Computer Applications and Quantitative Methods in Archaeology 1990 9780860547136, 9781407348629

Table of contents :
Front Cover
Copyright
Contents
List of Figures
Preface
1. Using public communication services for archaeological applications
2. SyGraf — resource based teaching with graphics
3. Every picture tells a story: 'The Archaeology Disc' and its implications
4. Putting the public in the picture: an interactive video applications generator
5. LIVE update: archaeological courseware using interactive video
6. Matrix processing of stratigraphic graphs: a new method
7. The computer representation of space in urban archaeology
8. SQL and hypertext generation of stratigraphic adjacency matrices
9. A new graph theoretic oriented program for Harris Matrix analysis
10. The ArcheoDATA System - towards a European archaeological document
11. Practical considerations for long term data conservation and analysis
12. How safe is your data?
13. Procuring medium-large systems in the public sector — the experience of the English Heritage Record of Scheduled Monuments
14. The operational requirement for a medium to large-scale system — the experience of the new English heritage record of scheduled monuments
15. A computational Bayes approach to some common archaeological problems
16. Some statistical problems arising in radiocarbon calibration
17. Graphical modelling of archaeological data
18. Statistical analysis of ceramic assemblages — a year's progress
19. A technique for reducing the size of sparse contingency tables
20. An approach to quantifying window glass
21. Towards a virtual archaeology
22. Furness Abbey survey project — The application of computer graphics and data visualisation to reconstruction modelling of an historic monument
23. Images, databases and edge detection for archaeological object drawings
24. Visualisation of sherd movement in the plough zone
25. Integrating spatial information in computerised Sites and Monuments Records: meeting archaeological requirements in the 1990s
26. Terrain modelling, deposit survival and urban archaeology
27. Computer and hardware modelling of archaeological sediment transport on hillslopes
28. Simulating coin hoard formation
29. Formal methods for the analysis of archaeological data: data analysis vs expert systems
30. Palamede — application of expert systems to the archaeology of prehistoric urban civilisations

Citation preview

Computer Applications and Quantitative Methods in Archaeology 1990 Edited by

Kris Lockyear & Sebastian Rahtz with Clive Orton Paul Reilly Gary Lock Julian Richards Nick Ryan

BAR Int ernat ional Series 565 1991

Published in 2019 by BAR Publishing, Oxford BAR International Series 565 Computer Applications and Quantitative Methods in Archaeology 1990 © The editors and contributors severally and the Publisher 1991 The authors’ moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher. ISBN 9780860547136 paperback ISBN 9781407348629 e-book DOI https://doi.org/10.30861/9780860547136 A catalogue record for this book is available from the British Library This book is available at www.barpublishing.com BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Tempvs Reparatvm in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 1991. This present volume is published by BAR Publishing, 2019.

BAR

PUBLISHING BAR titles are available from:

E MAIL P HONE F AX

BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK [email protected] +44 (0)1865 310431 +44 (0)1865 316916 www.barpublishing.com

Contents I 1

Using public communication services for archaeological applications

1.1 1.2 1.3 1.4 1.5 1.6 1.7

3

Klaus D. Kleefeld

Kai la/cobs

2

3

COMMUNICATING AND TEACHING

Intrcxluctionand motivation................................................................................................. ......... .... Retrieving archaeologicalinformation today- a nuisance.................................................... .............. Message handling and the directory.......................................................... ................ ........ ........ ......... Mcxlelingarchaeological information .. .... .... ... .. ........ ....... ...... .. .... ... ...... .. ...... . .... ...... .... ... .. .. .. .. .. .. ..... .. 1\vo additional benefits ............. ................. ......... .. .. .. ......... ................. ...... .. ....... .... ......... ................. Making use of public communication services - a scenario.................. .......... ............... ............... ...... Some final remarks and conclusion................. ............. ....... ................... ........ ...................... .............

SyGraf -

resource based teaching with graphics

3 3 4 5 5 6 7 9

David Wheatley

2.1 2.2 2.3 2.4 2.5 2.6 2.7 3

History of the SYASSproject .. ....... .. .... ................................................................ .. .. .. ... .. ...... .... ... .. .. Program philosophy and 'virtual archaeology' ................................................................................... Program design and implementation................................................................................ .................. The resource.................... ............................ ............................ ................................ ........ ......... ...... The program .................................................... ........ ............. ................. .... ..... .. .. .. ....... .......... ....... .. Practical concerns.............................................. ......................................... ............ ........ ................. Implications of SyGraf........ ......... ......................................................................................... ...... .....

Every picture tells a story: 'The Archaeology Disc' and its implications

9 9 10 10 11 11 12 15

Roger Martlew

3.1 3.2 3.3 4

Intrcxluction............................................................ ........ ................................ ................. ............... Interactive videodiscs in archaeology teaching.............. ............................. .............. .................... ...... Optical discs for image archives in archaeology ... .. .. .. . .. .. ... .. ..... .. .. ... ... ...... .. .. .. .. ... .. ... ... .. .. . .. ........ .. ... ..

Putting the public in the picture J. Maytom

4.1 4.2 4.3 5

LIVE update: archaeological courseware using interactive video

5.1 5.2 5.3 5.4 5.5

II 6

21

K. Torevell

Intrcxluction............................. ...... ..................................................................... ....... ..................... The Coppergate interactive video application ..... ... .. ....... .. .. ............... ... ... .... ... .. .. .. .. .... .. .... ..... .. .. .. .. .. .. . Conclusion........ ...................... ............... .................................................... ..... ......... ................ ...... .

Clive Ruggles

15 16 18

Jeremy Huggett

Steven Hayles

1/oward Pringle

21 21 22 23

Ian Lauder

Intrcxluction.... .............................. ................... .. ......... .......... ....... ...... ......... .... .......... ......... .. ........... Standardisation: issues and goals............................................... ..................... .................................. The LIVE structure model............................ ................................................................. .............. ..... The LIVE metafile definition............................................................................................................ Prospects for archaeology teaching using the LIVE system .. . .. .. ... .. .. .. ..... .... .. .. ... .. .. . .. .. .. .... .. ... .. ....... .. ..

STRATIGRAPHIC MATRIX PROCESSING

Matrix processing of stratigraphic graphs: a new method BrunoDesachy FrarzroisDjindjian 6.1 Introduction .. .... .. ....... .... .. ......... .... .. .... .... .. ... .. .. .. .... .. .. ..... .... .. .... ... .. .... .... .. ..... .. .. .. .. .... ..... ....... .. .. .. ... .. 6.2 Methcxl.......................... .............. ......... ...... ................ ............... .. ......................... ......... ...... .... ....... 6.3 Archaeological interpretationof a stratigraphic graph: from stratigraphy to chronology.........................

23 23 24 26 28 29

29 29 29 31

6.4 7

Conclusion......... ...... .... :.............. ..... ......................... ............... .......................................................

The computer representation or space in urban archaeology J. W. Huggett

7 .1 7 .2 7 .3 7 .4 7 .5 8

SQL and hypertext generation or stratigraphic adjacency matrices 8.1 8.2 8.3 8.4

9

39 40 40 41 41 43

Dave Chapman

Introduction .................................................................................................................................... System comJ)()nents... .. ......... .... .... ..... ............. .. .... ........... .. .. ..................... ......... .. .. .. ............... .... ... .. A RDBMS for archaeological stratigraphy. ........................................................................................ Conclusions...... ........................................................................................................................ ......

A new graph theoretic oriented program ror Harris Matrix analysis Irmela Herzog

9 .1 9 .2 9 .3

39

M.A. Cooper

Introduction .. .......... ..... ........ ........ . .. .. ...... .. ..... .. .... .. .. .. ..... .. .......... ....... ...... .. ........... .. .. .. .. ............. ..... Alternative representations .... ........ ..... .......... ... .. .. .. .. .... ..... ........ ............... .... .. ..... .... .. .... ............... .. ... Correlating seCS of hydrograph used in flume experiments ................................................................................. . 184 Theoretical Shields' diagram for large particles in shallow flows ......................................................... . 185 Observed conditions of entrainment of pot sherds in experiments ........................................................ . 186 Shields' diagram with relative roughness values forparticles repeatedly entrained in experiments (lines join values for the same particle) ...................................................................................................... . 187 Experimental results of distance moved for two particle size classes .................................................... . 188 Absolute probabilities of particle movement following a given number of storm events of equal magnitude (renewal probabilities) ..................................................................................................... . 189 Gamma distribution fit to experimental results for rest periods between separate particle movements ..... . 189 Gamma distribution fit to experimental results for distance moved ............... ....................................... . 190 Experimental results of particle dispersion across the slope during movement, with normal distribution fit 191 Schematic diagram of computer model structure ................................................................................ . 191 Initial particle distribution (upper) and example of distribution following ten simulated storm events (lower) ............................................................................................. ................................... ........... . 192 Histograms of initial and final distributions along slope to show the effects of selective entrainment .. .... . 193 Interactions to be considered in further development of the model ..... .............. .......... ........... ........... .... . 194

22.3 22.4

27.5 27.6 27.7

27.8 27.9 27.10 27.11 27.12 27.13

viii

......... ........................................................

145 146

164

181 182

182

LIST OF FIGURES

28.9 28.10 28.11 28.12 28.13 28.14

Histogrrun for Fiesole ................................................................................................................... .. .. Histogrrun for Monte Codruzzo ........................................................................................................ . Histogrrun for Casaleone ................................................................................................................. . Histogram for San Giuliano Vecchio ................................................................................................ .. Histograin for Alvignano ......................................... ..... ...................... ... .... ..................................... .. Syracuse, Berchidda, Monte Codruzzo . & Fiesole by percentage ........................ ...... ........................... .. Pontecorvo & Maccarese by percentage ........................................................................................... .. Casleone, San Giuliano, Alvignano & Avetrana by percentage ........................................................... .. Factors affecting coin hoard data ................. ................. ...................................................... ............. .. An Example Run of SIMULATE ........................................................................................................ .. Fiesole Hoard Simulation ............................................................................................................... .. San Giuliano hoard simulation ........................................................................................................ .. Differing Introduction Spans ........................................................................................................... .. Differing I>ecayRates ..................................................................................................................... .

197 197 198 198 198 199 202 202 203 204 204 205 205 206

30.1 30.2

Schema for PALAMEDE.......... .................. ... .................... .... ............ ...... ............ ................................ Thoory of the origin of the state......... ......... ............................. .... ................................. .......... ...... .....

212 212

28.1 28.2 28.3 28.4 28.5 28.6 28.7

28.8

ix

Preface Kris Lockyear (Institute of Archaeology, 31-34 Gordon Square, London WClH OPY)

Sebastian Rahtz (Department of Electronics and Computer Science, University, Southampton, S09 5Nlf) Computer Applications in Archaeology 1990 was held at the University of Southampton, 21st.-23rd. March. Some forty-seven papers were presented over the three days, frequently in two parallel sessions. There was also a variety of hardware and software on display, and tutorial sessions were held on GIS, quantitative methods and typesetting. Over 150 people attended the conference, including participants from as far apart as Rome and Troms0. CAA has increasingly become international and to reflect this CAA 92 will be held at Arhus in Denmark. This volume presents a selection of thirty papers from the conference, on a wide range of topics. They reflect the both the wide variety of papers that were presented, and the interests of participants. Topics include both the development of new techniques and the application of older, 'established' techniques to archaeological problems. Also included are more general discussions of various topics and techniques (e.g. expert systems, GIS, and long term data storage). Also of note are the substantial sections on communication and stratigraphic analysis. It had been hoped to produce an update to the bibliography of archaeological computing which Nick Ryan published in CAA 88, but this has now been published separately by Daniel Arroyo-Bishop as 'Une bibliographie sur l'application de l'informatique en archeologie' by D. Arroyo-Bishop and M. T. Lantada Zarzosa, published by CNRS GDR 880, September 1990. For a number of reasons, not all within our control, this years proceedings are a little later than we might have hoped. However, at the time of writing, it seems that the volume

will be produced in time for CAA 91 in Oxford. We feel that publication within a year is an acceptable delay, especially when compared with the other conference proceedings. This volume was to be produced in much the same format as the last three years. However, in an attempt to reduce the overall size, and hopefully cost, of the volume we have turned to a double column format. We hope that the final result is to everyone's satisfaction. During the last year there have been a number of other conferences which have had themes not foreign to CAA. The second World Archaeological Congress had a major session devoted to communication and information technology. Some papers from which are being published in the fifth World Archaeological Bulletin. There will also be a volume in the One World Archaeology series, Archaeology in the Information Age, with papers from this session. The Theoretical Archaeology Group conference held at Lampeter in Decemeber 1990 also contained a session on Geographical Information Systems. We feel that these are encouraging signs. Not only, via CAA and other fora, are archaeologists with an interest in computing and/or statistics (or vice versa) communicating more with eachother, but they are also willing to communicate to a wider archaeological audience. A frequent complaint within archaeology is that the various specialists do not communicate with, or understand eachothers work and perspectives. It is important to b~eak this trend and recently these barriers are being overcome. St. Valentines Day, 1991.

2

1 Using public communication services for archaeological applications Kai Jakobs (Technical University of Aachen, Computer Science Dept, lnformatik W, Ahornstr,55 ;51 Aachen, FRG)

Klaus D. Kleefeld (Cologne College, Dept. of Architecture, Study Group on Preservation of Monuments, Betzdorfer Str. 2, D-5 Koln 90)

1.1 Introduction and motivation

These points, together with some technical featureswhich are very attractive for archaeological applications, make these systems. especially the DS, really promising tools for the support of research work. Our basic idea is twofold: first, we want to stress the prospective benefits of the use of public communication services for archaeological applications. We propose to use the directory's globally distributed data base managed by various institutions and universities. This will provide easy access to archaeological information stored, for example, at university institutes, specialist journals and appropriate departments. Additionally, MHS may be used for interpersonal communication, group communication and bulletin boards. Relevant information may include excavation reports, scientific papers or any other form of documentation. Second, we will give an example how archaeological information may easily be mapped on the information structure of the directory, thus providing standardised access to this information. The remainder of the paper is structured as follows: Section 1.2 summarizes state-of-the-art in retrieving archaeological information (from a very personal view, admittedly). Section 1.3 briefly introduces the technical details of the Message Handling and Directory services, respectively. A heirarchical model of a broad range of archaeological information is presented in section1.4. In section 1.5 we describe two additional features which may turn out to be quite advantageous. Finally, section 1.6 sketches a scenario of how use of public communication services might be organised and managed, and how archaeological research may benefit it.

The quantity of information collected in various scientific disciplines has become really enormous and completely confusing, resulting in one of the most unpleasant effects in scientific work. Information on possibly closely related topics is stored at different sites, usually unaccessible from remote locations. Although this is becoming a common problem, archaeology and related disciplines particularly suffer from this type of unintentional 'information hiding'. Reports, even on important excavations, are very often published in newsletters or journals of only regional circulation. As a result, scientific work in this area is frequently slowed down or even blocked. On the other hand, a large number of different communication networks now bridge the continents. These networks include the 'classic' ones like phone, telex or facsimile, but also public or private data communication networks. As a matter of fact public data networks, especially, provide a very effective means of overcoming the rather frustrating situation outlined above. Currently, data networking is globally migrating towards OSI (Open Systems Interconnection). OSI stands for a set of standard communication protocols and service definitions intended to provide for (technically) unrestricted communication in a completely heterogeneous environment. OSI standards are currently under development at the International Standards Organisation (ISO) in the framework of the OSI Reference Model (OSI/RM) (International Standards Organisation 1984). For the scenario envisaged in this paper two OSI services are of particular interest: the electronic mail service Message Handling System (MHS) and the Directory Service (DS). Whereas the first offers a fast, worldwideexchange of electronic letters, the second provides for servicesanalogous to the 'White Pages' and the 'YellowPages' of the telephone service. We propose to use public communication services provided by PTTs or equivalent national organisations to make distributed information accessible. In particular, we propose to use the Directory Service in conjunction with public electronic mail systems as a basis of information exchange and storage. With this approach we take into consideration two major conditions:

1.2

Retrieving archaeological Information today - a nuisance

Publications in archaeology and related fields may, unfortunately, very often be characterised by big delays Years may pass between an excavation and its

publication. inaccessibility Much information is published only in

some internal reports. That which is published in journals, v ery often suffers from a very regional circulation Although the phenomenon is very well known, it can be illustrated by a rather impressive example: excavations have been undertaken beneath Cologne Cathedral since 1948. These excavations are ' .. . the best published scientific excavation in Germany since World War 2' (Doppefeld & Weyres 1980). But: most of the results have only

• MHS and DS will soon be available (more or less) worldwide. • Using, accessing and maintaining the directory data base will be cheap.

3

KAI JAKOBS AND KLAUS KLEEFELD

Discussing MHS we will have to pay attention to the directory service as well. The DS is supposed to considerably enhance user-friendliness of MHS. That is, in addition to the quite elementary services of MHS more sophisticated naming facilities, infonnation services (yellow pages) as well as mechanisms supporting group communication are provided.

been published in the 'KOlner Domblatt', a newsletter with a circulation of about 4,000, very few copies of which are available at public libraries.

Compiling relevant infonnation today is a real problem, for archaeologists as well as for all other people. However, in archaeology - and related disciplines - things seem to be especially fusttating. This is mainly due to the fact that there is no real international platf onn for publication: whereas in computer science, for example, there are a huge number of international journals and conferences providing a broad basis for inf onnation exchange. There are almost no comparable opportunities for archaeologists . What's more, almost no bibliographic data bases, which are for instance available for all natural sciences, are provided (at least as far as the authors know). 1 This situation may lead to considerable delays for scientific work, since it may be necessary to do a lot of travelling just to compile the relevant infonnation required to continue research work.

1.3.2 The directory The CCITT X.500 Directory Service (CCI 1988a) provides information about, and a unifonn naming scheme for, a network •s resources (including e.g. hosts, processes, devices and human users). In terms of the DS, these resources usually are referred to as Objects. A non-ambiguous name is assigned to every object. In general, a DS has to provide three types of service: • name -+ infonnation. For example, an objects name may be mapped on its network address. • name -+ set of names . A set of objects is identified by one single name. • information -+ set of names. This service establishes a 'Yellow Pages• function.

1.3 Some inevitable technical details: message handling and the directory - a (very) short introduction

Usually, the OS is described as a highly distributed ClientServer System. This may be characterised by a typically small number of hosts (the servers, the Directory System Agents (DSA)) providing callable services to the other hosts of the system (the clients, the Directory User Agents (DUA)). A name is assigned to every object. This name is called Relative Distinguished Name (RON). Every RON is nonambiguous relative to its immediate superior. The sequence of RDNs of an object plus those of its superiors forms the Distinguished Name (ON) of this object. The DN is globally unambiguous. The directory also provides for one kind of alternative name, called Aliases. Every object is represented by an Entry, all the entries forms the Directory Infonnation Base (DIB). An object's name and the information stored in an entry are composed of Attributes. The relationships between objects in a network is usually hierarchical. Thus, the DIB will be structured in a 'treeshaped • way. This tree is called Directory Information Tree (DIT). Infonnation about objects is stored in the DIT. Every DSA holds data about a subset of objects known by other DSAs (to provide for distributed operations). The directory's Schema specifies the structure of the DIT, defines Object Classes (e.g. Country, Organisation), Attribute Types (e.g. country name or telephone number) and Attribute Syntaxes (e.g. printable string or numeric string) permitted. The schema is composed form a number of subschemas each valid within one particular Administrative Authority. The possibility of defining new subschemas is of specific interest for our problem. Every area of application may define subschemas tailored for their very specific needs. We will come back to that point in section 1.4.

Within this section we will briefly describe basic Message Handling System (MHS) and Directory System (DS) models and services. It is intended to provide a very condensed overview of the subject; just to introduce it to the reader. Those who have a deeper interest in MHS's and the directory's functionality should, for example, refer to Craigie 1988 or Jakobs 1989. Roughly speaking, the MHS's major task is messagetransport oriented, whereas the DS deals with supplementary naming and management issues.

1.3.1 Message handling system The X.400 MHS recommendations (CCI 1988b) describe a vendor-independent electronic messaging system. MHS may be characterised as a stored-and-forward architecture (cf. Fig. 1.1). The recommendations specify a general MHS model as well as related services and protocols. In the near future, MHS will allow worldwide exchange of interpersonal messages in a rather simple and effective way. Fig. 1.1 shows the basic MHS model. The (human) user accesses the Message Transfer System (MTS) via his private User Agent (UA). A message is routed in a storeand-forward manner from MTA to MTA until it reaches its final destination MTA. It may now be stored in a Message Store (MS) associated with the recipients UA. Basically, a message is made up of two parts: an Envelope and the Contents (cf. Fig. 1.2). The envelope carries infonnation required by the system like, for example on 'real· envelopes - source address (the sender) and destination address (the addressee). The contents is the piece of infonnation to be delivered.

1 A data base named MONUDOC does provide information on preservation of historical monuments , but focusses on impairments caused by environmental influences.

4

USING PUBLIC COMMUNICATION SERVICES FOR ARCHAEOLOGICAL APPLICATIONS

Megg,age HandHng System

Mes&aQe

+

Transfer System

Basic Message Structure

VA = User Agent, MS - Message Store MTA= Message Transfer system

Figure 1.2: Message Structure

Figure 1.1: MHS Model

Directory

s«v Ice

~..,DSP'--~

L__J~

Client DAP DSA

DUA

EJ

DSP = Directory Service Protocol, DAP = Directory Access Protocol

Figure 1.3: Directory Model

1.4 Modeling archaeological information

1.5 Two additional benefits

Like every scientific discipline archaeology has to deal with different types of data. However, hierarchical interdependencies between these data may be identified. For example burial ground - grave - burial object or the scheme region - period - type of finding place. You may easily identify a whole set of such hierarchical interdependencies, temporally, spatially or possibly genealogically. At each level of this hierarchy additional information will have to be assigned to the respective objects. This may for instance include information of regional peculiarities, let's say some specific pottery type, or a detailed (verbal and graphic) description of an amphora Additionally, references to relevant literature might be included. Hierarchical relations makes applying the directory for information retrival look reasonable. The only thing that remains to be done is to specify new object classes and new attribute types. The directory standard provides appropriate mechanisms. In terms of the directory, such modifications will result in a new Subschema, valid for one, possibly more system agents. That is, this new subschema will have to be applied for those DSAs holding archaeological information.

Besides the possibility of simply mapping archaeological information onto the directory information tree, there are two more possibly very interesting features that should be mentioned: Distribution Lists and Bulletin Boards. • Distributions lists provide an effective basis for cooperative work; a predefined group of recipients (let's say a number of colleagues sharing the same research interests) may simply be addressed via just one logical group name. It is foreseeable that cooperative work of geographically separated partners will increase drastically. Obviously, this will require a sophisticated supporting communication system. Thus, mechanisms supporting group communication will become of major interest. • Bulletin boards, though not yet directly supported by MHS, may serve as a platform for international discussions on various topics and as medium for quick dissemination of information. For instance, today's largest data network, uucp, provides a wide variety of so called 'newsgroups' on topics including all aspects of computer science (naturally), but also history and social cultural issues. The German PTf's public mailbox system 'Telebox' for instance provides for

5

KAI JAKOBS AND KLAUS KLEEFELD

Root

RON

□

Country

Attribute Value

C-OE

Organization

0-RWTH Aachen

Org. Units

OU-Dept. Com. Sc.

□

Org. Person

OP-Kai Jakobs

Figure 1.5: A Distinguished Name

Figure 1.4: Model of an Directory Entry

DIT Elements

Definitions

belongs to

rules for Values

Figure 1.6: The Directory Schema

a similar service. So why not establish newsgroups on archaeology or related subjects?

1.6 Makinguse of publiccommunication services- a scenario Finally, we briefly want to outline a scenario whereby archaeological research may benefit from using the (public) directory and (public) electronic messaging systems. Directory system agents may be operated at a number of research institutes or universities. The user accesses the system via public data networks, providing (relatively) cheap and reliable access. Of course, every user is responsible for keeping the infonnation base up-to-date by submitting infonnation to the system and storing it accordingly. The different directory system agents (DSAs) are interconnected via public links and may be accessed either directly or via a message handling agent. DSAs are installed at a number of university computing centres where staff should be available to install, run and maintain the software .

6

The only prerequisities required to access this 'archaeo logical network' would be a personal computer, access to a public network (e.g. via modem or acoustic coupler) and the MHS/DS software necessary to access the network. Things may be that simple since a 'normal ' user will usually only need the respective user-agent software already available for personal computers. If such a scenario became reality, it would also quite helpful on the spot at excavations : DS/MHS user agent software running on portable PCs might assist in identifying a find or in obtaining information on related finds. By the way, the ISODE (ISO Development Environment) software package provides communication software compatible to ISO standards running on top of either X.25 or TCP/IP. Inter alia the package includes MHS software in line with the 1988 MHS version and the QUIPU directory service, a full X.500 implementation (excluding strong authentication). The package is now broadly accepted and widely used.

1.

USING PUBLIC COMMUNICATIONSERVICESFOR ARCHAEOLOGICALAPPLICATIONS

Figure 1.7: The Suggested DIT

Putting all this together, ISODE appears to be a very promising tool to realize our scenario without too high a financial overhead. If such a scenario could be accomplished, research in archaeology, which is to a high degree based on the availability of information, might be brought a considerable step forward.

only to read from the database, but also to add additional information to it. This simple e.quipment will (in the ideal case) then provide him with worldwide access to archaeological information as well as fast communication with colleagues, no matter where they reside.

Bibliography 1.7 Some final remarks and conclusion CCI 1988a. The Directory. Recommendation Series X.500. International Telegraph and Telephone Consultative Committee.

The problem of retrieving relevant information is well known and frustrating. This holds especially for archaeology and related disciplines. To remedy this matter, we propose to adopt public communication services as a platform for both information retrival and exchange. The directory service may easily be employed for the first task, the latter may be achieved using electronic mail services. We have pointed out functional advantages as well as the financial ones. Our scenario comprises a highly distributed data base, installed and maintained, for example, at universities and accessible by everyone all over the world. It is foreseeable that everyone who is able to afford a PC will be able not

CCI 1988b. Message Handling Systems. Recommendation Series X.400. International Telegraph and Telephone Consultative Committee. CRAIGIE.J. 1988. "X.400 (88)-A Tutorial", Compwer Networks and ISDN Systems, 16: 153-60. DoPPEFELD, 0. & W. WEYRF.S1980. Die Ausgrabungen im Dom

zu Koln . Mainz.

INTERNATIONAL STANDARDS ORGANISATION 1984. ISO IS 7498, Open Systems lnJerconnection-Basic Reference Model. JAKOBS,K. 1989. "ISO's Directory Proposal-Evolution, Current Status and Future Problems''.Canadianlournal oflnformlllion Sciences, 64 (1).

7

KAI JAKOBSAND KLAUS KLEEFELD

8

2 SyGraf-

resource based teaching with graphics

David Wheatley (Department of Archaeology, University of Southampton, High.field, Southampton S09 5NH)

2.1 History of the SYASS project

2.2

The Southampton York Archaeological Simulation System (SYASS) is a joint project between the Universities of Southampton and York, and is funded by the Computers in Teaching Initiative (CTI). The project aims to produce an excavation simulation system for use in the teaching of excavation strategy to undergraduate students. Use of a computer simulation was felt to be a promising means of effectively introducing students to the types of decisions that field archaeologists have to make as they compromise the abstract ideal of total recovery against the real world, where pressure from developers and superiors is to reduce costs while still answering archaeological questions. If students could be introduced to this type of decision-making process before (not instead) of letting them loose in the field then this was felt to be advantageous. As a secondary aim the project hopes to encourage awareness of computer technology amongst both students and staff. A number of similar products already exist including Oxfordshire County Council's Dig and Settle, CUP's Unearthing The Past, Digging Deeper into History by Roger Martlew for English Heritage and the American Fugawiland program. Many of these packages are excellent, but all except Fugawiland are aimed at schools, and not at universities. Fugawiland, while strong in some respects, is a directed exercise at a regional scale with no facilities for alteration of the scenario by teachers. A pilot project called CEMYSYASS(Rahtz 1988b) was undertaken in by Sebastian Rahtz. This comprised a simulation based on the database program INGRESin which a data set derived from the Protestant Cemetery in Rome was used as the resource. A simple set of routines were provided with which the user could record details of the gravestones at various levels of detail. The project employed a Research Fellow for eighteen months to investigate and implement further the aims of the project. Approaches to the simulation were discussed (O'Flaherty 1988a, O'Flaherty 1988b) . The SYGRAF program developed from a short research project undertaken as part of the M.Sc. in Scientific Archaeology (Archaeological Computing) at Southampton University. The program was intended as a re-write of the existing SYASSprogram, but concentrating effort on producing an intuitive and attractive interface (Wheatley 1989). Since October 1989, the original program has been substantially re-written, extended and de-bugged. Further work on the code was undertaken and a substantially improved version has undergone some testing with undergraduate students at Southampton.

It is worthwhile explaining, as the SYASSproject has striven to do from its inception, that the SYASSproduct is not intended to replace (or even to diminish) the fieldwork content of the undergraduate curriculum. Instead, the program is intended to better prepare students before and after they have been into the field, for the sort of decisions they will be required to make. Such an approach may in time reduce the time taken to train students outside the classroom, and therefore actually increase the time available for real excavation. The SYGRAF database does not attempt to represent an archaeological site as it occurs 'in the ground', but instead records subjective facts and geographical information about archaeological entities in a format which will be familiar to archaeologically literate students. The site representation adopted for the program has been compared to a level 2 or 3 archive of a real site, and this is essentially accurate - the site database contains information which, in field archaeology, would be included at the level 2 or 3 stage (the site is 'phased' for example, and the pottery sherds are identified). The SYGRAF resource cannot therefore be considered to be contributing to a 'virtual archaeology', and is not intended to do so. Such an essentially reductionist approach neglects the singular most important fact of archaeological interpretation, that of the ascription of meaning to archaeological objects and the theory-laden nature of archaeological data (e.g. Shanks & Tilley 1987). To claim superiority, therefore, by the standards of the ultimate empiricist dream of total recording, is to deny that archaeologists theorize and interpret not objects with no meaning, but theoretical objects. In database terminology, no entity can exist unless it is within a universe of discourse, and to include an entity in a data schema is to theorize it. A 'virtual archaeology' aims also to reduce the amount of practical fieldwork undertaken by field archaeologists by reducing the field project to the status of an experiment which can then be undertaken at a hypothetical level. This ignores the cultural nature of archaeological work, and is a clear attempt to gain credibility by recourse to crude scientistic values. SYGRAFshould be regarded as a tool for analysing and improving the awareness of students to specific aspects of the work undertaken by field archaeologists, and not as an intellectual 'damage limitation exercise'. The SYGRAF product recognises that archaeological fieldwork is as much a cultural activity as an intellectual one. This attitude is, it is to be hoped,reflectedin the SYGRAFprogram.

9

Program philosophy and 'virtual archaeology'

DAVIDWHEA1LEY

2.3 Programdesign and Implementation The general approachto the simulation has been the same as the CEMYSYASSprogram (Rahtz 1988b), in that it is a resource-based simulation system. 'Resource based' is here defined as a system whose primary constituent is a database (or resource), and where the program (or programs) used to manipulate the resource have a secondary function within the simulation. This philosophy allows teaching to follow a user-driven approach, which can be contrasted with a more program-directed approach. The latter would, I believe, have created a far less flexible result. Projects which have adopted a program-directed philosophy have frequently suffered from the 'Press return to continue' syndrome (Hawkins 1987). A resource based simulation is preferable, then, for the following reasons.

• It provides a betteranalogy with reality, where there exists a site (a resource) and a number of ways of extracting information from this resource within a given set of limitations. Decisions are made in reality in an active way; the excavator not only chooses which of the available options to take but decides what the options are in the first place. A programdirected simulation would have been far more limited in this respect. • It is more flexible. Once a data format for representation of an archaeological site has been adopted, other sites can be coded to act as the resource. By using the simulation program on other databases, different situations can be can be designed for students. In addition to these aims of the SYASSproject, the requirements of the new program were as follows: • That the graphics should be interesting and attractive , preferably operating in real-time and re-drawing the site as it excavated. • The interface should be an intuitive, Graphical User Interface using a mouse and buttons approach. It was felt that this type of interface was most likely to encourage non computer-using students. • That the new program be sufficiently modular to allow for the addition of new pieces of program code (in other words new excavation activities) at a later date. In order to implement the program, the underlying representation of the site needed some considerable thought. The primary requirements of the site database (resource) were: 1. The site representation should incorporate depth: that is the site should be layered. The number of layers should not be defined in the program but in the database. 2. The underlying representation of the site should include the graphical description of the site and that the operations which the program performed on the database simulated as closely as possible the real activity of excavation. 3. As much information as possible should be part of the site database, not the program. 'Hardwiring' facts about the site into the program was felt to be compromising the resource-based nature of the simulation .

The specifications discussed above have now been implemented for IBM PC and similar computers. The SYGRAF prototype is a CLIPPERapplication; CLIPPER is a dBase compiler. This makes it possible to build stand-alone applications for oos which can perform operations on standard dBase tables, by writing in the dBase language. In addition to this, CLIPPERallows user-defined functions and extensions to the dBase language to be written in C and in Assembler. It was this combination of facilities, (and personal preference of the C programming language) which led to the adoption of CLIPPERas the tool for prototyping the SYASSprogram. The framework of the program, the routines which control the database operations for excavation etc., are all written in CLIPPER. The graphics and mouse routines (in other words almost all 1/0 operations) have been written in Microsoft C (though many of the graphics routines, particularly those for text handling in graphics mode are not taken from the Microsoft C Graphics Library).

2.4 The resource The resource is a dBase database, consisting of fourteen tables and three indices. These constitute a relational database in which the site is represented in terms of two types of objects, contexts and finds. Any archaeological object which can be approximated in plan view as a polygon is deemed to be a context; while objects which can be sensibly considered to be spot-finds (in other words have their location recorded only as a point) are deemed to be finds. Contexts are typically archaeological entities such as pits, post-holes, ditches, banks, walls and so on while objects like pottery sherds, coins, flints and animal bones would most naturally be considered as finds. Finds are defined entirely by single entries in a finds table, and can be equated to point entities familiar to programmers and users of Geographic Information Systems (Burrough 1986, Wansleeben 1988). Contexts are recorded in two tables as simple polygons. Vertices of the context polygons are considered to be attributes of a context, a context may have anywhere between 3 and 99 vertices on its polygon. The vertices are stored in the points table, while the contexts table contains a single record for each context. This form of representation of polygons is the simplest, and so the fastest to perform real-time graphics operations on. The disadvantages of such schemes are, however, well discussed (e.g. Burrough 1986) and mainly stem from the fact that polygons in this schema are topologically independent from one another, and so cannot share edges. So-called weird polygons are also possible, although the requirement for the polygons to be closed precludes the possibility of deadends . Spatial analysis using such a representation would be virtually impossible. The finds and contexts tables also contain fields identifying the layer in which the object occurs and the descriptive attributes of the object. To further increase the speed of the program, the relational model was broken in one important regard - the vertex points stored in the points table are stored in the order they must be drawn to produce a polygon, and to recover from disasters an order attribute was included in the table . Without this attribute , or the assumption that

2.

SYGRAF -

the points are in the correct order, the representation is incomplete as the points could be drawn in any order. Importantly, the resource specification includes provision for extending the finds and contexts tables, by adding further descriptive fields to the database structure. These allow the descriptions of finds or contexts to be made quite comprehensive, and allow creators of site databases a degree of flexibility. These extra descriptive fields are then accessible to users of the program from the text-based database section of the program. The concept of depth is incorporated into the model by including layer attributes. Every find and context has the layer from which it originates as a field, and the layers of the site are considered to be overlain, from layer 1 at the top to however many layers there are, up to a nominal maximum of nine at present. During excavation by the program, the user is only allowed to remove one layer at a time, and is only allowed to excavate lower layers once he or she has removed the overlying layers. As discussed above, this database does not attempt to represent a site in an uninterpreted format, instead it may be equated with a level two or three archive of an archaeological site, which consists of context records, finds records and phase plans. It was convenient to represent the site in this structured, interpreted way because of the very practical, restrictions of processing speed and data access time - it would obviously have been possible to implement a simulation in which phase definitions were left to the excavator by recording the contexts as three-dimensional shapes. This is possible on mainframe or minicomputers, but is not yet feasible within the restrictions of the desktop microcomputers available to most Universities and other similar institutions. The program had to run at an acceptable speed on IBM Model 50s (and other Intel 286 based computers) and this ruled out real-time three dimensional graphics operations of the sort the program does in two dimensions. Recording the site as a series of phased objects has the additional benefit of making the process of creating new sites easier. Sites are often published in a series of phase plans which can be used as the basis of the stored site. This approach was used in the creation of the Winnal Down site for SYGRAF as used at Southampton University. Plans were digitised in a simplified form from the published report (Fasham 1985), and then further information was appended to the finds and context records later. Thus, instead of a three-dimensional representation of a site, the program deals with different levels, each represented as a two-dimensional phase plan. This twodimensionality can, to some extent be offset by the addition of descriptions and by the use of other sources of graphics, most notably the laser disk resource (see below), but these things will never alter the essentially two dimensional nature of the representation. A facility is provided to link pictures stored on laser disk to the program. This is simply implemented by including a 'frame number' field in both the context and finds database tables. If the laser disk is selected (by a command line option when the program is started) the program sends appropriate codes to the laser disk every time the textual description is requested by the user. As such, the laser disk picture frame

11

RESOURCE BASEDTEACHINGWITII GRAPHICS

can be regarded as an extra tuple of the relational database, and is treated in exactly this way.

2.5 The program The program provides facilities for two main activities, excavation and observation. It is mostly operated by a mouse and button interface, which allows the user to intuitively point at buttons on the screen in order to initiate actions, and to point at objects on the screen to simply request information. In addition to this, the program has a more traditional text-based database section, which allows simple queries to be made of the subset of data which represents the user's database. I do not propose to discuss in detail how the program works, though I wish to identify two features which are significant. Firstly, when the program 'excavates' an area of a layer, the program will convert finds in the resource database into find records in the users database. More importantly, however, it will convert context polygons in the resource database into context polygons in the user's database which represent only that area of the context which would be recovered by the trench. This is accomplished by using a clipping algorithm: the Sutherland-Hodgeman clip. Thus if two trenches each cut the same ditch, the program recovers two seperate contexts for the user, each with a unique context number. The only two ways the user has of connecting the contexts, are interpolation or further excavation. Budgetary control is built into the program. When a user excavates, the program calculates the cost of the excavation based on a simple relationship between area excavated, level of excavation and money. There are four levels of recovery, characterised by four tools, JCB, pickaxe, shovel and trowel. Each of these recover different amounts of detail and consequently are more or less expensive per unit area. The user first chooses which recovery level to use, then identifies the area to excavate. The program then recovers a proportion of the finds, according to the chosen recovery level and subtracts an appropriate amount from the budget. The user must weigh up the cost of excavating at the highest recovery level available (with a trowel), for example, against the destructive nature of excavating with a JCB every time a trench is dug.

2.6

Practical concerns

An exercise using the SYGRAF program was introduced into the curriculum of the first year undergraduate course at Southampton University at the start of 1990. There was, unfortunately, insufficient time to design a comprehensive evaluation strategy for the program, and instead an exercise was set, and this was followed up by a discussion of the program and exercise with the students. Thirty students were involved, each was required to attend two hour-long practical sessions. Because none of the students had used the program before, two site databases were needed during the exercise. One was reserved for the exercise itself, while the other was provided for the students

DAVID WHEATI..EY

to practice with. During the first of the two practical sessions students were introduced to the program, and shown how to operate the program and the associated hardware - most notably the mouse and keyboard. The students were then given a week to get used to the program by excavating the 'practice' site, with any of four computers available to them in the department. A number of problems occurred at this stage. All of these problems were unforseen, although they seem inevitable in retrospect. All also stem from an astounding lack of basic computer literacy amongst first year undergraduates.

questions about the site. These took the form of 'what are the proportions of coarse and fine pottery on the site and what does this mean?' and 'are there the remains of any structures on the site?' How many of these questions were answered, and in what detail was left to the students. A final question 'can you say anything else about the site?' was also included. It was impressed upon the students that they should take their time to excavate the site, that they should understand what the budget meant in real terms, and that they should use the program's print facilities to make plans as the excavation continued. It was also explained that they should attempt interpretation as well as observation of their results, bearing in mind problems of confidence, certainty and bias due to their chosen excavation strategy. It was also impressed upon them that although they had a series of questions to answer, there were no absolute right or wrong answers, and that they were free to investigate anything they pleased within the restraints of the budget. After completing the exercise the students submitted a report. The final reports submitted by students varied tremendously. Some chose to present a detailed discussion of both their strategy for excavation and the results obtained, while others took the path of least resistance and submitted answers to the eight questions and nothing else. This did not make marking the results easy. The final marks were intended to be representative not of the success in locating the archaeology, but of the level of thought given to the design of an excavation strategy, and on the level and quality of the interpretation. The aspects marked best were those which the students had been told to consider most closely, such as potential sources of bias in recovery and the nature of such bias; the problems of certainty in interpretation and the ways in which the chosen strategy for excavation may have effected the final interpretation. The discussion prompted a variety of responses, not all positive. Of the negative responses, one of the most common was the that the exercise was not sufficiently guided. This was, I believe, the natural reaction of students educated mostly in a traditionally passive manner, suddenly faced with an exercise which required them to take the initiative. Many students, unsurprisingly, wanted to be told in more detail what to do and were unhappy with the uncertainty of deciding for themselves what to investigate. Other reactions from students were less concerned with the program philosophy (of which most seemed to approve) and more about the specific implementation. One of the most positive responses was a suggestion that the exercise should contain other types of information such as geophysical data, topography or results from auger sampling.

1. There were too many switches on the computers. Many sat in front of a working computer, with the monitor switched off, for some time before plucking up the courage to ask someone. Many neglected to switch on the mains supply. 2. It proved necessary, in some cases, to explain that the wire led from the back of the mouse, and that it should be kept perpendicular to the computer to work properly. 3. Using the DOS command line to start the program proved too difficult for roughly half of the students. Although directories were discussed in the practical, many failed to understand the basic concepts of files and directories. This resulted in some students copying exactly what was shown in the handout (this meant that they typed in the DOS command prompt even though it already appeared on the screen). Fortunately a menu program is now installed on all of the computers, and the program was accessible from this. 4. Converting the budget from the arbitrary financial units used in the program into real terms (excavation capacity) was difficult, and had to be explained a number of times. 5. Roughly half the students failed to understand the form-based database query system after the practical, and had to be shown in more detail. Although these may seem banal they are listed here to illustrate that computer packages such as SYGRAF cannot be used in isolation from a more general programme of computer education. The first lesson of the SYGRAF teaching experience was that introducing a fundamental computing course to the first-year undergraduate curriculum must be a priority. There were, however, more positive aspects of the experience: most of the students seemed interested and none seemed totally out of their depth. Many quickly found errors in the handout which may have been misleading (the students were not alone in the learning process) and all mastered the basics of the program eventually. The second practical was an introduction to the exercise itself. The resource database, which comprised three layers of digitised information based loosely on the Bronze age and Iron age sites at Winnall Down (Fasham 1985) was installed on the computer, and the students were asked 'excavate' the site. The exercise itself was designed to minimise the amount of guidance given to students and so compel them to make active decisions during the exercise. It was apparent, however, that some guidance was needed with first year students, so the exercise included eight general

2.7

Implications of SyGraf

There are, I think, generally two models of the education process operating within archaeology and defining how the subject should be taught at University level. These can be characterised as teaching by passive learning, and teaching by active learning. In the former, the traditional form of education, the teacher is in control of the content and direction of the learning; the student is a passive receptacle

12

2.

SYGRAF -

RESOURCE BASED TEACHING Willi

GRAPHICS

it can be used badly. It can be used to perpetuate passive learning, should those who control access to it and create exercises to do with it choose so to do. I choose, however, to be optimistic and to claim that SYGRAF can contribute to an educational philosophy within which active students take decisions about their own education, and the rOle of the lecturer is altered from that of holder of power to facilitator and motivator. I do not claim that SYGRAF makes this easier, but it does perhaps suggest one way this can be approached. Chris Tilley has called for an 'archaeological glasnost'; is it then merely pure rhetoric to suggest that we first need an educational perestroika?

into which the wisdom of the teacher is poured. This form of teaching is characterised by lectures and traditional examinations or, at best, by demonstrations during which the student is required to passively receive information. This form of teaching is rooted firmly in the hierarchical nature of society and represents the way in which power is exerted within education. All control of the educational process in the passive learning model is in the hands of the lecturer, all decisions about the direction and content of learning are taken by the lecturer. This control of educational decisions not only imparts power to the teacher and denies power to the students but is also far easier for both the lecturer and the students. The lecturer does not have to respond to unforseen situations, the student does not have to generate any original thought or provide any real motivation for learning. Lecturing has justifiably been characterised as the least efficient way of getting information from the notes of a lecturer to the notes of a student without passing through the minds of either. Although there is good lecturing and bad lecturing, all lectures are part of the passive learning paradigm. There is, however, a growing awareness of an alternative to this passive process. SYGRAF attempts to contibute to a model of education which is characterised by an active student. In this educational model the student is given control of the decisions about what is learned, and how it is learned. This is not an easy task because traditional modes of education are seductive and, in their own terms, extremely successful - students who have sat doggedly through a series of lectures can retain the right information to pass examinations. If the mould of passive learning is broken, however, then the means of judging the success or failure must also change and it is my own belief that students who have been required to accept responsiblility for decisions about their own learning process, and who have been allowed to actively participate in the process will make better researchers, teachers and citizens. SYGRAF is nothing but a tool, and a prototype tool at that. This paper has described that tool and set out the motivation for building it. I hope (but by no means guarantee) that it reflects the views of all of those involved with the SYASS project. As a tool, however, SYGRAF can be used well or

Bibliography BURROUGH,P. A. 1986. Principles of Geographical Information Systems for land resources assessment. Clarendon Press, Oxford. FASIIAM,P. J. 1985. The Prehistoric Settlement al Winnal Down, Winchester. Monograph 2. Hampshire Field Club. HAWKINS, S. 1987. "An Examination of the Context of the Proposed Southampton-York Archaeological Simulation Sys tem", Syass Project Internal Document No. 6. O'FLAHERTY,B. 1988a. ''The Southampton -York Archaeological Simulation System", in Rahtz 1988, pp. 491-497. O'FLAHERTY, B. 1988b. ''The teaching of archaeological excavation-a new initiative", Archaeological CompUling Newsleller, 15. RAIITZ.S. P. Q., (ed.) 1988a. Computer and Quantitative Methods in Archaeology 1988, International Series 446, Oxford. British Archaeological Reports. RAHTZ.S. P. Q. 1988b. "A resource based archaeological simula tion", in Rahtz 1988, pp. 473-490. SHANKS,M. & C. TILI..EY1987. Re-Constructing Archaeology. Cambridge University Press. WANSLEEBEN,M. 1988. "Geographic Information Systems in archaeologicalresearch", in Rahtz 1988, pp. 433 - 51. WHEAn.EY, D. W. 1989. "SyGraf-A Prototype Graphically Based Excavation Simulation for Computer Aided Leaming", Master's thesis, Department of Archaeology, University of Southampton .

13

DAVID WHEA1LEY

14

3 Every picture tells a story: 'The Archaeology Disc' and its implications Roger Martlew (Department of Archaeology, University of Leicester, Leicester LE2 7l.X)

3.1 Introduction

which often present difficult material for undergraduates to assimilate. A visual record of such sites is important not only for showing the nature of site itself, but also its setting in the landscape. The contrasts between sites such as Stonehenge and Castlerigg, or Balnuaran of Clavaand and the Hurlers can be seen instantly on the screen, without any need for lengthy textual descriptions. Given the limitations of the budget it was to our advantage that finds from stone circles are relatively few. Stone circles represent an excellent example of the rapid fall off in the availability of visual information once the 'classic sites' have been covered in, for example, general textbooks on prehistory. The cost of publishing visual information by traditional means has obvious consequences: in seven general books on prehistory (one relating to Europe, one specifically to Scotland and the rest to Britain), Stonehenge receives most coverage in the form of halftone plates. The 12 pictures are mostly black and white, with notable exceptions in a particularly glossy production (Clarke et al 1985). A limited range of other sites appears less regularly: Avebury (8), the Ring of Brodgar and Callanish (6), Loanhead of Daviot (4), and Castlerigg and the Stones of Stenncss (3). Four other sites are shown in one plate each, spread between three books. The main specialist book on this subject (Burl 1976) contains 36 black and white plates. Students can expect to see other visual information on stone circles in their lectures, perhaps including unpublished material, but this is not information to which they have easy access (Martlew 1990). In contrast, an attempt was made on The Archaeology Disc to show at least one picture of every known stone circle, even if the site consisted of widely spaced, low stones set among tall heather. Over 2,500 still pictures are supplemented by a full motion video, and almost all of the images are in full colour . This is not to say that printed textbooks should be full of pictures, or should try to present a comprehensive visual catalogue . Each medium has a different rOle to perform, and has different strengths and weaknesses. The point of the comparison is to show the extent to which traditional printed media have failed to disseminate visual information, and that an extremely efficient medium is now available to do the job . Most of the material for the stone circle image databank on The Archaeology Disc came from the personal collections of academics and private individuals. The budget permitted only a limited amount of new material to be gathered, and this was aimed specifically at demonstrating innovations made possible by the technology. We are panicularl y grateful to Dr. Aubrey Burl who allowed us complete access to his extensive archive, and to others such as John Bamatt and Alison Haggarty who allowed us to use material in advance of publication. Most of the material consisted

A two and a half year project in the Archaeology Department at Leicester University on the use of videodiscs in archaeology is now finished. This paper presents the conclusions of the project, and discusses its implications for teaching archaeology, and for archaeological recording methods. The project was one of over 130 funded by the University Grants Committee (the predecessor of the Universities Funding Council) and the Computer Board for Research in Universities. The primary aim was to investigate and develop the use of new technology in undergraduate teaching. The background to the national Computers in Teaching Initiative (CTI) and details of the wide range of projects supported by it can be found in the various issues of its publication The CTISS File. As is often the case, the budget for the Leicester videodisc project was a compromise betwen what was actually needed, and the level of funding which was on offer. The project at Leicester University was inspired not only by the potential of videodiscs for teaching, but also by a wider interest in the use of this technology for recording the visual and graphical information which forms such an important part of archaeological records at all levels (Martlew 1988). Awareness was already being raised by demonstrations of videodiscs such as the BBC Domesday discs, but there was a limit to the level of understanding which professional archaeologists could gain by trying to retrieve often strange information about their home town from the Domesday Community Disc. It was felt that a videodisc containing specifically archaeological material was needed, both to investigate its potential as a recording technology and to develop innovatory applications for undergraduate teaching.

3.1.1

Compiling material for the videodisc

The videodisc produced by the project became known as 'The Archaeology Disc", and was designed from the start as a resource disc. Since it is currently not possible to change the contents of a laservision disc once it has been pressed, it was not thought advisable to enshrine any particular theories or interpretations on the disc. The next excavation or fieldwork project could easily overturn such theories and leave the disc as a museum exhibit itself, rather than as a tool of continuing value. The main subject for The Archaeology Disc was British stone circles, with some material on related sites. This provided a reasonably well defined core of field monuments, presenting a range of morphological characteristics and a wide geographical spread. Stone circles also support an interesting variety of theoretical interpretations, such as astronomical alignments and geometrical construction,

15

R. MARTLEW

of 35mm colour slides, but plans were also included as drawings, many of them being re-drawn in colour to make full use of the medium. One particular strength of videodisc technology is its capacity for combining still pictures, moving images and sound. Archaeological material is not noticeably mobile or noisy, but there are ways in which the techniques of television can supplement the information about artefacts or sites recorded conventionally by still photography or drawing. Videodiscs, however, off er better access to the images than linear video or film, and it was important to locate relevant material for The Archaeology Disc which would allow us to investigate the implications of this for archaeological recording. With the generous cooperation of Palladium International PLC it was possible to include material on stone circles from a television programme made by Simon Welfare for Yorkshire Television. This includes rare (and genuine) footage of the winter solsticial sunrise as seen from inside Newgrange tomb in Ireland, a powerful way of introducing the subject of archaeoastronomy to undergraduates. Since there is full control of the audio channels on the disc, the commentary, which is aimed at a lay audience, can be turned off. There is useful teaching material even here, however, with current interest in the public presentation of archaeological sites and monuments. Even with this moving video, only about half of one side of The Archaeology Disc was used, so spare space was sold to anyone who wanted to try out the technology for a share of the production costs. Some of this material has no connection with archaeology, such as that from the Institute of Irrigation Studies at Southampton University and the Royal United Hospital in Bath. With complete computer control over access to different parts of the disc, this subdivision represents efficient use of the resource and need not be apparent to the user. Several archaeological bodies also took the opportunity to test the technology with their own material, including the Royal Commission on the Historical Monuments of England, the National Monuments Record of Scotland and the York Archaeological Trust (Maytom, this volume). The transfer of slides, photographs and drawings to videotape was carried out by Ken Morse Rostrum Cameras, and the quality of the images on the disc has been admired by other videodisc producers. Final editing of the master tape was carried out by Sheffield University Television, and a master disc was produced by Philips at their plant in Blackburn. Thirty copies of the disc were pressed, the distribution and use of which are strictly limited by agreement with the owners of copyright material on the disc. The Archaeology Disc is intended purely for research and evaluation purposes, and copies have been distributed at a notional cost to cover pressing and replication charges. The intention of the Computers in Teaching Initiative was to encourage and develop the use of new technology in Higher Education. Most of the interest shown in The Archaeology Disc, however, came not from this sector but from, or on behalf of, the secondary sector. Discs have been sold to education resource centres and departments of education from Grampian to Exeter for evaluation and trials, but only two copies have actually been acquired by a university department of archaeology. The possible

reasons for this, and its implications, are discussed in the next section.

3.2 Interactive videodiscs In archaeology teaching The pedagogical background to the Leicester videodisc project has been discussed elsewhere (Martlew 1989, Martlew 1990). Other more general treatments of this aspect can be found in, for example, Barker 1989. The main points to be covered here concern practical issues of developing courseware associated specifically with The Archaeology Disc, and the identification of problems and productive lines of development for the future. Although careful attention has been paid to the effectiveness of interactive videodiscs in a training environment, research into their use in more general education is spread thinly among a wide range of specific applications covering different agegroups (Whiting 1989, p. 45; Mashiter 1989, p. 205). Videodisc projects are still very much at the leading edge of developments in university teaching, but the initiative should now be moving from proselytising to implementing videodisc applications (Clark 1989). The CTI, as a "pump-priming" exercise, has supported the initial creation of evangelists, and the CTI subject centres are now helping them spread their message. There is, however, little real support for the process which matters most: the production of materials which can be picked up and used straight away by university teachers and students around the country. This failure can only hold back the use of new technology in teaching. The current climate in Higher Education, with student numbers increasing while staff numbers and resources remain much the same, introduces a very real danger that new technology of any form will be seen as a cheap alternative to employing additional staff. The aim of the Leicester videodisc project was not to develop technology to replace human teachers, but to provide an additional tool to broaden the current learning environment for undergraduates. No single approach to teaching and learning can ever be expected to satsify all of the 'consumers' all of the time. An understanding of the processes of learning will enable effective strategies to be identified for particular individuals, and a good teacher will ensure that students can choose their strategy from a range of learning activities. Whiting, for example, found that small groups of students collaborated on a computer assisted learning (CAL) tutorial at a single terminal. There was interaction between the members of these groups which added more to the activity than would have arisen from a single user working at a computer. 'Holists' in the groups grasped the broad aims of the tutorial, and explained them to other members of the group who could be defined as 'serialists' (Pask 1976, p. 130). The serialists in return encouraged the holists to look at the work in greater detail. One of Whiting's conclusions is worth quoting in full:

16

3.

EVERY PICTURE TELLS A STORY: 'TIIE ARCHAEOLOGY DISC' AND ·ITS IMPLICATIONS

3.2.2 "careful design of such learning experiences can promote the autonomy of learners and simultaneously free the teacher to act in ways which encourage the development of their students' cognitive abilities, rather than rote learning and remembrance of facts." (Whiting 1989,

It is this low level, labour intensive work which is retarding the development of computers as teaching aids, in archaeology or in any other subject. Such work does not contribute to an academic's list of publications, and so will not be counted by the Universities Funding Council when it carries out its superficial yet influential assessment of the worth of university departments. The lack of institutional infrastructures to support the development of undergraduate teaching methods has been discussed elsewhere (Martlew 1990, p. 467). It is sufficient to report here that the presence at Leicester of two 'research projects' on the use of new technology in teaching has had as little institutional impact as any other research project in the university: courted while they were 'sexy' (to use the jargon of the eighties), they are now, in old age, quietly forgotten. Responses have varied from institution to institution, and from subject to subject. Archaeology has an advantage in that the use of computers by the profession will ensure them a place in the undergraduate curriculum. This is also, however, :1 disadvantage, in that it compromises the aim of the CTI to develop computers as tools for teaching and learning. This aim will have been met, for example, when students use computers to learn about Neolithic ritual monuments in a prehistory course, rather than databases or wordprocessing in a computing course (even if archaeological examples are used (Dobson 1989, p. 19)).

p.49). The Archaeology Disc is an attempt to provide support for teachers in the management of their students' learning activities. Visual information is central to the teaching of archaeology, and students' access to such information is currently restricted and outside their control (Martlew 1990).

3.2.1

Computers in Teaching: the. lack of initiative

A cautionary tale of software (non) development

Unfortunately, authoring software for the development of 'courseware' (packaged programs for use by students) only became available for the chosen hardware configuration as the project drew to a close. The criteria which had been applied in looking for hardware and software were portability and cheapness, on the grounds that it was the aim of the CTI to reach the widest possible audience. The hardware selected for the project was the Research Machines Nimbus PC, for which, in the early days, the necessary interface was being developed in response to the BBC Domesday videodisc project. This hardware was chosen in preferance to the BBC micro because of the greater potential and portability of MSDOS 16 bit software. IV authoring software running under MS DOS appeared to be forthcoming at the start of the project, but the people responsible for this work at Research Machines left to set up their own company, with different research and development priorities. Two years later a preproduction version of the software has been given to the project for betatesting, at the same time that a different company finally produced an MSDOS version of their BBC authoring package. The incomplete nature of the prototype software, and the late arrival of its alternative, will not support a serious evaluation of learning strategies using the videodisc within the original terms of the project. It has been said that "every single interactive video project ever undertaken is always in need of just a little more . . . " (Clark 1988, p. 43) and the Leicester videodisc project (for reasons outside our control) has maintained this tradition. In fact this would probably have been true even if the software had been available on time, since the resource which has been created could sustain the development of curriculum based teaching materials for at least the time of the project over again. Providing relatively simple interactive access to the 44 images of the sheep skeleton, for example (albeit with poorly documented and incomplete software), took about a week of concentrated effort. The work consisted mainly of the fairly mundane tasks of designing graphic overlays, and making sure that the various links from one frame to another operated correctly: actually designing the overall scheme took relatively little time.

3.2.3

The future for videodiscs in archaeology teaching

Videodiscs are likely to be used in teaching archaeology in the future, but mainly because of their use in institutions outside the education system rather than any pedagogically led developments in universities. Museums, government agencies and others connected with professional archaeology, tourism and the 'heritage industry' will be using videodiscs for their own archival and educational purposes, and these resources may filter into mainstream education. Any use of videodiscs , however, should not be seen in isolation. They will occupy a particular niche, performing the tasks to which they are best suited, alongside other technologies fulfilling different needs in a multimedia environment. "Our business is knowledge creation and the communication of that knowledge . Success as a university requires freedom to interact with knowledge and information sources wherever they are" (Jordan 1989, p. 33). One of the hardware implications of this need is the growing demand for broadband campus networks, so that facilities other than data communications can be made available. Aston University is installing a system which consists not only of two Ethernet channels carrying data at up to two megabits per second, but also includes eight oneway video channels and four twoway video channels. 2500 service points will be installed on the campus, the Science Park and in student residences (ibid). The inclusion of visual information in the services provided by such a network is a major area for development.

17

R.MAR1LEW

This, however, is about as far as videodisc technology currently goes. One significant factor counts against a videodisc becoming the final archive for visual information, and that is the problem of resolution. Analogue video screens contain 625 lines vertically in PAL format, with 768 square pixels on each line. At a distance of more than six times the height of the picture, the human eye cannot detect the difference between the lines (Clark 1987, p. 61). It is common for users of videodiscs, however, to be sitting closer to the screen than this in order to interact with the image via mouse, keyboard or touchscreen. For the serious, in-depth study of images such as aerial photographs, finely decorated artefacts or detailed drawings, analogue video does not yet provide a solution for the professional. It is arguable, however, that at this level of study there will never be any substitute for the real thing, and that what is important is that users can select relevant original material quickly and with maximum flexibility.

This level of provision is obviously capital intensive, and requires careful justification before launching into expenditure on the scale of the £4 million being spent by Aston University. Videodisc projects associated primarily with teaching, rather than research, will find such justification extremeley difficult in the university environment. However, the slow takeup of videodiscs, relying on developments outside the education system, may not be the worst of all possible evils. If university administrators decide that videodiscs and other computer related technologies are a way of increasing 'productivity', enabling the same number of staff to teach ever larger numbers of students, the right development may find itself being pushed along for the wrong reasons. It is hoped that the experience of the Leicester videodisc project will provide further ammunition for the educationalists against such cynical exploitation of new technology in teaching.

3.3

Optical discs for image archives in archaeology

3.3.2 Heritage documentation Videodiscs are being used in a number of projects to record standing buildings, works of art and other artefacts which come under the general heading of 'cultural heritage' . A recent report identified 16 'cultural catalogues' on videodisc in use in museums and libraries throughout the European Community (Commission of the European Communities 1988, p. 259). A major project began in Italy in 1986, with the government spending 600 billion Lire on combined text and image databanks to record the country's cultural heritage. 16 out of 19project examined for the DOCMIXreport are based on optical disc technology, with titles ranging from La presenza ebraica in Italia (an image databank of ancient Jewish artefacts on videodisc) to Torrie complessifortificati di Roma medioevale (towers and fortifications of medieval Rome). Another project, Verso Genova Medieval, allows surrogate travel through medieval Genoa using analogue images on videodisc and computer generated graphics (ibid, 244). The SIRIS project (Sistema lnformativo per la Riconstruzione dell' I nsediamento Storico) aims to create an integrated text, cartographic and image databank of historical settlements in the region of Emilia-Romagna. What is in effect a multimedia geographical information system consists of a microVAX running INFORMIX under XWindows,and controlling a videodisc player which displays images on a separate monitor. The project cost over a billion Lire, and employed 96 people for two years (SIRIS 1989). In France, 22 regional centres have been set up to make heritage documentation more accessible to the public. The level of documentation varies from region to region, but there is generalIya heavy reliance on microfilm. At least one region, Languedoc, is developing a videodisc application, in this case on medieval stainedglass windows (CenUivre et al 1989). In Denmark, videodiscs are being used to display 2,000 background maps on which the distribution of prehistoric sites and monuments can be plotted using overlaid computer graphics. A project at the National Museum in Copenhagen has stored 105,000 pictures of artefacts from the museum's

The second major aim of the Leicester project was to investigate the use of videodiscs for storing image archives. The need for improved storage and retrieval systems for image data in archaeology is not hard to define. Videodiscs offer tremendous potential for applications ranging from the tens of thousands of excavation slides held by an urban unit (Maytom, this volume), to the millions of slides, prints and artwork held in national collections such as the picture library of the Royal Commission on the Historical Monuments of England (RCHME). The inclusion of a range of material from the York Archaeological Trust, the RCHME and the National Monuments Record of Scotland on The Archaeology Disc enabled the staff of the various institutions to see their own familiar images as portrayed by video. This resulted in a useful definition of the best role for interactive video in an archival context, and the identification of the limitations of this particular medium.

3.3.1 Visual catalogues Videodiscs are already in use in some museums (most notably the Prins Henrik Maritime Museum in the Netherlands) as visual catalogues which can be searched by visiting scholars and members of the public. The videodisc serves as a quick and easy way of making a 'first pass' through the available information. Instead of just getting a printout listing the photographs or artefacts in a conection which match the user's requirements, the user can actually browse through the pictures or see the artefacts straight away. This is particularly useful for archaeologists when trying to locate, for example, the best selection of aerial photographs of a group of cropmark sites. As with text retrieval, locating the information by standard query on site name or location is relatively easy without a computer, since that approach is anticipated by the archive's manual cataloguing system. The real advantages arise when the power of the computer to search on a number of different cataloguing fields simultaneously is combined with the almost instant retrieval of the image itself, not just its storage location 18

3.

EVERY PICTIJRE TELLS A STORY: 'TIIE ARCHAEOLOGY DISC' AND ITS IMPLICATIONS

Bibliography

collection on videodisc, for use by visitors to the museum (Larsen 1989). The brief descriptions of these projects is not intended to be an exhaustive survey of archival videodisc applications in Europe, but is rather to show the scale at which videodisc technology comes into its own. The sheer volume of material which can be stored on a videodisc can in itself be a disincentive, both to potential developers and to funding agencies. The transfer of images to videodisc is labour intensive, and therefore expensive, but it comes nowhere near the labour and costs involved in cataloguing the images, and in developing support materials in the form of a userfriendly interface and documentation.

BARKER.P.G., (ed.) 1989. M ultUMdiacomputer assisted learning.

Kogan Page, London.

BURL,H. A. W. 1976. The sloM circles of the British Isles. Yale UniversityPress, London. CENTUVRE,B., 0. TOCHE,& Y. J. Riou 1989. "Description of France's experience of heritage documentationcentres", in Architectural Heritage: new technologies in docwnenlalion. Round table of experts organised by the Council of Europe and the Royal Commission on the Historical Monuments of England. CLARK,D. R. 1987. ''Twenty-first century books: an assessment of the role of videodisc in the next 25 years", in Laurillard, D., (ed.), lnleractive media, pp. ~73. Ellis Horwood, Chichester.

3.3.3 The future for videodisc archives

CLARK,D.R. 1988. "Hard lessons for interactive video", Educational Computing, 9 (2): 41-5.

The main consideration for the future development of videodisc applications is the dichotomy between current specifications of optical media. The potential rewards are great: a computer controlled delivery system which has sufficient image quality to replace photographic archives completely. On the analogue side, the laservision format for videodiscs may be superceded by the development of high definition television, while the digital side the Compact Disc (CD) format appears to have inspired the War of the Acronyms: CDI and DVI are fighting for a commercial foothold, and in the meantime CDXA offers a partial stage in the development of a single medium for mixed digital video, audio and data. Confusion over hardware standards, interfacing and software standards is adding to the inevitable claims and counterclaims in an industry which is still very young, as rival manufacturers battle for potentially lucrative markets. Quoting the editor of The Videodisc Monitor, Frenkel points out that there is still "a long way to go before digital formats are anywhere near as cost effective or data dense for full motion video and large databases" as analogue videodiscs (Frenkel 1989, p. 875). In the meantime, there are many important issues about the indexing of image databanks, and userfriendly interaction with them, which must be explored. Current technology is perfectly adequate for investigating these issues, and it would be shortsighted to defer this work in the expectation of technological changes some time over the next 10 years. Now that we have a videodisc containing relevant material, some of these problems can be tackled in the field of archaeology. If future work with optical discs of either analogue or digital format is led by the technology, laservision videodiscs will remain in a relatively small, specialised niche serving large scale stills archives, and commercial training needs which require full screen, full motion video. If, however, future work is applications led, videodiscs will form a significant part of a multimedia environment alongside smaller scale archives on CD (such as bibliographic and excavation archives), and ad hoc collections of mixed text and graphics on CD ROM discs. The Leicester videodisc project has shown that producing the discs is relatively straightforward, while developing the friendly and powerful interface which the users of such technology require is a much greater task. Work in this area can usefully proceed while the hardware manufacturers fight over the nuts and bolts of future deli very systems: it is, after all, the message which is important, not the medium.

CI.ARK,D. R. 1989. "Separating out the design and development of IV from its delivery", in Tucker, R. N., (ed.), lnleractive Media: the human issues, pp. 56-62. Kogan Page, London. CLARKE,D. V., .T G. COWIE,& A. FoxoN 1985. Symbols of power at the time of Stonehenge. HMSO, Edinburgh. COMMISSIONOF TIIE EUROPFANCOMMUNITIES1988. DOCMIX State of the art and market requiremenls in Europe: Electronic Image Banks. Report EUR 11736 EN. Directorate General Telecommunications, Information Industries and Innovation, Brussels, Luxembourg. DoBSON, M. 1989. ''The archaeology of Pallas", Archaeological Computing Newsletter, 18: 11-22.

K. A. 1989. ''The next generation of interactive technologies", Communications of the Associalionfor Computing Machinery, 32 (7): 872-82.

FRENKEL,

JORDAN,A. 1989. "Setting high standards", Educalional Computing, 10 (2): 32-5. LARSEN,C. U. 1989. Video disks and image processing. Round table of experts organised by the Council of Europe and the Royal Commission on the Historical Monuments of England. MARTI.EW, R. D. 1988. "Optical Storage: another can of WORMS?", in Ruggles, C. L. N. & Rahtz, S. P. Q., (eds.), Computer and Quanlilative Methods in Archaeology 1987, International Series 393, pp. 26-58. British Archaeological Reports, Oxford. MARTI.EW,R. D. 1989. "Pictures and Pedagogy: teaching archaeology using image archives on videodisc", Science and Archaeology, 31: 12-19. Special issue: Computer Archaeology: papers presented at The Dynamic Text conference, Toronto, June 1989. MARTI.EW,R. D. 1990. "Videodiscs and the politics of knowledge", in Miall, D., (ed.), Humanities and the computer: new directions, pp. 39-48. Oxford University Press. MASHITER,J. 1989. "Video discs for interactive learning", in Barker 1989, pp. 194-209. PASK,G. 1976. "Styles and strategies of learning", British Journal of Educational Psychology, 46: 128-48. SIRIS 1989. "SIRIS: Sistema Informativo per la Riconstruzione dell'lnsediamento Storico", in lnformazioni (Anno V, nuova serie, 1). Istituto Beni Culturali e Naturali della Regione Emilia Romagna, Bologna. J. 1989. ''The case for multimedia CAL", in Barker 1989, pp. 44-52.

WHITING,

19

R. MARTI..EW

20

4 Putting the public in the picture: an interactive video applications generator J. Maytom K. Torevell (York Archaeological Trust, 1 Pavement, York)

4.1 Introduction

pressed in 1988 by the Leicester Interactive Videodisc project (see Ruggles 1988 and Ruggles, this volume). Existing authoring software was either found to be unsuitable or very expensive. The Trust decided, therefore, to write its own applications generator using CLIPPER, the language in which all the Trust's existing in house database applications are written.

The York Archaeological Trust Interactive Videodisc project aims to allow the general public direct access to the Trust's pictorial archive. The project is based at the Trust's newly opened Archaeological Resource Centre (the ARC) at St. Saviour's church in York. The medieval church of St Saviour stands in what was once a fashionable Georgian area of York. It was made redundant in 1952 and from that time has been used by various organisations as a warehouse. The York Archaeological Trust acquired it for this purpose in 1974 as sub-tenants of the York Civic Trust. The building was already in a very poor condition and if demolition was to be avoided a programme of costly and extensive restoration needed to take place. A 75 year head lease was acquired from the York Civic Trust in 1984 for a pepper-com rent. The York Archaeological Trust was then in a position to put money into the building. To date, a total of £545,000 has been spent. It was the York Archaeological Trust's intention to renovate and save this historic building whilst at the same time provide storage and office space in the centre of York. St. Saviour's Church was also to house the ARC. The ARC is an extension of the Trust's charitable aims; to educate the public in archaeology. The ARC allows people of all ages to work alongside professional archaeologists, to handle objects from excavations and to explore how they were made as apposed to the traditional museum approach. Visitors are actively encouraged to touch the exhibits and to use reconstructed technologies such as Viking age weaving looms. It is into this environment that computers and interactive video have been introduced. The public are able to use three different computer applications: a finds recording system, a computer aided design (CAD) system and an interactive video. All three try to illustrate the different uses to which archaeologists put the computer. The finds systems mimics the computerised recording system in use by finds research staff on the first floor of the ARC. The public are able to enter information about a selection of small finds. The 'conversational' approach of the software combined with extensive data verification makes it impossible to enter incorrect information. They can then produce a report complete with their own name. The CAD system allows a less extensive level of interaction and is intended more as a demonstration of such a package. The interactive video application preempts the way in which archaeologists will handle site archives in Lhe future. Initial software development has taken place using an assemblage of excavation slides from the Coppergate excavation in York. These slides were included on the videodisc

4.2 The Coppergate interactive video application 4.2.1

Hardware

The equipment used for both development and applications is the Microvitec 'Insight' system, Incorporating a Videologic M.I.C. 3000 card and touch screen. The M.I.C. 3000 card allows the 'playing' of analogue images and has its own set of rather limited commands such as 'PLAY', 'STEP', 'FADE' etc. It does not allow the conversion of analogue images into digital form.

4.2.2

Initial design and linking of slides

The applications generator created is extremely flexible and allows any computer-generated images or text to be changed at any time during the running and building up of an application. It therefore enabled the Trust to make excellent use of a set of interesting but quite disjointed slides of the Coppergate excavation to build up an exciting application specifically for use in ARC. The Coppergate application allows the public to journey along different paths through the videodisc images of the Coppergate excavation. The pathways linking images are all predetermined but the choice of route is interactively decided by the users at a run time by touching the appropriate area of the screen. These pathways are depicted by symbols which overlay the video image, the symbols only appearing for existing relationships. The user may choose to go 'LEFT', 'RIGHT', see the 'NEXT' slide in a sequence or even 'ZOOM' to another slide. The 'ZOOM' gives a closer look at a particular detail of the previous slide. All 'ZOOMS' are remembered by the system, enabling the user to retrace his steps back through the slides; a 'MOVE BACK' symbol is displayed in the bottom left comer of the screen. As the user views the slides, relevant textual information is displayed, questions are asked, and featuresmeriting special attention are highlighted. A more advanced level of information can be accessed by the more specialist user.

21

J. MAY'IOM AND K. TOREVELL theme or just the next slide in a sequence. These two types of link are fixed and represented by the screen symbols 'RIGHT' and 'NEXT'. When created or altered by the edit facility, the target slide is automatically given the reciprocal relationship of 'LEFT' or 'LAST'. The reverse is also true. Other pairs of fixed relationships may also defined as required e.g. 'UP, 'DOWW. 'ZOOMS' and 'hidden touches' may be created or edited, allowing a maximum of 10 for each slide in this application. The system maximum is 1920. The size, location and target of a 'ZOOM' can all be determined. Text can be input for 'Hidden touches'. Slides may also be given a physical location on the site map.

It would even be possible to present such information in other languages.

4.2.3

1

MAP' faclllty

At key points, a simplified site map may be accessed indicating the users current visual position on the excavation. From here, the user may select a particular period that is of interest or choose to run an introductory sequence that sets both Coppergate and the interactive video in context. To recreate the idea of 'digging down' into history, the slides are separated into five different periods: modem, medieval, Viking, Anglian and Roman. A 'MAP' symbol at the bottom right of the screen indicates possible access to the map. Each period is given a brief introduction, and the user is then free to 'walk' around the slides as desired.

4.3 4.2.4

1

Hidden' zones

The flexible nature of the software also has many research and archive implications. The applications generator may be used by research staff to build their own links through archive material, each researcher constructing their own individual pathways. New pictures from the disc may be added to existing pathways at any time. This project, aimed primarily at schools and the general public, has already led archaeologists in York to take a new approach to the photographic recording of sites and objects and created an appreciation of the archive as a flexible research tool.

After undergoing trials with groups of 8-10 year old children, it became obvious that young children are often uninterested in reading any text at all, unless directed by an adult. They are more interested in touching the screen symbols and watching the computer respond. They are therefore learning how to use the system, but not gaining any knowledge of the archaeology. By introducing 'hidden touch' areas of the slide the children are unwittingly informed as to the content of the slides.

4.2.5

Conclusion

Applications Generator

Bibliography

An application may be developed by using the edit facility. In edit mode, links and text may be created, deleted or changed. Some links between slides represent physical, spatial movements whilst other links may be continuations of a

RUGGLES,C. L. N. 1988. "Software for the Leicester Interactive Videodisc project", in Rahtz, S. P. Q.• (ed.), Computer and Quanlilalive Melhods in Archaeology 1988, International Series 446, pp. 523-542. British Archaeological Reports, Oxford.

22

5 LIVE update: archaeological courseware using interactive video Clive Ruggles (School of Archaeological Studies, and, Department of Computing Studies, University of Leicester, Leicester LEJ 7RH)

Jeremy Huggett (Department of Archaeology, The University, Glasgow G12 8QQ)

Steven Hayles Howard Pringle Ian Lauder (Department of Pathology, University of Leicester, Leicester LE2 7LX)

5.1

Introduction

authoring systems. The philosophy of the project is that these aims can be achieved through standardisation. Hence the project is concerned with the standardisation of concepts for the development and execution of courseware using multi-media technology in general and IV technology in particular. 'Academic portability', that is, applicability to a wide variety of subject areas where teaching methodologies may be very different, is an important concern. Pathology is a training area where a rigid goal-oriented approach to computer-aided learning (CAL) prevails: archaeology is an area where open-ended exploration is generally much more valuable. The combination of these subject areas, together with others (notably geography) that are also involved in the current phase of LIVE, has helped considerably in preventing too discipline -specific an approach. The current phase of LIVE is sponsored by IBM UK Ltd., the Leverhulme Trust and the CEC COMETT initiative.

Interactive video (IV) technology, whether analogue- or digital-based, has huge potential in a wide variety of teaching and training applications where the skills of interpreting visual material are fundamentally important, as in the teaching of archaeology. It is ideal for independent student use or, with networked workstations sharing a single resource, for tutorial group teaching. The initial phase of the LIVE (Leicester Interactive Video in Education) project was sponsored by the DOC/Computer Board under their Computers in Teaching Initiative (CTI). Its aim was to produce a system using interactive video (IV) technology to teach archaeology to undergraduates. A laser videodisc ('The Archaeology Disc') was produced in 1988, together with basic authoring facilities enabling tutorials to be written for the RM Nimbus PC 1 computer linked to a Philips videodisc player (Ruggles 1988). The LIVE project was not the only project at Leicester University during the late 1980s concerned with the development of teaching and training applications using interac tive video. A collaboration between nine UK University Pathology Departments (Mercer et al 1988) produced two videodiscs for teaching human and vetinary pathology ('UK PATH 1' in 1987 and 'UK PATH 2 ' in 1989) together with the 'VIPA' authoring system which has enabled several hundred tutorials to be written for the BBC Master computer linked to a Philips videodisc player (CTICM 1989). Although there were considerable pedagogical differences of approach between two such diverse subject areas, interacti c2>=== =-=-=-==---~=--=-=--=--

~2 :

8

{( 2) 3):

. . r . (4): 2 :.,1: : .•-~·---;-c-,.:-·-

E F

(e)

= 1"'=

:

:

:

.

.

:

:

2 : 2 : create

2 3

5 5 5 6 7 8

4

9 9

0 1

1

1

2

1 l 2 2 2

2

3 3 3

3 4 4 5 5

2 3 3 3 4 5 5 6 6

1 1 1 1 2 2

1 1

selected.

Conclusions

This article presents only the first stage of a larger research project and the program is missing many desirable features. It will be necessary to add the ability to edit the matrix, move single contexts, or groups of contexts, without losing the definitive relationships. Much has already been done to present errors in the data, and the next stage of development will focus on these matrix editing features. However important these features are, it must be remembered that the fundamental importance of a computer generated stratigraphic matrix is not to duplicate present practice. The purpose of this matrix generation program is to provide a basic and rigorous stratigraphic model with an interface which integrates this model with other forms of information. Programs based, as this one is, on demonstrable adjacency and clear rules of relation provide a baseline for further interpretation of stratigraphic structure. Interpretive stratigraphic models based on such provisional matrices offer greater compatibility than interpretive models which develop from the primary record in that basic rules of redundancy may vary without documentation. Starting with a base model moves the debate from drawing conventions to issues of structural development and association. By using a Hypertext interface the stratigraphic model has a further interpretive advantage. Context nodes on

1 1

SQL>select from contexts where nochildren-1 or ovlevel - treelevel+l; 0

2 3

2

Thus the SQL command to select valid linkages to be drawn reduces to:

ID

1

1 1 1

TREELEVELOVLEVELNOVALIDLINKS

In order to enable the drafting of the resultant matrix these SQL commands are embedded in suitable SuperCard code, the final form of the output being shown in Fig. 8.7. Elementary drawing rules have been established which enable a relatively clear chart to be produced automatically. These rules obviously require further enhancement if they are to produce more than a provisional matrix. Provision must also be made for insertion of interpretive relationships and editing of the graphic representation. However, with the limitations outlined above, the matrix does seem to provide a useful tool for site interpretation. Testing of the procedure against two sites of differing complexity demonstrates the effectiveness ofthealgorithms,especially in the second case where more than 500 contexts are modelled.

• from contexts;

OVERLAYS TREELEVEL 1 0 1 2 3 1 1 4 5 1 1 6 7 1 1 8

0

12 records

The resultant table for the above data set is show in:

ID 0

OVERLAYS

4 5 6 6 7 8

view contexts

SQL> select

ID

view validlinks(id,nolinks) as select id,count(overlays) from contexts where ovlevel =treelevel+l or nochildren - 1 group by id;

View created.

47

ROBIN BOAST AND DAVE CHAPMAN

the matrix exist as objects which can be related to other Hypertext objects. This allows the interpretation of a matrix to be substantiated by reference to data, graphics, text or primary site records. The creation of the stratigraphic matrix in Supercard offers a presentation of the matrix, and subsequent interpretive matrices, that directly integrates the stratigraphic model with the information that supports or challenges that model. Therefore, the purpose of this research is not simply to provide a quick and easy means of drawing a Harris Matrix, but is to provide a means by which stratigraphic models may be developed from basic principles and integrated with relevant information. Such an environment allows for the critical relation of data and interpretation, and the ability to deconstruct arguments back to a basic record. Though we feel that we have achieved many of these goals, it is clear that much is yet to be done. However, the major focus of this research remains the integration of archaeological information and its effect on modes of interpretation.

Archaeology of the Museum of London (DUA) and the Department of Photogrammetry and Surveying at University College London (UCL P&S) under funding from the Autocarto/RICS education trust.

Bibliography BARKER,P. 1982. Techniques of Archaeological Excavation. Batsford, London. BISHOP, S. & J. D. WILCOCK 1976. "Archaeological Context Sorting by Computer: The Strata Program", Scknce and Archaeology, 17: 3-12. R. & C. OR'ION 1989. "Hypercard as a teaching tool", in Rahtz, S. P.Q. & Richards, J. D., (eds.), Computer Applications and Quantitalive Methods in Archaeology 1989, International Series 548, Oxford. British Archaeological Reports.

GRACE,

J. L. 1988. Relational database management for microcomputers: design and implementation. Holt, Rinehart and Winston, New York.

HARRINGTON,

Acknowledgements

The system described has been implemented as part of a research collaboration between the Department of Urban

Principles of Archaeological Stratigraphy. Academic Press, London, second edition.

HARRIS, E. 1989.

48

8.

SQL AND HYPERTEXT GENERATION OF STRATIGRAPHIC ADJACENCY MATRICES

Unexcavated

Surface 0

SQL> UpdAte context treelevel-0 where id •01

SQL> ••lect

SQL> update context aer Ht t reelevel-2 where id in (select overlay a f r0ffi context where treelevel-2 - 1)

SQL> update context aer Ht tr-level•! where id in (Hlect overlay• frca context_•er where treelevel•l-11

_mer Ht

• trOffl context_mer;

mer -

ID OVERLIIYS TREELEVEL

7 record•

updated.

16 record•

SQL> Hlect • frca where treelevel>llull:

context_mer

_••r

ID 0\/ERLIIYS TREELEVEL

ID OVERLIIYI TRSELEVEL

25 record•

updated.

SQL> Hlect • fr011 context where treelevel>Null:

Hlected.

Ht

., Furt.her laca•tion (lrE)

SQL> update context mer Ht treelevel•l when id in f••lect overlay• trOffl context where treelevel•3-1 J I record•

update treelevel•4 SQL>

_mer

upd.llted.

context

••t

mer set

IQL> update context mer treelevel•5 where id in (1elect overlay• from context where t reelevel•5 - 1J

-

(Hlect overlay• where treelevel-4

from - 1J

, record•

.

updated

context

mer -

3 record•

111er -

SQL> update context mer tnelevel•6 where id in t••l•ct overlay• tro,n contex.t mer where treelev;l•6 - 1J

updAted. 1 record

SQL> Hlect • froa when treelevel>llulll

conuxt

_ mer

SQL> select • !r0ffl context where tre•l•v•l>Null:

_m•r

SQL> •elect

where

••t

• frOffl

context

t reelevel>Null

updated

.

_ mer

:

SOL> Hlect

• fr0ffl

context _mer ID OVERLIIYS TREELEVEL

ID O\/ERL1'YS TREELEVEL

ID OVERLIIYS TREELEVEL

where

treelevel>Null: ID OVERLIIYS TREELEVEL

t

u,

ttt

t

ttt

"' SOL> update context_••r when id in Uelect overlay• fr.,.,

o record•

Ht

treelevel•7

context_mer

where

tnelevel•7

- 1J

updated.

Figure 8.4: Building on the stratigraphic sequence

49

ROBIN BOAST AND DAVE CHAPMAN

OnHcavated

surfac• 0

IQL> in1ert (id,overlay1) 1 r.cord

IQL> updat• cont .. t _,..r Ht t..-l•v•l•l id in (Ml.ct overlay, from contHt •r -r• trfflevel•l - 1) -

into contHt mer valuu (1, S)1

-r•

cr ... t•d .

SOL> Hl.ct

IQL> update contaxt_Nr Ht treelevel•2 where id in (HlKt overlay• frcontext ,.., wheN trNlevel•2 - ll -

• from contHt_,..r1 7 record•

updated .

17 NCOrd•

updated.

ID OVERLAYS TI\EELEVEL IQL> Hl.ct

• from contHt

_•ri

HlKt

IQL>

ID OVEIU.JIYS TI\EELEVEL

• frOlft context_1Mr1

ID OVERLAYS TIIEELEYEL

H9

26 record.a ••l•cted

2, record•

.

,elected

.

IO f"llrthU'

racewetion (In)

SOL> update where id In

contHt

9 record•

_,..r

trHl•v•l•l

Ht

SOL>

Updat•

contHt

where id in

( Hlect overlay, trHlev.l•l-1 l

from contnt

,.., -

updated .

vller•

( ■ elect overlay ■

trHlevel•4 9 r.cord1

... , Ht from context

- ll

trHlev

••t

e l-4

SQL> update context mer tr••l•v•l•S id In (Hl.ct overlay ■ from context _,.., vh•r• trHlevel•S - 1)

vhere

mer where

-

updated .

• from context_mer;

SQL> 1elect

record ■

update id in

contHt

overlay ■

where trffleve1-, reco rd ■

updated.

SOL> ■ elect • from context _mer1

• from context _Mr;

ID OVERLAYS TREELEVEL

ID OVf!RLAYS nEELEVEL ID OVEIU.JIYS nu:ELEVEL

Ht

2, record•

'

n,or Ht trNleve1-, tra.. conteJCt mer - 1) -

updated .

• Crom context _ mer;

SQL> aelect ID OVERLAYSTIU:ELEVEL

t••lect

9

9 SQL> ••l•ct

SOL> vllere

999

999

s

s

999

s

■ elected .

2, 2,

record ■

aelected.

Figure 8.5: The effect of coding errors on the stratigraphic sequence

50

record ■

••l•cted.

8.

SQL AND HYPERTEXTGENERATIONOF STRATIGRAPHICADJACENCYMA1RICES

re• u11•

Ca111

s,,..u

ck 11■91 •

■ W M ■ trlN

/\ ~.;:(,;.

~

• .•·•··•..·•'-··~•.v

,.......... ,...

••in•- ,......

( 1 COtllllCL- Hl trw11Ytl11li l

[jJ

Figure 8.6: SuperCard environment for Matrix manipulation

ID

DURRrchaeolo

database

=-=---==-~-----=-- - 01 -------

Figure 8.7: Adjacency matrix for sample dataset

51

-----

- --

ROBIN BOAST AND DAVE CHAPMAN

52

9 A new graph theoretic oriented program for Harris Matrix analysis lrmela Herzog Irwin Scollar (Rheinisches Amt fur Bodendenkmalpjlege, Colmantstr. 14, D 5300 Bonn 1, West Germany)

9.1 The Harris diagram

9.2

A relative chronology of a site may be obtained by analysing archaeological stratigraphy. This well-known method in archaeology was first put on a systematic basis by Harris (Harris 1975, Harris 1977). He also proposed visualising the relative chronology of a site with a diagram that shows all the stratigraphic relations. Harris called this diagram the Harris Winchester matrix, but mathematically speaking a matrix is a different thing so that we prefer the term Harris diagram. The following time relationships may exist between two layers 1 and 2 (Orton 1980, p. 65-80): 1. 2. 3. 4.

Existing programs for Harris diagram generation

Shortly after the Harris diagram had been invented first attempts were made to provide computer assisted generation of the diagram, because there is a great amount of manual work needed to establish a diagram for some hundred layers. The computt"Zprograms for Harris diagram generation which we have seen all have some disadvantages: • The STRATA program (Bishop & Wilcock 1976) produces a diagram that only shows the positions of the layers but not their relationships. Interactive modification of the data is not supported. • GAMP(Day 1987, program version 3.1, 5/4/88) does not support contemporary or equal relationships. No printable diagram output is created. • GNET (Ryan 1989) is designed for a Sun or a DEC VaxStation and a PostScript laser printer, which are not at many people's disposal. A PC version is available, which requires a graphic card (CGA or better) and a mouse. It supports equal but not contemporary relationships. (1be information on the PC version of GNETwas supplied by a referee.) • The ORPHEUSmatrix generator (Williams 1989) is a full screen editor for Harris diagrams. It was still being developed when its handbook was written. The example diagram only shows the layer positions but not their relations.

1 is later than 2 (above) 2 is earlier than 1 (below) 1 and 2 are contemporary based on a priori knowledge There is no direct relationship between 1 and 2.

Another relationship is useful in practice: 5. 1 and 2 are equal (equivalent). An example is a wall observed in two different cuttings.

Fig. 9.1 illustrates how the Harris diagram shows these relationships. Harris does not differentiate between equal and contemporary relationships. If layer 1 is later than 2 and 2 is later than 3, it follows that 1 is later than 3. In this case the relationship '1 is later than 3' is called indirect, because it is not based on direct observation. If a direct above- or below- relation can also be established indirectly, this relation is called redundant and for reasons of clarity it is normally not included in the Harris diagram. The layers and their relationships in general do not lead to a unique representation in Harris diagram form. There may be variations in the horizontal sequence of the layers as well as in their depth position, as Fig. 9.2 shows (cf. Dalland 1984 ). The excavator who draws the diagram tends to choose the horizontal sequence of the layers such that spatially close layers in the field appear near each other in the Harris diagram. Also crossings of lines indicating relationships are avoided as far as possible. However, there are certain situations where crossings cannot be prevented in the Harris diagram. For layers whose depth position may vary over a range when taking only the time relationships into account, a look at the layers' artefactual content may help to determine their depth position: For example two layers with approximately the same distribution of sherd types should be set approximately on the same horizontal layer.

Most of these computer programs require that the layers be identified by numbers only, but in practice it is often convenient to use alphanumeric identifiers like '45a' or '141 '. The most serious problem seems to be the layout of the diagram. This is often fixed, and does not take into account that relationships in general do not lead to a unique diagram. The problem of crossings and crossing minimization is not addressed. When the relations of the layers are shown at all, the layers are connected by straight lines so that in a situation such as in Fig. 9.1, (e) a crossing occurs that could be ~voided. Cycles are often resolved by deleting an arbitrary relation within the cycle.

9.3 The Bonn computer program for Harris diagram generation The Bonn computer program was developed for IBMcompatible PCs. Dynamic data structures are used so that small data sets need less memory than large data sets. It is independent of the computer's graphic card. Any printer supporting the IBM graphic character set may be used for diagram output.

53

IRMELA HERZOG AND IRWIN SCOI.LAR

[Q_qJ qJ_gJ

Q]====QJ

c5l±J dJBJ (a)

(b)

(c)

(d)

(e)

Figure 9 .1: Examples of Harris diagram representations of stratigraphic relations:

(a) 1 is above (later than) 2, 2 is below (earlier than) 1. (b) 1 is equal to (contemporary with) 2. (c) 1 is above 2 and 3. (d) 1 is above 3 and 4, 2 is above 4. (e) both 1 and 2 are above 3 and 4.

(c)

(b)

(a)

Figure 9 .2: Differing representations of the same data

(a) example of a small Harris matrix (b) same relationships as in (a), but changed horizontal sequences of the layers (c) same relationships as in (a), but depth of 3 changed.

9.3.1 Data entry and change

above, below and contemporary relations are merged and the layers are connected with a horizontal double bar, see Fig. 9.3. If the user enters the relation that 1 is above 2, the program automatically establishesthe relation that 2 is below 1. Similarly, if 1 and 2 are already set contemporary and the user sets 3 contemporary with 2, then the program knows that 1 and 3 are contemporary. Any direct relation maybe deleted. This is quite easy for above or below relationships. But if a contemporary relation is erased a more complex operation results: for exampleif 1, 2, 3 and 4 are contemporary and the relation 1 is contemporary with 2 is to be deleted, the user must decide whether 3 and 4 are contemporary with either I or 2. If an equal relationsay between 1 and 2 is deleted, then the user must decide for each above or below relation of 1 and 2, whether it belongs to only one or both layers. The contemporary relationsof two layers whose equalrelation is being deleted are dealt with in two different waysdepending on whether the layers remain contemporary or not. Most changes in the layer data base affect the corresponding relations. For example, if a layer is erased, all its relations are deleted, too. When two layers are merged, their relations must be merged as well. Conversely, if a layer is split, for each relation of the source layer the user is asked to choose whether it belongs to one or both new layers. Here

The program HARRIS developed in Bonn stores the layers in a data base. Each layer must have a unique short identifier (alphanumeric, up to 8 characters) and may have a label (up to 40 characters), for comments. The user of the program may choose to • • • • •

define a new layer change the name or label of a layer delete a layer split a layer so that two layers result merge two layers.

If the layer names are consecutive numbers then the program can generate the names automatically. The user may choose variable length names (example: 1, 2, . .. , 100) or fixed length names (example: 001, 002, .. . , 100). It is also possible to define the layers as one proceeds, i.e. when establishing a relation. Above, below, ·contemporary and equal relationships are supportedby the program. Since Hanis did not distinguish between contemporary and equal relationships, the difference between the two tenns must be explained: if two layers are contemporary they will be set on the same horizontal line of the Harris diagram. If two layers are set equal, their

54

9.

A NEW GRAPH TIIEORETICORIENTEDPROGRAMFOR HARRISMATIUXANALYSIS

(c)

(a)

Figure 9.3: Effects of contemporary or equal relations on Harris diagrams

(a) small Harris diagram without contemporaryor equal relations. (b) diagram as in (a), but layers 2 and 6 are set contemporary.

(c) diagram as in (a), but layers 2 and 6 are set equal. methods for answering the questions b) and c) may be found in almost any book with a chapter on graph theory (Aho et al 1983,Reingoldet al 1977), in the latter case these methods must be extended.

again different measures are taken depending on whether the newly created layers are contemporary. A protocol file listing the layers and their relations can be generated. A stratigraphic data set is stored by the program in several binary files. Some archaeologists are working with data bases storing layers and their relationships. To avoid enteringdata twice, a utility is provided for convertingAscii files in the HARRISprotocol file format to files readable by the program. As Ryan has mentioned,a Harris diagram has some similarity with a mathematical structure called a directed graph, especiallyif only the above and below relationshipsare considered. Therefore the data structuresproposed for directed graphs were used to store above and below relationships. This enabled us to use standard graph algorithms to solve most problems of data checking.

9.3.3 Automatic data checking

Since it is only possible to lay out the Harris diagram if the data set is consistent, the layout must be preceded by a suite of automatic checks. It is not sensible to create a Harris diagram showing two or more separate components. Therefore the program first makes sure that the data set is connected (a standard procedure in graph theory) and, if not so, a warning is issued and the components can be listed in a protocol file. With the help of the conversion utility separate data sets may be created for each component. It is possible to proceed with data checking and layout even if the diagram consists of more than one component. Afterwards the program looks for cycles, i.e. for layers that lie indirectly above or below themselves. If cycles are found, the program lists for each cycle the layers that form it. The program cannot decide which of the relations of the layers in the cycle is erroneous, therefore the user is asked to look for the error in his data and to erase the erroneous relation. A layer may be involved in more than one cycle, see Fig. 9.4. Again there are methods known in graph theory to solve problems concerning the cycle structure of a graph. The following problems may be solved (listing according to increasing complexityand difficultyof the problem (Reingold etal 1977)):

9.3.2 Posing questions about the stratigraphic data set

The user may check the stratigraphicdata set by asking the followingquestions which will be answeredby the program: a) Which layers are directlyabove, below,contemporary with or equal to a given layer? b) Which layers are above (below) a given layer? c) Does a given layer overlie (underlie) another given layer? If so, the direct relationsconnecting the source layer with the target layer may be listed by the program. d) Which layers have neither above nor below nor contemporary relations? To answer questions a) and d} is trivial from the programmer•s point of view. The answers of questions b) and c) come in two flavours: the user may or may not take contemporary relations into account. In the former case the

1. Determine whether a given graph contains a cycle 2. Determine a fundamentalset of cycles 3. Determine all cycles

55

lRMELA HERZOGAND IRWINSCOLLAR

, 1

·•

I

I

2

a

I

t 6

4

3

i

5

,.

,

•

,J i

8

7 •

1

Figure 9.4: Two cycles that have two layers in common:

Cycle a: 1, 4, 5, 7, 3, 1 Cycle b: 2, 4, 5, 8, 6, 2 from the preliminary depth coordinates. When checking whether A contemporary with B is a bad contemporary relation, then either:

The program lists the fundamental set of cycles. From these cycles all the other cycles in the graph may be reconstructed. For example, in Fig. 9 .4 two fundamental cycles are listed. Another cycle may be generated by combining these two cycles, namely

• A and B have the same depth coordinates. Then A cannot be above Band B cannot be above A, therefore

1, 4, 5, 8, 6, 2, 4, 5, 7, 3, 1.

the contemporary relation is not bad or • the depth coordinates are different, e.g. A's depth is less than B's depth. Then, as in redundancy checking, all paths from A down to the level of B are generated. If B is encountered, the contemporary relation is considered bad.

This example shows that a listing of the fundamental cycles is adequate. Listing all cycles will lead in general to less clarity. When checking for cycles, preliminary depth coordinates are assigned to the layers without taking the contemporary relations into account. These preliminary depths help to reduce the effort when looking for redundant links. For each below link (from layer A down to layer B) a check is made to find out if another path connects these two layers. The program follows all paths from source layer A downwards (except the direct path to B) until these paths hit the preliminary depth of B. If B is encountered, the relationship A above B is redundant and therefore erased. The program looks for bad contemporary relations that are in conflict with (indirect) above or below relations. For example, if the user defines the relations 1 above 2, 2 above 3 and 1 contemporary with 3, then the contemporary relati~nship will be considered bad by the program and will be deleted. Above or below relationships have higher priority

Conflicting contemporary relationships must also be resolved. If, as in Fig. 9.2, (c), it is requested that 2 be contemporary with 6 and 4 be contemporary with 5, then it is not possible to draw a diagram which allows these conditions. In this case the two contemporary relations are called conflicting and again the user is asked to choose the erroneous relation. In our example a cycle including contemporary relations results: 2 is above 5, 5 is contemporary with 4, 4 is above 6, and 6 is contemporary with 2. Therefore it is obvious that conflicting contemporary relations can be found via an extension of the nonnal fundamental cycle finder. The data checking phase may take several minutesfor a data set with several hundred layers. Therefore it is not possible to check after the establishment of each new relation whether or not the data set remains consistent. Only the direct relations are checked on entry, i.e. if 1 is directly

thancontemporary relationsbecausetheformercanbe observed directly by the excavator, whereas the latter are only subjective conclusions and therefore more open to error. When perfonning this check, the program again benefits 56

9.

A NEW GRAPH THEORETIC ORIENTED PROGRAM FOR HARRIS MATRIX ANALYSIS

above 2, it is not possible to establish the relationships 1 is below 2, 1 is contemporary with 2, or 1 is equal to 2.

Instead, the user is encouraged to establish contemporary relations. Hard-copy output may be obtained on a plotter or printer which supports the HPGL graphics language.

9.3.4 Laying out the Harris diagram The Harris diagram has some similarity with directed graphs, and especially with so called k-level hierarchies (Di Battista & Nardelli 1988, Tumassia et al 1988). The main difference is that a Harris diagram allows structures as in Fig. 9.1, (e), which we call ff-structures because of their fonn. In order to minimise crossings in the diagram, it is necessary to detect these structures. Unfortunately, combinations of H- structures may occur, so that the detection of ff-structures becomes quite difficult, see Fig. 9.5. At the time of writing, the program is able to detect simple H-structures only. Computationally the problem is solved by creating a pseudo-layer for each ff-structure which is represented as in Fig. 9.6 on output. It is quite easy to assign depth coordinates to the layers. There are methods known in the literature (Dal land 1984) which can be readily extended so that contemporary relations are taken into account. In general, more than one depth coordinate configuration is valid. The program positions the layers as high as possible in the diagram. The assignment of horizontal layer sequences in the diagram is by far the most difficult problem in automatic layout. If the diagram cannot be drawn without crossings, one may want to minimise the number of crossings. But it has been shown that this problem is NP-complete (Di Battista & Nardelli 1988). Therefore, we can only use heuristic methods to achieve this goal approximately. The user may choose to detennine the horizontal sequences of the layers manually or automatically. The manual layout is assisted by the program in that the user-selected sequence is saved and becomes the default in the next iteration. Automatic layout is achieved via the PQ-tree concept as proposed by Booth and Lueker (1976) and applied to k-level hierarchies by Di Battista and Nardelli (1988). Finally, the program computes horizontal positions for each layer. Then the printout showing the layers and their relations is prepared. The diagram is created using only the IBM character set, so that the program works independently of the graphic card, and the diagram may be printed on almost any printer. With most large data sets the breadth of the diagram will be greater than the maximum printer line length. Therefore sidewise output of the diagram is supported. It is quite difficult to include crossings in the diagram. There are cases when quite a few lines corr~sponding to crossing relations extend over the full breadth of the diagram. This certainly does not enhance the readability of the diagram. Therefore crossings are currently not plotted. Any layer that has a crossing relation is marked with a double frame. A list of relations which could not be displayed is generated at the bottom of the diagram. Once a diagram has been created, the user may list all layers at a given depth horizon. Also, the depth of layers may be changed. This means that from Fig. 9.2(b) 9.2(c) may be generated and vice versa. It is not recommended to use this feature extensively, because after a new layer or new relations have been entered, this depth information gets lost.

9.3.5 Practical experience The Bonn program was used to reproduce the diagram for the South Gate site as published by Harris (Harris 1975, Fig. 29). This data set consists of 406 strata and 856 above or below relations. The diagram in the Harris publication allows only rough guesses about contemporary relations, therefore these relations were omitted. Layers the frames of which are connected with a horizontal double bar were considered equal in the sense of our paper. One error was found (layer 266 occurs twice in the diagram). After correcting this error, no other problems were detected during the data checking phase. Computation times are given in table 9 .1. The diagram produced by the program is 110 cm long and 40 cm wide. Additionally, an excavation in the city of Xanten in the lower Rhineland was analysed. The data set consisted of about 1000 strata and 6900 relations. Some hundred redundant links were erased. Several cycles were found, and from one erroneous relation several cycles often resulted. Also about 50 bad contemporary and a very few conflicting contemporary relations were found. Because of the excellent standard of the excavation records, these errors could be corrected within a few hours. During the layout phase, the program detected 22 ff-structures. Computation times for a version of the Xanten data set with 881 redundant links and 40 bad contemporary relations are given in table 9.1. The resulting Harris diagram shows 67 depth horizons and is about 3.5 metres wide. In general, computation time depends very much on the speed of the disk storing the stratigraphic data set. Layout preparation speed is highly correlated with the number of layers having more than one above relation and on the number of crossings. A mathematic coprocessor does not decrease processing time.

9.3.6 Planned extensions A remaining problem is the detection of combined Hstructures. Also it may be possible to show at least some local crossings in the diagram. If cycles or conflicting contemporary relations are encountered, the program now lists the layers that are part of the cycle and asks the user to cut one of the relations. Only after all cycles have been resolved successfully, is it possible to start the diagram layout. An alternative currently not supported would be to lay out the diagram and replace the cycles with a special symbol. Another planned extension is the introduction of a phase concept. This enables the user to incorporate his knowledge and theories about the phases of the stratigraphic data set into the analysis. The administration of phases will include the generation of new phases, deletion, splitting and merging of existing phases and, of course, the attachmentor detachment of phases to layers. The phases have to be checked, too, in that any phase cycles must be removed. The indication of phases will be part of the final diagram.

57

lRMELA HERZOG AND IRWIN SCOLLAR

4

(a) 1 overlies: 2 overlies: 3 overlies: 4 overlies:

2

3

5

6

7

(b) 1 overlies: 4, 5, 6. 2 overlies: 4, 5, 6. 3 overlies: 5, 6, 7.

5, 6. 5, 6, 7, 8. 7, 8. 7, 8, 9.

Figure 9.5: Two examples of combinations of H-structures.

1

2

3

4

1

5

3

2

7

6

4

8

9

(b)

(a)

Figure 9.6: Program representation of H-structures:

(a) Fig. 9.l(e) (b) Fig. 9.5(a) A very useful extension would be the combination of the program with a finds' data base, so that the finds from the layers may be visualised in the Harris diagram. The program is distributed with Version 4 of the Bonn Seriation and Archaeological Statistics Package at no extra charge.

Electronics and Computer Science, University of Southampton. DI BATilSTA,G. & E. NARDELll1988. "Hierarchies and Planarity Theory", IEEE Transactions on Systems, Man, and Cybernetics, 18 (6): 1035-1046. HARRIS,E. C. 1975. ''The Stratigraphic Sequence: A Question of Time .., World Archaeology, 7: 109-121. HARRIS,E. C. 1977. "Units of archaeological stratification", Norw. Arch. Rev., 10: 84-94. with comments: 95-106.

Bibliography AHO, A. V., J.E.

HoPCROFf, & J. D. UU.MAN 1983.

ORTON,C. 1980. Mathematics in Archaeology. Collins, London.

Data

E. M., J. NIEVERGfl..T, & N. DEO 1977. Combinalorial Algorithms-Theory and Practice. Prentice Hall, Englewood Cliffs.

REINGOID,

Structures and Algorithms. Addison-Wesley. BISHOP,S. & J. D. Wn..cocK 1976. "Archaeological Context Sorting by Computer: The Strata Program .., Science and Archaeology, 17: 3-12. Boom, K. S. & G. S. LUEKER1976. "Testing for the Consecutive Ones Property, Interval Graphs, and Graph Planarity Using PQ-Tree Algorithms", Journal of Computer and System Sciences, 13: 335-379. DALI.AND,M. 1984. "A procedure for use in stratigraphical analysis", Scottish Archaeological Review, 3 (2): 116-127. with comment by E.C. Harris: 127-133. DAY,G. A. N. 1987. "GAMP- an archaeological matrix program ... Computer Science BSc dissertation, Department of

RYAN,N. S. 1989. "Browsing through the Stratigraphic Record", in Rahtz, S. P. Q. & Richards, J. D., (eds.), Computer Applications and Quantitative Methods in Archaeology 1989, International Series 548, pp. 327-334. British Archaeological Reports, Oxford. TAMASSIA,R., G. DI BATilSTA,& C. BATINI 1988. "Automatic Graph Drawing and Readability of Diagrams", IEEE Transactions on Systems, Man, and Cybernetics, 18 (1): 61-79. WlLUAMS, P. 1989. "Orpheus Matrix Generator. Handbook of Orpheus/Delilah".

58

9.

A NEW GRAPHTHEORETICORIENTEDPROGRAMFOR HARRISMA1RIXANALYSIS

Automatic data check

South AT-286, 8 Mhz. Data on Harddisk with 40 ms seek 1 min 17 s

Layout preparation

5 min 25 s

Creating matrix file Total

44s 7 min 26 s

Gate AT-386, 20 Mhz. cached Data on Ramdisk 10 s

AT-286 Harddisk

Xanten AT-386 Ramdisk

28min

6 min

45 s

1 h 32 min

17 min

9s 1 min 4 s

36min 2 h 36 min

5 min 28 min

Table 9 .1: Compu talion times of the program HARRIS for the data sets Xanten and South Gate, Xanten consisting of about 1000 layers and 6900 relations and South Gate of 406 layers and 856 relations.

59

IRMELA HERZOG AND IRWIN SCOLLAR

60

10 The ArcheoDATA System -towards

a European archaeological document

Daniel Arroyo- Bishop (GDR 880, "Terrains et Theories en Archeologie" du CNRS, Universite de Paris I, lnstitut d'Art et Archeologie, 3, rue Michelet, 75006-Paris, France)

10.1 Introduction

10.2 Site recording

The aim of the ArchooDATA System is to create a minimal context that permits not only the publication of better excavation reports, but also to improove the recording, analysis and conservation of archaeological data. To this end the ArchooDATA System contains a series of methodological elements for the organization of the archaeological record in its entirety, both on a national and international level, which reflects archaeological realities and assures a better opportunity for research. To achieve this, a system has been devised that copes with the problem of archaeological research in a global manner, and not as a myriad of isolated entities. In this way, we may one day be able to create a European archaeological document. ArchooDATA System is the combined effort of the GDR 880 of the Centre National de la Recherche Scientifique, the University of Paris, the Sous-Direction de l' ArcMologie of the Ministry of Culture and field archaeologists, in order to devise a system which will improve conditions for archaeological research. It is principally funded by the CNRS through it's Action Thematique Programoo. The theory behind, and the functional basis of, the ArchooDATA System have been published in the CAA 89 conference proceedings. Its consultation is indispensable to the comprehension of this second and complementary paper. We will begin with a rapid overvue of the ArcheoDATA System which incorporates the following:

We have adapted the Universal Transversal Mercator {UTM) coordinate system, which are used for the recording of site survey data, and applied them directly to the excavation grid itself. For fieldwork we have developed two different ways of spatially recording the same archaeological phenomenon. The Universal Metric Unit, which rationalizes data handling in a global sense, and the Relative Metric Unit, which although it uses the same basic units, structures the same space in a more natural and intuitive manner. The system used to identify any given point on the excavation grid should locate objects and phenomena in a natural way and the archaeologist should be able to easily visualise and utilise them spatially within the archaeological site. An effort has thus been made to take into account, in the grid numbering proposed, of everyday habits and routines. The unit numbers used are not too large so as not to be too complicated and difficult to retain. Instead, they rely on the successive re-use of the same basic block each time but on a smaller scale. Common to both systems is their structuring in area, subarea and square meter, with the UTM unit being reserved for the UMU system.

10.3 The RMU system As this was the basic system which was presented last year, we will only go over its underlying structure. (Fig. 10.1) The RMU grid system is considered 'relative' in that it is not directly based on the UTM coordinates, but on geographical points chosen independently by the excavator himself. Even though an exact North-South orientation is not obligatory, it is recommended, as this would give the added advantage of enabling a later conversion of the RMU's into UMU's. The Northern orientation can vary up to 44 degrees to either side of the true geographic position and as long as this orientation is found on top of the grid.

• A code which relates the archaeological site with an administrative space by the use of international postal and telecommunications codes. • A code that relates the archaeological site with geographic space through the use of the UTM coordinates. • An inventory system that relates the finds to the place where they were found and that simplifies the selection and grouping of the archaeological material. • Ten series of recording sheets and folders that cover different phases of archaeological work, starting with surveys, through excavation and finally to storage. • A recording system which can be used manually, but which has been archaeologicaly reasoned and structured, from the beginning, to be efficiently computerized.

10.4 The UMU system The numbering of the UMU grid is based on the international UTM coordinates of latitude and longitude (Fig. 10.2). Each grid block is comprised of one hundred units numbered from zero to ninety-nine and related directly to the hundreds, tens and units of the UTM coordinates. It is by orderingthe coordinates in pairs of longitude and latitud~ that any one point is designated. The ensuing number locates the exact unit directly on the earth's surface. When the excavation

61

DANIEL ARROYO-BISHOP -------

1 Metric Unit ---

lM

JOM -------. 6

---

7

10

11

21

·..:n

-------1 41 ··.

Subare:-------

SJ

61

71

123 074 1529

11 91

1

2

3

/ 4

..r ·

6

I

9

10

"'

ArcheoDATA l'jllmeut,U

v. 2.0

...

11

21

s

100

m

31 41

1 Areii SJ 61 71

81 100

91

i,

100M

------

-

Figure 10.1: The RMU grid structure based on independently chosen 'relative' coordinates (not strict UTM) and adapted to a more intuitive, left to right and top to bottom type of organisation recording

OPS (Global Positioning System) satellite positioning units which have recently become available.

grid is set up, the number of each square is determined directly by its UTM coordinates. Since no two places can have the same absolute coordinate, all excavation data is unique and it can be related to all other data recorded in the same way. The inventory number structure is exactly the same as for the RMU system (Fig. 10.1). The structure of both systems is identical as this is necessary for the future re-use of the data. Whatever system is chosen by the archaeologist, provision has been made for the automatic conversion of data from one system to the other (Fig. 10.3). This of course can be handled automatically in the background by the computer without any intervention by the archaeologist. If this system is without a doubt the most interesting for the future, it is reserved at present to large or long term excavations, as the system needs great precision when set up, and the means to do so, by an archaeologist, will not be generally available in the foreseeable future. The archaeologist will probably use one of the small hand held

10.5 New recording sheets After completing the recording sheets for the stratigraphic unit based recording system, work has been directed towards making up the three-dimensional recording sheets and folders used in the UMU and UMR systems. These sheets are innovative in that they clearly imply that, even in prehistoric period three-dimensional recording, the same archaeological hierarchy of feature and structure terms should be used, even if the stratigraphic unit is absent or difficult to define. A posthole made up of several stratigraphic units or one where it is barely discemable though texture differences, are both the same archaeologically: a feature. A semi-circle of postholes (features) should be interpreted as a structure, be it in a historic or a prehistoric excavation .

62

10.

TIIE ARCH11ODATA SYSTEM -

TOWARDSA EUROPEAN ARCHAEOLOGICALDOCUMENT

I ArcbeoDATA; I .,..,_....,

v. 2.0

Recordingby UMU 1. .......... -.---.-.......--.--.--...--.----.-.---.-~-.--.--...--~-----,---,

'

-

"

4612,00(

2672023 1529 4681.000_ :::::

,.

"

.. -,. 4680,270 -

-I

."

'

mm :urn:a::: W:'.:m nm w:::::::~:;;;;;=::::::;t::::~:;:: :::::::: ::::~::: :::: :::: ::::> :;:; ::::: ::~:: 4681,00
.

+

+ ··· ······

J

l,

j

!

l,

l,

I

I

L_

l,

l

·· · · ; ·· ··

l,

; , ,1111 11111111l1 11,1,,, ,1 ,,11 11, 11f 111111 , 11t 1111f 1111111111+1111111111111l11111111 1l1111l1t1 1!111,:

,

2

~

~

i

~

S

~

i

i

:

Figure 10.5: Part of the Metric Unit recording sheets

67

I

l,

'!

L

l, I

l

DANIEL ARROYO-BISHOP

Zone2

AreaI Subarea 147 63

64

65

66

64

65

66

Zone2 67

Theexcavation and recording of Zone2 bytheStratigraphic Unitsystem

Thechangeoverbetween th~two systems

Theexcavat ionand recording of l'Area/ Subarea147 bytheMetricUnit system

61

62

63

67

68

Figure 10.6: The two recording systems , by stratigraphic Unit and by metric Unit, their application and their complementarity in the field

Zone2

Zone2

AreaI Subarea 147 63

64

65

66

67

Excavation andrecording by theMetricUnit system withina Stratigraphic Unit system

Figure 10.7: The excavation of a pit by Metric Unit within a system recorded by Stratigraphic Units

68

69

70

THE ARCHEODATASYSTEM -

10.

TOWARDSA EUROPEAN ARCHAEOLOGICALDOCUMENT

4630.900

...

· .·:.,

-::1.:" : .. · ..

310

0, plots of the joint posterior density of 01 and 03, their marginal densities and the marginal density of 02 01 + 83 are given in Fig. 16.3. The marginal posterior density of 0 1 is symmetric with a mode at I 810; its expected value and standard deviation are 1807 and 62 respectively. The marginal posterior density of 02 has a mode at about 1860; its expected value and standard deviation are 1863 and 58 respectively. These results show good agreement with those of Vincent obtained by a slightly different method. If we follow the convention of reporting two-sigma intervals these are 1683 to 1931, and 1747 to 1979 for 0 1 and 02 respectively. But these intervals overlap which contradicts our prior knowledge that 01 is smaller than 02. Thus it is obvious that reporting in this way will be grossly misleading and further analysis is required. The contour plot of the joint posterior density of 8 1 and 03 is more informative, the mode is in the region of 0 1 1830 and 03 very close to zero. That is the most likely dates for events 1 and 2 are about 1830BP and a little before this, respectively. If we want to report an interval for 8 1 and 03, and we encourage this, we need a means of summarising the shape of the contours. This is a somewhat difficult task and we believe that a summary in the form of a contour plot together with some brief explanation is probably adequate. Turning to the marginal plot of 03 , this confirms that the

02

l(x; 0)p(0)

p(Blx)

= 1,2,

= fl(x; 0)p(0)d0

=

=

Little or no prior knowledege about the calendar date may be expressed by a 'vague prior', i.e., by letting p( 0) = 1 over the whole range of 0. Effectively this is saying that the event under consideration is, in the absence of any other information, equally likely to have occurred in any year. (This is, in fact, what the CALIB program assumes although it is not entirely clear that users of the program appreciate that this assumption has been made). Of course in some situations more prior information will be available. For example we may know that a date should be, based on other evidence, between two calendar dates 0a and 0b(0a < 0b) but that there is no evidence suggesting that one date within this range is more likely than any other. This may be captured by setting our prior density as

=

Another common problem is that we know a priori that one event must be later than another and we have radiocarbon results for both. Let 01 and 02(01 < 02) be the unknown calendar dates of the two events. Let x 1 and

104

16.

SOME STATISTICALPROBLEMS ARISING IN RADIOCARBONCALIBRATION

4000-------------,

3500 Radiocarbon years BP

3500

X

Radiocarbon years BP

3000

3000

2500

2500 2000

1500

1000

2000

500

1500

1000

500

Calibrated date BP (b)

Calibrated date BP (a) 4000-----------,

3500

Radiocarbon years BP

3000

x

2500 2000

1500

1000

500

Calibrated date BP (c) Figure 16.2: Plots of the likelihood using a simplified calibration curve (a) (b) (c)

X

X X

= 3500, ,G. E. M. 1989 ... Recognising and Using Geometric Features", in Woodwark,J. R., (ed.), Geometric Reasoning, pp.

Scientific Centre Report 185, Winchester.

169-188. Oxford University Press.

REILLY, P. 1989 ... Data Visualization in Archaeology", IBM Systems Journal, 28 (4): 569-579.

MARTIN,J. 1988. "Computing for Art's Sake", Datamation, December: 861')-86/13.

REILLY,P. & S. J. SHENNAN1989. "Applying Solid Modelling

MIYATA,K.1990. "A Method of Generating Stone Wall Patterns", ComplllerGraphics,24 (4): 387-394.

and Animated Three-Dimensional Graphics to Archaeological Problems", in Rahtz & Richards 1989, pp. 157-166.

M0SCATI,P. 1989 ... Archeologia e Informatica: l'esperienza di Neapolis", Rivista IBM. XXV, 1: 24-27.

SCOLI..AR,I. 1969. ..A Program for the Simulation of Magnetic Anomalies of Archaeological Origin in a Computer", ProspezioniArcheologiscM,4: 59-83.

NIKKEI C0MPUIElt GRAPIDCS 1989a. ..Art and Presentation", NiJc/cei Comp~r Graphics, 12: 74-79.

SMilH, I. 1985 ... Romans Make a High-Tech Comeback: Sid and Dora's bath show pulls in the crowd", Computing,June: 7-8.

NIKKEIC0MPUTF.RGRAPIDCS1989b ... Fujita Industrial Company and IBM Tokyo ResearchLaboratory Work Together to Reproduce Edo Casde u it was 400 Years Ago", NiJc/cei Complller

S0PRINTENDEN7.A ARCHF.OLOGIAPERI.E PR0VINCIEDI SASSARI E NU0R0 1989 ... SJ.Pl.A .- Progetto S.I.T.A.G. (Sistema lnformativo Territoriale Archeologico Gallura)", in Archae-

Graphics, 10: 102-103. O'F'LAHERTY,B., S. P. Q. RAHTz, J. RICHARDS,& S. SHENNAN 1990. 'The Development of Computer-Based Resources for Teaching Archaeology", in Cacaly, S. & Losfeld. G., (eds.),

ologiadel Terrilorio.Terriloriodell'ArcMologia: immaginidi wn'esperienzadi catalogazioneinformalicadei beni cllltwrali della Gallwra.Tempio Pausania. Chiarella-Sassari.

Sciences Historiqws, Sciences du Passe et Nouvelles Technologiesd'lnformalion: bilan et evalwaJion,Actes du Congres Internationalde Lah (lfr..17-18 Mars 1989). CREDO, Uni-

STANaf, Z. 1989. "Computervision: producing intra-site plans",

versite de Gaulle, Lille Ill, Lille.

TYREu., J.,

ArchaeologicalComplllingNewsletter, 20: 1-10.

F. YAZDY, M. RII.EY,& N. WINmRBOITOM 1990. ..CARVUPP: Computer-Assisted Radiological V1Sualisation Using Parallel Processor". Paper presented at the Transputer Applications '90, 11-13 July, 1990 at the University of Southampton.

RAHrz.S. P. Q. & J. D. RICHARDS, (eds.) 1989. Complller Applicationsand QwantitativeMethodsin Archaeology1989, International Series S48, Oxford. British Archaeological Reports. RAINS, M. 1989. ''Computerising the Plan Database: the Aegis system",ArchaeologicalComplllingNewsletter, 21: 1- 2.

WEISS, A. 1989. '"The Compass System", in Rahtz & Richards 1989,pp . 295-318.

138

21.

10WARDS A VIRTIJALARCHAEOLOGY

Figure 21.4: Grafland cut features progetto", Rivista IBM. XXV, 1: 30-39.

ZINGAREUJ,D. 1989. "Baldassare Conticallo: cronaca di un

139

PAULRBIILY

140

22 Furness Abbey survey project - The application of computer graphics and data visualisation to reconstruction modelling of an historic monument Ken Delooze (North Cheshire College, , Computer-Aided Engineering Centre, Warrington North Campus, Winwick Road, Warrington WA28QA, UK)

Jason Wood (Lancaster University Archaeological Unit, Physics Building, Lancaster University, Lancaster LAJ 4YW, UK)

22.1

Preamble

centre displays, audio-visual presentations and educational work-packs can be produced or existing ones enhanced, while marketable spin-offs might include popular interpretative mataial such as new guidebooks, broad sheets, model kits etc. In particular, the 20 historic reconstruction drawings can be used to generate full measured 3D perspectives and colour work showing development of the abbey far beyond the traditional artists impression . The initial project is now coming to an end and it is the purpose of this paper to propose what is considered to be the next logical step. Building on the basis of a very considerable amount of recording work that has been achieved during the past five years, it was felt appropriate to extend the use of the survey data and to generate a 3D electronic model of the abbey, using a 30 computer modelling system. To this end, a collaborative programme was established with North Cheshire College, ComputerAided Engineering Centre, to investigate the possibilities for the application of 3D computer modelling from which video sequences for presentation, and interactive graphics stations for research and educational use, could be produced.

The paper outlines the potential for the application of highly developed softwarewhich has been generated for the chemical, oil andgas industries, and how the archaeological world can benefit. This industrial development has been necessi tated by the need to create long term storage of engineering data and for maintaining lifetime records relating to the design construction and commissioning. The need to generate accurate libraries of engineering data so that information is only generated once, means that everyone uses exactly the same data across all disciplines and projects. The following describes how this has been achieved for engineering large and complex chemical plants, and then develops the theme in the possible application to the archaeologist.

22.2 Background to the Furness Abbey survey project Since 1985, the Lancaster University Archaeological Unit has been co-ordinating a large scale in-depth archaeological and architectural survey and analysis of Furness Abbey, Cumbri~ one of the most substantial surviving ruined Cistercian houses in Britain. The project is one of a number of historic fabric surveys set up in recent years by English Heritage to provide full basic recording of monuments prior to repair and conservation. Furness is the largest and most complex site for which complete survey has been attempted and the work required has necessarily been more detailed. The choice of photogrammetry as the basis of the recording scheme has made possible the study of the monument to far higher standards than would otherwise have been achievable. The history of the site is far more complex than has previously been realised and a rigorous approach to analysis has been adopted to identify and date constituent building periods. The format of the archive presentation has been designed for multi-purpose use. The basic stone-by-stone record drawings are distinct from colour-coded drawings generated from them which illustrate both the suggested historical development and reconstruction of the monument. Quite apart from the use of the archive for production of academic publications and as part of the planned programme of conservation, the accuracy of the survey data and fuller knowledge of the building history will lead to more informed presentation and marketing of the site to the visiting public. For example, on-site graphics schemes, visitor

22.3 Evaluation and Implementation of computer-aided engineering systems The past ten years have seen widespread application of computing methods to the engineering of large-scale chemical plants. The following provides a brief overview of how these were evaluated, implemented and developed to considerable advantage. One area which has benefited most from the implementation of computer systems is that of plant layout, pipe routing and pipework design. Ten years ago, there were a number of software packages that addressed the production of 2D draughting and 3D modelling oflarge and complex chemical plants and pipework fabrication and installation. Having short-listed a few, very detailed functional specifications outlining ideal systems were put together. Each system vendor responded, stating precisely how well their software would meet the specification. By this means the majority were rejected and the shortcomings of the most promising identified. One of the software packages selected was PDMS (Plant Design Management System). The advantage of PDMS was that it addressed the plant design aspects and would ultimately become the design database driving the drawing

141

KENDELOOZE AND JASON WOOD

and material conttol software, from which the fabrication drawings for complex piping systems were produced. The methods of actually determining the optimum plant layout are basically unchanged from all previous methods: Ten years aao Use of araph paper to put down fint thoughts Preliminary block models in wood andplutic Detailed desianmodels

Now Preliminary 2D electronic layouts Preliminary 3D electronic models The actual electronic design - plant built full-size in computer database

The key advantageof using computing techniques is integrity of design : • All designen use the standard data held in the database

• All designers use the same design procedures(many contractors and different companies use the same data)

• All designers take advantage of special programmes (macros) for producing designs (pumps, valves, standard equipment, stairs, cat ladders etc.). Such items only require a programme to be created once and stored in the macrolibrary for all to use. For a very large and complex electronic model, such as the THORP reprocessingplant at Sellafield (Cumbria, UK) being built for British Nuclear Fuels pie, there can be a great number of design databases which all must come together into one large single database for the complete project. For THORP, a multiplicity of design contractors worked in harmony to construct some 150 design databases, which, once assembled, created one huge electronic model. The creation of only one 'intelligent' model, from which all design and construction data can be interrogared and extracted by a large number of people, le.adsto greater efficiency and consistency, saving time and effort and reducing the possibility of errors. The computer model scores over the physical model in that the latter is a single entity which can only be in one place at any time and is non-intelligent, except for crude tagging. Physical models are also difficult to manage in tenns of accommodatingdesign changes. Finally, any data extracted from a physical model is only as accurate as the draughtsman can measure it to produce hand draughted drawings. Using physical models and hand drawings one can suffer 10-12% mors on pipework fabrication, resulting in high costs being incurred at the construction site. Using computer methods it is possible to achieve 0.1% errors, regardless of the overall plant size, thus eliminating many expensive on-site modifications and delays.

22.4 Plant Design ManagementSystem, dynamic InteractivemodelREVIEW and flash video Both PDMS and REVIEW were developed by the CADCentre Ltd in Cambridge. Put simply, the package comes in three essential modules : • PEGS (Project Engineaingand Graphics System), a 2D intelligent graphic, darabao.e • PDMS, a detailed design dalabue and 3D solid modeller • REVIEW,for realistic interactive model review. PDMS and REVIEW have proven track records and many of the world's largest engineering and construction companies in the powa generatingand process industries are major users of theIOftware package.An active usergroup helps bring about developmentsto suit particular needs. For example, British Nuclear Fuels pie have developed their own more sophisticated version of REVIEW known as EVS (Enhanced Visualisation Software). The CADCentre themselves are also about to launch a new module called DESIGN which can create colour-shaded pictures actually within the PDMS design database, without the need to first transfer data into REVIEW. PDMS is available on DEC, PRIME and IBM hardware. REVIEW runs on Silicon Graphics 4D series workstations and compatibles. Drawings within PDMS can be produced at any level of detail and at any desired scale. Fully annotated and dimensioned drawings are available and multiple views from any angle with hidden-line removal can be made, either true perspective, wide-angle or conventional orthographic. In addition, sections, isometric and cut-away views and views from within the model looking out can be produced. As all drawings are taken from the single model, any design changes introduced will appearon all subsequentdrawings. Data may also be taken from the PDMS model in the form of reports and materials take off, or for interfacing to other software packages such as stress analysis systems. The REVIEW and EVS visualisation modules take the model data from PDMS and produce quality colour-shaded pictures with multiview walk-through capability 'in seconds', whereas PDMS running on a PRIMEmini-computer can sometimes take hours to generate a single complex image. Quality mode pictures may be viewed as static images at the workstation and as hard copy thermal transfer print-outs, or as part of a sequential presentation or cartoonstyle animation. Real-time animation is available by going into wire-frame representation or fast mode (rapid travel) straight line component simplification. A display view control panel allows the user to define any view point within the model using view angle, position and orientation commands. 1be step distance and angle of rotation can also be adjusted as required. Any particular feature within the model can be called up by the observer and selectively viewed or 'clipped away' using a user-definableviewbox, while the interactive zoom facility allows for detailed enlargements. An ability to selectively switch colours and toggle individual colours on and off can be used to make elements easier to recognise and show 142

22.

~

specific areas which would otherwise be obscured. Tugging of individual features for verificationand identification purposes is also possible; the tag labels can remain in place even while the model is rotated. Model components and tag labels can be highlighted and made to flash against a duller backgroundusing the display element transparency feature. Both ambient and directional fixed light sources are available, as well as user-definable light direction control. Finally, an invaluable special feature of REVIEW and EVS is the interactive Scale Man. This figure can be manipulated within the model to help provide a sense of scale and perspective and can be moved independently of the model view. Of particular value is the ability to set eye position to the Scale Man. The potential for the generation of real-time, colourshade.danimation sequences on video film, direct from the PDMS model, is currently being explored. Colleagues from the Warrington-based firm HELP (Hazard Evaluation and Loss Prevention) have recently reformatted PDMS models into a format acceptable to Amazing Array Production's advanced WAVEFRONTgraphics package. Software such as WAVEFRONT gives more control over every object, including the positions of camera and lights, producing a photo-realistic image.

22.5

Establishment of the Furness Abbey 3D computer model project

It is clear that industry has committed considerable time and money and now has considerable expertise in using and developing CAE and CAD systems. It makes sense, therefore, that where possible, other disciplines should begin to benefit from these achievements. The purpose of the remainder of this paper is to show how these systems can be applied to reconstruction modelling of historic monuments, to the benefit of both the archaeologist and general public visiting such sites. Since 1988 North Cheshire College Computer-Aided Engineering Centte and Lancaster University Archaeological Unit have been working together to undertake research into the application of industrial 3D modelling techniques, using Furness Abbey as the basis of the project. PDMS was chosen for 3D modelling purposes and EVS for creating the colour-shaded pictures and walk-through sequences. Finally, an experimental flash video was produced using data taken directly from the PDMS design database. Initial work simply involved an extension to the existing teaching programme at North Cheshire College and the modelling, using PDMS primatives, of fake ashlar stone wall faces with arched openings in both 2D and 3D isometric images (Fig. 22.1), together with examples of EVS walkthrough views. The second phase of exploratory work was aimed at reconstruction modelling in PDMS on a Prime 9955 of an actual pan of Furness Abbey, based on outline dimensions talcendirectly from the 2D historic reconstruction drawings prepared by Lancaster University ArchaeologicalUnit. The area of the monument selected was that of the nave, aisles and west tower of the abbey church. It was decided to model the nave and aisles stone-by-stone. To make the

ABBEYSURVEYPROJECT

piers, for example, one stone was created and then copie.d and positioned until there were sufficient stones to form the first course around the pier. The course was then copie.dand the second positione.dabove it Once the whole column was complete, it too was copie.dand the piers positioned to form one side of the nave arcade. The piers of the opposite nave arcade were simply generated by mirror copying those of the first arcade. 1bis technique was applied to all the various architectural elements, including the arcade arches, windows, aisle reponds, vaults, roofs and parapets. Data from the PDMS model was then ttansfered to EVS to produce colour-shaded pictures. A selection of views containing different levels of information and varied colouring and light source arrangements were created and high quality hard copy thermal print-outs obtained (Figs. 22.2-22.5). A number of these static images were later captured on video film, usingthedatatakendirectly from the PDMS model.

22.6

Furness Abbey 3D computer model - proposed future work and benefits

It is intended that the next phase of work be the creation of a series of PDMS design databases of reconstructions of different parts of the abbey at different stages in its development. Once complete, thr. databases would be assembled to form one huge 3D intelligent model of the monastic complex in its entirety. As with the Sellafieldexample, the creation of only one model, from which all data can be interrogated and extracted, leads to greater efficiency and consistency. These are important considerationsas it is intended that the Furness project be set up in a very similar manner to that adopted for the design of the THORP plant, because of the size and variety of the desired workforce. The workforce will hopefully be drawn from a combination of the following: • Secondary school students • College students from related departments (mechanical and/or construction) • University students (undergraduate and/or postgraduate) • Trained students wishing to maintain or extend their skills (including staff of the Lancaster University Archaeological Unit) • Industrial input. The work to ere.atethe PDMS database will be achieved by proper project planning and the generation of well documented design procedures, using the management and computer facilities offered by North Cheshire College. Direction of the overall project would rest with Lancaster University Archaeological Unit. Benefits of using a multi-various workforce can be identified and listed under the following headings:

Schools • Early introduction to college environment • Early opportunity to use large CAD systems, not available to schools 143

KEN DELOOZE AND JASON WOOD

Figure 22.1: Isometric view of fake ashlar wall with arched openings, modelled using PDMSprimitives

• Early opportunity to learn high technology skills ('appetite whetter') • Early opportunity to learn how computers are used in many and various ways • Early opportunity to learn about the role of English Heritage and the importance of archaeological survey and analysis of historic buildings • Early exposure to industry through the college envi ronment.

This work will undoubtedly take several years to produce but it is felt that it is an excellent opportunity to seek to achieve what is considered to be a unique project, integrating virtually the full spectrum of Education and resulting in a product that can bring enjoyment to learning and increased understanding and publicity for all concerned. For the purposes of this project, it is intended that the input of nodal data into PDMS from the basic 2D outline reconstruction drawings will be manual, as keyboard education is a primary aim. There is a recognised need for research to develop interface technology for creating the PDMS database direct from the reconstruction drawings in AutoCAD, possibly via the photogrammetric survey DXF files. Once the PDMS model is complete, data can be transfered into either REVIEW or EVS in order to create the colour shaded pictures. These visualisation modules would offer the ability to adopt any viewing position from within and without the abbey, including both bird's eye and 'monk's' eye views (Figs. 22.2 and 22.3). The facility to be able to 'get inside and walk around' the reconstructed buildings would give a greater feeling of enclosed space and volume and enhance the sense of 'being there'. The user-definable viewbox and telescopic worn commands would make it possible to select certain detail from the model for particular inspection (e.g. Fig. 22.5) and to create 'cut away' crosssectional views through complex areasof the structure (e.g. Fig. 22.4). The ability to 'clip away' parts of the model would be extremely useful for showing obscured detail such as the arrangement of trusses below the roof coverings or above the vaults (e.g. Fig. 22.2), or for indicating the nature and position of temporary works such as scaffolding and centring. Colour switching and toggling would allow for

North Cheshire College • School leavers familiarity with the college • More effective use of the CAD system • Development of the use of software into other areas (i.e . not pipework design, as in the case of PDMS) • Introduction of the uses of major CAD systems into the curricula of building and/or construction students • Adult education (including industrial skills extensions).

Lancaster UniversityArchaeologicalUnit • Training of staff in the use of major CAD systems • Access for archaeological purposes to CAD technology and expertise unavailable within the university • Use of 3D computer graphics as a research tool for reconstruction ideas, concept design, testing of archaeological interpretation and simulation of struc tural sttess etc. • Use of 3D modelandvideo for presentation,education, communication and interaction • Publicity and marketing tool ('pretty pictures').

144

22.

FURNESS ABBEY SURVEYPROJECT r of l'ri" it IL•P rt'otl ls ?2UUU ...... ti,•, o, ,,.. , " , t wu , ,. t1 1, rz ou

"""b

rotnl """b

I

nu,.t,,,.. of r.,.1,.;11,,u,. ronrt i :nL'l'I r of 1•0 l11rJu11 tu t,p r 1111red I I'! I 169

Figure 22.2: Bird's eye view of nave and aisles produced by EVS, with roof coverings 'clipped away' to show arrangement of trusses. Note the display view control panel showing view angle, position and orientation commands

general public in the monument and its structural history. The chief fascination of the vast majority of people who visit monuments like Furness Abbey would appear to lie in a more visual interpretation of the function and original appearance of standing buildings and excavated remains. A realistic picture is immediately recognisable to anyone, even if they know nothing about architecture or archaeology . One possible theme might be the depiction of how the site developed from earliest times, explaining how the various buildings were erected and in what chronological sequence. Another potential subject could be a walk-through video following a typical monk's day, showing what individual parts of the monastic complex looked like and were used for at a particular period. Views taken from the computer model could also be interposed with still shots of the surviving ruins. Creation of a video will require considerable planning and forethought. Walk-through routes have to be pre-defined as the video sequence is taken directly from the PDMS database. The level of detail of the model and its individual components are other important considerations. Like most solid modellers, PDMS has difficulty dealing with complex, irregular shapes. Moreover, experimental modelling of the nave of the abbey church, stone-by-stone, although visually impressive, proved to be very expensive in terms of computer storage and led to problems and delays when the data was being made to generate a walk-through sequence on

differentiation between identifiable periods and phases of construction. Specific areas, such as floor surfaces, could also be made to fade out and become transparent, in order to view buried foundations or earlier remains. Model tagging and element highlighting could be used to identify certain features from within complex views, while the lighting control could be used to create the impression of depth and atmosphere within the reconstructed building (e.g. Fig. 22.3). Finally the interactive Scale 'Monk' would be an invaluable aid to help judge the sizes and scale of features in relation to reality. REVIEW and EVS have no facility for generating shadow or reflection effects and no surface modelling or texturing is available. The software is, therefore, limited in the context of creating a topographical landscape into which the abbey could be set. Such facilities are offered by other packages so that some research and development would be necessary to interface with these.

22.7

Furness Abbey Flash Video posed future work

pro-

It is intended that a video be created from the 3D computer model for presentation at the abbey site permanent exhibition. Such a production would undoubtedly help to stimulate a greater interest and appreciation amongst the

145

KEN DELOOZEAND JASON WOOD

Figure 22.3: 'Monks 's' eye view of nave produced by EVE, showing use of varied lighting control

Keen, D. Pride and B. Sykes at North Cheshire College; J. Williams and A. Olivier at Lancaster University Archaeological Unit, and D. Sherlock at English Heritage. R. Busby of North Cheshire College and J. Durwood built the Furness Abbey PDMS model using reconstruction drawings prepared by D. Cooper and R. Cooper of Lancaster University Archaeological Unit. The EVS pictures and video were created by K. Ryder and C. Williams of the Computer Applications Group at British Nuclear Fuels pie, under the direction of M. Hall-Wilton. Assistance was also provided by R. Longdon of the CADCentre Ltd, D. Grenfell of HELP and G. Chapman of Sheffield City Polytechnic. Further information regarding PDMS and REVIEW is available from the CAOCentre Ltd, High Cross, Madingley Road, Cambridge CB3 OHB.

video. Real time animation proved to be possible only by simplifying the model by reducing the number of primatives. Looking further ahead, production of an interactive videodisc, to allow the public to interrogate the model and position themselves within it, would require further outlay of hardware and software, together with reprogramming, although not necessarily re-inputting.

Acknowledgements Development of the project to date has been indirectly funded by the Department of Trade and Industry, British Nuclear Fuels pie and English Heritage. Since the project was first conceived, support has been welcomed from T.

146

_22.

FURNESS ABBEY SURVEYPROJECT

Figure 22.4: Cut-away view through nave and aisles produced by EVS, showing complexity of vault and roof construction, and addition of west tower

147

KEN DELOOZE AND JASON WOOD

Figure 22.5: Enlarged view of aisle vaulting produced by EVS using viewbox and telescopic zoom commands

148

23 Images , databases and edge detection for archaeological object drawin gs P.H. Lewis (Department of Electronics and Computer Science, University, Southampton S09 SNH)

K. J. Goodson (IBM UK Scientific Centre, Athelstan House, St. Clement Street, Winchester, S023 9DR)

23.1 Introduction

drawing images may be displayed in separate windows. The close analogy with the desktop means that archaeologists can transfer euily from traditional modes of working to the use of more powerful computer based tools. Some of the current features which the GOAD environmentprovides via menus and buttons in the window based interface are as follows:

The aims of this paper are to present a brief overview of the GOAD (Graphically Oriented ArchaeologicalDatabase Project) followed by a short discussion of two important issues for the development of sophisticated graphical databases, namely the automatic extractionof explicit shape infonnation from raster images of line drawings and techniques for representing shape for effective shape retrieval and classification.

• Tools for capturing images from TV cameras and scanning devices. • Tools for enhancing or editing images prior to storage in the database. • Database facilities for storing text, graphical and image data. • The ability to retrieve the above infonnation using SQL or specially developed fonns using the WIMP interface. Text, graphics and images may be displayed in separate windows and once displayed they may be manipulated independently of the database. • Tools for extracting vector shape representations of the line work in raster images.

23.2 The GOAD project The aim of the GOAD project was the development of an integrated text, graphical and image artefact database with particular emphasis on the storage, retrieval and dissemination strategies. It was initially funded by a two year grant from the Science Based Archaeology Committee of the SERC. Specific issues to be addressed in the project were:

Two exemplar databases have been constructed within the GOAD environment. The first consists of the textural and image data relating to a corpus of Roman amphori (Peacock & Williams 1986) while the second is a representative corpus of fonns of Roman pottery from the Oxford kiln sites collated with excavation and fabric data (materials supplied by the Oxford Archaeological Unit as part of a continuing project on the hospital site, after Young 1977). The dissemination of the database is probably the area of the project that is most affected by technologychanges. Our initial plan was to develop our own display routines and to port these to a microcomputer for distribution, but the rapid spread of Postscript as a standard for screen displays and not only printing devices means that we are now able to offer a single fonnat for storage, display and printing although there are currently no suitable display Postcript implementations for micros. We are confident that this device independent approach is correct for publication purposes.

• the development of techniques for the capture and storage of graphical and image data. • the design of database structures and retrieval methods to provide rapid retrieval and effective display of text, graphics and image data. • the implications for publication of such databases either electronically over networks or by more conventional paper-based methods. The strategy adopted in the development of the database system involved integrationof several important ingredients which are summarized as follows: 1. The Postscript page description language and inter-

2.

3. 4.

5.

preter. This was chosen to provide a versatile mechanism for communicating text, graphics and image infonnation in a device independent way. The NeWs network extensible window system which provides a WIMP interface based on modem bitmapped displays. The Ingres relational database to provide the central infonnation storage and retrieval capability. Tools for capturing and processing raster images from scanners and vidicon cameras. Extensive additional software written in-house in C and Postscript to complete and integrate the GOAD environment.

An example of a query session within the GOAD environment is shown in Fig. 23.2, illustrating the windowing facilities of the user interface where a variety of text and line

23.3 Automatic shape extraction One of the major problems when building a useful graphical database is that of entering the graphical infonnation itself. An abundance of line drawings of archaeological artefacts is available in paper fonn and scanning or frame-grabbing this into the computer as a bitmap or raster image is essentially straight forward. But if the drawing is to be used for anything other than for display purposes, for example if it is to be used for automatic shape retrieval or classification, 149

P. H. LEWISANDK. J. GooDSON

P...- - -- G-ra_p _h_i_c_a _lly _ O_r_ie_n_t _e_d_A _rc_h_a_e_o_log _ i_c_al_Da __ta_b _a_s_e~~====.==1:;s·~-~•fi_'•:··· _·:'s';..;.: i_·.~--:_ ·'·_;ii,;;i ··~m I-> I-> I->

---------------

--

Show single Show all (3.2) (3.1) I I --- -- select I I I 1--> Magnify/ I

1___

'

I

Display (3.3)

table

sherd

I-> average I-> frequency

Retain

Vector

form

-I

---

I I

(3. 4)

--1

dists. at

surface

1

V

(INTER-ACTIVE/

NON-INTERACTIVE mode) Non-Interactive (3.1.2) (3. 4. 2)

Interactive (3.1. 1 ) (3. 4 .1)

-- 1

< ---

Retu r n to mai n

V

(Select

i on of VIEWI NG PLANE)

P l an

El e va ti on

St and ar d cub i c

(3.1.1.1) (3.1.1.2) (3.1.1. ( 3 . 1.2. (3.1.2. 1 ) (3.1.2. 2) I - -> - - - > - - - - - Ma g nif y/ - < --- -I

---

--

---

Retain

-

b y s h ap e Dat a b y y ears Return I to main Squares only Sh i elds only Ci rcles only

Figure 24. l: Sequence of options

the first position and the largest numbe r representing the last position) . 6. Sherds that appear on the surface are outlined in white and all others are outlined in yellow .

Screen 12 shows a similar approach on the standard cubic output. Non-interactive output displays the whole set of data without a pause.

Show single (option 3.2) This allows the user to select one particular sherd record from the database and display the sherd 's movements using the standard or scaled cubic output. Viewing the movement of one specific sherd conveys more detailed information. Each year the sherd is characterized according to its location within the ground, the direction it points in, its tilting (deflection) in the ground and whether it is face-up, face-down or vertical. All this information is represented by the following rules:

Und er each diagram is placed a comm ent box conveying information about the display shown . For example, for a single sherd this information is about the sherd 's identifica tion number and shape . Screens 13 to 16 show examples of this form of output.

Display table (option 3.3) This choice provides the user with an overview of all the sherds held in the current database, associating each sherd with its recorded shape and the total distance it has moved during the years recorded in the E- W and N- S planes, as well as in terms of depth and the overall total distance moved by the sherd spatially. For each distance field within the table, two measurements are recorded and displayed. The first measurement represents the cumulative distance value, while the second measure ment, shown in brackets alongside the first measurement , represents the distance value from the original to the final position of the sherd. If, for any year within a sherd record, the data has been recorded as missing, then it is assumed that the sherd has not moved in any direction that year and its distance values are counted as zero in each direction. The purpose of displaying such a table lies in the provision of enabling the user to select sherds of particular statistical interest for later viewing.

1. The centre point for each shape gives the sherd's location . 2. The direction in which the sherd is pointing is shown by a small white circle on one of the vertices. This must be viewed relative to the compass. 3. Tilting of the sherd may only be judged by inspection (the vertical disposition, however , is clearly appar ent). 4. The sherd is drawn in its symbolic colour, with the following conventions: face-up sherds are shaded in solid colour, face-down sherds are line-filled and vertical sherds are drawn in outline. 5. The sequence of movements is represented by annotation of the sherd positions numerically (1 representing 157

MADHUMITA SEN-GUPTA, SUSAN LAFLIN

& PETERREYNOLDS

Vector form (option 3.4) This shows the movement of all sherds in a given year as though from a common starting point. This should demonstrate any overall movement even more clearly than the plan output. Screen 18 shows the result for the year 1982. Once again, the different colours indicate the different shapes of sherd and for this data set and mode of cultivation there is no overall movement. The program is being run interactively and so the user has the choice of viewing the results for the next year. Screen 17 shows the non-interactive use of this output for the range of years from 1981 to 1984 and here the different colours, superimposed upon one another, indicate the different years.

24.2.2 User Interface The interface has been designed to provide the user with as much flexibility as possible, and has been adopted because it makes the system highly robust. It is structured as a sequence of menus, each highlighting a set of options from which the user is required to make a selection. This selection may be done using only the keyboard, or using a mouse as well as the keyboard. The choice of operating mode is made by the user in the initial stage of executing the program. When the keyboard is used, a menu appears on the screen and the user selects an option following the rules outlined below: • For a menu appearing horizontally across the screen, right and left cursor keys are used to select the option, which is highlighted. • For a menu appearing vertically down the screen, up and down arrow keys are used to select the option, which is highlighted. In either case, the user must press the return key to indicate his/her choice and start operation of the selected option. When the mouse is used, a menu appears on the screen and the user selects an option by moving the mouse to the appropriate option box which becomes highlighted. Commencing execution of the selected option is simply done by pressing the right button of the mouse. Naturally, where the system requires typed input, only the keyboard may be used. Validation of user response in such a case is of prime importance, and hence the system has been thoroughly tested and developed to maximise this. A choice of sixteen colours including red, blue, black, white, yellow, cyan, magenta and light and dark shades of some of these seven colours available when developing the software, was utilised as comprehensively as possible (to enhance clarity and highlight important information in the graphical displays), and the .layout for the final schematic flow of options and menus was constructed in order to maximize flexibility to the user, hence making the system user-friendly. It should be noted that the use of colour and threedimensionality helped to make the screen outputs effective and useful in discriminating between patterns of movement.

Unfortunately the absence of colour has meant that these affects cannot be fully appreciated here.

24.3 Conclusions and Further research From the computer scientists• point of view, the requested software has been developed and tested to provide a satisfactory performance of the system required. However, further enhancements to the resulting software may be made. For example, the use of colour coding to symbolise depthvariation of a single sherd's location within a volume of ground. During development of the software, it was intended that varying shades of the same colour should be employed to symbolise this depth-variation. Unfortunately, this could not be implemented since the hardware constraint of using an EGA graphics card had imposed a restriction in the availability of colours. Another enhancement that would appear desirable is to provide the user with a facility to rotate a three dimensional view. Since this had not been specifically requested, the more necessary requirements were attended to instead. Nevertheless, this is all from the point of view of the Computer Scientists. Far more important is how useful has the software proved for the end users? And what are the possibilities of further developments using this software? These questions may only be answered by the actual end users. Peter Reynolds is both the commissioner and principal end user of this software, so the remainder of this section comprises his views and conclusions upon the resulting software and its use. The reason for the development of the software was specifically to create a graphics presentation of a large and cumbersome sequential data base in order to allow subsequent selection, analysis and comparison. In effect it comprises a relatively sophisticated three dimensional display of fixed points allowing adjustments to scale and, therefore, enhancement by distortion. Its greatest use is the facility to display both combined and individual sequences of fixed points. Built into the program as a basic requirement was the need to cross refer and provide comparative analyses between different phases of data. The overall objective is the analysis of the data bases achieved from the study of artefact movement in the topsoil (Reynolds 1989). The purpose of this study is to evaluate the topsoil as a potentially important archaeological layer worthy of detailed excavation in that the material evidence contained within it can be associated with underlying archaeological features. Artefact assemblages in and on the surface of the soil layer are argued as indicators of a site and are frequently used in field walking to plot locations. Given this, the hypothesis seeks to examine if the assemblages are infinitely spread by plough action or whether the concentration is real in the sense that travel is limited by the nature of the cultivation. Unidirectional ploughing will, in fact, move soil infinitely but, in practice, ploughing is carefully executed in a multidirectional manner in order not to move field areas more than the width of a furrow. Therefore artefact movement within the soil, while subject to a degree of spreading, an objective of this trial is to assess the amount of such spreading, is likely to remain within a relatively limited zone. Vertical movement

158

24.

VISUALISATIONOF SHERD MOVEMENTIN THE PLOUGHZONE

of artefacts within the soil is also in need of analysis since a correlation between surface finds and the finds within the ploughzonehas been conjectured (Reynolds 1987,Reynolds 1989) at circa 16%. The development of this program, therefore, is a critical adjunct to this fundamental research enquiry . In its present state it responds extremely well to the data and allows considerable flexibility of analysis with statistical routines incorporated within its framework. Future development, with a significantly increased data base, should ideally include a future projection to simulate tens and hundreds of years or sequences of ploughing activity. In its current form there is the possibility of using different presentations of the sequences already obtained to compare with excavated data to determine presence or absence of correlations . From the point of view of both the commissioner and end user, the program is proving particularly useful and efficacious. It has reduced the laborious hours of plotting out the sherd assemblages found in specific conditions greatly. Initially the scaling proved unusual but one quickly becomes

used to the screen image. Because both the horizontal scales were disproportionately large in comparison with the vertical scale the ability to expand the last proved extremely valuable. The colour coding was specifically requested and responds as required. In conclusion this program not only achieves its initial objective but also offers further speculative options.

Bibliography REYNOLDS,P.J. 1987. Sherdmovement in the ploughsoil. Buts er Ancient Farm Project Trust. REYNOLDS,P. J. 1988 . " Sherd movement in the plough zone physical data base into computer simulation", in Rahtz, S. P. Q ., (ed.), Computer and Quantitative Methods in Archaeology 1988, International Series 446. British Archaeologic al Reports, Oxford. REYNOLDS,P. J. 1989 . Sherd movement in the ploughsoil. Buts er Ancient Farm Project Trust.

159

~

>

0

::c:

i

~

Cl)

~

6

' Cl)

C

Cll

~

t""4

>

~

~

i

"Tl oo·

~ trJ

C

'"1

0

N

.... 0 °'

~

~

Screen 1: user has selected to view, non-interactively,all sherd movements in a scaled volume of ground

Screen 2: user has selected to view, non-interactively, all sherd movements in a standard structure of ground

~ 0

5 Cll

Cl) (')

~ 0

::s ,;,,

r

Screen 3 : user has selected to view, interactively, a range of sherd movements in elevation view

Screen 4: user has selected to view, non-interactively, all shield-shaped sherd movements in plan view

~

'Tl

OQ. C:

~

--

°'

Iv

~

~

Screen 5: user has selected to view, non-interactively, all square-shaped sherd movements in elevation view

Screen 6: user has selected to view, interactively, specific circular-shaped sherd movements in standard structure

Cl)

... 0

~ ~

~

,.:: . .-.:•:•::, . .

UlltWiMtiJmft-

< en

4144¥ 1981

::,

1982

C:

>

{fl

C en

V\

6o PLAN UIEI.I

. .. . • . • .. .. . .• ••. •.• •.• ••• •.. • . • .. . •

;

+

~ 0

'Tl

en

= ~

E:: 0


l

N

~

~ Cll

Screen 13: user has selected to view a single sherd and trace out its movements within a scaled vol. of ground

Screen 14: user has selected to view a single sherd and trace out its movements within a scaled vol. of ground

0




-l I

00

Screen 17: user has selected to view a vector representation of sherd-movements noninteractively

Screen 18: user has selected to view a vector representation of sherd movements interactively

~

0 t""' c:, en

25 Integrating spatial information in computerised Sites and Monuments Records: meeting archaeological requirements in the 1990s Gary R. Lock (Institute of Archaeology, University of Oxford, 36, Beaumont Street, Oxford, OXJ 2PG, UK)

Trevor M. Harris (Department of Geology and Geography, West Virginia University, 425, White Hall, Morgantown, WV 26506, USA)

25.1 Introduction Most county Sites and Monuments Records (SMRs) are now computerised to a greater or lesser extent. The value of these SMRs as repositories of the national archaeological record is considerable and they provide a rich resource for any analysis of the cultural landscape. Many SMRs are now reaching the end of the initial data collection and early computerisation phase. In this paper we review the current nature and status of SMRs and consider the implications for users, actual and potential, which have arisen from the present structure of SMR database design . In addition we discuss the emergence of new technology, Geographic Information Systems (GIS), which is now sufficiently developed to be particularly suited to possible further phases of SMR development. This technology provides a fully integrated spatial database management and analysis environment which would enable SMRs to go beyond the present limiting inventorying stage and provide a highly sophisticated information system. Such a system would be capable of greatly improved functionality, have advanced analysis capability, and be able to develop decision support roles in the management of the cultural environment. These technological advances necessitate the inclusion of the full archaeological record, including its spatial component, within the database environment. The underlying basis of SMR computerisation was formulated under the financial and technological exigencies of the late 1970s and early 1980s. The present structure of computerised SMRs must be seen in the context of these origins. These factors have resulted in the construction of databases which closely mirror the structure of earlier manual card index systems. Thus the database can be searched for archaeological sites and artefacts according to a combination of single or multiple attributes based on boolean logic and standard database interrogative procedures. One major difficulty which has arisen from the general adoption of the standardised database structure mainly promulgated by English Heritage, however, has been the problem of how to handle the locational and geographical information associated with the archaeological record. While all would accept that without knowing where an archaeological site or artefact exists on the ground the value of the information is greatly diminished, in most SMRs the spatial description of the phenomena has been reduced to either a single Ordnance Survey point coordinate or to a nominal representation such as the parish administrative area. In many instances of course the inability to locate a find or site accurately on the ground invalidates the allocation of precise coordinates. No amount of sophistication in information retrieval system

design will easily overcome the problems associated with the lack of information precision, though research within the field of GIS concerned with 'fuzzy space' may provide one very profitable avenue of enquiry. In nearly all SMRs the inability to incorporate the full spatial description of an archaeological record within the computerised database has resulted in the information having to be stored by hand on small scale Ordnance Survey topographic map sheets. In this paper our concern is focused primarily upon the handling of the full archaeological record within the computerised SMR. To a considerable extent the way in which the spatial description of an archaeological record is currently handled greatly constrains the effectiveness and utility of the SMR to service the needs of users. To this end three aspects of the handling of archaeological information within current computerised SMRs are considered in the light of the current inability of SMRs to maximise their full potential as they are presently structured; in the light of professional and societal trends toward more sophisticated uses of computerised information systems; in the light of current developments in Geographic Information Systems; and in the light of recent government enquiries and reports into the handling of spatial information and the role of GIS technology in many areas of UK society. The first concern arising from the structure of current SMR databases is that a vital component of the archaeological record, its spatial description, is reduced to a geometric form that in many instances clearly does not reflect reality. Thus linear features, such as embankments or ditches, and zonal areas, such as field systems, are simplistically recorded in the computerised database by the allocation of a single coordinate or a parish unit. It is usual for the coordinate to reflect the centroid of a structure. Some SMRs have attempted to describe the shape of such features with a series of grid references, using three for a linear feature for example. While this obviously contains added information it remains incapable of serious analysis. The implications of this, and other, database restrictions are discussed later. The second concern is that current SMR systems are unable to record the topological relationships which exist between archaeological, natural and human environmental phenomena. This rather technical aspect is also discussed later in the paper but the outcome of this omission is critical because it precludes any form of spatially oriented enquiry or spatial analysis being undertaken on the archaeological information within the database, other than that at a very elementary and crude level. This is especially relevant to the SMR user community for even basic requests for information concerning what archaeological sites exist within a certain distance of a proposed development, or seeking the relationships which 165

GARY R. LOCK AND TREVORM. HARRIS

comprehensive and frequently updated source of archaeological information available in the UK (Holman 1985, Chadburn 1989) and provide a unique potential for regional spatial research in archaeology as well as cultural resource management. Every county in England, Wales, Northern Ireland and most of Scotland now has SMR cover, together with one or more professional archaeologist(s) specifically dedicated to the management of that record. All types of information are entered into the record although their main importance lies in the inventorying of archaeological sites. Despite the potential of this resource for academic research at all levels, SMRs are heavily focused on servicing cultural resource management activities within regional and local planning. Given the origins of SMRs and the location of many SMRs within County Council Planning Departments (Chadburn 1989) this bias in application is not surprising. However, we shall argue that this apparent schism between an invaluable archaeological resource and a potential major academic user group is partly due to the unwieldy structure of SMRs and reflects the failure of SMRs to fully integrate spatial information within the computerised archaeological database. Furthermore, we also argue that the spatial shortcomings of existing SMRs also severely limits their ability to perform even their primary tasks of cultural resource management. The first SMRs of the late 1960s and early 1970s were based on manual card index systems with many adopting the Optical Co-Incidence punch card system (Benson 1974 ). Although the potential of computer-based SMRs was recognised in the late 1960s it was not until 1974 that the first application emerged (Benson 1985). Many SMR workers were slow to appreciate the advantages of computerised records. As late as 1978 the Association of County Archaeological Officers in their Guide to the Establishment of Sites and Monuments Records had to incorporate the Optical Coincidence card system as an option to the mainstream computerisation of the Record 'in order to satisfy those members who were perfectly happy to continue with a manual system ... and who were unwilling or unable to experiment or invest in the more effective technology then available' (Benson 1985, p. 33). The last decade has seen the slow and somewhat painful adoption of computers by all SMRs with much discussion centred around the standardisation of records and terminology (Chadburn 1988). A recent survey (Chadburn 1989) has shown that the 46 SMRs in England display a wide range of hardware and software types with 33 being microcomputer based while the rest employ either minicomputers or mainframe systems. It is perhaps surprising that in the late 1980s 'a few SMRs still rely heavily on Optical Co-Incidence Cards for retrieval, and ... can only undertake limited searches of their records' (Chadburn 1989, p. 22). It is important to acknowledge the influential role played by English Heritage in the computerisation process of SMRs. This influence has taken the form of financial assistance, software development and associated advice. In the year 1987/88, for example, of the £3.5 million distributed through English Heritage 10% went to SMR work (English Heritage 1987). This financial and computing support is reflected in the number of SMRs that use SAMSON; software written by English Heritage and based on the database

might exist between various archaeological sites, or the extent of linear or multiple relationships existing between archaeological and environmental phenomena, cannot be pursued without some basis of topological relationship encoded within the database. The third major concern arises from this inability to store anything but a crude locational description of an archaeological record within the database. The result of this has been that the full geographical location, spatial extent and implied topological relationship of an archaeological record is stored separately, outside of the computer database, on Ordnance Survey paper maps. This 'stop-gap' procedure effectively divorces the important spatial component of the archaeological record from the computerised record; a procedure which runs counter to all professional developments in efficient database design and management. This not only precludes full blown spatial queries from being entertained but prevents the adoption of digital mapping procedures or even the use of digital topographic maps from the Ordnance Survey. We suggest that the current inability of computerised SMRs to incorporate the full archaeological record and topological relationships within an integrated information system environment prevents the full potential of the SMR resource from being exploited. Our concern in this paper is that having reached this stage of development, the computerised SMRs are in danger of becoming fossilised and outmoded partial repositories of the archaeological record; archaeological gazetteers rather than the sophisticated information systems of which they are capable. To this end our focus is to review the present functional capability of SMRs and identify important future directions for system development which would enable SMRs to more fully meet the requirements of archaeologists and society well into the next millennium.

25.2

The development and functional capabilities of SMRs

25.2.1

SMR origins and early computerisation

The origin of SMRs lies in the strong British tradition of field work and site surveying going back to the beginning of this century (see Burrow 1985 for a detailed history). The appointment of 0. G. S. Crawford as the first Ordnance Survey Archaeology Officer inaugurated a systematic approach to archaeological mapping and recording. Indeed, Ordnance Survey record cards have formed the backbone of most SMRs. The findings of the Walsh Committee (Walsh 1969), set up to look into the protection of field monuments, recommended that every County Planning Authority should hold a record of all known field monuments and have suitable archaeological expertise on their staff. In view of the threatened, and indeed actual, destruction of the archaeological resource due to rapid urban development, the government of the day reacted favourably to these recommendations. The result is that nearly the whole of the UK is now covered by SMRs. Importantly, emphasis is placed as much on detailed local archaeological information as on sites of national interest. The county-based SMRs are thus the most

166

25. package Superfile. Of the 33 microcomputer based SMRs, 25 use SAMSON (Chadburn 1989, Table 3.3) despite it being based on a flat file database system lacking much of the sophistication and power of modem database packages. In return for this support English Heritage has been able to call upon the computerised SMR databases in the performance of its statutory duties. One such obligation is to keep a schedule of monuments of national importance which are protected under the 1979 Act. This schedule is currently being enhanced by the recently initiated Monuments Protection Programme (MPP) which is also going to be based on the interrogation of SMRs.

25.2.2 Spatial limitations of SMRs While English Heritage have supplied considerable aid and advice to SMRs concerning computerisation of the register, this has been aimed almost exclusively at establishing a standardised database structure closely replicating the structure of the earlier card index system. Because of the difficulties of defining spatial objects and spatial relationships within a computing environment this database design precluded the inclusion of any but a crude locational component within the database. The result is that todays computerised SMRs perform the same sort of analysis as the earlier manual systems, albeit much more quickly and with greater flexibility. To reach this national situation within two decades is a significant achievement which should not go unsung for it has provided an important early base for 'informed decisions to be made about our cultural heritage, and [provide] a tool for a range of activities such as education, research and planning' (Chadburn 1989, p. 13). We suggest, however, that the development of SMRs now stands at a critical crossroad in its development. In essence we see the paths bifurcating between the route of continued development of the system along the existing database road, flat file or relational, or of opting toward the potential offered by the new technology of GIS. The choice confronting the SMR community centres largely on the perception of that group of the role and functionality of SMRs in the 1990s and, importantly, beyond. To a large extent the choice of direction is intricately tied to the apparently superficial consideration of whether, and how, to integrate the full archaeological record, including its spatial and topological component, within a computerised database environment. This aspect lies at the very core of our questioning of current SMR database design because of the implications that arise concerning system functionality, the role of SMRs, and the nature of user requirements. At the moment the computerised SMR is built upon a standard database structure of records and fields which translate to a series of items (in this case usually archaeological sites) each with a list of descriptive attributes. This structure allows the usual kinds of SMR analysis such as the generation of catalogues by site type, parish, period or any other suitable field. A considerable amount of time and effort has been expended in standardising the fields within SMRs. Chadburn (1989, p. 14) has identified a 'record content standard', a series of data categories which most SMRs include in every record. Booth (1988) has suggested a standard for data transfer of site specific data intended for use between SMRs and other heritage databases, specifically the HBMC

INTEGRATINGSPATIALINFORMATIONIN COMPUTERISEDSMRS

SAM Record and the RCHME NAR. Within all of these the spatial component of the record is treated as one or more fields; the two most important being the nominally measured 'parish' (it is either in, or not in, a parish), and specific point data in the form of a National Grid Reference. The present situation has evolved over the past two decades from ad hoc developments with little standardisation in hardware, software or content. However, one theme common to all SMRs has been their inability to satisfactorily integrate the full locational component of an archaeological record within the computerised archaeological database. Operating parallel to the computerised database, the SMRs have a series of 1: I 0560 scale (or metric equivalent) Ordnance Survey paper maps onto which each computer database entry is plotted by hand and referenced by a unique Primary Reference Number (PRN). These base maps can vary considerably in their date of publication and thus the completeness of the topographic information which they possess. To update a series of these maps by hand onto current OS maps would be a very time consuming process. Occasionally these maps will be associated with various overlays, older maps of interest, or coverages of more sensitive areas at different scales. The spatial information associated with an archaeological record is thus split between two different storage media and, not uncommonly, different physical locations within the same building. This rather inflexible and quixotic situation greatly restricts the range of questions that can be asked of the data and the quality and nature of the response from an SMR. Archaeological enquiries with an emphasis upon location must be phrased in terms of either the fields within the database record structure (usually related to the parish) for computer-based output, or map sheet number for manual cartographic output. Catalogues by parish, and crude database interrogation by grid reference will satisfy some spatial queries currently put to SMRs but a fundamental component of the archaeological record is actually stored off-line. This not only greatly restricts the range of queries that can be directed at the database but runs counter to all developments in database management systems . The result is that the SMRs represent only partially computerised systems of the archaeological record . The SMR database field specified for the locational information pertaining to an archaeological record at the moment simply acts as a pointer to the main spatial information repositories which are the maps.

25.3

Handling spatial archaeological information within a computing environment: the need for GIS

Many of the problems encountered by archaeologists in the handling of spatial information within a computing environment as outlined above have been known to geographers for some time. By definition the focus of the geographer's interest is the spatial component. The handling of geographic information is fundamental to good management, planning, and decision-making within the natural and human environments. As the House of Commons Report into the Handling of Spatial Information comments, 'Most human activity depends on geographic information: on knowing where 167

GARY R. LOCK AND TREVORM. HARRIS

things are and understanding how they relate to each other' (DoE 1987, p. 7) In archaeology the geography of sites and artefacts and the relationships between them is vital in explaining and understanding past societies. Considerable research has been undertaken, primarily by geographers, to develop principles whereby geographical information could be stored, manipulated and integrated within a computing environment. It is contended here that these technological developments in spatial data handling and analysis have important implications for archaeology and especially SMRs. It is not the place in this paper to detail the nature of GIS. This has been undertaken with respect to archaeological applications in previous papers (Harris 1986, Harris 1988, Wansleeben 1988,Kvamme&Kohler 1988,Kvamme 1989, Allen et al 1990). A number of standard texts also exist in the GIS literature (Burrough 1986, Star & Estes 1990, Tomlin 1990). Suffice to say at this point that the traditional manual method of integrating spatial information has involved overlaying maps physically one on another and tracing the areas of intersection or union (McHarg 1969). The development of GIS has enabled the basic spatial primitives of point specific data, linear features and polygonal areas, as well as pixel based information, to be integrated and analyzed within a computing environment. What this involves is effectively the storage of phenomena as a series of layers of information, or coverages, within a database whereby each layer would represent a distribution traditionally represented in the form of a map (see Harris 1986). In this database the topological relationships which exist between the spatial entities are retained and are stored as part of the relational database. Thereafter requests for information concerning the location of archaeological sites or artefacts or their proximity one to another or to other features can be undertaken. Sophisticated spatial data handling or querying may be undertaken on the system. Thus requests for information concerning what sites existed at particular locations or specified areas may be entertained. Because any part of the database may be linked spatially with any other part map coverages may be overlaid and combined to produce new composite coverages which in tum could be stored separately within the database. Selected areas generated around designated features can be created by 'buffering' and used to extract portions of the database as required. By this method zones of a specified size or extent can be generated around a geographic feature or point, line or area and the buffered zone then overlaid on top of other map coverages to select and analyze phenomena falling within the designated area. Thus generating a corridor either side of a proposed road development and overlaying this buffered zone across a sequence of archaeological, or indeed ecological and socio-economic coverages, would enable the impact of such a development to be determined and evaluated against other route proposals. The outcome of this type of spatial query ability is that an archaeological GIS database would be capable of not only storing and retrieving the full geographical description of archaeological phenomena but of integrating, analyzing and subsequently digitally plotting archaeological phenomena and environmental variables held within the system. Importantly, the ability of the database to entertain full blown spatial queries would greatly improve the effectiveness of SMRs in their

task of aiding planning decisions. Thus questions to an SMR from a planner are concerned less with the specifics of the archaeological record than with the overall significance of an impact on the cultural resource arising frbm a proposed development. Similarly an archaeologist could investigate any number of relationships existing between archaeological phenomena by period or specific site type with environmental phenomena such as aspect or slope or soil type. This latter point opens up the discussion somewhat to consider the potential role of a comprehensive SMR information system possessing the full archaeological record and GIS capability for a range of applications ranging from cultural resource management to regional archaeological analysis. In the same way that the full spatial description of an archaeological site could be stored within a computerised GIS database, thereby enabling spatial linkages within the database to be explored, so other map coverages representing a variety of human and natural environmental phenomena, administrative and planning designations, topographic and infrastructural information may similarly be entered into the GIS database and integrated with the archaeological coverages. An SMR set up in this fashion would have almost limitless potential for managing the cultural resource and undertaking regional archaeological analysis. At this point the difference in system capability between the present SMR and a GIS based SMR in terms of functionality and flexibility in performing a range of applications is vast: the distinction between an archaeological information system capable of meeting the needs of society in the 21st century and that of a system struggling to inventory the full archaeological record. Given the increasing corporate nature of information in society, particularly in a planning environment where several departments share and make demands upon the same information sources, it is no surprise to learn that a number of government agencies and county planning authorities are currently considering or building a regional GIS database. The utility companies are already well advanced with such systems. A large number of these activities rely upon having a spatial querying capability because it is the distribution and variation of objects across space which forms the basis for analysis and decision making. The early computerisation of regional SMRs would provide a core around which these geographical information systems could be constructed. A clear distinction should be drawn at this stage between digital mapping or CADCAM systems and GIS (Cowan 1987, Dangermond 1986). We cannot stress this enough as our experiences within the SMR community show a widespread misunderstanding that database plus digital mapping equals GIS. This is not true. GIS do possess considerable mapping capabilities, for maps are the most effective medium for storing and interpreting complex spatial information. But while digital mapping systems have been in existence for many years these systems are without the capability of integrating or analyzing the full spatial range of spatial information. Thus while thematic information may be mapped by these systems, any integration of data must be based upon a common spatial unit or zone, such as the parish or county. Any variation from this geographical unit, for example to overlay and combine differing spatial units

168

25. such as soil type zones with later prehistoric field systems could not be accommodated. Similarly, while other features could be digitally encoded and drawn onto the maps, the systems possess no functional ability to combine, analyze or interrogate the spatial or topological relationships which exist between the features. GIS are an analytical engine with digital mapping being just one of several forms of output. GIS then are specialist computer information systems which enable spatial and thematic information to be digitally encoded, stored, manipulated, retrieved, analyzed and out put in a variety of forms. Handling geographic information within a computerised system has presented a number of problems though recent developments in the storage capacity and processing speed of computers, in tandem with significantly reduced hardware costs have overcome some of these difficulties. These improvements have stimulated the development and use of GIS in a wide range of application areas. Furthermore, the ease of use and the reduced cost of 'off the shelf' specialist GIS software has brought such technology within the purview of many non-specialist users. Already considerable amounts of geographical information, such as archaeological data, satellite imagery, Ordnance Survey topographic maps, and population census data exist in digital form. Together these developments have facilitated the development and application of GIS as a tool for undertaking advanced forms of computerised spatial data handling and analysis.

25.4 User requirements Some of the advantages to the potential user arising from the adoption of a GIS oriented approach to SMR database design have already been mentioned. It is noticeable that while much has been written about the development, structure and organisation of SMRs, the specific needs and information product requirements of the user community have not been as fully addressed. Caution must be exercised when assessing current SMR user requirements because long-standing users of SMR data are well aware of the limitations of SMRs. The nature of current user queries are thus more indicative of the known capabilities and limitations of the system than of their own information needs. To illustrate the shortcomings of the existing database structures and the potential of a fully integrated system we identify two major types of SMR user. The first group comprise ad hoc researchers, whether university, museum or unit based, who have site, period and/or locational specific archaeological interests and information requirements. The second user type involves those in the planning environment with much less specific archaeological interests but with as great a need to be able to manage and conserve the cultural resource and assess the possible impacts arising from development permits being granted. It is the latter group who dominate the SMRs at the moment. In both cases it is difficult to envisage a scenario where users would not be interested in the locational component of the information requested. It seems more than likely that if these needs could be assessed, spatial functionality would be high on their list. Personal experience of a research oriented request for information from an SMR for information concerning all

INTEGRATING SPATIAL INFORMATION IN COMPUTERISED SMRS

prehistoric sites in an area of five by ten kilometres, for example, typically generated a five centimetre thick computer printout and two large photocopies of maps (each covering a five kilometre square) covered in hand-drawn black markings and corresponding PRNs. The user in this case is left to extract the required information from the data contained in the printout and maps. In ess~nce this represents the inventory nature of the SMR rather than that of an information system capability to which we allude. GIS possess the ability to hold many map extents or partition s within its database. Most GIS have edge matching capabilities which enable them to generate apparently seamless map extents. Thus it is possible to generate a window over a particular geographical area and extract information accord ing to the specific areal requirements of the user. Similarly any combination of information within the database could be extracted for these specified areas. It is also virtually impossible at the moment to produce extract maps from the SMRs. The example request for prehistoric sites is likely to produce a map showing blanket coverage of the total SMR for the area in question leaving it up to the researcher to extract the relevant site locations . If the content of the database extended to include altitudinal information or soil type information then recall of archaeological information could as easily be based upon specified combinations of zones generated from these coverages as from the standard tile or OS map sheet. How more apposite it would be to have the capability to search a database using not just thematic criteria but spatial criteria. Thus if planning areas or soil type coverages existed within a system then requests for information combining specified sites falling within a planning zone or a set distance from deposits of clay with flint, or within a designated planning area, or within a certain distance of contemporary urban areas, could be a commonplace occurrence. This scenario does imply the availability of map coverages within the system other than just that of archaeological distributions . Given the high initial cost of data input this is no small concern. Howev er, the corporate nature of much of this information, the need to service many users, and the existing and future availability of vast quantities of geographic information in digital form such as satellite imagery, OS topographic maps, utility company maps, and census information, suggest that this is an achievable scenario now and not just far into the future. Evidence of such applications in North America demonstrably support such a claim. Needless to say computer listings could be generated and digital map output generated automatically by the GIS. To this extent maps could be output without recourse to extensive computer printouts or the photocopying of hand drawn maps. Output could also be sent to a user in digital form via electronic mail or disk storage and provide the basis for subsequent computer analysis. Such a system would be eminently suitable to meet ad hoc archaeological enquiries. Even obtaining copies of the maps currently linked to the SMRs introduce other problems related to the handling of spatial information in the archaeological record. Archaeological sites are currently spatially referenced on these maps by a symbol, the size and extent of which need bear no relationship to the actual size of the site on the ground . A small barrow, for instance, or even a single artefact are

169

GARY R. LOCK AND TREVOR M. HARRIS

archaeological information on the SMR maps, which can be added or altered manually either directly onto the base maps or using overlays, it does apply to the underlying OS topographic base maps which are often years out of date; a last major revision in the 1950s is not unusual. When a new version of a map is released by the OS a situation could arise where the whole SMR would have to be copied by hand from the old to the new maps if the latest topographic information were to be utilised. Further, the digital encoding of these maps is presently being undertaken by the OS to provide national topographic coverage. This digital resource will form a major topographic database for the UK which may be linked with other environmental databases. Without a full locational reference, however, the SMRs will only be able to tap a small part of the potential offered by the availability of this digital map information. The storage of map information in digital form would not only allow easier updating of both archaeological and background information it also reduces the dependency on the OS to produce updates. Local changes in data with a spatial component can be immediately incorporated into the database. Another aspect of SMRs which is severely limited at the moment is integration with other types of environmental data. Archaeology has been painfully slow to integrate with other areas of landscape conservation despite the fact that the archaeological record is a finite and rapidly disappearing resource. The greening of British politics over the last few years has included only a minimal archaeological input. A trend toward a coordinated and integrated approach to the many different strands of landscape conservation could be a major motivating factor for the adoption of GIS in archaeology on a large scale. Much of this integration would undoubtedly occur at the county level and the SMRs could be central to this process. This raises the interesting possibility of SMRs moving away from being independent, self contained units to being a part of a county wide 'corporate database' together with the associated questions of data access and data security. Returning to the earlier example of a planning enquiry, it is important to know whether the proposed route for a new road impacts on any environmentally sensitive areas. The ability to define a linear corridor through the landscape and see whether it includes any SSSis, nature reserves, historic buildings as well as archaeological sites is the level of integration which is now possible with GIS technology. It may be that many SMRs are introduced to this new technology on the coat-tails of county planning departments. GIS certainly offers the opportunity for archaeology to raise its public profile and integrate with not only other environmental databases but with conservation and resource management concerns as well. This potential has not gone unnoticed by some archaeologists working within planning departments where the establishment of GIS based 'Environmental Records' are being investigated. Certainly, GIS based SMRs enabling the integration of the spatial and thematic components of the archaeological record as well as other environmental data offer an attractive view of the future in which improved and more diverse uses of these archaeological databases are thereby encouraged. To capitalise fully on these new opportunities it is essential to be able to digitally record the extent, nature, and shape of

represented by a symbol which is probably hundreds of metres in size if taken to scale. It is the same in reverse for large sites; even if they are shown to size on the map, in the database they are recorded by a single point grid reference. A well preserved late prehistoric field system, for example, which may be many hectares in extent, is carefully drawn onto the OS base maps. In the database, however, it is represented by a single grid coordinate with no indication of shape. The size of a site may be recorded as a separate data category. Even greater problems arise when trying to record linear features because a grid reference of the centroid is completely inappropriate. As well as this inability to describe the basic spatial size and shape of data, current computerised SMRs are also incapable of analyzing spatial relationships. If a later prehistoric linear ditch system connected several field systems and possible contemporary settlements were within 2km of them, these spatial relationships could only be recorded digitally using crude nominal coding within existing computerised SMRs. Not only are the computerised databases unable to provide archaeological information in a form which satisfies many research enquiries but in their present computerised form they also fall short of satisfying the requirements of their major user group. Because of the ever present pressure on the historic environment in the UK from urban and rural development, the main application of SMRs has been geared firmly toward servicing planners and the extensive planning system in the UK. Most enquiries originating from planning authorities are concerned less with the archaeological content of the record per se and more with assessing the significance and possible impact on the cultural resource arising from the granting of planning permission to a developer to build at a given location. Many planning enquiries therefore are spatially oriented and geared toward knowing about what exists, or often what might exist, at a given location or the area immediately adjacent to it. In this case the computerised system is abandoned at the first hurdle to play a secondary role to that of the primary visual inspection of the hand-drawn maps. Through a laborious process of cross referencing the maps with the database it is eventually possible to estimate the total impact of any development and to present that impact in a graphical format. What is needed ideally is the ability to define a spatial unit of any size and shape, produce a map of that area showing spatial archaeological data together with any other spatial data of interest and automatically link sites within that area to the attribute database. Linked to this should be the possibility of modelling the computerised database or establishing a predictive capability based upon multivariate relationships for example, which is not possible at the moment because of the lack of the full archaeological record in the database.

25.5

SMRs in the context of broader developments in information handling

Besides the problems associated with analysis detailed above, there are major problems involved in the management and updating of data kept in manual map form. Maps are notoriously difficult to update and tend to fossilise information. While this does not necessarily apply to the

170

25. each site. A basic premise in archaeology is that human activity involves the ordering and use of space and that such activities are likely to be represented by patterning within the archaeological record. As such, archaeological phenomena are underpinned by their unique position in space and time and by the latent relationships existing between them. The spatial component in archaeological analysis is thus particularly important and yet has been excluded from the SMR computing environment because of the difficulties previously associated with integrating archaeological information which differ in their basic spatial unit. The fundamental decision facing SMRs is whether or not the integration of point, line, polygon and pixel data within a computing environment is important. It seems clear from both user and data management points of view that a move in this direction is necessary if SMRs are to meet future demands. GIS technology has been developing since the 1970s and several systems are now available which will handle the DBMS and the spatial/mapping requirements of SMRs. The cost of such systems are rapidly falling in real terms . Technological limitations determined the structure of earlier SMR database design in the pursuit of a faster computerised version of the manual card index system but now developments are such that a reconsideration of where SMRs are going is necessary. Any system that claims to be computerised and yet excludes a major element of its primary data from the computing environment must be at an early stage of its developmental path. We accept that GIS may still be an unrealistic option given the resources, support and training available to many SMRs but it must be acknowledged that GIS technology is available, that GIS could make a substantial contribution to SMR development, and that they are going to become more extensively involved in all aspects of society and especially in urban and regional planning activities.

25.6 The Chorley Report, GIS and archaeology Discussion of the application of GIS technology to SMRs cannot take place without placing it within the wider context of the diffusion of GIS technology through other sectors of society and archaeology in general. GIS are anticipated to have a major impact on the way in which geographic information is handled by society. The commercial importance of this technology has been recognised by the U. K. government. In 1987 the House of Commons Committee of Enquiry chaired by Lord Chorley published its report into the Handling of Geographic Information (DoE 1987). This enquiry, in turn, had arisen out of the 1983 House of Lords Select Committee report on Remote Sensing and Digital Mapping (H.M.S.O. 1983). These reports and subsequent Government responses (DTI 1984, DoE 1988) emphasise the government's concern to exploit the potential offered by these technological developments. These developments are not unique to the UK but are currently taking place in many countries and especially North America. The uptake of GIS in the UK to date has been limited. In archaeology it has been minimal (Harris & Lock 1990). Much of this

INTEGRATINGSPATIALINFORMATIONIN COMPUTERISEDSMRS

response can be directly attributed to the widespread lack of awareness of GIS capability and of potential application areas. In raising this user awareness the Chorley Report has important implications for the wider GIS user community and not least for archaeologists. The 1987 Commission had no doubt about the importance of GIS and its anticipated impact on society. The development of GIS, the report claimed, was as significant to spatial analysis as, "the invention of the microscope and the telescope were to science, the computer to economics, and the printing press to information dissemination. It is the biggest step forward in the handling of geographic information since the invention of the map." DoE 1987,p. 8 The report highlighted a number of barriers which prevented the full benefits of GIS being obtained. The principal barrier was the lack of user awareness about the central importance of these systems and of their potential benefits (DoE 1987, p. 1). The Committee acknowledged that the falling costs of computer processing power and memory, as well as GIS software, would greatly facilitate the adoption rate ofGIS. The general lack of awareness by many in the UK of GIS technology which the Chorley Report highlighted is particularly germane to this discussion about the possible nature of future SMR developments. This is because of the long term nature of financial investments being made by sponsoring organisations . Investment decisions concerning SMRs undertaken now will largely determine the future shape and direction of SMR development. Given the scarcity of resources such decisions will have to be lived with for some time. For this reason alone awareness of GIS and its potential for SMR development should be recognised by the sponsors and the SMR user community. Decisions concerning the SMRs should at least be made in the full knowledge of these technological developments. Elsewhere we have reviewed a number of possible scenarios for the adoption of GIS within UK archaeology (Harris & Lock 1990). A number of issues were identified concerning raising the awareness of archaeologists to the potential of GIS, as well as about educating and training personnel in their use. What became clear from this analysis was the central role which SMRs could play in this diffusion process. As a result of the rich detailed archaeological record maintained by the SMRs and of the national coverage, any decision by SMRs to adopt, or not to adopt, GIS technology will have major implications for the widespread adoption of this technology by UK archaeologists. As the Chorley Report stressed "The full benefits of sharing geographic information and Geographic Information Systems cannot be realised unless all the potential sharers are aware of them". Thus while some sections of the archaeological community may adopt GIS relatively quickly, the main benefits of adopting this technology would remain limited because of the lack of awareness or rejection of GIS by others and because of the constraints to developing archaeological GIS databases independent of the SMRs. Rejection or prolonged resistance to GIS by the SMRs would certainly affect to a major degree the ability of archaeologists and other archaeological data 171

GARY R. LOCK AND TREVORM. HARRIS

users to tap the full potential of these rich regional databases, undertake full spatial analysis, or exchange archaeological information. The decisions of government agencies such as RCHME and English Heritage could be expected to have an important rOle in this respect because these organisations have to date, and will have in the future, considerable influence over the purchase of SMR hardware and software and the direction of SMR database development. While both the RCHME and English Heritage have considered the Chorley Report and its recommendations in the light of their current working practices, neither organisation is formulating an official response for external publication. A note of optimism does exist from the knowledge that within the RCHME GIS is a 'live issue forming a part of the formulation of a new information strategy document' and may possibly comprise 'the core part of future retrieval systems' (Grant , pers. com.), especially when combined with the recent increase in RCHME funding which is partially earmarked for improving computer capacity, together with their new lead role for SMRs (RCHME 1990a, RCHME 1990b).

25.7

it has not done before, and in many cases does not understand" DoE 1987, p. 158 In the context of GIS and the SMRs, and indeed archaeology in general in the UK at the moment, such comments are particularly apposite . The application of this new technology within SMRs would have a substantial impact, we hesitate to use the word revolution, not only in the ways in which archaeologists handle and use spatial information but in the ways in which archaeological information itself is used within other sectors of society such as planning and cultural resource management.

Acknowledgements We would like to thank Amanda Chadburn for allowing us to use information prior to publication .

Bibliography ALLEN,K. M., S. GREEN,& E. ZUBROW1990. lnlerpreting space : GIS and archaeology. Taylor and Francis, Basingstoke.

Conclusion

SMRs originated as map based systems as part of the OS service and in some respect we are advocating a return to that spatial emphasis. The emphasis of the computerisation of SMRs has been to split the full archaeological record and create databases which mirror the structure of the antecedent card index systems. The spatial element is relegated to a separate storage medium about which little in the published work on SMRs ever seems to be written. Of concern to us is what appears to be an unwritten belief that the two will remain separate. In this paper we have suggested that there is a need for a fundamental shift in archaeological computer database philosophy toward utilisation of the new potential offered by technological developments in GIS. To this end emphasis in database development should be toward the integration of the full archaeological record and of the topological relationships between them within the database. To conclude we return to the Chorley Report and some illuminating evidence given to them by a company long involved in establishing GIS in North America:

BENSON,D. 1974. "A Sites and Monuments Record for the Oxford region", Oxoniensia, 37: 226-37 . BENSON,D. 1985. "Problems of data entry and retrieval", in Bur row, I., (ed.), Counly Archaeological Records: progress and polenlial, pp. 27-34. Association of County Archaeological Officers, Somerset. Boom, B. K. W. 1988. "Site specific data-A standard for data transfer", Archaeological Computing Newsletter, 16: 15-19. BURROUGH,P.A. 1986. Principles of Geographical Information Systems for land resources assessmenl . Clarendon Press, Ox ford. BURROW,I., (ed.) 1985. Counly archaeological records: progress and potenlial. Association of County Archaeological Ofncers, Somerset. CHADBURN,A. 1988. "Approaches to controlling archaeological vocabulary for data retrieval", in Rahtz 1988, pp. 389- 98 . CHADBURN, A. 1989. "Computerised county Sites and Monuments records in England: an overview of their structure, develop ment and progress", in Rahtz, S. P. Q. & Richards, J. D., (eds.), Computer Applications and Quanlitative Methods in Archaeology 1989, International Series 548, pp. 13-24 . British Archaeological Reports, Oxford.

"In dealing with a relatively new technol ogy such as GIS we have found over and over again in North America that the technical problems are minor in comparison with the human ones. The success or failure of a GIS effort has rarely depended on technical factors, and almost always on institutional or managerial ones" DoE 1987, p. 154

COWAN,D. 1987 . "GIS vs. CAD vs. DBMS: What are the differences", in Proceedings of the Second lnlernational Workshop on Geographic Information Systems, GJS '87, Vol. I. American Society for Photogrammetry and Remote Sensing, 45-56, Falls Church, Virginia. DANGERMOND, J. 1986. "CAD vs. GIS" , Computer Graphics World, 9 (10): 73- 74 . DoE 1987. Handling geographic information, Report to the Sec retary of Stale for the Environmenl from the Committee of Enquiry into the Handling of Geographic Information, chaired by Lord Chorley, Departmenl of the Environment . H.M.S .O ., London.

The Report goes on to note, "In short, we believe that the greatest obstacle to greater GIS use will continue to be the human problem of introducing a new technology which requires not only a new way of doing things, but whose main purpose is to permit the agency to do a host of things which

DoE 1988. Handling Geographic Information: the Governmenl's response to the Report of the Committee of Enquiry chaired by Lord Chorley, Deparlmenl of the Environment . H.M.S.0 ., London .

172

25.

INTEGRATINGSPATIALINFORMATIONIN COMPUTERISEDSMRS

DTI 1984. Remote Sensing and Digital Mapping: The Government's reply to the First Report from the House of Lords Select Committee on Science andTechnology,Cmd 9320. Department of Trade and Industry, H.M.S.O., London.

KVAMME,K. L. 1989. "Geographic Information Systems in regional research and data management", in Schiffer, M. B ., (ed.), Archaeological Method and Theory, 1, pp. 139- 203. University of Arizona Press, Tucson.

ENGLISHHERITAGE1987. Rescue archaeology funding in 198788. English Heritage, London.

KVAMME,K. L. & T. A. KOl-ll..ER1988. "Geographic Information Systems: technical aids for data collection, analysis, and displays", in Judge, J. W. & Sebastian, L., (eds.), Quantifying the present and predicting the past: theory, method and application of archaeological predictive modelling, pp. 493-547. U.S. Government Printing Office, Washington D.C.

HARR.Is,T. M. 1986. "Geographic Information System design for archaeological site information retrieval", in Computer Applications in Archaeology 1986, pp. 148-61. University of Birmingham, Birmingham.

McHARG, I. L. 1969. Design with nature. Natural History Press, New York.

HARR.Is,T. M. 1988. "Digital Terrain Modelling and three dimensional surface graphics for landscape and site analysis in archaeology and regional planning", in Ruggles, C. L. N. & Rahtz, S. P. Q., (eds.), Computer and Quantitative Methods in Archaeology 1987, International Series 393, pp. 161-72. British Archaeological Reports, Oxford.

RAH1Z,S. P. Q., (ed.) 1988. Computer and Quantitative Methods in Archaeology 1988, International Series 446, Oxford. British Archaeological Reports. RCHME 1990a. "Extra funds for the RCHME", RCHME Newsletter, 2. RCHME, London.

HARRIS, T. M. & G. R. LocK 1990. ''The diffusion of a new technology: a perspective on the adoption of Geographic In formation Systems within UK archaeology", in Allen, K. M., Green, S., & Zubrow, E., (eds.), lnterpreting space: GIS and Archaeology. Taylor and Francis.

RCHME 1990b. "RCHME's lead role for SMRs", RCHME Newsletter, 2. RCHME, London. STAR,J. & J. ESTES1990. Geographic Information Systems : an introduction. Prentice Hall, New Jersey. TOMUN,C. D. 1990. Geographic Information Systems and cartographic modelling. Prentice Hall, New Jersey.

H.M.S.O. 1983. Remote Sensing and Digital Mapping, Report to the House of Lords Select Committee on Science and Technology, chaired by Lord Shackleton. H.M.S.O., London. 2 volumes.

WALSH,D. 1969. Report of the Committee of Enquiry into the arrangements/or the protection of field monuments 1966-68. HMSO, London. Reprinted 1972, Command 3904.

HOLMAN,N. 1985. "Evaluating the contents of Sites and Mon uments Records: an alternative approach", Archaeological Review from Cambridge, 4 (1): 65-79.

WANSLEEBEN,M. 1988. "Geographic Information Systems in archaeological research", in Rahtz 1988, pp. 433-51.

173

GARY R. LOCK AND TREVOR M. HARRIS

174

26 Terrain modelling, deposit survival and urban archaeology J. D. Richards (Department of Archaeology, University of York, Micklegate House, Micklegate, York, YOJ JJ'Z)

26.1

Introduction

This paper describes an application of computers to urban site evaluation. It is based on work conducted during 198990 by the Department of Archaeology at York University in conjunction with civil engineering contractors Ove Arup for York City Council and English Heritage. The paper first describes the archaeological situation which provides the background to the problem, then looks briefly at the prediction of the deposit model; the rest of the paper concentrates on the computing aspects of the project undertaken in York. The renewed pace of urban development in the late 1980s has led to considerable destruction of archaeological deposits, particularly in the historic cores of many British towns. In 1989 the bulldozing, or piling, of celebrated sites made the headlines in the local and national press, and questions were asked in Parliament. Amongst the most notable examples, recalling the 1954 scandal over the Walbrook Mithraeum, were the Rose Theatre, London, and the Queen's Hotel site, York. As was originally observed in the 1950's, when such cases are presented in the media as confrontation there are no winners. In each case, the developers were forced to meet considerable extra costs, and were portrayed in the press as the villains of the piece, despite the fact that they had originally agreed to fund archaeological work. The archaeologists were made to look incompetent for their apparent inability to predict the presence of significant remains and, in the eyes of the press, appeared to be implicated in the destruction of important sites. Of course, excavators can never know exactly what they will find, although some attempt should be made to find out. Yet the speed of the required response and the nature of archaeological funding have combined to work against this. Urban units have often been forced to adopt an opportunist strategy. The scale of urban renewal has meant that they have had to be extremely selective in their choice of sites to excavate. All too often, this selection has been dictated by the availability of sites for which developer-funding was available, rather than by a research design taking account of the survival and quality of deposits likely to provide a rich information yield. Units have rarely been able to divert resources to an evaluation of urban centres as a whole, which would allow them to place the importance of individual sites in context. Such information is vital if well-informed decisions are to be taken on the research priorities in specific areas. It is particularly relevant with the introduction of contract archaeology and competitive tendering, when fixing the correct price for a job becomes doubly important. Recent commentators, notably Martin Biddle (Biddle 1989, p. 760), have stressed the importance of conducting a full site evaluation before archaeological advice is given.

English Heritage also now underline the need for full site evaluation before development is allowed to proceed (Wainwright 1989, p. 434). Archaeological site evaluation may soon be made a condition of planning consent in a number of cities. In the United States, it is widely accepted that there may be up to three stages of archaeological investigation beforedevelopment(Burrow& Hunter 1990,p.195). Phase 1 surveys are designed to establish if any cultural resources exist; Phase 2 surveys characterise the extent of the site, and demonstrate its significance. Preservation in situ will then always be the first option in the case of an important site. Only in a small number of cases will work proceed to Phase 3 operations, i.e. excavation, where it is not possible to avoid destruction of the deposit.

26.2 The deposit model A major component of any site evaluation has to be an assessment of the depth of surviving archaeological deposits. This is not a particularly new science. Carver has frequently advocated the importance of site evaluation in urban archaeology (e.g. 1978, 1980, 1983, 1987). A number of techniques, including archival study, contour and cellar depth surveys, allow the production of a deposit model. One of the best known archaeological evaluations was conducted for London at the height of the previous Rescue boom (Biddle & Hudson 1973). Maps showing the extent of known destruction from cellars, known thickness of deposits, and proposed developments provided graphic documentation of the threat to the capital's archaeology. Carver, working on several West Midlands towns, including Shrewsbury, Worcester and Stafford, demonstrated the use of mapping to illustrate the survival of relict features. The first systematic deposit-mapping project sponsored by central and local government was undertaken by Carver in Stafford in 1977 (Carver pers comm).

26.3 The York Archaeology Assessment Project The City of York is currently undergoing a development boom. High costs in London are forcing business to relocate in the provinces. York City Council is offering prime sites for redevelopment, many of them on land within the historic core of York. In 1989 they published a portfolio of 35 development sites which was widely circulated to national developers (York City Council 1989). One of these developments was the now infamous Queen's Hotel site which was to be occupied by the headquarters of the new National Curriculum Council. The site

175

J. D. RICHARDS had Iain derelict for fourteen years. When limited excavation revealed that well-preserved Viking remains comparable to those uncovered at Coppergate overlay massive Roman masonry, interpreted by some as belonging to an Imperial Palace, there was enormous media interest, and national calls for development to be halted. In the event most of the archaeology had already been destroyed, and most of the rest went in the bulldozer's bucket, once the fuss had died down, although concrete piles were redesigned to allow some of the principle Roman walls to be preserved in situ. Nevertheless there was concern that this situation should not be allowed to happen again, and a general feeling that it should have been possible to predict the existence of deep waterlogged archaeological deposits in advance. In 1982 Gill Andrews had completed a survey of the archaeology of York, for the DoE as it then was (Andrews 1984). This included maps showing known sites and areas of known destruction, but there had been no attempt to provide a deposit model. The City Council and English Heritage now contracted civil engineers Ove Arup to look at possible engineering solutions , including imaginative foun dation methods, to York's 35 development sites. Ove Arup sub-contracted the archaeological component of the project, including prediction of the deposit model, to the Department of Archaeology at York University.

Within the University the UNIMAP package from UNIRAS was chosen as the most appropriate mapping package for the job (ISG 1988). UNIMAP combines interactive development of maps with powerful two-, three- and fourD mapping techniques. Its default values allow the rapid generation of results, but with further work the package can be customised to specific user requirements. Within York, full colour maps can be developed and displayed at a graphics terminal; A3 full colour output is available on an ink-jet plotter; larger scale line maps may be plotted on an A 1 Calcomp plotter. In addition, the interpolation procedures, having been developed for the oil exploration industry, were thought to be sufficiently robust to deal with the sparse York data. Fixed length data files were output from dBase and input to UNIMAP. The first stage was simply to plot the data values, which cover an area slightly larger than the medieval walled city. UNIMAP allows the user to impose a grid, of any size, over the data, without interpolation . Interpolation may then be used to draw contours . The default UNIRAS method uses a refinement of the bi-linear method (cf. Haigh & Kelly 1987). If the original data points are posted, the black dots give some indication of the reliability of the contours. It is essential to remember the distribution of the original data throughout, as contours will be interpolated even when data is sparse, or non-existent. The maps can be upgraded, however, as more infonnation is added to the database.

26.4 The UNIRAS mapping system In order to provide an evaluation of the archaeology of York the first task was to compile a database of known archaeological observations. The major source was the archive of the York Archaeological Trust (YA'D, including information from excavations and from the numerous watching briefs conducted by YAT staff. Further information was extracted from the City Engineer's bore hole data, as well as from a variety of published sources, including the Royal Commission for Historical Monuments volumes for York. Data were collected on standardised pro-fonna and later input to a PC using dBaseIII+. The choice of database software was largely determined by the need for compatability with the City Council, who would take a copy of the database at the end of the project. The database will eventually provide the basis for the archaeological element of a Geographic Information System in the City Planning Department. Data were collected for five major period divisions: Natural, Roman, Anglian, Anglo-Scandinavian and Medieval. Recording fields included easting, northing, height and thickness of deposit, and information affecting quality of deposits, such as moisture content, degree of disturbance, residuality, anaerobic levels and so on. More information was stored as free-format text, including full description, comments and source data. By the end of the data gathering stage 1087 records had been entered on computer (Price 1989). The main aim of this aspect of the project was the pro duction of a series of five period maps of York, showing the topography at the end of the Prehistoric, Roman, Anglian, Anglo-Scandinavian and Medieval periods.

26.5

Preliminary results and discussion

In this section, the production of the Roman map will be examined in some detail, as a case study, before briefly presenting the other period maps. A number of features can be identified on the Roman period map, including the influence of the natural topography on the position of the Roman road system, bridge across the River Ouse, and legionary fortress (Fig. 26.1). It should be noted that this map was produced by combining the data for the depth of natural, and the depth of Roman levels; in effect adding the Roman deposits to the natural landscape. This technique was used for each of the period maps. If Roman points only were used then a less adequate representation is derived (Fig. 26.2). If a three-D wire-frame model is preferred the view is easily changed (Fig. 26.3). UNIMAP allows the user to select the angle and height of view, and nature of the projection. Archaeologists have long recognised the benefits of presenting survey data in the form of three dimensional surfaces (e.g. Lock 1980); these benefits extend to the study of historic landscapes (Reilly 1988). Users can use a topographic colour scale if they prefer; or define their own. Other variables in the data file, such as the moisture level of deposit, can be plotted as a fourth dimension on a threeD surface. In practice, wetness of deposit appeared to be a rather localised variable, not susceptible to interpolation . More informative maps were derived simply by plotting the data points, coded, for example, according to whether they were anaerobic or not (Fig. 26.4). Other features, such as the Roman fortress walls and road network , can also be

176

26.

TERRAIN MODELLING, DEPOSIT SURVIVAL AND URBAN ARCHAEOLOG Y

overlaid on the on the three-D topographic model of the Roman ground surface (Fig. 26.5). When a number of data sets are compared, the develop ment of York through the ages can be studied as a sequence of maps (Fig. 26.6 , Fig. 26. 1, Figs. 26.7- 26 .9) which can be overlaid on the modern street plan. Dump ing and landscaping is revealed, and the areas of concentrated human activity are identified. Thus one can observe the pattern of the shifting urban topography, from the prehistoric landscape to Eboracum to Eoforwic to Jorvik, as York develops from the Roman capital to Anglo-Saxon entrepot to AngloScandinavian metropolis. Finally, the likely archaeological impact of developments can be assessed by studying the depth of deposits in specific areas (Fig. 26.10). The maps therefore provide a predictive terrain model. Solid colour output is the most visually striking, but the illustrations accompanying this paper demonstrate that useful monochrome output can also be produced. The figures are UNIRAS-generated Postscript files which have been printed on a laser printer. Nevertheless, UNIMAP is a large and expensive mainframe mapping system which was chosen for its power. There are a number of PC-based mapping systems which may be adequate for some applications. For instance, a similar series of two-D maps was produced using the PCbased system, SURFER on an IBM PS2/50, as a pilot study for the York project. SURFER also allows the user to generate 3-D wireframe views. Finally, it is worth emphasising that the York data is not exceptional in its quantity or quality. It should be possible to produce deposit survival maps for many historic centres that have been the subject of extensive archaeological study. As I have attempted to show, the future development of urban archaeology is dependent upon predictive site evaluation.

Bibliography ANDREWS,G. 1984 . "Archaeology in York: an Ass ess ment. A survey prepar ed for the Ancient Monum ents In specto rate of the Department of the Env ironm ent", in Addyman, P. & Black, V., (eds .), Archaeological Papers fr om York pr esented to M. W. Barley, pp . 173-2 08 . York Archaeological Trust, York. BIDDLE,M. 1989. "The Rose reviewed: a comedy(?) Anliquily, 63: 753-60.

of errors",

BIDDLE,M. & D. M. HUDSON1973. The Future of London's Past. RESCUE, Worcester. BURROW,I. & R. HUNTER1990. "Contracting archaeology? Cul tural Resource Management in New Jersey, USA", The Field Archaeologist, 12: 194-200. CARVER,M. 0. H. 1978. "Early Shrewsbury: an archaeological definition in 1975", Transactions of the Shropshire Archaeo logical Society, 59: 225-63. CARVER,M. 0. H., (ed.) 1980 . Medieval Worcester: an archaeo logical framework. Transactions 7. Worcester Archaeological Society. CARVER,M. 0 . H. 1983. "Forty French Towns: an essay on archae logical site evaluation and historical aims", Oxford Journal of Archaeology, 2 (3): 339 - 78 . CARVER,M. 0. H. 1987. Underneath English Towns. B.T. Bats ford, London. HAIGH, J. G. B. & M.A. KELLY1987. "Contouring techniques for archaeological distributions", Journal of Archaeological Science, 14: 231-41. ISG 1988. UN/MAP Manual Version 5, 103. IUCC Information Systems Group. LoCK, G. 1980. "Some archaeological uses of the PICASO Graphics Package", Science and Archaeology, 22: 16-24. PRICE, N. S. 1989. "The York Archaeology Assessment Project 1989-1990: Report on data gathering completed to December 1989; Estimates of interrogation potential for archive records in the City of York". unpublished report .

Acknowledgements The archaeology component of the City of York Archaeology Assessment Project was undertaken in the Department of Archaeology, York University, directed by Professor Martin Carver. Neil Price and Mark Whyman collected and organised the data. Steve Roskams discussed aspects of its interpretation with me. I am also grateful to Rob Fletcher of York University Computing Service for assistance with UNIRAS, and York students Richard Kelly and Paul Miller for assisting in the production of some of the maps. The York deposit survival maps were first conceived in 1988 as an undergraduate dissertation undertaken by Pete Richardson (Richardson 1989).

Data Visualisation: Recenl Advances in the Application of Graphic Systems to Archaeology. Report 185.

REIU.Y, P. 1988.

IBM UK Scientific Centre, Winchester. RICHARDSON,P. 1989 . "An evaluation and assessment of the depth of natural in the City of York and of the SURFER computer application program", B.A. thesis, Department of Archaeology, University of York. WAINWRIGHT, G. J. 1989. "Saving the Rose",Anliquity, 63: 430-

5. YORKCITY COUNCIL1989. York: The Office Portfolio. York Area Economic Development Unit, York City Council, York.

177

J. D. RICHARDS

52400 52200 52000 51800 16.5 15.0 13.5 12.0 10.5 9.0 7.5 6.0

51600 51400 51200

4.5

3.0 1.5

59800 60000 60200 60400 60600 60800 61000 61200 Figure 26.1: Roman York, with position of data points plotted

52400

52200

52000

51800 16.5 15.0 13.5 12.0 10.5

9.0 7.5 6.0 4.5

51600

51400

51200

3.0 1.5

59800

60000

60200

60400 60600

Figure 26.2: Roman York: using Roman data only

178

60800

61000 61200

26.

TERRAIN MODELLING, DEPOSIT SURVIVAL AND URBAN ARCHAEOLOGY

----

ABOVE

15.0-

13.512.010.5-

16.5 16.5 15.0

13.5 12.0 9.0- 10.5 7.5- 9.0 6.07.5 4.56.0 3.04.5 1.5- 3.0 BELOW 1.5

Wt-t ~:1

Figure 26.3: Roman York

52470 52370

•

•

52270

I

52170

;\:::

:'

:•:•

52070

•:•: :-:,

:~.

51970 51770

~

51570 51470

•

• •

•

-

51670

•

·=::::,

::r·

,:,:

••

.{:

•:::•

,:::

51870

••

• \::

,· ::::

51370 51270 51170

-:} ... . ..•:•.

.._,, I.: ~

200

'-

0

z0

100 50 10 0 150

140

130

120

110

100

90

80

70

60

50

40

50

40

year (B.C.) Figure 28.1: Fiesole n

= 1976, max(y) = 700

700

400

-.

'S> .._,, ~

~

200

'0

z0

100 50 10 0 150

140

130

120

110

100

90

80

70

60

year (B.C.) Figure 28.2: Monte Codruzzo n = 4471, max(y)

random process. In this example, the coins being hoarded are all silver denarii and not the bronze denominations also being minted at the time. We also have the advantage that the coins were of a stable weight and fineness during this period. In other periods the selection of coins for hoarding is greatly influenced by the individual coins metallic content and weight. However, if all other factors are equal, the choice of coins for hoarding can be seen to be random selections from the coinage pool (e.g. Thordeman 1948). The program as it stands now was written with a very specific task in mind and therefore will only deal with the period 156 to 50 BC. Output is limited to a listing, and the data summarized as PJCTEXscattergrams for inclusion in k\TEXdocuments. The program is initialised with the simple command simulate. The user is prompted for a number of pieces of information (see Fig. 28.10). Having read in the die data, and the 'introduction delay' factor the program constructs a series of battleship curves for each

= 633

year's coinage which are fed into a two dimensional array. The coinage pool for each year from which coins will be collected is then calculated. The actual hoarding process is then simulated by simple random selection of coins from the pool. In the case of a Type One hoard the total number of coins requested will be collected as a series of random choices from the pool calculated for that year. For a Type Two hoard the opening and closing dates for the hoard are inputted, as well as the total number of coins that will be finally in the hoard. The program then collects the appropriate number of coins randomly from the coinage pool for each year.

28.5 The results As with all simulations the number of possible variations that could be tested is immense. It was decided therefore 197

KRIS LOCKYEAR

150

140

130

70

60

Figure 28.3: Casaleone n = 709, max(y)

= 49

120

110

100

90

80

50

40

year (B.C.) 200

~ ..._,

100

0

150

140

130

120

110

100

90

80

70

60

50

40

50

40

year (B.C.) Figure 28.4: San Giuliano Vecchio n = 1718, max(y)

= 150

400

-.

~ ..._,

~

c...,.

200 100

0

z0

50 10 0 150

140

130

120

110

100

90

80

70

60

year (B.C.) Figure 28.5: Alvignano n

= 2334, max(y) = 307

to concentrate on the effects of three factors, the coinage 'decay rate,' the 'introduction delay,' and the manner of hoarding. Each of these was varied whilst keeping the other factors constant. Results from a number of runs using the same parameters showed that there was remarkably little variation between each run.3 In order to enable a certain amount of comparisonthe size and date of two of the hoards studied were used. At this point I must emphasise a number of points. 1. As the die figures used in this program are derived from the hoards, any similaritybetween the simulated hoards and the real ones may contain a degree of circularity. 2. The simulationwill not explain the factors which produced the hoard structure observed. The simulation by necessity is a simplificationof the real situation,

and it would probably be possible to replicate the observed hoard structure in a number of ways. 3. As a result of the above, it is invalid to attempt any statisticalcomparisonor correlationbetween the real, and the simulatedhoards. It is not, therefore, worth writing the program in such a way that it alters its own parametersuntil it finds the closest 'fit' to a real hoard. 4. This programmustbe seenas a first step in the study of coin hoard fonnation, and not a definitivestatement. Firstly, the effects of altering the period of time taken to collect the hoard was examined. So that the simulated hoards could be comparedwith real hoards, the date and size of two of the hoards were used as parameters. These were the Fiesole hoard, as an extreme example of a hoard with high closing figures,and the San Giuliano Vecchiohoard as

3 The random number generation was checked very carefully in case it was inadequate. The similarity is easily explained when one notes the limited number of possible choices (107) and the large number of selections (minimum in this study of 1716). This effect is increased when the dominance of some issues is noted, and the fact that the coins are plotted in five year groups. Smaller hoards, or those plotted by individual year, show less similarity.

198

28.

SIMULATING COIN HOARD FORMATION

45%

+

o Syracuse • Berchidda ◊ Monte Codru zzo + Fiesole

40% 35%

◊

30% 25% 20% 15% 0 0 0

5% 0

i

0

' '' ~

0

0

~ +

•

•

•

$

• •

+

+ g

s

◊

i

0

~

• •

◊

+

i

~

◊

0 ----....-------.-----r----..-----~---......-----..--------. 150 140 130 120 110

+

•

100

90

80

year (B.C.) Figure 28.6: Hoard Comparisons by Percentage

the hoard which ends closest to 50 B.C. Fiesole has 1976 coins and closed in 89 B.C. and San Giuliano Vecchio had 1716 coins and closed in 48 B.C. A decay rate of 2% with an introduction delay of 5 years was used in these first simulations. The 'long savings' hoards were collected over the whole of the period represented by the hoard. The 'short savings' hoards were collected over the ten years before the hoard closed. The emergency hoard is obviously collected in the closing year of the hoard. Secondly, the effect of changing the decay rates was examined, and finally the effect of altering the 'introduction delay.' For these runs the Fiesole date and total of 1976 coins and 89 B.C. were used. The Type One, emergency hoard, model was used for these. The decay rates tried were ½%, 3% and 6% per annum. The introduction spans were 20 years, 10 years, and 1 year. To illustrate the results of changing the period of time over which the hoard was collected, one set of figures from each of the three simulations for the 'Fiesole' type hoard have been plotted on the same graph with the Fiesole figures (Fig. 28.11) and the same for San Giuliano (Fig. 28.12). Also presented are one set of figures for each of the differing decay rate simulations (Fig. 28.14) and the introduction delay time simulations (Fig. 28.13) along with the actual figures from Fiesole. In Fig. 28.11 it can be seen that the results for the Fiesole simulation unsurprisingly show that the emergency hoard was the nearest to the Fiesole hoard itself, but that even this was not anywhere as high as the real result for Fiesole in its final five year span. The other interesting thing to note is the

differences between the long and the short savings hoard. The very long time span ends up with a very high figures for the first five years. This is despite the 35 million denarii that I set as the coinage in circulation before 156 which is not plotted on these diagrams. 4 The representation of the relative numbers of dies per year is not very good in the long savings hoard; it fails to reflect the rise around 110115. However, the figures for the short savings hoard, the emergency hoard and the real hoard show a similar pattern of rise and fall up until the final five year span, as had been noted when comparing differing real hoards. The generally lower percentage for the real hoard is due to the affect of the very high percentage for the last five year span. The San Giuliano simulation shows a remarkable similarity between the short savings hoard and the San Giuliano hoard itself. Again, the long time span savings hoard has a very odd pattern to it, most unlike any of the real hoards looked at in the dissertation, but the short savings, emergency, and real hoard data are very similar until the last ten years. Again, this is very like comparing a number of real hoards. In figure 28.13 it is interesting to note that the introduction rates used did not greatly affect the pattern of the curve. This is possibly a function of the formula for the curves. However, the battleship curve that is employed so widely is only a theoretical shape for the introduction and decay of coinage, although it seems to be a very likely approximation of the real situation (Collis 1974b). The decay rates are interesting for the way that they affect the results. The figures for the last time bracket vary so

• There are a number of problems with this, but it is used as it is Hopkins' best guess at this figure (Hopkins 1980).

199

KRIS LOCKYEAR

28.6.2 The coinage

widely as a direct result of the fact that the amount of the earlier coinage still in circulation is quite large when the decay rate is low, and vice versa. In figure 28.14 can be seen that the very high figures for Fiesole could be explained by a decay rate of between 3% and 6% and still having the coins as a random selection of those in circulation. Preston's article (1983) suggests a rate somewhat higher than 2%, although his work has some serious flaws. A die study of the Fiesole hoard is really necessary in order to see if the large quantities ofRRC 341/2 are from a limited number of dies or not. For the earlier years, however, the hoards are again remarkably similar apart from the first date bracket. From this we can make a number of tentative assertions.

There are a number of numismatic problems with this work. The foremost is that the die figures used per year are derived from the hoard data itself. Crawford's method has been highly criticised on numismatic and statistical grounds (Mattingly 1977, Burnett 1987, Buttrey 1989, Lockyear 1989). The method used to calculate the figures used here are based on a modification of Crawford's method and removes many of the statistical problems by using regression analysis (Lockyear 1989, section 2.3), but few of the numismatic. Other methods for estimating the number of dies per annum combine a detailed die analysis and a statistical estimate of the number of dies. A variety of formulre have been proposed (e.g. Brown 1957, Esty 1984, Lyon 1965), and these have been compared using artificial data to assess their effectiveness (Esty 1986). A detailed die analysis for all issues of the Republic would be many lifetimes work and in general much reliance has been placed on those issues which have die marks, e.g. C. Calpurnius Piso L. F. Frugi (RRC 408, see Hersch 1976). This topic is currently under further investigation. The general agreement between the real hoards and the simulated ones could be argued as supporting the method of die estimation employed, but I feel that there are too many possible sources of error for this to have much validity. The method of collecting the hoards is also overly simplistic and a variety of other possible methods could be employed.

1. The 'introduction delay' does not effect the overall yearly pattern much in the opening years, and has a comparatively minor effect on the closing years of the hoard, although the difference between the curve used , and the possible 'real' curve must be noted. 2. The decay rate affects the pattern in the closing years quite significantly. This is important if an attempt is being made to try and interpret the hoard has either an 'emergency' or a 'savings' hoard. 3. The method of collection seems to have the greatest effect on the pattern. It seems unlikely, both from a common sense point of view, and as a result of this program, that most hoards were collected over very long time spans. There will of course be exceptions, the most likely being 'temple hoards' where votive deposits seem to have collected over a long period of time.

28.6

28.6.3 The program Following the criticisms of simulation studies in archaeology (Freeman 1988) the seeding of the random number generator could be changed from using the UNIX function time ( ) to manual input. Although some differing random number generators available in the SUN/UNIXlibrary were tried these ought to be compared more systematically. The formula for the battleship curve was a simple approximation and this could be improved along the lines of that used by Herzog and Scholar (1988). The program was also very specific to the period under consideration. Simulate v2 will hopefully be much more widely usable.

Problems, and areas for improvement

As this program was written as a small part of a M.Sc. dissertation it was not possible to test, and refine many factors. These points have to be noted and discussed.

28.6.1 The theoretical basis A criticism concerns the validity of simulation studies. Simulation was once seen as a sophisticated tool for trying to understand many processes (Doran 1970, Hodder 1978). Recently, the use of this technique has been implicitly criticized as being deterministic, scientistic and reductionalist (for a summary see Shanks & Tilley 1989). In this context it can be seen that coinage does generally act as a 'system.' The effect of 'agency' in the formation of hoards is not ignored. Hoards which have been subjected to unusual collection patterns can be seen clearly against the wider patterns. (In another period, the coin hoard associated with the Sutton Hoo ship can be clearly seen to be unusual). In order to work from the data to an understanding of the society which created it, we must have at least an idea of how the data were formed, and in the context of this hoard study simulation is a valid method. This is not to say that all simulations are valid, as with the application of any other tool. It also doesn't mean that the interpretations of the coin evidence in the light of the results has to be within a 'scientistic' framework.

Acknowledgments This article derives from part of the authors dissertation for the degree of M.Sc. submitted to the University of Southampton, (Lockyear 1989). Other sections of this dissertation are in preparation for publication. The author is currently engaged in further research on this topic at the Institute of Archaeology, University College London. The author would like to thank Richard Reece, Stephen Shennan, Sebastian Rahtz and Francis Wenban-Smith for their help. They are, however, not responsible for the ideas contained in this paper.

200

28.

SIMULATINGCOIN HOARD FORMATION

Bibliography

HENDY, F. 1984. Studies in the Byzantine Monetry Economy. Cambridge University Press.

AITCfflSON,N. B. 1988 . "Roman wealth, native ritual: coin hoards within and beyond Roman Britain", World Archaeology, 20 (2): 270-284 .

HERSCH, C. A. 1976. "A Study of the Coinage of the Moneyer C. Calpumius Piso L. F. Frugi", Numismatic Chronicle, 136: 7-63.

BECK, C. W. & S. J. SHENNAN1991. Amber in British Prehistory. Oxbow Books, Oxford.

HERSCH,C. A. 1977. "Notes on the Chronology and Interpretation of the Roman Republican Coinage", Numismatic Chronicle, 137: 19-36.

BROWN, I. D. 1957. "Some Notes on the Coinage of Elizabeth I with special reference to her Hammered Silver ", British Numismatic Journal, 28. BURNETT,A . 1987. "The Changing Face of Republican Numis matics", Journal of Roman Studies, 77: 177-183.

HERzoG, I. & I. SCOLI..AR1988. "A mathematical basis for simulation of seriatable data", in Rahtz, S. P. Q., (ed.), Computer and Quantilative Methods in Archaeology 1988, International Series 446, pp. 53-62. British Archaeological Reports, Oxford.

"Review of Crawford 1985", Classical

HODDER,I., (ed.) 1978. Simulation Studies in Archaeology. Cambridge University Press. New Directions in Archaeology.

CASEY, P. J. 1974. "The Interpretation of Romano-British site Finds", in Casey, P. J. & Reece, R., (eds.), Coins and the Archaeologist, pp. 37-51. British Archaeological Reports. British Series 4.

HODDER,I. & R. REECE 1977. "A Model for the Distribution of Coins in the Western Roman Empire", Journal of Archaeological Science, 4: 1-18.

BUTTREY,T. V. 1989. Philology, 84/1 .

HOPKINS, K. 1980.

"Taxes and Trade in the Roman Empire",

Journal of Roman Studies, 70: 101 -125.

CASEY,P. J. 1980. Coinage In Roman Britain. Shire. CASEY,P. J. 1986. Understanding Ancient Coins. Batsford.

KENT, J.P. C. 1956. "Gold Coinage in the Later Roman Empire", in Carson, R. A. G. & Sutherland, C. H. V., (eds.), Roman Coinage: Essays presented to Harold Mattingly, pp.190-204. Oxford University Press.

CASEY,P. J. & R. REECE, (eds.) 1974. Coins and the Archaeologist. British Archaeological Reports, first edition. British Series No. 4. CASEY,P.J.& R. REECE, (eds.) 1988. Coins and the Archaeologist. Seaby, second edition. COLLIS, J. 1974a. "A Functionlist Approach to pre -Roman Coinage", in Casey, P. J. & Reece, R., (eds.), Coins and the Archaeologist, pp. 1- 11. British Archaeological Reports. British Series 4.

KENT, J.P. C. 1974. "Interpreting Coin Finds", in Casey, P. J. & Reece, R., (eds.), Coins and the Archaeologist, pp. 184-200. British Archaeological Reports. British Series 4.

LoCKYEAR, K. 1989. "A Statistical Investigation of Roman Republican Coin Hoards". M.Sc. Dissertation, University of Southampton. LYON, C. S. S. 1965. "The Estimation of the Number of Dies Employed in a Coinage", Numismatic Circular, 73: 180-1.

COLLIS, J. 1974b. "Data for Dating", in Casey, P. J. & Reece, R., (eds.), Coins and the Archaeologist, pp. 173-183. British Archaeological Reports. British Series 4.

MATTINGLY,H.B. 1977. "Coinage and the Roman State", Numis matic Chronicle, 137: 199-215.

CRAWFORD,M. H. 1969. "Coin Hoards and the Pattern of Violence in the Late Republic", Papers of the British School at Rome, pp. 76-8 1.

OTTAWAY, B. & C. STRAHM1975. "Swiss Neolithic copper beads: currency, ornament or prestige items", World Archaeology, 6: 307-321.

CRAWFORD,M. H. 1970. "Money and Exchange in the Roman World", Journal of Roman Studies, 60.

PATTERSON,C. C. 1972. "Silver Stocks and Losses in Ancient and Modem Times", Economic History Review, 25: 207-10.

CRAWFORD,M. H. 1974. Roman Republican Coinage. Cambridge University Press.

PRESroN, H. S. 1983. "Roman Republican Coin Hoards: an Age Correction and Other Comments", Annali, 30: 83-93.

CRAWFORD,M. H. 1985. Coinage and Money under the Roman Republic. Methuen. CREIGHlUN, J. 1989. Paper delivered to the Conflict or Cooperation conference held at the Univerisity of Oxford Department of Extra -Mural Studies, October 1989.

REECE, R. 1974. "Numerical Aspects of Roman Coin Hoards in Britain", in Casey, P. J. & Reece, R., (eds.), Coins and the Archaeologist, pp. 78- 94. British Archaeological Reports. British Series 4.

CRUMP, T. 1981. The Phenomenon of Money. Kegan Paul, London, Boston and Henley.

Routledge and

REr:.CE,R. 1982. "Roman Coinage in the Western Mediterranean : A Quantitative Approach", Opus, 1: 341-50.

DoRAN, J. 1970. "Systems theory, computer simulations and archaeology", World Archaeology, 1: 289-298.

REECE, R. 1988. "Interpreting Roman hoards", World Archaeol ogy, 20 (2): 261-269.

Esrr, W.W. 1984. "Estimating the Size of a Coinage",Numismatic Chronicle, 144: 180-3.

RENFREW, C. & S. SHENNAN, (eds.) 1982. Ranking, resource

and exchange: aspects of the archaeology of early European society. Cambridge University Press. New Directions in Ar -

Esrr, W.W. 1986. "Estimation of the Size of a Coinage: a Survey and Comparison of Methods.", Numismatic Chronicle, 146: 185-215.

chaeology. SELLWOOD,D. 1963. "Some Experiments in Greek Minting Tech niques", Numismatic Chronicle, 123: 217-31.

FREEMAN,P. 1988. "How to simulate if you must", in Ruggles, C. L. N. & Rahtz, S. P. Q., (eds.), Computer and Quantitative Methods in Archaeology 1987, International Series 393, pp. 139-46. British Archaeological Reports, Oxford. HART, K. 1986. "Heads or Tails? Two sides of the Coin", 21: 637-56.

SHANKS, M. & C. TllLEY 1989. "Archaeology into the 1990's", Norwegian Archaeological Review, 22: 1-12.

Man,

201

THORDEMAN,B. 1948. "The Lohe hoard: a contribution to the methodology of numismatics", NumismaticChronicle,108: 188-204.

KRIS LOCKYEAR

25%

•

0

o Pontecorvo • Maccarese

•

20%

•

15% 0

10% 0 0

5%

0 0 0

0 _.__ _

• •

0

• • • • • 0

0

0

0

• • •

0 0

•

i __,,------=-...---~----r------,----.-----.------.-------,

140

150

120

130

llO

100

90

70

80

year (B.C.) Figure 28.7: Hoard Comparisons by Percentage

25%

•

o Casaleone • Alvignano ◊Avetrana

20%

* San Giuliano

* 0

i

10% 5%5

0

•

◊

15%

◊

8••

*0

i

•••i • ~ i It •t •••f i •~ ◊

~

◊

150 140 130 120 llO

100 90

80

70

◊

0

0

0

•

.

; i

◊

◊

60

◊

* 0

50

◊

40

year (B.C.) Figure 28.8: Hoard Comparisons by Percentage

202

•

0

i

28.

SIMULATINGCOIN HOARD FORMATION

BULLION.-------------------------,

! NUMBER OF DIES

'

NUMBER OF COINS MINTED PER DIE

NEW{Offi DELAY IN RELEASE OF COIN FROMMINT

TOT AL COINAGE POOL

DISTANCE FROM THE DISTRIBUfION POINT(S)

! SPEED OF CIRCULATION

REMINTED?

♦

LOCAL COINAGE POOL

COINS WITHDRAWN? COLLECI1ON FACTORS

TYPE ONE HOARD

TYPE 1WO HOARD

RECOVERY

NON-RECOVERY

+

LATER (ARCHAEOLOGICAL 1) RECOVERY

½

I

HOARDS AVAILABLE FOR STUDY

SITE ANDS

KL/990

Figure 28 .9 : Factors affecting coin hoard data.

203

KRIS LOCKYEAR

kl88a37%

simulate

HOARD FORMATIONSIMULATION PROGRAM Enter

a filename

for output test exists. OK to overwrite? (Y/N) : Enter filename with dies data Introduction delay Decay rate : Output per die: Calculating coinage curves .......... Would you like to simulate A type one (emergency) hoard or a type two (savings) hoard? Enter 1 or 2 : How many coins per year? Starting date (156 - 50) : Ending date (156 - 50) : Error. 156-50. Try again How many times? : PicTeX Output? : Again (Y/N) :

test y

dat 5 2

30000 done.

2 20 100 40 80 4 y

n

Figure 28.10: An Example Run of SIMULATE

45%

o Emergency • Short Savings ◊ Long Savings Ill Fiesole

40% 35% ◊

0

30% 25% 20% 15%

• ◊

10%

5%

• 0

114

0 150

•

~

◊

•

◊

I

◊

• ~

0

0

0

ill

~

Ill

•

B

0

0

• 0

~

0 ~

◊

◊

~

~

•

~

•

~

◊

140

130

120

110

year (B.C.) Figure 28 .11: Fiesole Hoard Simulation

204

•

0

◊

100

I

•

◊

90

28.

SIMULATINGCOIN HOARD FORMATION

30%

o Emergency • Short Savings ◊ Long Savings ~ San Giuliano

25%

• 0

◊

20%

IF

15%

0

e

◊

II!

10%

◊

IF

◊

5%

0

◊

◊

•

~

' 0

150

140

◊ IF

••• • 130

i

~

• •

◊

120

110

•

•

••

◊

◊

◊

g

'

100

i

•

◊

◊

90

80

I

0

IF

Ii

IF

◊

60

70

50

year (B.C.) Figure 28.12: San Giuliano hoard simulation

45%

o Short Introduction • Medium Introduction ◊ Long Introduction IF Fiesole

40%

0

•

35%

◊

30% 25% 20%

15% ~

10%

0

0

•

5%

o ......._ _ __

_ _ __i __ 150

140

__

~

i

0

~

IF

~

....,...... ___ 130

• ~

--,.-___ 120

•e

~

• ~

_ 110

100

year (B.C.) Figure 28.13: Differing Introduction Spans

205

IF

8 90

KRISLOCKYEAR

◊

50% 45%

o

½% per year

• 3% per year ◊6%

40%

per year

111Fiesole

•

35% 30% 25%

0

20% 15% 10% 5% 0

jJ4

0

•

0 0

••

0

••

0

◊

◊

150

140

•

0

0

~

•

I

~

130

•

~

◊

•

~

•

◊

i

i

◊

year (B.C.) Figure 28.14: Differing Decay Rates

206

I

0

110

120

◊

100

90

29 Formal methods for the analysis of archaeological data: data analysis vs expert systems Vanda Vitali (Royal Ontario Museum, Toronto, Canada, and, CN.R.S. -U.P.R. 315, Paris, France)

29.1

Introduction

Over the last two decades the use of computer-basedmethods in archaeologyhas experienceda considerableincrease (Doran & Hodson 1976, Orton 1980, Carr 1985, Ruggles & Rahtz 1988, Rahtz 1988). While at first cautious, the archaeologicalresearch communityhas now acceptedcomputers as one of its essential tools and, as in other fields of scientificendeavor, is exploring the applicationsof new computer-basedmethods in research,teachingand communication. This phenomenonhas resulted from two mutually interrelateddevelopments. On one hand, modem archaeology has become much more oriented towards the systematicaccumulationof data. Numerical data are now generated as a means of describing various purely archaeologicalparameters. Also, technical studies of archaeological material, which are being employed more and more in archaeology, are producing numerical information. On the other hand, the use of computers and computerbased methods has made possiblethe accumulation,storage and processing of large sets of data. Micro computers are now considered an essential part of any archaeological activity. In addition, developments in statistical software have resulted in techniques for the analysis of large data sets. Thus, the generation of more and more systematic data in archaeologyand the necessity to store and evaluate them have fed on each other and created a need for methods (tools and techniques) for the analysis end evaluation of archaeological data. However, these methods shape both the questions and the design of archaeological research. Therefore,as new approachesbecome available and accessible, an assessment of their r0ie and utility becomes of principal importance. The task that they can perform, their potential and their limitations,all have to be considered. This paperexploresthe r0ie of two computer-basedmethods used in archaeological research as formal approaches to the analysis of archaeological data. The two methods which will be under discussionare data analysis and expert systems.

29.2

Definitions

Data analysis can be defined as a set of mathematical/statistical procedures, generally used as computer programs, embracing elementary but particularly multidimensional statistical techniques that require an iterative application in order to statistically process the data and extract informationfrom the data set. This method involvesthe use of mathematical/statisticalrules generally applicable and

207

not subje.ctdependent as procedures for the assessmentof data and the acquisition of new information. An Expert system is a problem-solvingcomputer program utilizing a set of prescriptionsor 'rules' that providea model for the reasoningand thus for the solutionof a specific type of problem in a particular knowledge domain against which a data base pertaining to the same knowledgedomain is evaluated. When dealing with this method it is ne.cessary to distinguish two different aspects of it: the development of an expert system and its use. 1. The development or an expert system involves

the construction of rules from the already existing knowledgedescribing the solution of a specifictypeproblemand the coding of those rules into a computer program. 2. The use oran expert system involves the comparison and assessment of informationregarding a particular case study against the solution model establishedby the constructed rules. Therefore, an expert system method provides a model for reasoning based on existing knowledge in a specific field (i.e. it is subject dependent) for the assessmentof data regarding a case study.

29.3

Data analysis method

In general terms, the most important characteristic of the data analysis method is that it represents essentially an exploratorymode of analysis. The examination/exploration is done on the entire data set and an iterative manner of analysis allows for a relatively flexible way of determining the behaviour that best describes the data in question. Disagreements with the postulated general mathematical behaviour are easily visible and thus attention is drawn to 'misfit' or new information. This capacity to highlightnew informationmakes this method one of the most useful tools for analysis in scientificresearch. Data analysis procedures utilize numerical data (continuous, categoricalor nominal) or data that have been coded into some form of numerical data. Since the principal r0le of the data analysis method is to explore an entire set in order to extract a pattern of behaviour that the data exhibit, the process starts first with the examinationof the initial (input) data set. This set is tested for its validity and quality and then it is screened to determine if there is arly underlying structure of distribution properties for the data, such as normal, logarithmic, non-parametric,etc. On the basis of this information and considering the type of problem that is of archaeologicalinterest, a particular data

VANDA VITALI

analysis technique (protocol) is chosen to yield a general model against which the data is examined. For instance, when dealing with a data set describing several groups of archaeological ceramics in terms of stylistic and compositional variables, and wanting to establish a way of characterizing those groups and discriminating amongst them, an initial set would first be screened for each variable of each group to establish its range of values and its variance. Correlations between variables would then be examined. Finally, distribution properties of each variable would be detennined. On the basis of established characteristics, and in view of the problem, a suitable discriminant analysis procedure would be used to characterize the groups, determine if these groups can be separated from one another, and establish criteria on which the groups might be distinguished. The exploration and examination of the data is done in an iterative manner that allows for a refinement of the description of the behaviour of the data and the extraction of additional and new information from the data (including a rejection of the proposed behaviour) . For example, for the characterization and discrimination of ceramic groups, various parameters used in characterizing and discriminating between groups would be tested · for their validity and efficiency as well as various models for group definitions . In such a process, information about ceramic groups is obtained and refined. As a method, data analysis has been used in archaeological research for over two decades (Djiandjian forthcoming a). It is criticized principally because it relies on numerical information, because it imposes a relatively rigid, mathematical/statistical structure on the analysis, (Doran 1986) and because itis general (i.e. knowledge free) (Doran 1989). These aspects of the data analysis method are sometimes considered ill-suited to the nature of archaeological inves tigation. In response to some of these reservations, recent work has shown that as long as non-numerical descriptors can be unambiguously defined, the coding and subsequent use of this type of data do not represent a problem for the application of data analysis techniques (Djiandjian forthcoming b). Nevertheless , the inability of data analysis method to handle certain type of information, such as description of structures, remains. In terms of the considerations of the distribution properties, linking the distribution properties of the input data to the choice of the evaluative techniqu e may have considerable practical merit. If the data follows a certain type of distribution behaviour, then the models built on this type of behaviour are more likely to describe the archaeological reality with greater veracity. Also, today there are more and more data analysis techniques that use nonparametric approaches and require no assumptions about the distribution properties of the data set under investigation. The fact that this method is general rather than domain specific, can equally represent an advantage. The use of data analysis method, however, requires considerable expertise in mathematics/statistics as well as a knowledge of the archaeological problem.

208

29.4

Expert system method

29.4.1

Development of an expert system

This phase in the expert system approach involves the structuring of existing knowledge regarding a specific subject (problem) into a series of rule-like statements that provide a model of the task that the system is to perform. According to Reichgelt and van Harmelin (1986), expert systems can perform four single, primitive tasks: classification, monitoring, design, and simulation (or any combination of those tasks), and the formalism employed for expressing the constructed rules is to a certain degree task dependent. Expert systems can be applied to both numerical and non-numerical data as well as to incomplete data, an advantage when dealing with archaeological information. For example, when considering the problem of interpreting archaeological sites based on site finds (artifacts and ecofacts) and site features (man-made or natural such as walls, ditches , pits, etc.), a set of IF ... THEN ... statements (rules) is constructed for each type of find and type of feature (and a combination of finds and features) that defines the activity of the site at the time of its occupation {and thus provide a cultural interpretation of the site) in terms of the finds and features considered (Patel & Stutt 1988). This phase of the expert system approach forces the researcher to fully resolve and expose the logic of the arguments (rules) used. As a consequence, it can be extremely valuable for those who perform it as a means of structuring, standardizing and documenting their reasoning. At the same time, however, it is extremely time consuming. Construction of an expert system requires an expertise in the field of application, as well as in computer science.

29.4.2

Use of an expert system

This phase of an expert system approach involves the comparison and assessment of a data set pertaining to a specific case study against the reasoning model established by the constructed rules. When using an expert system, an input data set is not generally initially screened for its validity or for its structure in a way comparable to the first step in data analysis but, rather, data are directly tested against the expert system. For instance, when using an expert system for the inter pretation of archaeological sites on the basis of site finds and features, data on site finds and features characteristic of a specific site are compared to the IF . .. THEN ... rules and conclusions about the activity of the site and its cultural interpretation are reached on the basis of the stated rules. The principal focus of an expert-system-based analysis is on the agreement of the data with the prescribed model rather than on the exploration of the data themselves. Mathematical/statistical procedures can be incorporated into an expert system but then they take a form of rule-like statements and the r0le of these procedures ceases to be exploratory as is the case in data analysis. Finally, the use of an expert system, at least in principle, requires no particular expertise. Expert systems as tools are relatively recent additions to research methodology. Applications of expert system methodology in archaeology are few in number and the

29.

FORMAL METIIODS FOR TI-IEANALYSISOF ARCHAEOLOGICALDATA

and through L. D. Levine, Royal Ontario Museum, Toronto, Canada, is gratefully acknowledged.

r6le of the expert system approach in archaeology has been variously assessed (Wilcock 1985, Huggett & Baker 1985, Doran 1986, Doran 1988, Gardin et al 1987, Lagrange 1988, Vitali & Lagrange 1988, Vitali 1989). The principal general criticism refer to their inability to handle non-monotonic logic and their way of handling incomplete and uncertain information, all which are of importance to archaeology. Perhaps, with further research and the development of expert systems in general, better understanding and solutions to the issues raised by these criticisms will be found.

29.5

Bibliography CARR, C., (ed.) 1985. For Concordance in Archaeological Anal-

ysis: Bridging Dala Structures, Quantitative Technique, and Theory. Westport Publishers, Inc., Fayetteville. in cooperation with the Institute for Quantitative Archaeology, University of Arkansas. DJINDJIAN, F. forthcoming a. "Data Analysis in Archaeology", Science and Archaeology. forthcoming .

Conclusions

DnNDJIAN,F. forthcoming b.

Methodes pour l'Archaeologie.

Sociite Prehistorique Fran~e.

Therefore, it would appear that the two methods being considered here have different r6les in a research process. Data analysis, as a method, employs generally applicable mathematical/statistical techniques in order to explore a data base under consideration. It is essentially an inductive method in a sense that its principal contribution is in deriving a description of the behaviour of the data and thus a description of the problem under study. An expert system provides a model for the reasoning and the solution of a specific problem-type against which a data base is evaluated and, thus, its application is deductive in nature. Its principal contributions are first, in structuring the operational knowledge in a specific field and thus imposing the development of a coherent and formalized framework on the field under study, and second and more importantly, in performing routine analysis tasks in a standardized way. However, the quality of these analysis depends on the quality of the expert system and the previously established domain-specific knowledge. Thus, if a research process is described as a series of steps that begins with a question, continues through the acquisition of materials, and then through the assessment of acquired materials to a final proposition , the data analysis method occupies a place in the assessment of the materials (evidence). The expert system method, on the other hand, takes up the r0le of structuring the final propositions and of performing the routine solution tasks and, thus, does not play a part, strictly speaking, in the research process itself. Both formal methods, as long as they are given their proper r0le, have their place in the archaeological reconstruction of the past.

Paris. forthcoming.

DoRAN, J. 1986. "Formal methods and archaeological theory: a perspective", World Archaeology, 18: 21-37. DoRAN, J. 1988. "Expert systems and archaeology: ahead?", in Ruggles & Rahtz 1988, pp. 237- 241.

what lies

DoRAN, J. 1989. "Personal communication" . DoRAN, J. E. & F. R. HODSON1976. MathemaJics and CompUlers in Archaeology. Harvard University Press, Cambridge. GARDIN, J.C., 0. GUII.LAUME, P. Q.

HF.RMAN, A. HEsNARD, M.S.

LAGRANGE,M. RENAUD, & E. ZADoRA-RIO 1987. System.es experts et sciences humaines : Le cas de l' archeologie. Eyrolles, Paris. HUGGETT, J. & K. BAKER 1985. ''The Computerized Archaeologist: The Development of Expert Systems", Science and

Archaeology, 27: 3-7. LAGRANGE,M. S. 1988. "Expert Systems in Archaeology social sciences: A User 's View", Archeo -log, 3: 7-22.

and

ORTON,C. 1980. Mathematics in Archaeology. Collins, London . PATEL,J. & A. STIJTI 1988. "KIVA: an archaeological interpreter", Technical Report 35, Human Cognition Research Laboratory . RAHTZ, S. P. Q., (ed.) 1988. CompUler and Quantitative Metlwds in Archaeology 1988, International Series 446, Oxford. British Archaeological Reports . REICHGFLT,H. & F. VANHARMEuN 1986. "Criteria for Choosing Representation Languages and Control Regimes for Expert Systems", The Knowledge Engineering Review, 1: 2-17. RUGGLF.S,C. L. N. & S. P. Q . RAHTZ, (eds .) 1988. CompUler and Quantitative Methods in Archaeology 1987, International Series 393, Oxford . British Archaeological Reports . VITALI,V. 1989. "Expert Systems in Archaeometry", in Farquhar, R. M., Hancock, R. G. V., & Pavlish, L. A., (eds.), Proc.

26th lnternalional Archaeoml!try Symposium, University of Toronto, 16-20thMay 1988 .

Acknowledgements

VITALI, V. & M. S. LAGRANGE 1988. "VANDAL: an Expert System for the Provenance Determination of Archaeological Ceramics based on INAA Data", in Rahtz 1988, pp . 369-375.

The author is grateful to F. Djindjian, J. Doran, U. M. Franklin, J.-C. Gardin, M.-S Lagrange and H.-P. Francfort for their comments on the earlier version of the manuscript. The funding for the project, of which this article is one result, obtained through CNRS - UPR 315, Paris, France

WILCOCK,J. 1985. "A Review of Expert Systems: Their Shortcomings and Possible Applications in Archaeology", in CompUler Applications in Archaeology 1985, pp. 139-144, Lon don. Institute of Archaeology, University of London.

209

VANDAVITALI

210

30 Palamede-application civilisations

of expert systems to the archaeology of prehistoric urban

H. P. Francfort (CNRS, Equipe de Recherche no. 315, 23 rue du Maroc, 75940 Paris Cedex 19, France)

single fact), and Ant-hill and Termites (as their descriptive vocabulary is nicely anthropmorphic). Running the rules on such limit cases give interesting results (tables). The final fact-bases (initial fact-base + inferred facts) are consistent with the theories of the authors. They more or less rapidly infer the presence of State and administration from small and/or uncertain data bases. Generally speaking, the permissivity of the reasonings is great because they use concepts borrowed from more solidly founded disciplines of living societies, without proper archaeological evaluation. In order to check if simply going further into further details in the archaeological data before manipulating the socio-cultural concepts of USE can solve the problem, the second module of the doxography PALAMEDE,CIV, was built. The CIV module starts by analysing an article (Jacobson 1986) upon the State nature of the Indus Civilisation. The author's details are put into the archaeological data-base. Therefore, the PALAMEDEfact-base extracted (CIV$1ND) is purely archaeological and 43 rules model the reasoning of the author's conclusions about the presence of the State; this rule-based is named CIV$JAC (Table 30.2). This archaeological rule-base, tested with the sociocultural facts-base PAYS-TEST of module USE, remains inactive; reciprocally, the socio-cultural rules of USE do not react to the archaeological facts of CIV$1ND. Naturally, the results of running CIV$JAC on CIV$IND are very consistent with the author's conclusions and the final factsbase (called USE$IND) is consistent with the socio-cultural universe of USE. The run of the various USE rule-bases on the final facts-base USE$IND works perfectly. In order to test the permissivity of the CIV$JAC reasons or its restrictiveness, a facts-base at the limits was submitted to the rules: the Chalcolithic Civilisation of Palestine, extracted from a recent article where the authors conclude the presence of a chiefdom, not a State. Simply using the sociocultural conclusions of the authors (facts-base USE$PALO), 4 of the USE rule-bases infer a State. Using the archaeological data-base of the authors (facts-base CIV$PAL), the Indus reasoning CIV$JAC infers 29 or 18 new facts (if including as untrue 'craftsmen quarters' in the original facts-base). The final facts-base is, again, submitted to the socio-cultural rule-bases of USE, producing various sorts of State formations. Therefore, we can conclude that with almost any archaeological (but not only) data base, it is always possible to link data to concepts in order to produce plausible socio-cultural interpretations and that these interpretations can be easily used to infer the existence of State formations. It is just a matter of measure. The actual archaeological correlates of State formations have not been discovered because of

The PALAMEDEsystem is designed to evaluate the question of urbanisation supposed to reflect the Early State societies. In the East, between the 4th and the 2nd millenium, urbanisation is usually a clear concept, when cities are well developed and when a writing system keeps the administrative records. However, in many cases, like the incipient periods, or the in periphery of urbanised regions, the identification of the socio-economic system is not so easy. As we were working on such problems in Central Asia and Northern India, PALAMEDE has been conceived to help the researcher in some of such ambiguous cases. It has been written in SNARK,an inference engine developed by J. L. Lauri~re. It is the result of a collaboration with M. S. Lagrange and M. Renaud. PALAMEDE,in a first part called 'doxography' (meaning 'description of opinion'), models various types of reasoning about the State process, at a high interpretative level: socio-economic, socio-cultural, political. In a second part called 'physiography' (meaning 'the description of nature'), PALAMEDE concentrates upon economic questions by measures and calculations upon material remains, approaching the socio-economic level only through a meta-terminology. PALAMEDE is a complex of six modules (Fig. 30.1), running independently for simplicity reasons (a single 'expert' would have been quite heavy). The 'doxography' groups 2 modules; USE (Urbanisation, Socittt, Etat) and CIV (Civilisation). The 'physiography' groups 4 modules: TOP (Topography of activities), TEC (Technology of artefacts), ARC (Architectures) and SYN (Synthese of conclusions). The doxography, in its first module (USE), analyses (by Gardin 's logicist analysis and H. Wrights 's schematisations - Fig. 30.2) and models 10 sets of reasoning appertaining to existence of the State. The knowledge representation enumerates 140 concepts of the type: Society X Society X

Organisation.Soc Goods

Stratified Redistributed etc ...

written in the ternary SN ARK format. These concepts are all extracted from the inferences used by authors in their reasonings (Whittfogel, Diakonoff, Carneiro, Adams, Wright etc.) They are used in the rulebases of PALAMEDE. Many of these concepts are repetitive or synonyms up to a point that only 16 concepts are sufficient for triggering all the inferences of all the reasons analysed. This core of 16 concepts has been called the facts-base PAYS-TEST (Test-country) (Tobie 30.1). Besides the TEST-COUNTRY, four facts-bases have been constructed by analysing cases at the limits in order to test the good will of the reasoning analysed: Europe in the Iron Age, the Neolithic in Wessex, Large Scale Irrigation (a

211

H.

P. FRANCFORT MODULE

TOP MODULE CIV

MODULE TEC

MODULE USE MODULE ARC

TEXTES

DONNEESARCHAEOLOGIQUES

Figure 30.1: Schema for PALAMEDE

_____ War1are

War leaders

------War captives

Increased craft

as slaves

production

State

Herding

Collecting

l--t-------3

--i

Redistributional

Managerial 1-----,;-

institutions

officials

Produdive differentials

Wealth

Land

concentration

purchase

Hydraulic agricultlKal .._______

1----~--i

between trads and grol4)5

Figure 30.2: Theory of the origin of the state according to Adams after Wright 1978)

The 'physiography' groups the four modules: 10P, TEC, ARC, SYN. The module 10Ppography of activities analyses the intra-site spatial distribution of artefacts measured by 3 (meta) functions; 'domestic, craft, prestige'. Successive periods are compared 2 by 2. The archaeologist writes the fact-base TOP£SHOR listing all the finds in their architectural context with indications about the origin of raw material; the Shortughai base has 600 lines. The rule-base TOP£SHOR is founded on the simple principle of cumulative meters. The 100 rules of TOP attribute, according to the evaluation by the archaeologist, a value to the finds in tenns of their explicitly supposed 'domestic, craft or prestige' function. The values are accumulated by places (95) and periods (9) and synthethised by phases (2). The number and percentage of places of e.ach function at each phase are given. The results are written in words by PALAMEDE, such as:

three frequently ambiguous and tautological concepts: very big urban population concentration, very large territory of an archaeological culture, and a developed and readable writing system. The macro-problem is: under what conditions can we transfer the attributes of living societies to the archaeological material data and validate it? The second part of PALAMEDE - 'physiography' - is more constructive. It applies mostly to the data from a protohistoric site we excavated: Shortughai in N .E. Afghanistan. The 'physiography' goes back to the archaeological data, considering mainly the economy of production. This part of PALAMEDEis devoted to the relative comparative measure of successive stages of production economy under intrasite conditions. PALAMEDEbuilds conclusions by using for criteria a number of 'meta-notions' in a specific field (local). All the meta-notations are build upon evaluations of archaeological ~ at the most basic level. A meta-notation for example is the 'sophistication of craft production• which is precisely calculated in a limited sense and differs therefore from the common use concept.

212

PALAMEDE -

30.

BR FINALE NUAG ADAM CARN TOCH WIDI WRJO

EXPERT SYSTEMS AND PREHISTORIC URBAN CIVILISATIONS

NREGLES

NDECLEN

BF

73 15

72 13 10 16 13 18

62 30 26 26 28 33

11

27 15 23 Table 30.1:

REGLE: JACOBS I NATURE SI NATURE DIMENSION OBJ QUALIF POPULATION ALORS FR

CIVILISATION SITE GRAND SOPl-ll URBAN URBAN

(X)

(Y) (Y) (Y) (Y) (X)

REGLE: JACOBS2 NATURE SI NATURE DIMENSION QUALIF INCLUT QUALIF NOMBRE CULTURE ALORS POP -URB ETAT-POLIT FR