its name from the HIPS project - (Hyper-interaction within Physical Space). This prototype was funded by the European Commission to act as a hand-held electronic tour guide by providing personalised, contextual information based on the position of the user and his or her previous activity. Like many early prototypes, Hippie used infrared to obtain its location, but it also capitalised on internet connectivity to provide a greater amount of interpretation; it also sought to adapt to the user’s preferences.

display pertinent information. In discussing the future of ubiquitous computing, Mark Weiser (1991, 68) commented: ‘Little is more basic to human perception than physical juxtaposition, and so ubiquitous computers must know where they are.’ In the two intervening decades, computers have indeed taken steps towards the goal of becoming ubiquitous, seamlessly integrating within daily life, and, as discussed below, determining their location within the physical world is certainly one of their tasks. As a result, location-based technology has the potential to transform the presentation of interpretation at archaeological and heritage sites. It eliminates the need for signage; this can prevent a site from appearing cluttered and avoid damage to fragile structures and monuments. Likewise, updating virtual interpretation is much easier than republishing guidebooks and leaflets, or reinstalling defaced or out-of-date signs. Depending upon the device used, e.g. a visitor’s mobile, PDA, or MP3 player, the interpretation can be accessed anywhere, at any time; after all, if a site is accessible at any time, it stands to reason that the interpretation should be as well.

Prototypes using archaeological sites as testing grounds for augmented reality also began around this time, with creations such as ARCHEOguide (Vlahakis et al. 2002), and similar prototypes at the Early Iron Age site of Els Vilars in Catalonia (Yamane and Lorés 2004). Relying on the user’s location in order to present site-specific information in the form of virtual reconstructions of the past, this type of location-based media has only recently been made practical.

Unlike traditional audio guides, location-based media allows for true random access as users are not restricted to following a set trail (Reid et al. 2005a). Advantages such as these are what have attracted so many researchers, both within academia and the commercial realm, to explore the possibilities offered by locationbased media at a wide-range of visitor attractions, from aquariums and science centres (Bellotti et al. 2002; Fleck et al. 2002) to museums and heritage sites (Proctor 2005; Pert 2008a).

The current uses of location-based interpretation have moved beyond the experimental stages discussed in this section, and a combination of factors have allowed such interpretation to be easily accessible through both proprietary hardware systems and users’ personal devices. Node Explore was one of the first companies to investigate using the former method to provide locationbased interpretation at sites such as Ashton Court, Stourhead, and Culloden.

Although it is the latter with which this research was concerned, it is clear the technology is quite malleable and can be adapted to many types of locations. In the field of heritage interpretation in particular, being in the location where a past event occurred has resonance with visitors (Reid et al. 2005a); the development of increasingly accurate location-sensing technology in personal digital devices may revolutionise the way in which visitors are able to access such interpretation.

Regarding the latter method, that of using personal devices, the development of the Mobile Bristol toolkit, and the next generation Hewlett-Packard mScape Maker, allowed anyone to create their own mediascape; some of the first experiments were with historic subjects, e.g. Riot! 1831 (Reid et al. 2005a). Using mScaper, the World Heritage Site of Blaenavon, Wales, produced an ‘e-trail’ of the town, which it attempted to hire out from the Visitor Centre (Pert 2008b). The uptake of smartphones and other technology with location-sensing capabilities will likely see a rapid increase in the creation of similar tours.

History of location-based guides The first experiments in using location-sensing technology to produce a tour guide were carried out over a decade ago in the Graphics, Visualization, and Usability Centre (GVU) at the Georgia Institute of Technology (Abowd et al. 1997). This system, known as Cyberguide, utilised infrared and served to guide visitors through the laboratory. A more ambitious Cyberguide project was later undertaken, incorporating the nascent public GPS to extend the tour from the lab to the surrounding areas.

Location-sensing technology The key component of location-based media, that which sets it apart from traditional interpretive methods, is the underlying technology that senses the location of the user within the landscape. Anthony LaMarca and Eyal de Lara (2008, 1) concisely summarise the reason for the proliferation of so many technologies that are used for this one purpose:

The Distributed Multimedia Research Group at Lancaster University explored the possibilities on a larger scale, creating an all-encompassing tourist information unit, called GUIDE, which relied on a wireless network (Cheverst et al. 2000). The tablet device was distributed 171



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process has the same limitations as mentioned above, namely the addition of transponders to a site and the use of proprietary hardware.

‘Unfortunately, there is no single location technology that is good for every situation and exhibits high accuracy, low cost, and universal coverage. In fact, high accuracy and good coverage seldom coexist, and when they do, it comes at an extreme cost. Instead, the modern localization landscape is a kaleidoscope of location systems based on a multitude of different technologies including satellite, mobile telephony, 802.11, ultrasound, and infrared among others.’

Global Positioning System (GPS) Originating as a United States military program in the 1970s, the Global Positioning System (GPS) has emerged over the past decade as a synonym for location-sensing technology as a whole (Kaplan and Hegarty 2006, 3). Since the elimination of Selective Availability (SA) in 2000, GPS has moved quite rapidly from the realm of specialists to general usage (El-Rabbany 2006, 46).

As a result, a number of systems are currently in use, each with distinct advantages and disadvantages when used at archaeological or heritage sites. An overview of this technology is presented here.

As with infrared, GPS relies on a transmitter and a receiver to determine location. The receiver is the device held by the user; when activated, it searches for the constellation of twenty-four global positioning satellites that orbit the earth. The relative position of the satellites to each other is known, as is their position above the earth. Their distance from the receiving device is determined based on the time it takes for signals to travel between the device and satellite. Based on this data, the device is then able to calculate its location based on trilateration. At least four satellites are necessary to trilaterate a fix; six or more provide a more accurate reading (El-Rabbany 2006, 8-9). A clear line of sight between the satellites and the receiving device is essential to ensure that the fix remains stable; buildings, trees, and bad weather can obscure the signal. Likewise, a multipath error can occur when the receiver loses the GPS signal or the signal is reflected off surrounding objects, such as water or buildings, i.e. the ‘urban canyon effect’ (Borriello et al. 2005, 37; El-Rabbany 2006, 28). The reason for its informal name of ‘GPS bounce’ is evident in Figure 1.

Infrared The electromagnetic spectrum is composed of radio waves, microwaves, infrared, visible light, ultraviolet, xrays, and gamma rays (Campbell 2006, 31). The utilisation of the location-sensing capabilities of radio waves in everyday processes is a relatively recent development, and is discussed below. Infrared, however, has long been used in devices such as remote controls to allow data to be sent and received over short distances. Infrared devices work through the use of a transmitter and a receiver; the former sends out pulses of light to be received by the latter. The pulses have corresponding binary codes and commands that direct the behaviour of the end product. In the realm of location-based media, this typically consists of triggering a specific audio file based on the movement or location of the visitor. As discussed above, infrared was used in the first experiments with location-based media (Abowd et al. 1997), and although this technology can only be used at close range and requires the addition of separate hardware to the site, it is becoming prevalent at indoor locations where such limitations are of less importance.

GPS accuracy is not ideal at present (typically within 10 m), but it can be increased with the addition of differential GPS (DGPS), consisting of a base station with a known position (El-Rabbany 2006, 94). This is then able to correct for satellite and calculation errors, improving accuracy to within 1m or less.

Antenna Audio, long a leading producer of museum audio guides, is one such company that is beginning to employ infrared systems at heritage sites throughout England (e.g. Bristol’s ss Great Britain). Their X-plorer guide is billed as ‘bringing exhibits to life’, and these systems seek to capitalise on the immersive nature of the technology in order to present the past (Reid et al. 2005b; ‘Antenna Audio’, 2007). In this particular instance, the immersive quality arises in part from the potential freedom given to visitors: they do not have to consciously select a route, follow a structured trail, or stop to push buttons as on traditional audio guides (Reid et al. 2005a).

In his Introduction to GPS (2006, 11), Ahmed ElRabbany concludes his introductory chapter: ‘As for the future, it is said that the number of GPS applications will be limited only by one’s imagination.’ Indeed, for outdoor locations, it is GPS technology that has the potential to transform the way such sites are presented, as well as affect how visitors engage with and experience the past. First and foremost, the system does not require the addition of hardware or signage to a sensitive archaeological site. Rather, visitors bring the necessary devices to and from the site, precluding any negative environmental impact.

In contrast to this, Scandinavian company Dataton is also exploring the use of infrared systems through their PickUp audio guide. PickUp is designed to be intuitive; users point the device at a transponder marking an area of interest and click the ‘action button’ to access the audio commentary. The device can also be programmed to activate automatically when a visitor passes a transponder. Although certainly novel in form, PickUp

Although there are proprietary hardware systems using GPS to deliver interpretation, such as Node Explore’s Explorer unit, it is believed that the future of such technology as an interpretative method will rest with the visitor. The explosive growth of mobile phones with GPS 172



Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites capabilities (smartphones) has given rise to a number of applications that take advantage of this functionality, such as FourSquare and VoucherCloud, in addition to general mapping and navigational uses.

combining the benefits of a tour guide, i.e. the ability of the visitor to ask questions or get clarification, with the freedom and flexibility offered by a digital device. Using the RFID system outlined above, for example, it would be possible for a museum visitor to scan a tagged object with a reader to access additional details and communicate with the host machine, whether through an automated question/answer service (e.g. FAQs) or a manned station providing interpretation. As El-Rabbany (2006, 11) praised the ever-growing potential of GPS, Hunt et al. (2007, 1) likewise see RFID tags as providing a source for new location-based services: ‘In the coming years, new RFID applications will benefit a wide range of industries and government agencies in ways that no other technology has ever been able.’

Figure 1 An illustration of GPS bounce encountered during the prototype trials at the Clifton Suspension Bridge.

QR codes In a similar manner to RFID tags, Quick Response (QR) codes were initially used as part of inventory control, but soon gave rise to a variety of applications. They are known as object hyperlinking for the ability to connect a real world object to a virtual site, and in effect serve as barcodes encoded with specific data that can be accessed by mobile phone cameras (Rohs and Gfeller 2004, 265). The data most commonly encoded includes text, contact details, or a hyperlink to connect the user directly to a website. As a result, it has found a ready market as a form of advertising, allowing viewers to easily link to the product’s webpage from wherever they have internet access.

Radio Frequency Identification (RFID) Radio-frequency identification tags (RFID) have become a common, if invisible, feature in many aspects of day-today life, typically used for inventory control. The tags typically contain at least two parts: (1) an integrated circuit to store and process information; and (2) an antenna to receive and transmit the signal. In turn, there are two common types of RFID tags: (1) active, which contain a battery in order to transmit signals automatically; and (2) passive, which require an external source for signal transmission (Hunt et al. 2007, 6). As would be expected, the former are more expensive but can transmit data over a greater distance, whilst the latter cost less to produce and have a shorter range (Hunt et al. 2007, 7).

As a way of presenting heritage, a QR code could be added to a traditional interpretative sign, which is then scanned by the visitor’s mobile phone to call up an associated web page or trigger further content. Such a system was used in 2007 as part of the BBC television documentary Coast in order to provide information about Britain’s coastline, ranging from geology and botany, to archaeology and history.

The tags form just one part of an RFID system. The other two components are a reader (or interrogator) to access the data stored on the tag, and a controller (or host), which typically takes the form of a workstation. When the reader approaches the tagged object, it sends a signal to the tag that allows the data to be transmitted, and this in turn can be relayed to the host machine (Hunt et al. 2007, 6).

Although mobile phone technology has increased dramatically since then, a problem remains: the manner in which the data is encoded and read by a device varies from operating system to operating system. As a result, only those with particular devices can access the interpretation as intended, limiting its potential and functionality. As a solution for testing purposes, the studies run by O’Hara and Kindberg on the Coast tour (2008) provided visitors with trial phones with the application pre-installed, much like the testing at the Clifton Suspension Bridge discussed in Section 7.

RFID tags work at a greater distance than infrared due to the longer radio waves, but also relies on the installation of on-site hardware. However, by using this technology the University of Lancaster was able to successfully illustrate how radio frequency devices and similar wireless systems can be used to create an allencompassing tourist guide. For example, their GUIDE not only told visitors about sites of historic interest, but could communicate queries to the Tourist Information Office as well (Cheverst et al. 2000).

Bluetooth The initial aim of this technology was to achieve interconnectivity between any Bluetooth device regardless of brand, i.e. any Bluetooth device can connect to similar devices within range (Kostopoulous et al. 2008, 259). However, as a result of this functionality, it allows

The potential of two-way communication coupled with location-based devices has yet to be fully explored, but the GUIDE prototype hints at the possibility of 173



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process requires the installation of a grid of receivers in the ceiling in order to obtain the ultrasonic radio signal issued by the transceiver (Harter et al. 2002). As a result, devices would only be of use in an indoor setting, such as the ‘Walk in the Woods’ prototype studied by HewlettPackard (Hull et al. 2002).

for a relatively low-cost wireless network to be established for sensing the location of devices indoors (Zheng and Ni 2006, 425). In a similar way to RFID tags, a Bluetooth tag can emit data that is then read by a device with Bluetooth capabilities; most mobile phones, smartphones, and PDAs have this function as standard (Zheng and Ni 2006, 425).

At present, this technology requires the use of proprietary hardware. As discussed below, the future development of location-based media is likely to be device driven; as a result, GPS and Wi-Fi are the greatest contenders to be the technology to power such devices due to their widespread adoption in a variety of systems.

The city of Norwich used this technology as part of their Norwich 12 heritage trail, which highlights twelve architecturally and historically important buildings in the city. In 2008, the first of twelve intended ‘totems’ was placed at the site of Norwich Castle, and it used Bluetooth-enabled hypertags (produced by the eponymously named company Hypertag) to broadcast interpretive content to the digital devices of interested passers-by (Brinkley 2008).

Devices: proprietary Location-based devices such as Node’s Explorer unit, Antenna Audio’s X-plorer guide, and Dataton’s PickUp units adhere to a similar format. This method ensures that visitors have the same experience, i.e. the interpretation performs in the same way on each unit. Likewise, it is easy for staff to replace or repair a faulty device. From the visitor’s perspective, they do not have to plan ahead to download a guide, and the similarity to traditional audio guides likely adds a degree of familiarity that makes such guides easier to use.

Wireless / Wi-Fi Wi-Fi is the high bandwidth technology most commonly used for internet access, especially with regards to mobile computing. Hippie, discussed earlier, is one such tour that was designed with internet functionality as a driving feature (Oppermann and Specht 2000) and, more recently (2005-2006), Ruthin, Wales developed an e-trail that relied on a combination of GPS and internet access to provide interpretation to the town’s historic centre (Pert 2008a, 6). Using a map-based interface, users would tap the building about which they wanted interpretation, and the application would launch Coflein, the Royal Commission on the Ancient and Historical Monuments of Wales’ online database of historic sites and buildings, as well as provide archive material if available.

However, there are disadvantages as well: devices can only be used during set hours whilst the site is in operation; the site is required to purchase hardware devices and subsequent maintenance; and manpower is required for the distribution and collection of devices. Although negated at larger sites such as the ss Great Britain or Roman Baths, many smaller sites simply do not have the funding or personnel to invest in such systems.

Wi-Fi connectivity has the potential to be used as a location-sensing method by measuring the intensity of the signal in a user’s device in relation to the known location of the transmitter (Zheng and Ni 2006, 398). The company Skyhook Wireless has developed a Wi-Fi Positioning System (WPS) based on knowing the location of millions of Wi-Fi hotspots; with a claimed accuracy of 10-20 m, it is an ideal form of location sensing for indoor and urban environments. The company’s XPS system seeks to be a hybrid of the Wi-Fi Positioning System and GPS, allowing for a more accurate crossover between outdoor and indoor location finding.

Commercially, many companies who traditionally produced audio guides, e.g. Acoustiguide and Antenna Audio, have diversified to produce not only proprietary location-based systems, but also tours available for download to a visitor’s own device. Advertising literature for MobiTour, another audio guide company, extols the virtues of using a visitor’s mobile phone, highlighting that a shift to a new method of delivery can benefit both the site itself, as well as the visitor:

Another aspect of Wi-Fi technology is that it allows for communication over the internet, e.g. for asking queries at a manned location, as well as providing the potential to access additional information via a web browser. Although not tested as part of the prototype guides used in this study, the potential for two-way communication should not be overlooked.

‘Enhance the visitor experience at your historic site with a simple to setup and use mobile phone audio guide. MobiTour uses your visitor’s mobile phone as an audio guide ensuring their visit to your Historic House is a complete and entertaining experience. It also has the added benefit of being simple to setup, and simple to maintain. There is no hardware for you to purchase, maintain, issue or collect, freeing up staff for other duties.’

Ultrasonic

Devices: visitor owned

Although studied for nearly a decade (e.g. Hightower and Borriello 2001), ultrasonic methods of location sensing have yet to catch on in popular or commercial usage. Although accurate to within centimetres, this method

Mobile phones are rapidly becoming the standard bearer in the realm of ubiquitous computing. According to the International Telecommunications Union (Parkes 2010), 4.6 billion were owned worldwide in 2009, and the 174



Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites number was expected to pass 5 billion at the end of 2010. The development of the smartphone has given rise to a personal device that, for all intents and purposes, functions as a miniature computer. Such devices are commonly cited as the embodiment of pervasive computing (Hansmann et al. 2003; Zheng and Ni 2006), and a forum entitled ‘Cell Phone Culture’ given at the Massachusetts Institute of Technology in November 2005 further describes the many uses that mobile phones have taken on:

harness mobile technology lies in a realistic understanding of what activities people on a large scale are actually doing with their handsets now.’

‘Ubiquitous in most developed societies in Europe, the Americas and Asia, the cell phone has become a laboratory –some would say an asylum– for testing the limits of technological convergence. Less a telephone today than a multi-purpose computer, cell phones are game consoles, still cameras, email systems, text messengers, carriers of entertainment and business data, nodes of commerce.’

4. Methodology

Based on these figures, it is not unreasonable to surmise that if growth and user uptake continue as forecast, then the potential offered by mobile computing through smartphones will allow heritage interpretation in this manner—and any other type of interpretation that is location reliant—to become a reality.

David Uzzell (1998, 185) highlights the need for visitor evaluations when considering the effectiveness of traditional forms of interpretation: ‘If interpretation is to achieve the objectives to which most interpreters aspire, then simply throwing money at exhibitions, leaflets or staff has at best an uncertain outcome and at worst is profligate. Coming to grips with ‘‘doing it’’ has taken precedence over standing back and questioning whether it is actually working.’

In addition to the features described above, modern smartphones have two important components that are likely to play a role in the development of location-based interpretation: the presence of (1) wireless capabilities, and (2) GPS receivers. The former can allow for the download of applications (‘apps’) for an almost unimaginable number of subjects. It can also serve as a method of determining location indoors based on a measure of signal strength. Assisted GPS (A-GPS) for mobile phones can also help boost the weak satellite signal when a user is inside, allowing a GPS-based tour to cross from outdoors to indoors (El-Rabbany 2006, 135). The incorporation of a GPS receiver allows for apps to make the user’s location the driving force behind the functionality of the software.

Therefore, in order to determine what audiences desired from interpretation, a number of studies were carried out at heritage sites using both established methods of interpretation (Section 5: Roman Baths) and prototype guides designed by the author (Section 6: Prototype Development). Visitor data in the form of observations and self-administered questionnaires were collected in order to gain a better insight into audience preferences. The Roman Baths Museum was selected for these studies for several reasons: (1) the use of audio guides at the Baths is a well-established form of interpretation; (2) two new narratives were added shortly before the study was carried out, allowing for a greater range of narration to be studied and compared; and (3) at the time of the study, the Baths welcomed nearly one million visitors each year, making it one of the most visited archaeological sites in the UK (ALVA 2006 Visitor Statistics).

But first the devices must get into the hands of the end users. Less than a third of those surveyed during the prototype testing had a smartphone during the time the trials occurred (July 2009 – March 2010), supporting a statement from The Neilsen Company (‘Brits Increasing Wise to Smartphones...’, 2009) that significant opportunities for growth remained.

In a similar manner, the Clifton Suspension Bridge is a well-visited site, with approximately 400,000 visitors a year. However, it is here that the similarity ends. The Roman Baths charge an admission fee and is only open during set hours; the Suspension Bridge is a free site, accessible at any time by those who wish to see it. However, it lacks interpretation on the site itself, and those wishing to know more about the site and its history must visit during the Visitor Centre’s hours of operation. Such differences allow for a greater degree of visitor behaviour to be observed.

There are indications that such growth is proceeding rapidly: per a press release from consumer research organisation comScore (‘UK Leads . . .’, 2010), the UK is the fastest growing marketplace for smartphones in Europe, with 11 million subscribers and a 70% growth over 2009. In October 2009, The Economist predicted that most handsets sold will be ‘smart’ by 2015 (‘Cleverly Simple’ 2009). According to Edward Kershaw, the vice-president of Mobile Media at Neilsen:

The underlying aims for each of the evaluations carried out throughout the course of this research varied, but overall each sought to explore how visitors to archaeological and heritage sites interacted with and rated various methods of interpretation. The results were

‘Although there have been sizeable increases in the takeup of new mobile technologies such as video and location-based services, they remain niche forms of behaviour. The era of the handset as a truly multi-media device on a mass-market level lies somewhere on the horizon, and the key for companies to successfully 175



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Visitors were queried about a variety of subjects, such as tour preferences and awareness of tour availability, and asked to rate certain features: ability to move at one’s own pace, length of audio clips, sound effects, educational value, entertainment value, extra information provided, and ease of use of the device. Qualitative data was also collected from in-house sources at the Roman Baths, specifically a Visitors’ Book and Comment Cards that were available at the exit.

subjected to chi-square testing to determine if there was any significant difference between the tours used.

5. Roman Baths As discussed above, a key component of this research involved examining visitor reception to established audio tours in order to determine visitor preference with regards to aspects of content. This data was then extrapolated to a location-based method of delivery and tested with a prototype design, as described in Sections 6 and 7.

The audio guide device records aspects of usage such as the amount of a segment listened to, the frequency of the visitor stopping or pausing the device, and which tours visitors chose to access. This allowed for a comparison between the self-reported qualitative results of the visitor evaluations, and the quantitative, actual usage.

Two questionnaire-based studies were carried out at the Roman Baths Museum in Bath, England in March-April and August-September 2007, respectively referred to as the Spring and Autumn surveys. These surveys collected demographic data, audio guide and content preference, and queried visitors as to hypothetical situations regarding location-based interpretation.

Findings Over 500 visitors were surveyed as part of the Spring and Autumn Surveys at the Roman Baths, and in general, traditional audio guides were well-received, with 95% of those surveyed using it during their visit. From their responses, a number of factors were highlighted regarding visitor preferences when using this form of interpretation. These can broadly be divided into: (1) the content of the audio guide; (2) choice; and (3) visitor behaviour.

The audio guides examined as part of this study were first introduced at the Roman Baths in 1995. A key motivation for the museum’s implementation was a desire for ‘quietness’; in the past, there was a ‘cacophony’ of competing tour guides and talking visitors (K. Smith, pers. comm., January 2005). With the introduction of the audio guide, noise was greatly reduced (K. Smith, pers. comm., January 2005; Walter 1996).

Audio guide content

Today, audio guides are the primary method of delivering interpretative content at the Baths, and each visitor receives one as part of the admission price. Englishspeaking visitor can access three tours from the same device: a general tour, a children’s tour, and a tour of the Baths given by American travel-writer and humorist Bill Bryson. Regarding length of the audio content, the general tour averages 1:24 minutes per segment, the children’s tour 0:48, and Bryson occupies the middle ground with an average of 1:10 per segment.

The results regarding the audio guide content were a mixture of the surprising and expected. Based on the qualitative data obtained through both the surveys and Visitors’ Book comments, it was anticipated that there would be a significant difference regarding tour length, but this did not materialise in the chi-square tests performed on the rating of the Autumn Survey. However, over half of the respondents reported stopping or pausing the audio segments before completion. Likewise, the Spring Survey revealed a difference between the general and children’s tour, with indications that the shorter length of the latter is preferred. The rather vocal qualitative responses also serve to indicate that more can be done to shorten the audio segments, either through judicious editing or the use of layers (e.g. ‘Press x to hear more information’). Music and sound effects were rated quite low.

Some sections of the general tour have ‘layers’ or ‘fascinating facts’, which give the visitor the option of listening to more detailed information. These are made available at the end of the main audio segment, at which point listeners are advised to press a second number if they wish to hear further details. In this way, the audio tours are intended to be interactive: visitors not only choose which sections they want to listen to, but can delve deeper into a chosen subject area if desired.

SPRING #5

The guides themselves record visitor data, e.g. what sections were listened to, how much of a section was actually played, whether extra sections were accessed. Although individual audio guide data could not be matched to survey results during the course of this study, this data still provided an interesting comparison regarding actual usage and visitor reported usage. For example, although a majority of the visitors reported listening to the audio commentary at most available locations, on average only 20 minutes of the available 80 minutes of audio was accessed.

SPRING #23

Easier to listen to; simple description + not too long to listen to [regarding use of the children’s tour].

The audio guide sections are too long and bog down the pace of visiting everything here . . . Keep the sections brief + include information that isn’t already provided on the signs. While the information provided was useful and interesting, the audio effects (ie. splashing, laughing) are superfluous + a bit annoying.

176



Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites AUTUMN #173 Sounds effects annoying, just wanted info.

controlling the audio commentary. Visitors overestimate their consumption of interpretation.

SPRING #291

Of note are the responses to the questions that seek to gauge visitors’ interest to downloading guides to their own device and the use of post-visit websites to provide additional information. The results are inversely proportionate, with only one-third indicating willingness to download such a guide, but a 65% majority expressing interest in obtaining additional information. Based solely on these results, it appears that using one’s own device in a museum setting, where a number of interpretive methods are at the visitor’s disposal, has little interest.

Add more stories or points where you can find out a interesting fact.

AUTUMN #305 Take off some of the music + sound effects which drown the commentary. Also unexpected, there was no quantitative difference regarding the use of extra information, but in hindsight it appears that the way in which visitors may have interpreted this phrase varies, i.e. each tour added information in a different way: the general tour through the use of ‘fascinating facts’; Bryson at the Baths by using opinions rather than fact-based interpretation; and the children’s tour through the use of quizzes.

Overall, however, the contradictory nature of the public is evident throughout both the quantitative and qualitative results. For example, although ‘different types of tours’ was the second most common selection for improvement, it drops to last when visitors were asked to select which audio guide features they find most important. The comments, both on the surveys and in the Visitors’ Book/Comment Cards, reflect individual preferences: some feel there is too much overlap between interpretation, others would like to see additional methods used; the Bryson at the Baths tour is loved by some, loathed by others; the motivation of visitors can range over sixteen possible reasons for selecting a particular tour.

Of great interest—from both a commercial perspective and for heritage interpretation in general—is the number of visitors (57%) who reported being willing to return to the Baths to use a different tour. This indicates that visitors value the option to get additional, different interpretation; however, as discussed below, this option was not always taken. Choice Surprisingly, only half of the respondents took advantage of the multiple tours available and listened to more than one; it was subsequently shown that approximately onethird of the respondents were unaware that they had three options. Visitors cannot be expected to automatically discern the availability of options, especially when the data obtained from the guide itself shows that many visitors prematurely stop the audio guide instructions. In a similar manner, the Bill Bryson tour is in need of a better introduction; comments on the Visitors’ Book/Comment Cards questioning the use of an American accent clearly reveal that at least some of the audience is unaware of not only Bryson’s background, but also the reason his tour is included. SPRING #13

As a result, it is clear that a ‘one-size-fits-all’ approach, whether in the method of interpretation or the content presented therein, does not necessarily meet the needs of visitors. This underscores Ronald Hawkey’s assertion (2004, 4) that ‘The real key to future development [of interpretation] is likely to be personalisation...’

6. Prototype Development As a result of the visitor evaluations described in Section 5, specific features highlighted by visitors as necessary to include in interpretation (e.g. shorter, layered audio segments) were incorporated into the creation of three digital prototypes: (1) Audio-only; (2) Image-based; and (3) GPS-based. These were subsequently tested by visitors at the Clifton Suspension Bridge in order to determine whether the method of delivery affected the audience’s reception to the guide, i.e. although the audio content remained the same on all three tours, how visitors accessed the interpretation varied.

I was not aware there was a Bill Bryson Tour (unfortunately).

SPRING #22 Bryson was annoying! #244 Enjoy his books – speaks with more enthusiasm + accessible descriptions. SPRING #274 SPRING #309

Didn’t think much of Bill Bryson – either content or diction.

The software used for all prototypes was HewlettPackard’s mScape Maker, which allows for the creation of mediascapes that can be composed of audio, still images, and video; it also allows the importation of HTML and Flash files. Logging is also available to track the path of visitors through a site, as well as record their usage of the device. The mScape Maker Suite was previously available for free download, but is unsupported by HP at present.

Who is Bill Bryson?

Visitor Behaviour A number of issues of interest fall under the category of visitor behaviour. One such item is the use of audio controls: over half reported pausing or stopping the tour before completion. As mentioned above, this indicates that the tours are too long at present; it also shows the importance of providing visitors with a method of 177

 

also

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process The development of a mediascape begins with importing a map of the desired area and inputting the GPS coordinates for the location. Regions are then drawn on the map in the form of predefined polygons or by a freehand tool. Each region is then programmed to perform an action upon the entrance or exit of the user to that region.

Density of trigger locations In a related, but often-overlooked, aspect of providing interpretation is the density of points of interest. With the addition of the Bryson at the Baths and children’s audio tour, there are over one hundred signs within the Roman Baths competing for visitors’ attention. Three or four signs are often visible at any point, and, in an extreme instance, eight different audio guide signs were counted in one location.

mScape Maker also allows for the creation of ‘hotspots’ as overlays on .JPG images, which in turn can be programmed to carry out actions. Each button used in the prototypes was programmed in this way, e.g. for stopping or pausing an audio segment or for returning to a regional menu. However, limitations in the software precluded the development of a fast-forward or rewind button, although visitors could replay the segments.

It is felt that visitors should be given space and time to absorb the presented interpretation (Copeland 2006); what is ideal? Do visitors listen at points evenly throughout the complex, or do visitors listen heavily at the start of the tour and taper off near the end, as ‘museum fatigue’ (Gilman 1916) sets in? Questions such as these are important in the design of any interpretation, whether using traditional or digital methods. Music and sound effects Brian Bath (2000, 161) wrote in praise of ‘soundscapes’ at heritage sites, and cites a qualitative survey carried out in 1995 to support the claims that such use of sound increases visitor satisfaction. However, it is likely that the novelty of such tours at that time resulted in the data obtained. Fleck et al. (2002, 18) call this the ‘wow factor’, and it is worth considering when testing any new technology: novelty is likely to impact how users assess it, e.g. they may be more forgiving than if the device or system were in standard use.

Figure 2 The leaflet given to participants on the Audioonly and Image-based prototype tours.

The more recent visitor comments given on the surveys at the Roman Baths about this subject are especially telling, both those written on the questionnaires and the statements made directly to the author: several visitors specifically mentioned the low quality of the sound effects, and one was quite unhappy about the sounds of animals being led to sacrifice. Likewise, the answers to the Autumn Survey show that this aspect of the tour was rated the lowest overall. The negative response to the use of music and sound effects indicates that these features should be avoided unless it directly relates to the site or artefacts on display, e.g. a musical instrument.

As signs could not be placed on the bridge to indicate places of interest, as is done at the Roman Baths and many locations that rely on traditional audio guides, a map was created to suggest areas in which to play the interpretation (Figure 2). The path took visitors along both sides of the bridge in a circular trail, and the guides were given to all participants on the Audio-only and Image-based guides. The location-based tour relied purely on the internal GPS receiver and location-specific menus to provide visitors with tour options.

Images

Audio segment length

One of the most important issues facing the use of digital technology is whether to use images or video as part of the interpretation. Do images aid in the interpretation or merely distract the visitor from looking at the real object in front of him or her? Studies of the technology in different arenas, e.g. Bellotti et al. 2002 at the Genoa Aquarium, have shown the latter to be a very real pitfall. As a result, images should only be used when they can enhance the interpretation.

At the Roman Baths, negative comments about the general audio tour chiefly centred on its length, whereas the children’s tour was praised for its simplicity. Often discussed in the field of interpretation studies is the idea of width and depth, referring to how visitors typically fall into two general categories: those who prefer a general overview on subjects (‘skimmers’), and those who desire in-depth interpretation (‘divers’). As a result, one of the most important steps in the design process is determining how much content to create so that each visitor type is satisfied with the amount of interpretation provided.

Fortunately, this should be much more straightforward within the realm of archaeological interpretation; there is enormous potential to give shape to the invisible by showing reconstructions of ruined or long vanished 178

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites buildings, displaying the location of a now-excavated artefact in situ, or safely revealing underground sites. Observing how visitors use the digital display and react to these images is an essential step in the overall design of the interpretative content.

be off-putting at a site with an evocative atmosphere, such as a battlefield. O’Hara and Kindberg (2007, 272) cite this as an advantage to using 2D barcodes: ‘The deliberativeness of reading a barcode gives firm control over “when”, so the user can access the content when the time is right...’. The Dataton PickUp audio guide is likewise activated by the user, rather than triggered automatically (although such a function can be programmed into the device). In writing about Hippie, one of the earliest forms of location-based media, Oppermann and Specht (1999) describe how a blinking icon is used to alert the user that a new commentary is available, and ‘...the user is free to select the new presentation whenever he or she wants but is not patronised by automated presentations.’ Although not studied in detail as part of this research, this aspect of the technology must be considered in the design process.

Narrative types The flexibility of the technology is such that it is possible to present numerous narratives: a basic chronology of the site, a guide intended for children, a tour given from the point of view of someone who once lived or worked at the site. Yet will visitors take advantage of this wide range of options? In both surveys at the Roman Baths, an overwhelming majority of visitors listened to the general tour most often, and stated that it was their desire for facts that guided their decision. However, as discovered in the Autumn Survey, over onethird of visitors were unaware of the options available, making it difficult to determine whether this apparent preference was real or the result of poor publicity.

Admission cost Regarding admission costs and interpretation, the amount visitors would be willing to pay to access interpretation on their own device showed a marked increase at the Suspension Bridge when compared to the Roman Baths. This is likely to be attributed to the lack of an entry fee. Subsequent questions regarding the prototype guides supported this argument, with visitors commenting that their willingness to pay for interpretation was inversely proportionate to the entry fee charged. Additional testing at further sites would be needed to determine whether this assessment is accurate.

Choice As stated above, giving visitors control and choice are critical aspects of interpretive design, whether in the amount or type of information presented. Perhaps of even more importance is choice in the format the interpretation takes. Preferred learning styles vary (i.e. whether to learn through auditory, visual, or kinetic means), and both auditory and visual can be used with location-based technology; if possible, content using both methods should be provided. This will not only allow the interpretation to reach a wider audience, but also allow the most appropriate form of interpretation for the site’s context to be utilised.

Those offered the possibility of accessing interpretive material on a digital device as part of the No Interpretation survey (discussed below) showed a willingness to pay an average of £2.42. However, those who used the leaflet responded that it was not worth £1.00, despite it having a physical presence. This indicates that there is an inclination to not only pay for but also value interpretation that is provided digitally.

Another aspect of formatting is the approach taken to present the interpretation. Complete personalisation is not possible or practical, but the advantages of digital technology are such that a number of features can be catered to. For example, the same interpretation can be used, but users are given the choice in how to approach it, e.g. whether through a linear narrative or one that is conceptual or thematically-based. In a similar manner, users can choose whether or not to use automatic activation, or be guided from point to point with a location-based tour.

However, in light of previous research by Paulette McManus (2000), in which it was shown that 93% of visitors kept at least one guidebook from their visits to English Heritage sites, it raises the question as to how mobile digital interpretation would fare after the visit: would it be valued as a possession in a similar manner? Based on the overall responses to the query regarding a post-visit website, it appears that there is such an interest regardless of the media used.

Automatic activation Although visitors were not questioned about this feature during the surveys, this is an aspect of using locationbased guidance systems that must be taken into account: do visitors want the interpretation to automatically begin when they approach a trigger location, or do they prefer the option of starting it themselves? Most location-based devices employ the former method, e.g. the Antenna Audio Explorer guide at the ss Great Britain or Node Explore’s Ashton Court system. However, this removes a degree of control from the visitor, and could potentially

General conclusions Due to limitations of time and programming, only some of the features discussed here—such as shorter audio segments, layering, and the avoidance of sound effects— were incorporated into the design process used to create the prototype guides described in Section 7.

179

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Additional aspects of interest arose from the differences between the Roman Baths and Suspension Bridge. As intended, this allowed the sites to be more easily contrasted, and from this several factors became apparent: (1) the impact that an admission fee has on willingness to pay for interpretation; (2) visitors appear to value digital interpretation more than physical; and (3) visitor preferences seem fluid.

Hardware In regards to the devices themselves, screen glare is a barrier to the successful use of visual content on digital interpretation at outdoor locations. Visitors to the bridge often cupped the device to shade it or stood in the shade of the abutments in an effort to block the glare. Newer devices with brighter screens or electrophoretic displays, however, are likely to emerge as a solution to this issue.

7. Clifton Suspension Bridge

The other content used, that of audio commentaries, has highlighted two areas that must be dealt with: (1) audio methods are not the most appropriate method for urban contexts due to traffic noise, and (2) complete audio controls are necessary for pausing, stopping, fastforwarding, and rewinding. The latter can be easily overcome with more complex software. The former, however, would indicate that interpretation within an urban landscape should avoid audio commentaries, and instead rely on screen-based content and text. Coupled with the improvement of display screen technology, this is the most likely path to take for interpretation in urban environments, e.g. the Museum of London’s Streetmuseum.

Due to its nature as an iconic Bristol landmark, the Clifton Suspension Bridge was selected as a site to carry out a series of surveys to determine visitor preferences with regards to personal guidance systems. Although it has a free visitor centre, at the time of the initial No Interpretation survey, there were no methods available for accessing interpretation whilst visitors walked across the bridge. Soon after this initial evaluation, the visitor centre developed and sold a leaflet-based trail, and it was decided to query visitors regarding their experience with this method of interpretation. With the development of several working prototypes, these preliminary surveys would then allow visitors’ attitudes to be tracked along a spectrum of interpretation at one location: (1) no interpretation present; (2) written interpretation (leaflet); (3) Audio-only guide; (4) an Image-based guide; and (5) a guide activated by GPS. Each survey collected basic demographic data and asked visitors to rate the method of interpretation currently being tested; it concluded with general questions about the use of digital devices.

AUDIO #5

Traffic noise.

IMAGE #12 But the sound was too quiet at some points with the traffic. GPS #4

I think it would depend on the phone you have. Some have volume controls at the side which would make it easier.

Based on the results of the surveys, the null hypothesis, stating that there is no difference between user preference in method of delivery, cannot be rejected: statistical testing showed that there is an observable difference in only a very few situations. This indicates that although location-based media is not rated as a significant improvement as a method of delivering interpretation, it is also not significantly worse, but rather can be considered on par with more traditional methods.

Another aspect of the hardware used is the manner in which it was distributed, i.e. via a set point (the author) rather than using visitors’ personal devices. Although the majority of visitors found the iPAQs and HTC easy to use, several had difficulties that were only solved because the author was present and could provide assistance. Using a personal device would likely negate this issue as users have a better understanding of how to operate their own device.

However, there is a continued upward trajectory of visitors being willing to download tours to their own digital device. It is this factor, that of being able to access interpretation anywhere, anytime, on their own device, which may have more of an impact on the future of heritage interpretation than the specific method of delivery, i.e. location-based or otherwise. However, as intended, the prototype guides and associated visitor evaluations and observations flagged up a number of practical aspects of location-based media, and the use of digital interpretation in general, that must be taken into consideration if such methods are to be used successfully in the future.

Site In addition to the audio problems outlined above in regards to the context of the Suspension Bridge site, another environmental aspect that must be considered is the movement of visitors within a space. Visitors using the prototype guides on the bridge were often jostled by passers-by or found it difficult to navigate a crowded walkway whilst also listening to the commentaries and looking for the features it described. In a related issue, visitors were often unaware of other pedestrians; it is believed that the concentration required to use a digital guide could pose a safety risk, especially in high traffic areas. This seems to preclude the use of such guides in an urban environment until better solutions for safety and display are found.

180

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites IMAGE #12 Rushed by people crossing bridge. GPS #1

personal side of the location can make the interpretation more immediate and more intimate, which appears to have a positive correlation with visitor satisfaction.

Found middle section a bit uncomfortable as nowhere to stand out of the way of other visitors. Fine at each end as lots of room / bench.

As discussed previously, content can be presented in many forms, with audio and images/video being those used on the prototype guides. As would be expected, a number of factors appeared in relation to the content used. Regarding the audio commentaries and length of the tour, the GPS guide was rated higher than the Audioonly tours, despite the actual content being the same. It is possible that users found the location-based method of delivery less overwhelming than that of the Audio-only menu where all options were available at once. Based on both the surveys at the Roman Baths and those at the bridge, short, layered segments of audio are preferable to one long segment as it puts control into the hands of the user to decide whether or not to access additional interpretation.

Visitor behaviour Despite the outdoor nature of the site, visitor behaviour often mirrored that described in museum settings. For example, users typically had a brief orientation period on the bridge to familiarise themselves with the equipment as well as view the bridge and Avon Gorge; likewise, they rarely chose to backtrack to listen to audio segments they may have missed. Most apparent, visitors do not necessarily follow an intended route, regardless of the medium used. Visitors did not listen to interpretation on both sides of the bridge, and rarely crossed to the other side. It is because of this behavioural trait that locationbased media has an edge when compared to traditional methods that typically require visitors to access interpretation in a particular order for it to make sense. As the interpretation is tied to the location, rather than a predefined trail, the visitor’s choice of route has no impact.

Adding images to the audio commentaries saw an increase in visitors’ satisfaction with the amount of interpretation available and the length of the audio clips, despite the commentaries being the same. This is a strong indicator that the use of images is well received, and can alter the user’s perception of the amount of interpretation provided. The images and videos also allowed visitors to more easily find the areas described by the commentary, as well view certain areas, such as the interior of the abutment that would be otherwise hidden.

Isolation when using electronic guides has long been cited as a disruption to the social aspects of a museum visit (e.g. Falk and Dierking 1992; Aoki et al. 2002). However, based on the use of headphone splitters, such isolation seems self-imposed by the visitor rather than the technology. Splitters require visitors to keep the same pace and listen to the same commentaries; it was expected that there would be discussion between companions deciding which section to play or about the interpretation that had been listened to. However, this was rarely observed to occur. The social side of using one’s own device has yet to be fully explored, and there is enormous scope for development.

However, the use of illustrations must be done carefully, as visitors themselves commented they felt that they looked at the guide more than the bridge itself. These categories of images seem to be the most appropriate to illustrate at an archaeological or heritage site: (1) structures that are no longer extant or those in ruins; (2) underground aspects of a site; (3) changes to the usage of a site over time; and (4) the rapidly changing nature of an archaeological excavation.

The necessity of using multiple methods of data collection, ranging from qualitative comments and observations to device-recorded behaviour, has also been shown to reveal far more details about how visitors both use interpretation and their perception of said usage. Additional research would give interpretive designers a better insight into which areas of feedback visitors may exaggerate (e.g. the amount of interpretation accessed) or downplay (e.g. frequency of stopping or pausing the audio commentary).

As technology advances, so too do users’ expectations as to the content. For example, the option for more interactivity was only selected on the Image- and GPSbased guides. Yet with these greater expectations, users have also indicated a willingness to pay more for more complex tours. However, the higher ranking of the category ‘ease of use’ on the Audio-only indicates that simplicity of use is still preferred. Due to the difficulties with screen visibility, attempts were not made to provide in-depth textual interpretation, but further research into its use is necessary to completely explore the range of content available. An international visitor highlighted another use of text, that of providing foreign subtitles. This may be a simple solution if multilingual audio content is unavailable or cost prohibitive.

Content The use of location-based media and digital devices is merely a way of delivering content to the user. It is this content that is of importance and which has the greatest effect on the visitor’s experience. For example, visitors often commented to the author afterwards that they enjoyed the stories related to the bridge, such as Sarah Ann Henley who attempted to commit suicide in the 19th century, but was saved when her large, billowing skirts acted as a parachute to slow her fall. A focus on the

Finally, of extreme interest is that several visitors asked for the tours to be more location based, indicating that visitors are willing to adopt such technology. As with the change in interest to downloading a tour to one’s own 181

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process device, it is likely just a matter of giving users time and opportunity to adapt to new methods of interpretation.

The problem with using an audio commentary in a hightraffic environment has already been discussed, and another issue with urban sites is the safety of the visitor.

Choice As has already been touched upon, visitors tend to contradict one another. It is not possible to create a ‘onesize-fits-all’ method, hence personalisation through digital means, whether location-based or otherwise, is likely to be the way forward. This encompasses not only the approach taken to provide interpretation, whether linear or thematic, positivist or constructionist, but can include options such as greater (or no) interactivity, the ability to get directions to the next point of interest (or not), and even the method use, whether audio or text based. As discussed in the following section, the conclusions drawn from all of the visitors’ studies have been united into a framework that should be considered if locationbased media is to be deployed on a widespread basis. The possible future of the technology is also considered, as well as additional areas of research.

8. Conclusions

Figure 3 The Seven Cs framework.

As a result of the observations, quantitative data, and visitor comments obtained and briefly discussed throughout this paper, a framework termed the ‘Seven Cs’ has been developed to assist in producing a more positive visitor experience when using location-based media, although it can be adapted to other interpretive methods as well. It includes specific questions and aspects of both the technology and the site itself that should be taken into consideration before embarking on mediascape design. These Seven Cs are: (1) context; (2) clarity; (3) choice; (4) control(s); (5) cost; (6) contact; and (7) content.

Users tend to focus on the device at the expense of traffic safety by watching either the images or, if a map is used, the dot representing themselves moving across the landscape. With a greater degree of GPS accuracy it would perhaps be possible to flash the screen or use the vibrating function of a mobile phone as a warning to ensure that visitors were alerted when nearing the edge of the pavement or crossing roads.



It is a cyclical rather than linear design, although ‘context’ serves as a logical starting point. As illustrated in Figure 3, all points within this framework are interrelated, with each factor influencing the design and development of the others. The inner circle is an iterative process of developing a prototype or feature and constantly testing and refining it, as was done in the development of the Suspension Bridge prototypes.



Assess the technical limitations of the location for whichever method of location sensing is desired, e.g. GPS accuracy in an urban environment.



Assess which type of content is best suited to the site.



Create a basic prototype and test the environment.



Design for the environment, working with the heritage that is in or can be seen from the location.

Context Assessing the context in which the interpretation is to be used is vital, and a practical first step before the other factors are considered. Is interpretation using locationbased media suited to the site, or would it be better served by another method? Is it practical to rely on locationsensing technology, whether GPS or another method? How will visitors move through the space? Is it practical to use the desired content, i.e. will it be audible or visible?

C1. CONTEXT RECOMMENDATIONS Observe the location, both how visitors use it and the environment itself, on different days of the week, at different times of the day, and in different weather conditions.

Figure 4 Recommendations for working with the context of a site; the context typically has an impact on the content that can be used. In discussing heritage interpretation in general, David Uzzell (1998, 186) states that, ‘The type of interpretive solutions adopted ought to reflect the particular 182

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites opportunities and constraints inherent in the heritage and its location.’ Although location-sensing technology is rapidly finding a place in museums and locations of historical importance (e.g. Antenna Audio’s use of the infrared X-plorer system at the ss Great Britain), it is felt that open, non-urban locations provide more opportunities for the use of GPS, at least as it stands at present.



Clarity Clarity has two components: (1) the technology (i.e. hardware and software used), and (2) general design (e.g. tour awareness and density of points of interest). Regarding the former, is the device easy to use and can the visitor easily navigate through any options provided? This is a strong case for using the visitor’s personal device, which they already know how to use, rather than relying on separate equipment as was done for the trials discussed. It is also a practical step for smaller sites that do not have the funding to invest in hardware, or the manpower to hand out and collect guides. In discussing Sotto Voce, an audio guide that allows users to eavesdrop on each other’s audio commentary, Rebecca Grinter et al. (2002, 154) highlighted this as one of the issues with using proprietary hardware: ‘Unlike a mobile phone, Sotto Voce is a loaned and not owned technology, so it may never achieve enough use for people to evolve and adapt their behaviours based on extended ‘‘familiarity’’ with the technology.’

C2. CLARITY RECOMMENDATIONS Determine what device is going to be used: personal or proprietary.



Ensure that the device is suited to the context, e.g. the content is audible/visible in all situations and weather if an outdoor site.



The controls of a proprietary device should be intuitive for the user.



If a menu-based design is used, navigation between options or screens should be straightforward.



Make any available options explicit; do not assume that a visitor will listen to the instructions.



Limit the available points of interest in a particular area.



Provide feedback so that users are aware that device is working.

Figure 5 Recommendations for ensuring that visitors find clarity, rather than confusion, when using a device. Choice Both at the Baths and the Suspension Bridge, visitors have shown a preference for having different types of tours available and being able to get additional information if desired. This is where digital guides, and location-based media in particular, could enhance a user’s experience by offering a far greater degree of personalisation than currently available.

As discussed in Context above, traffic noise at the bridge could overwhelm the audio, leading to a negative experience. Sound issues also revolved around the onscreen controls (Control(s) below); using physical controls on the device would likely be more effective as users would then have more confidence that their adjustments were registering on the device, and have a better idea if the sound is turned up (or down) completely, especially if coupled with an on-screen volume gauge.

The use of layering is one such method already in use with traditional audio guides. This was well received at the Roman Baths and Suspension Bridge trials, and past studies such as Bellotti et al. (2002, 37) have concluded in a similar fashion that the response of visitors ‘... highlights the need for layering the available information based on user preferences and needs.’ In describing the audio technology at Plas Mawr, Peter Humphries (2006, 26) concisely summarises the benefits of using layers: ‘Visitors can choose how much of the tour they wish to listen to, according to their particular interests or time available.’

The second aspect of Clarity to be considered is in regards to general design of content and choice: are visitors made aware of the options that are available to them? As discussed below, choice is likely to become a vital component of both location-based media and personal digital interpretation, and therefore visitors must have a clear idea of what their choices entail.

However, layers are just one possible method of giving visitors the choice of the amount of interpretation to access. Digital guides with Wi-Fi connectivity can tap into online resources to provide additional options, allowing those who wish to dig deeper into the interpretation the opportunity to do; by using locationbased technology, visitors can do so easily whilst at the site itself, rather than remembering to further investigate at a later time.

As discussed in Section 5, the density of points of interest must also be considered to ensure that visitors are not overwhelmed by their options. The Suspension Bridge prototypes indicate that a location-based system is better at doing this than a traditional audio guide, likely due to a limited number of points of interest being available at any one time.

Different types of tours can likewise be presented, catering to Richard Prentice’s division (1991, 298) of 183

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process general visitor, school children, and informed visitor (e.g. an in-depth engineering guide as sought by GPS #13). Taking things a step further, the opinions of noted personalities could also be of interest to the visitor, as done with the Bryson at the Baths tour. English Heritage has produced a series of podcasts in this format, with ‘Bill Bailey’s Stonehenge’ proving the most popular of all the podcasts English Heritage provides (November 2010). However, as noted at the Baths in relation to Bill Bryson, such tours are not for everyone, hence the necessity of having multiple options.

Control(s) By giving visitors the choice of how to access interpretation, they have a far greater degree of control over their experience. Such control extends to not only what they listen to and how much, but also the physical controls available. Although basic audio controls were programmed into the Suspension Bridge tours to allow the user to pause or stop at any time, coding limitations in mScape Maker prevented rewind and fast-forward from being used. However, as seen by requests on the Audioonly tour surveys, these features must be included if any digital tour is to be successful.

Visitors can also be given the choice to the approach used to provide interpretation, whether in a linear format, or based on concepts and themes; a ‘just the facts’ positivistic approach, or through constructionist ‘selfinterpretation’ (Copeland 2006, 83). As discussed below (Content) it is the content that is of importance; giving visitors the choice of how to access it, however, can improve their experience and satisfaction.



Features can likewise be personalised on a practical level as the preference of one visitor can often contradict the preference of another. This was made very clear during the testing of the Image-based prototype where one visitor (Image #2) felt that the guide was ‘too device led.’ Using the same guide, the next visitor thought it was too vague and wanted more guidance (Image #3: ‘Let the tour guide me, not me guide the tour.’). In this instance, simply providing users with the choice of having walking directions to the next point of interest (or not) would enhance the experience. Similarly, an on/off option in relation to automatic activation gives users the control necessary to use this feature of location-based technology when, and if, it suits them. Although the public is typically referred to as a single entity, it cannot be forgotten that it is composed of individuals, each with their own interests, preferences, and idiosyncrasies.

C3. CHOICE RECOMMENDATIONS Decide what options visitors are to be provided with: format, amount, and method of presenting content; auto activation on/off; guidance directions on/off, etc.



Although a number of options are possible, keep the context in mind. Certain options may be possible but not practical!



Layers are a simple way to give users the choice of how much content to access.



As discussed in ‘Clarity’, make sure visitors are aware of the options available to them.

Figure 6 Recommendations for providing visitors with choice; this typically influences the design of the content, and is turn influenced by the context of the interpretation. Difficulties with audio controls were also experienced during the prototype testing due to the screen-based nature of the controls, especially as regards the volume. Although sound was an issue due to the site’s context, visitors had no feedback as to whether their selections registered. A physical control button should be used if possible, and a graphical representation of an increase or decrease in volume should be standard, as well as a timeline of the audio track to give the user control over fast-forwarding or rewinding the segment. The latter is only now becoming common in audio guides, but is likely to be expected by the audience due to familiarity with audio devices such as MP3 players.

On a larger scale, users can also be provided with the choice as to which format the interpretation should take, whether through auditory or visual means. Although not quite yet practical for long text, advances in display screen technology make the latter option a very real possibility in the foreseeable future. The emergence of smartphones and downloadable apps has also given visitors the choice of when and where to access interpretation. By quite literally putting interpretive tools into the hands of the visitor, users are able to create their own unique experience in the manner they most prefer. It may seem daunting to a heritage site to provide such a wide-range of options, but a key advantage of the technology is that it is easy to update, and options can be added over time. Indeed, a gradual approach is recommended to allow both users and sites to gain familiarity and confidence in using the technology.

Such familiarity should be kept in mind with regards to the symbols used to denote audio functions. The design of the audio buttons was based on previous studies of user preferences (Fennel 2007), and these indeed seemed to work well during the visitor trials without any confusion. Modern designs of audio guides, however, appear to be moving to a more streamlined and individualised approach; as a result, it is not always possible to identify what a control button signifies (Figure 7). Additional research is necessary to determine to what degree this impacts the visitor’s experience.

184

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites Control also encompasses the pace of the user’s visit; this feature was consistently ranked by visitors as one of the most important features in an audio guide. By providing the options discussed above (Choice), in addition to the ability to easily adjust one’s pace through the use of audio controls (e.g. rewinding to hear a section of interest, rather than replaying the section in its entirety), visitor satisfaction is likely to increase.

which symbols are most frequently used and easily understood.

Cost Writing about the industrial site of New Lanark, James Arnold (1993, 219) is explicit about the connection between interpretive systems and profit: ‘When we introduced interesting media systems to explain particular stories to people, we were able to increase the numbers to 150,000 at more than quintuple the original charges.’

(a)

(c)





A timeline of content playback provides users with finer control over fast-forward/rewinding, and allows them to gauge the amount of content available.

All testing at the bridge was done free of charge to visitors, but they were queried as to whether or not they would pay to download such a tour at a later time. Initial results show that they would, and the price visitors are willing to pay increases with the complexity of the tour. However, most also commented that their willingness to pay is inversely proportionate to the entry fee charged. Such a factor is encouraging news for smaller sites (e.g. unmanned sites in the care of English Heritage; Anderson 2010) that do not have the finances or personnel necessary to invest in proprietary hardware, as it indicates that visitors would be willing to pay to download interpretation through an application. As discussed in Section 3, the introduction of apps has changed not only the delivery of interpretation, but also the way in which visitors must assess the value of said interpretation. Although the surface cost may be less than that of a traditional guidebook, updates, accessibility on multiple devices, and accuracy (or lack thereof) may lead to visitors paying far more than intended. That being said, current app prices are in line with visitors’ responses, i.e. less than £2.00 on average.

(b)

Cost can, however, serve as a limiting factor, preventing visitors from accessing interpretation if they feel it exceeds their predetermined idea of value. For example, the growth of mobile internet usage may see an alteration in the valuation of apps, i.e. if visitors are already paying to access the internet on the go, why should they bother with an app when they can search Google for details about the site? This bears watching to determine whether it has any influence on price and visitor usage of apps.

(d)

In discussing the importance of providing interpretation, Binks et al. (1988, 2-3) comment: ‘Good on-site presentation of the archaeological dig is good for raising awareness, good public relations and good for generating income and support for continued work ... your visitors should go away interested in and understanding the value of digging up the past, and appreciative of the role of the archaeologist ... Heritage is now a considerable marketable commodity. [Emphasis added]’

C4. CONTROL(S) RECOMMENDATIONS The following controls should be included: pause, stop, fast-forward, rewind, play, volume, and help. Avoid using the same keys for multiple functions if possible. All controls should be clearly labelled; investigate other devices as well as literature to determine 185

 

If possible, a physical volume control should be used in tandem with an onscreen graphical representation.

Figure 8 Recommendations to ensure that the visitor feels in control of his or her experience, and can easily control the content on the device.

Figure 7 Audio guides produced by: (a) Acoustiguide; (b) Orpheo; (c) Guide ID; and (d) Dataton. Each has different controls available, as well as different ways of accessing audio controls. 



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Heritage is indeed marketable; how a site is interpreted has become part of the publicity and the attraction. Interpretation is no longer merely a passive way of explaining a site, but is instead being pushed to the forefront of advertising and marketing campaigns. For example, the audio guide at the Roman Baths is used in the site’s publicity leaflets, as well as publicised to visitors upon their arrival in the Reception Hall. In the competition to attract visitors to museum and heritage sites, it is likely that interpretation will become a vital component, whether it includes apps, websites to be accessed post-visit, or features that have yet to be developed. 

dependent on site access, and the use of a familiar piece of equipment should limit hardware problems. However, the lack of platform interoperability at present, e.g. an app programmed for iPhone may not necessarily be available for an Android phone and vice versa, means that visitors may not be able to access interpretation on their device, and staffed sites may not be able to provide support for multiple devices if there is a software glitch.



C5. COST RECOMMENDATIONS If an admission fee is charged, include the cost of using the interpretation in it, rather than charging separately.



If charging solely for the interpretation, £2.00 or less is, at present, a reasonable amount.



Visitors are willing to pay more for a more complex guide.

C6. CONTACT RECOMMENDATIONS Based on the context, decide how visitors are to access the interpretation: through a distribution point or on their own.



If possible, visitors should be able to access the interpretation without needing to plan ahead.



If a proprietary system, staff must be able to provide assistance.

Figure 10 Recommendations for considering how contact is made with the visitor, both in terms of distribution and assistance.

Figure 9 Recommendations for determining cost of interpretation.

Content Finally, it is a cliché, but true that content is king (Morris 1988, 76; Zimmerman 2003, 140; Bath 2006, 172). Or as described in a slightly different way by Tim Copeland (2006, 97), interpretation can use low- or high-tech means ‘as it is the ‘‘message’’ not the ‘‘medium’’ that is fundamental to understanding ...’

Contact The setup of testing at the bridge was straightforward, with the author both distributing and collecting the guides, as well as following the visitors to observe their behaviour and provide help if required. In non-trial situations, it must be considered how contact is made with the user to receive the tour, whether it is done from a set distribution point or through a download to their own device. If the latter, will the download only be available in advance, or can the user obtain it while visiting the site? Perhaps even more importantly, how will the visitor get help if there are problems?

Despite the different methods used in delivering the interpretation, much of the feedback revolved around the actual information imparted. Although not recorded on the questionnaires, many visitors reported to the author afterwards that they enjoyed the stories of people’s involvement in the bridge, such the vignette about Sarah Ann Henley. A mix of these stories plus typical facts was well received, but even more options were desired.

Hardware can either be distributed by the site itself, as is done with traditional audio guides, or a visitor’s personal device can be used. Each has its own advantages and drawbacks that must be taken into consideration. The former is more cost and labour intensive: a location must initially invest in the device and the associated maintenance and upkeep, and staff must be available to distribute and collect the guide, as well as be on hand to provide assistance. As a result of these limitations, interpretation can only be provided during set hours, and typically only at larger archaeological sites and museums. However, by providing visitors with identical devices, one can be assured of users having the same experience, i.e. the content will be displayed in an identical manner. If help is needed, trained staff will be familiar with the device.

All of this relates back to choice, and giving visitors enough information and enough variety to choose from. For example, the prototype guides at the Suspension Bridge used shorter audio segments and the layering of information, both of which received high marks, and there were far fewer complaints about the length of audio segments. Oberlander et al. (1997, 8) confirm this observation in their own studies: ‘Indeed, keeping it short is probably more important in the speech-delivery version of our system ... people don’t want to stand around listening to paragraphs of speech; it is more forgiving to deliver short segments, and leave it up to the user to request more, if they want it.’ However, any method of interpretation is only as good as the content it imparts. Those who develop the interpretative content, whether for distribution by a sign

Conversely, a visitor using his or her own device can theoretically access interpretation anytime, anywhere, 186

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites or a machine, are ultimately responsible for the success or failure of the interpretation. In describing the use of new technology at heritage sites, Brian Bath (2006, 172) highlights this issue: ‘As always, the imaginative use of the potential of the technology is far more important than the technology itself. A well-written guidebook or an excellent personal guided tour is far better than a bad audio or tour virtual model.’ Conversely, you cannot blame the technology for poorly presented interpretation and design decisions. One such example of this is the use of music and sound effects: although audio guide technology makes it possible to use these features, they are often labelled a distraction and unnecessary.

discusses the people involved in the location. Architecture and artefacts are sterile without their makers and consumers.

Rather than creating a large number of points of interest for each topic, use layers if possible.



Use a chime or other signal at the end of an audio file so that the user knows it has finished.



If at the site itself, use images and video sparingly to avoid distraction; ensure that the illustrations chosen add to the visitors’ understanding. Do not show illustrations of what can be seen.



Music and sound effects should only be used if they add to the interpretation. They should not be layered behind spoken commentaries.



In designing content, the three most basic members of the public to consider are children, general visitors, and informed visitors. If a standalone tour for the latter cannot be created, providing more indepth interpretation with layers is suitable.



Whenever possible, ensure that the interpretation

Visitors seldom access interpretation in the way in which the designer intends: ensure that it is logical no matter which order is used, or which choices are provided.

At the time of this study’s inception, it was envisioned that guides relying on location-based technology would follow a similar format to pre-existing audio tours, a format also co-opted by podcasts and existing locationbased tours. However, the growing presence of Wi-Fi may result in a wildly different design; for example, it is not difficult to imagine a smartphone application allowing for the display of a related Wikipedia page based on the user’s location. Specially created ‘tours’ and ‘guides’ are no longer a requirement. In 1993, James Arnold wrote (232) that ‘The art of interpretation does not stand still.’ Ultimately, the methods used to provide interpretation are guaranteed to change and advance. Continued studies into the usability of such technology—whether it involves the incorporation of social media, post-visit interpretive access, augmented reality, or something yet to be invented—are necessary to ensure that visitors receive the interpretation necessary to understand, appreciate, and, most importantly, protect their heritage.

Acknowledgements My most sincere thanks are owed to the many people who assisted in this research: PhD supervisors at the University of Bristol, Dr Kirsten Cater and Professor Mark Horton; Ki in particular provided the necessary nudges to see me through to completion, and helped me navigate the complex maze of research, prototype design, and writing. Katie Smith at the Roman Baths and Mike Rowland at the Clifton Suspension allowed me to carry out research at their respective sites, and were always willing to answer my numerous questions with extreme patience. Junction K granted permission for the use and editing their Suspension Bridge material. Jon Paget provided his voice to the prototype guides; that he also chose to marry me was a nice bonus. A special thank you to the hundreds of members of the public who took the time to share their opinion with a clipboard-wielding American and Angeliki Chrysanthi and Constantinos

187

 



Final thoughts

C7. CONTENT RECOMMENDATIONS If using audio, keep each section as short as possible; 30-45 seconds is ideal, one minute maximum. Use layers to provide additional detail and depth.



Content recommendations for traditional methods of media apply, e.g. make interpretation relevant to the visitor.

Figure 11 Recommendations for the design of interpretive content. The form it takes is reliant on the context of the site.

This paper begins with a poetic description of the aims of heritage interpretation (Tilden 1977, 38), and concludes with one slightly more practical. Ian Shepherd (1993, 175) describes the goals of heritage interpretation as follows, which are not limited by the media used to convey it; all methods of interpretation should be able to achieve these aims. It is a matter of how the content is created in order to fulfil this role, and how the public responds to the content that is of importance, and therefore which determines the success or failure of the interpretation.  an aim to increase enjoyment  an element of discovery  an addition to understanding  answers when questions arise  inspiration/aesthetic awareness  significance of the site; why is it important?  a conservation message 



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Papadopoulos for their patience in waiting for this long overdue article.

Campbell, J. B. 2006. Introduction to remote sensing. London, Taylor and Francis.

References

Cell Phone Culture, 7 November 2005, Massachusetts Institute of Technology. Available at: http://web.mit.edu/ comm-forum/forums/cell_phone_cultu re.htm.

Abowd, G., Atkeson, C., Hong, J., Long, S., Kooper, R., and Pinkerton, M. 1997. Cyberguide: a mobile contextaware tour guide. ACM Wireless Networks 3, 421-433.

Cheverst, K., Davies, N., Mitchell, K., Friday, A., and Efstratiou, C. 2000. Developing a context-aware electronic tourist guide: some issues and experiences, in The proceedings of the SIGCHI conference on Human factors in computing systems 2 (1): 17-24.

Addis, M. 2005. New technologies and cultural consumption – edutainment in born! European Journal of Marketing 39(7/8), 729-736.

‘Cleverly Simple’, The Economist, 1 October 2009.

Anderson, J. 2010. The past in your pocket: mobile media and interactive interpretation. English Heritage’s Research News 13, 17-19.

Copeland, T. 2004. Presenting archaeology to the public, in Merriman, N. (ed.), Public archaeology, Abingdon, Routledge, 132-144.

Aoki, M., Grinter, R. E., Hurst, A., Szymanski, M. H., Thornton, J. D., and Woodruff, A. 2002. Sotto Voce: exploring the interplay of conversation and mobile audio spaces, in Proceedings of the SIGCHI conference on human factors in computing systems 2002 4(1), 431-438.

Copeland, T. 2006. Constructing pasts: interpreting the historic environment, in Hems, A. and Blockley, M. (eds), Heritage interpretation. London, Routledge, 83-95. El-Rabbany, A. 2006. Introduction to GPS: The global positioning system. London, Artech House.

Arnold, J. 1993. The new Lanark story, in Fladmark, J. M. (ed.), Heritage: conservation, interpretation and enterprise. London, Donhead Publishing, 215-234.

Falk, J. and Dierking, L. 1992. The museum experience. Washington, D.C., Whalesback Books.

Bath, B. 2006. The use of new technology in the interpretation of historic landscapes, in Hems, A. and Blockley, M. (eds), Heritage interpretation London, Routledge, 163-172.

Fennel, A. 2007. An investigation into audio device icon preference. Available at: http://www.tiresias.org/research/ reports/icon_preference.html.

Bellotti, F., Berta, R., de Gloria, A., and Margarone, M. 2002. Testing a hypermedia tour guide Pervasive Computing 1(2), 33-41.

Fleck, M., Frid, M., Kindberg, T., O’Brien-Strain, E., Rajani, R., and Spasojevic, M. 2002. From informing to remembering: ubiquitous systems in interactive museums, Pervasive Computing, 13-21.

Binks, G., Dyke, J., and Dagnall, P. 1988. Visitors welcome: a manual on the presentation and interpretation of archaeological excavations. London, HMSO.

Gilman, B.I. 1916. Museum fatigue. Scientific Monthly 12, 62-74. Grinter, R. E., Aoki, M., Hurst, A., Szymanski, M. H., Thornton, J. D., and Woodruff, A. 2002. Revisiting the visit: understanding how technology can shape the museum visit, in The Proceedings of the 2002 ACM conference on computer supported cooperative work, 146-155.

Birney, B. 1986. A comparative study of children’s perceptions and knowledge. Unpublished PhD Dissertation, University of California, Los Angeles. Bitgood, S. 2006. An analysis of visitor circulation: movement patterns and the general value principle, Curator 49(4), 463-475.

Hansmann, U., Merk, L., Nicklous, M.S., and Stober, T. 2003. Pervasive Computing (2nd edition). Berlin, Springer-Verlag.

Borriello, G., Chalmers, M., LaMarca, A., and Nixon, P. 2005. Delivering real-world ubiquitous location systems, Communications of the ACM 48(3), 36-41.

Harter, A., Hopper, A., Steggles, P. Ward, A., and Webster, 2002. The anatomy of a context-aware application, Wireless Networks 8(2-3), 187-197.

Brinkley, K. 2008. Public explore Norwich heritage via interactive bluetooth ‘Totems’, Available at: http://www.culture24.org.uk/science+%26+nature/techno logy/art59825>

Hawkey, R. 2004. Learning with digital technologies in museums, science centres and galleries. FutureLab Series, Report 9.

Brits increasingly wise to smartphones as 10 million+ browse the mobile Web, Nielsen Wire, 16 November 2009.

Hightower, J. and Borriello, G. 2001. A survey and taxonomy of location systems for ubiquitous computing, Technical Report UW-CSE 01-08-03. 188

 

Massung: Visitor Reception to Location-based Interpretation at Archaeological and Heritage Sites Hill, A. 1993. The North East of Scotland Agricultural Heritage Centre: interpretation at Aden, in Fladmark, J. M. (ed.), Heritage: conservation, interpretation and enterprise. London, Donhead Publishing, 203-213.

Morris, M. 1988. Changing perceptions of the past: the Bronze Age – a case study, in Bintliff, J. (ed.) Extracting meaning from the past, Oxford, Oxbow Books, 69-85. Oberlander, J., Mellish, C., O’Donnell, M., and Knott, A. 1997. Exploring a gallery with intelligent labels, in The proceedings of the fourth international conference on hypermedia and interactivity in museums.

Hodder, I. 2008. Multivocality and social archaeology, in Habu, J., Fawcett, C., and Matsunaga, J.M. (eds), Evaluating multiple narratives: beyond nationalist, colonialist, imperialist archaeologies. Springer, 196-200.

O’Hara, K. and Kindberg, T. 2008. Understanding user engagement with bardcoded signs in the ‘‘Coast’’ location-based experience, in The Journal of Locationbased Services 1 (4), 256-273.

Hull, R., Reid, J., and Geelhoed, E. 2002. Creating experiences with wearable computing, Pervasive Computing, IEEE 1(4), 56-61.

Oppermann, R. and Specht, M. 2000. A context-sensitive nomadic exhibition guide, in Thomas, P. and Gellersen, H.-W. (eds), Handheld and ubiquitous computing 2000, 127-142.

Humphries, P. 2006. Heritage interpretation and CADW, in Hems, A. and Blockley, M. (eds), Heritage interpretation, London, Routledge, 71-82. Hunt, V. D., Puglia, A., and Puglia, M. 2007. RFID: A guide to radio frequency identification. Hoboken, NJ, Wiley.

Oppermann, R. and Specht, M. 1999. Adaptive mobile guide for information and learning on demand, in The Proceedings of the Eighth International Conference on Human-Computer Interaction, 642-646.

James, N. 1986. Leaving it to the experts, in Hughes, M. and Rowley, L. (eds), The management and preservation of field monuments, Oxford, Oxford University Department for External Studies, 47-57.

Parkes, S. 2010. ITU sees 5 billion mobile subscriptions globally in 2010. Available at:http://www.itu.int/net/ pressoffice/ press_releases/2010/06.aspx.

Jones, M. 1993. The elusive reality of landscape: concepts and approaches in research, in Fladmark, J. M. (ed.), Heritage: conservation, interpretation and enterprise, London, Donhead Publishing, 17-42.

Pearce, S. 1993. Museum archaeology, in Hunter, J. and Ralson, I. (eds) Archaeological resource management in the UK: an introduction, Stroud, Sutton Publishing, 232242.

Kaplan, E. D. and Hegarty, C. J. 2006. Understanding GPS: principles and applications (2nd edition), London, Artech House.

Pert, T. 2008a. History in your hands: using mobile devices in heritage interpretation. Aberystwyth, Royal Commission on the Ancient and Historical Monuments of Wales.

Kostopoulos, G., Kitsos, P. and Koufopavlou, O. 2008. Pervasive computing security: bluetooth example, in Yan, L., Zhang, Y., Yang, L. T., and Ning, H. (eds), The Internet of things: From RFID to the next-generation pervasive networked systems, London, Auerbach Publications, 257-286.

Pert, T. 2008b. Hand-held heritage: the development of mobile applications by The Royal Commission on The Ancient and Historical Monuments of Wales. Available at: http://www.archimuse.com/mw2008/papers/pert/pert. html>

LaMarca, A. and de Lara, E. 2008. Location systems: an introduction to the technology behind location awareness, Synthesis Lectures on Mobile and Pervasive Computing, Vol. 3(1), 1-122.

Prentice, R. 1991. Measuring the education effectiveness of on-site interpretation designed for tourists: an assessment of student recall from geographical field visits to Kidwelly Castle, Dyfed, Area 23(4), 297-308.

McManus, M. 2000. A Visitor’s guide to the contents and use of guidebooks, in McManus, M. (ed.) Archaeological displays and the public: museology and interpretation, London, Archetype Publications, 165-168.

Proctor, N. 2005. Off base or on target? Pros and cons of wireless and location-aware applications in the museum, ICHIM (Paris).

Melton, A. W. 1935. Problems of installation in museums of art. American Association of Museums Monograph New Series No. 14. Washington, DC, American Association of Museums.

Reid, J., Hull, H., Cater, K., and Fleuriot, C. 2005a. Magic moments in situated mediscapes, in ACM SIGCHI international conference on advances in computer entertainment technology ACE 2005.

Melton, A. W. 1972. Visitor behavior in museums: some early research in environmental design Human Factors 14(5), 393-403.

Reid, J., Geelhoed, E., Hull, R., Cater, K., and Clayton B. 2005b. Parallel worlds: immersion in location-based experiences, in ACM SIGGRAPH 2005, Conference Abstracts and Applications.

189

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Robinson, E. 1928. The behavior of the museum visitor. American Association of Museums Monograph New Series No. 5. Washington, DC, American Association of Museums.

and environmental interpretation: problems prospects, London, The Stationery Office, 185-201.

and

Vlahakis, V., Ioannidis, N., Karigiannis, J., Tsotros, M., Gounaris, M., Stricker, D., Gleue, T., Daehne, P. Almeida, L. 2002. Archeoguide: an augmented reality guide for archaeological sites, Computer Graphics and Applications 22 (5), 52-60.

Rohs, M. and Gfeller, B. 2004. Using camera-equipped mobile phones for interacting with real word objects, in Ferscha, A., Hoertner, H., and Kotsis, G. (eds) Advances in pervasive computing, Austria Computing Society, 265–271.

Walter, T. 1996. From museum to morgue? Electronic guides in Roman Bath, Tourism Management 17 (4), 241245.

Shephard, I. 1993. Explaining the local heritage: the role of a regional archaeologist, in Fladmark, J. M. (ed.), Heritage: conservation, interpretation and enterprise, London, Donhead Publishing, 175-184.

Weiser, M. 1991. The computer for the 21st century, Scientific American 265 (September), 66-75. Yamane, L. and Lorés, J. 2004. Els Vilars: a study of a cultural heritage augmented reality device’, Selection of human-computer interaction related papers of Interacción 2004.

Tilden, F. 1977. Interpreting our heritage (3rd edition). Chapel Hill, NC, University of North Carolina Press. UK leads European countries in smartphone adoption, ComScore, 31 March 2010.

Zheng, P. and Ni, L. M. 2006. Smart phone and next generation mobile computing. London, Elsevier.

Uzzell, D. 1998. Strategic considerations and practical approaches to the evaluation of heritage and environmental interpretation, in Uzzell, D. and Ballantyne, R. (eds), Contemporary issues in heritage

Zimmermann, L. 2003. Presenting the past. Walnut Creek, CA, Altamira Press.

190

 

12 Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology Tom Brughmans Archaeological Computing Research Group, University of Southampton Abstract Facebook currently has over 500 million active users, only six years after its launch in 2004. The social networking website’s viral spread and its direct influence on the everyday lives of its users troubles some and intrigues others. It derives its strength in popularity and influence through its ability to provide a digital medium for social relationships. This paper is not about Facebook at all. Rather, through this analogy the strength of relationships between people becomes apparent most dramatically. Undoubtedly social relationships were as crucial to stimulating human actions in the past as they are in the present. In fact, much of what we do as archaeologists aims at understanding such relationships. But how are they reflected in the material record? And do social network analysis techniques aimed at understanding such relationships help archaeologists understand past social relationships? This paper explores the assumptions and issues involved in applying a social network perspective in archaeology. It argues that the nature of archaeological data makes its application in archaeology fundamentally different from that in the social and behavioural sciences. As a first step to solving the identified issues it will suggest an integrated approach using ego-networks, popular whole-network models, multiple networks and affiliation networks, in an analytical process that goes from method to phenomena and back again. Keywords: Social Network Analysis; Complex Systems; Social Relationships; Archaeological Data Critique; Graph Theory; Archaeological Networks. ________________________________________________________________________________________________ potential of other network-based approaches has been published recently (Brughmans 2010).

1. Introduction Does social network analysis allow archaeologists to understand past social relationships? The social network perspective is based on the assumption that relationships between individuals shape their actions and it offers a set of theories and techniques for understanding human behaviour through relationships between individuals or communities and their affiliations. But can this perspective just be adopted from the social and behavioural sciences by archaeologists and be applied to archaeological data? Does it succeed in explaining the full complexity of past social relationships? This paper aims at surfacing fundamental issues with the archaeological application of social network analysis which have been largely ignored in previous applications (Coward 2010; Graham 2006a; Graham 2006b; Graham 2009; Isaksen 2008; Mizoguchi 2009; Munson and Macri 2009; Terrell 2010a; Terrell 2010b). A number of suggestions will be made as a first step to overcoming these issues.

2. Once upon a time… A short fiction about a network and a politician1 Once upon a time in Rome there was a man Called Marcus Tullius Cicero. He was a great orator and one of the best lawyers in the city. In fact, only one other lawyer was said to be his superior, a man called Quintus Hortensius Hortalus. As Cicero had achieved almost all he could within the boundaries of his profession, he decided to take up politics. It was his lifelong dream to become a Consul of Rome, so Cicero rose to the challenge and signed himself up as a candidate for the coming elections. Consular elections in Rome, as you may know, worked according to a very familiar political principle: the most popular individual gets the job. Popularity, however, is not exempt from another and

 1 This short fiction is very loosely based on Robert Harris’ novel Imperium (2006). It is adapted by the author and does not aim at being historically accurate in any way.

This paper purposefully only covers social network analysis. A general discussion of the archaeological

191



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process they liked to discuss. Armed with this knowledge Cicero gave public talks in some of the most popular meeting places in the north about the issues his new Facebook friends cared about. Every day more and more northerners added Cicero as a friend on Facebook, so that their support in the coming elections was as good as guaranteed.

possibly even more familiar human principle: everything has a price. In Rome votes could be bought, so in the end the richest person would be elected. Our friend Cicero might not have been counted among the wealthiest men in Rome and neither was he a member of the established political families. There was no doubt in his mind that the elections would be the most demanding struggle in his life so far. He did have one advantage however Cicero was quite a popular man. In fact, according to his Facebook profile (Figure 1) he had over one thousand friends. Everyone even remotely familiar with Facebook, or indeed the idea of friendship, will know that that is a large number of friends to have. Among those friends were a couple of very influential and popular men like Titus Pomponius Atticus, who frequented in many different social circles including the highest echelons of Roman aristocracy. Cicero himself was mostly popular with the rural elite as well as with some groups of the Roman plebs.

Cicero decided to spend the last few weeks before the elections in Rome, so his supporters there would not have the feeling he abandoned them. But on his way back from the north he received a disturbing e-mail. Gaius Antonius Hybrida apparently removed Cicero as a friend on Facebook. This was a real blow for Cicero, because Hybrida was one of the only aristocrats that publicly supported Cicero. Indeed, not too long ago our friend successfully defended Hybrida in court when he was accused for his inappropriate lifestyle. Cicero knew all too well that the man was an alcoholic and a brute, but an aristocratic alcoholic and a popular brute would still be a valuable ally for Cicero in the elections. Now Cicero was puzzled why he lost the little support he had from the aristocrats. He decided to send his secretary Marcus Tullius Tiro to Rome ahead of him to find out why he was betrayed by Hybrida. And sure enough the reliable Tiro presented him with valuable information when Cicero arrived in Rome. Apparently four rich and mighty aristocrats, Marcus Licinius Crassus, Gaius Iulius Caesar, Lucius Sergius Catilina and Hybrida, had their eyes on becoming the most powerful men in the empire. Catilina and Hybrida would be their candidates for the consulate and the four of them conspired against all their rivals. As a consequence, Hybrida had to remove Cicero as a friend on Facebook. At hearing the news Cicero was devastated. If filthy rich people like Crassus conspired against him, his chances of becoming a consul of Rome were less than those of a crippled Phoenician basket weaver surviving the ‘rabid wild animals matinee’ Sunday afternoons in the amphitheatre.

Figure 1 Cicero’s Facebook profile. Cicero knew, however, that one thousand friends would not be enough to ensure victory in the coming elections. If he was to have a real chance at actually becoming the next consul of Rome, Cicero would need to distinguish himself in some way from the other candidates. To do this, Cicero did not invest too much effort in the electorate of Rome, with its established political affiliations and corruption scandals that would even disgust the most inhumane of persian kings. Instead he turned his attention to the communities living to the north of the city, who could cast their vote in this year’s election for the very first time. By browsing through some public Facebook profiles Cicero found out that these new voters were very different from the Romans, as they shared different Facebook pages and visited each other’s farms in Farmville and so on. Cicero did not have any friends from these communities himself, however, so he asked some of his Facebook friends who lived closer to them to find out who among the new voters were the most popular and influential people. Cicero visited these people personally and added them as friends on Facebook. By doing this Cicero became part of a totally new and isolated network within the electorate. He explored who were friends with whom and what topics

Cicero knew, however, that the other aristocrats would never allow for all the power in the empire to be concentrated in the hands of the conspirators. As a last resort he decided to turn to his old nemesis Hortensius for help, in a desperate attempt to still get some support from the established political families. At hearing the news of the conspiracy, Hortensius and his aristocratic friends pretended not to be very impressed and sent Cicero home without much hope of any support. The next day Cicero went to the elections a broken man, thinking the only thing he would win that day was public humiliation. As the first results came in, his fears were confirmed. Time after time Catilina and Hybrida were the top ranking candidates, with Cicero dangling somewhere at the bottom-end with the truly hopeless. Just when Cicero wanted to retire to his home and friends to save what was still left of his dignity, Hortensius appeared. His nemesis made the public statement of walking up to Cicero, who was standing on a stage with the other candidates. Without averting his eyes from Cicero, Hortensius reached behind him and was handed his 192



Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology applications have a number of principles in common, as summarized by Wasserman and Faust (1994, 4): x Actors and their actions are viewed as interdependent rather than independent, autonomous units

iPhone 4S by his personal slave. With this powerful device he immediately wrote on his Twitter page ‘Cicero is a cool guy. Just what Rome needs! Vote for him’ (Figure 2). As you may know, Twitter profiles are public and virtually everyone in Rome read Hortensius’ message immediately. Many retweeted the message spreading the word almost instantaneously to nearly every member of the electorate. Then Hortensius removed his friendship with Hybrida and the other conspirators on Facebook, and many of his friends did the same. As a result of Hortensius’ public action all the remaining citizens who could still vote swarmed to support Cicero, turning the tables in his favour. And so it was that Cicero became consul of Rome as a new man without any family or financial support, but thanks to Facebook and Twitter.

x

Relational ties (linkages) between actors are channels for transfer or ‘flow’ of resources (either material or nonmaterial)

x

Network models focusing on individuals view the network structural environment as providing opportunities for or constraints on individual action

x

Network models conceptualize structure (social, economic, political, and so forth) as lasting patterns of relations among actors

These principles make the social network a useful perspective for understanding a diversity of research questions including diffusion and adoption of innovations (Rogers 1979; Valente 1995; Valente 2005), belief systems (Erickson 1988), markets (White 1981), exchange and power (Markovsky et al. 1988) and occupational mobility (Breiger 1981). Its full potential for the archaeological discipline has still to be explored (Brughmans 2010) and the discussion here of issues surrounding the archaeological use of social network analysis can be seen as a step in this direction. Figure 2 Hortensius’ Twitter profile announcing his support to Cicero.

Social network analysis methods are rooted in mathematics (in particular graph theory (Barnes and Harary 1983; Harary 1969; Harary and Norman 1953; Harary et al. 1965), statistical and probability theory, and algebraic models) from which techniques are adopted for identifying, examining and visualising patterns of relationships. Visualization of social data is a crucial component of social network analysis, as it facilitates an intuitive understanding of network concepts (Freeman 2005; Nooy et al. 2005, 14). A graph represents the structure of a network of relationships, while a network consists of a graph and additional information on the vertices or the lines of the graph (Nooy et al. 2005, 6-7). It consists of a set of vertices (also called points or nodes) which represent the smallest units in the analysis, and a set of lines (or ties) between these vertices which represent their relationships.

3. Social network analysis and Cicero Undoubtedly this piece of fiction is not the story we want to write in the history books. Yet by making the analogy with modern social media and exaggerating it to a ridiculous extent we can imagine what the effects are of thinking about past social relationships through modern social network analysis terminology. There are a number of problems related to using social network analysis techniques for understanding social relationships in the past. Underlying these problems is a specifically archaeological issue of the nature of archaeological data and how they reflect social relationships in the past. These issues will be discussed in more detail below. But let us first explore what social network analysis is through our example of Cicero’s rise to power.

Figures 3 to 6 are examples of social network visualisations. They show a minimal abstraction of the evolving fictitious Facebook friendships of the Roman electorate as discussed in the short story above. In these social networks individuals are represented by nodes and the lines between them indicate friendship ties. On each network the location of Cicero is indicated. The nodes are given a number to clearly distinguish the different groups within the Roman electorate mentioned in the story. The new voters living north of Rome are the first group. Cicero and his closest friends are the second group, and his supporters in Rome are the fifth. The aristocrats

Social network analysis is used in the social and behavioural sciences as a set of theories, models and applications that focus on the relationships among social entities, and on the patterns and implications of these relationships. As such it cannot be seen as a single homogeneous method as its name suggests. It is a distinct research perspective within the social and behavioural sciences, however, because social network analysis is based on an assumption of the importance of relationships among interacting units (Wasserman and Faust 1994, 3-4). In addition, social network analysis 193



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process (including Hortensius, Hybrida, Catilina, Crassus and Caesar) are the third and their supporters in Rome are the fourth group. A typical social network analysis of this evolving network would reveal a number of interesting aspects of the structure of the Roman electorate’s fictitious Facebook friendships, as well as the role and position of individuals within this structure. At the start of the story (Figure 3) Cicero was a popular man with friends from different social groups in the city of Rome, including the aristocracy thanks to his connection with Atticus. In this situation, however, Cicero’s structural position is more or less equal to his aristocratic rivals as he is influential to a similar number of people. Cicero has the enormous disadvantage of not being strongly connected with the established aristocratic political families. In this initial network we can also see that the communities to the north of Rome are isolated, i.e. these individuals are not Facebook friends with any of the Roman voters. At this stage neither Cicero nor his rivals enjoy the support of these voters. Cicero thought this might have been the case and decided to invest in these new voters. However, Cicero could not have done this successfully on his own as he did not have friends in these communities. He had to ask some of his friends to identify the most influential people and infiltrate the communities with their help. The result can be seen in Figure 4. The northerners are now connected to the rest of the electorate and Cicero became their ‘bridge’ to communities in the city of Rome. At this stage Cicero’s position is more favourable. Not only does he occupy a very central position in the entire social network, having many friends himself and tying into many and diverse social groups, but he is also highly influential to the social actions and flow of material and immaterial resources between the city population and the northerners.

Figure 4 Minimal social network of fictitious Facebook friendships of the Roman electorate as discussed in the short story, after Cicero became friends with the most influential people of the Northern communities. By losing his friendship with Hybrida, however, Cicero’s position immediately becomes much less favourable, as can be seen in Figure 5. He is still an important gobetween for the northern communities but became much more peripheral in the network as a whole. He has no direct support from the aristocracy and has no influence on their actions, nor on the actions of their followers. The tables turn when Hortensius publicly announces his support to Cicero, however (Figure 6). Cicero is highly central in this network, being directly influential to all social groups within the electorate. He dominates the flow of resources as a necessary link between most groups. Moreover, by losing their friendship with the other aristocrats and their supporters, the conspirators become isolated and lose all the influence they had on anyone outside their own group.

Figure 5 Minimal social network of fictitious Facebook friendships of the Roman electorate as discussed in the short story, after Hybrida removed Cicero as a friend on Facebook. Figure 3 Minimal social network of fictitious Facebook friendships of the Roman electorate as discussed in the short story, at the start of the story.

194



Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology ability to interview individuals in interaction and observe human behaviour directly is an advantage most of the social sciences have over archaeology, until time travel becomes a possibility that is. The added challenge of having to derive such human behaviour indirectly through material remains, and probably most importantly our awareness of this challenge and the necessary critical attitude when confronting it, makes the archaeological use of social network analysis techniques less straightforward from that in other disciplines. So let us have a look at some of the issues with using social network analysis as archaeologists. Firstly, social network analysis methods reduce the complexity of social interaction to a limited number of variables. In the case of our story about Cicero, for example, to its attestation in one medium: fictitious Facebook friendships. Relationships between individuals are shaped by a diverse range of often overlapping factors. Individuals and social phenomena are embedded in a web of social relationships (Granovetter 1985) which are themselves constantly negotiated against a background of multiple contexts or domains (e.g. political, economic, cultural) (Mische and White 1998). In this sense, a social network itself is no more than one attribute of an individual who is also related or affiliated to a number of contexts (Knox et al. 2006, 118; Watts 2003, 114-121). A single network cannot capture this complexity of social relationships, especially in intimate networks which by their very nature are multi-layered (Gamble 1999, 58). In addition, a network’s visualization as a graph contributes in part to this problem. The number of dimensions (or variables) a graph can represent is limited. Indeed, ‘the world is not a graph’ as Knox et al. (2006, 135) rightly put it. Being a potent exploratory tool, however, social network visualizations are quite influential to the way social relationships are interpreted (Freeman 2005). Network analysts try to overcome this issue in part by confronting multiple networks (Koehly and Pattison 2005), exploring them as parts of ‘whole networks’ through popular network models (Barabási and Albert 1999; Watts and Strogatz 1998), or by giving them a fuller cultural foundation (Mische and White 1998). We will discuss the potential of these approaches for the archaeological discipline in more detail below.

Figure 6 Minimal social network of fictitious Facebook friendships of the Roman electorate as discussed in the short story, after Hortensius announced his support to Cicero.

4. Some issues with social network analysis in archaeology From this fictitious example of Cicero’s evolving social networks we can tell that social network analysis allows for a new look at old data, one that focuses on how relationships between individuals or communities influence their actions and the flow of resources. It shows how Cicero’s initial network of friendships, although it was extensive, prohibited him from being elected as he did not have relationships with the right groups of people. However, the networks illustrated how this situation changed rapidly by forging a relationship with Hortensius, a man with many friends and influential within the right social circles. However interesting these conclusions might sound, they are not the results we wish to achieve with our archaeological research efforts. Such statements do not fully explain social relationships between people in the past for a number of reasons discussed below. I should stress that the issues mentioned here are not exclusive to archaeology. Indeed, these and many more problems with social network analysis are widely recognised in the social and behavioural sciences (Knox et al. 2006). Archaeologists are, however, confronted with an additional complicating factor which sheds a whole new light on these issues: the nature of archaeological data. As archaeologists our data consists of the materialised remains of human actions, which by their very nature are fragmentary samples of an unknown whole. Indeed, David Clarke (1973, 17) famously stated that ‘Archaeology … is the discipline with the theory and practice for the recovery of unobservable hominid behaviour patterns from indirect traces in bad samples’. This is not to say that sampling issues are a purely archaeological phenomenon. Data collection and sampling are no less problematic in the social and behavioural sciences and require solid strategies (Frank 2005; Marsden 2005). Collecting data on entire populations and defining the boundaries of such whole networks is considered particularly problematic (Knox et al. 2006, 120-121; Marsden 2005, 9-10). However, the

Although a number of archaeological applications of network analysis are essentially limited to exploring a single network (Graham 2006a; Isaksen 2008; Mizoguchi 2009), some archaeologists and historians do stress the need to explore multiple facets of social relationships through overlapping networks (Munson and Macri 2009; Preiser-Kapeller 2010). Munson and Macri (2009) adopted a network approach to understand changing patterns of social and political interaction in Classic Maya society. They drew upon an extensive epigraphic database to create networks with 101 sites as nodes and 1044 contextually defined place-name phrases or statements as arcs (directed relationships) or loops (relationships from one node to itself). The renormalized degree centralization index (Butts 2006) is used which, as it includes variables for network size and density, allows 195



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Most archaeological applications of network analysis consider it as a set of ideas and methods for analysis rather than an actual social form. In fact, some (e.g. Graham 2006b; Mizoguchi 2009) put a clear focus on network techniques without adequately discussing the implications these techniques entail concerning networks as a social form. How did people in the past consider the boundaries of their social networks and how are they reflected in material remains? When questions like these are not addressed, networks are seen as an explanatory tool and the danger exists that the assumptions inherent in social network techniques will be reflected in the interpretations. As I will argue below, an awareness of a healthy balance between network techniques and phenomena should be explicitly present in archaeological applications of social network analysis.

one to compare networks of different types of relationships. Munson and Macri confronted antagonistic, diplomatic, subordinate and kinship networks. Moreover, they rightly stress that one of the advantages of networkbased approaches is that they do not assume geographical structure to be significant (Munson and Macri 2009, 429). One can ignore relationships based on proximity and choose to explore the structure of different types of relationships, or do both. Munson and Macri did just that. They confronted their networks with the average geographic distance between paired sites for each subnetwork, adding yet another layer to an already complex web of interdependent relationships. Munson and Macri’s pioneering work on multiple archaeological networks is a first step towards confronting the complex nature of social relationships embedded in evolving and diverse contexts. But even if we represent all the archaeological data we have as a series of overlapping networks, we would still not be able to understand this complexity because of the fragmentary nature of our sources and our ignorance of the entire population our sample is derived from. Some aspects of social relationships just do not leave a trace in the archaeological record. Moreover, social network analysis does not help us to clearly identify those aspects our sources might potentially reveal to us. What past social networks we think the data reflects is entirely up to the archaeologist, who is oneself embedded in a complex web of social relationships within a set of contexts.

Thirdly, human actions are limited by a strictly local knowledge of the networks they belong to and influenced by a general ignorance of the social system as a whole (Watts et al. 2002). In our fictitious story for example, Cicero decided to win over new voters because he considered his own social network too small, not because he knew the exact size and structure of his competitors’ networks and the impact his efforts would have on them. Also, Cicero did not have knowledge of the social networks of these new voters. He had to rely on the local knowledge of other individuals to direct his efforts for his campaign. This notion is crucial when we want to understand human actions in a social network from an individual’s point of view, as for the ‘search in networks’ problem for example (for an overview see Watts 2003, 130-161). Imagine one individual is given the task to pass an object on to another individual unknown to him or her (for the original experiment see Korte and Milgram 1970; Milgram 1967; Milgram 1992). What the first person will do is pass the object on to someone he or she believes to be closer (in any conceivable way: geographically, socially, professionally etc.) to the recipient. What the first person will not do, however, is identify the shortest possible path to deliver the message, as this individual does not have the necessary knowledge of the social network as a whole to do this. This means that at every step in the process of moving an object from one part of a social network to another, a decision will be made to act, motived by an individual’s local knowledge (Kleinberg 2000; Watts et al. 2002). This notion can be explored through the well-established distinction between ‘whole network’ and ‘ego network’ methods in social network analysis (Knox et al. 2006, 118; Marsden 2002; Newman 2003; Wellman 1988). The former cover entire populations whilst the latter include only one node (or the ‘ego’), its neighbours and all lines among these selected nodes (Nooy et al. 2005, 145). A number of network analytical techniques have been developed for both ‘whole networks’ (Wasserman and Faust 1994) and ‘egonetworks’ (Marsden 2002).

Secondly, following the argument voiced by Riles (2001) and Knox et al. (2006), recasting social relationships in a network form also reveals the significant methodological issue that the phenomenon we are interested in has essentially become the same things as the technique we use (Riles 2001, 172). In Riles’ (2001) words, it can be turned inside out: ‘the inside of the networks (the social relationships of which it is composed) is at the same time the outside (the representations or visualisation)’ (Knox et al. 2006, 133). By performing a social network analysis we assume that the network is an existing social form which ‘poses methodological dilemmas in relation to the establishment of an analytical position’ according to Knox et al. (2006, 115). Indeed, anthropologists have shown that the network is an ethnographically significant form (Green 2002; Riles 2001; Strathern 1996). However, keeping network-techniques and network-phenomena apart as two separate things is crucial but not easy in practice, given the rigidity (and indeed the institutionalisation as Knox et al. (2006, 115) call it) of established social network analysis methods. The phenomena challenge this rigidity and reveal inherent assumptions social theorists have imposed by using the network metaphor as an explanatory device (Knox et al. 2006, 134). In fact, the use of network analysis as an explanatory tool is limited as I have argued elsewhere (Brughmans 2010, 298). The social network metaphor should challenge questions of social relatedness, but ‘as soon as it stops challenging and starts prescribing, then the productive capacity of the network is diminished’ (Knox et al. 2006, 134).

These two distinct network perspectives allow for topdown as well as bottom-up approaches in archaeology (Coward 2010, 458; Gamble 1999, 33-36; Earl et al. 2011). In light of the nature of archaeological data, 196



Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology behaviour through an archaeological record that is itself created by people. But archaeologists should not assume that the structure of such social implications is examined directly through any type of network analysis. I will not discuss this point any further here, as this paper is concerned with the potential and issues surrounding the archaeological use of social network analysis techniques.

however, this limitation to local knowledge of the network becomes a determining issue in the selection of a specific social network analysis approach. As archaeologists, our data are typically the material residues of individuals’ actions influenced by the local knowledge of the social networks they were embedded in. It can only inform us of parts of social networks. As we are not in a position to explore social relationships directly, but rather through its reflection in material culture, exploring entire social networks as a patchwork of ‘local knowledge’ becomes problematic. Moreover, many social network analysis techniques assume knowledge of the entire social network, like some of the popular centrality measures for example (Everett and Borgatti 2005; Freeman 1979). The selection of quantitative techniques should happen with the nature of the data in mind. Munson and Macri (2008, 426-427) have argued that archaeological data are often not fine-grained enough to identify humans as nodes. Although this might often be the case, and without disregarding the potential for network-based approaches with larger entities (e.g. sites, households, communities, artefact assemblages), I believe archaeologists should rise to this challenge if we are to understand past social networks. The nature of our data demands it. All this does not mean that top-down approaches are useless in archaeology. As Knappett et al. (2008) have shown, confronting hypothetical social networks with the archaeological record is informative and can lead to new ideas on how communities interacted. A combination of top-down and bottom-up approaches is the most promising solution to this issue, as we will argue in more detail below.

5. The problem Social network analysis has been introduced in the archaeological discipline as a potent set of ideas and tools to explore the social relationships between people in the past. The above discussion, however, identified a few issues with this: 1. The full complexity of past social interactions is not reflected in the archaeological record, and social network analysis does not succeed in representing this complexity. 2. The use of social network analysis as an explanatory tool is limited and it implies the danger that the network as a social phenomenon and as an analytical tool is confused. 3. Human actions are based on local knowledge of social networks, which makes the task of deriving entire past social networks from particular material remains problematic. The key problem underlying these issues, and what makes archaeological applications of social network analysis fundamentally different from its use in social and behavioural sciences, is the nature of archaeological data. How does it reflect past social relationships? And can social network analysis actually be used to visualise and explore these relationships, and to ultimately improve our understanding of past social relationships? I argue that it can, as long as we keep the above mentioned issues in mind. To do this, however, a significant modification of social network analysis techniques is required and a specifically archaeological approach should be developed. In the remainder of this paper I will make some suggestions as a first step towards solving this issue.

Lastly, I would like to turn the discussion away from social network analysis for a brief moment and reveal the wealth of potential archaeological use of ‘the “new” science of networks’ as it has been termed by Duncan Watts (2004). This allegedly new field has emerged in recent years as the result of interdisciplinary efforts and a mutual interest in the idea of networks in disciplines like physics, sociology, mathematics, computer science, biology and economics. Of particular interest to archaeologists are those network approaches that, contrary to social network analysis, do not require social entities to be the nodes in the network. From methods to explore relations between physical objects, like the system of linked routers that makes up the internet (Yook et al. 2002), archaeologists can develop new approaches for understanding the many ways in which material culture relates. Although archaeologists like Shawn Graham (2006b; 2009) and Jessica Munson with Martha Macri (2009) have shown the potential of mainstream social network analysis techniques for the archaeological discipline, I would like to stress that our interest in network analysis should not be limited to its social application. Because, as I have argued elsewhere (Brughmans 2010, 282), the danger exists that the insistence on humanizing networks will lead to the misconception that all archaeological network analysis is social network analysis. That most archaeological relationships have social implications is obvious, as archaeologists are concerned with studying past human

6. Thinking beyond the tool: the critical archaeological application of SNA The first issue that needs to be addressed is the second one mentioned above: the use of social network analysis as an explanatory tool is limited and it implies the danger that the network as a social phenomenon and as an analytical tool are confused. At a very basic level, this statement forces us to be thoroughly conscious about the social network analysis techniques we use and how we use them. We do not want the network to become an aim in itself. Rather, it should be considered a way to think about past social relatedness. However, applying a social network analysis approach implies that we think through the metaphor of the network, which is itself a cultural 197



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process where the social context cannot be established, (2) data where the social context can only be established on a macro-level, and (3) data that can be ascribed to individuals. I consider SNA insightful as a research perspective on all three levels but in very different ways. When dealing with the first data type social networks can only be used as hypotheses, as qualitative tools forcing the researcher to think through social relationships until new data becomes available that allows for validation of the hypotheses. In the second case some general statements can be made on the level of group structure drawing on SNA concepts. Only for the third data type, however, will a quantitative social network analysis be able to move beyond mere hypothetical structures (Bergs 2005, 51-52). The possibility of tracing the direct relations of an individual is often a prerequisite for applying any quantitative SNA tools. There are examples of archaeologists being able to reconstruct part of the social networks of individuals. In his work on the Roman brick industry in central Italy, Shawn Graham (2006b; 2009) combined information on brick producing centres, derived from an archaeometrical analysis of clay sources, with names of individuals appearing on brick stamps. He was able to construct a social network of individuals where Domitia Lucilla, mother of Marcus Aurelius, occupied a structurally favourable position through which she was able to control the flow of information in the brick trade (Graham 2006b, 93-114; 2009, 681). This type of data has the potential to explore individuals' position on and knowledge of part of past social networks. Such a local approach is to be encouraged in particular because we will never be informed on the entire population and their relationships, the boundaries of these networks are always artificial creations as a direct consequence of the archaeological record.

construction (Knox et al. 2006, 129). This is an obstacle we cannot avoid and all archaeologists using networkbased techniques should be conscious about. This is an issue most of archaeology has to deal with. We try to understand the past by imposing contemporary classifications and terms to its material reflections. Often such modern constructions receive an aura of having been something very real in the past, like the identification of cultural groups or time periods. This is no different for social networks. Apart from stressing the familiar argument that archaeologists should be aware of their own contexts and modern perspective, I would like to suggest a methodological approach that might help archaeologists deal with disambiguating the social network as a past phenomenon and as a technique. When we continue following Riles’ (2001) and Knox et al.’s (2006) arguments, we might confront this issue by turning ‘the network from the form of analysis to the focus of analysis and back again to turn the network inside out in Riles’ terms, in a self-reflective form of engagement’ (Knox et al. 2006, 134). On a practical level this can be done by ‘rather than begin[ning] with a whole population defined by an organizational boundary, and using network methods to assess how this population is structured, one starts from discursive unities in the form of stories to consider how far they lead to organizational boundaries’ (Knox et al. 2006, 130). As I mentioned above, in archaeology we are only confronted with such particular stories and they should form the basis of our analyses. Archaeological networks can be built from particular aspects of social relationships reflected in material remains. In doing so we allow the archaeological record to direct the creation of networks and to define its own boundaries. From there we need to think explicitly in terms of past social networks, and consider how our archaeological networks inform us of different types of social networks that actually existed in the past. A critical discussion of past social networks based on a recontextualisation of the identified archaeological networks. From this, hypotheses of whole past social networks might emerge which can in turn be submitted to a network analysis. This process of going from method to phenomena and back again forces archaeologists to acknowledge the existence of the network metaphor and their specific use of it. The next issue, of individuals only having local knowledge of social networks, is probably the most challenging to solve, as it puts strict limitations on the scale of past social networks archaeologists are able to analyse. As we mentioned above, the resolution of archaeological data might very often prevent us from getting the necessary information on a select group of individuals (Munson and Macri 2009, 426-427), not mentioning the difficulties we are faced with when trying to find out more than one single aspect of the social relationships of the same group of people. The problem of collecting data on individuals has been stressed by Bergs (2005, 51) for the case of historical sociolinguistics and led him to suggest three basic data types: (1) data

Figure 7 Cicero’s minimal ego-network of fictitious Facebook friendships of the Roman electorate as discussed in the final stage of the short story. The numbers represent the different groups within the electorate. The network clearly shows how Cicero is directly tied into each of these groups. From data like this a true bottom-up approach can follow, where the individual’s point of view is the main focus. I 198



Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology First of all, as many different aspects of past social relationships as possible should be represented as networks. This could be done through the creation of multiple networks as Munson and Macri (2009) illustrated. Metrics to compare such multiple networks should be adopted from the social sciences like those suggested by Koehly and Pattison (2005) or those temporal metrics by Santoro et al. (2011), and critically modified for the archaeological discipline. Like the renormalized degree centralization index used by Munson and Macri (2009), these metrics could be based on scaleinvariable measures like the number of nodes, average path length, clustering coefficient and average degree. However, this approach still leaves two issues unresolved. As I mentioned before, it still leaves a big part of the complexity of social relationships unspoken for. And secondly, it does not include the many contexts that I argued were so influential to shaping social networks. To tackle this second issue, I suggest developing an archaeological method for using affiliation networks. This is a well-established approach in social network analysis where the membership of an organisation or the participation in an event is considered a source of social ties. Affiliation networks are typically represented as two-mode networks, where one group of nodes represents the individuals/communities and another group the organisations. Figures 8, 9 and 10 represent the affiliation network of the Roman electorate in the final stage of the story. The two-mode network (Figure 8) shows how each individual is affiliated with one or more groups of voters. This two-mode network can be divided into a one-mode network representing how these individuals are related to each other based on the similarity of their affiliations (Figure 9) and a one-mode network showing the similarity of groups within the electorate based on how many individuals are affiliated to two or more groups (Figure 10). ‘The affiliation network in fact becomes a substrate on which the actual network of social ties is enacted’ (Watts 2003, 118). In fact, the affiliation network in archaeology can be used to map broad generic or small specific contexts explicitly. Contexts could be broad known social, geographic or political entities. Stratigraphic contexts or typologies could also be used as contexts in affiliation networks. A growing set of metrics to analyse two-mode affiliation networks is being developed (Everett and Borgatti 2005; Faust 2005).

believe ego-network techniques developed by social network analysts are most promising. As mentioned earlier, these ego-networks only discuss the ego, its neighbours and the relationships between them. As an example, consider Cicero’s ego-network in the final stage of the story (Figure 7). In this way, the data we need to collect becomes much smaller and we will be able to consciously treat archaeological data for what it is: material reflections of an individual’s actions motivated by local knowledge of the social networks. It allows us to compare networks of the material evidence with hypothesised partial social (ego-) networks, to remain conscious about the things we believe the archaeological record has to reveal to us. But how about entire systems? What about the top-down approach? It should be stressed that an ego approach and a whole network approach are not mutually exclusive (Marsden 2005). Bentley and Maschner (2001, 38) argue that a complex systems approach does not remove the individual from consideration or dehumanise change. The system implies that it is shaped by local action, just as much as ego’s with local knowledge imply that there is a wider social network. As I argued before, it is problematic to add up parts of social networks to get a better idea of the system. I believe the way we should approach this issue of how to link ego- and whole networks is the approach to complex systems in archaeology as described by Bentley and Maschner (2001, 2003a). Underlying this approach is the familiar idea that the whole is greater than the sum of its parts. The strength of a complex systems approach, as Bentley and Maschner (2003b, 1) rightly argue, lies in its ability to ‘offer a scientific method for bridging the reductionist study of parts … to the constructionist study of the related whole’. As such, it does not aim to understand every individual aspect of a complex system, as in complex systems new properties and new behaviour appear constantly (Anderson 1972, 393). Rather, it specifically considers complexity as being complex. This idea might sound self-evident but its relevance becomes clear in light of the complexity of social relationships mentioned above. A complex systems approach, I believe, might provide a powerful scientific method to make up for the reductionist perspective of ego-networks in social network analysis and confront them with the complex whole of which they are part. In practice, archaeological networks can be confronted with the known features and behaviour of complex systems models (like the scale-free (Barabási and Albert 1999) and small-world (Watts and Strogatz 1998) models, or alternative frameworks for social systems (Geard and Bullock 2007; Geard and Bullock 2010). For overviews see Albert and Barabási 2002; Barabási 2002; Newman et al. 2006; Watts 2003; Watts 2004. For critiques on these models see Mitchell 2009, 253-255; Shalizi 2011).

The first issue of not being able to attest for the full complexity of social relationships is more problematic, however. The multiple layers of social networks and the contexts with which they engage are obviously not discrete entities but have a considerable degree of overlap. This overlap in a sense is reflected through individuals having multiple affiliations, yet it is limited to the size of our dataset. To deal with this issue, Mische and White (1998) argue that ‘as we live in overlapping and multiple networks, we need to focus on the “switching processes” in which we move from one network to another. These are the “public” arenas in which multiple stories coexist and jostle alongside each other’ (Knox et al. 2006, 130). In archaeology this is

Lastly we will discuss the first mentioned issue: The full complexity of past social interactions is not reflected in the archaeological record, and social network analysis does not succeed in representing this complexity. The key thing I argue to do is contextualise. 199



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process easier said than done. Comparing different aspects of past social relationships reflected in different data types already involves plenty of issues. But to explore exactly how these different aspects relate to each other requires data the archaeological record arguably cannot offer and introduces methodological challenges social network analysis cannot help us with. As mentioned above when discussing the potential of a complex systems perspective, hypothetical network models of these ‘switching processes’ could be constructed and tested against our data. But other than hypothesising, I believe that here the role of a social network approach for understanding past social relationships ends.

archaeological data. I argued that if social network analysis is to be of any use in archaeology these issues will need to be addressed and archaeological data critique should be explicitly incorporated in its methodologies.

Figure 9 Minimal one-mode affiliation network of fictitious Facebook friendships of the Roman electorate as discussed in the final stage of the short story. All nodes represent individuals. Numbers and size of nodes represent the number of other individuals they share an affiliation with. Width of lines represents the number of affiliations individuals have in common. Figure 8 Minimal two-mode affiliation network of fictitious Facebook friendships of the Roman electorate as discussed in the final stage of the short story. Numbers 1 to 5 represent the different groups, number 6 are individuals.

7. Conclusion In this paper I have argued that a social network analysis approach holds great potential for understanding aspects of past social relationships. Social network theories, analytical techniques and applications cannot be adopted in the archaeological discipline without question, however. Three issues were revealed that make such an adoption problematic. Firstly, the full complexity of past social interactions is not reflected in the archaeological record, and social network analysis does not succeed in representing this complexity. Secondly, the use of social network analysis as an explanatory tool is limited and it implies the danger that the network as a social phenomenon and as an analytical tool are confused. Thirdly, human actions are based on local knowledge of social networks, which makes the task of deriving entire past social networks from particular material remains problematic. A key problem underlying these three issues, and what makes archaeological applications of social network analysis fundamentally different from its use in the social and behavioural sciences, is the nature of

Figure 10 minimal one-mode affiliation network of fictitious Facebook friendships of the Roman electorate as discussed in the final stage of the short story. All nodes represent affiliations. Numbers and size of nodes represent the number of other affiliations they are connected to through individuals. Width of lines represents the number of individuals these affiliations have in common.

200

 

Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology Barabási, A.-L. and Albert, R. 1999. Emergence of scaling in random networks. Science 286 (5439), 509512.

As a first step to solving these issues I have suggested a number of approaches that I believe are particularly promising for archaeological applications. Firstly, we should turn the network from the form of analysis to the focus of analysis and back again. This process of going from method to phenomena and back again forces archaeologists to acknowledge the existence of the network metaphor and their specific use of it. Secondly, an individual’s perspective with only local knowledge of social networks can be explored through ego-network techniques. A complex systems approach on the other hand, could bridge the gap between ego-networks and the complex whole-networks of which they are part. Such a combination of a bottom-up and top-down approach will allow us to acknowledge the existence of the complex whole as well as the parts, whilst avoiding the whole becoming merely a sum of its parts. Thirdly, we will need to contextualise explicitly by confronting as many archaeological networks as possible, using techniques to compare multiple networks. In addition, these multiple networks can be positioned within affiliation networks that represent the many contexts with which past social relationships, but also the archaeological data and the archaeologists engage.

Barnes, J. and Harary, F. 1983. Graph theory in network analysis. Social Networks 5 (2), 235-244. Bentley, R. A. and Maschner, H. 2001. Stylistic change as a self-organized critical phenomenon: an archaeological study in complexity. Journal of Archaeological Method and Theory 8 (1), 35-66. Bentley, R. A. and Maschner, H. 2003a. Complex systems and archaeology. Salt Lake City, University of Utah Press. Bentley, R. A. and Maschner, H. 2003b. Preface: considering complexity theory in archaeology, in Bentley, R. A. and Maschner, H. (eds), Complex systems and archaeology 1-8. Salt Lake City, University of Utah Press. Bergs, A., 2005. Social networks and historical sociolinguistics. Studies in morphosyntactic variation in the Paston letters (1421-1503). (Topics in English Linguistics 51), Berlin/New York, Mouton De Gruyter.

Such an aggregated approach draws upon a number of network analytical techniques, a necessary evil if we aim to understand past social relationships through a social network perspective. I have argued, however, that we will never be informed about the full complexity of past social relationships. Moreover, social network analysis techniques would not even succeed in understanding this complexity if we would be informed about it. ‘Facebooking the past’ only makes sense when acknowledging and critically addressing the issues involved.

Breiger, R. L. 1981. The social class structure of occupational mobility. American Journal of Sociology 87, 578-611. Brughmans, T. 2010. Connecting the dots: towards archaeological network analysis. Oxford Journal of Archaeology 29 (3), 277-303. Butts, C. 2006. Exact bounds for degree centralization. Social Networks 28 (4), 283-296.

Acknowledgements

Clarke, D. L. 1973. Archaeology: the loss of innocence. Antiquity 47 (185), 6–18.

I would like to thank Prof. Simon Keay, Prof. Dr. Jeroen Poblome and two anonymous reviewers for their valuable comments, which have been incorporated in the final version of this text. I would also like to thank the organisers of the TAG session and editors of this volume for inviting me to contribute and for their time and efforts spent in bringing these papers together.

Coward, F. 2010. Small worlds, material culture and ancient Near Eastern social networks. Proceedings of the British Academy 158, 453-484. Earl, G. P., Isaksen, L., Keay, S., Brughmans, T. and Potts, D. 2011, in press. Computational methods on the Roman Port networks project, in Keay, S. (ed.), Rome, Portus and the Mediterranean. Archaeological Monographs of the British School at Rome. London.

References Albert, R. and Barabási, A.-L. 2002. Statistical mechanics of complex networks. Reviews of Modern Physics 74 (January), 47-97.

Erickson, B. 1988. The relational basis of attitudes, in Wellman, B. and Berkowitz S. D. (eds), Social structures: a network approach, 99-121. Cambridge, Cambridge University Press.

Anderson, W. 1972. More is different. Science 177 (4047), 393-396.

Everett, M. G. and Borgatti, S. 2005. Extending centrality, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis, 57-76. Cambridge, Cambridge University Press.

Barabàsi, A.-L. 2002. Linked: the new science of networks. Cambridge, Massachusetts, Perseus.

201



Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Harary, F. and Norman, R. Z. 1953. Graph theory as a mathematical model in social science. Ann Arbor, University of Michigan, Institute for Social Research.

Faust, K. 2005. Using correspondence analysis for joint displays of affiliation networks, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis, 117-147. Cambridge, Cambridge University Press.

Harary, F., Norman, R. Z., and Cartwright, D. 1965. Structural models: an introduction to the theory of directed graphs. New York, Wiley.

Frank, O. 2005. Network sampling and model fitting, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis, 31-56. Cambridge, Cambridge University Press.

Harris, R. 2006. Imperium. New York, Hutchinson. Isaksen, L. 2008. The application of network analysis to ancient transport geography: a case study of Roman Baetica. Digital Medievalist 4. Available at: http://www.digital medievalist.org/journal/4/isaksen/.

Freeman, L. 1979. Centrality in social networks conceptual clarification. Social Networks 1 (3), 215-239. Freeman, L. C. 2005. Graphic techniques for exploring social network data, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis, 248–268. Cambridge, Cambridge University Press.

Kleinberg, J. M. 2000. Navigation in a small world. Nature 406 (August), 845. Knappett, C., Evans, T. and Rivers, R. 2008. Modelling maritime interaction in the Aegean Bronze Age. Antiquity 82, 1009–1024.

Gamble, C. 1999. The Palaeolithic societies of Europe. Cambridge, Cambridge University Press.

Knox, H., Savage, M. and Harvey, 2006. Social networks and the study of relations: networks as method, metaphor and form. Economy and Society 35 (1), 113-140.

Graham, S. 2006a. Networks, agent-based models and the Antonine itineraries: implications for Roman archaeology. Journal of Mediterranean Archaeology 19 (1), 45-64.

Koehly, L. M. and Pattison, P. 2005. Random graph models for social network analysis: multiple relations or Graham, S. 2006b. EX FIGLINIS, the network dynamics multiple raters, in Carrington, P. J. Scott, J. and of the Tiber valley brick industry in the hinterland of Wasserman, S. (eds), Models and methods in social Rome, British Archaeological Reports International network analysis, 162-191. Cambridge, Cambridge Series 1486. Oxford, BAR Publishing. University Press.

Graham, S. 2009. The space between: the geography of social networks in the Tiber valley, in F. Coarelli and Patterson, H. (eds), Mercator Placidissimus: The Tiber Valley in Antiquity New research in the upper and middle river valley. Rome, Edizioni Quasar.

Korte, C. and Milgram, S. 1970. Acquaintance networks between racial groups - application of the small world method. Journal of personality and social psychology 15 (2), 101. Markovsky, B., Willer, D. and Patton, T. 1988. Power relations in exchange networks. American Sociological Review 53, 220-236.

Geard, N. and Bullock, S. 2007. Milieu and function: toward a multilayer framework for understanding social networks in the Ninth European conference on artificial life; workshop: the emergence of social behaviour. Lisbon.

Marsden, V. 2002. Egocentric and sociocentric measures of network centrality. Social Networks 24 (4), 407–422.

Geard, N. and Bullock, S. 2010. Competition and the dynamics of group affiliation. Advances in Complex Systems, 13 (04), 501-517.

Marsden, V. 2005. Recent developments in network measurement, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis, 8-30. Cambridge, Cambridge University Press.

Granovetter, M. 1985. Economic action and social structure: The problem of embeddedness. American Journal of Sociology 91 (3), 481-510.

Milgram, S. 1967. The small world problem. Psychology Today, 2, p 60-67.

Green, S. 2002. Culture in a network: dykes, webs and women in London and Manchester, in Rapport, N. (ed.), British subjects: an anthropology of Britain, 181-202. Oxford, Berg.

Milgram, S. 1992. The individual in a social world: essays and experiments, 2nd edition. New York, McGraw-Hill.

Harary, F. 1969. Graph theory. Reading (Mass) London, Addison - Wesley.

Mische, A. and White, H. 1998. Between conversation and situation: public switching dynamics across network domains. Social Research 65, 695-724.

202



Brughmans: Facebooking the Past: a Critical Social Network Analysis Approach for Archaeology analysis to simulate, graph, identify, and analyze social and cultural boundaries between communities. The Journal of Island and Coastal Archaeology 5 (1), 3-32.

Mitchell, M. 2009. Complexity: a guided tour. New York, Oxford Univeristy Press. Mizoguchi, K. 2009. Nodes and edges: A network approach to hierarchisation and state formation in Japan. Journal of Anthropological Archaeology 28 (1), 14-26.

Terrell, J. E. 2010b. Social network analysis of the genetic structure of Pacific islanders. Annals of human genetics, 74 (3), 211-232.

Munson, J. L. and Macri, M. J. 2009. Sociopolitical network interactions: a case study of the Classic Maya. Journal of Anthropological Archaeology 28 (4), 424-438.

Valente, T.W. 1995. Network models of the diffusion of innovations. Creskill, NJ, Hampton Press.

Newman, M. et al. 2006. Structure and dynamics of networks. Princeton, Princeton University Press.

Valente, T. W. 2005. Network models and methods for studying the diffusion of innovations, in Carrington, P. J. Scott, J. and Wasserman, S. (eds), Models and methods in social network analysis. Cambridge, Cambridge University Press, 98-116.

Nooy, W. de, Mrvar, A. and Batagelj, V. 2005. Exploratory social network analysis with Pajek. Cambridge - New York, Cambridge University Press.

Wasserman, S. and Faust, K. 1994. Social network analysis: methods and applications. Cambridge, Cambridge University Press.

Preiser-kapeller, J. 2010. Calculating Byzantium ? Social network analysis and complexity sciences as tools for the exploration of Medieval social dynamics (working paper). Available at: http://www.oeaw.ac.at/byzanz/repository/ Preiser_WorkingPapersIV_ComplexCrisis.pdf.

Watts, D. J. 2003. Six Degrees: The Science of a Connected Age. London, Vintage.

Riles, A. 2001. The Network inside Out. Ann Arbor, MI, University of Michigan Press.

Watts, D. J. and Strogatz, S. H. 1998. Collective dynamics of “small-world” networks. Nature 393 (6684), 440-442.

Newman, M. 2003. Ego-centered networks and the ripple effect. Social Networks, 25 (1), 83-95.

Watts, D. J. 2004. The “New” Science of Networks. Annual Review of Sociology 30 (1), 243-270.

Rogers, E. M. 1979. Network analysis of the diffusion of innovations, in Holland, W. and Leinhardt, S. (eds), Perspectives on social network research, 137-164. New York, Academic Press.

Watts, D. J., Dodds, P. S. and Newman, M. E. J. 2002. Identity and search in social networks. Science 296 (5571), 1302-1305.

Santoro, N., Quattrociocchi, W., Flocchini, P., Casteigts, A. and Amblard, F. 2011. Time-varying graphs and Social Network Analysis: temporal indicators and metrics. arXiv:1102.0629v1.

Wellman, B. 1988. Structural analysis: from method and metaphor to theory and substance, in Wellman, B. and Berkowitz, S. D. (eds), Social structures: a network approach, 19-61. Cambridge, Massachusetts, Cambridge University Press.

Shalizi, C. 2011. Power law distributions, 1/f Noise, Long-memory Time Series. Available at: http://cscs. umich.edu/~crshalizi/notebooks/power-laws.html.

White, H. C. 1981. Where do markets come from? American Journal of Sociology 87, 517-547.

Strathern, M. 1996. Cutting the network. Journal of the Royal Anthropological Institute 2 (3), 517–535.

Yook, S. H., Jeong, H. and Barabási, A.-L. 2002. Modeling the Internet’s large-scale topology. Proceedings of the National Academy of Sciences of the United States of America 99 (21), 13382-13386.

Terrell, J. E. 2010a. Language and material culture on the Sepik Coast of Papua New Guinea: using social network

203



Commentary What Lies Beneath: Lifting the Lid on Archaeological Computing Jeremy Huggett Archaeology, School of Humanities, University of Glasgow ________________________________________________________________________________ Keywords: Technology; Tools; McLuhan’s Laws of Media; Tetradic Analysis; GIS

________________________________________________________________________________ in archaeology. These quite properly consider questions of application and the nature of the underlying model as part of the critique of the research results generated by the analysis. More rarely they go ‘beyond the tool’ and consider the wider implications of the application, the constraints it may have imposed, the way the research questions may have been shaped by the tool, how the tool may have structured the research, and so on. Level 3 approaches are rarer still. As the common denominator between the three publications mentioned above and the present volume, I have sought various ways of looking at this subject in my own research, whether it is considering the implications of technological determinism (Huggett 2000), thinking about issues surrounding information handling and processing (Huggett 2004a), or using the metaphor of fetishism as a medium for thinking through some of the characteristics and implications of the technological tools archaeologists use, such as beguilement, disguise, mystery, and gender (Huggett 2004b).

‘We shape our tools and thereafter our tools shape us’ (Culkin 1968, 60). ‘The most important thing about a technology is how it changes people’ (Lenier 2011, 4).

1. Thinking beyond the tool What do we mean when we talk about ‘thinking beyond the tool’ in the context of archaeological computing? The brief for this volume was refreshingly clear and it is worth reiterating it here. The challenge was: 

to discuss the underlying theoretical concepts behind our methodological tools;



to examine the extent to which constraints of these tools alter our perceptions and interpretations about the past;



to investigate future directions from a theoretical perspective.

On the face of it, therefore, aspects of ‘thinking beyond the tool’ are to some extent a characteristic of virtually all reflective computer-using archaeologists - if only because demonstrating the validity of our results is in part dependent on demonstrating a knowledgeable application of the tools we use. But is this sufficient? Focus on the application level risks losing sight of the broader context from which these applications are derived and within which they are used. Information technologies are socially charged: there are issues of control, surveillance, power, politics, order, and structure associated with them - which in turn influence how archaeological knowledge is created, represented, manipulated, modelled, and understood.

Earlier attempts to do likewise have met with mixed success (for example, Cooper and Richards 1985; Lock and Brown 2000; Huggett and Ross 2004a) and, amongst other things, demonstrated that thinking ‘beyond tools’ is not an easy thing to do. In the commentary introducing the 2004 collection, three different levels of approach were identified (Huggett and Ross 2004b): 1. 2. 3.

the specific application, its implementation and use (for example, databases, GIS and viewshed analysis, agent-based modelling); the origins and prospects of larger-scale systems (for example, online National Monuments Records, digital data archives); the broader implications of information technologies within archaeology and how they are integrated into the subject.

To some, this seems essentially irrelevant, either because it is believed to be self-evident or because the implications of the tools are not appreciated. For instance, in response to a paper arguing that computer-using archaeologists had largely ignored the consequences of the new information and communication technologies we have absorbed within our subject (Huggett 2004b), it was suggested that this approach ‘reflects more closely those environments in which the technological contribution is

By far the most frequent are levels 1 and 2, which are typically found throughout the broad range of publications relating to aspects of computer applications 204

 

Huggett: What Lies Beneath: Lifting the Lid on Archeological Computing considered predominant and computer science, which is often the source of financial resources, is confined to the role of a tool, albeit one with wonderful solutions’ (Moscati 2004, 13), and went on to argue that there has always been a clear vision of the effects of computer methodologies represented in the published range of papers and articles. In many respects, this underlined the very issues of beguilement and mystery that my paper sought (and hence presumably failed) to present: this broader approach is seen to be only relevant in circumstances where technology is the driver and the computer is reduced to a mere tool.

In terms of the archaeological technological tree, most computer-using archaeologists can demonstrate that they are already thinking about their tools as they use them; however the top-down approach is largely absent and may be thought to be futile or irrelevant in the face of the bottom-up approaches which are already evident. Why might this be the case? Part of the answer may be found in considering what we mean by ‘thinking beyond the tool’ and how a ‘tool’ is defined.

To others, this approach seems futile and pointless. As archaeologists, we work within a global environment predicated increasingly upon ubiquitous information technologies. Since we cannot avoid this, we cannot change this, cannot step aside from this, and are powerless to influence this, what then is the value or purpose of a broader investigation of the impact of these tools when we should be concentrating on how best to use them to enhance our understanding of the past?

‘Tool’ is a term often encountered in relation to archaeological use of computers - not least in the title of this volume - and the context within which the term is used can be very revealing about attitudes to information technologies. In one sense, computers are referred to as tools alongside trowels, spades, and wheelbarrows. The computer forms part of the general tool-kit used by archaeologists, albeit a tool which functions in a variety of different ways: simultaneously a tool for writing, reading, drawing, mapping, storing, retrieving, communicating, analysing, reconstructing, integrating, and disseminating our data and information. Alternatively (or indeed, at the same time), ‘tool’ can be used in the context of software or a particular application - a database, a GIS, a 3D modeller, each of which will typically contain its own specialist tools for handling, processing, manipulating, and presenting data. ‘Tool’ may therefore refer to the hardware and all it consists of, or specifically to the software which runs on the hardware, or indeed to specific elements of the software itself - the term is frequently used in an ambiguous and fluid manner which gives rise to different interpretations of what it means to think ‘beyond the tool’ and whether or not this is a worthwhile endeavour.

2. An excursus on tools and technology

I propose that the importance of standing back and taking this broader perspective is that, if the tools we use do change us, if they do affect what we do and how we do it, we should be paying attention to this. Culkin’s statement cited at the outset of this paper (commonly erroneously associated with Marshall McLuhan) goes on to suggest that our tools shape the way we organise reality - and so by archaeological extension, how we organise and understand the past. As I’ve suggested elsewhere, through understanding how these technologies operate on us as well as for us, we can seek to ensure that they serve us better in what as archaeologists we already do, and help us initiate new and innovative ways of thinking about the past (Huggett 2004a). Indeed, I have argued that we have a responsibility to do this, since not to make the attempt means that we remain powerless consumers in the face of otherwise autonomous technologies (Huggett 2004b, 89).

This ambiguity in relation to the definition of a ‘tool’ in the context of information technology is not peculiar to archaeology. For example, Feibleman proposes that the design and construction of tools are the consequence of technology, and the subsequent use of the tool is associated with additional technology which determines how the tool is used (1967, 330). This additional technology includes instructions for the operation of the tool, and - crucially - directions and formulae for using it within the context for which it was designed. Although he is writing before the microcomputer revolution, Feibleman is aware of microcircuits (1967, 334), but significantly sees computers in the same light as gyroscopes and thermostats - instructions about the operation of the tool are built into it (1967, 335). This is in some respects reminiscent of Mumford’s classic definition of the essential distinction between a machine and a tool as ‘the degree of independence in the operation from the skill and motive power of the operator: the tool lends itself to manipulation, the machine to automatic action’ (Mumford 1934, 10). Neither definition of tool fits the case of the computer particularly well, but a key distinction is that while Feibleman’s definition places limitations on the function of the tool, Mumford allows

Floridi offers an elegant analogy in the context of the growth of information and the need for a philosophical approach to it. which seems equally relevant in the context of archaeology: ‘Our technological tree has been growing its far-reaching branches much more widely, rapidly, and chaotically than its conceptual, ethical, and cultural roots...The risk is that, like a tree with weak roots, further and healthier growth at the top might be impaired by a fragile foundation at the bottom...while technology keeps growing bottom-up, it is high time we start digging deeper, top-down, in order to expand and reinforce our conceptual understanding of our information age, of its nature, its less visible implications, and its impact on human and environmental welfare, and thus give ourselves a chance to anticipate difficulties, identify opportunities, and resolve problems, conflicts, and dilemmas’ (Floridi 2009, 154).

205

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process the use of the tool to be limited only by the skill and know-how of the operator. On the other hand, Feibleman employs a very broad concept of technology, rather than seeing it simply in more mechanistic terms. In both cases, the tool is more than simply a technical object: it achieves functionality through the skill of the user (Mumford) or the instructions embedded within it (Feibleman). However, the ‘tool’ approach carries within it a tendency to make technology seem neutral - the equivalent of the aphorism that guns do not kill people, people kill people. A tool is neither good nor bad, but is capable of being put to good or bad use. As Mumford originally emphasised, what gives the tool direction is the user: the tool ‘is in no position to give itself ends and is only the means of realising ends provided by human beings’ (Verbeek 2005, 39). Consequently, the tool or technology is not itself at issue, only the way in which it has been implemented or used. This encourages a view of computer-based tools as theoretically neutral, essentially empty vessels into which data are poured, with interpretation only starting as analysis commences. Correspondingly, evaluations of the application of such tools and their results will tend to be limited by a lack of consideration of the social, cultural and philosophical issues associated with their use. Consequently the focus will be on a limited perspective of the tool or application rather than the wider picture.

people learn and internalise rules in such a way that they become essentially instinctive, but may be recoverable by the conscious mind (the improvisation of a jazz musician, for example);



people behave in appropriate ways but this does not involve consciously or sub-consciously ‘knowing’ the rules (for instance, people conform with rules of grammar without being able to articulate them).

Identifying and deconstructing which of these may apply is important in understanding the context of a particular application. Technological activity is, according to Mitcham, ‘that pivotal event in which knowledge and volition unite to bring artefacts into existence or to use them; it is likewise the occasion for artefacts themselves to influence the mind and will’ (1994, 209). Since technological activities include categories such as crafting, designing, manufacturing, and working, they are not strangers to archaeological investigation. The difference, perhaps, is that we are ourselves implicated in these activities which makes it difficult to stand as dispassionate, objective observers - indeed, turning the gaze inwards upon ourselves can quickly become an exercise in navelgazing. However, Ihde argues that to some degree we must ‘go native’ and actually become informed participants, since this enables us to consider and perhaps influence the developmental phases of technologies, as well as already extant philosophies and their effects (2004, 91).

Mitcham (1994, 156ff) provides a broader and more flexible framework for considering technology, building on the definitions of Marx, Heidegger, Ellul, Dessauer and others: technology as knowledge (techniques and know-how), as activity (making and using), and as object (the material technical artefacts themselves). A similar framework was earlier provided by MacKenzie and Wajcman (1985) in which technology comprises artefacts and technical systems, knowledge about these artefacts and systems, and the practices of handling them. Breaking the subject down like this provides a convenient structure for considering the technological computer tools archaeologists use. The categories are not mutually exclusive; they can provide different perspectives on the problem of thinking ‘beyond the tool’.

Technological artefacts should put archaeologists on familiar ground, accustomed as we are to artefact studies and the way that artefacts influence and structure society through human agency. Although technological artefacts can cover a whole realm of objects ranging from clothing and utensils through to machines and automata (Mitcham 1994, 162), we are concerned here with computer hardware and software and the way the availability, presence, and use of these can impact on the practice of archaeology. This goes beyond the physical nature of the artefact itself - for instance, it is more than a question of considering the impact of miniaturisation and mobility on application and use. For example, Kroes and Meijers (2006, 2) argue that technical artefacts have a dual nature - they can be described in complementary ways, neither of which can be subsumed beneath the other. There is the physical object itself, but importantly there are also the intended functions associated with that object (similar to Feibleman’s ‘additional’ technology, for instance). However, there is also a third dimension: unintended functions - the use of the technical artefact in ways which differ from what was originally intended (Carlson et al. 2010, 210). These are at least as important as the intended or designed functionality as they add a degree of unpredictability into the equation: the impacts and influences may be hidden but these are also where serendipity and inspiration can live.

Technological knowledge can constitute the body of knowledge which gives rise to technological artefacts hence the computer as technological artefact is an implementation of computer technology (Carlson et al. 2010, 210). However, it can also be the know-how, rules of thumb, routines, sets of rules, and theories which in combination develop and shape the use of the technical artefact. These may give rise to regularities in use (customary behaviour and application) reinforced by training, teaching, and experience but may also be subverted and circumvented. Nor need they necessarily be evident. For example, Faulkner and Runde (2009, 446) distinguish three ways that rules may contribute causally to the determination of behaviour: 



people follow rules in a deliberate, conscious way (such as a recipe, a manual, etc.);

206

 

Huggett: What Lies Beneath: Lifting the Lid on Archeological Computing So thinking ‘beyond tools’ should involve more than simply accounting for and justifying the use of a particular application. Thinking about tools in terms of technological knowledge, activities, and objects emphasises the range of implicit, explicit, and tacit assumptions and beliefs wrapped within a social, political, and technical environment. Ultimately, these tools do not yet create themselves - we create them, improve them, refine them and ultimately accept or reject them. They augment and scaffold our thought and analysis, and consequently need to be approached in a considered, aware, and knowledgeable manner. With that in mind, to what extent do the papers in this volume make us, as archaeologists, and the IT tools we employ the focus of study?

comment which is developed further in this commentary. He recognises the way in which practitioners from other disciplines draw on examples from archaeology while archaeologists take advantage of new developments in other fields, but the potential implications of this are not developed at this time. His suggestion that modern mass field data capture techniques represent a shift to indiscriminate data capture and consequently shifts interpretation from the point of data collection to a postsurvey activity could represent a change in the kind of technological activities referred to above. However, mass field data capture still requires a range of decisions to be made about aspects of the survey procedure which, even if considered to be essentially practical, nevertheless have potential implications for the interpretative process. He offers the interesting prospect of complex spatial analysis tools becoming as commonplace as word processors and accessible to non-specialists: this raises the question of the extent to which the technological knowledge associated with these tools becomes regularised, tacit, with applications becoming procedural, even rule-based? The image of ‘thinking spaces’ constructed from semantically linked datasets is presented as a democratising tool, enhancing multivocality and allowing data and interpretations to be reviewed and challenged, although this follows a warning about the limitations of linked data, which, like all forms of data, are dependent on the intentions of the originator. As Cripps observes, this raises problems for the downstream consumers of such data, and it has close parallels with the idea of intended and unintended functions of technology which is worthy of further consideration.

3. A commentary on the papers Watterson looks at how the act of assembling and presenting an archaeological reconstruction functions as an interpretative process at every level, from the development and creation of the models through to audience consumption and presentation. In the process, she argues that archaeologists on the whole are not exploiting the interpretative values of creative technologies to their full potential, and virtual reconstructions often appear as an afterthought with little research potential - they act more as a case study for the technology. Although the criticism has been made before, her reflective approach offers the prospect of a less mechanistic approach to reconstruction models. Frankland picks up the theme of public consumption of virtual reconstructions and sets out to test the view that photo-realistic rendering removes the visual cues that the reconstruction is an interpretation and hence increases its authority. Three styles of presentation (hand-drawn sketch, watercolour, and photo-realistic) with three levels of detail were prepared and viewers’ responses evaluated. Archaeological concerns seemed to be largely restricted to specialists - perhaps unsurprisingly the popular vote was generally for photo-realistic presentations and there seems to be a reasonable appreciation of the limits of interpretation. The most obvious question - which of the three styles was ‘better’ (by whatever gauge) - is not asked directly, only in conjunction with the three levels of detail, which perhaps suggests an area for further consideration. The analysis provides a welcome consideration of how we present reconstructions and an evaluation of the extent to which archaeological assumptions about public presentation methods are borne out.

Katsianis turns to the use of GIS for archaeological excavations, suggesting that digital technologies have really contributed successfully only to better management of excavation archives rather than to the collection and processing of data on site (for an alternative perspective, see, for instance, Warwick et al. 2009). In a review of excavation as knowledge production, digital information systems in excavation, and knowledge production practices in excavation, a range of issues are flagged including the ‘distancing’ effect of technology, changing techniques (such as the shift away from excavation diaries), and the quality of the excavation record and the consequences for subsequent analysis. All these have implications for both excavation practice and the use of digital technologies, and would be worthy of more detailed consideration than is possible here. He presents a conceptual model based on the ‘What’, ‘Where’, ‘When’ triptych which is already used across a variety of different application areas (including archaeological archives). Some of the issues raised about knowledge representation have been addressed by, for example, LP Archaeology’s Archaeological Recording Kit (ARK)1, as used by Dufton and Fenwick elsewhere in this volume. For example, in the ‘What’ discussion, the question of new entities arising that cannot be integrated into predefined categories can be handled in ARK through its ‘fragments’ model. The

Cripps responds directly to the editors’ challenge, looking at a range of developments in archaeological spatial technologies in order to show how archaeological theory and technological practice can be reflexively related. He points to the way that archaeology has been quick to take up new technologies and techniques and suggests that this can sometimes be seen as neutral but may be problematic in other cases - a tantalising

                                                             1

207

 

http://ark.lparchaeology.com/

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process ‘Where’ discussion presents examples of 3D grids and brep models of excavation deposits, but there is no reference to the recording limitations that he refers to earlier which could have significant consequences for the resulting models. Clearly, the final implementation in ArcScene has some limitations in terms of its presentational and 3D capabilities, but the potential is clearly demonstrated. A detailed consideration of value would be worthwhile, especially given the conclusion that ‘the effective introduction of new technologies in excavation is directly related to their role in reassessing and improving the ways we interact with reality to create knowledge about the past’.

properties alone would be to attempt to record ‘everything’. Wu and Lock focus on the spatial construction of social relations, examining the relationship between the spatial and the social, applying Ingold’s ‘wayfaring’ theory to demonstrate how social relations may be influenced by human agents moving around a settlement. The model incorporates access diagrams and space syntax as the basis for developing a meshwork of social relations. Least cost path analysis is used to reconstruct the pathways from a household which are then accumulated together to highlight the pathway density, with high density pathways suggesting more social interactions and vice versa. It is an elegant approach, and in terms of ‘thinking beyond the tool’ the paper is more a case of ‘having thought beyond the tool’, presenting what is effectively a new tool. Wu and Lock suggest that the focus of many spatial technologies is on location and order, and rarely are they used to investigate how people interact with the surrounding spatial structure, although much agent-based modelling seeks to do precisely that.

González-Pérez presents an approach to the description of archaeological entities that avoids classification as an a priori mechanism, using instead the entity’s properties as units of description. He argues that this avoids category bias and that classification can still be subsequently applied if desired. The discussion of inflexibility of category systems, in that they are incapable of recognising new variants and hence wrongly force entities into inappropriate categories is perfectly valid, and could be usefully compared with the capabilities of the ARK system referred to above. A typeless information modelling approach is outlined, which moves directly from the entity itself to identifying its properties, omitting its classification. The question therefore becomes which relevant properties define the entity rather what the entity actually is in the first place, and if required sets of properties can be linked subsequently to a classification. González-Pérez recognises that the determination of properties and the description of their values is a complex mix of cognitive processes, perhaps omitting that these properties do not necessarily arise naturally in the absence of categorisation. The determination of the properties of an object will, at least in part, be derived from past experience, a priori knowledge, comparison with others, and hence the typeless approach is not as truly objective as it might appear. In other words, some aspects of category will tend to be captured within the definition of properties. Whether we can really avoid classification like this remains questionable - arguably consciously or subconsciously we classify or categorise what we see, identifying an object for what (we think) it is. Only when it is not recognised is recourse made to description in terms of properties alone, and even then these may arise through comparison with something which is known. González-Pérez recognises the problem that the cognitive processes employed in constructing properties are subject to bias to the same extent as those involved in categorisation, arguing that the problem is less since properties are atomised relative to category and do not determine the structure of the information in the way that categorisation does. However true this might be, there remains the issue that we categorise and identify things all the time and to do otherwise goes against natural thought processes and risks making early categorisation implicit rather than explicit. Ultimately, it would seem that the only way to record objects reliably through

Pethen looks to develop an approach for integrating GIS and phenomenology in the analysis of ancient Egyptian ritual landscapes and mining sites. In a discussion of the relationship between GIS and phenomenological approaches to landscape, she suggests that the deterministic perspective of GIS is more to do with the way in which it is presented: a tendency for archaeological research questions to be used as GIS ‘case studies’, hence suggesting to phenomenologists who are predisposed to be distrustful of GIS that the technology is more important than the archaeological question. Furthermore, limitations in understanding landscape tend to be addressed by seeking more technological solutions as is demonstrated here by Wu and Lock in a different context. She recognises the limitations of phenomenology and argues that both GIS and phenomenology require additional material to explore visual experience fully and to grapple with ancient perceptions of contemporary worlds. The method described relies on the presence of textual material to provide information on ancient attitudes and experience of landscape to ensure the work is theoretically sound - in this respect, what is being developed sounds not too dissimilar to the kind of historical GIS which would be recognised by GIS-using historians. The paper is explicitly a work in progress and hence steps back from the point of actual application which means that so far the strength of the argument remains untested, although one might suppose that the integration of textual sources within historical GIS more generally would provide some room for optimism. Verhagen and Jeneson summarise an attempt to predict the course of a stretch of Roman road using least cost path analysis to find the optimal connections between two or more locations based on distance and effort. They suggest that in most cases it is hard to judge whether these models are good at predicting past routes since the physical evidence is usually limited. Indeed, in a detailed and introspective discussion, the result generated 208

 

Huggett: What Lies Beneath: Lifting the Lid on Archeological Computing conforms broadly to expectations although they express a lack of confidence in the results. Nevertheless, they see the positive benefits of attempting to model in terms of comparing different outcomes. Verhagen and Jeneson explicitly address ‘thinking beyond the tool’ although they felt that they were forced to think more about the tool than beyond it. They suggest we are still lacking the appropriate ‘spatial language’ and accompanying toolboxes, and feel that more energy should be invested in developing software tools that will allow us to more easily compare different theoretical perspectives. The emphasis is therefore once more on seeking technological solutions for archaeological problems.

Massung examines the response of visitors to locationbased media presentations at heritage sites. She recognises that prior research has typically concentrated on the capabilities and implementation of the technology, and instead focuses on whether there is actual visitor demand for such interpretations and what might be necessary to move the experience beyond experimental novelty. Questionnaire-based analysis at Bath’s Roman Baths (in relation to existing audio tours) and a prototype location-based design at the Clifton Suspension Bridge showed that there was no difference between user preferences in terms of method of delivery. Although location-based media were not rated as a significant improvement in the delivery of information, neither were they significantly worse. Given the increasing use of location-aware smartphones, this is an interesting demonstration that new technological solutions will not necessarily be preferred, perhaps until they become more common elsewhere. Indeed, as Massung observes, the increased availability of mobile internet may mean visitors no longer need specially created ‘tours’ and ‘guides’, which would demonstrate an unanticipated outcome of the technology.

Fiz, Subias and Cuesta outline an approach to modelling the landscape of Oxyrhynchus using a range of different sources. They describe the detailed collection and assembly of a variety of textual, cartographic, and satellite datasets in order to reconstruct the nineteenthcentury landscape around ancient Oxyrhnchus in order to attempt to identify earlier features. An essentially descriptive account, this is an interesting exercise in source analysis and methodology and their results underline the importance of ground-truthing data.

Brughmans discusses the application of social network analysis in archaeology. In the process, he highlights a range of issues including the challenge of deriving individual human behaviour indirectly through material remains, the reduction of social interactions to a limited number of variables, the inherent tautology in which the phenomenon under study becomes the technique used to study it, and the way that the methodology generally assumes knowledge of the whole network when reality limits this to local awareness of a restricted network. He proposes some archaeological solutions to these problems and argues that social network analysis holds considerable potential for understanding aspects of past societies, although a concrete archaeological example of the implementation of the tool would be welcome. Indeed, ‘thinking beyond the tool’ in some respects relates to ways of improving the tool in terms of dealing with its potential archaeological use.

Dufton and Fenwick see the development of web-based technologies as a way of integrating and disseminating burial and funerary data, as well as providing a tool for complex and spatial chronological analyses. Amongst other things, they argue that the standardisation of recording methodologies has created a significant gap between data and interpretation in site practice - as they say, restricting the contribution of the excavator to a series of prompted pro-forma fields often also has the unintended effect of denying them the interpretative voice. In many respects, therefore, their contribution aligns with those of Katsianis and González-Pérez in that they are seeking ways of modelling and structuring complex data. Their solution is found in the application of LP Archaeology’s Archaeological Recording Kit, which employs a hybrid Entity-Attribute-Value data model grouping different categories of data ‘fragments’ (text strings, images, dates, actions, plans, locations, etc.) under a primary record identifier within a web-based environment. Given the other contributors elsewhere in this volume who share their search for greater flexibility in data representation, this model could have benefitted from a more detailed explanation of the methodology. Although Dufton and Fenwick note that other systems are under development, there is a high-profile alternative that is already highly developed. The Integrated Archaeological Data Base (IADB) has been in use for many years and developed as a comprehensive web-based excavation recording and post-excavation processing system through the Silchester VERA project (for instance, see Warwick et al. 2009 and example outcomes in Clarke et al. 2007). It would be interesting to see a comparison at different levels between IADB/VERA and ARK, since they represent two of the most developed systems in general archaeological use.

4. Digging deeper Robert Zemeckis’ thriller/horror film What Lies Beneath received a mixed critical reception on its release in 2000: Empire magazine rather magnificently described it as ‘... an enjoyably giddy ride, certainly, but once you’re back from the edge of your seat, you realise most of the creaks and groans are from the decomposing script’2. This collection of papers may provide a giddy ride but the scripts are rather better than the film in this case! However, revisiting the original objectives and considering the success of ‘thinking beyond the tool’, it is fair to say that the results are mixed. In many cases we tread familiar ground in the sense of reviews of specific applications and their implementation and use - only a

                                                             2

http://www.empireonline.com/reviews/ReviewComplete. asp?FID=6237 209

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process few go beyond to consider the wider implications, constraints, effects, and impacts of these tools. Novelty of application and new technological solutions, commendable though they may be, are only part of ‘thinking beyond the tool’. So how might we proceed? How can we break out of this cycle that has been evident for twenty-five years or more, in which we review our tools within the context of their application, but rarely look beyond them?

emphasises, what this demonstrates is that the four laws rarely generate singular effects - radio cannot be reduced to the four impacts identified above: we can also see radio as making obsolete face-to-face conversation as well as the wiring and cables of the telephone and telegraph, for example. In the same way, depending on the starting point the computer can seen as the outcome of the reversal of the television, the book, the film theatre, and so on. Consequently, there is greater depth and complexity to the application of these heuristics than might at first be apparent. Levinson points to a resemblance with Hegel’s dialectic (McLuhan 1977, 173, Levinson 1999, 192-3) in which a thesis or position is countered with an antithesis or opposite idea, with the best of both being brought together in a synthesis. Where McLuhan departs is in his use of retrieval, which provides a historical basis for predicting future impacts. In the process, Hegel’s triad is developed into a tetrad by McLuhan (Figure 1).

To reiterate Floridi’s argument quoted earlier, we need to start digging deeper if we are to gain a greater understanding of the tools we use in the search for understanding the past. One approach would be to draw upon the kind of framework discussed above for thinking about technology. However, while it provides a useful means of considering technological knowledge, activities, and artefacts, it is perhaps insufficiently problem-oriented to help drive this question forward, especially if we are seeking to break out of a cycle in which these bigger issues are consistently not addressed. Of course, the risk in doing so is that we sail off onto a sea of generalisations about society and lose sight of the stuff of archaeology, in the process laying ourselves open to accusations of irrelevance. Somehow we need a means of digger deeper into the wider nature of our tools, tacit and implicit, while at the same time experience suggests we are most comfortable thinking about our tools within the immediate context of their applications. One way of developing a greater critical scrutiny is to consider Marshall McLuhan’s Laws of Media (presented posthumously in McLuhan and McLuhan 1988, but initially presented in McLuhan 1975, 1977) as a means of understanding the effects of our technological artefacts on ourselves and our practice. These are not laws as such McLuhan saw them as exploratory tools or ‘probes’ that provide insights into the effects of a technology: they are more heuristic device than scientific method. According to McLuhan (1977, 175), there are four ‘laws’ or observations on the impacts of human artefacts on us: 

Amplification: what aspects of human function does it enhance or amplify?



Obsolescence: what does it eclipse or supersede that had previously been extensively used?



Retrieval: what was previously obsolescent but now comes back into use?



Reversal: what does it reverse or flip into when developed to its full potential?

The tetrad model brings out a series of inherent oppositions within the laws: as McLuhan observes, ‘Retrieval is to Obsolescence as Amplification is to Reversal - and - Retrieval is to Amplification as Obsolescence is to Reversal’ (1977, 177). Understanding what a tool enhances is increased by an appreciation of what it retrieves from the past; perceiving what a tool supersedes is heightened by thinking about what it reintroduces and builds upon.

These ‘laws’ effectively captured and summarised all of McLuhan’s earlier work and insights (Grosswiler 1996; Levinson 1999, 187). For example, using these laws, radio can be seen as enhancing access to mass audiences, it makes print obsolescent, it retrieves the town crier who had been largely obsolesced by print, and, when pushed to its technological limits, it flips into the audiovisual medium of television. As Levinson (1999, 188)

This is only a very brief outline of McLuhan’s laws and it is not claimed here that their application will create some kind of ‘truth’ - indeed, as Tyler (2008) emphasises, the laws of media do not lend themselves to falsifiability since illustrative examples can be readily stretched to fit. McLuhan himself saw the laws simply as a means of investigation, and it is in this light that they are presented here. McLuhan claimed that the laws represent ‘an order

Figure 22Thinking about tools using McLuhan’s tetrad.

210

 

Huggett: What Lies Beneath: Lifting the Lid on Archeological Computing of thought and experience’ (1975, 75), while Sui and Goodchild observe that ‘As an exploratory probe resting on a set of questions, instead of a bounded theory, the tetrad will facilitate our simultaneous understanding and integral awareness’ (2003, 10). The proposal here, therefore, is that answering these four questions aids in bringing the kind of critical scrutiny that is required in ‘thinking beyond the tool’.

undertaken with a simple query to a database, followed by some manipulating and massaging to get the data into the desired form. More controversially, perhaps, Sui and Goodchild suggest that some analytical skills have also become obsolete through the use of GIS - the black box effect of button-clicking means that complex tasks can be undertaken with relatively little understanding of what is actually taking place – ‘one of the consequences of access to powerful GIS tools may be a greater interest in doing the thing right rather than in doing the right thing’ (2003, 11).

5. Thinking with tetrads Although not widely applied beyond McLuhanite circles, these laws have been used as a means of addressing varied fields ranging from digital game theory (Tyler 2008), blogging (MacDougall 2005), hypertext (Moulthrop 1993), location-based services (Sui 2004), and social GIS (Sui and Goodchild 2003). By way of an example, the following discussion takes the tetradic analysis of GIS by Sui and Goodchild (2003) as a starting point, whilst applying an archaeological perspective.

GIS retrieve Classically GIS have been associated with a resurgence of processual and deterministic approaches in archaeology, revisiting the quantitative and scientific methods of the 1960s and 1970s and hence were, and in some respects still are, locked into methodologies from the past (see also Pethen, this volume). Additionally, Sui and Goodchild point to the use of GIS as bringing a reemphasis on narrative and rhetoric - the ease of creation and availability of maps and images provide more than just visual cues since behind them are stories to be told: they have become ‘rhetorical devices to create meaning and discourse’ (2003, 12).

GIS enhance The introduction and use of GIS within archaeology has transformed our ability to handle locational data and in the process influenced everything from the management of excavation data through to National Monument Records. GIS make the integration of different categories of data feasible, even if collected under different spatial systems, and in the conjunction of these different data new information can be revealed. The business of creating maps has been made more accessible, even democratised. GIS also bring with them access to a host of different spatial and statistical analytical tools, placing a varied and complex toolbox at our fingertips and providing the facilities to expand the tool-set by developing new instruments for archaeological application. In many respects, the capability of GIS to incorporate and manipulate the range of data can be seen as a way of helping us handle the tide of information that threatens to swamp us. Ultimately, one might claim, with some justification, that GIS have been implicated in the transformation of the subject itself.

GIS reverse into The way GIS model data imposes constraints and potentially excludes different representations of knowledge. Lock and Harris (2000, xvii) criticised the spatial determinism imposed by the requirements of points, lines, polygons, and pixels, for example, and complex data are required to be partitioned into layers. Important considerations such as positional uncertainty, boundary uncertainty, thematic uncertainty, and temporal and cognitive representations are handled with only limited degrees of success. Despite this, as many have commented, the outputs of GIS - like those of 3D modelling - can carry connotations of truth or fact, generating a false sense of what is real. Consequently there is a risk that ‘computerisation of the natural and cultural may inevitably lead to the naturalisation and culturalisation of the computerised’ (Sui and Goodchild 2003, 12) and we increasingly become trapped in particular modes of knowing, separated from the real world as well as the world of the past that we seek to access.

GIS make obsolete GIS have made traditional map-making almost redundant - we rarely spend time preparing maps at drawing tables with permatrace, while letraset and leaky or blocked rotring pens are nostalgic memories. (Indeed, coincidentally, I have just converted my own traditional drafting table into a computer desk as a diversion whilst writing this paper). The widespread availability of digital data in repositories and archives makes it less necessary to seek out the data personally - Sui and Goodchild go further and point to a decline in traditional field techniques, reducing the current collection of primary data since so many secondary datasets are available (2003, 11), though this is perhaps rather less true of archaeology. Certainly data collection tasks that would previously have taken weeks or months in libraries trawling through journals and reports can now often be

Doubtless there is room for discussion about what is and is not included - even which heading certain statements appear under - and it is clear there is no single answer to each question, but that is the point of the exercise. It also becomes apparent that, despite McLuhan’s claim that all four laws apply simultaneously, it is difficult to respond fully to the law of reversal since we cannot yet know how things will develop in the future - we can only really comment on possible trajectories while at the same time attempting to avoid excessively dystopian overtones.

211

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Having characterised a response to the four laws, we can then consider examples of the oppositions between them.

tools and, at the same time, helps to understand those effects.

Retrieval: Obsolescence

6. Conclusions

An example here is the retrieval of processual, quantitative techniques versus the obsolescence of analytical skills (the development of the black box). As discussed elsewhere (Huggett 2004a), the conjunction of systems theory and the New Archaeology in the 1970s encouraged a mechanistic view of the world as represented most clearly in the computers which required a formalised, algorithmic, definition of tasks and data in order to facilitate the analyses. Increasingly userfriendliness, ease of data input, push-button applications, and the guarantee of (usually) attractive outputs can lead inexorably to a black box approach in which there is little understanding of the underlying systems required in order to use the tool, and which may in turn lead to invalid or inappropriate outcomes going unrecognised.

The GIS case study summarised here focuses on a tool in the sense of a single category of software used by archaeologists, and others could equally well be presented - databases or 3D modellers, for example. The same technique can also be applied to tools within such software packages (for example, constructing digital elevation models, or generating viewshed analyses) and to hardware applications (for example, the use of on-site tablet computers for recording, laser scanning for survey). For instance, using a tablet computer onsite can be seen to enhance through simplified interaction, and all data captured is born digital with potentially instant automatic data exchange and communication; it would obsolesce the laptop computer, the planning board, the context file and the day book; it would retrieve the drawing slate, the chalkboard, even the clay or stone tablet; and it would reverse through the risk of diversion - surreptitious use of Twitter and Facebook, games of Angry Birds, and so on. It is left to the reader to fill in the gaps or extend the example further, rather than labour the point that the technique is applicable at all levels and to all types of tool.

Enhance: Reversal An obvious opposition is on the one hand the benefits of data management and manipulation, but on the other hand, the constraints imposed as a consequence of the data models applied. GIS facilitate handling of multiple large complex datasets, but they do so by requiring that data to be presented in particular ways. Consequently the application of computers in archaeology is predicated on the properties of the digital model (Lock 1995, 14) and that digital model is itself predicated on the properties of computers.

At the outset of this paper, it was argued that, although archaeologists are capable of thinking about their tools, and frequently do so in the context of specific applications or case studies (as exemplified by many of the cogent contributions in this volume), we seem to find it more difficult to think beyond the tool in order to consider the ways in which these tools impact upon us and our practice, shape our perceptions and alter our interpretations. One means of overcoming this limitation is to seek inspiration outside of archaeology itself and to recognise that other fields exist for the very purpose of considering the development, effects, and implications of new technologies. This in itself is not a novel proposal: for example, Denning (2000) highlights the potential of History and Philosophy of Science and Science and Technology Studies as sources of provocation for archaeologists. At the same time, however, it has to be recognised that it is possible to go too far in this respect to lose sight of the archaeological objective. It is also possible to be overly utopian or dystopian in considering the impact and potential of these technologies and indeed utopianism is a frequent characteristic of archaeological discussions of technological tools. However, one of the advantages of the approach outlined here is the way in which utopian/dystopian perspectives are tempered through the application of these laws - a potential for utopianism in discussing enhancement and dystopianism in discussing reversal is balanced to a degree by consideration of retrieval and obsolescence.

Retrieval: Enhance Here the re-emphasis on storytelling and the use of imaging as rhetorical devices can be set against the enhancement of access to data and the ease with which it can be mapped and presented. GIS can arguably refocus away from the limitations of manual map-making and facilitate the communication of transformations and change through the transformation and changes between multiple images of data. Obsolescence: Reversal The obsolescence of traditional cartography versus the limitations of the means of representation exchanges the restrictions of manual illustration for the flexibility of computer-generated images, but in the process reveals the limits of representation which fall short of what we intend or require. In a similar manner, the obsolescence of lengthy and complex data collection processes as a result of the increased availability of digital data can be contrasted with the implications raised by increasing use and reliance on secondary computerised datasets which may be far removed from their original context of creation and use.

No grand claims are made for the approach proposed here; the argument is simply that by applying these questions, or questions like them, it is possible to break

Considering oppositions like these draws to the surface often paradoxical or contradictory effects of using the

212

 

Huggett: What Lies Beneath: Lifting the Lid on Archeological Computing the mould which has been evident for the past twenty-five years in terms of published contributions that have sought to address this issue. In short, thinking outside the box can help us think beyond the tool.

Huggett, J. and Ross, S. (eds) 2004a. Archaeological informatics: beyond technology. Special issue, Internet Archaeology 15. Available at: http://intarch.ac.uk/journal/ issue15/index.html

References

Huggett, J. and Ross, S. 2004b. Introduction. Internet Archaeology 15. Available at: http://intarch.ac.uk/ journal/issue15/inf_index.html

Carlsen, H., Dreborg, K., Godman, M., Hansson, S., Johansson, L. and Wikman-Svahn. P. 2010. Assessing socially disruptive technological change. Technology in Society 32, 209-18.

Ihde, D. 2004. Philosophy of technology, in Kemp, P. (ed.) World and Worldhood. 91-108. Netherlands, Springer.

Clarke, A., Fulford, M., Rains, M. and Tootell, K. 2007. Silchester Roman town insula IX: the development of an urban property c. AD 40-50 - c. AD 250. Internet Archaeology 21. Available at: http://intarch.ac.uk /journal/issue21/silchester_index.html).

Kroes, P. and Meijers, A. 2006. The dual nature of technical artefacts. Studies in History and Philosophy of Science 37, 1-4. Lenier, J. 2011. You are not a gadget: a manifesto. London, Penguin Group.

Cooper, M. and Richards, J. (eds) 1985. Current Issues in Archaeological Computing. Oxford.

Levinson, P. 1999. Digital McLuhan: a guide to the information millennium. London, Routledge.

Culkin, J. 1968. Each culture develops its own senseratio to meet the demands of its environment, in Stearn, G. (ed.) McLuhan Hot and Cool. 57-66. Harmondsworth.

Lock, G. 1995. Archaeological computing, archaeological theory, and moves towards contextualism, in Huggett, J. and Ryan, N. (eds) Proceeding of the conference computer applications and quantitative methods in archaeology (CAA 1994). 13-18. Oxford.

Denning, K. 2000. Fuller’s social epistemology: applied philosophy for archaeologists?, in Holtorf, C. and Karlsson, H. (eds) Philosophy and archaeological practice: Perspectives for the 21st Century. 203-214. Göteborg.

Lock, G. and Brown, K. (eds) 2000. On the theory and practice of archaeological computing. Oxford.

Faulkner, P. and Runde, J. 2009. On the identity of technological objects and user innovations in function. Academy of Management Review 34 (3), 442-462.

Lock, G. and Harris, T. 2000. Introduction: return to Ravello, in Lock, G. (ed.) Beyond the map: archaeology and spatial technologies. xiii-xxv. Amsterdam

Feibleman, J. 1967. The Philosophy of tools. Social Forces 45 (3), 329-37.

MacDougall , R. 2005. Identity, electronic ethos, and blogs: a technologic analysis of symbolic exchange on the new news medium. American Behavioral Scientist 49 (4), 575-599.

Floridi, L. 2009. The information society and its philosophy: introduction to the special Issue on the philosophy of information. The Information Society 25 (3), 153-58.

MacKenzie, D. and Wajcman, J. (eds) 1985. The social shaping of technology. London.

Grosswiler, P. 1996. The dialectical methods of Marshall McLuhan, Marxism, and critical theory. Canadian Journal of Communication 21 (1). Available at: http://www.cjc-online.ca/index.php/journal/article/view Article/925/831.

McLuhan, M. 1975. McLuhan's laws of the media. Technology and Culture 16 (1), 74-78.

Huggett, J. 2000. Computers and archaeological culture change, in Lock, G. and Brown, K. (eds), On the Theory and Practice of Archaeological Computing. 5-22. Oxford.

McLuhan, M. and McLuhan, E. 1988. Laws of media: the new science. Toronto.

McLuhan, M. 1977. Laws of the media. Et cetera 34 (2), 173-179.

Moscati, P. 2004. Languages, communication, information technology: an introduction. Archeologia e Calcolatori 15, 11-22.

Huggett, J. 2004a. The Past in bits: towards an archaeology of information technology? Internet Archaeology 15. Available at:http://intarch.ac.uk/journal/ issue15/Huggett_index.html.

Moulthrop, S. 1993. You say you want a revolution? Hypertext and the laws of media, in Amiran, E. and Unsworth, J. (eds) Essays in Postmodern Culture. 69-97. Oxford.

Huggett, J. 2004b. Archaeology and the new technological fetishism. Archeologia e Calcolatori 15, 81-92.

Mumford, L. 1934. Technics and civilisation. New York.

213

 

Thinking beyond the Tool: Archaeological Computing & the Interpretive Process Sui, D. 2004. The media and the messages of locationbased services (LBS): death of distance or the revenge of geography? Annals of GIS 10 (2), 166-174.

Verbeek, P. 2005. What things do: philosophical reflections on technology, agency, and design. Pennsylvania.

Sui, D. and Goodchild, M. 2003. A tetradic analysis of GIS and society using McLuhan’s law of the media. The Canadian Geographer 47 (1), 5-17.

Warwick, C., Terras, M., Fisher, C., Baker, M., O'Riordan, E., Grove, M., Fulford, M., Clarke, A. and Rains, M. 2009. iTrench: a study of user reactions to the use of information technology in field archaeology. Literary and Linguistic Computing 24 (2), 211-223.

Tyler, T. 2008. A procrustean probe, Game Studies 8 (2). Available at: http://gamestudies.org/0802/articles/tyler).

214